summaryrefslogtreecommitdiffstats
path: root/src
diff options
context:
space:
mode:
Diffstat (limited to 'src')
-rw-r--r--src/CMakeLists.txt1
-rw-r--r--src/audio_core/audio_renderer.cpp13
-rw-r--r--src/audio_core/audio_renderer.h8
-rw-r--r--src/audio_core/cubeb_sink.cpp1
-rw-r--r--src/audio_core/stream.cpp12
-rw-r--r--src/common/CMakeLists.txt5
-rw-r--r--src/common/bit_set.h99
-rw-r--r--src/common/concepts.h4
-rw-r--r--src/common/multi_level_queue.h345
-rw-r--r--src/common/scope_exit.h2
-rw-r--r--src/common/telemetry.h4
-rw-r--r--src/common/thread_worker.cpp58
-rw-r--r--src/common/thread_worker.h30
-rw-r--r--src/common/x64/xbyak_abi.h20
-rw-r--r--src/core/CMakeLists.txt17
-rw-r--r--src/core/arm/dynarmic/arm_dynarmic_32.cpp3
-rw-r--r--src/core/arm/dynarmic/arm_dynarmic_64.cpp5
-rw-r--r--src/core/core.cpp34
-rw-r--r--src/core/core.h20
-rw-r--r--src/core/cpu_manager.cpp98
-rw-r--r--src/core/file_sys/common_funcs.h56
-rw-r--r--src/core/file_sys/content_archive.cpp31
-rw-r--r--src/core/file_sys/content_archive.h8
-rw-r--r--src/core/file_sys/nca_patch.cpp2
-rw-r--r--src/core/file_sys/nca_patch.h2
-rw-r--r--src/core/file_sys/patch_manager.cpp4
-rw-r--r--src/core/file_sys/romfs_factory.cpp22
-rw-r--r--src/core/file_sys/romfs_factory.h4
-rw-r--r--src/core/file_sys/system_archive/data/font_nintendo_extended.cpp555
-rw-r--r--src/core/file_sys/system_archive/data/font_nintendo_extended.h2
-rw-r--r--src/core/file_sys/system_archive/system_version.cpp12
-rw-r--r--src/core/file_sys/vfs.cpp32
-rw-r--r--src/core/file_sys/vfs.h44
-rw-r--r--src/core/file_sys/vfs_concat.cpp18
-rw-r--r--src/core/file_sys/vfs_concat.h2
-rw-r--r--src/core/file_sys/vfs_layered.cpp24
-rw-r--r--src/core/file_sys/vfs_layered.h18
-rw-r--r--src/core/file_sys/vfs_offset.cpp4
-rw-r--r--src/core/file_sys/vfs_offset.h6
-rw-r--r--src/core/file_sys/vfs_real.cpp24
-rw-r--r--src/core/file_sys/vfs_real.h24
-rw-r--r--src/core/file_sys/vfs_static.h2
-rw-r--r--src/core/file_sys/vfs_vector.cpp12
-rw-r--r--src/core/file_sys/vfs_vector.h26
-rw-r--r--src/core/file_sys/xts_archive.cpp6
-rw-r--r--src/core/file_sys/xts_archive.h6
-rw-r--r--src/core/frontend/applets/error.cpp7
-rw-r--r--src/core/frontend/applets/general_frontend.cpp68
-rw-r--r--src/core/frontend/applets/general_frontend.h51
-rw-r--r--src/core/frontend/applets/web_browser.cpp24
-rw-r--r--src/core/frontend/applets/web_browser.h20
-rw-r--r--src/core/frontend/input_interpreter.cpp45
-rw-r--r--src/core/frontend/input_interpreter.h120
-rw-r--r--src/core/hle/ipc_helpers.h67
-rw-r--r--src/core/hle/kernel/address_arbiter.cpp21
-rw-r--r--src/core/hle/kernel/global_scheduler_context.cpp52
-rw-r--r--src/core/hle/kernel/global_scheduler_context.h81
-rw-r--r--src/core/hle/kernel/handle_table.cpp4
-rw-r--r--src/core/hle/kernel/hle_ipc.cpp41
-rw-r--r--src/core/hle/kernel/hle_ipc.h17
-rw-r--r--src/core/hle/kernel/k_affinity_mask.h58
-rw-r--r--src/core/hle/kernel/k_priority_queue.h451
-rw-r--r--src/core/hle/kernel/k_scheduler.cpp784
-rw-r--r--src/core/hle/kernel/k_scheduler.h201
-rw-r--r--src/core/hle/kernel/k_scheduler_lock.h75
-rw-r--r--src/core/hle/kernel/k_scoped_lock.h41
-rw-r--r--src/core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h50
-rw-r--r--src/core/hle/kernel/kernel.cpp174
-rw-r--r--src/core/hle/kernel/kernel.h34
-rw-r--r--src/core/hle/kernel/memory/memory_block.h6
-rw-r--r--src/core/hle/kernel/memory/memory_block_manager.h4
-rw-r--r--src/core/hle/kernel/mutex.cpp12
-rw-r--r--src/core/hle/kernel/physical_core.cpp8
-rw-r--r--src/core/hle/kernel/physical_core.h15
-rw-r--r--src/core/hle/kernel/process.cpp14
-rw-r--r--src/core/hle/kernel/process.h13
-rw-r--r--src/core/hle/kernel/process_capability.cpp2
-rw-r--r--src/core/hle/kernel/readable_event.cpp4
-rw-r--r--src/core/hle/kernel/resource_limit.cpp4
-rw-r--r--src/core/hle/kernel/scheduler.cpp819
-rw-r--r--src/core/hle/kernel/scheduler.h320
-rw-r--r--src/core/hle/kernel/server_session.cpp36
-rw-r--r--src/core/hle/kernel/server_session.h12
-rw-r--r--src/core/hle/kernel/service_thread.cpp110
-rw-r--r--src/core/hle/kernel/service_thread.h28
-rw-r--r--src/core/hle/kernel/svc.cpp140
-rw-r--r--src/core/hle/kernel/synchronization.cpp11
-rw-r--r--src/core/hle/kernel/thread.cpp79
-rw-r--r--src/core/hle/kernel/thread.h114
-rw-r--r--src/core/hle/kernel/time_manager.cpp17
-rw-r--r--src/core/hle/service/am/am.cpp12
-rw-r--r--src/core/hle/service/am/applets/applets.cpp35
-rw-r--r--src/core/hle/service/am/applets/applets.h20
-rw-r--r--src/core/hle/service/am/applets/controller.cpp23
-rw-r--r--src/core/hle/service/am/applets/error.cpp4
-rw-r--r--src/core/hle/service/am/applets/general_backend.cpp6
-rw-r--r--src/core/hle/service/am/applets/web_browser.cpp792
-rw-r--r--src/core/hle/service/am/applets/web_browser.h80
-rw-r--r--src/core/hle/service/am/applets/web_types.h178
-rw-r--r--src/core/hle/service/aoc/aoc_u.cpp88
-rw-r--r--src/core/hle/service/aoc/aoc_u.h2
-rw-r--r--src/core/hle/service/apm/controller.cpp3
-rw-r--r--src/core/hle/service/apm/interface.cpp7
-rw-r--r--src/core/hle/service/audio/audout_u.cpp6
-rw-r--r--src/core/hle/service/audio/audren_u.cpp14
-rw-r--r--src/core/hle/service/bcat/backend/boxcat.cpp2
-rw-r--r--src/core/hle/service/fatal/fatal.cpp5
-rw-r--r--src/core/hle/service/filesystem/filesystem.cpp41
-rw-r--r--src/core/hle/service/filesystem/filesystem.h4
-rw-r--r--src/core/hle/service/filesystem/fsp_srv.cpp57
-rw-r--r--src/core/hle/service/filesystem/fsp_srv.h1
-rw-r--r--src/core/hle/service/friend/friend.cpp3
-rw-r--r--src/core/hle/service/hid/controllers/npad.cpp43
-rw-r--r--src/core/hle/service/hid/controllers/npad.h7
-rw-r--r--src/core/hle/service/hid/hid.cpp2
-rw-r--r--src/core/hle/service/lm/lm.cpp2
-rw-r--r--src/core/hle/service/ncm/ncm.cpp2
-rw-r--r--src/core/hle/service/nim/nim.cpp14
-rw-r--r--src/core/hle/service/ns/ns.cpp11
-rw-r--r--src/core/hle/service/ns/pl_u.cpp39
-rw-r--r--src/core/hle/service/ns/pl_u.h19
-rw-r--r--src/core/hle/service/nvdrv/devices/nvdevice.h9
-rw-r--r--src/core/hle/service/nvdrv/devices/nvdisp_disp0.cpp9
-rw-r--r--src/core/hle/service/nvdrv/devices/nvdisp_disp0.h8
-rw-r--r--src/core/hle/service/nvdrv/devices/nvhost_as_gpu.cpp9
-rw-r--r--src/core/hle/service/nvdrv/devices/nvhost_as_gpu.h8
-rw-r--r--src/core/hle/service/nvdrv/devices/nvhost_ctrl.cpp20
-rw-r--r--src/core/hle/service/nvdrv/devices/nvhost_ctrl.h11
-rw-r--r--src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.cpp8
-rw-r--r--src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.h8
-rw-r--r--src/core/hle/service/nvdrv/devices/nvhost_gpu.cpp8
-rw-r--r--src/core/hle/service/nvdrv/devices/nvhost_gpu.h8
-rw-r--r--src/core/hle/service/nvdrv/devices/nvhost_nvdec.cpp9
-rw-r--r--src/core/hle/service/nvdrv/devices/nvhost_nvdec.h8
-rw-r--r--src/core/hle/service/nvdrv/devices/nvhost_nvdec_common.h33
-rw-r--r--src/core/hle/service/nvdrv/devices/nvhost_nvjpg.cpp9
-rw-r--r--src/core/hle/service/nvdrv/devices/nvhost_nvjpg.h8
-rw-r--r--src/core/hle/service/nvdrv/devices/nvhost_vic.cpp8
-rw-r--r--src/core/hle/service/nvdrv/devices/nvhost_vic.h8
-rw-r--r--src/core/hle/service/nvdrv/devices/nvmap.cpp8
-rw-r--r--src/core/hle/service/nvdrv/devices/nvmap.h8
-rw-r--r--src/core/hle/service/nvdrv/interface.cpp92
-rw-r--r--src/core/hle/service/nvdrv/nvdata.h11
-rw-r--r--src/core/hle/service/nvdrv/nvdrv.cpp13
-rw-r--r--src/core/hle/service/nvdrv/nvdrv.h6
-rw-r--r--src/core/hle/service/nvflinger/buffer_queue.cpp152
-rw-r--r--src/core/hle/service/nvflinger/buffer_queue.h20
-rw-r--r--src/core/hle/service/nvflinger/nvflinger.cpp34
-rw-r--r--src/core/hle/service/nvflinger/nvflinger.h9
-rw-r--r--src/core/hle/service/pcie/pcie.cpp2
-rw-r--r--src/core/hle/service/prepo/prepo.cpp4
-rw-r--r--src/core/hle/service/service.cpp30
-rw-r--r--src/core/hle/service/service.h16
-rw-r--r--src/core/hle/service/set/set_sys.cpp4
-rw-r--r--src/core/hle/service/sockets/blocking_worker.h161
-rw-r--r--src/core/hle/service/sockets/bsd.cpp138
-rw-r--r--src/core/hle/service/sockets/bsd.h9
-rw-r--r--src/core/hle/service/sockets/sockets.h23
-rw-r--r--src/core/hle/service/sockets/sockets_translate.cpp62
-rw-r--r--src/core/hle/service/sockets/sockets_translate.h4
-rw-r--r--src/core/hle/service/time/time.cpp2
-rw-r--r--src/core/hle/service/vi/vi.cpp75
-rw-r--r--src/core/loader/deconstructed_rom_directory.h2
-rw-r--r--src/core/loader/elf.h2
-rw-r--r--src/core/loader/kip.h2
-rw-r--r--src/core/loader/nax.h2
-rw-r--r--src/core/loader/nca.h2
-rw-r--r--src/core/loader/nro.h2
-rw-r--r--src/core/loader/nso.h2
-rw-r--r--src/core/loader/nsp.h2
-rw-r--r--src/core/loader/xci.h2
-rw-r--r--src/core/memory.cpp54
-rw-r--r--src/core/network/network.cpp62
-rw-r--r--src/core/network/network.h24
-rw-r--r--src/core/network/sockets.h4
-rw-r--r--src/core/settings.cpp7
-rw-r--r--src/core/settings.h5
-rwxr-xr-xsrc/input_common/analog_from_button.cpp22
-rw-r--r--src/input_common/gcadapter/gc_poller.cpp28
-rw-r--r--src/input_common/mouse/mouse_poller.cpp25
-rw-r--r--src/input_common/sdl/sdl_impl.cpp26
-rw-r--r--src/input_common/udp/client.cpp5
-rw-r--r--src/tests/CMakeLists.txt1
-rw-r--r--src/tests/common/multi_level_queue.cpp55
-rw-r--r--src/video_core/CMakeLists.txt192
-rw-r--r--src/video_core/buffer_cache/buffer_block.h19
-rw-r--r--src/video_core/buffer_cache/buffer_cache.h21
-rw-r--r--src/video_core/buffer_cache/map_interval.h3
-rw-r--r--src/video_core/cdma_pusher.cpp30
-rw-r--r--src/video_core/cdma_pusher.h16
-rw-r--r--src/video_core/command_classes/codecs/codec.cpp4
-rw-r--r--src/video_core/command_classes/codecs/vp9.cpp42
-rw-r--r--src/video_core/command_classes/codecs/vp9.h2
-rw-r--r--src/video_core/command_classes/vic.cpp10
-rw-r--r--src/video_core/compatible_formats.cpp142
-rw-r--r--src/video_core/compatible_formats.h23
-rw-r--r--src/video_core/delayed_destruction_ring.h32
-rw-r--r--src/video_core/dirty_flags.cpp9
-rw-r--r--src/video_core/dirty_flags.h3
-rw-r--r--src/video_core/dma_pusher.cpp9
-rw-r--r--src/video_core/dma_pusher.h14
-rw-r--r--src/video_core/engines/engine_upload.cpp8
-rw-r--r--src/video_core/engines/engine_upload.h4
-rw-r--r--src/video_core/engines/fermi_2d.cpp90
-rw-r--r--src/video_core/engines/fermi_2d.h331
-rw-r--r--src/video_core/engines/kepler_compute.cpp26
-rw-r--r--src/video_core/engines/kepler_compute.h5
-rw-r--r--src/video_core/engines/kepler_memory.cpp4
-rw-r--r--src/video_core/engines/kepler_memory.h2
-rw-r--r--src/video_core/engines/maxwell_3d.cpp56
-rw-r--r--src/video_core/engines/maxwell_3d.h194
-rw-r--r--src/video_core/engines/maxwell_dma.cpp9
-rw-r--r--src/video_core/engines/maxwell_dma.h16
-rw-r--r--src/video_core/engines/shader_bytecode.h6
-rw-r--r--src/video_core/fence_manager.h25
-rw-r--r--src/video_core/framebuffer_config.h31
-rw-r--r--src/video_core/gpu.cpp95
-rw-r--r--src/video_core/gpu.h67
-rw-r--r--src/video_core/gpu_asynch.cpp86
-rw-r--r--src/video_core/gpu_asynch.h47
-rw-r--r--src/video_core/gpu_synch.cpp61
-rw-r--r--src/video_core/gpu_synch.h41
-rw-r--r--src/video_core/gpu_thread.cpp50
-rw-r--r--src/video_core/gpu_thread.h33
-rw-r--r--src/video_core/guest_driver.h4
-rw-r--r--src/video_core/host_shaders/CMakeLists.txt67
-rw-r--r--src/video_core/host_shaders/block_linear_unswizzle_2d.comp122
-rw-r--r--src/video_core/host_shaders/block_linear_unswizzle_3d.comp125
-rw-r--r--src/video_core/host_shaders/convert_depth_to_float.frag13
-rw-r--r--src/video_core/host_shaders/convert_float_to_depth.frag13
-rw-r--r--src/video_core/host_shaders/full_screen_triangle.vert29
-rw-r--r--src/video_core/host_shaders/opengl_copy_bc4.comp70
-rw-r--r--src/video_core/host_shaders/opengl_present.frag4
-rw-r--r--src/video_core/host_shaders/opengl_present.vert4
-rw-r--r--src/video_core/host_shaders/pitch_unswizzle.comp86
-rw-r--r--src/video_core/host_shaders/vulkan_blit_color_float.frag14
-rw-r--r--src/video_core/host_shaders/vulkan_blit_depth_stencil.frag16
-rw-r--r--src/video_core/host_shaders/vulkan_present.frag (renamed from src/video_core/renderer_vulkan/shaders/blit.frag)9
-rw-r--r--src/video_core/host_shaders/vulkan_present.vert (renamed from src/video_core/renderer_vulkan/shaders/blit.vert)9
-rw-r--r--src/video_core/host_shaders/vulkan_quad_array.comp (renamed from src/video_core/renderer_vulkan/shaders/quad_array.comp)9
-rw-r--r--src/video_core/host_shaders/vulkan_quad_indexed.comp (renamed from src/video_core/renderer_vulkan/shaders/quad_indexed.comp)9
-rw-r--r--src/video_core/host_shaders/vulkan_uint8.comp (renamed from src/video_core/renderer_vulkan/shaders/uint8.comp)9
-rw-r--r--src/video_core/macro/macro_hle.cpp6
-rw-r--r--src/video_core/macro/macro_hle.h2
-rw-r--r--src/video_core/macro/macro_interpreter.cpp27
-rw-r--r--src/video_core/macro/macro_interpreter.h10
-rw-r--r--src/video_core/macro/macro_jit_x64.cpp21
-rw-r--r--src/video_core/macro/macro_jit_x64.h4
-rw-r--r--src/video_core/memory_manager.cpp5
-rw-r--r--src/video_core/memory_manager.h4
-rw-r--r--src/video_core/morton.cpp250
-rw-r--r--src/video_core/morton.h18
-rw-r--r--src/video_core/query_cache.h8
-rw-r--r--src/video_core/rasterizer_interface.h12
-rw-r--r--src/video_core/renderer_opengl/gl_arb_decompiler.cpp60
-rw-r--r--src/video_core/renderer_opengl/gl_buffer_cache.cpp33
-rw-r--r--src/video_core/renderer_opengl/gl_buffer_cache.h12
-rw-r--r--src/video_core/renderer_opengl/gl_device.cpp64
-rw-r--r--src/video_core/renderer_opengl/gl_device.h13
-rw-r--r--src/video_core/renderer_opengl/gl_fence_manager.cpp14
-rw-r--r--src/video_core/renderer_opengl/gl_fence_manager.h12
-rw-r--r--src/video_core/renderer_opengl/gl_framebuffer_cache.cpp85
-rw-r--r--src/video_core/renderer_opengl/gl_framebuffer_cache.h68
-rw-r--r--src/video_core/renderer_opengl/gl_query_cache.cpp24
-rw-r--r--src/video_core/renderer_opengl/gl_query_cache.h12
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer.cpp514
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer.h71
-rw-r--r--src/video_core/renderer_opengl/gl_resource_manager.cpp2
-rw-r--r--src/video_core/renderer_opengl/gl_sampler_cache.cpp52
-rw-r--r--src/video_core/renderer_opengl/gl_sampler_cache.h25
-rw-r--r--src/video_core/renderer_opengl/gl_shader_cache.cpp16
-rw-r--r--src/video_core/renderer_opengl/gl_shader_cache.h11
-rw-r--r--src/video_core/renderer_opengl/gl_shader_decompiler.cpp55
-rw-r--r--src/video_core/renderer_opengl/gl_shader_decompiler.h16
-rw-r--r--src/video_core/renderer_opengl/gl_shader_disk_cache.cpp2
-rw-r--r--src/video_core/renderer_opengl/gl_shader_manager.cpp15
-rw-r--r--src/video_core/renderer_opengl/gl_shader_manager.h6
-rw-r--r--src/video_core/renderer_opengl/gl_state_tracker.cpp9
-rw-r--r--src/video_core/renderer_opengl/gl_state_tracker.h15
-rw-r--r--src/video_core/renderer_opengl/gl_stream_buffer.cpp32
-rw-r--r--src/video_core/renderer_opengl/gl_stream_buffer.h19
-rw-r--r--src/video_core/renderer_opengl/gl_texture_cache.cpp1455
-rw-r--r--src/video_core/renderer_opengl/gl_texture_cache.h285
-rw-r--r--src/video_core/renderer_opengl/maxwell_to_gl.h43
-rw-r--r--src/video_core/renderer_opengl/renderer_opengl.cpp61
-rw-r--r--src/video_core/renderer_opengl/renderer_opengl.h9
-rw-r--r--src/video_core/renderer_opengl/util_shaders.cpp224
-rw-r--r--src/video_core/renderer_opengl/util_shaders.h51
-rw-r--r--src/video_core/renderer_opengl/utils.cpp42
-rw-r--r--src/video_core/renderer_opengl/utils.h16
-rw-r--r--src/video_core/renderer_vulkan/blit_image.cpp624
-rw-r--r--src/video_core/renderer_vulkan/blit_image.h97
-rw-r--r--src/video_core/renderer_vulkan/fixed_pipeline_state.cpp1
-rw-r--r--src/video_core/renderer_vulkan/fixed_pipeline_state.h1
-rw-r--r--src/video_core/renderer_vulkan/maxwell_to_vk.cpp69
-rw-r--r--src/video_core/renderer_vulkan/maxwell_to_vk.h2
-rw-r--r--src/video_core/renderer_vulkan/renderer_vulkan.cpp17
-rw-r--r--src/video_core/renderer_vulkan/renderer_vulkan.h9
-rw-r--r--src/video_core/renderer_vulkan/vk_blit_screen.cpp301
-rw-r--r--src/video_core/renderer_vulkan/vk_blit_screen.h4
-rw-r--r--src/video_core/renderer_vulkan/vk_buffer_cache.cpp117
-rw-r--r--src/video_core/renderer_vulkan/vk_buffer_cache.h12
-rw-r--r--src/video_core/renderer_vulkan/vk_command_pool.cpp4
-rw-r--r--src/video_core/renderer_vulkan/vk_command_pool.h2
-rw-r--r--src/video_core/renderer_vulkan/vk_compute_pass.cpp359
-rw-r--r--src/video_core/renderer_vulkan/vk_compute_pass.h27
-rw-r--r--src/video_core/renderer_vulkan/vk_compute_pipeline.cpp16
-rw-r--r--src/video_core/renderer_vulkan/vk_compute_pipeline.h8
-rw-r--r--src/video_core/renderer_vulkan/vk_device.cpp126
-rw-r--r--src/video_core/renderer_vulkan/vk_device.h34
-rw-r--r--src/video_core/renderer_vulkan/vk_fence_manager.cpp19
-rw-r--r--src/video_core/renderer_vulkan/vk_fence_manager.h20
-rw-r--r--src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp78
-rw-r--r--src/video_core/renderer_vulkan/vk_graphics_pipeline.h24
-rw-r--r--src/video_core/renderer_vulkan/vk_image.cpp135
-rw-r--r--src/video_core/renderer_vulkan/vk_image.h84
-rw-r--r--src/video_core/renderer_vulkan/vk_memory_manager.cpp20
-rw-r--r--src/video_core/renderer_vulkan/vk_memory_manager.h26
-rw-r--r--src/video_core/renderer_vulkan/vk_pipeline_cache.cpp39
-rw-r--r--src/video_core/renderer_vulkan/vk_pipeline_cache.h12
-rw-r--r--src/video_core/renderer_vulkan/vk_query_cache.cpp36
-rw-r--r--src/video_core/renderer_vulkan/vk_query_cache.h14
-rw-r--r--src/video_core/renderer_vulkan/vk_rasterizer.cpp707
-rw-r--r--src/video_core/renderer_vulkan/vk_rasterizer.h139
-rw-r--r--src/video_core/renderer_vulkan/vk_renderpass_cache.cpp158
-rw-r--r--src/video_core/renderer_vulkan/vk_renderpass_cache.h70
-rw-r--r--src/video_core/renderer_vulkan/vk_sampler_cache.cpp83
-rw-r--r--src/video_core/renderer_vulkan/vk_sampler_cache.h29
-rw-r--r--src/video_core/renderer_vulkan/vk_scheduler.cpp79
-rw-r--r--src/video_core/renderer_vulkan/vk_scheduler.h16
-rw-r--r--src/video_core/renderer_vulkan/vk_shader_decompiler.cpp46
-rw-r--r--src/video_core/renderer_vulkan/vk_shader_decompiler.h16
-rw-r--r--src/video_core/renderer_vulkan/vk_shader_util.cpp11
-rw-r--r--src/video_core/renderer_vulkan/vk_shader_util.h4
-rw-r--r--src/video_core/renderer_vulkan/vk_state_tracker.cpp25
-rw-r--r--src/video_core/renderer_vulkan/vk_state_tracker.h8
-rw-r--r--src/video_core/renderer_vulkan/vk_stream_buffer.cpp20
-rw-r--r--src/video_core/renderer_vulkan/vk_stream_buffer.h12
-rw-r--r--src/video_core/renderer_vulkan/vk_texture_cache.cpp1472
-rw-r--r--src/video_core/renderer_vulkan/vk_texture_cache.h327
-rw-r--r--src/video_core/renderer_vulkan/vk_update_descriptor.cpp4
-rw-r--r--src/video_core/renderer_vulkan/vk_update_descriptor.h32
-rw-r--r--src/video_core/renderer_vulkan/wrapper.cpp85
-rw-r--r--src/video_core/renderer_vulkan/wrapper.h142
-rw-r--r--src/video_core/sampler_cache.cpp21
-rw-r--r--src/video_core/sampler_cache.h60
-rw-r--r--src/video_core/shader/ast.cpp13
-rw-r--r--src/video_core/shader/ast.h31
-rw-r--r--src/video_core/shader/async_shaders.cpp11
-rw-r--r--src/video_core/shader/async_shaders.h8
-rw-r--r--src/video_core/shader/control_flow.cpp20
-rw-r--r--src/video_core/shader/control_flow.h14
-rw-r--r--src/video_core/shader/decode.cpp12
-rw-r--r--src/video_core/shader/decode/arithmetic.cpp3
-rw-r--r--src/video_core/shader/decode/arithmetic_integer.cpp9
-rw-r--r--src/video_core/shader/decode/arithmetic_integer_immediate.cpp5
-rw-r--r--src/video_core/shader/decode/conversion.cpp4
-rw-r--r--src/video_core/shader/decode/half_set.cpp14
-rw-r--r--src/video_core/shader/decode/image.cpp21
-rw-r--r--src/video_core/shader/decode/memory.cpp25
-rw-r--r--src/video_core/shader/decode/other.cpp44
-rw-r--r--src/video_core/shader/decode/shift.cpp2
-rw-r--r--src/video_core/shader/decode/texture.cpp65
-rw-r--r--src/video_core/shader/decode/warp.cpp2
-rw-r--r--src/video_core/shader/expr.h6
-rw-r--r--src/video_core/shader/node.h31
-rw-r--r--src/video_core/shader/node_helper.cpp2
-rw-r--r--src/video_core/shader/shader_ir.cpp21
-rw-r--r--src/video_core/shader/shader_ir.h26
-rw-r--r--src/video_core/surface.cpp14
-rw-r--r--src/video_core/surface.h152
-rw-r--r--src/video_core/texture_cache/accelerated_swizzle.cpp70
-rw-r--r--src/video_core/texture_cache/accelerated_swizzle.h45
-rw-r--r--src/video_core/texture_cache/copy_params.h36
-rw-r--r--src/video_core/texture_cache/decode_bc4.cpp97
-rw-r--r--src/video_core/texture_cache/decode_bc4.h16
-rw-r--r--src/video_core/texture_cache/descriptor_table.h82
-rw-r--r--src/video_core/texture_cache/format_lookup_table.cpp380
-rw-r--r--src/video_core/texture_cache/format_lookup_table.h42
-rw-r--r--src/video_core/texture_cache/formatter.cpp95
-rw-r--r--src/video_core/texture_cache/formatter.h263
-rw-r--r--src/video_core/texture_cache/image_base.cpp216
-rw-r--r--src/video_core/texture_cache/image_base.h83
-rw-r--r--src/video_core/texture_cache/image_info.cpp189
-rw-r--r--src/video_core/texture_cache/image_info.h38
-rw-r--r--src/video_core/texture_cache/image_view_base.cpp41
-rw-r--r--src/video_core/texture_cache/image_view_base.h47
-rw-r--r--src/video_core/texture_cache/image_view_info.cpp88
-rw-r--r--src/video_core/texture_cache/image_view_info.h50
-rw-r--r--src/video_core/texture_cache/render_targets.h51
-rw-r--r--src/video_core/texture_cache/samples_helper.h55
-rw-r--r--src/video_core/texture_cache/slot_vector.h156
-rw-r--r--src/video_core/texture_cache/surface_base.cpp298
-rw-r--r--src/video_core/texture_cache/surface_base.h333
-rw-r--r--src/video_core/texture_cache/surface_params.cpp445
-rw-r--r--src/video_core/texture_cache/surface_params.h294
-rw-r--r--src/video_core/texture_cache/surface_view.cpp27
-rw-r--r--src/video_core/texture_cache/surface_view.h68
-rw-r--r--src/video_core/texture_cache/texture_cache.h2398
-rw-r--r--src/video_core/texture_cache/types.h140
-rw-r--r--src/video_core/texture_cache/util.cpp1232
-rw-r--r--src/video_core/texture_cache/util.h107
-rw-r--r--src/video_core/textures/astc.cpp58
-rw-r--r--src/video_core/textures/astc.h5
-rw-r--r--src/video_core/textures/convert.cpp93
-rw-r--r--src/video_core/textures/convert.h22
-rw-r--r--src/video_core/textures/decoders.cpp249
-rw-r--r--src/video_core/textures/decoders.h44
-rw-r--r--src/video_core/textures/texture.cpp16
-rw-r--r--src/video_core/textures/texture.h239
-rw-r--r--src/video_core/video_core.cpp14
-rw-r--r--src/yuzu/CMakeLists.txt10
-rw-r--r--src/yuzu/applets/controller.cpp4
-rw-r--r--src/yuzu/applets/controller.h2
-rw-r--r--src/yuzu/applets/error.cpp6
-rw-r--r--src/yuzu/applets/profile_select.cpp4
-rw-r--r--src/yuzu/applets/profile_select.h2
-rw-r--r--src/yuzu/applets/software_keyboard.cpp4
-rw-r--r--src/yuzu/applets/software_keyboard.h2
-rw-r--r--src/yuzu/applets/web_browser.cpp443
-rw-r--r--src/yuzu/applets/web_browser.h191
-rw-r--r--src/yuzu/applets/web_browser_scripts.h193
-rw-r--r--src/yuzu/bootmanager.cpp16
-rw-r--r--src/yuzu/bootmanager.h2
-rw-r--r--src/yuzu/configuration/config.cpp11
-rw-r--r--src/yuzu/configuration/configure_debug.cpp5
-rw-r--r--src/yuzu/configuration/configure_debug.ui74
-rw-r--r--src/yuzu/configuration/configure_graphics.cpp11
-rw-r--r--src/yuzu/configuration/configure_input_advanced.cpp2
-rw-r--r--src/yuzu/configuration/configure_input_advanced.ui31
-rw-r--r--src/yuzu/configuration/configure_input_player.cpp85
-rw-r--r--src/yuzu/configuration/configure_motion_touch.cpp4
-rw-r--r--src/yuzu/debugger/profiler.cpp2
-rw-r--r--src/yuzu/debugger/wait_tree.cpp12
-rw-r--r--src/yuzu/game_list_p.h3
-rw-r--r--src/yuzu/main.cpp333
-rw-r--r--src/yuzu/main.h18
-rw-r--r--src/yuzu/main.ui78
-rw-r--r--src/yuzu/util/url_request_interceptor.cpp32
-rw-r--r--src/yuzu/util/url_request_interceptor.h30
-rw-r--r--src/yuzu_cmd/CMakeLists.txt15
-rw-r--r--src/yuzu_cmd/config.cpp4
-rw-r--r--src/yuzu_cmd/default_ini.h3
-rw-r--r--src/yuzu_cmd/yuzu.cpp31
-rw-r--r--src/yuzu_tester/config.cpp1
-rw-r--r--src/yuzu_tester/yuzu.cpp1
446 files changed, 17970 insertions, 13564 deletions
diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt
index a22b564d6..8777df751 100644
--- a/src/CMakeLists.txt
+++ b/src/CMakeLists.txt
@@ -62,6 +62,7 @@ else()
-Werror=implicit-fallthrough
-Werror=missing-declarations
-Werror=reorder
+ -Werror=uninitialized
-Werror=unused-result
-Wextra
-Wmissing-declarations
diff --git a/src/audio_core/audio_renderer.cpp b/src/audio_core/audio_renderer.cpp
index 179560cd7..d2ce8c814 100644
--- a/src/audio_core/audio_renderer.cpp
+++ b/src/audio_core/audio_renderer.cpp
@@ -11,7 +11,6 @@
#include "audio_core/info_updater.h"
#include "audio_core/voice_context.h"
#include "common/logging/log.h"
-#include "core/hle/kernel/writable_event.h"
#include "core/memory.h"
#include "core/settings.h"
@@ -71,10 +70,9 @@ namespace {
namespace AudioCore {
AudioRenderer::AudioRenderer(Core::Timing::CoreTiming& core_timing, Core::Memory::Memory& memory_,
AudioCommon::AudioRendererParameter params,
- std::shared_ptr<Kernel::WritableEvent> buffer_event_,
+ Stream::ReleaseCallback&& release_callback,
std::size_t instance_number)
- : worker_params{params}, buffer_event{buffer_event_},
- memory_pool_info(params.effect_count + params.voice_count * 4),
+ : worker_params{params}, memory_pool_info(params.effect_count + params.voice_count * 4),
voice_context(params.voice_count), effect_context(params.effect_count), mix_context(),
sink_context(params.sink_count), splitter_context(),
voices(params.voice_count), memory{memory_},
@@ -85,10 +83,9 @@ AudioRenderer::AudioRenderer(Core::Timing::CoreTiming& core_timing, Core::Memory
params.num_splitter_send_channels);
mix_context.Initialize(behavior_info, params.submix_count + 1, params.effect_count);
audio_out = std::make_unique<AudioCore::AudioOut>();
- stream =
- audio_out->OpenStream(core_timing, params.sample_rate, AudioCommon::STREAM_NUM_CHANNELS,
- fmt::format("AudioRenderer-Instance{}", instance_number),
- [=]() { buffer_event_->Signal(); });
+ stream = audio_out->OpenStream(
+ core_timing, params.sample_rate, AudioCommon::STREAM_NUM_CHANNELS,
+ fmt::format("AudioRenderer-Instance{}", instance_number), std::move(release_callback));
audio_out->StartStream(stream);
QueueMixedBuffer(0);
diff --git a/src/audio_core/audio_renderer.h b/src/audio_core/audio_renderer.h
index 90f7eafa4..18567f618 100644
--- a/src/audio_core/audio_renderer.h
+++ b/src/audio_core/audio_renderer.h
@@ -27,10 +27,6 @@ namespace Core::Timing {
class CoreTiming;
}
-namespace Kernel {
-class WritableEvent;
-}
-
namespace Core::Memory {
class Memory;
}
@@ -44,8 +40,7 @@ class AudioRenderer {
public:
AudioRenderer(Core::Timing::CoreTiming& core_timing, Core::Memory::Memory& memory_,
AudioCommon::AudioRendererParameter params,
- std::shared_ptr<Kernel::WritableEvent> buffer_event_,
- std::size_t instance_number);
+ Stream::ReleaseCallback&& release_callback, std::size_t instance_number);
~AudioRenderer();
[[nodiscard]] ResultCode UpdateAudioRenderer(const std::vector<u8>& input_params,
@@ -61,7 +56,6 @@ private:
BehaviorInfo behavior_info{};
AudioCommon::AudioRendererParameter worker_params;
- std::shared_ptr<Kernel::WritableEvent> buffer_event;
std::vector<ServerMemoryPoolInfo> memory_pool_info;
VoiceContext voice_context;
EffectContext effect_context;
diff --git a/src/audio_core/cubeb_sink.cpp b/src/audio_core/cubeb_sink.cpp
index cf7b186e4..043447eaa 100644
--- a/src/audio_core/cubeb_sink.cpp
+++ b/src/audio_core/cubeb_sink.cpp
@@ -30,6 +30,7 @@ public:
params.rate = sample_rate;
params.channels = num_channels;
params.format = CUBEB_SAMPLE_S16NE;
+ params.prefs = CUBEB_STREAM_PREF_PERSIST;
switch (num_channels) {
case 1:
params.layout = CUBEB_LAYOUT_MONO;
diff --git a/src/audio_core/stream.cpp b/src/audio_core/stream.cpp
index eca296589..afe68c9ed 100644
--- a/src/audio_core/stream.cpp
+++ b/src/audio_core/stream.cpp
@@ -130,7 +130,11 @@ bool Stream::ContainsBuffer([[maybe_unused]] Buffer::Tag tag) const {
std::vector<Buffer::Tag> Stream::GetTagsAndReleaseBuffers(std::size_t max_count) {
std::vector<Buffer::Tag> tags;
for (std::size_t count = 0; count < max_count && !released_buffers.empty(); ++count) {
- tags.push_back(released_buffers.front()->GetTag());
+ if (released_buffers.front()) {
+ tags.push_back(released_buffers.front()->GetTag());
+ } else {
+ ASSERT_MSG(false, "Invalid tag in released_buffers!");
+ }
released_buffers.pop();
}
return tags;
@@ -140,7 +144,11 @@ std::vector<Buffer::Tag> Stream::GetTagsAndReleaseBuffers() {
std::vector<Buffer::Tag> tags;
tags.reserve(released_buffers.size());
while (!released_buffers.empty()) {
- tags.push_back(released_buffers.front()->GetTag());
+ if (released_buffers.front()) {
+ tags.push_back(released_buffers.front()->GetTag());
+ } else {
+ ASSERT_MSG(false, "Invalid tag in released_buffers!");
+ }
released_buffers.pop();
}
return tags;
diff --git a/src/common/CMakeLists.txt b/src/common/CMakeLists.txt
index 56c7e21f5..5c8003eb1 100644
--- a/src/common/CMakeLists.txt
+++ b/src/common/CMakeLists.txt
@@ -104,6 +104,7 @@ add_library(common STATIC
detached_tasks.h
bit_cast.h
bit_field.h
+ bit_set.h
bit_util.h
cityhash.cpp
cityhash.h
@@ -140,7 +141,6 @@ add_library(common STATIC
microprofile.h
microprofileui.h
misc.cpp
- multi_level_queue.h
page_table.cpp
page_table.h
param_package.cpp
@@ -162,6 +162,8 @@ add_library(common STATIC
thread.cpp
thread.h
thread_queue_list.h
+ thread_worker.cpp
+ thread_worker.h
threadsafe_queue.h
time_zone.cpp
time_zone.h
@@ -209,7 +211,6 @@ else()
endif()
create_target_directory_groups(common)
-find_package(Boost 1.71 COMPONENTS context headers REQUIRED)
target_link_libraries(common PUBLIC ${Boost_LIBRARIES} fmt::fmt microprofile)
target_link_libraries(common PRIVATE lz4::lz4 xbyak)
diff --git a/src/common/bit_set.h b/src/common/bit_set.h
new file mode 100644
index 000000000..9235ad412
--- /dev/null
+++ b/src/common/bit_set.h
@@ -0,0 +1,99 @@
+/*
+ * Copyright (c) 2018-2020 Atmosphère-NX
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#pragma once
+
+#include <array>
+#include <bit>
+
+#include "common/alignment.h"
+#include "common/bit_util.h"
+#include "common/common_types.h"
+
+namespace Common {
+
+namespace impl {
+
+template <typename Storage, size_t N>
+class BitSet {
+
+public:
+ constexpr BitSet() = default;
+
+ constexpr void SetBit(size_t i) {
+ this->words[i / FlagsPerWord] |= GetBitMask(i % FlagsPerWord);
+ }
+
+ constexpr void ClearBit(size_t i) {
+ this->words[i / FlagsPerWord] &= ~GetBitMask(i % FlagsPerWord);
+ }
+
+ constexpr size_t CountLeadingZero() const {
+ for (size_t i = 0; i < NumWords; i++) {
+ if (this->words[i]) {
+ return FlagsPerWord * i + CountLeadingZeroImpl(this->words[i]);
+ }
+ }
+ return FlagsPerWord * NumWords;
+ }
+
+ constexpr size_t GetNextSet(size_t n) const {
+ for (size_t i = (n + 1) / FlagsPerWord; i < NumWords; i++) {
+ Storage word = this->words[i];
+ if (!IsAligned(n + 1, FlagsPerWord)) {
+ word &= GetBitMask(n % FlagsPerWord) - 1;
+ }
+ if (word) {
+ return FlagsPerWord * i + CountLeadingZeroImpl(word);
+ }
+ }
+ return FlagsPerWord * NumWords;
+ }
+
+private:
+ static_assert(std::is_unsigned_v<Storage>);
+ static_assert(sizeof(Storage) <= sizeof(u64));
+
+ static constexpr size_t FlagsPerWord = BitSize<Storage>();
+ static constexpr size_t NumWords = AlignUp(N, FlagsPerWord) / FlagsPerWord;
+
+ static constexpr auto CountLeadingZeroImpl(Storage word) {
+ return std::countl_zero(static_cast<unsigned long long>(word)) -
+ (BitSize<unsigned long long>() - FlagsPerWord);
+ }
+
+ static constexpr Storage GetBitMask(size_t bit) {
+ return Storage(1) << (FlagsPerWord - 1 - bit);
+ }
+
+ std::array<Storage, NumWords> words{};
+};
+
+} // namespace impl
+
+template <size_t N>
+using BitSet8 = impl::BitSet<u8, N>;
+
+template <size_t N>
+using BitSet16 = impl::BitSet<u16, N>;
+
+template <size_t N>
+using BitSet32 = impl::BitSet<u32, N>;
+
+template <size_t N>
+using BitSet64 = impl::BitSet<u64, N>;
+
+} // namespace Common
diff --git a/src/common/concepts.h b/src/common/concepts.h
index 5bef3ad67..aa08065a7 100644
--- a/src/common/concepts.h
+++ b/src/common/concepts.h
@@ -31,4 +31,8 @@ concept DerivedFrom = requires {
std::is_convertible_v<const volatile Derived*, const volatile Base*>;
};
+// TODO: Replace with std::convertible_to when libc++ implements it.
+template <typename From, typename To>
+concept ConvertibleTo = std::is_convertible_v<From, To>;
+
} // namespace Common
diff --git a/src/common/multi_level_queue.h b/src/common/multi_level_queue.h
deleted file mode 100644
index 4b305bf40..000000000
--- a/src/common/multi_level_queue.h
+++ /dev/null
@@ -1,345 +0,0 @@
-// Copyright 2019 TuxSH
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <array>
-#include <iterator>
-#include <list>
-#include <utility>
-
-#include "common/bit_util.h"
-#include "common/common_types.h"
-
-namespace Common {
-
-/**
- * A MultiLevelQueue is a type of priority queue which has the following characteristics:
- * - iteratable through each of its elements.
- * - back can be obtained.
- * - O(1) add, lookup (both front and back)
- * - discrete priorities and a max of 64 priorities (limited domain)
- * This type of priority queue is normaly used for managing threads within an scheduler
- */
-template <typename T, std::size_t Depth>
-class MultiLevelQueue {
-public:
- using value_type = T;
- using reference = value_type&;
- using const_reference = const value_type&;
- using pointer = value_type*;
- using const_pointer = const value_type*;
-
- using difference_type = typename std::pointer_traits<pointer>::difference_type;
- using size_type = std::size_t;
-
- template <bool is_constant>
- class iterator_impl {
- public:
- using iterator_category = std::bidirectional_iterator_tag;
- using value_type = T;
- using pointer = std::conditional_t<is_constant, T*, const T*>;
- using reference = std::conditional_t<is_constant, const T&, T&>;
- using difference_type = typename std::pointer_traits<pointer>::difference_type;
-
- friend bool operator==(const iterator_impl& lhs, const iterator_impl& rhs) {
- if (lhs.IsEnd() && rhs.IsEnd())
- return true;
- return std::tie(lhs.current_priority, lhs.it) == std::tie(rhs.current_priority, rhs.it);
- }
-
- friend bool operator!=(const iterator_impl& lhs, const iterator_impl& rhs) {
- return !operator==(lhs, rhs);
- }
-
- reference operator*() const {
- return *it;
- }
-
- pointer operator->() const {
- return it.operator->();
- }
-
- iterator_impl& operator++() {
- if (IsEnd()) {
- return *this;
- }
-
- ++it;
-
- if (it == GetEndItForPrio()) {
- u64 prios = mlq.used_priorities;
- prios &= ~((1ULL << (current_priority + 1)) - 1);
- if (prios == 0) {
- current_priority = static_cast<u32>(mlq.depth());
- } else {
- current_priority = CountTrailingZeroes64(prios);
- it = GetBeginItForPrio();
- }
- }
- return *this;
- }
-
- iterator_impl& operator--() {
- if (IsEnd()) {
- if (mlq.used_priorities != 0) {
- current_priority = 63 - CountLeadingZeroes64(mlq.used_priorities);
- it = GetEndItForPrio();
- --it;
- }
- } else if (it == GetBeginItForPrio()) {
- u64 prios = mlq.used_priorities;
- prios &= (1ULL << current_priority) - 1;
- if (prios != 0) {
- current_priority = CountTrailingZeroes64(prios);
- it = GetEndItForPrio();
- --it;
- }
- } else {
- --it;
- }
- return *this;
- }
-
- iterator_impl operator++(int) {
- const iterator_impl v{*this};
- ++(*this);
- return v;
- }
-
- iterator_impl operator--(int) {
- const iterator_impl v{*this};
- --(*this);
- return v;
- }
-
- // allow implicit const->non-const
- iterator_impl(const iterator_impl<false>& other)
- : mlq(other.mlq), it(other.it), current_priority(other.current_priority) {}
-
- iterator_impl(const iterator_impl<true>& other)
- : mlq(other.mlq), it(other.it), current_priority(other.current_priority) {}
-
- iterator_impl& operator=(const iterator_impl<false>& other) {
- mlq = other.mlq;
- it = other.it;
- current_priority = other.current_priority;
- return *this;
- }
-
- friend class iterator_impl<true>;
- iterator_impl() = default;
-
- private:
- friend class MultiLevelQueue;
- using container_ref =
- std::conditional_t<is_constant, const MultiLevelQueue&, MultiLevelQueue&>;
- using list_iterator = std::conditional_t<is_constant, typename std::list<T>::const_iterator,
- typename std::list<T>::iterator>;
-
- explicit iterator_impl(container_ref mlq, list_iterator it, u32 current_priority)
- : mlq(mlq), it(it), current_priority(current_priority) {}
- explicit iterator_impl(container_ref mlq, u32 current_priority)
- : mlq(mlq), it(), current_priority(current_priority) {}
-
- bool IsEnd() const {
- return current_priority == mlq.depth();
- }
-
- list_iterator GetBeginItForPrio() const {
- return mlq.levels[current_priority].begin();
- }
-
- list_iterator GetEndItForPrio() const {
- return mlq.levels[current_priority].end();
- }
-
- container_ref mlq;
- list_iterator it;
- u32 current_priority;
- };
-
- using iterator = iterator_impl<false>;
- using const_iterator = iterator_impl<true>;
-
- void add(const T& element, u32 priority, bool send_back = true) {
- if (send_back)
- levels[priority].push_back(element);
- else
- levels[priority].push_front(element);
- used_priorities |= 1ULL << priority;
- }
-
- void remove(const T& element, u32 priority) {
- auto it = ListIterateTo(levels[priority], element);
- if (it == levels[priority].end())
- return;
- levels[priority].erase(it);
- if (levels[priority].empty()) {
- used_priorities &= ~(1ULL << priority);
- }
- }
-
- void adjust(const T& element, u32 old_priority, u32 new_priority, bool adjust_front = false) {
- remove(element, old_priority);
- add(element, new_priority, !adjust_front);
- }
- void adjust(const_iterator it, u32 old_priority, u32 new_priority, bool adjust_front = false) {
- adjust(*it, old_priority, new_priority, adjust_front);
- }
-
- void transfer_to_front(const T& element, u32 priority, MultiLevelQueue& other) {
- ListSplice(other.levels[priority], other.levels[priority].begin(), levels[priority],
- ListIterateTo(levels[priority], element));
-
- other.used_priorities |= 1ULL << priority;
-
- if (levels[priority].empty()) {
- used_priorities &= ~(1ULL << priority);
- }
- }
-
- void transfer_to_front(const_iterator it, u32 priority, MultiLevelQueue& other) {
- transfer_to_front(*it, priority, other);
- }
-
- void transfer_to_back(const T& element, u32 priority, MultiLevelQueue& other) {
- ListSplice(other.levels[priority], other.levels[priority].end(), levels[priority],
- ListIterateTo(levels[priority], element));
-
- other.used_priorities |= 1ULL << priority;
-
- if (levels[priority].empty()) {
- used_priorities &= ~(1ULL << priority);
- }
- }
-
- void transfer_to_back(const_iterator it, u32 priority, MultiLevelQueue& other) {
- transfer_to_back(*it, priority, other);
- }
-
- void yield(u32 priority, std::size_t n = 1) {
- ListShiftForward(levels[priority], n);
- }
-
- [[nodiscard]] std::size_t depth() const {
- return Depth;
- }
-
- [[nodiscard]] std::size_t size(u32 priority) const {
- return levels[priority].size();
- }
-
- [[nodiscard]] std::size_t size() const {
- u64 priorities = used_priorities;
- std::size_t size = 0;
- while (priorities != 0) {
- const u64 current_priority = CountTrailingZeroes64(priorities);
- size += levels[current_priority].size();
- priorities &= ~(1ULL << current_priority);
- }
- return size;
- }
-
- [[nodiscard]] bool empty() const {
- return used_priorities == 0;
- }
-
- [[nodiscard]] bool empty(u32 priority) const {
- return (used_priorities & (1ULL << priority)) == 0;
- }
-
- [[nodiscard]] u32 highest_priority_set(u32 max_priority = 0) const {
- const u64 priorities =
- max_priority == 0 ? used_priorities : (used_priorities & ~((1ULL << max_priority) - 1));
- return priorities == 0 ? Depth : static_cast<u32>(CountTrailingZeroes64(priorities));
- }
-
- [[nodiscard]] u32 lowest_priority_set(u32 min_priority = Depth - 1) const {
- const u64 priorities = min_priority >= Depth - 1
- ? used_priorities
- : (used_priorities & ((1ULL << (min_priority + 1)) - 1));
- return priorities == 0 ? Depth : 63 - CountLeadingZeroes64(priorities);
- }
-
- [[nodiscard]] const_iterator cbegin(u32 max_prio = 0) const {
- const u32 priority = highest_priority_set(max_prio);
- return priority == Depth ? cend()
- : const_iterator{*this, levels[priority].cbegin(), priority};
- }
- [[nodiscard]] const_iterator begin(u32 max_prio = 0) const {
- return cbegin(max_prio);
- }
- [[nodiscard]] iterator begin(u32 max_prio = 0) {
- const u32 priority = highest_priority_set(max_prio);
- return priority == Depth ? end() : iterator{*this, levels[priority].begin(), priority};
- }
-
- [[nodiscard]] const_iterator cend(u32 min_prio = Depth - 1) const {
- return min_prio == Depth - 1 ? const_iterator{*this, Depth} : cbegin(min_prio + 1);
- }
- [[nodiscard]] const_iterator end(u32 min_prio = Depth - 1) const {
- return cend(min_prio);
- }
- [[nodiscard]] iterator end(u32 min_prio = Depth - 1) {
- return min_prio == Depth - 1 ? iterator{*this, Depth} : begin(min_prio + 1);
- }
-
- [[nodiscard]] T& front(u32 max_priority = 0) {
- const u32 priority = highest_priority_set(max_priority);
- return levels[priority == Depth ? 0 : priority].front();
- }
- [[nodiscard]] const T& front(u32 max_priority = 0) const {
- const u32 priority = highest_priority_set(max_priority);
- return levels[priority == Depth ? 0 : priority].front();
- }
-
- [[nodiscard]] T& back(u32 min_priority = Depth - 1) {
- const u32 priority = lowest_priority_set(min_priority); // intended
- return levels[priority == Depth ? 63 : priority].back();
- }
- [[nodiscard]] const T& back(u32 min_priority = Depth - 1) const {
- const u32 priority = lowest_priority_set(min_priority); // intended
- return levels[priority == Depth ? 63 : priority].back();
- }
-
- void clear() {
- used_priorities = 0;
- for (std::size_t i = 0; i < Depth; i++) {
- levels[i].clear();
- }
- }
-
-private:
- using const_list_iterator = typename std::list<T>::const_iterator;
-
- static void ListShiftForward(std::list<T>& list, const std::size_t shift = 1) {
- if (shift >= list.size()) {
- return;
- }
-
- const auto begin_range = list.begin();
- const auto end_range = std::next(begin_range, shift);
- list.splice(list.end(), list, begin_range, end_range);
- }
-
- static void ListSplice(std::list<T>& in_list, const_list_iterator position,
- std::list<T>& out_list, const_list_iterator element) {
- in_list.splice(position, out_list, element);
- }
-
- [[nodiscard]] static const_list_iterator ListIterateTo(const std::list<T>& list,
- const T& element) {
- auto it = list.cbegin();
- while (it != list.cend() && *it != element) {
- ++it;
- }
- return it;
- }
-
- std::array<std::list<T>, Depth> levels;
- u64 used_priorities = 0;
-};
-
-} // namespace Common
diff --git a/src/common/scope_exit.h b/src/common/scope_exit.h
index 68ef5f197..fa46cb394 100644
--- a/src/common/scope_exit.h
+++ b/src/common/scope_exit.h
@@ -10,7 +10,7 @@
namespace detail {
template <typename Func>
struct ScopeExitHelper {
- explicit ScopeExitHelper(Func&& func) : func(std::move(func)) {}
+ explicit ScopeExitHelper(Func&& func_) : func(std::move(func_)) {}
~ScopeExitHelper() {
if (active) {
func();
diff --git a/src/common/telemetry.h b/src/common/telemetry.h
index a50c5d1de..49186e848 100644
--- a/src/common/telemetry.h
+++ b/src/common/telemetry.h
@@ -52,8 +52,8 @@ public:
template <typename T>
class Field : public FieldInterface {
public:
- Field(FieldType type, std::string name, T value)
- : name(std::move(name)), type(type), value(std::move(value)) {}
+ Field(FieldType type_, std::string name_, T value_)
+ : name(std::move(name_)), type(type_), value(std::move(value_)) {}
Field(const Field&) = default;
Field& operator=(const Field&) = default;
diff --git a/src/common/thread_worker.cpp b/src/common/thread_worker.cpp
new file mode 100644
index 000000000..8f9bf447a
--- /dev/null
+++ b/src/common/thread_worker.cpp
@@ -0,0 +1,58 @@
+// Copyright 2020 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "common/thread.h"
+#include "common/thread_worker.h"
+
+namespace Common {
+
+ThreadWorker::ThreadWorker(std::size_t num_workers, const std::string& name) {
+ for (std::size_t i = 0; i < num_workers; ++i)
+ threads.emplace_back([this, thread_name{std::string{name}}] {
+ Common::SetCurrentThreadName(thread_name.c_str());
+
+ // Wait for first request
+ {
+ std::unique_lock lock{queue_mutex};
+ condition.wait(lock, [this] { return stop || !requests.empty(); });
+ }
+
+ while (true) {
+ std::function<void()> task;
+
+ {
+ std::unique_lock lock{queue_mutex};
+ condition.wait(lock, [this] { return stop || !requests.empty(); });
+ if (stop || requests.empty()) {
+ return;
+ }
+ task = std::move(requests.front());
+ requests.pop();
+ }
+
+ task();
+ }
+ });
+}
+
+ThreadWorker::~ThreadWorker() {
+ {
+ std::unique_lock lock{queue_mutex};
+ stop = true;
+ }
+ condition.notify_all();
+ for (std::thread& thread : threads) {
+ thread.join();
+ }
+}
+
+void ThreadWorker::QueueWork(std::function<void()>&& work) {
+ {
+ std::unique_lock lock{queue_mutex};
+ requests.emplace(work);
+ }
+ condition.notify_one();
+}
+
+} // namespace Common
diff --git a/src/common/thread_worker.h b/src/common/thread_worker.h
new file mode 100644
index 000000000..f1859971f
--- /dev/null
+++ b/src/common/thread_worker.h
@@ -0,0 +1,30 @@
+// Copyright 2020 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <atomic>
+#include <functional>
+#include <mutex>
+#include <string>
+#include <vector>
+#include <queue>
+
+namespace Common {
+
+class ThreadWorker final {
+public:
+ explicit ThreadWorker(std::size_t num_workers, const std::string& name);
+ ~ThreadWorker();
+ void QueueWork(std::function<void()>&& work);
+
+private:
+ std::vector<std::thread> threads;
+ std::queue<std::function<void()>> requests;
+ std::mutex queue_mutex;
+ std::condition_variable condition;
+ std::atomic_bool stop{};
+};
+
+} // namespace Common
diff --git a/src/common/x64/xbyak_abi.h b/src/common/x64/xbyak_abi.h
index 26e4bfda5..c2c9b6134 100644
--- a/src/common/x64/xbyak_abi.h
+++ b/src/common/x64/xbyak_abi.h
@@ -11,25 +11,25 @@
namespace Common::X64 {
-constexpr std::size_t RegToIndex(const Xbyak::Reg& reg) {
+constexpr size_t RegToIndex(const Xbyak::Reg& reg) {
using Kind = Xbyak::Reg::Kind;
ASSERT_MSG((reg.getKind() & (Kind::REG | Kind::XMM)) != 0,
"RegSet only support GPRs and XMM registers.");
ASSERT_MSG(reg.getIdx() < 16, "RegSet only supports XXM0-15.");
- return reg.getIdx() + (reg.getKind() == Kind::REG ? 0 : 16);
+ return static_cast<size_t>(reg.getIdx()) + (reg.getKind() == Kind::REG ? 0 : 16);
}
-constexpr Xbyak::Reg64 IndexToReg64(std::size_t reg_index) {
+constexpr Xbyak::Reg64 IndexToReg64(size_t reg_index) {
ASSERT(reg_index < 16);
return Xbyak::Reg64(static_cast<int>(reg_index));
}
-constexpr Xbyak::Xmm IndexToXmm(std::size_t reg_index) {
+constexpr Xbyak::Xmm IndexToXmm(size_t reg_index) {
ASSERT(reg_index >= 16 && reg_index < 32);
return Xbyak::Xmm(static_cast<int>(reg_index - 16));
}
-constexpr Xbyak::Reg IndexToReg(std::size_t reg_index) {
+constexpr Xbyak::Reg IndexToReg(size_t reg_index) {
if (reg_index < 16) {
return IndexToReg64(reg_index);
} else {
@@ -182,7 +182,7 @@ inline size_t ABI_PushRegistersAndAdjustStack(Xbyak::CodeGenerator& code, std::b
size_t rsp_alignment, size_t needed_frame_size = 0) {
auto frame_info = ABI_CalculateFrameSize(regs, rsp_alignment, needed_frame_size);
- for (std::size_t i = 0; i < regs.size(); ++i) {
+ for (size_t i = 0; i < regs.size(); ++i) {
if (regs[i] && ABI_ALL_GPRS[i]) {
code.push(IndexToReg64(i));
}
@@ -192,7 +192,7 @@ inline size_t ABI_PushRegistersAndAdjustStack(Xbyak::CodeGenerator& code, std::b
code.sub(code.rsp, frame_info.subtraction);
}
- for (std::size_t i = 0; i < regs.size(); ++i) {
+ for (size_t i = 0; i < regs.size(); ++i) {
if (regs[i] && ABI_ALL_XMMS[i]) {
code.movaps(code.xword[code.rsp + frame_info.xmm_offset], IndexToXmm(i));
frame_info.xmm_offset += 0x10;
@@ -206,7 +206,7 @@ inline void ABI_PopRegistersAndAdjustStack(Xbyak::CodeGenerator& code, std::bits
size_t rsp_alignment, size_t needed_frame_size = 0) {
auto frame_info = ABI_CalculateFrameSize(regs, rsp_alignment, needed_frame_size);
- for (std::size_t i = 0; i < regs.size(); ++i) {
+ for (size_t i = 0; i < regs.size(); ++i) {
if (regs[i] && ABI_ALL_XMMS[i]) {
code.movaps(IndexToXmm(i), code.xword[code.rsp + frame_info.xmm_offset]);
frame_info.xmm_offset += 0x10;
@@ -218,8 +218,8 @@ inline void ABI_PopRegistersAndAdjustStack(Xbyak::CodeGenerator& code, std::bits
}
// GPRs need to be popped in reverse order
- for (std::size_t j = 0; j < regs.size(); ++j) {
- const std::size_t i = regs.size() - j - 1;
+ for (size_t j = 0; j < regs.size(); ++j) {
+ const size_t i = regs.size() - j - 1;
if (regs[i] && ABI_ALL_GPRS[i]) {
code.pop(IndexToReg64(i));
}
diff --git a/src/core/CMakeLists.txt b/src/core/CMakeLists.txt
index 66de33799..01f3e9419 100644
--- a/src/core/CMakeLists.txt
+++ b/src/core/CMakeLists.txt
@@ -41,6 +41,7 @@ add_library(core STATIC
file_sys/bis_factory.h
file_sys/card_image.cpp
file_sys/card_image.h
+ file_sys/common_funcs.h
file_sys/content_archive.cpp
file_sys/content_archive.h
file_sys/control_metadata.cpp
@@ -134,6 +135,8 @@ add_library(core STATIC
frontend/emu_window.h
frontend/framebuffer_layout.cpp
frontend/framebuffer_layout.h
+ frontend/input_interpreter.cpp
+ frontend/input_interpreter.h
frontend/input.h
hardware_interrupt_manager.cpp
hardware_interrupt_manager.h
@@ -148,10 +151,19 @@ add_library(core STATIC
hle/kernel/code_set.cpp
hle/kernel/code_set.h
hle/kernel/errors.h
+ hle/kernel/global_scheduler_context.cpp
+ hle/kernel/global_scheduler_context.h
hle/kernel/handle_table.cpp
hle/kernel/handle_table.h
hle/kernel/hle_ipc.cpp
hle/kernel/hle_ipc.h
+ hle/kernel/k_affinity_mask.h
+ hle/kernel/k_priority_queue.h
+ hle/kernel/k_scheduler.cpp
+ hle/kernel/k_scheduler.h
+ hle/kernel/k_scheduler_lock.h
+ hle/kernel/k_scoped_lock.h
+ hle/kernel/k_scoped_scheduler_lock_and_sleep.h
hle/kernel/kernel.cpp
hle/kernel/kernel.h
hle/kernel/memory/address_space_info.cpp
@@ -186,12 +198,12 @@ add_library(core STATIC
hle/kernel/readable_event.h
hle/kernel/resource_limit.cpp
hle/kernel/resource_limit.h
- hle/kernel/scheduler.cpp
- hle/kernel/scheduler.h
hle/kernel/server_port.cpp
hle/kernel/server_port.h
hle/kernel/server_session.cpp
hle/kernel/server_session.h
+ hle/kernel/service_thread.cpp
+ hle/kernel/service_thread.h
hle/kernel/session.cpp
hle/kernel/session.h
hle/kernel/shared_memory.cpp
@@ -490,7 +502,6 @@ add_library(core STATIC
hle/service/sm/controller.h
hle/service/sm/sm.cpp
hle/service/sm/sm.h
- hle/service/sockets/blocking_worker.h
hle/service/sockets/bsd.cpp
hle/service/sockets/bsd.h
hle/service/sockets/ethc.cpp
diff --git a/src/core/arm/dynarmic/arm_dynarmic_32.cpp b/src/core/arm/dynarmic/arm_dynarmic_32.cpp
index 193fd7d62..e9c74b1a6 100644
--- a/src/core/arm/dynarmic/arm_dynarmic_32.cpp
+++ b/src/core/arm/dynarmic/arm_dynarmic_32.cpp
@@ -294,6 +294,9 @@ void ARM_Dynarmic_32::InvalidateCacheRange(VAddr addr, std::size_t size) {
}
void ARM_Dynarmic_32::ClearExclusiveState() {
+ if (!jit) {
+ return;
+ }
jit->ClearExclusiveState();
}
diff --git a/src/core/arm/dynarmic/arm_dynarmic_64.cpp b/src/core/arm/dynarmic/arm_dynarmic_64.cpp
index 0f0585d0f..7a4eb88a2 100644
--- a/src/core/arm/dynarmic/arm_dynarmic_64.cpp
+++ b/src/core/arm/dynarmic/arm_dynarmic_64.cpp
@@ -15,8 +15,8 @@
#include "core/core.h"
#include "core/core_timing.h"
#include "core/hardware_properties.h"
+#include "core/hle/kernel/k_scheduler.h"
#include "core/hle/kernel/process.h"
-#include "core/hle/kernel/scheduler.h"
#include "core/hle/kernel/svc.h"
#include "core/memory.h"
#include "core/settings.h"
@@ -330,6 +330,9 @@ void ARM_Dynarmic_64::InvalidateCacheRange(VAddr addr, std::size_t size) {
}
void ARM_Dynarmic_64::ClearExclusiveState() {
+ if (!jit) {
+ return;
+ }
jit->ClearExclusiveState();
}
diff --git a/src/core/core.cpp b/src/core/core.cpp
index 01e4faac8..1a2002dec 100644
--- a/src/core/core.cpp
+++ b/src/core/core.cpp
@@ -27,10 +27,10 @@
#include "core/file_sys/vfs_real.h"
#include "core/hardware_interrupt_manager.h"
#include "core/hle/kernel/client_port.h"
+#include "core/hle/kernel/k_scheduler.h"
#include "core/hle/kernel/kernel.h"
#include "core/hle/kernel/physical_core.h"
#include "core/hle/kernel/process.h"
-#include "core/hle/kernel/scheduler.h"
#include "core/hle/kernel/thread.h"
#include "core/hle/service/am/applets/applets.h"
#include "core/hle/service/apm/controller.h"
@@ -159,7 +159,7 @@ struct System::Impl {
device_memory = std::make_unique<Core::DeviceMemory>();
is_multicore = Settings::values.use_multi_core.GetValue();
- is_async_gpu = is_multicore || Settings::values.use_asynchronous_gpu_emulation.GetValue();
+ is_async_gpu = Settings::values.use_asynchronous_gpu_emulation.GetValue();
kernel.SetMulticore(is_multicore);
cpu_manager.SetMulticore(is_multicore);
@@ -237,7 +237,7 @@ struct System::Impl {
Kernel::Process::Create(system, "main", Kernel::Process::ProcessType::Userland);
const auto [load_result, load_parameters] = app_loader->Load(*main_process, system);
if (load_result != Loader::ResultStatus::Success) {
- LOG_CRITICAL(Core, "Failed to load ROM (Error {})!", static_cast<int>(load_result));
+ LOG_CRITICAL(Core, "Failed to load ROM (Error {})!", load_result);
Shutdown();
return static_cast<ResultStatus>(static_cast<u32>(ResultStatus::ErrorLoader) +
@@ -267,8 +267,7 @@ struct System::Impl {
u64 title_id{0};
if (app_loader->ReadProgramId(title_id) != Loader::ResultStatus::Success) {
- LOG_ERROR(Core, "Failed to find title id for ROM (Error {})",
- static_cast<u32>(load_result));
+ LOG_ERROR(Core, "Failed to find title id for ROM (Error {})", load_result);
}
perf_stats = std::make_unique<PerfStats>(title_id);
// Reset counters and set time origin to current frame
@@ -308,7 +307,6 @@ struct System::Impl {
service_manager.reset();
cheat_engine.reset();
telemetry_session.reset();
- device_memory.reset();
// Close all CPU/threading state
cpu_manager.Shutdown();
@@ -508,14 +506,6 @@ std::size_t System::CurrentCoreIndex() const {
return core;
}
-Kernel::Scheduler& System::CurrentScheduler() {
- return impl->kernel.CurrentScheduler();
-}
-
-const Kernel::Scheduler& System::CurrentScheduler() const {
- return impl->kernel.CurrentScheduler();
-}
-
Kernel::PhysicalCore& System::CurrentPhysicalCore() {
return impl->kernel.CurrentPhysicalCore();
}
@@ -524,22 +514,14 @@ const Kernel::PhysicalCore& System::CurrentPhysicalCore() const {
return impl->kernel.CurrentPhysicalCore();
}
-Kernel::Scheduler& System::Scheduler(std::size_t core_index) {
- return impl->kernel.Scheduler(core_index);
-}
-
-const Kernel::Scheduler& System::Scheduler(std::size_t core_index) const {
- return impl->kernel.Scheduler(core_index);
-}
-
/// Gets the global scheduler
-Kernel::GlobalScheduler& System::GlobalScheduler() {
- return impl->kernel.GlobalScheduler();
+Kernel::GlobalSchedulerContext& System::GlobalSchedulerContext() {
+ return impl->kernel.GlobalSchedulerContext();
}
/// Gets the global scheduler
-const Kernel::GlobalScheduler& System::GlobalScheduler() const {
- return impl->kernel.GlobalScheduler();
+const Kernel::GlobalSchedulerContext& System::GlobalSchedulerContext() const {
+ return impl->kernel.GlobalSchedulerContext();
}
Kernel::Process* System::CurrentProcess() {
diff --git a/src/core/core.h b/src/core/core.h
index 29b8fb92a..579a774e4 100644
--- a/src/core/core.h
+++ b/src/core/core.h
@@ -26,11 +26,11 @@ class VfsFilesystem;
} // namespace FileSys
namespace Kernel {
-class GlobalScheduler;
+class GlobalSchedulerContext;
class KernelCore;
class PhysicalCore;
class Process;
-class Scheduler;
+class KScheduler;
} // namespace Kernel
namespace Loader {
@@ -213,12 +213,6 @@ public:
/// Gets the index of the currently running CPU core
[[nodiscard]] std::size_t CurrentCoreIndex() const;
- /// Gets the scheduler for the CPU core that is currently running
- [[nodiscard]] Kernel::Scheduler& CurrentScheduler();
-
- /// Gets the scheduler for the CPU core that is currently running
- [[nodiscard]] const Kernel::Scheduler& CurrentScheduler() const;
-
/// Gets the physical core for the CPU core that is currently running
[[nodiscard]] Kernel::PhysicalCore& CurrentPhysicalCore();
@@ -261,17 +255,11 @@ public:
/// Gets an immutable reference to the renderer.
[[nodiscard]] const VideoCore::RendererBase& Renderer() const;
- /// Gets the scheduler for the CPU core with the specified index
- [[nodiscard]] Kernel::Scheduler& Scheduler(std::size_t core_index);
-
- /// Gets the scheduler for the CPU core with the specified index
- [[nodiscard]] const Kernel::Scheduler& Scheduler(std::size_t core_index) const;
-
/// Gets the global scheduler
- [[nodiscard]] Kernel::GlobalScheduler& GlobalScheduler();
+ [[nodiscard]] Kernel::GlobalSchedulerContext& GlobalSchedulerContext();
/// Gets the global scheduler
- [[nodiscard]] const Kernel::GlobalScheduler& GlobalScheduler() const;
+ [[nodiscard]] const Kernel::GlobalSchedulerContext& GlobalSchedulerContext() const;
/// Gets the manager for the guest device memory
[[nodiscard]] Core::DeviceMemory& DeviceMemory();
diff --git a/src/core/cpu_manager.cpp b/src/core/cpu_manager.cpp
index 0cff985e9..373395047 100644
--- a/src/core/cpu_manager.cpp
+++ b/src/core/cpu_manager.cpp
@@ -10,9 +10,9 @@
#include "core/core.h"
#include "core/core_timing.h"
#include "core/cpu_manager.h"
+#include "core/hle/kernel/k_scheduler.h"
#include "core/hle/kernel/kernel.h"
#include "core/hle/kernel/physical_core.h"
-#include "core/hle/kernel/scheduler.h"
#include "core/hle/kernel/thread.h"
#include "video_core/gpu.h"
@@ -109,11 +109,8 @@ void* CpuManager::GetStartFuncParamater() {
void CpuManager::MultiCoreRunGuestThread() {
auto& kernel = system.Kernel();
- {
- auto& sched = kernel.CurrentScheduler();
- sched.OnThreadStart();
- }
- auto* thread = kernel.CurrentScheduler().GetCurrentThread();
+ kernel.CurrentScheduler()->OnThreadStart();
+ auto* thread = kernel.CurrentScheduler()->GetCurrentThread();
auto& host_context = thread->GetHostContext();
host_context->SetRewindPoint(GuestRewindFunction, this);
MultiCoreRunGuestLoop();
@@ -130,8 +127,8 @@ void CpuManager::MultiCoreRunGuestLoop() {
physical_core = &kernel.CurrentPhysicalCore();
}
system.ExitDynarmicProfile();
- auto& scheduler = kernel.CurrentScheduler();
- scheduler.TryDoContextSwitch();
+ physical_core->ArmInterface().ClearExclusiveState();
+ kernel.CurrentScheduler()->RescheduleCurrentCore();
}
}
@@ -140,25 +137,21 @@ void CpuManager::MultiCoreRunIdleThread() {
while (true) {
auto& physical_core = kernel.CurrentPhysicalCore();
physical_core.Idle();
- auto& scheduler = kernel.CurrentScheduler();
- scheduler.TryDoContextSwitch();
+ kernel.CurrentScheduler()->RescheduleCurrentCore();
}
}
void CpuManager::MultiCoreRunSuspendThread() {
auto& kernel = system.Kernel();
- {
- auto& sched = kernel.CurrentScheduler();
- sched.OnThreadStart();
- }
+ kernel.CurrentScheduler()->OnThreadStart();
while (true) {
auto core = kernel.GetCurrentHostThreadID();
- auto& scheduler = kernel.CurrentScheduler();
+ auto& scheduler = *kernel.CurrentScheduler();
Kernel::Thread* current_thread = scheduler.GetCurrentThread();
Common::Fiber::YieldTo(current_thread->GetHostContext(), core_data[core].host_context);
ASSERT(scheduler.ContextSwitchPending());
ASSERT(core == kernel.GetCurrentHostThreadID());
- scheduler.TryDoContextSwitch();
+ scheduler.RescheduleCurrentCore();
}
}
@@ -206,11 +199,8 @@ void CpuManager::MultiCorePause(bool paused) {
void CpuManager::SingleCoreRunGuestThread() {
auto& kernel = system.Kernel();
- {
- auto& sched = kernel.CurrentScheduler();
- sched.OnThreadStart();
- }
- auto* thread = kernel.CurrentScheduler().GetCurrentThread();
+ kernel.CurrentScheduler()->OnThreadStart();
+ auto* thread = kernel.CurrentScheduler()->GetCurrentThread();
auto& host_context = thread->GetHostContext();
host_context->SetRewindPoint(GuestRewindFunction, this);
SingleCoreRunGuestLoop();
@@ -218,7 +208,7 @@ void CpuManager::SingleCoreRunGuestThread() {
void CpuManager::SingleCoreRunGuestLoop() {
auto& kernel = system.Kernel();
- auto* thread = kernel.CurrentScheduler().GetCurrentThread();
+ auto* thread = kernel.CurrentScheduler()->GetCurrentThread();
while (true) {
auto* physical_core = &kernel.CurrentPhysicalCore();
system.EnterDynarmicProfile();
@@ -230,9 +220,10 @@ void CpuManager::SingleCoreRunGuestLoop() {
thread->SetPhantomMode(true);
system.CoreTiming().Advance();
thread->SetPhantomMode(false);
+ physical_core->ArmInterface().ClearExclusiveState();
PreemptSingleCore();
auto& scheduler = kernel.Scheduler(current_core);
- scheduler.TryDoContextSwitch();
+ scheduler.RescheduleCurrentCore();
}
}
@@ -244,51 +235,53 @@ void CpuManager::SingleCoreRunIdleThread() {
system.CoreTiming().AddTicks(1000U);
idle_count++;
auto& scheduler = physical_core.Scheduler();
- scheduler.TryDoContextSwitch();
+ scheduler.RescheduleCurrentCore();
}
}
void CpuManager::SingleCoreRunSuspendThread() {
auto& kernel = system.Kernel();
- {
- auto& sched = kernel.CurrentScheduler();
- sched.OnThreadStart();
- }
+ kernel.CurrentScheduler()->OnThreadStart();
while (true) {
auto core = kernel.GetCurrentHostThreadID();
- auto& scheduler = kernel.CurrentScheduler();
+ auto& scheduler = *kernel.CurrentScheduler();
Kernel::Thread* current_thread = scheduler.GetCurrentThread();
Common::Fiber::YieldTo(current_thread->GetHostContext(), core_data[0].host_context);
ASSERT(scheduler.ContextSwitchPending());
ASSERT(core == kernel.GetCurrentHostThreadID());
- scheduler.TryDoContextSwitch();
+ scheduler.RescheduleCurrentCore();
}
}
void CpuManager::PreemptSingleCore(bool from_running_enviroment) {
- std::size_t old_core = current_core;
- auto& scheduler = system.Kernel().Scheduler(old_core);
- Kernel::Thread* current_thread = scheduler.GetCurrentThread();
- if (idle_count >= 4 || from_running_enviroment) {
- if (!from_running_enviroment) {
- system.CoreTiming().Idle();
- idle_count = 0;
+ {
+ auto& scheduler = system.Kernel().Scheduler(current_core);
+ Kernel::Thread* current_thread = scheduler.GetCurrentThread();
+ if (idle_count >= 4 || from_running_enviroment) {
+ if (!from_running_enviroment) {
+ system.CoreTiming().Idle();
+ idle_count = 0;
+ }
+ current_thread->SetPhantomMode(true);
+ system.CoreTiming().Advance();
+ current_thread->SetPhantomMode(false);
}
- current_thread->SetPhantomMode(true);
- system.CoreTiming().Advance();
- current_thread->SetPhantomMode(false);
+ current_core.store((current_core + 1) % Core::Hardware::NUM_CPU_CORES);
+ system.CoreTiming().ResetTicks();
+ scheduler.Unload(scheduler.GetCurrentThread());
+
+ auto& next_scheduler = system.Kernel().Scheduler(current_core);
+ Common::Fiber::YieldTo(current_thread->GetHostContext(), next_scheduler.ControlContext());
}
- current_core.store((current_core + 1) % Core::Hardware::NUM_CPU_CORES);
- system.CoreTiming().ResetTicks();
- scheduler.Unload();
- auto& next_scheduler = system.Kernel().Scheduler(current_core);
- Common::Fiber::YieldTo(current_thread->GetHostContext(), next_scheduler.ControlContext());
- /// May have changed scheduler
- auto& current_scheduler = system.Kernel().Scheduler(current_core);
- current_scheduler.Reload();
- auto* currrent_thread2 = current_scheduler.GetCurrentThread();
- if (!currrent_thread2->IsIdleThread()) {
- idle_count = 0;
+
+ // May have changed scheduler
+ {
+ auto& scheduler = system.Kernel().Scheduler(current_core);
+ scheduler.Reload(scheduler.GetCurrentThread());
+ auto* currrent_thread2 = scheduler.GetCurrentThread();
+ if (!currrent_thread2->IsIdleThread()) {
+ idle_count = 0;
+ }
}
}
@@ -369,8 +362,7 @@ void CpuManager::RunThread(std::size_t core) {
return;
}
- auto& scheduler = system.Kernel().CurrentScheduler();
- Kernel::Thread* current_thread = scheduler.GetCurrentThread();
+ auto current_thread = system.Kernel().CurrentScheduler()->GetCurrentThread();
data.is_running = true;
Common::Fiber::YieldTo(data.host_context, current_thread->GetHostContext());
data.is_running = false;
diff --git a/src/core/file_sys/common_funcs.h b/src/core/file_sys/common_funcs.h
new file mode 100644
index 000000000..7ed97aa50
--- /dev/null
+++ b/src/core/file_sys/common_funcs.h
@@ -0,0 +1,56 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "common/common_types.h"
+
+namespace FileSys {
+
+constexpr u64 AOC_TITLE_ID_MASK = 0x7FF;
+constexpr u64 AOC_TITLE_ID_OFFSET = 0x1000;
+constexpr u64 BASE_TITLE_ID_MASK = 0xFFFFFFFFFFFFE000;
+
+/**
+ * Gets the base title ID from a given title ID.
+ *
+ * @param title_id The title ID.
+ * @returns The base title ID.
+ */
+[[nodiscard]] constexpr u64 GetBaseTitleID(u64 title_id) {
+ return title_id & BASE_TITLE_ID_MASK;
+}
+
+/**
+ * Gets the base title ID with a program index offset from a given title ID.
+ *
+ * @param title_id The title ID.
+ * @param program_index The program index.
+ * @returns The base title ID with a program index offset.
+ */
+[[nodiscard]] constexpr u64 GetBaseTitleIDWithProgramIndex(u64 title_id, u64 program_index) {
+ return GetBaseTitleID(title_id) + program_index;
+}
+
+/**
+ * Gets the AOC (Add-On Content) base title ID from a given title ID.
+ *
+ * @param title_id The title ID.
+ * @returns The AOC base title ID.
+ */
+[[nodiscard]] constexpr u64 GetAOCBaseTitleID(u64 title_id) {
+ return GetBaseTitleID(title_id) + AOC_TITLE_ID_OFFSET;
+}
+
+/**
+ * Gets the AOC (Add-On Content) ID from a given AOC title ID.
+ *
+ * @param aoc_title_id The AOC title ID.
+ * @returns The AOC ID.
+ */
+[[nodiscard]] constexpr u64 GetAOCID(u64 aoc_title_id) {
+ return aoc_title_id & AOC_TITLE_ID_MASK;
+}
+
+} // namespace FileSys
diff --git a/src/core/file_sys/content_archive.cpp b/src/core/file_sys/content_archive.cpp
index 76af47ff9..a6c0337fa 100644
--- a/src/core/file_sys/content_archive.cpp
+++ b/src/core/file_sys/content_archive.cpp
@@ -410,8 +410,9 @@ u8 NCA::GetCryptoRevision() const {
std::optional<Core::Crypto::Key128> NCA::GetKeyAreaKey(NCASectionCryptoType type) const {
const auto master_key_id = GetCryptoRevision();
- if (!keys.HasKey(Core::Crypto::S128KeyType::KeyArea, master_key_id, header.key_index))
- return {};
+ if (!keys.HasKey(Core::Crypto::S128KeyType::KeyArea, master_key_id, header.key_index)) {
+ return std::nullopt;
+ }
std::vector<u8> key_area(header.key_area.begin(), header.key_area.end());
Core::Crypto::AESCipher<Core::Crypto::Key128> cipher(
@@ -420,15 +421,17 @@ std::optional<Core::Crypto::Key128> NCA::GetKeyAreaKey(NCASectionCryptoType type
cipher.Transcode(key_area.data(), key_area.size(), key_area.data(), Core::Crypto::Op::Decrypt);
Core::Crypto::Key128 out;
- if (type == NCASectionCryptoType::XTS)
+ if (type == NCASectionCryptoType::XTS) {
std::copy(key_area.begin(), key_area.begin() + 0x10, out.begin());
- else if (type == NCASectionCryptoType::CTR || type == NCASectionCryptoType::BKTR)
+ } else if (type == NCASectionCryptoType::CTR || type == NCASectionCryptoType::BKTR) {
std::copy(key_area.begin() + 0x20, key_area.begin() + 0x30, out.begin());
- else
+ } else {
LOG_CRITICAL(Crypto, "Called GetKeyAreaKey on invalid NCASectionCryptoType type={:02X}",
- static_cast<u8>(type));
+ type);
+ }
+
u128 out_128{};
- memcpy(out_128.data(), out.data(), 16);
+ std::memcpy(out_128.data(), out.data(), sizeof(u128));
LOG_TRACE(Crypto, "called with crypto_rev={:02X}, kak_index={:02X}, key={:016X}{:016X}",
master_key_id, header.key_index, out_128[1], out_128[0]);
@@ -507,7 +510,7 @@ VirtualFile NCA::Decrypt(const NCASectionHeader& s_header, VirtualFile in, u64 s
// TODO(DarkLordZach): Find a test case for XTS-encrypted NCAs
default:
LOG_ERROR(Crypto, "called with unhandled crypto type={:02X}",
- static_cast<u8>(s_header.raw.header.crypto_type));
+ s_header.raw.header.crypto_type);
return nullptr;
}
}
@@ -516,15 +519,17 @@ Loader::ResultStatus NCA::GetStatus() const {
return status;
}
-std::vector<std::shared_ptr<VfsFile>> NCA::GetFiles() const {
- if (status != Loader::ResultStatus::Success)
+std::vector<VirtualFile> NCA::GetFiles() const {
+ if (status != Loader::ResultStatus::Success) {
return {};
+ }
return files;
}
-std::vector<std::shared_ptr<VfsDirectory>> NCA::GetSubdirectories() const {
- if (status != Loader::ResultStatus::Success)
+std::vector<VirtualDir> NCA::GetSubdirectories() const {
+ if (status != Loader::ResultStatus::Success) {
return {};
+ }
return dirs;
}
@@ -532,7 +537,7 @@ std::string NCA::GetName() const {
return file->GetName();
}
-std::shared_ptr<VfsDirectory> NCA::GetParentDirectory() const {
+VirtualDir NCA::GetParentDirectory() const {
return file->GetContainingDirectory();
}
diff --git a/src/core/file_sys/content_archive.h b/src/core/file_sys/content_archive.h
index 69292232a..e9eccdea3 100644
--- a/src/core/file_sys/content_archive.h
+++ b/src/core/file_sys/content_archive.h
@@ -82,7 +82,7 @@ struct NCAHeader {
};
static_assert(sizeof(NCAHeader) == 0x400, "NCAHeader has incorrect size.");
-inline bool IsDirectoryExeFS(const std::shared_ptr<VfsDirectory>& pfs) {
+inline bool IsDirectoryExeFS(const VirtualDir& pfs) {
// According to switchbrew, an exefs must only contain these two files:
return pfs->GetFile("main") != nullptr && pfs->GetFile("main.npdm") != nullptr;
}
@@ -104,10 +104,10 @@ public:
Loader::ResultStatus GetStatus() const;
- std::vector<std::shared_ptr<VfsFile>> GetFiles() const override;
- std::vector<std::shared_ptr<VfsDirectory>> GetSubdirectories() const override;
+ std::vector<VirtualFile> GetFiles() const override;
+ std::vector<VirtualDir> GetSubdirectories() const override;
std::string GetName() const override;
- std::shared_ptr<VfsDirectory> GetParentDirectory() const override;
+ VirtualDir GetParentDirectory() const override;
NCAContentType GetType() const;
u64 GetTitleId() const;
diff --git a/src/core/file_sys/nca_patch.cpp b/src/core/file_sys/nca_patch.cpp
index 5990a2fd5..adcf0732f 100644
--- a/src/core/file_sys/nca_patch.cpp
+++ b/src/core/file_sys/nca_patch.cpp
@@ -191,7 +191,7 @@ bool BKTR::Resize(std::size_t new_size) {
return false;
}
-std::shared_ptr<VfsDirectory> BKTR::GetContainingDirectory() const {
+VirtualDir BKTR::GetContainingDirectory() const {
return base_romfs->GetContainingDirectory();
}
diff --git a/src/core/file_sys/nca_patch.h b/src/core/file_sys/nca_patch.h
index 60c544f8e..503cf473e 100644
--- a/src/core/file_sys/nca_patch.h
+++ b/src/core/file_sys/nca_patch.h
@@ -106,7 +106,7 @@ public:
bool Resize(std::size_t new_size) override;
- std::shared_ptr<VfsDirectory> GetContainingDirectory() const override;
+ VirtualDir GetContainingDirectory() const override;
bool IsWritable() const override;
diff --git a/src/core/file_sys/patch_manager.cpp b/src/core/file_sys/patch_manager.cpp
index e9d1607d0..7c3284df8 100644
--- a/src/core/file_sys/patch_manager.cpp
+++ b/src/core/file_sys/patch_manager.cpp
@@ -12,6 +12,7 @@
#include "common/logging/log.h"
#include "common/string_util.h"
#include "core/core.h"
+#include "core/file_sys/common_funcs.h"
#include "core/file_sys/content_archive.h"
#include "core/file_sys/control_metadata.h"
#include "core/file_sys/ips_layer.h"
@@ -30,7 +31,6 @@ namespace FileSys {
namespace {
constexpr u32 SINGLE_BYTE_MODULUS = 0x100;
-constexpr u64 DLC_BASE_TITLE_ID_MASK = 0xFFFFFFFFFFFFE000;
constexpr std::array<const char*, 14> EXEFS_FILE_NAMES{
"main", "main.npdm", "rtld", "sdk", "subsdk0", "subsdk1", "subsdk2",
@@ -532,7 +532,7 @@ PatchManager::PatchVersionNames PatchManager::GetPatchVersionNames(VirtualFile u
dlc_match.reserve(dlc_entries.size());
std::copy_if(dlc_entries.begin(), dlc_entries.end(), std::back_inserter(dlc_match),
[this](const ContentProviderEntry& entry) {
- return (entry.title_id & DLC_BASE_TITLE_ID_MASK) == title_id &&
+ return GetBaseTitleID(entry.title_id) == title_id &&
content_provider.GetEntry(entry)->GetStatus() ==
Loader::ResultStatus::Success;
});
diff --git a/src/core/file_sys/romfs_factory.cpp b/src/core/file_sys/romfs_factory.cpp
index 987199747..f4e16e4be 100644
--- a/src/core/file_sys/romfs_factory.cpp
+++ b/src/core/file_sys/romfs_factory.cpp
@@ -7,6 +7,7 @@
#include "common/common_types.h"
#include "common/logging/log.h"
#include "core/file_sys/card_image.h"
+#include "core/file_sys/common_funcs.h"
#include "core/file_sys/content_archive.h"
#include "core/file_sys/nca_metadata.h"
#include "core/file_sys/patch_manager.h"
@@ -47,6 +48,27 @@ ResultVal<VirtualFile> RomFSFactory::OpenCurrentProcess(u64 current_process_titl
patch_manager.PatchRomFS(file, ivfc_offset, ContentRecordType::Program, update_raw));
}
+ResultVal<VirtualFile> RomFSFactory::OpenPatchedRomFS(u64 title_id, ContentRecordType type) const {
+ auto nca = content_provider.GetEntry(title_id, type);
+
+ if (nca == nullptr) {
+ // TODO: Find the right error code to use here
+ return RESULT_UNKNOWN;
+ }
+
+ const PatchManager patch_manager{title_id, filesystem_controller, content_provider};
+
+ return MakeResult<VirtualFile>(
+ patch_manager.PatchRomFS(nca->GetRomFS(), nca->GetBaseIVFCOffset(), type));
+}
+
+ResultVal<VirtualFile> RomFSFactory::OpenPatchedRomFSWithProgramIndex(
+ u64 title_id, u8 program_index, ContentRecordType type) const {
+ const auto res_title_id = GetBaseTitleIDWithProgramIndex(title_id, program_index);
+
+ return OpenPatchedRomFS(res_title_id, type);
+}
+
ResultVal<VirtualFile> RomFSFactory::Open(u64 title_id, StorageId storage,
ContentRecordType type) const {
const std::shared_ptr<NCA> res = GetEntry(title_id, storage, type);
diff --git a/src/core/file_sys/romfs_factory.h b/src/core/file_sys/romfs_factory.h
index ec704dfa8..96dd0d578 100644
--- a/src/core/file_sys/romfs_factory.h
+++ b/src/core/file_sys/romfs_factory.h
@@ -42,6 +42,10 @@ public:
void SetPackedUpdate(VirtualFile update_raw);
[[nodiscard]] ResultVal<VirtualFile> OpenCurrentProcess(u64 current_process_title_id) const;
+ [[nodiscard]] ResultVal<VirtualFile> OpenPatchedRomFS(u64 title_id,
+ ContentRecordType type) const;
+ [[nodiscard]] ResultVal<VirtualFile> OpenPatchedRomFSWithProgramIndex(
+ u64 title_id, u8 program_index, ContentRecordType type) const;
[[nodiscard]] ResultVal<VirtualFile> Open(u64 title_id, StorageId storage,
ContentRecordType type) const;
diff --git a/src/core/file_sys/system_archive/data/font_nintendo_extended.cpp b/src/core/file_sys/system_archive/data/font_nintendo_extended.cpp
index 69d62ce8f..29ef110a6 100644
--- a/src/core/file_sys/system_archive/data/font_nintendo_extended.cpp
+++ b/src/core/file_sys/system_archive/data/font_nintendo_extended.cpp
@@ -6,191 +6,384 @@
namespace FileSys::SystemArchive::SharedFontData {
-const std::array<unsigned char, 2932> FONT_NINTENDO_EXTENDED{{
- 0x00, 0x01, 0x00, 0x00, 0x00, 0x0f, 0x00, 0x80, 0x00, 0x03, 0x00, 0x70, 0x44, 0x53, 0x49, 0x47,
- 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x0b, 0x6c, 0x00, 0x00, 0x00, 0x08, 0x4f, 0x53, 0x2f, 0x32,
- 0x33, 0x86, 0x1d, 0x9b, 0x00, 0x00, 0x01, 0x78, 0x00, 0x00, 0x00, 0x60, 0x63, 0x6d, 0x61, 0x70,
- 0xc2, 0x06, 0x20, 0xde, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00, 0x00, 0xa0, 0x63, 0x76, 0x74, 0x20,
- 0x00, 0x14, 0x00, 0x00, 0x00, 0x00, 0x04, 0x2c, 0x00, 0x00, 0x00, 0x06, 0x66, 0x70, 0x67, 0x6d,
- 0x06, 0x59, 0x9c, 0x37, 0x00, 0x00, 0x02, 0xa0, 0x00, 0x00, 0x01, 0x73, 0x67, 0x61, 0x73, 0x70,
- 0x00, 0x00, 0x00, 0x10, 0x00, 0x00, 0x0b, 0x64, 0x00, 0x00, 0x00, 0x08, 0x67, 0x6c, 0x79, 0x66,
- 0x10, 0x31, 0x88, 0x00, 0x00, 0x00, 0x04, 0x34, 0x00, 0x00, 0x04, 0x64, 0x68, 0x65, 0x61, 0x64,
- 0x15, 0x9d, 0xef, 0x91, 0x00, 0x00, 0x00, 0xfc, 0x00, 0x00, 0x00, 0x36, 0x68, 0x68, 0x65, 0x61,
- 0x09, 0x60, 0x03, 0x71, 0x00, 0x00, 0x01, 0x34, 0x00, 0x00, 0x00, 0x24, 0x68, 0x6d, 0x74, 0x78,
- 0x0d, 0x2e, 0x03, 0xa7, 0x00, 0x00, 0x01, 0xd8, 0x00, 0x00, 0x00, 0x26, 0x6c, 0x6f, 0x63, 0x61,
- 0x05, 0xc0, 0x04, 0x6c, 0x00, 0x00, 0x08, 0x98, 0x00, 0x00, 0x00, 0x1e, 0x6d, 0x61, 0x78, 0x70,
- 0x02, 0x1c, 0x00, 0x5f, 0x00, 0x00, 0x01, 0x58, 0x00, 0x00, 0x00, 0x20, 0x6e, 0x61, 0x6d, 0x65,
- 0x7c, 0xe0, 0x84, 0x5c, 0x00, 0x00, 0x08, 0xb8, 0x00, 0x00, 0x02, 0x09, 0x70, 0x6f, 0x73, 0x74,
- 0x47, 0x4e, 0x74, 0x19, 0x00, 0x00, 0x0a, 0xc4, 0x00, 0x00, 0x00, 0x9e, 0x70, 0x72, 0x65, 0x70,
- 0x1c, 0xfc, 0x7d, 0x9c, 0x00, 0x00, 0x04, 0x14, 0x00, 0x00, 0x00, 0x16, 0x00, 0x01, 0x00, 0x00,
- 0x00, 0x01, 0x00, 0x00, 0x7c, 0xc7, 0xb1, 0x63, 0x5f, 0x0f, 0x3c, 0xf5, 0x00, 0x1b, 0x03, 0xe8,
- 0x00, 0x00, 0x00, 0x00, 0xd9, 0x44, 0x2f, 0x5d, 0x00, 0x00, 0x00, 0x00, 0xd9, 0x45, 0x7b, 0x69,
- 0x00, 0x00, 0x00, 0x00, 0x03, 0xe6, 0x03, 0xe8, 0x00, 0x00, 0x00, 0x06, 0x00, 0x02, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x03, 0x84, 0xff, 0x83, 0x01, 0xf4, 0x03, 0xe8,
- 0x00, 0x00, 0x00, 0x00, 0x03, 0xe6, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x05, 0x00, 0x01, 0x00, 0x00, 0x00, 0x0e, 0x00, 0x5e,
- 0x00, 0x03, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x0a, 0x00, 0x00,
- 0x02, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x04, 0x03, 0x74, 0x01, 0x90, 0x00, 0x05,
- 0x00, 0x04, 0x00, 0xcd, 0x00, 0xcd, 0x00, 0x00, 0x01, 0x1f, 0x00, 0xcd, 0x00, 0xcd, 0x00, 0x00,
- 0x03, 0xc3, 0x00, 0x66, 0x02, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+const std::array<unsigned char, 6024> FONT_NINTENDO_EXTENDED{{
+ 0x00, 0x01, 0x00, 0x00, 0x00, 0x0E, 0x00, 0x80, 0x00, 0x03, 0x00, 0x60, 0x4F, 0x53, 0x2F, 0x32,
+ 0x34, 0x00, 0x1E, 0x26, 0x00, 0x00, 0x01, 0x68, 0x00, 0x00, 0x00, 0x60, 0x63, 0x6D, 0x61, 0x70,
+ 0xC1, 0xE7, 0xC8, 0xF3, 0x00, 0x00, 0x02, 0x0C, 0x00, 0x00, 0x01, 0x72, 0x63, 0x76, 0x74, 0x20,
+ 0x00, 0x14, 0x00, 0x00, 0x00, 0x00, 0x05, 0x0C, 0x00, 0x00, 0x00, 0x06, 0x66, 0x70, 0x67, 0x6D,
+ 0x06, 0x59, 0x9C, 0x37, 0x00, 0x00, 0x03, 0x80, 0x00, 0x00, 0x01, 0x73, 0x67, 0x61, 0x73, 0x70,
+ 0x00, 0x00, 0x00, 0x10, 0x00, 0x00, 0x17, 0x80, 0x00, 0x00, 0x00, 0x08, 0x67, 0x6C, 0x79, 0x66,
+ 0x50, 0x0B, 0xEA, 0xFA, 0x00, 0x00, 0x05, 0x50, 0x00, 0x00, 0x0F, 0x04, 0x68, 0x65, 0x61, 0x64,
+ 0x18, 0x65, 0x81, 0x09, 0x00, 0x00, 0x00, 0xEC, 0x00, 0x00, 0x00, 0x36, 0x68, 0x68, 0x65, 0x61,
+ 0x09, 0x88, 0x03, 0x86, 0x00, 0x00, 0x01, 0x24, 0x00, 0x00, 0x00, 0x24, 0x68, 0x6D, 0x74, 0x78,
+ 0x0A, 0xF0, 0x01, 0x94, 0x00, 0x00, 0x01, 0xC8, 0x00, 0x00, 0x00, 0x42, 0x6C, 0x6F, 0x63, 0x61,
+ 0x34, 0x80, 0x30, 0x6E, 0x00, 0x00, 0x05, 0x14, 0x00, 0x00, 0x00, 0x3A, 0x6D, 0x61, 0x78, 0x70,
+ 0x02, 0x2C, 0x00, 0x72, 0x00, 0x00, 0x01, 0x48, 0x00, 0x00, 0x00, 0x20, 0x6E, 0x61, 0x6D, 0x65,
+ 0xDB, 0xC5, 0x42, 0x4D, 0x00, 0x00, 0x14, 0x54, 0x00, 0x00, 0x01, 0xFE, 0x70, 0x6F, 0x73, 0x74,
+ 0xF4, 0xB4, 0xAC, 0xAB, 0x00, 0x00, 0x16, 0x54, 0x00, 0x00, 0x01, 0x2A, 0x70, 0x72, 0x65, 0x70,
+ 0x1C, 0xFC, 0x7D, 0x9C, 0x00, 0x00, 0x04, 0xF4, 0x00, 0x00, 0x00, 0x16, 0x00, 0x01, 0x00, 0x00,
+ 0x00, 0x01, 0x00, 0x00, 0xC9, 0x16, 0x5B, 0x71, 0x5F, 0x0F, 0x3C, 0xF5, 0x00, 0x0B, 0x04, 0x00,
+ 0x00, 0x00, 0x00, 0x00, 0xD9, 0x44, 0x2F, 0x5D, 0x00, 0x00, 0x00, 0x00, 0xDC, 0x02, 0x0D, 0xA7,
+ 0x00, 0x14, 0xFF, 0x98, 0x03, 0xEC, 0x03, 0x70, 0x00, 0x00, 0x00, 0x08, 0x00, 0x02, 0x00, 0x00,
+ 0x00, 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x03, 0x9A, 0xFF, 0x80, 0x02, 0x00, 0x04, 0x00,
+ 0x00, 0x00, 0x00, 0x00, 0x03, 0xEC, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x05, 0x00, 0x01, 0x00, 0x00, 0x00, 0x1C, 0x00, 0x71,
+ 0x00, 0x05, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x0A, 0x00, 0x00,
+ 0x02, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x04, 0x03, 0xC4, 0x01, 0x90, 0x00, 0x05,
+ 0x00, 0x04, 0x00, 0xD2, 0x00, 0xD2, 0x00, 0x00, 0x01, 0x26, 0x00, 0xD2, 0x00, 0xD2, 0x00, 0x00,
+ 0x03, 0xDA, 0x00, 0x68, 0x02, 0x0C, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x01, 0x10, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x20, 0x20, 0x20, 0x20, 0x00, 0xc0, 0x00, 0x00, 0xe0, 0xe9, 0x03, 0x84, 0xff, 0x83,
- 0x01, 0xf4, 0x02, 0xee, 0x00, 0xfa, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x03, 0xe8,
- 0x02, 0xbc, 0x00, 0x00, 0x00, 0x20, 0x00, 0x00, 0x03, 0xe8, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0x00, 0xfa, 0x00, 0x00, 0x00, 0xfa, 0x00, 0x00, 0x03, 0xe8, 0x00, 0xeb, 0x01, 0x21, 0x00, 0xff,
- 0x00, 0xff, 0x01, 0x3d, 0x01, 0x17, 0x00, 0x42, 0x00, 0x1c, 0x00, 0x3e, 0x00, 0x17, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00, 0x68, 0x00, 0x01, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x1c, 0x00, 0x03, 0x00, 0x01, 0x00, 0x00, 0x00, 0x68, 0x00, 0x06, 0x00, 0x4c,
- 0x00, 0x00, 0x00, 0x00, 0x00, 0x21, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+ 0x00, 0x00, 0x20, 0x20, 0x20, 0x20, 0x00, 0xC0, 0x00, 0x0D, 0xE0, 0xF0, 0x03, 0x9A, 0xFF, 0x80,
+ 0x02, 0x00, 0x03, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x04, 0x00,
+ 0x02, 0xCD, 0x00, 0x00, 0x00, 0x20, 0x00, 0x01, 0x04, 0x00, 0x00, 0xA4, 0x00, 0x00, 0x00, 0x00,
+ 0x01, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x14, 0x00, 0x14, 0x00, 0x14,
+ 0x00, 0x14, 0x00, 0x14, 0x00, 0x14, 0x00, 0x14, 0x00, 0x14, 0x00, 0x14, 0x00, 0x14, 0x00, 0x14,
+ 0x00, 0x14, 0x00, 0x14, 0x00, 0x14, 0x00, 0x14, 0x00, 0x14, 0x00, 0x14, 0x00, 0x14, 0x00, 0x14,
+ 0x00, 0x14, 0x00, 0x14, 0x00, 0x14, 0x00, 0x14, 0x00, 0x14, 0x00, 0x00, 0x00, 0x00, 0x00, 0x03,
+ 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00, 0x1C, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x6C,
+ 0x00, 0x03, 0x00, 0x01, 0x00, 0x00, 0x00, 0x1C, 0x00, 0x04, 0x00, 0x50, 0x00, 0x00, 0x00, 0x10,
+ 0x00, 0x10, 0x00, 0x03, 0x00, 0x00, 0x00, 0x00, 0x00, 0x0D, 0x00, 0x20, 0xE0, 0xA9, 0xE0, 0xB4,
+ 0xE0, 0xE9, 0xE0, 0xF0, 0xFF, 0xFF, 0x00, 0x00, 0x00, 0x00, 0x00, 0x0D, 0x00, 0x20, 0xE0, 0xA0,
+ 0xE0, 0xB3, 0xE0, 0xE0, 0xE0, 0xEF, 0xFF, 0xFF, 0x00, 0x01, 0xFF, 0xF5, 0xFF, 0xE3, 0x1F, 0x64,
+ 0x1F, 0x5B, 0x1F, 0x30, 0x1F, 0x2B, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01, 0x06, 0x00, 0x00, 0x01, 0x00,
+ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01, 0x02, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00, 0x00,
+ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x03, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0x00, 0x02, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x03, 0x00, 0x04, 0x00, 0x38, 0x00, 0x00, 0x00, 0x0a,
- 0x00, 0x08, 0x00, 0x02, 0x00, 0x02, 0x00, 0x00, 0x00, 0x0d, 0x00, 0x20, 0xe0, 0xe9, 0xff, 0xff,
- 0x00, 0x00, 0x00, 0x00, 0x00, 0x0d, 0x00, 0x20, 0xe0, 0xe0, 0xff, 0xff, 0x00, 0x01, 0xff, 0xf5,
- 0xff, 0xe3, 0x1f, 0x24, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0xb8, 0x00, 0x00, 0x2c, 0x4b, 0xb8, 0x00, 0x09, 0x50, 0x58, 0xb1, 0x01, 0x01, 0x8e, 0x59, 0xb8,
- 0x01, 0xff, 0x85, 0xb8, 0x00, 0x44, 0x1d, 0xb9, 0x00, 0x09, 0x00, 0x03, 0x5f, 0x5e, 0x2d, 0xb8,
- 0x00, 0x01, 0x2c, 0x20, 0x20, 0x45, 0x69, 0x44, 0xb0, 0x01, 0x60, 0x2d, 0xb8, 0x00, 0x02, 0x2c,
- 0xb8, 0x00, 0x01, 0x2a, 0x21, 0x2d, 0xb8, 0x00, 0x03, 0x2c, 0x20, 0x46, 0xb0, 0x03, 0x25, 0x46,
- 0x52, 0x58, 0x23, 0x59, 0x20, 0x8a, 0x20, 0x8a, 0x49, 0x64, 0x8a, 0x20, 0x46, 0x20, 0x68, 0x61,
- 0x64, 0xb0, 0x04, 0x25, 0x46, 0x20, 0x68, 0x61, 0x64, 0x52, 0x58, 0x23, 0x65, 0x8a, 0x59, 0x2f,
- 0x20, 0xb0, 0x00, 0x53, 0x58, 0x69, 0x20, 0xb0, 0x00, 0x54, 0x58, 0x21, 0xb0, 0x40, 0x59, 0x1b,
- 0x69, 0x20, 0xb0, 0x00, 0x54, 0x58, 0x21, 0xb0, 0x40, 0x65, 0x59, 0x59, 0x3a, 0x2d, 0xb8, 0x00,
- 0x04, 0x2c, 0x20, 0x46, 0xb0, 0x04, 0x25, 0x46, 0x52, 0x58, 0x23, 0x8a, 0x59, 0x20, 0x46, 0x20,
- 0x6a, 0x61, 0x64, 0xb0, 0x04, 0x25, 0x46, 0x20, 0x6a, 0x61, 0x64, 0x52, 0x58, 0x23, 0x8a, 0x59,
- 0x2f, 0xfd, 0x2d, 0xb8, 0x00, 0x05, 0x2c, 0x4b, 0x20, 0xb0, 0x03, 0x26, 0x50, 0x58, 0x51, 0x58,
- 0xb0, 0x80, 0x44, 0x1b, 0xb0, 0x40, 0x44, 0x59, 0x1b, 0x21, 0x21, 0x20, 0x45, 0xb0, 0xc0, 0x50,
- 0x58, 0xb0, 0xc0, 0x44, 0x1b, 0x21, 0x59, 0x59, 0x2d, 0xb8, 0x00, 0x06, 0x2c, 0x20, 0x20, 0x45,
- 0x69, 0x44, 0xb0, 0x01, 0x60, 0x20, 0x20, 0x45, 0x7d, 0x69, 0x18, 0x44, 0xb0, 0x01, 0x60, 0x2d,
- 0xb8, 0x00, 0x07, 0x2c, 0xb8, 0x00, 0x06, 0x2a, 0x2d, 0xb8, 0x00, 0x08, 0x2c, 0x4b, 0x20, 0xb0,
- 0x03, 0x26, 0x53, 0x58, 0xb0, 0x40, 0x1b, 0xb0, 0x00, 0x59, 0x8a, 0x8a, 0x20, 0xb0, 0x03, 0x26,
- 0x53, 0x58, 0x23, 0x21, 0xb0, 0x80, 0x8a, 0x8a, 0x1b, 0x8a, 0x23, 0x59, 0x20, 0xb0, 0x03, 0x26,
- 0x53, 0x58, 0x23, 0x21, 0xb8, 0x00, 0xc0, 0x8a, 0x8a, 0x1b, 0x8a, 0x23, 0x59, 0x20, 0xb0, 0x03,
- 0x26, 0x53, 0x58, 0x23, 0x21, 0xb8, 0x01, 0x00, 0x8a, 0x8a, 0x1b, 0x8a, 0x23, 0x59, 0x20, 0xb0,
- 0x03, 0x26, 0x53, 0x58, 0x23, 0x21, 0xb8, 0x01, 0x40, 0x8a, 0x8a, 0x1b, 0x8a, 0x23, 0x59, 0x20,
- 0xb8, 0x00, 0x03, 0x26, 0x53, 0x58, 0xb0, 0x03, 0x25, 0x45, 0xb8, 0x01, 0x80, 0x50, 0x58, 0x23,
- 0x21, 0xb8, 0x01, 0x80, 0x23, 0x21, 0x1b, 0xb0, 0x03, 0x25, 0x45, 0x23, 0x21, 0x23, 0x21, 0x59,
- 0x1b, 0x21, 0x59, 0x44, 0x2d, 0xb8, 0x00, 0x09, 0x2c, 0x4b, 0x53, 0x58, 0x45, 0x44, 0x1b, 0x21,
- 0x21, 0x59, 0x2d, 0x00, 0xb8, 0x00, 0x00, 0x2b, 0x00, 0xba, 0x00, 0x01, 0x00, 0x01, 0x00, 0x07,
- 0x2b, 0xb8, 0x00, 0x00, 0x20, 0x45, 0x7d, 0x69, 0x18, 0x44, 0x00, 0x00, 0x00, 0x14, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x02, 0x03, 0xe6, 0x03, 0xe8, 0x00, 0x06,
- 0x00, 0x00, 0x35, 0x01, 0x33, 0x15, 0x01, 0x23, 0x35, 0x03, 0x52, 0x94, 0xfc, 0xa6, 0x8c, 0x90,
- 0x03, 0x58, 0x86, 0xfc, 0xa0, 0x8e, 0x00, 0x00, 0x00, 0x02, 0x00, 0xeb, 0x00, 0xcc, 0x02, 0xfb,
- 0x03, 0x1e, 0x00, 0x08, 0x00, 0x0f, 0x00, 0x00, 0x01, 0x33, 0x13, 0x23, 0x27, 0x23, 0x07, 0x23,
- 0x13, 0x17, 0x07, 0x06, 0x15, 0x33, 0x27, 0x07, 0x01, 0xbc, 0x6d, 0xd2, 0x7c, 0x26, 0xcc, 0x26,
- 0x7c, 0xd1, 0x35, 0x40, 0x02, 0x89, 0x45, 0x02, 0x03, 0x1e, 0xfd, 0xae, 0x77, 0x77, 0x02, 0x52,
- 0x9b, 0xcc, 0x08, 0x04, 0xda, 0x02, 0x00, 0x00, 0x00, 0x03, 0x01, 0x21, 0x00, 0xcc, 0x02, 0xc5,
- 0x03, 0x1e, 0x00, 0x15, 0x00, 0x1f, 0x00, 0x2b, 0x00, 0x00, 0x25, 0x11, 0x33, 0x32, 0x1e, 0x02,
- 0x15, 0x14, 0x0e, 0x02, 0x07, 0x1e, 0x01, 0x15, 0x14, 0x0e, 0x02, 0x2b, 0x01, 0x13, 0x33, 0x32,
- 0x36, 0x35, 0x34, 0x26, 0x2b, 0x01, 0x1d, 0x01, 0x33, 0x32, 0x3e, 0x02, 0x35, 0x34, 0x26, 0x2b,
- 0x01, 0x15, 0x01, 0x21, 0xea, 0x25, 0x3f, 0x2e, 0x1a, 0x0e, 0x15, 0x1b, 0x0e, 0x2d, 0x2d, 0x1a,
- 0x2e, 0x3f, 0x25, 0xf8, 0x76, 0x62, 0x20, 0x2a, 0x28, 0x22, 0x62, 0x76, 0x10, 0x18, 0x11, 0x09,
- 0x22, 0x22, 0x74, 0xcc, 0x02, 0x52, 0x18, 0x2b, 0x3c, 0x24, 0x1d, 0x1f, 0x17, 0x17, 0x14, 0x0f,
- 0x48, 0x2f, 0x24, 0x3f, 0x2e, 0x1a, 0x01, 0x5b, 0x29, 0x20, 0x20, 0x2b, 0x94, 0xf8, 0x0e, 0x16,
- 0x1c, 0x0e, 0x1f, 0x31, 0x9e, 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0xff, 0x00, 0xcc, 0x02, 0xe7,
- 0x03, 0x1e, 0x00, 0x0c, 0x00, 0x00, 0x01, 0x33, 0x17, 0x37, 0x33, 0x03, 0x13, 0x23, 0x27, 0x07,
- 0x23, 0x13, 0x03, 0x01, 0x04, 0x86, 0x69, 0x69, 0x86, 0xa3, 0xa8, 0x88, 0x6c, 0x6c, 0x88, 0xa8,
- 0xa3, 0x03, 0x1e, 0xcb, 0xcb, 0xfe, 0xda, 0xfe, 0xd4, 0xcf, 0xcf, 0x01, 0x2c, 0x01, 0x26, 0x00,
- 0x00, 0x01, 0x00, 0xff, 0x00, 0xcc, 0x02, 0xe7, 0x03, 0x1e, 0x00, 0x0f, 0x00, 0x00, 0x01, 0x03,
- 0x33, 0x17, 0x32, 0x15, 0x1e, 0x01, 0x15, 0x1b, 0x01, 0x33, 0x03, 0x15, 0x23, 0x35, 0x01, 0xb8,
- 0xb9, 0x7e, 0x01, 0x01, 0x01, 0x03, 0x70, 0x75, 0x7f, 0xb9, 0x76, 0x01, 0xa3, 0x01, 0x7b, 0x01,
- 0x01, 0x01, 0x05, 0x02, 0xff, 0x00, 0x01, 0x0a, 0xfe, 0x85, 0xd7, 0xd7, 0x00, 0x01, 0x01, 0x3d,
- 0x00, 0xcc, 0x02, 0xa9, 0x03, 0x1e, 0x00, 0x06, 0x00, 0x00, 0x25, 0x11, 0x33, 0x11, 0x33, 0x15,
- 0x21, 0x01, 0x3d, 0x75, 0xf7, 0xfe, 0x94, 0xcc, 0x02, 0x52, 0xfe, 0x10, 0x62, 0x00, 0x00, 0x00,
- 0x00, 0x02, 0x01, 0x17, 0x00, 0xbc, 0x02, 0xcf, 0x03, 0x0e, 0x00, 0x15, 0x00, 0x21, 0x00, 0x00,
- 0x25, 0x11, 0x33, 0x32, 0x1e, 0x02, 0x1d, 0x01, 0x0e, 0x03, 0x1d, 0x01, 0x17, 0x15, 0x23, 0x27,
- 0x23, 0x15, 0x23, 0x13, 0x33, 0x32, 0x3e, 0x02, 0x35, 0x34, 0x26, 0x2b, 0x01, 0x15, 0x01, 0x17,
- 0xf4, 0x27, 0x40, 0x2e, 0x19, 0x01, 0x1f, 0x24, 0x1e, 0x78, 0x7d, 0x6a, 0x5c, 0x75, 0x76, 0x72,
- 0x12, 0x19, 0x11, 0x08, 0x26, 0x26, 0x6a, 0xbc, 0x02, 0x52, 0x1d, 0x31, 0x42, 0x25, 0x16, 0x18,
- 0x32, 0x2a, 0x1b, 0x02, 0x01, 0xef, 0x06, 0xd7, 0xd7, 0x01, 0x3f, 0x10, 0x1a, 0x1e, 0x0f, 0x23,
- 0x36, 0xb0, 0x00, 0x00, 0x00, 0x02, 0x00, 0x42, 0x00, 0xbc, 0x03, 0xa4, 0x03, 0x0e, 0x00, 0x0a,
- 0x00, 0x11, 0x00, 0x00, 0x13, 0x35, 0x21, 0x15, 0x01, 0x21, 0x15, 0x21, 0x35, 0x01, 0x21, 0x01,
- 0x11, 0x33, 0x11, 0x33, 0x15, 0x21, 0x42, 0x01, 0xa7, 0xfe, 0xeb, 0x01, 0x1b, 0xfe, 0x53, 0x01,
- 0x15, 0xfe, 0xeb, 0x01, 0xf7, 0x75, 0xf6, 0xfe, 0x95, 0x02, 0xac, 0x62, 0x45, 0xfe, 0x55, 0x62,
- 0x47, 0x01, 0xa9, 0xfe, 0x10, 0x02, 0x52, 0xfe, 0x10, 0x62, 0x00, 0x00, 0x00, 0x03, 0x00, 0x1c,
- 0x00, 0xbc, 0x03, 0xca, 0x03, 0x0e, 0x00, 0x0a, 0x00, 0x21, 0x00, 0x2f, 0x00, 0x00, 0x13, 0x35,
- 0x21, 0x15, 0x01, 0x21, 0x15, 0x21, 0x35, 0x01, 0x21, 0x01, 0x11, 0x33, 0x32, 0x1e, 0x02, 0x15,
- 0x14, 0x06, 0x07, 0x0e, 0x03, 0x15, 0x17, 0x15, 0x23, 0x27, 0x23, 0x15, 0x23, 0x13, 0x33, 0x32,
- 0x3e, 0x02, 0x35, 0x34, 0x2e, 0x02, 0x2b, 0x01, 0x15, 0x1c, 0x01, 0xa7, 0xfe, 0xeb, 0x01, 0x1b,
- 0xfe, 0x53, 0x01, 0x15, 0xfe, 0xeb, 0x01, 0xf7, 0xf3, 0x27, 0x41, 0x2d, 0x19, 0x1c, 0x20, 0x01,
- 0x0d, 0x0e, 0x0a, 0x78, 0x7d, 0x69, 0x5c, 0x75, 0x76, 0x71, 0x11, 0x1a, 0x12, 0x09, 0x0a, 0x14,
- 0x1d, 0x13, 0x69, 0x02, 0xac, 0x62, 0x45, 0xfe, 0x55, 0x62, 0x47, 0x01, 0xa9, 0xfe, 0x10, 0x02,
- 0x52, 0x1d, 0x31, 0x42, 0x25, 0x2b, 0x44, 0x1d, 0x01, 0x08, 0x09, 0x07, 0x01, 0xf1, 0x06, 0xd7,
- 0xd7, 0x01, 0x3f, 0x11, 0x19, 0x1f, 0x0e, 0x11, 0x20, 0x19, 0x0f, 0xb0, 0x00, 0x02, 0x00, 0x3e,
- 0x00, 0xb3, 0x03, 0xa8, 0x03, 0x17, 0x00, 0x3a, 0x00, 0x41, 0x00, 0x00, 0x13, 0x34, 0x3e, 0x02,
- 0x33, 0x32, 0x1e, 0x02, 0x15, 0x23, 0x27, 0x34, 0x27, 0x2e, 0x01, 0x23, 0x22, 0x0e, 0x02, 0x15,
- 0x14, 0x16, 0x15, 0x1e, 0x05, 0x15, 0x14, 0x0e, 0x02, 0x23, 0x22, 0x2e, 0x02, 0x35, 0x33, 0x1e,
- 0x01, 0x33, 0x32, 0x3e, 0x02, 0x35, 0x34, 0x2e, 0x04, 0x35, 0x01, 0x11, 0x33, 0x11, 0x33, 0x15,
- 0x21, 0x50, 0x24, 0x3b, 0x4a, 0x27, 0x28, 0x4b, 0x39, 0x22, 0x73, 0x01, 0x01, 0x08, 0x2b, 0x29,
- 0x10, 0x20, 0x19, 0x0f, 0x01, 0x0b, 0x35, 0x41, 0x46, 0x3b, 0x25, 0x23, 0x3a, 0x4b, 0x27, 0x2b,
- 0x50, 0x3f, 0x26, 0x74, 0x05, 0x34, 0x33, 0x10, 0x20, 0x1a, 0x11, 0x2c, 0x42, 0x4d, 0x42, 0x2c,
- 0x01, 0xef, 0x73, 0xf6, 0xfe, 0x97, 0x02, 0x70, 0x2a, 0x3f, 0x2a, 0x14, 0x18, 0x2e, 0x44, 0x2c,
- 0x02, 0x03, 0x01, 0x27, 0x27, 0x07, 0x10, 0x1a, 0x12, 0x02, 0x0b, 0x02, 0x1f, 0x22, 0x19, 0x17,
- 0x27, 0x3f, 0x34, 0x2c, 0x3e, 0x28, 0x13, 0x1a, 0x32, 0x48, 0x2e, 0x30, 0x30, 0x06, 0x0f, 0x1a,
- 0x13, 0x21, 0x27, 0x1e, 0x1b, 0x29, 0x3e, 0x31, 0xfe, 0x4c, 0x02, 0x53, 0xfe, 0x10, 0x63, 0x00,
- 0x00, 0x03, 0x00, 0x17, 0x00, 0xb3, 0x03, 0xce, 0x03, 0x17, 0x00, 0x38, 0x00, 0x4f, 0x00, 0x5d,
- 0x00, 0x00, 0x13, 0x34, 0x3e, 0x02, 0x33, 0x32, 0x1e, 0x02, 0x15, 0x23, 0x27, 0x34, 0x23, 0x2e,
- 0x01, 0x23, 0x22, 0x0e, 0x02, 0x15, 0x14, 0x1e, 0x04, 0x15, 0x14, 0x0e, 0x02, 0x23, 0x22, 0x2e,
- 0x02, 0x35, 0x33, 0x1e, 0x01, 0x33, 0x32, 0x3e, 0x02, 0x35, 0x34, 0x26, 0x27, 0x2e, 0x03, 0x35,
- 0x01, 0x11, 0x33, 0x32, 0x1e, 0x02, 0x15, 0x14, 0x06, 0x07, 0x30, 0x0e, 0x02, 0x31, 0x17, 0x15,
- 0x23, 0x27, 0x23, 0x15, 0x23, 0x13, 0x33, 0x32, 0x3e, 0x02, 0x35, 0x34, 0x2e, 0x02, 0x2b, 0x01,
- 0x15, 0x2a, 0x24, 0x3a, 0x4a, 0x26, 0x29, 0x4b, 0x39, 0x23, 0x73, 0x01, 0x01, 0x08, 0x2a, 0x2a,
- 0x10, 0x1f, 0x1a, 0x10, 0x2c, 0x42, 0x4d, 0x42, 0x2c, 0x23, 0x39, 0x4b, 0x27, 0x2b, 0x51, 0x3f,
- 0x27, 0x75, 0x05, 0x34, 0x33, 0x10, 0x20, 0x1a, 0x10, 0x1f, 0x1c, 0x25, 0x53, 0x47, 0x2e, 0x01,
- 0xed, 0xf3, 0x27, 0x41, 0x2d, 0x19, 0x1c, 0x20, 0x0c, 0x0e, 0x0c, 0x78, 0x7d, 0x68, 0x5d, 0x75,
- 0x76, 0x71, 0x11, 0x1a, 0x12, 0x09, 0x0a, 0x14, 0x1d, 0x13, 0x69, 0x02, 0x71, 0x2a, 0x3e, 0x2a,
- 0x14, 0x18, 0x2e, 0x44, 0x2c, 0x02, 0x02, 0x27, 0x29, 0x07, 0x11, 0x1a, 0x12, 0x1d, 0x24, 0x1c,
- 0x1d, 0x2b, 0x40, 0x32, 0x2c, 0x3f, 0x29, 0x13, 0x1a, 0x31, 0x49, 0x2e, 0x30, 0x30, 0x06, 0x0f,
- 0x19, 0x13, 0x1e, 0x22, 0x0b, 0x0e, 0x20, 0x2f, 0x43, 0x30, 0xfe, 0x4b, 0x02, 0x52, 0x1d, 0x32,
- 0x42, 0x25, 0x2c, 0x42, 0x1d, 0x08, 0x0a, 0x08, 0xf1, 0x06, 0xd7, 0xd7, 0x01, 0x3f, 0x11, 0x19,
- 0x1f, 0x0e, 0x11, 0x20, 0x19, 0x0f, 0xb0, 0x00, 0x00, 0x00, 0x00, 0x12, 0x00, 0x12, 0x00, 0x12,
- 0x00, 0x12, 0x00, 0x32, 0x00, 0x72, 0x00, 0x8e, 0x00, 0xac, 0x00, 0xbe, 0x00, 0xf0, 0x01, 0x14,
- 0x01, 0x5c, 0x01, 0xb6, 0x02, 0x32, 0x00, 0x00, 0x00, 0x00, 0x00, 0x0d, 0x00, 0xa2, 0x00, 0x01,
- 0x00, 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x10, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00,
- 0x00, 0x02, 0x00, 0x07, 0x00, 0x10, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x03, 0x00, 0x2f,
- 0x00, 0x17, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x04, 0x00, 0x12, 0x00, 0x46, 0x00, 0x01,
- 0x00, 0x00, 0x00, 0x00, 0x00, 0x05, 0x00, 0x0d, 0x00, 0x58, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00,
- 0x00, 0x06, 0x00, 0x12, 0x00, 0x65, 0x00, 0x03, 0x00, 0x01, 0x04, 0x09, 0x00, 0x01, 0x00, 0x20,
- 0x00, 0x77, 0x00, 0x03, 0x00, 0x01, 0x04, 0x09, 0x00, 0x02, 0x00, 0x0e, 0x00, 0x97, 0x00, 0x03,
- 0x00, 0x01, 0x04, 0x09, 0x00, 0x03, 0x00, 0x5e, 0x00, 0xa5, 0x00, 0x03, 0x00, 0x01, 0x04, 0x09,
- 0x00, 0x04, 0x00, 0x24, 0x01, 0x03, 0x00, 0x03, 0x00, 0x01, 0x04, 0x09, 0x00, 0x05, 0x00, 0x1a,
- 0x01, 0x27, 0x00, 0x03, 0x00, 0x01, 0x04, 0x09, 0x00, 0x06, 0x00, 0x24, 0x01, 0x41, 0x00, 0x03,
- 0x00, 0x01, 0x04, 0x09, 0x00, 0x11, 0x00, 0x02, 0x01, 0x65, 0x59, 0x75, 0x7a, 0x75, 0x4f, 0x53,
- 0x53, 0x45, 0x78, 0x74, 0x65, 0x6e, 0x73, 0x69, 0x6f, 0x6e, 0x52, 0x65, 0x67, 0x75, 0x6c, 0x61,
- 0x72, 0x56, 0x65, 0x72, 0x73, 0x69, 0x6f, 0x6e, 0x20, 0x31, 0x2e, 0x30, 0x30, 0x30, 0x3b, 0x3b,
- 0x59, 0x75, 0x7a, 0x75, 0x4f, 0x53, 0x53, 0x45, 0x78, 0x74, 0x65, 0x6e, 0x73, 0x69, 0x6f, 0x6e,
- 0x2d, 0x52, 0x3b, 0x32, 0x30, 0x31, 0x39, 0x3b, 0x46, 0x4c, 0x56, 0x49, 0x2d, 0x36, 0x31, 0x34,
- 0x59, 0x75, 0x7a, 0x75, 0x4f, 0x53, 0x53, 0x45, 0x78, 0x74, 0x65, 0x6e, 0x73, 0x69, 0x6f, 0x6e,
- 0x20, 0x52, 0x56, 0x65, 0x72, 0x73, 0x69, 0x6f, 0x6e, 0x20, 0x31, 0x2e, 0x30, 0x30, 0x30, 0x59,
- 0x75, 0x7a, 0x75, 0x4f, 0x53, 0x53, 0x45, 0x78, 0x74, 0x65, 0x6e, 0x73, 0x69, 0x6f, 0x6e, 0x2d,
- 0x52, 0x00, 0x59, 0x00, 0x75, 0x00, 0x7a, 0x00, 0x75, 0x00, 0x4f, 0x00, 0x53, 0x00, 0x53, 0x00,
- 0x45, 0x00, 0x78, 0x00, 0x74, 0x00, 0x65, 0x00, 0x6e, 0x00, 0x73, 0x00, 0x69, 0x00, 0x6f, 0x00,
- 0x6e, 0x00, 0x52, 0x00, 0x65, 0x00, 0x67, 0x00, 0x75, 0x00, 0x6c, 0x00, 0x61, 0x00, 0x72, 0x00,
- 0x56, 0x00, 0x65, 0x00, 0x72, 0x00, 0x73, 0x00, 0x69, 0x00, 0x6f, 0x00, 0x6e, 0x00, 0x20, 0x00,
- 0x31, 0x00, 0x2e, 0x00, 0x30, 0x00, 0x30, 0x00, 0x30, 0x00, 0x3b, 0x00, 0x3b, 0x00, 0x59, 0x00,
- 0x75, 0x00, 0x7a, 0x00, 0x75, 0x00, 0x4f, 0x00, 0x53, 0x00, 0x53, 0x00, 0x45, 0x00, 0x78, 0x00,
- 0x74, 0x00, 0x65, 0x00, 0x6e, 0x00, 0x73, 0x00, 0x69, 0x00, 0x6f, 0x00, 0x6e, 0x00, 0x2d, 0x00,
- 0x52, 0x00, 0x3b, 0x00, 0x32, 0x00, 0x30, 0x00, 0x31, 0x00, 0x39, 0x00, 0x3b, 0x00, 0x46, 0x00,
- 0x4c, 0x00, 0x56, 0x00, 0x49, 0x00, 0x2d, 0x00, 0x36, 0x00, 0x31, 0x00, 0x34, 0x00, 0x59, 0x00,
- 0x75, 0x00, 0x7a, 0x00, 0x75, 0x00, 0x4f, 0x00, 0x53, 0x00, 0x53, 0x00, 0x45, 0x00, 0x78, 0x00,
- 0x74, 0x00, 0x65, 0x00, 0x6e, 0x00, 0x73, 0x00, 0x69, 0x00, 0x6f, 0x00, 0x6e, 0x00, 0x20, 0x00,
- 0x52, 0x00, 0x56, 0x00, 0x65, 0x00, 0x72, 0x00, 0x73, 0x00, 0x69, 0x00, 0x6f, 0x00, 0x6e, 0x00,
- 0x20, 0x00, 0x31, 0x00, 0x2e, 0x00, 0x30, 0x00, 0x30, 0x00, 0x30, 0x00, 0x59, 0x00, 0x75, 0x00,
- 0x7a, 0x00, 0x75, 0x00, 0x4f, 0x00, 0x53, 0x00, 0x53, 0x00, 0x45, 0x00, 0x78, 0x00, 0x74, 0x00,
- 0x65, 0x00, 0x6e, 0x00, 0x73, 0x00, 0x69, 0x00, 0x6f, 0x00, 0x6e, 0x00, 0x2d, 0x00, 0x52, 0x00,
- 0x52, 0x00, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xff, 0x9c, 0x00, 0x32,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x00, 0x0e, 0x00, 0x00, 0x01, 0x02, 0x01, 0x03, 0x00, 0x03, 0x01, 0x04,
- 0x01, 0x05, 0x01, 0x06, 0x01, 0x07, 0x01, 0x08, 0x01, 0x09, 0x01, 0x0a, 0x01, 0x0b, 0x01, 0x0c,
- 0x01, 0x0d, 0x07, 0x75, 0x6e, 0x69, 0x30, 0x30, 0x30, 0x30, 0x07, 0x75, 0x6e, 0x69, 0x30, 0x30,
- 0x30, 0x44, 0x07, 0x75, 0x6e, 0x69, 0x45, 0x30, 0x45, 0x30, 0x07, 0x75, 0x6e, 0x69, 0x45, 0x30,
- 0x45, 0x31, 0x07, 0x75, 0x6e, 0x69, 0x45, 0x30, 0x45, 0x32, 0x07, 0x75, 0x6e, 0x69, 0x45, 0x30,
- 0x45, 0x33, 0x07, 0x75, 0x6e, 0x69, 0x45, 0x30, 0x45, 0x34, 0x07, 0x75, 0x6e, 0x69, 0x45, 0x30,
- 0x45, 0x35, 0x07, 0x75, 0x6e, 0x69, 0x45, 0x30, 0x45, 0x36, 0x07, 0x75, 0x6e, 0x69, 0x45, 0x30,
- 0x45, 0x37, 0x07, 0x75, 0x6e, 0x69, 0x45, 0x30, 0x45, 0x38, 0x07, 0x75, 0x6e, 0x69, 0x45, 0x30,
- 0x45, 0x39, 0x00, 0x00, 0x00, 0x01, 0x00, 0x01, 0xff, 0xff, 0x00, 0x0f, 0x00, 0x00, 0x00, 0x01,
- 0x00, 0x00, 0x00, 0x00,
+ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+ 0xB8, 0x00, 0x00, 0x2C, 0x4B, 0xB8, 0x00, 0x09, 0x50, 0x58, 0xB1, 0x01, 0x01, 0x8E, 0x59, 0xB8,
+ 0x01, 0xFF, 0x85, 0xB8, 0x00, 0x44, 0x1D, 0xB9, 0x00, 0x09, 0x00, 0x03, 0x5F, 0x5E, 0x2D, 0xB8,
+ 0x00, 0x01, 0x2C, 0x20, 0x20, 0x45, 0x69, 0x44, 0xB0, 0x01, 0x60, 0x2D, 0xB8, 0x00, 0x02, 0x2C,
+ 0xB8, 0x00, 0x01, 0x2A, 0x21, 0x2D, 0xB8, 0x00, 0x03, 0x2C, 0x20, 0x46, 0xB0, 0x03, 0x25, 0x46,
+ 0x52, 0x58, 0x23, 0x59, 0x20, 0x8A, 0x20, 0x8A, 0x49, 0x64, 0x8A, 0x20, 0x46, 0x20, 0x68, 0x61,
+ 0x64, 0xB0, 0x04, 0x25, 0x46, 0x20, 0x68, 0x61, 0x64, 0x52, 0x58, 0x23, 0x65, 0x8A, 0x59, 0x2F,
+ 0x20, 0xB0, 0x00, 0x53, 0x58, 0x69, 0x20, 0xB0, 0x00, 0x54, 0x58, 0x21, 0xB0, 0x40, 0x59, 0x1B,
+ 0x69, 0x20, 0xB0, 0x00, 0x54, 0x58, 0x21, 0xB0, 0x40, 0x65, 0x59, 0x59, 0x3A, 0x2D, 0xB8, 0x00,
+ 0x04, 0x2C, 0x20, 0x46, 0xB0, 0x04, 0x25, 0x46, 0x52, 0x58, 0x23, 0x8A, 0x59, 0x20, 0x46, 0x20,
+ 0x6A, 0x61, 0x64, 0xB0, 0x04, 0x25, 0x46, 0x20, 0x6A, 0x61, 0x64, 0x52, 0x58, 0x23, 0x8A, 0x59,
+ 0x2F, 0xFD, 0x2D, 0xB8, 0x00, 0x05, 0x2C, 0x4B, 0x20, 0xB0, 0x03, 0x26, 0x50, 0x58, 0x51, 0x58,
+ 0xB0, 0x80, 0x44, 0x1B, 0xB0, 0x40, 0x44, 0x59, 0x1B, 0x21, 0x21, 0x20, 0x45, 0xB0, 0xC0, 0x50,
+ 0x58, 0xB0, 0xC0, 0x44, 0x1B, 0x21, 0x59, 0x59, 0x2D, 0xB8, 0x00, 0x06, 0x2C, 0x20, 0x20, 0x45,
+ 0x69, 0x44, 0xB0, 0x01, 0x60, 0x20, 0x20, 0x45, 0x7D, 0x69, 0x18, 0x44, 0xB0, 0x01, 0x60, 0x2D,
+ 0xB8, 0x00, 0x07, 0x2C, 0xB8, 0x00, 0x06, 0x2A, 0x2D, 0xB8, 0x00, 0x08, 0x2C, 0x4B, 0x20, 0xB0,
+ 0x03, 0x26, 0x53, 0x58, 0xB0, 0x40, 0x1B, 0xB0, 0x00, 0x59, 0x8A, 0x8A, 0x20, 0xB0, 0x03, 0x26,
+ 0x53, 0x58, 0x23, 0x21, 0xB0, 0x80, 0x8A, 0x8A, 0x1B, 0x8A, 0x23, 0x59, 0x20, 0xB0, 0x03, 0x26,
+ 0x53, 0x58, 0x23, 0x21, 0xB8, 0x00, 0xC0, 0x8A, 0x8A, 0x1B, 0x8A, 0x23, 0x59, 0x20, 0xB0, 0x03,
+ 0x26, 0x53, 0x58, 0x23, 0x21, 0xB8, 0x01, 0x00, 0x8A, 0x8A, 0x1B, 0x8A, 0x23, 0x59, 0x20, 0xB0,
+ 0x03, 0x26, 0x53, 0x58, 0x23, 0x21, 0xB8, 0x01, 0x40, 0x8A, 0x8A, 0x1B, 0x8A, 0x23, 0x59, 0x20,
+ 0xB8, 0x00, 0x03, 0x26, 0x53, 0x58, 0xB0, 0x03, 0x25, 0x45, 0xB8, 0x01, 0x80, 0x50, 0x58, 0x23,
+ 0x21, 0xB8, 0x01, 0x80, 0x23, 0x21, 0x1B, 0xB0, 0x03, 0x25, 0x45, 0x23, 0x21, 0x23, 0x21, 0x59,
+ 0x1B, 0x21, 0x59, 0x44, 0x2D, 0xB8, 0x00, 0x09, 0x2C, 0x4B, 0x53, 0x58, 0x45, 0x44, 0x1B, 0x21,
+ 0x21, 0x59, 0x2D, 0x00, 0xB8, 0x00, 0x00, 0x2B, 0x00, 0xBA, 0x00, 0x01, 0x00, 0x01, 0x00, 0x07,
+ 0x2B, 0xB8, 0x00, 0x00, 0x20, 0x45, 0x7D, 0x69, 0x18, 0x44, 0x00, 0x00, 0x00, 0x14, 0x00, 0x00,
+ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x16, 0x00, 0x16, 0x00, 0x16, 0x00, 0x16, 0x00, 0x70,
+ 0x00, 0xDC, 0x01, 0x34, 0x01, 0x7C, 0x01, 0xA2, 0x01, 0xF4, 0x02, 0x3C, 0x02, 0xA8, 0x03, 0x4C,
+ 0x03, 0xE2, 0x04, 0x20, 0x04, 0x58, 0x04, 0x9A, 0x04, 0xEE, 0x05, 0x32, 0x05, 0x64, 0x05, 0x80,
+ 0x05, 0xC6, 0x05, 0xF6, 0x06, 0x54, 0x06, 0xB2, 0x07, 0x38, 0x07, 0x60, 0x07, 0x82, 0x00, 0x00,
+ 0x00, 0x02, 0x00, 0xA4, 0xFF, 0xFF, 0x03, 0x5C, 0x03, 0x09, 0x00, 0x03, 0x00, 0x07, 0x00, 0x00,
+ 0x13, 0x11, 0x21, 0x11, 0x25, 0x21, 0x11, 0x21, 0xCD, 0x02, 0x66, 0xFD, 0x71, 0x02, 0xB8, 0xFD,
+ 0x48, 0x02, 0xE0, 0xFD, 0x48, 0x02, 0xB8, 0x29, 0xFC, 0xF6, 0x00, 0x00, 0x00, 0x04, 0x00, 0x14,
+ 0xFF, 0x98, 0x03, 0xEC, 0x03, 0x70, 0x00, 0x0F, 0x00, 0x1F, 0x00, 0x2F, 0x00, 0x39, 0x00, 0x00,
+ 0x00, 0x22, 0x0E, 0x02, 0x14, 0x1E, 0x02, 0x32, 0x3E, 0x02, 0x34, 0x2E, 0x01, 0x24, 0x32, 0x1E,
+ 0x02, 0x14, 0x0E, 0x02, 0x22, 0x2E, 0x02, 0x34, 0x3E, 0x01, 0x13, 0x12, 0x37, 0x33, 0x13, 0x12,
+ 0x15, 0x16, 0x23, 0x2F, 0x01, 0x23, 0x07, 0x23, 0x22, 0x26, 0x25, 0x30, 0x27, 0x26, 0x2F, 0x01,
+ 0x06, 0x07, 0x06, 0x32, 0x02, 0x5A, 0xB4, 0xA4, 0x77, 0x46, 0x46, 0x77, 0xA4, 0xB4, 0xA4, 0x77,
+ 0x46, 0x46, 0x77, 0xFE, 0x9E, 0xC8, 0xB7, 0x83, 0x4E, 0x4E, 0x83, 0xB7, 0xC8, 0xB7, 0x83, 0x4E,
+ 0x4E, 0x83, 0x23, 0x6C, 0x5E, 0x6D, 0x68, 0x68, 0x01, 0x39, 0x38, 0x2E, 0xD1, 0x2B, 0x37, 0x33,
+ 0x04, 0x01, 0x48, 0x1D, 0x1C, 0x0A, 0x05, 0x01, 0x45, 0x01, 0x89, 0x03, 0x3F, 0x46, 0x77, 0xA4,
+ 0xB4, 0xA4, 0x77, 0x46, 0x46, 0x77, 0xA4, 0xB4, 0xA4, 0x77, 0x77, 0x4E, 0x83, 0xB7, 0xC8, 0xB7,
+ 0x83, 0x4E, 0x4E, 0x83, 0xB7, 0xC8, 0xB7, 0x83, 0xFD, 0x64, 0x01, 0x1A, 0xEB, 0xFE, 0xFE, 0xFE,
+ 0xFD, 0x03, 0x01, 0x01, 0x77, 0x78, 0x01, 0xCF, 0x4C, 0x4C, 0x1C, 0x0C, 0x02, 0xBE, 0x02, 0x00,
+ 0x00, 0x05, 0x00, 0x14, 0xFF, 0x98, 0x03, 0xEC, 0x03, 0x70, 0x00, 0x0F, 0x00, 0x1B, 0x00, 0x2F,
+ 0x00, 0x3A, 0x00, 0x44, 0x00, 0x00, 0x12, 0x14, 0x1E, 0x02, 0x32, 0x3E, 0x02, 0x34, 0x2E, 0x02,
+ 0x22, 0x0E, 0x01, 0x02, 0x10, 0x3E, 0x01, 0x20, 0x1E, 0x01, 0x10, 0x0E, 0x01, 0x20, 0x26, 0x01,
+ 0x16, 0x17, 0x14, 0x06, 0x07, 0x06, 0x2B, 0x01, 0x19, 0x01, 0x17, 0x32, 0x17, 0x16, 0x17, 0x16,
+ 0x07, 0x06, 0x0F, 0x01, 0x36, 0x37, 0x34, 0x2E, 0x01, 0x27, 0x23, 0x15, 0x33, 0x32, 0x27, 0x32,
+ 0x37, 0x36, 0x26, 0x27, 0x26, 0x2B, 0x01, 0x15, 0x45, 0x46, 0x77, 0xA4, 0xB4, 0xA4, 0x77, 0x46,
+ 0x46, 0x77, 0xA4, 0xB4, 0xA4, 0x77, 0x77, 0x84, 0xE2, 0x01, 0x0C, 0xE2, 0x84, 0x84, 0xE2, 0xFE,
+ 0xF4, 0xE2, 0x01, 0xF7, 0x61, 0x01, 0x4E, 0x3E, 0x29, 0xAF, 0x4E, 0x81, 0x8B, 0x1D, 0x3C, 0x1F,
+ 0x19, 0x04, 0x06, 0x39, 0x57, 0x44, 0x01, 0x1B, 0x2D, 0x51, 0x46, 0x46, 0x47, 0x66, 0x70, 0x16,
+ 0x1F, 0x01, 0x2C, 0x08, 0x4B, 0x4C, 0x01, 0xDE, 0xB4, 0xA4, 0x77, 0x46, 0x46, 0x77, 0xA4, 0xB4,
+ 0xA4, 0x77, 0x46, 0x46, 0x77, 0xFE, 0x7C, 0x01, 0x0C, 0xE2, 0x84, 0x84, 0xE2, 0xFE, 0xF4, 0xE2,
+ 0x84, 0x84, 0x01, 0x6D, 0x21, 0x5B, 0x40, 0x50, 0x05, 0x03, 0x01, 0x03, 0x01, 0x05, 0x01, 0x05,
+ 0x09, 0x30, 0x25, 0x29, 0x40, 0x21, 0xC2, 0x06, 0x3E, 0x1A, 0x21, 0x0B, 0x01, 0x8C, 0xE1, 0x0A,
+ 0x0E, 0x54, 0x0B, 0x02, 0x79, 0x00, 0x00, 0x00, 0x00, 0x03, 0x00, 0x14, 0xFF, 0x98, 0x03, 0xEC,
+ 0x03, 0x70, 0x00, 0x0F, 0x00, 0x1B, 0x00, 0x38, 0x00, 0x00, 0x12, 0x14, 0x1E, 0x02, 0x32, 0x3E,
+ 0x02, 0x34, 0x2E, 0x02, 0x22, 0x0E, 0x01, 0x02, 0x10, 0x3E, 0x01, 0x20, 0x1E, 0x01, 0x10, 0x0E,
+ 0x01, 0x20, 0x26, 0x36, 0x34, 0x3F, 0x01, 0x27, 0x26, 0x27, 0x33, 0x17, 0x16, 0x33, 0x36, 0x3F,
+ 0x02, 0x32, 0x14, 0x06, 0x16, 0x12, 0x14, 0x2B, 0x01, 0x27, 0x26, 0x06, 0x0F, 0x01, 0x23, 0x45,
+ 0x46, 0x77, 0xA4, 0xB4, 0xA4, 0x77, 0x46, 0x46, 0x77, 0xA4, 0xB4, 0xA4, 0x77, 0x77, 0x84, 0xE2,
+ 0x01, 0x0C, 0xE2, 0x84, 0x84, 0xE2, 0xFE, 0xF4, 0xE2, 0x7B, 0x58, 0x58, 0x4D, 0x4F, 0x05, 0x7A,
+ 0x34, 0x34, 0x02, 0x01, 0x33, 0x32, 0x3C, 0x3C, 0xA1, 0x01, 0xB0, 0x3E, 0x3F, 0x39, 0x3B, 0x02,
+ 0x3A, 0x38, 0x3F, 0x01, 0xDE, 0xB4, 0xA4, 0x77, 0x46, 0x46, 0x77, 0xA4, 0xB4, 0xA4, 0x77, 0x46,
+ 0x46, 0x77, 0xFE, 0x7C, 0x01, 0x0C, 0xE2, 0x84, 0x84, 0xE2, 0xFE, 0xF4, 0xE2, 0x84, 0x84, 0x60,
+ 0x02, 0x87, 0x88, 0x79, 0x7A, 0x06, 0x54, 0x54, 0x01, 0x53, 0x53, 0x01, 0x01, 0xFB, 0x04, 0xFE,
+ 0xF8, 0x02, 0x5B, 0x5A, 0x03, 0x59, 0x59, 0x00, 0x00, 0x03, 0x00, 0x14, 0xFF, 0x98, 0x03, 0xEC,
+ 0x03, 0x70, 0x00, 0x0F, 0x00, 0x1B, 0x00, 0x2B, 0x00, 0x00, 0x00, 0x22, 0x0E, 0x02, 0x14, 0x1E,
+ 0x02, 0x32, 0x3E, 0x02, 0x34, 0x2E, 0x01, 0x24, 0x20, 0x1E, 0x01, 0x10, 0x0E, 0x01, 0x20, 0x2E,
+ 0x01, 0x10, 0x36, 0x01, 0x35, 0x27, 0x26, 0x34, 0x3B, 0x01, 0x17, 0x16, 0x36, 0x3F, 0x01, 0x33,
+ 0x03, 0x15, 0x23, 0x02, 0x5A, 0xB4, 0xA4, 0x77, 0x46, 0x46, 0x77, 0xA4, 0xB4, 0xA4, 0x77, 0x46,
+ 0x46, 0x77, 0xFE, 0x7C, 0x01, 0x0C, 0xE2, 0x84, 0x84, 0xE2, 0xFE, 0xF4, 0xE2, 0x84, 0x84, 0x01,
+ 0x36, 0x5E, 0x5F, 0x3C, 0x3D, 0x3D, 0x3D, 0x03, 0x3B, 0x3B, 0x77, 0xBE, 0x68, 0x03, 0x3F, 0x46,
+ 0x77, 0xA4, 0xB4, 0xA4, 0x77, 0x46, 0x46, 0x77, 0xA4, 0xB4, 0xA4, 0x77, 0x77, 0x84, 0xE2, 0xFE,
+ 0xF4, 0xE2, 0x84, 0x84, 0xE2, 0x01, 0x0C, 0xE2, 0xFD, 0xF9, 0x6E, 0x96, 0x95, 0x01, 0x67, 0x67,
+ 0x03, 0x66, 0x65, 0xFE, 0xD3, 0xDA, 0x00, 0x00, 0x00, 0x03, 0x00, 0x14, 0xFF, 0xBD, 0x03, 0xEC,
+ 0x03, 0x4B, 0x00, 0x06, 0x00, 0x0C, 0x00, 0x12, 0x00, 0x00, 0x01, 0x21, 0x22, 0x15, 0x30, 0x11,
+ 0x21, 0x17, 0x21, 0x11, 0x10, 0x25, 0x21, 0x01, 0x11, 0x33, 0x11, 0x21, 0x15, 0x03, 0xBB, 0xFD,
+ 0x77, 0xED, 0x03, 0x76, 0x31, 0xFC, 0x28, 0x01, 0x1E, 0x02, 0xBA, 0xFD, 0x5C, 0x68, 0x01, 0x08,
+ 0x03, 0x1A, 0xEE, 0xFD, 0xC2, 0x31, 0x02, 0x6F, 0x01, 0x1E, 0x01, 0xFD, 0x36, 0x02, 0x07, 0xFE,
+ 0x50, 0x57, 0x00, 0x00, 0x00, 0x04, 0x00, 0x14, 0xFF, 0xBD, 0x03, 0xEC, 0x03, 0x4B, 0x00, 0x06,
+ 0x00, 0x0C, 0x00, 0x27, 0x00, 0x32, 0x00, 0x00, 0x05, 0x11, 0x34, 0x27, 0x30, 0x21, 0x11, 0x07,
+ 0x11, 0x21, 0x20, 0x19, 0x01, 0x25, 0x11, 0x33, 0x32, 0x17, 0x16, 0x17, 0x16, 0x17, 0x16, 0x07,
+ 0x06, 0x07, 0x06, 0x07, 0x1E, 0x02, 0x15, 0x07, 0x23, 0x27, 0x2E, 0x01, 0x2F, 0x01, 0x15, 0x13,
+ 0x36, 0x35, 0x34, 0x27, 0x26, 0x27, 0x23, 0x15, 0x33, 0x36, 0x03, 0xBB, 0xED, 0xFD, 0x77, 0x31,
+ 0x02, 0xBA, 0x01, 0x1E, 0xFD, 0x2A, 0x77, 0x76, 0x15, 0x49, 0x20, 0x35, 0x08, 0x04, 0x06, 0x13,
+ 0x66, 0x0C, 0x01, 0x1F, 0x2E, 0x65, 0x3D, 0x3D, 0x2A, 0x56, 0x28, 0x2E, 0x19, 0x99, 0x3C, 0x20,
+ 0x10, 0x56, 0x4F, 0x46, 0x47, 0x12, 0x02, 0x3E, 0xED, 0x01, 0xFC, 0xD4, 0x31, 0x03, 0x8E, 0xFE,
+ 0xE1, 0xFD, 0x91, 0xC4, 0x02, 0x07, 0x01, 0x04, 0x13, 0x21, 0x44, 0x1D, 0x19, 0x58, 0x15, 0x02,
+ 0x01, 0x13, 0x2D, 0xA2, 0x01, 0x01, 0x3D, 0x81, 0x1A, 0x01, 0x01, 0xDA, 0x01, 0x2D, 0x08, 0x3A,
+ 0x29, 0x0F, 0x08, 0x01, 0x85, 0x01, 0x00, 0x00, 0x00, 0x04, 0x00, 0x14, 0xFF, 0xF5, 0x03, 0xEC,
+ 0x03, 0x13, 0x00, 0x09, 0x00, 0x11, 0x00, 0x26, 0x00, 0x32, 0x00, 0x00, 0x37, 0x21, 0x34, 0x10,
+ 0x35, 0x34, 0x27, 0x21, 0x04, 0x11, 0x23, 0x10, 0x25, 0x21, 0x16, 0x15, 0x11, 0x21, 0x37, 0x35,
+ 0x37, 0x36, 0x22, 0x2B, 0x01, 0x3D, 0x01, 0x3B, 0x01, 0x1D, 0x01, 0x0F, 0x01, 0x3B, 0x01, 0x1D,
+ 0x01, 0x2B, 0x01, 0x25, 0x35, 0x3B, 0x01, 0x1D, 0x01, 0x3B, 0x01, 0x1D, 0x01, 0x2B, 0x01, 0x45,
+ 0x03, 0x76, 0x45, 0xFE, 0x2D, 0xFE, 0xA2, 0x31, 0x01, 0x8F, 0x01, 0xD3, 0x76, 0xFC, 0x28, 0xA7,
+ 0x68, 0x68, 0x01, 0x5B, 0x5D, 0x90, 0x91, 0x6C, 0x6D, 0x71, 0x70, 0xA0, 0xA0, 0x01, 0x75, 0x27,
+ 0x28, 0x63, 0x63, 0x8B, 0x8A, 0x27, 0x69, 0x01, 0xA4, 0x69, 0x44, 0x01, 0x02, 0xFE, 0xA4, 0x01,
+ 0x8C, 0x03, 0x01, 0x75, 0xFD, 0x58, 0xBB, 0x24, 0x80, 0x80, 0x21, 0x21, 0x1F, 0x1E, 0x85, 0x86,
+ 0x20, 0x22, 0xC3, 0xC3, 0xA1, 0xA3, 0x20, 0x22, 0x00, 0x05, 0x00, 0x14, 0xFF, 0xF5, 0x03, 0xEC,
+ 0x03, 0x13, 0x00, 0x08, 0x00, 0x10, 0x00, 0x2B, 0x00, 0x37, 0x00, 0x44, 0x00, 0x00, 0x37, 0x21,
+ 0x11, 0x10, 0x25, 0x30, 0x21, 0x06, 0x15, 0x03, 0x11, 0x34, 0x37, 0x21, 0x04, 0x19, 0x01, 0x01,
+ 0x35, 0x17, 0x32, 0x17, 0x16, 0x17, 0x16, 0x07, 0x06, 0x07, 0x06, 0x17, 0x16, 0x17, 0x16, 0x17,
+ 0x16, 0x23, 0x2F, 0x01, 0x2E, 0x01, 0x2F, 0x01, 0x15, 0x23, 0x37, 0x32, 0x36, 0x37, 0x36, 0x35,
+ 0x26, 0x27, 0x26, 0x2B, 0x01, 0x15, 0x05, 0x35, 0x37, 0x36, 0x26, 0x2B, 0x01, 0x35, 0x21, 0x15,
+ 0x03, 0x17, 0x15, 0x45, 0x03, 0x76, 0xFE, 0xA2, 0xFE, 0x2D, 0x45, 0x31, 0x76, 0x01, 0xD3, 0x01,
+ 0x8F, 0xFE, 0x1E, 0x65, 0x6F, 0x15, 0x46, 0x10, 0x05, 0x04, 0x0D, 0x4F, 0x09, 0x09, 0x1F, 0x1D,
+ 0x3A, 0x06, 0x01, 0x30, 0x2F, 0x22, 0x37, 0x1E, 0x29, 0x14, 0x4E, 0x82, 0x34, 0x19, 0x0E, 0x13,
+ 0x0A, 0x22, 0x07, 0x38, 0x37, 0xFE, 0x3E, 0x68, 0x68, 0x01, 0x5C, 0x5C, 0x01, 0x20, 0xD8, 0xE1,
+ 0x27, 0x01, 0x5D, 0x01, 0x5B, 0x03, 0x01, 0x44, 0xFD, 0x58, 0x02, 0xA8, 0x75, 0x01, 0x03, 0xFE,
+ 0x74, 0xFE, 0x71, 0x01, 0x5C, 0xC5, 0x01, 0x04, 0x0C, 0x43, 0x15, 0x1D, 0x44, 0x10, 0x04, 0x06,
+ 0x14, 0x2B, 0x56, 0x10, 0x01, 0x01, 0x34, 0x52, 0x1C, 0x01, 0x01, 0xA5, 0xE3, 0x04, 0x06, 0x0A,
+ 0x20, 0x2C, 0x04, 0x01, 0x65, 0xE3, 0x47, 0x80, 0x80, 0x01, 0x42, 0x3D, 0xFE, 0xF5, 0x01, 0x41,
+ 0x00, 0x04, 0x00, 0x14, 0x00, 0x52, 0x03, 0xEC, 0x02, 0xB6, 0x00, 0x08, 0x00, 0x16, 0x00, 0x64,
+ 0x00, 0x70, 0x00, 0x00, 0x25, 0x11, 0x21, 0x22, 0x15, 0x30, 0x15, 0x14, 0x33, 0x11, 0x21, 0x32,
+ 0x15, 0x11, 0x14, 0x27, 0x21, 0x22, 0x26, 0x3D, 0x01, 0x34, 0x36, 0x13, 0x26, 0x27, 0x26, 0x27,
+ 0x26, 0x37, 0x33, 0x36, 0x37, 0x36, 0x33, 0x16, 0x17, 0x16, 0x17, 0x16, 0x37, 0x36, 0x37, 0x36,
+ 0x35, 0x34, 0x27, 0x26, 0x27, 0x26, 0x27, 0x26, 0x27, 0x26, 0x27, 0x26, 0x34, 0x37, 0x36, 0x37,
+ 0x36, 0x37, 0x36, 0x17, 0x16, 0x17, 0x16, 0x17, 0x16, 0x17, 0x16, 0x0F, 0x01, 0x22, 0x06, 0x23,
+ 0x27, 0x26, 0x27, 0x26, 0x23, 0x22, 0x07, 0x06, 0x07, 0x06, 0x17, 0x16, 0x17, 0x16, 0x17, 0x16,
+ 0x17, 0x16, 0x17, 0x16, 0x07, 0x06, 0x07, 0x06, 0x27, 0x37, 0x35, 0x3B, 0x01, 0x1D, 0x01, 0x3B,
+ 0x01, 0x1D, 0x01, 0x2B, 0x01, 0x03, 0xBB, 0xFD, 0x2A, 0xA0, 0xA0, 0x02, 0xEE, 0x19, 0x19, 0xFD,
+ 0x12, 0x57, 0x7A, 0x7A, 0xCA, 0x38, 0x1D, 0x16, 0x08, 0x03, 0x01, 0x02, 0x0F, 0x0C, 0x1E, 0x01,
+ 0x02, 0x04, 0x0C, 0x2B, 0x0F, 0x0E, 0x18, 0x0C, 0x09, 0x04, 0x15, 0x32, 0x23, 0x12, 0x1C, 0x0E,
+ 0x09, 0x03, 0x01, 0x01, 0x09, 0x21, 0x0F, 0x14, 0x2E, 0x2A, 0x13, 0x0F, 0x0C, 0x08, 0x0B, 0x05,
+ 0x02, 0x01, 0x02, 0x03, 0x36, 0x03, 0x02, 0x03, 0x08, 0x0D, 0x23, 0x16, 0x0E, 0x10, 0x01, 0x01,
+ 0x07, 0x0B, 0x32, 0x25, 0x13, 0x26, 0x0F, 0x09, 0x01, 0x01, 0x0F, 0x11, 0x24, 0x21, 0x2A, 0xE3,
+ 0x20, 0x20, 0x52, 0x50, 0x71, 0x71, 0x84, 0x02, 0x00, 0xAF, 0xA2, 0xAF, 0x02, 0x32, 0x19, 0xFD,
+ 0xCE, 0x19, 0x01, 0x84, 0x5C, 0xA2, 0x5C, 0x85, 0xFE, 0x29, 0x04, 0x1E, 0x18, 0x26, 0x0F, 0x01,
+ 0x02, 0x01, 0x03, 0x05, 0x0B, 0x29, 0x06, 0x02, 0x03, 0x04, 0x11, 0x0B, 0x0D, 0x0A, 0x06, 0x12,
+ 0x0D, 0x0A, 0x07, 0x0C, 0x18, 0x0D, 0x10, 0x06, 0x18, 0x05, 0x27, 0x14, 0x09, 0x03, 0x0A, 0x0D,
+ 0x06, 0x09, 0x09, 0x0D, 0x0F, 0x14, 0x0C, 0x06, 0x03, 0x02, 0x04, 0x10, 0x0A, 0x11, 0x08, 0x09,
+ 0x0E, 0x0C, 0x07, 0x0C, 0x0C, 0x0A, 0x07, 0x0F, 0x20, 0x11, 0x18, 0x1E, 0x1A, 0x1E, 0x0C, 0x0B,
+ 0x03, 0xAA, 0xA5, 0x89, 0x8A, 0x1C, 0x1B, 0x00, 0x00, 0x05, 0x00, 0x14, 0x00, 0x53, 0x03, 0xEC,
+ 0x02, 0xB6, 0x00, 0x08, 0x00, 0x16, 0x00, 0x2E, 0x00, 0x38, 0x00, 0x65, 0x00, 0x00, 0x01, 0x30,
+ 0x21, 0x11, 0x21, 0x32, 0x3D, 0x01, 0x34, 0x27, 0x32, 0x16, 0x1D, 0x01, 0x14, 0x06, 0x23, 0x21,
+ 0x26, 0x35, 0x11, 0x34, 0x33, 0x01, 0x11, 0x33, 0x32, 0x17, 0x16, 0x17, 0x16, 0x07, 0x06, 0x07,
+ 0x17, 0x1E, 0x01, 0x1F, 0x01, 0x23, 0x2A, 0x01, 0x2E, 0x01, 0x23, 0x27, 0x15, 0x37, 0x32, 0x37,
+ 0x36, 0x27, 0x2E, 0x01, 0x2B, 0x01, 0x15, 0x05, 0x26, 0x27, 0x37, 0x32, 0x3F, 0x01, 0x16, 0x17,
+ 0x1E, 0x01, 0x37, 0x36, 0x27, 0x2E, 0x04, 0x37, 0x3E, 0x01, 0x33, 0x32, 0x17, 0x16, 0x17, 0x14,
+ 0x06, 0x27, 0x26, 0x27, 0x26, 0x0E, 0x01, 0x1E, 0x02, 0x17, 0x16, 0x06, 0x07, 0x06, 0x07, 0x06,
+ 0x03, 0x1B, 0xFD, 0x2A, 0x02, 0xD6, 0xA0, 0xA0, 0x57, 0x7A, 0x7A, 0x57, 0xFD, 0x12, 0x19, 0x19,
+ 0x01, 0xD3, 0x47, 0x44, 0x11, 0x3E, 0x18, 0x21, 0x0B, 0x0C, 0x43, 0x04, 0x17, 0x1C, 0x1E, 0x16,
+ 0x26, 0x26, 0x03, 0x4D, 0x18, 0x1E, 0x11, 0x25, 0x3A, 0x0C, 0x22, 0x08, 0x03, 0x1B, 0x3E, 0x29,
+ 0xFE, 0xAC, 0x0D, 0x04, 0x02, 0x02, 0x1E, 0x1D, 0x03, 0x02, 0x0C, 0x4C, 0x13, 0x20, 0x07, 0x04,
+ 0x1B, 0x56, 0x2D, 0x1C, 0x01, 0x02, 0x44, 0x35, 0x49, 0x1F, 0x10, 0x03, 0x41, 0x01, 0x06, 0x0A,
+ 0x16, 0x3C, 0x18, 0x0C, 0x16, 0x5D, 0x15, 0x33, 0x03, 0x2B, 0x1E, 0x34, 0x59, 0x02, 0x84, 0xFE,
+ 0x00, 0xAF, 0xA2, 0xAF, 0x32, 0x85, 0x5C, 0xA2, 0x5C, 0x84, 0x01, 0x17, 0x02, 0x32, 0x19, 0xFE,
+ 0x2F, 0x01, 0x45, 0x01, 0x02, 0x19, 0x22, 0x32, 0x39, 0x0B, 0x08, 0x0F, 0x27, 0x2F, 0x24, 0x75,
+ 0x12, 0x01, 0x88, 0xBB, 0x04, 0x09, 0x2A, 0x0F, 0x0D, 0x53, 0x8A, 0x17, 0x1E, 0x04, 0x03, 0x03,
+ 0x0C, 0x04, 0x26, 0x0E, 0x0C, 0x14, 0x1A, 0x0E, 0x0E, 0x16, 0x16, 0x2C, 0x1A, 0x2D, 0x2D, 0x2A,
+ 0x16, 0x1D, 0x06, 0x04, 0x01, 0x1A, 0x09, 0x11, 0x09, 0x17, 0x18, 0x0D, 0x17, 0x0C, 0x1B, 0x71,
+ 0x1B, 0x12, 0x01, 0x03, 0x00, 0x03, 0x00, 0x14, 0xFF, 0x98, 0x03, 0xEC, 0x03, 0x70, 0x00, 0x0F,
+ 0x00, 0x1B, 0x00, 0x27, 0x00, 0x00, 0x00, 0x22, 0x0E, 0x02, 0x14, 0x1E, 0x02, 0x32, 0x3E, 0x02,
+ 0x34, 0x2E, 0x01, 0x24, 0x20, 0x1E, 0x01, 0x10, 0x0E, 0x01, 0x20, 0x2E, 0x01, 0x10, 0x36, 0x13,
+ 0x33, 0x35, 0x33, 0x15, 0x33, 0x15, 0x23, 0x15, 0x23, 0x35, 0x23, 0x02, 0x5A, 0xB4, 0xA4, 0x77,
+ 0x46, 0x46, 0x77, 0xA4, 0xB4, 0xA4, 0x77, 0x46, 0x46, 0x77, 0xFE, 0x7C, 0x01, 0x0C, 0xE2, 0x84,
+ 0x84, 0xE2, 0xFE, 0xF4, 0xE2, 0x84, 0x84, 0x7C, 0xC5, 0x4E, 0xC5, 0xC4, 0x50, 0xC4, 0x03, 0x3F,
+ 0x46, 0x77, 0xA4, 0xB4, 0xA4, 0x77, 0x46, 0x46, 0x77, 0xA4, 0xB4, 0xA4, 0x77, 0x77, 0x84, 0xE2,
+ 0xFE, 0xF4, 0xE2, 0x84, 0x84, 0xE2, 0x01, 0x0C, 0xE2, 0xFE, 0xC0, 0xC4, 0xC5, 0x4E, 0xC5, 0xC5,
+ 0x00, 0x03, 0x00, 0x14, 0xFF, 0x98, 0x03, 0xEC, 0x03, 0x70, 0x00, 0x0F, 0x00, 0x1B, 0x00, 0x1F,
+ 0x00, 0x00, 0x00, 0x22, 0x0E, 0x02, 0x14, 0x1E, 0x02, 0x32, 0x3E, 0x02, 0x34, 0x2E, 0x01, 0x24,
+ 0x20, 0x1E, 0x01, 0x10, 0x0E, 0x01, 0x20, 0x2E, 0x01, 0x10, 0x36, 0x13, 0x35, 0x21, 0x15, 0x02,
+ 0x5A, 0xB4, 0xA4, 0x77, 0x46, 0x46, 0x77, 0xA4, 0xB4, 0xA4, 0x77, 0x46, 0x46, 0x77, 0xFE, 0x7C,
+ 0x01, 0x0C, 0xE2, 0x84, 0x84, 0xE2, 0xFE, 0xF4, 0xE2, 0x84, 0x84, 0x7C, 0x01, 0xD8, 0x03, 0x3F,
+ 0x46, 0x77, 0xA4, 0xB4, 0xA4, 0x77, 0x46, 0x46, 0x77, 0xA4, 0xB4, 0xA4, 0x77, 0x77, 0x84, 0xE2,
+ 0xFE, 0xF4, 0xE2, 0x84, 0x84, 0xE2, 0x01, 0x0C, 0xE2, 0xFE, 0x71, 0x4E, 0x4E, 0x00, 0x00, 0x00,
+ 0x00, 0x03, 0x00, 0x14, 0xFF, 0x98, 0x03, 0xEC, 0x03, 0x70, 0x00, 0x0B, 0x00, 0x1B, 0x00, 0x25,
+ 0x00, 0x00, 0x00, 0x20, 0x0E, 0x01, 0x10, 0x1E, 0x01, 0x20, 0x3E, 0x01, 0x10, 0x26, 0x01, 0x12,
+ 0x37, 0x33, 0x13, 0x12, 0x15, 0x16, 0x23, 0x2F, 0x01, 0x23, 0x07, 0x23, 0x22, 0x26, 0x25, 0x30,
+ 0x27, 0x26, 0x2F, 0x01, 0x06, 0x07, 0x06, 0x32, 0x02, 0x86, 0xFE, 0xF4, 0xE2, 0x84, 0x84, 0xE2,
+ 0x01, 0x0C, 0xE2, 0x84, 0x84, 0xFD, 0xA0, 0x6C, 0x5E, 0x6D, 0x68, 0x68, 0x01, 0x39, 0x38, 0x2E,
+ 0xD1, 0x2B, 0x37, 0x33, 0x04, 0x01, 0x48, 0x1D, 0x1C, 0x0A, 0x05, 0x01, 0x45, 0x01, 0x89, 0x03,
+ 0x70, 0x84, 0xE2, 0xFE, 0xF4, 0xE2, 0x84, 0x84, 0xE2, 0x01, 0x0C, 0xE2, 0xFD, 0x9A, 0x01, 0x1A,
+ 0xEB, 0xFE, 0xFE, 0xFE, 0xFD, 0x03, 0x01, 0x01, 0x77, 0x78, 0x01, 0xCF, 0x4C, 0x4C, 0x1C, 0x0C,
+ 0x02, 0xBE, 0x02, 0x00, 0x00, 0x04, 0x00, 0x14, 0xFF, 0x98, 0x03, 0xEC, 0x03, 0x70, 0x00, 0x0B,
+ 0x00, 0x20, 0x00, 0x2B, 0x00, 0x35, 0x00, 0x00, 0x36, 0x10, 0x3E, 0x01, 0x20, 0x1E, 0x01, 0x10,
+ 0x0E, 0x01, 0x20, 0x26, 0x01, 0x30, 0x37, 0x36, 0x37, 0x36, 0x27, 0x26, 0x27, 0x26, 0x23, 0x27,
+ 0x19, 0x01, 0x33, 0x32, 0x37, 0x3E, 0x01, 0x35, 0x26, 0x07, 0x06, 0x2B, 0x01, 0x35, 0x33, 0x1E,
+ 0x02, 0x15, 0x06, 0x27, 0x23, 0x35, 0x33, 0x16, 0x17, 0x16, 0x14, 0x07, 0x06, 0x14, 0x84, 0xE2,
+ 0x01, 0x0C, 0xE2, 0x84, 0x84, 0xE2, 0xFE, 0xF4, 0xE2, 0x01, 0xF7, 0x0A, 0x3A, 0x05, 0x04, 0x19,
+ 0x20, 0x3B, 0x1D, 0x8B, 0x81, 0x4E, 0xAF, 0x29, 0x3E, 0x4E, 0x01, 0xAE, 0x0D, 0x47, 0x46, 0x46,
+ 0x52, 0x2C, 0x1B, 0x01, 0xB7, 0x27, 0x4C, 0x4C, 0x07, 0x2C, 0x1E, 0x16, 0xFE, 0x01, 0x0C, 0xE2,
+ 0x84, 0x84, 0xE2, 0xFE, 0xF4, 0xE2, 0x84, 0x84, 0x01, 0x6D, 0x06, 0x21, 0x40, 0x2A, 0x24, 0x30,
+ 0x09, 0x05, 0x01, 0xFE, 0xFB, 0xFE, 0xFD, 0x03, 0x05, 0x4F, 0x41, 0x5B, 0x9B, 0x01, 0x8C, 0x01,
+ 0x0B, 0x21, 0x1A, 0x3E, 0xDA, 0x79, 0x01, 0x01, 0x0B, 0x54, 0x0E, 0x0A, 0x00, 0x02, 0x00, 0x14,
+ 0xFF, 0x98, 0x03, 0xEC, 0x03, 0x70, 0x00, 0x0B, 0x00, 0x29, 0x00, 0x00, 0x36, 0x10, 0x3E, 0x01,
+ 0x20, 0x1E, 0x01, 0x10, 0x0E, 0x01, 0x20, 0x26, 0x36, 0x14, 0x3B, 0x01, 0x37, 0x36, 0x37, 0x36,
+ 0x1F, 0x01, 0x33, 0x32, 0x34, 0x02, 0x26, 0x36, 0x34, 0x23, 0x0F, 0x01, 0x06, 0x07, 0x22, 0x2F,
+ 0x01, 0x23, 0x16, 0x1F, 0x01, 0x07, 0x14, 0x84, 0xE2, 0x01, 0x0C, 0xE2, 0x84, 0x84, 0xE2, 0xFE,
+ 0xF4, 0xE2, 0x7B, 0x3D, 0x3F, 0x38, 0x3A, 0x01, 0x02, 0x3A, 0x39, 0x3F, 0x3E, 0xB0, 0x01, 0xA1,
+ 0x3C, 0x3C, 0x32, 0x33, 0x01, 0x02, 0x34, 0x34, 0x7A, 0x05, 0x4F, 0x4D, 0x58, 0xFE, 0x01, 0x0C,
+ 0xE2, 0x84, 0x84, 0xE2, 0xFE, 0xF4, 0xE2, 0x84, 0x84, 0x62, 0x02, 0x59, 0x59, 0x02, 0x01, 0x5A,
+ 0x5B, 0x02, 0x01, 0x08, 0x04, 0xFB, 0x01, 0x01, 0x53, 0x53, 0x01, 0x54, 0x54, 0x06, 0x7A, 0x79,
+ 0x88, 0x00, 0x00, 0x00, 0x00, 0x02, 0x00, 0x14, 0xFF, 0x98, 0x03, 0xEC, 0x03, 0x70, 0x00, 0x0B,
+ 0x00, 0x1B, 0x00, 0x00, 0x00, 0x20, 0x1E, 0x01, 0x10, 0x0E, 0x01, 0x20, 0x2E, 0x01, 0x10, 0x36,
+ 0x01, 0x15, 0x33, 0x35, 0x13, 0x23, 0x07, 0x0E, 0x01, 0x2F, 0x01, 0x23, 0x22, 0x16, 0x1F, 0x01,
+ 0x01, 0x7A, 0x01, 0x0C, 0xE2, 0x84, 0x84, 0xE2, 0xFE, 0xF4, 0xE2, 0x84, 0x84, 0x01, 0x36, 0x68,
+ 0xBE, 0x77, 0x3B, 0x3C, 0x02, 0x3D, 0x3D, 0x3D, 0x3D, 0x01, 0x5F, 0x5E, 0x03, 0x70, 0x84, 0xE2,
+ 0xFE, 0xF4, 0xE2, 0x84, 0x84, 0xE2, 0x01, 0x0C, 0xE2, 0xFD, 0xF9, 0x6D, 0xDA, 0x01, 0x2D, 0x65,
+ 0x66, 0x03, 0x67, 0x67, 0x01, 0x95, 0x96, 0x00, 0x00, 0x02, 0x00, 0x14, 0xFF, 0xBF, 0x03, 0xEC,
+ 0x03, 0x4A, 0x00, 0x05, 0x00, 0x0B, 0x00, 0x00, 0x05, 0x21, 0x11, 0x10, 0x05, 0x21, 0x01, 0x21,
+ 0x35, 0x21, 0x11, 0x23, 0x03, 0xEC, 0xFC, 0x28, 0x01, 0x14, 0x02, 0xC4, 0xFD, 0x5C, 0x01, 0x70,
+ 0xFE, 0xF8, 0x68, 0x41, 0x02, 0x77, 0x01, 0x14, 0x01, 0xFD, 0x38, 0x57, 0x01, 0xB0, 0x00, 0x00,
+ 0x00, 0x03, 0x00, 0x14, 0xFF, 0xBF, 0x03, 0xEC, 0x03, 0x49, 0x00, 0x05, 0x00, 0x20, 0x00, 0x2B,
+ 0x00, 0x00, 0x17, 0x11, 0x21, 0x20, 0x19, 0x01, 0x25, 0x33, 0x35, 0x17, 0x1E, 0x01, 0x1F, 0x01,
+ 0x33, 0x37, 0x2E, 0x02, 0x27, 0x34, 0x37, 0x36, 0x37, 0x36, 0x27, 0x26, 0x27, 0x26, 0x27, 0x26,
+ 0x2B, 0x01, 0x05, 0x06, 0x2B, 0x01, 0x35, 0x33, 0x16, 0x17, 0x16, 0x15, 0x14, 0x14, 0x02, 0xC4,
+ 0x01, 0x14, 0xFD, 0x2A, 0x69, 0x19, 0x2E, 0x28, 0x56, 0x2A, 0x3D, 0x3D, 0x01, 0x65, 0x2C, 0x20,
+ 0x0D, 0x66, 0x13, 0x06, 0x04, 0x09, 0x34, 0x20, 0x49, 0x15, 0x76, 0x77, 0x01, 0x02, 0x0C, 0x47,
+ 0x46, 0x4F, 0x56, 0x10, 0x20, 0x41, 0x03, 0x8A, 0xFE, 0xED, 0xFD, 0x89, 0xC2, 0xDA, 0x01, 0x01,
+ 0x1A, 0x81, 0x3D, 0x01, 0x01, 0xA3, 0x2C, 0x13, 0x01, 0x02, 0x13, 0x5A, 0x1A, 0x1C, 0x44, 0x21,
+ 0x13, 0x04, 0x01, 0xDA, 0x02, 0x85, 0x01, 0x08, 0x0F, 0x29, 0x3A, 0x00, 0x00, 0x03, 0x00, 0x14,
+ 0xFF, 0xFB, 0x03, 0xEC, 0x03, 0x0E, 0x00, 0x08, 0x00, 0x15, 0x00, 0x1B, 0x00, 0x00, 0x05, 0x21,
+ 0x11, 0x10, 0x21, 0x30, 0x21, 0x32, 0x15, 0x01, 0x21, 0x35, 0x23, 0x13, 0x35, 0x21, 0x15, 0x33,
+ 0x32, 0x22, 0x0F, 0x01, 0x05, 0x21, 0x35, 0x23, 0x11, 0x23, 0x03, 0xEC, 0xFC, 0x28, 0x01, 0x8A,
+ 0x01, 0xEC, 0x62, 0xFC, 0xCF, 0x01, 0x40, 0xE1, 0xD9, 0xFE, 0xDF, 0x5D, 0x5C, 0x01, 0x67, 0x68,
+ 0x01, 0x75, 0x01, 0x15, 0xC6, 0x4F, 0x05, 0x01, 0x89, 0x01, 0x8A, 0x63, 0xFD, 0xE1, 0x42, 0x01,
+ 0x0B, 0x3D, 0x42, 0x80, 0x80, 0x48, 0x42, 0x01, 0x44, 0x00, 0x00, 0x00, 0x00, 0x04, 0x00, 0x14,
+ 0xFF, 0xFB, 0x03, 0xEC, 0x03, 0x0E, 0x00, 0x07, 0x00, 0x22, 0x00, 0x2F, 0x00, 0x3C, 0x00, 0x00,
+ 0x17, 0x11, 0x34, 0x37, 0x21, 0x20, 0x19, 0x01, 0x01, 0x15, 0x33, 0x35, 0x17, 0x1E, 0x01, 0x1F,
+ 0x02, 0x32, 0x35, 0x26, 0x27, 0x26, 0x27, 0x26, 0x37, 0x36, 0x37, 0x36, 0x27, 0x26, 0x27, 0x26,
+ 0x23, 0x27, 0x17, 0x30, 0x23, 0x35, 0x33, 0x32, 0x17, 0x16, 0x17, 0x14, 0x07, 0x0E, 0x01, 0x05,
+ 0x21, 0x35, 0x27, 0x13, 0x35, 0x21, 0x15, 0x33, 0x32, 0x14, 0x0F, 0x01, 0x14, 0x62, 0x01, 0xEC,
+ 0x01, 0x8A, 0xFE, 0x1E, 0x4E, 0x14, 0x29, 0x1E, 0x37, 0x22, 0x2F, 0x2F, 0x06, 0x3A, 0x1D, 0x1F,
+ 0x09, 0x09, 0x4E, 0x0E, 0x04, 0x05, 0x0F, 0x47, 0x15, 0x6F, 0x65, 0x82, 0x34, 0x37, 0x38, 0x07,
+ 0x23, 0x09, 0x13, 0x0D, 0x1A, 0xFD, 0xD6, 0x01, 0x40, 0xE1, 0xD8, 0xFE, 0xE0, 0x5C, 0x5C, 0x67,
+ 0x68, 0x05, 0x02, 0xB0, 0x62, 0x01, 0xFE, 0x76, 0xFE, 0x77, 0x01, 0x56, 0xC5, 0xA5, 0x01, 0x01,
+ 0x1C, 0x52, 0x34, 0x01, 0x01, 0x0E, 0x58, 0x2C, 0x13, 0x06, 0x04, 0x0F, 0x45, 0x1E, 0x14, 0x42,
+ 0x0D, 0x04, 0x01, 0xA7, 0x65, 0x01, 0x04, 0x2C, 0x21, 0x09, 0x07, 0x03, 0xE3, 0x41, 0x01, 0x01,
+ 0x0B, 0x3D, 0x42, 0x01, 0x80, 0x80, 0x00, 0x00, 0x00, 0x03, 0x00, 0x14, 0x00, 0x5D, 0x03, 0xEC,
+ 0x02, 0xAB, 0x00, 0x08, 0x00, 0x37, 0x00, 0x3D, 0x00, 0x00, 0x13, 0x30, 0x21, 0x11, 0x21, 0x22,
+ 0x3D, 0x01, 0x34, 0x05, 0x37, 0x34, 0x27, 0x26, 0x27, 0x26, 0x07, 0x06, 0x07, 0x0E, 0x01, 0x17,
+ 0x1E, 0x01, 0x17, 0x16, 0x14, 0x07, 0x06, 0x26, 0x27, 0x26, 0x27, 0x22, 0x06, 0x07, 0x22, 0x17,
+ 0x1E, 0x01, 0x17, 0x16, 0x37, 0x36, 0x27, 0x26, 0x27, 0x2E, 0x02, 0x37, 0x36, 0x33, 0x32, 0x1F,
+ 0x02, 0x33, 0x35, 0x23, 0x11, 0x23, 0xD6, 0x03, 0x16, 0xFC, 0xEA, 0xC2, 0x01, 0xC6, 0x02, 0x01,
+ 0x0C, 0x3A, 0x2B, 0x2D, 0x13, 0x10, 0x2B, 0x01, 0x33, 0x17, 0x55, 0x15, 0x04, 0x09, 0x14, 0x58,
+ 0x0C, 0x04, 0x02, 0x02, 0x26, 0x14, 0x01, 0x03, 0x08, 0x33, 0x38, 0x5F, 0x20, 0x10, 0x01, 0x03,
+ 0x3C, 0x12, 0x59, 0x11, 0x01, 0x02, 0x39, 0x2C, 0x09, 0x02, 0x9D, 0xE2, 0xA2, 0x40, 0x02, 0xAB,
+ 0xFD, 0xB2, 0xD2, 0xAA, 0xD2, 0xDC, 0x03, 0x07, 0x0B, 0x38, 0x10, 0x0C, 0x09, 0x04, 0x08, 0x19,
+ 0x6C, 0x17, 0x0B, 0x17, 0x11, 0x07, 0x17, 0x0A, 0x1A, 0x0A, 0x29, 0x0C, 0x04, 0x04, 0x02, 0x10,
+ 0x25, 0x37, 0x04, 0x06, 0x37, 0x1D, 0x1C, 0x3F, 0x19, 0x08, 0x16, 0x13, 0x0B, 0x1F, 0x2B, 0x04,
+ 0xE9, 0x37, 0x01, 0x13, 0x00, 0x04, 0x00, 0x14, 0x00, 0x5D, 0x03, 0xEC, 0x02, 0xAB, 0x00, 0x07,
+ 0x00, 0x1F, 0x00, 0x2A, 0x00, 0x58, 0x00, 0x00, 0x01, 0x32, 0x1D, 0x01, 0x14, 0x23, 0x21, 0x11,
+ 0x01, 0x33, 0x35, 0x17, 0x1E, 0x03, 0x3B, 0x01, 0x27, 0x2E, 0x01, 0x2F, 0x01, 0x36, 0x37, 0x36,
+ 0x27, 0x26, 0x27, 0x26, 0x2B, 0x01, 0x17, 0x30, 0x23, 0x35, 0x33, 0x32, 0x16, 0x17, 0x16, 0x07,
+ 0x06, 0x05, 0x16, 0x37, 0x36, 0x37, 0x3E, 0x01, 0x27, 0x2E, 0x03, 0x3E, 0x01, 0x17, 0x16, 0x17,
+ 0x30, 0x37, 0x36, 0x27, 0x26, 0x27, 0x26, 0x27, 0x22, 0x06, 0x07, 0x06, 0x1E, 0x03, 0x17, 0x16,
+ 0x07, 0x06, 0x26, 0x27, 0x26, 0x27, 0x07, 0x06, 0x23, 0x07, 0x16, 0x03, 0x2A, 0xC2, 0xC2, 0xFC,
+ 0xEA, 0x01, 0xEC, 0x41, 0x11, 0x1F, 0x17, 0x4D, 0x02, 0x27, 0x26, 0x16, 0x1E, 0x1C, 0x17, 0x04,
+ 0x43, 0x0C, 0x0B, 0x21, 0x18, 0x3E, 0x0F, 0x46, 0x47, 0x66, 0x25, 0x29, 0x3E, 0x1B, 0x03, 0x08,
+ 0x22, 0x0C, 0xFE, 0x4D, 0x22, 0x59, 0x34, 0x1E, 0x2B, 0x03, 0x33, 0x16, 0x5C, 0x16, 0x0C, 0x18,
+ 0x3C, 0x16, 0x0B, 0x05, 0x22, 0x21, 0x01, 0x03, 0x10, 0x1F, 0x49, 0x36, 0x43, 0x02, 0x01, 0x1C,
+ 0x2D, 0x56, 0x1B, 0x04, 0x07, 0x20, 0x13, 0x4B, 0x0D, 0x01, 0x04, 0x1D, 0x1E, 0x02, 0x02, 0x04,
+ 0x02, 0xAB, 0xD2, 0xAA, 0xD2, 0x02, 0x4E, 0xFE, 0x39, 0x89, 0x01, 0x01, 0x11, 0x75, 0x01, 0x25,
+ 0x2F, 0x27, 0x0F, 0x08, 0x0C, 0x38, 0x33, 0x21, 0x19, 0x02, 0x01, 0x8A, 0x53, 0x0D, 0x0F, 0x2A,
+ 0x09, 0x04, 0x8A, 0x3A, 0x03, 0x01, 0x12, 0x1B, 0x71, 0x1B, 0x0C, 0x17, 0x0D, 0x18, 0x17, 0x09,
+ 0x11, 0x09, 0x1A, 0x01, 0x01, 0x07, 0x1E, 0x15, 0x29, 0x01, 0x2D, 0x2D, 0x1A, 0x2C, 0x16, 0x16,
+ 0x0D, 0x0F, 0x1A, 0x14, 0x0C, 0x0D, 0x27, 0x04, 0x0C, 0x03, 0x03, 0x04, 0x1E, 0x00, 0x00, 0x00,
+ 0x00, 0x02, 0x00, 0x14, 0xFF, 0x98, 0x03, 0xEC, 0x03, 0x70, 0x00, 0x0B, 0x00, 0x17, 0x00, 0x00,
+ 0x00, 0x20, 0x1E, 0x01, 0x10, 0x0E, 0x01, 0x20, 0x2E, 0x01, 0x10, 0x36, 0x13, 0x15, 0x33, 0x15,
+ 0x33, 0x35, 0x33, 0x35, 0x23, 0x35, 0x23, 0x15, 0x01, 0x7A, 0x01, 0x0C, 0xE2, 0x84, 0x84, 0xE2,
+ 0xFE, 0xF4, 0xE2, 0x84, 0x84, 0x7C, 0xC4, 0x50, 0xC4, 0xC5, 0x4E, 0x03, 0x70, 0x84, 0xE2, 0xFE,
+ 0xF4, 0xE2, 0x84, 0x84, 0xE2, 0x01, 0x0C, 0xE2, 0xFE, 0xC0, 0x4F, 0xC5, 0xC5, 0x4E, 0xC5, 0xC4,
+ 0x00, 0x02, 0x00, 0x14, 0xFF, 0x98, 0x03, 0xEC, 0x03, 0x70, 0x00, 0x0B, 0x00, 0x0F, 0x00, 0x00,
+ 0x00, 0x20, 0x1E, 0x01, 0x10, 0x0E, 0x01, 0x20, 0x2E, 0x01, 0x10, 0x36, 0x13, 0x21, 0x35, 0x21,
+ 0x01, 0x7A, 0x01, 0x0C, 0xE2, 0x84, 0x84, 0xE2, 0xFE, 0xF4, 0xE2, 0x84, 0x84, 0x7C, 0x01, 0xD8,
+ 0xFE, 0x28, 0x03, 0x70, 0x84, 0xE2, 0xFE, 0xF4, 0xE2, 0x84, 0x84, 0xE2, 0x01, 0x0C, 0xE2, 0xFE,
+ 0x71, 0x4E, 0x00, 0x00, 0x00, 0x00, 0x00, 0x0E, 0x00, 0xAE, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00,
+ 0x00, 0x00, 0x00, 0x15, 0x00, 0x2C, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x10,
+ 0x00, 0x64, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x02, 0x00, 0x07, 0x00, 0x85, 0x00, 0x01,
+ 0x00, 0x00, 0x00, 0x00, 0x00, 0x03, 0x00, 0x10, 0x00, 0xAF, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00,
+ 0x00, 0x04, 0x00, 0x10, 0x00, 0xE2, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x05, 0x00, 0x0D,
+ 0x01, 0x0F, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x06, 0x00, 0x10, 0x01, 0x3F, 0x00, 0x03,
+ 0x00, 0x01, 0x04, 0x09, 0x00, 0x00, 0x00, 0x2A, 0x00, 0x00, 0x00, 0x03, 0x00, 0x01, 0x04, 0x09,
+ 0x00, 0x01, 0x00, 0x20, 0x00, 0x42, 0x00, 0x03, 0x00, 0x01, 0x04, 0x09, 0x00, 0x02, 0x00, 0x0E,
+ 0x00, 0x75, 0x00, 0x03, 0x00, 0x01, 0x04, 0x09, 0x00, 0x03, 0x00, 0x20, 0x00, 0x8D, 0x00, 0x03,
+ 0x00, 0x01, 0x04, 0x09, 0x00, 0x04, 0x00, 0x20, 0x00, 0xC0, 0x00, 0x03, 0x00, 0x01, 0x04, 0x09,
+ 0x00, 0x05, 0x00, 0x1A, 0x00, 0xF3, 0x00, 0x03, 0x00, 0x01, 0x04, 0x09, 0x00, 0x06, 0x00, 0x20,
+ 0x01, 0x1D, 0x00, 0x59, 0x00, 0x75, 0x00, 0x7A, 0x00, 0x75, 0x00, 0x20, 0x00, 0x45, 0x00, 0x6D,
+ 0x00, 0x75, 0x00, 0x6C, 0x00, 0x61, 0x00, 0x74, 0x00, 0x6F, 0x00, 0x72, 0x00, 0x20, 0x00, 0x50,
+ 0x00, 0x72, 0x00, 0x6F, 0x00, 0x6A, 0x00, 0x65, 0x00, 0x63, 0x00, 0x74, 0x00, 0x00, 0x59, 0x75,
+ 0x7A, 0x75, 0x20, 0x45, 0x6D, 0x75, 0x6C, 0x61, 0x74, 0x6F, 0x72, 0x20, 0x50, 0x72, 0x6F, 0x6A,
+ 0x65, 0x63, 0x74, 0x00, 0x00, 0x59, 0x00, 0x75, 0x00, 0x7A, 0x00, 0x75, 0x00, 0x4F, 0x00, 0x53,
+ 0x00, 0x53, 0x00, 0x45, 0x00, 0x78, 0x00, 0x74, 0x00, 0x65, 0x00, 0x6E, 0x00, 0x73, 0x00, 0x69,
+ 0x00, 0x6F, 0x00, 0x6E, 0x00, 0x00, 0x59, 0x75, 0x7A, 0x75, 0x4F, 0x53, 0x53, 0x45, 0x78, 0x74,
+ 0x65, 0x6E, 0x73, 0x69, 0x6F, 0x6E, 0x00, 0x00, 0x52, 0x00, 0x65, 0x00, 0x67, 0x00, 0x75, 0x00,
+ 0x6C, 0x00, 0x61, 0x00, 0x72, 0x00, 0x00, 0x52, 0x65, 0x67, 0x75, 0x6C, 0x61, 0x72, 0x00, 0x00,
+ 0x59, 0x00, 0x75, 0x00, 0x7A, 0x00, 0x75, 0x00, 0x4F, 0x00, 0x53, 0x00, 0x53, 0x00, 0x45, 0x00,
+ 0x78, 0x00, 0x74, 0x00, 0x65, 0x00, 0x6E, 0x00, 0x73, 0x00, 0x69, 0x00, 0x6F, 0x00, 0x6E, 0x00,
+ 0x00, 0x59, 0x75, 0x7A, 0x75, 0x4F, 0x53, 0x53, 0x45, 0x78, 0x74, 0x65, 0x6E, 0x73, 0x69, 0x6F,
+ 0x6E, 0x00, 0x00, 0x59, 0x00, 0x75, 0x00, 0x7A, 0x00, 0x75, 0x00, 0x4F, 0x00, 0x53, 0x00, 0x53,
+ 0x00, 0x45, 0x00, 0x78, 0x00, 0x74, 0x00, 0x65, 0x00, 0x6E, 0x00, 0x73, 0x00, 0x69, 0x00, 0x6F,
+ 0x00, 0x6E, 0x00, 0x00, 0x59, 0x75, 0x7A, 0x75, 0x4F, 0x53, 0x53, 0x45, 0x78, 0x74, 0x65, 0x6E,
+ 0x73, 0x69, 0x6F, 0x6E, 0x00, 0x00, 0x56, 0x00, 0x65, 0x00, 0x72, 0x00, 0x73, 0x00, 0x69, 0x00,
+ 0x6F, 0x00, 0x6E, 0x00, 0x20, 0x00, 0x31, 0x00, 0x2E, 0x00, 0x30, 0x00, 0x30, 0x00, 0x30, 0x00,
+ 0x00, 0x56, 0x65, 0x72, 0x73, 0x69, 0x6F, 0x6E, 0x20, 0x31, 0x2E, 0x30, 0x30, 0x30, 0x00, 0x00,
+ 0x59, 0x00, 0x75, 0x00, 0x7A, 0x00, 0x75, 0x00, 0x4F, 0x00, 0x53, 0x00, 0x53, 0x00, 0x45, 0x00,
+ 0x78, 0x00, 0x74, 0x00, 0x65, 0x00, 0x6E, 0x00, 0x73, 0x00, 0x69, 0x00, 0x6F, 0x00, 0x6E, 0x00,
+ 0x00, 0x59, 0x75, 0x7A, 0x75, 0x4F, 0x53, 0x53, 0x45, 0x78, 0x74, 0x65, 0x6E, 0x73, 0x69, 0x6F,
+ 0x6E, 0x00, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xFF, 0xB5, 0x00, 0x32,
+ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+ 0x00, 0x00, 0x00, 0x00, 0x00, 0x1C, 0x00, 0x00, 0x01, 0x02, 0x01, 0x03, 0x00, 0x03, 0x01, 0x04,
+ 0x01, 0x05, 0x01, 0x06, 0x01, 0x07, 0x01, 0x08, 0x01, 0x09, 0x01, 0x0A, 0x01, 0x0B, 0x01, 0x0C,
+ 0x01, 0x0D, 0x01, 0x0E, 0x01, 0x0F, 0x01, 0x10, 0x01, 0x11, 0x01, 0x12, 0x01, 0x13, 0x01, 0x14,
+ 0x01, 0x15, 0x01, 0x16, 0x01, 0x17, 0x01, 0x18, 0x01, 0x19, 0x01, 0x1A, 0x01, 0x1B, 0x07, 0x75,
+ 0x6E, 0x69, 0x30, 0x30, 0x30, 0x30, 0x07, 0x75, 0x6E, 0x69, 0x30, 0x30, 0x30, 0x44, 0x07, 0x75,
+ 0x6E, 0x69, 0x45, 0x30, 0x41, 0x30, 0x07, 0x75, 0x6E, 0x69, 0x45, 0x30, 0x41, 0x31, 0x07, 0x75,
+ 0x6E, 0x69, 0x45, 0x30, 0x41, 0x32, 0x07, 0x75, 0x6E, 0x69, 0x45, 0x30, 0x41, 0x33, 0x07, 0x75,
+ 0x6E, 0x69, 0x45, 0x30, 0x41, 0x34, 0x07, 0x75, 0x6E, 0x69, 0x45, 0x30, 0x41, 0x35, 0x07, 0x75,
+ 0x6E, 0x69, 0x45, 0x30, 0x41, 0x36, 0x07, 0x75, 0x6E, 0x69, 0x45, 0x30, 0x41, 0x37, 0x07, 0x75,
+ 0x6E, 0x69, 0x45, 0x30, 0x41, 0x38, 0x07, 0x75, 0x6E, 0x69, 0x45, 0x30, 0x41, 0x39, 0x07, 0x75,
+ 0x6E, 0x69, 0x45, 0x30, 0x42, 0x33, 0x07, 0x75, 0x6E, 0x69, 0x45, 0x30, 0x42, 0x34, 0x07, 0x75,
+ 0x6E, 0x69, 0x45, 0x30, 0x45, 0x30, 0x07, 0x75, 0x6E, 0x69, 0x45, 0x30, 0x45, 0x31, 0x07, 0x75,
+ 0x6E, 0x69, 0x45, 0x30, 0x45, 0x32, 0x07, 0x75, 0x6E, 0x69, 0x45, 0x30, 0x45, 0x33, 0x07, 0x75,
+ 0x6E, 0x69, 0x45, 0x30, 0x45, 0x34, 0x07, 0x75, 0x6E, 0x69, 0x45, 0x30, 0x45, 0x35, 0x07, 0x75,
+ 0x6E, 0x69, 0x45, 0x30, 0x45, 0x36, 0x07, 0x75, 0x6E, 0x69, 0x45, 0x30, 0x45, 0x37, 0x07, 0x75,
+ 0x6E, 0x69, 0x45, 0x30, 0x45, 0x38, 0x07, 0x75, 0x6E, 0x69, 0x45, 0x30, 0x45, 0x39, 0x07, 0x75,
+ 0x6E, 0x69, 0x45, 0x30, 0x45, 0x46, 0x07, 0x75, 0x6E, 0x69, 0x45, 0x30, 0x46, 0x30, 0x00, 0x00,
+ 0x00, 0x01, 0x00, 0x01, 0xFF, 0xFF, 0x00, 0x0F,
}};
} // namespace FileSys::SystemArchive::SharedFontData
diff --git a/src/core/file_sys/system_archive/data/font_nintendo_extended.h b/src/core/file_sys/system_archive/data/font_nintendo_extended.h
index 2089f3db9..edb9df914 100644
--- a/src/core/file_sys/system_archive/data/font_nintendo_extended.h
+++ b/src/core/file_sys/system_archive/data/font_nintendo_extended.h
@@ -8,6 +8,6 @@
namespace FileSys::SystemArchive::SharedFontData {
-extern const std::array<unsigned char, 2932> FONT_NINTENDO_EXTENDED;
+extern const std::array<unsigned char, 6024> FONT_NINTENDO_EXTENDED;
} // namespace FileSys::SystemArchive::SharedFontData
diff --git a/src/core/file_sys/system_archive/system_version.cpp b/src/core/file_sys/system_archive/system_version.cpp
index aa313de66..7bfbc9a67 100644
--- a/src/core/file_sys/system_archive/system_version.cpp
+++ b/src/core/file_sys/system_archive/system_version.cpp
@@ -12,17 +12,17 @@ namespace SystemVersionData {
// This section should reflect the best system version to describe yuzu's HLE api.
// TODO(DarkLordZach): Update when HLE gets better.
-constexpr u8 VERSION_MAJOR = 10;
+constexpr u8 VERSION_MAJOR = 11;
constexpr u8 VERSION_MINOR = 0;
-constexpr u8 VERSION_MICRO = 2;
+constexpr u8 VERSION_MICRO = 0;
-constexpr u8 REVISION_MAJOR = 1;
+constexpr u8 REVISION_MAJOR = 5;
constexpr u8 REVISION_MINOR = 0;
constexpr char PLATFORM_STRING[] = "NX";
-constexpr char VERSION_HASH[] = "f90143fa8bbc061d4f68c35f95f04f8080c0ecdc";
-constexpr char DISPLAY_VERSION[] = "10.0.2";
-constexpr char DISPLAY_TITLE[] = "NintendoSDK Firmware for NX 10.0.2-1.0";
+constexpr char VERSION_HASH[] = "34197eba8810e2edd5e9dfcfbde7b340882e856d";
+constexpr char DISPLAY_VERSION[] = "11.0.0";
+constexpr char DISPLAY_TITLE[] = "NintendoSDK Firmware for NX 11.0.0-5.0";
} // namespace SystemVersionData
diff --git a/src/core/file_sys/vfs.cpp b/src/core/file_sys/vfs.cpp
index b2f026b6d..f497e9396 100644
--- a/src/core/file_sys/vfs.cpp
+++ b/src/core/file_sys/vfs.cpp
@@ -203,7 +203,7 @@ std::string VfsFile::GetFullPath() const {
return GetContainingDirectory()->GetFullPath() + "/" + GetName();
}
-std::shared_ptr<VfsFile> VfsDirectory::GetFileRelative(std::string_view path) const {
+VirtualFile VfsDirectory::GetFileRelative(std::string_view path) const {
auto vec = Common::FS::SplitPathComponents(path);
vec.erase(std::remove_if(vec.begin(), vec.end(), [](const auto& str) { return str.empty(); }),
vec.end());
@@ -231,7 +231,7 @@ std::shared_ptr<VfsFile> VfsDirectory::GetFileRelative(std::string_view path) co
return dir->GetFile(vec.back());
}
-std::shared_ptr<VfsFile> VfsDirectory::GetFileAbsolute(std::string_view path) const {
+VirtualFile VfsDirectory::GetFileAbsolute(std::string_view path) const {
if (IsRoot()) {
return GetFileRelative(path);
}
@@ -239,7 +239,7 @@ std::shared_ptr<VfsFile> VfsDirectory::GetFileAbsolute(std::string_view path) co
return GetParentDirectory()->GetFileAbsolute(path);
}
-std::shared_ptr<VfsDirectory> VfsDirectory::GetDirectoryRelative(std::string_view path) const {
+VirtualDir VfsDirectory::GetDirectoryRelative(std::string_view path) const {
auto vec = Common::FS::SplitPathComponents(path);
vec.erase(std::remove_if(vec.begin(), vec.end(), [](const auto& str) { return str.empty(); }),
vec.end());
@@ -261,7 +261,7 @@ std::shared_ptr<VfsDirectory> VfsDirectory::GetDirectoryRelative(std::string_vie
return dir;
}
-std::shared_ptr<VfsDirectory> VfsDirectory::GetDirectoryAbsolute(std::string_view path) const {
+VirtualDir VfsDirectory::GetDirectoryAbsolute(std::string_view path) const {
if (IsRoot()) {
return GetDirectoryRelative(path);
}
@@ -269,14 +269,14 @@ std::shared_ptr<VfsDirectory> VfsDirectory::GetDirectoryAbsolute(std::string_vie
return GetParentDirectory()->GetDirectoryAbsolute(path);
}
-std::shared_ptr<VfsFile> VfsDirectory::GetFile(std::string_view name) const {
+VirtualFile VfsDirectory::GetFile(std::string_view name) const {
const auto& files = GetFiles();
const auto iter = std::find_if(files.begin(), files.end(),
[&name](const auto& file1) { return name == file1->GetName(); });
return iter == files.end() ? nullptr : *iter;
}
-std::shared_ptr<VfsDirectory> VfsDirectory::GetSubdirectory(std::string_view name) const {
+VirtualDir VfsDirectory::GetSubdirectory(std::string_view name) const {
const auto& subs = GetSubdirectories();
const auto iter = std::find_if(subs.begin(), subs.end(),
[&name](const auto& file1) { return name == file1->GetName(); });
@@ -301,7 +301,7 @@ std::size_t VfsDirectory::GetSize() const {
return file_total + subdir_total;
}
-std::shared_ptr<VfsFile> VfsDirectory::CreateFileRelative(std::string_view path) {
+VirtualFile VfsDirectory::CreateFileRelative(std::string_view path) {
auto vec = Common::FS::SplitPathComponents(path);
vec.erase(std::remove_if(vec.begin(), vec.end(), [](const auto& str) { return str.empty(); }),
vec.end());
@@ -324,7 +324,7 @@ std::shared_ptr<VfsFile> VfsDirectory::CreateFileRelative(std::string_view path)
return dir->CreateFileRelative(Common::FS::GetPathWithoutTop(path));
}
-std::shared_ptr<VfsFile> VfsDirectory::CreateFileAbsolute(std::string_view path) {
+VirtualFile VfsDirectory::CreateFileAbsolute(std::string_view path) {
if (IsRoot()) {
return CreateFileRelative(path);
}
@@ -332,7 +332,7 @@ std::shared_ptr<VfsFile> VfsDirectory::CreateFileAbsolute(std::string_view path)
return GetParentDirectory()->CreateFileAbsolute(path);
}
-std::shared_ptr<VfsDirectory> VfsDirectory::CreateDirectoryRelative(std::string_view path) {
+VirtualDir VfsDirectory::CreateDirectoryRelative(std::string_view path) {
auto vec = Common::FS::SplitPathComponents(path);
vec.erase(std::remove_if(vec.begin(), vec.end(), [](const auto& str) { return str.empty(); }),
vec.end());
@@ -355,7 +355,7 @@ std::shared_ptr<VfsDirectory> VfsDirectory::CreateDirectoryRelative(std::string_
return dir->CreateDirectoryRelative(Common::FS::GetPathWithoutTop(path));
}
-std::shared_ptr<VfsDirectory> VfsDirectory::CreateDirectoryAbsolute(std::string_view path) {
+VirtualDir VfsDirectory::CreateDirectoryAbsolute(std::string_view path) {
if (IsRoot()) {
return CreateDirectoryRelative(path);
}
@@ -446,27 +446,27 @@ bool ReadOnlyVfsDirectory::IsReadable() const {
return true;
}
-std::shared_ptr<VfsDirectory> ReadOnlyVfsDirectory::CreateSubdirectory(std::string_view name) {
+VirtualDir ReadOnlyVfsDirectory::CreateSubdirectory(std::string_view name) {
return nullptr;
}
-std::shared_ptr<VfsFile> ReadOnlyVfsDirectory::CreateFile(std::string_view name) {
+VirtualFile ReadOnlyVfsDirectory::CreateFile(std::string_view name) {
return nullptr;
}
-std::shared_ptr<VfsFile> ReadOnlyVfsDirectory::CreateFileAbsolute(std::string_view path) {
+VirtualFile ReadOnlyVfsDirectory::CreateFileAbsolute(std::string_view path) {
return nullptr;
}
-std::shared_ptr<VfsFile> ReadOnlyVfsDirectory::CreateFileRelative(std::string_view path) {
+VirtualFile ReadOnlyVfsDirectory::CreateFileRelative(std::string_view path) {
return nullptr;
}
-std::shared_ptr<VfsDirectory> ReadOnlyVfsDirectory::CreateDirectoryAbsolute(std::string_view path) {
+VirtualDir ReadOnlyVfsDirectory::CreateDirectoryAbsolute(std::string_view path) {
return nullptr;
}
-std::shared_ptr<VfsDirectory> ReadOnlyVfsDirectory::CreateDirectoryRelative(std::string_view path) {
+VirtualDir ReadOnlyVfsDirectory::CreateDirectoryRelative(std::string_view path) {
return nullptr;
}
diff --git a/src/core/file_sys/vfs.h b/src/core/file_sys/vfs.h
index 954094772..afd64e95c 100644
--- a/src/core/file_sys/vfs.h
+++ b/src/core/file_sys/vfs.h
@@ -91,7 +91,7 @@ public:
// Resizes the file to new_size. Returns whether or not the operation was successful.
virtual bool Resize(std::size_t new_size) = 0;
// Gets a pointer to the directory containing this file, returning nullptr if there is none.
- virtual std::shared_ptr<VfsDirectory> GetContainingDirectory() const = 0;
+ virtual VirtualDir GetContainingDirectory() const = 0;
// Returns whether or not the file can be written to.
virtual bool IsWritable() const = 0;
@@ -183,27 +183,27 @@ public:
// Retrives the file located at path as if the current directory was root. Returns nullptr if
// not found.
- virtual std::shared_ptr<VfsFile> GetFileRelative(std::string_view path) const;
+ virtual VirtualFile GetFileRelative(std::string_view path) const;
// Calls GetFileRelative(path) on the root of the current directory.
- virtual std::shared_ptr<VfsFile> GetFileAbsolute(std::string_view path) const;
+ virtual VirtualFile GetFileAbsolute(std::string_view path) const;
// Retrives the directory located at path as if the current directory was root. Returns nullptr
// if not found.
- virtual std::shared_ptr<VfsDirectory> GetDirectoryRelative(std::string_view path) const;
+ virtual VirtualDir GetDirectoryRelative(std::string_view path) const;
// Calls GetDirectoryRelative(path) on the root of the current directory.
- virtual std::shared_ptr<VfsDirectory> GetDirectoryAbsolute(std::string_view path) const;
+ virtual VirtualDir GetDirectoryAbsolute(std::string_view path) const;
// Returns a vector containing all of the files in this directory.
- virtual std::vector<std::shared_ptr<VfsFile>> GetFiles() const = 0;
+ virtual std::vector<VirtualFile> GetFiles() const = 0;
// Returns the file with filename matching name. Returns nullptr if directory dosen't have a
// file with name.
- virtual std::shared_ptr<VfsFile> GetFile(std::string_view name) const;
+ virtual VirtualFile GetFile(std::string_view name) const;
// Returns a vector containing all of the subdirectories in this directory.
- virtual std::vector<std::shared_ptr<VfsDirectory>> GetSubdirectories() const = 0;
+ virtual std::vector<VirtualDir> GetSubdirectories() const = 0;
// Returns the directory with name matching name. Returns nullptr if directory dosen't have a
// directory with name.
- virtual std::shared_ptr<VfsDirectory> GetSubdirectory(std::string_view name) const;
+ virtual VirtualDir GetSubdirectory(std::string_view name) const;
// Returns whether or not the directory can be written to.
virtual bool IsWritable() const = 0;
@@ -219,31 +219,31 @@ public:
virtual std::size_t GetSize() const;
// Returns the parent directory of this directory. Returns nullptr if this directory is root or
// has no parent.
- virtual std::shared_ptr<VfsDirectory> GetParentDirectory() const = 0;
+ virtual VirtualDir GetParentDirectory() const = 0;
// Creates a new subdirectory with name name. Returns a pointer to the new directory or nullptr
// if the operation failed.
- virtual std::shared_ptr<VfsDirectory> CreateSubdirectory(std::string_view name) = 0;
+ virtual VirtualDir CreateSubdirectory(std::string_view name) = 0;
// Creates a new file with name name. Returns a pointer to the new file or nullptr if the
// operation failed.
- virtual std::shared_ptr<VfsFile> CreateFile(std::string_view name) = 0;
+ virtual VirtualFile CreateFile(std::string_view name) = 0;
// Creates a new file at the path relative to this directory. Also creates directories if
// they do not exist and is supported by this implementation. Returns nullptr on any failure.
- virtual std::shared_ptr<VfsFile> CreateFileRelative(std::string_view path);
+ virtual VirtualFile CreateFileRelative(std::string_view path);
// Creates a new file at the path relative to root of this directory. Also creates directories
// if they do not exist and is supported by this implementation. Returns nullptr on any failure.
- virtual std::shared_ptr<VfsFile> CreateFileAbsolute(std::string_view path);
+ virtual VirtualFile CreateFileAbsolute(std::string_view path);
// Creates a new directory at the path relative to this directory. Also creates directories if
// they do not exist and is supported by this implementation. Returns nullptr on any failure.
- virtual std::shared_ptr<VfsDirectory> CreateDirectoryRelative(std::string_view path);
+ virtual VirtualDir CreateDirectoryRelative(std::string_view path);
// Creates a new directory at the path relative to root of this directory. Also creates
// directories if they do not exist and is supported by this implementation. Returns nullptr on
// any failure.
- virtual std::shared_ptr<VfsDirectory> CreateDirectoryAbsolute(std::string_view path);
+ virtual VirtualDir CreateDirectoryAbsolute(std::string_view path);
// Deletes the subdirectory with the given name and returns true on success.
virtual bool DeleteSubdirectory(std::string_view name) = 0;
@@ -280,12 +280,12 @@ class ReadOnlyVfsDirectory : public VfsDirectory {
public:
bool IsWritable() const override;
bool IsReadable() const override;
- std::shared_ptr<VfsDirectory> CreateSubdirectory(std::string_view name) override;
- std::shared_ptr<VfsFile> CreateFile(std::string_view name) override;
- std::shared_ptr<VfsFile> CreateFileAbsolute(std::string_view path) override;
- std::shared_ptr<VfsFile> CreateFileRelative(std::string_view path) override;
- std::shared_ptr<VfsDirectory> CreateDirectoryAbsolute(std::string_view path) override;
- std::shared_ptr<VfsDirectory> CreateDirectoryRelative(std::string_view path) override;
+ VirtualDir CreateSubdirectory(std::string_view name) override;
+ VirtualFile CreateFile(std::string_view name) override;
+ VirtualFile CreateFileAbsolute(std::string_view path) override;
+ VirtualFile CreateFileRelative(std::string_view path) override;
+ VirtualDir CreateDirectoryAbsolute(std::string_view path) override;
+ VirtualDir CreateDirectoryRelative(std::string_view path) override;
bool DeleteSubdirectory(std::string_view name) override;
bool DeleteSubdirectoryRecursive(std::string_view name) override;
bool CleanSubdirectoryRecursive(std::string_view name) override;
diff --git a/src/core/file_sys/vfs_concat.cpp b/src/core/file_sys/vfs_concat.cpp
index e0ff70174..3c5a7d87a 100644
--- a/src/core/file_sys/vfs_concat.cpp
+++ b/src/core/file_sys/vfs_concat.cpp
@@ -46,7 +46,7 @@ VirtualFile ConcatenatedVfsFile::MakeConcatenatedFile(std::vector<VirtualFile> f
if (files.size() == 1)
return files[0];
- return std::shared_ptr<VfsFile>(new ConcatenatedVfsFile(std::move(files), std::move(name)));
+ return VirtualFile(new ConcatenatedVfsFile(std::move(files), std::move(name)));
}
VirtualFile ConcatenatedVfsFile::MakeConcatenatedFile(u8 filler_byte,
@@ -71,20 +71,23 @@ VirtualFile ConcatenatedVfsFile::MakeConcatenatedFile(u8 filler_byte,
if (files.begin()->first != 0)
files.emplace(0, std::make_shared<StaticVfsFile>(filler_byte, files.begin()->first));
- return std::shared_ptr<VfsFile>(new ConcatenatedVfsFile(std::move(files), std::move(name)));
+ return VirtualFile(new ConcatenatedVfsFile(std::move(files), std::move(name)));
}
std::string ConcatenatedVfsFile::GetName() const {
- if (files.empty())
+ if (files.empty()) {
return "";
- if (!name.empty())
+ }
+ if (!name.empty()) {
return name;
+ }
return files.begin()->second->GetName();
}
std::size_t ConcatenatedVfsFile::GetSize() const {
- if (files.empty())
+ if (files.empty()) {
return 0;
+ }
return files.rbegin()->first + files.rbegin()->second->GetSize();
}
@@ -92,9 +95,10 @@ bool ConcatenatedVfsFile::Resize(std::size_t new_size) {
return false;
}
-std::shared_ptr<VfsDirectory> ConcatenatedVfsFile::GetContainingDirectory() const {
- if (files.empty())
+VirtualDir ConcatenatedVfsFile::GetContainingDirectory() const {
+ if (files.empty()) {
return nullptr;
+ }
return files.begin()->second->GetContainingDirectory();
}
diff --git a/src/core/file_sys/vfs_concat.h b/src/core/file_sys/vfs_concat.h
index 7a26343c0..287c72555 100644
--- a/src/core/file_sys/vfs_concat.h
+++ b/src/core/file_sys/vfs_concat.h
@@ -31,7 +31,7 @@ public:
std::string GetName() const override;
std::size_t GetSize() const override;
bool Resize(std::size_t new_size) override;
- std::shared_ptr<VfsDirectory> GetContainingDirectory() const override;
+ VirtualDir GetContainingDirectory() const override;
bool IsWritable() const override;
bool IsReadable() const override;
std::size_t Read(u8* data, std::size_t length, std::size_t offset) const override;
diff --git a/src/core/file_sys/vfs_layered.cpp b/src/core/file_sys/vfs_layered.cpp
index 338e398da..434b03cec 100644
--- a/src/core/file_sys/vfs_layered.cpp
+++ b/src/core/file_sys/vfs_layered.cpp
@@ -20,10 +20,10 @@ VirtualDir LayeredVfsDirectory::MakeLayeredDirectory(std::vector<VirtualDir> dir
if (dirs.size() == 1)
return dirs[0];
- return std::shared_ptr<VfsDirectory>(new LayeredVfsDirectory(std::move(dirs), std::move(name)));
+ return VirtualDir(new LayeredVfsDirectory(std::move(dirs), std::move(name)));
}
-std::shared_ptr<VfsFile> LayeredVfsDirectory::GetFileRelative(std::string_view path) const {
+VirtualFile LayeredVfsDirectory::GetFileRelative(std::string_view path) const {
for (const auto& layer : dirs) {
const auto file = layer->GetFileRelative(path);
if (file != nullptr)
@@ -33,23 +33,23 @@ std::shared_ptr<VfsFile> LayeredVfsDirectory::GetFileRelative(std::string_view p
return nullptr;
}
-std::shared_ptr<VfsDirectory> LayeredVfsDirectory::GetDirectoryRelative(
- std::string_view path) const {
+VirtualDir LayeredVfsDirectory::GetDirectoryRelative(std::string_view path) const {
std::vector<VirtualDir> out;
for (const auto& layer : dirs) {
auto dir = layer->GetDirectoryRelative(path);
- if (dir != nullptr)
+ if (dir != nullptr) {
out.push_back(std::move(dir));
+ }
}
return MakeLayeredDirectory(std::move(out));
}
-std::shared_ptr<VfsFile> LayeredVfsDirectory::GetFile(std::string_view name) const {
+VirtualFile LayeredVfsDirectory::GetFile(std::string_view name) const {
return GetFileRelative(name);
}
-std::shared_ptr<VfsDirectory> LayeredVfsDirectory::GetSubdirectory(std::string_view name) const {
+VirtualDir LayeredVfsDirectory::GetSubdirectory(std::string_view name) const {
return GetDirectoryRelative(name);
}
@@ -57,7 +57,7 @@ std::string LayeredVfsDirectory::GetFullPath() const {
return dirs[0]->GetFullPath();
}
-std::vector<std::shared_ptr<VfsFile>> LayeredVfsDirectory::GetFiles() const {
+std::vector<VirtualFile> LayeredVfsDirectory::GetFiles() const {
std::vector<VirtualFile> out;
for (const auto& layer : dirs) {
for (const auto& file : layer->GetFiles()) {
@@ -72,7 +72,7 @@ std::vector<std::shared_ptr<VfsFile>> LayeredVfsDirectory::GetFiles() const {
return out;
}
-std::vector<std::shared_ptr<VfsDirectory>> LayeredVfsDirectory::GetSubdirectories() const {
+std::vector<VirtualDir> LayeredVfsDirectory::GetSubdirectories() const {
std::vector<std::string> names;
for (const auto& layer : dirs) {
for (const auto& sd : layer->GetSubdirectories()) {
@@ -101,15 +101,15 @@ std::string LayeredVfsDirectory::GetName() const {
return name.empty() ? dirs[0]->GetName() : name;
}
-std::shared_ptr<VfsDirectory> LayeredVfsDirectory::GetParentDirectory() const {
+VirtualDir LayeredVfsDirectory::GetParentDirectory() const {
return dirs[0]->GetParentDirectory();
}
-std::shared_ptr<VfsDirectory> LayeredVfsDirectory::CreateSubdirectory(std::string_view name) {
+VirtualDir LayeredVfsDirectory::CreateSubdirectory(std::string_view name) {
return nullptr;
}
-std::shared_ptr<VfsFile> LayeredVfsDirectory::CreateFile(std::string_view name) {
+VirtualFile LayeredVfsDirectory::CreateFile(std::string_view name) {
return nullptr;
}
diff --git a/src/core/file_sys/vfs_layered.h b/src/core/file_sys/vfs_layered.h
index 8a25c3428..6d7513ac6 100644
--- a/src/core/file_sys/vfs_layered.h
+++ b/src/core/file_sys/vfs_layered.h
@@ -21,20 +21,20 @@ public:
/// Wrapper function to allow for more efficient handling of dirs.size() == 0, 1 cases.
static VirtualDir MakeLayeredDirectory(std::vector<VirtualDir> dirs, std::string name = "");
- std::shared_ptr<VfsFile> GetFileRelative(std::string_view path) const override;
- std::shared_ptr<VfsDirectory> GetDirectoryRelative(std::string_view path) const override;
- std::shared_ptr<VfsFile> GetFile(std::string_view name) const override;
- std::shared_ptr<VfsDirectory> GetSubdirectory(std::string_view name) const override;
+ VirtualFile GetFileRelative(std::string_view path) const override;
+ VirtualDir GetDirectoryRelative(std::string_view path) const override;
+ VirtualFile GetFile(std::string_view name) const override;
+ VirtualDir GetSubdirectory(std::string_view name) const override;
std::string GetFullPath() const override;
- std::vector<std::shared_ptr<VfsFile>> GetFiles() const override;
- std::vector<std::shared_ptr<VfsDirectory>> GetSubdirectories() const override;
+ std::vector<VirtualFile> GetFiles() const override;
+ std::vector<VirtualDir> GetSubdirectories() const override;
bool IsWritable() const override;
bool IsReadable() const override;
std::string GetName() const override;
- std::shared_ptr<VfsDirectory> GetParentDirectory() const override;
- std::shared_ptr<VfsDirectory> CreateSubdirectory(std::string_view name) override;
- std::shared_ptr<VfsFile> CreateFile(std::string_view name) override;
+ VirtualDir GetParentDirectory() const override;
+ VirtualDir CreateSubdirectory(std::string_view name) override;
+ VirtualFile CreateFile(std::string_view name) override;
bool DeleteSubdirectory(std::string_view name) override;
bool DeleteFile(std::string_view name) override;
bool Rename(std::string_view name) override;
diff --git a/src/core/file_sys/vfs_offset.cpp b/src/core/file_sys/vfs_offset.cpp
index 7714d3de5..056737b54 100644
--- a/src/core/file_sys/vfs_offset.cpp
+++ b/src/core/file_sys/vfs_offset.cpp
@@ -9,7 +9,7 @@
namespace FileSys {
-OffsetVfsFile::OffsetVfsFile(std::shared_ptr<VfsFile> file_, std::size_t size_, std::size_t offset_,
+OffsetVfsFile::OffsetVfsFile(VirtualFile file_, std::size_t size_, std::size_t offset_,
std::string name_, VirtualDir parent_)
: file(file_), offset(offset_), size(size_), name(std::move(name_)),
parent(parent_ == nullptr ? file->GetContainingDirectory() : std::move(parent_)) {}
@@ -37,7 +37,7 @@ bool OffsetVfsFile::Resize(std::size_t new_size) {
return true;
}
-std::shared_ptr<VfsDirectory> OffsetVfsFile::GetContainingDirectory() const {
+VirtualDir OffsetVfsFile::GetContainingDirectory() const {
return parent;
}
diff --git a/src/core/file_sys/vfs_offset.h b/src/core/file_sys/vfs_offset.h
index f7b7a3256..b2ccc5c7b 100644
--- a/src/core/file_sys/vfs_offset.h
+++ b/src/core/file_sys/vfs_offset.h
@@ -17,14 +17,14 @@ namespace FileSys {
// the size of this wrapper.
class OffsetVfsFile : public VfsFile {
public:
- OffsetVfsFile(std::shared_ptr<VfsFile> file, std::size_t size, std::size_t offset = 0,
+ OffsetVfsFile(VirtualFile file, std::size_t size, std::size_t offset = 0,
std::string new_name = "", VirtualDir new_parent = nullptr);
~OffsetVfsFile() override;
std::string GetName() const override;
std::size_t GetSize() const override;
bool Resize(std::size_t new_size) override;
- std::shared_ptr<VfsDirectory> GetContainingDirectory() const override;
+ VirtualDir GetContainingDirectory() const override;
bool IsWritable() const override;
bool IsReadable() const override;
std::size_t Read(u8* data, std::size_t length, std::size_t offset) const override;
@@ -42,7 +42,7 @@ public:
private:
std::size_t TrimToFit(std::size_t r_size, std::size_t r_offset) const;
- std::shared_ptr<VfsFile> file;
+ VirtualFile file;
std::size_t offset;
std::size_t size;
std::string name;
diff --git a/src/core/file_sys/vfs_real.cpp b/src/core/file_sys/vfs_real.cpp
index 488687ba9..a287eebe3 100644
--- a/src/core/file_sys/vfs_real.cpp
+++ b/src/core/file_sys/vfs_real.cpp
@@ -263,7 +263,7 @@ bool RealVfsFile::Resize(std::size_t new_size) {
return backing->Resize(new_size);
}
-std::shared_ptr<VfsDirectory> RealVfsFile::GetContainingDirectory() const {
+VirtualDir RealVfsFile::GetContainingDirectory() const {
return base.OpenDirectory(parent_path, perms);
}
@@ -352,7 +352,7 @@ RealVfsDirectory::RealVfsDirectory(RealVfsFilesystem& base_, const std::string&
RealVfsDirectory::~RealVfsDirectory() = default;
-std::shared_ptr<VfsFile> RealVfsDirectory::GetFileRelative(std::string_view path) const {
+VirtualFile RealVfsDirectory::GetFileRelative(std::string_view path) const {
const auto full_path = FS::SanitizePath(this->path + DIR_SEP + std::string(path));
if (!FS::Exists(full_path) || FS::IsDirectory(full_path)) {
return nullptr;
@@ -360,7 +360,7 @@ std::shared_ptr<VfsFile> RealVfsDirectory::GetFileRelative(std::string_view path
return base.OpenFile(full_path, perms);
}
-std::shared_ptr<VfsDirectory> RealVfsDirectory::GetDirectoryRelative(std::string_view path) const {
+VirtualDir RealVfsDirectory::GetDirectoryRelative(std::string_view path) const {
const auto full_path = FS::SanitizePath(this->path + DIR_SEP + std::string(path));
if (!FS::Exists(full_path) || !FS::IsDirectory(full_path)) {
return nullptr;
@@ -368,20 +368,20 @@ std::shared_ptr<VfsDirectory> RealVfsDirectory::GetDirectoryRelative(std::string
return base.OpenDirectory(full_path, perms);
}
-std::shared_ptr<VfsFile> RealVfsDirectory::GetFile(std::string_view name) const {
+VirtualFile RealVfsDirectory::GetFile(std::string_view name) const {
return GetFileRelative(name);
}
-std::shared_ptr<VfsDirectory> RealVfsDirectory::GetSubdirectory(std::string_view name) const {
+VirtualDir RealVfsDirectory::GetSubdirectory(std::string_view name) const {
return GetDirectoryRelative(name);
}
-std::shared_ptr<VfsFile> RealVfsDirectory::CreateFileRelative(std::string_view path) {
+VirtualFile RealVfsDirectory::CreateFileRelative(std::string_view path) {
const auto full_path = FS::SanitizePath(this->path + DIR_SEP + std::string(path));
return base.CreateFile(full_path, perms);
}
-std::shared_ptr<VfsDirectory> RealVfsDirectory::CreateDirectoryRelative(std::string_view path) {
+VirtualDir RealVfsDirectory::CreateDirectoryRelative(std::string_view path) {
const auto full_path = FS::SanitizePath(this->path + DIR_SEP + std::string(path));
return base.CreateDirectory(full_path, perms);
}
@@ -391,11 +391,11 @@ bool RealVfsDirectory::DeleteSubdirectoryRecursive(std::string_view name) {
return base.DeleteDirectory(full_path);
}
-std::vector<std::shared_ptr<VfsFile>> RealVfsDirectory::GetFiles() const {
+std::vector<VirtualFile> RealVfsDirectory::GetFiles() const {
return IterateEntries<RealVfsFile, VfsFile>();
}
-std::vector<std::shared_ptr<VfsDirectory>> RealVfsDirectory::GetSubdirectories() const {
+std::vector<VirtualDir> RealVfsDirectory::GetSubdirectories() const {
return IterateEntries<RealVfsDirectory, VfsDirectory>();
}
@@ -411,7 +411,7 @@ std::string RealVfsDirectory::GetName() const {
return path_components.back();
}
-std::shared_ptr<VfsDirectory> RealVfsDirectory::GetParentDirectory() const {
+VirtualDir RealVfsDirectory::GetParentDirectory() const {
if (path_components.size() <= 1) {
return nullptr;
}
@@ -419,12 +419,12 @@ std::shared_ptr<VfsDirectory> RealVfsDirectory::GetParentDirectory() const {
return base.OpenDirectory(parent_path, perms);
}
-std::shared_ptr<VfsDirectory> RealVfsDirectory::CreateSubdirectory(std::string_view name) {
+VirtualDir RealVfsDirectory::CreateSubdirectory(std::string_view name) {
const std::string subdir_path = (path + DIR_SEP).append(name);
return base.CreateDirectory(subdir_path, perms);
}
-std::shared_ptr<VfsFile> RealVfsDirectory::CreateFile(std::string_view name) {
+VirtualFile RealVfsDirectory::CreateFile(std::string_view name) {
const std::string file_path = (path + DIR_SEP).append(name);
return base.CreateFile(file_path, perms);
}
diff --git a/src/core/file_sys/vfs_real.h b/src/core/file_sys/vfs_real.h
index 0b537b22c..23e99865e 100644
--- a/src/core/file_sys/vfs_real.h
+++ b/src/core/file_sys/vfs_real.h
@@ -50,7 +50,7 @@ public:
std::string GetName() const override;
std::size_t GetSize() const override;
bool Resize(std::size_t new_size) override;
- std::shared_ptr<VfsDirectory> GetContainingDirectory() const override;
+ VirtualDir GetContainingDirectory() const override;
bool IsWritable() const override;
bool IsReadable() const override;
std::size_t Read(u8* data, std::size_t length, std::size_t offset) const override;
@@ -79,21 +79,21 @@ class RealVfsDirectory : public VfsDirectory {
public:
~RealVfsDirectory() override;
- std::shared_ptr<VfsFile> GetFileRelative(std::string_view path) const override;
- std::shared_ptr<VfsDirectory> GetDirectoryRelative(std::string_view path) const override;
- std::shared_ptr<VfsFile> GetFile(std::string_view name) const override;
- std::shared_ptr<VfsDirectory> GetSubdirectory(std::string_view name) const override;
- std::shared_ptr<VfsFile> CreateFileRelative(std::string_view path) override;
- std::shared_ptr<VfsDirectory> CreateDirectoryRelative(std::string_view path) override;
+ VirtualFile GetFileRelative(std::string_view path) const override;
+ VirtualDir GetDirectoryRelative(std::string_view path) const override;
+ VirtualFile GetFile(std::string_view name) const override;
+ VirtualDir GetSubdirectory(std::string_view name) const override;
+ VirtualFile CreateFileRelative(std::string_view path) override;
+ VirtualDir CreateDirectoryRelative(std::string_view path) override;
bool DeleteSubdirectoryRecursive(std::string_view name) override;
- std::vector<std::shared_ptr<VfsFile>> GetFiles() const override;
- std::vector<std::shared_ptr<VfsDirectory>> GetSubdirectories() const override;
+ std::vector<VirtualFile> GetFiles() const override;
+ std::vector<VirtualDir> GetSubdirectories() const override;
bool IsWritable() const override;
bool IsReadable() const override;
std::string GetName() const override;
- std::shared_ptr<VfsDirectory> GetParentDirectory() const override;
- std::shared_ptr<VfsDirectory> CreateSubdirectory(std::string_view name) override;
- std::shared_ptr<VfsFile> CreateFile(std::string_view name) override;
+ VirtualDir GetParentDirectory() const override;
+ VirtualDir CreateSubdirectory(std::string_view name) override;
+ VirtualFile CreateFile(std::string_view name) override;
bool DeleteSubdirectory(std::string_view name) override;
bool DeleteFile(std::string_view name) override;
bool Rename(std::string_view name) override;
diff --git a/src/core/file_sys/vfs_static.h b/src/core/file_sys/vfs_static.h
index 8b27c30fa..c840b24b9 100644
--- a/src/core/file_sys/vfs_static.h
+++ b/src/core/file_sys/vfs_static.h
@@ -31,7 +31,7 @@ public:
return true;
}
- std::shared_ptr<VfsDirectory> GetContainingDirectory() const override {
+ VirtualDir GetContainingDirectory() const override {
return parent;
}
diff --git a/src/core/file_sys/vfs_vector.cpp b/src/core/file_sys/vfs_vector.cpp
index 75fc04302..c1ec1e645 100644
--- a/src/core/file_sys/vfs_vector.cpp
+++ b/src/core/file_sys/vfs_vector.cpp
@@ -25,7 +25,7 @@ bool VectorVfsFile::Resize(size_t new_size) {
return true;
}
-std::shared_ptr<VfsDirectory> VectorVfsFile::GetContainingDirectory() const {
+VirtualDir VectorVfsFile::GetContainingDirectory() const {
return parent;
}
@@ -68,11 +68,11 @@ VectorVfsDirectory::VectorVfsDirectory(std::vector<VirtualFile> files_,
VectorVfsDirectory::~VectorVfsDirectory() = default;
-std::vector<std::shared_ptr<VfsFile>> VectorVfsDirectory::GetFiles() const {
+std::vector<VirtualFile> VectorVfsDirectory::GetFiles() const {
return files;
}
-std::vector<std::shared_ptr<VfsDirectory>> VectorVfsDirectory::GetSubdirectories() const {
+std::vector<VirtualDir> VectorVfsDirectory::GetSubdirectories() const {
return dirs;
}
@@ -88,7 +88,7 @@ std::string VectorVfsDirectory::GetName() const {
return name;
}
-std::shared_ptr<VfsDirectory> VectorVfsDirectory::GetParentDirectory() const {
+VirtualDir VectorVfsDirectory::GetParentDirectory() const {
return parent;
}
@@ -116,11 +116,11 @@ bool VectorVfsDirectory::Rename(std::string_view name_) {
return true;
}
-std::shared_ptr<VfsDirectory> VectorVfsDirectory::CreateSubdirectory(std::string_view name) {
+VirtualDir VectorVfsDirectory::CreateSubdirectory(std::string_view name) {
return nullptr;
}
-std::shared_ptr<VfsFile> VectorVfsDirectory::CreateFile(std::string_view name) {
+VirtualFile VectorVfsDirectory::CreateFile(std::string_view name) {
return nullptr;
}
diff --git a/src/core/file_sys/vfs_vector.h b/src/core/file_sys/vfs_vector.h
index 95d3da2f2..2aff9ca34 100644
--- a/src/core/file_sys/vfs_vector.h
+++ b/src/core/file_sys/vfs_vector.h
@@ -17,9 +17,9 @@ namespace FileSys {
template <std::size_t size>
class ArrayVfsFile : public VfsFile {
public:
- explicit ArrayVfsFile(const std::array<u8, size>& data, std::string name = "",
- VirtualDir parent = nullptr)
- : data(data), name(std::move(name)), parent(std::move(parent)) {}
+ explicit ArrayVfsFile(const std::array<u8, size>& data_, std::string name_ = "",
+ VirtualDir parent_ = nullptr)
+ : data(data_), name(std::move(name_)), parent(std::move(parent_)) {}
std::string GetName() const override {
return name;
@@ -33,7 +33,7 @@ public:
return false;
}
- std::shared_ptr<VfsDirectory> GetContainingDirectory() const override {
+ VirtualDir GetContainingDirectory() const override {
return parent;
}
@@ -51,12 +51,12 @@ public:
return read;
}
- std::size_t Write(const u8* data, std::size_t length, std::size_t offset) override {
+ std::size_t Write(const u8* data_, std::size_t length, std::size_t offset) override {
return 0;
}
- bool Rename(std::string_view name) override {
- this->name = name;
+ bool Rename(std::string_view new_name) override {
+ name = new_name;
return true;
}
@@ -82,7 +82,7 @@ public:
std::string GetName() const override;
std::size_t GetSize() const override;
bool Resize(std::size_t new_size) override;
- std::shared_ptr<VfsDirectory> GetContainingDirectory() const override;
+ VirtualDir GetContainingDirectory() const override;
bool IsWritable() const override;
bool IsReadable() const override;
std::size_t Read(u8* data, std::size_t length, std::size_t offset) const override;
@@ -106,17 +106,17 @@ public:
VirtualDir parent = nullptr);
~VectorVfsDirectory() override;
- std::vector<std::shared_ptr<VfsFile>> GetFiles() const override;
- std::vector<std::shared_ptr<VfsDirectory>> GetSubdirectories() const override;
+ std::vector<VirtualFile> GetFiles() const override;
+ std::vector<VirtualDir> GetSubdirectories() const override;
bool IsWritable() const override;
bool IsReadable() const override;
std::string GetName() const override;
- std::shared_ptr<VfsDirectory> GetParentDirectory() const override;
+ VirtualDir GetParentDirectory() const override;
bool DeleteSubdirectory(std::string_view name) override;
bool DeleteFile(std::string_view name) override;
bool Rename(std::string_view name) override;
- std::shared_ptr<VfsDirectory> CreateSubdirectory(std::string_view name) override;
- std::shared_ptr<VfsFile> CreateFile(std::string_view name) override;
+ VirtualDir CreateSubdirectory(std::string_view name) override;
+ VirtualFile CreateFile(std::string_view name) override;
virtual void AddFile(VirtualFile file);
virtual void AddDirectory(VirtualDir dir);
diff --git a/src/core/file_sys/xts_archive.cpp b/src/core/file_sys/xts_archive.cpp
index 24c58e7ae..814fd5680 100644
--- a/src/core/file_sys/xts_archive.cpp
+++ b/src/core/file_sys/xts_archive.cpp
@@ -152,11 +152,11 @@ NAXContentType NAX::GetContentType() const {
return type;
}
-std::vector<std::shared_ptr<VfsFile>> NAX::GetFiles() const {
+std::vector<VirtualFile> NAX::GetFiles() const {
return {dec_file};
}
-std::vector<std::shared_ptr<VfsDirectory>> NAX::GetSubdirectories() const {
+std::vector<VirtualDir> NAX::GetSubdirectories() const {
return {};
}
@@ -164,7 +164,7 @@ std::string NAX::GetName() const {
return file->GetName();
}
-std::shared_ptr<VfsDirectory> NAX::GetParentDirectory() const {
+VirtualDir NAX::GetParentDirectory() const {
return file->GetContainingDirectory();
}
diff --git a/src/core/file_sys/xts_archive.h b/src/core/file_sys/xts_archive.h
index c472e226e..63a032b68 100644
--- a/src/core/file_sys/xts_archive.h
+++ b/src/core/file_sys/xts_archive.h
@@ -47,13 +47,13 @@ public:
NAXContentType GetContentType() const;
- std::vector<std::shared_ptr<VfsFile>> GetFiles() const override;
+ std::vector<VirtualFile> GetFiles() const override;
- std::vector<std::shared_ptr<VfsDirectory>> GetSubdirectories() const override;
+ std::vector<VirtualDir> GetSubdirectories() const override;
std::string GetName() const override;
- std::shared_ptr<VfsDirectory> GetParentDirectory() const override;
+ VirtualDir GetParentDirectory() const override;
private:
Loader::ResultStatus Parse(std::string_view path);
diff --git a/src/core/frontend/applets/error.cpp b/src/core/frontend/applets/error.cpp
index 4002a9211..dceb20ff8 100644
--- a/src/core/frontend/applets/error.cpp
+++ b/src/core/frontend/applets/error.cpp
@@ -2,6 +2,7 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
+#include "common/logging/log.h"
#include "core/frontend/applets/error.h"
namespace Core::Frontend {
@@ -10,7 +11,7 @@ ErrorApplet::~ErrorApplet() = default;
void DefaultErrorApplet::ShowError(ResultCode error, std::function<void()> finished) const {
LOG_CRITICAL(Service_Fatal, "Application requested error display: {:04}-{:04} (raw={:08X})",
- static_cast<u32>(error.module.Value()), error.description.Value(), error.raw);
+ error.module.Value(), error.description.Value(), error.raw);
}
void DefaultErrorApplet::ShowErrorWithTimestamp(ResultCode error, std::chrono::seconds time,
@@ -18,7 +19,7 @@ void DefaultErrorApplet::ShowErrorWithTimestamp(ResultCode error, std::chrono::s
LOG_CRITICAL(
Service_Fatal,
"Application requested error display: {:04X}-{:04X} (raw={:08X}) with timestamp={:016X}",
- static_cast<u32>(error.module.Value()), error.description.Value(), error.raw, time.count());
+ error.module.Value(), error.description.Value(), error.raw, time.count());
}
void DefaultErrorApplet::ShowCustomErrorText(ResultCode error, std::string main_text,
@@ -26,7 +27,7 @@ void DefaultErrorApplet::ShowCustomErrorText(ResultCode error, std::string main_
std::function<void()> finished) const {
LOG_CRITICAL(Service_Fatal,
"Application requested custom error with error_code={:04X}-{:04X} (raw={:08X})",
- static_cast<u32>(error.module.Value()), error.description.Value(), error.raw);
+ error.module.Value(), error.description.Value(), error.raw);
LOG_CRITICAL(Service_Fatal, " Main Text: {}", main_text);
LOG_CRITICAL(Service_Fatal, " Detail Text: {}", detail_text);
}
diff --git a/src/core/frontend/applets/general_frontend.cpp b/src/core/frontend/applets/general_frontend.cpp
index c30b36de7..7483ffb76 100644
--- a/src/core/frontend/applets/general_frontend.cpp
+++ b/src/core/frontend/applets/general_frontend.cpp
@@ -53,72 +53,4 @@ void DefaultPhotoViewerApplet::ShowAllPhotos(std::function<void()> finished) con
finished();
}
-ECommerceApplet::~ECommerceApplet() = default;
-
-DefaultECommerceApplet::~DefaultECommerceApplet() = default;
-
-void DefaultECommerceApplet::ShowApplicationInformation(
- std::function<void()> finished, u64 title_id, std::optional<u128> user_id,
- std::optional<bool> full_display, std::optional<std::string> extra_parameter) {
- const auto value = user_id.value_or(u128{});
- LOG_INFO(Service_AM,
- "Application requested frontend show application information for EShop, "
- "title_id={:016X}, user_id={:016X}{:016X}, full_display={}, extra_parameter={}",
- title_id, value[1], value[0],
- full_display.has_value() ? fmt::format("{}", *full_display) : "null",
- extra_parameter.value_or("null"));
- finished();
-}
-
-void DefaultECommerceApplet::ShowAddOnContentList(std::function<void()> finished, u64 title_id,
- std::optional<u128> user_id,
- std::optional<bool> full_display) {
- const auto value = user_id.value_or(u128{});
- LOG_INFO(Service_AM,
- "Application requested frontend show add on content list for EShop, "
- "title_id={:016X}, user_id={:016X}{:016X}, full_display={}",
- title_id, value[1], value[0],
- full_display.has_value() ? fmt::format("{}", *full_display) : "null");
- finished();
-}
-
-void DefaultECommerceApplet::ShowSubscriptionList(std::function<void()> finished, u64 title_id,
- std::optional<u128> user_id) {
- const auto value = user_id.value_or(u128{});
- LOG_INFO(Service_AM,
- "Application requested frontend show subscription list for EShop, title_id={:016X}, "
- "user_id={:016X}{:016X}",
- title_id, value[1], value[0]);
- finished();
-}
-
-void DefaultECommerceApplet::ShowConsumableItemList(std::function<void()> finished, u64 title_id,
- std::optional<u128> user_id) {
- const auto value = user_id.value_or(u128{});
- LOG_INFO(
- Service_AM,
- "Application requested frontend show consumable item list for EShop, title_id={:016X}, "
- "user_id={:016X}{:016X}",
- title_id, value[1], value[0]);
- finished();
-}
-
-void DefaultECommerceApplet::ShowShopHome(std::function<void()> finished, u128 user_id,
- bool full_display) {
- LOG_INFO(Service_AM,
- "Application requested frontend show home menu for EShop, user_id={:016X}{:016X}, "
- "full_display={}",
- user_id[1], user_id[0], full_display);
- finished();
-}
-
-void DefaultECommerceApplet::ShowSettings(std::function<void()> finished, u128 user_id,
- bool full_display) {
- LOG_INFO(Service_AM,
- "Application requested frontend show settings menu for EShop, user_id={:016X}{:016X}, "
- "full_display={}",
- user_id[1], user_id[0], full_display);
- finished();
-}
-
} // namespace Core::Frontend
diff --git a/src/core/frontend/applets/general_frontend.h b/src/core/frontend/applets/general_frontend.h
index 4b63f828e..b713b14ee 100644
--- a/src/core/frontend/applets/general_frontend.h
+++ b/src/core/frontend/applets/general_frontend.h
@@ -58,55 +58,4 @@ public:
void ShowAllPhotos(std::function<void()> finished) const override;
};
-class ECommerceApplet {
-public:
- virtual ~ECommerceApplet();
-
- // Shows a page with application icons, description, name, and price.
- virtual void ShowApplicationInformation(std::function<void()> finished, u64 title_id,
- std::optional<u128> user_id = {},
- std::optional<bool> full_display = {},
- std::optional<std::string> extra_parameter = {}) = 0;
-
- // Shows a page with all of the add on content available for a game, with name, description, and
- // price.
- virtual void ShowAddOnContentList(std::function<void()> finished, u64 title_id,
- std::optional<u128> user_id = {},
- std::optional<bool> full_display = {}) = 0;
-
- // Shows a page with all of the subscriptions (recurring payments) for a game, with name,
- // description, price, and renewal period.
- virtual void ShowSubscriptionList(std::function<void()> finished, u64 title_id,
- std::optional<u128> user_id = {}) = 0;
-
- // Shows a page with a list of any additional game related purchasable items (DLC,
- // subscriptions, etc) for a particular game, with name, description, type, and price.
- virtual void ShowConsumableItemList(std::function<void()> finished, u64 title_id,
- std::optional<u128> user_id = {}) = 0;
-
- // Shows the home page of the shop.
- virtual void ShowShopHome(std::function<void()> finished, u128 user_id, bool full_display) = 0;
-
- // Shows the user settings page of the shop.
- virtual void ShowSettings(std::function<void()> finished, u128 user_id, bool full_display) = 0;
-};
-
-class DefaultECommerceApplet : public ECommerceApplet {
-public:
- ~DefaultECommerceApplet() override;
-
- void ShowApplicationInformation(std::function<void()> finished, u64 title_id,
- std::optional<u128> user_id, std::optional<bool> full_display,
- std::optional<std::string> extra_parameter) override;
- void ShowAddOnContentList(std::function<void()> finished, u64 title_id,
- std::optional<u128> user_id,
- std::optional<bool> full_display) override;
- void ShowSubscriptionList(std::function<void()> finished, u64 title_id,
- std::optional<u128> user_id) override;
- void ShowConsumableItemList(std::function<void()> finished, u64 title_id,
- std::optional<u128> user_id) override;
- void ShowShopHome(std::function<void()> finished, u128 user_id, bool full_display) override;
- void ShowSettings(std::function<void()> finished, u128 user_id, bool full_display) override;
-};
-
} // namespace Core::Frontend
diff --git a/src/core/frontend/applets/web_browser.cpp b/src/core/frontend/applets/web_browser.cpp
index 528295ffc..50db6a654 100644
--- a/src/core/frontend/applets/web_browser.cpp
+++ b/src/core/frontend/applets/web_browser.cpp
@@ -11,14 +11,22 @@ WebBrowserApplet::~WebBrowserApplet() = default;
DefaultWebBrowserApplet::~DefaultWebBrowserApplet() = default;
-void DefaultWebBrowserApplet::OpenPageLocal(std::string_view filename,
- std::function<void()> unpack_romfs_callback,
- std::function<void()> finished_callback) {
- LOG_INFO(Service_AM,
- "(STUBBED) called - No suitable web browser implementation found to open website page "
- "at '{}'!",
- filename);
- finished_callback();
+void DefaultWebBrowserApplet::OpenLocalWebPage(
+ std::string_view local_url, std::function<void()> extract_romfs_callback,
+ std::function<void(Service::AM::Applets::WebExitReason, std::string)> callback) const {
+ LOG_WARNING(Service_AM, "(STUBBED) called, backend requested to open local web page at {}",
+ local_url);
+
+ callback(Service::AM::Applets::WebExitReason::WindowClosed, "http://localhost/");
+}
+
+void DefaultWebBrowserApplet::OpenExternalWebPage(
+ std::string_view external_url,
+ std::function<void(Service::AM::Applets::WebExitReason, std::string)> callback) const {
+ LOG_WARNING(Service_AM, "(STUBBED) called, backend requested to open external web page at {}",
+ external_url);
+
+ callback(Service::AM::Applets::WebExitReason::WindowClosed, "http://localhost/");
}
} // namespace Core::Frontend
diff --git a/src/core/frontend/applets/web_browser.h b/src/core/frontend/applets/web_browser.h
index 110e33bc4..1c5ef19a9 100644
--- a/src/core/frontend/applets/web_browser.h
+++ b/src/core/frontend/applets/web_browser.h
@@ -7,22 +7,34 @@
#include <functional>
#include <string_view>
+#include "core/hle/service/am/applets/web_types.h"
+
namespace Core::Frontend {
class WebBrowserApplet {
public:
virtual ~WebBrowserApplet();
- virtual void OpenPageLocal(std::string_view url, std::function<void()> unpack_romfs_callback,
- std::function<void()> finished_callback) = 0;
+ virtual void OpenLocalWebPage(
+ std::string_view local_url, std::function<void()> extract_romfs_callback,
+ std::function<void(Service::AM::Applets::WebExitReason, std::string)> callback) const = 0;
+
+ virtual void OpenExternalWebPage(
+ std::string_view external_url,
+ std::function<void(Service::AM::Applets::WebExitReason, std::string)> callback) const = 0;
};
class DefaultWebBrowserApplet final : public WebBrowserApplet {
public:
~DefaultWebBrowserApplet() override;
- void OpenPageLocal(std::string_view url, std::function<void()> unpack_romfs_callback,
- std::function<void()> finished_callback) override;
+ void OpenLocalWebPage(std::string_view local_url, std::function<void()> extract_romfs_callback,
+ std::function<void(Service::AM::Applets::WebExitReason, std::string)>
+ callback) const override;
+
+ void OpenExternalWebPage(std::string_view external_url,
+ std::function<void(Service::AM::Applets::WebExitReason, std::string)>
+ callback) const override;
};
} // namespace Core::Frontend
diff --git a/src/core/frontend/input_interpreter.cpp b/src/core/frontend/input_interpreter.cpp
new file mode 100644
index 000000000..66ae506cd
--- /dev/null
+++ b/src/core/frontend/input_interpreter.cpp
@@ -0,0 +1,45 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/core.h"
+#include "core/frontend/input_interpreter.h"
+#include "core/hle/service/hid/controllers/npad.h"
+#include "core/hle/service/hid/hid.h"
+#include "core/hle/service/sm/sm.h"
+
+InputInterpreter::InputInterpreter(Core::System& system)
+ : npad{system.ServiceManager()
+ .GetService<Service::HID::Hid>("hid")
+ ->GetAppletResource()
+ ->GetController<Service::HID::Controller_NPad>(Service::HID::HidController::NPad)} {}
+
+InputInterpreter::~InputInterpreter() = default;
+
+void InputInterpreter::PollInput() {
+ const u32 button_state = npad.GetAndResetPressState();
+
+ previous_index = current_index;
+ current_index = (current_index + 1) % button_states.size();
+
+ button_states[current_index] = button_state;
+}
+
+bool InputInterpreter::IsButtonPressedOnce(HIDButton button) const {
+ const bool current_press =
+ (button_states[current_index] & (1U << static_cast<u8>(button))) != 0;
+ const bool previous_press =
+ (button_states[previous_index] & (1U << static_cast<u8>(button))) != 0;
+
+ return current_press && !previous_press;
+}
+
+bool InputInterpreter::IsButtonHeld(HIDButton button) const {
+ u32 held_buttons{button_states[0]};
+
+ for (std::size_t i = 1; i < button_states.size(); ++i) {
+ held_buttons &= button_states[i];
+ }
+
+ return (held_buttons & (1U << static_cast<u8>(button))) != 0;
+}
diff --git a/src/core/frontend/input_interpreter.h b/src/core/frontend/input_interpreter.h
new file mode 100644
index 000000000..fea9aebe6
--- /dev/null
+++ b/src/core/frontend/input_interpreter.h
@@ -0,0 +1,120 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+
+#include "common/common_types.h"
+
+namespace Core {
+class System;
+}
+
+namespace Service::HID {
+class Controller_NPad;
+}
+
+enum class HIDButton : u8 {
+ A,
+ B,
+ X,
+ Y,
+ LStick,
+ RStick,
+ L,
+ R,
+ ZL,
+ ZR,
+ Plus,
+ Minus,
+
+ DLeft,
+ DUp,
+ DRight,
+ DDown,
+
+ LStickLeft,
+ LStickUp,
+ LStickRight,
+ LStickDown,
+
+ RStickLeft,
+ RStickUp,
+ RStickRight,
+ RStickDown,
+
+ LeftSL,
+ LeftSR,
+
+ RightSL,
+ RightSR,
+};
+
+/**
+ * The InputInterpreter class interfaces with HID to retrieve button press states.
+ * Input is intended to be polled every 50ms so that a button is considered to be
+ * held down after 400ms has elapsed since the initial button press and subsequent
+ * repeated presses occur every 50ms.
+ */
+class InputInterpreter {
+public:
+ explicit InputInterpreter(Core::System& system);
+ virtual ~InputInterpreter();
+
+ /// Gets a button state from HID and inserts it into the array of button states.
+ void PollInput();
+
+ /**
+ * The specified button is considered to be pressed once
+ * if it is currently pressed and not pressed previously.
+ *
+ * @param button The button to check.
+ *
+ * @returns True when the button is pressed once.
+ */
+ [[nodiscard]] bool IsButtonPressedOnce(HIDButton button) const;
+
+ /**
+ * Checks whether any of the buttons in the parameter list is pressed once.
+ *
+ * @tparam HIDButton The buttons to check.
+ *
+ * @returns True when at least one of the buttons is pressed once.
+ */
+ template <HIDButton... T>
+ [[nodiscard]] bool IsAnyButtonPressedOnce() {
+ return (IsButtonPressedOnce(T) || ...);
+ }
+
+ /**
+ * The specified button is considered to be held down if it is pressed in all 9 button states.
+ *
+ * @param button The button to check.
+ *
+ * @returns True when the button is held down.
+ */
+ [[nodiscard]] bool IsButtonHeld(HIDButton button) const;
+
+ /**
+ * Checks whether any of the buttons in the parameter list is held down.
+ *
+ * @tparam HIDButton The buttons to check.
+ *
+ * @returns True when at least one of the buttons is held down.
+ */
+ template <HIDButton... T>
+ [[nodiscard]] bool IsAnyButtonHeld() {
+ return (IsButtonHeld(T) || ...);
+ }
+
+private:
+ Service::HID::Controller_NPad& npad;
+
+ /// Stores 9 consecutive button states polled from HID.
+ std::array<u32, 9> button_states{};
+
+ std::size_t previous_index{};
+ std::size_t current_index{};
+};
diff --git a/src/core/hle/ipc_helpers.h b/src/core/hle/ipc_helpers.h
index d57776ce9..56cc911d1 100644
--- a/src/core/hle/ipc_helpers.h
+++ b/src/core/hle/ipc_helpers.h
@@ -166,8 +166,23 @@ public:
ValidateHeader();
}
+ void PushImpl(s8 value);
+ void PushImpl(s16 value);
+ void PushImpl(s32 value);
+ void PushImpl(s64 value);
+ void PushImpl(u8 value);
+ void PushImpl(u16 value);
+ void PushImpl(u32 value);
+ void PushImpl(u64 value);
+ void PushImpl(float value);
+ void PushImpl(double value);
+ void PushImpl(bool value);
+ void PushImpl(ResultCode value);
+
template <typename T>
- void Push(T value);
+ void Push(T value) {
+ return PushImpl(value);
+ }
template <typename First, typename... Other>
void Push(const First& first_value, const Other&... other_values);
@@ -215,13 +230,11 @@ private:
/// Push ///
-template <>
-inline void ResponseBuilder::Push(s32 value) {
+inline void ResponseBuilder::PushImpl(s32 value) {
cmdbuf[index++] = static_cast<u32>(value);
}
-template <>
-inline void ResponseBuilder::Push(u32 value) {
+inline void ResponseBuilder::PushImpl(u32 value) {
cmdbuf[index++] = value;
}
@@ -233,62 +246,52 @@ void ResponseBuilder::PushRaw(const T& value) {
index += (sizeof(T) + 3) / 4; // round up to word length
}
-template <>
-inline void ResponseBuilder::Push(ResultCode value) {
+inline void ResponseBuilder::PushImpl(ResultCode value) {
// Result codes are actually 64-bit in the IPC buffer, but only the high part is discarded.
Push(value.raw);
Push<u32>(0);
}
-template <>
-inline void ResponseBuilder::Push(s8 value) {
+inline void ResponseBuilder::PushImpl(s8 value) {
PushRaw(value);
}
-template <>
-inline void ResponseBuilder::Push(s16 value) {
+inline void ResponseBuilder::PushImpl(s16 value) {
PushRaw(value);
}
-template <>
-inline void ResponseBuilder::Push(s64 value) {
- Push(static_cast<u32>(value));
- Push(static_cast<u32>(value >> 32));
+inline void ResponseBuilder::PushImpl(s64 value) {
+ PushImpl(static_cast<u32>(value));
+ PushImpl(static_cast<u32>(value >> 32));
}
-template <>
-inline void ResponseBuilder::Push(u8 value) {
+inline void ResponseBuilder::PushImpl(u8 value) {
PushRaw(value);
}
-template <>
-inline void ResponseBuilder::Push(u16 value) {
+inline void ResponseBuilder::PushImpl(u16 value) {
PushRaw(value);
}
-template <>
-inline void ResponseBuilder::Push(u64 value) {
- Push(static_cast<u32>(value));
- Push(static_cast<u32>(value >> 32));
+inline void ResponseBuilder::PushImpl(u64 value) {
+ PushImpl(static_cast<u32>(value));
+ PushImpl(static_cast<u32>(value >> 32));
}
-template <>
-inline void ResponseBuilder::Push(float value) {
+inline void ResponseBuilder::PushImpl(float value) {
u32 integral;
std::memcpy(&integral, &value, sizeof(u32));
- Push(integral);
+ PushImpl(integral);
}
-template <>
-inline void ResponseBuilder::Push(double value) {
+inline void ResponseBuilder::PushImpl(double value) {
u64 integral;
std::memcpy(&integral, &value, sizeof(u64));
- Push(integral);
+ PushImpl(integral);
}
-template <>
-inline void ResponseBuilder::Push(bool value) {
- Push(static_cast<u8>(value));
+inline void ResponseBuilder::PushImpl(bool value) {
+ PushImpl(static_cast<u8>(value));
}
template <typename First, typename... Other>
diff --git a/src/core/hle/kernel/address_arbiter.cpp b/src/core/hle/kernel/address_arbiter.cpp
index 048acd30e..20ffa7d47 100644
--- a/src/core/hle/kernel/address_arbiter.cpp
+++ b/src/core/hle/kernel/address_arbiter.cpp
@@ -12,8 +12,9 @@
#include "core/hle/kernel/address_arbiter.h"
#include "core/hle/kernel/errors.h"
#include "core/hle/kernel/handle_table.h"
+#include "core/hle/kernel/k_scheduler.h"
+#include "core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h"
#include "core/hle/kernel/kernel.h"
-#include "core/hle/kernel/scheduler.h"
#include "core/hle/kernel/thread.h"
#include "core/hle/kernel/time_manager.h"
#include "core/hle/result.h"
@@ -58,7 +59,7 @@ ResultCode AddressArbiter::SignalToAddress(VAddr address, SignalType type, s32 v
}
ResultCode AddressArbiter::SignalToAddressOnly(VAddr address, s32 num_to_wake) {
- SchedulerLock lock(system.Kernel());
+ KScopedSchedulerLock lock(system.Kernel());
const std::vector<std::shared_ptr<Thread>> waiting_threads =
GetThreadsWaitingOnAddress(address);
WakeThreads(waiting_threads, num_to_wake);
@@ -67,7 +68,7 @@ ResultCode AddressArbiter::SignalToAddressOnly(VAddr address, s32 num_to_wake) {
ResultCode AddressArbiter::IncrementAndSignalToAddressIfEqual(VAddr address, s32 value,
s32 num_to_wake) {
- SchedulerLock lock(system.Kernel());
+ KScopedSchedulerLock lock(system.Kernel());
auto& memory = system.Memory();
// Ensure that we can write to the address.
@@ -92,7 +93,7 @@ ResultCode AddressArbiter::IncrementAndSignalToAddressIfEqual(VAddr address, s32
ResultCode AddressArbiter::ModifyByWaitingCountAndSignalToAddressIfEqual(VAddr address, s32 value,
s32 num_to_wake) {
- SchedulerLock lock(system.Kernel());
+ KScopedSchedulerLock lock(system.Kernel());
auto& memory = system.Memory();
// Ensure that we can write to the address.
@@ -153,11 +154,11 @@ ResultCode AddressArbiter::WaitForAddressIfLessThan(VAddr address, s32 value, s6
bool should_decrement) {
auto& memory = system.Memory();
auto& kernel = system.Kernel();
- Thread* current_thread = system.CurrentScheduler().GetCurrentThread();
+ Thread* current_thread = kernel.CurrentScheduler()->GetCurrentThread();
Handle event_handle = InvalidHandle;
{
- SchedulerLockAndSleep lock(kernel, event_handle, current_thread, timeout);
+ KScopedSchedulerLockAndSleep lock(kernel, event_handle, current_thread, timeout);
if (current_thread->IsPendingTermination()) {
lock.CancelSleep();
@@ -210,7 +211,7 @@ ResultCode AddressArbiter::WaitForAddressIfLessThan(VAddr address, s32 value, s6
}
{
- SchedulerLock lock(kernel);
+ KScopedSchedulerLock lock(kernel);
if (current_thread->IsWaitingForArbitration()) {
RemoveThread(SharedFrom(current_thread));
current_thread->WaitForArbitration(false);
@@ -223,11 +224,11 @@ ResultCode AddressArbiter::WaitForAddressIfLessThan(VAddr address, s32 value, s6
ResultCode AddressArbiter::WaitForAddressIfEqual(VAddr address, s32 value, s64 timeout) {
auto& memory = system.Memory();
auto& kernel = system.Kernel();
- Thread* current_thread = system.CurrentScheduler().GetCurrentThread();
+ Thread* current_thread = kernel.CurrentScheduler()->GetCurrentThread();
Handle event_handle = InvalidHandle;
{
- SchedulerLockAndSleep lock(kernel, event_handle, current_thread, timeout);
+ KScopedSchedulerLockAndSleep lock(kernel, event_handle, current_thread, timeout);
if (current_thread->IsPendingTermination()) {
lock.CancelSleep();
@@ -265,7 +266,7 @@ ResultCode AddressArbiter::WaitForAddressIfEqual(VAddr address, s32 value, s64 t
}
{
- SchedulerLock lock(kernel);
+ KScopedSchedulerLock lock(kernel);
if (current_thread->IsWaitingForArbitration()) {
RemoveThread(SharedFrom(current_thread));
current_thread->WaitForArbitration(false);
diff --git a/src/core/hle/kernel/global_scheduler_context.cpp b/src/core/hle/kernel/global_scheduler_context.cpp
new file mode 100644
index 000000000..a133e8ed0
--- /dev/null
+++ b/src/core/hle/kernel/global_scheduler_context.cpp
@@ -0,0 +1,52 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <mutex>
+
+#include "common/assert.h"
+#include "core/core.h"
+#include "core/hle/kernel/global_scheduler_context.h"
+#include "core/hle/kernel/k_scheduler.h"
+#include "core/hle/kernel/kernel.h"
+
+namespace Kernel {
+
+GlobalSchedulerContext::GlobalSchedulerContext(KernelCore& kernel)
+ : kernel{kernel}, scheduler_lock{kernel} {}
+
+GlobalSchedulerContext::~GlobalSchedulerContext() = default;
+
+void GlobalSchedulerContext::AddThread(std::shared_ptr<Thread> thread) {
+ std::scoped_lock lock{global_list_guard};
+ thread_list.push_back(std::move(thread));
+}
+
+void GlobalSchedulerContext::RemoveThread(std::shared_ptr<Thread> thread) {
+ std::scoped_lock lock{global_list_guard};
+ thread_list.erase(std::remove(thread_list.begin(), thread_list.end(), thread),
+ thread_list.end());
+}
+
+void GlobalSchedulerContext::PreemptThreads() {
+ // The priority levels at which the global scheduler preempts threads every 10 ms. They are
+ // ordered from Core 0 to Core 3.
+ static constexpr std::array<u32, Core::Hardware::NUM_CPU_CORES> preemption_priorities{
+ 59,
+ 59,
+ 59,
+ 63,
+ };
+
+ ASSERT(IsLocked());
+ for (u32 core_id = 0; core_id < Core::Hardware::NUM_CPU_CORES; core_id++) {
+ const u32 priority = preemption_priorities[core_id];
+ kernel.Scheduler(core_id).RotateScheduledQueue(core_id, priority);
+ }
+}
+
+bool GlobalSchedulerContext::IsLocked() const {
+ return scheduler_lock.IsLockedByCurrentThread();
+}
+
+} // namespace Kernel
diff --git a/src/core/hle/kernel/global_scheduler_context.h b/src/core/hle/kernel/global_scheduler_context.h
new file mode 100644
index 000000000..5c7b89290
--- /dev/null
+++ b/src/core/hle/kernel/global_scheduler_context.h
@@ -0,0 +1,81 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <atomic>
+#include <vector>
+
+#include "common/common_types.h"
+#include "common/spin_lock.h"
+#include "core/hardware_properties.h"
+#include "core/hle/kernel/k_priority_queue.h"
+#include "core/hle/kernel/k_scheduler_lock.h"
+#include "core/hle/kernel/thread.h"
+
+namespace Kernel {
+
+class KernelCore;
+class SchedulerLock;
+
+using KSchedulerPriorityQueue =
+ KPriorityQueue<Thread, Core::Hardware::NUM_CPU_CORES, THREADPRIO_LOWEST, THREADPRIO_HIGHEST>;
+constexpr s32 HighestCoreMigrationAllowedPriority = 2;
+
+class GlobalSchedulerContext final {
+ friend class KScheduler;
+
+public:
+ using LockType = KAbstractSchedulerLock<KScheduler>;
+
+ explicit GlobalSchedulerContext(KernelCore& kernel);
+ ~GlobalSchedulerContext();
+
+ /// Adds a new thread to the scheduler
+ void AddThread(std::shared_ptr<Thread> thread);
+
+ /// Removes a thread from the scheduler
+ void RemoveThread(std::shared_ptr<Thread> thread);
+
+ /// Returns a list of all threads managed by the scheduler
+ [[nodiscard]] const std::vector<std::shared_ptr<Thread>>& GetThreadList() const {
+ return thread_list;
+ }
+
+ /**
+ * Rotates the scheduling queues of threads at a preemption priority and then does
+ * some core rebalancing. Preemption priorities can be found in the array
+ * 'preemption_priorities'.
+ *
+ * @note This operation happens every 10ms.
+ */
+ void PreemptThreads();
+
+ /// Returns true if the global scheduler lock is acquired
+ bool IsLocked() const;
+
+ [[nodiscard]] LockType& SchedulerLock() {
+ return scheduler_lock;
+ }
+
+ [[nodiscard]] const LockType& SchedulerLock() const {
+ return scheduler_lock;
+ }
+
+private:
+ friend class KScopedSchedulerLock;
+ friend class KScopedSchedulerLockAndSleep;
+
+ KernelCore& kernel;
+
+ std::atomic_bool scheduler_update_needed{};
+ KSchedulerPriorityQueue priority_queue;
+ LockType scheduler_lock;
+
+ /// Lists all thread ids that aren't deleted/etc.
+ std::vector<std::shared_ptr<Thread>> thread_list;
+ Common::SpinLock global_list_guard{};
+};
+
+} // namespace Kernel
diff --git a/src/core/hle/kernel/handle_table.cpp b/src/core/hle/kernel/handle_table.cpp
index 3e745c18b..40988b0fd 100644
--- a/src/core/hle/kernel/handle_table.cpp
+++ b/src/core/hle/kernel/handle_table.cpp
@@ -8,9 +8,9 @@
#include "core/core.h"
#include "core/hle/kernel/errors.h"
#include "core/hle/kernel/handle_table.h"
+#include "core/hle/kernel/k_scheduler.h"
#include "core/hle/kernel/kernel.h"
#include "core/hle/kernel/process.h"
-#include "core/hle/kernel/scheduler.h"
#include "core/hle/kernel/thread.h"
namespace Kernel {
@@ -105,7 +105,7 @@ bool HandleTable::IsValid(Handle handle) const {
std::shared_ptr<Object> HandleTable::GetGeneric(Handle handle) const {
if (handle == CurrentThread) {
- return SharedFrom(kernel.CurrentScheduler().GetCurrentThread());
+ return SharedFrom(kernel.CurrentScheduler()->GetCurrentThread());
} else if (handle == CurrentProcess) {
return SharedFrom(kernel.CurrentProcess());
}
diff --git a/src/core/hle/kernel/hle_ipc.cpp b/src/core/hle/kernel/hle_ipc.cpp
index 81f85643b..83decf6cf 100644
--- a/src/core/hle/kernel/hle_ipc.cpp
+++ b/src/core/hle/kernel/hle_ipc.cpp
@@ -17,11 +17,12 @@
#include "core/hle/kernel/errors.h"
#include "core/hle/kernel/handle_table.h"
#include "core/hle/kernel/hle_ipc.h"
+#include "core/hle/kernel/k_scheduler.h"
+#include "core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h"
#include "core/hle/kernel/kernel.h"
#include "core/hle/kernel/object.h"
#include "core/hle/kernel/process.h"
#include "core/hle/kernel/readable_event.h"
-#include "core/hle/kernel/scheduler.h"
#include "core/hle/kernel/server_session.h"
#include "core/hle/kernel/thread.h"
#include "core/hle/kernel/time_manager.h"
@@ -45,44 +46,6 @@ void SessionRequestHandler::ClientDisconnected(
boost::range::remove_erase(connected_sessions, server_session);
}
-std::shared_ptr<WritableEvent> HLERequestContext::SleepClientThread(
- const std::string& reason, u64 timeout, WakeupCallback&& callback,
- std::shared_ptr<WritableEvent> writable_event) {
- // Put the client thread to sleep until the wait event is signaled or the timeout expires.
-
- if (!writable_event) {
- // Create event if not provided
- const auto pair = WritableEvent::CreateEventPair(kernel, "HLE Pause Event: " + reason);
- writable_event = pair.writable;
- }
-
- {
- Handle event_handle = InvalidHandle;
- SchedulerLockAndSleep lock(kernel, event_handle, thread.get(), timeout);
- thread->SetHLECallback(
- [context = *this, callback](std::shared_ptr<Thread> thread) mutable -> bool {
- ThreadWakeupReason reason = thread->GetSignalingResult() == RESULT_TIMEOUT
- ? ThreadWakeupReason::Timeout
- : ThreadWakeupReason::Signal;
- callback(thread, context, reason);
- context.WriteToOutgoingCommandBuffer(*thread);
- return true;
- });
- const auto readable_event{writable_event->GetReadableEvent()};
- writable_event->Clear();
- thread->SetHLESyncObject(readable_event.get());
- thread->SetStatus(ThreadStatus::WaitHLEEvent);
- thread->SetSynchronizationResults(nullptr, RESULT_TIMEOUT);
- readable_event->AddWaitingThread(thread);
- lock.Release();
- thread->SetHLETimeEvent(event_handle);
- }
-
- is_thread_waiting = true;
-
- return writable_event;
-}
-
HLERequestContext::HLERequestContext(KernelCore& kernel, Core::Memory::Memory& memory,
std::shared_ptr<ServerSession> server_session,
std::shared_ptr<Thread> thread)
diff --git a/src/core/hle/kernel/hle_ipc.h b/src/core/hle/kernel/hle_ipc.h
index c31a65476..b112e1ebd 100644
--- a/src/core/hle/kernel/hle_ipc.h
+++ b/src/core/hle/kernel/hle_ipc.h
@@ -129,23 +129,6 @@ public:
using WakeupCallback = std::function<void(
std::shared_ptr<Thread> thread, HLERequestContext& context, ThreadWakeupReason reason)>;
- /**
- * Puts the specified guest thread to sleep until the returned event is signaled or until the
- * specified timeout expires.
- * @param reason Reason for pausing the thread, to be used for debugging purposes.
- * @param timeout Timeout in nanoseconds after which the thread will be awoken and the callback
- * invoked with a Timeout reason.
- * @param callback Callback to be invoked when the thread is resumed. This callback must write
- * the entire command response once again, regardless of the state of it before this function
- * was called.
- * @param writable_event Event to use to wake up the thread. If unspecified, an event will be
- * created.
- * @returns Event that when signaled will resume the thread and call the callback function.
- */
- std::shared_ptr<WritableEvent> SleepClientThread(
- const std::string& reason, u64 timeout, WakeupCallback&& callback,
- std::shared_ptr<WritableEvent> writable_event = nullptr);
-
/// Populates this context with data from the requesting process/thread.
ResultCode PopulateFromIncomingCommandBuffer(const HandleTable& handle_table,
u32_le* src_cmdbuf);
diff --git a/src/core/hle/kernel/k_affinity_mask.h b/src/core/hle/kernel/k_affinity_mask.h
new file mode 100644
index 000000000..dd73781cd
--- /dev/null
+++ b/src/core/hle/kernel/k_affinity_mask.h
@@ -0,0 +1,58 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+// This file references various implementation details from Atmosphere, an open-source firmware for
+// the Nintendo Switch. Copyright 2018-2020 Atmosphere-NX.
+
+#pragma once
+
+#include "common/assert.h"
+#include "common/common_types.h"
+#include "core/hardware_properties.h"
+
+namespace Kernel {
+
+class KAffinityMask {
+public:
+ constexpr KAffinityMask() = default;
+
+ [[nodiscard]] constexpr u64 GetAffinityMask() const {
+ return this->mask;
+ }
+
+ constexpr void SetAffinityMask(u64 new_mask) {
+ ASSERT((new_mask & ~AllowedAffinityMask) == 0);
+ this->mask = new_mask;
+ }
+
+ [[nodiscard]] constexpr bool GetAffinity(s32 core) const {
+ return this->mask & GetCoreBit(core);
+ }
+
+ constexpr void SetAffinity(s32 core, bool set) {
+ ASSERT(0 <= core && core < static_cast<s32>(Core::Hardware::NUM_CPU_CORES));
+
+ if (set) {
+ this->mask |= GetCoreBit(core);
+ } else {
+ this->mask &= ~GetCoreBit(core);
+ }
+ }
+
+ constexpr void SetAll() {
+ this->mask = AllowedAffinityMask;
+ }
+
+private:
+ [[nodiscard]] static constexpr u64 GetCoreBit(s32 core) {
+ ASSERT(0 <= core && core < static_cast<s32>(Core::Hardware::NUM_CPU_CORES));
+ return (1ULL << core);
+ }
+
+ static constexpr u64 AllowedAffinityMask = (1ULL << Core::Hardware::NUM_CPU_CORES) - 1;
+
+ u64 mask{};
+};
+
+} // namespace Kernel
diff --git a/src/core/hle/kernel/k_priority_queue.h b/src/core/hle/kernel/k_priority_queue.h
new file mode 100644
index 000000000..99fb8fe93
--- /dev/null
+++ b/src/core/hle/kernel/k_priority_queue.h
@@ -0,0 +1,451 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+// This file references various implementation details from Atmosphere, an open-source firmware for
+// the Nintendo Switch. Copyright 2018-2020 Atmosphere-NX.
+
+#pragma once
+
+#include <array>
+#include <concepts>
+
+#include "common/assert.h"
+#include "common/bit_set.h"
+#include "common/bit_util.h"
+#include "common/common_types.h"
+#include "common/concepts.h"
+
+namespace Kernel {
+
+class Thread;
+
+template <typename T>
+concept KPriorityQueueAffinityMask = !std::is_reference_v<T> && requires(T & t) {
+ { t.GetAffinityMask() }
+ ->Common::ConvertibleTo<u64>;
+ {t.SetAffinityMask(std::declval<u64>())};
+
+ { t.GetAffinity(std::declval<int32_t>()) }
+ ->std::same_as<bool>;
+ {t.SetAffinity(std::declval<int32_t>(), std::declval<bool>())};
+ {t.SetAll()};
+};
+
+template <typename T>
+concept KPriorityQueueMember = !std::is_reference_v<T> && requires(T & t) {
+ {typename T::QueueEntry()};
+ {(typename T::QueueEntry()).Initialize()};
+ {(typename T::QueueEntry()).SetPrev(std::addressof(t))};
+ {(typename T::QueueEntry()).SetNext(std::addressof(t))};
+ { (typename T::QueueEntry()).GetNext() }
+ ->std::same_as<T*>;
+ { (typename T::QueueEntry()).GetPrev() }
+ ->std::same_as<T*>;
+ { t.GetPriorityQueueEntry(std::declval<s32>()) }
+ ->std::same_as<typename T::QueueEntry&>;
+
+ {t.GetAffinityMask()};
+ { typename std::remove_cvref<decltype(t.GetAffinityMask())>::type() }
+ ->KPriorityQueueAffinityMask;
+
+ { t.GetActiveCore() }
+ ->Common::ConvertibleTo<s32>;
+ { t.GetPriority() }
+ ->Common::ConvertibleTo<s32>;
+};
+
+template <typename Member, size_t _NumCores, int LowestPriority, int HighestPriority>
+requires KPriorityQueueMember<Member> class KPriorityQueue {
+public:
+ using AffinityMaskType = typename std::remove_cv_t<
+ typename std::remove_reference<decltype(std::declval<Member>().GetAffinityMask())>::type>;
+
+ static_assert(LowestPriority >= 0);
+ static_assert(HighestPriority >= 0);
+ static_assert(LowestPriority >= HighestPriority);
+ static constexpr size_t NumPriority = LowestPriority - HighestPriority + 1;
+ static constexpr size_t NumCores = _NumCores;
+
+ static constexpr bool IsValidCore(s32 core) {
+ return 0 <= core && core < static_cast<s32>(NumCores);
+ }
+
+ static constexpr bool IsValidPriority(s32 priority) {
+ return HighestPriority <= priority && priority <= LowestPriority + 1;
+ }
+
+private:
+ using Entry = typename Member::QueueEntry;
+
+public:
+ class KPerCoreQueue {
+ private:
+ std::array<Entry, NumCores> root{};
+
+ public:
+ constexpr KPerCoreQueue() {
+ for (auto& per_core_root : root) {
+ per_core_root.Initialize();
+ }
+ }
+
+ constexpr bool PushBack(s32 core, Member* member) {
+ // Get the entry associated with the member.
+ Entry& member_entry = member->GetPriorityQueueEntry(core);
+
+ // Get the entry associated with the end of the queue.
+ Member* tail = this->root[core].GetPrev();
+ Entry& tail_entry =
+ (tail != nullptr) ? tail->GetPriorityQueueEntry(core) : this->root[core];
+
+ // Link the entries.
+ member_entry.SetPrev(tail);
+ member_entry.SetNext(nullptr);
+ tail_entry.SetNext(member);
+ this->root[core].SetPrev(member);
+
+ return tail == nullptr;
+ }
+
+ constexpr bool PushFront(s32 core, Member* member) {
+ // Get the entry associated with the member.
+ Entry& member_entry = member->GetPriorityQueueEntry(core);
+
+ // Get the entry associated with the front of the queue.
+ Member* head = this->root[core].GetNext();
+ Entry& head_entry =
+ (head != nullptr) ? head->GetPriorityQueueEntry(core) : this->root[core];
+
+ // Link the entries.
+ member_entry.SetPrev(nullptr);
+ member_entry.SetNext(head);
+ head_entry.SetPrev(member);
+ this->root[core].SetNext(member);
+
+ return (head == nullptr);
+ }
+
+ constexpr bool Remove(s32 core, Member* member) {
+ // Get the entry associated with the member.
+ Entry& member_entry = member->GetPriorityQueueEntry(core);
+
+ // Get the entries associated with next and prev.
+ Member* prev = member_entry.GetPrev();
+ Member* next = member_entry.GetNext();
+ Entry& prev_entry =
+ (prev != nullptr) ? prev->GetPriorityQueueEntry(core) : this->root[core];
+ Entry& next_entry =
+ (next != nullptr) ? next->GetPriorityQueueEntry(core) : this->root[core];
+
+ // Unlink.
+ prev_entry.SetNext(next);
+ next_entry.SetPrev(prev);
+
+ return (this->GetFront(core) == nullptr);
+ }
+
+ constexpr Member* GetFront(s32 core) const {
+ return this->root[core].GetNext();
+ }
+ };
+
+ class KPriorityQueueImpl {
+ public:
+ constexpr KPriorityQueueImpl() = default;
+
+ constexpr void PushBack(s32 priority, s32 core, Member* member) {
+ ASSERT(IsValidCore(core));
+ ASSERT(IsValidPriority(priority));
+
+ if (priority > LowestPriority) {
+ return;
+ }
+
+ if (this->queues[priority].PushBack(core, member)) {
+ this->available_priorities[core].SetBit(priority);
+ }
+ }
+
+ constexpr void PushFront(s32 priority, s32 core, Member* member) {
+ ASSERT(IsValidCore(core));
+ ASSERT(IsValidPriority(priority));
+
+ if (priority > LowestPriority) {
+ return;
+ }
+
+ if (this->queues[priority].PushFront(core, member)) {
+ this->available_priorities[core].SetBit(priority);
+ }
+ }
+
+ constexpr void Remove(s32 priority, s32 core, Member* member) {
+ ASSERT(IsValidCore(core));
+ ASSERT(IsValidPriority(priority));
+
+ if (priority > LowestPriority) {
+ return;
+ }
+
+ if (this->queues[priority].Remove(core, member)) {
+ this->available_priorities[core].ClearBit(priority);
+ }
+ }
+
+ constexpr Member* GetFront(s32 core) const {
+ ASSERT(IsValidCore(core));
+
+ const s32 priority =
+ static_cast<s32>(this->available_priorities[core].CountLeadingZero());
+ if (priority <= LowestPriority) {
+ return this->queues[priority].GetFront(core);
+ } else {
+ return nullptr;
+ }
+ }
+
+ constexpr Member* GetFront(s32 priority, s32 core) const {
+ ASSERT(IsValidCore(core));
+ ASSERT(IsValidPriority(priority));
+
+ if (priority <= LowestPriority) {
+ return this->queues[priority].GetFront(core);
+ } else {
+ return nullptr;
+ }
+ }
+
+ constexpr Member* GetNext(s32 core, const Member* member) const {
+ ASSERT(IsValidCore(core));
+
+ Member* next = member->GetPriorityQueueEntry(core).GetNext();
+ if (next == nullptr) {
+ const s32 priority = static_cast<s32>(
+ this->available_priorities[core].GetNextSet(member->GetPriority()));
+ if (priority <= LowestPriority) {
+ next = this->queues[priority].GetFront(core);
+ }
+ }
+ return next;
+ }
+
+ constexpr void MoveToFront(s32 priority, s32 core, Member* member) {
+ ASSERT(IsValidCore(core));
+ ASSERT(IsValidPriority(priority));
+
+ if (priority <= LowestPriority) {
+ this->queues[priority].Remove(core, member);
+ this->queues[priority].PushFront(core, member);
+ }
+ }
+
+ constexpr Member* MoveToBack(s32 priority, s32 core, Member* member) {
+ ASSERT(IsValidCore(core));
+ ASSERT(IsValidPriority(priority));
+
+ if (priority <= LowestPriority) {
+ this->queues[priority].Remove(core, member);
+ this->queues[priority].PushBack(core, member);
+ return this->queues[priority].GetFront(core);
+ } else {
+ return nullptr;
+ }
+ }
+
+ private:
+ std::array<KPerCoreQueue, NumPriority> queues{};
+ std::array<Common::BitSet64<NumPriority>, NumCores> available_priorities{};
+ };
+
+private:
+ KPriorityQueueImpl scheduled_queue;
+ KPriorityQueueImpl suggested_queue;
+
+private:
+ constexpr void ClearAffinityBit(u64& affinity, s32 core) {
+ affinity &= ~(u64(1) << core);
+ }
+
+ constexpr s32 GetNextCore(u64& affinity) {
+ const s32 core = Common::CountTrailingZeroes64(affinity);
+ ClearAffinityBit(affinity, core);
+ return core;
+ }
+
+ constexpr void PushBack(s32 priority, Member* member) {
+ ASSERT(IsValidPriority(priority));
+
+ // Push onto the scheduled queue for its core, if we can.
+ u64 affinity = member->GetAffinityMask().GetAffinityMask();
+ if (const s32 core = member->GetActiveCore(); core >= 0) {
+ this->scheduled_queue.PushBack(priority, core, member);
+ ClearAffinityBit(affinity, core);
+ }
+
+ // And suggest the thread for all other cores.
+ while (affinity) {
+ this->suggested_queue.PushBack(priority, GetNextCore(affinity), member);
+ }
+ }
+
+ constexpr void PushFront(s32 priority, Member* member) {
+ ASSERT(IsValidPriority(priority));
+
+ // Push onto the scheduled queue for its core, if we can.
+ u64 affinity = member->GetAffinityMask().GetAffinityMask();
+ if (const s32 core = member->GetActiveCore(); core >= 0) {
+ this->scheduled_queue.PushFront(priority, core, member);
+ ClearAffinityBit(affinity, core);
+ }
+
+ // And suggest the thread for all other cores.
+ // Note: Nintendo pushes onto the back of the suggested queue, not the front.
+ while (affinity) {
+ this->suggested_queue.PushBack(priority, GetNextCore(affinity), member);
+ }
+ }
+
+ constexpr void Remove(s32 priority, Member* member) {
+ ASSERT(IsValidPriority(priority));
+
+ // Remove from the scheduled queue for its core.
+ u64 affinity = member->GetAffinityMask().GetAffinityMask();
+ if (const s32 core = member->GetActiveCore(); core >= 0) {
+ this->scheduled_queue.Remove(priority, core, member);
+ ClearAffinityBit(affinity, core);
+ }
+
+ // Remove from the suggested queue for all other cores.
+ while (affinity) {
+ this->suggested_queue.Remove(priority, GetNextCore(affinity), member);
+ }
+ }
+
+public:
+ constexpr KPriorityQueue() = default;
+
+ // Getters.
+ constexpr Member* GetScheduledFront(s32 core) const {
+ return this->scheduled_queue.GetFront(core);
+ }
+
+ constexpr Member* GetScheduledFront(s32 core, s32 priority) const {
+ return this->scheduled_queue.GetFront(priority, core);
+ }
+
+ constexpr Member* GetSuggestedFront(s32 core) const {
+ return this->suggested_queue.GetFront(core);
+ }
+
+ constexpr Member* GetSuggestedFront(s32 core, s32 priority) const {
+ return this->suggested_queue.GetFront(priority, core);
+ }
+
+ constexpr Member* GetScheduledNext(s32 core, const Member* member) const {
+ return this->scheduled_queue.GetNext(core, member);
+ }
+
+ constexpr Member* GetSuggestedNext(s32 core, const Member* member) const {
+ return this->suggested_queue.GetNext(core, member);
+ }
+
+ constexpr Member* GetSamePriorityNext(s32 core, const Member* member) const {
+ return member->GetPriorityQueueEntry(core).GetNext();
+ }
+
+ // Mutators.
+ constexpr void PushBack(Member* member) {
+ this->PushBack(member->GetPriority(), member);
+ }
+
+ constexpr void Remove(Member* member) {
+ this->Remove(member->GetPriority(), member);
+ }
+
+ constexpr void MoveToScheduledFront(Member* member) {
+ this->scheduled_queue.MoveToFront(member->GetPriority(), member->GetActiveCore(), member);
+ }
+
+ constexpr Thread* MoveToScheduledBack(Member* member) {
+ return this->scheduled_queue.MoveToBack(member->GetPriority(), member->GetActiveCore(),
+ member);
+ }
+
+ // First class fancy operations.
+ constexpr void ChangePriority(s32 prev_priority, bool is_running, Member* member) {
+ ASSERT(IsValidPriority(prev_priority));
+
+ // Remove the member from the queues.
+ const s32 new_priority = member->GetPriority();
+ this->Remove(prev_priority, member);
+
+ // And enqueue. If the member is running, we want to keep it running.
+ if (is_running) {
+ this->PushFront(new_priority, member);
+ } else {
+ this->PushBack(new_priority, member);
+ }
+ }
+
+ constexpr void ChangeAffinityMask(s32 prev_core, const AffinityMaskType& prev_affinity,
+ Member* member) {
+ // Get the new information.
+ const s32 priority = member->GetPriority();
+ const AffinityMaskType& new_affinity = member->GetAffinityMask();
+ const s32 new_core = member->GetActiveCore();
+
+ // Remove the member from all queues it was in before.
+ for (s32 core = 0; core < static_cast<s32>(NumCores); core++) {
+ if (prev_affinity.GetAffinity(core)) {
+ if (core == prev_core) {
+ this->scheduled_queue.Remove(priority, core, member);
+ } else {
+ this->suggested_queue.Remove(priority, core, member);
+ }
+ }
+ }
+
+ // And add the member to all queues it should be in now.
+ for (s32 core = 0; core < static_cast<s32>(NumCores); core++) {
+ if (new_affinity.GetAffinity(core)) {
+ if (core == new_core) {
+ this->scheduled_queue.PushBack(priority, core, member);
+ } else {
+ this->suggested_queue.PushBack(priority, core, member);
+ }
+ }
+ }
+ }
+
+ constexpr void ChangeCore(s32 prev_core, Member* member, bool to_front = false) {
+ // Get the new information.
+ const s32 new_core = member->GetActiveCore();
+ const s32 priority = member->GetPriority();
+
+ // We don't need to do anything if the core is the same.
+ if (prev_core != new_core) {
+ // Remove from the scheduled queue for the previous core.
+ if (prev_core >= 0) {
+ this->scheduled_queue.Remove(priority, prev_core, member);
+ }
+
+ // Remove from the suggested queue and add to the scheduled queue for the new core.
+ if (new_core >= 0) {
+ this->suggested_queue.Remove(priority, new_core, member);
+ if (to_front) {
+ this->scheduled_queue.PushFront(priority, new_core, member);
+ } else {
+ this->scheduled_queue.PushBack(priority, new_core, member);
+ }
+ }
+
+ // Add to the suggested queue for the previous core.
+ if (prev_core >= 0) {
+ this->suggested_queue.PushBack(priority, prev_core, member);
+ }
+ }
+ }
+};
+
+} // namespace Kernel
diff --git a/src/core/hle/kernel/k_scheduler.cpp b/src/core/hle/kernel/k_scheduler.cpp
new file mode 100644
index 000000000..c5fd82a6b
--- /dev/null
+++ b/src/core/hle/kernel/k_scheduler.cpp
@@ -0,0 +1,784 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+// This file references various implementation details from Atmosphere, an open-source firmware for
+// the Nintendo Switch. Copyright 2018-2020 Atmosphere-NX.
+
+#include "common/assert.h"
+#include "common/bit_util.h"
+#include "common/fiber.h"
+#include "common/logging/log.h"
+#include "core/arm/arm_interface.h"
+#include "core/core.h"
+#include "core/core_timing.h"
+#include "core/cpu_manager.h"
+#include "core/hle/kernel/k_scheduler.h"
+#include "core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h"
+#include "core/hle/kernel/kernel.h"
+#include "core/hle/kernel/physical_core.h"
+#include "core/hle/kernel/process.h"
+#include "core/hle/kernel/thread.h"
+#include "core/hle/kernel/time_manager.h"
+
+namespace Kernel {
+
+static void IncrementScheduledCount(Kernel::Thread* thread) {
+ if (auto process = thread->GetOwnerProcess(); process) {
+ process->IncrementScheduledCount();
+ }
+}
+
+void KScheduler::RescheduleCores(KernelCore& kernel, u64 cores_pending_reschedule,
+ Core::EmuThreadHandle global_thread) {
+ u32 current_core = global_thread.host_handle;
+ bool must_context_switch = global_thread.guest_handle != InvalidHandle &&
+ (current_core < Core::Hardware::NUM_CPU_CORES);
+
+ while (cores_pending_reschedule != 0) {
+ u32 core = Common::CountTrailingZeroes64(cores_pending_reschedule);
+ ASSERT(core < Core::Hardware::NUM_CPU_CORES);
+ if (!must_context_switch || core != current_core) {
+ auto& phys_core = kernel.PhysicalCore(core);
+ phys_core.Interrupt();
+ } else {
+ must_context_switch = true;
+ }
+ cores_pending_reschedule &= ~(1ULL << core);
+ }
+ if (must_context_switch) {
+ auto core_scheduler = kernel.CurrentScheduler();
+ kernel.ExitSVCProfile();
+ core_scheduler->RescheduleCurrentCore();
+ kernel.EnterSVCProfile();
+ }
+}
+
+u64 KScheduler::UpdateHighestPriorityThread(Thread* highest_thread) {
+ std::scoped_lock lock{guard};
+ if (Thread* prev_highest_thread = this->state.highest_priority_thread;
+ prev_highest_thread != highest_thread) {
+ if (prev_highest_thread != nullptr) {
+ IncrementScheduledCount(prev_highest_thread);
+ prev_highest_thread->SetLastScheduledTick(system.CoreTiming().GetCPUTicks());
+ }
+ if (this->state.should_count_idle) {
+ if (highest_thread != nullptr) {
+ // if (Process* process = highest_thread->GetOwnerProcess(); process != nullptr) {
+ // process->SetRunningThread(this->core_id, highest_thread,
+ // this->state.idle_count);
+ //}
+ } else {
+ this->state.idle_count++;
+ }
+ }
+
+ this->state.highest_priority_thread = highest_thread;
+ this->state.needs_scheduling = true;
+ return (1ULL << this->core_id);
+ } else {
+ return 0;
+ }
+}
+
+u64 KScheduler::UpdateHighestPriorityThreadsImpl(KernelCore& kernel) {
+ ASSERT(kernel.GlobalSchedulerContext().IsLocked());
+
+ // Clear that we need to update.
+ ClearSchedulerUpdateNeeded(kernel);
+
+ u64 cores_needing_scheduling = 0, idle_cores = 0;
+ Thread* top_threads[Core::Hardware::NUM_CPU_CORES];
+ auto& priority_queue = GetPriorityQueue(kernel);
+
+ /// We want to go over all cores, finding the highest priority thread and determining if
+ /// scheduling is needed for that core.
+ for (size_t core_id = 0; core_id < Core::Hardware::NUM_CPU_CORES; core_id++) {
+ Thread* top_thread = priority_queue.GetScheduledFront(static_cast<s32>(core_id));
+ if (top_thread != nullptr) {
+ // If the thread has no waiters, we need to check if the process has a thread pinned.
+ // TODO(bunnei): Implement thread pinning
+ } else {
+ idle_cores |= (1ULL << core_id);
+ }
+
+ top_threads[core_id] = top_thread;
+ cores_needing_scheduling |=
+ kernel.Scheduler(core_id).UpdateHighestPriorityThread(top_threads[core_id]);
+ }
+
+ // Idle cores are bad. We're going to try to migrate threads to each idle core in turn.
+ while (idle_cores != 0) {
+ u32 core_id = Common::CountTrailingZeroes64(idle_cores);
+ if (Thread* suggested = priority_queue.GetSuggestedFront(core_id); suggested != nullptr) {
+ s32 migration_candidates[Core::Hardware::NUM_CPU_CORES];
+ size_t num_candidates = 0;
+
+ // While we have a suggested thread, try to migrate it!
+ while (suggested != nullptr) {
+ // Check if the suggested thread is the top thread on its core.
+ const s32 suggested_core = suggested->GetActiveCore();
+ if (Thread* top_thread =
+ (suggested_core >= 0) ? top_threads[suggested_core] : nullptr;
+ top_thread != suggested) {
+ // Make sure we're not dealing with threads too high priority for migration.
+ if (top_thread != nullptr &&
+ top_thread->GetPriority() < HighestCoreMigrationAllowedPriority) {
+ break;
+ }
+
+ // The suggested thread isn't bound to its core, so we can migrate it!
+ suggested->SetActiveCore(core_id);
+ priority_queue.ChangeCore(suggested_core, suggested);
+
+ top_threads[core_id] = suggested;
+ cores_needing_scheduling |=
+ kernel.Scheduler(core_id).UpdateHighestPriorityThread(top_threads[core_id]);
+ break;
+ }
+
+ // Note this core as a candidate for migration.
+ ASSERT(num_candidates < Core::Hardware::NUM_CPU_CORES);
+ migration_candidates[num_candidates++] = suggested_core;
+ suggested = priority_queue.GetSuggestedNext(core_id, suggested);
+ }
+
+ // If suggested is nullptr, we failed to migrate a specific thread. So let's try all our
+ // candidate cores' top threads.
+ if (suggested == nullptr) {
+ for (size_t i = 0; i < num_candidates; i++) {
+ // Check if there's some other thread that can run on the candidate core.
+ const s32 candidate_core = migration_candidates[i];
+ suggested = top_threads[candidate_core];
+ if (Thread* next_on_candidate_core =
+ priority_queue.GetScheduledNext(candidate_core, suggested);
+ next_on_candidate_core != nullptr) {
+ // The candidate core can run some other thread! We'll migrate its current
+ // top thread to us.
+ top_threads[candidate_core] = next_on_candidate_core;
+ cores_needing_scheduling |=
+ kernel.Scheduler(candidate_core)
+ .UpdateHighestPriorityThread(top_threads[candidate_core]);
+
+ // Perform the migration.
+ suggested->SetActiveCore(core_id);
+ priority_queue.ChangeCore(candidate_core, suggested);
+
+ top_threads[core_id] = suggested;
+ cores_needing_scheduling |=
+ kernel.Scheduler(core_id).UpdateHighestPriorityThread(
+ top_threads[core_id]);
+ break;
+ }
+ }
+ }
+ }
+
+ idle_cores &= ~(1ULL << core_id);
+ }
+
+ return cores_needing_scheduling;
+}
+
+void KScheduler::OnThreadStateChanged(KernelCore& kernel, Thread* thread, u32 old_state) {
+ ASSERT(kernel.GlobalSchedulerContext().IsLocked());
+
+ // Check if the state has changed, because if it hasn't there's nothing to do.
+ const auto cur_state = thread->scheduling_state;
+ if (cur_state == old_state) {
+ return;
+ }
+
+ // Update the priority queues.
+ if (old_state == static_cast<u32>(ThreadSchedStatus::Runnable)) {
+ // If we were previously runnable, then we're not runnable now, and we should remove.
+ GetPriorityQueue(kernel).Remove(thread);
+ IncrementScheduledCount(thread);
+ SetSchedulerUpdateNeeded(kernel);
+ } else if (cur_state == static_cast<u32>(ThreadSchedStatus::Runnable)) {
+ // If we're now runnable, then we weren't previously, and we should add.
+ GetPriorityQueue(kernel).PushBack(thread);
+ IncrementScheduledCount(thread);
+ SetSchedulerUpdateNeeded(kernel);
+ }
+}
+
+void KScheduler::OnThreadPriorityChanged(KernelCore& kernel, Thread* thread, Thread* current_thread,
+ u32 old_priority) {
+
+ ASSERT(kernel.GlobalSchedulerContext().IsLocked());
+
+ // If the thread is runnable, we want to change its priority in the queue.
+ if (thread->scheduling_state == static_cast<u32>(ThreadSchedStatus::Runnable)) {
+ GetPriorityQueue(kernel).ChangePriority(
+ old_priority, thread == kernel.CurrentScheduler()->GetCurrentThread(), thread);
+ IncrementScheduledCount(thread);
+ SetSchedulerUpdateNeeded(kernel);
+ }
+}
+
+void KScheduler::OnThreadAffinityMaskChanged(KernelCore& kernel, Thread* thread,
+ const KAffinityMask& old_affinity, s32 old_core) {
+ ASSERT(kernel.GlobalSchedulerContext().IsLocked());
+
+ // If the thread is runnable, we want to change its affinity in the queue.
+ if (thread->scheduling_state == static_cast<u32>(ThreadSchedStatus::Runnable)) {
+ GetPriorityQueue(kernel).ChangeAffinityMask(old_core, old_affinity, thread);
+ IncrementScheduledCount(thread);
+ SetSchedulerUpdateNeeded(kernel);
+ }
+}
+
+void KScheduler::RotateScheduledQueue(s32 core_id, s32 priority) {
+ ASSERT(system.GlobalSchedulerContext().IsLocked());
+
+ // Get a reference to the priority queue.
+ auto& kernel = system.Kernel();
+ auto& priority_queue = GetPriorityQueue(kernel);
+
+ // Rotate the front of the queue to the end.
+ Thread* top_thread = priority_queue.GetScheduledFront(core_id, priority);
+ Thread* next_thread = nullptr;
+ if (top_thread != nullptr) {
+ next_thread = priority_queue.MoveToScheduledBack(top_thread);
+ if (next_thread != top_thread) {
+ IncrementScheduledCount(top_thread);
+ IncrementScheduledCount(next_thread);
+ }
+ }
+
+ // While we have a suggested thread, try to migrate it!
+ {
+ Thread* suggested = priority_queue.GetSuggestedFront(core_id, priority);
+ while (suggested != nullptr) {
+ // Check if the suggested thread is the top thread on its core.
+ const s32 suggested_core = suggested->GetActiveCore();
+ if (Thread* top_on_suggested_core =
+ (suggested_core >= 0) ? priority_queue.GetScheduledFront(suggested_core)
+ : nullptr;
+ top_on_suggested_core != suggested) {
+ // If the next thread is a new thread that has been waiting longer than our
+ // suggestion, we prefer it to our suggestion.
+ if (top_thread != next_thread && next_thread != nullptr &&
+ next_thread->GetLastScheduledTick() < suggested->GetLastScheduledTick()) {
+ suggested = nullptr;
+ break;
+ }
+
+ // If we're allowed to do a migration, do one.
+ // NOTE: Unlike migrations in UpdateHighestPriorityThread, this moves the suggestion
+ // to the front of the queue.
+ if (top_on_suggested_core == nullptr ||
+ top_on_suggested_core->GetPriority() >= HighestCoreMigrationAllowedPriority) {
+ suggested->SetActiveCore(core_id);
+ priority_queue.ChangeCore(suggested_core, suggested, true);
+ IncrementScheduledCount(suggested);
+ break;
+ }
+ }
+
+ // Get the next suggestion.
+ suggested = priority_queue.GetSamePriorityNext(core_id, suggested);
+ }
+ }
+
+ // Now that we might have migrated a thread with the same priority, check if we can do better.
+
+ {
+ Thread* best_thread = priority_queue.GetScheduledFront(core_id);
+ if (best_thread == GetCurrentThread()) {
+ best_thread = priority_queue.GetScheduledNext(core_id, best_thread);
+ }
+
+ // If the best thread we can choose has a priority the same or worse than ours, try to
+ // migrate a higher priority thread.
+ if (best_thread != nullptr && best_thread->GetPriority() >= static_cast<u32>(priority)) {
+ Thread* suggested = priority_queue.GetSuggestedFront(core_id);
+ while (suggested != nullptr) {
+ // If the suggestion's priority is the same as ours, don't bother.
+ if (suggested->GetPriority() >= best_thread->GetPriority()) {
+ break;
+ }
+
+ // Check if the suggested thread is the top thread on its core.
+ const s32 suggested_core = suggested->GetActiveCore();
+ if (Thread* top_on_suggested_core =
+ (suggested_core >= 0) ? priority_queue.GetScheduledFront(suggested_core)
+ : nullptr;
+ top_on_suggested_core != suggested) {
+ // If we're allowed to do a migration, do one.
+ // NOTE: Unlike migrations in UpdateHighestPriorityThread, this moves the
+ // suggestion to the front of the queue.
+ if (top_on_suggested_core == nullptr ||
+ top_on_suggested_core->GetPriority() >=
+ HighestCoreMigrationAllowedPriority) {
+ suggested->SetActiveCore(core_id);
+ priority_queue.ChangeCore(suggested_core, suggested, true);
+ IncrementScheduledCount(suggested);
+ break;
+ }
+ }
+
+ // Get the next suggestion.
+ suggested = priority_queue.GetSuggestedNext(core_id, suggested);
+ }
+ }
+ }
+
+ // After a rotation, we need a scheduler update.
+ SetSchedulerUpdateNeeded(kernel);
+}
+
+bool KScheduler::CanSchedule(KernelCore& kernel) {
+ return kernel.CurrentScheduler()->GetCurrentThread()->GetDisableDispatchCount() <= 1;
+}
+
+bool KScheduler::IsSchedulerUpdateNeeded(const KernelCore& kernel) {
+ return kernel.GlobalSchedulerContext().scheduler_update_needed.load(std::memory_order_acquire);
+}
+
+void KScheduler::SetSchedulerUpdateNeeded(KernelCore& kernel) {
+ kernel.GlobalSchedulerContext().scheduler_update_needed.store(true, std::memory_order_release);
+}
+
+void KScheduler::ClearSchedulerUpdateNeeded(KernelCore& kernel) {
+ kernel.GlobalSchedulerContext().scheduler_update_needed.store(false, std::memory_order_release);
+}
+
+void KScheduler::DisableScheduling(KernelCore& kernel) {
+ if (auto* scheduler = kernel.CurrentScheduler(); scheduler) {
+ ASSERT(scheduler->GetCurrentThread()->GetDisableDispatchCount() >= 0);
+ scheduler->GetCurrentThread()->DisableDispatch();
+ }
+}
+
+void KScheduler::EnableScheduling(KernelCore& kernel, u64 cores_needing_scheduling,
+ Core::EmuThreadHandle global_thread) {
+ if (auto* scheduler = kernel.CurrentScheduler(); scheduler) {
+ scheduler->GetCurrentThread()->EnableDispatch();
+ }
+ RescheduleCores(kernel, cores_needing_scheduling, global_thread);
+}
+
+u64 KScheduler::UpdateHighestPriorityThreads(KernelCore& kernel) {
+ if (IsSchedulerUpdateNeeded(kernel)) {
+ return UpdateHighestPriorityThreadsImpl(kernel);
+ } else {
+ return 0;
+ }
+}
+
+KSchedulerPriorityQueue& KScheduler::GetPriorityQueue(KernelCore& kernel) {
+ return kernel.GlobalSchedulerContext().priority_queue;
+}
+
+void KScheduler::YieldWithoutCoreMigration() {
+ auto& kernel = system.Kernel();
+
+ // Validate preconditions.
+ ASSERT(CanSchedule(kernel));
+ ASSERT(kernel.CurrentProcess() != nullptr);
+
+ // Get the current thread and process.
+ Thread& cur_thread = *GetCurrentThread();
+ Process& cur_process = *kernel.CurrentProcess();
+
+ // If the thread's yield count matches, there's nothing for us to do.
+ if (cur_thread.GetYieldScheduleCount() == cur_process.GetScheduledCount()) {
+ return;
+ }
+
+ // Get a reference to the priority queue.
+ auto& priority_queue = GetPriorityQueue(kernel);
+
+ // Perform the yield.
+ {
+ KScopedSchedulerLock lock(kernel);
+
+ const auto cur_state = cur_thread.scheduling_state;
+ if (cur_state == static_cast<u32>(ThreadSchedStatus::Runnable)) {
+ // Put the current thread at the back of the queue.
+ Thread* next_thread = priority_queue.MoveToScheduledBack(std::addressof(cur_thread));
+ IncrementScheduledCount(std::addressof(cur_thread));
+
+ // If the next thread is different, we have an update to perform.
+ if (next_thread != std::addressof(cur_thread)) {
+ SetSchedulerUpdateNeeded(kernel);
+ } else {
+ // Otherwise, set the thread's yield count so that we won't waste work until the
+ // process is scheduled again.
+ cur_thread.SetYieldScheduleCount(cur_process.GetScheduledCount());
+ }
+ }
+ }
+}
+
+void KScheduler::YieldWithCoreMigration() {
+ auto& kernel = system.Kernel();
+
+ // Validate preconditions.
+ ASSERT(CanSchedule(kernel));
+ ASSERT(kernel.CurrentProcess() != nullptr);
+
+ // Get the current thread and process.
+ Thread& cur_thread = *GetCurrentThread();
+ Process& cur_process = *kernel.CurrentProcess();
+
+ // If the thread's yield count matches, there's nothing for us to do.
+ if (cur_thread.GetYieldScheduleCount() == cur_process.GetScheduledCount()) {
+ return;
+ }
+
+ // Get a reference to the priority queue.
+ auto& priority_queue = GetPriorityQueue(kernel);
+
+ // Perform the yield.
+ {
+ KScopedSchedulerLock lock(kernel);
+
+ const auto cur_state = cur_thread.scheduling_state;
+ if (cur_state == static_cast<u32>(ThreadSchedStatus::Runnable)) {
+ // Get the current active core.
+ const s32 core_id = cur_thread.GetActiveCore();
+
+ // Put the current thread at the back of the queue.
+ Thread* next_thread = priority_queue.MoveToScheduledBack(std::addressof(cur_thread));
+ IncrementScheduledCount(std::addressof(cur_thread));
+
+ // While we have a suggested thread, try to migrate it!
+ bool recheck = false;
+ Thread* suggested = priority_queue.GetSuggestedFront(core_id);
+ while (suggested != nullptr) {
+ // Check if the suggested thread is the thread running on its core.
+ const s32 suggested_core = suggested->GetActiveCore();
+
+ if (Thread* running_on_suggested_core =
+ (suggested_core >= 0)
+ ? kernel.Scheduler(suggested_core).state.highest_priority_thread
+ : nullptr;
+ running_on_suggested_core != suggested) {
+ // If the current thread's priority is higher than our suggestion's we prefer
+ // the next thread to the suggestion. We also prefer the next thread when the
+ // current thread's priority is equal to the suggestions, but the next thread
+ // has been waiting longer.
+ if ((suggested->GetPriority() > cur_thread.GetPriority()) ||
+ (suggested->GetPriority() == cur_thread.GetPriority() &&
+ next_thread != std::addressof(cur_thread) &&
+ next_thread->GetLastScheduledTick() < suggested->GetLastScheduledTick())) {
+ suggested = nullptr;
+ break;
+ }
+
+ // If we're allowed to do a migration, do one.
+ // NOTE: Unlike migrations in UpdateHighestPriorityThread, this moves the
+ // suggestion to the front of the queue.
+ if (running_on_suggested_core == nullptr ||
+ running_on_suggested_core->GetPriority() >=
+ HighestCoreMigrationAllowedPriority) {
+ suggested->SetActiveCore(core_id);
+ priority_queue.ChangeCore(suggested_core, suggested, true);
+ IncrementScheduledCount(suggested);
+ break;
+ } else {
+ // We couldn't perform a migration, but we should check again on a future
+ // yield.
+ recheck = true;
+ }
+ }
+
+ // Get the next suggestion.
+ suggested = priority_queue.GetSuggestedNext(core_id, suggested);
+ }
+
+ // If we still have a suggestion or the next thread is different, we have an update to
+ // perform.
+ if (suggested != nullptr || next_thread != std::addressof(cur_thread)) {
+ SetSchedulerUpdateNeeded(kernel);
+ } else if (!recheck) {
+ // Otherwise if we don't need to re-check, set the thread's yield count so that we
+ // won't waste work until the process is scheduled again.
+ cur_thread.SetYieldScheduleCount(cur_process.GetScheduledCount());
+ }
+ }
+ }
+}
+
+void KScheduler::YieldToAnyThread() {
+ auto& kernel = system.Kernel();
+
+ // Validate preconditions.
+ ASSERT(CanSchedule(kernel));
+ ASSERT(kernel.CurrentProcess() != nullptr);
+
+ // Get the current thread and process.
+ Thread& cur_thread = *GetCurrentThread();
+ Process& cur_process = *kernel.CurrentProcess();
+
+ // If the thread's yield count matches, there's nothing for us to do.
+ if (cur_thread.GetYieldScheduleCount() == cur_process.GetScheduledCount()) {
+ return;
+ }
+
+ // Get a reference to the priority queue.
+ auto& priority_queue = GetPriorityQueue(kernel);
+
+ // Perform the yield.
+ {
+ KScopedSchedulerLock lock(kernel);
+
+ const auto cur_state = cur_thread.scheduling_state;
+ if (cur_state == static_cast<u32>(ThreadSchedStatus::Runnable)) {
+ // Get the current active core.
+ const s32 core_id = cur_thread.GetActiveCore();
+
+ // Migrate the current thread to core -1.
+ cur_thread.SetActiveCore(-1);
+ priority_queue.ChangeCore(core_id, std::addressof(cur_thread));
+ IncrementScheduledCount(std::addressof(cur_thread));
+
+ // If there's nothing scheduled, we can try to perform a migration.
+ if (priority_queue.GetScheduledFront(core_id) == nullptr) {
+ // While we have a suggested thread, try to migrate it!
+ Thread* suggested = priority_queue.GetSuggestedFront(core_id);
+ while (suggested != nullptr) {
+ // Check if the suggested thread is the top thread on its core.
+ const s32 suggested_core = suggested->GetActiveCore();
+ if (Thread* top_on_suggested_core =
+ (suggested_core >= 0) ? priority_queue.GetScheduledFront(suggested_core)
+ : nullptr;
+ top_on_suggested_core != suggested) {
+ // If we're allowed to do a migration, do one.
+ if (top_on_suggested_core == nullptr ||
+ top_on_suggested_core->GetPriority() >=
+ HighestCoreMigrationAllowedPriority) {
+ suggested->SetActiveCore(core_id);
+ priority_queue.ChangeCore(suggested_core, suggested);
+ IncrementScheduledCount(suggested);
+ }
+
+ // Regardless of whether we migrated, we had a candidate, so we're done.
+ break;
+ }
+
+ // Get the next suggestion.
+ suggested = priority_queue.GetSuggestedNext(core_id, suggested);
+ }
+
+ // If the suggestion is different from the current thread, we need to perform an
+ // update.
+ if (suggested != std::addressof(cur_thread)) {
+ SetSchedulerUpdateNeeded(kernel);
+ } else {
+ // Otherwise, set the thread's yield count so that we won't waste work until the
+ // process is scheduled again.
+ cur_thread.SetYieldScheduleCount(cur_process.GetScheduledCount());
+ }
+ } else {
+ // Otherwise, we have an update to perform.
+ SetSchedulerUpdateNeeded(kernel);
+ }
+ }
+ }
+}
+
+KScheduler::KScheduler(Core::System& system, std::size_t core_id)
+ : system(system), core_id(core_id) {
+ switch_fiber = std::make_shared<Common::Fiber>(OnSwitch, this);
+ this->state.needs_scheduling = true;
+ this->state.interrupt_task_thread_runnable = false;
+ this->state.should_count_idle = false;
+ this->state.idle_count = 0;
+ this->state.idle_thread_stack = nullptr;
+ this->state.highest_priority_thread = nullptr;
+}
+
+KScheduler::~KScheduler() = default;
+
+Thread* KScheduler::GetCurrentThread() const {
+ if (current_thread) {
+ return current_thread;
+ }
+ return idle_thread;
+}
+
+u64 KScheduler::GetLastContextSwitchTicks() const {
+ return last_context_switch_time;
+}
+
+void KScheduler::RescheduleCurrentCore() {
+ ASSERT(GetCurrentThread()->GetDisableDispatchCount() == 1);
+
+ auto& phys_core = system.Kernel().PhysicalCore(core_id);
+ if (phys_core.IsInterrupted()) {
+ phys_core.ClearInterrupt();
+ }
+ guard.lock();
+ if (this->state.needs_scheduling) {
+ Schedule();
+ } else {
+ guard.unlock();
+ }
+}
+
+void KScheduler::OnThreadStart() {
+ SwitchContextStep2();
+}
+
+void KScheduler::Unload(Thread* thread) {
+ if (thread) {
+ thread->SetIsRunning(false);
+ if (thread->IsContinuousOnSVC() && !thread->IsHLEThread()) {
+ system.ArmInterface(core_id).ExceptionalExit();
+ thread->SetContinuousOnSVC(false);
+ }
+ if (!thread->IsHLEThread() && !thread->HasExited()) {
+ Core::ARM_Interface& cpu_core = system.ArmInterface(core_id);
+ cpu_core.SaveContext(thread->GetContext32());
+ cpu_core.SaveContext(thread->GetContext64());
+ // Save the TPIDR_EL0 system register in case it was modified.
+ thread->SetTPIDR_EL0(cpu_core.GetTPIDR_EL0());
+ cpu_core.ClearExclusiveState();
+ }
+ thread->context_guard.unlock();
+ }
+}
+
+void KScheduler::Reload(Thread* thread) {
+ if (thread) {
+ ASSERT_MSG(thread->GetSchedulingStatus() == ThreadSchedStatus::Runnable,
+ "Thread must be runnable.");
+
+ // Cancel any outstanding wakeup events for this thread
+ thread->SetIsRunning(true);
+ thread->SetWasRunning(false);
+
+ auto* const thread_owner_process = thread->GetOwnerProcess();
+ if (thread_owner_process != nullptr) {
+ system.Kernel().MakeCurrentProcess(thread_owner_process);
+ }
+ if (!thread->IsHLEThread()) {
+ Core::ARM_Interface& cpu_core = system.ArmInterface(core_id);
+ cpu_core.LoadContext(thread->GetContext32());
+ cpu_core.LoadContext(thread->GetContext64());
+ cpu_core.SetTlsAddress(thread->GetTLSAddress());
+ cpu_core.SetTPIDR_EL0(thread->GetTPIDR_EL0());
+ cpu_core.ClearExclusiveState();
+ }
+ }
+}
+
+void KScheduler::SwitchContextStep2() {
+ // Load context of new thread
+ Reload(current_thread);
+
+ RescheduleCurrentCore();
+}
+
+void KScheduler::ScheduleImpl() {
+ Thread* previous_thread = current_thread;
+ current_thread = state.highest_priority_thread;
+
+ this->state.needs_scheduling = false;
+
+ if (current_thread == previous_thread) {
+ guard.unlock();
+ return;
+ }
+
+ Process* const previous_process = system.Kernel().CurrentProcess();
+
+ UpdateLastContextSwitchTime(previous_thread, previous_process);
+
+ // Save context for previous thread
+ Unload(previous_thread);
+
+ std::shared_ptr<Common::Fiber>* old_context;
+ if (previous_thread != nullptr) {
+ old_context = &previous_thread->GetHostContext();
+ } else {
+ old_context = &idle_thread->GetHostContext();
+ }
+ guard.unlock();
+
+ Common::Fiber::YieldTo(*old_context, switch_fiber);
+ /// When a thread wakes up, the scheduler may have changed to other in another core.
+ auto& next_scheduler = *system.Kernel().CurrentScheduler();
+ next_scheduler.SwitchContextStep2();
+}
+
+void KScheduler::OnSwitch(void* this_scheduler) {
+ KScheduler* sched = static_cast<KScheduler*>(this_scheduler);
+ sched->SwitchToCurrent();
+}
+
+void KScheduler::SwitchToCurrent() {
+ while (true) {
+ {
+ std::scoped_lock lock{guard};
+ current_thread = state.highest_priority_thread;
+ this->state.needs_scheduling = false;
+ }
+ const auto is_switch_pending = [this] {
+ std::scoped_lock lock{guard};
+ return state.needs_scheduling.load(std::memory_order_relaxed);
+ };
+ do {
+ if (current_thread != nullptr && !current_thread->IsHLEThread()) {
+ current_thread->context_guard.lock();
+ if (!current_thread->IsRunnable()) {
+ current_thread->context_guard.unlock();
+ break;
+ }
+ if (static_cast<u32>(current_thread->GetProcessorID()) != core_id) {
+ current_thread->context_guard.unlock();
+ break;
+ }
+ }
+ std::shared_ptr<Common::Fiber>* next_context;
+ if (current_thread != nullptr) {
+ next_context = &current_thread->GetHostContext();
+ } else {
+ next_context = &idle_thread->GetHostContext();
+ }
+ Common::Fiber::YieldTo(switch_fiber, *next_context);
+ } while (!is_switch_pending());
+ }
+}
+
+void KScheduler::UpdateLastContextSwitchTime(Thread* thread, Process* process) {
+ const u64 prev_switch_ticks = last_context_switch_time;
+ const u64 most_recent_switch_ticks = system.CoreTiming().GetCPUTicks();
+ const u64 update_ticks = most_recent_switch_ticks - prev_switch_ticks;
+
+ if (thread != nullptr) {
+ thread->UpdateCPUTimeTicks(update_ticks);
+ }
+
+ if (process != nullptr) {
+ process->UpdateCPUTimeTicks(update_ticks);
+ }
+
+ last_context_switch_time = most_recent_switch_ticks;
+}
+
+void KScheduler::Initialize() {
+ std::string name = "Idle Thread Id:" + std::to_string(core_id);
+ std::function<void(void*)> init_func = Core::CpuManager::GetIdleThreadStartFunc();
+ void* init_func_parameter = system.GetCpuManager().GetStartFuncParamater();
+ ThreadType type = static_cast<ThreadType>(THREADTYPE_KERNEL | THREADTYPE_HLE | THREADTYPE_IDLE);
+ auto thread_res = Thread::Create(system, type, name, 0, 64, 0, static_cast<u32>(core_id), 0,
+ nullptr, std::move(init_func), init_func_parameter);
+ idle_thread = thread_res.Unwrap().get();
+
+ {
+ KScopedSchedulerLock lock{system.Kernel()};
+ idle_thread->SetStatus(ThreadStatus::Ready);
+ }
+}
+
+KScopedSchedulerLock::KScopedSchedulerLock(KernelCore& kernel)
+ : KScopedLock(kernel.GlobalSchedulerContext().SchedulerLock()) {}
+
+KScopedSchedulerLock::~KScopedSchedulerLock() = default;
+
+} // namespace Kernel
diff --git a/src/core/hle/kernel/k_scheduler.h b/src/core/hle/kernel/k_scheduler.h
new file mode 100644
index 000000000..e84abc84c
--- /dev/null
+++ b/src/core/hle/kernel/k_scheduler.h
@@ -0,0 +1,201 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+// This file references various implementation details from Atmosphere, an open-source firmware for
+// the Nintendo Switch. Copyright 2018-2020 Atmosphere-NX.
+
+#pragma once
+
+#include <atomic>
+
+#include "common/common_types.h"
+#include "common/spin_lock.h"
+#include "core/hle/kernel/global_scheduler_context.h"
+#include "core/hle/kernel/k_priority_queue.h"
+#include "core/hle/kernel/k_scheduler_lock.h"
+#include "core/hle/kernel/k_scoped_lock.h"
+
+namespace Common {
+class Fiber;
+}
+
+namespace Core {
+class System;
+}
+
+namespace Kernel {
+
+class KernelCore;
+class Process;
+class SchedulerLock;
+class Thread;
+
+class KScheduler final {
+public:
+ explicit KScheduler(Core::System& system, std::size_t core_id);
+ ~KScheduler();
+
+ /// Reschedules to the next available thread (call after current thread is suspended)
+ void RescheduleCurrentCore();
+
+ /// Reschedules cores pending reschedule, to be called on EnableScheduling.
+ static void RescheduleCores(KernelCore& kernel, u64 cores_pending_reschedule,
+ Core::EmuThreadHandle global_thread);
+
+ /// The next two are for SingleCore Only.
+ /// Unload current thread before preempting core.
+ void Unload(Thread* thread);
+
+ /// Reload current thread after core preemption.
+ void Reload(Thread* thread);
+
+ /// Gets the current running thread
+ [[nodiscard]] Thread* GetCurrentThread() const;
+
+ /// Gets the timestamp for the last context switch in ticks.
+ [[nodiscard]] u64 GetLastContextSwitchTicks() const;
+
+ [[nodiscard]] bool ContextSwitchPending() const {
+ return state.needs_scheduling.load(std::memory_order_relaxed);
+ }
+
+ void Initialize();
+
+ void OnThreadStart();
+
+ [[nodiscard]] std::shared_ptr<Common::Fiber>& ControlContext() {
+ return switch_fiber;
+ }
+
+ [[nodiscard]] const std::shared_ptr<Common::Fiber>& ControlContext() const {
+ return switch_fiber;
+ }
+
+ [[nodiscard]] u64 UpdateHighestPriorityThread(Thread* highest_thread);
+
+ /**
+ * Takes a thread and moves it to the back of the it's priority list.
+ *
+ * @note This operation can be redundant and no scheduling is changed if marked as so.
+ */
+ void YieldWithoutCoreMigration();
+
+ /**
+ * Takes a thread and moves it to the back of the it's priority list.
+ * Afterwards, tries to pick a suggested thread from the suggested queue that has worse time or
+ * a better priority than the next thread in the core.
+ *
+ * @note This operation can be redundant and no scheduling is changed if marked as so.
+ */
+ void YieldWithCoreMigration();
+
+ /**
+ * Takes a thread and moves it out of the scheduling queue.
+ * and into the suggested queue. If no thread can be scheduled afterwards in that core,
+ * a suggested thread is obtained instead.
+ *
+ * @note This operation can be redundant and no scheduling is changed if marked as so.
+ */
+ void YieldToAnyThread();
+
+ /// Notify the scheduler a thread's status has changed.
+ static void OnThreadStateChanged(KernelCore& kernel, Thread* thread, u32 old_state);
+
+ /// Notify the scheduler a thread's priority has changed.
+ static void OnThreadPriorityChanged(KernelCore& kernel, Thread* thread, Thread* current_thread,
+ u32 old_priority);
+
+ /// Notify the scheduler a thread's core and/or affinity mask has changed.
+ static void OnThreadAffinityMaskChanged(KernelCore& kernel, Thread* thread,
+ const KAffinityMask& old_affinity, s32 old_core);
+
+ static bool CanSchedule(KernelCore& kernel);
+ static bool IsSchedulerUpdateNeeded(const KernelCore& kernel);
+ static void SetSchedulerUpdateNeeded(KernelCore& kernel);
+ static void ClearSchedulerUpdateNeeded(KernelCore& kernel);
+ static void DisableScheduling(KernelCore& kernel);
+ static void EnableScheduling(KernelCore& kernel, u64 cores_needing_scheduling,
+ Core::EmuThreadHandle global_thread);
+ [[nodiscard]] static u64 UpdateHighestPriorityThreads(KernelCore& kernel);
+
+private:
+ friend class GlobalSchedulerContext;
+
+ /**
+ * Takes care of selecting the new scheduled threads in three steps:
+ *
+ * 1. First a thread is selected from the top of the priority queue. If no thread
+ * is obtained then we move to step two, else we are done.
+ *
+ * 2. Second we try to get a suggested thread that's not assigned to any core or
+ * that is not the top thread in that core.
+ *
+ * 3. Third is no suggested thread is found, we do a second pass and pick a running
+ * thread in another core and swap it with its current thread.
+ *
+ * returns the cores needing scheduling.
+ */
+ [[nodiscard]] static u64 UpdateHighestPriorityThreadsImpl(KernelCore& kernel);
+
+ [[nodiscard]] static KSchedulerPriorityQueue& GetPriorityQueue(KernelCore& kernel);
+
+ void RotateScheduledQueue(s32 core_id, s32 priority);
+
+ void Schedule() {
+ ASSERT(GetCurrentThread()->GetDisableDispatchCount() == 1);
+ this->ScheduleImpl();
+ }
+
+ /// Switches the CPU's active thread context to that of the specified thread
+ void ScheduleImpl();
+
+ /// When a thread wakes up, it must run this through it's new scheduler
+ void SwitchContextStep2();
+
+ /**
+ * Called on every context switch to update the internal timestamp
+ * This also updates the running time ticks for the given thread and
+ * process using the following difference:
+ *
+ * ticks += most_recent_ticks - last_context_switch_ticks
+ *
+ * The internal tick timestamp for the scheduler is simply the
+ * most recent tick count retrieved. No special arithmetic is
+ * applied to it.
+ */
+ void UpdateLastContextSwitchTime(Thread* thread, Process* process);
+
+ static void OnSwitch(void* this_scheduler);
+ void SwitchToCurrent();
+
+ Thread* current_thread{};
+ Thread* idle_thread{};
+
+ std::shared_ptr<Common::Fiber> switch_fiber{};
+
+ struct SchedulingState {
+ std::atomic<bool> needs_scheduling;
+ bool interrupt_task_thread_runnable{};
+ bool should_count_idle{};
+ u64 idle_count{};
+ Thread* highest_priority_thread{};
+ void* idle_thread_stack{};
+ };
+
+ SchedulingState state;
+
+ Core::System& system;
+ u64 last_context_switch_time{};
+ const std::size_t core_id;
+
+ Common::SpinLock guard{};
+};
+
+class KScopedSchedulerLock : KScopedLock<GlobalSchedulerContext::LockType> {
+public:
+ explicit KScopedSchedulerLock(KernelCore& kernel);
+ ~KScopedSchedulerLock();
+};
+
+} // namespace Kernel
diff --git a/src/core/hle/kernel/k_scheduler_lock.h b/src/core/hle/kernel/k_scheduler_lock.h
new file mode 100644
index 000000000..2f1c1f691
--- /dev/null
+++ b/src/core/hle/kernel/k_scheduler_lock.h
@@ -0,0 +1,75 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+// This file references various implementation details from Atmosphere, an open-source firmware for
+// the Nintendo Switch. Copyright 2018-2020 Atmosphere-NX.
+
+#pragma once
+
+#include "common/assert.h"
+#include "common/spin_lock.h"
+#include "core/hardware_properties.h"
+#include "core/hle/kernel/kernel.h"
+
+namespace Kernel {
+
+class KernelCore;
+
+template <typename SchedulerType>
+class KAbstractSchedulerLock {
+public:
+ explicit KAbstractSchedulerLock(KernelCore& kernel) : kernel{kernel} {}
+
+ bool IsLockedByCurrentThread() const {
+ return this->owner_thread == kernel.GetCurrentEmuThreadID();
+ }
+
+ void Lock() {
+ if (this->IsLockedByCurrentThread()) {
+ // If we already own the lock, we can just increment the count.
+ ASSERT(this->lock_count > 0);
+ this->lock_count++;
+ } else {
+ // Otherwise, we want to disable scheduling and acquire the spinlock.
+ SchedulerType::DisableScheduling(kernel);
+ this->spin_lock.lock();
+
+ // For debug, ensure that our state is valid.
+ ASSERT(this->lock_count == 0);
+ ASSERT(this->owner_thread == Core::EmuThreadHandle::InvalidHandle());
+
+ // Increment count, take ownership.
+ this->lock_count = 1;
+ this->owner_thread = kernel.GetCurrentEmuThreadID();
+ }
+ }
+
+ void Unlock() {
+ ASSERT(this->IsLockedByCurrentThread());
+ ASSERT(this->lock_count > 0);
+
+ // Release an instance of the lock.
+ if ((--this->lock_count) == 0) {
+ // We're no longer going to hold the lock. Take note of what cores need scheduling.
+ const u64 cores_needing_scheduling =
+ SchedulerType::UpdateHighestPriorityThreads(kernel);
+ Core::EmuThreadHandle leaving_thread = owner_thread;
+
+ // Note that we no longer hold the lock, and unlock the spinlock.
+ this->owner_thread = Core::EmuThreadHandle::InvalidHandle();
+ this->spin_lock.unlock();
+
+ // Enable scheduling, and perform a rescheduling operation.
+ SchedulerType::EnableScheduling(kernel, cores_needing_scheduling, leaving_thread);
+ }
+ }
+
+private:
+ KernelCore& kernel;
+ Common::SpinLock spin_lock{};
+ s32 lock_count{};
+ Core::EmuThreadHandle owner_thread{Core::EmuThreadHandle::InvalidHandle()};
+};
+
+} // namespace Kernel
diff --git a/src/core/hle/kernel/k_scoped_lock.h b/src/core/hle/kernel/k_scoped_lock.h
new file mode 100644
index 000000000..d7cc557b2
--- /dev/null
+++ b/src/core/hle/kernel/k_scoped_lock.h
@@ -0,0 +1,41 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+// This file references various implementation details from Atmosphere, an open-source firmware for
+// the Nintendo Switch. Copyright 2018-2020 Atmosphere-NX.
+
+#pragma once
+
+#include "common/common_types.h"
+
+namespace Kernel {
+
+template <typename T>
+concept KLockable = !std::is_reference_v<T> && requires(T & t) {
+ { t.Lock() }
+ ->std::same_as<void>;
+ { t.Unlock() }
+ ->std::same_as<void>;
+};
+
+template <typename T>
+requires KLockable<T> class KScopedLock {
+public:
+ explicit KScopedLock(T* l) : lock_ptr(l) {
+ this->lock_ptr->Lock();
+ }
+ explicit KScopedLock(T& l) : KScopedLock(std::addressof(l)) { /* ... */
+ }
+ ~KScopedLock() {
+ this->lock_ptr->Unlock();
+ }
+
+ KScopedLock(const KScopedLock&) = delete;
+ KScopedLock(KScopedLock&&) = delete;
+
+private:
+ T* lock_ptr;
+};
+
+} // namespace Kernel
diff --git a/src/core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h b/src/core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h
new file mode 100644
index 000000000..2bb3817fa
--- /dev/null
+++ b/src/core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h
@@ -0,0 +1,50 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+// This file references various implementation details from Atmosphere, an open-source firmware for
+// the Nintendo Switch. Copyright 2018-2020 Atmosphere-NX.
+
+#pragma once
+
+#include "common/common_types.h"
+#include "core/hle/kernel/handle_table.h"
+#include "core/hle/kernel/kernel.h"
+#include "core/hle/kernel/thread.h"
+#include "core/hle/kernel/time_manager.h"
+
+namespace Kernel {
+
+class KScopedSchedulerLockAndSleep {
+public:
+ explicit KScopedSchedulerLockAndSleep(KernelCore& kernel, Handle& event_handle, Thread* t,
+ s64 timeout)
+ : kernel(kernel), event_handle(event_handle), thread(t), timeout_tick(timeout) {
+ event_handle = InvalidHandle;
+
+ // Lock the scheduler.
+ kernel.GlobalSchedulerContext().scheduler_lock.Lock();
+ }
+
+ ~KScopedSchedulerLockAndSleep() {
+ // Register the sleep.
+ if (this->timeout_tick > 0) {
+ kernel.TimeManager().ScheduleTimeEvent(event_handle, this->thread, this->timeout_tick);
+ }
+
+ // Unlock the scheduler.
+ kernel.GlobalSchedulerContext().scheduler_lock.Unlock();
+ }
+
+ void CancelSleep() {
+ this->timeout_tick = 0;
+ }
+
+private:
+ KernelCore& kernel;
+ Handle& event_handle;
+ Thread* thread{};
+ s64 timeout_tick{};
+};
+
+} // namespace Kernel
diff --git a/src/core/hle/kernel/kernel.cpp b/src/core/hle/kernel/kernel.cpp
index 929db696d..e8ece8164 100644
--- a/src/core/hle/kernel/kernel.cpp
+++ b/src/core/hle/kernel/kernel.cpp
@@ -8,13 +8,14 @@
#include <functional>
#include <memory>
#include <thread>
-#include <unordered_map>
+#include <unordered_set>
#include <utility>
#include "common/assert.h"
#include "common/logging/log.h"
#include "common/microprofile.h"
#include "common/thread.h"
+#include "common/thread_worker.h"
#include "core/arm/arm_interface.h"
#include "core/arm/cpu_interrupt_handler.h"
#include "core/arm/exclusive_monitor.h"
@@ -27,6 +28,7 @@
#include "core/hle/kernel/client_port.h"
#include "core/hle/kernel/errors.h"
#include "core/hle/kernel/handle_table.h"
+#include "core/hle/kernel/k_scheduler.h"
#include "core/hle/kernel/kernel.h"
#include "core/hle/kernel/memory/memory_layout.h"
#include "core/hle/kernel/memory/memory_manager.h"
@@ -34,7 +36,7 @@
#include "core/hle/kernel/physical_core.h"
#include "core/hle/kernel/process.h"
#include "core/hle/kernel/resource_limit.h"
-#include "core/hle/kernel/scheduler.h"
+#include "core/hle/kernel/service_thread.h"
#include "core/hle/kernel/shared_memory.h"
#include "core/hle/kernel/synchronization.h"
#include "core/hle/kernel/thread.h"
@@ -49,17 +51,20 @@ namespace Kernel {
struct KernelCore::Impl {
explicit Impl(Core::System& system, KernelCore& kernel)
- : global_scheduler{kernel}, synchronization{system}, time_manager{system},
- global_handle_table{kernel}, system{system} {}
+ : synchronization{system}, time_manager{system}, global_handle_table{kernel}, system{
+ system} {}
void SetMulticore(bool is_multicore) {
this->is_multicore = is_multicore;
}
void Initialize(KernelCore& kernel) {
- Shutdown();
RegisterHostThread();
+ global_scheduler_context = std::make_unique<Kernel::GlobalSchedulerContext>(kernel);
+ service_thread_manager =
+ std::make_unique<Common::ThreadWorker>(1, "yuzu:ServiceThreadManager");
+
InitializePhysicalCores();
InitializeSystemResourceLimit(kernel);
InitializeMemoryLayout();
@@ -75,6 +80,12 @@ struct KernelCore::Impl {
}
void Shutdown() {
+ process_list.clear();
+
+ // Ensures all service threads gracefully shutdown
+ service_thread_manager.reset();
+ service_threads.clear();
+
next_object_id = 0;
next_kernel_process_id = Process::InitialKIPIDMin;
next_user_process_id = Process::ProcessIDMin;
@@ -86,42 +97,29 @@ struct KernelCore::Impl {
}
}
- for (std::size_t i = 0; i < cores.size(); i++) {
- cores[i].Shutdown();
- schedulers[i].reset();
- }
cores.clear();
- process_list.clear();
current_process = nullptr;
system_resource_limit = nullptr;
global_handle_table.Clear();
- preemption_event = nullptr;
- global_scheduler.Shutdown();
+ preemption_event = nullptr;
named_ports.clear();
- for (auto& core : cores) {
- core.Shutdown();
- }
- cores.clear();
-
exclusive_monitor.reset();
- num_host_threads = 0;
- std::fill(register_host_thread_keys.begin(), register_host_thread_keys.end(),
- std::thread::id{});
- std::fill(register_host_thread_values.begin(), register_host_thread_values.end(), 0);
+ // Next host thead ID to use, 0-3 IDs represent core threads, >3 represent others
+ next_host_thread_id = Core::Hardware::NUM_CPU_CORES;
}
void InitializePhysicalCores() {
exclusive_monitor =
Core::MakeExclusiveMonitor(system.Memory(), Core::Hardware::NUM_CPU_CORES);
for (std::size_t i = 0; i < Core::Hardware::NUM_CPU_CORES; i++) {
- schedulers[i] = std::make_unique<Kernel::Scheduler>(system, i);
+ schedulers[i] = std::make_unique<Kernel::KScheduler>(system, i);
cores.emplace_back(i, system, *schedulers[i], interrupts);
}
}
@@ -154,8 +152,8 @@ struct KernelCore::Impl {
preemption_event = Core::Timing::CreateEvent(
"PreemptionCallback", [this, &kernel](std::uintptr_t, std::chrono::nanoseconds) {
{
- SchedulerLock lock(kernel);
- global_scheduler.PreemptThreads();
+ KScopedSchedulerLock lock(kernel);
+ global_scheduler_context->PreemptThreads();
}
const auto time_interval = std::chrono::nanoseconds{
Core::Timing::msToCycles(std::chrono::milliseconds(10))};
@@ -194,58 +192,52 @@ struct KernelCore::Impl {
}
}
+ /// Creates a new host thread ID, should only be called by GetHostThreadId
+ u32 AllocateHostThreadId(std::optional<std::size_t> core_id) {
+ if (core_id) {
+ // The first for slots are reserved for CPU core threads
+ ASSERT(*core_id < Core::Hardware::NUM_CPU_CORES);
+ return static_cast<u32>(*core_id);
+ } else {
+ return next_host_thread_id++;
+ }
+ }
+
+ /// Gets the host thread ID for the caller, allocating a new one if this is the first time
+ u32 GetHostThreadId(std::optional<std::size_t> core_id = std::nullopt) {
+ const thread_local auto host_thread_id{AllocateHostThreadId(core_id)};
+ return host_thread_id;
+ }
+
+ /// Registers a CPU core thread by allocating a host thread ID for it
void RegisterCoreThread(std::size_t core_id) {
- const std::thread::id this_id = std::this_thread::get_id();
+ ASSERT(core_id < Core::Hardware::NUM_CPU_CORES);
+ const auto this_id = GetHostThreadId(core_id);
if (!is_multicore) {
single_core_thread_id = this_id;
}
- const auto end =
- register_host_thread_keys.begin() + static_cast<ptrdiff_t>(num_host_threads);
- const auto it = std::find(register_host_thread_keys.begin(), end, this_id);
- ASSERT(core_id < Core::Hardware::NUM_CPU_CORES);
- ASSERT(it == end);
- InsertHostThread(static_cast<u32>(core_id));
}
+ /// Registers a new host thread by allocating a host thread ID for it
void RegisterHostThread() {
- const std::thread::id this_id = std::this_thread::get_id();
- const auto end =
- register_host_thread_keys.begin() + static_cast<ptrdiff_t>(num_host_threads);
- const auto it = std::find(register_host_thread_keys.begin(), end, this_id);
- if (it == end) {
- InsertHostThread(registered_thread_ids++);
- }
+ [[maybe_unused]] const auto this_id = GetHostThreadId();
}
- void InsertHostThread(u32 value) {
- const size_t index = num_host_threads++;
- ASSERT_MSG(index < NUM_REGISTRABLE_HOST_THREADS, "Too many host threads");
- register_host_thread_values[index] = value;
- register_host_thread_keys[index] = std::this_thread::get_id();
- }
-
- [[nodiscard]] u32 GetCurrentHostThreadID() const {
- const std::thread::id this_id = std::this_thread::get_id();
+ [[nodiscard]] u32 GetCurrentHostThreadID() {
+ const auto this_id = GetHostThreadId();
if (!is_multicore && single_core_thread_id == this_id) {
return static_cast<u32>(system.GetCpuManager().CurrentCore());
}
- const auto end =
- register_host_thread_keys.begin() + static_cast<ptrdiff_t>(num_host_threads);
- const auto it = std::find(register_host_thread_keys.begin(), end, this_id);
- if (it == end) {
- return Core::INVALID_HOST_THREAD_ID;
- }
- return register_host_thread_values[static_cast<size_t>(
- std::distance(register_host_thread_keys.begin(), it))];
+ return this_id;
}
- Core::EmuThreadHandle GetCurrentEmuThreadID() const {
+ [[nodiscard]] Core::EmuThreadHandle GetCurrentEmuThreadID() {
Core::EmuThreadHandle result = Core::EmuThreadHandle::InvalidHandle();
result.host_handle = GetCurrentHostThreadID();
if (result.host_handle >= Core::Hardware::NUM_CPU_CORES) {
return result;
}
- const Kernel::Scheduler& sched = cores[result.host_handle].Scheduler();
+ const Kernel::KScheduler& sched = cores[result.host_handle].Scheduler();
const Kernel::Thread* current = sched.GetCurrentThread();
if (current != nullptr && !current->IsPhantomMode()) {
result.guest_handle = current->GetGlobalHandle();
@@ -314,7 +306,7 @@ struct KernelCore::Impl {
// Lists all processes that exist in the current session.
std::vector<std::shared_ptr<Process>> process_list;
Process* current_process = nullptr;
- Kernel::GlobalScheduler global_scheduler;
+ std::unique_ptr<Kernel::GlobalSchedulerContext> global_scheduler_context;
Kernel::Synchronization synchronization;
Kernel::TimeManager time_manager;
@@ -333,15 +325,8 @@ struct KernelCore::Impl {
std::unique_ptr<Core::ExclusiveMonitor> exclusive_monitor;
std::vector<Kernel::PhysicalCore> cores;
- // 0-3 IDs represent core threads, >3 represent others
- std::atomic<u32> registered_thread_ids{Core::Hardware::NUM_CPU_CORES};
-
- // Number of host threads is a relatively high number to avoid overflowing
- static constexpr size_t NUM_REGISTRABLE_HOST_THREADS = 64;
- std::atomic<size_t> num_host_threads{0};
- std::array<std::atomic<std::thread::id>, NUM_REGISTRABLE_HOST_THREADS>
- register_host_thread_keys{};
- std::array<std::atomic<u32>, NUM_REGISTRABLE_HOST_THREADS> register_host_thread_values{};
+ // Next host thead ID to use, 0-3 IDs represent core threads, >3 represent others
+ std::atomic<u32> next_host_thread_id{Core::Hardware::NUM_CPU_CORES};
// Kernel memory management
std::unique_ptr<Memory::MemoryManager> memory_manager;
@@ -353,12 +338,19 @@ struct KernelCore::Impl {
std::shared_ptr<Kernel::SharedMemory> irs_shared_mem;
std::shared_ptr<Kernel::SharedMemory> time_shared_mem;
+ // Threads used for services
+ std::unordered_set<std::shared_ptr<Kernel::ServiceThread>> service_threads;
+
+ // Service threads are managed by a worker thread, so that a calling service thread can queue up
+ // the release of itself
+ std::unique_ptr<Common::ThreadWorker> service_thread_manager;
+
std::array<std::shared_ptr<Thread>, Core::Hardware::NUM_CPU_CORES> suspend_threads{};
std::array<Core::CPUInterruptHandler, Core::Hardware::NUM_CPU_CORES> interrupts{};
- std::array<std::unique_ptr<Kernel::Scheduler>, Core::Hardware::NUM_CPU_CORES> schedulers{};
+ std::array<std::unique_ptr<Kernel::KScheduler>, Core::Hardware::NUM_CPU_CORES> schedulers{};
bool is_multicore{};
- std::thread::id single_core_thread_id{};
+ u32 single_core_thread_id{};
std::array<u64, Core::Hardware::NUM_CPU_CORES> svc_ticks{};
@@ -415,19 +407,19 @@ const std::vector<std::shared_ptr<Process>>& KernelCore::GetProcessList() const
return impl->process_list;
}
-Kernel::GlobalScheduler& KernelCore::GlobalScheduler() {
- return impl->global_scheduler;
+Kernel::GlobalSchedulerContext& KernelCore::GlobalSchedulerContext() {
+ return *impl->global_scheduler_context;
}
-const Kernel::GlobalScheduler& KernelCore::GlobalScheduler() const {
- return impl->global_scheduler;
+const Kernel::GlobalSchedulerContext& KernelCore::GlobalSchedulerContext() const {
+ return *impl->global_scheduler_context;
}
-Kernel::Scheduler& KernelCore::Scheduler(std::size_t id) {
+Kernel::KScheduler& KernelCore::Scheduler(std::size_t id) {
return *impl->schedulers[id];
}
-const Kernel::Scheduler& KernelCore::Scheduler(std::size_t id) const {
+const Kernel::KScheduler& KernelCore::Scheduler(std::size_t id) const {
return *impl->schedulers[id];
}
@@ -451,16 +443,13 @@ const Kernel::PhysicalCore& KernelCore::CurrentPhysicalCore() const {
return impl->cores[core_id];
}
-Kernel::Scheduler& KernelCore::CurrentScheduler() {
+Kernel::KScheduler* KernelCore::CurrentScheduler() {
u32 core_id = impl->GetCurrentHostThreadID();
- ASSERT(core_id < Core::Hardware::NUM_CPU_CORES);
- return *impl->schedulers[core_id];
-}
-
-const Kernel::Scheduler& KernelCore::CurrentScheduler() const {
- u32 core_id = impl->GetCurrentHostThreadID();
- ASSERT(core_id < Core::Hardware::NUM_CPU_CORES);
- return *impl->schedulers[core_id];
+ if (core_id >= Core::Hardware::NUM_CPU_CORES) {
+ // This is expected when called from not a guest thread
+ return {};
+ }
+ return impl->schedulers[core_id].get();
}
std::array<Core::CPUInterruptHandler, Core::Hardware::NUM_CPU_CORES>& KernelCore::Interrupts() {
@@ -623,7 +612,7 @@ const Kernel::SharedMemory& KernelCore::GetTimeSharedMem() const {
void KernelCore::Suspend(bool in_suspention) {
const bool should_suspend = exception_exited || in_suspention;
{
- SchedulerLock lock(*this);
+ KScopedSchedulerLock lock(*this);
ThreadStatus status = should_suspend ? ThreadStatus::Ready : ThreadStatus::WaitSleep;
for (std::size_t i = 0; i < Core::Hardware::NUM_CPU_CORES; i++) {
impl->suspend_threads[i]->SetStatus(status);
@@ -650,4 +639,19 @@ void KernelCore::ExitSVCProfile() {
MicroProfileLeave(MICROPROFILE_TOKEN(Kernel_SVC), impl->svc_ticks[core]);
}
+std::weak_ptr<Kernel::ServiceThread> KernelCore::CreateServiceThread(const std::string& name) {
+ auto service_thread = std::make_shared<Kernel::ServiceThread>(*this, 1, name);
+ impl->service_thread_manager->QueueWork(
+ [this, service_thread] { impl->service_threads.emplace(service_thread); });
+ return service_thread;
+}
+
+void KernelCore::ReleaseServiceThread(std::weak_ptr<Kernel::ServiceThread> service_thread) {
+ impl->service_thread_manager->QueueWork([this, service_thread] {
+ if (auto strong_ptr = service_thread.lock()) {
+ impl->service_threads.erase(strong_ptr);
+ }
+ });
+}
+
} // namespace Kernel
diff --git a/src/core/hle/kernel/kernel.h b/src/core/hle/kernel/kernel.h
index a73a93039..e3169f5a7 100644
--- a/src/core/hle/kernel/kernel.h
+++ b/src/core/hle/kernel/kernel.h
@@ -35,13 +35,14 @@ class SlabHeap;
class AddressArbiter;
class ClientPort;
-class GlobalScheduler;
+class GlobalSchedulerContext;
class HandleTable;
class PhysicalCore;
class Process;
class ResourceLimit;
-class Scheduler;
+class KScheduler;
class SharedMemory;
+class ServiceThread;
class Synchronization;
class Thread;
class TimeManager;
@@ -102,16 +103,16 @@ public:
const std::vector<std::shared_ptr<Process>>& GetProcessList() const;
/// Gets the sole instance of the global scheduler
- Kernel::GlobalScheduler& GlobalScheduler();
+ Kernel::GlobalSchedulerContext& GlobalSchedulerContext();
/// Gets the sole instance of the global scheduler
- const Kernel::GlobalScheduler& GlobalScheduler() const;
+ const Kernel::GlobalSchedulerContext& GlobalSchedulerContext() const;
/// Gets the sole instance of the Scheduler assoviated with cpu core 'id'
- Kernel::Scheduler& Scheduler(std::size_t id);
+ Kernel::KScheduler& Scheduler(std::size_t id);
/// Gets the sole instance of the Scheduler assoviated with cpu core 'id'
- const Kernel::Scheduler& Scheduler(std::size_t id) const;
+ const Kernel::KScheduler& Scheduler(std::size_t id) const;
/// Gets the an instance of the respective physical CPU core.
Kernel::PhysicalCore& PhysicalCore(std::size_t id);
@@ -120,10 +121,7 @@ public:
const Kernel::PhysicalCore& PhysicalCore(std::size_t id) const;
/// Gets the sole instance of the Scheduler at the current running core.
- Kernel::Scheduler& CurrentScheduler();
-
- /// Gets the sole instance of the Scheduler at the current running core.
- const Kernel::Scheduler& CurrentScheduler() const;
+ Kernel::KScheduler* CurrentScheduler();
/// Gets the an instance of the current physical CPU core.
Kernel::PhysicalCore& CurrentPhysicalCore();
@@ -230,6 +228,22 @@ public:
void ExitSVCProfile();
+ /**
+ * Creates an HLE service thread, which are used to execute service routines asynchronously.
+ * While these are allocated per ServerSession, these need to be owned and managed outside of
+ * ServerSession to avoid a circular dependency.
+ * @param name String name for the ServerSession creating this thread, used for debug purposes.
+ * @returns The a weak pointer newly created service thread.
+ */
+ std::weak_ptr<Kernel::ServiceThread> CreateServiceThread(const std::string& name);
+
+ /**
+ * Releases a HLE service thread, instructing KernelCore to free it. This should be called when
+ * the ServerSession associated with the thread is destroyed.
+ * @param service_thread Service thread to release.
+ */
+ void ReleaseServiceThread(std::weak_ptr<Kernel::ServiceThread> service_thread);
+
private:
friend class Object;
friend class Process;
diff --git a/src/core/hle/kernel/memory/memory_block.h b/src/core/hle/kernel/memory/memory_block.h
index 9d7839d08..37fe19916 100644
--- a/src/core/hle/kernel/memory/memory_block.h
+++ b/src/core/hle/kernel/memory/memory_block.h
@@ -222,9 +222,9 @@ public:
public:
constexpr MemoryBlock() = default;
- constexpr MemoryBlock(VAddr addr, std::size_t num_pages, MemoryState state,
- MemoryPermission perm, MemoryAttribute attribute)
- : addr{addr}, num_pages(num_pages), state{state}, perm{perm}, attribute{attribute} {}
+ constexpr MemoryBlock(VAddr addr_, std::size_t num_pages_, MemoryState state_,
+ MemoryPermission perm_, MemoryAttribute attribute_)
+ : addr{addr_}, num_pages(num_pages_), state{state_}, perm{perm_}, attribute{attribute_} {}
constexpr VAddr GetAddress() const {
return addr;
diff --git a/src/core/hle/kernel/memory/memory_block_manager.h b/src/core/hle/kernel/memory/memory_block_manager.h
index 6e1d41075..f57d1bbcc 100644
--- a/src/core/hle/kernel/memory/memory_block_manager.h
+++ b/src/core/hle/kernel/memory/memory_block_manager.h
@@ -57,8 +57,8 @@ public:
private:
void MergeAdjacent(iterator it, iterator& next_it);
- const VAddr start_addr;
- const VAddr end_addr;
+ [[maybe_unused]] const VAddr start_addr;
+ [[maybe_unused]] const VAddr end_addr;
MemoryBlockTree memory_block_tree;
};
diff --git a/src/core/hle/kernel/mutex.cpp b/src/core/hle/kernel/mutex.cpp
index 8f6c944d1..4f8075e0e 100644
--- a/src/core/hle/kernel/mutex.cpp
+++ b/src/core/hle/kernel/mutex.cpp
@@ -11,11 +11,11 @@
#include "core/core.h"
#include "core/hle/kernel/errors.h"
#include "core/hle/kernel/handle_table.h"
+#include "core/hle/kernel/k_scheduler.h"
#include "core/hle/kernel/kernel.h"
#include "core/hle/kernel/mutex.h"
#include "core/hle/kernel/object.h"
#include "core/hle/kernel/process.h"
-#include "core/hle/kernel/scheduler.h"
#include "core/hle/kernel/thread.h"
#include "core/hle/result.h"
#include "core/memory.h"
@@ -73,9 +73,9 @@ ResultCode Mutex::TryAcquire(VAddr address, Handle holding_thread_handle,
auto& kernel = system.Kernel();
std::shared_ptr<Thread> current_thread =
- SharedFrom(kernel.CurrentScheduler().GetCurrentThread());
+ SharedFrom(kernel.CurrentScheduler()->GetCurrentThread());
{
- SchedulerLock lock(kernel);
+ KScopedSchedulerLock lock(kernel);
// The mutex address must be 4-byte aligned
if ((address % sizeof(u32)) != 0) {
return ERR_INVALID_ADDRESS;
@@ -114,7 +114,7 @@ ResultCode Mutex::TryAcquire(VAddr address, Handle holding_thread_handle,
}
{
- SchedulerLock lock(kernel);
+ KScopedSchedulerLock lock(kernel);
auto* owner = current_thread->GetLockOwner();
if (owner != nullptr) {
owner->RemoveMutexWaiter(current_thread);
@@ -153,10 +153,10 @@ std::pair<ResultCode, std::shared_ptr<Thread>> Mutex::Unlock(std::shared_ptr<Thr
ResultCode Mutex::Release(VAddr address) {
auto& kernel = system.Kernel();
- SchedulerLock lock(kernel);
+ KScopedSchedulerLock lock(kernel);
std::shared_ptr<Thread> current_thread =
- SharedFrom(kernel.CurrentScheduler().GetCurrentThread());
+ SharedFrom(kernel.CurrentScheduler()->GetCurrentThread());
auto [result, new_owner] = Unlock(current_thread, address);
diff --git a/src/core/hle/kernel/physical_core.cpp b/src/core/hle/kernel/physical_core.cpp
index 50aca5752..7fea45f96 100644
--- a/src/core/hle/kernel/physical_core.cpp
+++ b/src/core/hle/kernel/physical_core.cpp
@@ -7,14 +7,14 @@
#include "core/arm/dynarmic/arm_dynarmic_32.h"
#include "core/arm/dynarmic/arm_dynarmic_64.h"
#include "core/core.h"
+#include "core/hle/kernel/k_scheduler.h"
#include "core/hle/kernel/kernel.h"
#include "core/hle/kernel/physical_core.h"
-#include "core/hle/kernel/scheduler.h"
namespace Kernel {
PhysicalCore::PhysicalCore(std::size_t core_index, Core::System& system,
- Kernel::Scheduler& scheduler, Core::CPUInterrupts& interrupts)
+ Kernel::KScheduler& scheduler, Core::CPUInterrupts& interrupts)
: core_index{core_index}, system{system}, scheduler{scheduler},
interrupts{interrupts}, guard{std::make_unique<Common::SpinLock>()} {}
@@ -43,10 +43,6 @@ void PhysicalCore::Idle() {
interrupts[core_index].AwaitInterrupt();
}
-void PhysicalCore::Shutdown() {
- scheduler.Shutdown();
-}
-
bool PhysicalCore::IsInterrupted() const {
return interrupts[core_index].IsInterrupted();
}
diff --git a/src/core/hle/kernel/physical_core.h b/src/core/hle/kernel/physical_core.h
index 37513130a..f2b0911aa 100644
--- a/src/core/hle/kernel/physical_core.h
+++ b/src/core/hle/kernel/physical_core.h
@@ -15,7 +15,7 @@ class SpinLock;
}
namespace Kernel {
-class Scheduler;
+class KScheduler;
} // namespace Kernel
namespace Core {
@@ -28,7 +28,7 @@ namespace Kernel {
class PhysicalCore {
public:
- PhysicalCore(std::size_t core_index, Core::System& system, Kernel::Scheduler& scheduler,
+ PhysicalCore(std::size_t core_index, Core::System& system, Kernel::KScheduler& scheduler,
Core::CPUInterrupts& interrupts);
~PhysicalCore();
@@ -36,7 +36,7 @@ public:
PhysicalCore& operator=(const PhysicalCore&) = delete;
PhysicalCore(PhysicalCore&&) = default;
- PhysicalCore& operator=(PhysicalCore&&) = default;
+ PhysicalCore& operator=(PhysicalCore&&) = delete;
/// Initialize the core for the specified parameters.
void Initialize(bool is_64_bit);
@@ -55,9 +55,6 @@ public:
/// Check if this core is interrupted
bool IsInterrupted() const;
- // Shutdown this physical core.
- void Shutdown();
-
bool IsInitialized() const {
return arm_interface != nullptr;
}
@@ -82,18 +79,18 @@ public:
return core_index;
}
- Kernel::Scheduler& Scheduler() {
+ Kernel::KScheduler& Scheduler() {
return scheduler;
}
- const Kernel::Scheduler& Scheduler() const {
+ const Kernel::KScheduler& Scheduler() const {
return scheduler;
}
private:
const std::size_t core_index;
Core::System& system;
- Kernel::Scheduler& scheduler;
+ Kernel::KScheduler& scheduler;
Core::CPUInterrupts& interrupts;
std::unique_ptr<Common::SpinLock> guard;
std::unique_ptr<Core::ARM_Interface> arm_interface;
diff --git a/src/core/hle/kernel/process.cpp b/src/core/hle/kernel/process.cpp
index b17529dee..b905b486a 100644
--- a/src/core/hle/kernel/process.cpp
+++ b/src/core/hle/kernel/process.cpp
@@ -15,13 +15,13 @@
#include "core/file_sys/program_metadata.h"
#include "core/hle/kernel/code_set.h"
#include "core/hle/kernel/errors.h"
+#include "core/hle/kernel/k_scheduler.h"
#include "core/hle/kernel/kernel.h"
#include "core/hle/kernel/memory/memory_block_manager.h"
#include "core/hle/kernel/memory/page_table.h"
#include "core/hle/kernel/memory/slab_heap.h"
#include "core/hle/kernel/process.h"
#include "core/hle/kernel/resource_limit.h"
-#include "core/hle/kernel/scheduler.h"
#include "core/hle/kernel/thread.h"
#include "core/hle/lock.h"
#include "core/memory.h"
@@ -54,7 +54,7 @@ void SetupMainThread(Core::System& system, Process& owner_process, u32 priority,
auto& kernel = system.Kernel();
// Threads by default are dormant, wake up the main thread so it runs when the scheduler fires
{
- SchedulerLock lock{kernel};
+ KScopedSchedulerLock lock{kernel};
thread->SetStatus(ThreadStatus::Ready);
}
}
@@ -213,7 +213,7 @@ void Process::UnregisterThread(const Thread* thread) {
}
ResultCode Process::ClearSignalState() {
- SchedulerLock lock(system.Kernel());
+ KScopedSchedulerLock lock(system.Kernel());
if (status == ProcessStatus::Exited) {
LOG_ERROR(Kernel, "called on a terminated process instance.");
return ERR_INVALID_STATE;
@@ -314,7 +314,7 @@ void Process::PrepareForTermination() {
if (thread->GetOwnerProcess() != this)
continue;
- if (thread.get() == system.CurrentScheduler().GetCurrentThread())
+ if (thread.get() == kernel.CurrentScheduler()->GetCurrentThread())
continue;
// TODO(Subv): When are the other running/ready threads terminated?
@@ -325,7 +325,7 @@ void Process::PrepareForTermination() {
}
};
- stop_threads(system.GlobalScheduler().GetThreadList());
+ stop_threads(system.GlobalSchedulerContext().GetThreadList());
FreeTLSRegion(tls_region_address);
tls_region_address = 0;
@@ -347,7 +347,7 @@ static auto FindTLSPageWithAvailableSlots(std::vector<TLSPage>& tls_pages) {
}
VAddr Process::CreateTLSRegion() {
- SchedulerLock lock(system.Kernel());
+ KScopedSchedulerLock lock(system.Kernel());
if (auto tls_page_iter{FindTLSPageWithAvailableSlots(tls_pages)};
tls_page_iter != tls_pages.cend()) {
return *tls_page_iter->ReserveSlot();
@@ -378,7 +378,7 @@ VAddr Process::CreateTLSRegion() {
}
void Process::FreeTLSRegion(VAddr tls_address) {
- SchedulerLock lock(system.Kernel());
+ KScopedSchedulerLock lock(system.Kernel());
const VAddr aligned_address = Common::AlignDown(tls_address, Core::Memory::PAGE_SIZE);
auto iter =
std::find_if(tls_pages.begin(), tls_pages.end(), [aligned_address](const auto& page) {
diff --git a/src/core/hle/kernel/process.h b/src/core/hle/kernel/process.h
index f45cb5674..e412e58aa 100644
--- a/src/core/hle/kernel/process.h
+++ b/src/core/hle/kernel/process.h
@@ -216,6 +216,16 @@ public:
total_process_running_time_ticks += ticks;
}
+ /// Gets the process schedule count, used for thread yelding
+ s64 GetScheduledCount() const {
+ return schedule_count;
+ }
+
+ /// Increments the process schedule count, used for thread yielding.
+ void IncrementScheduledCount() {
+ ++schedule_count;
+ }
+
/// Gets 8 bytes of random data for svcGetInfo RandomEntropy
u64 GetRandomEntropy(std::size_t index) const {
return random_entropy.at(index);
@@ -397,6 +407,9 @@ private:
/// Name of this process
std::string name;
+ /// Schedule count of this process
+ s64 schedule_count{};
+
/// System context
Core::System& system;
};
diff --git a/src/core/hle/kernel/process_capability.cpp b/src/core/hle/kernel/process_capability.cpp
index 63880f13d..0f128c586 100644
--- a/src/core/hle/kernel/process_capability.cpp
+++ b/src/core/hle/kernel/process_capability.cpp
@@ -199,7 +199,7 @@ ResultCode ProcessCapabilities::ParseSingleFlagCapability(u32& set_flags, u32& s
break;
}
- LOG_ERROR(Kernel, "Invalid capability type! type={}", static_cast<u32>(type));
+ LOG_ERROR(Kernel, "Invalid capability type! type={}", type);
return ERR_INVALID_CAPABILITY_DESCRIPTOR;
}
diff --git a/src/core/hle/kernel/readable_event.cpp b/src/core/hle/kernel/readable_event.cpp
index 6e286419e..cea262ce0 100644
--- a/src/core/hle/kernel/readable_event.cpp
+++ b/src/core/hle/kernel/readable_event.cpp
@@ -6,10 +6,10 @@
#include "common/assert.h"
#include "common/logging/log.h"
#include "core/hle/kernel/errors.h"
+#include "core/hle/kernel/k_scheduler.h"
#include "core/hle/kernel/kernel.h"
#include "core/hle/kernel/object.h"
#include "core/hle/kernel/readable_event.h"
-#include "core/hle/kernel/scheduler.h"
#include "core/hle/kernel/thread.h"
namespace Kernel {
@@ -39,7 +39,7 @@ void ReadableEvent::Clear() {
}
ResultCode ReadableEvent::Reset() {
- SchedulerLock lock(kernel);
+ KScopedSchedulerLock lock(kernel);
if (!is_signaled) {
LOG_TRACE(Kernel, "Handle is not signaled! object_id={}, object_type={}, object_name={}",
GetObjectId(), GetTypeName(), GetName());
diff --git a/src/core/hle/kernel/resource_limit.cpp b/src/core/hle/kernel/resource_limit.cpp
index 212e442f4..7bf50339d 100644
--- a/src/core/hle/kernel/resource_limit.cpp
+++ b/src/core/hle/kernel/resource_limit.cpp
@@ -65,8 +65,8 @@ ResultCode ResourceLimit::SetLimitValue(ResourceType resource, s64 value) {
limit[index] = value;
return RESULT_SUCCESS;
} else {
- LOG_ERROR(Kernel, "Limit value is too large! resource={}, value={}, index={}",
- static_cast<u32>(resource), value, index);
+ LOG_ERROR(Kernel, "Limit value is too large! resource={}, value={}, index={}", resource,
+ value, index);
return ERR_INVALID_STATE;
}
}
diff --git a/src/core/hle/kernel/scheduler.cpp b/src/core/hle/kernel/scheduler.cpp
deleted file mode 100644
index 5c63b0b4a..000000000
--- a/src/core/hle/kernel/scheduler.cpp
+++ /dev/null
@@ -1,819 +0,0 @@
-// Copyright 2018 yuzu emulator team
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-//
-// SelectThreads, Yield functions originally by TuxSH.
-// licensed under GPLv2 or later under exception provided by the author.
-
-#include <algorithm>
-#include <mutex>
-#include <set>
-#include <unordered_set>
-#include <utility>
-
-#include "common/assert.h"
-#include "common/bit_util.h"
-#include "common/fiber.h"
-#include "common/logging/log.h"
-#include "core/arm/arm_interface.h"
-#include "core/core.h"
-#include "core/core_timing.h"
-#include "core/cpu_manager.h"
-#include "core/hle/kernel/kernel.h"
-#include "core/hle/kernel/physical_core.h"
-#include "core/hle/kernel/process.h"
-#include "core/hle/kernel/scheduler.h"
-#include "core/hle/kernel/time_manager.h"
-
-namespace Kernel {
-
-GlobalScheduler::GlobalScheduler(KernelCore& kernel) : kernel{kernel} {}
-
-GlobalScheduler::~GlobalScheduler() = default;
-
-void GlobalScheduler::AddThread(std::shared_ptr<Thread> thread) {
- std::scoped_lock lock{global_list_guard};
- thread_list.push_back(std::move(thread));
-}
-
-void GlobalScheduler::RemoveThread(std::shared_ptr<Thread> thread) {
- std::scoped_lock lock{global_list_guard};
- thread_list.erase(std::remove(thread_list.begin(), thread_list.end(), thread),
- thread_list.end());
-}
-
-u32 GlobalScheduler::SelectThreads() {
- ASSERT(is_locked);
- const auto update_thread = [](Thread* thread, Scheduler& sched) {
- std::scoped_lock lock{sched.guard};
- if (thread != sched.selected_thread_set.get()) {
- if (thread == nullptr) {
- ++sched.idle_selection_count;
- }
- sched.selected_thread_set = SharedFrom(thread);
- }
- const bool reschedule_pending =
- sched.is_context_switch_pending || (sched.selected_thread_set != sched.current_thread);
- sched.is_context_switch_pending = reschedule_pending;
- std::atomic_thread_fence(std::memory_order_seq_cst);
- return reschedule_pending;
- };
- if (!is_reselection_pending.load()) {
- return 0;
- }
- std::array<Thread*, Core::Hardware::NUM_CPU_CORES> top_threads{};
-
- u32 idle_cores{};
-
- // Step 1: Get top thread in schedule queue.
- for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
- Thread* top_thread =
- scheduled_queue[core].empty() ? nullptr : scheduled_queue[core].front();
- if (top_thread != nullptr) {
- // TODO(Blinkhawk): Implement Thread Pinning
- } else {
- idle_cores |= (1U << core);
- }
- top_threads[core] = top_thread;
- }
-
- while (idle_cores != 0) {
- u32 core_id = Common::CountTrailingZeroes32(idle_cores);
-
- if (!suggested_queue[core_id].empty()) {
- std::array<s32, Core::Hardware::NUM_CPU_CORES> migration_candidates{};
- std::size_t num_candidates = 0;
- auto iter = suggested_queue[core_id].begin();
- Thread* suggested = nullptr;
- // Step 2: Try selecting a suggested thread.
- while (iter != suggested_queue[core_id].end()) {
- suggested = *iter;
- iter++;
- s32 suggested_core_id = suggested->GetProcessorID();
- Thread* top_thread =
- suggested_core_id >= 0 ? top_threads[suggested_core_id] : nullptr;
- if (top_thread != suggested) {
- if (top_thread != nullptr &&
- top_thread->GetPriority() < THREADPRIO_MAX_CORE_MIGRATION) {
- suggested = nullptr;
- break;
- // There's a too high thread to do core migration, cancel
- }
- TransferToCore(suggested->GetPriority(), static_cast<s32>(core_id), suggested);
- break;
- }
- suggested = nullptr;
- migration_candidates[num_candidates++] = suggested_core_id;
- }
- // Step 3: Select a suggested thread from another core
- if (suggested == nullptr) {
- for (std::size_t i = 0; i < num_candidates; i++) {
- s32 candidate_core = migration_candidates[i];
- suggested = top_threads[candidate_core];
- auto it = scheduled_queue[candidate_core].begin();
- it++;
- Thread* next = it != scheduled_queue[candidate_core].end() ? *it : nullptr;
- if (next != nullptr) {
- TransferToCore(suggested->GetPriority(), static_cast<s32>(core_id),
- suggested);
- top_threads[candidate_core] = next;
- break;
- } else {
- suggested = nullptr;
- }
- }
- }
- top_threads[core_id] = suggested;
- }
-
- idle_cores &= ~(1U << core_id);
- }
- u32 cores_needing_context_switch{};
- for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
- Scheduler& sched = kernel.Scheduler(core);
- ASSERT(top_threads[core] == nullptr ||
- static_cast<u32>(top_threads[core]->GetProcessorID()) == core);
- if (update_thread(top_threads[core], sched)) {
- cores_needing_context_switch |= (1U << core);
- }
- }
- return cores_needing_context_switch;
-}
-
-bool GlobalScheduler::YieldThread(Thread* yielding_thread) {
- ASSERT(is_locked);
- // Note: caller should use critical section, etc.
- if (!yielding_thread->IsRunnable()) {
- // Normally this case shouldn't happen except for SetThreadActivity.
- is_reselection_pending.store(true, std::memory_order_release);
- return false;
- }
- const u32 core_id = static_cast<u32>(yielding_thread->GetProcessorID());
- const u32 priority = yielding_thread->GetPriority();
-
- // Yield the thread
- Reschedule(priority, core_id, yielding_thread);
- const Thread* const winner = scheduled_queue[core_id].front();
- if (kernel.GetCurrentHostThreadID() != core_id) {
- is_reselection_pending.store(true, std::memory_order_release);
- }
-
- return AskForReselectionOrMarkRedundant(yielding_thread, winner);
-}
-
-bool GlobalScheduler::YieldThreadAndBalanceLoad(Thread* yielding_thread) {
- ASSERT(is_locked);
- // Note: caller should check if !thread.IsSchedulerOperationRedundant and use critical section,
- // etc.
- if (!yielding_thread->IsRunnable()) {
- // Normally this case shouldn't happen except for SetThreadActivity.
- is_reselection_pending.store(true, std::memory_order_release);
- return false;
- }
- const u32 core_id = static_cast<u32>(yielding_thread->GetProcessorID());
- const u32 priority = yielding_thread->GetPriority();
-
- // Yield the thread
- Reschedule(priority, core_id, yielding_thread);
-
- std::array<Thread*, Core::Hardware::NUM_CPU_CORES> current_threads;
- for (std::size_t i = 0; i < current_threads.size(); i++) {
- current_threads[i] = scheduled_queue[i].empty() ? nullptr : scheduled_queue[i].front();
- }
-
- Thread* next_thread = scheduled_queue[core_id].front(priority);
- Thread* winner = nullptr;
- for (auto& thread : suggested_queue[core_id]) {
- const s32 source_core = thread->GetProcessorID();
- if (source_core >= 0) {
- if (current_threads[source_core] != nullptr) {
- if (thread == current_threads[source_core] ||
- current_threads[source_core]->GetPriority() < min_regular_priority) {
- continue;
- }
- }
- }
- if (next_thread->GetLastRunningTicks() >= thread->GetLastRunningTicks() ||
- next_thread->GetPriority() < thread->GetPriority()) {
- if (thread->GetPriority() <= priority) {
- winner = thread;
- break;
- }
- }
- }
-
- if (winner != nullptr) {
- if (winner != yielding_thread) {
- TransferToCore(winner->GetPriority(), s32(core_id), winner);
- }
- } else {
- winner = next_thread;
- }
-
- if (kernel.GetCurrentHostThreadID() != core_id) {
- is_reselection_pending.store(true, std::memory_order_release);
- }
-
- return AskForReselectionOrMarkRedundant(yielding_thread, winner);
-}
-
-bool GlobalScheduler::YieldThreadAndWaitForLoadBalancing(Thread* yielding_thread) {
- ASSERT(is_locked);
- // Note: caller should check if !thread.IsSchedulerOperationRedundant and use critical section,
- // etc.
- if (!yielding_thread->IsRunnable()) {
- // Normally this case shouldn't happen except for SetThreadActivity.
- is_reselection_pending.store(true, std::memory_order_release);
- return false;
- }
- Thread* winner = nullptr;
- const u32 core_id = static_cast<u32>(yielding_thread->GetProcessorID());
-
- // Remove the thread from its scheduled mlq, put it on the corresponding "suggested" one instead
- TransferToCore(yielding_thread->GetPriority(), -1, yielding_thread);
-
- // If the core is idle, perform load balancing, excluding the threads that have just used this
- // function...
- if (scheduled_queue[core_id].empty()) {
- // Here, "current_threads" is calculated after the ""yield"", unlike yield -1
- std::array<Thread*, Core::Hardware::NUM_CPU_CORES> current_threads;
- for (std::size_t i = 0; i < current_threads.size(); i++) {
- current_threads[i] = scheduled_queue[i].empty() ? nullptr : scheduled_queue[i].front();
- }
- for (auto& thread : suggested_queue[core_id]) {
- const s32 source_core = thread->GetProcessorID();
- if (source_core < 0 || thread == current_threads[source_core]) {
- continue;
- }
- if (current_threads[source_core] == nullptr ||
- current_threads[source_core]->GetPriority() >= min_regular_priority) {
- winner = thread;
- }
- break;
- }
- if (winner != nullptr) {
- if (winner != yielding_thread) {
- TransferToCore(winner->GetPriority(), static_cast<s32>(core_id), winner);
- }
- } else {
- winner = yielding_thread;
- }
- } else {
- winner = scheduled_queue[core_id].front();
- }
-
- if (kernel.GetCurrentHostThreadID() != core_id) {
- is_reselection_pending.store(true, std::memory_order_release);
- }
-
- return AskForReselectionOrMarkRedundant(yielding_thread, winner);
-}
-
-void GlobalScheduler::PreemptThreads() {
- ASSERT(is_locked);
- for (std::size_t core_id = 0; core_id < Core::Hardware::NUM_CPU_CORES; core_id++) {
- const u32 priority = preemption_priorities[core_id];
-
- if (scheduled_queue[core_id].size(priority) > 0) {
- if (scheduled_queue[core_id].size(priority) > 1) {
- scheduled_queue[core_id].front(priority)->IncrementYieldCount();
- }
- scheduled_queue[core_id].yield(priority);
- if (scheduled_queue[core_id].size(priority) > 1) {
- scheduled_queue[core_id].front(priority)->IncrementYieldCount();
- }
- }
-
- Thread* current_thread =
- scheduled_queue[core_id].empty() ? nullptr : scheduled_queue[core_id].front();
- Thread* winner = nullptr;
- for (auto& thread : suggested_queue[core_id]) {
- const s32 source_core = thread->GetProcessorID();
- if (thread->GetPriority() != priority) {
- continue;
- }
- if (source_core >= 0) {
- Thread* next_thread = scheduled_queue[source_core].empty()
- ? nullptr
- : scheduled_queue[source_core].front();
- if (next_thread != nullptr && next_thread->GetPriority() < 2) {
- break;
- }
- if (next_thread == thread) {
- continue;
- }
- }
- if (current_thread != nullptr &&
- current_thread->GetLastRunningTicks() >= thread->GetLastRunningTicks()) {
- winner = thread;
- break;
- }
- }
-
- if (winner != nullptr) {
- TransferToCore(winner->GetPriority(), s32(core_id), winner);
- current_thread =
- winner->GetPriority() <= current_thread->GetPriority() ? winner : current_thread;
- }
-
- if (current_thread != nullptr && current_thread->GetPriority() > priority) {
- for (auto& thread : suggested_queue[core_id]) {
- const s32 source_core = thread->GetProcessorID();
- if (thread->GetPriority() < priority) {
- continue;
- }
- if (source_core >= 0) {
- Thread* next_thread = scheduled_queue[source_core].empty()
- ? nullptr
- : scheduled_queue[source_core].front();
- if (next_thread != nullptr && next_thread->GetPriority() < 2) {
- break;
- }
- if (next_thread == thread) {
- continue;
- }
- }
- if (current_thread != nullptr &&
- current_thread->GetLastRunningTicks() >= thread->GetLastRunningTicks()) {
- winner = thread;
- break;
- }
- }
-
- if (winner != nullptr) {
- TransferToCore(winner->GetPriority(), s32(core_id), winner);
- current_thread = winner;
- }
- }
-
- is_reselection_pending.store(true, std::memory_order_release);
- }
-}
-
-void GlobalScheduler::EnableInterruptAndSchedule(u32 cores_pending_reschedule,
- Core::EmuThreadHandle global_thread) {
- u32 current_core = global_thread.host_handle;
- bool must_context_switch = global_thread.guest_handle != InvalidHandle &&
- (current_core < Core::Hardware::NUM_CPU_CORES);
- while (cores_pending_reschedule != 0) {
- u32 core = Common::CountTrailingZeroes32(cores_pending_reschedule);
- ASSERT(core < Core::Hardware::NUM_CPU_CORES);
- if (!must_context_switch || core != current_core) {
- auto& phys_core = kernel.PhysicalCore(core);
- phys_core.Interrupt();
- } else {
- must_context_switch = true;
- }
- cores_pending_reschedule &= ~(1U << core);
- }
- if (must_context_switch) {
- auto& core_scheduler = kernel.CurrentScheduler();
- kernel.ExitSVCProfile();
- core_scheduler.TryDoContextSwitch();
- kernel.EnterSVCProfile();
- }
-}
-
-void GlobalScheduler::Suggest(u32 priority, std::size_t core, Thread* thread) {
- ASSERT(is_locked);
- suggested_queue[core].add(thread, priority);
-}
-
-void GlobalScheduler::Unsuggest(u32 priority, std::size_t core, Thread* thread) {
- ASSERT(is_locked);
- suggested_queue[core].remove(thread, priority);
-}
-
-void GlobalScheduler::Schedule(u32 priority, std::size_t core, Thread* thread) {
- ASSERT(is_locked);
- ASSERT_MSG(thread->GetProcessorID() == s32(core), "Thread must be assigned to this core.");
- scheduled_queue[core].add(thread, priority);
-}
-
-void GlobalScheduler::SchedulePrepend(u32 priority, std::size_t core, Thread* thread) {
- ASSERT(is_locked);
- ASSERT_MSG(thread->GetProcessorID() == s32(core), "Thread must be assigned to this core.");
- scheduled_queue[core].add(thread, priority, false);
-}
-
-void GlobalScheduler::Reschedule(u32 priority, std::size_t core, Thread* thread) {
- ASSERT(is_locked);
- scheduled_queue[core].remove(thread, priority);
- scheduled_queue[core].add(thread, priority);
-}
-
-void GlobalScheduler::Unschedule(u32 priority, std::size_t core, Thread* thread) {
- ASSERT(is_locked);
- scheduled_queue[core].remove(thread, priority);
-}
-
-void GlobalScheduler::TransferToCore(u32 priority, s32 destination_core, Thread* thread) {
- ASSERT(is_locked);
- const bool schedulable = thread->GetPriority() < THREADPRIO_COUNT;
- const s32 source_core = thread->GetProcessorID();
- if (source_core == destination_core || !schedulable) {
- return;
- }
- thread->SetProcessorID(destination_core);
- if (source_core >= 0) {
- Unschedule(priority, static_cast<u32>(source_core), thread);
- }
- if (destination_core >= 0) {
- Unsuggest(priority, static_cast<u32>(destination_core), thread);
- Schedule(priority, static_cast<u32>(destination_core), thread);
- }
- if (source_core >= 0) {
- Suggest(priority, static_cast<u32>(source_core), thread);
- }
-}
-
-bool GlobalScheduler::AskForReselectionOrMarkRedundant(Thread* current_thread,
- const Thread* winner) {
- if (current_thread == winner) {
- current_thread->IncrementYieldCount();
- return true;
- } else {
- is_reselection_pending.store(true, std::memory_order_release);
- return false;
- }
-}
-
-void GlobalScheduler::AdjustSchedulingOnStatus(Thread* thread, u32 old_flags) {
- if (old_flags == thread->scheduling_state) {
- return;
- }
- ASSERT(is_locked);
-
- if (old_flags == static_cast<u32>(ThreadSchedStatus::Runnable)) {
- // In this case the thread was running, now it's pausing/exitting
- if (thread->processor_id >= 0) {
- Unschedule(thread->current_priority, static_cast<u32>(thread->processor_id), thread);
- }
-
- for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
- if (core != static_cast<u32>(thread->processor_id) &&
- ((thread->affinity_mask >> core) & 1) != 0) {
- Unsuggest(thread->current_priority, core, thread);
- }
- }
- } else if (thread->scheduling_state == static_cast<u32>(ThreadSchedStatus::Runnable)) {
- // The thread is now set to running from being stopped
- if (thread->processor_id >= 0) {
- Schedule(thread->current_priority, static_cast<u32>(thread->processor_id), thread);
- }
-
- for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
- if (core != static_cast<u32>(thread->processor_id) &&
- ((thread->affinity_mask >> core) & 1) != 0) {
- Suggest(thread->current_priority, core, thread);
- }
- }
- }
-
- SetReselectionPending();
-}
-
-void GlobalScheduler::AdjustSchedulingOnPriority(Thread* thread, u32 old_priority) {
- if (thread->scheduling_state != static_cast<u32>(ThreadSchedStatus::Runnable)) {
- return;
- }
- ASSERT(is_locked);
- if (thread->processor_id >= 0) {
- Unschedule(old_priority, static_cast<u32>(thread->processor_id), thread);
- }
-
- for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
- if (core != static_cast<u32>(thread->processor_id) &&
- ((thread->affinity_mask >> core) & 1) != 0) {
- Unsuggest(old_priority, core, thread);
- }
- }
-
- if (thread->processor_id >= 0) {
- if (thread == kernel.CurrentScheduler().GetCurrentThread()) {
- SchedulePrepend(thread->current_priority, static_cast<u32>(thread->processor_id),
- thread);
- } else {
- Schedule(thread->current_priority, static_cast<u32>(thread->processor_id), thread);
- }
- }
-
- for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
- if (core != static_cast<u32>(thread->processor_id) &&
- ((thread->affinity_mask >> core) & 1) != 0) {
- Suggest(thread->current_priority, core, thread);
- }
- }
- thread->IncrementYieldCount();
- SetReselectionPending();
-}
-
-void GlobalScheduler::AdjustSchedulingOnAffinity(Thread* thread, u64 old_affinity_mask,
- s32 old_core) {
- if (thread->scheduling_state != static_cast<u32>(ThreadSchedStatus::Runnable) ||
- thread->current_priority >= THREADPRIO_COUNT) {
- return;
- }
- ASSERT(is_locked);
-
- for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
- if (((old_affinity_mask >> core) & 1) != 0) {
- if (core == static_cast<u32>(old_core)) {
- Unschedule(thread->current_priority, core, thread);
- } else {
- Unsuggest(thread->current_priority, core, thread);
- }
- }
- }
-
- for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
- if (((thread->affinity_mask >> core) & 1) != 0) {
- if (core == static_cast<u32>(thread->processor_id)) {
- Schedule(thread->current_priority, core, thread);
- } else {
- Suggest(thread->current_priority, core, thread);
- }
- }
- }
-
- thread->IncrementYieldCount();
- SetReselectionPending();
-}
-
-void GlobalScheduler::Shutdown() {
- for (std::size_t core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
- scheduled_queue[core].clear();
- suggested_queue[core].clear();
- }
- thread_list.clear();
-}
-
-void GlobalScheduler::Lock() {
- Core::EmuThreadHandle current_thread = kernel.GetCurrentEmuThreadID();
- ASSERT(!current_thread.IsInvalid());
- if (current_thread == current_owner) {
- ++scope_lock;
- } else {
- inner_lock.lock();
- is_locked = true;
- current_owner = current_thread;
- ASSERT(current_owner != Core::EmuThreadHandle::InvalidHandle());
- scope_lock = 1;
- }
-}
-
-void GlobalScheduler::Unlock() {
- if (--scope_lock != 0) {
- ASSERT(scope_lock > 0);
- return;
- }
- u32 cores_pending_reschedule = SelectThreads();
- Core::EmuThreadHandle leaving_thread = current_owner;
- current_owner = Core::EmuThreadHandle::InvalidHandle();
- scope_lock = 1;
- is_locked = false;
- inner_lock.unlock();
- EnableInterruptAndSchedule(cores_pending_reschedule, leaving_thread);
-}
-
-Scheduler::Scheduler(Core::System& system, std::size_t core_id) : system(system), core_id(core_id) {
- switch_fiber = std::make_shared<Common::Fiber>(std::function<void(void*)>(OnSwitch), this);
-}
-
-Scheduler::~Scheduler() = default;
-
-bool Scheduler::HaveReadyThreads() const {
- return system.GlobalScheduler().HaveReadyThreads(core_id);
-}
-
-Thread* Scheduler::GetCurrentThread() const {
- if (current_thread) {
- return current_thread.get();
- }
- return idle_thread.get();
-}
-
-Thread* Scheduler::GetSelectedThread() const {
- return selected_thread.get();
-}
-
-u64 Scheduler::GetLastContextSwitchTicks() const {
- return last_context_switch_time;
-}
-
-void Scheduler::TryDoContextSwitch() {
- auto& phys_core = system.Kernel().CurrentPhysicalCore();
- if (phys_core.IsInterrupted()) {
- phys_core.ClearInterrupt();
- }
- guard.lock();
- if (is_context_switch_pending) {
- SwitchContext();
- } else {
- guard.unlock();
- }
-}
-
-void Scheduler::OnThreadStart() {
- SwitchContextStep2();
-}
-
-void Scheduler::Unload(Thread* thread) {
- if (thread) {
- thread->last_running_ticks = system.CoreTiming().GetCPUTicks();
- thread->SetIsRunning(false);
- if (thread->IsContinuousOnSVC() && !thread->IsHLEThread()) {
- system.ArmInterface(core_id).ExceptionalExit();
- thread->SetContinuousOnSVC(false);
- }
- if (!thread->IsHLEThread() && !thread->HasExited()) {
- Core::ARM_Interface& cpu_core = system.ArmInterface(core_id);
- cpu_core.SaveContext(thread->GetContext32());
- cpu_core.SaveContext(thread->GetContext64());
- // Save the TPIDR_EL0 system register in case it was modified.
- thread->SetTPIDR_EL0(cpu_core.GetTPIDR_EL0());
- cpu_core.ClearExclusiveState();
- }
- thread->context_guard.unlock();
- }
-}
-
-void Scheduler::Unload() {
- Unload(current_thread.get());
-}
-
-void Scheduler::Reload(Thread* thread) {
- if (thread) {
- ASSERT_MSG(thread->GetSchedulingStatus() == ThreadSchedStatus::Runnable,
- "Thread must be runnable.");
-
- // Cancel any outstanding wakeup events for this thread
- thread->SetIsRunning(true);
- thread->SetWasRunning(false);
- thread->last_running_ticks = system.CoreTiming().GetCPUTicks();
-
- auto* const thread_owner_process = thread->GetOwnerProcess();
- if (thread_owner_process != nullptr) {
- system.Kernel().MakeCurrentProcess(thread_owner_process);
- }
- if (!thread->IsHLEThread()) {
- Core::ARM_Interface& cpu_core = system.ArmInterface(core_id);
- cpu_core.LoadContext(thread->GetContext32());
- cpu_core.LoadContext(thread->GetContext64());
- cpu_core.SetTlsAddress(thread->GetTLSAddress());
- cpu_core.SetTPIDR_EL0(thread->GetTPIDR_EL0());
- cpu_core.ClearExclusiveState();
- }
- }
-}
-
-void Scheduler::Reload() {
- Reload(current_thread.get());
-}
-
-void Scheduler::SwitchContextStep2() {
- // Load context of new thread
- Reload(selected_thread.get());
-
- TryDoContextSwitch();
-}
-
-void Scheduler::SwitchContext() {
- current_thread_prev = current_thread;
- selected_thread = selected_thread_set;
- Thread* previous_thread = current_thread_prev.get();
- Thread* new_thread = selected_thread.get();
- current_thread = selected_thread;
-
- is_context_switch_pending = false;
-
- if (new_thread == previous_thread) {
- guard.unlock();
- return;
- }
-
- Process* const previous_process = system.Kernel().CurrentProcess();
-
- UpdateLastContextSwitchTime(previous_thread, previous_process);
-
- // Save context for previous thread
- Unload(previous_thread);
-
- std::shared_ptr<Common::Fiber>* old_context;
- if (previous_thread != nullptr) {
- old_context = &previous_thread->GetHostContext();
- } else {
- old_context = &idle_thread->GetHostContext();
- }
- guard.unlock();
-
- Common::Fiber::YieldTo(*old_context, switch_fiber);
- /// When a thread wakes up, the scheduler may have changed to other in another core.
- auto& next_scheduler = system.Kernel().CurrentScheduler();
- next_scheduler.SwitchContextStep2();
-}
-
-void Scheduler::OnSwitch(void* this_scheduler) {
- Scheduler* sched = static_cast<Scheduler*>(this_scheduler);
- sched->SwitchToCurrent();
-}
-
-void Scheduler::SwitchToCurrent() {
- while (true) {
- {
- std::scoped_lock lock{guard};
- selected_thread = selected_thread_set;
- current_thread = selected_thread;
- is_context_switch_pending = false;
- }
- const auto is_switch_pending = [this] {
- std::scoped_lock lock{guard};
- return is_context_switch_pending;
- };
- do {
- if (current_thread != nullptr && !current_thread->IsHLEThread()) {
- current_thread->context_guard.lock();
- if (!current_thread->IsRunnable()) {
- current_thread->context_guard.unlock();
- break;
- }
- if (static_cast<u32>(current_thread->GetProcessorID()) != core_id) {
- current_thread->context_guard.unlock();
- break;
- }
- }
- std::shared_ptr<Common::Fiber>* next_context;
- if (current_thread != nullptr) {
- next_context = &current_thread->GetHostContext();
- } else {
- next_context = &idle_thread->GetHostContext();
- }
- Common::Fiber::YieldTo(switch_fiber, *next_context);
- } while (!is_switch_pending());
- }
-}
-
-void Scheduler::UpdateLastContextSwitchTime(Thread* thread, Process* process) {
- const u64 prev_switch_ticks = last_context_switch_time;
- const u64 most_recent_switch_ticks = system.CoreTiming().GetCPUTicks();
- const u64 update_ticks = most_recent_switch_ticks - prev_switch_ticks;
-
- if (thread != nullptr) {
- thread->UpdateCPUTimeTicks(update_ticks);
- }
-
- if (process != nullptr) {
- process->UpdateCPUTimeTicks(update_ticks);
- }
-
- last_context_switch_time = most_recent_switch_ticks;
-}
-
-void Scheduler::Initialize() {
- std::string name = "Idle Thread Id:" + std::to_string(core_id);
- std::function<void(void*)> init_func = Core::CpuManager::GetIdleThreadStartFunc();
- void* init_func_parameter = system.GetCpuManager().GetStartFuncParamater();
- ThreadType type = static_cast<ThreadType>(THREADTYPE_KERNEL | THREADTYPE_HLE | THREADTYPE_IDLE);
- auto thread_res = Thread::Create(system, type, name, 0, 64, 0, static_cast<u32>(core_id), 0,
- nullptr, std::move(init_func), init_func_parameter);
- idle_thread = std::move(thread_res).Unwrap();
-}
-
-void Scheduler::Shutdown() {
- current_thread = nullptr;
- selected_thread = nullptr;
-}
-
-SchedulerLock::SchedulerLock(KernelCore& kernel) : kernel{kernel} {
- kernel.GlobalScheduler().Lock();
-}
-
-SchedulerLock::~SchedulerLock() {
- kernel.GlobalScheduler().Unlock();
-}
-
-SchedulerLockAndSleep::SchedulerLockAndSleep(KernelCore& kernel, Handle& event_handle,
- Thread* time_task, s64 nanoseconds)
- : SchedulerLock{kernel}, event_handle{event_handle}, time_task{time_task}, nanoseconds{
- nanoseconds} {
- event_handle = InvalidHandle;
-}
-
-SchedulerLockAndSleep::~SchedulerLockAndSleep() {
- if (sleep_cancelled) {
- return;
- }
- auto& time_manager = kernel.TimeManager();
- time_manager.ScheduleTimeEvent(event_handle, time_task, nanoseconds);
-}
-
-void SchedulerLockAndSleep::Release() {
- if (sleep_cancelled) {
- return;
- }
- auto& time_manager = kernel.TimeManager();
- time_manager.ScheduleTimeEvent(event_handle, time_task, nanoseconds);
- sleep_cancelled = true;
-}
-
-} // namespace Kernel
diff --git a/src/core/hle/kernel/scheduler.h b/src/core/hle/kernel/scheduler.h
deleted file mode 100644
index 68db4a5ef..000000000
--- a/src/core/hle/kernel/scheduler.h
+++ /dev/null
@@ -1,320 +0,0 @@
-// Copyright 2018 yuzu emulator team
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <atomic>
-#include <memory>
-#include <mutex>
-#include <vector>
-
-#include "common/common_types.h"
-#include "common/multi_level_queue.h"
-#include "common/spin_lock.h"
-#include "core/hardware_properties.h"
-#include "core/hle/kernel/thread.h"
-
-namespace Common {
-class Fiber;
-}
-
-namespace Core {
-class ARM_Interface;
-class System;
-} // namespace Core
-
-namespace Kernel {
-
-class KernelCore;
-class Process;
-class SchedulerLock;
-
-class GlobalScheduler final {
-public:
- explicit GlobalScheduler(KernelCore& kernel);
- ~GlobalScheduler();
-
- /// Adds a new thread to the scheduler
- void AddThread(std::shared_ptr<Thread> thread);
-
- /// Removes a thread from the scheduler
- void RemoveThread(std::shared_ptr<Thread> thread);
-
- /// Returns a list of all threads managed by the scheduler
- const std::vector<std::shared_ptr<Thread>>& GetThreadList() const {
- return thread_list;
- }
-
- /// Notify the scheduler a thread's status has changed.
- void AdjustSchedulingOnStatus(Thread* thread, u32 old_flags);
-
- /// Notify the scheduler a thread's priority has changed.
- void AdjustSchedulingOnPriority(Thread* thread, u32 old_priority);
-
- /// Notify the scheduler a thread's core and/or affinity mask has changed.
- void AdjustSchedulingOnAffinity(Thread* thread, u64 old_affinity_mask, s32 old_core);
-
- /**
- * Takes care of selecting the new scheduled threads in three steps:
- *
- * 1. First a thread is selected from the top of the priority queue. If no thread
- * is obtained then we move to step two, else we are done.
- *
- * 2. Second we try to get a suggested thread that's not assigned to any core or
- * that is not the top thread in that core.
- *
- * 3. Third is no suggested thread is found, we do a second pass and pick a running
- * thread in another core and swap it with its current thread.
- *
- * returns the cores needing scheduling.
- */
- u32 SelectThreads();
-
- bool HaveReadyThreads(std::size_t core_id) const {
- return !scheduled_queue[core_id].empty();
- }
-
- /**
- * Takes a thread and moves it to the back of the it's priority list.
- *
- * @note This operation can be redundant and no scheduling is changed if marked as so.
- */
- bool YieldThread(Thread* thread);
-
- /**
- * Takes a thread and moves it to the back of the it's priority list.
- * Afterwards, tries to pick a suggested thread from the suggested queue that has worse time or
- * a better priority than the next thread in the core.
- *
- * @note This operation can be redundant and no scheduling is changed if marked as so.
- */
- bool YieldThreadAndBalanceLoad(Thread* thread);
-
- /**
- * Takes a thread and moves it out of the scheduling queue.
- * and into the suggested queue. If no thread can be scheduled afterwards in that core,
- * a suggested thread is obtained instead.
- *
- * @note This operation can be redundant and no scheduling is changed if marked as so.
- */
- bool YieldThreadAndWaitForLoadBalancing(Thread* thread);
-
- /**
- * Rotates the scheduling queues of threads at a preemption priority and then does
- * some core rebalancing. Preemption priorities can be found in the array
- * 'preemption_priorities'.
- *
- * @note This operation happens every 10ms.
- */
- void PreemptThreads();
-
- u32 CpuCoresCount() const {
- return Core::Hardware::NUM_CPU_CORES;
- }
-
- void SetReselectionPending() {
- is_reselection_pending.store(true, std::memory_order_release);
- }
-
- bool IsReselectionPending() const {
- return is_reselection_pending.load(std::memory_order_acquire);
- }
-
- void Shutdown();
-
-private:
- friend class SchedulerLock;
-
- /// Lock the scheduler to the current thread.
- void Lock();
-
- /// Unlocks the scheduler, reselects threads, interrupts cores for rescheduling
- /// and reschedules current core if needed.
- void Unlock();
-
- void EnableInterruptAndSchedule(u32 cores_pending_reschedule,
- Core::EmuThreadHandle global_thread);
-
- /**
- * Add a thread to the suggested queue of a cpu core. Suggested threads may be
- * picked if no thread is scheduled to run on the core.
- */
- void Suggest(u32 priority, std::size_t core, Thread* thread);
-
- /**
- * Remove a thread to the suggested queue of a cpu core. Suggested threads may be
- * picked if no thread is scheduled to run on the core.
- */
- void Unsuggest(u32 priority, std::size_t core, Thread* thread);
-
- /**
- * Add a thread to the scheduling queue of a cpu core. The thread is added at the
- * back the queue in its priority level.
- */
- void Schedule(u32 priority, std::size_t core, Thread* thread);
-
- /**
- * Add a thread to the scheduling queue of a cpu core. The thread is added at the
- * front the queue in its priority level.
- */
- void SchedulePrepend(u32 priority, std::size_t core, Thread* thread);
-
- /// Reschedule an already scheduled thread based on a new priority
- void Reschedule(u32 priority, std::size_t core, Thread* thread);
-
- /// Unschedules a thread.
- void Unschedule(u32 priority, std::size_t core, Thread* thread);
-
- /**
- * Transfers a thread into an specific core. If the destination_core is -1
- * it will be unscheduled from its source code and added into its suggested
- * queue.
- */
- void TransferToCore(u32 priority, s32 destination_core, Thread* thread);
-
- bool AskForReselectionOrMarkRedundant(Thread* current_thread, const Thread* winner);
-
- static constexpr u32 min_regular_priority = 2;
- std::array<Common::MultiLevelQueue<Thread*, THREADPRIO_COUNT>, Core::Hardware::NUM_CPU_CORES>
- scheduled_queue;
- std::array<Common::MultiLevelQueue<Thread*, THREADPRIO_COUNT>, Core::Hardware::NUM_CPU_CORES>
- suggested_queue;
- std::atomic<bool> is_reselection_pending{false};
-
- // The priority levels at which the global scheduler preempts threads every 10 ms. They are
- // ordered from Core 0 to Core 3.
- std::array<u32, Core::Hardware::NUM_CPU_CORES> preemption_priorities = {59, 59, 59, 62};
-
- /// Scheduler lock mechanisms.
- bool is_locked{};
- std::mutex inner_lock;
- std::atomic<s64> scope_lock{};
- Core::EmuThreadHandle current_owner{Core::EmuThreadHandle::InvalidHandle()};
-
- Common::SpinLock global_list_guard{};
-
- /// Lists all thread ids that aren't deleted/etc.
- std::vector<std::shared_ptr<Thread>> thread_list;
- KernelCore& kernel;
-};
-
-class Scheduler final {
-public:
- explicit Scheduler(Core::System& system, std::size_t core_id);
- ~Scheduler();
-
- /// Returns whether there are any threads that are ready to run.
- bool HaveReadyThreads() const;
-
- /// Reschedules to the next available thread (call after current thread is suspended)
- void TryDoContextSwitch();
-
- /// The next two are for SingleCore Only.
- /// Unload current thread before preempting core.
- void Unload(Thread* thread);
- void Unload();
- /// Reload current thread after core preemption.
- void Reload(Thread* thread);
- void Reload();
-
- /// Gets the current running thread
- Thread* GetCurrentThread() const;
-
- /// Gets the currently selected thread from the top of the multilevel queue
- Thread* GetSelectedThread() const;
-
- /// Gets the timestamp for the last context switch in ticks.
- u64 GetLastContextSwitchTicks() const;
-
- bool ContextSwitchPending() const {
- return is_context_switch_pending;
- }
-
- void Initialize();
-
- /// Shutdowns the scheduler.
- void Shutdown();
-
- void OnThreadStart();
-
- std::shared_ptr<Common::Fiber>& ControlContext() {
- return switch_fiber;
- }
-
- const std::shared_ptr<Common::Fiber>& ControlContext() const {
- return switch_fiber;
- }
-
-private:
- friend class GlobalScheduler;
-
- /// Switches the CPU's active thread context to that of the specified thread
- void SwitchContext();
-
- /// When a thread wakes up, it must run this through it's new scheduler
- void SwitchContextStep2();
-
- /**
- * Called on every context switch to update the internal timestamp
- * This also updates the running time ticks for the given thread and
- * process using the following difference:
- *
- * ticks += most_recent_ticks - last_context_switch_ticks
- *
- * The internal tick timestamp for the scheduler is simply the
- * most recent tick count retrieved. No special arithmetic is
- * applied to it.
- */
- void UpdateLastContextSwitchTime(Thread* thread, Process* process);
-
- static void OnSwitch(void* this_scheduler);
- void SwitchToCurrent();
-
- std::shared_ptr<Thread> current_thread = nullptr;
- std::shared_ptr<Thread> selected_thread = nullptr;
- std::shared_ptr<Thread> current_thread_prev = nullptr;
- std::shared_ptr<Thread> selected_thread_set = nullptr;
- std::shared_ptr<Thread> idle_thread = nullptr;
-
- std::shared_ptr<Common::Fiber> switch_fiber = nullptr;
-
- Core::System& system;
- u64 last_context_switch_time = 0;
- u64 idle_selection_count = 0;
- const std::size_t core_id;
-
- Common::SpinLock guard{};
-
- bool is_context_switch_pending = false;
-};
-
-class SchedulerLock {
-public:
- [[nodiscard]] explicit SchedulerLock(KernelCore& kernel);
- ~SchedulerLock();
-
-protected:
- KernelCore& kernel;
-};
-
-class SchedulerLockAndSleep : public SchedulerLock {
-public:
- explicit SchedulerLockAndSleep(KernelCore& kernel, Handle& event_handle, Thread* time_task,
- s64 nanoseconds);
- ~SchedulerLockAndSleep();
-
- void CancelSleep() {
- sleep_cancelled = true;
- }
-
- void Release();
-
-private:
- Handle& event_handle;
- Thread* time_task;
- s64 nanoseconds;
- bool sleep_cancelled{};
-};
-
-} // namespace Kernel
diff --git a/src/core/hle/kernel/server_session.cpp b/src/core/hle/kernel/server_session.cpp
index 8c19f2534..b40fe3916 100644
--- a/src/core/hle/kernel/server_session.cpp
+++ b/src/core/hle/kernel/server_session.cpp
@@ -14,9 +14,9 @@
#include "core/hle/kernel/client_session.h"
#include "core/hle/kernel/handle_table.h"
#include "core/hle/kernel/hle_ipc.h"
+#include "core/hle/kernel/k_scheduler.h"
#include "core/hle/kernel/kernel.h"
#include "core/hle/kernel/process.h"
-#include "core/hle/kernel/scheduler.h"
#include "core/hle/kernel/server_session.h"
#include "core/hle/kernel/session.h"
#include "core/hle/kernel/thread.h"
@@ -25,19 +25,19 @@
namespace Kernel {
ServerSession::ServerSession(KernelCore& kernel) : SynchronizationObject{kernel} {}
-ServerSession::~ServerSession() = default;
+
+ServerSession::~ServerSession() {
+ kernel.ReleaseServiceThread(service_thread);
+}
ResultVal<std::shared_ptr<ServerSession>> ServerSession::Create(KernelCore& kernel,
std::shared_ptr<Session> parent,
std::string name) {
std::shared_ptr<ServerSession> session{std::make_shared<ServerSession>(kernel)};
- session->request_event =
- Core::Timing::CreateEvent(name, [session](std::uintptr_t, std::chrono::nanoseconds) {
- session->CompleteSyncRequest();
- });
session->name = std::move(name);
session->parent = std::move(parent);
+ session->service_thread = kernel.CreateServiceThread(session->name);
return MakeResult(std::move(session));
}
@@ -130,8 +130,7 @@ ResultCode ServerSession::HandleDomainSyncRequest(Kernel::HLERequestContext& con
}
}
- LOG_CRITICAL(IPC, "Unknown domain command={}",
- static_cast<int>(domain_message_header.command.Value()));
+ LOG_CRITICAL(IPC, "Unknown domain command={}", domain_message_header.command.Value());
ASSERT(false);
return RESULT_SUCCESS;
}
@@ -143,16 +142,16 @@ ResultCode ServerSession::QueueSyncRequest(std::shared_ptr<Thread> thread,
std::make_shared<HLERequestContext>(kernel, memory, SharedFrom(this), std::move(thread));
context->PopulateFromIncomingCommandBuffer(kernel.CurrentProcess()->GetHandleTable(), cmd_buf);
- request_queue.Push(std::move(context));
+
+ if (auto strong_ptr = service_thread.lock()) {
+ strong_ptr->QueueSyncRequest(*this, std::move(context));
+ return RESULT_SUCCESS;
+ }
return RESULT_SUCCESS;
}
-ResultCode ServerSession::CompleteSyncRequest() {
- ASSERT(!request_queue.Empty());
-
- auto& context = *request_queue.Front();
-
+ResultCode ServerSession::CompleteSyncRequest(HLERequestContext& context) {
ResultCode result = RESULT_SUCCESS;
// If the session has been converted to a domain, handle the domain request
if (IsDomain() && context.HasDomainMessageHeader()) {
@@ -171,25 +170,20 @@ ResultCode ServerSession::CompleteSyncRequest() {
// Some service requests require the thread to block
{
- SchedulerLock lock(kernel);
+ KScopedSchedulerLock lock(kernel);
if (!context.IsThreadWaiting()) {
context.GetThread().ResumeFromWait();
context.GetThread().SetSynchronizationResults(nullptr, result);
}
}
- request_queue.Pop();
-
return result;
}
ResultCode ServerSession::HandleSyncRequest(std::shared_ptr<Thread> thread,
Core::Memory::Memory& memory,
Core::Timing::CoreTiming& core_timing) {
- const ResultCode result = QueueSyncRequest(std::move(thread), memory);
- const auto delay = std::chrono::nanoseconds{kernel.IsMulticore() ? 0 : 20000};
- core_timing.ScheduleEvent(delay, request_event, {});
- return result;
+ return QueueSyncRequest(std::move(thread), memory);
}
} // namespace Kernel
diff --git a/src/core/hle/kernel/server_session.h b/src/core/hle/kernel/server_session.h
index d23e9ec68..e8d1d99ea 100644
--- a/src/core/hle/kernel/server_session.h
+++ b/src/core/hle/kernel/server_session.h
@@ -10,6 +10,7 @@
#include <vector>
#include "common/threadsafe_queue.h"
+#include "core/hle/kernel/service_thread.h"
#include "core/hle/kernel/synchronization_object.h"
#include "core/hle/result.h"
@@ -43,6 +44,8 @@ class Thread;
* TLS buffer and control is transferred back to it.
*/
class ServerSession final : public SynchronizationObject {
+ friend class ServiceThread;
+
public:
explicit ServerSession(KernelCore& kernel);
~ServerSession() override;
@@ -132,7 +135,7 @@ private:
ResultCode QueueSyncRequest(std::shared_ptr<Thread> thread, Core::Memory::Memory& memory);
/// Completes a sync request from the emulated application.
- ResultCode CompleteSyncRequest();
+ ResultCode CompleteSyncRequest(HLERequestContext& context);
/// Handles a SyncRequest to a domain, forwarding the request to the proper object or closing an
/// object handle.
@@ -163,11 +166,8 @@ private:
/// The name of this session (optional)
std::string name;
- /// Core timing event used to schedule the service request at some point in the future
- std::shared_ptr<Core::Timing::EventType> request_event;
-
- /// Queue of scheduled service requests
- Common::MPSCQueue<std::shared_ptr<Kernel::HLERequestContext>> request_queue;
+ /// Thread to dispatch service requests
+ std::weak_ptr<ServiceThread> service_thread;
};
} // namespace Kernel
diff --git a/src/core/hle/kernel/service_thread.cpp b/src/core/hle/kernel/service_thread.cpp
new file mode 100644
index 000000000..ee46f3e21
--- /dev/null
+++ b/src/core/hle/kernel/service_thread.cpp
@@ -0,0 +1,110 @@
+// Copyright 2020 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <condition_variable>
+#include <functional>
+#include <mutex>
+#include <thread>
+#include <vector>
+#include <queue>
+
+#include "common/assert.h"
+#include "common/scope_exit.h"
+#include "common/thread.h"
+#include "core/core.h"
+#include "core/hle/kernel/kernel.h"
+#include "core/hle/kernel/server_session.h"
+#include "core/hle/kernel/service_thread.h"
+#include "core/hle/lock.h"
+#include "video_core/renderer_base.h"
+
+namespace Kernel {
+
+class ServiceThread::Impl final {
+public:
+ explicit Impl(KernelCore& kernel, std::size_t num_threads, const std::string& name);
+ ~Impl();
+
+ void QueueSyncRequest(ServerSession& session, std::shared_ptr<HLERequestContext>&& context);
+
+private:
+ std::vector<std::thread> threads;
+ std::queue<std::function<void()>> requests;
+ std::mutex queue_mutex;
+ std::condition_variable condition;
+ const std::string service_name;
+ bool stop{};
+};
+
+ServiceThread::Impl::Impl(KernelCore& kernel, std::size_t num_threads, const std::string& name)
+ : service_name{name} {
+ for (std::size_t i = 0; i < num_threads; ++i)
+ threads.emplace_back([this, &kernel] {
+ Common::SetCurrentThreadName(std::string{"yuzu:HleService:" + service_name}.c_str());
+
+ // Wait for first request before trying to acquire a render context
+ {
+ std::unique_lock lock{queue_mutex};
+ condition.wait(lock, [this] { return stop || !requests.empty(); });
+ }
+
+ kernel.RegisterHostThread();
+
+ while (true) {
+ std::function<void()> task;
+
+ {
+ std::unique_lock lock{queue_mutex};
+ condition.wait(lock, [this] { return stop || !requests.empty(); });
+ if (stop || requests.empty()) {
+ return;
+ }
+ task = std::move(requests.front());
+ requests.pop();
+ }
+
+ task();
+ }
+ });
+}
+
+void ServiceThread::Impl::QueueSyncRequest(ServerSession& session,
+ std::shared_ptr<HLERequestContext>&& context) {
+ {
+ std::unique_lock lock{queue_mutex};
+
+ // ServerSession owns the service thread, so we cannot caption a strong pointer here in the
+ // event that the ServerSession is terminated.
+ std::weak_ptr<ServerSession> weak_ptr{SharedFrom(&session)};
+ requests.emplace([weak_ptr, context{std::move(context)}]() {
+ if (auto strong_ptr = weak_ptr.lock()) {
+ strong_ptr->CompleteSyncRequest(*context);
+ }
+ });
+ }
+ condition.notify_one();
+}
+
+ServiceThread::Impl::~Impl() {
+ {
+ std::unique_lock lock{queue_mutex};
+ stop = true;
+ }
+ condition.notify_all();
+ for (std::thread& thread : threads) {
+ thread.join();
+ }
+}
+
+ServiceThread::ServiceThread(KernelCore& kernel, std::size_t num_threads, const std::string& name)
+ : impl{std::make_unique<Impl>(kernel, num_threads, name)} {}
+
+ServiceThread::~ServiceThread() = default;
+
+void ServiceThread::QueueSyncRequest(ServerSession& session,
+ std::shared_ptr<HLERequestContext>&& context) {
+ impl->QueueSyncRequest(session, std::move(context));
+}
+
+} // namespace Kernel
diff --git a/src/core/hle/kernel/service_thread.h b/src/core/hle/kernel/service_thread.h
new file mode 100644
index 000000000..025ab8fb5
--- /dev/null
+++ b/src/core/hle/kernel/service_thread.h
@@ -0,0 +1,28 @@
+// Copyright 2020 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <memory>
+#include <string>
+
+namespace Kernel {
+
+class HLERequestContext;
+class KernelCore;
+class ServerSession;
+
+class ServiceThread final {
+public:
+ explicit ServiceThread(KernelCore& kernel, std::size_t num_threads, const std::string& name);
+ ~ServiceThread();
+
+ void QueueSyncRequest(ServerSession& session, std::shared_ptr<HLERequestContext>&& context);
+
+private:
+ class Impl;
+ std::unique_ptr<Impl> impl;
+};
+
+} // namespace Kernel
diff --git a/src/core/hle/kernel/svc.cpp b/src/core/hle/kernel/svc.cpp
index 95d6e2b4d..de3ed25da 100644
--- a/src/core/hle/kernel/svc.cpp
+++ b/src/core/hle/kernel/svc.cpp
@@ -24,6 +24,8 @@
#include "core/hle/kernel/client_session.h"
#include "core/hle/kernel/errors.h"
#include "core/hle/kernel/handle_table.h"
+#include "core/hle/kernel/k_scheduler.h"
+#include "core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h"
#include "core/hle/kernel/kernel.h"
#include "core/hle/kernel/memory/memory_block.h"
#include "core/hle/kernel/memory/page_table.h"
@@ -32,7 +34,6 @@
#include "core/hle/kernel/process.h"
#include "core/hle/kernel/readable_event.h"
#include "core/hle/kernel/resource_limit.h"
-#include "core/hle/kernel/scheduler.h"
#include "core/hle/kernel/shared_memory.h"
#include "core/hle/kernel/svc.h"
#include "core/hle/kernel/svc_types.h"
@@ -234,8 +235,7 @@ static ResultCode SetMemoryAttribute(Core::System& system, VAddr address, u64 si
static ResultCode SetMemoryAttribute32(Core::System& system, u32 address, u32 size, u32 mask,
u32 attribute) {
- return SetMemoryAttribute(system, static_cast<VAddr>(address), static_cast<std::size_t>(size),
- mask, attribute);
+ return SetMemoryAttribute(system, address, size, mask, attribute);
}
/// Maps a memory range into a different range.
@@ -255,8 +255,7 @@ static ResultCode MapMemory(Core::System& system, VAddr dst_addr, VAddr src_addr
}
static ResultCode MapMemory32(Core::System& system, u32 dst_addr, u32 src_addr, u32 size) {
- return MapMemory(system, static_cast<VAddr>(dst_addr), static_cast<VAddr>(src_addr),
- static_cast<std::size_t>(size));
+ return MapMemory(system, dst_addr, src_addr, size);
}
/// Unmaps a region that was previously mapped with svcMapMemory
@@ -276,8 +275,7 @@ static ResultCode UnmapMemory(Core::System& system, VAddr dst_addr, VAddr src_ad
}
static ResultCode UnmapMemory32(Core::System& system, u32 dst_addr, u32 src_addr, u32 size) {
- return UnmapMemory(system, static_cast<VAddr>(dst_addr), static_cast<VAddr>(src_addr),
- static_cast<std::size_t>(size));
+ return UnmapMemory(system, dst_addr, src_addr, size);
}
/// Connect to an OS service given the port name, returns the handle to the port to out
@@ -332,7 +330,8 @@ static ResultCode ConnectToNamedPort32(Core::System& system, Handle* out_handle,
/// Makes a blocking IPC call to an OS service.
static ResultCode SendSyncRequest(Core::System& system, Handle handle) {
- const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable();
+ auto& kernel = system.Kernel();
+ const auto& handle_table = kernel.CurrentProcess()->GetHandleTable();
std::shared_ptr<ClientSession> session = handle_table.Get<ClientSession>(handle);
if (!session) {
LOG_ERROR(Kernel_SVC, "called with invalid handle=0x{:08X}", handle);
@@ -341,9 +340,9 @@ static ResultCode SendSyncRequest(Core::System& system, Handle handle) {
LOG_TRACE(Kernel_SVC, "called handle=0x{:08X}({})", handle, session->GetName());
- auto thread = system.CurrentScheduler().GetCurrentThread();
+ auto thread = kernel.CurrentScheduler()->GetCurrentThread();
{
- SchedulerLock lock(system.Kernel());
+ KScopedSchedulerLock lock(kernel);
thread->InvalidateHLECallback();
thread->SetStatus(ThreadStatus::WaitIPC);
session->SendSyncRequest(SharedFrom(thread), system.Memory(), system.CoreTiming());
@@ -352,12 +351,12 @@ static ResultCode SendSyncRequest(Core::System& system, Handle handle) {
if (thread->HasHLECallback()) {
Handle event_handle = thread->GetHLETimeEvent();
if (event_handle != InvalidHandle) {
- auto& time_manager = system.Kernel().TimeManager();
+ auto& time_manager = kernel.TimeManager();
time_manager.UnscheduleTimeEvent(event_handle);
}
{
- SchedulerLock lock(system.Kernel());
+ KScopedSchedulerLock lock(kernel);
auto* sync_object = thread->GetHLESyncObject();
sync_object->RemoveWaitingThread(SharedFrom(thread));
}
@@ -531,8 +530,7 @@ static ResultCode ArbitrateLock(Core::System& system, Handle holding_thread_hand
static ResultCode ArbitrateLock32(Core::System& system, Handle holding_thread_handle,
u32 mutex_addr, Handle requesting_thread_handle) {
- return ArbitrateLock(system, holding_thread_handle, static_cast<VAddr>(mutex_addr),
- requesting_thread_handle);
+ return ArbitrateLock(system, holding_thread_handle, mutex_addr, requesting_thread_handle);
}
/// Unlock a mutex
@@ -555,7 +553,7 @@ static ResultCode ArbitrateUnlock(Core::System& system, VAddr mutex_addr) {
}
static ResultCode ArbitrateUnlock32(Core::System& system, u32 mutex_addr) {
- return ArbitrateUnlock(system, static_cast<VAddr>(mutex_addr));
+ return ArbitrateUnlock(system, mutex_addr);
}
enum class BreakType : u32 {
@@ -658,7 +656,6 @@ static void Break(Core::System& system, u32 reason, u64 info1, u64 info2) {
info2, has_dumped_buffer ? std::make_optional(debug_buffer) : std::nullopt);
if (!break_reason.signal_debugger) {
- SchedulerLock lock(system.Kernel());
LOG_CRITICAL(
Debug_Emulated,
"Emulated program broke execution! reason=0x{:016X}, info1=0x{:016X}, info2=0x{:016X}",
@@ -666,18 +663,14 @@ static void Break(Core::System& system, u32 reason, u64 info1, u64 info2) {
handle_debug_buffer(info1, info2);
- auto* const current_thread = system.CurrentScheduler().GetCurrentThread();
+ auto* const current_thread = system.Kernel().CurrentScheduler()->GetCurrentThread();
const auto thread_processor_id = current_thread->GetProcessorID();
system.ArmInterface(static_cast<std::size_t>(thread_processor_id)).LogBacktrace();
-
- // Kill the current thread
- system.Kernel().ExceptionalExit();
- current_thread->Stop();
}
}
static void Break32(Core::System& system, u32 reason, u32 info1, u32 info2) {
- Break(system, reason, static_cast<u64>(info1), static_cast<u64>(info2));
+ Break(system, reason, info1, info2);
}
/// Used to output a message on a debug hardware unit - does nothing on a retail unit
@@ -922,7 +915,7 @@ static ResultCode GetInfo(Core::System& system, u64* result, u64 info_id, u64 ha
}
const auto& core_timing = system.CoreTiming();
- const auto& scheduler = system.CurrentScheduler();
+ const auto& scheduler = *system.Kernel().CurrentScheduler();
const auto* const current_thread = scheduler.GetCurrentThread();
const bool same_thread = current_thread == thread.get();
@@ -948,7 +941,7 @@ static ResultCode GetInfo(Core::System& system, u64* result, u64 info_id, u64 ha
static ResultCode GetInfo32(Core::System& system, u32* result_low, u32* result_high, u32 sub_id_low,
u32 info_id, u32 handle, u32 sub_id_high) {
- const u64 sub_id{static_cast<u64>(sub_id_low | (static_cast<u64>(sub_id_high) << 32))};
+ const u64 sub_id{u64{sub_id_low} | (u64{sub_id_high} << 32)};
u64 res_value{};
const ResultCode result{GetInfo(system, &res_value, info_id, handle, sub_id)};
@@ -1009,7 +1002,7 @@ static ResultCode MapPhysicalMemory(Core::System& system, VAddr addr, u64 size)
}
static ResultCode MapPhysicalMemory32(Core::System& system, u32 addr, u32 size) {
- return MapPhysicalMemory(system, static_cast<VAddr>(addr), static_cast<std::size_t>(size));
+ return MapPhysicalMemory(system, addr, size);
}
/// Unmaps memory previously mapped via MapPhysicalMemory
@@ -1063,7 +1056,7 @@ static ResultCode UnmapPhysicalMemory(Core::System& system, VAddr addr, u64 size
}
static ResultCode UnmapPhysicalMemory32(Core::System& system, u32 addr, u32 size) {
- return UnmapPhysicalMemory(system, static_cast<VAddr>(addr), static_cast<std::size_t>(size));
+ return UnmapPhysicalMemory(system, addr, size);
}
/// Sets the thread activity
@@ -1090,7 +1083,7 @@ static ResultCode SetThreadActivity(Core::System& system, Handle handle, u32 act
return ERR_INVALID_HANDLE;
}
- if (thread.get() == system.CurrentScheduler().GetCurrentThread()) {
+ if (thread.get() == system.Kernel().CurrentScheduler()->GetCurrentThread()) {
LOG_ERROR(Kernel_SVC, "The thread handle specified is the current running thread");
return ERR_BUSY;
}
@@ -1123,7 +1116,7 @@ static ResultCode GetThreadContext(Core::System& system, VAddr thread_context, H
return ERR_INVALID_HANDLE;
}
- if (thread.get() == system.CurrentScheduler().GetCurrentThread()) {
+ if (thread.get() == system.Kernel().CurrentScheduler()->GetCurrentThread()) {
LOG_ERROR(Kernel_SVC, "The thread handle specified is the current running thread");
return ERR_BUSY;
}
@@ -1144,7 +1137,7 @@ static ResultCode GetThreadContext(Core::System& system, VAddr thread_context, H
}
static ResultCode GetThreadContext32(Core::System& system, u32 thread_context, Handle handle) {
- return GetThreadContext(system, static_cast<VAddr>(thread_context), handle);
+ return GetThreadContext(system, thread_context, handle);
}
/// Gets the priority for the specified thread
@@ -1281,8 +1274,7 @@ static ResultCode MapSharedMemory(Core::System& system, Handle shared_memory_han
static ResultCode MapSharedMemory32(Core::System& system, Handle shared_memory_handle, u32 addr,
u32 size, u32 permissions) {
- return MapSharedMemory(system, shared_memory_handle, static_cast<VAddr>(addr),
- static_cast<std::size_t>(size), permissions);
+ return MapSharedMemory(system, shared_memory_handle, addr, size, permissions);
}
static ResultCode QueryProcessMemory(Core::System& system, VAddr memory_info_address,
@@ -1480,7 +1472,7 @@ static void ExitProcess(Core::System& system) {
current_process->PrepareForTermination();
// Kill the current thread
- system.CurrentScheduler().GetCurrentThread()->Stop();
+ system.Kernel().CurrentScheduler()->GetCurrentThread()->Stop();
}
static void ExitProcess32(Core::System& system) {
@@ -1552,8 +1544,7 @@ static ResultCode CreateThread(Core::System& system, Handle* out_handle, VAddr e
static ResultCode CreateThread32(Core::System& system, Handle* out_handle, u32 priority,
u32 entry_point, u32 arg, u32 stack_top, s32 processor_id) {
- return CreateThread(system, out_handle, static_cast<VAddr>(entry_point), static_cast<u64>(arg),
- static_cast<VAddr>(stack_top), priority, processor_id);
+ return CreateThread(system, out_handle, entry_point, arg, stack_top, priority, processor_id);
}
/// Starts the thread for the provided handle
@@ -1581,8 +1572,8 @@ static ResultCode StartThread32(Core::System& system, Handle thread_handle) {
static void ExitThread(Core::System& system) {
LOG_DEBUG(Kernel_SVC, "called, pc=0x{:08X}", system.CurrentArmInterface().GetPC());
- auto* const current_thread = system.CurrentScheduler().GetCurrentThread();
- system.GlobalScheduler().RemoveThread(SharedFrom(current_thread));
+ auto* const current_thread = system.Kernel().CurrentScheduler()->GetCurrentThread();
+ system.GlobalSchedulerContext().RemoveThread(SharedFrom(current_thread));
current_thread->Stop();
}
@@ -1592,53 +1583,39 @@ static void ExitThread32(Core::System& system) {
/// Sleep the current thread
static void SleepThread(Core::System& system, s64 nanoseconds) {
- LOG_DEBUG(Kernel_SVC, "called nanoseconds={}", nanoseconds);
+ LOG_TRACE(Kernel_SVC, "called nanoseconds={}", nanoseconds);
enum class SleepType : s64 {
- YieldWithoutLoadBalancing = 0,
- YieldWithLoadBalancing = -1,
+ YieldWithoutCoreMigration = 0,
+ YieldWithCoreMigration = -1,
YieldAndWaitForLoadBalancing = -2,
};
- auto& scheduler = system.CurrentScheduler();
- auto* const current_thread = scheduler.GetCurrentThread();
- bool is_redundant = false;
-
+ auto& scheduler = *system.Kernel().CurrentScheduler();
if (nanoseconds <= 0) {
switch (static_cast<SleepType>(nanoseconds)) {
- case SleepType::YieldWithoutLoadBalancing: {
- auto pair = current_thread->YieldSimple();
- is_redundant = pair.second;
+ case SleepType::YieldWithoutCoreMigration: {
+ scheduler.YieldWithoutCoreMigration();
break;
}
- case SleepType::YieldWithLoadBalancing: {
- auto pair = current_thread->YieldAndBalanceLoad();
- is_redundant = pair.second;
+ case SleepType::YieldWithCoreMigration: {
+ scheduler.YieldWithCoreMigration();
break;
}
case SleepType::YieldAndWaitForLoadBalancing: {
- auto pair = current_thread->YieldAndWaitForLoadBalancing();
- is_redundant = pair.second;
+ scheduler.YieldToAnyThread();
break;
}
default:
UNREACHABLE_MSG("Unimplemented sleep yield type '{:016X}'!", nanoseconds);
}
} else {
- current_thread->Sleep(nanoseconds);
- }
-
- if (is_redundant && !system.Kernel().IsMulticore()) {
- system.Kernel().ExitSVCProfile();
- system.CoreTiming().AddTicks(1000U);
- system.GetCpuManager().PreemptSingleCore();
- system.Kernel().EnterSVCProfile();
+ scheduler.GetCurrentThread()->Sleep(nanoseconds);
}
}
static void SleepThread32(Core::System& system, u32 nanoseconds_low, u32 nanoseconds_high) {
- const s64 nanoseconds = static_cast<s64>(static_cast<u64>(nanoseconds_low) |
- (static_cast<u64>(nanoseconds_high) << 32));
+ const auto nanoseconds = static_cast<s64>(u64{nanoseconds_low} | (u64{nanoseconds_high} << 32));
SleepThread(system, nanoseconds);
}
@@ -1668,10 +1645,10 @@ static ResultCode WaitProcessWideKeyAtomic(Core::System& system, VAddr mutex_add
ASSERT(condition_variable_addr == Common::AlignDown(condition_variable_addr, 4));
auto& kernel = system.Kernel();
Handle event_handle;
- Thread* current_thread = system.CurrentScheduler().GetCurrentThread();
- auto* const current_process = system.Kernel().CurrentProcess();
+ Thread* current_thread = kernel.CurrentScheduler()->GetCurrentThread();
+ auto* const current_process = kernel.CurrentProcess();
{
- SchedulerLockAndSleep lock(kernel, event_handle, current_thread, nano_seconds);
+ KScopedSchedulerLockAndSleep lock(kernel, event_handle, current_thread, nano_seconds);
const auto& handle_table = current_process->GetHandleTable();
std::shared_ptr<Thread> thread = handle_table.Get<Thread>(thread_handle);
ASSERT(thread);
@@ -1707,7 +1684,7 @@ static ResultCode WaitProcessWideKeyAtomic(Core::System& system, VAddr mutex_add
}
{
- SchedulerLock lock(kernel);
+ KScopedSchedulerLock lock(kernel);
auto* owner = current_thread->GetLockOwner();
if (owner != nullptr) {
@@ -1724,10 +1701,8 @@ static ResultCode WaitProcessWideKeyAtomic(Core::System& system, VAddr mutex_add
static ResultCode WaitProcessWideKeyAtomic32(Core::System& system, u32 mutex_addr,
u32 condition_variable_addr, Handle thread_handle,
u32 nanoseconds_low, u32 nanoseconds_high) {
- const s64 nanoseconds =
- static_cast<s64>(nanoseconds_low | (static_cast<u64>(nanoseconds_high) << 32));
- return WaitProcessWideKeyAtomic(system, static_cast<VAddr>(mutex_addr),
- static_cast<VAddr>(condition_variable_addr), thread_handle,
+ const auto nanoseconds = static_cast<s64>(nanoseconds_low | (u64{nanoseconds_high} << 32));
+ return WaitProcessWideKeyAtomic(system, mutex_addr, condition_variable_addr, thread_handle,
nanoseconds);
}
@@ -1740,7 +1715,7 @@ static void SignalProcessWideKey(Core::System& system, VAddr condition_variable_
// Retrieve a list of all threads that are waiting for this condition variable.
auto& kernel = system.Kernel();
- SchedulerLock lock(kernel);
+ KScopedSchedulerLock lock(kernel);
auto* const current_process = kernel.CurrentProcess();
std::vector<std::shared_ptr<Thread>> waiting_threads =
current_process->GetConditionVariableThreads(condition_variable_addr);
@@ -1833,8 +1808,8 @@ static ResultCode WaitForAddress(Core::System& system, VAddr address, u32 type,
static ResultCode WaitForAddress32(Core::System& system, u32 address, u32 type, s32 value,
u32 timeout_low, u32 timeout_high) {
- s64 timeout = static_cast<s64>(timeout_low | (static_cast<u64>(timeout_high) << 32));
- return WaitForAddress(system, static_cast<VAddr>(address), type, value, timeout);
+ const auto timeout = static_cast<s64>(timeout_low | (u64{timeout_high} << 32));
+ return WaitForAddress(system, address, type, value, timeout);
}
// Signals to an address (via Address Arbiter)
@@ -1862,7 +1837,7 @@ static ResultCode SignalToAddress(Core::System& system, VAddr address, u32 type,
static ResultCode SignalToAddress32(Core::System& system, u32 address, u32 type, s32 value,
s32 num_to_wake) {
- return SignalToAddress(system, static_cast<VAddr>(address), type, value, num_to_wake);
+ return SignalToAddress(system, address, type, value, num_to_wake);
}
static void KernelDebug([[maybe_unused]] Core::System& system,
@@ -1893,7 +1868,7 @@ static u64 GetSystemTick(Core::System& system) {
}
static void GetSystemTick32(Core::System& system, u32* time_low, u32* time_high) {
- u64 time = GetSystemTick(system);
+ const auto time = GetSystemTick(system);
*time_low = static_cast<u32>(time);
*time_high = static_cast<u32>(time >> 32);
}
@@ -1984,8 +1959,7 @@ static ResultCode CreateTransferMemory(Core::System& system, Handle* handle, VAd
static ResultCode CreateTransferMemory32(Core::System& system, Handle* handle, u32 addr, u32 size,
u32 permissions) {
- return CreateTransferMemory(system, handle, static_cast<VAddr>(addr),
- static_cast<std::size_t>(size), permissions);
+ return CreateTransferMemory(system, handle, addr, size, permissions);
}
static ResultCode GetThreadCoreMask(Core::System& system, Handle thread_handle, u32* core,
@@ -2003,7 +1977,7 @@ static ResultCode GetThreadCoreMask(Core::System& system, Handle thread_handle,
}
*core = thread->GetIdealCore();
- *mask = thread->GetAffinityMask();
+ *mask = thread->GetAffinityMask().GetAffinityMask();
return RESULT_SUCCESS;
}
@@ -2075,8 +2049,7 @@ static ResultCode SetThreadCoreMask(Core::System& system, Handle thread_handle,
static ResultCode SetThreadCoreMask32(Core::System& system, Handle thread_handle, u32 core,
u32 affinity_mask_low, u32 affinity_mask_high) {
- const u64 affinity_mask =
- static_cast<u64>(affinity_mask_low) | (static_cast<u64>(affinity_mask_high) << 32);
+ const auto affinity_mask = u64{affinity_mask_low} | (u64{affinity_mask_high} << 32);
return SetThreadCoreMask(system, thread_handle, core, affinity_mask);
}
@@ -2341,9 +2314,10 @@ static ResultCode GetThreadList(Core::System& system, u32* out_num_threads, VAdd
return RESULT_SUCCESS;
}
-static ResultCode FlushProcessDataCache32(Core::System& system, Handle handle, u32 address,
- u32 size) {
- // Note(Blinkhawk): For emulation purposes of the data cache this is mostly a nope
+static ResultCode FlushProcessDataCache32([[maybe_unused]] Core::System& system,
+ [[maybe_unused]] Handle handle,
+ [[maybe_unused]] u32 address, [[maybe_unused]] u32 size) {
+ // Note(Blinkhawk): For emulation purposes of the data cache this is mostly a no-op,
// as all emulation is done in the same cache level in host architecture, thus data cache
// does not need flushing.
LOG_DEBUG(Kernel_SVC, "called");
@@ -2639,7 +2613,7 @@ void Call(Core::System& system, u32 immediate) {
auto& kernel = system.Kernel();
kernel.EnterSVCProfile();
- auto* thread = system.CurrentScheduler().GetCurrentThread();
+ auto* thread = kernel.CurrentScheduler()->GetCurrentThread();
thread->SetContinuousOnSVC(true);
const FunctionDef* info = system.CurrentProcess()->Is64BitProcess() ? GetSVCInfo64(immediate)
diff --git a/src/core/hle/kernel/synchronization.cpp b/src/core/hle/kernel/synchronization.cpp
index 8b875d853..d3f520ea2 100644
--- a/src/core/hle/kernel/synchronization.cpp
+++ b/src/core/hle/kernel/synchronization.cpp
@@ -5,8 +5,9 @@
#include "core/core.h"
#include "core/hle/kernel/errors.h"
#include "core/hle/kernel/handle_table.h"
+#include "core/hle/kernel/k_scheduler.h"
+#include "core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h"
#include "core/hle/kernel/kernel.h"
-#include "core/hle/kernel/scheduler.h"
#include "core/hle/kernel/synchronization.h"
#include "core/hle/kernel/synchronization_object.h"
#include "core/hle/kernel/thread.h"
@@ -18,7 +19,7 @@ Synchronization::Synchronization(Core::System& system) : system{system} {}
void Synchronization::SignalObject(SynchronizationObject& obj) const {
auto& kernel = system.Kernel();
- SchedulerLock lock(kernel);
+ KScopedSchedulerLock lock(kernel);
if (obj.IsSignaled()) {
for (auto thread : obj.GetWaitingThreads()) {
if (thread->GetSchedulingStatus() == ThreadSchedStatus::Paused) {
@@ -37,10 +38,10 @@ void Synchronization::SignalObject(SynchronizationObject& obj) const {
std::pair<ResultCode, Handle> Synchronization::WaitFor(
std::vector<std::shared_ptr<SynchronizationObject>>& sync_objects, s64 nano_seconds) {
auto& kernel = system.Kernel();
- auto* const thread = system.CurrentScheduler().GetCurrentThread();
+ auto* const thread = kernel.CurrentScheduler()->GetCurrentThread();
Handle event_handle = InvalidHandle;
{
- SchedulerLockAndSleep lock(kernel, event_handle, thread, nano_seconds);
+ KScopedSchedulerLockAndSleep lock(kernel, event_handle, thread, nano_seconds);
const auto itr =
std::find_if(sync_objects.begin(), sync_objects.end(),
[thread](const std::shared_ptr<SynchronizationObject>& object) {
@@ -89,7 +90,7 @@ std::pair<ResultCode, Handle> Synchronization::WaitFor(
}
{
- SchedulerLock lock(kernel);
+ KScopedSchedulerLock lock(kernel);
ResultCode signaling_result = thread->GetSignalingResult();
SynchronizationObject* signaling_object = thread->GetSignalingObject();
thread->SetSynchronizationObjects(nullptr);
diff --git a/src/core/hle/kernel/thread.cpp b/src/core/hle/kernel/thread.cpp
index 7d1eb2c6e..a4f9e0d97 100644
--- a/src/core/hle/kernel/thread.cpp
+++ b/src/core/hle/kernel/thread.cpp
@@ -17,10 +17,11 @@
#include "core/hardware_properties.h"
#include "core/hle/kernel/errors.h"
#include "core/hle/kernel/handle_table.h"
+#include "core/hle/kernel/k_scheduler.h"
+#include "core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h"
#include "core/hle/kernel/kernel.h"
#include "core/hle/kernel/object.h"
#include "core/hle/kernel/process.h"
-#include "core/hle/kernel/scheduler.h"
#include "core/hle/kernel/thread.h"
#include "core/hle/kernel/time_manager.h"
#include "core/hle/result.h"
@@ -50,7 +51,7 @@ Thread::~Thread() = default;
void Thread::Stop() {
{
- SchedulerLock lock(kernel);
+ KScopedSchedulerLock lock(kernel);
SetStatus(ThreadStatus::Dead);
Signal();
kernel.GlobalHandleTable().Close(global_handle);
@@ -67,7 +68,7 @@ void Thread::Stop() {
}
void Thread::ResumeFromWait() {
- SchedulerLock lock(kernel);
+ KScopedSchedulerLock lock(kernel);
switch (status) {
case ThreadStatus::Paused:
case ThreadStatus::WaitSynch:
@@ -99,19 +100,18 @@ void Thread::ResumeFromWait() {
}
void Thread::OnWakeUp() {
- SchedulerLock lock(kernel);
-
+ KScopedSchedulerLock lock(kernel);
SetStatus(ThreadStatus::Ready);
}
ResultCode Thread::Start() {
- SchedulerLock lock(kernel);
+ KScopedSchedulerLock lock(kernel);
SetStatus(ThreadStatus::Ready);
return RESULT_SUCCESS;
}
void Thread::CancelWait() {
- SchedulerLock lock(kernel);
+ KScopedSchedulerLock lock(kernel);
if (GetSchedulingStatus() != ThreadSchedStatus::Paused || !is_waiting_on_sync) {
is_sync_cancelled = true;
return;
@@ -186,12 +186,14 @@ ResultVal<std::shared_ptr<Thread>> Thread::Create(Core::System& system, ThreadTy
thread->status = ThreadStatus::Dormant;
thread->entry_point = entry_point;
thread->stack_top = stack_top;
+ thread->disable_count = 1;
thread->tpidr_el0 = 0;
thread->nominal_priority = thread->current_priority = priority;
- thread->last_running_ticks = 0;
+ thread->schedule_count = -1;
+ thread->last_scheduled_tick = 0;
thread->processor_id = processor_id;
thread->ideal_core = processor_id;
- thread->affinity_mask = 1ULL << processor_id;
+ thread->affinity_mask.SetAffinity(processor_id, true);
thread->wait_objects = nullptr;
thread->mutex_wait_address = 0;
thread->condvar_wait_address = 0;
@@ -201,7 +203,7 @@ ResultVal<std::shared_ptr<Thread>> Thread::Create(Core::System& system, ThreadTy
thread->owner_process = owner_process;
thread->type = type_flags;
if ((type_flags & THREADTYPE_IDLE) == 0) {
- auto& scheduler = kernel.GlobalScheduler();
+ auto& scheduler = kernel.GlobalSchedulerContext();
scheduler.AddThread(thread);
}
if (owner_process) {
@@ -225,7 +227,7 @@ ResultVal<std::shared_ptr<Thread>> Thread::Create(Core::System& system, ThreadTy
}
void Thread::SetPriority(u32 priority) {
- SchedulerLock lock(kernel);
+ KScopedSchedulerLock lock(kernel);
ASSERT_MSG(priority <= THREADPRIO_LOWEST && priority >= THREADPRIO_HIGHEST,
"Invalid priority value.");
nominal_priority = priority;
@@ -362,7 +364,7 @@ bool Thread::InvokeHLECallback(std::shared_ptr<Thread> thread) {
}
ResultCode Thread::SetActivity(ThreadActivity value) {
- SchedulerLock lock(kernel);
+ KScopedSchedulerLock lock(kernel);
auto sched_status = GetSchedulingStatus();
@@ -391,7 +393,7 @@ ResultCode Thread::SetActivity(ThreadActivity value) {
ResultCode Thread::Sleep(s64 nanoseconds) {
Handle event_handle{};
{
- SchedulerLockAndSleep lock(kernel, event_handle, this, nanoseconds);
+ KScopedSchedulerLockAndSleep lock(kernel, event_handle, this, nanoseconds);
SetStatus(ThreadStatus::WaitSleep);
}
@@ -402,39 +404,12 @@ ResultCode Thread::Sleep(s64 nanoseconds) {
return RESULT_SUCCESS;
}
-std::pair<ResultCode, bool> Thread::YieldSimple() {
- bool is_redundant = false;
- {
- SchedulerLock lock(kernel);
- is_redundant = kernel.GlobalScheduler().YieldThread(this);
- }
- return {RESULT_SUCCESS, is_redundant};
-}
-
-std::pair<ResultCode, bool> Thread::YieldAndBalanceLoad() {
- bool is_redundant = false;
- {
- SchedulerLock lock(kernel);
- is_redundant = kernel.GlobalScheduler().YieldThreadAndBalanceLoad(this);
- }
- return {RESULT_SUCCESS, is_redundant};
-}
-
-std::pair<ResultCode, bool> Thread::YieldAndWaitForLoadBalancing() {
- bool is_redundant = false;
- {
- SchedulerLock lock(kernel);
- is_redundant = kernel.GlobalScheduler().YieldThreadAndWaitForLoadBalancing(this);
- }
- return {RESULT_SUCCESS, is_redundant};
-}
-
void Thread::AddSchedulingFlag(ThreadSchedFlags flag) {
const u32 old_state = scheduling_state;
pausing_state |= static_cast<u32>(flag);
const u32 base_scheduling = static_cast<u32>(GetSchedulingStatus());
scheduling_state = base_scheduling | pausing_state;
- kernel.GlobalScheduler().AdjustSchedulingOnStatus(this, old_state);
+ KScheduler::OnThreadStateChanged(kernel, this, old_state);
}
void Thread::RemoveSchedulingFlag(ThreadSchedFlags flag) {
@@ -442,23 +417,24 @@ void Thread::RemoveSchedulingFlag(ThreadSchedFlags flag) {
pausing_state &= ~static_cast<u32>(flag);
const u32 base_scheduling = static_cast<u32>(GetSchedulingStatus());
scheduling_state = base_scheduling | pausing_state;
- kernel.GlobalScheduler().AdjustSchedulingOnStatus(this, old_state);
+ KScheduler::OnThreadStateChanged(kernel, this, old_state);
}
void Thread::SetSchedulingStatus(ThreadSchedStatus new_status) {
const u32 old_state = scheduling_state;
scheduling_state = (scheduling_state & static_cast<u32>(ThreadSchedMasks::HighMask)) |
static_cast<u32>(new_status);
- kernel.GlobalScheduler().AdjustSchedulingOnStatus(this, old_state);
+ KScheduler::OnThreadStateChanged(kernel, this, old_state);
}
void Thread::SetCurrentPriority(u32 new_priority) {
const u32 old_priority = std::exchange(current_priority, new_priority);
- kernel.GlobalScheduler().AdjustSchedulingOnPriority(this, old_priority);
+ KScheduler::OnThreadPriorityChanged(kernel, this, kernel.CurrentScheduler()->GetCurrentThread(),
+ old_priority);
}
ResultCode Thread::SetCoreAndAffinityMask(s32 new_core, u64 new_affinity_mask) {
- SchedulerLock lock(kernel);
+ KScopedSchedulerLock lock(kernel);
const auto HighestSetCore = [](u64 mask, u32 max_cores) {
for (s32 core = static_cast<s32>(max_cores - 1); core >= 0; core--) {
if (((mask >> core) & 1) != 0) {
@@ -479,20 +455,21 @@ ResultCode Thread::SetCoreAndAffinityMask(s32 new_core, u64 new_affinity_mask) {
}
if (use_override) {
ideal_core_override = new_core;
- affinity_mask_override = new_affinity_mask;
} else {
- const u64 old_affinity_mask = std::exchange(affinity_mask, new_affinity_mask);
+ const auto old_affinity_mask = affinity_mask;
+ affinity_mask.SetAffinityMask(new_affinity_mask);
ideal_core = new_core;
- if (old_affinity_mask != new_affinity_mask) {
+ if (old_affinity_mask.GetAffinityMask() != new_affinity_mask) {
const s32 old_core = processor_id;
- if (processor_id >= 0 && ((affinity_mask >> processor_id) & 1) == 0) {
+ if (processor_id >= 0 && !affinity_mask.GetAffinity(processor_id)) {
if (static_cast<s32>(ideal_core) < 0) {
- processor_id = HighestSetCore(affinity_mask, Core::Hardware::NUM_CPU_CORES);
+ processor_id = HighestSetCore(affinity_mask.GetAffinityMask(),
+ Core::Hardware::NUM_CPU_CORES);
} else {
processor_id = ideal_core;
}
}
- kernel.GlobalScheduler().AdjustSchedulingOnAffinity(this, old_affinity_mask, old_core);
+ KScheduler::OnThreadAffinityMaskChanged(kernel, this, old_affinity_mask, old_core);
}
}
return RESULT_SUCCESS;
diff --git a/src/core/hle/kernel/thread.h b/src/core/hle/kernel/thread.h
index a75071e9b..11ef29888 100644
--- a/src/core/hle/kernel/thread.h
+++ b/src/core/hle/kernel/thread.h
@@ -4,6 +4,7 @@
#pragma once
+#include <array>
#include <functional>
#include <string>
#include <utility>
@@ -12,6 +13,7 @@
#include "common/common_types.h"
#include "common/spin_lock.h"
#include "core/arm/arm_interface.h"
+#include "core/hle/kernel/k_affinity_mask.h"
#include "core/hle/kernel/object.h"
#include "core/hle/kernel/synchronization_object.h"
#include "core/hle/result.h"
@@ -27,10 +29,10 @@ class System;
namespace Kernel {
-class GlobalScheduler;
+class GlobalSchedulerContext;
class KernelCore;
class Process;
-class Scheduler;
+class KScheduler;
enum ThreadPriority : u32 {
THREADPRIO_HIGHEST = 0, ///< Highest thread priority
@@ -345,8 +347,12 @@ public:
void SetStatus(ThreadStatus new_status);
- u64 GetLastRunningTicks() const {
- return last_running_ticks;
+ s64 GetLastScheduledTick() const {
+ return this->last_scheduled_tick;
+ }
+
+ void SetLastScheduledTick(s64 tick) {
+ this->last_scheduled_tick = tick;
}
u64 GetTotalCPUTimeTicks() const {
@@ -361,10 +367,18 @@ public:
return processor_id;
}
+ s32 GetActiveCore() const {
+ return GetProcessorID();
+ }
+
void SetProcessorID(s32 new_core) {
processor_id = new_core;
}
+ void SetActiveCore(s32 new_core) {
+ processor_id = new_core;
+ }
+
Process* GetOwnerProcess() {
return owner_process;
}
@@ -469,7 +483,7 @@ public:
return ideal_core;
}
- u64 GetAffinityMask() const {
+ const KAffinityMask& GetAffinityMask() const {
return affinity_mask;
}
@@ -478,21 +492,12 @@ public:
/// Sleeps this thread for the given amount of nanoseconds.
ResultCode Sleep(s64 nanoseconds);
- /// Yields this thread without rebalancing loads.
- std::pair<ResultCode, bool> YieldSimple();
-
- /// Yields this thread and does a load rebalancing.
- std::pair<ResultCode, bool> YieldAndBalanceLoad();
-
- /// Yields this thread and if the core is left idle, loads are rebalanced
- std::pair<ResultCode, bool> YieldAndWaitForLoadBalancing();
-
- void IncrementYieldCount() {
- yield_count++;
+ s64 GetYieldScheduleCount() const {
+ return this->schedule_count;
}
- u64 GetYieldCount() const {
- return yield_count;
+ void SetYieldScheduleCount(s64 count) {
+ this->schedule_count = count;
}
ThreadSchedStatus GetSchedulingStatus() const {
@@ -568,9 +573,59 @@ public:
return has_exited;
}
+ class QueueEntry {
+ public:
+ constexpr QueueEntry() = default;
+
+ constexpr void Initialize() {
+ this->prev = nullptr;
+ this->next = nullptr;
+ }
+
+ constexpr Thread* GetPrev() const {
+ return this->prev;
+ }
+ constexpr Thread* GetNext() const {
+ return this->next;
+ }
+ constexpr void SetPrev(Thread* thread) {
+ this->prev = thread;
+ }
+ constexpr void SetNext(Thread* thread) {
+ this->next = thread;
+ }
+
+ private:
+ Thread* prev{};
+ Thread* next{};
+ };
+
+ QueueEntry& GetPriorityQueueEntry(s32 core) {
+ return this->per_core_priority_queue_entry[core];
+ }
+
+ const QueueEntry& GetPriorityQueueEntry(s32 core) const {
+ return this->per_core_priority_queue_entry[core];
+ }
+
+ s32 GetDisableDispatchCount() const {
+ return disable_count;
+ }
+
+ void DisableDispatch() {
+ ASSERT(GetDisableDispatchCount() >= 0);
+ disable_count++;
+ }
+
+ void EnableDispatch() {
+ ASSERT(GetDisableDispatchCount() > 0);
+ disable_count--;
+ }
+
private:
- friend class GlobalScheduler;
- friend class Scheduler;
+ friend class GlobalSchedulerContext;
+ friend class KScheduler;
+ friend class Process;
void SetSchedulingStatus(ThreadSchedStatus new_status);
void AddSchedulingFlag(ThreadSchedFlags flag);
@@ -583,12 +638,14 @@ private:
ThreadContext64 context_64{};
std::shared_ptr<Common::Fiber> host_context{};
- u64 thread_id = 0;
-
ThreadStatus status = ThreadStatus::Dormant;
+ u32 scheduling_state = 0;
+
+ u64 thread_id = 0;
VAddr entry_point = 0;
VAddr stack_top = 0;
+ std::atomic_int disable_count = 0;
ThreadType type;
@@ -602,9 +659,8 @@ private:
u32 current_priority = 0;
u64 total_cpu_time_ticks = 0; ///< Total CPU running ticks.
- u64 last_running_ticks = 0; ///< CPU tick when thread was last running
- u64 yield_count = 0; ///< Number of redundant yields carried by this thread.
- ///< a redundant yield is one where no scheduling is changed
+ s64 schedule_count{};
+ s64 last_scheduled_tick{};
s32 processor_id = 0;
@@ -646,16 +702,16 @@ private:
Handle hle_time_event;
SynchronizationObject* hle_object;
- Scheduler* scheduler = nullptr;
+ KScheduler* scheduler = nullptr;
+
+ std::array<QueueEntry, Core::Hardware::NUM_CPU_CORES> per_core_priority_queue_entry{};
u32 ideal_core{0xFFFFFFFF};
- u64 affinity_mask{0x1};
+ KAffinityMask affinity_mask{};
s32 ideal_core_override = -1;
- u64 affinity_mask_override = 0x1;
u32 affinity_override_count = 0;
- u32 scheduling_state = 0;
u32 pausing_state = 0;
bool is_running = false;
bool is_waiting_on_sync = false;
diff --git a/src/core/hle/kernel/time_manager.cpp b/src/core/hle/kernel/time_manager.cpp
index caf329bfb..79628e2b4 100644
--- a/src/core/hle/kernel/time_manager.cpp
+++ b/src/core/hle/kernel/time_manager.cpp
@@ -7,8 +7,8 @@
#include "core/core_timing.h"
#include "core/core_timing_util.h"
#include "core/hle/kernel/handle_table.h"
+#include "core/hle/kernel/k_scheduler.h"
#include "core/hle/kernel/kernel.h"
-#include "core/hle/kernel/scheduler.h"
#include "core/hle/kernel/thread.h"
#include "core/hle/kernel/time_manager.h"
@@ -18,12 +18,18 @@ TimeManager::TimeManager(Core::System& system_) : system{system_} {
time_manager_event_type = Core::Timing::CreateEvent(
"Kernel::TimeManagerCallback",
[this](std::uintptr_t thread_handle, std::chrono::nanoseconds) {
- const SchedulerLock lock(system.Kernel());
+ const KScopedSchedulerLock lock(system.Kernel());
const auto proper_handle = static_cast<Handle>(thread_handle);
- if (cancelled_events[proper_handle]) {
- return;
+
+ std::shared_ptr<Thread> thread;
+ {
+ std::lock_guard lock{mutex};
+ if (cancelled_events[proper_handle]) {
+ return;
+ }
+ thread = system.Kernel().RetrieveThreadFromGlobalHandleTable(proper_handle);
}
- auto thread = this->system.Kernel().RetrieveThreadFromGlobalHandleTable(proper_handle);
+
if (thread) {
// Thread can be null if process has exited
thread->OnWakeUp();
@@ -56,6 +62,7 @@ void TimeManager::UnscheduleTimeEvent(Handle event_handle) {
}
void TimeManager::CancelTimeEvent(Thread* time_task) {
+ std::lock_guard lock{mutex};
const Handle event_handle = time_task->GetGlobalHandle();
UnscheduleTimeEvent(event_handle);
}
diff --git a/src/core/hle/service/am/am.cpp b/src/core/hle/service/am/am.cpp
index 38d877f6e..cb13210e5 100644
--- a/src/core/hle/service/am/am.cpp
+++ b/src/core/hle/service/am/am.cpp
@@ -1092,14 +1092,14 @@ void ILibraryAppletCreator::CreateLibraryApplet(Kernel::HLERequestContext& ctx)
const auto applet_id = rp.PopRaw<Applets::AppletId>();
const auto applet_mode = rp.PopRaw<u32>();
- LOG_DEBUG(Service_AM, "called with applet_id={:08X}, applet_mode={:08X}",
- static_cast<u32>(applet_id), applet_mode);
+ LOG_DEBUG(Service_AM, "called with applet_id={:08X}, applet_mode={:08X}", applet_id,
+ applet_mode);
const auto& applet_manager{system.GetAppletManager()};
const auto applet = applet_manager.GetApplet(applet_id);
if (applet == nullptr) {
- LOG_ERROR(Service_AM, "Applet doesn't exist! applet_id={}", static_cast<u32>(applet_id));
+ LOG_ERROR(Service_AM, "Applet doesn't exist! applet_id={}", applet_id);
IPC::ResponseBuilder rb{ctx, 2};
rb.Push(RESULT_UNKNOWN);
@@ -1290,7 +1290,7 @@ void IApplicationFunctions::PopLaunchParameter(Kernel::HLERequestContext& ctx) {
IPC::RequestParser rp{ctx};
const auto kind = rp.PopEnum<LaunchParameterKind>();
- LOG_DEBUG(Service_AM, "called, kind={:08X}", static_cast<u8>(kind));
+ LOG_DEBUG(Service_AM, "called, kind={:08X}", kind);
if (kind == LaunchParameterKind::ApplicationSpecific && !launch_popped_application_specific) {
const auto backend = BCAT::CreateBackendFromSettings(system, [this](u64 tid) {
@@ -1537,8 +1537,8 @@ void IApplicationFunctions::GetSaveDataSize(Kernel::HLERequestContext& ctx) {
IPC::RequestParser rp{ctx};
const auto [type, user_id] = rp.PopRaw<Parameters>();
- LOG_DEBUG(Service_AM, "called with type={:02X}, user_id={:016X}{:016X}", static_cast<u8>(type),
- user_id[1], user_id[0]);
+ LOG_DEBUG(Service_AM, "called with type={:02X}, user_id={:016X}{:016X}", type, user_id[1],
+ user_id[0]);
const auto size = system.GetFileSystemController().ReadSaveDataSize(
type, system.CurrentProcess()->GetTitleID(), user_id);
diff --git a/src/core/hle/service/am/applets/applets.cpp b/src/core/hle/service/am/applets/applets.cpp
index 2b626bb40..08676c3fc 100644
--- a/src/core/hle/service/am/applets/applets.cpp
+++ b/src/core/hle/service/am/applets/applets.cpp
@@ -142,14 +142,14 @@ void Applet::Initialize() {
AppletFrontendSet::AppletFrontendSet() = default;
-AppletFrontendSet::AppletFrontendSet(ControllerApplet controller, ECommerceApplet e_commerce,
- ErrorApplet error, ParentalControlsApplet parental_controls,
- PhotoViewer photo_viewer, ProfileSelect profile_select,
- SoftwareKeyboard software_keyboard, WebBrowser web_browser)
- : controller{std::move(controller)}, e_commerce{std::move(e_commerce)}, error{std::move(error)},
- parental_controls{std::move(parental_controls)}, photo_viewer{std::move(photo_viewer)},
- profile_select{std::move(profile_select)}, software_keyboard{std::move(software_keyboard)},
- web_browser{std::move(web_browser)} {}
+AppletFrontendSet::AppletFrontendSet(ControllerApplet controller_applet, ErrorApplet error_applet,
+ ParentalControlsApplet parental_controls_applet,
+ PhotoViewer photo_viewer_, ProfileSelect profile_select_,
+ SoftwareKeyboard software_keyboard_, WebBrowser web_browser_)
+ : controller{std::move(controller_applet)}, error{std::move(error_applet)},
+ parental_controls{std::move(parental_controls_applet)},
+ photo_viewer{std::move(photo_viewer_)}, profile_select{std::move(profile_select_)},
+ software_keyboard{std::move(software_keyboard_)}, web_browser{std::move(web_browser_)} {}
AppletFrontendSet::~AppletFrontendSet() = default;
@@ -170,10 +170,6 @@ void AppletManager::SetAppletFrontendSet(AppletFrontendSet set) {
frontend.controller = std::move(set.controller);
}
- if (set.e_commerce != nullptr) {
- frontend.e_commerce = std::move(set.e_commerce);
- }
-
if (set.error != nullptr) {
frontend.error = std::move(set.error);
}
@@ -210,10 +206,6 @@ void AppletManager::SetDefaultAppletsIfMissing() {
std::make_unique<Core::Frontend::DefaultControllerApplet>(system.ServiceManager());
}
- if (frontend.e_commerce == nullptr) {
- frontend.e_commerce = std::make_unique<Core::Frontend::DefaultECommerceApplet>();
- }
-
if (frontend.error == nullptr) {
frontend.error = std::make_unique<Core::Frontend::DefaultErrorApplet>();
}
@@ -257,13 +249,14 @@ std::shared_ptr<Applet> AppletManager::GetApplet(AppletId id) const {
return std::make_shared<ProfileSelect>(system, *frontend.profile_select);
case AppletId::SoftwareKeyboard:
return std::make_shared<SoftwareKeyboard>(system, *frontend.software_keyboard);
+ case AppletId::Web:
+ case AppletId::Shop:
+ case AppletId::OfflineWeb:
+ case AppletId::LoginShare:
+ case AppletId::WebAuth:
+ return std::make_shared<WebBrowser>(system, *frontend.web_browser);
case AppletId::PhotoViewer:
return std::make_shared<PhotoViewer>(system, *frontend.photo_viewer);
- case AppletId::LibAppletShop:
- return std::make_shared<WebBrowser>(system, *frontend.web_browser,
- frontend.e_commerce.get());
- case AppletId::LibAppletOff:
- return std::make_shared<WebBrowser>(system, *frontend.web_browser);
default:
UNIMPLEMENTED_MSG(
"No backend implementation exists for applet_id={:02X}! Falling back to stub applet.",
diff --git a/src/core/hle/service/am/applets/applets.h b/src/core/hle/service/am/applets/applets.h
index a1f4cf897..4fd792c05 100644
--- a/src/core/hle/service/am/applets/applets.h
+++ b/src/core/hle/service/am/applets/applets.h
@@ -50,13 +50,13 @@ enum class AppletId : u32 {
ProfileSelect = 0x10,
SoftwareKeyboard = 0x11,
MiiEdit = 0x12,
- LibAppletWeb = 0x13,
- LibAppletShop = 0x14,
+ Web = 0x13,
+ Shop = 0x14,
PhotoViewer = 0x15,
Settings = 0x16,
- LibAppletOff = 0x17,
- LibAppletWhitelisted = 0x18,
- LibAppletAuth = 0x19,
+ OfflineWeb = 0x17,
+ LoginShare = 0x18,
+ WebAuth = 0x19,
MyPage = 0x1A,
};
@@ -157,7 +157,6 @@ protected:
struct AppletFrontendSet {
using ControllerApplet = std::unique_ptr<Core::Frontend::ControllerApplet>;
- using ECommerceApplet = std::unique_ptr<Core::Frontend::ECommerceApplet>;
using ErrorApplet = std::unique_ptr<Core::Frontend::ErrorApplet>;
using ParentalControlsApplet = std::unique_ptr<Core::Frontend::ParentalControlsApplet>;
using PhotoViewer = std::unique_ptr<Core::Frontend::PhotoViewerApplet>;
@@ -166,10 +165,10 @@ struct AppletFrontendSet {
using WebBrowser = std::unique_ptr<Core::Frontend::WebBrowserApplet>;
AppletFrontendSet();
- AppletFrontendSet(ControllerApplet controller, ECommerceApplet e_commerce, ErrorApplet error,
- ParentalControlsApplet parental_controls, PhotoViewer photo_viewer,
- ProfileSelect profile_select, SoftwareKeyboard software_keyboard,
- WebBrowser web_browser);
+ AppletFrontendSet(ControllerApplet controller_applet, ErrorApplet error_applet,
+ ParentalControlsApplet parental_controls_applet, PhotoViewer photo_viewer_,
+ ProfileSelect profile_select_, SoftwareKeyboard software_keyboard_,
+ WebBrowser web_browser_);
~AppletFrontendSet();
AppletFrontendSet(const AppletFrontendSet&) = delete;
@@ -179,7 +178,6 @@ struct AppletFrontendSet {
AppletFrontendSet& operator=(AppletFrontendSet&&) noexcept;
ControllerApplet controller;
- ECommerceApplet e_commerce;
ErrorApplet error;
ParentalControlsApplet parental_controls;
PhotoViewer photo_viewer;
diff --git a/src/core/hle/service/am/applets/controller.cpp b/src/core/hle/service/am/applets/controller.cpp
index e8ea4248b..7edfca64e 100644
--- a/src/core/hle/service/am/applets/controller.cpp
+++ b/src/core/hle/service/am/applets/controller.cpp
@@ -29,14 +29,14 @@ static Core::Frontend::ControllerParameters ConvertToFrontendParameters(
npad_style_set.raw = private_arg.style_set;
return {
- .min_players = std::max(s8(1), header.player_count_min),
+ .min_players = std::max(s8{1}, header.player_count_min),
.max_players = header.player_count_max,
.keep_controllers_connected = header.enable_take_over_connection,
.enable_single_mode = header.enable_single_mode,
.enable_border_color = header.enable_identification_color,
- .border_colors = identification_colors,
+ .border_colors = std::move(identification_colors),
.enable_explain_text = enable_text,
- .explain_text = text,
+ .explain_text = std::move(text),
.allow_pro_controller = npad_style_set.pro_controller == 1,
.allow_handheld = npad_style_set.handheld == 1,
.allow_dual_joycons = npad_style_set.joycon_dual == 1,
@@ -227,15 +227,14 @@ void Controller::ConfigurationComplete() {
// If enable_single_mode is enabled, player_count is 1 regardless of any other parameters.
// Otherwise, only count connected players from P1-P8.
result_info.player_count =
- is_single_mode ? 1
- : static_cast<s8>(std::count_if(
- players.begin(), players.end() - 2,
- [](Settings::PlayerInput player) { return player.connected; }));
-
- result_info.selected_id = HID::Controller_NPad::IndexToNPad(
- std::distance(players.begin(),
- std::find_if(players.begin(), players.end(),
- [](Settings::PlayerInput player) { return player.connected; })));
+ is_single_mode
+ ? 1
+ : static_cast<s8>(std::count_if(players.begin(), players.end() - 2,
+ [](const auto& player) { return player.connected; }));
+
+ result_info.selected_id = HID::Controller_NPad::IndexToNPad(std::distance(
+ players.begin(), std::find_if(players.begin(), players.end(),
+ [](const auto& player) { return player.connected; })));
result_info.result = 0;
diff --git a/src/core/hle/service/am/applets/error.cpp b/src/core/hle/service/am/applets/error.cpp
index dcd4b2a35..d85505082 100644
--- a/src/core/hle/service/am/applets/error.cpp
+++ b/src/core/hle/service/am/applets/error.cpp
@@ -125,7 +125,7 @@ void Error::Initialize() {
error_code = Decode64BitError(args->error_record.error_code_64);
break;
default:
- UNIMPLEMENTED_MSG("Unimplemented LibAppletError mode={:02X}!", static_cast<u8>(mode));
+ UNIMPLEMENTED_MSG("Unimplemented LibAppletError mode={:02X}!", mode);
}
}
@@ -179,7 +179,7 @@ void Error::Execute() {
error_code, std::chrono::seconds{args->error_record.posix_time}, callback);
break;
default:
- UNIMPLEMENTED_MSG("Unimplemented LibAppletError mode={:02X}!", static_cast<u8>(mode));
+ UNIMPLEMENTED_MSG("Unimplemented LibAppletError mode={:02X}!", mode);
DisplayCompleted();
}
}
diff --git a/src/core/hle/service/am/applets/general_backend.cpp b/src/core/hle/service/am/applets/general_backend.cpp
index bdb6fd464..4d1df5cbe 100644
--- a/src/core/hle/service/am/applets/general_backend.cpp
+++ b/src/core/hle/service/am/applets/general_backend.cpp
@@ -90,7 +90,7 @@ void Auth::Execute() {
const auto unimplemented_log = [this] {
UNIMPLEMENTED_MSG("Unimplemented Auth applet type for type={:08X}, arg0={:02X}, "
"arg1={:02X}, arg2={:02X}",
- static_cast<u32>(type), arg0, arg1, arg2);
+ type, arg0, arg1, arg2);
};
switch (type) {
@@ -136,7 +136,7 @@ void Auth::Execute() {
}
void Auth::AuthFinished(bool is_successful) {
- this->successful = is_successful;
+ successful = is_successful;
struct Return {
ResultCode result_code;
@@ -193,7 +193,7 @@ void PhotoViewer::Execute() {
frontend.ShowAllPhotos(callback);
break;
default:
- UNIMPLEMENTED_MSG("Unimplemented PhotoViewer applet mode={:02X}!", static_cast<u8>(mode));
+ UNIMPLEMENTED_MSG("Unimplemented PhotoViewer applet mode={:02X}!", mode);
}
}
diff --git a/src/core/hle/service/am/applets/web_browser.cpp b/src/core/hle/service/am/applets/web_browser.cpp
index c3b6b706a..2ab420789 100644
--- a/src/core/hle/service/am/applets/web_browser.cpp
+++ b/src/core/hle/service/am/applets/web_browser.cpp
@@ -1,558 +1,478 @@
-// Copyright 2018 yuzu emulator team
+// Copyright 2020 yuzu Emulator Project
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
-#include <array>
-#include <cstring>
-#include <vector>
-
#include "common/assert.h"
-#include "common/common_funcs.h"
#include "common/common_paths.h"
#include "common/file_util.h"
-#include "common/hex_util.h"
#include "common/logging/log.h"
#include "common/string_util.h"
#include "core/core.h"
#include "core/file_sys/content_archive.h"
#include "core/file_sys/mode.h"
#include "core/file_sys/nca_metadata.h"
+#include "core/file_sys/patch_manager.h"
#include "core/file_sys/registered_cache.h"
#include "core/file_sys/romfs.h"
#include "core/file_sys/system_archive/system_archive.h"
-#include "core/file_sys/vfs_types.h"
-#include "core/frontend/applets/general_frontend.h"
+#include "core/file_sys/vfs_vector.h"
#include "core/frontend/applets/web_browser.h"
#include "core/hle/kernel/process.h"
+#include "core/hle/result.h"
+#include "core/hle/service/am/am.h"
#include "core/hle/service/am/applets/web_browser.h"
#include "core/hle/service/filesystem/filesystem.h"
-#include "core/loader/loader.h"
+#include "core/hle/service/ns/pl_u.h"
namespace Service::AM::Applets {
-enum class WebArgTLVType : u16 {
- InitialURL = 0x1,
- ShopArgumentsURL = 0x2, ///< TODO(DarkLordZach): This is not the official name.
- CallbackURL = 0x3,
- CallbackableURL = 0x4,
- ApplicationID = 0x5,
- DocumentPath = 0x6,
- DocumentKind = 0x7,
- SystemDataID = 0x8,
- ShareStartPage = 0x9,
- Whitelist = 0xA,
- News = 0xB,
- UserID = 0xE,
- AlbumEntry0 = 0xF,
- ScreenShotEnabled = 0x10,
- EcClientCertEnabled = 0x11,
- Unk12 = 0x12,
- PlayReportEnabled = 0x13,
- Unk14 = 0x14,
- Unk15 = 0x15,
- BootDisplayKind = 0x17,
- BackgroundKind = 0x18,
- FooterEnabled = 0x19,
- PointerEnabled = 0x1A,
- LeftStickMode = 0x1B,
- KeyRepeatFrame1 = 0x1C,
- KeyRepeatFrame2 = 0x1D,
- BootAsMediaPlayerInv = 0x1E,
- DisplayUrlKind = 0x1F,
- BootAsMediaPlayer = 0x21,
- ShopJumpEnabled = 0x22,
- MediaAutoPlayEnabled = 0x23,
- LobbyParameter = 0x24,
- ApplicationAlbumEntry = 0x26,
- JsExtensionEnabled = 0x27,
- AdditionalCommentText = 0x28,
- TouchEnabledOnContents = 0x29,
- UserAgentAdditionalString = 0x2A,
- AdditionalMediaData0 = 0x2B,
- MediaPlayerAutoCloseEnabled = 0x2C,
- PageCacheEnabled = 0x2D,
- WebAudioEnabled = 0x2E,
- Unk2F = 0x2F,
- YouTubeVideoWhitelist = 0x31,
- FooterFixedKind = 0x32,
- PageFadeEnabled = 0x33,
- MediaCreatorApplicationRatingAge = 0x34,
- BootLoadingIconEnabled = 0x35,
- PageScrollIndicationEnabled = 0x36,
- MediaPlayerSpeedControlEnabled = 0x37,
- AlbumEntry1 = 0x38,
- AlbumEntry2 = 0x39,
- AlbumEntry3 = 0x3A,
- AdditionalMediaData1 = 0x3B,
- AdditionalMediaData2 = 0x3C,
- AdditionalMediaData3 = 0x3D,
- BootFooterButton = 0x3E,
- OverrideWebAudioVolume = 0x3F,
- OverrideMediaAudioVolume = 0x40,
- BootMode = 0x41,
- WebSessionEnabled = 0x42,
-};
-
-enum class ShimKind : u32 {
- Shop = 1,
- Login = 2,
- Offline = 3,
- Share = 4,
- Web = 5,
- Wifi = 6,
- Lobby = 7,
-};
-
-enum class ShopWebTarget {
- ApplicationInfo,
- AddOnContentList,
- SubscriptionList,
- ConsumableItemList,
- Home,
- Settings,
-};
-
namespace {
-constexpr std::size_t SHIM_KIND_COUNT = 0x8;
-
-struct WebArgHeader {
- u16 count;
- INSERT_PADDING_BYTES(2);
- ShimKind kind;
-};
-static_assert(sizeof(WebArgHeader) == 0x8, "WebArgHeader has incorrect size.");
-
-struct WebArgTLV {
- WebArgTLVType type;
- u16 size;
- u32 offset;
-};
-static_assert(sizeof(WebArgTLV) == 0x8, "WebArgTLV has incorrect size.");
-
-struct WebCommonReturnValue {
- u32 result_code;
- INSERT_PADDING_BYTES(0x4);
- std::array<char, 0x1000> last_url;
- u64 last_url_size;
-};
-static_assert(sizeof(WebCommonReturnValue) == 0x1010, "WebCommonReturnValue has incorrect size.");
-
-struct WebWifiPageArg {
- INSERT_PADDING_BYTES(4);
- std::array<char, 0x100> connection_test_url;
- std::array<char, 0x400> initial_url;
- std::array<u8, 0x10> nifm_network_uuid;
- u32 nifm_requirement;
-};
-static_assert(sizeof(WebWifiPageArg) == 0x518, "WebWifiPageArg has incorrect size.");
-
-struct WebWifiReturnValue {
- INSERT_PADDING_BYTES(4);
- u32 result;
-};
-static_assert(sizeof(WebWifiReturnValue) == 0x8, "WebWifiReturnValue has incorrect size.");
-
-enum class OfflineWebSource : u32 {
- OfflineHtmlPage = 0x1,
- ApplicationLegalInformation = 0x2,
- SystemDataPage = 0x3,
-};
-
-std::map<WebArgTLVType, std::vector<u8>> GetWebArguments(const std::vector<u8>& arg) {
- if (arg.size() < sizeof(WebArgHeader))
- return {};
-
- WebArgHeader header{};
- std::memcpy(&header, arg.data(), sizeof(WebArgHeader));
-
- std::map<WebArgTLVType, std::vector<u8>> out;
- u64 offset = sizeof(WebArgHeader);
- for (std::size_t i = 0; i < header.count; ++i) {
- if (arg.size() < (offset + sizeof(WebArgTLV)))
- return out;
+template <typename T>
+void ParseRawValue(T& value, const std::vector<u8>& data) {
+ static_assert(std::is_trivially_copyable_v<T>,
+ "It's undefined behavior to use memcpy with non-trivially copyable objects");
+ std::memcpy(&value, data.data(), data.size());
+}
- WebArgTLV tlv{};
- std::memcpy(&tlv, arg.data() + offset, sizeof(WebArgTLV));
- offset += sizeof(WebArgTLV);
+template <typename T>
+T ParseRawValue(const std::vector<u8>& data) {
+ T value;
+ ParseRawValue(value, data);
+ return value;
+}
- offset += tlv.offset;
- if (arg.size() < (offset + tlv.size))
- return out;
+std::string ParseStringValue(const std::vector<u8>& data) {
+ return Common::StringFromFixedZeroTerminatedBuffer(reinterpret_cast<const char*>(data.data()),
+ data.size());
+}
- std::vector<u8> data(tlv.size);
- std::memcpy(data.data(), arg.data() + offset, tlv.size);
- offset += tlv.size;
+std::string GetMainURL(const std::string& url) {
+ const auto index = url.find('?');
- out.insert_or_assign(tlv.type, data);
+ if (index == std::string::npos) {
+ return url;
}
- return out;
+ return url.substr(0, index);
}
-FileSys::VirtualFile GetApplicationRomFS(const Core::System& system, u64 title_id,
- FileSys::ContentRecordType type) {
- const auto& installed{system.GetContentProvider()};
- const auto res = installed.GetEntry(title_id, type);
+WebArgInputTLVMap ReadWebArgs(const std::vector<u8>& web_arg, WebArgHeader& web_arg_header) {
+ std::memcpy(&web_arg_header, web_arg.data(), sizeof(WebArgHeader));
- if (res != nullptr) {
- return res->GetRomFS();
+ if (web_arg.size() == sizeof(WebArgHeader)) {
+ return {};
}
- if (type == FileSys::ContentRecordType::Data) {
- return FileSys::SystemArchive::SynthesizeSystemArchive(title_id);
+ WebArgInputTLVMap input_tlv_map;
+
+ u64 current_offset = sizeof(WebArgHeader);
+
+ for (std::size_t i = 0; i < web_arg_header.total_tlv_entries; ++i) {
+ if (web_arg.size() < current_offset + sizeof(WebArgInputTLV)) {
+ return input_tlv_map;
+ }
+
+ WebArgInputTLV input_tlv;
+ std::memcpy(&input_tlv, web_arg.data() + current_offset, sizeof(WebArgInputTLV));
+
+ current_offset += sizeof(WebArgInputTLV);
+
+ if (web_arg.size() < current_offset + input_tlv.arg_data_size) {
+ return input_tlv_map;
+ }
+
+ std::vector<u8> data(input_tlv.arg_data_size);
+ std::memcpy(data.data(), web_arg.data() + current_offset, input_tlv.arg_data_size);
+
+ current_offset += input_tlv.arg_data_size;
+
+ input_tlv_map.insert_or_assign(input_tlv.input_tlv_type, std::move(data));
}
- return nullptr;
+ return input_tlv_map;
}
-} // Anonymous namespace
+FileSys::VirtualFile GetOfflineRomFS(Core::System& system, u64 title_id,
+ FileSys::ContentRecordType nca_type) {
+ if (nca_type == FileSys::ContentRecordType::Data) {
+ const auto nca =
+ system.GetFileSystemController().GetSystemNANDContents()->GetEntry(title_id, nca_type);
+
+ if (nca == nullptr) {
+ LOG_ERROR(Service_AM,
+ "NCA of type={} with title_id={:016X} is not found in the System NAND!",
+ nca_type, title_id);
+ return FileSys::SystemArchive::SynthesizeSystemArchive(title_id);
+ }
-WebBrowser::WebBrowser(Core::System& system_, Core::Frontend::WebBrowserApplet& frontend_,
- Core::Frontend::ECommerceApplet* frontend_e_commerce_)
- : Applet{system_.Kernel()}, frontend(frontend_),
- frontend_e_commerce(frontend_e_commerce_), system{system_} {}
+ return nca->GetRomFS();
+ } else {
+ const auto nca = system.GetContentProvider().GetEntry(title_id, nca_type);
-WebBrowser::~WebBrowser() = default;
+ if (nca == nullptr) {
+ LOG_ERROR(Service_AM,
+ "NCA of type={} with title_id={:016X} is not found in the ContentProvider!",
+ nca_type, title_id);
+ return nullptr;
+ }
-void WebBrowser::Initialize() {
- Applet::Initialize();
+ const FileSys::PatchManager pm{title_id, system.GetFileSystemController(),
+ system.GetContentProvider()};
- complete = false;
- temporary_dir.clear();
- filename.clear();
- status = RESULT_SUCCESS;
+ return pm.PatchRomFS(nca->GetRomFS(), nca->GetBaseIVFCOffset(), nca_type);
+ }
+}
- const auto web_arg_storage = broker.PopNormalDataToApplet();
- ASSERT(web_arg_storage != nullptr);
- const auto& web_arg = web_arg_storage->GetData();
+void ExtractSharedFonts(Core::System& system) {
+ static constexpr std::array<const char*, 7> DECRYPTED_SHARED_FONTS{
+ "FontStandard.ttf",
+ "FontChineseSimplified.ttf",
+ "FontExtendedChineseSimplified.ttf",
+ "FontChineseTraditional.ttf",
+ "FontKorean.ttf",
+ "FontNintendoExtended.ttf",
+ "FontNintendoExtended2.ttf",
+ };
- ASSERT(web_arg.size() >= 0x8);
- std::memcpy(&kind, web_arg.data() + 0x4, sizeof(ShimKind));
+ for (std::size_t i = 0; i < NS::SHARED_FONTS.size(); ++i) {
+ const auto fonts_dir = Common::FS::SanitizePath(
+ fmt::format("{}/fonts", Common::FS::GetUserPath(Common::FS::UserPath::CacheDir)),
+ Common::FS::DirectorySeparator::PlatformDefault);
- args = GetWebArguments(web_arg);
+ const auto font_file_path =
+ Common::FS::SanitizePath(fmt::format("{}/{}", fonts_dir, DECRYPTED_SHARED_FONTS[i]),
+ Common::FS::DirectorySeparator::PlatformDefault);
- InitializeInternal();
-}
+ if (Common::FS::Exists(font_file_path)) {
+ continue;
+ }
-bool WebBrowser::TransactionComplete() const {
- return complete;
-}
+ const auto font = NS::SHARED_FONTS[i];
+ const auto font_title_id = static_cast<u64>(font.first);
-ResultCode WebBrowser::GetStatus() const {
- return status;
-}
+ const auto nca = system.GetFileSystemController().GetSystemNANDContents()->GetEntry(
+ font_title_id, FileSys::ContentRecordType::Data);
-void WebBrowser::ExecuteInteractive() {
- UNIMPLEMENTED_MSG("Unexpected interactive data recieved!");
-}
+ FileSys::VirtualFile romfs;
-void WebBrowser::Execute() {
- if (complete) {
- return;
- }
+ if (!nca) {
+ romfs = FileSys::SystemArchive::SynthesizeSystemArchive(font_title_id);
+ } else {
+ romfs = nca->GetRomFS();
+ }
- if (status != RESULT_SUCCESS) {
- complete = true;
+ if (!romfs) {
+ LOG_ERROR(Service_AM, "SharedFont RomFS with title_id={:016X} cannot be extracted!",
+ font_title_id);
+ continue;
+ }
- // This is a workaround in order not to softlock yuzu when an error happens during the
- // webapplet init. In order to avoid an svcBreak, the status is set to RESULT_SUCCESS
- Finalize();
- status = RESULT_SUCCESS;
+ const auto extracted_romfs = FileSys::ExtractRomFS(romfs);
- return;
- }
+ if (!extracted_romfs) {
+ LOG_ERROR(Service_AM, "SharedFont RomFS with title_id={:016X} failed to extract!",
+ font_title_id);
+ continue;
+ }
- ExecuteInternal();
-}
+ const auto font_file = extracted_romfs->GetFile(font.second);
-void WebBrowser::UnpackRomFS() {
- if (unpacked)
- return;
+ if (!font_file) {
+ LOG_ERROR(Service_AM, "SharedFont RomFS with title_id={:016X} has no font file \"{}\"!",
+ font_title_id, font.second);
+ continue;
+ }
- ASSERT(offline_romfs != nullptr);
- const auto dir =
- FileSys::ExtractRomFS(offline_romfs, FileSys::RomFSExtractionType::SingleDiscard);
- const auto& vfs{system.GetFilesystem()};
- const auto temp_dir = vfs->CreateDirectory(temporary_dir, FileSys::Mode::ReadWrite);
- FileSys::VfsRawCopyD(dir, temp_dir);
+ std::vector<u32> font_data_u32(font_file->GetSize() / sizeof(u32));
+ font_file->ReadBytes<u32>(font_data_u32.data(), font_file->GetSize());
- unpacked = true;
-}
+ std::transform(font_data_u32.begin(), font_data_u32.end(), font_data_u32.begin(),
+ Common::swap32);
-void WebBrowser::Finalize() {
- complete = true;
+ std::vector<u8> decrypted_data(font_file->GetSize() - 8);
- WebCommonReturnValue out{};
- out.result_code = 0;
- out.last_url_size = 0;
+ NS::DecryptSharedFontToTTF(font_data_u32, decrypted_data);
- std::vector<u8> data(sizeof(WebCommonReturnValue));
- std::memcpy(data.data(), &out, sizeof(WebCommonReturnValue));
+ FileSys::VirtualFile decrypted_font = std::make_shared<FileSys::VectorVfsFile>(
+ std::move(decrypted_data), DECRYPTED_SHARED_FONTS[i]);
- broker.PushNormalDataFromApplet(std::make_shared<IStorage>(system, std::move(data)));
- broker.SignalStateChanged();
+ const auto temp_dir =
+ system.GetFilesystem()->CreateDirectory(fonts_dir, FileSys::Mode::ReadWrite);
+
+ const auto out_file = temp_dir->CreateFile(DECRYPTED_SHARED_FONTS[i]);
- if (!temporary_dir.empty() && Common::FS::IsDirectory(temporary_dir)) {
- Common::FS::DeleteDirRecursively(temporary_dir);
+ FileSys::VfsRawCopy(decrypted_font, out_file);
}
}
-void WebBrowser::InitializeInternal() {
- using WebAppletInitializer = void (WebBrowser::*)();
+} // namespace
- constexpr std::array<WebAppletInitializer, SHIM_KIND_COUNT> functions{
- nullptr, &WebBrowser::InitializeShop,
- nullptr, &WebBrowser::InitializeOffline,
- nullptr, nullptr,
- nullptr, nullptr,
- };
+WebBrowser::WebBrowser(Core::System& system_, const Core::Frontend::WebBrowserApplet& frontend_)
+ : Applet{system_.Kernel()}, frontend(frontend_), system{system_} {}
- const auto index = static_cast<u32>(kind);
+WebBrowser::~WebBrowser() = default;
- if (index > functions.size() || functions[index] == nullptr) {
- LOG_ERROR(Service_AM, "Invalid shim_kind={:08X}", index);
- return;
- }
+void WebBrowser::Initialize() {
+ Applet::Initialize();
- const auto function = functions[index];
- (this->*function)();
-}
+ LOG_INFO(Service_AM, "Initializing Web Browser Applet.");
-void WebBrowser::ExecuteInternal() {
- using WebAppletExecutor = void (WebBrowser::*)();
+ LOG_DEBUG(Service_AM,
+ "Initializing Applet with common_args: arg_version={}, lib_version={}, "
+ "play_startup_sound={}, size={}, system_tick={}, theme_color={}",
+ common_args.arguments_version, common_args.library_version,
+ common_args.play_startup_sound, common_args.size, common_args.system_tick,
+ common_args.theme_color);
- constexpr std::array<WebAppletExecutor, SHIM_KIND_COUNT> functions{
- nullptr, &WebBrowser::ExecuteShop,
- nullptr, &WebBrowser::ExecuteOffline,
- nullptr, nullptr,
- nullptr, nullptr,
- };
+ web_applet_version = WebAppletVersion{common_args.library_version};
- const auto index = static_cast<u32>(kind);
+ const auto web_arg_storage = broker.PopNormalDataToApplet();
+ ASSERT(web_arg_storage != nullptr);
- if (index > functions.size() || functions[index] == nullptr) {
- LOG_ERROR(Service_AM, "Invalid shim_kind={:08X}", index);
- return;
- }
+ const auto& web_arg = web_arg_storage->GetData();
+ ASSERT_OR_EXECUTE(web_arg.size() >= sizeof(WebArgHeader), { return; });
- const auto function = functions[index];
- (this->*function)();
-}
+ web_arg_input_tlv_map = ReadWebArgs(web_arg, web_arg_header);
-void WebBrowser::InitializeShop() {
- if (frontend_e_commerce == nullptr) {
- LOG_ERROR(Service_AM, "Missing ECommerce Applet frontend!");
- status = RESULT_UNKNOWN;
- return;
- }
+ LOG_DEBUG(Service_AM, "WebArgHeader: total_tlv_entries={}, shim_kind={}",
+ web_arg_header.total_tlv_entries, web_arg_header.shim_kind);
- const auto user_id_data = args.find(WebArgTLVType::UserID);
+ ExtractSharedFonts(system);
- user_id = std::nullopt;
- if (user_id_data != args.end()) {
- user_id = u128{};
- std::memcpy(user_id->data(), user_id_data->second.data(), sizeof(u128));
+ switch (web_arg_header.shim_kind) {
+ case ShimKind::Shop:
+ InitializeShop();
+ break;
+ case ShimKind::Login:
+ InitializeLogin();
+ break;
+ case ShimKind::Offline:
+ InitializeOffline();
+ break;
+ case ShimKind::Share:
+ InitializeShare();
+ break;
+ case ShimKind::Web:
+ InitializeWeb();
+ break;
+ case ShimKind::Wifi:
+ InitializeWifi();
+ break;
+ case ShimKind::Lobby:
+ InitializeLobby();
+ break;
+ default:
+ UNREACHABLE_MSG("Invalid ShimKind={}", web_arg_header.shim_kind);
+ break;
}
+}
- const auto url = args.find(WebArgTLVType::ShopArgumentsURL);
+bool WebBrowser::TransactionComplete() const {
+ return complete;
+}
- if (url == args.end()) {
- LOG_ERROR(Service_AM, "Missing EShop Arguments URL for initialization!");
- status = RESULT_UNKNOWN;
- return;
- }
+ResultCode WebBrowser::GetStatus() const {
+ return status;
+}
- std::vector<std::string> split_query;
- Common::SplitString(Common::StringFromFixedZeroTerminatedBuffer(
- reinterpret_cast<const char*>(url->second.data()), url->second.size()),
- '?', split_query);
-
- // 2 -> Main URL '?' Query Parameters
- // Less is missing info, More is malformed
- if (split_query.size() != 2) {
- LOG_ERROR(Service_AM, "EShop Arguments has more than one question mark, malformed");
- status = RESULT_UNKNOWN;
- return;
- }
+void WebBrowser::ExecuteInteractive() {
+ UNIMPLEMENTED_MSG("WebSession is not implemented");
+}
- std::vector<std::string> queries;
- Common::SplitString(split_query[1], '&', queries);
+void WebBrowser::Execute() {
+ switch (web_arg_header.shim_kind) {
+ case ShimKind::Shop:
+ ExecuteShop();
+ break;
+ case ShimKind::Login:
+ ExecuteLogin();
+ break;
+ case ShimKind::Offline:
+ ExecuteOffline();
+ break;
+ case ShimKind::Share:
+ ExecuteShare();
+ break;
+ case ShimKind::Web:
+ ExecuteWeb();
+ break;
+ case ShimKind::Wifi:
+ ExecuteWifi();
+ break;
+ case ShimKind::Lobby:
+ ExecuteLobby();
+ break;
+ default:
+ UNREACHABLE_MSG("Invalid ShimKind={}", web_arg_header.shim_kind);
+ WebBrowserExit(WebExitReason::EndButtonPressed);
+ break;
+ }
+}
- const auto split_single_query =
- [](const std::string& in) -> std::pair<std::string, std::string> {
- const auto index = in.find('=');
- if (index == std::string::npos || index == in.size() - 1) {
- return {in, ""};
- }
+void WebBrowser::ExtractOfflineRomFS() {
+ LOG_DEBUG(Service_AM, "Extracting RomFS to {}", offline_cache_dir);
- return {in.substr(0, index), in.substr(index + 1)};
- };
+ const auto extracted_romfs_dir =
+ FileSys::ExtractRomFS(offline_romfs, FileSys::RomFSExtractionType::SingleDiscard);
- std::transform(queries.begin(), queries.end(),
- std::inserter(shop_query, std::next(shop_query.begin())), split_single_query);
+ const auto temp_dir =
+ system.GetFilesystem()->CreateDirectory(offline_cache_dir, FileSys::Mode::ReadWrite);
- const auto scene = shop_query.find("scene");
+ FileSys::VfsRawCopyD(extracted_romfs_dir, temp_dir);
+}
- if (scene == shop_query.end()) {
- LOG_ERROR(Service_AM, "No scene parameter was passed via shop query!");
- status = RESULT_UNKNOWN;
- return;
+void WebBrowser::WebBrowserExit(WebExitReason exit_reason, std::string last_url) {
+ if ((web_arg_header.shim_kind == ShimKind::Share &&
+ web_applet_version >= WebAppletVersion::Version196608) ||
+ (web_arg_header.shim_kind == ShimKind::Web &&
+ web_applet_version >= WebAppletVersion::Version524288)) {
+ // TODO: Push Output TLVs instead of a WebCommonReturnValue
}
- const std::map<std::string, ShopWebTarget, std::less<>> target_map{
- {"product_detail", ShopWebTarget::ApplicationInfo},
- {"aocs", ShopWebTarget::AddOnContentList},
- {"subscriptions", ShopWebTarget::SubscriptionList},
- {"consumption", ShopWebTarget::ConsumableItemList},
- {"settings", ShopWebTarget::Settings},
- {"top", ShopWebTarget::Home},
- };
+ WebCommonReturnValue web_common_return_value;
- const auto target = target_map.find(scene->second);
- if (target == target_map.end()) {
- LOG_ERROR(Service_AM, "Scene for shop query is invalid! (scene={})", scene->second);
- status = RESULT_UNKNOWN;
- return;
- }
+ web_common_return_value.exit_reason = exit_reason;
+ std::memcpy(&web_common_return_value.last_url, last_url.data(), last_url.size());
+ web_common_return_value.last_url_size = last_url.size();
- shop_web_target = target->second;
+ LOG_DEBUG(Service_AM, "WebCommonReturnValue: exit_reason={}, last_url={}, last_url_size={}",
+ exit_reason, last_url, last_url.size());
- const auto title_id_data = shop_query.find("dst_app_id");
- if (title_id_data != shop_query.end()) {
- title_id = std::stoull(title_id_data->second, nullptr, 0x10);
- }
+ complete = true;
+ std::vector<u8> out_data(sizeof(WebCommonReturnValue));
+ std::memcpy(out_data.data(), &web_common_return_value, out_data.size());
+ broker.PushNormalDataFromApplet(std::make_shared<IStorage>(system, std::move(out_data)));
+ broker.SignalStateChanged();
+}
- const auto mode_data = shop_query.find("mode");
- if (mode_data != shop_query.end()) {
- shop_full_display = mode_data->second == "full";
- }
+bool WebBrowser::InputTLVExistsInMap(WebArgInputTLVType input_tlv_type) const {
+ return web_arg_input_tlv_map.find(input_tlv_type) != web_arg_input_tlv_map.end();
}
-void WebBrowser::InitializeOffline() {
- if (args.find(WebArgTLVType::DocumentPath) == args.end() ||
- args.find(WebArgTLVType::DocumentKind) == args.end() ||
- args.find(WebArgTLVType::ApplicationID) == args.end()) {
- status = RESULT_UNKNOWN;
- LOG_ERROR(Service_AM, "Missing necessary parameters for initialization!");
+std::optional<std::vector<u8>> WebBrowser::GetInputTLVData(WebArgInputTLVType input_tlv_type) {
+ const auto map_it = web_arg_input_tlv_map.find(input_tlv_type);
+
+ if (map_it == web_arg_input_tlv_map.end()) {
+ return std::nullopt;
}
- const auto url_data = args[WebArgTLVType::DocumentPath];
- filename = Common::StringFromFixedZeroTerminatedBuffer(
- reinterpret_cast<const char*>(url_data.data()), url_data.size());
+ return map_it->second;
+}
- OfflineWebSource source;
- ASSERT(args[WebArgTLVType::DocumentKind].size() >= 4);
- std::memcpy(&source, args[WebArgTLVType::DocumentKind].data(), sizeof(OfflineWebSource));
+void WebBrowser::InitializeShop() {}
- constexpr std::array<const char*, 3> WEB_SOURCE_NAMES{
- "manual",
- "legal",
- "system",
- };
+void WebBrowser::InitializeLogin() {}
+
+void WebBrowser::InitializeOffline() {
+ const auto document_path =
+ ParseStringValue(GetInputTLVData(WebArgInputTLVType::DocumentPath).value());
+
+ const auto document_kind =
+ ParseRawValue<DocumentKind>(GetInputTLVData(WebArgInputTLVType::DocumentKind).value());
+
+ std::string additional_paths;
- temporary_dir =
- Common::FS::SanitizePath(Common::FS::GetUserPath(Common::FS::UserPath::CacheDir) +
- "web_applet_" + WEB_SOURCE_NAMES[static_cast<u32>(source) - 1],
- Common::FS::DirectorySeparator::PlatformDefault);
- Common::FS::DeleteDirRecursively(temporary_dir);
-
- u64 title_id = 0; // 0 corresponds to current process
- ASSERT(args[WebArgTLVType::ApplicationID].size() >= 0x8);
- std::memcpy(&title_id, args[WebArgTLVType::ApplicationID].data(), sizeof(u64));
- FileSys::ContentRecordType type = FileSys::ContentRecordType::Data;
-
- switch (source) {
- case OfflineWebSource::OfflineHtmlPage:
- // While there is an AppID TLV field, in official SW this is always ignored.
- title_id = 0;
- type = FileSys::ContentRecordType::HtmlDocument;
+ switch (document_kind) {
+ case DocumentKind::OfflineHtmlPage:
+ default:
+ title_id = system.CurrentProcess()->GetTitleID();
+ nca_type = FileSys::ContentRecordType::HtmlDocument;
+ additional_paths = "html-document";
break;
- case OfflineWebSource::ApplicationLegalInformation:
- type = FileSys::ContentRecordType::LegalInformation;
+ case DocumentKind::ApplicationLegalInformation:
+ title_id = ParseRawValue<u64>(GetInputTLVData(WebArgInputTLVType::ApplicationID).value());
+ nca_type = FileSys::ContentRecordType::LegalInformation;
break;
- case OfflineWebSource::SystemDataPage:
- type = FileSys::ContentRecordType::Data;
+ case DocumentKind::SystemDataPage:
+ title_id = ParseRawValue<u64>(GetInputTLVData(WebArgInputTLVType::SystemDataID).value());
+ nca_type = FileSys::ContentRecordType::Data;
break;
}
- if (title_id == 0) {
- title_id = system.CurrentProcess()->GetTitleID();
- }
+ static constexpr std::array<const char*, 3> RESOURCE_TYPES{
+ "manual",
+ "legal_information",
+ "system_data",
+ };
- offline_romfs = GetApplicationRomFS(system, title_id, type);
- if (offline_romfs == nullptr) {
- status = RESULT_UNKNOWN;
- LOG_ERROR(Service_AM, "Failed to find offline data for request!");
- }
+ offline_cache_dir = Common::FS::SanitizePath(
+ fmt::format("{}/offline_web_applet_{}/{:016X}",
+ Common::FS::GetUserPath(Common::FS::UserPath::CacheDir),
+ RESOURCE_TYPES[static_cast<u32>(document_kind) - 1], title_id),
+ Common::FS::DirectorySeparator::PlatformDefault);
- std::string path_additional_directory;
- if (source == OfflineWebSource::OfflineHtmlPage) {
- path_additional_directory = std::string(DIR_SEP).append("html-document");
- }
+ offline_document = Common::FS::SanitizePath(
+ fmt::format("{}/{}/{}", offline_cache_dir, additional_paths, document_path),
+ Common::FS::DirectorySeparator::PlatformDefault);
+}
+
+void WebBrowser::InitializeShare() {}
- filename =
- Common::FS::SanitizePath(temporary_dir + path_additional_directory + DIR_SEP + filename,
- Common::FS::DirectorySeparator::PlatformDefault);
+void WebBrowser::InitializeWeb() {
+ external_url = ParseStringValue(GetInputTLVData(WebArgInputTLVType::InitialURL).value());
}
+void WebBrowser::InitializeWifi() {}
+
+void WebBrowser::InitializeLobby() {}
+
void WebBrowser::ExecuteShop() {
- const auto callback = [this]() { Finalize(); };
+ LOG_WARNING(Service_AM, "(STUBBED) called, Shop Applet is not implemented");
+ WebBrowserExit(WebExitReason::EndButtonPressed);
+}
- const auto check_optional_parameter = [this](const auto& p) {
- if (!p.has_value()) {
- LOG_ERROR(Service_AM, "Missing one or more necessary parameters for execution!");
- status = RESULT_UNKNOWN;
- return false;
- }
+void WebBrowser::ExecuteLogin() {
+ LOG_WARNING(Service_AM, "(STUBBED) called, Login Applet is not implemented");
+ WebBrowserExit(WebExitReason::EndButtonPressed);
+}
- return true;
- };
+void WebBrowser::ExecuteOffline() {
+ const auto main_url = Common::FS::SanitizePath(GetMainURL(offline_document),
+ Common::FS::DirectorySeparator::PlatformDefault);
- switch (shop_web_target) {
- case ShopWebTarget::ApplicationInfo:
- if (!check_optional_parameter(title_id))
- return;
- frontend_e_commerce->ShowApplicationInformation(callback, *title_id, user_id,
- shop_full_display, shop_extra_parameter);
- break;
- case ShopWebTarget::AddOnContentList:
- if (!check_optional_parameter(title_id))
- return;
- frontend_e_commerce->ShowAddOnContentList(callback, *title_id, user_id, shop_full_display);
- break;
- case ShopWebTarget::ConsumableItemList:
- if (!check_optional_parameter(title_id))
- return;
- frontend_e_commerce->ShowConsumableItemList(callback, *title_id, user_id);
- break;
- case ShopWebTarget::Home:
- if (!check_optional_parameter(user_id))
- return;
- if (!check_optional_parameter(shop_full_display))
- return;
- frontend_e_commerce->ShowShopHome(callback, *user_id, *shop_full_display);
- break;
- case ShopWebTarget::Settings:
- if (!check_optional_parameter(user_id))
- return;
- if (!check_optional_parameter(shop_full_display))
- return;
- frontend_e_commerce->ShowSettings(callback, *user_id, *shop_full_display);
- break;
- case ShopWebTarget::SubscriptionList:
- if (!check_optional_parameter(title_id))
+ if (!Common::FS::Exists(main_url)) {
+ offline_romfs = GetOfflineRomFS(system, title_id, nca_type);
+
+ if (offline_romfs == nullptr) {
+ LOG_ERROR(Service_AM,
+ "RomFS with title_id={:016X} and nca_type={} cannot be extracted!", title_id,
+ nca_type);
+ WebBrowserExit(WebExitReason::WindowClosed);
return;
- frontend_e_commerce->ShowSubscriptionList(callback, *title_id, user_id);
- break;
- default:
- UNREACHABLE();
+ }
}
+
+ LOG_INFO(Service_AM, "Opening offline document at {}", offline_document);
+
+ frontend.OpenLocalWebPage(
+ offline_document, [this] { ExtractOfflineRomFS(); },
+ [this](WebExitReason exit_reason, std::string last_url) {
+ WebBrowserExit(exit_reason, last_url);
+ });
}
-void WebBrowser::ExecuteOffline() {
- frontend.OpenPageLocal(
- filename, [this] { UnpackRomFS(); }, [this] { Finalize(); });
+void WebBrowser::ExecuteShare() {
+ LOG_WARNING(Service_AM, "(STUBBED) called, Share Applet is not implemented");
+ WebBrowserExit(WebExitReason::EndButtonPressed);
+}
+
+void WebBrowser::ExecuteWeb() {
+ LOG_INFO(Service_AM, "Opening external URL at {}", external_url);
+
+ frontend.OpenExternalWebPage(external_url,
+ [this](WebExitReason exit_reason, std::string last_url) {
+ WebBrowserExit(exit_reason, last_url);
+ });
}
+void WebBrowser::ExecuteWifi() {
+ LOG_WARNING(Service_AM, "(STUBBED) called, Wifi Applet is not implemented");
+ WebBrowserExit(WebExitReason::EndButtonPressed);
+}
+
+void WebBrowser::ExecuteLobby() {
+ LOG_WARNING(Service_AM, "(STUBBED) called, Lobby Applet is not implemented");
+ WebBrowserExit(WebExitReason::EndButtonPressed);
+}
} // namespace Service::AM::Applets
diff --git a/src/core/hle/service/am/applets/web_browser.h b/src/core/hle/service/am/applets/web_browser.h
index 8d4027411..04c274754 100644
--- a/src/core/hle/service/am/applets/web_browser.h
+++ b/src/core/hle/service/am/applets/web_browser.h
@@ -1,28 +1,31 @@
-// Copyright 2018 yuzu emulator team
+// Copyright 2020 yuzu Emulator Project
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
#pragma once
-#include <map>
+#include <optional>
+
+#include "common/common_funcs.h"
+#include "common/common_types.h"
#include "core/file_sys/vfs_types.h"
-#include "core/hle/service/am/am.h"
+#include "core/hle/result.h"
#include "core/hle/service/am/applets/applets.h"
+#include "core/hle/service/am/applets/web_types.h"
namespace Core {
class System;
}
-namespace Service::AM::Applets {
+namespace FileSys {
+enum class ContentRecordType : u8;
+}
-enum class ShimKind : u32;
-enum class ShopWebTarget;
-enum class WebArgTLVType : u16;
+namespace Service::AM::Applets {
class WebBrowser final : public Applet {
public:
- WebBrowser(Core::System& system_, Core::Frontend::WebBrowserApplet& frontend_,
- Core::Frontend::ECommerceApplet* frontend_e_commerce_ = nullptr);
+ WebBrowser(Core::System& system_, const Core::Frontend::WebBrowserApplet& frontend_);
~WebBrowser() override;
@@ -33,49 +36,50 @@ public:
void ExecuteInteractive() override;
void Execute() override;
- // Callback to be fired when the frontend needs the manual RomFS unpacked to temporary
- // directory. This is a blocking call and may take a while as some manuals can be up to 100MB in
- // size. Attempting to access files at filename before invocation is likely to not work.
- void UnpackRomFS();
+ void ExtractOfflineRomFS();
- // Callback to be fired when the frontend is finished browsing. This will delete the temporary
- // manual RomFS extracted files, so ensure this is only called at actual finalization.
- void Finalize();
+ void WebBrowserExit(WebExitReason exit_reason, std::string last_url = "");
private:
- void InitializeInternal();
- void ExecuteInternal();
+ bool InputTLVExistsInMap(WebArgInputTLVType input_tlv_type) const;
- // Specific initializers for the types of web applets
+ std::optional<std::vector<u8>> GetInputTLVData(WebArgInputTLVType input_tlv_type);
+
+ // Initializers for the various types of browser applets
void InitializeShop();
+ void InitializeLogin();
void InitializeOffline();
+ void InitializeShare();
+ void InitializeWeb();
+ void InitializeWifi();
+ void InitializeLobby();
- // Specific executors for the types of web applets
+ // Executors for the various types of browser applets
void ExecuteShop();
+ void ExecuteLogin();
void ExecuteOffline();
+ void ExecuteShare();
+ void ExecuteWeb();
+ void ExecuteWifi();
+ void ExecuteLobby();
- Core::Frontend::WebBrowserApplet& frontend;
-
- // Extra frontends for specialized functions
- Core::Frontend::ECommerceApplet* frontend_e_commerce;
+ const Core::Frontend::WebBrowserApplet& frontend;
- bool complete = false;
- bool unpacked = false;
- ResultCode status = RESULT_SUCCESS;
+ bool complete{false};
+ ResultCode status{RESULT_SUCCESS};
- ShimKind kind;
- std::map<WebArgTLVType, std::vector<u8>> args;
+ WebAppletVersion web_applet_version;
+ WebExitReason web_exit_reason;
+ WebArgHeader web_arg_header;
+ WebArgInputTLVMap web_arg_input_tlv_map;
+ u64 title_id;
+ FileSys::ContentRecordType nca_type;
+ std::string offline_cache_dir;
+ std::string offline_document;
FileSys::VirtualFile offline_romfs;
- std::string temporary_dir;
- std::string filename;
-
- ShopWebTarget shop_web_target;
- std::map<std::string, std::string, std::less<>> shop_query;
- std::optional<u64> title_id = 0;
- std::optional<u128> user_id;
- std::optional<bool> shop_full_display;
- std::string shop_extra_parameter;
+
+ std::string external_url;
Core::System& system;
};
diff --git a/src/core/hle/service/am/applets/web_types.h b/src/core/hle/service/am/applets/web_types.h
new file mode 100644
index 000000000..419c2bf79
--- /dev/null
+++ b/src/core/hle/service/am/applets/web_types.h
@@ -0,0 +1,178 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include <unordered_map>
+#include <vector>
+
+#include "common/common_funcs.h"
+#include "common/common_types.h"
+#include "common/swap.h"
+
+namespace Service::AM::Applets {
+
+enum class WebAppletVersion : u32_le {
+ Version0 = 0x0, // Only used by WifiWebAuthApplet
+ Version131072 = 0x20000, // 1.0.0 - 2.3.0
+ Version196608 = 0x30000, // 3.0.0 - 4.1.0
+ Version327680 = 0x50000, // 5.0.0 - 5.1.0
+ Version393216 = 0x60000, // 6.0.0 - 7.0.1
+ Version524288 = 0x80000, // 8.0.0+
+};
+
+enum class ShimKind : u32 {
+ Shop = 1,
+ Login = 2,
+ Offline = 3,
+ Share = 4,
+ Web = 5,
+ Wifi = 6,
+ Lobby = 7,
+};
+
+enum class WebExitReason : u32 {
+ EndButtonPressed = 0,
+ BackButtonPressed = 1,
+ ExitRequested = 2,
+ CallbackURL = 3,
+ WindowClosed = 4,
+ ErrorDialog = 7,
+};
+
+enum class WebArgInputTLVType : u16 {
+ InitialURL = 0x1,
+ CallbackURL = 0x3,
+ CallbackableURL = 0x4,
+ ApplicationID = 0x5,
+ DocumentPath = 0x6,
+ DocumentKind = 0x7,
+ SystemDataID = 0x8,
+ ShareStartPage = 0x9,
+ Whitelist = 0xA,
+ News = 0xB,
+ UserID = 0xE,
+ AlbumEntry0 = 0xF,
+ ScreenShotEnabled = 0x10,
+ EcClientCertEnabled = 0x11,
+ PlayReportEnabled = 0x13,
+ BootDisplayKind = 0x17,
+ BackgroundKind = 0x18,
+ FooterEnabled = 0x19,
+ PointerEnabled = 0x1A,
+ LeftStickMode = 0x1B,
+ KeyRepeatFrame1 = 0x1C,
+ KeyRepeatFrame2 = 0x1D,
+ BootAsMediaPlayerInverted = 0x1E,
+ DisplayURLKind = 0x1F,
+ BootAsMediaPlayer = 0x21,
+ ShopJumpEnabled = 0x22,
+ MediaAutoPlayEnabled = 0x23,
+ LobbyParameter = 0x24,
+ ApplicationAlbumEntry = 0x26,
+ JsExtensionEnabled = 0x27,
+ AdditionalCommentText = 0x28,
+ TouchEnabledOnContents = 0x29,
+ UserAgentAdditionalString = 0x2A,
+ AdditionalMediaData0 = 0x2B,
+ MediaPlayerAutoCloseEnabled = 0x2C,
+ PageCacheEnabled = 0x2D,
+ WebAudioEnabled = 0x2E,
+ YouTubeVideoWhitelist = 0x31,
+ FooterFixedKind = 0x32,
+ PageFadeEnabled = 0x33,
+ MediaCreatorApplicationRatingAge = 0x34,
+ BootLoadingIconEnabled = 0x35,
+ PageScrollIndicatorEnabled = 0x36,
+ MediaPlayerSpeedControlEnabled = 0x37,
+ AlbumEntry1 = 0x38,
+ AlbumEntry2 = 0x39,
+ AlbumEntry3 = 0x3A,
+ AdditionalMediaData1 = 0x3B,
+ AdditionalMediaData2 = 0x3C,
+ AdditionalMediaData3 = 0x3D,
+ BootFooterButton = 0x3E,
+ OverrideWebAudioVolume = 0x3F,
+ OverrideMediaAudioVolume = 0x40,
+ BootMode = 0x41,
+ WebSessionEnabled = 0x42,
+ MediaPlayerOfflineEnabled = 0x43,
+};
+
+enum class WebArgOutputTLVType : u16 {
+ ShareExitReason = 0x1,
+ LastURL = 0x2,
+ LastURLSize = 0x3,
+ SharePostResult = 0x4,
+ PostServiceName = 0x5,
+ PostServiceNameSize = 0x6,
+ PostID = 0x7,
+ PostIDSize = 0x8,
+ MediaPlayerAutoClosedByCompletion = 0x9,
+};
+
+enum class DocumentKind : u32 {
+ OfflineHtmlPage = 1,
+ ApplicationLegalInformation = 2,
+ SystemDataPage = 3,
+};
+
+enum class ShareStartPage : u32 {
+ Default,
+ Settings,
+};
+
+enum class BootDisplayKind : u32 {
+ Default,
+ White,
+ Black,
+};
+
+enum class BackgroundKind : u32 {
+ Default,
+};
+
+enum class LeftStickMode : u32 {
+ Pointer,
+ Cursor,
+};
+
+enum class WebSessionBootMode : u32 {
+ AllForeground,
+ AllForegroundInitiallyHidden,
+};
+
+struct WebArgHeader {
+ u16 total_tlv_entries{};
+ INSERT_PADDING_BYTES(2);
+ ShimKind shim_kind{};
+};
+static_assert(sizeof(WebArgHeader) == 0x8, "WebArgHeader has incorrect size.");
+
+struct WebArgInputTLV {
+ WebArgInputTLVType input_tlv_type{};
+ u16 arg_data_size{};
+ INSERT_PADDING_WORDS(1);
+};
+static_assert(sizeof(WebArgInputTLV) == 0x8, "WebArgInputTLV has incorrect size.");
+
+struct WebArgOutputTLV {
+ WebArgOutputTLVType output_tlv_type{};
+ u16 arg_data_size{};
+ INSERT_PADDING_WORDS(1);
+};
+static_assert(sizeof(WebArgOutputTLV) == 0x8, "WebArgOutputTLV has incorrect size.");
+
+struct WebCommonReturnValue {
+ WebExitReason exit_reason{};
+ INSERT_PADDING_WORDS(1);
+ std::array<char, 0x1000> last_url{};
+ u64 last_url_size{};
+};
+static_assert(sizeof(WebCommonReturnValue) == 0x1010, "WebCommonReturnValue has incorrect size.");
+
+using WebArgInputTLVMap = std::unordered_map<WebArgInputTLVType, std::vector<u8>>;
+
+} // namespace Service::AM::Applets
diff --git a/src/core/hle/service/aoc/aoc_u.cpp b/src/core/hle/service/aoc/aoc_u.cpp
index 6abac3f78..23e28565b 100644
--- a/src/core/hle/service/aoc/aoc_u.cpp
+++ b/src/core/hle/service/aoc/aoc_u.cpp
@@ -7,6 +7,7 @@
#include <vector>
#include "common/logging/log.h"
#include "core/core.h"
+#include "core/file_sys/common_funcs.h"
#include "core/file_sys/content_archive.h"
#include "core/file_sys/control_metadata.h"
#include "core/file_sys/nca_metadata.h"
@@ -23,11 +24,8 @@
namespace Service::AOC {
-constexpr u64 DLC_BASE_TITLE_ID_MASK = 0xFFFFFFFFFFFFE000;
-constexpr u64 DLC_BASE_TO_AOC_ID = 0x1000;
-
static bool CheckAOCTitleIDMatchesBase(u64 title_id, u64 base) {
- return (title_id & DLC_BASE_TITLE_ID_MASK) == base;
+ return FileSys::GetBaseTitleID(title_id) == base;
}
static std::vector<u64> AccumulateAOCTitleIDs(Core::System& system) {
@@ -48,6 +46,62 @@ static std::vector<u64> AccumulateAOCTitleIDs(Core::System& system) {
return add_on_content;
}
+class IPurchaseEventManager final : public ServiceFramework<IPurchaseEventManager> {
+public:
+ explicit IPurchaseEventManager(Core::System& system_)
+ : ServiceFramework{system_, "IPurchaseEventManager"} {
+ // clang-format off
+ static const FunctionInfo functions[] = {
+ {0, &IPurchaseEventManager::SetDefaultDeliveryTarget, "SetDefaultDeliveryTarget"},
+ {1, &IPurchaseEventManager::SetDeliveryTarget, "SetDeliveryTarget"},
+ {2, &IPurchaseEventManager::GetPurchasedEventReadableHandle, "GetPurchasedEventReadableHandle"},
+ {3, nullptr, "PopPurchasedProductInfo"},
+ {4, nullptr, "PopPurchasedProductInfoWithUid"},
+ };
+ // clang-format on
+
+ RegisterHandlers(functions);
+
+ purchased_event = Kernel::WritableEvent::CreateEventPair(
+ system.Kernel(), "IPurchaseEventManager:PurchasedEvent");
+ }
+
+private:
+ void SetDefaultDeliveryTarget(Kernel::HLERequestContext& ctx) {
+ IPC::RequestParser rp{ctx};
+
+ const auto unknown_1 = rp.Pop<u64>();
+ [[maybe_unused]] const auto unknown_2 = ctx.ReadBuffer();
+
+ LOG_WARNING(Service_AOC, "(STUBBED) called, unknown_1={}", unknown_1);
+
+ IPC::ResponseBuilder rb{ctx, 2};
+ rb.Push(RESULT_SUCCESS);
+ }
+
+ void SetDeliveryTarget(Kernel::HLERequestContext& ctx) {
+ IPC::RequestParser rp{ctx};
+
+ const auto unknown_1 = rp.Pop<u64>();
+ [[maybe_unused]] const auto unknown_2 = ctx.ReadBuffer();
+
+ LOG_WARNING(Service_AOC, "(STUBBED) called, unknown_1={}", unknown_1);
+
+ IPC::ResponseBuilder rb{ctx, 2};
+ rb.Push(RESULT_SUCCESS);
+ }
+
+ void GetPurchasedEventReadableHandle(Kernel::HLERequestContext& ctx) {
+ LOG_WARNING(Service_AOC, "called");
+
+ IPC::ResponseBuilder rb{ctx, 2, 1};
+ rb.Push(RESULT_SUCCESS);
+ rb.PushCopyObjects(purchased_event.readable);
+ }
+
+ Kernel::EventPair purchased_event;
+};
+
AOC_U::AOC_U(Core::System& system_)
: ServiceFramework{system_, "aoc:u"}, add_on_content{AccumulateAOCTitleIDs(system)} {
// clang-format off
@@ -62,8 +116,8 @@ AOC_U::AOC_U(Core::System& system_)
{7, &AOC_U::PrepareAddOnContent, "PrepareAddOnContent"},
{8, &AOC_U::GetAddOnContentListChangedEvent, "GetAddOnContentListChangedEvent"},
{9, nullptr, "GetAddOnContentLostErrorCode"},
- {100, nullptr, "CreateEcPurchasedEventManager"},
- {101, nullptr, "CreatePermanentEcPurchasedEventManager"},
+ {100, &AOC_U::CreateEcPurchasedEventManager, "CreateEcPurchasedEventManager"},
+ {101, &AOC_U::CreatePermanentEcPurchasedEventManager, "CreatePermanentEcPurchasedEventManager"},
};
// clang-format on
@@ -123,11 +177,11 @@ void AOC_U::ListAddOnContent(Kernel::HLERequestContext& ctx) {
const auto& disabled = Settings::values.disabled_addons[current];
if (std::find(disabled.begin(), disabled.end(), "DLC") == disabled.end()) {
for (u64 content_id : add_on_content) {
- if ((content_id & DLC_BASE_TITLE_ID_MASK) != current) {
+ if (FileSys::GetBaseTitleID(content_id) != current) {
continue;
}
- out.push_back(static_cast<u32>(content_id & 0x7FF));
+ out.push_back(static_cast<u32>(FileSys::GetAOCID(content_id)));
}
}
@@ -169,7 +223,7 @@ void AOC_U::GetAddOnContentBaseId(Kernel::HLERequestContext& ctx) {
const auto res = pm.GetControlMetadata();
if (res.first == nullptr) {
- rb.Push(title_id + DLC_BASE_TO_AOC_ID);
+ rb.Push(FileSys::GetAOCBaseTitleID(title_id));
return;
}
@@ -201,6 +255,22 @@ void AOC_U::GetAddOnContentListChangedEvent(Kernel::HLERequestContext& ctx) {
rb.PushCopyObjects(aoc_change_event.readable);
}
+void AOC_U::CreateEcPurchasedEventManager(Kernel::HLERequestContext& ctx) {
+ LOG_WARNING(Service_AOC, "(STUBBED) called");
+
+ IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+ rb.Push(RESULT_SUCCESS);
+ rb.PushIpcInterface<IPurchaseEventManager>(system);
+}
+
+void AOC_U::CreatePermanentEcPurchasedEventManager(Kernel::HLERequestContext& ctx) {
+ LOG_WARNING(Service_AOC, "(STUBBED) called");
+
+ IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+ rb.Push(RESULT_SUCCESS);
+ rb.PushIpcInterface<IPurchaseEventManager>(system);
+}
+
void InstallInterfaces(SM::ServiceManager& service_manager, Core::System& system) {
std::make_shared<AOC_U>(system)->InstallAsService(service_manager);
}
diff --git a/src/core/hle/service/aoc/aoc_u.h b/src/core/hle/service/aoc/aoc_u.h
index 7628f4568..26ee51be0 100644
--- a/src/core/hle/service/aoc/aoc_u.h
+++ b/src/core/hle/service/aoc/aoc_u.h
@@ -27,6 +27,8 @@ private:
void GetAddOnContentBaseId(Kernel::HLERequestContext& ctx);
void PrepareAddOnContent(Kernel::HLERequestContext& ctx);
void GetAddOnContentListChangedEvent(Kernel::HLERequestContext& ctx);
+ void CreateEcPurchasedEventManager(Kernel::HLERequestContext& ctx);
+ void CreatePermanentEcPurchasedEventManager(Kernel::HLERequestContext& ctx);
std::vector<u64> add_on_content;
Kernel::EventPair aoc_change_event;
diff --git a/src/core/hle/service/apm/controller.cpp b/src/core/hle/service/apm/controller.cpp
index ce993bad3..03636642b 100644
--- a/src/core/hle/service/apm/controller.cpp
+++ b/src/core/hle/service/apm/controller.cpp
@@ -48,8 +48,7 @@ void Controller::SetPerformanceConfiguration(PerformanceMode mode,
[config](const auto& entry) { return entry.first == config; });
if (iter == config_to_speed.cend()) {
- LOG_ERROR(Service_APM, "Invalid performance configuration value provided: {}",
- static_cast<u32>(config));
+ LOG_ERROR(Service_APM, "Invalid performance configuration value provided: {}", config);
return;
}
diff --git a/src/core/hle/service/apm/interface.cpp b/src/core/hle/service/apm/interface.cpp
index 89442e21e..298f6d520 100644
--- a/src/core/hle/service/apm/interface.cpp
+++ b/src/core/hle/service/apm/interface.cpp
@@ -28,8 +28,7 @@ private:
const auto mode = rp.PopEnum<PerformanceMode>();
const auto config = rp.PopEnum<PerformanceConfiguration>();
- LOG_DEBUG(Service_APM, "called mode={} config={}", static_cast<u32>(mode),
- static_cast<u32>(config));
+ LOG_DEBUG(Service_APM, "called mode={} config={}", mode, config);
controller.SetPerformanceConfiguration(mode, config);
@@ -41,7 +40,7 @@ private:
IPC::RequestParser rp{ctx};
const auto mode = rp.PopEnum<PerformanceMode>();
- LOG_DEBUG(Service_APM, "called mode={}", static_cast<u32>(mode));
+ LOG_DEBUG(Service_APM, "called mode={}", mode);
IPC::ResponseBuilder rb{ctx, 3};
rb.Push(RESULT_SUCCESS);
@@ -111,7 +110,7 @@ void APM_Sys::SetCpuBoostMode(Kernel::HLERequestContext& ctx) {
IPC::RequestParser rp{ctx};
const auto mode = rp.PopEnum<CpuBoostMode>();
- LOG_DEBUG(Service_APM, "called, mode={:08X}", static_cast<u32>(mode));
+ LOG_DEBUG(Service_APM, "called, mode={:08X}", mode);
controller.SetFromCpuBoostMode(mode);
diff --git a/src/core/hle/service/audio/audout_u.cpp b/src/core/hle/service/audio/audout_u.cpp
index 145f47ee2..0cd797109 100644
--- a/src/core/hle/service/audio/audout_u.cpp
+++ b/src/core/hle/service/audio/audout_u.cpp
@@ -70,8 +70,10 @@ public:
Kernel::WritableEvent::CreateEventPair(system.Kernel(), "IAudioOutBufferReleased");
stream = audio_core.OpenStream(system.CoreTiming(), audio_params.sample_rate,
- audio_params.channel_count, std::move(unique_name),
- [this] { buffer_event.writable->Signal(); });
+ audio_params.channel_count, std::move(unique_name), [this] {
+ const auto guard = LockService();
+ buffer_event.writable->Signal();
+ });
}
private:
diff --git a/src/core/hle/service/audio/audren_u.cpp b/src/core/hle/service/audio/audren_u.cpp
index 6e7b7316c..c5c22d053 100644
--- a/src/core/hle/service/audio/audren_u.cpp
+++ b/src/core/hle/service/audio/audren_u.cpp
@@ -49,16 +49,16 @@ public:
system_event =
Kernel::WritableEvent::CreateEventPair(system.Kernel(), "IAudioRenderer:SystemEvent");
- renderer = std::make_unique<AudioCore::AudioRenderer>(system.CoreTiming(), system.Memory(),
- audren_params, system_event.writable,
- instance_number);
+ renderer = std::make_unique<AudioCore::AudioRenderer>(
+ system.CoreTiming(), system.Memory(), audren_params,
+ [this]() {
+ const auto guard = LockService();
+ system_event.writable->Signal();
+ },
+ instance_number);
}
private:
- void UpdateAudioCallback() {
- system_event.writable->Signal();
- }
-
void GetSampleRate(Kernel::HLERequestContext& ctx) {
LOG_DEBUG(Service_Audio, "called");
diff --git a/src/core/hle/service/bcat/backend/boxcat.cpp b/src/core/hle/service/bcat/backend/boxcat.cpp
index 3b6f7498e..e43f3f47f 100644
--- a/src/core/hle/service/bcat/backend/boxcat.cpp
+++ b/src/core/hle/service/bcat/backend/boxcat.cpp
@@ -483,7 +483,7 @@ Boxcat::StatusResult Boxcat::GetStatus(std::optional<std::string>& global,
global = json["global"].get<std::string>();
if (json["games"].is_array()) {
- for (const auto object : json["games"]) {
+ for (const auto& object : json["games"]) {
if (object.is_object() && object.find("name") != object.end()) {
EventStatus detail{};
if (object["header"].is_string()) {
diff --git a/src/core/hle/service/fatal/fatal.cpp b/src/core/hle/service/fatal/fatal.cpp
index 9b7672a91..13147472e 100644
--- a/src/core/hle/service/fatal/fatal.cpp
+++ b/src/core/hle/service/fatal/fatal.cpp
@@ -111,8 +111,9 @@ static void GenerateErrorReport(Core::System& system, ResultCode error_code,
static void ThrowFatalError(Core::System& system, ResultCode error_code, FatalType fatal_type,
const FatalInfo& info) {
- LOG_ERROR(Service_Fatal, "Threw fatal error type {} with error code 0x{:X}",
- static_cast<u32>(fatal_type), error_code.raw);
+ LOG_ERROR(Service_Fatal, "Threw fatal error type {} with error code 0x{:X}", fatal_type,
+ error_code.raw);
+
switch (fatal_type) {
case FatalType::ErrorReportAndScreen:
GenerateErrorReport(system, error_code, info);
diff --git a/src/core/hle/service/filesystem/filesystem.cpp b/src/core/hle/service/filesystem/filesystem.cpp
index ca93062cf..b15c737e1 100644
--- a/src/core/hle/service/filesystem/filesystem.cpp
+++ b/src/core/hle/service/filesystem/filesystem.cpp
@@ -298,10 +298,35 @@ ResultVal<FileSys::VirtualFile> FileSystemController::OpenRomFSCurrentProcess()
return romfs_factory->OpenCurrentProcess(system.CurrentProcess()->GetTitleID());
}
+ResultVal<FileSys::VirtualFile> FileSystemController::OpenPatchedRomFS(
+ u64 title_id, FileSys::ContentRecordType type) const {
+ LOG_TRACE(Service_FS, "Opening patched RomFS for title_id={:016X}", title_id);
+
+ if (romfs_factory == nullptr) {
+ // TODO: Find a better error code for this
+ return RESULT_UNKNOWN;
+ }
+
+ return romfs_factory->OpenPatchedRomFS(title_id, type);
+}
+
+ResultVal<FileSys::VirtualFile> FileSystemController::OpenPatchedRomFSWithProgramIndex(
+ u64 title_id, u8 program_index, FileSys::ContentRecordType type) const {
+ LOG_TRACE(Service_FS, "Opening patched RomFS for title_id={:016X}, program_index={}", title_id,
+ program_index);
+
+ if (romfs_factory == nullptr) {
+ // TODO: Find a better error code for this
+ return RESULT_UNKNOWN;
+ }
+
+ return romfs_factory->OpenPatchedRomFSWithProgramIndex(title_id, program_index, type);
+}
+
ResultVal<FileSys::VirtualFile> FileSystemController::OpenRomFS(
u64 title_id, FileSys::StorageId storage_id, FileSys::ContentRecordType type) const {
LOG_TRACE(Service_FS, "Opening RomFS for title_id={:016X}, storage_id={:02X}, type={:02X}",
- title_id, static_cast<u8>(storage_id), static_cast<u8>(type));
+ title_id, storage_id, type);
if (romfs_factory == nullptr) {
// TODO(bunnei): Find a better error code for this
@@ -313,8 +338,8 @@ ResultVal<FileSys::VirtualFile> FileSystemController::OpenRomFS(
ResultVal<FileSys::VirtualDir> FileSystemController::CreateSaveData(
FileSys::SaveDataSpaceId space, const FileSys::SaveDataAttribute& save_struct) const {
- LOG_TRACE(Service_FS, "Creating Save Data for space_id={:01X}, save_struct={}",
- static_cast<u8>(space), save_struct.DebugInfo());
+ LOG_TRACE(Service_FS, "Creating Save Data for space_id={:01X}, save_struct={}", space,
+ save_struct.DebugInfo());
if (save_data_factory == nullptr) {
return FileSys::ERROR_ENTITY_NOT_FOUND;
@@ -325,8 +350,8 @@ ResultVal<FileSys::VirtualDir> FileSystemController::CreateSaveData(
ResultVal<FileSys::VirtualDir> FileSystemController::OpenSaveData(
FileSys::SaveDataSpaceId space, const FileSys::SaveDataAttribute& attribute) const {
- LOG_TRACE(Service_FS, "Opening Save Data for space_id={:01X}, save_struct={}",
- static_cast<u8>(space), attribute.DebugInfo());
+ LOG_TRACE(Service_FS, "Opening Save Data for space_id={:01X}, save_struct={}", space,
+ attribute.DebugInfo());
if (save_data_factory == nullptr) {
return FileSys::ERROR_ENTITY_NOT_FOUND;
@@ -337,7 +362,7 @@ ResultVal<FileSys::VirtualDir> FileSystemController::OpenSaveData(
ResultVal<FileSys::VirtualDir> FileSystemController::OpenSaveDataSpace(
FileSys::SaveDataSpaceId space) const {
- LOG_TRACE(Service_FS, "Opening Save Data Space for space_id={:01X}", static_cast<u8>(space));
+ LOG_TRACE(Service_FS, "Opening Save Data Space for space_id={:01X}", space);
if (save_data_factory == nullptr) {
return FileSys::ERROR_ENTITY_NOT_FOUND;
@@ -358,7 +383,7 @@ ResultVal<FileSys::VirtualDir> FileSystemController::OpenSDMC() const {
ResultVal<FileSys::VirtualDir> FileSystemController::OpenBISPartition(
FileSys::BisPartitionId id) const {
- LOG_TRACE(Service_FS, "Opening BIS Partition with id={:08X}", static_cast<u32>(id));
+ LOG_TRACE(Service_FS, "Opening BIS Partition with id={:08X}", id);
if (bis_factory == nullptr) {
return FileSys::ERROR_ENTITY_NOT_FOUND;
@@ -374,7 +399,7 @@ ResultVal<FileSys::VirtualDir> FileSystemController::OpenBISPartition(
ResultVal<FileSys::VirtualFile> FileSystemController::OpenBISPartitionStorage(
FileSys::BisPartitionId id) const {
- LOG_TRACE(Service_FS, "Opening BIS Partition Storage with id={:08X}", static_cast<u32>(id));
+ LOG_TRACE(Service_FS, "Opening BIS Partition Storage with id={:08X}", id);
if (bis_factory == nullptr) {
return FileSys::ERROR_ENTITY_NOT_FOUND;
diff --git a/src/core/hle/service/filesystem/filesystem.h b/src/core/hle/service/filesystem/filesystem.h
index 6dbbf0b2b..7102d3f9a 100644
--- a/src/core/hle/service/filesystem/filesystem.h
+++ b/src/core/hle/service/filesystem/filesystem.h
@@ -66,6 +66,10 @@ public:
void SetPackedUpdate(FileSys::VirtualFile update_raw);
ResultVal<FileSys::VirtualFile> OpenRomFSCurrentProcess() const;
+ ResultVal<FileSys::VirtualFile> OpenPatchedRomFS(u64 title_id,
+ FileSys::ContentRecordType type) const;
+ ResultVal<FileSys::VirtualFile> OpenPatchedRomFSWithProgramIndex(
+ u64 title_id, u8 program_index, FileSys::ContentRecordType type) const;
ResultVal<FileSys::VirtualFile> OpenRomFS(u64 title_id, FileSys::StorageId storage_id,
FileSys::ContentRecordType type) const;
ResultVal<FileSys::VirtualDir> CreateSaveData(
diff --git a/src/core/hle/service/filesystem/fsp_srv.cpp b/src/core/hle/service/filesystem/fsp_srv.cpp
index b3480494c..9cc260515 100644
--- a/src/core/hle/service/filesystem/fsp_srv.cpp
+++ b/src/core/hle/service/filesystem/fsp_srv.cpp
@@ -413,7 +413,7 @@ public:
const auto mode = static_cast<FileSys::Mode>(rp.Pop<u32>());
- LOG_DEBUG(Service_FS, "called. file={}, mode={}", name, static_cast<u32>(mode));
+ LOG_DEBUG(Service_FS, "called. file={}, mode={}", name, mode);
auto result = backend.OpenFile(name, mode);
if (result.Failed()) {
@@ -553,8 +553,7 @@ private:
const auto save_root = fsc.OpenSaveDataSpace(space);
if (save_root.Failed() || *save_root == nullptr) {
- LOG_ERROR(Service_FS, "The save root for the space_id={:02X} was invalid!",
- static_cast<u8>(space));
+ LOG_ERROR(Service_FS, "The save root for the space_id={:02X} was invalid!", space);
return;
}
@@ -718,7 +717,7 @@ FSP_SRV::FSP_SRV(Core::System& system_)
{202, &FSP_SRV::OpenDataStorageByDataId, "OpenDataStorageByDataId"},
{203, &FSP_SRV::OpenPatchDataStorageByCurrentProcess, "OpenPatchDataStorageByCurrentProcess"},
{204, nullptr, "OpenDataFileSystemByProgramIndex"},
- {205, nullptr, "OpenDataStorageByProgramIndex"},
+ {205, &FSP_SRV::OpenDataStorageWithProgramIndex, "OpenDataStorageWithProgramIndex"},
{400, nullptr, "OpenDeviceOperator"},
{500, nullptr, "OpenSdCardDetectionEventNotifier"},
{501, nullptr, "OpenGameCardDetectionEventNotifier"},
@@ -795,8 +794,7 @@ void FSP_SRV::OpenFileSystemWithPatch(Kernel::HLERequestContext& ctx) {
const auto type = rp.PopRaw<FileSystemType>();
const auto title_id = rp.PopRaw<u64>();
- LOG_WARNING(Service_FS, "(STUBBED) called with type={}, title_id={:016X}",
- static_cast<u8>(type), title_id);
+ LOG_WARNING(Service_FS, "(STUBBED) called with type={}, title_id={:016X}", type, title_id);
IPC::ResponseBuilder rb{ctx, 2, 0, 0};
rb.Push(RESULT_UNKNOWN);
@@ -883,7 +881,7 @@ void FSP_SRV::OpenReadOnlySaveDataFileSystem(Kernel::HLERequestContext& ctx) {
void FSP_SRV::OpenSaveDataInfoReaderBySaveDataSpaceId(Kernel::HLERequestContext& ctx) {
IPC::RequestParser rp{ctx};
const auto space = rp.PopRaw<FileSys::SaveDataSpaceId>();
- LOG_INFO(Service_FS, "called, space={}", static_cast<u8>(space));
+ LOG_INFO(Service_FS, "called, space={}", space);
IPC::ResponseBuilder rb{ctx, 2, 0, 1};
rb.Push(RESULT_SUCCESS);
@@ -915,10 +913,10 @@ void FSP_SRV::ReadSaveDataFileSystemExtraDataWithMaskBySaveDataAttribute(
"(STUBBED) called, flags={}, space_id={}, attribute.title_id={:016X}\n"
"attribute.user_id={:016X}{:016X}, attribute.save_id={:016X}\n"
"attribute.type={}, attribute.rank={}, attribute.index={}",
- flags, static_cast<u32>(parameters.space_id), parameters.attribute.title_id,
+ flags, parameters.space_id, parameters.attribute.title_id,
parameters.attribute.user_id[1], parameters.attribute.user_id[0],
- parameters.attribute.save_id, static_cast<u32>(parameters.attribute.type),
- static_cast<u32>(parameters.attribute.rank), parameters.attribute.index);
+ parameters.attribute.save_id, parameters.attribute.type, parameters.attribute.rank,
+ parameters.attribute.index);
IPC::ResponseBuilder rb{ctx, 3};
rb.Push(RESULT_SUCCESS);
@@ -951,7 +949,7 @@ void FSP_SRV::OpenDataStorageByDataId(Kernel::HLERequestContext& ctx) {
const auto title_id = rp.PopRaw<u64>();
LOG_DEBUG(Service_FS, "called with storage_id={:02X}, unknown={:08X}, title_id={:016X}",
- static_cast<u8>(storage_id), unknown, title_id);
+ storage_id, unknown, title_id);
auto data = fsc.OpenRomFS(title_id, storage_id, FileSys::ContentRecordType::Data);
@@ -968,7 +966,7 @@ void FSP_SRV::OpenDataStorageByDataId(Kernel::HLERequestContext& ctx) {
// TODO(DarkLordZach): Find the right error code to use here
LOG_ERROR(Service_FS,
"could not open data storage with title_id={:016X}, storage_id={:02X}", title_id,
- static_cast<u8>(storage_id));
+ storage_id);
IPC::ResponseBuilder rb{ctx, 2};
rb.Push(RESULT_UNKNOWN);
return;
@@ -987,21 +985,46 @@ void FSP_SRV::OpenDataStorageByDataId(Kernel::HLERequestContext& ctx) {
void FSP_SRV::OpenPatchDataStorageByCurrentProcess(Kernel::HLERequestContext& ctx) {
IPC::RequestParser rp{ctx};
- auto storage_id = rp.PopRaw<FileSys::StorageId>();
- auto title_id = rp.PopRaw<u64>();
+ const auto storage_id = rp.PopRaw<FileSys::StorageId>();
+ const auto title_id = rp.PopRaw<u64>();
- LOG_DEBUG(Service_FS, "called with storage_id={:02X}, title_id={:016X}",
- static_cast<u8>(storage_id), title_id);
+ LOG_DEBUG(Service_FS, "called with storage_id={:02X}, title_id={:016X}", storage_id, title_id);
IPC::ResponseBuilder rb{ctx, 2};
rb.Push(FileSys::ERROR_ENTITY_NOT_FOUND);
}
+void FSP_SRV::OpenDataStorageWithProgramIndex(Kernel::HLERequestContext& ctx) {
+ IPC::RequestParser rp{ctx};
+
+ const auto program_index = rp.PopRaw<u8>();
+
+ LOG_DEBUG(Service_FS, "called, program_index={}", program_index);
+
+ auto romfs = fsc.OpenPatchedRomFSWithProgramIndex(
+ system.CurrentProcess()->GetTitleID(), program_index, FileSys::ContentRecordType::Program);
+
+ if (romfs.Failed()) {
+ // TODO: Find the right error code to use here
+ LOG_ERROR(Service_FS, "could not open storage with program_index={}", program_index);
+
+ IPC::ResponseBuilder rb{ctx, 2};
+ rb.Push(RESULT_UNKNOWN);
+ return;
+ }
+
+ auto storage = std::make_shared<IStorage>(system, std::move(romfs.Unwrap()));
+
+ IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+ rb.Push(RESULT_SUCCESS);
+ rb.PushIpcInterface<IStorage>(std::move(storage));
+}
+
void FSP_SRV::SetGlobalAccessLogMode(Kernel::HLERequestContext& ctx) {
IPC::RequestParser rp{ctx};
log_mode = rp.PopEnum<LogMode>();
- LOG_DEBUG(Service_FS, "called, log_mode={:08X}", static_cast<u32>(log_mode));
+ LOG_DEBUG(Service_FS, "called, log_mode={:08X}", log_mode);
IPC::ResponseBuilder rb{ctx, 2};
rb.Push(RESULT_SUCCESS);
diff --git a/src/core/hle/service/filesystem/fsp_srv.h b/src/core/hle/service/filesystem/fsp_srv.h
index 472286d6e..8ed933279 100644
--- a/src/core/hle/service/filesystem/fsp_srv.h
+++ b/src/core/hle/service/filesystem/fsp_srv.h
@@ -49,6 +49,7 @@ private:
void OpenDataStorageByCurrentProcess(Kernel::HLERequestContext& ctx);
void OpenDataStorageByDataId(Kernel::HLERequestContext& ctx);
void OpenPatchDataStorageByCurrentProcess(Kernel::HLERequestContext& ctx);
+ void OpenDataStorageWithProgramIndex(Kernel::HLERequestContext& ctx);
void SetGlobalAccessLogMode(Kernel::HLERequestContext& ctx);
void GetGlobalAccessLogMode(Kernel::HLERequestContext& ctx);
void OutputAccessLogToSdCard(Kernel::HLERequestContext& ctx);
diff --git a/src/core/hle/service/friend/friend.cpp b/src/core/hle/service/friend/friend.cpp
index 40a289594..c5b053c31 100644
--- a/src/core/hle/service/friend/friend.cpp
+++ b/src/core/hle/service/friend/friend.cpp
@@ -229,8 +229,7 @@ private:
break;
default:
// HOS seems not have an error case for an unknown notification
- LOG_WARNING(Service_ACC, "Unknown notification {:08X}",
- static_cast<u32>(notification.notification_type));
+ LOG_WARNING(Service_ACC, "Unknown notification {:08X}", notification.notification_type);
break;
}
diff --git a/src/core/hle/service/hid/controllers/npad.cpp b/src/core/hle/service/hid/controllers/npad.cpp
index 66c4fe60a..d280e7caf 100644
--- a/src/core/hle/service/hid/controllers/npad.cpp
+++ b/src/core/hle/service/hid/controllers/npad.cpp
@@ -116,6 +116,31 @@ u32 Controller_NPad::IndexToNPad(std::size_t index) {
}
}
+bool Controller_NPad::IsNpadIdValid(u32 npad_id) {
+ switch (npad_id) {
+ case 0:
+ case 1:
+ case 2:
+ case 3:
+ case 4:
+ case 5:
+ case 6:
+ case 7:
+ case NPAD_UNKNOWN:
+ case NPAD_HANDHELD:
+ return true;
+ default:
+ LOG_ERROR(Service_HID, "Invalid npad id {}", npad_id);
+ return false;
+ }
+}
+
+bool Controller_NPad::IsDeviceHandleValid(const DeviceHandle& device_handle) {
+ return IsNpadIdValid(device_handle.npad_id) &&
+ device_handle.npad_type < NpadType::MaxNpadType &&
+ device_handle.device_index < DeviceIndex::MaxDeviceIndex;
+}
+
Controller_NPad::Controller_NPad(Core::System& system) : ControllerBase(system), system(system) {}
Controller_NPad::~Controller_NPad() {
@@ -742,6 +767,10 @@ bool Controller_NPad::VibrateControllerAtIndex(std::size_t npad_index, std::size
void Controller_NPad::VibrateController(const DeviceHandle& vibration_device_handle,
const VibrationValue& vibration_value) {
+ if (!IsDeviceHandleValid(vibration_device_handle)) {
+ return;
+ }
+
if (!Settings::values.vibration_enabled.GetValue() && !permit_vibration_session_enabled) {
return;
}
@@ -798,12 +827,20 @@ void Controller_NPad::VibrateControllers(const std::vector<DeviceHandle>& vibrat
Controller_NPad::VibrationValue Controller_NPad::GetLastVibration(
const DeviceHandle& vibration_device_handle) const {
+ if (!IsDeviceHandleValid(vibration_device_handle)) {
+ return {};
+ }
+
const auto npad_index = NPadIdToIndex(vibration_device_handle.npad_id);
const auto device_index = static_cast<std::size_t>(vibration_device_handle.device_index);
return latest_vibration_values[npad_index][device_index];
}
void Controller_NPad::InitializeVibrationDevice(const DeviceHandle& vibration_device_handle) {
+ if (!IsDeviceHandleValid(vibration_device_handle)) {
+ return;
+ }
+
const auto npad_index = NPadIdToIndex(vibration_device_handle.npad_id);
const auto device_index = static_cast<std::size_t>(vibration_device_handle.device_index);
InitializeVibrationDeviceAtIndex(npad_index, device_index);
@@ -824,6 +861,10 @@ void Controller_NPad::SetPermitVibrationSession(bool permit_vibration_session) {
}
bool Controller_NPad::IsVibrationDeviceMounted(const DeviceHandle& vibration_device_handle) const {
+ if (!IsDeviceHandleValid(vibration_device_handle)) {
+ return false;
+ }
+
const auto npad_index = NPadIdToIndex(vibration_device_handle.npad_id);
const auto device_index = static_cast<std::size_t>(vibration_device_handle.device_index);
return vibration_devices_mounted[npad_index][device_index];
@@ -1017,7 +1058,7 @@ void Controller_NPad::ClearAllControllers() {
}
u32 Controller_NPad::GetAndResetPressState() {
- return std::exchange(press_state, 0);
+ return press_state.exchange(0);
}
bool Controller_NPad::IsControllerSupported(NPadControllerType controller) const {
diff --git a/src/core/hle/service/hid/controllers/npad.h b/src/core/hle/service/hid/controllers/npad.h
index 96f319294..e2e826623 100644
--- a/src/core/hle/service/hid/controllers/npad.h
+++ b/src/core/hle/service/hid/controllers/npad.h
@@ -5,6 +5,7 @@
#pragma once
#include <array>
+#include <atomic>
#include "common/bit_field.h"
#include "common/common_types.h"
#include "core/frontend/input.h"
@@ -56,12 +57,14 @@ public:
JoyconLeft = 6,
JoyconRight = 7,
Pokeball = 9,
+ MaxNpadType = 10,
};
enum class DeviceIndex : u8 {
Left = 0,
Right = 1,
None = 2,
+ MaxDeviceIndex = 3,
};
enum class GyroscopeZeroDriftMode : u32 {
@@ -213,6 +216,8 @@ public:
static Settings::ControllerType MapNPadToSettingsType(Controller_NPad::NPadControllerType type);
static std::size_t NPadIdToIndex(u32 npad_id);
static u32 IndexToNPad(std::size_t index);
+ static bool IsNpadIdValid(u32 npad_id);
+ static bool IsDeviceHandleValid(const DeviceHandle& device_handle);
private:
struct CommonHeader {
@@ -411,7 +416,7 @@ private:
bool IsControllerSupported(NPadControllerType controller) const;
void RequestPadStateUpdate(u32 npad_id);
- u32 press_state{};
+ std::atomic<u32> press_state{};
NpadStyleSet style{};
std::array<NPadEntry, 10> shared_memory_entries{};
diff --git a/src/core/hle/service/hid/hid.cpp b/src/core/hle/service/hid/hid.cpp
index b3c7234e1..8d95f74e6 100644
--- a/src/core/hle/service/hid/hid.cpp
+++ b/src/core/hle/service/hid/hid.cpp
@@ -78,11 +78,13 @@ IAppletResource::IAppletResource(Core::System& system_)
pad_update_event = Core::Timing::CreateEvent(
"HID::UpdatePadCallback",
[this](std::uintptr_t user_data, std::chrono::nanoseconds ns_late) {
+ const auto guard = LockService();
UpdateControllers(user_data, ns_late);
});
motion_update_event = Core::Timing::CreateEvent(
"HID::MotionPadCallback",
[this](std::uintptr_t user_data, std::chrono::nanoseconds ns_late) {
+ const auto guard = LockService();
UpdateMotion(user_data, ns_late);
});
diff --git a/src/core/hle/service/lm/lm.cpp b/src/core/hle/service/lm/lm.cpp
index f884b2735..8e49b068c 100644
--- a/src/core/hle/service/lm/lm.cpp
+++ b/src/core/hle/service/lm/lm.cpp
@@ -68,7 +68,7 @@ private:
IPC::RequestParser rp{ctx};
const auto destination = rp.PopEnum<DestinationFlag>();
- LOG_DEBUG(Service_LM, "called, destination={:08X}", static_cast<u32>(destination));
+ LOG_DEBUG(Service_LM, "called, destination={:08X}", destination);
manager.SetDestination(destination);
diff --git a/src/core/hle/service/ncm/ncm.cpp b/src/core/hle/service/ncm/ncm.cpp
index b8d627ca8..2dcda16f6 100644
--- a/src/core/hle/service/ncm/ncm.cpp
+++ b/src/core/hle/service/ncm/ncm.cpp
@@ -45,7 +45,7 @@ public:
}
private:
- FileSys::StorageId storage;
+ [[maybe_unused]] FileSys::StorageId storage;
};
class IRegisteredLocationResolver final : public ServiceFramework<IRegisteredLocationResolver> {
diff --git a/src/core/hle/service/nim/nim.cpp b/src/core/hle/service/nim/nim.cpp
index d33b26129..d16223064 100644
--- a/src/core/hle/service/nim/nim.cpp
+++ b/src/core/hle/service/nim/nim.cpp
@@ -217,7 +217,7 @@ public:
{1, nullptr, "RefreshDebugAvailability"},
{2, nullptr, "ClearDebugResponse"},
{3, nullptr, "RegisterDebugResponse"},
- {4, nullptr, "IsLargeResourceAvailable"},
+ {4, &NIM_ECA::IsLargeResourceAvailable, "IsLargeResourceAvailable"},
};
// clang-format on
@@ -231,6 +231,18 @@ private:
rb.Push(RESULT_SUCCESS);
rb.PushIpcInterface<IShopServiceAccessServer>(system);
}
+
+ void IsLargeResourceAvailable(Kernel::HLERequestContext& ctx) {
+ IPC::RequestParser rp{ctx};
+
+ const auto unknown{rp.Pop<u64>()};
+
+ LOG_INFO(Service_NIM, "(STUBBED) called, unknown={}", unknown);
+
+ IPC::ResponseBuilder rb{ctx, 3};
+ rb.Push(RESULT_SUCCESS);
+ rb.Push(false);
+ }
};
class NIM_SHP final : public ServiceFramework<NIM_SHP> {
diff --git a/src/core/hle/service/ns/ns.cpp b/src/core/hle/service/ns/ns.cpp
index ef7584641..6ccf8995c 100644
--- a/src/core/hle/service/ns/ns.cpp
+++ b/src/core/hle/service/ns/ns.cpp
@@ -673,7 +673,7 @@ public:
explicit NS_VM(Core::System& system_) : ServiceFramework{system_, "ns:vm"} {
// clang-format off
static const FunctionInfo functions[] = {
- {1200, nullptr, "NeedsUpdateVulnerability"},
+ {1200, &NS_VM::NeedsUpdateVulnerability, "NeedsUpdateVulnerability"},
{1201, nullptr, "UpdateSafeSystemVersionForDebug"},
{1202, nullptr, "GetSafeSystemVersion"},
};
@@ -681,6 +681,15 @@ public:
RegisterHandlers(functions);
}
+
+private:
+ void NeedsUpdateVulnerability(Kernel::HLERequestContext& ctx) {
+ LOG_WARNING(Service_NS, "(STUBBED) called");
+
+ IPC::ResponseBuilder rb{ctx, 3};
+ rb.Push(RESULT_SUCCESS);
+ rb.Push(false);
+ }
};
void InstallInterfaces(SM::ServiceManager& service_manager, Core::System& system) {
diff --git a/src/core/hle/service/ns/pl_u.cpp b/src/core/hle/service/ns/pl_u.cpp
index ccc137e40..71c7587db 100644
--- a/src/core/hle/service/ns/pl_u.cpp
+++ b/src/core/hle/service/ns/pl_u.cpp
@@ -27,29 +27,11 @@
namespace Service::NS {
-enum class FontArchives : u64 {
- Extension = 0x0100000000000810,
- Standard = 0x0100000000000811,
- Korean = 0x0100000000000812,
- ChineseTraditional = 0x0100000000000813,
- ChineseSimple = 0x0100000000000814,
-};
-
struct FontRegion {
u32 offset;
u32 size;
};
-constexpr std::array<std::pair<FontArchives, const char*>, 7> SHARED_FONTS{
- std::make_pair(FontArchives::Standard, "nintendo_udsg-r_std_003.bfttf"),
- std::make_pair(FontArchives::ChineseSimple, "nintendo_udsg-r_org_zh-cn_003.bfttf"),
- std::make_pair(FontArchives::ChineseSimple, "nintendo_udsg-r_ext_zh-cn_003.bfttf"),
- std::make_pair(FontArchives::ChineseTraditional, "nintendo_udjxh-db_zh-tw_003.bfttf"),
- std::make_pair(FontArchives::Korean, "nintendo_udsg-r_ko_003.bfttf"),
- std::make_pair(FontArchives::Extension, "nintendo_ext_003.bfttf"),
- std::make_pair(FontArchives::Extension, "nintendo_ext2_003.bfttf"),
-};
-
// The below data is specific to shared font data dumped from Switch on f/w 2.2
// Virtual address and offsets/sizes likely will vary by dump
[[maybe_unused]] constexpr VAddr SHARED_FONT_MEM_VADDR{0x00000009d3016000ULL};
@@ -80,6 +62,18 @@ static void DecryptSharedFont(const std::vector<u32>& input, Kernel::PhysicalMem
offset += transformed_font.size() * sizeof(u32);
}
+void DecryptSharedFontToTTF(const std::vector<u32>& input, std::vector<u8>& output) {
+ ASSERT_MSG(input[0] == EXPECTED_MAGIC, "Failed to derive key, unexpected magic number");
+
+ const u32 KEY = input[0] ^ EXPECTED_RESULT; // Derive key using an inverse xor
+ std::vector<u32> transformed_font(input.size());
+ // TODO(ogniK): Figure out a better way to do this
+ std::transform(input.begin(), input.end(), transformed_font.begin(),
+ [&KEY](u32 font_data) { return Common::swap32(font_data ^ KEY); });
+ transformed_font[1] = Common::swap32(transformed_font[1]) ^ KEY; // "re-encrypt" the size
+ std::memcpy(output.data(), transformed_font.data() + 2, transformed_font.size() * sizeof(u32));
+}
+
void EncryptSharedFont(const std::vector<u32>& input, std::vector<u8>& output,
std::size_t& offset) {
ASSERT_MSG(offset + (input.size() * sizeof(u32)) < SHARED_FONT_MEM_SIZE,
@@ -182,21 +176,18 @@ PL_U::PL_U(Core::System& system_)
}
if (!romfs) {
- LOG_ERROR(Service_NS, "Failed to find or synthesize {:016X}! Skipping",
- static_cast<u64>(font.first));
+ LOG_ERROR(Service_NS, "Failed to find or synthesize {:016X}! Skipping", font.first);
continue;
}
const auto extracted_romfs = FileSys::ExtractRomFS(romfs);
if (!extracted_romfs) {
- LOG_ERROR(Service_NS, "Failed to extract RomFS for {:016X}! Skipping",
- static_cast<u64>(font.first));
+ LOG_ERROR(Service_NS, "Failed to extract RomFS for {:016X}! Skipping", font.first);
continue;
}
const auto font_fp = extracted_romfs->GetFile(font.second);
if (!font_fp) {
- LOG_ERROR(Service_NS, "{:016X} has no file \"{}\"! Skipping",
- static_cast<u64>(font.first), font.second);
+ LOG_ERROR(Service_NS, "{:016X} has no file \"{}\"! Skipping", font.first, font.second);
continue;
}
std::vector<u32> font_data_u32(font_fp->GetSize() / sizeof(u32));
diff --git a/src/core/hle/service/ns/pl_u.h b/src/core/hle/service/ns/pl_u.h
index 224dcb997..f920c7f69 100644
--- a/src/core/hle/service/ns/pl_u.h
+++ b/src/core/hle/service/ns/pl_u.h
@@ -16,6 +16,25 @@ class FileSystemController;
namespace NS {
+enum class FontArchives : u64 {
+ Extension = 0x0100000000000810,
+ Standard = 0x0100000000000811,
+ Korean = 0x0100000000000812,
+ ChineseTraditional = 0x0100000000000813,
+ ChineseSimple = 0x0100000000000814,
+};
+
+constexpr std::array<std::pair<FontArchives, const char*>, 7> SHARED_FONTS{
+ std::make_pair(FontArchives::Standard, "nintendo_udsg-r_std_003.bfttf"),
+ std::make_pair(FontArchives::ChineseSimple, "nintendo_udsg-r_org_zh-cn_003.bfttf"),
+ std::make_pair(FontArchives::ChineseSimple, "nintendo_udsg-r_ext_zh-cn_003.bfttf"),
+ std::make_pair(FontArchives::ChineseTraditional, "nintendo_udjxh-db_zh-tw_003.bfttf"),
+ std::make_pair(FontArchives::Korean, "nintendo_udsg-r_ko_003.bfttf"),
+ std::make_pair(FontArchives::Extension, "nintendo_ext_003.bfttf"),
+ std::make_pair(FontArchives::Extension, "nintendo_ext2_003.bfttf"),
+};
+
+void DecryptSharedFontToTTF(const std::vector<u32>& input, std::vector<u8>& output);
void EncryptSharedFont(const std::vector<u32>& input, std::vector<u8>& output, std::size_t& offset);
class PL_U final : public ServiceFramework<PL_U> {
diff --git a/src/core/hle/service/nvdrv/devices/nvdevice.h b/src/core/hle/service/nvdrv/devices/nvdevice.h
index 44a8bc060..5681599ba 100644
--- a/src/core/hle/service/nvdrv/devices/nvdevice.h
+++ b/src/core/hle/service/nvdrv/devices/nvdevice.h
@@ -31,8 +31,8 @@ public:
* @param output A buffer where the output data will be written to.
* @returns The result code of the ioctl.
*/
- virtual NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- IoctlCtrl& ctrl) = 0;
+ virtual NvResult Ioctl1(Ioctl command, const std::vector<u8>& input,
+ std::vector<u8>& output) = 0;
/**
* Handles an ioctl2 request.
@@ -43,8 +43,7 @@ public:
* @returns The result code of the ioctl.
*/
virtual NvResult Ioctl2(Ioctl command, const std::vector<u8>& input,
- const std::vector<u8>& inline_input, std::vector<u8>& output,
- IoctlCtrl& ctrl) = 0;
+ const std::vector<u8>& inline_input, std::vector<u8>& output) = 0;
/**
* Handles an ioctl3 request.
@@ -55,7 +54,7 @@ public:
* @returns The result code of the ioctl.
*/
virtual NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- std::vector<u8>& inline_output, IoctlCtrl& ctrl) = 0;
+ std::vector<u8>& inline_output) = 0;
protected:
Core::System& system;
diff --git a/src/core/hle/service/nvdrv/devices/nvdisp_disp0.cpp b/src/core/hle/service/nvdrv/devices/nvdisp_disp0.cpp
index 170a7c9a0..ce615c758 100644
--- a/src/core/hle/service/nvdrv/devices/nvdisp_disp0.cpp
+++ b/src/core/hle/service/nvdrv/devices/nvdisp_disp0.cpp
@@ -18,21 +18,20 @@ nvdisp_disp0::nvdisp_disp0(Core::System& system, std::shared_ptr<nvmap> nvmap_de
: nvdevice(system), nvmap_dev(std::move(nvmap_dev)) {}
nvdisp_disp0 ::~nvdisp_disp0() = default;
-NvResult nvdisp_disp0::Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- IoctlCtrl& ctrl) {
+NvResult nvdisp_disp0::Ioctl1(Ioctl command, const std::vector<u8>& input,
+ std::vector<u8>& output) {
UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
return NvResult::NotImplemented;
}
NvResult nvdisp_disp0::Ioctl2(Ioctl command, const std::vector<u8>& input,
- const std::vector<u8>& inline_input, std::vector<u8>& output,
- IoctlCtrl& ctrl) {
+ const std::vector<u8>& inline_input, std::vector<u8>& output) {
UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
return NvResult::NotImplemented;
}
NvResult nvdisp_disp0::Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- std::vector<u8>& inline_output, IoctlCtrl& ctrl) {
+ std::vector<u8>& inline_output) {
UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
return NvResult::NotImplemented;
}
diff --git a/src/core/hle/service/nvdrv/devices/nvdisp_disp0.h b/src/core/hle/service/nvdrv/devices/nvdisp_disp0.h
index eb7575e40..55a33b7e4 100644
--- a/src/core/hle/service/nvdrv/devices/nvdisp_disp0.h
+++ b/src/core/hle/service/nvdrv/devices/nvdisp_disp0.h
@@ -20,13 +20,11 @@ public:
explicit nvdisp_disp0(Core::System& system, std::shared_ptr<nvmap> nvmap_dev);
~nvdisp_disp0() override;
- NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- IoctlCtrl& ctrl) override;
+ NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override;
NvResult Ioctl2(Ioctl command, const std::vector<u8>& input,
- const std::vector<u8>& inline_input, std::vector<u8>& output,
- IoctlCtrl& ctrl) override;
+ const std::vector<u8>& inline_input, std::vector<u8>& output) override;
NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- std::vector<u8>& inline_output, IoctlCtrl& ctrl) override;
+ std::vector<u8>& inline_output) override;
/// Performs a screen flip, drawing the buffer pointed to by the handle.
void flip(u32 buffer_handle, u32 offset, u32 format, u32 width, u32 height, u32 stride,
diff --git a/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.cpp b/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.cpp
index 4e0652c39..6b062e10e 100644
--- a/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.cpp
+++ b/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.cpp
@@ -21,8 +21,8 @@ nvhost_as_gpu::nvhost_as_gpu(Core::System& system, std::shared_ptr<nvmap> nvmap_
: nvdevice(system), nvmap_dev(std::move(nvmap_dev)) {}
nvhost_as_gpu::~nvhost_as_gpu() = default;
-NvResult nvhost_as_gpu::Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- IoctlCtrl& ctrl) {
+NvResult nvhost_as_gpu::Ioctl1(Ioctl command, const std::vector<u8>& input,
+ std::vector<u8>& output) {
switch (command.group) {
case 'A':
switch (command.cmd) {
@@ -55,14 +55,13 @@ NvResult nvhost_as_gpu::Ioctl1(Ioctl command, const std::vector<u8>& input, std:
}
NvResult nvhost_as_gpu::Ioctl2(Ioctl command, const std::vector<u8>& input,
- const std::vector<u8>& inline_input, std::vector<u8>& output,
- IoctlCtrl& ctrl) {
+ const std::vector<u8>& inline_input, std::vector<u8>& output) {
UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
return NvResult::NotImplemented;
}
NvResult nvhost_as_gpu::Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- std::vector<u8>& inline_output, IoctlCtrl& ctrl) {
+ std::vector<u8>& inline_output) {
switch (command.group) {
case 'A':
switch (command.cmd) {
diff --git a/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.h b/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.h
index 2bd355af9..08035fa0e 100644
--- a/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.h
+++ b/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.h
@@ -30,13 +30,11 @@ public:
explicit nvhost_as_gpu(Core::System& system, std::shared_ptr<nvmap> nvmap_dev);
~nvhost_as_gpu() override;
- NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- IoctlCtrl& ctrl) override;
+ NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override;
NvResult Ioctl2(Ioctl command, const std::vector<u8>& input,
- const std::vector<u8>& inline_input, std::vector<u8>& output,
- IoctlCtrl& ctrl) override;
+ const std::vector<u8>& inline_input, std::vector<u8>& output) override;
NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- std::vector<u8>& inline_output, IoctlCtrl& ctrl) override;
+ std::vector<u8>& inline_output) override;
private:
class BufferMap final {
diff --git a/src/core/hle/service/nvdrv/devices/nvhost_ctrl.cpp b/src/core/hle/service/nvdrv/devices/nvhost_ctrl.cpp
index 92d31b620..fea3b7b9f 100644
--- a/src/core/hle/service/nvdrv/devices/nvhost_ctrl.cpp
+++ b/src/core/hle/service/nvdrv/devices/nvhost_ctrl.cpp
@@ -20,8 +20,7 @@ nvhost_ctrl::nvhost_ctrl(Core::System& system, EventInterface& events_interface,
: nvdevice(system), events_interface{events_interface}, syncpoint_manager{syncpoint_manager} {}
nvhost_ctrl::~nvhost_ctrl() = default;
-NvResult nvhost_ctrl::Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- IoctlCtrl& ctrl) {
+NvResult nvhost_ctrl::Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) {
switch (command.group) {
case 0x0:
switch (command.cmd) {
@@ -30,9 +29,9 @@ NvResult nvhost_ctrl::Ioctl1(Ioctl command, const std::vector<u8>& input, std::v
case 0x1c:
return IocCtrlClearEventWait(input, output);
case 0x1d:
- return IocCtrlEventWait(input, output, false, ctrl);
+ return IocCtrlEventWait(input, output, false);
case 0x1e:
- return IocCtrlEventWait(input, output, true, ctrl);
+ return IocCtrlEventWait(input, output, true);
case 0x1f:
return IocCtrlEventRegister(input, output);
case 0x20:
@@ -48,14 +47,13 @@ NvResult nvhost_ctrl::Ioctl1(Ioctl command, const std::vector<u8>& input, std::v
}
NvResult nvhost_ctrl::Ioctl2(Ioctl command, const std::vector<u8>& input,
- const std::vector<u8>& inline_input, std::vector<u8>& output,
- IoctlCtrl& ctrl) {
+ const std::vector<u8>& inline_input, std::vector<u8>& output) {
UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
return NvResult::NotImplemented;
}
NvResult nvhost_ctrl::Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- std::vector<u8>& inline_output, IoctlCtrl& ctrl) {
+ std::vector<u8>& inline_outpu) {
UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
return NvResult::NotImplemented;
}
@@ -69,7 +67,7 @@ NvResult nvhost_ctrl::NvOsGetConfigU32(const std::vector<u8>& input, std::vector
}
NvResult nvhost_ctrl::IocCtrlEventWait(const std::vector<u8>& input, std::vector<u8>& output,
- bool is_async, IoctlCtrl& ctrl) {
+ bool is_async) {
IocCtrlEventWaitParams params{};
std::memcpy(&params, input.data(), sizeof(params));
LOG_DEBUG(Service_NVDRV, "syncpt_id={}, threshold={}, timeout={}, is_async={}",
@@ -141,12 +139,6 @@ NvResult nvhost_ctrl::IocCtrlEventWait(const std::vector<u8>& input, std::vector
params.value |= event_id;
event.event.writable->Clear();
gpu.RegisterSyncptInterrupt(params.syncpt_id, target_value);
- if (!is_async && ctrl.fresh_call) {
- ctrl.must_delay = true;
- ctrl.timeout = params.timeout;
- ctrl.event_id = event_id;
- return NvResult::Timeout;
- }
std::memcpy(output.data(), &params, sizeof(params));
return NvResult::Timeout;
}
diff --git a/src/core/hle/service/nvdrv/devices/nvhost_ctrl.h b/src/core/hle/service/nvdrv/devices/nvhost_ctrl.h
index 107168e21..c5aa1362a 100644
--- a/src/core/hle/service/nvdrv/devices/nvhost_ctrl.h
+++ b/src/core/hle/service/nvdrv/devices/nvhost_ctrl.h
@@ -18,13 +18,11 @@ public:
SyncpointManager& syncpoint_manager);
~nvhost_ctrl() override;
- NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- IoctlCtrl& ctrl) override;
+ NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override;
NvResult Ioctl2(Ioctl command, const std::vector<u8>& input,
- const std::vector<u8>& inline_input, std::vector<u8>& output,
- IoctlCtrl& ctrl) override;
+ const std::vector<u8>& inline_input, std::vector<u8>& output) override;
NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- std::vector<u8>& inline_output, IoctlCtrl& ctrl) override;
+ std::vector<u8>& inline_output) override;
private:
struct IocSyncptReadParams {
@@ -123,8 +121,7 @@ private:
static_assert(sizeof(IocCtrlEventKill) == 8, "IocCtrlEventKill is incorrect size");
NvResult NvOsGetConfigU32(const std::vector<u8>& input, std::vector<u8>& output);
- NvResult IocCtrlEventWait(const std::vector<u8>& input, std::vector<u8>& output, bool is_async,
- IoctlCtrl& ctrl);
+ NvResult IocCtrlEventWait(const std::vector<u8>& input, std::vector<u8>& output, bool is_async);
NvResult IocCtrlEventRegister(const std::vector<u8>& input, std::vector<u8>& output);
NvResult IocCtrlEventUnregister(const std::vector<u8>& input, std::vector<u8>& output);
NvResult IocCtrlClearEventWait(const std::vector<u8>& input, std::vector<u8>& output);
diff --git a/src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.cpp b/src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.cpp
index 647f5907e..0320d3ae2 100644
--- a/src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.cpp
+++ b/src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.cpp
@@ -16,7 +16,7 @@ nvhost_ctrl_gpu::nvhost_ctrl_gpu(Core::System& system) : nvdevice(system) {}
nvhost_ctrl_gpu::~nvhost_ctrl_gpu() = default;
NvResult nvhost_ctrl_gpu::Ioctl1(Ioctl command, const std::vector<u8>& input,
- std::vector<u8>& output, IoctlCtrl& ctrl) {
+ std::vector<u8>& output) {
switch (command.group) {
case 'G':
switch (command.cmd) {
@@ -48,15 +48,13 @@ NvResult nvhost_ctrl_gpu::Ioctl1(Ioctl command, const std::vector<u8>& input,
}
NvResult nvhost_ctrl_gpu::Ioctl2(Ioctl command, const std::vector<u8>& input,
- const std::vector<u8>& inline_input, std::vector<u8>& output,
- IoctlCtrl& ctrl) {
+ const std::vector<u8>& inline_input, std::vector<u8>& output) {
UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
return NvResult::NotImplemented;
}
NvResult nvhost_ctrl_gpu::Ioctl3(Ioctl command, const std::vector<u8>& input,
- std::vector<u8>& output, std::vector<u8>& inline_output,
- IoctlCtrl& ctrl) {
+ std::vector<u8>& output, std::vector<u8>& inline_output) {
switch (command.group) {
case 'G':
switch (command.cmd) {
diff --git a/src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.h b/src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.h
index c2fffe734..137b88238 100644
--- a/src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.h
+++ b/src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.h
@@ -16,13 +16,11 @@ public:
explicit nvhost_ctrl_gpu(Core::System& system);
~nvhost_ctrl_gpu() override;
- NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- IoctlCtrl& ctrl) override;
+ NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override;
NvResult Ioctl2(Ioctl command, const std::vector<u8>& input,
- const std::vector<u8>& inline_input, std::vector<u8>& output,
- IoctlCtrl& ctrl) override;
+ const std::vector<u8>& inline_input, std::vector<u8>& output) override;
NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- std::vector<u8>& inline_output, IoctlCtrl& ctrl) override;
+ std::vector<u8>& inline_output) override;
private:
struct IoctlGpuCharacteristics {
diff --git a/src/core/hle/service/nvdrv/devices/nvhost_gpu.cpp b/src/core/hle/service/nvdrv/devices/nvhost_gpu.cpp
index b0c2caba5..af8b3d9f1 100644
--- a/src/core/hle/service/nvdrv/devices/nvhost_gpu.cpp
+++ b/src/core/hle/service/nvdrv/devices/nvhost_gpu.cpp
@@ -23,8 +23,7 @@ nvhost_gpu::nvhost_gpu(Core::System& system, std::shared_ptr<nvmap> nvmap_dev,
nvhost_gpu::~nvhost_gpu() = default;
-NvResult nvhost_gpu::Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- IoctlCtrl& ctrl) {
+NvResult nvhost_gpu::Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) {
switch (command.group) {
case 0x0:
switch (command.cmd) {
@@ -76,8 +75,7 @@ NvResult nvhost_gpu::Ioctl1(Ioctl command, const std::vector<u8>& input, std::ve
};
NvResult nvhost_gpu::Ioctl2(Ioctl command, const std::vector<u8>& input,
- const std::vector<u8>& inline_input, std::vector<u8>& output,
- IoctlCtrl& ctrl) {
+ const std::vector<u8>& inline_input, std::vector<u8>& output) {
switch (command.group) {
case 'H':
switch (command.cmd) {
@@ -91,7 +89,7 @@ NvResult nvhost_gpu::Ioctl2(Ioctl command, const std::vector<u8>& input,
}
NvResult nvhost_gpu::Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- std::vector<u8>& inline_output, IoctlCtrl& ctrl) {
+ std::vector<u8>& inline_output) {
UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
return NvResult::NotImplemented;
}
diff --git a/src/core/hle/service/nvdrv/devices/nvhost_gpu.h b/src/core/hle/service/nvdrv/devices/nvhost_gpu.h
index aa0048a9d..e0298b4fe 100644
--- a/src/core/hle/service/nvdrv/devices/nvhost_gpu.h
+++ b/src/core/hle/service/nvdrv/devices/nvhost_gpu.h
@@ -26,13 +26,11 @@ public:
SyncpointManager& syncpoint_manager);
~nvhost_gpu() override;
- NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- IoctlCtrl& ctrl) override;
+ NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override;
NvResult Ioctl2(Ioctl command, const std::vector<u8>& input,
- const std::vector<u8>& inline_input, std::vector<u8>& output,
- IoctlCtrl& ctrl) override;
+ const std::vector<u8>& inline_input, std::vector<u8>& output) override;
NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- std::vector<u8>& inline_output, IoctlCtrl& ctrl) override;
+ std::vector<u8>& inline_output) override;
private:
enum class CtxObjects : u32_le {
diff --git a/src/core/hle/service/nvdrv/devices/nvhost_nvdec.cpp b/src/core/hle/service/nvdrv/devices/nvhost_nvdec.cpp
index b8328c314..d8735491c 100644
--- a/src/core/hle/service/nvdrv/devices/nvhost_nvdec.cpp
+++ b/src/core/hle/service/nvdrv/devices/nvhost_nvdec.cpp
@@ -15,8 +15,8 @@ nvhost_nvdec::nvhost_nvdec(Core::System& system, std::shared_ptr<nvmap> nvmap_de
: nvhost_nvdec_common(system, std::move(nvmap_dev)) {}
nvhost_nvdec::~nvhost_nvdec() = default;
-NvResult nvhost_nvdec::Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- IoctlCtrl& ctrl) {
+NvResult nvhost_nvdec::Ioctl1(Ioctl command, const std::vector<u8>& input,
+ std::vector<u8>& output) {
switch (command.group) {
case 0x0:
switch (command.cmd) {
@@ -58,14 +58,13 @@ NvResult nvhost_nvdec::Ioctl1(Ioctl command, const std::vector<u8>& input, std::
}
NvResult nvhost_nvdec::Ioctl2(Ioctl command, const std::vector<u8>& input,
- const std::vector<u8>& inline_input, std::vector<u8>& output,
- IoctlCtrl& ctrl) {
+ const std::vector<u8>& inline_input, std::vector<u8>& output) {
UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
return NvResult::NotImplemented;
}
NvResult nvhost_nvdec::Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- std::vector<u8>& inline_output, IoctlCtrl& ctrl) {
+ std::vector<u8>& inline_output) {
UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
return NvResult::NotImplemented;
}
diff --git a/src/core/hle/service/nvdrv/devices/nvhost_nvdec.h b/src/core/hle/service/nvdrv/devices/nvhost_nvdec.h
index 884ed6c5b..79b8b6de1 100644
--- a/src/core/hle/service/nvdrv/devices/nvhost_nvdec.h
+++ b/src/core/hle/service/nvdrv/devices/nvhost_nvdec.h
@@ -14,13 +14,11 @@ public:
explicit nvhost_nvdec(Core::System& system, std::shared_ptr<nvmap> nvmap_dev);
~nvhost_nvdec() override;
- NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- IoctlCtrl& ctrl) override;
+ NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override;
NvResult Ioctl2(Ioctl command, const std::vector<u8>& input,
- const std::vector<u8>& inline_input, std::vector<u8>& output,
- IoctlCtrl& ctrl) override;
+ const std::vector<u8>& inline_input, std::vector<u8>& output) override;
NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- std::vector<u8>& inline_output, IoctlCtrl& ctrl) override;
+ std::vector<u8>& inline_output) override;
};
} // namespace Service::Nvidia::Devices
diff --git a/src/core/hle/service/nvdrv/devices/nvhost_nvdec_common.h b/src/core/hle/service/nvdrv/devices/nvhost_nvdec_common.h
index ab152bf0e..d9f95ba58 100644
--- a/src/core/hle/service/nvdrv/devices/nvhost_nvdec_common.h
+++ b/src/core/hle/service/nvdrv/devices/nvhost_nvdec_common.h
@@ -18,39 +18,6 @@ public:
explicit nvhost_nvdec_common(Core::System& system, std::shared_ptr<nvmap> nvmap_dev);
~nvhost_nvdec_common() override;
- /**
- * Handles an ioctl1 request.
- * @param command The ioctl command id.
- * @param input A buffer containing the input data for the ioctl.
- * @param output A buffer where the output data will be written to.
- * @returns The result code of the ioctl.
- */
- virtual NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- IoctlCtrl& ctrl) = 0;
-
- /**
- * Handles an ioctl2 request.
- * @param command The ioctl command id.
- * @param input A buffer containing the input data for the ioctl.
- * @param inline_input A buffer containing the input data for the ioctl which has been inlined.
- * @param output A buffer where the output data will be written to.
- * @returns The result code of the ioctl.
- */
- virtual NvResult Ioctl2(Ioctl command, const std::vector<u8>& input,
- const std::vector<u8>& inline_input, std::vector<u8>& output,
- IoctlCtrl& ctrl) = 0;
-
- /**
- * Handles an ioctl3 request.
- * @param command The ioctl command id.
- * @param input A buffer containing the input data for the ioctl.
- * @param output A buffer where the output data will be written to.
- * @param inline_output A buffer where the inlined output data will be written to.
- * @returns The result code of the ioctl.
- */
- virtual NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- std::vector<u8>& inline_output, IoctlCtrl& ctrl) = 0;
-
protected:
class BufferMap final {
public:
diff --git a/src/core/hle/service/nvdrv/devices/nvhost_nvjpg.cpp b/src/core/hle/service/nvdrv/devices/nvhost_nvjpg.cpp
index 6f4ab0ab3..2d06955c0 100644
--- a/src/core/hle/service/nvdrv/devices/nvhost_nvjpg.cpp
+++ b/src/core/hle/service/nvdrv/devices/nvhost_nvjpg.cpp
@@ -13,8 +13,8 @@ namespace Service::Nvidia::Devices {
nvhost_nvjpg::nvhost_nvjpg(Core::System& system) : nvdevice(system) {}
nvhost_nvjpg::~nvhost_nvjpg() = default;
-NvResult nvhost_nvjpg::Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- IoctlCtrl& ctrl) {
+NvResult nvhost_nvjpg::Ioctl1(Ioctl command, const std::vector<u8>& input,
+ std::vector<u8>& output) {
switch (command.group) {
case 'H':
switch (command.cmd) {
@@ -33,14 +33,13 @@ NvResult nvhost_nvjpg::Ioctl1(Ioctl command, const std::vector<u8>& input, std::
}
NvResult nvhost_nvjpg::Ioctl2(Ioctl command, const std::vector<u8>& input,
- const std::vector<u8>& inline_input, std::vector<u8>& output,
- IoctlCtrl& ctrl) {
+ const std::vector<u8>& inline_input, std::vector<u8>& output) {
UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
return NvResult::NotImplemented;
}
NvResult nvhost_nvjpg::Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- std::vector<u8>& inline_output, IoctlCtrl& ctrl) {
+ std::vector<u8>& inline_output) {
UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
return NvResult::NotImplemented;
}
diff --git a/src/core/hle/service/nvdrv/devices/nvhost_nvjpg.h b/src/core/hle/service/nvdrv/devices/nvhost_nvjpg.h
index 6fb99d959..43948d18d 100644
--- a/src/core/hle/service/nvdrv/devices/nvhost_nvjpg.h
+++ b/src/core/hle/service/nvdrv/devices/nvhost_nvjpg.h
@@ -16,13 +16,11 @@ public:
explicit nvhost_nvjpg(Core::System& system);
~nvhost_nvjpg() override;
- NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- IoctlCtrl& ctrl) override;
+ NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override;
NvResult Ioctl2(Ioctl command, const std::vector<u8>& input,
- const std::vector<u8>& inline_input, std::vector<u8>& output,
- IoctlCtrl& ctrl) override;
+ const std::vector<u8>& inline_input, std::vector<u8>& output) override;
NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- std::vector<u8>& inline_output, IoctlCtrl& ctrl) override;
+ std::vector<u8>& inline_output) override;
private:
struct IoctlSetNvmapFD {
diff --git a/src/core/hle/service/nvdrv/devices/nvhost_vic.cpp b/src/core/hle/service/nvdrv/devices/nvhost_vic.cpp
index 55a17f423..805fe86ae 100644
--- a/src/core/hle/service/nvdrv/devices/nvhost_vic.cpp
+++ b/src/core/hle/service/nvdrv/devices/nvhost_vic.cpp
@@ -15,8 +15,7 @@ nvhost_vic::nvhost_vic(Core::System& system, std::shared_ptr<nvmap> nvmap_dev)
nvhost_vic::~nvhost_vic() = default;
-NvResult nvhost_vic::Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- IoctlCtrl& ctrl) {
+NvResult nvhost_vic::Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) {
switch (command.group) {
case 0x0:
switch (command.cmd) {
@@ -51,14 +50,13 @@ NvResult nvhost_vic::Ioctl1(Ioctl command, const std::vector<u8>& input, std::ve
}
NvResult nvhost_vic::Ioctl2(Ioctl command, const std::vector<u8>& input,
- const std::vector<u8>& inline_input, std::vector<u8>& output,
- IoctlCtrl& ctrl) {
+ const std::vector<u8>& inline_input, std::vector<u8>& output) {
UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
return NvResult::NotImplemented;
}
NvResult nvhost_vic::Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- std::vector<u8>& inline_output, IoctlCtrl& ctrl) {
+ std::vector<u8>& inline_output) {
UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
return NvResult::NotImplemented;
}
diff --git a/src/core/hle/service/nvdrv/devices/nvhost_vic.h b/src/core/hle/service/nvdrv/devices/nvhost_vic.h
index 7f4858cd4..b2e11f4d4 100644
--- a/src/core/hle/service/nvdrv/devices/nvhost_vic.h
+++ b/src/core/hle/service/nvdrv/devices/nvhost_vic.h
@@ -14,12 +14,10 @@ public:
explicit nvhost_vic(Core::System& system, std::shared_ptr<nvmap> nvmap_dev);
~nvhost_vic();
- NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- IoctlCtrl& ctrl) override;
+ NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override;
NvResult Ioctl2(Ioctl command, const std::vector<u8>& input,
- const std::vector<u8>& inline_input, std::vector<u8>& output,
- IoctlCtrl& ctrl) override;
+ const std::vector<u8>& inline_input, std::vector<u8>& output) override;
NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- std::vector<u8>& inline_output, IoctlCtrl& ctrl) override;
+ std::vector<u8>& inline_output) override;
};
} // namespace Service::Nvidia::Devices
diff --git a/src/core/hle/service/nvdrv/devices/nvmap.cpp b/src/core/hle/service/nvdrv/devices/nvmap.cpp
index 910cfee51..4015a2740 100644
--- a/src/core/hle/service/nvdrv/devices/nvmap.cpp
+++ b/src/core/hle/service/nvdrv/devices/nvmap.cpp
@@ -19,8 +19,7 @@ nvmap::nvmap(Core::System& system) : nvdevice(system) {
nvmap::~nvmap() = default;
-NvResult nvmap::Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- IoctlCtrl& ctrl) {
+NvResult nvmap::Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) {
switch (command.group) {
case 0x1:
switch (command.cmd) {
@@ -49,14 +48,13 @@ NvResult nvmap::Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<
}
NvResult nvmap::Ioctl2(Ioctl command, const std::vector<u8>& input,
- const std::vector<u8>& inline_input, std::vector<u8>& output,
- IoctlCtrl& ctrl) {
+ const std::vector<u8>& inline_input, std::vector<u8>& output) {
UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
return NvResult::NotImplemented;
}
NvResult nvmap::Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- std::vector<u8>& inline_output, IoctlCtrl& ctrl) {
+ std::vector<u8>& inline_output) {
UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
return NvResult::NotImplemented;
}
diff --git a/src/core/hle/service/nvdrv/devices/nvmap.h b/src/core/hle/service/nvdrv/devices/nvmap.h
index c0c2fa5eb..4484bd79f 100644
--- a/src/core/hle/service/nvdrv/devices/nvmap.h
+++ b/src/core/hle/service/nvdrv/devices/nvmap.h
@@ -19,13 +19,11 @@ public:
explicit nvmap(Core::System& system);
~nvmap() override;
- NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- IoctlCtrl& ctrl) override;
+ NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override;
NvResult Ioctl2(Ioctl command, const std::vector<u8>& input,
- const std::vector<u8>& inline_input, std::vector<u8>& output,
- IoctlCtrl& ctrl) override;
+ const std::vector<u8>& inline_input, std::vector<u8>& output) override;
NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
- std::vector<u8>& inline_output, IoctlCtrl& ctrl) override;
+ std::vector<u8>& inline_output) override;
/// Returns the allocated address of an nvmap object given its handle.
VAddr GetObjectAddress(u32 handle) const;
diff --git a/src/core/hle/service/nvdrv/interface.cpp b/src/core/hle/service/nvdrv/interface.cpp
index d72c531f6..cc23b001c 100644
--- a/src/core/hle/service/nvdrv/interface.cpp
+++ b/src/core/hle/service/nvdrv/interface.cpp
@@ -61,32 +61,9 @@ void NVDRV::Ioctl1(Kernel::HLERequestContext& ctx) {
std::vector<u8> output_buffer(ctx.GetWriteBufferSize(0));
const auto input_buffer = ctx.ReadBuffer(0);
- IoctlCtrl ctrl{};
-
- const auto nv_result = nvdrv->Ioctl1(fd, command, input_buffer, output_buffer, ctrl);
- if (ctrl.must_delay) {
- ctrl.fresh_call = false;
- ctx.SleepClientThread(
- "NVServices::DelayedResponse", ctrl.timeout,
- [=, this](std::shared_ptr<Kernel::Thread> thread, Kernel::HLERequestContext& ctx_,
- Kernel::ThreadWakeupReason reason) {
- IoctlCtrl ctrl2{ctrl};
- std::vector<u8> tmp_output = output_buffer;
- const auto nv_result2 = nvdrv->Ioctl1(fd, command, input_buffer, tmp_output, ctrl2);
-
- if (command.is_out != 0) {
- ctx.WriteBuffer(tmp_output);
- }
-
- IPC::ResponseBuilder rb{ctx_, 3};
- rb.Push(RESULT_SUCCESS);
- rb.PushEnum(nv_result2);
- },
- nvdrv->GetEventWriteable(ctrl.event_id));
- } else {
- if (command.is_out != 0) {
- ctx.WriteBuffer(output_buffer);
- }
+ const auto nv_result = nvdrv->Ioctl1(fd, command, input_buffer, output_buffer);
+ if (command.is_out != 0) {
+ ctx.WriteBuffer(output_buffer);
}
IPC::ResponseBuilder rb{ctx, 3};
@@ -110,36 +87,8 @@ void NVDRV::Ioctl2(Kernel::HLERequestContext& ctx) {
const auto input_inlined_buffer = ctx.ReadBuffer(1);
std::vector<u8> output_buffer(ctx.GetWriteBufferSize(0));
- IoctlCtrl ctrl{};
-
const auto nv_result =
- nvdrv->Ioctl2(fd, command, input_buffer, input_inlined_buffer, output_buffer, ctrl);
- if (ctrl.must_delay) {
- ctrl.fresh_call = false;
- ctx.SleepClientThread(
- "NVServices::DelayedResponse", ctrl.timeout,
- [=, this](std::shared_ptr<Kernel::Thread> thread, Kernel::HLERequestContext& ctx_,
- Kernel::ThreadWakeupReason reason) {
- IoctlCtrl ctrl2{ctrl};
- std::vector<u8> tmp_output = output_buffer;
- const auto nv_result2 = nvdrv->Ioctl2(fd, command, input_buffer,
- input_inlined_buffer, tmp_output, ctrl2);
-
- if (command.is_out != 0) {
- ctx.WriteBuffer(tmp_output);
- }
-
- IPC::ResponseBuilder rb{ctx_, 3};
- rb.Push(RESULT_SUCCESS);
- rb.PushEnum(nv_result2);
- },
- nvdrv->GetEventWriteable(ctrl.event_id));
- } else {
- if (command.is_out != 0) {
- ctx.WriteBuffer(output_buffer);
- }
- }
-
+ nvdrv->Ioctl2(fd, command, input_buffer, input_inlined_buffer, output_buffer);
if (command.is_out != 0) {
ctx.WriteBuffer(output_buffer);
}
@@ -165,36 +114,11 @@ void NVDRV::Ioctl3(Kernel::HLERequestContext& ctx) {
std::vector<u8> output_buffer(ctx.GetWriteBufferSize(0));
std::vector<u8> output_buffer_inline(ctx.GetWriteBufferSize(1));
- IoctlCtrl ctrl{};
const auto nv_result =
- nvdrv->Ioctl3(fd, command, input_buffer, output_buffer, output_buffer_inline, ctrl);
- if (ctrl.must_delay) {
- ctrl.fresh_call = false;
- ctx.SleepClientThread(
- "NVServices::DelayedResponse", ctrl.timeout,
- [=, this](std::shared_ptr<Kernel::Thread> thread, Kernel::HLERequestContext& ctx_,
- Kernel::ThreadWakeupReason reason) {
- IoctlCtrl ctrl2{ctrl};
- std::vector<u8> tmp_output = output_buffer;
- std::vector<u8> tmp_output2 = output_buffer;
- const auto nv_result2 =
- nvdrv->Ioctl3(fd, command, input_buffer, tmp_output, tmp_output2, ctrl2);
-
- if (command.is_out != 0) {
- ctx.WriteBuffer(tmp_output, 0);
- ctx.WriteBuffer(tmp_output2, 1);
- }
-
- IPC::ResponseBuilder rb{ctx_, 3};
- rb.Push(RESULT_SUCCESS);
- rb.PushEnum(nv_result2);
- },
- nvdrv->GetEventWriteable(ctrl.event_id));
- } else {
- if (command.is_out != 0) {
- ctx.WriteBuffer(output_buffer, 0);
- ctx.WriteBuffer(output_buffer_inline, 1);
- }
+ nvdrv->Ioctl3(fd, command, input_buffer, output_buffer, output_buffer_inline);
+ if (command.is_out != 0) {
+ ctx.WriteBuffer(output_buffer, 0);
+ ctx.WriteBuffer(output_buffer_inline, 1);
}
IPC::ResponseBuilder rb{ctx, 3};
diff --git a/src/core/hle/service/nvdrv/nvdata.h b/src/core/hle/service/nvdrv/nvdata.h
index a3c4ecd85..3294bc0e7 100644
--- a/src/core/hle/service/nvdrv/nvdata.h
+++ b/src/core/hle/service/nvdrv/nvdata.h
@@ -97,15 +97,4 @@ union Ioctl {
BitField<31, 1, u32> is_out;
};
-struct IoctlCtrl {
- // First call done to the servioce for services that call itself again after a call.
- bool fresh_call{true};
- // Tells the Ioctl Wrapper that it must delay the IPC response and send the thread to sleep
- bool must_delay{};
- // Timeout for the delay
- s64 timeout{};
- // NV Event Id
- s32 event_id{-1};
-};
-
} // namespace Service::Nvidia
diff --git a/src/core/hle/service/nvdrv/nvdrv.cpp b/src/core/hle/service/nvdrv/nvdrv.cpp
index 8e0c9f093..e03195afe 100644
--- a/src/core/hle/service/nvdrv/nvdrv.cpp
+++ b/src/core/hle/service/nvdrv/nvdrv.cpp
@@ -91,7 +91,7 @@ DeviceFD Module::Open(const std::string& device_name) {
}
NvResult Module::Ioctl1(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
- std::vector<u8>& output, IoctlCtrl& ctrl) {
+ std::vector<u8>& output) {
if (fd < 0) {
LOG_ERROR(Service_NVDRV, "Invalid DeviceFD={}!", fd);
return NvResult::InvalidState;
@@ -104,12 +104,11 @@ NvResult Module::Ioctl1(DeviceFD fd, Ioctl command, const std::vector<u8>& input
return NvResult::NotImplemented;
}
- return itr->second->Ioctl1(command, input, output, ctrl);
+ return itr->second->Ioctl1(command, input, output);
}
NvResult Module::Ioctl2(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
- const std::vector<u8>& inline_input, std::vector<u8>& output,
- IoctlCtrl& ctrl) {
+ const std::vector<u8>& inline_input, std::vector<u8>& output) {
if (fd < 0) {
LOG_ERROR(Service_NVDRV, "Invalid DeviceFD={}!", fd);
return NvResult::InvalidState;
@@ -122,11 +121,11 @@ NvResult Module::Ioctl2(DeviceFD fd, Ioctl command, const std::vector<u8>& input
return NvResult::NotImplemented;
}
- return itr->second->Ioctl2(command, input, inline_input, output, ctrl);
+ return itr->second->Ioctl2(command, input, inline_input, output);
}
NvResult Module::Ioctl3(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
- std::vector<u8>& output, std::vector<u8>& inline_output, IoctlCtrl& ctrl) {
+ std::vector<u8>& output, std::vector<u8>& inline_output) {
if (fd < 0) {
LOG_ERROR(Service_NVDRV, "Invalid DeviceFD={}!", fd);
return NvResult::InvalidState;
@@ -139,7 +138,7 @@ NvResult Module::Ioctl3(DeviceFD fd, Ioctl command, const std::vector<u8>& input
return NvResult::NotImplemented;
}
- return itr->second->Ioctl3(command, input, output, inline_output, ctrl);
+ return itr->second->Ioctl3(command, input, output, inline_output);
}
NvResult Module::Close(DeviceFD fd) {
diff --git a/src/core/hle/service/nvdrv/nvdrv.h b/src/core/hle/service/nvdrv/nvdrv.h
index 5985d2179..144e657e5 100644
--- a/src/core/hle/service/nvdrv/nvdrv.h
+++ b/src/core/hle/service/nvdrv/nvdrv.h
@@ -119,13 +119,13 @@ public:
/// Sends an ioctl command to the specified file descriptor.
NvResult Ioctl1(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
- std::vector<u8>& output, IoctlCtrl& ctrl);
+ std::vector<u8>& output);
NvResult Ioctl2(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
- const std::vector<u8>& inline_input, std::vector<u8>& output, IoctlCtrl& ctrl);
+ const std::vector<u8>& inline_input, std::vector<u8>& output);
NvResult Ioctl3(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
- std::vector<u8>& output, std::vector<u8>& inline_output, IoctlCtrl& ctrl);
+ std::vector<u8>& output, std::vector<u8>& inline_output);
/// Closes a device file descriptor and returns operation success.
NvResult Close(DeviceFD fd);
diff --git a/src/core/hle/service/nvflinger/buffer_queue.cpp b/src/core/hle/service/nvflinger/buffer_queue.cpp
index b89a2d41b..c8c6a4d64 100644
--- a/src/core/hle/service/nvflinger/buffer_queue.cpp
+++ b/src/core/hle/service/nvflinger/buffer_queue.cpp
@@ -22,10 +22,16 @@ BufferQueue::BufferQueue(Kernel::KernelCore& kernel, u32 id, u64 layer_id)
BufferQueue::~BufferQueue() = default;
void BufferQueue::SetPreallocatedBuffer(u32 slot, const IGBPBuffer& igbp_buffer) {
+ ASSERT(slot < buffer_slots);
LOG_WARNING(Service, "Adding graphics buffer {}", slot);
- free_buffers.push_back(slot);
- queue.push_back({
+ {
+ std::unique_lock lock{queue_mutex};
+ free_buffers.push_back(slot);
+ }
+ condition.notify_one();
+
+ buffers[slot] = {
.slot = slot,
.status = Buffer::Status::Free,
.igbp_buffer = igbp_buffer,
@@ -33,127 +39,139 @@ void BufferQueue::SetPreallocatedBuffer(u32 slot, const IGBPBuffer& igbp_buffer)
.crop_rect = {},
.swap_interval = 0,
.multi_fence = {},
- });
+ };
buffer_wait_event.writable->Signal();
}
std::optional<std::pair<u32, Service::Nvidia::MultiFence*>> BufferQueue::DequeueBuffer(u32 width,
u32 height) {
+ // Wait for first request before trying to dequeue
+ {
+ std::unique_lock lock{queue_mutex};
+ condition.wait(lock, [this] { return !free_buffers.empty() || !is_connect; });
+ }
- if (free_buffers.empty()) {
+ if (!is_connect) {
+ // Buffer was disconnected while the thread was blocked, this is most likely due to
+ // emulation being stopped
return std::nullopt;
}
+ std::unique_lock lock{queue_mutex};
+
auto f_itr = free_buffers.begin();
- auto itr = queue.end();
+ auto slot = buffers.size();
while (f_itr != free_buffers.end()) {
- auto slot = *f_itr;
- itr = std::find_if(queue.begin(), queue.end(), [&](const Buffer& buffer) {
- // Only consider free buffers. Buffers become free once again after they've been
- // Acquired and Released by the compositor, see the NVFlinger::Compose method.
- if (buffer.status != Buffer::Status::Free) {
- return false;
- }
-
- if (buffer.slot != slot) {
- return false;
- }
-
- // Make sure that the parameters match.
- return buffer.igbp_buffer.width == width && buffer.igbp_buffer.height == height;
- });
-
- if (itr != queue.end()) {
+ const Buffer& buffer = buffers[*f_itr];
+ if (buffer.status == Buffer::Status::Free && buffer.igbp_buffer.width == width &&
+ buffer.igbp_buffer.height == height) {
+ slot = *f_itr;
free_buffers.erase(f_itr);
break;
}
++f_itr;
}
-
- if (itr == queue.end()) {
+ if (slot == buffers.size()) {
return std::nullopt;
}
-
- itr->status = Buffer::Status::Dequeued;
- return {{itr->slot, &itr->multi_fence}};
+ buffers[slot].status = Buffer::Status::Dequeued;
+ return {{buffers[slot].slot, &buffers[slot].multi_fence}};
}
const IGBPBuffer& BufferQueue::RequestBuffer(u32 slot) const {
- auto itr = std::find_if(queue.begin(), queue.end(),
- [&](const Buffer& buffer) { return buffer.slot == slot; });
- ASSERT(itr != queue.end());
- ASSERT(itr->status == Buffer::Status::Dequeued);
- return itr->igbp_buffer;
+ ASSERT(slot < buffers.size());
+ ASSERT(buffers[slot].status == Buffer::Status::Dequeued);
+ ASSERT(buffers[slot].slot == slot);
+
+ return buffers[slot].igbp_buffer;
}
void BufferQueue::QueueBuffer(u32 slot, BufferTransformFlags transform,
const Common::Rectangle<int>& crop_rect, u32 swap_interval,
Service::Nvidia::MultiFence& multi_fence) {
- auto itr = std::find_if(queue.begin(), queue.end(),
- [&](const Buffer& buffer) { return buffer.slot == slot; });
- ASSERT(itr != queue.end());
- ASSERT(itr->status == Buffer::Status::Dequeued);
- itr->status = Buffer::Status::Queued;
- itr->transform = transform;
- itr->crop_rect = crop_rect;
- itr->swap_interval = swap_interval;
- itr->multi_fence = multi_fence;
+ ASSERT(slot < buffers.size());
+ ASSERT(buffers[slot].status == Buffer::Status::Dequeued);
+ ASSERT(buffers[slot].slot == slot);
+
+ buffers[slot].status = Buffer::Status::Queued;
+ buffers[slot].transform = transform;
+ buffers[slot].crop_rect = crop_rect;
+ buffers[slot].swap_interval = swap_interval;
+ buffers[slot].multi_fence = multi_fence;
queue_sequence.push_back(slot);
}
void BufferQueue::CancelBuffer(u32 slot, const Service::Nvidia::MultiFence& multi_fence) {
- const auto itr = std::find_if(queue.begin(), queue.end(),
- [slot](const Buffer& buffer) { return buffer.slot == slot; });
- ASSERT(itr != queue.end());
- ASSERT(itr->status != Buffer::Status::Free);
- itr->status = Buffer::Status::Free;
- itr->multi_fence = multi_fence;
- itr->swap_interval = 0;
+ ASSERT(slot < buffers.size());
+ ASSERT(buffers[slot].status != Buffer::Status::Free);
+ ASSERT(buffers[slot].slot == slot);
- free_buffers.push_back(slot);
+ buffers[slot].status = Buffer::Status::Free;
+ buffers[slot].multi_fence = multi_fence;
+ buffers[slot].swap_interval = 0;
+
+ {
+ std::unique_lock lock{queue_mutex};
+ free_buffers.push_back(slot);
+ }
+ condition.notify_one();
buffer_wait_event.writable->Signal();
}
std::optional<std::reference_wrapper<const BufferQueue::Buffer>> BufferQueue::AcquireBuffer() {
- auto itr = queue.end();
+ std::size_t buffer_slot = buffers.size();
// Iterate to find a queued buffer matching the requested slot.
- while (itr == queue.end() && !queue_sequence.empty()) {
- const u32 slot = queue_sequence.front();
- itr = std::find_if(queue.begin(), queue.end(), [&slot](const Buffer& buffer) {
- return buffer.status == Buffer::Status::Queued && buffer.slot == slot;
- });
+ while (buffer_slot == buffers.size() && !queue_sequence.empty()) {
+ const auto slot = static_cast<std::size_t>(queue_sequence.front());
+ ASSERT(slot < buffers.size());
+ if (buffers[slot].status == Buffer::Status::Queued) {
+ ASSERT(buffers[slot].slot == slot);
+ buffer_slot = slot;
+ }
queue_sequence.pop_front();
}
- if (itr == queue.end()) {
+ if (buffer_slot == buffers.size()) {
return std::nullopt;
}
- itr->status = Buffer::Status::Acquired;
- return *itr;
+ buffers[buffer_slot].status = Buffer::Status::Acquired;
+ return {{buffers[buffer_slot]}};
}
void BufferQueue::ReleaseBuffer(u32 slot) {
- auto itr = std::find_if(queue.begin(), queue.end(),
- [&](const Buffer& buffer) { return buffer.slot == slot; });
- ASSERT(itr != queue.end());
- ASSERT(itr->status == Buffer::Status::Acquired);
- itr->status = Buffer::Status::Free;
- free_buffers.push_back(slot);
+ ASSERT(slot < buffers.size());
+ ASSERT(buffers[slot].status == Buffer::Status::Acquired);
+ ASSERT(buffers[slot].slot == slot);
+
+ buffers[slot].status = Buffer::Status::Free;
+ {
+ std::unique_lock lock{queue_mutex};
+ free_buffers.push_back(slot);
+ }
+ condition.notify_one();
buffer_wait_event.writable->Signal();
}
-void BufferQueue::Disconnect() {
- queue.clear();
+void BufferQueue::Connect() {
queue_sequence.clear();
id = 1;
layer_id = 1;
+ is_connect = true;
+}
+
+void BufferQueue::Disconnect() {
+ buffers.fill({});
+ queue_sequence.clear();
+ buffer_wait_event.writable->Signal();
+ is_connect = false;
+ condition.notify_one();
}
u32 BufferQueue::Query(QueryType type) {
- LOG_WARNING(Service, "(STUBBED) called type={}", static_cast<u32>(type));
+ LOG_WARNING(Service, "(STUBBED) called type={}", type);
switch (type) {
case QueryType::NativeWindowFormat:
diff --git a/src/core/hle/service/nvflinger/buffer_queue.h b/src/core/hle/service/nvflinger/buffer_queue.h
index e7517c7e1..a2f60d9eb 100644
--- a/src/core/hle/service/nvflinger/buffer_queue.h
+++ b/src/core/hle/service/nvflinger/buffer_queue.h
@@ -4,7 +4,9 @@
#pragma once
+#include <condition_variable>
#include <list>
+#include <mutex>
#include <optional>
#include <vector>
@@ -21,6 +23,7 @@ class KernelCore;
namespace Service::NVFlinger {
+constexpr u32 buffer_slots = 0x40;
struct IGBPBuffer {
u32_le magic;
u32_le width;
@@ -98,6 +101,7 @@ public:
void CancelBuffer(u32 slot, const Service::Nvidia::MultiFence& multi_fence);
std::optional<std::reference_wrapper<const Buffer>> AcquireBuffer();
void ReleaseBuffer(u32 slot);
+ void Connect();
void Disconnect();
u32 Query(QueryType type);
@@ -105,18 +109,28 @@ public:
return id;
}
+ bool IsConnected() const {
+ return is_connect;
+ }
+
std::shared_ptr<Kernel::WritableEvent> GetWritableBufferWaitEvent() const;
std::shared_ptr<Kernel::ReadableEvent> GetBufferWaitEvent() const;
private:
- u32 id;
- u64 layer_id;
+ BufferQueue(const BufferQueue&) = delete;
+
+ u32 id{};
+ u64 layer_id{};
+ std::atomic_bool is_connect{};
std::list<u32> free_buffers;
- std::vector<Buffer> queue;
+ std::array<Buffer, buffer_slots> buffers;
std::list<u32> queue_sequence;
Kernel::EventPair buffer_wait_event;
+
+ std::mutex queue_mutex;
+ std::condition_variable condition;
};
} // namespace Service::NVFlinger
diff --git a/src/core/hle/service/nvflinger/nvflinger.cpp b/src/core/hle/service/nvflinger/nvflinger.cpp
index 44aa2bdae..4b3581949 100644
--- a/src/core/hle/service/nvflinger/nvflinger.cpp
+++ b/src/core/hle/service/nvflinger/nvflinger.cpp
@@ -88,6 +88,10 @@ NVFlinger::NVFlinger(Core::System& system) : system(system) {
}
NVFlinger::~NVFlinger() {
+ for (auto& buffer_queue : buffer_queues) {
+ buffer_queue->Disconnect();
+ }
+
if (system.IsMulticore()) {
is_running = false;
wait_event->Set();
@@ -104,6 +108,8 @@ void NVFlinger::SetNVDrvInstance(std::shared_ptr<Nvidia::Module> instance) {
}
std::optional<u64> NVFlinger::OpenDisplay(std::string_view name) {
+ const auto guard = Lock();
+
LOG_DEBUG(Service, "Opening \"{}\" display", name);
// TODO(Subv): Currently we only support the Default display.
@@ -121,6 +127,7 @@ std::optional<u64> NVFlinger::OpenDisplay(std::string_view name) {
}
std::optional<u64> NVFlinger::CreateLayer(u64 display_id) {
+ const auto guard = Lock();
auto* const display = FindDisplay(display_id);
if (display == nullptr) {
@@ -129,18 +136,22 @@ std::optional<u64> NVFlinger::CreateLayer(u64 display_id) {
const u64 layer_id = next_layer_id++;
const u32 buffer_queue_id = next_buffer_queue_id++;
- buffer_queues.emplace_back(system.Kernel(), buffer_queue_id, layer_id);
- display->CreateLayer(layer_id, buffer_queues.back());
+ buffer_queues.emplace_back(
+ std::make_unique<BufferQueue>(system.Kernel(), buffer_queue_id, layer_id));
+ display->CreateLayer(layer_id, *buffer_queues.back());
return layer_id;
}
void NVFlinger::CloseLayer(u64 layer_id) {
+ const auto guard = Lock();
+
for (auto& display : displays) {
display.CloseLayer(layer_id);
}
}
std::optional<u32> NVFlinger::FindBufferQueueId(u64 display_id, u64 layer_id) const {
+ const auto guard = Lock();
const auto* const layer = FindLayer(display_id, layer_id);
if (layer == nullptr) {
@@ -151,6 +162,7 @@ std::optional<u32> NVFlinger::FindBufferQueueId(u64 display_id, u64 layer_id) co
}
std::shared_ptr<Kernel::ReadableEvent> NVFlinger::FindVsyncEvent(u64 display_id) const {
+ const auto guard = Lock();
auto* const display = FindDisplay(display_id);
if (display == nullptr) {
@@ -160,20 +172,16 @@ std::shared_ptr<Kernel::ReadableEvent> NVFlinger::FindVsyncEvent(u64 display_id)
return display->GetVSyncEvent();
}
-BufferQueue& NVFlinger::FindBufferQueue(u32 id) {
+BufferQueue* NVFlinger::FindBufferQueue(u32 id) {
+ const auto guard = Lock();
const auto itr = std::find_if(buffer_queues.begin(), buffer_queues.end(),
- [id](const auto& queue) { return queue.GetId() == id; });
+ [id](const auto& queue) { return queue->GetId() == id; });
- ASSERT(itr != buffer_queues.end());
- return *itr;
-}
-
-const BufferQueue& NVFlinger::FindBufferQueue(u32 id) const {
- const auto itr = std::find_if(buffer_queues.begin(), buffer_queues.end(),
- [id](const auto& queue) { return queue.GetId() == id; });
+ if (itr == buffer_queues.end()) {
+ return nullptr;
+ }
- ASSERT(itr != buffer_queues.end());
- return *itr;
+ return itr->get();
}
VI::Display* NVFlinger::FindDisplay(u64 display_id) {
diff --git a/src/core/hle/service/nvflinger/nvflinger.h b/src/core/hle/service/nvflinger/nvflinger.h
index 1ebe949c0..c6765259f 100644
--- a/src/core/hle/service/nvflinger/nvflinger.h
+++ b/src/core/hle/service/nvflinger/nvflinger.h
@@ -75,10 +75,7 @@ public:
[[nodiscard]] std::shared_ptr<Kernel::ReadableEvent> FindVsyncEvent(u64 display_id) const;
/// Obtains a buffer queue identified by the ID.
- [[nodiscard]] BufferQueue& FindBufferQueue(u32 id);
-
- /// Obtains a buffer queue identified by the ID.
- [[nodiscard]] const BufferQueue& FindBufferQueue(u32 id) const;
+ [[nodiscard]] BufferQueue* FindBufferQueue(u32 id);
/// Performs a composition request to the emulated nvidia GPU and triggers the vsync events when
/// finished.
@@ -86,11 +83,11 @@ public:
[[nodiscard]] s64 GetNextTicks() const;
+private:
[[nodiscard]] std::unique_lock<std::mutex> Lock() const {
return std::unique_lock{*guard};
}
-private:
/// Finds the display identified by the specified ID.
[[nodiscard]] VI::Display* FindDisplay(u64 display_id);
@@ -110,7 +107,7 @@ private:
std::shared_ptr<Nvidia::Module> nvdrv;
std::vector<VI::Display> displays;
- std::vector<BufferQueue> buffer_queues;
+ std::vector<std::unique_ptr<BufferQueue>> buffer_queues;
/// Id to use for the next layer that is created, this counter is shared among all displays.
u64 next_layer_id = 1;
diff --git a/src/core/hle/service/pcie/pcie.cpp b/src/core/hle/service/pcie/pcie.cpp
index 80c0fc7ac..f6686fc4d 100644
--- a/src/core/hle/service/pcie/pcie.cpp
+++ b/src/core/hle/service/pcie/pcie.cpp
@@ -48,7 +48,7 @@ public:
class PCIe final : public ServiceFramework<PCIe> {
public:
- explicit PCIe(Core::System& system_) : ServiceFramework{system, "pcie"} {
+ explicit PCIe(Core::System& system_) : ServiceFramework{system_, "pcie"} {
// clang-format off
static const FunctionInfo functions[] = {
{0, nullptr, "RegisterClassDriver"},
diff --git a/src/core/hle/service/prepo/prepo.cpp b/src/core/hle/service/prepo/prepo.cpp
index 392fda73e..b417624c9 100644
--- a/src/core/hle/service/prepo/prepo.cpp
+++ b/src/core/hle/service/prepo/prepo.cpp
@@ -65,7 +65,7 @@ private:
}
LOG_DEBUG(Service_PREPO, "called, type={:02X}, process_id={:016X}, data1_size={:016X}",
- static_cast<u8>(Type), process_id, data[0].size());
+ Type, process_id, data[0].size());
const auto& reporter{system.GetReporter()};
reporter.SavePlayReport(Type, system.CurrentProcess()->GetTitleID(), data, process_id);
@@ -92,7 +92,7 @@ private:
LOG_DEBUG(
Service_PREPO,
"called, type={:02X}, user_id={:016X}{:016X}, process_id={:016X}, data1_size={:016X}",
- static_cast<u8>(Type), user_id[1], user_id[0], process_id, data[0].size());
+ Type, user_id[1], user_id[0], process_id, data[0].size());
const auto& reporter{system.GetReporter()};
reporter.SavePlayReport(Type, system.CurrentProcess()->GetTitleID(), data, process_id,
diff --git a/src/core/hle/service/service.cpp b/src/core/hle/service/service.cpp
index 360e0bf37..ff2a5b1db 100644
--- a/src/core/hle/service/service.cpp
+++ b/src/core/hle/service/service.cpp
@@ -95,9 +95,14 @@ ServiceFrameworkBase::ServiceFrameworkBase(Core::System& system_, const char* se
: system{system_}, service_name{service_name_}, max_sessions{max_sessions_},
handler_invoker{handler_invoker_} {}
-ServiceFrameworkBase::~ServiceFrameworkBase() = default;
+ServiceFrameworkBase::~ServiceFrameworkBase() {
+ // Wait for other threads to release access before destroying
+ const auto guard = LockService();
+}
void ServiceFrameworkBase::InstallAsService(SM::ServiceManager& service_manager) {
+ const auto guard = LockService();
+
ASSERT(!port_installed);
auto port = service_manager.RegisterService(service_name, max_sessions).Unwrap();
@@ -106,6 +111,8 @@ void ServiceFrameworkBase::InstallAsService(SM::ServiceManager& service_manager)
}
void ServiceFrameworkBase::InstallAsNamedPort(Kernel::KernelCore& kernel) {
+ const auto guard = LockService();
+
ASSERT(!port_installed);
auto [server_port, client_port] =
@@ -115,17 +122,6 @@ void ServiceFrameworkBase::InstallAsNamedPort(Kernel::KernelCore& kernel) {
port_installed = true;
}
-std::shared_ptr<Kernel::ClientPort> ServiceFrameworkBase::CreatePort(Kernel::KernelCore& kernel) {
- ASSERT(!port_installed);
-
- auto [server_port, client_port] =
- Kernel::ServerPort::CreatePortPair(kernel, max_sessions, service_name);
- auto port = MakeResult(std::move(server_port)).Unwrap();
- port->SetHleHandler(shared_from_this());
- port_installed = true;
- return client_port;
-}
-
void ServiceFrameworkBase::RegisterHandlersBase(const FunctionInfoBase* functions, std::size_t n) {
handlers.reserve(handlers.size() + n);
for (std::size_t i = 0; i < n; ++i) {
@@ -164,6 +160,8 @@ void ServiceFrameworkBase::InvokeRequest(Kernel::HLERequestContext& ctx) {
}
ResultCode ServiceFrameworkBase::HandleSyncRequest(Kernel::HLERequestContext& context) {
+ const auto guard = LockService();
+
switch (context.GetCommandType()) {
case IPC::CommandType::Close: {
IPC::ResponseBuilder rb{context, 2};
@@ -181,10 +179,14 @@ ResultCode ServiceFrameworkBase::HandleSyncRequest(Kernel::HLERequestContext& co
break;
}
default:
- UNIMPLEMENTED_MSG("command_type={}", static_cast<int>(context.GetCommandType()));
+ UNIMPLEMENTED_MSG("command_type={}", context.GetCommandType());
}
- context.WriteToOutgoingCommandBuffer(context.GetThread());
+ // If emulation was shutdown, we are closing service threads, do not write the response back to
+ // memory that may be shutting down as well.
+ if (system.IsPoweredOn()) {
+ context.WriteToOutgoingCommandBuffer(context.GetThread());
+ }
return RESULT_SUCCESS;
}
diff --git a/src/core/hle/service/service.h b/src/core/hle/service/service.h
index 62a182310..916445517 100644
--- a/src/core/hle/service/service.h
+++ b/src/core/hle/service/service.h
@@ -5,9 +5,11 @@
#pragma once
#include <cstddef>
+#include <mutex>
#include <string>
#include <boost/container/flat_map.hpp>
#include "common/common_types.h"
+#include "common/spin_lock.h"
#include "core/hle/kernel/hle_ipc.h"
#include "core/hle/kernel/object.h"
@@ -68,11 +70,9 @@ public:
void InstallAsService(SM::ServiceManager& service_manager);
/// Creates a port pair and registers it on the kernel's global port registry.
void InstallAsNamedPort(Kernel::KernelCore& kernel);
- /// Creates and returns an unregistered port for the service.
- std::shared_ptr<Kernel::ClientPort> CreatePort(Kernel::KernelCore& kernel);
-
+ /// Invokes a service request routine.
void InvokeRequest(Kernel::HLERequestContext& ctx);
-
+ /// Handles a synchronization request for the service.
ResultCode HandleSyncRequest(Kernel::HLERequestContext& context) override;
protected:
@@ -80,6 +80,11 @@ protected:
template <typename Self>
using HandlerFnP = void (Self::*)(Kernel::HLERequestContext&);
+ /// Used to gain exclusive access to the service members, e.g. from CoreTiming thread.
+ [[nodiscard]] std::scoped_lock<Common::SpinLock> LockService() {
+ return std::scoped_lock{lock_service};
+ }
+
/// System context that the service operates under.
Core::System& system;
@@ -115,6 +120,9 @@ private:
/// Function used to safely up-cast pointers to the derived class before invoking a handler.
InvokerFn* handler_invoker;
boost::container::flat_map<u32, FunctionInfoBase> handlers;
+
+ /// Used to gain exclusive access to the service members, e.g. from CoreTiming thread.
+ Common::SpinLock lock_service;
};
/**
diff --git a/src/core/hle/service/set/set_sys.cpp b/src/core/hle/service/set/set_sys.cpp
index 19b8f113d..b58b2c8c5 100644
--- a/src/core/hle/service/set/set_sys.cpp
+++ b/src/core/hle/service/set/set_sys.cpp
@@ -34,9 +34,9 @@ void GetFirmwareVersionImpl(Kernel::HLERequestContext& ctx, GetFirmwareVersionTy
// consistence (currently reports as 5.1.0-0.0)
const auto archive = FileSys::SystemArchive::SystemVersion();
- const auto early_exit_failure = [&ctx](const std::string& desc, ResultCode code) {
+ const auto early_exit_failure = [&ctx](std::string_view desc, ResultCode code) {
LOG_ERROR(Service_SET, "General failure while attempting to resolve firmware version ({}).",
- desc.c_str());
+ desc);
IPC::ResponseBuilder rb{ctx, 2};
rb.Push(code);
};
diff --git a/src/core/hle/service/sockets/blocking_worker.h b/src/core/hle/service/sockets/blocking_worker.h
deleted file mode 100644
index 2d53e52b6..000000000
--- a/src/core/hle/service/sockets/blocking_worker.h
+++ /dev/null
@@ -1,161 +0,0 @@
-// Copyright 2020 yuzu emulator team
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <atomic>
-#include <memory>
-#include <string>
-#include <string_view>
-#include <thread>
-#include <variant>
-#include <vector>
-
-#include <fmt/format.h>
-
-#include "common/assert.h"
-#include "common/microprofile.h"
-#include "common/thread.h"
-#include "core/core.h"
-#include "core/hle/kernel/hle_ipc.h"
-#include "core/hle/kernel/kernel.h"
-#include "core/hle/kernel/thread.h"
-#include "core/hle/kernel/writable_event.h"
-
-namespace Service::Sockets {
-
-/**
- * Worker abstraction to execute blocking calls on host without blocking the guest thread
- *
- * @tparam Service Service where the work is executed
- * @tparam Types Types of work to execute
- */
-template <class Service, class... Types>
-class BlockingWorker {
- using This = BlockingWorker<Service, Types...>;
- using WorkVariant = std::variant<std::monostate, Types...>;
-
-public:
- /// Create a new worker
- static std::unique_ptr<This> Create(Core::System& system, Service* service,
- std::string_view name) {
- return std::unique_ptr<This>(new This(system, service, name));
- }
-
- ~BlockingWorker() {
- while (!is_available.load(std::memory_order_relaxed)) {
- // Busy wait until work is finished
- std::this_thread::yield();
- }
- // Monostate means to exit the thread
- work = std::monostate{};
- work_event.Set();
- thread.join();
- }
-
- /**
- * Try to capture the worker to send work after a success
- * @returns True when the worker has been successfully captured
- */
- bool TryCapture() {
- bool expected = true;
- return is_available.compare_exchange_weak(expected, false, std::memory_order_relaxed,
- std::memory_order_relaxed);
- }
-
- /**
- * Send work to this worker abstraction
- * @see TryCapture must be called before attempting to call this function
- */
- template <class Work>
- void SendWork(Work new_work) {
- ASSERT_MSG(!is_available, "Trying to send work on a worker that's not captured");
- work = std::move(new_work);
- work_event.Set();
- }
-
- /// Generate a callback for @see SleepClientThread
- template <class Work>
- auto Callback() {
- return [this](std::shared_ptr<Kernel::Thread>, Kernel::HLERequestContext& ctx,
- Kernel::ThreadWakeupReason reason) {
- ASSERT(reason == Kernel::ThreadWakeupReason::Signal);
- std::get<Work>(work).Response(ctx);
- is_available.store(true);
- };
- }
-
- /// Get kernel event that will be signalled by the worker when the host operation finishes
- std::shared_ptr<Kernel::WritableEvent> KernelEvent() const {
- return kernel_event;
- }
-
-private:
- explicit BlockingWorker(Core::System& system, Service* service, std::string_view name) {
- auto pair = Kernel::WritableEvent::CreateEventPair(system.Kernel(), std::string(name));
- kernel_event = std::move(pair.writable);
- thread = std::thread([this, &system, service, name] { Run(system, service, name); });
- }
-
- void Run(Core::System& system, Service* service, std::string_view name) {
- system.RegisterHostThread();
-
- const std::string thread_name = fmt::format("yuzu:{}", name);
- MicroProfileOnThreadCreate(thread_name.c_str());
- Common::SetCurrentThreadName(thread_name.c_str());
-
- bool keep_running = true;
- while (keep_running) {
- work_event.Wait();
-
- const auto visit_fn = [service, &keep_running]<typename T>(T&& w) {
- if constexpr (std::is_same_v<std::decay_t<T>, std::monostate>) {
- keep_running = false;
- } else {
- w.Execute(service);
- }
- };
- std::visit(visit_fn, work);
-
- kernel_event->Signal();
- }
- }
-
- std::thread thread;
- WorkVariant work;
- Common::Event work_event;
- std::shared_ptr<Kernel::WritableEvent> kernel_event;
- std::atomic_bool is_available{true};
-};
-
-template <class Service, class... Types>
-class BlockingWorkerPool {
- using Worker = BlockingWorker<Service, Types...>;
-
-public:
- explicit BlockingWorkerPool(Core::System& system_, Service* service_)
- : system{system_}, service{service_} {}
-
- /// Returns a captured worker thread, creating new ones if necessary
- Worker* CaptureWorker() {
- for (auto& worker : workers) {
- if (worker->TryCapture()) {
- return worker.get();
- }
- }
- auto new_worker = Worker::Create(system, service, fmt::format("BSD:{}", workers.size()));
- [[maybe_unused]] const bool success = new_worker->TryCapture();
- ASSERT(success);
-
- return workers.emplace_back(std::move(new_worker)).get();
- }
-
-private:
- Core::System& system;
- Service* const service;
-
- std::vector<std::unique_ptr<Worker>> workers;
-};
-
-} // namespace Service::Sockets
diff --git a/src/core/hle/service/sockets/bsd.cpp b/src/core/hle/service/sockets/bsd.cpp
index a9875b9a6..2b824059d 100644
--- a/src/core/hle/service/sockets/bsd.cpp
+++ b/src/core/hle/service/sockets/bsd.cpp
@@ -30,7 +30,7 @@ bool IsConnectionBased(Type type) {
case Type::DGRAM:
return false;
default:
- UNIMPLEMENTED_MSG("Unimplemented type={}", static_cast<int>(type));
+ UNIMPLEMENTED_MSG("Unimplemented type={}", type);
return false;
}
}
@@ -178,13 +178,12 @@ void BSD::Poll(Kernel::HLERequestContext& ctx) {
LOG_DEBUG(Service, "called. nfds={} timeout={}", nfds, timeout);
- ExecuteWork(ctx, "BSD:Poll", timeout != 0,
- PollWork{
- .nfds = nfds,
- .timeout = timeout,
- .read_buffer = ctx.ReadBuffer(),
- .write_buffer = std::vector<u8>(ctx.GetWriteBufferSize()),
- });
+ ExecuteWork(ctx, PollWork{
+ .nfds = nfds,
+ .timeout = timeout,
+ .read_buffer = ctx.ReadBuffer(),
+ .write_buffer = std::vector<u8>(ctx.GetWriteBufferSize()),
+ });
}
void BSD::Accept(Kernel::HLERequestContext& ctx) {
@@ -193,11 +192,10 @@ void BSD::Accept(Kernel::HLERequestContext& ctx) {
LOG_DEBUG(Service, "called. fd={}", fd);
- ExecuteWork(ctx, "BSD:Accept", IsBlockingSocket(fd),
- AcceptWork{
- .fd = fd,
- .write_buffer = std::vector<u8>(ctx.GetWriteBufferSize()),
- });
+ ExecuteWork(ctx, AcceptWork{
+ .fd = fd,
+ .write_buffer = std::vector<u8>(ctx.GetWriteBufferSize()),
+ });
}
void BSD::Bind(Kernel::HLERequestContext& ctx) {
@@ -215,11 +213,10 @@ void BSD::Connect(Kernel::HLERequestContext& ctx) {
LOG_DEBUG(Service, "called. fd={} addrlen={}", fd, ctx.GetReadBufferSize());
- ExecuteWork(ctx, "BSD:Connect", IsBlockingSocket(fd),
- ConnectWork{
- .fd = fd,
- .addr = ctx.ReadBuffer(),
- });
+ ExecuteWork(ctx, ConnectWork{
+ .fd = fd,
+ .addr = ctx.ReadBuffer(),
+ });
}
void BSD::GetPeerName(Kernel::HLERequestContext& ctx) {
@@ -327,12 +324,11 @@ void BSD::Recv(Kernel::HLERequestContext& ctx) {
LOG_DEBUG(Service, "called. fd={} flags=0x{:x} len={}", fd, flags, ctx.GetWriteBufferSize());
- ExecuteWork(ctx, "BSD:Recv", IsBlockingSocket(fd),
- RecvWork{
- .fd = fd,
- .flags = flags,
- .message = std::vector<u8>(ctx.GetWriteBufferSize()),
- });
+ ExecuteWork(ctx, RecvWork{
+ .fd = fd,
+ .flags = flags,
+ .message = std::vector<u8>(ctx.GetWriteBufferSize()),
+ });
}
void BSD::RecvFrom(Kernel::HLERequestContext& ctx) {
@@ -344,13 +340,12 @@ void BSD::RecvFrom(Kernel::HLERequestContext& ctx) {
LOG_DEBUG(Service, "called. fd={} flags=0x{:x} len={} addrlen={}", fd, flags,
ctx.GetWriteBufferSize(0), ctx.GetWriteBufferSize(1));
- ExecuteWork(ctx, "BSD:RecvFrom", IsBlockingSocket(fd),
- RecvFromWork{
- .fd = fd,
- .flags = flags,
- .message = std::vector<u8>(ctx.GetWriteBufferSize(0)),
- .addr = std::vector<u8>(ctx.GetWriteBufferSize(1)),
- });
+ ExecuteWork(ctx, RecvFromWork{
+ .fd = fd,
+ .flags = flags,
+ .message = std::vector<u8>(ctx.GetWriteBufferSize(0)),
+ .addr = std::vector<u8>(ctx.GetWriteBufferSize(1)),
+ });
}
void BSD::Send(Kernel::HLERequestContext& ctx) {
@@ -361,12 +356,11 @@ void BSD::Send(Kernel::HLERequestContext& ctx) {
LOG_DEBUG(Service, "called. fd={} flags=0x{:x} len={}", fd, flags, ctx.GetReadBufferSize());
- ExecuteWork(ctx, "BSD:Send", IsBlockingSocket(fd),
- SendWork{
- .fd = fd,
- .flags = flags,
- .message = ctx.ReadBuffer(),
- });
+ ExecuteWork(ctx, SendWork{
+ .fd = fd,
+ .flags = flags,
+ .message = ctx.ReadBuffer(),
+ });
}
void BSD::SendTo(Kernel::HLERequestContext& ctx) {
@@ -377,13 +371,12 @@ void BSD::SendTo(Kernel::HLERequestContext& ctx) {
LOG_DEBUG(Service, "called. fd={} flags=0x{} len={} addrlen={}", fd, flags,
ctx.GetReadBufferSize(0), ctx.GetReadBufferSize(1));
- ExecuteWork(ctx, "BSD:SendTo", IsBlockingSocket(fd),
- SendToWork{
- .fd = fd,
- .flags = flags,
- .message = ctx.ReadBuffer(0),
- .addr = ctx.ReadBuffer(1),
- });
+ ExecuteWork(ctx, SendToWork{
+ .fd = fd,
+ .flags = flags,
+ .message = ctx.ReadBuffer(0),
+ .addr = ctx.ReadBuffer(1),
+ });
}
void BSD::Write(Kernel::HLERequestContext& ctx) {
@@ -392,12 +385,11 @@ void BSD::Write(Kernel::HLERequestContext& ctx) {
LOG_DEBUG(Service, "called. fd={} len={}", fd, ctx.GetReadBufferSize());
- ExecuteWork(ctx, "BSD:Write", IsBlockingSocket(fd),
- SendWork{
- .fd = fd,
- .flags = 0,
- .message = ctx.ReadBuffer(),
- });
+ ExecuteWork(ctx, SendWork{
+ .fd = fd,
+ .flags = 0,
+ .message = ctx.ReadBuffer(),
+ });
}
void BSD::Close(Kernel::HLERequestContext& ctx) {
@@ -410,24 +402,9 @@ void BSD::Close(Kernel::HLERequestContext& ctx) {
}
template <typename Work>
-void BSD::ExecuteWork(Kernel::HLERequestContext& ctx, std::string_view sleep_reason,
- bool is_blocking, Work work) {
- if (!is_blocking) {
- work.Execute(this);
- work.Response(ctx);
- return;
- }
-
- // Signal a dummy response to make IPC validation happy
- // This will be overwritten by the SleepClientThread callback
+void BSD::ExecuteWork(Kernel::HLERequestContext& ctx, Work work) {
+ work.Execute(this);
work.Response(ctx);
-
- auto worker = worker_pool.CaptureWorker();
-
- ctx.SleepClientThread(std::string(sleep_reason), std::numeric_limits<u64>::max(),
- worker->Callback<Work>(), worker->KernelEvent());
-
- worker->SendWork(std::move(work));
}
std::pair<s32, Errno> BSD::SocketImpl(Domain domain, Type type, Protocol protocol) {
@@ -489,18 +466,18 @@ std::pair<s32, Errno> BSD::PollImpl(std::vector<u8>& write_buffer, std::vector<u
}
for (PollFD& pollfd : fds) {
- ASSERT(pollfd.revents == 0);
+ ASSERT(False(pollfd.revents));
if (pollfd.fd > static_cast<s32>(MAX_FD) || pollfd.fd < 0) {
LOG_ERROR(Service, "File descriptor handle={} is invalid", pollfd.fd);
- pollfd.revents = 0;
+ pollfd.revents = PollEvents{};
return {0, Errno::SUCCESS};
}
const std::optional<FileDescriptor>& descriptor = file_descriptors[pollfd.fd];
if (!descriptor) {
LOG_ERROR(Service, "File descriptor handle={} is not allocated", pollfd.fd);
- pollfd.revents = POLL_NVAL;
+ pollfd.revents = PollEvents::Nval;
return {0, Errno::SUCCESS};
}
}
@@ -510,7 +487,7 @@ std::pair<s32, Errno> BSD::PollImpl(std::vector<u8>& write_buffer, std::vector<u
Network::PollFD result;
result.socket = file_descriptors[pollfd.fd]->socket.get();
result.events = TranslatePollEventsToHost(pollfd.events);
- result.revents = 0;
+ result.revents = Network::PollEvents{};
return result;
});
@@ -636,7 +613,7 @@ std::pair<s32, Errno> BSD::FcntlImpl(s32 fd, FcntlCmd cmd, s32 arg) {
return {0, Errno::SUCCESS};
}
default:
- UNIMPLEMENTED_MSG("Unimplemented cmd={}", static_cast<int>(cmd));
+ UNIMPLEMENTED_MSG("Unimplemented cmd={}", cmd);
return {-1, Errno::SUCCESS};
}
}
@@ -679,7 +656,7 @@ Errno BSD::SetSockOptImpl(s32 fd, u32 level, OptName optname, size_t optlen, con
case OptName::RCVTIMEO:
return Translate(socket->SetRcvTimeo(value));
default:
- UNIMPLEMENTED_MSG("Unimplemented optname={}", static_cast<int>(optname));
+ UNIMPLEMENTED_MSG("Unimplemented optname={}", optname);
return Errno::SUCCESS;
}
}
@@ -807,18 +784,6 @@ bool BSD::IsFileDescriptorValid(s32 fd) const noexcept {
return true;
}
-bool BSD::IsBlockingSocket(s32 fd) const noexcept {
- // Inform invalid sockets as non-blocking
- // This way we avoid using a worker thread as it will fail without blocking host
- if (fd > static_cast<s32>(MAX_FD) || fd < 0) {
- return false;
- }
- if (!file_descriptors[fd]) {
- return false;
- }
- return (file_descriptors[fd]->flags & FLAG_O_NONBLOCK) != 0;
-}
-
void BSD::BuildErrnoResponse(Kernel::HLERequestContext& ctx, Errno bsd_errno) const noexcept {
IPC::ResponseBuilder rb{ctx, 4};
@@ -827,8 +792,7 @@ void BSD::BuildErrnoResponse(Kernel::HLERequestContext& ctx, Errno bsd_errno) co
rb.PushEnum(bsd_errno);
}
-BSD::BSD(Core::System& system_, const char* name)
- : ServiceFramework{system_, name}, worker_pool{system_, this} {
+BSD::BSD(Core::System& system_, const char* name) : ServiceFramework{system_, name} {
// clang-format off
static const FunctionInfo functions[] = {
{0, &BSD::RegisterClient, "RegisterClient"},
diff --git a/src/core/hle/service/sockets/bsd.h b/src/core/hle/service/sockets/bsd.h
index f14713fc4..6da0bfeb2 100644
--- a/src/core/hle/service/sockets/bsd.h
+++ b/src/core/hle/service/sockets/bsd.h
@@ -11,7 +11,6 @@
#include "common/common_types.h"
#include "core/hle/kernel/hle_ipc.h"
#include "core/hle/service/service.h"
-#include "core/hle/service/sockets/blocking_worker.h"
#include "core/hle/service/sockets/sockets.h"
namespace Core {
@@ -138,8 +137,7 @@ private:
void Close(Kernel::HLERequestContext& ctx);
template <typename Work>
- void ExecuteWork(Kernel::HLERequestContext& ctx, std::string_view sleep_reason,
- bool is_blocking, Work work);
+ void ExecuteWork(Kernel::HLERequestContext& ctx, Work work);
std::pair<s32, Errno> SocketImpl(Domain domain, Type type, Protocol protocol);
std::pair<s32, Errno> PollImpl(std::vector<u8>& write_buffer, std::vector<u8> read_buffer,
@@ -163,15 +161,10 @@ private:
s32 FindFreeFileDescriptorHandle() noexcept;
bool IsFileDescriptorValid(s32 fd) const noexcept;
- bool IsBlockingSocket(s32 fd) const noexcept;
void BuildErrnoResponse(Kernel::HLERequestContext& ctx, Errno bsd_errno) const noexcept;
std::array<std::optional<FileDescriptor>, MAX_FD> file_descriptors;
-
- BlockingWorkerPool<BSD, PollWork, AcceptWork, ConnectWork, RecvWork, RecvFromWork, SendWork,
- SendToWork>
- worker_pool;
};
class BSDCFG final : public ServiceFramework<BSDCFG> {
diff --git a/src/core/hle/service/sockets/sockets.h b/src/core/hle/service/sockets/sockets.h
index 89a410076..5a65ed2a9 100644
--- a/src/core/hle/service/sockets/sockets.h
+++ b/src/core/hle/service/sockets/sockets.h
@@ -69,10 +69,22 @@ struct SockAddrIn {
std::array<u8, 8> zeroes;
};
+enum class PollEvents : u16 {
+ // Using Pascal case because IN is a macro on Windows.
+ In = 1 << 0,
+ Pri = 1 << 1,
+ Out = 1 << 2,
+ Err = 1 << 3,
+ Hup = 1 << 4,
+ Nval = 1 << 5,
+};
+
+DECLARE_ENUM_FLAG_OPERATORS(PollEvents);
+
struct PollFD {
s32 fd;
- u16 events;
- u16 revents;
+ PollEvents events;
+ PollEvents revents;
};
struct Linger {
@@ -80,13 +92,6 @@ struct Linger {
u32 linger;
};
-constexpr u16 POLL_IN = 0x01;
-constexpr u16 POLL_PRI = 0x02;
-constexpr u16 POLL_OUT = 0x04;
-constexpr u16 POLL_ERR = 0x08;
-constexpr u16 POLL_HUP = 0x10;
-constexpr u16 POLL_NVAL = 0x20;
-
constexpr u32 FLAG_MSG_DONTWAIT = 0x80;
constexpr u32 FLAG_O_NONBLOCK = 0x800;
diff --git a/src/core/hle/service/sockets/sockets_translate.cpp b/src/core/hle/service/sockets/sockets_translate.cpp
index 2e626fd86..c822d21b8 100644
--- a/src/core/hle/service/sockets/sockets_translate.cpp
+++ b/src/core/hle/service/sockets/sockets_translate.cpp
@@ -27,7 +27,7 @@ Errno Translate(Network::Errno value) {
case Network::Errno::NOTCONN:
return Errno::NOTCONN;
default:
- UNIMPLEMENTED_MSG("Unimplemented errno={}", static_cast<int>(value));
+ UNIMPLEMENTED_MSG("Unimplemented errno={}", value);
return Errno::SUCCESS;
}
}
@@ -41,7 +41,7 @@ Network::Domain Translate(Domain domain) {
case Domain::INET:
return Network::Domain::INET;
default:
- UNIMPLEMENTED_MSG("Unimplemented domain={}", static_cast<int>(domain));
+ UNIMPLEMENTED_MSG("Unimplemented domain={}", domain);
return {};
}
}
@@ -51,7 +51,7 @@ Domain Translate(Network::Domain domain) {
case Network::Domain::INET:
return Domain::INET;
default:
- UNIMPLEMENTED_MSG("Unimplemented domain={}", static_cast<int>(domain));
+ UNIMPLEMENTED_MSG("Unimplemented domain={}", domain);
return {};
}
}
@@ -63,7 +63,7 @@ Network::Type Translate(Type type) {
case Type::DGRAM:
return Network::Type::DGRAM;
default:
- UNIMPLEMENTED_MSG("Unimplemented type={}", static_cast<int>(type));
+ UNIMPLEMENTED_MSG("Unimplemented type={}", type);
}
}
@@ -84,48 +84,48 @@ Network::Protocol Translate(Type type, Protocol protocol) {
case Protocol::UDP:
return Network::Protocol::UDP;
default:
- UNIMPLEMENTED_MSG("Unimplemented protocol={}", static_cast<int>(protocol));
+ UNIMPLEMENTED_MSG("Unimplemented protocol={}", protocol);
return Network::Protocol::TCP;
}
}
-u16 TranslatePollEventsToHost(u32 flags) {
- u32 result = 0;
- const auto translate = [&result, &flags](u32 from, u32 to) {
- if ((flags & from) != 0) {
+Network::PollEvents TranslatePollEventsToHost(PollEvents flags) {
+ Network::PollEvents result{};
+ const auto translate = [&result, &flags](PollEvents from, Network::PollEvents to) {
+ if (True(flags & from)) {
flags &= ~from;
result |= to;
}
};
- translate(POLL_IN, Network::POLL_IN);
- translate(POLL_PRI, Network::POLL_PRI);
- translate(POLL_OUT, Network::POLL_OUT);
- translate(POLL_ERR, Network::POLL_ERR);
- translate(POLL_HUP, Network::POLL_HUP);
- translate(POLL_NVAL, Network::POLL_NVAL);
-
- UNIMPLEMENTED_IF_MSG(flags != 0, "Unimplemented flags={}", flags);
- return static_cast<u16>(result);
+ translate(PollEvents::In, Network::PollEvents::In);
+ translate(PollEvents::Pri, Network::PollEvents::Pri);
+ translate(PollEvents::Out, Network::PollEvents::Out);
+ translate(PollEvents::Err, Network::PollEvents::Err);
+ translate(PollEvents::Hup, Network::PollEvents::Hup);
+ translate(PollEvents::Nval, Network::PollEvents::Nval);
+
+ UNIMPLEMENTED_IF_MSG((u16)flags != 0, "Unimplemented flags={}", (u16)flags);
+ return result;
}
-u16 TranslatePollEventsToGuest(u32 flags) {
- u32 result = 0;
- const auto translate = [&result, &flags](u32 from, u32 to) {
- if ((flags & from) != 0) {
+PollEvents TranslatePollEventsToGuest(Network::PollEvents flags) {
+ PollEvents result{};
+ const auto translate = [&result, &flags](Network::PollEvents from, PollEvents to) {
+ if (True(flags & from)) {
flags &= ~from;
result |= to;
}
};
- translate(Network::POLL_IN, POLL_IN);
- translate(Network::POLL_PRI, POLL_PRI);
- translate(Network::POLL_OUT, POLL_OUT);
- translate(Network::POLL_ERR, POLL_ERR);
- translate(Network::POLL_HUP, POLL_HUP);
- translate(Network::POLL_NVAL, POLL_NVAL);
+ translate(Network::PollEvents::In, PollEvents::In);
+ translate(Network::PollEvents::Pri, PollEvents::Pri);
+ translate(Network::PollEvents::Out, PollEvents::Out);
+ translate(Network::PollEvents::Err, PollEvents::Err);
+ translate(Network::PollEvents::Hup, PollEvents::Hup);
+ translate(Network::PollEvents::Nval, PollEvents::Nval);
- UNIMPLEMENTED_IF_MSG(flags != 0, "Unimplemented flags={}", flags);
- return static_cast<u16>(result);
+ UNIMPLEMENTED_IF_MSG((u16)flags != 0, "Unimplemented flags={}", (u16)flags);
+ return result;
}
Network::SockAddrIn Translate(SockAddrIn value) {
@@ -157,7 +157,7 @@ Network::ShutdownHow Translate(ShutdownHow how) {
case ShutdownHow::RDWR:
return Network::ShutdownHow::RDWR;
default:
- UNIMPLEMENTED_MSG("Unimplemented how={}", static_cast<int>(how));
+ UNIMPLEMENTED_MSG("Unimplemented how={}", how);
return {};
}
}
diff --git a/src/core/hle/service/sockets/sockets_translate.h b/src/core/hle/service/sockets/sockets_translate.h
index e498913d4..057d1ff22 100644
--- a/src/core/hle/service/sockets/sockets_translate.h
+++ b/src/core/hle/service/sockets/sockets_translate.h
@@ -31,10 +31,10 @@ Network::Type Translate(Type type);
Network::Protocol Translate(Type type, Protocol protocol);
/// Translate abstract poll event flags to guest poll event flags
-u16 TranslatePollEventsToHost(u32 flags);
+Network::PollEvents TranslatePollEventsToHost(PollEvents flags);
/// Translate guest poll event flags to abstract poll event flags
-u16 TranslatePollEventsToGuest(u32 flags);
+PollEvents TranslatePollEventsToGuest(Network::PollEvents flags);
/// Translate guest socket address structure to abstract socket address structure
Network::SockAddrIn Translate(SockAddrIn value);
diff --git a/src/core/hle/service/time/time.cpp b/src/core/hle/service/time/time.cpp
index 7b7ac282d..abc753d5d 100644
--- a/src/core/hle/service/time/time.cpp
+++ b/src/core/hle/service/time/time.cpp
@@ -10,8 +10,8 @@
#include "core/hle/ipc_helpers.h"
#include "core/hle/kernel/client_port.h"
#include "core/hle/kernel/client_session.h"
+#include "core/hle/kernel/k_scheduler.h"
#include "core/hle/kernel/kernel.h"
-#include "core/hle/kernel/scheduler.h"
#include "core/hle/service/time/interface.h"
#include "core/hle/service/time/time.h"
#include "core/hle/service/time/time_sharedmemory.h"
diff --git a/src/core/hle/service/vi/vi.cpp b/src/core/hle/service/vi/vi.cpp
index af5b8b0b9..968cd16b6 100644
--- a/src/core/hle/service/vi/vi.cpp
+++ b/src/core/hle/service/vi/vi.cpp
@@ -282,18 +282,24 @@ public:
void DeserializeData() override {
[[maybe_unused]] const std::u16string token = ReadInterfaceToken();
data = Read<Data>();
- buffer = Read<NVFlinger::IGBPBuffer>();
+ if (data.contains_object != 0) {
+ buffer_container = Read<BufferContainer>();
+ }
}
struct Data {
u32_le slot;
- INSERT_PADDING_WORDS(1);
+ u32_le contains_object;
+ };
+
+ struct BufferContainer {
u32_le graphic_buffer_length;
INSERT_PADDING_WORDS(1);
+ NVFlinger::IGBPBuffer buffer{};
};
- Data data;
- NVFlinger::IGBPBuffer buffer;
+ Data data{};
+ BufferContainer buffer_container{};
};
class IGBPSetPreallocatedBufferResponseParcel : public Parcel {
@@ -528,10 +534,9 @@ private:
const u32 flags = rp.Pop<u32>();
LOG_DEBUG(Service_VI, "called. id=0x{:08X} transaction={:X}, flags=0x{:08X}", id,
- static_cast<u32>(transaction), flags);
+ transaction, flags);
- const auto guard = nv_flinger.Lock();
- auto& buffer_queue = nv_flinger.FindBufferQueue(id);
+ auto& buffer_queue = *nv_flinger.FindBufferQueue(id);
switch (transaction) {
case TransactionId::Connect: {
@@ -541,13 +546,16 @@ private:
Settings::values.resolution_factor.GetValue()),
static_cast<u32>(static_cast<u32>(DisplayResolution::UndockedHeight) *
Settings::values.resolution_factor.GetValue())};
+
+ buffer_queue.Connect();
+
ctx.WriteBuffer(response.Serialize());
break;
}
case TransactionId::SetPreallocatedBuffer: {
IGBPSetPreallocatedBufferRequestParcel request{ctx.ReadBuffer()};
- buffer_queue.SetPreallocatedBuffer(request.data.slot, request.buffer);
+ buffer_queue.SetPreallocatedBuffer(request.data.slot, request.buffer_container.buffer);
IGBPSetPreallocatedBufferResponseParcel response{};
ctx.WriteBuffer(response.Serialize());
@@ -557,40 +565,25 @@ private:
IGBPDequeueBufferRequestParcel request{ctx.ReadBuffer()};
const u32 width{request.data.width};
const u32 height{request.data.height};
- auto result = buffer_queue.DequeueBuffer(width, height);
-
- if (result) {
- // Buffer is available
- IGBPDequeueBufferResponseParcel response{result->first, *result->second};
- ctx.WriteBuffer(response.Serialize());
- } else {
- // Wait the current thread until a buffer becomes available
- ctx.SleepClientThread(
- "IHOSBinderDriver::DequeueBuffer", UINT64_MAX,
- [=, this](std::shared_ptr<Kernel::Thread> thread,
- Kernel::HLERequestContext& ctx, Kernel::ThreadWakeupReason reason) {
- // Repeat TransactParcel DequeueBuffer when a buffer is available
- const auto guard = nv_flinger.Lock();
- auto& buffer_queue = nv_flinger.FindBufferQueue(id);
- auto result = buffer_queue.DequeueBuffer(width, height);
- ASSERT_MSG(result != std::nullopt, "Could not dequeue buffer.");
-
- IGBPDequeueBufferResponseParcel response{result->first, *result->second};
- ctx.WriteBuffer(response.Serialize());
- IPC::ResponseBuilder rb{ctx, 2};
- rb.Push(RESULT_SUCCESS);
- },
- buffer_queue.GetWritableBufferWaitEvent());
- }
+
+ do {
+ if (auto result = buffer_queue.DequeueBuffer(width, height); result) {
+ // Buffer is available
+ IGBPDequeueBufferResponseParcel response{result->first, *result->second};
+ ctx.WriteBuffer(response.Serialize());
+ break;
+ }
+ } while (buffer_queue.IsConnected());
+
break;
}
case TransactionId::RequestBuffer: {
IGBPRequestBufferRequestParcel request{ctx.ReadBuffer()};
auto& buffer = buffer_queue.RequestBuffer(request.slot);
-
IGBPRequestBufferResponseParcel response{buffer};
ctx.WriteBuffer(response.Serialize());
+
break;
}
case TransactionId::QueueBuffer: {
@@ -676,7 +669,7 @@ private:
LOG_WARNING(Service_VI, "(STUBBED) called id={}, unknown={:08X}", id, unknown);
- const auto& buffer_queue = nv_flinger.FindBufferQueue(id);
+ const auto& buffer_queue = *nv_flinger.FindBufferQueue(id);
// TODO(Subv): Find out what this actually is.
IPC::ResponseBuilder rb{ctx, 2, 1};
@@ -1066,8 +1059,8 @@ private:
const auto scaling_mode = rp.PopEnum<NintendoScaleMode>();
const u64 unknown = rp.Pop<u64>();
- LOG_DEBUG(Service_VI, "called. scaling_mode=0x{:08X}, unknown=0x{:016X}",
- static_cast<u32>(scaling_mode), unknown);
+ LOG_DEBUG(Service_VI, "called. scaling_mode=0x{:08X}, unknown=0x{:016X}", scaling_mode,
+ unknown);
IPC::ResponseBuilder rb{ctx, 2};
@@ -1210,7 +1203,7 @@ private:
void ConvertScalingMode(Kernel::HLERequestContext& ctx) {
IPC::RequestParser rp{ctx};
const auto mode = rp.PopEnum<NintendoScaleMode>();
- LOG_DEBUG(Service_VI, "called mode={}", static_cast<u32>(mode));
+ LOG_DEBUG(Service_VI, "called mode={}", mode);
const auto converted_mode = ConvertScalingModeImpl(mode);
@@ -1230,8 +1223,8 @@ private:
const auto height = rp.Pop<u64>();
LOG_DEBUG(Service_VI, "called width={}, height={}", width, height);
- constexpr std::size_t base_size = 0x20000;
- constexpr std::size_t alignment = 0x1000;
+ constexpr u64 base_size = 0x20000;
+ constexpr u64 alignment = 0x1000;
const auto texture_size = width * height * 4;
const auto out_size = (texture_size + base_size - 1) / base_size * base_size;
@@ -1311,7 +1304,7 @@ void detail::GetDisplayServiceImpl(Kernel::HLERequestContext& ctx, Core::System&
const auto policy = rp.PopEnum<Policy>();
if (!IsValidServiceAccess(permission, policy)) {
- LOG_ERROR(Service_VI, "Permission denied for policy {}", static_cast<u32>(policy));
+ LOG_ERROR(Service_VI, "Permission denied for policy {}", policy);
IPC::ResponseBuilder rb{ctx, 2};
rb.Push(ERR_PERMISSION_DENIED);
return;
diff --git a/src/core/loader/deconstructed_rom_directory.h b/src/core/loader/deconstructed_rom_directory.h
index 35d340317..3c968580f 100644
--- a/src/core/loader/deconstructed_rom_directory.h
+++ b/src/core/loader/deconstructed_rom_directory.h
@@ -32,7 +32,7 @@ public:
/**
* Returns the type of the file
- * @param file std::shared_ptr<VfsFile> open file
+ * @param file open file
* @return FileType found, or FileType::Error if this loader doesn't know it
*/
static FileType IdentifyType(const FileSys::VirtualFile& file);
diff --git a/src/core/loader/elf.h b/src/core/loader/elf.h
index 3527933ad..2067932c7 100644
--- a/src/core/loader/elf.h
+++ b/src/core/loader/elf.h
@@ -21,7 +21,7 @@ public:
/**
* Returns the type of the file
- * @param file std::shared_ptr<VfsFile> open file
+ * @param file open file
* @return FileType found, or FileType::Error if this loader doesn't know it
*/
static FileType IdentifyType(const FileSys::VirtualFile& file);
diff --git a/src/core/loader/kip.h b/src/core/loader/kip.h
index dee05a7b5..14a85e295 100644
--- a/src/core/loader/kip.h
+++ b/src/core/loader/kip.h
@@ -23,7 +23,7 @@ public:
/**
* Returns the type of the file
- * @param file std::shared_ptr<VfsFile> open file
+ * @param file open file
* @return FileType found, or FileType::Error if this loader doesn't know it
*/
static FileType IdentifyType(const FileSys::VirtualFile& file);
diff --git a/src/core/loader/nax.h b/src/core/loader/nax.h
index c2b7722b5..a5b5e2ae1 100644
--- a/src/core/loader/nax.h
+++ b/src/core/loader/nax.h
@@ -28,7 +28,7 @@ public:
/**
* Returns the type of the file
- * @param file std::shared_ptr<VfsFile> open file
+ * @param file open file
* @return FileType found, or FileType::Error if this loader doesn't know it
*/
static FileType IdentifyType(const FileSys::VirtualFile& file);
diff --git a/src/core/loader/nca.h b/src/core/loader/nca.h
index 711070294..918792800 100644
--- a/src/core/loader/nca.h
+++ b/src/core/loader/nca.h
@@ -28,7 +28,7 @@ public:
/**
* Returns the type of the file
- * @param file std::shared_ptr<VfsFile> open file
+ * @param file open file
* @return FileType found, or FileType::Error if this loader doesn't know it
*/
static FileType IdentifyType(const FileSys::VirtualFile& file);
diff --git a/src/core/loader/nro.h b/src/core/loader/nro.h
index a2aab2ecc..a82b66221 100644
--- a/src/core/loader/nro.h
+++ b/src/core/loader/nro.h
@@ -32,7 +32,7 @@ public:
/**
* Returns the type of the file
- * @param file std::shared_ptr<VfsFile> open file
+ * @param file open file
* @return FileType found, or FileType::Error if this loader doesn't know it
*/
static FileType IdentifyType(const FileSys::VirtualFile& file);
diff --git a/src/core/loader/nso.h b/src/core/loader/nso.h
index d331096ae..3af461b5f 100644
--- a/src/core/loader/nso.h
+++ b/src/core/loader/nso.h
@@ -75,7 +75,7 @@ public:
/**
* Returns the type of the file
- * @param file std::shared_ptr<VfsFile> open file
+ * @param file open file
* @return FileType found, or FileType::Error if this loader doesn't know it
*/
static FileType IdentifyType(const FileSys::VirtualFile& file);
diff --git a/src/core/loader/nsp.h b/src/core/loader/nsp.h
index f0518ac47..d48d87f2c 100644
--- a/src/core/loader/nsp.h
+++ b/src/core/loader/nsp.h
@@ -34,7 +34,7 @@ public:
/**
* Returns the type of the file
- * @param file std::shared_ptr<VfsFile> open file
+ * @param file open file
* @return FileType found, or FileType::Error if this loader doesn't know it
*/
static FileType IdentifyType(const FileSys::VirtualFile& file);
diff --git a/src/core/loader/xci.h b/src/core/loader/xci.h
index 764dc8328..9f0ceb5ef 100644
--- a/src/core/loader/xci.h
+++ b/src/core/loader/xci.h
@@ -34,7 +34,7 @@ public:
/**
* Returns the type of the file
- * @param file std::shared_ptr<VfsFile> open file
+ * @param file open file
* @return FileType found, or FileType::Error if this loader doesn't know it
*/
static FileType IdentifyType(const FileSys::VirtualFile& file);
diff --git a/src/core/memory.cpp b/src/core/memory.cpp
index b88aa5c40..54a848936 100644
--- a/src/core/memory.cpp
+++ b/src/core/memory.cpp
@@ -4,6 +4,7 @@
#include <algorithm>
#include <cstring>
+#include <mutex>
#include <optional>
#include <utility>
@@ -497,7 +498,21 @@ struct Memory::Impl {
return CopyBlock(*system.CurrentProcess(), dest_addr, src_addr, size);
}
+ struct PageEntry {
+ u8* const pointer;
+ const Common::PageType attribute;
+ };
+
+ PageEntry SafePageEntry(std::size_t base) const {
+ std::lock_guard lock{rasterizer_cache_guard};
+ return {
+ .pointer = current_page_table->pointers[base],
+ .attribute = current_page_table->attributes[base],
+ };
+ }
+
void RasterizerMarkRegionCached(VAddr vaddr, u64 size, bool cached) {
+ std::lock_guard lock{rasterizer_cache_guard};
if (vaddr == 0) {
return;
}
@@ -630,16 +645,22 @@ struct Memory::Impl {
*/
template <typename T>
T Read(const VAddr vaddr) {
- const u8* const page_pointer = current_page_table->pointers[vaddr >> PAGE_BITS];
- if (page_pointer != nullptr) {
- // NOTE: Avoid adding any extra logic to this fast-path block
+ // Avoid adding any extra logic to this fast-path block
+ if (const u8* const pointer = current_page_table->pointers[vaddr >> PAGE_BITS]) {
T value;
- std::memcpy(&value, &page_pointer[vaddr], sizeof(T));
+ std::memcpy(&value, &pointer[vaddr], sizeof(T));
return value;
}
- const Common::PageType type = current_page_table->attributes[vaddr >> PAGE_BITS];
- switch (type) {
+ // Otherwise, we need to grab the page with a lock, in case it is currently being modified
+ const auto entry = SafePageEntry(vaddr >> PAGE_BITS);
+ if (entry.pointer) {
+ T value;
+ std::memcpy(&value, &entry.pointer[vaddr], sizeof(T));
+ return value;
+ }
+
+ switch (entry.attribute) {
case Common::PageType::Unmapped:
LOG_ERROR(HW_Memory, "Unmapped Read{} @ 0x{:08X}", sizeof(T) * 8, vaddr);
return 0;
@@ -667,20 +688,24 @@ struct Memory::Impl {
* @tparam T The data type to write to memory. This type *must* be
* trivially copyable, otherwise the behavior of this function
* is undefined.
- *
- * @returns The instance of T write to the specified virtual address.
*/
template <typename T>
void Write(const VAddr vaddr, const T data) {
- u8* const page_pointer = current_page_table->pointers[vaddr >> PAGE_BITS];
- if (page_pointer != nullptr) {
- // NOTE: Avoid adding any extra logic to this fast-path block
- std::memcpy(&page_pointer[vaddr], &data, sizeof(T));
+ // Avoid adding any extra logic to this fast-path block
+ if (u8* const pointer = current_page_table->pointers[vaddr >> PAGE_BITS]) {
+ std::memcpy(&pointer[vaddr], &data, sizeof(T));
return;
}
- const Common::PageType type = current_page_table->attributes[vaddr >> PAGE_BITS];
- switch (type) {
+ // Otherwise, we need to grab the page with a lock, in case it is currently being modified
+ const auto entry = SafePageEntry(vaddr >> PAGE_BITS);
+ if (entry.pointer) {
+ // Memory was mapped, we are done
+ std::memcpy(&entry.pointer[vaddr], &data, sizeof(T));
+ return;
+ }
+
+ switch (entry.attribute) {
case Common::PageType::Unmapped:
LOG_ERROR(HW_Memory, "Unmapped Write{} 0x{:08X} @ 0x{:016X}", sizeof(data) * 8,
static_cast<u32>(data), vaddr);
@@ -758,6 +783,7 @@ struct Memory::Impl {
return true;
}
+ mutable std::mutex rasterizer_cache_guard;
Common::PageTable* current_page_table = nullptr;
Core::System& system;
};
diff --git a/src/core/network/network.cpp b/src/core/network/network.cpp
index 5ef2e8511..681e93468 100644
--- a/src/core/network/network.cpp
+++ b/src/core/network/network.cpp
@@ -11,7 +11,7 @@
#ifdef _WIN32
#define _WINSOCK_DEPRECATED_NO_WARNINGS // gethostname
#include <winsock2.h>
-#elif __unix__
+#elif YUZU_UNIX
#include <errno.h>
#include <fcntl.h>
#include <netdb.h>
@@ -54,7 +54,7 @@ constexpr IPv4Address TranslateIPv4(in_addr addr) {
sockaddr TranslateFromSockAddrIn(SockAddrIn input) {
sockaddr_in result;
-#ifdef __unix__
+#if YUZU_UNIX
result.sin_len = sizeof(result);
#endif
@@ -63,7 +63,7 @@ sockaddr TranslateFromSockAddrIn(SockAddrIn input) {
result.sin_family = AF_INET;
break;
default:
- UNIMPLEMENTED_MSG("Unhandled sockaddr family={}", static_cast<int>(input.family));
+ UNIMPLEMENTED_MSG("Unhandled sockaddr family={}", input.family);
result.sin_family = AF_INET;
break;
}
@@ -99,7 +99,7 @@ bool EnableNonBlock(SOCKET fd, bool enable) {
return ioctlsocket(fd, FIONBIO, &value) != SOCKET_ERROR;
}
-#elif __unix__ // ^ _WIN32 v __unix__
+#elif YUZU_UNIX // ^ _WIN32 v YUZU_UNIX
using SOCKET = int;
using WSAPOLLFD = pollfd;
@@ -133,7 +133,7 @@ sockaddr TranslateFromSockAddrIn(SockAddrIn input) {
result.sin_family = AF_INET;
break;
default:
- UNIMPLEMENTED_MSG("Unhandled sockaddr family={}", static_cast<int>(input.family));
+ UNIMPLEMENTED_MSG("Unhandled sockaddr family={}", input.family);
result.sin_family = AF_INET;
break;
}
@@ -186,7 +186,7 @@ int TranslateDomain(Domain domain) {
case Domain::INET:
return AF_INET;
default:
- UNIMPLEMENTED_MSG("Unimplemented domain={}", static_cast<int>(domain));
+ UNIMPLEMENTED_MSG("Unimplemented domain={}", domain);
return 0;
}
}
@@ -198,7 +198,7 @@ int TranslateType(Type type) {
case Type::DGRAM:
return SOCK_DGRAM;
default:
- UNIMPLEMENTED_MSG("Unimplemented type={}", static_cast<int>(type));
+ UNIMPLEMENTED_MSG("Unimplemented type={}", type);
return 0;
}
}
@@ -210,7 +210,7 @@ int TranslateProtocol(Protocol protocol) {
case Protocol::UDP:
return IPPROTO_UDP;
default:
- UNIMPLEMENTED_MSG("Unimplemented protocol={}", static_cast<int>(protocol));
+ UNIMPLEMENTED_MSG("Unimplemented protocol={}", protocol);
return 0;
}
}
@@ -238,49 +238,49 @@ SockAddrIn TranslateToSockAddrIn(sockaddr input_) {
return result;
}
-u16 TranslatePollEvents(u32 events) {
- u32 result = 0;
+short TranslatePollEvents(PollEvents events) {
+ short result = 0;
- if ((events & POLL_IN) != 0) {
- events &= ~POLL_IN;
+ if (True(events & PollEvents::In)) {
+ events &= ~PollEvents::In;
result |= POLLIN;
}
- if ((events & POLL_PRI) != 0) {
- events &= ~POLL_PRI;
+ if (True(events & PollEvents::Pri)) {
+ events &= ~PollEvents::Pri;
#ifdef _WIN32
LOG_WARNING(Service, "Winsock doesn't support POLLPRI");
#else
- result |= POLL_PRI;
+ result |= POLLPRI;
#endif
}
- if ((events & POLL_OUT) != 0) {
- events &= ~POLL_OUT;
+ if (True(events & PollEvents::Out)) {
+ events &= ~PollEvents::Out;
result |= POLLOUT;
}
- UNIMPLEMENTED_IF_MSG(events != 0, "Unhandled guest events=0x{:x}", events);
+ UNIMPLEMENTED_IF_MSG((u16)events != 0, "Unhandled guest events=0x{:x}", (u16)events);
- return static_cast<u16>(result);
+ return result;
}
-u16 TranslatePollRevents(u32 revents) {
- u32 result = 0;
- const auto translate = [&result, &revents](u32 host, u32 guest) {
+PollEvents TranslatePollRevents(short revents) {
+ PollEvents result{};
+ const auto translate = [&result, &revents](short host, PollEvents guest) {
if ((revents & host) != 0) {
- revents &= ~host;
+ revents &= static_cast<short>(~host);
result |= guest;
}
};
- translate(POLLIN, POLL_IN);
- translate(POLLPRI, POLL_PRI);
- translate(POLLOUT, POLL_OUT);
- translate(POLLERR, POLL_ERR);
- translate(POLLHUP, POLL_HUP);
+ translate(POLLIN, PollEvents::In);
+ translate(POLLPRI, PollEvents::Pri);
+ translate(POLLOUT, PollEvents::Out);
+ translate(POLLERR, PollEvents::Err);
+ translate(POLLHUP, PollEvents::Hup);
UNIMPLEMENTED_IF_MSG(revents != 0, "Unhandled host revents=0x{:x}", revents);
- return static_cast<u16>(result);
+ return result;
}
template <typename T>
@@ -350,7 +350,7 @@ std::pair<s32, Errno> Poll(std::vector<PollFD>& pollfds, s32 timeout) {
}
for (size_t i = 0; i < num; ++i) {
- pollfds[i].revents = TranslatePollRevents(static_cast<u32>(host_pollfds[i].revents));
+ pollfds[i].revents = TranslatePollRevents(host_pollfds[i].revents);
}
if (result > 0) {
@@ -482,7 +482,7 @@ Errno Socket::Shutdown(ShutdownHow how) {
host_how = SD_BOTH;
break;
default:
- UNIMPLEMENTED_MSG("Unimplemented flag how={}", static_cast<int>(how));
+ UNIMPLEMENTED_MSG("Unimplemented flag how={}", how);
return Errno::SUCCESS;
}
if (shutdown(fd, host_how) != SOCKET_ERROR) {
diff --git a/src/core/network/network.h b/src/core/network/network.h
index 0622e4593..76b2821f2 100644
--- a/src/core/network/network.h
+++ b/src/core/network/network.h
@@ -61,19 +61,25 @@ struct SockAddrIn {
};
/// Cross-platform poll fd structure
+
+enum class PollEvents : u16 {
+ // Using Pascal case because IN is a macro on Windows.
+ In = 1 << 0,
+ Pri = 1 << 1,
+ Out = 1 << 2,
+ Err = 1 << 3,
+ Hup = 1 << 4,
+ Nval = 1 << 5,
+};
+
+DECLARE_ENUM_FLAG_OPERATORS(PollEvents);
+
struct PollFD {
Socket* socket;
- u16 events;
- u16 revents;
+ PollEvents events;
+ PollEvents revents;
};
-constexpr u16 POLL_IN = 1 << 0;
-constexpr u16 POLL_PRI = 1 << 1;
-constexpr u16 POLL_OUT = 1 << 2;
-constexpr u16 POLL_ERR = 1 << 3;
-constexpr u16 POLL_HUP = 1 << 4;
-constexpr u16 POLL_NVAL = 1 << 5;
-
class NetworkInstance {
public:
explicit NetworkInstance();
diff --git a/src/core/network/sockets.h b/src/core/network/sockets.h
index 7bdff0fe4..a44393325 100644
--- a/src/core/network/sockets.h
+++ b/src/core/network/sockets.h
@@ -9,7 +9,7 @@
#if defined(_WIN32)
#include <winsock.h>
-#elif !defined(__unix__)
+#elif !YUZU_UNIX
#error "Platform not implemented"
#endif
@@ -84,7 +84,7 @@ public:
#if defined(_WIN32)
SOCKET fd = INVALID_SOCKET;
-#elif defined(__unix__)
+#elif YUZU_UNIX
int fd = -1;
#endif
};
diff --git a/src/core/settings.cpp b/src/core/settings.cpp
index e9997a263..39306509a 100644
--- a/src/core/settings.cpp
+++ b/src/core/settings.cpp
@@ -72,8 +72,6 @@ void LogSettings() {
log_setting("DataStorage_UseVirtualSd", values.use_virtual_sd);
log_setting("DataStorage_NandDir", Common::FS::GetUserPath(Common::FS::UserPath::NANDDir));
log_setting("DataStorage_SdmcDir", Common::FS::GetUserPath(Common::FS::UserPath::SDMCDir));
- log_setting("Debugging_UseGdbstub", values.use_gdbstub);
- log_setting("Debugging_GdbstubPort", values.gdbstub_port);
log_setting("Debugging_ProgramArgs", values.program_args);
log_setting("Services_BCATBackend", values.bcat_backend);
log_setting("Services_BCATBoxcatLocal", values.bcat_boxcat_local);
@@ -150,9 +148,4 @@ void RestoreGlobalState(bool is_powered_on) {
values.motion_enabled.SetGlobal(true);
}
-void Sanitize() {
- values.use_asynchronous_gpu_emulation.SetValue(
- values.use_asynchronous_gpu_emulation.GetValue() || values.use_multi_core.GetValue());
-}
-
} // namespace Settings
diff --git a/src/core/settings.h b/src/core/settings.h
index 8e076f7ef..0cd3c0c84 100644
--- a/src/core/settings.h
+++ b/src/core/settings.h
@@ -180,6 +180,8 @@ struct Values {
std::string motion_device;
std::string udp_input_servers;
+ bool emulate_analog_keyboard;
+
bool mouse_enabled;
std::string mouse_device;
MouseButtonsRaw mouse_buttons;
@@ -255,7 +257,4 @@ void LogSettings();
// Restore the global state of all applicable settings in the Values struct
void RestoreGlobalState(bool is_powered_on);
-// Fixes settings that are known to cause issues with the emulator
-void Sanitize();
-
} // namespace Settings
diff --git a/src/input_common/analog_from_button.cpp b/src/input_common/analog_from_button.cpp
index d748c1c04..40b516f85 100755
--- a/src/input_common/analog_from_button.cpp
+++ b/src/input_common/analog_from_button.cpp
@@ -6,6 +6,7 @@
#include <cmath>
#include <thread>
#include "common/math_util.h"
+#include "core/settings.h"
#include "input_common/analog_from_button.h"
namespace InputCommon {
@@ -112,7 +113,26 @@ public:
}
std::tuple<float, float> GetStatus() const override {
- return std::make_tuple(std::cos(angle) * amplitude, std::sin(angle) * amplitude);
+ if (Settings::values.emulate_analog_keyboard) {
+ return std::make_tuple(std::cos(angle) * amplitude, std::sin(angle) * amplitude);
+ }
+ constexpr float SQRT_HALF = 0.707106781f;
+ int x = 0, y = 0;
+ if (right->GetStatus()) {
+ ++x;
+ }
+ if (left->GetStatus()) {
+ --x;
+ }
+ if (up->GetStatus()) {
+ ++y;
+ }
+ if (down->GetStatus()) {
+ --y;
+ }
+ const float coef = modifier->GetStatus() ? modifier_scale : 1.0f;
+ return std::make_tuple(static_cast<float>(x) * coef * (y == 0 ? 1.0f : SQRT_HALF),
+ static_cast<float>(y) * coef * (x == 0 ? 1.0f : SQRT_HALF));
}
bool GetAnalogDirectionStatus(Input::AnalogDirection direction) const override {
diff --git a/src/input_common/gcadapter/gc_poller.cpp b/src/input_common/gcadapter/gc_poller.cpp
index 4d1052414..9670bdeb2 100644
--- a/src/input_common/gcadapter/gc_poller.cpp
+++ b/src/input_common/gcadapter/gc_poller.cpp
@@ -139,10 +139,10 @@ void GCButtonFactory::EndConfiguration() {
class GCAnalog final : public Input::AnalogDevice {
public:
- explicit GCAnalog(u32 port_, u32 axis_x_, u32 axis_y_, float deadzone_,
- const GCAdapter::Adapter* adapter, float range_)
- : port(port_), axis_x(axis_x_), axis_y(axis_y_), deadzone(deadzone_), gcadapter(adapter),
- range(range_) {}
+ explicit GCAnalog(u32 port_, u32 axis_x_, u32 axis_y_, bool invert_x_, bool invert_y_,
+ float deadzone_, float range_, const GCAdapter::Adapter* adapter)
+ : port(port_), axis_x(axis_x_), axis_y(axis_y_), invert_x(invert_x_), invert_y(invert_y_),
+ deadzone(deadzone_), range(range_), gcadapter(adapter) {}
float GetAxis(u32 axis) const {
if (gcadapter->DeviceConnected(port)) {
@@ -157,7 +157,12 @@ public:
std::pair<float, float> GetAnalog(u32 analog_axis_x, u32 analog_axis_y) const {
float x = GetAxis(analog_axis_x);
float y = GetAxis(analog_axis_y);
-
+ if (invert_x) {
+ x = -x;
+ }
+ if (invert_y) {
+ y = -y;
+ }
// Make sure the coordinates are in the unit circle,
// otherwise normalize it.
float r = x * x + y * y;
@@ -200,9 +205,11 @@ private:
const u32 port;
const u32 axis_x;
const u32 axis_y;
+ const bool invert_x;
+ const bool invert_y;
const float deadzone;
- const GCAdapter::Adapter* gcadapter;
const float range;
+ const GCAdapter::Adapter* gcadapter;
mutable std::mutex mutex;
};
@@ -223,8 +230,13 @@ std::unique_ptr<Input::AnalogDevice> GCAnalogFactory::Create(const Common::Param
const auto axis_y = static_cast<u32>(params.Get("axis_y", 1));
const auto deadzone = std::clamp(params.Get("deadzone", 0.0f), 0.0f, 1.0f);
const auto range = std::clamp(params.Get("range", 1.0f), 0.50f, 1.50f);
+ const std::string invert_x_value = params.Get("invert_x", "+");
+ const std::string invert_y_value = params.Get("invert_y", "+");
+ const bool invert_x = invert_x_value == "-";
+ const bool invert_y = invert_y_value == "-";
- return std::make_unique<GCAnalog>(port, axis_x, axis_y, deadzone, adapter.get(), range);
+ return std::make_unique<GCAnalog>(port, axis_x, axis_y, invert_x, invert_y, deadzone, range,
+ adapter.get());
}
void GCAnalogFactory::BeginConfiguration() {
@@ -282,6 +294,8 @@ Common::ParamPackage GCAnalogFactory::GetNextInput() {
params.Set("port", controller_number);
params.Set("axis_x", analog_x_axis);
params.Set("axis_y", analog_y_axis);
+ params.Set("invert_x", "+");
+ params.Set("invert_y", "+");
analog_x_axis = -1;
analog_y_axis = -1;
controller_number = -1;
diff --git a/src/input_common/mouse/mouse_poller.cpp b/src/input_common/mouse/mouse_poller.cpp
index 7445ad3ad..508eb0c7d 100644
--- a/src/input_common/mouse/mouse_poller.cpp
+++ b/src/input_common/mouse/mouse_poller.cpp
@@ -62,10 +62,10 @@ void MouseButtonFactory::EndConfiguration() {
class MouseAnalog final : public Input::AnalogDevice {
public:
- explicit MouseAnalog(u32 port_, u32 axis_x_, u32 axis_y_, float deadzone_, float range_,
- const MouseInput::Mouse* mouse_input_)
- : button(port_), axis_x(axis_x_), axis_y(axis_y_), deadzone(deadzone_), range(range_),
- mouse_input(mouse_input_) {}
+ explicit MouseAnalog(u32 port_, u32 axis_x_, u32 axis_y_, bool invert_x_, bool invert_y_,
+ float deadzone_, float range_, const MouseInput::Mouse* mouse_input_)
+ : button(port_), axis_x(axis_x_), axis_y(axis_y_), invert_x(invert_x_), invert_y(invert_y_),
+ deadzone(deadzone_), range(range_), mouse_input(mouse_input_) {}
float GetAxis(u32 axis) const {
std::lock_guard lock{mutex};
@@ -77,6 +77,12 @@ public:
std::pair<float, float> GetAnalog(u32 analog_axis_x, u32 analog_axis_y) const {
float x = GetAxis(analog_axis_x);
float y = GetAxis(analog_axis_y);
+ if (invert_x) {
+ x = -x;
+ }
+ if (invert_y) {
+ y = -y;
+ }
// Make sure the coordinates are in the unit circle,
// otherwise normalize it.
@@ -104,6 +110,8 @@ private:
const u32 button;
const u32 axis_x;
const u32 axis_y;
+ const bool invert_x;
+ const bool invert_y;
const float deadzone;
const float range;
const MouseInput::Mouse* mouse_input;
@@ -128,8 +136,13 @@ std::unique_ptr<Input::AnalogDevice> MouseAnalogFactory::Create(
const auto axis_y = static_cast<u32>(params.Get("axis_y", 1));
const auto deadzone = std::clamp(params.Get("deadzone", 0.0f), 0.0f, 1.0f);
const auto range = std::clamp(params.Get("range", 1.0f), 0.50f, 1.50f);
+ const std::string invert_x_value = params.Get("invert_x", "+");
+ const std::string invert_y_value = params.Get("invert_y", "+");
+ const bool invert_x = invert_x_value == "-";
+ const bool invert_y = invert_y_value == "-";
- return std::make_unique<MouseAnalog>(port, axis_x, axis_y, deadzone, range, mouse_input.get());
+ return std::make_unique<MouseAnalog>(port, axis_x, axis_y, invert_x, invert_y, deadzone, range,
+ mouse_input.get());
}
void MouseAnalogFactory::BeginConfiguration() {
@@ -153,6 +166,8 @@ Common::ParamPackage MouseAnalogFactory::GetNextInput() const {
params.Set("port", static_cast<u16>(pad.button));
params.Set("axis_x", 0);
params.Set("axis_y", 1);
+ params.Set("invert_x", "+");
+ params.Set("invert_y", "+");
return params;
}
}
diff --git a/src/input_common/sdl/sdl_impl.cpp b/src/input_common/sdl/sdl_impl.cpp
index d56b7587b..d32eb732a 100644
--- a/src/input_common/sdl/sdl_impl.cpp
+++ b/src/input_common/sdl/sdl_impl.cpp
@@ -352,13 +352,20 @@ private:
class SDLAnalog final : public Input::AnalogDevice {
public:
explicit SDLAnalog(std::shared_ptr<SDLJoystick> joystick_, int axis_x_, int axis_y_,
- float deadzone_, float range_)
- : joystick(std::move(joystick_)), axis_x(axis_x_), axis_y(axis_y_), deadzone(deadzone_),
- range(range_) {}
+ bool invert_x_, bool invert_y_, float deadzone_, float range_)
+ : joystick(std::move(joystick_)), axis_x(axis_x_), axis_y(axis_y_), invert_x(invert_x_),
+ invert_y(invert_y_), deadzone(deadzone_), range(range_) {}
std::tuple<float, float> GetStatus() const override {
- const auto [x, y] = joystick->GetAnalog(axis_x, axis_y, range);
+ auto [x, y] = joystick->GetAnalog(axis_x, axis_y, range);
const float r = std::sqrt((x * x) + (y * y));
+ if (invert_x) {
+ x = -x;
+ }
+ if (invert_y) {
+ y = -y;
+ }
+
if (r > deadzone) {
return std::make_tuple(x / r * (r - deadzone) / (1 - deadzone),
y / r * (r - deadzone) / (1 - deadzone));
@@ -386,6 +393,8 @@ private:
std::shared_ptr<SDLJoystick> joystick;
const int axis_x;
const int axis_y;
+ const bool invert_x;
+ const bool invert_y;
const float deadzone;
const float range;
};
@@ -572,12 +581,17 @@ public:
const int axis_y = params.Get("axis_y", 1);
const float deadzone = std::clamp(params.Get("deadzone", 0.0f), 0.0f, 1.0f);
const float range = std::clamp(params.Get("range", 1.0f), 0.50f, 1.50f);
+ const std::string invert_x_value = params.Get("invert_x", "+");
+ const std::string invert_y_value = params.Get("invert_y", "+");
+ const bool invert_x = invert_x_value == "-";
+ const bool invert_y = invert_y_value == "-";
auto joystick = state.GetSDLJoystickByGUID(guid, port);
// This is necessary so accessing GetAxis with axis_x and axis_y won't crash
joystick->SetAxis(axis_x, 0);
joystick->SetAxis(axis_y, 0);
- return std::make_unique<SDLAnalog>(joystick, axis_x, axis_y, deadzone, range);
+ return std::make_unique<SDLAnalog>(joystick, axis_x, axis_y, invert_x, invert_y, deadzone,
+ range);
}
private:
@@ -886,6 +900,8 @@ Common::ParamPackage BuildParamPackageForAnalog(int port, const std::string& gui
params.Set("guid", guid);
params.Set("axis_x", axis_x);
params.Set("axis_y", axis_y);
+ params.Set("invert_x", "+");
+ params.Set("invert_y", "+");
return params;
}
} // Anonymous namespace
diff --git a/src/input_common/udp/client.cpp b/src/input_common/udp/client.cpp
index 17a9225d7..412d57896 100644
--- a/src/input_common/udp/client.cpp
+++ b/src/input_common/udp/client.cpp
@@ -225,6 +225,11 @@ void Client::OnPortInfo([[maybe_unused]] Response::PortInfo data) {
}
void Client::OnPadData(Response::PadData data, std::size_t client) {
+ // Accept packets only for the correct pad
+ if (static_cast<u8>(clients[client].pad_index) != data.info.id) {
+ return;
+ }
+
LOG_TRACE(Input, "PadData packet received");
if (data.packet_counter == clients[client].packet_sequence) {
LOG_WARNING(
diff --git a/src/tests/CMakeLists.txt b/src/tests/CMakeLists.txt
index 47ef30aa9..d80b0b688 100644
--- a/src/tests/CMakeLists.txt
+++ b/src/tests/CMakeLists.txt
@@ -2,7 +2,6 @@ add_executable(tests
common/bit_field.cpp
common/bit_utils.cpp
common/fibers.cpp
- common/multi_level_queue.cpp
common/param_package.cpp
common/ring_buffer.cpp
core/arm/arm_test_common.cpp
diff --git a/src/tests/common/multi_level_queue.cpp b/src/tests/common/multi_level_queue.cpp
deleted file mode 100644
index cca7ec7da..000000000
--- a/src/tests/common/multi_level_queue.cpp
+++ /dev/null
@@ -1,55 +0,0 @@
-// Copyright 2019 Yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include <catch2/catch.hpp>
-#include <math.h>
-#include "common/common_types.h"
-#include "common/multi_level_queue.h"
-
-namespace Common {
-
-TEST_CASE("MultiLevelQueue", "[common]") {
- std::array<f32, 8> values = {0.0, 5.0, 1.0, 9.0, 8.0, 2.0, 6.0, 7.0};
- Common::MultiLevelQueue<f32, 64> mlq;
- REQUIRE(mlq.empty());
- mlq.add(values[2], 2);
- mlq.add(values[7], 7);
- mlq.add(values[3], 3);
- mlq.add(values[4], 4);
- mlq.add(values[0], 0);
- mlq.add(values[5], 5);
- mlq.add(values[6], 6);
- mlq.add(values[1], 1);
- u32 index = 0;
- bool all_set = true;
- for (auto& f : mlq) {
- all_set &= (f == values[index]);
- index++;
- }
- REQUIRE(all_set);
- REQUIRE(!mlq.empty());
- f32 v = 8.0;
- mlq.add(v, 2);
- v = -7.0;
- mlq.add(v, 2, false);
- REQUIRE(mlq.front(2) == -7.0);
- mlq.yield(2);
- REQUIRE(mlq.front(2) == values[2]);
- REQUIRE(mlq.back(2) == -7.0);
- REQUIRE(mlq.empty(8));
- v = 10.0;
- mlq.add(v, 8);
- mlq.adjust(v, 8, 9);
- REQUIRE(mlq.front(9) == v);
- REQUIRE(mlq.empty(8));
- REQUIRE(!mlq.empty(9));
- mlq.adjust(values[0], 0, 9);
- REQUIRE(mlq.highest_priority_set() == 1);
- REQUIRE(mlq.lowest_priority_set() == 9);
- mlq.remove(values[1], 1);
- REQUIRE(mlq.highest_priority_set() == 2);
- REQUIRE(mlq.empty(1));
-}
-
-} // namespace Common
diff --git a/src/video_core/CMakeLists.txt b/src/video_core/CMakeLists.txt
index abcee2a1c..e050f9aed 100644
--- a/src/video_core/CMakeLists.txt
+++ b/src/video_core/CMakeLists.txt
@@ -25,6 +25,7 @@ add_library(video_core STATIC
command_classes/vic.h
compatible_formats.cpp
compatible_formats.h
+ delayed_destruction_ring.h
dirty_flags.cpp
dirty_flags.h
dma_pusher.cpp
@@ -47,6 +48,7 @@ add_library(video_core STATIC
engines/shader_bytecode.h
engines/shader_header.h
engines/shader_type.h
+ framebuffer_config.h
macro/macro.cpp
macro/macro.h
macro/macro_hle.cpp
@@ -58,10 +60,6 @@ add_library(video_core STATIC
fence_manager.h
gpu.cpp
gpu.h
- gpu_asynch.cpp
- gpu_asynch.h
- gpu_synch.cpp
- gpu_synch.h
gpu_thread.cpp
gpu_thread.h
guest_driver.cpp
@@ -84,14 +82,10 @@ add_library(video_core STATIC
renderer_opengl/gl_device.h
renderer_opengl/gl_fence_manager.cpp
renderer_opengl/gl_fence_manager.h
- renderer_opengl/gl_framebuffer_cache.cpp
- renderer_opengl/gl_framebuffer_cache.h
renderer_opengl/gl_rasterizer.cpp
renderer_opengl/gl_rasterizer.h
renderer_opengl/gl_resource_manager.cpp
renderer_opengl/gl_resource_manager.h
- renderer_opengl/gl_sampler_cache.cpp
- renderer_opengl/gl_sampler_cache.h
renderer_opengl/gl_shader_cache.cpp
renderer_opengl/gl_shader_cache.h
renderer_opengl/gl_shader_decompiler.cpp
@@ -113,10 +107,68 @@ add_library(video_core STATIC
renderer_opengl/maxwell_to_gl.h
renderer_opengl/renderer_opengl.cpp
renderer_opengl/renderer_opengl.h
- renderer_opengl/utils.cpp
- renderer_opengl/utils.h
- sampler_cache.cpp
- sampler_cache.h
+ renderer_opengl/util_shaders.cpp
+ renderer_opengl/util_shaders.h
+ renderer_vulkan/blit_image.cpp
+ renderer_vulkan/blit_image.h
+ renderer_vulkan/fixed_pipeline_state.cpp
+ renderer_vulkan/fixed_pipeline_state.h
+ renderer_vulkan/maxwell_to_vk.cpp
+ renderer_vulkan/maxwell_to_vk.h
+ renderer_vulkan/nsight_aftermath_tracker.cpp
+ renderer_vulkan/nsight_aftermath_tracker.h
+ renderer_vulkan/renderer_vulkan.h
+ renderer_vulkan/renderer_vulkan.cpp
+ renderer_vulkan/vk_blit_screen.cpp
+ renderer_vulkan/vk_blit_screen.h
+ renderer_vulkan/vk_buffer_cache.cpp
+ renderer_vulkan/vk_buffer_cache.h
+ renderer_vulkan/vk_command_pool.cpp
+ renderer_vulkan/vk_command_pool.h
+ renderer_vulkan/vk_compute_pass.cpp
+ renderer_vulkan/vk_compute_pass.h
+ renderer_vulkan/vk_compute_pipeline.cpp
+ renderer_vulkan/vk_compute_pipeline.h
+ renderer_vulkan/vk_descriptor_pool.cpp
+ renderer_vulkan/vk_descriptor_pool.h
+ renderer_vulkan/vk_device.cpp
+ renderer_vulkan/vk_device.h
+ renderer_vulkan/vk_fence_manager.cpp
+ renderer_vulkan/vk_fence_manager.h
+ renderer_vulkan/vk_graphics_pipeline.cpp
+ renderer_vulkan/vk_graphics_pipeline.h
+ renderer_vulkan/vk_master_semaphore.cpp
+ renderer_vulkan/vk_master_semaphore.h
+ renderer_vulkan/vk_memory_manager.cpp
+ renderer_vulkan/vk_memory_manager.h
+ renderer_vulkan/vk_pipeline_cache.cpp
+ renderer_vulkan/vk_pipeline_cache.h
+ renderer_vulkan/vk_query_cache.cpp
+ renderer_vulkan/vk_query_cache.h
+ renderer_vulkan/vk_rasterizer.cpp
+ renderer_vulkan/vk_rasterizer.h
+ renderer_vulkan/vk_resource_pool.cpp
+ renderer_vulkan/vk_resource_pool.h
+ renderer_vulkan/vk_scheduler.cpp
+ renderer_vulkan/vk_scheduler.h
+ renderer_vulkan/vk_shader_decompiler.cpp
+ renderer_vulkan/vk_shader_decompiler.h
+ renderer_vulkan/vk_shader_util.cpp
+ renderer_vulkan/vk_shader_util.h
+ renderer_vulkan/vk_staging_buffer_pool.cpp
+ renderer_vulkan/vk_staging_buffer_pool.h
+ renderer_vulkan/vk_state_tracker.cpp
+ renderer_vulkan/vk_state_tracker.h
+ renderer_vulkan/vk_stream_buffer.cpp
+ renderer_vulkan/vk_stream_buffer.h
+ renderer_vulkan/vk_swapchain.cpp
+ renderer_vulkan/vk_swapchain.h
+ renderer_vulkan/vk_texture_cache.cpp
+ renderer_vulkan/vk_texture_cache.h
+ renderer_vulkan/vk_update_descriptor.cpp
+ renderer_vulkan/vk_update_descriptor.h
+ renderer_vulkan/wrapper.cpp
+ renderer_vulkan/wrapper.h
shader_cache.h
shader_notify.cpp
shader_notify.h
@@ -173,19 +225,32 @@ add_library(video_core STATIC
shader/transform_feedback.h
surface.cpp
surface.h
+ texture_cache/accelerated_swizzle.cpp
+ texture_cache/accelerated_swizzle.h
+ texture_cache/decode_bc4.cpp
+ texture_cache/decode_bc4.h
+ texture_cache/descriptor_table.h
+ texture_cache/formatter.cpp
+ texture_cache/formatter.h
texture_cache/format_lookup_table.cpp
texture_cache/format_lookup_table.h
- texture_cache/surface_base.cpp
- texture_cache/surface_base.h
- texture_cache/surface_params.cpp
- texture_cache/surface_params.h
- texture_cache/surface_view.cpp
- texture_cache/surface_view.h
+ texture_cache/image_base.cpp
+ texture_cache/image_base.h
+ texture_cache/image_info.cpp
+ texture_cache/image_info.h
+ texture_cache/image_view_base.cpp
+ texture_cache/image_view_base.h
+ texture_cache/image_view_info.cpp
+ texture_cache/image_view_info.h
+ texture_cache/render_targets.h
+ texture_cache/samples_helper.h
+ texture_cache/slot_vector.h
texture_cache/texture_cache.h
+ texture_cache/types.h
+ texture_cache/util.cpp
+ texture_cache/util.h
textures/astc.cpp
textures/astc.h
- textures/convert.cpp
- textures/convert.h
textures/decoders.cpp
textures/decoders.h
textures/texture.cpp
@@ -194,75 +259,6 @@ add_library(video_core STATIC
video_core.h
)
-if (ENABLE_VULKAN)
- target_sources(video_core PRIVATE
- renderer_vulkan/fixed_pipeline_state.cpp
- renderer_vulkan/fixed_pipeline_state.h
- renderer_vulkan/maxwell_to_vk.cpp
- renderer_vulkan/maxwell_to_vk.h
- renderer_vulkan/nsight_aftermath_tracker.cpp
- renderer_vulkan/nsight_aftermath_tracker.h
- renderer_vulkan/renderer_vulkan.h
- renderer_vulkan/renderer_vulkan.cpp
- renderer_vulkan/vk_blit_screen.cpp
- renderer_vulkan/vk_blit_screen.h
- renderer_vulkan/vk_buffer_cache.cpp
- renderer_vulkan/vk_buffer_cache.h
- renderer_vulkan/vk_command_pool.cpp
- renderer_vulkan/vk_command_pool.h
- renderer_vulkan/vk_compute_pass.cpp
- renderer_vulkan/vk_compute_pass.h
- renderer_vulkan/vk_compute_pipeline.cpp
- renderer_vulkan/vk_compute_pipeline.h
- renderer_vulkan/vk_descriptor_pool.cpp
- renderer_vulkan/vk_descriptor_pool.h
- renderer_vulkan/vk_device.cpp
- renderer_vulkan/vk_device.h
- renderer_vulkan/vk_fence_manager.cpp
- renderer_vulkan/vk_fence_manager.h
- renderer_vulkan/vk_graphics_pipeline.cpp
- renderer_vulkan/vk_graphics_pipeline.h
- renderer_vulkan/vk_image.cpp
- renderer_vulkan/vk_image.h
- renderer_vulkan/vk_master_semaphore.cpp
- renderer_vulkan/vk_master_semaphore.h
- renderer_vulkan/vk_memory_manager.cpp
- renderer_vulkan/vk_memory_manager.h
- renderer_vulkan/vk_pipeline_cache.cpp
- renderer_vulkan/vk_pipeline_cache.h
- renderer_vulkan/vk_query_cache.cpp
- renderer_vulkan/vk_query_cache.h
- renderer_vulkan/vk_rasterizer.cpp
- renderer_vulkan/vk_rasterizer.h
- renderer_vulkan/vk_renderpass_cache.cpp
- renderer_vulkan/vk_renderpass_cache.h
- renderer_vulkan/vk_resource_pool.cpp
- renderer_vulkan/vk_resource_pool.h
- renderer_vulkan/vk_sampler_cache.cpp
- renderer_vulkan/vk_sampler_cache.h
- renderer_vulkan/vk_scheduler.cpp
- renderer_vulkan/vk_scheduler.h
- renderer_vulkan/vk_shader_decompiler.cpp
- renderer_vulkan/vk_shader_decompiler.h
- renderer_vulkan/vk_shader_util.cpp
- renderer_vulkan/vk_shader_util.h
- renderer_vulkan/vk_staging_buffer_pool.cpp
- renderer_vulkan/vk_staging_buffer_pool.h
- renderer_vulkan/vk_state_tracker.cpp
- renderer_vulkan/vk_state_tracker.h
- renderer_vulkan/vk_stream_buffer.cpp
- renderer_vulkan/vk_stream_buffer.h
- renderer_vulkan/vk_swapchain.cpp
- renderer_vulkan/vk_swapchain.h
- renderer_vulkan/vk_texture_cache.cpp
- renderer_vulkan/vk_texture_cache.h
- renderer_vulkan/vk_update_descriptor.cpp
- renderer_vulkan/vk_update_descriptor.h
- renderer_vulkan/wrapper.cpp
- renderer_vulkan/wrapper.h
- )
-endif()
-
create_target_directory_groups(video_core)
target_link_libraries(video_core PUBLIC common core)
@@ -278,12 +274,8 @@ endif()
add_dependencies(video_core host_shaders)
target_include_directories(video_core PRIVATE ${HOST_SHADERS_INCLUDE})
-
-if (ENABLE_VULKAN)
- target_include_directories(video_core PRIVATE sirit ../../externals/Vulkan-Headers/include)
- target_compile_definitions(video_core PRIVATE HAS_VULKAN)
- target_link_libraries(video_core PRIVATE sirit)
-endif()
+target_include_directories(video_core PRIVATE sirit ../../externals/Vulkan-Headers/include)
+target_link_libraries(video_core PRIVATE sirit)
if (ENABLE_NSIGHT_AFTERMATH)
if (NOT DEFINED ENV{NSIGHT_AFTERMATH_SDK})
@@ -297,13 +289,21 @@ if (ENABLE_NSIGHT_AFTERMATH)
endif()
if (MSVC)
- target_compile_options(video_core PRIVATE /we4267)
+ target_compile_options(video_core PRIVATE
+ /we4267 # 'var' : conversion from 'size_t' to 'type', possible loss of data
+ /we4456 # Declaration of 'identifier' hides previous local declaration
+ /we4457 # Declaration of 'identifier' hides function parameter
+ /we4458 # Declaration of 'identifier' hides class member
+ /we4459 # Declaration of 'identifier' hides global declaration
+ /we4715 # 'function' : not all control paths return a value
+ )
else()
target_compile_options(video_core PRIVATE
-Werror=conversion
-Wno-error=sign-conversion
-Werror=pessimizing-move
-Werror=redundant-move
+ -Werror=shadow
-Werror=switch
-Werror=type-limits
-Werror=unused-variable
diff --git a/src/video_core/buffer_cache/buffer_block.h b/src/video_core/buffer_cache/buffer_block.h
index e64170e66..e9306194a 100644
--- a/src/video_core/buffer_cache/buffer_block.h
+++ b/src/video_core/buffer_cache/buffer_block.h
@@ -4,34 +4,29 @@
#pragma once
-#include <unordered_set>
-#include <utility>
-
-#include "common/alignment.h"
#include "common/common_types.h"
-#include "video_core/gpu.h"
namespace VideoCommon {
class BufferBlock {
public:
- bool Overlaps(VAddr start, VAddr end) const {
+ [[nodiscard]] bool Overlaps(VAddr start, VAddr end) const {
return (cpu_addr < end) && (cpu_addr_end > start);
}
- bool IsInside(VAddr other_start, VAddr other_end) const {
+ [[nodiscard]] bool IsInside(VAddr other_start, VAddr other_end) const {
return cpu_addr <= other_start && other_end <= cpu_addr_end;
}
- std::size_t Offset(VAddr in_addr) const {
+ [[nodiscard]] std::size_t Offset(VAddr in_addr) const {
return static_cast<std::size_t>(in_addr - cpu_addr);
}
- VAddr CpuAddr() const {
+ [[nodiscard]] VAddr CpuAddr() const {
return cpu_addr;
}
- VAddr CpuAddrEnd() const {
+ [[nodiscard]] VAddr CpuAddrEnd() const {
return cpu_addr_end;
}
@@ -40,11 +35,11 @@ public:
cpu_addr_end = new_addr + size;
}
- std::size_t Size() const {
+ [[nodiscard]] std::size_t Size() const {
return size;
}
- u64 Epoch() const {
+ [[nodiscard]] u64 Epoch() const {
return epoch;
}
diff --git a/src/video_core/buffer_cache/buffer_cache.h b/src/video_core/buffer_cache/buffer_cache.h
index e7edd733f..83b9ee871 100644
--- a/src/video_core/buffer_cache/buffer_cache.h
+++ b/src/video_core/buffer_cache/buffer_cache.h
@@ -118,20 +118,17 @@ public:
/// Prepares the buffer cache for data uploading
/// @param max_size Maximum number of bytes that will be uploaded
/// @return True when a stream buffer invalidation was required, false otherwise
- bool Map(std::size_t max_size) {
+ void Map(std::size_t max_size) {
std::lock_guard lock{mutex};
- bool invalidated;
- std::tie(buffer_ptr, buffer_offset_base, invalidated) = stream_buffer->Map(max_size, 4);
+ std::tie(buffer_ptr, buffer_offset_base) = stream_buffer.Map(max_size, 4);
buffer_offset = buffer_offset_base;
-
- return invalidated;
}
/// Finishes the upload stream
void Unmap() {
std::lock_guard lock{mutex};
- stream_buffer->Unmap(buffer_offset - buffer_offset_base);
+ stream_buffer.Unmap(buffer_offset - buffer_offset_base);
}
/// Function called at the end of each frame, inteded for deferred operations
@@ -261,9 +258,9 @@ public:
protected:
explicit BufferCache(VideoCore::RasterizerInterface& rasterizer_,
Tegra::MemoryManager& gpu_memory_, Core::Memory::Memory& cpu_memory_,
- std::unique_ptr<StreamBuffer> stream_buffer_)
+ StreamBuffer& stream_buffer_)
: rasterizer{rasterizer_}, gpu_memory{gpu_memory_}, cpu_memory{cpu_memory_},
- stream_buffer{std::move(stream_buffer_)}, stream_buffer_handle{stream_buffer->Handle()} {}
+ stream_buffer{stream_buffer_} {}
~BufferCache() = default;
@@ -441,7 +438,7 @@ private:
buffer_ptr += size;
buffer_offset += size;
- return BufferInfo{stream_buffer->Handle(), uploaded_offset, stream_buffer->Address()};
+ return BufferInfo{stream_buffer.Handle(), uploaded_offset, stream_buffer.Address()};
}
void AlignBuffer(std::size_t alignment) {
@@ -545,7 +542,7 @@ private:
bool IsRegionWritten(VAddr start, VAddr end) const {
const u64 page_end = end >> WRITE_PAGE_BIT;
for (u64 page_start = start >> WRITE_PAGE_BIT; page_start <= page_end; ++page_start) {
- if (written_pages.count(page_start) > 0) {
+ if (written_pages.contains(page_start)) {
return true;
}
}
@@ -567,9 +564,7 @@ private:
VideoCore::RasterizerInterface& rasterizer;
Tegra::MemoryManager& gpu_memory;
Core::Memory::Memory& cpu_memory;
-
- std::unique_ptr<StreamBuffer> stream_buffer;
- BufferType stream_buffer_handle;
+ StreamBuffer& stream_buffer;
u8* buffer_ptr = nullptr;
u64 buffer_offset = 0;
diff --git a/src/video_core/buffer_cache/map_interval.h b/src/video_core/buffer_cache/map_interval.h
index fe0bcd1d8..ef974b08a 100644
--- a/src/video_core/buffer_cache/map_interval.h
+++ b/src/video_core/buffer_cache/map_interval.h
@@ -84,9 +84,10 @@ private:
void FillFreeList(Chunk& chunk);
std::vector<MapInterval*> free_list;
- std::unique_ptr<Chunk>* new_chunk = &first_chunk.next;
Chunk first_chunk;
+
+ std::unique_ptr<Chunk>* new_chunk = &first_chunk.next;
};
} // namespace VideoCommon
diff --git a/src/video_core/cdma_pusher.cpp b/src/video_core/cdma_pusher.cpp
index b60f86260..e3e7432f7 100644
--- a/src/video_core/cdma_pusher.cpp
+++ b/src/video_core/cdma_pusher.cpp
@@ -29,8 +29,8 @@
#include "video_core/memory_manager.h"
namespace Tegra {
-CDmaPusher::CDmaPusher(GPU& gpu)
- : gpu(gpu), nvdec_processor(std::make_shared<Nvdec>(gpu)),
+CDmaPusher::CDmaPusher(GPU& gpu_)
+ : gpu{gpu_}, nvdec_processor(std::make_shared<Nvdec>(gpu)),
vic_processor(std::make_unique<Vic>(gpu, nvdec_processor)),
host1x_processor(std::make_unique<Host1x>(gpu)),
nvdec_sync(std::make_unique<SyncptIncrManager>(gpu)),
@@ -100,11 +100,11 @@ void CDmaPusher::Step() {
}
}
-void CDmaPusher::ExecuteCommand(u32 offset, u32 data) {
+void CDmaPusher::ExecuteCommand(u32 state_offset, u32 data) {
switch (current_class) {
case ChClassId::NvDec:
- ThiStateWrite(nvdec_thi_state, offset, {data});
- switch (static_cast<ThiMethod>(offset)) {
+ ThiStateWrite(nvdec_thi_state, state_offset, {data});
+ switch (static_cast<ThiMethod>(state_offset)) {
case ThiMethod::IncSyncpt: {
LOG_DEBUG(Service_NVDRV, "NVDEC Class IncSyncpt Method");
const auto syncpoint_id = static_cast<u32>(data & 0xFF);
@@ -120,16 +120,16 @@ void CDmaPusher::ExecuteCommand(u32 offset, u32 data) {
case ThiMethod::SetMethod1:
LOG_DEBUG(Service_NVDRV, "NVDEC method 0x{:X}",
static_cast<u32>(nvdec_thi_state.method_0));
- nvdec_processor->ProcessMethod(
- static_cast<Tegra::Nvdec::Method>(nvdec_thi_state.method_0), {data});
+ nvdec_processor->ProcessMethod(static_cast<Nvdec::Method>(nvdec_thi_state.method_0),
+ {data});
break;
default:
break;
}
break;
case ChClassId::GraphicsVic:
- ThiStateWrite(vic_thi_state, static_cast<u32>(offset), {data});
- switch (static_cast<ThiMethod>(offset)) {
+ ThiStateWrite(vic_thi_state, static_cast<u32>(state_offset), {data});
+ switch (static_cast<ThiMethod>(state_offset)) {
case ThiMethod::IncSyncpt: {
LOG_DEBUG(Service_NVDRV, "VIC Class IncSyncpt Method");
const auto syncpoint_id = static_cast<u32>(data & 0xFF);
@@ -145,8 +145,7 @@ void CDmaPusher::ExecuteCommand(u32 offset, u32 data) {
case ThiMethod::SetMethod1:
LOG_DEBUG(Service_NVDRV, "VIC method 0x{:X}, Args=({})",
static_cast<u32>(vic_thi_state.method_0), data);
- vic_processor->ProcessMethod(static_cast<Tegra::Vic::Method>(vic_thi_state.method_0),
- {data});
+ vic_processor->ProcessMethod(static_cast<Vic::Method>(vic_thi_state.method_0), {data});
break;
default:
break;
@@ -155,7 +154,7 @@ void CDmaPusher::ExecuteCommand(u32 offset, u32 data) {
case ChClassId::Host1x:
// This device is mainly for syncpoint synchronization
LOG_DEBUG(Service_NVDRV, "Host1X Class Method");
- host1x_processor->ProcessMethod(static_cast<Tegra::Host1x::Method>(offset), {data});
+ host1x_processor->ProcessMethod(static_cast<Host1x::Method>(state_offset), {data});
break;
default:
UNIMPLEMENTED_MSG("Current class not implemented {:X}", static_cast<u32>(current_class));
@@ -163,9 +162,10 @@ void CDmaPusher::ExecuteCommand(u32 offset, u32 data) {
}
}
-void CDmaPusher::ThiStateWrite(ThiRegisters& state, u32 offset, const std::vector<u32>& arguments) {
- u8* const state_offset = reinterpret_cast<u8*>(&state) + sizeof(u32) * offset;
- std::memcpy(state_offset, arguments.data(), sizeof(u32) * arguments.size());
+void CDmaPusher::ThiStateWrite(ThiRegisters& state, u32 state_offset,
+ const std::vector<u32>& arguments) {
+ u8* const state_offset_ptr = reinterpret_cast<u8*>(&state) + sizeof(u32) * state_offset;
+ std::memcpy(state_offset_ptr, arguments.data(), sizeof(u32) * arguments.size());
}
} // namespace Tegra
diff --git a/src/video_core/cdma_pusher.h b/src/video_core/cdma_pusher.h
index 982f309c5..0db1cd646 100644
--- a/src/video_core/cdma_pusher.h
+++ b/src/video_core/cdma_pusher.h
@@ -68,8 +68,8 @@ struct ChCommand {
std::vector<u32> arguments;
};
-using ChCommandHeaderList = std::vector<Tegra::ChCommandHeader>;
-using ChCommandList = std::vector<Tegra::ChCommand>;
+using ChCommandHeaderList = std::vector<ChCommandHeader>;
+using ChCommandList = std::vector<ChCommand>;
struct ThiRegisters {
u32_le increment_syncpt{};
@@ -96,7 +96,7 @@ enum class ThiMethod : u32 {
class CDmaPusher {
public:
- explicit CDmaPusher(GPU& gpu);
+ explicit CDmaPusher(GPU& gpu_);
~CDmaPusher();
/// Push NVDEC command buffer entries into queue
@@ -109,17 +109,17 @@ public:
void Step();
/// Invoke command class devices to execute the command based on the current state
- void ExecuteCommand(u32 offset, u32 data);
+ void ExecuteCommand(u32 state_offset, u32 data);
private:
/// Write arguments value to the ThiRegisters member at the specified offset
- void ThiStateWrite(ThiRegisters& state, u32 offset, const std::vector<u32>& arguments);
+ void ThiStateWrite(ThiRegisters& state, u32 state_offset, const std::vector<u32>& arguments);
GPU& gpu;
- std::shared_ptr<Tegra::Nvdec> nvdec_processor;
- std::unique_ptr<Tegra::Vic> vic_processor;
- std::unique_ptr<Tegra::Host1x> host1x_processor;
+ std::shared_ptr<Nvdec> nvdec_processor;
+ std::unique_ptr<Vic> vic_processor;
+ std::unique_ptr<Host1x> host1x_processor;
std::unique_ptr<SyncptIncrManager> nvdec_sync;
std::unique_ptr<SyncptIncrManager> vic_sync;
ChClassId current_class{};
diff --git a/src/video_core/command_classes/codecs/codec.cpp b/src/video_core/command_classes/codecs/codec.cpp
index f547f5bd4..39bc923a5 100644
--- a/src/video_core/command_classes/codecs/codec.cpp
+++ b/src/video_core/command_classes/codecs/codec.cpp
@@ -44,7 +44,7 @@ Codec::~Codec() {
}
void Codec::SetTargetCodec(NvdecCommon::VideoCodec codec) {
- LOG_INFO(Service_NVDRV, "NVDEC video codec initialized to {}", static_cast<u32>(codec));
+ LOG_INFO(Service_NVDRV, "NVDEC video codec initialized to {}", codec);
current_codec = codec;
}
@@ -62,7 +62,7 @@ void Codec::Decode() {
} else if (current_codec == NvdecCommon::VideoCodec::Vp9) {
av_codec = avcodec_find_decoder(AV_CODEC_ID_VP9);
} else {
- LOG_ERROR(Service_NVDRV, "Unknown video codec {}", static_cast<u32>(current_codec));
+ LOG_ERROR(Service_NVDRV, "Unknown video codec {}", current_codec);
return;
}
diff --git a/src/video_core/command_classes/codecs/vp9.cpp b/src/video_core/command_classes/codecs/vp9.cpp
index 7d8d6ee3c..59e586695 100644
--- a/src/video_core/command_classes/codecs/vp9.cpp
+++ b/src/video_core/command_classes/codecs/vp9.cpp
@@ -233,7 +233,7 @@ constexpr std::array<s32, 254> map_lut{
}
} // Anonymous namespace
-VP9::VP9(GPU& gpu) : gpu(gpu) {}
+VP9::VP9(GPU& gpu_) : gpu{gpu_} {}
VP9::~VP9() = default;
@@ -374,43 +374,43 @@ void VP9::InsertEntropy(u64 offset, Vp9EntropyProbs& dst) {
}
Vp9FrameContainer VP9::GetCurrentFrame(const NvdecCommon::NvdecRegisters& state) {
- Vp9FrameContainer frame{};
+ Vp9FrameContainer current_frame{};
{
gpu.SyncGuestHost();
- frame.info = GetVp9PictureInfo(state);
- frame.bit_stream.resize(frame.info.bitstream_size);
- gpu.MemoryManager().ReadBlock(state.frame_bitstream_offset, frame.bit_stream.data(),
- frame.info.bitstream_size);
+ current_frame.info = GetVp9PictureInfo(state);
+ current_frame.bit_stream.resize(current_frame.info.bitstream_size);
+ gpu.MemoryManager().ReadBlock(state.frame_bitstream_offset, current_frame.bit_stream.data(),
+ current_frame.info.bitstream_size);
}
// Buffer two frames, saving the last show frame info
if (!next_next_frame.bit_stream.empty()) {
Vp9FrameContainer temp{
- .info = frame.info,
- .bit_stream = std::move(frame.bit_stream),
+ .info = current_frame.info,
+ .bit_stream = std::move(current_frame.bit_stream),
};
- next_next_frame.info.show_frame = frame.info.last_frame_shown;
- frame.info = next_next_frame.info;
- frame.bit_stream = std::move(next_next_frame.bit_stream);
+ next_next_frame.info.show_frame = current_frame.info.last_frame_shown;
+ current_frame.info = next_next_frame.info;
+ current_frame.bit_stream = std::move(next_next_frame.bit_stream);
next_next_frame = std::move(temp);
if (!next_frame.bit_stream.empty()) {
Vp9FrameContainer temp2{
- .info = frame.info,
- .bit_stream = std::move(frame.bit_stream),
+ .info = current_frame.info,
+ .bit_stream = std::move(current_frame.bit_stream),
};
- next_frame.info.show_frame = frame.info.last_frame_shown;
- frame.info = next_frame.info;
- frame.bit_stream = std::move(next_frame.bit_stream);
+ next_frame.info.show_frame = current_frame.info.last_frame_shown;
+ current_frame.info = next_frame.info;
+ current_frame.bit_stream = std::move(next_frame.bit_stream);
next_frame = std::move(temp2);
} else {
- next_frame.info = frame.info;
- next_frame.bit_stream = std::move(frame.bit_stream);
+ next_frame.info = current_frame.info;
+ next_frame.bit_stream = std::move(current_frame.bit_stream);
}
} else {
- next_next_frame.info = frame.info;
- next_next_frame.bit_stream = std::move(frame.bit_stream);
+ next_next_frame.info = current_frame.info;
+ next_next_frame.bit_stream = std::move(current_frame.bit_stream);
}
- return frame;
+ return current_frame;
}
std::vector<u8> VP9::ComposeCompressedHeader() {
diff --git a/src/video_core/command_classes/codecs/vp9.h b/src/video_core/command_classes/codecs/vp9.h
index 9ebbbf59e..8396c8105 100644
--- a/src/video_core/command_classes/codecs/vp9.h
+++ b/src/video_core/command_classes/codecs/vp9.h
@@ -108,7 +108,7 @@ private:
class VP9 {
public:
- explicit VP9(GPU& gpu);
+ explicit VP9(GPU& gpu_);
~VP9();
VP9(const VP9&) = delete;
diff --git a/src/video_core/command_classes/vic.cpp b/src/video_core/command_classes/vic.cpp
index 6cfc193fa..aa8c9f9de 100644
--- a/src/video_core/command_classes/vic.cpp
+++ b/src/video_core/command_classes/vic.cpp
@@ -9,7 +9,7 @@
#include "video_core/engines/maxwell_3d.h"
#include "video_core/gpu.h"
#include "video_core/memory_manager.h"
-#include "video_core/texture_cache/surface_params.h"
+#include "video_core/textures/decoders.h"
extern "C" {
#include <libswscale/swscale.h>
@@ -27,7 +27,7 @@ void Vic::VicStateWrite(u32 offset, u32 arguments) {
}
void Vic::ProcessMethod(Method method, const std::vector<u32>& arguments) {
- LOG_DEBUG(HW_GPU, "Vic method 0x{:X}", static_cast<u32>(method));
+ LOG_DEBUG(HW_GPU, "Vic method 0x{:X}", method);
VicStateWrite(static_cast<u32>(method), arguments[0]);
const u64 arg = static_cast<u64>(arguments[0]) << 8;
switch (method) {
@@ -105,9 +105,9 @@ void Vic::Execute() {
const auto size = Tegra::Texture::CalculateSize(true, 4, frame->width, frame->height, 1,
block_height, 0);
std::vector<u8> swizzled_data(size);
- Tegra::Texture::CopySwizzledData(frame->width, frame->height, 1, 4, 4,
- swizzled_data.data(), converted_frame_buffer.get(),
- false, block_height, 0, 1);
+ Tegra::Texture::SwizzleSubrect(frame->width, frame->height, frame->width * 4,
+ frame->width, 4, swizzled_data.data(),
+ converted_frame_buffer.get(), block_height, 0, 0);
gpu.MemoryManager().WriteBlock(output_surface_luma_address, swizzled_data.data(), size);
gpu.Maxwell3D().OnMemoryWrite();
diff --git a/src/video_core/compatible_formats.cpp b/src/video_core/compatible_formats.cpp
index b06c32c84..1619d8664 100644
--- a/src/video_core/compatible_formats.cpp
+++ b/src/video_core/compatible_formats.cpp
@@ -3,9 +3,9 @@
// Refer to the license.txt file included.
#include <array>
-#include <bitset>
#include <cstddef>
+#include "common/common_types.h"
#include "video_core/compatible_formats.h"
#include "video_core/surface.h"
@@ -13,23 +13,25 @@ namespace VideoCore::Surface {
namespace {
+using Table = std::array<std::array<u64, 2>, MaxPixelFormat>;
+
// Compatibility table taken from Table 3.X.2 in:
// https://www.khronos.org/registry/OpenGL/extensions/ARB/ARB_texture_view.txt
-constexpr std::array VIEW_CLASS_128_BITS = {
+constexpr std::array VIEW_CLASS_128_BITS{
PixelFormat::R32G32B32A32_FLOAT,
PixelFormat::R32G32B32A32_UINT,
PixelFormat::R32G32B32A32_SINT,
};
-constexpr std::array VIEW_CLASS_96_BITS = {
+constexpr std::array VIEW_CLASS_96_BITS{
PixelFormat::R32G32B32_FLOAT,
};
// Missing formats:
// PixelFormat::RGB32UI,
// PixelFormat::RGB32I,
-constexpr std::array VIEW_CLASS_64_BITS = {
+constexpr std::array VIEW_CLASS_64_BITS{
PixelFormat::R32G32_FLOAT, PixelFormat::R32G32_UINT,
PixelFormat::R32G32_SINT, PixelFormat::R16G16B16A16_FLOAT,
PixelFormat::R16G16B16A16_UNORM, PixelFormat::R16G16B16A16_SNORM,
@@ -38,7 +40,7 @@ constexpr std::array VIEW_CLASS_64_BITS = {
// TODO: How should we handle 48 bits?
-constexpr std::array VIEW_CLASS_32_BITS = {
+constexpr std::array VIEW_CLASS_32_BITS{
PixelFormat::R16G16_FLOAT, PixelFormat::B10G11R11_FLOAT, PixelFormat::R32_FLOAT,
PixelFormat::A2B10G10R10_UNORM, PixelFormat::R16G16_UINT, PixelFormat::R32_UINT,
PixelFormat::R16G16_SINT, PixelFormat::R32_SINT, PixelFormat::A8B8G8R8_UNORM,
@@ -50,43 +52,105 @@ constexpr std::array VIEW_CLASS_32_BITS = {
// TODO: How should we handle 24 bits?
-constexpr std::array VIEW_CLASS_16_BITS = {
+constexpr std::array VIEW_CLASS_16_BITS{
PixelFormat::R16_FLOAT, PixelFormat::R8G8_UINT, PixelFormat::R16_UINT,
PixelFormat::R16_SINT, PixelFormat::R8G8_UNORM, PixelFormat::R16_UNORM,
PixelFormat::R8G8_SNORM, PixelFormat::R16_SNORM, PixelFormat::R8G8_SINT,
};
-constexpr std::array VIEW_CLASS_8_BITS = {
+constexpr std::array VIEW_CLASS_8_BITS{
PixelFormat::R8_UINT,
PixelFormat::R8_UNORM,
PixelFormat::R8_SINT,
PixelFormat::R8_SNORM,
};
-constexpr std::array VIEW_CLASS_RGTC1_RED = {
+constexpr std::array VIEW_CLASS_RGTC1_RED{
PixelFormat::BC4_UNORM,
PixelFormat::BC4_SNORM,
};
-constexpr std::array VIEW_CLASS_RGTC2_RG = {
+constexpr std::array VIEW_CLASS_RGTC2_RG{
PixelFormat::BC5_UNORM,
PixelFormat::BC5_SNORM,
};
-constexpr std::array VIEW_CLASS_BPTC_UNORM = {
+constexpr std::array VIEW_CLASS_BPTC_UNORM{
PixelFormat::BC7_UNORM,
PixelFormat::BC7_SRGB,
};
-constexpr std::array VIEW_CLASS_BPTC_FLOAT = {
+constexpr std::array VIEW_CLASS_BPTC_FLOAT{
PixelFormat::BC6H_SFLOAT,
PixelFormat::BC6H_UFLOAT,
};
+constexpr std::array VIEW_CLASS_ASTC_4x4_RGBA{
+ PixelFormat::ASTC_2D_4X4_UNORM,
+ PixelFormat::ASTC_2D_4X4_SRGB,
+};
+
+constexpr std::array VIEW_CLASS_ASTC_5x4_RGBA{
+ PixelFormat::ASTC_2D_5X4_UNORM,
+ PixelFormat::ASTC_2D_5X4_SRGB,
+};
+
+constexpr std::array VIEW_CLASS_ASTC_5x5_RGBA{
+ PixelFormat::ASTC_2D_5X5_UNORM,
+ PixelFormat::ASTC_2D_5X5_SRGB,
+};
+
+constexpr std::array VIEW_CLASS_ASTC_6x5_RGBA{
+ PixelFormat::ASTC_2D_6X5_UNORM,
+ PixelFormat::ASTC_2D_6X5_SRGB,
+};
+
+constexpr std::array VIEW_CLASS_ASTC_6x6_RGBA{
+ PixelFormat::ASTC_2D_6X6_UNORM,
+ PixelFormat::ASTC_2D_6X6_SRGB,
+};
+
+constexpr std::array VIEW_CLASS_ASTC_8x5_RGBA{
+ PixelFormat::ASTC_2D_8X5_UNORM,
+ PixelFormat::ASTC_2D_8X5_SRGB,
+};
+
+constexpr std::array VIEW_CLASS_ASTC_8x8_RGBA{
+ PixelFormat::ASTC_2D_8X8_UNORM,
+ PixelFormat::ASTC_2D_8X8_SRGB,
+};
+
+// Missing formats:
+// PixelFormat::ASTC_2D_10X5_UNORM
+// PixelFormat::ASTC_2D_10X5_SRGB
+
+// Missing formats:
+// PixelFormat::ASTC_2D_10X6_UNORM
+// PixelFormat::ASTC_2D_10X6_SRGB
+
+constexpr std::array VIEW_CLASS_ASTC_10x8_RGBA{
+ PixelFormat::ASTC_2D_10X8_UNORM,
+ PixelFormat::ASTC_2D_10X8_SRGB,
+};
+
+constexpr std::array VIEW_CLASS_ASTC_10x10_RGBA{
+ PixelFormat::ASTC_2D_10X10_UNORM,
+ PixelFormat::ASTC_2D_10X10_SRGB,
+};
+
+// Missing formats
+// ASTC_2D_12X10_UNORM,
+// ASTC_2D_12X10_SRGB,
+
+constexpr std::array VIEW_CLASS_ASTC_12x12_RGBA{
+ PixelFormat::ASTC_2D_12X12_UNORM,
+ PixelFormat::ASTC_2D_12X12_SRGB,
+};
+
// Compatibility table taken from Table 4.X.1 in:
// https://www.khronos.org/registry/OpenGL/extensions/ARB/ARB_copy_image.txt
-constexpr std::array COPY_CLASS_128_BITS = {
+constexpr std::array COPY_CLASS_128_BITS{
PixelFormat::R32G32B32A32_UINT, PixelFormat::R32G32B32A32_FLOAT, PixelFormat::R32G32B32A32_SINT,
PixelFormat::BC2_UNORM, PixelFormat::BC2_SRGB, PixelFormat::BC3_UNORM,
PixelFormat::BC3_SRGB, PixelFormat::BC5_UNORM, PixelFormat::BC5_SNORM,
@@ -97,7 +161,7 @@ constexpr std::array COPY_CLASS_128_BITS = {
// PixelFormat::RGBA32I
// COMPRESSED_RG_RGTC2
-constexpr std::array COPY_CLASS_64_BITS = {
+constexpr std::array COPY_CLASS_64_BITS{
PixelFormat::R16G16B16A16_FLOAT, PixelFormat::R16G16B16A16_UINT,
PixelFormat::R16G16B16A16_UNORM, PixelFormat::R16G16B16A16_SNORM,
PixelFormat::R16G16B16A16_SINT, PixelFormat::R32G32_UINT,
@@ -110,32 +174,36 @@ constexpr std::array COPY_CLASS_64_BITS = {
// COMPRESSED_RGBA_S3TC_DXT1_EXT
// COMPRESSED_SIGNED_RED_RGTC1
-void Enable(FormatCompatibility::Table& compatiblity, size_t format_a, size_t format_b) {
- compatiblity[format_a][format_b] = true;
- compatiblity[format_b][format_a] = true;
+constexpr void Enable(Table& table, size_t format_a, size_t format_b) {
+ table[format_a][format_b / 64] |= u64(1) << (format_b % 64);
+ table[format_b][format_a / 64] |= u64(1) << (format_a % 64);
}
-void Enable(FormatCompatibility::Table& compatibility, PixelFormat format_a, PixelFormat format_b) {
- Enable(compatibility, static_cast<size_t>(format_a), static_cast<size_t>(format_b));
+constexpr void Enable(Table& table, PixelFormat format_a, PixelFormat format_b) {
+ Enable(table, static_cast<size_t>(format_a), static_cast<size_t>(format_b));
}
template <typename Range>
-void EnableRange(FormatCompatibility::Table& compatibility, const Range& range) {
+constexpr void EnableRange(Table& table, const Range& range) {
for (auto it_a = range.begin(); it_a != range.end(); ++it_a) {
for (auto it_b = it_a; it_b != range.end(); ++it_b) {
- Enable(compatibility, *it_a, *it_b);
+ Enable(table, *it_a, *it_b);
}
}
}
-} // Anonymous namespace
+constexpr bool IsSupported(const Table& table, PixelFormat format_a, PixelFormat format_b) {
+ const size_t a = static_cast<size_t>(format_a);
+ const size_t b = static_cast<size_t>(format_b);
+ return ((table[a][b / 64] >> (b % 64)) & 1) != 0;
+}
-FormatCompatibility::FormatCompatibility() {
+constexpr Table MakeViewTable() {
+ Table view{};
for (size_t i = 0; i < MaxPixelFormat; ++i) {
// Identity is allowed
Enable(view, i, i);
}
-
EnableRange(view, VIEW_CLASS_128_BITS);
EnableRange(view, VIEW_CLASS_96_BITS);
EnableRange(view, VIEW_CLASS_64_BITS);
@@ -146,10 +214,36 @@ FormatCompatibility::FormatCompatibility() {
EnableRange(view, VIEW_CLASS_RGTC2_RG);
EnableRange(view, VIEW_CLASS_BPTC_UNORM);
EnableRange(view, VIEW_CLASS_BPTC_FLOAT);
+ EnableRange(view, VIEW_CLASS_ASTC_4x4_RGBA);
+ EnableRange(view, VIEW_CLASS_ASTC_5x4_RGBA);
+ EnableRange(view, VIEW_CLASS_ASTC_5x5_RGBA);
+ EnableRange(view, VIEW_CLASS_ASTC_6x5_RGBA);
+ EnableRange(view, VIEW_CLASS_ASTC_6x6_RGBA);
+ EnableRange(view, VIEW_CLASS_ASTC_8x5_RGBA);
+ EnableRange(view, VIEW_CLASS_ASTC_8x8_RGBA);
+ EnableRange(view, VIEW_CLASS_ASTC_10x8_RGBA);
+ EnableRange(view, VIEW_CLASS_ASTC_10x10_RGBA);
+ EnableRange(view, VIEW_CLASS_ASTC_12x12_RGBA);
+ return view;
+}
- copy = view;
+constexpr Table MakeCopyTable() {
+ Table copy = MakeViewTable();
EnableRange(copy, COPY_CLASS_128_BITS);
EnableRange(copy, COPY_CLASS_64_BITS);
+ return copy;
+}
+
+} // Anonymous namespace
+
+bool IsViewCompatible(PixelFormat format_a, PixelFormat format_b) {
+ static constexpr Table TABLE = MakeViewTable();
+ return IsSupported(TABLE, format_a, format_b);
+}
+
+bool IsCopyCompatible(PixelFormat format_a, PixelFormat format_b) {
+ static constexpr Table TABLE = MakeCopyTable();
+ return IsSupported(TABLE, format_a, format_b);
}
} // namespace VideoCore::Surface
diff --git a/src/video_core/compatible_formats.h b/src/video_core/compatible_formats.h
index 51766349b..b5eb03bea 100644
--- a/src/video_core/compatible_formats.h
+++ b/src/video_core/compatible_formats.h
@@ -4,31 +4,12 @@
#pragma once
-#include <array>
-#include <bitset>
-#include <cstddef>
-
#include "video_core/surface.h"
namespace VideoCore::Surface {
-class FormatCompatibility {
-public:
- using Table = std::array<std::bitset<MaxPixelFormat>, MaxPixelFormat>;
-
- explicit FormatCompatibility();
-
- bool TestView(PixelFormat format_a, PixelFormat format_b) const noexcept {
- return view[static_cast<size_t>(format_a)][static_cast<size_t>(format_b)];
- }
-
- bool TestCopy(PixelFormat format_a, PixelFormat format_b) const noexcept {
- return copy[static_cast<size_t>(format_a)][static_cast<size_t>(format_b)];
- }
+bool IsViewCompatible(PixelFormat format_a, PixelFormat format_b);
-private:
- Table view;
- Table copy;
-};
+bool IsCopyCompatible(PixelFormat format_a, PixelFormat format_b);
} // namespace VideoCore::Surface
diff --git a/src/video_core/delayed_destruction_ring.h b/src/video_core/delayed_destruction_ring.h
new file mode 100644
index 000000000..4f1d29c04
--- /dev/null
+++ b/src/video_core/delayed_destruction_ring.h
@@ -0,0 +1,32 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include <cstddef>
+#include <utility>
+#include <vector>
+
+namespace VideoCommon {
+
+/// Container to push objects to be destroyed a few ticks in the future
+template <typename T, size_t TICKS_TO_DESTROY>
+class DelayedDestructionRing {
+public:
+ void Tick() {
+ index = (index + 1) % TICKS_TO_DESTROY;
+ elements[index].clear();
+ }
+
+ void Push(T&& object) {
+ elements[index].push_back(std::move(object));
+ }
+
+private:
+ size_t index = 0;
+ std::array<std::vector<T>, TICKS_TO_DESTROY> elements;
+};
+
+} // namespace VideoCommon
diff --git a/src/video_core/dirty_flags.cpp b/src/video_core/dirty_flags.cpp
index e16075993..b1eaac00c 100644
--- a/src/video_core/dirty_flags.cpp
+++ b/src/video_core/dirty_flags.cpp
@@ -9,13 +9,16 @@
#include "video_core/dirty_flags.h"
#define OFF(field_name) MAXWELL3D_REG_INDEX(field_name)
-#define NUM(field_name) (sizeof(::Tegra::Engines::Maxwell3D::Regs::field_name) / sizeof(u32))
+#define NUM(field_name) (sizeof(::Tegra::Engines::Maxwell3D::Regs::field_name) / (sizeof(u32)))
namespace VideoCommon::Dirty {
using Tegra::Engines::Maxwell3D;
void SetupDirtyRenderTargets(Tegra::Engines::Maxwell3D::DirtyState::Tables& tables) {
+ FillBlock(tables[0], OFF(tic), NUM(tic), Descriptors);
+ FillBlock(tables[0], OFF(tsc), NUM(tsc), Descriptors);
+
static constexpr std::size_t num_per_rt = NUM(rt[0]);
static constexpr std::size_t begin = OFF(rt);
static constexpr std::size_t num = num_per_rt * Maxwell3D::Regs::NumRenderTargets;
@@ -23,6 +26,10 @@ void SetupDirtyRenderTargets(Tegra::Engines::Maxwell3D::DirtyState::Tables& tabl
FillBlock(tables[0], begin + rt * num_per_rt, num_per_rt, ColorBuffer0 + rt);
}
FillBlock(tables[1], begin, num, RenderTargets);
+ FillBlock(tables[0], OFF(render_area), NUM(render_area), RenderTargets);
+
+ tables[0][OFF(rt_control)] = RenderTargets;
+ tables[1][OFF(rt_control)] = RenderTargetControl;
static constexpr std::array zeta_flags{ZetaBuffer, RenderTargets};
for (std::size_t i = 0; i < std::size(zeta_flags); ++i) {
diff --git a/src/video_core/dirty_flags.h b/src/video_core/dirty_flags.h
index 3f6c1d83a..875527ddd 100644
--- a/src/video_core/dirty_flags.h
+++ b/src/video_core/dirty_flags.h
@@ -16,7 +16,10 @@ namespace VideoCommon::Dirty {
enum : u8 {
NullEntry = 0,
+ Descriptors,
+
RenderTargets,
+ RenderTargetControl,
ColorBuffer0,
ColorBuffer1,
ColorBuffer2,
diff --git a/src/video_core/dma_pusher.cpp b/src/video_core/dma_pusher.cpp
index d8801b1f5..2c8b20024 100644
--- a/src/video_core/dma_pusher.cpp
+++ b/src/video_core/dma_pusher.cpp
@@ -13,7 +13,7 @@
namespace Tegra {
-DmaPusher::DmaPusher(Core::System& system, GPU& gpu) : gpu{gpu}, system{system} {}
+DmaPusher::DmaPusher(Core::System& system_, GPU& gpu_) : gpu{gpu_}, system{system_} {}
DmaPusher::~DmaPusher() = default;
@@ -152,7 +152,12 @@ void DmaPusher::SetState(const CommandHeader& command_header) {
void DmaPusher::CallMethod(u32 argument) const {
if (dma_state.method < non_puller_methods) {
- gpu.CallMethod({dma_state.method, argument, dma_state.subchannel, dma_state.method_count});
+ gpu.CallMethod(GPU::MethodCall{
+ dma_state.method,
+ argument,
+ dma_state.subchannel,
+ dma_state.method_count,
+ });
} else {
subchannels[dma_state.subchannel]->CallMethod(dma_state.method, argument,
dma_state.is_last_call);
diff --git a/src/video_core/dma_pusher.h b/src/video_core/dma_pusher.h
index 96ac267f7..19f286fa7 100644
--- a/src/video_core/dma_pusher.h
+++ b/src/video_core/dma_pusher.h
@@ -87,11 +87,11 @@ inline CommandHeader BuildCommandHeader(BufferMethods method, u32 arg_count, Sub
struct CommandList final {
CommandList() = default;
explicit CommandList(std::size_t size) : command_lists(size) {}
- explicit CommandList(std::vector<Tegra::CommandHeader>&& prefetch_command_list)
- : prefetch_command_list{std::move(prefetch_command_list)} {}
+ explicit CommandList(std::vector<CommandHeader>&& prefetch_command_list_)
+ : prefetch_command_list{std::move(prefetch_command_list_)} {}
- std::vector<Tegra::CommandListHeader> command_lists;
- std::vector<Tegra::CommandHeader> prefetch_command_list;
+ std::vector<CommandListHeader> command_lists;
+ std::vector<CommandHeader> prefetch_command_list;
};
/**
@@ -103,7 +103,7 @@ struct CommandList final {
*/
class DmaPusher final {
public:
- explicit DmaPusher(Core::System& system, GPU& gpu);
+ explicit DmaPusher(Core::System& system_, GPU& gpu_);
~DmaPusher();
void Push(CommandList&& entries) {
@@ -112,7 +112,7 @@ public:
void DispatchCalls();
- void BindSubchannel(Tegra::Engines::EngineInterface* engine, u32 subchannel_id) {
+ void BindSubchannel(Engines::EngineInterface* engine, u32 subchannel_id) {
subchannels[subchannel_id] = engine;
}
@@ -145,7 +145,7 @@ private:
bool ib_enable{true}; ///< IB mode enabled
- std::array<Tegra::Engines::EngineInterface*, max_subchannels> subchannels{};
+ std::array<Engines::EngineInterface*, max_subchannels> subchannels{};
GPU& gpu;
Core::System& system;
diff --git a/src/video_core/engines/engine_upload.cpp b/src/video_core/engines/engine_upload.cpp
index d44ad0cd8..71d7e1473 100644
--- a/src/video_core/engines/engine_upload.cpp
+++ b/src/video_core/engines/engine_upload.cpp
@@ -11,16 +11,16 @@
namespace Tegra::Engines::Upload {
-State::State(MemoryManager& memory_manager, Registers& regs)
- : regs{regs}, memory_manager{memory_manager} {}
+State::State(MemoryManager& memory_manager_, Registers& regs_)
+ : regs{regs_}, memory_manager{memory_manager_} {}
State::~State() = default;
-void State::ProcessExec(const bool is_linear) {
+void State::ProcessExec(const bool is_linear_) {
write_offset = 0;
copy_size = regs.line_length_in * regs.line_count;
inner_buffer.resize(copy_size);
- this->is_linear = is_linear;
+ is_linear = is_linear_;
}
void State::ProcessData(const u32 data, const bool is_last_call) {
diff --git a/src/video_core/engines/engine_upload.h b/src/video_core/engines/engine_upload.h
index 462da419e..1c7f1effa 100644
--- a/src/video_core/engines/engine_upload.h
+++ b/src/video_core/engines/engine_upload.h
@@ -54,10 +54,10 @@ struct Registers {
class State {
public:
- State(MemoryManager& memory_manager, Registers& regs);
+ explicit State(MemoryManager& memory_manager_, Registers& regs_);
~State();
- void ProcessExec(bool is_linear);
+ void ProcessExec(bool is_linear_);
void ProcessData(u32 data, bool is_last_call);
private:
diff --git a/src/video_core/engines/fermi_2d.cpp b/src/video_core/engines/fermi_2d.cpp
index 9409c4075..a01d334ad 100644
--- a/src/video_core/engines/fermi_2d.cpp
+++ b/src/video_core/engines/fermi_2d.cpp
@@ -10,7 +10,11 @@
namespace Tegra::Engines {
-Fermi2D::Fermi2D() = default;
+Fermi2D::Fermi2D() {
+ // Nvidia's OpenGL driver seems to assume these values
+ regs.src.depth = 1;
+ regs.dst.depth = 1;
+}
Fermi2D::~Fermi2D() = default;
@@ -21,79 +25,43 @@ void Fermi2D::BindRasterizer(VideoCore::RasterizerInterface& rasterizer_) {
void Fermi2D::CallMethod(u32 method, u32 method_argument, bool is_last_call) {
ASSERT_MSG(method < Regs::NUM_REGS,
"Invalid Fermi2D register, increase the size of the Regs structure");
-
regs.reg_array[method] = method_argument;
- switch (method) {
- // Trigger the surface copy on the last register write. This is blit_src_y, but this is 64-bit,
- // so trigger on the second 32-bit write.
- case FERMI2D_REG_INDEX(blit_src_y) + 1: {
- HandleSurfaceCopy();
- break;
- }
+ if (method == FERMI2D_REG_INDEX(pixels_from_memory.src_y0) + 1) {
+ Blit();
}
}
void Fermi2D::CallMultiMethod(u32 method, const u32* base_start, u32 amount, u32 methods_pending) {
- for (std::size_t i = 0; i < amount; i++) {
- CallMethod(method, base_start[i], methods_pending - static_cast<u32>(i) <= 1);
+ for (u32 i = 0; i < amount; ++i) {
+ CallMethod(method, base_start[i], methods_pending - i <= 1);
}
}
-static std::pair<u32, u32> DelimitLine(u32 src_1, u32 src_2, u32 dst_1, u32 dst_2, u32 src_line) {
- const u32 line_a = src_2 - src_1;
- const u32 line_b = dst_2 - dst_1;
- const u32 excess = std::max<s32>(0, line_a - src_line + src_1);
- return {line_b - (excess * line_b) / line_a, excess};
-}
-
-void Fermi2D::HandleSurfaceCopy() {
- LOG_DEBUG(HW_GPU, "Requested a surface copy with operation {}",
- static_cast<u32>(regs.operation));
+void Fermi2D::Blit() {
+ LOG_DEBUG(HW_GPU, "called. source address=0x{:x}, destination address=0x{:x}",
+ regs.src.Address(), regs.dst.Address());
- // TODO(Subv): Only raw copies are implemented.
- ASSERT(regs.operation == Operation::SrcCopy);
+ UNIMPLEMENTED_IF_MSG(regs.operation != Operation::SrcCopy, "Operation is not copy");
+ UNIMPLEMENTED_IF_MSG(regs.src.layer != 0, "Source layer is not zero");
+ UNIMPLEMENTED_IF_MSG(regs.dst.layer != 0, "Destination layer is not zero");
+ UNIMPLEMENTED_IF_MSG(regs.src.depth != 1, "Source depth is not one");
+ UNIMPLEMENTED_IF_MSG(regs.clip_enable != 0, "Clipped blit enabled");
- const u32 src_blit_x1{static_cast<u32>(regs.blit_src_x >> 32)};
- const u32 src_blit_y1{static_cast<u32>(regs.blit_src_y >> 32)};
- u32 src_blit_x2, src_blit_y2;
- if (regs.blit_control.origin == Origin::Corner) {
- src_blit_x2 =
- static_cast<u32>((regs.blit_src_x + (regs.blit_du_dx * regs.blit_dst_width)) >> 32);
- src_blit_y2 =
- static_cast<u32>((regs.blit_src_y + (regs.blit_dv_dy * regs.blit_dst_height)) >> 32);
- } else {
- src_blit_x2 = static_cast<u32>((regs.blit_src_x >> 32) + regs.blit_dst_width);
- src_blit_y2 = static_cast<u32>((regs.blit_src_y >> 32) + regs.blit_dst_height);
- }
- u32 dst_blit_x2 = regs.blit_dst_x + regs.blit_dst_width;
- u32 dst_blit_y2 = regs.blit_dst_y + regs.blit_dst_height;
- const auto [new_dst_w, src_excess_x] =
- DelimitLine(src_blit_x1, src_blit_x2, regs.blit_dst_x, dst_blit_x2, regs.src.width);
- const auto [new_dst_h, src_excess_y] =
- DelimitLine(src_blit_y1, src_blit_y2, regs.blit_dst_y, dst_blit_y2, regs.src.height);
- dst_blit_x2 = new_dst_w + regs.blit_dst_x;
- src_blit_x2 = src_blit_x2 - src_excess_x;
- dst_blit_y2 = new_dst_h + regs.blit_dst_y;
- src_blit_y2 = src_blit_y2 - src_excess_y;
- const auto [new_src_w, dst_excess_x] =
- DelimitLine(regs.blit_dst_x, dst_blit_x2, src_blit_x1, src_blit_x2, regs.dst.width);
- const auto [new_src_h, dst_excess_y] =
- DelimitLine(regs.blit_dst_y, dst_blit_y2, src_blit_y1, src_blit_y2, regs.dst.height);
- src_blit_x2 = new_src_w + src_blit_x1;
- dst_blit_x2 = dst_blit_x2 - dst_excess_x;
- src_blit_y2 = new_src_h + src_blit_y1;
- dst_blit_y2 = dst_blit_y2 - dst_excess_y;
- const Common::Rectangle<u32> src_rect{src_blit_x1, src_blit_y1, src_blit_x2, src_blit_y2};
- const Common::Rectangle<u32> dst_rect{regs.blit_dst_x, regs.blit_dst_y, dst_blit_x2,
- dst_blit_y2};
- const Config copy_config{
+ const auto& args = regs.pixels_from_memory;
+ const Config config{
.operation = regs.operation,
- .filter = regs.blit_control.filter,
- .src_rect = src_rect,
- .dst_rect = dst_rect,
+ .filter = args.sample_mode.filter,
+ .dst_x0 = args.dst_x0,
+ .dst_y0 = args.dst_y0,
+ .dst_x1 = args.dst_x0 + args.dst_width,
+ .dst_y1 = args.dst_y0 + args.dst_height,
+ .src_x0 = static_cast<s32>(args.src_x0 >> 32),
+ .src_y0 = static_cast<s32>(args.src_y0 >> 32),
+ .src_x1 = static_cast<s32>((args.du_dx * args.dst_width + args.src_x0) >> 32),
+ .src_y1 = static_cast<s32>((args.dv_dy * args.dst_height + args.src_y0) >> 32),
};
- if (!rasterizer->AccelerateSurfaceCopy(regs.src, regs.dst, copy_config)) {
+ if (!rasterizer->AccelerateSurfaceCopy(regs.src, regs.dst, config)) {
UNIMPLEMENTED();
}
}
diff --git a/src/video_core/engines/fermi_2d.h b/src/video_core/engines/fermi_2d.h
index 0909709ec..81522988e 100644
--- a/src/video_core/engines/fermi_2d.h
+++ b/src/video_core/engines/fermi_2d.h
@@ -53,8 +53,8 @@ public:
};
enum class Filter : u32 {
- PointSample = 0, // Nearest
- Linear = 1,
+ Point = 0,
+ Bilinear = 1,
};
enum class Operation : u32 {
@@ -67,88 +67,235 @@ public:
BlendPremult = 6,
};
- struct Regs {
- static constexpr std::size_t NUM_REGS = 0x258;
+ enum class MemoryLayout : u32 {
+ BlockLinear = 0,
+ Pitch = 1,
+ };
- struct Surface {
- RenderTargetFormat format;
- BitField<0, 1, u32> linear;
- union {
- BitField<0, 4, u32> block_width;
- BitField<4, 4, u32> block_height;
- BitField<8, 4, u32> block_depth;
- };
- u32 depth;
- u32 layer;
- u32 pitch;
- u32 width;
- u32 height;
- u32 address_high;
- u32 address_low;
-
- GPUVAddr Address() const {
- return static_cast<GPUVAddr>((static_cast<GPUVAddr>(address_high) << 32) |
- address_low);
- }
-
- u32 BlockWidth() const {
- return block_width.Value();
- }
-
- u32 BlockHeight() const {
- return block_height.Value();
- }
-
- u32 BlockDepth() const {
- return block_depth.Value();
- }
- };
- static_assert(sizeof(Surface) == 0x28, "Surface has incorrect size");
+ enum class CpuIndexWrap : u32 {
+ Wrap = 0,
+ NoWrap = 1,
+ };
+ struct Surface {
+ RenderTargetFormat format;
+ MemoryLayout linear;
union {
- struct {
- INSERT_UNION_PADDING_WORDS(0x80);
+ BitField<0, 4, u32> block_width;
+ BitField<4, 4, u32> block_height;
+ BitField<8, 4, u32> block_depth;
+ };
+ u32 depth;
+ u32 layer;
+ u32 pitch;
+ u32 width;
+ u32 height;
+ u32 addr_upper;
+ u32 addr_lower;
+
+ [[nodiscard]] constexpr GPUVAddr Address() const noexcept {
+ return (static_cast<GPUVAddr>(addr_upper) << 32) | static_cast<GPUVAddr>(addr_lower);
+ }
+ };
+ static_assert(sizeof(Surface) == 0x28, "Surface has incorrect size");
- Surface dst;
+ enum class SectorPromotion : u32 {
+ NoPromotion = 0,
+ PromoteTo2V = 1,
+ PromoteTo2H = 2,
+ PromoteTo4 = 3,
+ };
+
+ enum class NumTpcs : u32 {
+ All = 0,
+ One = 1,
+ };
- INSERT_UNION_PADDING_WORDS(2);
+ enum class RenderEnableMode : u32 {
+ False = 0,
+ True = 1,
+ Conditional = 2,
+ RenderIfEqual = 3,
+ RenderIfNotEqual = 4,
+ };
- Surface src;
+ enum class ColorKeyFormat : u32 {
+ A16R56G6B5 = 0,
+ A1R5G55B5 = 1,
+ A8R8G8B8 = 2,
+ A2R10G10B10 = 3,
+ Y8 = 4,
+ Y16 = 5,
+ Y32 = 6,
+ };
- INSERT_UNION_PADDING_WORDS(0x15);
+ union Beta4 {
+ BitField<0, 8, u32> b;
+ BitField<8, 8, u32> g;
+ BitField<16, 8, u32> r;
+ BitField<24, 8, u32> a;
+ };
- Operation operation;
+ struct Point {
+ u32 x;
+ u32 y;
+ };
- INSERT_UNION_PADDING_WORDS(0x177);
+ enum class PatternSelect : u32 {
+ MonoChrome8x8 = 0,
+ MonoChrome64x1 = 1,
+ MonoChrome1x64 = 2,
+ Color = 3,
+ };
+ enum class NotifyType : u32 {
+ WriteOnly = 0,
+ WriteThenAwaken = 1,
+ };
+
+ enum class MonochromePatternColorFormat : u32 {
+ A8X8R8G6B5 = 0,
+ A1R5G5B5 = 1,
+ A8R8G8B8 = 2,
+ A8Y8 = 3,
+ A8X8Y16 = 4,
+ Y32 = 5,
+ };
+
+ enum class MonochromePatternFormat : u32 {
+ CGA6_M1 = 0,
+ LE_M1 = 1,
+ };
+
+ union Regs {
+ static constexpr std::size_t NUM_REGS = 0x258;
+ struct {
+ u32 object;
+ INSERT_UNION_PADDING_WORDS(0x3F);
+ u32 no_operation;
+ NotifyType notify;
+ INSERT_UNION_PADDING_WORDS(0x2);
+ u32 wait_for_idle;
+ INSERT_UNION_PADDING_WORDS(0xB);
+ u32 pm_trigger;
+ INSERT_UNION_PADDING_WORDS(0xF);
+ u32 context_dma_notify;
+ u32 dst_context_dma;
+ u32 src_context_dma;
+ u32 semaphore_context_dma;
+ INSERT_UNION_PADDING_WORDS(0x1C);
+ Surface dst;
+ CpuIndexWrap pixels_from_cpu_index_wrap;
+ u32 kind2d_check_enable;
+ Surface src;
+ SectorPromotion pixels_from_memory_sector_promotion;
+ INSERT_UNION_PADDING_WORDS(0x1);
+ NumTpcs num_tpcs;
+ u32 render_enable_addr_upper;
+ u32 render_enable_addr_lower;
+ RenderEnableMode render_enable_mode;
+ INSERT_UNION_PADDING_WORDS(0x4);
+ u32 clip_x0;
+ u32 clip_y0;
+ u32 clip_width;
+ u32 clip_height;
+ BitField<0, 1, u32> clip_enable;
+ BitField<0, 3, ColorKeyFormat> color_key_format;
+ u32 color_key;
+ BitField<0, 1, u32> color_key_enable;
+ BitField<0, 8, u32> rop;
+ u32 beta1;
+ Beta4 beta4;
+ Operation operation;
+ union {
+ BitField<0, 6, u32> x;
+ BitField<8, 6, u32> y;
+ } pattern_offset;
+ BitField<0, 2, PatternSelect> pattern_select;
+ INSERT_UNION_PADDING_WORDS(0xC);
+ struct {
+ BitField<0, 3, MonochromePatternColorFormat> color_format;
+ BitField<0, 1, MonochromePatternFormat> format;
+ u32 color0;
+ u32 color1;
+ u32 pattern0;
+ u32 pattern1;
+ } monochrome_pattern;
+ struct {
+ std::array<u32, 0x40> X8R8G8B8;
+ std::array<u32, 0x20> R5G6B5;
+ std::array<u32, 0x20> X1R5G5B5;
+ std::array<u32, 0x10> Y8;
+ } color_pattern;
+ INSERT_UNION_PADDING_WORDS(0x10);
+ struct {
+ u32 prim_mode;
+ u32 prim_color_format;
+ u32 prim_color;
+ u32 line_tie_break_bits;
+ INSERT_UNION_PADDING_WORDS(0x14);
+ u32 prim_point_xy;
+ INSERT_UNION_PADDING_WORDS(0x7);
+ std::array<Point, 0x40> prim_point;
+ } render_solid;
+ struct {
+ u32 data_type;
+ u32 color_format;
+ u32 index_format;
+ u32 mono_format;
+ u32 wrap;
+ u32 color0;
+ u32 color1;
+ u32 mono_opacity;
+ INSERT_UNION_PADDING_WORDS(0x6);
+ u32 src_width;
+ u32 src_height;
+ u32 dx_du_frac;
+ u32 dx_du_int;
+ u32 dx_dv_frac;
+ u32 dy_dv_int;
+ u32 dst_x0_frac;
+ u32 dst_x0_int;
+ u32 dst_y0_frac;
+ u32 dst_y0_int;
+ u32 data;
+ } pixels_from_cpu;
+ INSERT_UNION_PADDING_WORDS(0x3);
+ u32 big_endian_control;
+ INSERT_UNION_PADDING_WORDS(0x3);
+ struct {
+ BitField<0, 3, u32> block_shape;
+ BitField<0, 5, u32> corral_size;
+ BitField<0, 1, u32> safe_overlap;
union {
- u32 raw;
BitField<0, 1, Origin> origin;
BitField<4, 1, Filter> filter;
- } blit_control;
-
+ } sample_mode;
INSERT_UNION_PADDING_WORDS(0x8);
-
- u32 blit_dst_x;
- u32 blit_dst_y;
- u32 blit_dst_width;
- u32 blit_dst_height;
- u64 blit_du_dx;
- u64 blit_dv_dy;
- u64 blit_src_x;
- u64 blit_src_y;
-
- INSERT_UNION_PADDING_WORDS(0x21);
- };
- std::array<u32, NUM_REGS> reg_array;
+ s32 dst_x0;
+ s32 dst_y0;
+ s32 dst_width;
+ s32 dst_height;
+ s64 du_dx;
+ s64 dv_dy;
+ s64 src_x0;
+ s64 src_y0;
+ } pixels_from_memory;
};
+ std::array<u32, NUM_REGS> reg_array;
} regs{};
struct Config {
- Operation operation{};
- Filter filter{};
- Common::Rectangle<u32> src_rect;
- Common::Rectangle<u32> dst_rect;
+ Operation operation;
+ Filter filter;
+ s32 dst_x0;
+ s32 dst_y0;
+ s32 dst_x1;
+ s32 dst_y1;
+ s32 src_x0;
+ s32 src_y0;
+ s32 src_x1;
+ s32 src_y1;
};
private:
@@ -156,25 +303,49 @@ private:
/// Performs the copy from the source surface to the destination surface as configured in the
/// registers.
- void HandleSurfaceCopy();
+ void Blit();
};
#define ASSERT_REG_POSITION(field_name, position) \
- static_assert(offsetof(Fermi2D::Regs, field_name) == position * 4, \
+ static_assert(offsetof(Fermi2D::Regs, field_name) == position, \
"Field " #field_name " has invalid position")
-ASSERT_REG_POSITION(dst, 0x80);
-ASSERT_REG_POSITION(src, 0x8C);
-ASSERT_REG_POSITION(operation, 0xAB);
-ASSERT_REG_POSITION(blit_control, 0x223);
-ASSERT_REG_POSITION(blit_dst_x, 0x22c);
-ASSERT_REG_POSITION(blit_dst_y, 0x22d);
-ASSERT_REG_POSITION(blit_dst_width, 0x22e);
-ASSERT_REG_POSITION(blit_dst_height, 0x22f);
-ASSERT_REG_POSITION(blit_du_dx, 0x230);
-ASSERT_REG_POSITION(blit_dv_dy, 0x232);
-ASSERT_REG_POSITION(blit_src_x, 0x234);
-ASSERT_REG_POSITION(blit_src_y, 0x236);
+ASSERT_REG_POSITION(object, 0x0);
+ASSERT_REG_POSITION(no_operation, 0x100);
+ASSERT_REG_POSITION(notify, 0x104);
+ASSERT_REG_POSITION(wait_for_idle, 0x110);
+ASSERT_REG_POSITION(pm_trigger, 0x140);
+ASSERT_REG_POSITION(context_dma_notify, 0x180);
+ASSERT_REG_POSITION(dst_context_dma, 0x184);
+ASSERT_REG_POSITION(src_context_dma, 0x188);
+ASSERT_REG_POSITION(semaphore_context_dma, 0x18C);
+ASSERT_REG_POSITION(dst, 0x200);
+ASSERT_REG_POSITION(pixels_from_cpu_index_wrap, 0x228);
+ASSERT_REG_POSITION(kind2d_check_enable, 0x22C);
+ASSERT_REG_POSITION(src, 0x230);
+ASSERT_REG_POSITION(pixels_from_memory_sector_promotion, 0x258);
+ASSERT_REG_POSITION(num_tpcs, 0x260);
+ASSERT_REG_POSITION(render_enable_addr_upper, 0x264);
+ASSERT_REG_POSITION(render_enable_addr_lower, 0x268);
+ASSERT_REG_POSITION(clip_x0, 0x280);
+ASSERT_REG_POSITION(clip_y0, 0x284);
+ASSERT_REG_POSITION(clip_width, 0x288);
+ASSERT_REG_POSITION(clip_height, 0x28c);
+ASSERT_REG_POSITION(clip_enable, 0x290);
+ASSERT_REG_POSITION(color_key_format, 0x294);
+ASSERT_REG_POSITION(color_key, 0x298);
+ASSERT_REG_POSITION(rop, 0x2A0);
+ASSERT_REG_POSITION(beta1, 0x2A4);
+ASSERT_REG_POSITION(beta4, 0x2A8);
+ASSERT_REG_POSITION(operation, 0x2AC);
+ASSERT_REG_POSITION(pattern_offset, 0x2B0);
+ASSERT_REG_POSITION(pattern_select, 0x2B4);
+ASSERT_REG_POSITION(monochrome_pattern, 0x2E8);
+ASSERT_REG_POSITION(color_pattern, 0x300);
+ASSERT_REG_POSITION(render_solid, 0x580);
+ASSERT_REG_POSITION(pixels_from_cpu, 0x800);
+ASSERT_REG_POSITION(big_endian_control, 0x870);
+ASSERT_REG_POSITION(pixels_from_memory, 0x880);
#undef ASSERT_REG_POSITION
diff --git a/src/video_core/engines/kepler_compute.cpp b/src/video_core/engines/kepler_compute.cpp
index 898370739..ba387506e 100644
--- a/src/video_core/engines/kepler_compute.cpp
+++ b/src/video_core/engines/kepler_compute.cpp
@@ -58,24 +58,6 @@ void KeplerCompute::CallMultiMethod(u32 method, const u32* base_start, u32 amoun
}
}
-Texture::FullTextureInfo KeplerCompute::GetTexture(std::size_t offset) const {
- const std::bitset<8> cbuf_mask = launch_description.const_buffer_enable_mask.Value();
- ASSERT(cbuf_mask[regs.tex_cb_index]);
-
- const auto& texinfo = launch_description.const_buffer_config[regs.tex_cb_index];
- ASSERT(texinfo.Address() != 0);
-
- const GPUVAddr address = texinfo.Address() + offset * sizeof(Texture::TextureHandle);
- ASSERT(address < texinfo.Address() + texinfo.size);
-
- const Texture::TextureHandle tex_handle{memory_manager.Read<u32>(address)};
- return GetTextureInfo(tex_handle);
-}
-
-Texture::FullTextureInfo KeplerCompute::GetTextureInfo(Texture::TextureHandle tex_handle) const {
- return Texture::FullTextureInfo{GetTICEntry(tex_handle.tic_id), GetTSCEntry(tex_handle.tsc_id)};
-}
-
u32 KeplerCompute::AccessConstBuffer32(ShaderType stage, u64 const_buffer, u64 offset) const {
ASSERT(stage == ShaderType::Compute);
const auto& buffer = launch_description.const_buffer_config[const_buffer];
@@ -98,9 +80,11 @@ SamplerDescriptor KeplerCompute::AccessBindlessSampler(ShaderType stage, u64 con
SamplerDescriptor KeplerCompute::AccessSampler(u32 handle) const {
const Texture::TextureHandle tex_handle{handle};
- const Texture::FullTextureInfo tex_info = GetTextureInfo(tex_handle);
- SamplerDescriptor result = SamplerDescriptor::FromTIC(tex_info.tic);
- result.is_shadow.Assign(tex_info.tsc.depth_compare_enabled.Value());
+ const Texture::TICEntry tic = GetTICEntry(tex_handle.tic_id);
+ const Texture::TSCEntry tsc = GetTSCEntry(tex_handle.tsc_id);
+
+ SamplerDescriptor result = SamplerDescriptor::FromTIC(tic);
+ result.is_shadow.Assign(tsc.depth_compare_enabled.Value());
return result;
}
diff --git a/src/video_core/engines/kepler_compute.h b/src/video_core/engines/kepler_compute.h
index 7f2500aab..51a041202 100644
--- a/src/video_core/engines/kepler_compute.h
+++ b/src/video_core/engines/kepler_compute.h
@@ -209,11 +209,6 @@ public:
void CallMultiMethod(u32 method, const u32* base_start, u32 amount,
u32 methods_pending) override;
- Texture::FullTextureInfo GetTexture(std::size_t offset) const;
-
- /// Given a texture handle, returns the TSC and TIC entries.
- Texture::FullTextureInfo GetTextureInfo(Texture::TextureHandle tex_handle) const;
-
u32 AccessConstBuffer32(ShaderType stage, u64 const_buffer, u64 offset) const override;
SamplerDescriptor AccessBoundSampler(ShaderType stage, u64 offset) const override;
diff --git a/src/video_core/engines/kepler_memory.cpp b/src/video_core/engines/kepler_memory.cpp
index dc71b2eec..9911140e9 100644
--- a/src/video_core/engines/kepler_memory.cpp
+++ b/src/video_core/engines/kepler_memory.cpp
@@ -14,8 +14,8 @@
namespace Tegra::Engines {
-KeplerMemory::KeplerMemory(Core::System& system, MemoryManager& memory_manager)
- : system{system}, upload_state{memory_manager, regs.upload} {}
+KeplerMemory::KeplerMemory(Core::System& system_, MemoryManager& memory_manager)
+ : system{system_}, upload_state{memory_manager, regs.upload} {}
KeplerMemory::~KeplerMemory() = default;
diff --git a/src/video_core/engines/kepler_memory.h b/src/video_core/engines/kepler_memory.h
index 5b7f71a00..62483589e 100644
--- a/src/video_core/engines/kepler_memory.h
+++ b/src/video_core/engines/kepler_memory.h
@@ -35,7 +35,7 @@ namespace Tegra::Engines {
class KeplerMemory final : public EngineInterface {
public:
- KeplerMemory(Core::System& system, MemoryManager& memory_manager);
+ explicit KeplerMemory(Core::System& system_, MemoryManager& memory_manager);
~KeplerMemory();
/// Write the value to the register identified by method.
diff --git a/src/video_core/engines/maxwell_3d.cpp b/src/video_core/engines/maxwell_3d.cpp
index 6287df633..9be651e24 100644
--- a/src/video_core/engines/maxwell_3d.cpp
+++ b/src/video_core/engines/maxwell_3d.cpp
@@ -2,7 +2,6 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
-#include <cinttypes>
#include <cstring>
#include <optional>
#include "common/assert.h"
@@ -227,6 +226,10 @@ void Maxwell3D::ProcessMethodCall(u32 method, u32 argument, u32 nonshadow_argume
OnMemoryWrite();
}
return;
+ case MAXWELL3D_REG_INDEX(fragment_barrier):
+ return rasterizer->FragmentBarrier();
+ case MAXWELL3D_REG_INDEX(tiled_cache_barrier):
+ return rasterizer->TiledCacheBarrier();
}
}
@@ -359,7 +362,7 @@ void Maxwell3D::CallMethodFromMME(u32 method, u32 method_argument) {
}
void Maxwell3D::FlushMMEInlineDraw() {
- LOG_TRACE(HW_GPU, "called, topology={}, count={}", static_cast<u32>(regs.draw.topology.Value()),
+ LOG_TRACE(HW_GPU, "called, topology={}, count={}", regs.draw.topology.Value(),
regs.vertex_buffer.count);
ASSERT_MSG(!(regs.index_array.count && regs.vertex_buffer.count), "Both indexed and direct?");
ASSERT(mme_draw.instance_count == mme_draw.gl_end_count);
@@ -504,8 +507,7 @@ void Maxwell3D::ProcessCounterReset() {
rasterizer->ResetCounter(QueryType::SamplesPassed);
break;
default:
- LOG_DEBUG(Render_OpenGL, "Unimplemented counter reset={}",
- static_cast<int>(regs.counter_reset));
+ LOG_DEBUG(Render_OpenGL, "Unimplemented counter reset={}", regs.counter_reset);
break;
}
}
@@ -520,7 +522,7 @@ void Maxwell3D::ProcessSyncPoint() {
}
void Maxwell3D::DrawArrays() {
- LOG_TRACE(HW_GPU, "called, topology={}, count={}", static_cast<u32>(regs.draw.topology.Value()),
+ LOG_TRACE(HW_GPU, "called, topology={}, count={}", regs.draw.topology.Value(),
regs.vertex_buffer.count);
ASSERT_MSG(!(regs.index_array.count && regs.vertex_buffer.count), "Both indexed and direct?");
@@ -558,12 +560,12 @@ std::optional<u64> Maxwell3D::GetQueryResult() {
return 0;
case Regs::QuerySelect::SamplesPassed:
// Deferred.
- rasterizer->Query(regs.query.QueryAddress(), VideoCore::QueryType::SamplesPassed,
+ rasterizer->Query(regs.query.QueryAddress(), QueryType::SamplesPassed,
system.GPU().GetTicks());
return std::nullopt;
default:
LOG_DEBUG(HW_GPU, "Unimplemented query select type {}",
- static_cast<u32>(regs.query.query_get.select.Value()));
+ regs.query.query_get.select.Value());
return 1;
}
}
@@ -640,7 +642,7 @@ void Maxwell3D::FinishCBData() {
}
Texture::TICEntry Maxwell3D::GetTICEntry(u32 tic_index) const {
- const GPUVAddr tic_address_gpu{regs.tic.TICAddress() + tic_index * sizeof(Texture::TICEntry)};
+ const GPUVAddr tic_address_gpu{regs.tic.Address() + tic_index * sizeof(Texture::TICEntry)};
Texture::TICEntry tic_entry;
memory_manager.ReadBlockUnsafe(tic_address_gpu, &tic_entry, sizeof(Texture::TICEntry));
@@ -649,43 +651,19 @@ Texture::TICEntry Maxwell3D::GetTICEntry(u32 tic_index) const {
}
Texture::TSCEntry Maxwell3D::GetTSCEntry(u32 tsc_index) const {
- const GPUVAddr tsc_address_gpu{regs.tsc.TSCAddress() + tsc_index * sizeof(Texture::TSCEntry)};
+ const GPUVAddr tsc_address_gpu{regs.tsc.Address() + tsc_index * sizeof(Texture::TSCEntry)};
Texture::TSCEntry tsc_entry;
memory_manager.ReadBlockUnsafe(tsc_address_gpu, &tsc_entry, sizeof(Texture::TSCEntry));
return tsc_entry;
}
-Texture::FullTextureInfo Maxwell3D::GetTextureInfo(Texture::TextureHandle tex_handle) const {
- return Texture::FullTextureInfo{GetTICEntry(tex_handle.tic_id), GetTSCEntry(tex_handle.tsc_id)};
-}
-
-Texture::FullTextureInfo Maxwell3D::GetStageTexture(ShaderType stage, std::size_t offset) const {
- const auto stage_index = static_cast<std::size_t>(stage);
- const auto& shader = state.shader_stages[stage_index];
- const auto& tex_info_buffer = shader.const_buffers[regs.tex_cb_index];
- ASSERT(tex_info_buffer.enabled && tex_info_buffer.address != 0);
-
- const GPUVAddr tex_info_address =
- tex_info_buffer.address + offset * sizeof(Texture::TextureHandle);
-
- ASSERT(tex_info_address < tex_info_buffer.address + tex_info_buffer.size);
-
- const Texture::TextureHandle tex_handle{memory_manager.Read<u32>(tex_info_address)};
-
- return GetTextureInfo(tex_handle);
-}
-
u32 Maxwell3D::GetRegisterValue(u32 method) const {
ASSERT_MSG(method < Regs::NUM_REGS, "Invalid Maxwell3D register");
return regs.reg_array[method];
}
void Maxwell3D::ProcessClearBuffers() {
- ASSERT(regs.clear_buffers.R == regs.clear_buffers.G &&
- regs.clear_buffers.R == regs.clear_buffers.B &&
- regs.clear_buffers.R == regs.clear_buffers.A);
-
rasterizer->Clear();
}
@@ -693,9 +671,7 @@ u32 Maxwell3D::AccessConstBuffer32(ShaderType stage, u64 const_buffer, u64 offse
ASSERT(stage != ShaderType::Compute);
const auto& shader_stage = state.shader_stages[static_cast<std::size_t>(stage)];
const auto& buffer = shader_stage.const_buffers[const_buffer];
- u32 result;
- std::memcpy(&result, memory_manager.GetPointer(buffer.address + offset), sizeof(u32));
- return result;
+ return memory_manager.Read<u32>(buffer.address + offset);
}
SamplerDescriptor Maxwell3D::AccessBoundSampler(ShaderType stage, u64 offset) const {
@@ -713,9 +689,11 @@ SamplerDescriptor Maxwell3D::AccessBindlessSampler(ShaderType stage, u64 const_b
SamplerDescriptor Maxwell3D::AccessSampler(u32 handle) const {
const Texture::TextureHandle tex_handle{handle};
- const Texture::FullTextureInfo tex_info = GetTextureInfo(tex_handle);
- SamplerDescriptor result = SamplerDescriptor::FromTIC(tex_info.tic);
- result.is_shadow.Assign(tex_info.tsc.depth_compare_enabled.Value());
+ const Texture::TICEntry tic = GetTICEntry(tex_handle.tic_id);
+ const Texture::TSCEntry tsc = GetTSCEntry(tex_handle.tsc_id);
+
+ SamplerDescriptor result = SamplerDescriptor::FromTIC(tic);
+ result.is_shadow.Assign(tsc.depth_compare_enabled.Value());
return result;
}
diff --git a/src/video_core/engines/maxwell_3d.h b/src/video_core/engines/maxwell_3d.h
index b0d9559d0..bf9e07c9b 100644
--- a/src/video_core/engines/maxwell_3d.h
+++ b/src/video_core/engines/maxwell_3d.h
@@ -438,16 +438,6 @@ public:
DecrWrapOGL = 0x8508,
};
- enum class MemoryLayout : u32 {
- Linear = 0,
- BlockLinear = 1,
- };
-
- enum class InvMemoryLayout : u32 {
- BlockLinear = 0,
- Linear = 1,
- };
-
enum class CounterReset : u32 {
SampleCnt = 0x01,
Unk02 = 0x02,
@@ -589,21 +579,31 @@ public:
NegativeW = 7,
};
+ enum class SamplerIndex : u32 {
+ Independently = 0,
+ ViaHeaderIndex = 1,
+ };
+
+ struct TileMode {
+ union {
+ BitField<0, 4, u32> block_width;
+ BitField<4, 4, u32> block_height;
+ BitField<8, 4, u32> block_depth;
+ BitField<12, 1, u32> is_pitch_linear;
+ BitField<16, 1, u32> is_3d;
+ };
+ };
+ static_assert(sizeof(TileMode) == 4);
+
struct RenderTargetConfig {
u32 address_high;
u32 address_low;
u32 width;
u32 height;
Tegra::RenderTargetFormat format;
+ TileMode tile_mode;
union {
- BitField<0, 3, u32> block_width;
- BitField<4, 3, u32> block_height;
- BitField<8, 3, u32> block_depth;
- BitField<12, 1, InvMemoryLayout> type;
- BitField<16, 1, u32> is_3d;
- } memory_layout;
- union {
- BitField<0, 16, u32> layers;
+ BitField<0, 16, u32> depth;
BitField<16, 1, u32> volume;
};
u32 layer_stride;
@@ -832,7 +832,11 @@ public:
u32 patch_vertices;
- INSERT_UNION_PADDING_WORDS(0xC);
+ INSERT_UNION_PADDING_WORDS(0x4);
+
+ u32 fragment_barrier;
+
+ INSERT_UNION_PADDING_WORDS(0x7);
std::array<ScissorTest, NumViewports> scissor_test;
@@ -842,7 +846,15 @@ public:
u32 stencil_back_mask;
u32 stencil_back_func_mask;
- INSERT_UNION_PADDING_WORDS(0xC);
+ INSERT_UNION_PADDING_WORDS(0x5);
+
+ u32 invalidate_texture_data_cache;
+
+ INSERT_UNION_PADDING_WORDS(0x1);
+
+ u32 tiled_cache_barrier;
+
+ INSERT_UNION_PADDING_WORDS(0x4);
u32 color_mask_common;
@@ -866,12 +878,7 @@ public:
u32 address_high;
u32 address_low;
Tegra::DepthFormat format;
- union {
- BitField<0, 4, u32> block_width;
- BitField<4, 4, u32> block_height;
- BitField<8, 4, u32> block_depth;
- BitField<20, 1, InvMemoryLayout> type;
- } memory_layout;
+ TileMode tile_mode;
u32 layer_stride;
GPUVAddr Address() const {
@@ -880,7 +887,18 @@ public:
}
} zeta;
- INSERT_UNION_PADDING_WORDS(0x41);
+ struct {
+ union {
+ BitField<0, 16, u32> x;
+ BitField<16, 16, u32> width;
+ };
+ union {
+ BitField<0, 16, u32> y;
+ BitField<16, 16, u32> height;
+ };
+ } render_area;
+
+ INSERT_UNION_PADDING_WORDS(0x3F);
union {
BitField<0, 4, u32> stencil;
@@ -921,7 +939,7 @@ public:
BitField<25, 3, u32> map_7;
};
- u32 GetMap(std::size_t index) const {
+ u32 Map(std::size_t index) const {
const std::array<u32, NumRenderTargets> maps{map_0, map_1, map_2, map_3,
map_4, map_5, map_6, map_7};
ASSERT(index < maps.size());
@@ -934,11 +952,13 @@ public:
u32 zeta_width;
u32 zeta_height;
union {
- BitField<0, 16, u32> zeta_layers;
+ BitField<0, 16, u32> zeta_depth;
BitField<16, 1, u32> zeta_volume;
};
- INSERT_UNION_PADDING_WORDS(0x26);
+ SamplerIndex sampler_index;
+
+ INSERT_UNION_PADDING_WORDS(0x25);
u32 depth_test_enable;
@@ -964,6 +984,7 @@ public:
float b;
float a;
} blend_color;
+
INSERT_UNION_PADDING_WORDS(0x4);
struct {
@@ -1001,7 +1022,12 @@ public:
float line_width_smooth;
float line_width_aliased;
- INSERT_UNION_PADDING_WORDS(0x1F);
+ INSERT_UNION_PADDING_WORDS(0x1B);
+
+ u32 invalidate_sampler_cache_no_wfi;
+ u32 invalidate_texture_header_cache_no_wfi;
+
+ INSERT_UNION_PADDING_WORDS(0x2);
u32 vb_element_base;
u32 vb_base_instance;
@@ -1045,13 +1071,13 @@ public:
} condition;
struct {
- u32 tsc_address_high;
- u32 tsc_address_low;
- u32 tsc_limit;
+ u32 address_high;
+ u32 address_low;
+ u32 limit;
- GPUVAddr TSCAddress() const {
- return static_cast<GPUVAddr>(
- (static_cast<GPUVAddr>(tsc_address_high) << 32) | tsc_address_low);
+ GPUVAddr Address() const {
+ return static_cast<GPUVAddr>((static_cast<GPUVAddr>(address_high) << 32) |
+ address_low);
}
} tsc;
@@ -1062,13 +1088,13 @@ public:
u32 line_smooth_enable;
struct {
- u32 tic_address_high;
- u32 tic_address_low;
- u32 tic_limit;
+ u32 address_high;
+ u32 address_low;
+ u32 limit;
- GPUVAddr TICAddress() const {
- return static_cast<GPUVAddr>(
- (static_cast<GPUVAddr>(tic_address_high) << 32) | tic_address_low);
+ GPUVAddr Address() const {
+ return static_cast<GPUVAddr>((static_cast<GPUVAddr>(address_high) << 32) |
+ address_low);
}
} tic;
@@ -1397,12 +1423,6 @@ public:
void FlushMMEInlineDraw();
- /// Given a texture handle, returns the TSC and TIC entries.
- Texture::FullTextureInfo GetTextureInfo(Texture::TextureHandle tex_handle) const;
-
- /// Returns the texture information for a specific texture in a specific shader stage.
- Texture::FullTextureInfo GetStageTexture(ShaderType stage, std::size_t offset) const;
-
u32 AccessConstBuffer32(ShaderType stage, u64 const_buffer, u64 offset) const override;
SamplerDescriptor AccessBoundSampler(ShaderType stage, u64 offset) const override;
@@ -1473,39 +1493,6 @@ private:
void ProcessMethodCall(u32 method, u32 argument, u32 nonshadow_argument, bool is_last_call);
- Core::System& system;
- MemoryManager& memory_manager;
-
- VideoCore::RasterizerInterface* rasterizer = nullptr;
-
- /// Start offsets of each macro in macro_memory
- std::array<u32, 0x80> macro_positions = {};
-
- std::array<bool, Regs::NUM_REGS> mme_inline{};
-
- /// Macro method that is currently being executed / being fed parameters.
- u32 executing_macro = 0;
- /// Parameters that have been submitted to the macro call so far.
- std::vector<u32> macro_params;
-
- /// Interpreter for the macro codes uploaded to the GPU.
- std::unique_ptr<MacroEngine> macro_engine;
-
- static constexpr u32 null_cb_data = 0xFFFFFFFF;
- struct {
- std::array<std::array<u32, 0x4000>, 16> buffer;
- u32 current{null_cb_data};
- u32 id{null_cb_data};
- u32 start_pos{};
- u32 counter{};
- } cb_data_state;
-
- Upload::State upload_state;
-
- bool execute_on{true};
-
- std::array<u8, Regs::NUM_REGS> dirty_pointers{};
-
/// Retrieves information about a specific TIC entry from the TIC buffer.
Texture::TICEntry GetTICEntry(u32 tic_index) const;
@@ -1514,8 +1501,8 @@ private:
/**
* Call a macro on this engine.
+ *
* @param method Method to call
- * @param num_parameters Number of arguments
* @param parameters Arguments to the method call
*/
void CallMacroMethod(u32 method, const std::vector<u32>& parameters);
@@ -1564,6 +1551,38 @@ private:
/// Returns a query's value or an empty object if the value will be deferred through a cache.
std::optional<u64> GetQueryResult();
+
+ Core::System& system;
+ MemoryManager& memory_manager;
+
+ VideoCore::RasterizerInterface* rasterizer = nullptr;
+
+ /// Start offsets of each macro in macro_memory
+ std::array<u32, 0x80> macro_positions{};
+
+ std::array<bool, Regs::NUM_REGS> mme_inline{};
+
+ /// Macro method that is currently being executed / being fed parameters.
+ u32 executing_macro = 0;
+ /// Parameters that have been submitted to the macro call so far.
+ std::vector<u32> macro_params;
+
+ /// Interpreter for the macro codes uploaded to the GPU.
+ std::unique_ptr<MacroEngine> macro_engine;
+
+ static constexpr u32 null_cb_data = 0xFFFFFFFF;
+ struct CBDataState {
+ std::array<std::array<u32, 0x4000>, 16> buffer;
+ u32 current{null_cb_data};
+ u32 id{null_cb_data};
+ u32 start_pos{};
+ u32 counter{};
+ };
+ CBDataState cb_data_state;
+
+ Upload::State upload_state;
+
+ bool execute_on{true};
};
#define ASSERT_REG_POSITION(field_name, position) \
@@ -1599,10 +1618,13 @@ ASSERT_REG_POSITION(polygon_offset_point_enable, 0x370);
ASSERT_REG_POSITION(polygon_offset_line_enable, 0x371);
ASSERT_REG_POSITION(polygon_offset_fill_enable, 0x372);
ASSERT_REG_POSITION(patch_vertices, 0x373);
+ASSERT_REG_POSITION(fragment_barrier, 0x378);
ASSERT_REG_POSITION(scissor_test, 0x380);
ASSERT_REG_POSITION(stencil_back_func_ref, 0x3D5);
ASSERT_REG_POSITION(stencil_back_mask, 0x3D6);
ASSERT_REG_POSITION(stencil_back_func_mask, 0x3D7);
+ASSERT_REG_POSITION(invalidate_texture_data_cache, 0x3DD);
+ASSERT_REG_POSITION(tiled_cache_barrier, 0x3DF);
ASSERT_REG_POSITION(color_mask_common, 0x3E4);
ASSERT_REG_POSITION(depth_bounds, 0x3E7);
ASSERT_REG_POSITION(rt_separate_frag_data, 0x3EB);
@@ -1610,6 +1632,7 @@ ASSERT_REG_POSITION(multisample_raster_enable, 0x3ED);
ASSERT_REG_POSITION(multisample_raster_samples, 0x3EE);
ASSERT_REG_POSITION(multisample_sample_mask, 0x3EF);
ASSERT_REG_POSITION(zeta, 0x3F8);
+ASSERT_REG_POSITION(render_area, 0x3FD);
ASSERT_REG_POSITION(clear_flags, 0x43E);
ASSERT_REG_POSITION(fill_rectangle, 0x44F);
ASSERT_REG_POSITION(vertex_attrib_format, 0x458);
@@ -1618,7 +1641,8 @@ ASSERT_REG_POSITION(multisample_coverage_to_color, 0x47E);
ASSERT_REG_POSITION(rt_control, 0x487);
ASSERT_REG_POSITION(zeta_width, 0x48a);
ASSERT_REG_POSITION(zeta_height, 0x48b);
-ASSERT_REG_POSITION(zeta_layers, 0x48c);
+ASSERT_REG_POSITION(zeta_depth, 0x48c);
+ASSERT_REG_POSITION(sampler_index, 0x48D);
ASSERT_REG_POSITION(depth_test_enable, 0x4B3);
ASSERT_REG_POSITION(independent_blend_enable, 0x4B9);
ASSERT_REG_POSITION(depth_write_enabled, 0x4BA);
@@ -1642,6 +1666,8 @@ ASSERT_REG_POSITION(frag_color_clamp, 0x4EA);
ASSERT_REG_POSITION(screen_y_control, 0x4EB);
ASSERT_REG_POSITION(line_width_smooth, 0x4EC);
ASSERT_REG_POSITION(line_width_aliased, 0x4ED);
+ASSERT_REG_POSITION(invalidate_sampler_cache_no_wfi, 0x509);
+ASSERT_REG_POSITION(invalidate_texture_header_cache_no_wfi, 0x50A);
ASSERT_REG_POSITION(vb_element_base, 0x50D);
ASSERT_REG_POSITION(vb_base_instance, 0x50E);
ASSERT_REG_POSITION(clip_distance_enabled, 0x544);
diff --git a/src/video_core/engines/maxwell_dma.cpp b/src/video_core/engines/maxwell_dma.cpp
index 8fa359d0a..ba750748c 100644
--- a/src/video_core/engines/maxwell_dma.cpp
+++ b/src/video_core/engines/maxwell_dma.cpp
@@ -16,8 +16,10 @@ namespace Tegra::Engines {
using namespace Texture;
-MaxwellDMA::MaxwellDMA(Core::System& system, MemoryManager& memory_manager)
- : system{system}, memory_manager{memory_manager} {}
+MaxwellDMA::MaxwellDMA(Core::System& system_, MemoryManager& memory_manager_)
+ : system{system_}, memory_manager{memory_manager_} {}
+
+MaxwellDMA::~MaxwellDMA() = default;
void MaxwellDMA::CallMethod(u32 method, u32 method_argument, bool is_last_call) {
ASSERT_MSG(method < NUM_REGS, "Invalid MaxwellDMA register");
@@ -94,6 +96,7 @@ void MaxwellDMA::CopyPitchToPitch() {
}
void MaxwellDMA::CopyBlockLinearToPitch() {
+ UNIMPLEMENTED_IF(regs.src_params.block_size.width != 0);
UNIMPLEMENTED_IF(regs.src_params.block_size.depth != 0);
UNIMPLEMENTED_IF(regs.src_params.layer != 0);
@@ -133,6 +136,8 @@ void MaxwellDMA::CopyBlockLinearToPitch() {
}
void MaxwellDMA::CopyPitchToBlockLinear() {
+ UNIMPLEMENTED_IF_MSG(regs.dst_params.block_size.width != 0, "Block width is not one");
+
const auto& dst_params = regs.dst_params;
const u32 bytes_per_pixel = regs.pitch_in / regs.line_length_in;
const u32 width = dst_params.width;
diff --git a/src/video_core/engines/maxwell_dma.h b/src/video_core/engines/maxwell_dma.h
index 50f445efc..3c59eeb13 100644
--- a/src/video_core/engines/maxwell_dma.h
+++ b/src/video_core/engines/maxwell_dma.h
@@ -72,11 +72,13 @@ public:
struct RenderEnable {
enum class Mode : u32 {
- FALSE = 0,
- TRUE = 1,
- CONDITIONAL = 2,
- RENDER_IF_EQUAL = 3,
- RENDER_IF_NOT_EQUAL = 4,
+ // Note: This uses Pascal case in order to avoid the identifiers
+ // FALSE and TRUE, which are reserved on Darwin.
+ False = 0,
+ True = 1,
+ Conditional = 2,
+ RenderIfEqual = 3,
+ RenderIfNotEqual = 4,
};
PackedGPUVAddr address;
@@ -185,8 +187,8 @@ public:
};
static_assert(sizeof(RemapConst) == 12);
- explicit MaxwellDMA(Core::System& system, MemoryManager& memory_manager);
- ~MaxwellDMA() = default;
+ explicit MaxwellDMA(Core::System& system_, MemoryManager& memory_manager_);
+ ~MaxwellDMA();
/// Write the value to the register identified by method.
void CallMethod(u32 method, u32 method_argument, bool is_last_call) override;
diff --git a/src/video_core/engines/shader_bytecode.h b/src/video_core/engines/shader_bytecode.h
index 37d17efdc..8b45f1b62 100644
--- a/src/video_core/engines/shader_bytecode.h
+++ b/src/video_core/engines/shader_bytecode.h
@@ -1437,8 +1437,7 @@ union Instruction {
return TextureType::TextureCube;
}
- LOG_CRITICAL(HW_GPU, "Unhandled texture_info: {}",
- static_cast<u32>(texture_info.Value()));
+ LOG_CRITICAL(HW_GPU, "Unhandled texture_info: {}", texture_info.Value());
UNREACHABLE();
return TextureType::Texture1D;
}
@@ -1533,8 +1532,7 @@ union Instruction {
return TextureType::Texture3D;
}
- LOG_CRITICAL(HW_GPU, "Unhandled texture_info: {}",
- static_cast<u32>(texture_info.Value()));
+ LOG_CRITICAL(HW_GPU, "Unhandled texture_info: {}", texture_info.Value());
UNREACHABLE();
return TextureType::Texture1D;
}
diff --git a/src/video_core/fence_manager.h b/src/video_core/fence_manager.h
index de6991ef6..3512283ff 100644
--- a/src/video_core/fence_manager.h
+++ b/src/video_core/fence_manager.h
@@ -9,6 +9,7 @@
#include "common/common_types.h"
#include "core/core.h"
+#include "video_core/delayed_destruction_ring.h"
#include "video_core/gpu.h"
#include "video_core/memory_manager.h"
#include "video_core/rasterizer_interface.h"
@@ -17,11 +18,11 @@ namespace VideoCommon {
class FenceBase {
public:
- FenceBase(u32 payload, bool is_stubbed)
- : address{}, payload{payload}, is_semaphore{false}, is_stubbed{is_stubbed} {}
+ explicit FenceBase(u32 payload_, bool is_stubbed_)
+ : address{}, payload{payload_}, is_semaphore{false}, is_stubbed{is_stubbed_} {}
- FenceBase(GPUVAddr address, u32 payload, bool is_stubbed)
- : address{address}, payload{payload}, is_semaphore{true}, is_stubbed{is_stubbed} {}
+ explicit FenceBase(GPUVAddr address_, u32 payload_, bool is_stubbed_)
+ : address{address_}, payload{payload_}, is_semaphore{true}, is_stubbed{is_stubbed_} {}
GPUVAddr GetAddress() const {
return address;
@@ -47,6 +48,11 @@ protected:
template <typename TFence, typename TTextureCache, typename TTBufferCache, typename TQueryCache>
class FenceManager {
public:
+ /// Notify the fence manager about a new frame
+ void TickFrame() {
+ delayed_destruction_ring.Tick();
+ }
+
void SignalSemaphore(GPUVAddr addr, u32 value) {
TryReleasePendingFences();
const bool should_flush = ShouldFlush();
@@ -86,7 +92,7 @@ public:
} else {
gpu.IncrementSyncPoint(current_fence->GetPayload());
}
- fences.pop();
+ PopFence();
}
}
@@ -132,7 +138,7 @@ private:
} else {
gpu.IncrementSyncPoint(current_fence->GetPayload());
}
- fences.pop();
+ PopFence();
}
}
@@ -158,7 +164,14 @@ private:
query_cache.CommitAsyncFlushes();
}
+ void PopFence() {
+ delayed_destruction_ring.Push(std::move(fences.front()));
+ fences.pop();
+ }
+
std::queue<TFence> fences;
+
+ DelayedDestructionRing<TFence, 6> delayed_destruction_ring;
};
} // namespace VideoCommon
diff --git a/src/video_core/framebuffer_config.h b/src/video_core/framebuffer_config.h
new file mode 100644
index 000000000..b86c3a757
--- /dev/null
+++ b/src/video_core/framebuffer_config.h
@@ -0,0 +1,31 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+namespace Tegra {
+
+/**
+ * Struct describing framebuffer configuration
+ */
+struct FramebufferConfig {
+ enum class PixelFormat : u32 {
+ A8B8G8R8_UNORM = 1,
+ RGB565_UNORM = 4,
+ B8G8R8A8_UNORM = 5,
+ };
+
+ VAddr address{};
+ u32 offset{};
+ u32 width{};
+ u32 height{};
+ u32 stride{};
+ PixelFormat pixel_format{};
+
+ using TransformFlags = Service::NVFlinger::BufferQueue::BufferTransformFlags;
+ TransformFlags transform_flags{};
+ Common::Rectangle<int> crop_rect;
+};
+
+} // namespace Tegra
diff --git a/src/video_core/gpu.cpp b/src/video_core/gpu.cpp
index e91f52938..6ab06775f 100644
--- a/src/video_core/gpu.cpp
+++ b/src/video_core/gpu.cpp
@@ -10,6 +10,7 @@
#include "core/core_timing.h"
#include "core/core_timing_util.h"
#include "core/frontend/emu_window.h"
+#include "core/hardware_interrupt_manager.h"
#include "core/memory.h"
#include "core/settings.h"
#include "video_core/engines/fermi_2d.h"
@@ -36,7 +37,8 @@ GPU::GPU(Core::System& system_, bool is_async_, bool use_nvdec_)
kepler_compute{std::make_unique<Engines::KeplerCompute>(system, *memory_manager)},
maxwell_dma{std::make_unique<Engines::MaxwellDMA>(system, *memory_manager)},
kepler_memory{std::make_unique<Engines::KeplerMemory>(system, *memory_manager)},
- shader_notify{std::make_unique<VideoCore::ShaderNotify>()}, is_async{is_async_} {}
+ shader_notify{std::make_unique<VideoCore::ShaderNotify>()}, is_async{is_async_},
+ gpu_thread{system_, is_async_} {}
GPU::~GPU() = default;
@@ -198,10 +200,6 @@ void GPU::SyncGuestHost() {
renderer->Rasterizer().SyncGuestHost();
}
-void GPU::OnCommandListEnd() {
- renderer->Rasterizer().ReleaseFences();
-}
-
enum class GpuSemaphoreOperation {
AcquireEqual = 0x1,
WriteLong = 0x2,
@@ -232,8 +230,12 @@ void GPU::CallMultiMethod(u32 method, u32 subchannel, const u32* base_start, u32
CallEngineMultiMethod(method, subchannel, base_start, amount, methods_pending);
} else {
for (std::size_t i = 0; i < amount; i++) {
- CallPullerMethod(
- {method, base_start[i], subchannel, methods_pending - static_cast<u32>(i)});
+ CallPullerMethod(MethodCall{
+ method,
+ base_start[i],
+ subchannel,
+ methods_pending - static_cast<u32>(i),
+ });
}
}
}
@@ -295,8 +297,7 @@ void GPU::CallPullerMethod(const MethodCall& method_call) {
break;
}
default:
- LOG_ERROR(HW_GPU, "Special puller engine method {:X} not implemented",
- static_cast<u32>(method));
+ LOG_ERROR(HW_GPU, "Special puller engine method {:X} not implemented", method);
break;
}
}
@@ -375,7 +376,7 @@ void GPU::ProcessBindMethod(const MethodCall& method_call) {
dma_pusher->BindSubchannel(kepler_memory.get(), method_call.subchannel);
break;
default:
- UNIMPLEMENTED_MSG("Unimplemented engine {:04X}", static_cast<u32>(engine_id));
+ UNIMPLEMENTED_MSG("Unimplemented engine {:04X}", engine_id);
}
}
@@ -388,8 +389,7 @@ void GPU::ProcessFenceActionMethod() {
IncrementSyncPoint(regs.fence_action.syncpoint_id);
break;
default:
- UNIMPLEMENTED_MSG("Unimplemented operation {}",
- static_cast<u32>(regs.fence_action.op.Value()));
+ UNIMPLEMENTED_MSG("Unimplemented operation {}", regs.fence_action.op.Value());
}
}
@@ -459,4 +459,75 @@ void GPU::ProcessSemaphoreAcquire() {
}
}
+void GPU::Start() {
+ gpu_thread.StartThread(*renderer, renderer->Context(), *dma_pusher, *cdma_pusher);
+ cpu_context = renderer->GetRenderWindow().CreateSharedContext();
+ cpu_context->MakeCurrent();
+}
+
+void GPU::ObtainContext() {
+ cpu_context->MakeCurrent();
+}
+
+void GPU::ReleaseContext() {
+ cpu_context->DoneCurrent();
+}
+
+void GPU::PushGPUEntries(Tegra::CommandList&& entries) {
+ gpu_thread.SubmitList(std::move(entries));
+}
+
+void GPU::PushCommandBuffer(Tegra::ChCommandHeaderList& entries) {
+ if (!use_nvdec) {
+ return;
+ }
+ // This condition fires when a video stream ends, clear all intermediary data
+ if (entries[0].raw == 0xDEADB33F) {
+ cdma_pusher.reset();
+ return;
+ }
+ if (!cdma_pusher) {
+ cdma_pusher = std::make_unique<Tegra::CDmaPusher>(*this);
+ }
+
+ // SubmitCommandBuffer would make the nvdec operations async, this is not currently working
+ // TODO(ameerj): RE proper async nvdec operation
+ // gpu_thread.SubmitCommandBuffer(std::move(entries));
+
+ cdma_pusher->Push(std::move(entries));
+ cdma_pusher->DispatchCalls();
+}
+
+void GPU::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
+ gpu_thread.SwapBuffers(framebuffer);
+}
+
+void GPU::FlushRegion(VAddr addr, u64 size) {
+ gpu_thread.FlushRegion(addr, size);
+}
+
+void GPU::InvalidateRegion(VAddr addr, u64 size) {
+ gpu_thread.InvalidateRegion(addr, size);
+}
+
+void GPU::FlushAndInvalidateRegion(VAddr addr, u64 size) {
+ gpu_thread.FlushAndInvalidateRegion(addr, size);
+}
+
+void GPU::TriggerCpuInterrupt(const u32 syncpoint_id, const u32 value) const {
+ auto& interrupt_manager = system.InterruptManager();
+ interrupt_manager.GPUInterruptSyncpt(syncpoint_id, value);
+}
+
+void GPU::WaitIdle() const {
+ gpu_thread.WaitIdle();
+}
+
+void GPU::OnCommandListEnd() {
+ if (is_async) {
+ // This command only applies to asynchronous GPU mode
+ gpu_thread.OnCommandListEnd();
+ }
+}
+
} // namespace Tegra
diff --git a/src/video_core/gpu.h b/src/video_core/gpu.h
index 21410e125..d81e38680 100644
--- a/src/video_core/gpu.h
+++ b/src/video_core/gpu.h
@@ -15,6 +15,8 @@
#include "core/hle/service/nvflinger/buffer_queue.h"
#include "video_core/cdma_pusher.h"
#include "video_core/dma_pusher.h"
+#include "video_core/framebuffer_config.h"
+#include "video_core/gpu_thread.h"
using CacheAddr = std::uintptr_t;
[[nodiscard]] inline CacheAddr ToCacheAddr(const void* host_ptr) {
@@ -101,28 +103,6 @@ enum class DepthFormat : u32 {
struct CommandListHeader;
class DebugContext;
-/**
- * Struct describing framebuffer configuration
- */
-struct FramebufferConfig {
- enum class PixelFormat : u32 {
- A8B8G8R8_UNORM = 1,
- RGB565_UNORM = 4,
- B8G8R8A8_UNORM = 5,
- };
-
- VAddr address;
- u32 offset;
- u32 width;
- u32 height;
- u32 stride;
- PixelFormat pixel_format;
-
- using TransformFlags = Service::NVFlinger::BufferQueue::BufferTransformFlags;
- TransformFlags transform_flags;
- Common::Rectangle<int> crop_rect;
-};
-
namespace Engines {
class Fermi2D;
class Maxwell3D;
@@ -141,7 +121,7 @@ enum class EngineID {
class MemoryManager;
-class GPU {
+class GPU final {
public:
struct MethodCall {
u32 method{};
@@ -149,17 +129,17 @@ public:
u32 subchannel{};
u32 method_count{};
- MethodCall(u32 method, u32 argument, u32 subchannel = 0, u32 method_count = 0)
- : method(method), argument(argument), subchannel(subchannel),
- method_count(method_count) {}
+ explicit MethodCall(u32 method_, u32 argument_, u32 subchannel_ = 0, u32 method_count_ = 0)
+ : method(method_), argument(argument_), subchannel(subchannel_),
+ method_count(method_count_) {}
[[nodiscard]] bool IsLastCall() const {
return method_count <= 1;
}
};
- explicit GPU(Core::System& system, bool is_async, bool use_nvdec);
- virtual ~GPU();
+ explicit GPU(Core::System& system_, bool is_async_, bool use_nvdec_);
+ ~GPU();
/// Binds a renderer to the GPU.
void BindRenderer(std::unique_ptr<VideoCore::RendererBase> renderer);
@@ -176,7 +156,7 @@ public:
/// Synchronizes CPU writes with Host GPU memory.
void SyncGuestHost();
/// Signal the ending of command list.
- virtual void OnCommandListEnd();
+ void OnCommandListEnd();
/// Request a host GPU memory flush from the CPU.
[[nodiscard]] u64 RequestFlush(VAddr addr, std::size_t size);
@@ -240,7 +220,7 @@ public:
}
// Waits for the GPU to finish working
- virtual void WaitIdle() const = 0;
+ void WaitIdle() const;
/// Allows the CPU/NvFlinger to wait on the GPU before presenting a frame.
void WaitFence(u32 syncpoint_id, u32 value);
@@ -330,34 +310,34 @@ public:
/// Performs any additional setup necessary in order to begin GPU emulation.
/// This can be used to launch any necessary threads and register any necessary
/// core timing events.
- virtual void Start() = 0;
+ void Start();
/// Obtain the CPU Context
- virtual void ObtainContext() = 0;
+ void ObtainContext();
/// Release the CPU Context
- virtual void ReleaseContext() = 0;
+ void ReleaseContext();
/// Push GPU command entries to be processed
- virtual void PushGPUEntries(Tegra::CommandList&& entries) = 0;
+ void PushGPUEntries(Tegra::CommandList&& entries);
/// Push GPU command buffer entries to be processed
- virtual void PushCommandBuffer(Tegra::ChCommandHeaderList& entries) = 0;
+ void PushCommandBuffer(Tegra::ChCommandHeaderList& entries);
/// Swap buffers (render frame)
- virtual void SwapBuffers(const Tegra::FramebufferConfig* framebuffer) = 0;
+ void SwapBuffers(const Tegra::FramebufferConfig* framebuffer);
/// Notify rasterizer that any caches of the specified region should be flushed to Switch memory
- virtual void FlushRegion(VAddr addr, u64 size) = 0;
+ void FlushRegion(VAddr addr, u64 size);
/// Notify rasterizer that any caches of the specified region should be invalidated
- virtual void InvalidateRegion(VAddr addr, u64 size) = 0;
+ void InvalidateRegion(VAddr addr, u64 size);
/// Notify rasterizer that any caches of the specified region should be flushed and invalidated
- virtual void FlushAndInvalidateRegion(VAddr addr, u64 size) = 0;
+ void FlushAndInvalidateRegion(VAddr addr, u64 size);
protected:
- virtual void TriggerCpuInterrupt(u32 syncpoint_id, u32 value) const = 0;
+ void TriggerCpuInterrupt(u32 syncpoint_id, u32 value) const;
private:
void ProcessBindMethod(const MethodCall& method_call);
@@ -414,8 +394,8 @@ private:
std::condition_variable sync_cv;
struct FlushRequest {
- FlushRequest(u64 fence, VAddr addr, std::size_t size)
- : fence{fence}, addr{addr}, size{size} {}
+ explicit FlushRequest(u64 fence_, VAddr addr_, std::size_t size_)
+ : fence{fence_}, addr{addr_}, size{size_} {}
u64 fence;
VAddr addr;
std::size_t size;
@@ -427,6 +407,9 @@ private:
std::mutex flush_request_mutex;
const bool is_async;
+
+ VideoCommon::GPUThread::ThreadManager gpu_thread;
+ std::unique_ptr<Core::Frontend::GraphicsContext> cpu_context;
};
#define ASSERT_REG_POSITION(field_name, position) \
diff --git a/src/video_core/gpu_asynch.cpp b/src/video_core/gpu_asynch.cpp
deleted file mode 100644
index a9baaf7ef..000000000
--- a/src/video_core/gpu_asynch.cpp
+++ /dev/null
@@ -1,86 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include "core/core.h"
-#include "core/hardware_interrupt_manager.h"
-#include "video_core/gpu_asynch.h"
-#include "video_core/gpu_thread.h"
-#include "video_core/renderer_base.h"
-
-namespace VideoCommon {
-
-GPUAsynch::GPUAsynch(Core::System& system, bool use_nvdec)
- : GPU{system, true, use_nvdec}, gpu_thread{system} {}
-
-GPUAsynch::~GPUAsynch() = default;
-
-void GPUAsynch::Start() {
- gpu_thread.StartThread(*renderer, renderer->Context(), *dma_pusher, *cdma_pusher);
- cpu_context = renderer->GetRenderWindow().CreateSharedContext();
- cpu_context->MakeCurrent();
-}
-
-void GPUAsynch::ObtainContext() {
- cpu_context->MakeCurrent();
-}
-
-void GPUAsynch::ReleaseContext() {
- cpu_context->DoneCurrent();
-}
-
-void GPUAsynch::PushGPUEntries(Tegra::CommandList&& entries) {
- gpu_thread.SubmitList(std::move(entries));
-}
-
-void GPUAsynch::PushCommandBuffer(Tegra::ChCommandHeaderList& entries) {
- if (!use_nvdec) {
- return;
- }
- // This condition fires when a video stream ends, clear all intermediary data
- if (entries[0].raw == 0xDEADB33F) {
- cdma_pusher.reset();
- return;
- }
- if (!cdma_pusher) {
- cdma_pusher = std::make_unique<Tegra::CDmaPusher>(*this);
- }
-
- // SubmitCommandBuffer would make the nvdec operations async, this is not currently working
- // TODO(ameerj): RE proper async nvdec operation
- // gpu_thread.SubmitCommandBuffer(std::move(entries));
-
- cdma_pusher->Push(std::move(entries));
- cdma_pusher->DispatchCalls();
-}
-
-void GPUAsynch::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
- gpu_thread.SwapBuffers(framebuffer);
-}
-
-void GPUAsynch::FlushRegion(VAddr addr, u64 size) {
- gpu_thread.FlushRegion(addr, size);
-}
-
-void GPUAsynch::InvalidateRegion(VAddr addr, u64 size) {
- gpu_thread.InvalidateRegion(addr, size);
-}
-
-void GPUAsynch::FlushAndInvalidateRegion(VAddr addr, u64 size) {
- gpu_thread.FlushAndInvalidateRegion(addr, size);
-}
-
-void GPUAsynch::TriggerCpuInterrupt(const u32 syncpoint_id, const u32 value) const {
- auto& interrupt_manager = system.InterruptManager();
- interrupt_manager.GPUInterruptSyncpt(syncpoint_id, value);
-}
-
-void GPUAsynch::WaitIdle() const {
- gpu_thread.WaitIdle();
-}
-
-void GPUAsynch::OnCommandListEnd() {
- gpu_thread.OnCommandListEnd();
-}
-
-} // namespace VideoCommon
diff --git a/src/video_core/gpu_asynch.h b/src/video_core/gpu_asynch.h
deleted file mode 100644
index 0c0872e73..000000000
--- a/src/video_core/gpu_asynch.h
+++ /dev/null
@@ -1,47 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include "video_core/gpu.h"
-#include "video_core/gpu_thread.h"
-
-namespace Core::Frontend {
-class GraphicsContext;
-}
-
-namespace VideoCore {
-class RendererBase;
-} // namespace VideoCore
-
-namespace VideoCommon {
-
-/// Implementation of GPU interface that runs the GPU asynchronously
-class GPUAsynch final : public Tegra::GPU {
-public:
- explicit GPUAsynch(Core::System& system, bool use_nvdec);
- ~GPUAsynch() override;
-
- void Start() override;
- void ObtainContext() override;
- void ReleaseContext() override;
- void PushGPUEntries(Tegra::CommandList&& entries) override;
- void PushCommandBuffer(Tegra::ChCommandHeaderList& entries) override;
- void SwapBuffers(const Tegra::FramebufferConfig* framebuffer) override;
- void FlushRegion(VAddr addr, u64 size) override;
- void InvalidateRegion(VAddr addr, u64 size) override;
- void FlushAndInvalidateRegion(VAddr addr, u64 size) override;
- void WaitIdle() const override;
-
- void OnCommandListEnd() override;
-
-protected:
- void TriggerCpuInterrupt(u32 syncpoint_id, u32 value) const override;
-
-private:
- GPUThread::ThreadManager gpu_thread;
- std::unique_ptr<Core::Frontend::GraphicsContext> cpu_context;
-};
-
-} // namespace VideoCommon
diff --git a/src/video_core/gpu_synch.cpp b/src/video_core/gpu_synch.cpp
deleted file mode 100644
index ecf7bbdf3..000000000
--- a/src/video_core/gpu_synch.cpp
+++ /dev/null
@@ -1,61 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include "video_core/gpu_synch.h"
-#include "video_core/renderer_base.h"
-
-namespace VideoCommon {
-
-GPUSynch::GPUSynch(Core::System& system, bool use_nvdec) : GPU{system, false, use_nvdec} {}
-
-GPUSynch::~GPUSynch() = default;
-
-void GPUSynch::Start() {}
-
-void GPUSynch::ObtainContext() {
- renderer->Context().MakeCurrent();
-}
-
-void GPUSynch::ReleaseContext() {
- renderer->Context().DoneCurrent();
-}
-
-void GPUSynch::PushGPUEntries(Tegra::CommandList&& entries) {
- dma_pusher->Push(std::move(entries));
- dma_pusher->DispatchCalls();
-}
-
-void GPUSynch::PushCommandBuffer(Tegra::ChCommandHeaderList& entries) {
- if (!use_nvdec) {
- return;
- }
- // This condition fires when a video stream ends, clears all intermediary data
- if (entries[0].raw == 0xDEADB33F) {
- cdma_pusher.reset();
- return;
- }
- if (!cdma_pusher) {
- cdma_pusher = std::make_unique<Tegra::CDmaPusher>(*this);
- }
- cdma_pusher->Push(std::move(entries));
- cdma_pusher->DispatchCalls();
-}
-
-void GPUSynch::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
- renderer->SwapBuffers(framebuffer);
-}
-
-void GPUSynch::FlushRegion(VAddr addr, u64 size) {
- renderer->Rasterizer().FlushRegion(addr, size);
-}
-
-void GPUSynch::InvalidateRegion(VAddr addr, u64 size) {
- renderer->Rasterizer().InvalidateRegion(addr, size);
-}
-
-void GPUSynch::FlushAndInvalidateRegion(VAddr addr, u64 size) {
- renderer->Rasterizer().FlushAndInvalidateRegion(addr, size);
-}
-
-} // namespace VideoCommon
diff --git a/src/video_core/gpu_synch.h b/src/video_core/gpu_synch.h
deleted file mode 100644
index 9d778c71a..000000000
--- a/src/video_core/gpu_synch.h
+++ /dev/null
@@ -1,41 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include "video_core/gpu.h"
-
-namespace Core::Frontend {
-class GraphicsContext;
-}
-
-namespace VideoCore {
-class RendererBase;
-} // namespace VideoCore
-
-namespace VideoCommon {
-
-/// Implementation of GPU interface that runs the GPU synchronously
-class GPUSynch final : public Tegra::GPU {
-public:
- explicit GPUSynch(Core::System& system, bool use_nvdec);
- ~GPUSynch() override;
-
- void Start() override;
- void ObtainContext() override;
- void ReleaseContext() override;
- void PushGPUEntries(Tegra::CommandList&& entries) override;
- void PushCommandBuffer(Tegra::ChCommandHeaderList& entries) override;
- void SwapBuffers(const Tegra::FramebufferConfig* framebuffer) override;
- void FlushRegion(VAddr addr, u64 size) override;
- void InvalidateRegion(VAddr addr, u64 size) override;
- void FlushAndInvalidateRegion(VAddr addr, u64 size) override;
- void WaitIdle() const override {}
-
-protected:
- void TriggerCpuInterrupt([[maybe_unused]] u32 syncpoint_id,
- [[maybe_unused]] u32 value) const override {}
-};
-
-} // namespace VideoCommon
diff --git a/src/video_core/gpu_thread.cpp b/src/video_core/gpu_thread.cpp
index 4b8f58283..7e490bcc3 100644
--- a/src/video_core/gpu_thread.cpp
+++ b/src/video_core/gpu_thread.cpp
@@ -4,6 +4,7 @@
#include "common/assert.h"
#include "common/microprofile.h"
+#include "common/scope_exit.h"
#include "common/thread.h"
#include "core/core.h"
#include "core/frontend/emu_window.h"
@@ -21,6 +22,8 @@ static void RunThread(Core::System& system, VideoCore::RendererBase& renderer,
SynchState& state, Tegra::CDmaPusher& cdma_pusher) {
std::string name = "yuzu:GPU";
MicroProfileOnThreadCreate(name.c_str());
+ SCOPE_EXIT({ MicroProfileOnThreadExit(); });
+
Common::SetCurrentThreadName(name.c_str());
Common::SetCurrentThreadPriority(Common::ThreadPriority::High);
system.RegisterHostThread();
@@ -39,23 +42,23 @@ static void RunThread(Core::System& system, VideoCore::RendererBase& renderer,
CommandDataContainer next;
while (state.is_running) {
next = state.queue.PopWait();
- if (const auto submit_list = std::get_if<SubmitListCommand>(&next.data)) {
+ if (auto* submit_list = std::get_if<SubmitListCommand>(&next.data)) {
dma_pusher.Push(std::move(submit_list->entries));
dma_pusher.DispatchCalls();
- } else if (const auto command_list = std::get_if<SubmitChCommandEntries>(&next.data)) {
+ } else if (auto* command_list = std::get_if<SubmitChCommandEntries>(&next.data)) {
// NVDEC
cdma_pusher.Push(std::move(command_list->entries));
cdma_pusher.DispatchCalls();
- } else if (const auto data = std::get_if<SwapBuffersCommand>(&next.data)) {
+ } else if (const auto* data = std::get_if<SwapBuffersCommand>(&next.data)) {
renderer.SwapBuffers(data->framebuffer ? &*data->framebuffer : nullptr);
} else if (std::holds_alternative<OnCommandListEndCommand>(next.data)) {
renderer.Rasterizer().ReleaseFences();
} else if (std::holds_alternative<GPUTickCommand>(next.data)) {
system.GPU().TickWork();
- } else if (const auto data = std::get_if<FlushRegionCommand>(&next.data)) {
- renderer.Rasterizer().FlushRegion(data->addr, data->size);
- } else if (const auto data = std::get_if<InvalidateRegionCommand>(&next.data)) {
- renderer.Rasterizer().OnCPUWrite(data->addr, data->size);
+ } else if (const auto* flush = std::get_if<FlushRegionCommand>(&next.data)) {
+ renderer.Rasterizer().FlushRegion(flush->addr, flush->size);
+ } else if (const auto* invalidate = std::get_if<InvalidateRegionCommand>(&next.data)) {
+ renderer.Rasterizer().OnCPUWrite(invalidate->addr, invalidate->size);
} else if (std::holds_alternative<EndProcessingCommand>(next.data)) {
return;
} else {
@@ -65,7 +68,8 @@ static void RunThread(Core::System& system, VideoCore::RendererBase& renderer,
}
}
-ThreadManager::ThreadManager(Core::System& system) : system{system} {}
+ThreadManager::ThreadManager(Core::System& system_, bool is_async_)
+ : system{system_}, is_async{is_async_} {}
ThreadManager::~ThreadManager() {
if (!thread.joinable()) {
@@ -97,19 +101,30 @@ void ThreadManager::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
}
void ThreadManager::FlushRegion(VAddr addr, u64 size) {
- if (!Settings::IsGPULevelHigh()) {
+ if (!is_async) {
+ // Always flush with synchronous GPU mode
PushCommand(FlushRegionCommand(addr, size));
return;
}
- if (!Settings::IsGPULevelExtreme()) {
- return;
- }
- if (system.Renderer().Rasterizer().MustFlushRegion(addr, size)) {
+
+ // Asynchronous GPU mode
+ switch (Settings::values.gpu_accuracy.GetValue()) {
+ case Settings::GPUAccuracy::Normal:
+ PushCommand(FlushRegionCommand(addr, size));
+ break;
+ case Settings::GPUAccuracy::High:
+ // TODO(bunnei): Is this right? Preserving existing behavior for now
+ break;
+ case Settings::GPUAccuracy::Extreme: {
auto& gpu = system.GPU();
u64 fence = gpu.RequestFlush(addr, size);
PushCommand(GPUTickCommand());
while (fence > gpu.CurrentFlushRequestFence()) {
}
+ break;
+ }
+ default:
+ UNIMPLEMENTED_MSG("Unsupported gpu_accuracy {}", Settings::values.gpu_accuracy.GetValue());
}
}
@@ -123,7 +138,8 @@ void ThreadManager::FlushAndInvalidateRegion(VAddr addr, u64 size) {
}
void ThreadManager::WaitIdle() const {
- while (state.last_fence > state.signaled_fence.load(std::memory_order_relaxed)) {
+ while (state.last_fence > state.signaled_fence.load(std::memory_order_relaxed) &&
+ system.IsPoweredOn()) {
}
}
@@ -134,6 +150,12 @@ void ThreadManager::OnCommandListEnd() {
u64 ThreadManager::PushCommand(CommandData&& command_data) {
const u64 fence{++state.last_fence};
state.queue.Push(CommandDataContainer(std::move(command_data), fence));
+
+ if (!is_async) {
+ // In synchronous GPU mode, block the caller until the command has executed
+ WaitIdle();
+ }
+
return fence;
}
diff --git a/src/video_core/gpu_thread.h b/src/video_core/gpu_thread.h
index 32a34e3a7..2775629e7 100644
--- a/src/video_core/gpu_thread.h
+++ b/src/video_core/gpu_thread.h
@@ -10,8 +10,9 @@
#include <optional>
#include <thread>
#include <variant>
+
#include "common/threadsafe_queue.h"
-#include "video_core/gpu.h"
+#include "video_core/framebuffer_config.h"
namespace Tegra {
struct FramebufferConfig;
@@ -25,6 +26,10 @@ class GraphicsContext;
class System;
} // namespace Core
+namespace VideoCore {
+class RendererBase;
+} // namespace VideoCore
+
namespace VideoCommon::GPUThread {
/// Command to signal to the GPU thread that processing has ended
@@ -32,30 +37,30 @@ struct EndProcessingCommand final {};
/// Command to signal to the GPU thread that a command list is ready for processing
struct SubmitListCommand final {
- explicit SubmitListCommand(Tegra::CommandList&& entries) : entries{std::move(entries)} {}
+ explicit SubmitListCommand(Tegra::CommandList&& entries_) : entries{std::move(entries_)} {}
Tegra::CommandList entries;
};
/// Command to signal to the GPU thread that a cdma command list is ready for processing
struct SubmitChCommandEntries final {
- explicit SubmitChCommandEntries(Tegra::ChCommandHeaderList&& entries)
- : entries{std::move(entries)} {}
+ explicit SubmitChCommandEntries(Tegra::ChCommandHeaderList&& entries_)
+ : entries{std::move(entries_)} {}
Tegra::ChCommandHeaderList entries;
};
/// Command to signal to the GPU thread that a swap buffers is pending
struct SwapBuffersCommand final {
- explicit SwapBuffersCommand(std::optional<const Tegra::FramebufferConfig> framebuffer)
- : framebuffer{std::move(framebuffer)} {}
+ explicit SwapBuffersCommand(std::optional<const Tegra::FramebufferConfig> framebuffer_)
+ : framebuffer{std::move(framebuffer_)} {}
std::optional<Tegra::FramebufferConfig> framebuffer;
};
/// Command to signal to the GPU thread to flush a region
struct FlushRegionCommand final {
- explicit constexpr FlushRegionCommand(VAddr addr, u64 size) : addr{addr}, size{size} {}
+ explicit constexpr FlushRegionCommand(VAddr addr_, u64 size_) : addr{addr_}, size{size_} {}
VAddr addr;
u64 size;
@@ -63,7 +68,7 @@ struct FlushRegionCommand final {
/// Command to signal to the GPU thread to invalidate a region
struct InvalidateRegionCommand final {
- explicit constexpr InvalidateRegionCommand(VAddr addr, u64 size) : addr{addr}, size{size} {}
+ explicit constexpr InvalidateRegionCommand(VAddr addr_, u64 size_) : addr{addr_}, size{size_} {}
VAddr addr;
u64 size;
@@ -71,8 +76,8 @@ struct InvalidateRegionCommand final {
/// Command to signal to the GPU thread to flush and invalidate a region
struct FlushAndInvalidateRegionCommand final {
- explicit constexpr FlushAndInvalidateRegionCommand(VAddr addr, u64 size)
- : addr{addr}, size{size} {}
+ explicit constexpr FlushAndInvalidateRegionCommand(VAddr addr_, u64 size_)
+ : addr{addr_}, size{size_} {}
VAddr addr;
u64 size;
@@ -92,8 +97,8 @@ using CommandData =
struct CommandDataContainer {
CommandDataContainer() = default;
- CommandDataContainer(CommandData&& data, u64 next_fence)
- : data{std::move(data)}, fence{next_fence} {}
+ explicit CommandDataContainer(CommandData&& data_, u64 next_fence_)
+ : data{std::move(data_)}, fence{next_fence_} {}
CommandData data;
u64 fence{};
@@ -112,7 +117,7 @@ struct SynchState final {
/// Class used to manage the GPU thread
class ThreadManager final {
public:
- explicit ThreadManager(Core::System& system);
+ explicit ThreadManager(Core::System& system_, bool is_async_);
~ThreadManager();
/// Creates and starts the GPU thread.
@@ -146,11 +151,11 @@ private:
/// Pushes a command to be executed by the GPU thread
u64 PushCommand(CommandData&& command_data);
-private:
SynchState state;
Core::System& system;
std::thread thread;
std::thread::id thread_id;
+ const bool is_async;
};
} // namespace VideoCommon::GPUThread
diff --git a/src/video_core/guest_driver.h b/src/video_core/guest_driver.h
index 99450777e..21e569ba1 100644
--- a/src/video_core/guest_driver.h
+++ b/src/video_core/guest_driver.h
@@ -19,8 +19,8 @@ namespace VideoCore {
class GuestDriverProfile {
public:
explicit GuestDriverProfile() = default;
- explicit GuestDriverProfile(std::optional<u32> texture_handler_size)
- : texture_handler_size{texture_handler_size} {}
+ explicit GuestDriverProfile(std::optional<u32> texture_handler_size_)
+ : texture_handler_size{texture_handler_size_} {}
void DeduceTextureHandlerSize(std::vector<u32> bound_offsets);
diff --git a/src/video_core/host_shaders/CMakeLists.txt b/src/video_core/host_shaders/CMakeLists.txt
index c157724a9..4c7399d5a 100644
--- a/src/video_core/host_shaders/CMakeLists.txt
+++ b/src/video_core/host_shaders/CMakeLists.txt
@@ -1,8 +1,26 @@
-set(SHADER_SOURCES
+set(SHADER_FILES
+ block_linear_unswizzle_2d.comp
+ block_linear_unswizzle_3d.comp
+ convert_depth_to_float.frag
+ convert_float_to_depth.frag
+ full_screen_triangle.vert
+ opengl_copy_bc4.comp
opengl_present.frag
opengl_present.vert
+ pitch_unswizzle.comp
+ vulkan_blit_color_float.frag
+ vulkan_blit_depth_stencil.frag
+ vulkan_present.frag
+ vulkan_present.vert
+ vulkan_quad_array.comp
+ vulkan_quad_indexed.comp
+ vulkan_uint8.comp
)
+find_program(GLSLANGVALIDATOR "glslangValidator" REQUIRED)
+
+set(GLSL_FLAGS "")
+
set(SHADER_INCLUDE ${CMAKE_CURRENT_BINARY_DIR}/include)
set(SHADER_DIR ${SHADER_INCLUDE}/video_core/host_shaders)
set(HOST_SHADERS_INCLUDE ${SHADER_INCLUDE} PARENT_SCOPE)
@@ -10,27 +28,44 @@ set(HOST_SHADERS_INCLUDE ${SHADER_INCLUDE} PARENT_SCOPE)
set(INPUT_FILE ${CMAKE_CURRENT_SOURCE_DIR}/source_shader.h.in)
set(HEADER_GENERATOR ${CMAKE_CURRENT_SOURCE_DIR}/StringShaderHeader.cmake)
-foreach(FILENAME IN ITEMS ${SHADER_SOURCES})
+foreach(FILENAME IN ITEMS ${SHADER_FILES})
string(REPLACE "." "_" SHADER_NAME ${FILENAME})
set(SOURCE_FILE ${CMAKE_CURRENT_SOURCE_DIR}/${FILENAME})
- set(HEADER_FILE ${SHADER_DIR}/${SHADER_NAME}.h)
- add_custom_command(
- OUTPUT
- ${HEADER_FILE}
- COMMAND
- ${CMAKE_COMMAND} -P ${HEADER_GENERATOR} ${SOURCE_FILE} ${HEADER_FILE} ${INPUT_FILE}
- MAIN_DEPENDENCY
- ${SOURCE_FILE}
- DEPENDS
- ${INPUT_FILE}
- # HEADER_GENERATOR should be included here but msbuild seems to assume it's always modified
- )
- set(SHADER_HEADERS ${SHADER_HEADERS} ${HEADER_FILE})
+ # Skip generating source headers on Vulkan exclusive files
+ if (NOT ${FILENAME} MATCHES "vulkan.*")
+ set(SOURCE_HEADER_FILE ${SHADER_DIR}/${SHADER_NAME}.h)
+ add_custom_command(
+ OUTPUT
+ ${SOURCE_HEADER_FILE}
+ COMMAND
+ ${CMAKE_COMMAND} -P ${HEADER_GENERATOR} ${SOURCE_FILE} ${SOURCE_HEADER_FILE} ${INPUT_FILE}
+ MAIN_DEPENDENCY
+ ${SOURCE_FILE}
+ DEPENDS
+ ${INPUT_FILE}
+ # HEADER_GENERATOR should be included here but msbuild seems to assume it's always modified
+ )
+ set(SHADER_HEADERS ${SHADER_HEADERS} ${SOURCE_HEADER_FILE})
+ endif()
+ # Skip compiling to SPIR-V OpenGL exclusive files
+ if (NOT ${FILENAME} MATCHES "opengl.*")
+ string(TOUPPER ${SHADER_NAME}_SPV SPIRV_VARIABLE_NAME)
+ set(SPIRV_HEADER_FILE ${SHADER_DIR}/${SHADER_NAME}_spv.h)
+ add_custom_command(
+ OUTPUT
+ ${SPIRV_HEADER_FILE}
+ COMMAND
+ ${GLSLANGVALIDATOR} -V ${GLSL_FLAGS} --variable-name ${SPIRV_VARIABLE_NAME} -o ${SPIRV_HEADER_FILE} ${SOURCE_FILE}
+ MAIN_DEPENDENCY
+ ${SOURCE_FILE}
+ )
+ set(SHADER_HEADERS ${SHADER_HEADERS} ${SPIRV_HEADER_FILE})
+ endif()
endforeach()
add_custom_target(host_shaders
DEPENDS
${SHADER_HEADERS}
SOURCES
- ${SHADER_SOURCES}
+ ${SHADER_FILES}
)
diff --git a/src/video_core/host_shaders/block_linear_unswizzle_2d.comp b/src/video_core/host_shaders/block_linear_unswizzle_2d.comp
new file mode 100644
index 000000000..a131be79e
--- /dev/null
+++ b/src/video_core/host_shaders/block_linear_unswizzle_2d.comp
@@ -0,0 +1,122 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#version 430
+
+#ifdef VULKAN
+
+#extension GL_EXT_shader_16bit_storage : require
+#extension GL_EXT_shader_8bit_storage : require
+#define HAS_EXTENDED_TYPES 1
+#define BEGIN_PUSH_CONSTANTS layout(push_constant) uniform PushConstants {
+#define END_PUSH_CONSTANTS };
+#define UNIFORM(n)
+#define BINDING_SWIZZLE_BUFFER 0
+#define BINDING_INPUT_BUFFER 1
+#define BINDING_OUTPUT_IMAGE 2
+
+#else // ^^^ Vulkan ^^^ // vvv OpenGL vvv
+
+#extension GL_NV_gpu_shader5 : enable
+#ifdef GL_NV_gpu_shader5
+#define HAS_EXTENDED_TYPES 1
+#else
+#define HAS_EXTENDED_TYPES 0
+#endif
+#define BEGIN_PUSH_CONSTANTS
+#define END_PUSH_CONSTANTS
+#define UNIFORM(n) layout (location = n) uniform
+#define BINDING_SWIZZLE_BUFFER 0
+#define BINDING_INPUT_BUFFER 1
+#define BINDING_OUTPUT_IMAGE 0
+
+#endif
+
+BEGIN_PUSH_CONSTANTS
+UNIFORM(0) uvec3 origin;
+UNIFORM(1) ivec3 destination;
+UNIFORM(2) uint bytes_per_block_log2;
+UNIFORM(3) uint layer_stride;
+UNIFORM(4) uint block_size;
+UNIFORM(5) uint x_shift;
+UNIFORM(6) uint block_height;
+UNIFORM(7) uint block_height_mask;
+END_PUSH_CONSTANTS
+
+layout(binding = BINDING_SWIZZLE_BUFFER, std430) readonly buffer SwizzleTable {
+ uint swizzle_table[];
+};
+
+#if HAS_EXTENDED_TYPES
+layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU8 { uint8_t u8data[]; };
+layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU16 { uint16_t u16data[]; };
+#endif
+layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU32 { uint u32data[]; };
+layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU64 { uvec2 u64data[]; };
+layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU128 { uvec4 u128data[]; };
+
+layout(binding = BINDING_OUTPUT_IMAGE) uniform writeonly uimage2DArray output_image;
+
+layout(local_size_x = 32, local_size_y = 32, local_size_z = 1) in;
+
+const uint GOB_SIZE_X = 64;
+const uint GOB_SIZE_Y = 8;
+const uint GOB_SIZE_Z = 1;
+const uint GOB_SIZE = GOB_SIZE_X * GOB_SIZE_Y * GOB_SIZE_Z;
+
+const uint GOB_SIZE_X_SHIFT = 6;
+const uint GOB_SIZE_Y_SHIFT = 3;
+const uint GOB_SIZE_Z_SHIFT = 0;
+const uint GOB_SIZE_SHIFT = GOB_SIZE_X_SHIFT + GOB_SIZE_Y_SHIFT + GOB_SIZE_Z_SHIFT;
+
+const uvec2 SWIZZLE_MASK = uvec2(GOB_SIZE_X - 1, GOB_SIZE_Y - 1);
+
+uint SwizzleOffset(uvec2 pos) {
+ pos = pos & SWIZZLE_MASK;
+ return swizzle_table[pos.y * 64 + pos.x];
+}
+
+uvec4 ReadTexel(uint offset) {
+ switch (bytes_per_block_log2) {
+#if HAS_EXTENDED_TYPES
+ case 0:
+ return uvec4(u8data[offset], 0, 0, 0);
+ case 1:
+ return uvec4(u16data[offset / 2], 0, 0, 0);
+#else
+ case 0:
+ return uvec4(bitfieldExtract(u32data[offset / 4], int((offset * 8) & 24), 8), 0, 0, 0);
+ case 1:
+ return uvec4(bitfieldExtract(u32data[offset / 4], int((offset * 8) & 16), 16), 0, 0, 0);
+#endif
+ case 2:
+ return uvec4(u32data[offset / 4], 0, 0, 0);
+ case 3:
+ return uvec4(u64data[offset / 8], 0, 0);
+ case 4:
+ return u128data[offset / 16];
+ }
+ return uvec4(0);
+}
+
+void main() {
+ uvec3 pos = gl_GlobalInvocationID + origin;
+ pos.x <<= bytes_per_block_log2;
+
+ // Read as soon as possible due to its latency
+ const uint swizzle = SwizzleOffset(pos.xy);
+
+ const uint block_y = pos.y >> GOB_SIZE_Y_SHIFT;
+
+ uint offset = 0;
+ offset += pos.z * layer_stride;
+ offset += (block_y >> block_height) * block_size;
+ offset += (block_y & block_height_mask) << GOB_SIZE_SHIFT;
+ offset += (pos.x >> GOB_SIZE_X_SHIFT) << x_shift;
+ offset += swizzle;
+
+ const uvec4 texel = ReadTexel(offset);
+ const ivec3 coord = ivec3(gl_GlobalInvocationID) + destination;
+ imageStore(output_image, coord, texel);
+}
diff --git a/src/video_core/host_shaders/block_linear_unswizzle_3d.comp b/src/video_core/host_shaders/block_linear_unswizzle_3d.comp
new file mode 100644
index 000000000..bb6872e6b
--- /dev/null
+++ b/src/video_core/host_shaders/block_linear_unswizzle_3d.comp
@@ -0,0 +1,125 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#version 430
+
+#ifdef VULKAN
+
+#extension GL_EXT_shader_16bit_storage : require
+#extension GL_EXT_shader_8bit_storage : require
+#define HAS_EXTENDED_TYPES 1
+#define BEGIN_PUSH_CONSTANTS layout(push_constant) uniform PushConstants {
+#define END_PUSH_CONSTANTS };
+#define UNIFORM(n)
+#define BINDING_SWIZZLE_BUFFER 0
+#define BINDING_INPUT_BUFFER 1
+#define BINDING_OUTPUT_IMAGE 2
+
+#else // ^^^ Vulkan ^^^ // vvv OpenGL vvv
+
+#extension GL_NV_gpu_shader5 : enable
+#ifdef GL_NV_gpu_shader5
+#define HAS_EXTENDED_TYPES 1
+#else
+#define HAS_EXTENDED_TYPES 0
+#endif
+#define BEGIN_PUSH_CONSTANTS
+#define END_PUSH_CONSTANTS
+#define UNIFORM(n) layout (location = n) uniform
+#define BINDING_SWIZZLE_BUFFER 0
+#define BINDING_INPUT_BUFFER 1
+#define BINDING_OUTPUT_IMAGE 0
+
+#endif
+
+BEGIN_PUSH_CONSTANTS
+UNIFORM(0) uvec3 origin;
+UNIFORM(1) ivec3 destination;
+UNIFORM(2) uint bytes_per_block_log2;
+UNIFORM(3) uint slice_size;
+UNIFORM(4) uint block_size;
+UNIFORM(5) uint x_shift;
+UNIFORM(6) uint block_height;
+UNIFORM(7) uint block_height_mask;
+UNIFORM(8) uint block_depth;
+UNIFORM(9) uint block_depth_mask;
+END_PUSH_CONSTANTS
+
+layout(binding = BINDING_SWIZZLE_BUFFER, std430) readonly buffer SwizzleTable {
+ uint swizzle_table[];
+};
+
+#if HAS_EXTENDED_TYPES
+layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU8 { uint8_t u8data[]; };
+layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU16 { uint16_t u16data[]; };
+#endif
+layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU32 { uint u32data[]; };
+layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU64 { uvec2 u64data[]; };
+layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU128 { uvec4 u128data[]; };
+
+layout(binding = BINDING_OUTPUT_IMAGE) uniform writeonly uimage3D output_image;
+
+layout(local_size_x = 16, local_size_y = 8, local_size_z = 8) in;
+
+const uint GOB_SIZE_X = 64;
+const uint GOB_SIZE_Y = 8;
+const uint GOB_SIZE_Z = 1;
+const uint GOB_SIZE = GOB_SIZE_X * GOB_SIZE_Y * GOB_SIZE_Z;
+
+const uint GOB_SIZE_X_SHIFT = 6;
+const uint GOB_SIZE_Y_SHIFT = 3;
+const uint GOB_SIZE_Z_SHIFT = 0;
+const uint GOB_SIZE_SHIFT = GOB_SIZE_X_SHIFT + GOB_SIZE_Y_SHIFT + GOB_SIZE_Z_SHIFT;
+
+const uvec2 SWIZZLE_MASK = uvec2(GOB_SIZE_X - 1, GOB_SIZE_Y - 1);
+
+uint SwizzleOffset(uvec2 pos) {
+ pos = pos & SWIZZLE_MASK;
+ return swizzle_table[pos.y * 64 + pos.x];
+}
+
+uvec4 ReadTexel(uint offset) {
+ switch (bytes_per_block_log2) {
+#if HAS_EXTENDED_TYPES
+ case 0:
+ return uvec4(u8data[offset], 0, 0, 0);
+ case 1:
+ return uvec4(u16data[offset / 2], 0, 0, 0);
+#else
+ case 0:
+ return uvec4(bitfieldExtract(u32data[offset / 4], int((offset * 8) & 24), 8), 0, 0, 0);
+ case 1:
+ return uvec4(bitfieldExtract(u32data[offset / 4], int((offset * 8) & 16), 16), 0, 0, 0);
+#endif
+ case 2:
+ return uvec4(u32data[offset / 4], 0, 0, 0);
+ case 3:
+ return uvec4(u64data[offset / 8], 0, 0);
+ case 4:
+ return u128data[offset / 16];
+ }
+ return uvec4(0);
+}
+
+void main() {
+ uvec3 pos = gl_GlobalInvocationID + origin;
+ pos.x <<= bytes_per_block_log2;
+
+ // Read as soon as possible due to its latency
+ const uint swizzle = SwizzleOffset(pos.xy);
+
+ const uint block_y = pos.y >> GOB_SIZE_Y_SHIFT;
+
+ uint offset = 0;
+ offset += (pos.z >> block_depth) * slice_size;
+ offset += (pos.z & block_depth_mask) << (GOB_SIZE_SHIFT + block_height);
+ offset += (block_y >> block_height) * block_size;
+ offset += (block_y & block_height_mask) << GOB_SIZE_SHIFT;
+ offset += (pos.x >> GOB_SIZE_X_SHIFT) << x_shift;
+ offset += swizzle;
+
+ const uvec4 texel = ReadTexel(offset);
+ const ivec3 coord = ivec3(gl_GlobalInvocationID) + destination;
+ imageStore(output_image, coord, texel);
+}
diff --git a/src/video_core/host_shaders/convert_depth_to_float.frag b/src/video_core/host_shaders/convert_depth_to_float.frag
new file mode 100644
index 000000000..624c58509
--- /dev/null
+++ b/src/video_core/host_shaders/convert_depth_to_float.frag
@@ -0,0 +1,13 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#version 450
+
+layout(binding = 0) uniform sampler2D depth_texture;
+layout(location = 0) out float output_color;
+
+void main() {
+ ivec2 coord = ivec2(gl_FragCoord.xy);
+ output_color = texelFetch(depth_texture, coord, 0).r;
+}
diff --git a/src/video_core/host_shaders/convert_float_to_depth.frag b/src/video_core/host_shaders/convert_float_to_depth.frag
new file mode 100644
index 000000000..d86c795f4
--- /dev/null
+++ b/src/video_core/host_shaders/convert_float_to_depth.frag
@@ -0,0 +1,13 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#version 450
+
+layout(binding = 0) uniform sampler2D color_texture;
+
+void main() {
+ ivec2 coord = ivec2(gl_FragCoord.xy);
+ float color = texelFetch(color_texture, coord, 0).r;
+ gl_FragDepth = color;
+}
diff --git a/src/video_core/host_shaders/full_screen_triangle.vert b/src/video_core/host_shaders/full_screen_triangle.vert
new file mode 100644
index 000000000..452ad6502
--- /dev/null
+++ b/src/video_core/host_shaders/full_screen_triangle.vert
@@ -0,0 +1,29 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#version 450
+
+#ifdef VULKAN
+#define BEGIN_PUSH_CONSTANTS layout(push_constant) uniform PushConstants {
+#define END_PUSH_CONSTANTS };
+#define UNIFORM(n)
+#else // ^^^ Vulkan ^^^ // vvv OpenGL vvv
+#define BEGIN_PUSH_CONSTANTS
+#define END_PUSH_CONSTANTS
+#define UNIFORM(n) layout (location = n) uniform
+#endif
+
+BEGIN_PUSH_CONSTANTS
+UNIFORM(0) vec2 tex_scale;
+UNIFORM(1) vec2 tex_offset;
+END_PUSH_CONSTANTS
+
+layout(location = 0) out vec2 texcoord;
+
+void main() {
+ float x = float((gl_VertexIndex & 1) << 2);
+ float y = float((gl_VertexIndex & 2) << 1);
+ gl_Position = vec4(x - 1.0, y - 1.0, 0.0, 1.0);
+ texcoord = fma(vec2(x, y) / 2.0, tex_scale, tex_offset);
+}
diff --git a/src/video_core/host_shaders/opengl_copy_bc4.comp b/src/video_core/host_shaders/opengl_copy_bc4.comp
new file mode 100644
index 000000000..7b8e20fbe
--- /dev/null
+++ b/src/video_core/host_shaders/opengl_copy_bc4.comp
@@ -0,0 +1,70 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#version 430 core
+#extension GL_ARB_gpu_shader_int64 : require
+
+layout (local_size_x = 4, local_size_y = 4) in;
+
+layout(binding = 0, rg32ui) readonly uniform uimage3D bc4_input;
+layout(binding = 1, rgba8ui) writeonly uniform uimage3D bc4_output;
+
+layout(location = 0) uniform uvec3 src_offset;
+layout(location = 1) uniform uvec3 dst_offset;
+
+// https://www.khronos.org/registry/OpenGL/extensions/ARB/ARB_texture_compression_rgtc.txt
+uint DecompressBlock(uint64_t bits, uvec2 coord) {
+ const uint code_offset = 16 + 3 * (4 * coord.y + coord.x);
+ const uint code = uint(bits >> code_offset) & 7;
+ const uint red0 = uint(bits >> 0) & 0xff;
+ const uint red1 = uint(bits >> 8) & 0xff;
+ if (red0 > red1) {
+ switch (code) {
+ case 0:
+ return red0;
+ case 1:
+ return red1;
+ case 2:
+ return (6 * red0 + 1 * red1) / 7;
+ case 3:
+ return (5 * red0 + 2 * red1) / 7;
+ case 4:
+ return (4 * red0 + 3 * red1) / 7;
+ case 5:
+ return (3 * red0 + 4 * red1) / 7;
+ case 6:
+ return (2 * red0 + 5 * red1) / 7;
+ case 7:
+ return (1 * red0 + 6 * red1) / 7;
+ }
+ } else {
+ switch (code) {
+ case 0:
+ return red0;
+ case 1:
+ return red1;
+ case 2:
+ return (4 * red0 + 1 * red1) / 5;
+ case 3:
+ return (3 * red0 + 2 * red1) / 5;
+ case 4:
+ return (2 * red0 + 3 * red1) / 5;
+ case 5:
+ return (1 * red0 + 4 * red1) / 5;
+ case 6:
+ return 0;
+ case 7:
+ return 0xff;
+ }
+ }
+ return 0;
+}
+
+void main() {
+ uvec2 packed_bits = imageLoad(bc4_input, ivec3(gl_WorkGroupID + src_offset)).rg;
+ uint64_t bits = packUint2x32(packed_bits);
+ uint red = DecompressBlock(bits, gl_LocalInvocationID.xy);
+ uvec4 color = uvec4(red & 0xff, 0, 0, 0xff);
+ imageStore(bc4_output, ivec3(gl_GlobalInvocationID + dst_offset), color);
+}
diff --git a/src/video_core/host_shaders/opengl_present.frag b/src/video_core/host_shaders/opengl_present.frag
index 8a4cb024b..84b818227 100644
--- a/src/video_core/host_shaders/opengl_present.frag
+++ b/src/video_core/host_shaders/opengl_present.frag
@@ -1,3 +1,7 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
#version 430 core
layout (location = 0) in vec2 frag_tex_coord;
diff --git a/src/video_core/host_shaders/opengl_present.vert b/src/video_core/host_shaders/opengl_present.vert
index 2235d31a4..c3b5adbba 100644
--- a/src/video_core/host_shaders/opengl_present.vert
+++ b/src/video_core/host_shaders/opengl_present.vert
@@ -1,3 +1,7 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
#version 430 core
out gl_PerVertex {
diff --git a/src/video_core/host_shaders/pitch_unswizzle.comp b/src/video_core/host_shaders/pitch_unswizzle.comp
new file mode 100644
index 000000000..cb48ec170
--- /dev/null
+++ b/src/video_core/host_shaders/pitch_unswizzle.comp
@@ -0,0 +1,86 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#version 430
+
+#ifdef VULKAN
+
+#extension GL_EXT_shader_16bit_storage : require
+#extension GL_EXT_shader_8bit_storage : require
+#define HAS_EXTENDED_TYPES 1
+#define BEGIN_PUSH_CONSTANTS layout(push_constant) uniform PushConstants {
+#define END_PUSH_CONSTANTS };
+#define UNIFORM(n)
+#define BINDING_INPUT_BUFFER 0
+#define BINDING_OUTPUT_IMAGE 1
+
+#else // ^^^ Vulkan ^^^ // vvv OpenGL vvv
+
+#extension GL_NV_gpu_shader5 : enable
+#ifdef GL_NV_gpu_shader5
+#define HAS_EXTENDED_TYPES 1
+#else
+#define HAS_EXTENDED_TYPES 0
+#endif
+#define BEGIN_PUSH_CONSTANTS
+#define END_PUSH_CONSTANTS
+#define UNIFORM(n) layout (location = n) uniform
+#define BINDING_INPUT_BUFFER 0
+#define BINDING_OUTPUT_IMAGE 0
+
+#endif
+
+BEGIN_PUSH_CONSTANTS
+UNIFORM(0) uvec2 origin;
+UNIFORM(1) ivec2 destination;
+UNIFORM(2) uint bytes_per_block;
+UNIFORM(3) uint pitch;
+END_PUSH_CONSTANTS
+
+#if HAS_EXTENDED_TYPES
+layout(binding = BINDING_INPUT_BUFFER, std430) readonly buffer InputBufferU8 { uint8_t u8data[]; };
+layout(binding = BINDING_INPUT_BUFFER, std430) readonly buffer InputBufferU16 { uint16_t u16data[]; };
+#endif
+layout(binding = BINDING_INPUT_BUFFER, std430) readonly buffer InputBufferU32 { uint u32data[]; };
+layout(binding = BINDING_INPUT_BUFFER, std430) readonly buffer InputBufferU64 { uvec2 u64data[]; };
+layout(binding = BINDING_INPUT_BUFFER, std430) readonly buffer InputBufferU128 { uvec4 u128data[]; };
+
+layout(binding = BINDING_OUTPUT_IMAGE) writeonly uniform uimage2D output_image;
+
+layout(local_size_x = 32, local_size_y = 32, local_size_z = 1) in;
+
+uvec4 ReadTexel(uint offset) {
+ switch (bytes_per_block) {
+#if HAS_EXTENDED_TYPES
+ case 1:
+ return uvec4(u8data[offset], 0, 0, 0);
+ case 2:
+ return uvec4(u16data[offset / 2], 0, 0, 0);
+#else
+ case 1:
+ return uvec4(bitfieldExtract(u32data[offset / 4], int((offset * 8) & 24), 8), 0, 0, 0);
+ case 2:
+ return uvec4(bitfieldExtract(u32data[offset / 4], int((offset * 8) & 16), 16), 0, 0, 0);
+#endif
+ case 4:
+ return uvec4(u32data[offset / 4], 0, 0, 0);
+ case 8:
+ return uvec4(u64data[offset / 8], 0, 0);
+ case 16:
+ return u128data[offset / 16];
+ }
+ return uvec4(0);
+}
+
+void main() {
+ uvec2 pos = gl_GlobalInvocationID.xy + origin;
+
+ uint offset = 0;
+ offset += pos.x * bytes_per_block;
+ offset += pos.y * pitch;
+
+ const uvec4 texel = ReadTexel(offset);
+ const ivec2 coord = ivec2(gl_GlobalInvocationID.xy) + destination;
+ imageStore(output_image, coord, texel);
+}
diff --git a/src/video_core/host_shaders/vulkan_blit_color_float.frag b/src/video_core/host_shaders/vulkan_blit_color_float.frag
new file mode 100644
index 000000000..4a6aae410
--- /dev/null
+++ b/src/video_core/host_shaders/vulkan_blit_color_float.frag
@@ -0,0 +1,14 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#version 450
+
+layout(binding = 0) uniform sampler2D tex;
+
+layout(location = 0) in vec2 texcoord;
+layout(location = 0) out vec4 color;
+
+void main() {
+ color = textureLod(tex, texcoord, 0);
+}
diff --git a/src/video_core/host_shaders/vulkan_blit_depth_stencil.frag b/src/video_core/host_shaders/vulkan_blit_depth_stencil.frag
new file mode 100644
index 000000000..19bb23a5a
--- /dev/null
+++ b/src/video_core/host_shaders/vulkan_blit_depth_stencil.frag
@@ -0,0 +1,16 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#version 450
+#extension GL_ARB_shader_stencil_export : require
+
+layout(binding = 0) uniform sampler2D depth_tex;
+layout(binding = 1) uniform isampler2D stencil_tex;
+
+layout(location = 0) in vec2 texcoord;
+
+void main() {
+ gl_FragDepth = textureLod(depth_tex, texcoord, 0).r;
+ gl_FragStencilRefARB = textureLod(stencil_tex, texcoord, 0).r;
+}
diff --git a/src/video_core/renderer_vulkan/shaders/blit.frag b/src/video_core/host_shaders/vulkan_present.frag
index a06ecd24a..0979ff3e6 100644
--- a/src/video_core/renderer_vulkan/shaders/blit.frag
+++ b/src/video_core/host_shaders/vulkan_present.frag
@@ -2,15 +2,6 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
-/*
- * Build instructions:
- * $ glslangValidator -V $THIS_FILE -o output.spv
- * $ spirv-opt -O --strip-debug output.spv -o optimized.spv
- * $ xxd -i optimized.spv
- *
- * Then copy that bytecode to the C++ file
- */
-
#version 460 core
layout (location = 0) in vec2 frag_tex_coord;
diff --git a/src/video_core/renderer_vulkan/shaders/blit.vert b/src/video_core/host_shaders/vulkan_present.vert
index c64d9235a..00b868958 100644
--- a/src/video_core/renderer_vulkan/shaders/blit.vert
+++ b/src/video_core/host_shaders/vulkan_present.vert
@@ -2,15 +2,6 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
-/*
- * Build instructions:
- * $ glslangValidator -V $THIS_FILE -o output.spv
- * $ spirv-opt -O --strip-debug output.spv -o optimized.spv
- * $ xxd -i optimized.spv
- *
- * Then copy that bytecode to the C++ file
- */
-
#version 460 core
layout (location = 0) in vec2 vert_position;
diff --git a/src/video_core/renderer_vulkan/shaders/quad_array.comp b/src/video_core/host_shaders/vulkan_quad_array.comp
index 5a5703308..212f4e998 100644
--- a/src/video_core/renderer_vulkan/shaders/quad_array.comp
+++ b/src/video_core/host_shaders/vulkan_quad_array.comp
@@ -2,15 +2,6 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
-/*
- * Build instructions:
- * $ glslangValidator -V $THIS_FILE -o output.spv
- * $ spirv-opt -O --strip-debug output.spv -o optimized.spv
- * $ xxd -i optimized.spv
- *
- * Then copy that bytecode to the C++ file
- */
-
#version 460 core
layout (local_size_x = 1024) in;
diff --git a/src/video_core/renderer_vulkan/shaders/quad_indexed.comp b/src/video_core/host_shaders/vulkan_quad_indexed.comp
index 5a472ba9b..8655591d0 100644
--- a/src/video_core/renderer_vulkan/shaders/quad_indexed.comp
+++ b/src/video_core/host_shaders/vulkan_quad_indexed.comp
@@ -2,15 +2,6 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
-/*
- * Build instructions:
- * $ glslangValidator -V quad_indexed.comp -o output.spv
- * $ spirv-opt -O --strip-debug output.spv -o optimized.spv
- * $ xxd -i optimized.spv
- *
- * Then copy that bytecode to the C++ file
- */
-
#version 460 core
layout (local_size_x = 1024) in;
diff --git a/src/video_core/renderer_vulkan/shaders/uint8.comp b/src/video_core/host_shaders/vulkan_uint8.comp
index a320f3ae0..ad74d7af9 100644
--- a/src/video_core/renderer_vulkan/shaders/uint8.comp
+++ b/src/video_core/host_shaders/vulkan_uint8.comp
@@ -2,15 +2,6 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
-/*
- * Build instructions:
- * $ glslangValidator -V $THIS_FILE -o output.spv
- * $ spirv-opt -O --strip-debug output.spv -o optimized.spv
- * $ xxd -i optimized.spv
- *
- * Then copy that bytecode to the C++ file
- */
-
#version 460 core
#extension GL_EXT_shader_16bit_storage : require
#extension GL_EXT_shader_8bit_storage : require
diff --git a/src/video_core/macro/macro_hle.cpp b/src/video_core/macro/macro_hle.cpp
index df00b57df..70ac7c620 100644
--- a/src/video_core/macro/macro_hle.cpp
+++ b/src/video_core/macro/macro_hle.cpp
@@ -85,7 +85,7 @@ constexpr std::array<std::pair<u64, HLEFunction>, 3> hle_funcs{{
{0x0217920100488FF7, &HLE_0217920100488FF7},
}};
-HLEMacro::HLEMacro(Engines::Maxwell3D& maxwell3d) : maxwell3d(maxwell3d) {}
+HLEMacro::HLEMacro(Engines::Maxwell3D& maxwell3d_) : maxwell3d{maxwell3d_} {}
HLEMacro::~HLEMacro() = default;
std::optional<std::unique_ptr<CachedMacro>> HLEMacro::GetHLEProgram(u64 hash) const {
@@ -99,8 +99,8 @@ std::optional<std::unique_ptr<CachedMacro>> HLEMacro::GetHLEProgram(u64 hash) co
HLEMacroImpl::~HLEMacroImpl() = default;
-HLEMacroImpl::HLEMacroImpl(Engines::Maxwell3D& maxwell3d, HLEFunction func)
- : maxwell3d(maxwell3d), func(func) {}
+HLEMacroImpl::HLEMacroImpl(Engines::Maxwell3D& maxwell3d_, HLEFunction func_)
+ : maxwell3d{maxwell3d_}, func{func_} {}
void HLEMacroImpl::Execute(const std::vector<u32>& parameters, u32 method) {
func(maxwell3d, parameters);
diff --git a/src/video_core/macro/macro_hle.h b/src/video_core/macro/macro_hle.h
index 37af875a0..cb3bd1600 100644
--- a/src/video_core/macro/macro_hle.h
+++ b/src/video_core/macro/macro_hle.h
@@ -20,7 +20,7 @@ using HLEFunction = void (*)(Engines::Maxwell3D& maxwell3d, const std::vector<u3
class HLEMacro {
public:
- explicit HLEMacro(Engines::Maxwell3D& maxwell3d);
+ explicit HLEMacro(Engines::Maxwell3D& maxwell3d_);
~HLEMacro();
std::optional<std::unique_ptr<CachedMacro>> GetHLEProgram(u64 hash) const;
diff --git a/src/video_core/macro/macro_interpreter.cpp b/src/video_core/macro/macro_interpreter.cpp
index bd01fd1f2..8da26fd59 100644
--- a/src/video_core/macro/macro_interpreter.cpp
+++ b/src/video_core/macro/macro_interpreter.cpp
@@ -11,29 +11,29 @@
MICROPROFILE_DEFINE(MacroInterp, "GPU", "Execute macro interpreter", MP_RGB(128, 128, 192));
namespace Tegra {
-MacroInterpreter::MacroInterpreter(Engines::Maxwell3D& maxwell3d)
- : MacroEngine::MacroEngine(maxwell3d), maxwell3d(maxwell3d) {}
+MacroInterpreter::MacroInterpreter(Engines::Maxwell3D& maxwell3d_)
+ : MacroEngine{maxwell3d_}, maxwell3d{maxwell3d_} {}
std::unique_ptr<CachedMacro> MacroInterpreter::Compile(const std::vector<u32>& code) {
return std::make_unique<MacroInterpreterImpl>(maxwell3d, code);
}
-MacroInterpreterImpl::MacroInterpreterImpl(Engines::Maxwell3D& maxwell3d,
- const std::vector<u32>& code)
- : maxwell3d(maxwell3d), code(code) {}
+MacroInterpreterImpl::MacroInterpreterImpl(Engines::Maxwell3D& maxwell3d_,
+ const std::vector<u32>& code_)
+ : maxwell3d{maxwell3d_}, code{code_} {}
-void MacroInterpreterImpl::Execute(const std::vector<u32>& parameters, u32 method) {
+void MacroInterpreterImpl::Execute(const std::vector<u32>& params, u32 method) {
MICROPROFILE_SCOPE(MacroInterp);
Reset();
- registers[1] = parameters[0];
- num_parameters = parameters.size();
+ registers[1] = params[0];
+ num_parameters = params.size();
if (num_parameters > parameters_capacity) {
parameters_capacity = num_parameters;
- this->parameters = std::make_unique<u32[]>(num_parameters);
+ parameters = std::make_unique<u32[]>(num_parameters);
}
- std::memcpy(this->parameters.get(), parameters.data(), num_parameters * sizeof(u32));
+ std::memcpy(parameters.get(), params.data(), num_parameters * sizeof(u32));
// Execute the code until we hit an exit condition.
bool keep_executing = true;
@@ -133,8 +133,7 @@ bool MacroInterpreterImpl::Step(bool is_delay_slot) {
break;
}
default:
- UNIMPLEMENTED_MSG("Unimplemented macro operation {}",
- static_cast<u32>(opcode.operation.Value()));
+ UNIMPLEMENTED_MSG("Unimplemented macro operation {}", opcode.operation.Value());
}
// An instruction with the Exit flag will not actually
@@ -182,7 +181,7 @@ u32 MacroInterpreterImpl::GetALUResult(Macro::ALUOperation operation, u32 src_a,
return ~(src_a & src_b);
default:
- UNIMPLEMENTED_MSG("Unimplemented ALU operation {}", static_cast<u32>(operation));
+ UNIMPLEMENTED_MSG("Unimplemented ALU operation {}", operation);
return 0;
}
}
@@ -230,7 +229,7 @@ void MacroInterpreterImpl::ProcessResult(Macro::ResultOperation operation, u32 r
Send((result >> 12) & 0b111111);
break;
default:
- UNIMPLEMENTED_MSG("Unimplemented result operation {}", static_cast<u32>(operation));
+ UNIMPLEMENTED_MSG("Unimplemented result operation {}", operation);
}
}
diff --git a/src/video_core/macro/macro_interpreter.h b/src/video_core/macro/macro_interpreter.h
index 90217fc89..d50c619ce 100644
--- a/src/video_core/macro/macro_interpreter.h
+++ b/src/video_core/macro/macro_interpreter.h
@@ -17,7 +17,7 @@ class Maxwell3D;
class MacroInterpreter final : public MacroEngine {
public:
- explicit MacroInterpreter(Engines::Maxwell3D& maxwell3d);
+ explicit MacroInterpreter(Engines::Maxwell3D& maxwell3d_);
protected:
std::unique_ptr<CachedMacro> Compile(const std::vector<u32>& code) override;
@@ -28,8 +28,8 @@ private:
class MacroInterpreterImpl : public CachedMacro {
public:
- MacroInterpreterImpl(Engines::Maxwell3D& maxwell3d, const std::vector<u32>& code);
- void Execute(const std::vector<u32>& parameters, u32 method) override;
+ explicit MacroInterpreterImpl(Engines::Maxwell3D& maxwell3d_, const std::vector<u32>& code_);
+ void Execute(const std::vector<u32>& params, u32 method) override;
private:
/// Resets the execution engine state, zeroing registers, etc.
@@ -38,9 +38,9 @@ private:
/**
* Executes a single macro instruction located at the current program counter. Returns whether
* the interpreter should keep running.
- * @param offset Offset to start execution at.
+ *
* @param is_delay_slot Whether the current step is being executed due to a delay slot in a
- * previous instruction.
+ * previous instruction.
*/
bool Step(bool is_delay_slot);
diff --git a/src/video_core/macro/macro_jit_x64.cpp b/src/video_core/macro/macro_jit_x64.cpp
index 954b87515..c6b2b2109 100644
--- a/src/video_core/macro/macro_jit_x64.cpp
+++ b/src/video_core/macro/macro_jit_x64.cpp
@@ -28,15 +28,15 @@ static const std::bitset<32> PERSISTENT_REGISTERS = Common::X64::BuildRegSet({
BRANCH_HOLDER,
});
-MacroJITx64::MacroJITx64(Engines::Maxwell3D& maxwell3d)
- : MacroEngine::MacroEngine(maxwell3d), maxwell3d(maxwell3d) {}
+MacroJITx64::MacroJITx64(Engines::Maxwell3D& maxwell3d_)
+ : MacroEngine{maxwell3d_}, maxwell3d{maxwell3d_} {}
std::unique_ptr<CachedMacro> MacroJITx64::Compile(const std::vector<u32>& code) {
return std::make_unique<MacroJITx64Impl>(maxwell3d, code);
}
-MacroJITx64Impl::MacroJITx64Impl(Engines::Maxwell3D& maxwell3d, const std::vector<u32>& code)
- : Xbyak::CodeGenerator(MAX_CODE_SIZE), code(code), maxwell3d(maxwell3d) {
+MacroJITx64Impl::MacroJITx64Impl(Engines::Maxwell3D& maxwell3d_, const std::vector<u32>& code_)
+ : CodeGenerator{MAX_CODE_SIZE}, code{code_}, maxwell3d{maxwell3d_} {
Compile();
}
@@ -165,8 +165,7 @@ void MacroJITx64Impl::Compile_ALU(Macro::Opcode opcode) {
}
break;
default:
- UNIMPLEMENTED_MSG("Unimplemented ALU operation {}",
- static_cast<std::size_t>(opcode.alu_operation.Value()));
+ UNIMPLEMENTED_MSG("Unimplemented ALU operation {}", opcode.alu_operation.Value());
break;
}
Compile_ProcessResult(opcode.result_operation, opcode.dst);
@@ -553,15 +552,15 @@ Xbyak::Reg32 MacroJITx64Impl::Compile_GetRegister(u32 index, Xbyak::Reg32 dst) {
}
void MacroJITx64Impl::Compile_ProcessResult(Macro::ResultOperation operation, u32 reg) {
- const auto SetRegister = [this](u32 reg, const Xbyak::Reg32& result) {
+ const auto SetRegister = [this](u32 reg_index, const Xbyak::Reg32& result) {
// Register 0 is supposed to always return 0. NOP is implemented as a store to the zero
// register.
- if (reg == 0) {
+ if (reg_index == 0) {
return;
}
- mov(dword[STATE + offsetof(JITState, registers) + reg * sizeof(u32)], result);
+ mov(dword[STATE + offsetof(JITState, registers) + reg_index * sizeof(u32)], result);
};
- const auto SetMethodAddress = [this](const Xbyak::Reg32& reg) { mov(METHOD_ADDRESS, reg); };
+ const auto SetMethodAddress = [this](const Xbyak::Reg32& reg32) { mov(METHOD_ADDRESS, reg32); };
switch (operation) {
case Macro::ResultOperation::IgnoreAndFetch:
@@ -604,7 +603,7 @@ void MacroJITx64Impl::Compile_ProcessResult(Macro::ResultOperation operation, u3
Compile_Send(RESULT);
break;
default:
- UNIMPLEMENTED_MSG("Unimplemented macro operation {}", static_cast<std::size_t>(operation));
+ UNIMPLEMENTED_MSG("Unimplemented macro operation {}", operation);
}
}
diff --git a/src/video_core/macro/macro_jit_x64.h b/src/video_core/macro/macro_jit_x64.h
index a180e7428..7f50ac2f8 100644
--- a/src/video_core/macro/macro_jit_x64.h
+++ b/src/video_core/macro/macro_jit_x64.h
@@ -23,7 +23,7 @@ constexpr size_t MAX_CODE_SIZE = 0x10000;
class MacroJITx64 final : public MacroEngine {
public:
- explicit MacroJITx64(Engines::Maxwell3D& maxwell3d);
+ explicit MacroJITx64(Engines::Maxwell3D& maxwell3d_);
protected:
std::unique_ptr<CachedMacro> Compile(const std::vector<u32>& code) override;
@@ -34,7 +34,7 @@ private:
class MacroJITx64Impl : public Xbyak::CodeGenerator, public CachedMacro {
public:
- MacroJITx64Impl(Engines::Maxwell3D& maxwell3d, const std::vector<u32>& code);
+ explicit MacroJITx64Impl(Engines::Maxwell3D& maxwell3d_, const std::vector<u32>& code_);
~MacroJITx64Impl();
void Execute(const std::vector<u32>& parameters, u32 method) override;
diff --git a/src/video_core/memory_manager.cpp b/src/video_core/memory_manager.cpp
index 6e70bd362..65feff588 100644
--- a/src/video_core/memory_manager.cpp
+++ b/src/video_core/memory_manager.cpp
@@ -57,7 +57,10 @@ void MemoryManager::Unmap(GPUVAddr gpu_addr, std::size_t size) {
}
// Flush and invalidate through the GPU interface, to be asynchronous if possible.
- system.GPU().FlushAndInvalidateRegion(*GpuToCpuAddress(gpu_addr), size);
+ const std::optional<VAddr> cpu_addr = GpuToCpuAddress(gpu_addr);
+ ASSERT(cpu_addr);
+
+ rasterizer->UnmapMemory(*cpu_addr, size);
UpdateRange(gpu_addr, PageEntry::State::Unmapped, size);
}
diff --git a/src/video_core/memory_manager.h b/src/video_core/memory_manager.h
index c078193d9..c35e57689 100644
--- a/src/video_core/memory_manager.h
+++ b/src/video_core/memory_manager.h
@@ -28,7 +28,7 @@ public:
};
constexpr PageEntry() = default;
- constexpr PageEntry(State state) : state{state} {}
+ constexpr PageEntry(State state_) : state{state_} {}
constexpr PageEntry(VAddr addr) : state{static_cast<State>(addr >> ShiftBits)} {}
[[nodiscard]] constexpr bool IsUnmapped() const {
@@ -68,7 +68,7 @@ static_assert(sizeof(PageEntry) == 4, "PageEntry is too large");
class MemoryManager final {
public:
- explicit MemoryManager(Core::System& system);
+ explicit MemoryManager(Core::System& system_);
~MemoryManager();
/// Binds a renderer to the memory manager.
diff --git a/src/video_core/morton.cpp b/src/video_core/morton.cpp
index 9da9fb4ff..e69de29bb 100644
--- a/src/video_core/morton.cpp
+++ b/src/video_core/morton.cpp
@@ -1,250 +0,0 @@
-// Copyright 2018 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include <array>
-#include <cstring>
-#include "common/assert.h"
-#include "common/common_types.h"
-#include "video_core/morton.h"
-#include "video_core/surface.h"
-#include "video_core/textures/decoders.h"
-
-namespace VideoCore {
-
-using Surface::GetBytesPerPixel;
-using Surface::PixelFormat;
-
-using MortonCopyFn = void (*)(u32, u32, u32, u32, u32, u32, u8*, u8*);
-using ConversionArray = std::array<MortonCopyFn, Surface::MaxPixelFormat>;
-
-template <bool morton_to_linear, PixelFormat format>
-static void MortonCopy(u32 stride, u32 block_height, u32 height, u32 block_depth, u32 depth,
- u32 tile_width_spacing, u8* buffer, u8* addr) {
- constexpr u32 bytes_per_pixel = GetBytesPerPixel(format);
-
- // With the BCn formats (DXT and DXN), each 4x4 tile is swizzled instead of just individual
- // pixel values.
- constexpr u32 tile_size_x{GetDefaultBlockWidth(format)};
- constexpr u32 tile_size_y{GetDefaultBlockHeight(format)};
-
- if constexpr (morton_to_linear) {
- Tegra::Texture::UnswizzleTexture(buffer, addr, tile_size_x, tile_size_y, bytes_per_pixel,
- stride, height, depth, block_height, block_depth,
- tile_width_spacing);
- } else {
- Tegra::Texture::CopySwizzledData((stride + tile_size_x - 1) / tile_size_x,
- (height + tile_size_y - 1) / tile_size_y, depth,
- bytes_per_pixel, bytes_per_pixel, addr, buffer, false,
- block_height, block_depth, tile_width_spacing);
- }
-}
-
-static constexpr ConversionArray morton_to_linear_fns = {
- MortonCopy<true, PixelFormat::A8B8G8R8_UNORM>,
- MortonCopy<true, PixelFormat::A8B8G8R8_SNORM>,
- MortonCopy<true, PixelFormat::A8B8G8R8_SINT>,
- MortonCopy<true, PixelFormat::A8B8G8R8_UINT>,
- MortonCopy<true, PixelFormat::R5G6B5_UNORM>,
- MortonCopy<true, PixelFormat::B5G6R5_UNORM>,
- MortonCopy<true, PixelFormat::A1R5G5B5_UNORM>,
- MortonCopy<true, PixelFormat::A2B10G10R10_UNORM>,
- MortonCopy<true, PixelFormat::A2B10G10R10_UINT>,
- MortonCopy<true, PixelFormat::A1B5G5R5_UNORM>,
- MortonCopy<true, PixelFormat::R8_UNORM>,
- MortonCopy<true, PixelFormat::R8_SNORM>,
- MortonCopy<true, PixelFormat::R8_SINT>,
- MortonCopy<true, PixelFormat::R8_UINT>,
- MortonCopy<true, PixelFormat::R16G16B16A16_FLOAT>,
- MortonCopy<true, PixelFormat::R16G16B16A16_UNORM>,
- MortonCopy<true, PixelFormat::R16G16B16A16_SNORM>,
- MortonCopy<true, PixelFormat::R16G16B16A16_SINT>,
- MortonCopy<true, PixelFormat::R16G16B16A16_UINT>,
- MortonCopy<true, PixelFormat::B10G11R11_FLOAT>,
- MortonCopy<true, PixelFormat::R32G32B32A32_UINT>,
- MortonCopy<true, PixelFormat::BC1_RGBA_UNORM>,
- MortonCopy<true, PixelFormat::BC2_UNORM>,
- MortonCopy<true, PixelFormat::BC3_UNORM>,
- MortonCopy<true, PixelFormat::BC4_UNORM>,
- MortonCopy<true, PixelFormat::BC4_SNORM>,
- MortonCopy<true, PixelFormat::BC5_UNORM>,
- MortonCopy<true, PixelFormat::BC5_SNORM>,
- MortonCopy<true, PixelFormat::BC7_UNORM>,
- MortonCopy<true, PixelFormat::BC6H_UFLOAT>,
- MortonCopy<true, PixelFormat::BC6H_SFLOAT>,
- MortonCopy<true, PixelFormat::ASTC_2D_4X4_UNORM>,
- MortonCopy<true, PixelFormat::B8G8R8A8_UNORM>,
- MortonCopy<true, PixelFormat::R32G32B32A32_FLOAT>,
- MortonCopy<true, PixelFormat::R32G32B32A32_SINT>,
- MortonCopy<true, PixelFormat::R32G32_FLOAT>,
- MortonCopy<true, PixelFormat::R32G32_SINT>,
- MortonCopy<true, PixelFormat::R32_FLOAT>,
- MortonCopy<true, PixelFormat::R16_FLOAT>,
- MortonCopy<true, PixelFormat::R16_UNORM>,
- MortonCopy<true, PixelFormat::R16_SNORM>,
- MortonCopy<true, PixelFormat::R16_UINT>,
- MortonCopy<true, PixelFormat::R16_SINT>,
- MortonCopy<true, PixelFormat::R16G16_UNORM>,
- MortonCopy<true, PixelFormat::R16G16_FLOAT>,
- MortonCopy<true, PixelFormat::R16G16_UINT>,
- MortonCopy<true, PixelFormat::R16G16_SINT>,
- MortonCopy<true, PixelFormat::R16G16_SNORM>,
- MortonCopy<true, PixelFormat::R32G32B32_FLOAT>,
- MortonCopy<true, PixelFormat::A8B8G8R8_SRGB>,
- MortonCopy<true, PixelFormat::R8G8_UNORM>,
- MortonCopy<true, PixelFormat::R8G8_SNORM>,
- MortonCopy<true, PixelFormat::R8G8_SINT>,
- MortonCopy<true, PixelFormat::R8G8_UINT>,
- MortonCopy<true, PixelFormat::R32G32_UINT>,
- MortonCopy<true, PixelFormat::R16G16B16X16_FLOAT>,
- MortonCopy<true, PixelFormat::R32_UINT>,
- MortonCopy<true, PixelFormat::R32_SINT>,
- MortonCopy<true, PixelFormat::ASTC_2D_8X8_UNORM>,
- MortonCopy<true, PixelFormat::ASTC_2D_8X5_UNORM>,
- MortonCopy<true, PixelFormat::ASTC_2D_5X4_UNORM>,
- MortonCopy<true, PixelFormat::B8G8R8A8_SRGB>,
- MortonCopy<true, PixelFormat::BC1_RGBA_SRGB>,
- MortonCopy<true, PixelFormat::BC2_SRGB>,
- MortonCopy<true, PixelFormat::BC3_SRGB>,
- MortonCopy<true, PixelFormat::BC7_SRGB>,
- MortonCopy<true, PixelFormat::A4B4G4R4_UNORM>,
- MortonCopy<true, PixelFormat::ASTC_2D_4X4_SRGB>,
- MortonCopy<true, PixelFormat::ASTC_2D_8X8_SRGB>,
- MortonCopy<true, PixelFormat::ASTC_2D_8X5_SRGB>,
- MortonCopy<true, PixelFormat::ASTC_2D_5X4_SRGB>,
- MortonCopy<true, PixelFormat::ASTC_2D_5X5_UNORM>,
- MortonCopy<true, PixelFormat::ASTC_2D_5X5_SRGB>,
- MortonCopy<true, PixelFormat::ASTC_2D_10X8_UNORM>,
- MortonCopy<true, PixelFormat::ASTC_2D_10X8_SRGB>,
- MortonCopy<true, PixelFormat::ASTC_2D_6X6_UNORM>,
- MortonCopy<true, PixelFormat::ASTC_2D_6X6_SRGB>,
- MortonCopy<true, PixelFormat::ASTC_2D_10X10_UNORM>,
- MortonCopy<true, PixelFormat::ASTC_2D_10X10_SRGB>,
- MortonCopy<true, PixelFormat::ASTC_2D_12X12_UNORM>,
- MortonCopy<true, PixelFormat::ASTC_2D_12X12_SRGB>,
- MortonCopy<true, PixelFormat::ASTC_2D_8X6_UNORM>,
- MortonCopy<true, PixelFormat::ASTC_2D_8X6_SRGB>,
- MortonCopy<true, PixelFormat::ASTC_2D_6X5_UNORM>,
- MortonCopy<true, PixelFormat::ASTC_2D_6X5_SRGB>,
- MortonCopy<true, PixelFormat::E5B9G9R9_FLOAT>,
- MortonCopy<true, PixelFormat::D32_FLOAT>,
- MortonCopy<true, PixelFormat::D16_UNORM>,
- MortonCopy<true, PixelFormat::D24_UNORM_S8_UINT>,
- MortonCopy<true, PixelFormat::S8_UINT_D24_UNORM>,
- MortonCopy<true, PixelFormat::D32_FLOAT_S8_UINT>,
-};
-
-static constexpr ConversionArray linear_to_morton_fns = {
- MortonCopy<false, PixelFormat::A8B8G8R8_UNORM>,
- MortonCopy<false, PixelFormat::A8B8G8R8_SNORM>,
- MortonCopy<false, PixelFormat::A8B8G8R8_SINT>,
- MortonCopy<false, PixelFormat::A8B8G8R8_UINT>,
- MortonCopy<false, PixelFormat::R5G6B5_UNORM>,
- MortonCopy<false, PixelFormat::B5G6R5_UNORM>,
- MortonCopy<false, PixelFormat::A1R5G5B5_UNORM>,
- MortonCopy<false, PixelFormat::A2B10G10R10_UNORM>,
- MortonCopy<false, PixelFormat::A2B10G10R10_UINT>,
- MortonCopy<false, PixelFormat::A1B5G5R5_UNORM>,
- MortonCopy<false, PixelFormat::R8_UNORM>,
- MortonCopy<false, PixelFormat::R8_SNORM>,
- MortonCopy<false, PixelFormat::R8_SINT>,
- MortonCopy<false, PixelFormat::R8_UINT>,
- MortonCopy<false, PixelFormat::R16G16B16A16_FLOAT>,
- MortonCopy<false, PixelFormat::R16G16B16A16_SNORM>,
- MortonCopy<false, PixelFormat::R16G16B16A16_SINT>,
- MortonCopy<false, PixelFormat::R16G16B16A16_UNORM>,
- MortonCopy<false, PixelFormat::R16G16B16A16_UINT>,
- MortonCopy<false, PixelFormat::B10G11R11_FLOAT>,
- MortonCopy<false, PixelFormat::R32G32B32A32_UINT>,
- MortonCopy<false, PixelFormat::BC1_RGBA_UNORM>,
- MortonCopy<false, PixelFormat::BC2_UNORM>,
- MortonCopy<false, PixelFormat::BC3_UNORM>,
- MortonCopy<false, PixelFormat::BC4_UNORM>,
- MortonCopy<false, PixelFormat::BC4_SNORM>,
- MortonCopy<false, PixelFormat::BC5_UNORM>,
- MortonCopy<false, PixelFormat::BC5_SNORM>,
- MortonCopy<false, PixelFormat::BC7_UNORM>,
- MortonCopy<false, PixelFormat::BC6H_UFLOAT>,
- MortonCopy<false, PixelFormat::BC6H_SFLOAT>,
- // TODO(Subv): Swizzling ASTC formats are not supported
- nullptr,
- MortonCopy<false, PixelFormat::B8G8R8A8_UNORM>,
- MortonCopy<false, PixelFormat::R32G32B32A32_FLOAT>,
- MortonCopy<false, PixelFormat::R32G32B32A32_SINT>,
- MortonCopy<false, PixelFormat::R32G32_FLOAT>,
- MortonCopy<false, PixelFormat::R32G32_SINT>,
- MortonCopy<false, PixelFormat::R32_FLOAT>,
- MortonCopy<false, PixelFormat::R16_FLOAT>,
- MortonCopy<false, PixelFormat::R16_UNORM>,
- MortonCopy<false, PixelFormat::R16_SNORM>,
- MortonCopy<false, PixelFormat::R16_UINT>,
- MortonCopy<false, PixelFormat::R16_SINT>,
- MortonCopy<false, PixelFormat::R16G16_UNORM>,
- MortonCopy<false, PixelFormat::R16G16_FLOAT>,
- MortonCopy<false, PixelFormat::R16G16_UINT>,
- MortonCopy<false, PixelFormat::R16G16_SINT>,
- MortonCopy<false, PixelFormat::R16G16_SNORM>,
- MortonCopy<false, PixelFormat::R32G32B32_FLOAT>,
- MortonCopy<false, PixelFormat::A8B8G8R8_SRGB>,
- MortonCopy<false, PixelFormat::R8G8_UNORM>,
- MortonCopy<false, PixelFormat::R8G8_SNORM>,
- MortonCopy<false, PixelFormat::R8G8_SINT>,
- MortonCopy<false, PixelFormat::R8G8_UINT>,
- MortonCopy<false, PixelFormat::R32G32_UINT>,
- MortonCopy<false, PixelFormat::R16G16B16X16_FLOAT>,
- MortonCopy<false, PixelFormat::R32_UINT>,
- MortonCopy<false, PixelFormat::R32_SINT>,
- nullptr,
- nullptr,
- nullptr,
- MortonCopy<false, PixelFormat::B8G8R8A8_SRGB>,
- MortonCopy<false, PixelFormat::BC1_RGBA_SRGB>,
- MortonCopy<false, PixelFormat::BC2_SRGB>,
- MortonCopy<false, PixelFormat::BC3_SRGB>,
- MortonCopy<false, PixelFormat::BC7_SRGB>,
- MortonCopy<false, PixelFormat::A4B4G4R4_UNORM>,
- nullptr,
- nullptr,
- nullptr,
- nullptr,
- nullptr,
- nullptr,
- nullptr,
- nullptr,
- nullptr,
- nullptr,
- nullptr,
- nullptr,
- nullptr,
- nullptr,
- nullptr,
- nullptr,
- nullptr,
- nullptr,
- MortonCopy<false, PixelFormat::E5B9G9R9_FLOAT>,
- MortonCopy<false, PixelFormat::D32_FLOAT>,
- MortonCopy<false, PixelFormat::D16_UNORM>,
- MortonCopy<false, PixelFormat::D24_UNORM_S8_UINT>,
- MortonCopy<false, PixelFormat::S8_UINT_D24_UNORM>,
- MortonCopy<false, PixelFormat::D32_FLOAT_S8_UINT>,
-};
-
-static MortonCopyFn GetSwizzleFunction(MortonSwizzleMode mode, Surface::PixelFormat format) {
- switch (mode) {
- case MortonSwizzleMode::MortonToLinear:
- return morton_to_linear_fns[static_cast<std::size_t>(format)];
- case MortonSwizzleMode::LinearToMorton:
- return linear_to_morton_fns[static_cast<std::size_t>(format)];
- }
- UNREACHABLE();
- return morton_to_linear_fns[static_cast<std::size_t>(format)];
-}
-
-void MortonSwizzle(MortonSwizzleMode mode, Surface::PixelFormat format, u32 stride,
- u32 block_height, u32 height, u32 block_depth, u32 depth, u32 tile_width_spacing,
- u8* buffer, u8* addr) {
- GetSwizzleFunction(mode, format)(stride, block_height, height, block_depth, depth,
- tile_width_spacing, buffer, addr);
-}
-
-} // namespace VideoCore
diff --git a/src/video_core/morton.h b/src/video_core/morton.h
index b714a7e3f..e69de29bb 100644
--- a/src/video_core/morton.h
+++ b/src/video_core/morton.h
@@ -1,18 +0,0 @@
-// Copyright 2018 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include "common/common_types.h"
-#include "video_core/surface.h"
-
-namespace VideoCore {
-
-enum class MortonSwizzleMode { MortonToLinear, LinearToMorton };
-
-void MortonSwizzle(MortonSwizzleMode mode, VideoCore::Surface::PixelFormat format, u32 stride,
- u32 block_height, u32 height, u32 block_depth, u32 depth, u32 tile_width_spacing,
- u8* buffer, u8* addr);
-
-} // namespace VideoCore
diff --git a/src/video_core/query_cache.h b/src/video_core/query_cache.h
index fc54ca0ef..203f2af05 100644
--- a/src/video_core/query_cache.h
+++ b/src/video_core/query_cache.h
@@ -28,8 +28,8 @@ namespace VideoCommon {
template <class QueryCache, class HostCounter>
class CounterStreamBase {
public:
- explicit CounterStreamBase(QueryCache& cache, VideoCore::QueryType type)
- : cache{cache}, type{type} {}
+ explicit CounterStreamBase(QueryCache& cache_, VideoCore::QueryType type_)
+ : cache{cache_}, type{type_} {}
/// Updates the state of the stream, enabling or disabling as needed.
void Update(bool enabled) {
@@ -334,8 +334,8 @@ private:
template <class HostCounter>
class CachedQueryBase {
public:
- explicit CachedQueryBase(VAddr cpu_addr, u8* host_ptr)
- : cpu_addr{cpu_addr}, host_ptr{host_ptr} {}
+ explicit CachedQueryBase(VAddr cpu_addr_, u8* host_ptr_)
+ : cpu_addr{cpu_addr_}, host_ptr{host_ptr_} {}
virtual ~CachedQueryBase() = default;
CachedQueryBase(CachedQueryBase&&) noexcept = default;
diff --git a/src/video_core/rasterizer_interface.h b/src/video_core/rasterizer_interface.h
index 27ef4c69a..0cb0f387d 100644
--- a/src/video_core/rasterizer_interface.h
+++ b/src/video_core/rasterizer_interface.h
@@ -76,6 +76,9 @@ public:
/// Sync memory between guest and host.
virtual void SyncGuestHost() = 0;
+ /// Unmap memory range
+ virtual void UnmapMemory(VAddr addr, u64 size) = 0;
+
/// Notify rasterizer that any caches of the specified region should be flushed to Switch memory
/// and invalidated
virtual void FlushAndInvalidateRegion(VAddr addr, u64 size) = 0;
@@ -83,6 +86,12 @@ public:
/// Notify the host renderer to wait for previous primitive and compute operations.
virtual void WaitForIdle() = 0;
+ /// Notify the host renderer to wait for reads and writes to render targets and flush caches.
+ virtual void FragmentBarrier() = 0;
+
+ /// Notify the host renderer to make available previous render target writes.
+ virtual void TiledCacheBarrier() = 0;
+
/// Notify the rasterizer to send all written commands to the host GPU.
virtual void FlushCommands() = 0;
@@ -91,8 +100,7 @@ public:
/// Attempt to use a faster method to perform a surface copy
[[nodiscard]] virtual bool AccelerateSurfaceCopy(
- const Tegra::Engines::Fermi2D::Regs::Surface& src,
- const Tegra::Engines::Fermi2D::Regs::Surface& dst,
+ const Tegra::Engines::Fermi2D::Surface& src, const Tegra::Engines::Fermi2D::Surface& dst,
const Tegra::Engines::Fermi2D::Config& copy_config) {
return false;
}
diff --git a/src/video_core/renderer_opengl/gl_arb_decompiler.cpp b/src/video_core/renderer_opengl/gl_arb_decompiler.cpp
index d6120c23e..3e4d88c30 100644
--- a/src/video_core/renderer_opengl/gl_arb_decompiler.cpp
+++ b/src/video_core/renderer_opengl/gl_arb_decompiler.cpp
@@ -71,7 +71,7 @@ std::string_view GetInputFlags(PixelImap attribute) {
case PixelImap::Unused:
break;
}
- UNIMPLEMENTED_MSG("Unknown attribute usage index={}", static_cast<int>(attribute));
+ UNIMPLEMENTED_MSG("Unknown attribute usage index={}", attribute);
return {};
}
@@ -123,7 +123,7 @@ std::string_view PrimitiveDescription(Tegra::Engines::Maxwell3D::Regs::Primitive
case Tegra::Engines::Maxwell3D::Regs::PrimitiveTopology::TriangleStripAdjacency:
return "TRIANGLES_ADJACENCY";
default:
- UNIMPLEMENTED_MSG("topology={}", static_cast<int>(topology));
+ UNIMPLEMENTED_MSG("topology={}", topology);
return "POINTS";
}
}
@@ -137,7 +137,7 @@ std::string_view TopologyName(Tegra::Shader::OutputTopology topology) {
case Tegra::Shader::OutputTopology::TriangleStrip:
return "TRIANGLE_STRIP";
default:
- UNIMPLEMENTED_MSG("Unknown output topology: {}", static_cast<u32>(topology));
+ UNIMPLEMENTED_MSG("Unknown output topology: {}", topology);
return "points";
}
}
@@ -187,8 +187,8 @@ std::string TextureType(const MetaTexture& meta) {
class ARBDecompiler final {
public:
- explicit ARBDecompiler(const Device& device, const ShaderIR& ir, const Registry& registry,
- ShaderType stage, std::string_view identifier);
+ explicit ARBDecompiler(const Device& device_, const ShaderIR& ir_, const Registry& registry_,
+ ShaderType stage_, std::string_view identifier);
std::string Code() const {
return shader_source;
@@ -802,9 +802,9 @@ private:
};
};
-ARBDecompiler::ARBDecompiler(const Device& device, const ShaderIR& ir, const Registry& registry,
- ShaderType stage, std::string_view identifier)
- : device{device}, ir{ir}, registry{registry}, stage{stage} {
+ARBDecompiler::ARBDecompiler(const Device& device_, const ShaderIR& ir_, const Registry& registry_,
+ ShaderType stage_, std::string_view identifier)
+ : device{device_}, ir{ir_}, registry{registry_}, stage{stage_} {
DefineGlobalMemory();
AddLine("TEMP RC;");
@@ -1134,44 +1134,44 @@ void ARBDecompiler::VisitAST(const ASTNode& node) {
for (ASTNode current = ast->nodes.GetFirst(); current; current = current->GetNext()) {
VisitAST(current);
}
- } else if (const auto ast = std::get_if<ASTIfThen>(&*node->GetInnerData())) {
- const std::string condition = VisitExpression(ast->condition);
+ } else if (const auto if_then = std::get_if<ASTIfThen>(&*node->GetInnerData())) {
+ const std::string condition = VisitExpression(if_then->condition);
ResetTemporaries();
AddLine("MOVC.U RC.x, {};", condition);
AddLine("IF NE.x;");
- for (ASTNode current = ast->nodes.GetFirst(); current; current = current->GetNext()) {
+ for (ASTNode current = if_then->nodes.GetFirst(); current; current = current->GetNext()) {
VisitAST(current);
}
AddLine("ENDIF;");
- } else if (const auto ast = std::get_if<ASTIfElse>(&*node->GetInnerData())) {
+ } else if (const auto if_else = std::get_if<ASTIfElse>(&*node->GetInnerData())) {
AddLine("ELSE;");
- for (ASTNode current = ast->nodes.GetFirst(); current; current = current->GetNext()) {
+ for (ASTNode current = if_else->nodes.GetFirst(); current; current = current->GetNext()) {
VisitAST(current);
}
- } else if (const auto ast = std::get_if<ASTBlockDecoded>(&*node->GetInnerData())) {
- VisitBlock(ast->nodes);
- } else if (const auto ast = std::get_if<ASTVarSet>(&*node->GetInnerData())) {
- AddLine("MOV.U F{}, {};", ast->index, VisitExpression(ast->condition));
+ } else if (const auto decoded = std::get_if<ASTBlockDecoded>(&*node->GetInnerData())) {
+ VisitBlock(decoded->nodes);
+ } else if (const auto var_set = std::get_if<ASTVarSet>(&*node->GetInnerData())) {
+ AddLine("MOV.U F{}, {};", var_set->index, VisitExpression(var_set->condition));
ResetTemporaries();
- } else if (const auto ast = std::get_if<ASTDoWhile>(&*node->GetInnerData())) {
- const std::string condition = VisitExpression(ast->condition);
+ } else if (const auto do_while = std::get_if<ASTDoWhile>(&*node->GetInnerData())) {
+ const std::string condition = VisitExpression(do_while->condition);
ResetTemporaries();
AddLine("REP;");
- for (ASTNode current = ast->nodes.GetFirst(); current; current = current->GetNext()) {
+ for (ASTNode current = do_while->nodes.GetFirst(); current; current = current->GetNext()) {
VisitAST(current);
}
AddLine("MOVC.U RC.x, {};", condition);
AddLine("BRK (NE.x);");
AddLine("ENDREP;");
- } else if (const auto ast = std::get_if<ASTReturn>(&*node->GetInnerData())) {
- const bool is_true = ExprIsTrue(ast->condition);
+ } else if (const auto ast_return = std::get_if<ASTReturn>(&*node->GetInnerData())) {
+ const bool is_true = ExprIsTrue(ast_return->condition);
if (!is_true) {
- AddLine("MOVC.U RC.x, {};", VisitExpression(ast->condition));
+ AddLine("MOVC.U RC.x, {};", VisitExpression(ast_return->condition));
AddLine("IF NE.x;");
ResetTemporaries();
}
- if (ast->kills) {
+ if (ast_return->kills) {
AddLine("KIL TR;");
} else {
Exit();
@@ -1179,11 +1179,11 @@ void ARBDecompiler::VisitAST(const ASTNode& node) {
if (!is_true) {
AddLine("ENDIF;");
}
- } else if (const auto ast = std::get_if<ASTBreak>(&*node->GetInnerData())) {
- if (ExprIsTrue(ast->condition)) {
+ } else if (const auto ast_break = std::get_if<ASTBreak>(&*node->GetInnerData())) {
+ if (ExprIsTrue(ast_break->condition)) {
AddLine("BRK;");
} else {
- AddLine("MOVC.U RC.x, {};", VisitExpression(ast->condition));
+ AddLine("MOVC.U RC.x, {};", VisitExpression(ast_break->condition));
AddLine("BRK (NE.x);");
ResetTemporaries();
}
@@ -1351,7 +1351,7 @@ std::string ARBDecompiler::Visit(const Node& node) {
GetGenericAttributeIndex(index), swizzle);
}
}
- UNIMPLEMENTED_MSG("Unimplemented input attribute={}", static_cast<int>(index));
+ UNIMPLEMENTED_MSG("Unimplemented input attribute={}", index);
break;
}
return "{0, 0, 0, 0}.x";
@@ -1485,9 +1485,7 @@ void ARBDecompiler::Exit() {
}
const auto safe_get_register = [this](u32 reg) -> std::string {
- // TODO(Rodrigo): Replace with contains once C++20 releases
- const auto& used_registers = ir.GetRegisters();
- if (used_registers.find(reg) != used_registers.end()) {
+ if (ir.GetRegisters().contains(reg)) {
return fmt::format("R{}.x", reg);
}
return "{0, 0, 0, 0}.x";
diff --git a/src/video_core/renderer_opengl/gl_buffer_cache.cpp b/src/video_core/renderer_opengl/gl_buffer_cache.cpp
index b1c4cd62f..5772cad87 100644
--- a/src/video_core/renderer_opengl/gl_buffer_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_buffer_cache.cpp
@@ -22,11 +22,11 @@ using Maxwell = Tegra::Engines::Maxwell3D::Regs;
MICROPROFILE_DEFINE(OpenGL_Buffer_Download, "OpenGL", "Buffer Download", MP_RGB(192, 192, 128));
-Buffer::Buffer(const Device& device, VAddr cpu_addr, std::size_t size)
- : VideoCommon::BufferBlock{cpu_addr, size} {
+Buffer::Buffer(const Device& device_, VAddr cpu_addr_, std::size_t size_)
+ : BufferBlock{cpu_addr_, size_} {
gl_buffer.Create();
- glNamedBufferData(gl_buffer.handle, static_cast<GLsizeiptr>(size), nullptr, GL_DYNAMIC_DRAW);
- if (device.UseAssemblyShaders() || device.HasVertexBufferUnifiedMemory()) {
+ glNamedBufferData(gl_buffer.handle, static_cast<GLsizeiptr>(size_), nullptr, GL_DYNAMIC_DRAW);
+ if (device_.UseAssemblyShaders() || device_.HasVertexBufferUnifiedMemory()) {
glMakeNamedBufferResidentNV(gl_buffer.handle, GL_READ_WRITE);
glGetNamedBufferParameterui64vNV(gl_buffer.handle, GL_BUFFER_GPU_ADDRESS_NV, &gpu_address);
}
@@ -34,14 +34,14 @@ Buffer::Buffer(const Device& device, VAddr cpu_addr, std::size_t size)
Buffer::~Buffer() = default;
-void Buffer::Upload(std::size_t offset, std::size_t size, const u8* data) {
- glNamedBufferSubData(Handle(), static_cast<GLintptr>(offset), static_cast<GLsizeiptr>(size),
- data);
+void Buffer::Upload(std::size_t offset, std::size_t data_size, const u8* data) {
+ glNamedBufferSubData(Handle(), static_cast<GLintptr>(offset),
+ static_cast<GLsizeiptr>(data_size), data);
}
-void Buffer::Download(std::size_t offset, std::size_t size, u8* data) {
+void Buffer::Download(std::size_t offset, std::size_t data_size, u8* data) {
MICROPROFILE_SCOPE(OpenGL_Buffer_Download);
- const GLsizeiptr gl_size = static_cast<GLsizeiptr>(size);
+ const GLsizeiptr gl_size = static_cast<GLsizeiptr>(data_size);
const GLintptr gl_offset = static_cast<GLintptr>(offset);
if (read_buffer.handle == 0) {
read_buffer.Create();
@@ -54,17 +54,16 @@ void Buffer::Download(std::size_t offset, std::size_t size, u8* data) {
}
void Buffer::CopyFrom(const Buffer& src, std::size_t src_offset, std::size_t dst_offset,
- std::size_t size) {
+ std::size_t copy_size) {
glCopyNamedBufferSubData(src.Handle(), Handle(), static_cast<GLintptr>(src_offset),
- static_cast<GLintptr>(dst_offset), static_cast<GLsizeiptr>(size));
+ static_cast<GLintptr>(dst_offset), static_cast<GLsizeiptr>(copy_size));
}
-OGLBufferCache::OGLBufferCache(VideoCore::RasterizerInterface& rasterizer,
- Tegra::MemoryManager& gpu_memory, Core::Memory::Memory& cpu_memory,
- const Device& device_, std::size_t stream_size)
- : GenericBufferCache{rasterizer, gpu_memory, cpu_memory,
- std::make_unique<OGLStreamBuffer>(device_, stream_size, true)},
- device{device_} {
+OGLBufferCache::OGLBufferCache(VideoCore::RasterizerInterface& rasterizer_,
+ Tegra::MemoryManager& gpu_memory_, Core::Memory::Memory& cpu_memory_,
+ const Device& device_, OGLStreamBuffer& stream_buffer_,
+ StateTracker& state_tracker)
+ : GenericBufferCache{rasterizer_, gpu_memory_, cpu_memory_, stream_buffer_}, device{device_} {
if (!device.HasFastBufferSubData()) {
return;
}
diff --git a/src/video_core/renderer_opengl/gl_buffer_cache.h b/src/video_core/renderer_opengl/gl_buffer_cache.h
index f75b32e31..17ee90316 100644
--- a/src/video_core/renderer_opengl/gl_buffer_cache.h
+++ b/src/video_core/renderer_opengl/gl_buffer_cache.h
@@ -22,18 +22,19 @@ namespace OpenGL {
class Device;
class OGLStreamBuffer;
class RasterizerOpenGL;
+class StateTracker;
class Buffer : public VideoCommon::BufferBlock {
public:
- explicit Buffer(const Device& device, VAddr cpu_addr, std::size_t size);
+ explicit Buffer(const Device& device_, VAddr cpu_addr_, std::size_t size_);
~Buffer();
- void Upload(std::size_t offset, std::size_t size, const u8* data);
+ void Upload(std::size_t offset, std::size_t data_size, const u8* data);
- void Download(std::size_t offset, std::size_t size, u8* data);
+ void Download(std::size_t offset, std::size_t data_size, u8* data);
void CopyFrom(const Buffer& src, std::size_t src_offset, std::size_t dst_offset,
- std::size_t size);
+ std::size_t copy_size);
GLuint Handle() const noexcept {
return gl_buffer.handle;
@@ -54,7 +55,8 @@ class OGLBufferCache final : public GenericBufferCache {
public:
explicit OGLBufferCache(VideoCore::RasterizerInterface& rasterizer,
Tegra::MemoryManager& gpu_memory, Core::Memory::Memory& cpu_memory,
- const Device& device, std::size_t stream_size);
+ const Device& device, OGLStreamBuffer& stream_buffer,
+ StateTracker& state_tracker);
~OGLBufferCache();
BufferInfo GetEmptyBuffer(std::size_t) override;
diff --git a/src/video_core/renderer_opengl/gl_device.cpp b/src/video_core/renderer_opengl/gl_device.cpp
index a94e4f72e..b24179d59 100644
--- a/src/video_core/renderer_opengl/gl_device.cpp
+++ b/src/video_core/renderer_opengl/gl_device.cpp
@@ -5,9 +5,11 @@
#include <algorithm>
#include <array>
#include <cstddef>
+#include <cstdlib>
#include <cstring>
#include <limits>
#include <optional>
+#include <span>
#include <vector>
#include <glad/glad.h>
@@ -27,27 +29,29 @@ constexpr u32 ReservedUniformBlocks = 1;
constexpr u32 NumStages = 5;
-constexpr std::array LimitUBOs = {
+constexpr std::array LIMIT_UBOS = {
GL_MAX_VERTEX_UNIFORM_BLOCKS, GL_MAX_TESS_CONTROL_UNIFORM_BLOCKS,
GL_MAX_TESS_EVALUATION_UNIFORM_BLOCKS, GL_MAX_GEOMETRY_UNIFORM_BLOCKS,
- GL_MAX_FRAGMENT_UNIFORM_BLOCKS, GL_MAX_COMPUTE_UNIFORM_BLOCKS};
-
-constexpr std::array LimitSSBOs = {
+ GL_MAX_FRAGMENT_UNIFORM_BLOCKS, GL_MAX_COMPUTE_UNIFORM_BLOCKS,
+};
+constexpr std::array LIMIT_SSBOS = {
GL_MAX_VERTEX_SHADER_STORAGE_BLOCKS, GL_MAX_TESS_CONTROL_SHADER_STORAGE_BLOCKS,
GL_MAX_TESS_EVALUATION_SHADER_STORAGE_BLOCKS, GL_MAX_GEOMETRY_SHADER_STORAGE_BLOCKS,
- GL_MAX_FRAGMENT_SHADER_STORAGE_BLOCKS, GL_MAX_COMPUTE_SHADER_STORAGE_BLOCKS};
-
-constexpr std::array LimitSamplers = {GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS,
- GL_MAX_TESS_CONTROL_TEXTURE_IMAGE_UNITS,
- GL_MAX_TESS_EVALUATION_TEXTURE_IMAGE_UNITS,
- GL_MAX_GEOMETRY_TEXTURE_IMAGE_UNITS,
- GL_MAX_TEXTURE_IMAGE_UNITS,
- GL_MAX_COMPUTE_TEXTURE_IMAGE_UNITS};
-
-constexpr std::array LimitImages = {
+ GL_MAX_FRAGMENT_SHADER_STORAGE_BLOCKS, GL_MAX_COMPUTE_SHADER_STORAGE_BLOCKS,
+};
+constexpr std::array LIMIT_SAMPLERS = {
+ GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS,
+ GL_MAX_TESS_CONTROL_TEXTURE_IMAGE_UNITS,
+ GL_MAX_TESS_EVALUATION_TEXTURE_IMAGE_UNITS,
+ GL_MAX_GEOMETRY_TEXTURE_IMAGE_UNITS,
+ GL_MAX_TEXTURE_IMAGE_UNITS,
+ GL_MAX_COMPUTE_TEXTURE_IMAGE_UNITS,
+};
+constexpr std::array LIMIT_IMAGES = {
GL_MAX_VERTEX_IMAGE_UNIFORMS, GL_MAX_TESS_CONTROL_IMAGE_UNIFORMS,
GL_MAX_TESS_EVALUATION_IMAGE_UNIFORMS, GL_MAX_GEOMETRY_IMAGE_UNIFORMS,
- GL_MAX_FRAGMENT_IMAGE_UNIFORMS, GL_MAX_COMPUTE_IMAGE_UNIFORMS};
+ GL_MAX_FRAGMENT_IMAGE_UNIFORMS, GL_MAX_COMPUTE_IMAGE_UNIFORMS,
+};
template <typename T>
T GetInteger(GLenum pname) {
@@ -76,8 +80,8 @@ std::vector<std::string_view> GetExtensions() {
return extensions;
}
-bool HasExtension(const std::vector<std::string_view>& images, std::string_view extension) {
- return std::find(images.begin(), images.end(), extension) != images.end();
+bool HasExtension(std::span<const std::string_view> extensions, std::string_view extension) {
+ return std::ranges::find(extensions, extension) != extensions.end();
}
u32 Extract(u32& base, u32& num, u32 amount, std::optional<GLenum> limit = {}) {
@@ -91,8 +95,8 @@ u32 Extract(u32& base, u32& num, u32 amount, std::optional<GLenum> limit = {}) {
std::array<u32, Tegra::Engines::MaxShaderTypes> BuildMaxUniformBuffers() noexcept {
std::array<u32, Tegra::Engines::MaxShaderTypes> max;
- std::transform(LimitUBOs.begin(), LimitUBOs.end(), max.begin(),
- [](GLenum pname) { return GetInteger<u32>(pname); });
+ std::ranges::transform(LIMIT_UBOS, max.begin(),
+ [](GLenum pname) { return GetInteger<u32>(pname); });
return max;
}
@@ -115,9 +119,10 @@ std::array<Device::BaseBindings, Tegra::Engines::MaxShaderTypes> BuildBaseBindin
for (std::size_t i = 0; i < NumStages; ++i) {
const std::size_t stage = stage_swizzle[i];
bindings[stage] = {
- Extract(base_ubo, num_ubos, total_ubos / NumStages, LimitUBOs[stage]),
- Extract(base_ssbo, num_ssbos, total_ssbos / NumStages, LimitSSBOs[stage]),
- Extract(base_samplers, num_samplers, total_samplers / NumStages, LimitSamplers[stage])};
+ Extract(base_ubo, num_ubos, total_ubos / NumStages, LIMIT_UBOS[stage]),
+ Extract(base_ssbo, num_ssbos, total_ssbos / NumStages, LIMIT_SSBOS[stage]),
+ Extract(base_samplers, num_samplers, total_samplers / NumStages,
+ LIMIT_SAMPLERS[stage])};
}
u32 num_images = GetInteger<u32>(GL_MAX_IMAGE_UNITS);
@@ -130,7 +135,7 @@ std::array<Device::BaseBindings, Tegra::Engines::MaxShaderTypes> BuildBaseBindin
// Reserve at least 4 image bindings on the fragment stage.
bindings[4].image =
- Extract(base_images, num_images, std::max(4U, num_images / NumStages), LimitImages[4]);
+ Extract(base_images, num_images, std::max(4U, num_images / NumStages), LIMIT_IMAGES[4]);
// This is guaranteed to be at least 1.
const u32 total_extracted_images = num_images / (NumStages - 1);
@@ -142,7 +147,7 @@ std::array<Device::BaseBindings, Tegra::Engines::MaxShaderTypes> BuildBaseBindin
continue;
}
bindings[stage].image =
- Extract(base_images, num_images, total_extracted_images, LimitImages[stage]);
+ Extract(base_images, num_images, total_extracted_images, LIMIT_IMAGES[stage]);
}
// Compute doesn't care about any of this.
@@ -188,6 +193,11 @@ bool IsASTCSupported() {
return true;
}
+[[nodiscard]] bool IsDebugToolAttached(std::span<const std::string_view> extensions) {
+ const bool nsight = std::getenv("NVTX_INJECTION64_PATH") || std::getenv("NSIGHT_LAUNCHED");
+ return nsight || HasExtension(extensions, "GL_EXT_debug_tool");
+}
+
} // Anonymous namespace
Device::Device()
@@ -206,9 +216,8 @@ Device::Device()
"Beta driver 443.24 is known to have issues. There might be performance issues.");
disable_fast_buffer_sub_data = true;
}
-
- uniform_buffer_alignment = GetInteger<std::size_t>(GL_UNIFORM_BUFFER_OFFSET_ALIGNMENT);
- shader_storage_alignment = GetInteger<std::size_t>(GL_SHADER_STORAGE_BUFFER_OFFSET_ALIGNMENT);
+ uniform_buffer_alignment = GetInteger<size_t>(GL_UNIFORM_BUFFER_OFFSET_ALIGNMENT);
+ shader_storage_alignment = GetInteger<size_t>(GL_SHADER_STORAGE_BUFFER_OFFSET_ALIGNMENT);
max_vertex_attributes = GetInteger<u32>(GL_MAX_VERTEX_ATTRIBS);
max_varyings = GetInteger<u32>(GL_MAX_VARYING_VECTORS);
max_compute_shared_memory_size = GetInteger<u32>(GL_MAX_COMPUTE_SHARED_MEMORY_SIZE);
@@ -224,6 +233,7 @@ Device::Device()
has_precise_bug = TestPreciseBug();
has_nv_viewport_array2 = GLAD_GL_NV_viewport_array2;
has_vertex_buffer_unified_memory = GLAD_GL_NV_vertex_buffer_unified_memory;
+ has_debugging_tool_attached = IsDebugToolAttached(extensions);
// At the moment of writing this, only Nvidia's driver optimizes BufferSubData on exclusive
// uniform buffers as "push constants"
diff --git a/src/video_core/renderer_opengl/gl_device.h b/src/video_core/renderer_opengl/gl_device.h
index 8a4b6b9fc..13e66846c 100644
--- a/src/video_core/renderer_opengl/gl_device.h
+++ b/src/video_core/renderer_opengl/gl_device.h
@@ -36,11 +36,11 @@ public:
return GetBaseBindings(static_cast<std::size_t>(shader_type));
}
- std::size_t GetUniformBufferAlignment() const {
+ size_t GetUniformBufferAlignment() const {
return uniform_buffer_alignment;
}
- std::size_t GetShaderStorageBufferAlignment() const {
+ size_t GetShaderStorageBufferAlignment() const {
return shader_storage_alignment;
}
@@ -104,6 +104,10 @@ public:
return has_nv_viewport_array2;
}
+ bool HasDebuggingToolAttached() const {
+ return has_debugging_tool_attached;
+ }
+
bool UseAssemblyShaders() const {
return use_assembly_shaders;
}
@@ -118,8 +122,8 @@ private:
std::array<u32, Tegra::Engines::MaxShaderTypes> max_uniform_buffers{};
std::array<BaseBindings, Tegra::Engines::MaxShaderTypes> base_bindings{};
- std::size_t uniform_buffer_alignment{};
- std::size_t shader_storage_alignment{};
+ size_t uniform_buffer_alignment{};
+ size_t shader_storage_alignment{};
u32 max_vertex_attributes{};
u32 max_varyings{};
u32 max_compute_shared_memory_size{};
@@ -135,6 +139,7 @@ private:
bool has_precise_bug{};
bool has_fast_buffer_sub_data{};
bool has_nv_viewport_array2{};
+ bool has_debugging_tool_attached{};
bool use_assembly_shaders{};
bool use_asynchronous_shaders{};
};
diff --git a/src/video_core/renderer_opengl/gl_fence_manager.cpp b/src/video_core/renderer_opengl/gl_fence_manager.cpp
index b532fdcc2..3e9c922f5 100644
--- a/src/video_core/renderer_opengl/gl_fence_manager.cpp
+++ b/src/video_core/renderer_opengl/gl_fence_manager.cpp
@@ -11,10 +11,10 @@
namespace OpenGL {
-GLInnerFence::GLInnerFence(u32 payload, bool is_stubbed) : FenceBase(payload, is_stubbed) {}
+GLInnerFence::GLInnerFence(u32 payload_, bool is_stubbed_) : FenceBase{payload_, is_stubbed_} {}
-GLInnerFence::GLInnerFence(GPUVAddr address, u32 payload, bool is_stubbed)
- : FenceBase(address, payload, is_stubbed) {}
+GLInnerFence::GLInnerFence(GPUVAddr address_, u32 payload_, bool is_stubbed_)
+ : FenceBase{address_, payload_, is_stubbed_} {}
GLInnerFence::~GLInnerFence() = default;
@@ -45,10 +45,10 @@ void GLInnerFence::Wait() {
glClientWaitSync(sync_object.handle, 0, GL_TIMEOUT_IGNORED);
}
-FenceManagerOpenGL::FenceManagerOpenGL(VideoCore::RasterizerInterface& rasterizer, Tegra::GPU& gpu,
- TextureCacheOpenGL& texture_cache,
- OGLBufferCache& buffer_cache, QueryCache& query_cache)
- : GenericFenceManager{rasterizer, gpu, texture_cache, buffer_cache, query_cache} {}
+FenceManagerOpenGL::FenceManagerOpenGL(VideoCore::RasterizerInterface& rasterizer_,
+ Tegra::GPU& gpu_, TextureCache& texture_cache_,
+ OGLBufferCache& buffer_cache_, QueryCache& query_cache_)
+ : GenericFenceManager{rasterizer_, gpu_, texture_cache_, buffer_cache_, query_cache_} {}
Fence FenceManagerOpenGL::CreateFence(u32 value, bool is_stubbed) {
return std::make_shared<GLInnerFence>(value, is_stubbed);
diff --git a/src/video_core/renderer_opengl/gl_fence_manager.h b/src/video_core/renderer_opengl/gl_fence_manager.h
index da1dcdace..30dbee613 100644
--- a/src/video_core/renderer_opengl/gl_fence_manager.h
+++ b/src/video_core/renderer_opengl/gl_fence_manager.h
@@ -17,8 +17,8 @@ namespace OpenGL {
class GLInnerFence : public VideoCommon::FenceBase {
public:
- GLInnerFence(u32 payload, bool is_stubbed);
- GLInnerFence(GPUVAddr address, u32 payload, bool is_stubbed);
+ explicit GLInnerFence(u32 payload_, bool is_stubbed_);
+ explicit GLInnerFence(GPUVAddr address_, u32 payload_, bool is_stubbed_);
~GLInnerFence();
void Queue();
@@ -33,13 +33,13 @@ private:
using Fence = std::shared_ptr<GLInnerFence>;
using GenericFenceManager =
- VideoCommon::FenceManager<Fence, TextureCacheOpenGL, OGLBufferCache, QueryCache>;
+ VideoCommon::FenceManager<Fence, TextureCache, OGLBufferCache, QueryCache>;
class FenceManagerOpenGL final : public GenericFenceManager {
public:
- explicit FenceManagerOpenGL(VideoCore::RasterizerInterface& rasterizer, Tegra::GPU& gpu,
- TextureCacheOpenGL& texture_cache, OGLBufferCache& buffer_cache,
- QueryCache& query_cache);
+ explicit FenceManagerOpenGL(VideoCore::RasterizerInterface& rasterizer_, Tegra::GPU& gpu_,
+ TextureCache& texture_cache_, OGLBufferCache& buffer_cache_,
+ QueryCache& query_cache_);
protected:
Fence CreateFence(u32 value, bool is_stubbed) override;
diff --git a/src/video_core/renderer_opengl/gl_framebuffer_cache.cpp b/src/video_core/renderer_opengl/gl_framebuffer_cache.cpp
deleted file mode 100644
index b8a512cb6..000000000
--- a/src/video_core/renderer_opengl/gl_framebuffer_cache.cpp
+++ /dev/null
@@ -1,85 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include <tuple>
-#include <unordered_map>
-#include <utility>
-
-#include <glad/glad.h>
-
-#include "common/common_types.h"
-#include "video_core/engines/maxwell_3d.h"
-#include "video_core/renderer_opengl/gl_framebuffer_cache.h"
-
-namespace OpenGL {
-
-using Maxwell = Tegra::Engines::Maxwell3D::Regs;
-using VideoCore::Surface::SurfaceType;
-
-FramebufferCacheOpenGL::FramebufferCacheOpenGL() = default;
-
-FramebufferCacheOpenGL::~FramebufferCacheOpenGL() = default;
-
-GLuint FramebufferCacheOpenGL::GetFramebuffer(const FramebufferCacheKey& key) {
- const auto [entry, is_cache_miss] = cache.try_emplace(key);
- auto& framebuffer{entry->second};
- if (is_cache_miss) {
- framebuffer = CreateFramebuffer(key);
- }
- return framebuffer.handle;
-}
-
-OGLFramebuffer FramebufferCacheOpenGL::CreateFramebuffer(const FramebufferCacheKey& key) {
- OGLFramebuffer framebuffer;
- framebuffer.Create();
-
- // TODO(Rodrigo): Use DSA here after Nvidia fixes their framebuffer DSA bugs.
- glBindFramebuffer(GL_DRAW_FRAMEBUFFER, framebuffer.handle);
-
- if (key.zeta) {
- const bool stencil = key.zeta->GetSurfaceParams().type == SurfaceType::DepthStencil;
- const GLenum attach_target = stencil ? GL_DEPTH_STENCIL_ATTACHMENT : GL_DEPTH_ATTACHMENT;
- key.zeta->Attach(attach_target, GL_DRAW_FRAMEBUFFER);
- }
-
- std::size_t num_buffers = 0;
- std::array<GLenum, Maxwell::NumRenderTargets> targets;
-
- for (std::size_t index = 0; index < Maxwell::NumRenderTargets; ++index) {
- if (!key.colors[index]) {
- targets[index] = GL_NONE;
- continue;
- }
- const GLenum attach_target = GL_COLOR_ATTACHMENT0 + static_cast<GLenum>(index);
- key.colors[index]->Attach(attach_target, GL_DRAW_FRAMEBUFFER);
-
- const u32 attachment = (key.color_attachments >> (BitsPerAttachment * index)) & 0b1111;
- targets[index] = GL_COLOR_ATTACHMENT0 + attachment;
- num_buffers = index + 1;
- }
-
- if (num_buffers > 0) {
- glDrawBuffers(static_cast<GLsizei>(num_buffers), std::data(targets));
- } else {
- glDrawBuffer(GL_NONE);
- }
-
- return framebuffer;
-}
-
-std::size_t FramebufferCacheKey::Hash() const noexcept {
- std::size_t hash = std::hash<View>{}(zeta);
- for (const auto& color : colors) {
- hash ^= std::hash<View>{}(color);
- }
- hash ^= static_cast<std::size_t>(color_attachments) << 16;
- return hash;
-}
-
-bool FramebufferCacheKey::operator==(const FramebufferCacheKey& rhs) const noexcept {
- return std::tie(colors, zeta, color_attachments) ==
- std::tie(rhs.colors, rhs.zeta, rhs.color_attachments);
-}
-
-} // namespace OpenGL
diff --git a/src/video_core/renderer_opengl/gl_framebuffer_cache.h b/src/video_core/renderer_opengl/gl_framebuffer_cache.h
deleted file mode 100644
index 8f698fee0..000000000
--- a/src/video_core/renderer_opengl/gl_framebuffer_cache.h
+++ /dev/null
@@ -1,68 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <array>
-#include <cstddef>
-#include <unordered_map>
-
-#include <glad/glad.h>
-
-#include "common/common_types.h"
-#include "video_core/engines/maxwell_3d.h"
-#include "video_core/renderer_opengl/gl_resource_manager.h"
-#include "video_core/renderer_opengl/gl_texture_cache.h"
-
-namespace OpenGL {
-
-constexpr std::size_t BitsPerAttachment = 4;
-
-struct FramebufferCacheKey {
- View zeta;
- std::array<View, Tegra::Engines::Maxwell3D::Regs::NumRenderTargets> colors;
- u32 color_attachments = 0;
-
- std::size_t Hash() const noexcept;
-
- bool operator==(const FramebufferCacheKey& rhs) const noexcept;
-
- bool operator!=(const FramebufferCacheKey& rhs) const noexcept {
- return !operator==(rhs);
- }
-
- void SetAttachment(std::size_t index, u32 attachment) {
- color_attachments |= attachment << (BitsPerAttachment * index);
- }
-};
-
-} // namespace OpenGL
-
-namespace std {
-
-template <>
-struct hash<OpenGL::FramebufferCacheKey> {
- std::size_t operator()(const OpenGL::FramebufferCacheKey& k) const noexcept {
- return k.Hash();
- }
-};
-
-} // namespace std
-
-namespace OpenGL {
-
-class FramebufferCacheOpenGL {
-public:
- FramebufferCacheOpenGL();
- ~FramebufferCacheOpenGL();
-
- GLuint GetFramebuffer(const FramebufferCacheKey& key);
-
-private:
- OGLFramebuffer CreateFramebuffer(const FramebufferCacheKey& key);
-
- std::unordered_map<FramebufferCacheKey, OGLFramebuffer> cache;
-};
-
-} // namespace OpenGL
diff --git a/src/video_core/renderer_opengl/gl_query_cache.cpp b/src/video_core/renderer_opengl/gl_query_cache.cpp
index 1a3d9720e..acebbf5f4 100644
--- a/src/video_core/renderer_opengl/gl_query_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_query_cache.cpp
@@ -30,11 +30,9 @@ constexpr GLenum GetTarget(VideoCore::QueryType type) {
} // Anonymous namespace
-QueryCache::QueryCache(RasterizerOpenGL& rasterizer, Tegra::Engines::Maxwell3D& maxwell3d,
- Tegra::MemoryManager& gpu_memory)
- : VideoCommon::QueryCacheBase<QueryCache, CachedQuery, CounterStream, HostCounter>(
- rasterizer, maxwell3d, gpu_memory),
- gl_rasterizer{rasterizer} {}
+QueryCache::QueryCache(RasterizerOpenGL& rasterizer_, Tegra::Engines::Maxwell3D& maxwell3d_,
+ Tegra::MemoryManager& gpu_memory_)
+ : QueryCacheBase(rasterizer_, maxwell3d_, gpu_memory_), gl_rasterizer{rasterizer_} {}
QueryCache::~QueryCache() = default;
@@ -59,10 +57,11 @@ bool QueryCache::AnyCommandQueued() const noexcept {
return gl_rasterizer.AnyCommandQueued();
}
-HostCounter::HostCounter(QueryCache& cache, std::shared_ptr<HostCounter> dependency,
- VideoCore::QueryType type)
- : VideoCommon::HostCounterBase<QueryCache, HostCounter>{std::move(dependency)}, cache{cache},
- type{type}, query{cache.AllocateQuery(type)} {
+HostCounter::HostCounter(QueryCache& cache_, std::shared_ptr<HostCounter> dependency_,
+ VideoCore::QueryType type_)
+ : HostCounterBase{std::move(dependency_)}, cache{cache_}, type{type_}, query{
+ cache.AllocateQuery(
+ type)} {
glBeginQuery(GetTarget(type), query.handle);
}
@@ -86,13 +85,14 @@ u64 HostCounter::BlockingQuery() const {
return static_cast<u64>(value);
}
-CachedQuery::CachedQuery(QueryCache& cache, VideoCore::QueryType type, VAddr cpu_addr, u8* host_ptr)
- : VideoCommon::CachedQueryBase<HostCounter>{cpu_addr, host_ptr}, cache{&cache}, type{type} {}
+CachedQuery::CachedQuery(QueryCache& cache_, VideoCore::QueryType type_, VAddr cpu_addr_,
+ u8* host_ptr_)
+ : CachedQueryBase{cpu_addr_, host_ptr_}, cache{&cache_}, type{type_} {}
CachedQuery::~CachedQuery() = default;
CachedQuery::CachedQuery(CachedQuery&& rhs) noexcept
- : VideoCommon::CachedQueryBase<HostCounter>(std::move(rhs)), cache{rhs.cache}, type{rhs.type} {}
+ : CachedQueryBase(std::move(rhs)), cache{rhs.cache}, type{rhs.type} {}
CachedQuery& CachedQuery::operator=(CachedQuery&& rhs) noexcept {
cache = rhs.cache;
diff --git a/src/video_core/renderer_opengl/gl_query_cache.h b/src/video_core/renderer_opengl/gl_query_cache.h
index 82cac51ee..7bbe5cfe9 100644
--- a/src/video_core/renderer_opengl/gl_query_cache.h
+++ b/src/video_core/renderer_opengl/gl_query_cache.h
@@ -29,8 +29,8 @@ using CounterStream = VideoCommon::CounterStreamBase<QueryCache, HostCounter>;
class QueryCache final
: public VideoCommon::QueryCacheBase<QueryCache, CachedQuery, CounterStream, HostCounter> {
public:
- explicit QueryCache(RasterizerOpenGL& rasterizer, Tegra::Engines::Maxwell3D& maxwell3d,
- Tegra::MemoryManager& gpu_memory);
+ explicit QueryCache(RasterizerOpenGL& rasterizer_, Tegra::Engines::Maxwell3D& maxwell3d_,
+ Tegra::MemoryManager& gpu_memory_);
~QueryCache();
OGLQuery AllocateQuery(VideoCore::QueryType type);
@@ -46,8 +46,8 @@ private:
class HostCounter final : public VideoCommon::HostCounterBase<QueryCache, HostCounter> {
public:
- explicit HostCounter(QueryCache& cache, std::shared_ptr<HostCounter> dependency,
- VideoCore::QueryType type);
+ explicit HostCounter(QueryCache& cache_, std::shared_ptr<HostCounter> dependency_,
+ VideoCore::QueryType type_);
~HostCounter();
void EndQuery();
@@ -62,8 +62,8 @@ private:
class CachedQuery final : public VideoCommon::CachedQueryBase<HostCounter> {
public:
- explicit CachedQuery(QueryCache& cache, VideoCore::QueryType type, VAddr cpu_addr,
- u8* host_ptr);
+ explicit CachedQuery(QueryCache& cache_, VideoCore::QueryType type_, VAddr cpu_addr_,
+ u8* host_ptr_);
~CachedQuery() override;
CachedQuery(CachedQuery&& rhs) noexcept;
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp
index cfddbde5d..8aa63d329 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.cpp
+++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp
@@ -25,12 +25,15 @@
#include "video_core/engines/maxwell_3d.h"
#include "video_core/engines/shader_type.h"
#include "video_core/memory_manager.h"
+#include "video_core/renderer_opengl/gl_device.h"
#include "video_core/renderer_opengl/gl_query_cache.h"
#include "video_core/renderer_opengl/gl_rasterizer.h"
#include "video_core/renderer_opengl/gl_shader_cache.h"
+#include "video_core/renderer_opengl/gl_texture_cache.h"
#include "video_core/renderer_opengl/maxwell_to_gl.h"
#include "video_core/renderer_opengl/renderer_opengl.h"
#include "video_core/shader_cache.h"
+#include "video_core/texture_cache/texture_cache.h"
namespace OpenGL {
@@ -55,18 +58,32 @@ MICROPROFILE_DEFINE(OpenGL_PrimitiveAssembly, "OpenGL", "Prim Asmbl", MP_RGB(255
namespace {
-constexpr std::size_t NUM_CONST_BUFFERS_PER_STAGE = 18;
-constexpr std::size_t NUM_CONST_BUFFERS_BYTES_PER_STAGE =
+constexpr size_t NUM_CONST_BUFFERS_PER_STAGE = 18;
+constexpr size_t NUM_CONST_BUFFERS_BYTES_PER_STAGE =
NUM_CONST_BUFFERS_PER_STAGE * Maxwell::MaxConstBufferSize;
-constexpr std::size_t TOTAL_CONST_BUFFER_BYTES =
+constexpr size_t TOTAL_CONST_BUFFER_BYTES =
NUM_CONST_BUFFERS_BYTES_PER_STAGE * Maxwell::MaxShaderStage;
-constexpr std::size_t NUM_SUPPORTED_VERTEX_ATTRIBUTES = 16;
-constexpr std::size_t NUM_SUPPORTED_VERTEX_BINDINGS = 16;
+constexpr size_t NUM_SUPPORTED_VERTEX_ATTRIBUTES = 16;
+constexpr size_t NUM_SUPPORTED_VERTEX_BINDINGS = 16;
+
+constexpr size_t MAX_TEXTURES = 192;
+constexpr size_t MAX_IMAGES = 48;
+
+struct TextureHandle {
+ constexpr TextureHandle(u32 data, bool via_header_index) {
+ const Tegra::Texture::TextureHandle handle{data};
+ image = handle.tic_id;
+ sampler = via_header_index ? image : handle.tsc_id.Value();
+ }
+
+ u32 image;
+ u32 sampler;
+};
template <typename Engine, typename Entry>
-Tegra::Texture::FullTextureInfo GetTextureInfo(const Engine& engine, const Entry& entry,
- ShaderType shader_type, std::size_t index = 0) {
+TextureHandle GetTextureInfo(const Engine& engine, bool via_header_index, const Entry& entry,
+ ShaderType shader_type, size_t index = 0) {
if constexpr (std::is_same_v<Entry, SamplerEntry>) {
if (entry.is_separated) {
const u32 buffer_1 = entry.buffer;
@@ -75,21 +92,16 @@ Tegra::Texture::FullTextureInfo GetTextureInfo(const Engine& engine, const Entry
const u32 offset_2 = entry.secondary_offset;
const u32 handle_1 = engine.AccessConstBuffer32(shader_type, buffer_1, offset_1);
const u32 handle_2 = engine.AccessConstBuffer32(shader_type, buffer_2, offset_2);
- return engine.GetTextureInfo(handle_1 | handle_2);
+ return TextureHandle(handle_1 | handle_2, via_header_index);
}
}
if (entry.is_bindless) {
- const u32 handle = engine.AccessConstBuffer32(shader_type, entry.buffer, entry.offset);
- return engine.GetTextureInfo(handle);
- }
-
- const auto& gpu_profile = engine.AccessGuestDriverProfile();
- const u32 offset = entry.offset + static_cast<u32>(index * gpu_profile.GetTextureHandlerSize());
- if constexpr (std::is_same_v<Engine, Tegra::Engines::Maxwell3D>) {
- return engine.GetStageTexture(shader_type, offset);
- } else {
- return engine.GetTexture(offset);
+ const u32 raw = engine.AccessConstBuffer32(shader_type, entry.buffer, entry.offset);
+ return TextureHandle(raw, via_header_index);
}
+ const u32 buffer = engine.GetBoundBuffer();
+ const u64 offset = (entry.offset + index) * sizeof(u32);
+ return TextureHandle(engine.AccessConstBuffer32(shader_type, buffer, offset), via_header_index);
}
std::size_t GetConstBufferSize(const Tegra::Engines::ConstBufferInfo& buffer,
@@ -97,7 +109,6 @@ std::size_t GetConstBufferSize(const Tegra::Engines::ConstBufferInfo& buffer,
if (!entry.IsIndirect()) {
return entry.GetSize();
}
-
if (buffer.size > Maxwell::MaxConstBufferSize) {
LOG_WARNING(Render_OpenGL, "Indirect constbuffer size {} exceeds maximum {}", buffer.size,
Maxwell::MaxConstBufferSize);
@@ -131,7 +142,7 @@ std::pair<GLint, GLint> TransformFeedbackEnum(u8 location) {
case 43:
return {GL_BACK_SECONDARY_COLOR_NV, 0};
}
- UNIMPLEMENTED_MSG("index={}", static_cast<int>(index));
+ UNIMPLEMENTED_MSG("index={}", index);
return {GL_POSITION, 0};
}
@@ -147,23 +158,60 @@ void UpdateBindlessSSBOs(GLenum target, const BindlessSSBO* ssbos, size_t num_ss
reinterpret_cast<const GLuint*>(ssbos));
}
+ImageViewType ImageViewTypeFromEntry(const SamplerEntry& entry) {
+ if (entry.is_buffer) {
+ return ImageViewType::Buffer;
+ }
+ switch (entry.type) {
+ case Tegra::Shader::TextureType::Texture1D:
+ return entry.is_array ? ImageViewType::e1DArray : ImageViewType::e1D;
+ case Tegra::Shader::TextureType::Texture2D:
+ return entry.is_array ? ImageViewType::e2DArray : ImageViewType::e2D;
+ case Tegra::Shader::TextureType::Texture3D:
+ return ImageViewType::e3D;
+ case Tegra::Shader::TextureType::TextureCube:
+ return entry.is_array ? ImageViewType::CubeArray : ImageViewType::Cube;
+ }
+ UNREACHABLE();
+ return ImageViewType::e2D;
+}
+
+ImageViewType ImageViewTypeFromEntry(const ImageEntry& entry) {
+ switch (entry.type) {
+ case Tegra::Shader::ImageType::Texture1D:
+ return ImageViewType::e1D;
+ case Tegra::Shader::ImageType::Texture1DArray:
+ return ImageViewType::e1DArray;
+ case Tegra::Shader::ImageType::Texture2D:
+ return ImageViewType::e2D;
+ case Tegra::Shader::ImageType::Texture2DArray:
+ return ImageViewType::e2DArray;
+ case Tegra::Shader::ImageType::Texture3D:
+ return ImageViewType::e3D;
+ case Tegra::Shader::ImageType::TextureBuffer:
+ return ImageViewType::Buffer;
+ }
+ UNREACHABLE();
+ return ImageViewType::e2D;
+}
+
} // Anonymous namespace
-RasterizerOpenGL::RasterizerOpenGL(Core::Frontend::EmuWindow& emu_window, Tegra::GPU& gpu_,
- Core::Memory::Memory& cpu_memory, const Device& device_,
+RasterizerOpenGL::RasterizerOpenGL(Core::Frontend::EmuWindow& emu_window_, Tegra::GPU& gpu_,
+ Core::Memory::Memory& cpu_memory_, const Device& device_,
ScreenInfo& screen_info_, ProgramManager& program_manager_,
StateTracker& state_tracker_)
- : RasterizerAccelerated{cpu_memory}, gpu(gpu_), maxwell3d(gpu.Maxwell3D()),
+ : RasterizerAccelerated(cpu_memory_), gpu(gpu_), maxwell3d(gpu.Maxwell3D()),
kepler_compute(gpu.KeplerCompute()), gpu_memory(gpu.MemoryManager()), device(device_),
screen_info(screen_info_), program_manager(program_manager_), state_tracker(state_tracker_),
- texture_cache(*this, maxwell3d, gpu_memory, device, state_tracker),
- shader_cache(*this, emu_window, gpu, maxwell3d, kepler_compute, gpu_memory, device),
+ stream_buffer(device, state_tracker),
+ texture_cache_runtime(device, program_manager, state_tracker),
+ texture_cache(texture_cache_runtime, *this, maxwell3d, kepler_compute, gpu_memory),
+ shader_cache(*this, emu_window_, gpu, maxwell3d, kepler_compute, gpu_memory, device),
query_cache(*this, maxwell3d, gpu_memory),
- buffer_cache(*this, gpu_memory, cpu_memory, device, STREAM_BUFFER_SIZE),
+ buffer_cache(*this, gpu_memory, cpu_memory_, device, stream_buffer, state_tracker),
fence_manager(*this, gpu, texture_cache, buffer_cache, query_cache),
- async_shaders(emu_window) {
- CheckExtensions();
-
+ async_shaders(emu_window_) {
unified_uniform_buffer.Create();
glNamedBufferStorage(unified_uniform_buffer.handle, TOTAL_CONST_BUFFER_BYTES, nullptr, 0);
@@ -174,7 +222,6 @@ RasterizerOpenGL::RasterizerOpenGL(Core::Frontend::EmuWindow& emu_window, Tegra:
nullptr, 0);
}
}
-
if (device.UseAsynchronousShaders()) {
async_shaders.AllocateWorkers();
}
@@ -186,14 +233,6 @@ RasterizerOpenGL::~RasterizerOpenGL() {
}
}
-void RasterizerOpenGL::CheckExtensions() {
- if (!GLAD_GL_ARB_texture_filter_anisotropic && !GLAD_GL_EXT_texture_filter_anisotropic) {
- LOG_WARNING(
- Render_OpenGL,
- "Anisotropic filter is not supported! This can cause graphical issues in some games.");
- }
-}
-
void RasterizerOpenGL::SetupVertexFormat() {
auto& flags = maxwell3d.dirty.flags;
if (!flags[Dirty::VertexFormats]) {
@@ -316,10 +355,16 @@ GLintptr RasterizerOpenGL::SetupIndexBuffer() {
return info.offset;
}
-void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
+void RasterizerOpenGL::SetupShaders() {
MICROPROFILE_SCOPE(OpenGL_Shader);
u32 clip_distances = 0;
+ std::array<Shader*, Maxwell::MaxShaderStage> shaders{};
+ image_view_indices.clear();
+ sampler_handles.clear();
+
+ texture_cache.SynchronizeGraphicsDescriptors();
+
for (std::size_t index = 0; index < Maxwell::MaxShaderProgram; ++index) {
const auto& shader_config = maxwell3d.regs.shader_config[index];
const auto program{static_cast<Maxwell::ShaderProgram>(index)};
@@ -338,7 +383,6 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
}
continue;
}
-
// Currently this stages are not supported in the OpenGL backend.
// TODO(Blinkhawk): Port tesselation shaders from Vulkan to OpenGL
if (program == Maxwell::ShaderProgram::TesselationControl ||
@@ -347,7 +391,6 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
}
Shader* const shader = shader_cache.GetStageProgram(program, async_shaders);
-
const GLuint program_handle = shader->IsBuilt() ? shader->GetHandle() : 0;
switch (program) {
case Maxwell::ShaderProgram::VertexA:
@@ -363,14 +406,17 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
default:
UNIMPLEMENTED_MSG("Unimplemented shader index={}, enable={}, offset=0x{:08X}", index,
shader_config.enable.Value(), shader_config.offset);
+ break;
}
// Stage indices are 0 - 5
- const std::size_t stage = index == 0 ? 0 : index - 1;
+ const size_t stage = index == 0 ? 0 : index - 1;
+ shaders[stage] = shader;
+
SetupDrawConstBuffers(stage, shader);
SetupDrawGlobalMemory(stage, shader);
- SetupDrawTextures(stage, shader);
- SetupDrawImages(stage, shader);
+ SetupDrawTextures(shader, stage);
+ SetupDrawImages(shader, stage);
// Workaround for Intel drivers.
// When a clip distance is enabled but not set in the shader it crops parts of the screen
@@ -384,9 +430,23 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
++index;
}
}
-
SyncClipEnabled(clip_distances);
maxwell3d.dirty.flags[Dirty::Shaders] = false;
+
+ const std::span indices_span(image_view_indices.data(), image_view_indices.size());
+ texture_cache.FillGraphicsImageViews(indices_span, image_view_ids);
+
+ size_t image_view_index = 0;
+ size_t texture_index = 0;
+ size_t image_index = 0;
+ for (size_t stage = 0; stage < Maxwell::MaxShaderStage; ++stage) {
+ const Shader* const shader = shaders[stage];
+ if (shader) {
+ const auto base = device.GetBaseBindings(stage);
+ BindTextures(shader->GetEntries(), base.sampler, base.image, image_view_index,
+ texture_index, image_index);
+ }
+ }
}
std::size_t RasterizerOpenGL::CalculateVertexArraysSize() const {
@@ -417,98 +477,6 @@ void RasterizerOpenGL::LoadDiskResources(u64 title_id, const std::atomic_bool& s
shader_cache.LoadDiskCache(title_id, stop_loading, callback);
}
-void RasterizerOpenGL::ConfigureFramebuffers() {
- MICROPROFILE_SCOPE(OpenGL_Framebuffer);
- if (!maxwell3d.dirty.flags[VideoCommon::Dirty::RenderTargets]) {
- return;
- }
- maxwell3d.dirty.flags[VideoCommon::Dirty::RenderTargets] = false;
-
- texture_cache.GuardRenderTargets(true);
-
- View depth_surface = texture_cache.GetDepthBufferSurface(true);
-
- const auto& regs = maxwell3d.regs;
- UNIMPLEMENTED_IF(regs.rt_separate_frag_data == 0);
-
- // Bind the framebuffer surfaces
- FramebufferCacheKey key;
- const auto colors_count = static_cast<std::size_t>(regs.rt_control.count);
- for (std::size_t index = 0; index < colors_count; ++index) {
- View color_surface{texture_cache.GetColorBufferSurface(index, true)};
- if (!color_surface) {
- continue;
- }
- // Assume that a surface will be written to if it is used as a framebuffer, even
- // if the shader doesn't actually write to it.
- texture_cache.MarkColorBufferInUse(index);
-
- key.SetAttachment(index, regs.rt_control.GetMap(index));
- key.colors[index] = std::move(color_surface);
- }
-
- if (depth_surface) {
- // Assume that a surface will be written to if it is used as a framebuffer, even if
- // the shader doesn't actually write to it.
- texture_cache.MarkDepthBufferInUse();
- key.zeta = std::move(depth_surface);
- }
-
- texture_cache.GuardRenderTargets(false);
-
- glBindFramebuffer(GL_DRAW_FRAMEBUFFER, framebuffer_cache.GetFramebuffer(key));
-}
-
-void RasterizerOpenGL::ConfigureClearFramebuffer(bool using_color, bool using_depth_stencil) {
- const auto& regs = maxwell3d.regs;
-
- texture_cache.GuardRenderTargets(true);
- View color_surface;
-
- if (using_color) {
- // Determine if we have to preserve the contents.
- // First we have to make sure all clear masks are enabled.
- bool preserve_contents = !regs.clear_buffers.R || !regs.clear_buffers.G ||
- !regs.clear_buffers.B || !regs.clear_buffers.A;
- const std::size_t index = regs.clear_buffers.RT;
- if (regs.clear_flags.scissor) {
- // Then we have to confirm scissor testing clears the whole image.
- const auto& scissor = regs.scissor_test[0];
- preserve_contents |= scissor.min_x > 0;
- preserve_contents |= scissor.min_y > 0;
- preserve_contents |= scissor.max_x < regs.rt[index].width;
- preserve_contents |= scissor.max_y < regs.rt[index].height;
- }
-
- color_surface = texture_cache.GetColorBufferSurface(index, preserve_contents);
- texture_cache.MarkColorBufferInUse(index);
- }
-
- View depth_surface;
- if (using_depth_stencil) {
- bool preserve_contents = false;
- if (regs.clear_flags.scissor) {
- // For depth stencil clears we only have to confirm scissor test covers the whole image.
- const auto& scissor = regs.scissor_test[0];
- preserve_contents |= scissor.min_x > 0;
- preserve_contents |= scissor.min_y > 0;
- preserve_contents |= scissor.max_x < regs.zeta_width;
- preserve_contents |= scissor.max_y < regs.zeta_height;
- }
-
- depth_surface = texture_cache.GetDepthBufferSurface(preserve_contents);
- texture_cache.MarkDepthBufferInUse();
- }
- texture_cache.GuardRenderTargets(false);
-
- FramebufferCacheKey key;
- key.colors[0] = std::move(color_surface);
- key.zeta = std::move(depth_surface);
-
- state_tracker.NotifyFramebuffer();
- glBindFramebuffer(GL_DRAW_FRAMEBUFFER, framebuffer_cache.GetFramebuffer(key));
-}
-
void RasterizerOpenGL::Clear() {
if (!maxwell3d.ShouldExecute()) {
return;
@@ -523,8 +491,9 @@ void RasterizerOpenGL::Clear() {
regs.clear_buffers.A) {
use_color = true;
- state_tracker.NotifyColorMask0();
- glColorMaski(0, regs.clear_buffers.R != 0, regs.clear_buffers.G != 0,
+ const GLuint index = regs.clear_buffers.RT;
+ state_tracker.NotifyColorMask(index);
+ glColorMaski(index, regs.clear_buffers.R != 0, regs.clear_buffers.G != 0,
regs.clear_buffers.B != 0, regs.clear_buffers.A != 0);
// TODO(Rodrigo): Determine if clamping is used on clears
@@ -557,15 +526,17 @@ void RasterizerOpenGL::Clear() {
state_tracker.NotifyScissor0();
glDisablei(GL_SCISSOR_TEST, 0);
}
-
UNIMPLEMENTED_IF(regs.clear_flags.viewport);
- ConfigureClearFramebuffer(use_color, use_depth || use_stencil);
+ {
+ auto lock = texture_cache.AcquireLock();
+ texture_cache.UpdateRenderTargets(true);
+ state_tracker.BindFramebuffer(texture_cache.GetFramebuffer()->Handle());
+ }
if (use_color) {
- glClearBufferfv(GL_COLOR, 0, regs.clear_color);
+ glClearBufferfv(GL_COLOR, regs.clear_buffers.RT, regs.clear_color);
}
-
if (use_depth && use_stencil) {
glClearBufferfi(GL_DEPTH_STENCIL, 0, regs.clear_depth, regs.clear_stencil);
} else if (use_depth) {
@@ -622,16 +593,7 @@ void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) {
(Maxwell::MaxConstBufferSize + device.GetUniformBufferAlignment());
// Prepare the vertex array.
- const bool invalidated = buffer_cache.Map(buffer_size);
-
- if (invalidated) {
- // When the stream buffer has been invalidated, we have to consider vertex buffers as dirty
- auto& dirty = maxwell3d.dirty.flags;
- dirty[Dirty::VertexBuffers] = true;
- for (int index = Dirty::VertexBuffer0; index <= Dirty::VertexBuffer31; ++index) {
- dirty[index] = true;
- }
- }
+ buffer_cache.Map(buffer_size);
// Prepare vertex array format.
SetupVertexFormat();
@@ -655,22 +617,16 @@ void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) {
}
// Setup shaders and their used resources.
- texture_cache.GuardSamplers(true);
- const GLenum primitive_mode = MaxwellToGL::PrimitiveTopology(maxwell3d.regs.draw.topology);
- SetupShaders(primitive_mode);
- texture_cache.GuardSamplers(false);
-
- ConfigureFramebuffers();
+ auto lock = texture_cache.AcquireLock();
+ SetupShaders();
// Signal the buffer cache that we are not going to upload more things.
buffer_cache.Unmap();
-
+ texture_cache.UpdateRenderTargets(false);
+ state_tracker.BindFramebuffer(texture_cache.GetFramebuffer()->Handle());
program_manager.BindGraphicsPipeline();
- if (texture_cache.TextureBarrier()) {
- glTextureBarrier();
- }
-
+ const GLenum primitive_mode = MaxwellToGL::PrimitiveTopology(maxwell3d.regs.draw.topology);
BeginTransformFeedback(primitive_mode);
const GLuint base_instance = static_cast<GLuint>(maxwell3d.regs.vb_base_instance);
@@ -722,15 +678,13 @@ void RasterizerOpenGL::DispatchCompute(GPUVAddr code_addr) {
buffer_cache.Acquire();
current_cbuf = 0;
- auto kernel = shader_cache.GetComputeKernel(code_addr);
- program_manager.BindCompute(kernel->GetHandle());
+ Shader* const kernel = shader_cache.GetComputeKernel(code_addr);
- SetupComputeTextures(kernel);
- SetupComputeImages(kernel);
+ auto lock = texture_cache.AcquireLock();
+ BindComputeTextures(kernel);
- const std::size_t buffer_size =
- Tegra::Engines::KeplerCompute::NumConstBuffers *
- (Maxwell::MaxConstBufferSize + device.GetUniformBufferAlignment());
+ const size_t buffer_size = Tegra::Engines::KeplerCompute::NumConstBuffers *
+ (Maxwell::MaxConstBufferSize + device.GetUniformBufferAlignment());
buffer_cache.Map(buffer_size);
SetupComputeConstBuffers(kernel);
@@ -739,7 +693,6 @@ void RasterizerOpenGL::DispatchCompute(GPUVAddr code_addr) {
buffer_cache.Unmap();
const auto& launch_desc = kepler_compute.launch_description;
- program_manager.BindCompute(kernel->GetHandle());
glDispatchCompute(launch_desc.grid_dim_x, launch_desc.grid_dim_y, launch_desc.grid_dim_z);
++num_queued_commands;
}
@@ -760,7 +713,10 @@ void RasterizerOpenGL::FlushRegion(VAddr addr, u64 size) {
if (addr == 0 || size == 0) {
return;
}
- texture_cache.FlushRegion(addr, size);
+ {
+ auto lock = texture_cache.AcquireLock();
+ texture_cache.DownloadMemory(addr, size);
+ }
buffer_cache.FlushRegion(addr, size);
query_cache.FlushRegion(addr, size);
}
@@ -769,7 +725,8 @@ bool RasterizerOpenGL::MustFlushRegion(VAddr addr, u64 size) {
if (!Settings::IsGPULevelHigh()) {
return buffer_cache.MustFlushRegion(addr, size);
}
- return texture_cache.MustFlushRegion(addr, size) || buffer_cache.MustFlushRegion(addr, size);
+ return texture_cache.IsRegionGpuModified(addr, size) ||
+ buffer_cache.MustFlushRegion(addr, size);
}
void RasterizerOpenGL::InvalidateRegion(VAddr addr, u64 size) {
@@ -777,7 +734,10 @@ void RasterizerOpenGL::InvalidateRegion(VAddr addr, u64 size) {
if (addr == 0 || size == 0) {
return;
}
- texture_cache.InvalidateRegion(addr, size);
+ {
+ auto lock = texture_cache.AcquireLock();
+ texture_cache.WriteMemory(addr, size);
+ }
shader_cache.InvalidateRegion(addr, size);
buffer_cache.InvalidateRegion(addr, size);
query_cache.InvalidateRegion(addr, size);
@@ -788,18 +748,29 @@ void RasterizerOpenGL::OnCPUWrite(VAddr addr, u64 size) {
if (addr == 0 || size == 0) {
return;
}
- texture_cache.OnCPUWrite(addr, size);
+ {
+ auto lock = texture_cache.AcquireLock();
+ texture_cache.WriteMemory(addr, size);
+ }
shader_cache.OnCPUWrite(addr, size);
buffer_cache.OnCPUWrite(addr, size);
}
void RasterizerOpenGL::SyncGuestHost() {
MICROPROFILE_SCOPE(OpenGL_CacheManagement);
- texture_cache.SyncGuestHost();
buffer_cache.SyncGuestHost();
shader_cache.SyncGuestHost();
}
+void RasterizerOpenGL::UnmapMemory(VAddr addr, u64 size) {
+ {
+ auto lock = texture_cache.AcquireLock();
+ texture_cache.UnmapMemory(addr, size);
+ }
+ buffer_cache.OnCPUWrite(addr, size);
+ shader_cache.OnCPUWrite(addr, size);
+}
+
void RasterizerOpenGL::SignalSemaphore(GPUVAddr addr, u32 value) {
if (!gpu.IsAsync()) {
gpu_memory.Write<u32>(addr, value);
@@ -841,6 +812,14 @@ void RasterizerOpenGL::WaitForIdle() {
GL_SHADER_STORAGE_BARRIER_BIT | GL_QUERY_BUFFER_BARRIER_BIT);
}
+void RasterizerOpenGL::FragmentBarrier() {
+ glMemoryBarrier(GL_FRAMEBUFFER_BARRIER_BIT);
+}
+
+void RasterizerOpenGL::TiledCacheBarrier() {
+ glTextureBarrier();
+}
+
void RasterizerOpenGL::FlushCommands() {
// Only flush when we have commands queued to OpenGL.
if (num_queued_commands == 0) {
@@ -854,45 +833,95 @@ void RasterizerOpenGL::TickFrame() {
// Ticking a frame means that buffers will be swapped, calling glFlush implicitly.
num_queued_commands = 0;
+ fence_manager.TickFrame();
buffer_cache.TickFrame();
+ {
+ auto lock = texture_cache.AcquireLock();
+ texture_cache.TickFrame();
+ }
}
-bool RasterizerOpenGL::AccelerateSurfaceCopy(const Tegra::Engines::Fermi2D::Regs::Surface& src,
- const Tegra::Engines::Fermi2D::Regs::Surface& dst,
+bool RasterizerOpenGL::AccelerateSurfaceCopy(const Tegra::Engines::Fermi2D::Surface& src,
+ const Tegra::Engines::Fermi2D::Surface& dst,
const Tegra::Engines::Fermi2D::Config& copy_config) {
MICROPROFILE_SCOPE(OpenGL_Blits);
- texture_cache.DoFermiCopy(src, dst, copy_config);
+ auto lock = texture_cache.AcquireLock();
+ texture_cache.BlitImage(dst, src, copy_config);
return true;
}
bool RasterizerOpenGL::AccelerateDisplay(const Tegra::FramebufferConfig& config,
VAddr framebuffer_addr, u32 pixel_stride) {
- if (!framebuffer_addr) {
- return {};
+ if (framebuffer_addr == 0) {
+ return false;
}
-
MICROPROFILE_SCOPE(OpenGL_CacheManagement);
- const auto surface{texture_cache.TryFindFramebufferSurface(framebuffer_addr)};
- if (!surface) {
- return {};
+ auto lock = texture_cache.AcquireLock();
+ ImageView* const image_view{texture_cache.TryFindFramebufferImageView(framebuffer_addr)};
+ if (!image_view) {
+ return false;
}
-
// Verify that the cached surface is the same size and format as the requested framebuffer
- const auto& params{surface->GetSurfaceParams()};
- const auto& pixel_format{
- VideoCore::Surface::PixelFormatFromGPUPixelFormat(config.pixel_format)};
- ASSERT_MSG(params.width == config.width, "Framebuffer width is different");
- ASSERT_MSG(params.height == config.height, "Framebuffer height is different");
+ // ASSERT_MSG(image_view->size.width == config.width, "Framebuffer width is different");
+ // ASSERT_MSG(image_view->size.height == config.height, "Framebuffer height is different");
- if (params.pixel_format != pixel_format) {
- LOG_DEBUG(Render_OpenGL, "Framebuffer pixel_format is different");
- }
+ screen_info.display_texture = image_view->Handle(ImageViewType::e2D);
+ screen_info.display_srgb = VideoCore::Surface::IsPixelFormatSRGB(image_view->format);
+ return true;
+}
- screen_info.display_texture = surface->GetTexture();
- screen_info.display_srgb = surface->GetSurfaceParams().srgb_conversion;
+void RasterizerOpenGL::BindComputeTextures(Shader* kernel) {
+ image_view_indices.clear();
+ sampler_handles.clear();
- return true;
+ texture_cache.SynchronizeComputeDescriptors();
+
+ SetupComputeTextures(kernel);
+ SetupComputeImages(kernel);
+
+ const std::span indices_span(image_view_indices.data(), image_view_indices.size());
+ texture_cache.FillComputeImageViews(indices_span, image_view_ids);
+
+ program_manager.BindCompute(kernel->GetHandle());
+ size_t image_view_index = 0;
+ size_t texture_index = 0;
+ size_t image_index = 0;
+ BindTextures(kernel->GetEntries(), 0, 0, image_view_index, texture_index, image_index);
+}
+
+void RasterizerOpenGL::BindTextures(const ShaderEntries& entries, GLuint base_texture,
+ GLuint base_image, size_t& image_view_index,
+ size_t& texture_index, size_t& image_index) {
+ const GLuint* const samplers = sampler_handles.data() + texture_index;
+ const GLuint* const textures = texture_handles.data() + texture_index;
+ const GLuint* const images = image_handles.data() + image_index;
+
+ const size_t num_samplers = entries.samplers.size();
+ for (const auto& sampler : entries.samplers) {
+ for (size_t i = 0; i < sampler.size; ++i) {
+ const ImageViewId image_view_id = image_view_ids[image_view_index++];
+ const ImageView& image_view = texture_cache.GetImageView(image_view_id);
+ const GLuint handle = image_view.Handle(ImageViewTypeFromEntry(sampler));
+ texture_handles[texture_index++] = handle;
+ }
+ }
+ const size_t num_images = entries.images.size();
+ for (size_t unit = 0; unit < num_images; ++unit) {
+ // TODO: Mark as modified
+ const ImageViewId image_view_id = image_view_ids[image_view_index++];
+ const ImageView& image_view = texture_cache.GetImageView(image_view_id);
+ const GLuint handle = image_view.Handle(ImageViewTypeFromEntry(entries.images[unit]));
+ image_handles[image_index] = handle;
+ ++image_index;
+ }
+ if (num_samplers > 0) {
+ glBindSamplers(base_texture, static_cast<GLsizei>(num_samplers), samplers);
+ glBindTextures(base_texture, static_cast<GLsizei>(num_samplers), textures);
+ }
+ if (num_images > 0) {
+ glBindImageTextures(base_image, static_cast<GLsizei>(num_images), images);
+ }
}
void RasterizerOpenGL::SetupDrawConstBuffers(std::size_t stage_index, Shader* shader) {
@@ -999,7 +1028,6 @@ void RasterizerOpenGL::SetupDrawGlobalMemory(std::size_t stage_index, Shader* sh
GL_VERTEX_PROGRAM_NV, GL_TESS_CONTROL_PROGRAM_NV, GL_TESS_EVALUATION_PROGRAM_NV,
GL_GEOMETRY_PROGRAM_NV, GL_FRAGMENT_PROGRAM_NV,
};
-
const auto& cbufs{maxwell3d.state.shader_stages[stage_index]};
const auto& entries{shader->GetEntries().global_memory_entries};
@@ -1056,77 +1084,53 @@ void RasterizerOpenGL::SetupGlobalMemory(u32 binding, const GlobalMemoryEntry& e
}
}
-void RasterizerOpenGL::SetupDrawTextures(std::size_t stage_index, Shader* shader) {
- MICROPROFILE_SCOPE(OpenGL_Texture);
- u32 binding = device.GetBaseBindings(stage_index).sampler;
+void RasterizerOpenGL::SetupDrawTextures(const Shader* shader, size_t stage_index) {
+ const bool via_header_index =
+ maxwell3d.regs.sampler_index == Maxwell::SamplerIndex::ViaHeaderIndex;
for (const auto& entry : shader->GetEntries().samplers) {
const auto shader_type = static_cast<ShaderType>(stage_index);
- for (std::size_t i = 0; i < entry.size; ++i) {
- const auto texture = GetTextureInfo(maxwell3d, entry, shader_type, i);
- SetupTexture(binding++, texture, entry);
+ for (size_t index = 0; index < entry.size; ++index) {
+ const auto handle =
+ GetTextureInfo(maxwell3d, via_header_index, entry, shader_type, index);
+ const Sampler* const sampler = texture_cache.GetGraphicsSampler(handle.sampler);
+ sampler_handles.push_back(sampler->Handle());
+ image_view_indices.push_back(handle.image);
}
}
}
-void RasterizerOpenGL::SetupComputeTextures(Shader* kernel) {
- MICROPROFILE_SCOPE(OpenGL_Texture);
- u32 binding = 0;
+void RasterizerOpenGL::SetupComputeTextures(const Shader* kernel) {
+ const bool via_header_index = kepler_compute.launch_description.linked_tsc;
for (const auto& entry : kernel->GetEntries().samplers) {
- for (std::size_t i = 0; i < entry.size; ++i) {
- const auto texture = GetTextureInfo(kepler_compute, entry, ShaderType::Compute, i);
- SetupTexture(binding++, texture, entry);
+ for (size_t i = 0; i < entry.size; ++i) {
+ const auto handle =
+ GetTextureInfo(kepler_compute, via_header_index, entry, ShaderType::Compute, i);
+ const Sampler* const sampler = texture_cache.GetComputeSampler(handle.sampler);
+ sampler_handles.push_back(sampler->Handle());
+ image_view_indices.push_back(handle.image);
}
}
}
-void RasterizerOpenGL::SetupTexture(u32 binding, const Tegra::Texture::FullTextureInfo& texture,
- const SamplerEntry& entry) {
- const auto view = texture_cache.GetTextureSurface(texture.tic, entry);
- if (!view) {
- // Can occur when texture addr is null or its memory is unmapped/invalid
- glBindSampler(binding, 0);
- glBindTextureUnit(binding, 0);
- return;
- }
- const GLuint handle = view->GetTexture(texture.tic.x_source, texture.tic.y_source,
- texture.tic.z_source, texture.tic.w_source);
- glBindTextureUnit(binding, handle);
- if (!view->GetSurfaceParams().IsBuffer()) {
- glBindSampler(binding, sampler_cache.GetSampler(texture.tsc));
- }
-}
-
-void RasterizerOpenGL::SetupDrawImages(std::size_t stage_index, Shader* shader) {
- u32 binding = device.GetBaseBindings(stage_index).image;
+void RasterizerOpenGL::SetupDrawImages(const Shader* shader, size_t stage_index) {
+ const bool via_header_index =
+ maxwell3d.regs.sampler_index == Maxwell::SamplerIndex::ViaHeaderIndex;
for (const auto& entry : shader->GetEntries().images) {
const auto shader_type = static_cast<ShaderType>(stage_index);
- const auto tic = GetTextureInfo(maxwell3d, entry, shader_type).tic;
- SetupImage(binding++, tic, entry);
+ const auto handle = GetTextureInfo(maxwell3d, via_header_index, entry, shader_type);
+ image_view_indices.push_back(handle.image);
}
}
-void RasterizerOpenGL::SetupComputeImages(Shader* shader) {
- u32 binding = 0;
+void RasterizerOpenGL::SetupComputeImages(const Shader* shader) {
+ const bool via_header_index = kepler_compute.launch_description.linked_tsc;
for (const auto& entry : shader->GetEntries().images) {
- const auto tic = GetTextureInfo(kepler_compute, entry, ShaderType::Compute).tic;
- SetupImage(binding++, tic, entry);
+ const auto handle =
+ GetTextureInfo(kepler_compute, via_header_index, entry, ShaderType::Compute);
+ image_view_indices.push_back(handle.image);
}
}
-void RasterizerOpenGL::SetupImage(u32 binding, const Tegra::Texture::TICEntry& tic,
- const ImageEntry& entry) {
- const auto view = texture_cache.GetImageSurface(tic, entry);
- if (!view) {
- glBindImageTexture(binding, 0, 0, GL_FALSE, 0, GL_READ_ONLY, GL_R8);
- return;
- }
- if (entry.is_written) {
- view->MarkAsModified(texture_cache.Tick());
- }
- const GLuint handle = view->GetTexture(tic.x_source, tic.y_source, tic.z_source, tic.w_source);
- glBindImageTexture(binding, handle, 0, GL_TRUE, 0, GL_READ_WRITE, view->GetFormat());
-}
-
void RasterizerOpenGL::SyncViewport() {
auto& flags = maxwell3d.dirty.flags;
const auto& regs = maxwell3d.regs;
@@ -1526,17 +1530,9 @@ void RasterizerOpenGL::SyncPointState() {
flags[Dirty::PointSize] = false;
oglEnable(GL_POINT_SPRITE, maxwell3d.regs.point_sprite_enable);
+ oglEnable(GL_PROGRAM_POINT_SIZE, maxwell3d.regs.vp_point_size.enable);
- if (maxwell3d.regs.vp_point_size.enable) {
- // By definition of GL_POINT_SIZE, it only matters if GL_PROGRAM_POINT_SIZE is disabled.
- glEnable(GL_PROGRAM_POINT_SIZE);
- return;
- }
-
- // Limit the point size to 1 since nouveau sometimes sets a point size of 0 (and that's invalid
- // in OpenGL).
glPointSize(std::max(1.0f, maxwell3d.regs.point_size));
- glDisable(GL_PROGRAM_POINT_SIZE);
}
void RasterizerOpenGL::SyncLineState() {
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.h b/src/video_core/renderer_opengl/gl_rasterizer.h
index 1d0f585fa..82e03e677 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.h
+++ b/src/video_core/renderer_opengl/gl_rasterizer.h
@@ -7,12 +7,13 @@
#include <array>
#include <atomic>
#include <cstddef>
-#include <map>
#include <memory>
#include <optional>
#include <tuple>
#include <utility>
+#include <boost/container/static_vector.hpp>
+
#include <glad/glad.h>
#include "common/common_types.h"
@@ -23,16 +24,14 @@
#include "video_core/renderer_opengl/gl_buffer_cache.h"
#include "video_core/renderer_opengl/gl_device.h"
#include "video_core/renderer_opengl/gl_fence_manager.h"
-#include "video_core/renderer_opengl/gl_framebuffer_cache.h"
#include "video_core/renderer_opengl/gl_query_cache.h"
#include "video_core/renderer_opengl/gl_resource_manager.h"
-#include "video_core/renderer_opengl/gl_sampler_cache.h"
#include "video_core/renderer_opengl/gl_shader_cache.h"
#include "video_core/renderer_opengl/gl_shader_decompiler.h"
#include "video_core/renderer_opengl/gl_shader_manager.h"
#include "video_core/renderer_opengl/gl_state_tracker.h"
+#include "video_core/renderer_opengl/gl_stream_buffer.h"
#include "video_core/renderer_opengl/gl_texture_cache.h"
-#include "video_core/renderer_opengl/utils.h"
#include "video_core/shader/async_shaders.h"
#include "video_core/textures/texture.h"
@@ -51,7 +50,7 @@ class MemoryManager;
namespace OpenGL {
struct ScreenInfo;
-struct DrawParameters;
+struct ShaderEntries;
struct BindlessSSBO {
GLuint64EXT address;
@@ -62,10 +61,10 @@ static_assert(sizeof(BindlessSSBO) * CHAR_BIT == 128);
class RasterizerOpenGL : public VideoCore::RasterizerAccelerated {
public:
- explicit RasterizerOpenGL(Core::Frontend::EmuWindow& emu_window, Tegra::GPU& gpu,
- Core::Memory::Memory& cpu_memory, const Device& device,
- ScreenInfo& screen_info, ProgramManager& program_manager,
- StateTracker& state_tracker);
+ explicit RasterizerOpenGL(Core::Frontend::EmuWindow& emu_window_, Tegra::GPU& gpu_,
+ Core::Memory::Memory& cpu_memory_, const Device& device_,
+ ScreenInfo& screen_info_, ProgramManager& program_manager_,
+ StateTracker& state_tracker_);
~RasterizerOpenGL() override;
void Draw(bool is_indexed, bool is_instanced) override;
@@ -79,15 +78,18 @@ public:
void InvalidateRegion(VAddr addr, u64 size) override;
void OnCPUWrite(VAddr addr, u64 size) override;
void SyncGuestHost() override;
+ void UnmapMemory(VAddr addr, u64 size) override;
void SignalSemaphore(GPUVAddr addr, u32 value) override;
void SignalSyncPoint(u32 value) override;
void ReleaseFences() override;
void FlushAndInvalidateRegion(VAddr addr, u64 size) override;
void WaitForIdle() override;
+ void FragmentBarrier() override;
+ void TiledCacheBarrier() override;
void FlushCommands() override;
void TickFrame() override;
- bool AccelerateSurfaceCopy(const Tegra::Engines::Fermi2D::Regs::Surface& src,
- const Tegra::Engines::Fermi2D::Regs::Surface& dst,
+ bool AccelerateSurfaceCopy(const Tegra::Engines::Fermi2D::Surface& src,
+ const Tegra::Engines::Fermi2D::Surface& dst,
const Tegra::Engines::Fermi2D::Config& copy_config) override;
bool AccelerateDisplay(const Tegra::FramebufferConfig& config, VAddr framebuffer_addr,
u32 pixel_stride) override;
@@ -108,11 +110,14 @@ public:
}
private:
- /// Configures the color and depth framebuffer states.
- void ConfigureFramebuffers();
+ static constexpr size_t MAX_TEXTURES = 192;
+ static constexpr size_t MAX_IMAGES = 48;
+ static constexpr size_t MAX_IMAGE_VIEWS = MAX_TEXTURES + MAX_IMAGES;
+
+ void BindComputeTextures(Shader* kernel);
- /// Configures the color and depth framebuffer for clearing.
- void ConfigureClearFramebuffer(bool using_color, bool using_depth_stencil);
+ void BindTextures(const ShaderEntries& entries, GLuint base_texture, GLuint base_image,
+ size_t& image_view_index, size_t& texture_index, size_t& image_index);
/// Configures the current constbuffers to use for the draw command.
void SetupDrawConstBuffers(std::size_t stage_index, Shader* shader);
@@ -136,23 +141,16 @@ private:
size_t size, BindlessSSBO* ssbo);
/// Configures the current textures to use for the draw command.
- void SetupDrawTextures(std::size_t stage_index, Shader* shader);
+ void SetupDrawTextures(const Shader* shader, size_t stage_index);
/// Configures the textures used in a compute shader.
- void SetupComputeTextures(Shader* kernel);
-
- /// Configures a texture.
- void SetupTexture(u32 binding, const Tegra::Texture::FullTextureInfo& texture,
- const SamplerEntry& entry);
+ void SetupComputeTextures(const Shader* kernel);
/// Configures images in a graphics shader.
- void SetupDrawImages(std::size_t stage_index, Shader* shader);
+ void SetupDrawImages(const Shader* shader, size_t stage_index);
/// Configures images in a compute shader.
- void SetupComputeImages(Shader* shader);
-
- /// Configures an image.
- void SetupImage(u32 binding, const Tegra::Texture::TICEntry& tic, const ImageEntry& entry);
+ void SetupComputeImages(const Shader* shader);
/// Syncs the viewport and depth range to match the guest state
void SyncViewport();
@@ -227,9 +225,6 @@ private:
/// End a transform feedback
void EndTransformFeedback();
- /// Check for extension that are not strictly required but are needed for correct emulation
- void CheckExtensions();
-
std::size_t CalculateVertexArraysSize() const;
std::size_t CalculateIndexBufferSize() const;
@@ -242,7 +237,7 @@ private:
GLintptr SetupIndexBuffer();
- void SetupShaders(GLenum primitive_mode);
+ void SetupShaders();
Tegra::GPU& gpu;
Tegra::Engines::Maxwell3D& maxwell3d;
@@ -254,19 +249,21 @@ private:
ProgramManager& program_manager;
StateTracker& state_tracker;
- TextureCacheOpenGL texture_cache;
+ OGLStreamBuffer stream_buffer;
+ TextureCacheRuntime texture_cache_runtime;
+ TextureCache texture_cache;
ShaderCacheOpenGL shader_cache;
- SamplerCacheOpenGL sampler_cache;
- FramebufferCacheOpenGL framebuffer_cache;
QueryCache query_cache;
OGLBufferCache buffer_cache;
FenceManagerOpenGL fence_manager;
VideoCommon::Shader::AsyncShaders async_shaders;
- static constexpr std::size_t STREAM_BUFFER_SIZE = 128 * 1024 * 1024;
-
- GLint vertex_binding = 0;
+ boost::container::static_vector<u32, MAX_IMAGE_VIEWS> image_view_indices;
+ std::array<ImageViewId, MAX_IMAGE_VIEWS> image_view_ids;
+ boost::container::static_vector<GLuint, MAX_TEXTURES> sampler_handles;
+ std::array<GLuint, MAX_TEXTURES> texture_handles;
+ std::array<GLuint, MAX_IMAGES> image_handles;
std::array<OGLBuffer, Tegra::Engines::Maxwell3D::Regs::NumTransformFeedbackBuffers>
transform_feedback_buffers;
@@ -280,7 +277,7 @@ private:
std::size_t current_cbuf = 0;
OGLBuffer unified_uniform_buffer;
- /// Number of commands queued to the OpenGL driver. Reseted on flush.
+ /// Number of commands queued to the OpenGL driver. Resetted on flush.
std::size_t num_queued_commands = 0;
u32 last_clip_distance_mask = 0;
diff --git a/src/video_core/renderer_opengl/gl_resource_manager.cpp b/src/video_core/renderer_opengl/gl_resource_manager.cpp
index 0ebcec427..0e34a0f20 100644
--- a/src/video_core/renderer_opengl/gl_resource_manager.cpp
+++ b/src/video_core/renderer_opengl/gl_resource_manager.cpp
@@ -71,7 +71,7 @@ void OGLSampler::Create() {
return;
MICROPROFILE_SCOPE(OpenGL_ResourceCreation);
- glGenSamplers(1, &handle);
+ glCreateSamplers(1, &handle);
}
void OGLSampler::Release() {
diff --git a/src/video_core/renderer_opengl/gl_sampler_cache.cpp b/src/video_core/renderer_opengl/gl_sampler_cache.cpp
deleted file mode 100644
index 5c174879a..000000000
--- a/src/video_core/renderer_opengl/gl_sampler_cache.cpp
+++ /dev/null
@@ -1,52 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include "common/logging/log.h"
-#include "video_core/renderer_opengl/gl_resource_manager.h"
-#include "video_core/renderer_opengl/gl_sampler_cache.h"
-#include "video_core/renderer_opengl/maxwell_to_gl.h"
-
-namespace OpenGL {
-
-SamplerCacheOpenGL::SamplerCacheOpenGL() = default;
-
-SamplerCacheOpenGL::~SamplerCacheOpenGL() = default;
-
-OGLSampler SamplerCacheOpenGL::CreateSampler(const Tegra::Texture::TSCEntry& tsc) const {
- OGLSampler sampler;
- sampler.Create();
-
- const GLuint sampler_id{sampler.handle};
- glSamplerParameteri(
- sampler_id, GL_TEXTURE_MAG_FILTER,
- MaxwellToGL::TextureFilterMode(tsc.mag_filter, Tegra::Texture::TextureMipmapFilter::None));
- glSamplerParameteri(sampler_id, GL_TEXTURE_MIN_FILTER,
- MaxwellToGL::TextureFilterMode(tsc.min_filter, tsc.mipmap_filter));
- glSamplerParameteri(sampler_id, GL_TEXTURE_WRAP_S, MaxwellToGL::WrapMode(tsc.wrap_u));
- glSamplerParameteri(sampler_id, GL_TEXTURE_WRAP_T, MaxwellToGL::WrapMode(tsc.wrap_v));
- glSamplerParameteri(sampler_id, GL_TEXTURE_WRAP_R, MaxwellToGL::WrapMode(tsc.wrap_p));
- glSamplerParameteri(sampler_id, GL_TEXTURE_COMPARE_MODE,
- tsc.depth_compare_enabled == 1 ? GL_COMPARE_REF_TO_TEXTURE : GL_NONE);
- glSamplerParameteri(sampler_id, GL_TEXTURE_COMPARE_FUNC,
- MaxwellToGL::DepthCompareFunc(tsc.depth_compare_func));
- glSamplerParameterfv(sampler_id, GL_TEXTURE_BORDER_COLOR, tsc.GetBorderColor().data());
- glSamplerParameterf(sampler_id, GL_TEXTURE_MIN_LOD, tsc.GetMinLod());
- glSamplerParameterf(sampler_id, GL_TEXTURE_MAX_LOD, tsc.GetMaxLod());
- glSamplerParameterf(sampler_id, GL_TEXTURE_LOD_BIAS, tsc.GetLodBias());
- if (GLAD_GL_ARB_texture_filter_anisotropic) {
- glSamplerParameterf(sampler_id, GL_TEXTURE_MAX_ANISOTROPY, tsc.GetMaxAnisotropy());
- } else if (GLAD_GL_EXT_texture_filter_anisotropic) {
- glSamplerParameterf(sampler_id, GL_TEXTURE_MAX_ANISOTROPY_EXT, tsc.GetMaxAnisotropy());
- } else {
- LOG_WARNING(Render_OpenGL, "Anisotropy not supported by host GPU driver");
- }
-
- return sampler;
-}
-
-GLuint SamplerCacheOpenGL::ToSamplerType(const OGLSampler& sampler) const {
- return sampler.handle;
-}
-
-} // namespace OpenGL
diff --git a/src/video_core/renderer_opengl/gl_sampler_cache.h b/src/video_core/renderer_opengl/gl_sampler_cache.h
deleted file mode 100644
index 34ee37f00..000000000
--- a/src/video_core/renderer_opengl/gl_sampler_cache.h
+++ /dev/null
@@ -1,25 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <glad/glad.h>
-
-#include "video_core/renderer_opengl/gl_resource_manager.h"
-#include "video_core/sampler_cache.h"
-
-namespace OpenGL {
-
-class SamplerCacheOpenGL final : public VideoCommon::SamplerCache<GLuint, OGLSampler> {
-public:
- explicit SamplerCacheOpenGL();
- ~SamplerCacheOpenGL();
-
-protected:
- OGLSampler CreateSampler(const Tegra::Texture::TSCEntry& tsc) const override;
-
- GLuint ToSamplerType(const OGLSampler& sampler) const override;
-};
-
-} // namespace OpenGL
diff --git a/src/video_core/renderer_opengl/gl_shader_cache.cpp b/src/video_core/renderer_opengl/gl_shader_cache.cpp
index bd56bed0c..d4841fdb7 100644
--- a/src/video_core/renderer_opengl/gl_shader_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_cache.cpp
@@ -27,7 +27,6 @@
#include "video_core/renderer_opengl/gl_shader_decompiler.h"
#include "video_core/renderer_opengl/gl_shader_disk_cache.h"
#include "video_core/renderer_opengl/gl_state_tracker.h"
-#include "video_core/renderer_opengl/utils.h"
#include "video_core/shader/memory_util.h"
#include "video_core/shader/registry.h"
#include "video_core/shader/shader_ir.h"
@@ -198,10 +197,10 @@ ProgramSharedPtr BuildShader(const Device& device, ShaderType shader_type, u64 u
return program;
}
-Shader::Shader(std::shared_ptr<VideoCommon::Shader::Registry> registry_, ShaderEntries entries_,
- ProgramSharedPtr program_, bool is_built)
+Shader::Shader(std::shared_ptr<Registry> registry_, ShaderEntries entries_,
+ ProgramSharedPtr program_, bool is_built_)
: registry{std::move(registry_)}, entries{std::move(entries_)}, program{std::move(program_)},
- is_built(is_built) {
+ is_built{is_built_} {
handle = program->assembly_program.handle;
if (handle == 0) {
handle = program->source_program.handle;
@@ -318,14 +317,13 @@ std::unique_ptr<Shader> Shader::CreateFromCache(const ShaderParameters& params,
precompiled_shader.registry, precompiled_shader.entries, precompiled_shader.program));
}
-ShaderCacheOpenGL::ShaderCacheOpenGL(RasterizerOpenGL& rasterizer,
+ShaderCacheOpenGL::ShaderCacheOpenGL(RasterizerOpenGL& rasterizer_,
Core::Frontend::EmuWindow& emu_window_, Tegra::GPU& gpu_,
Tegra::Engines::Maxwell3D& maxwell3d_,
Tegra::Engines::KeplerCompute& kepler_compute_,
Tegra::MemoryManager& gpu_memory_, const Device& device_)
- : VideoCommon::ShaderCache<Shader>{rasterizer}, emu_window{emu_window_}, gpu{gpu_},
- gpu_memory{gpu_memory_}, maxwell3d{maxwell3d_},
- kepler_compute{kepler_compute_}, device{device_} {}
+ : ShaderCache{rasterizer_}, emu_window{emu_window_}, gpu{gpu_}, gpu_memory{gpu_memory_},
+ maxwell3d{maxwell3d_}, kepler_compute{kepler_compute_}, device{device_} {}
ShaderCacheOpenGL::~ShaderCacheOpenGL() = default;
@@ -460,7 +458,7 @@ void ShaderCacheOpenGL::LoadDiskCache(u64 title_id, const std::atomic_bool& stop
ProgramSharedPtr ShaderCacheOpenGL::GeneratePrecompiledProgram(
const ShaderDiskCacheEntry& entry, const ShaderDiskCachePrecompiled& precompiled_entry,
const std::unordered_set<GLenum>& supported_formats) {
- if (supported_formats.find(precompiled_entry.binary_format) == supported_formats.end()) {
+ if (!supported_formats.contains(precompiled_entry.binary_format)) {
LOG_INFO(Render_OpenGL, "Precompiled cache entry with unsupported format, removing");
return {};
}
diff --git a/src/video_core/renderer_opengl/gl_shader_cache.h b/src/video_core/renderer_opengl/gl_shader_cache.h
index 1708af06a..2aed0697e 100644
--- a/src/video_core/renderer_opengl/gl_shader_cache.h
+++ b/src/video_core/renderer_opengl/gl_shader_cache.h
@@ -108,7 +108,7 @@ public:
private:
explicit Shader(std::shared_ptr<VideoCommon::Shader::Registry> registry, ShaderEntries entries,
- ProgramSharedPtr program, bool is_built = true);
+ ProgramSharedPtr program, bool is_built_ = true);
std::shared_ptr<VideoCommon::Shader::Registry> registry;
ShaderEntries entries;
@@ -119,10 +119,11 @@ private:
class ShaderCacheOpenGL final : public VideoCommon::ShaderCache<Shader> {
public:
- explicit ShaderCacheOpenGL(RasterizerOpenGL& rasterizer, Core::Frontend::EmuWindow& emu_window,
- Tegra::GPU& gpu, Tegra::Engines::Maxwell3D& maxwell3d,
- Tegra::Engines::KeplerCompute& kepler_compute,
- Tegra::MemoryManager& gpu_memory, const Device& device);
+ explicit ShaderCacheOpenGL(RasterizerOpenGL& rasterizer_,
+ Core::Frontend::EmuWindow& emu_window_, Tegra::GPU& gpu,
+ Tegra::Engines::Maxwell3D& maxwell3d_,
+ Tegra::Engines::KeplerCompute& kepler_compute_,
+ Tegra::MemoryManager& gpu_memory_, const Device& device_);
~ShaderCacheOpenGL() override;
/// Loads disk cache for the current game
diff --git a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
index 95ca96c8e..2e1fa252d 100644
--- a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
@@ -38,11 +38,9 @@ using Tegra::Shader::IpaSampleMode;
using Tegra::Shader::PixelImap;
using Tegra::Shader::Register;
using Tegra::Shader::TextureType;
-using VideoCommon::Shader::BuildTransformFeedback;
-using VideoCommon::Shader::Registry;
-using namespace std::string_literals;
using namespace VideoCommon::Shader;
+using namespace std::string_literals;
using Maxwell = Tegra::Engines::Maxwell3D::Regs;
using Operation = const OperationNode&;
@@ -131,7 +129,7 @@ private:
class Expression final {
public:
- Expression(std::string code, Type type) : code{std::move(code)}, type{type} {
+ Expression(std::string code_, Type type_) : code{std::move(code_)}, type{type_} {
ASSERT(type != Type::Void);
}
Expression() : type{Type::Void} {}
@@ -148,8 +146,8 @@ public:
ASSERT(type == Type::Void);
}
- std::string As(Type type) const {
- switch (type) {
+ std::string As(Type type_) const {
+ switch (type_) {
case Type::Bool:
return AsBool();
case Type::Bool2:
@@ -316,7 +314,7 @@ std::pair<const char*, u32> GetPrimitiveDescription(Maxwell::PrimitiveTopology t
case Maxwell::PrimitiveTopology::TriangleStripAdjacency:
return {"triangles_adjacency", 6};
default:
- UNIMPLEMENTED_MSG("topology={}", static_cast<int>(topology));
+ UNIMPLEMENTED_MSG("topology={}", topology);
return {"points", 1};
}
}
@@ -342,7 +340,7 @@ std::string GetTopologyName(Tegra::Shader::OutputTopology topology) {
case Tegra::Shader::OutputTopology::TriangleStrip:
return "triangle_strip";
default:
- UNIMPLEMENTED_MSG("Unknown output topology: {}", static_cast<u32>(topology));
+ UNIMPLEMENTED_MSG("Unknown output topology: {}", topology);
return "points";
}
}
@@ -418,11 +416,12 @@ struct GenericVaryingDescription {
class GLSLDecompiler final {
public:
- explicit GLSLDecompiler(const Device& device, const ShaderIR& ir, const Registry& registry,
- ShaderType stage, std::string_view identifier, std::string_view suffix)
- : device{device}, ir{ir}, registry{registry}, stage{stage}, identifier{identifier},
- suffix{suffix}, header{ir.GetHeader()}, use_unified_uniforms{
- UseUnifiedUniforms(device, ir, stage)} {
+ explicit GLSLDecompiler(const Device& device_, const ShaderIR& ir_, const Registry& registry_,
+ ShaderType stage_, std::string_view identifier_,
+ std::string_view suffix_)
+ : device{device_}, ir{ir_}, registry{registry_}, stage{stage_}, identifier{identifier_},
+ suffix{suffix_}, header{ir.GetHeader()}, use_unified_uniforms{
+ UseUnifiedUniforms(device_, ir_, stage_)} {
if (stage != ShaderType::Compute) {
transform_feedback = BuildTransformFeedback(registry.GetGraphicsInfo());
}
@@ -744,7 +743,7 @@ private:
case PixelImap::Unused:
break;
}
- UNIMPLEMENTED_MSG("Unknown attribute usage index={}", static_cast<int>(attribute));
+ UNIMPLEMENTED_MSG("Unknown attribute usage index={}", attribute);
return {};
}
@@ -777,16 +776,16 @@ private:
name = "gs_" + name + "[]";
}
- std::string suffix;
+ std::string suffix_;
if (stage == ShaderType::Fragment) {
const auto input_mode{header.ps.GetPixelImap(location)};
if (input_mode == PixelImap::Unused) {
return;
}
- suffix = GetInputFlags(input_mode);
+ suffix_ = GetInputFlags(input_mode);
}
- code.AddLine("layout (location = {}) {} in vec4 {};", location, suffix, name);
+ code.AddLine("layout (location = {}) {} in vec4 {};", location, suffix_, name);
}
void DeclareOutputAttributes() {
@@ -877,7 +876,7 @@ private:
}
u32 binding = device.GetBaseBindings(stage).uniform_buffer;
- for (const auto [index, info] : ir.GetConstantBuffers()) {
+ for (const auto& [index, info] : ir.GetConstantBuffers()) {
const u32 num_elements = Common::AlignUp(info.GetSize(), 4) / 4;
const u32 size = info.IsIndirect() ? MAX_CONSTBUFFER_ELEMENTS : num_elements;
code.AddLine("layout (std140, binding = {}) uniform {} {{", binding++,
@@ -1251,7 +1250,7 @@ private:
}
break;
}
- UNIMPLEMENTED_MSG("Unhandled input attribute: {}", static_cast<u32>(attribute));
+ UNIMPLEMENTED_MSG("Unhandled input attribute: {}", attribute);
return {"0", Type::Int};
}
@@ -1331,7 +1330,7 @@ private:
GetSwizzle(element)),
Type::Float}};
}
- UNIMPLEMENTED_MSG("Unhandled output attribute: {}", static_cast<u32>(attribute));
+ UNIMPLEMENTED_MSG("Unhandled output attribute: {}", attribute);
return std::nullopt;
}
}
@@ -2100,13 +2099,13 @@ private:
const auto type = meta.sampler.is_shadow ? Type::Float : Type::Int;
const bool separate_dc = meta.sampler.is_shadow;
- std::vector<TextureIR> ir;
+ std::vector<TextureIR> ir_;
if (meta.sampler.is_shadow) {
- ir = {TextureOffset{}};
+ ir_ = {TextureOffset{}};
} else {
- ir = {TextureOffset{}, TextureArgument{type, meta.component}};
+ ir_ = {TextureOffset{}, TextureArgument{type, meta.component}};
}
- return {GenerateTexture(operation, "Gather", ir, separate_dc) + GetSwizzle(meta.element),
+ return {GenerateTexture(operation, "Gather", ir_, separate_dc) + GetSwizzle(meta.element),
Type::Float};
}
@@ -2752,11 +2751,11 @@ private:
}
}
- std::string GetSampler(const Sampler& sampler) const {
+ std::string GetSampler(const SamplerEntry& sampler) const {
return AppendSuffix(sampler.index, "sampler");
}
- std::string GetImage(const Image& image) const {
+ std::string GetImage(const ImageEntry& image) const {
return AppendSuffix(image.index, "image");
}
@@ -2801,7 +2800,7 @@ std::string GetFlowVariable(u32 index) {
class ExprDecompiler {
public:
- explicit ExprDecompiler(GLSLDecompiler& decomp) : decomp{decomp} {}
+ explicit ExprDecompiler(GLSLDecompiler& decomp_) : decomp{decomp_} {}
void operator()(const ExprAnd& expr) {
inner += '(';
@@ -2856,7 +2855,7 @@ private:
class ASTDecompiler {
public:
- explicit ASTDecompiler(GLSLDecompiler& decomp) : decomp{decomp} {}
+ explicit ASTDecompiler(GLSLDecompiler& decomp_) : decomp{decomp_} {}
void operator()(const ASTProgram& ast) {
ASTNode current = ast.nodes.GetFirst();
diff --git a/src/video_core/renderer_opengl/gl_shader_decompiler.h b/src/video_core/renderer_opengl/gl_shader_decompiler.h
index 451c9689a..be68994bb 100644
--- a/src/video_core/renderer_opengl/gl_shader_decompiler.h
+++ b/src/video_core/renderer_opengl/gl_shader_decompiler.h
@@ -20,13 +20,13 @@ namespace OpenGL {
class Device;
using Maxwell = Tegra::Engines::Maxwell3D::Regs;
-using SamplerEntry = VideoCommon::Shader::Sampler;
-using ImageEntry = VideoCommon::Shader::Image;
+using SamplerEntry = VideoCommon::Shader::SamplerEntry;
+using ImageEntry = VideoCommon::Shader::ImageEntry;
class ConstBufferEntry : public VideoCommon::Shader::ConstBuffer {
public:
- explicit ConstBufferEntry(u32 max_offset, bool is_indirect, u32 index)
- : VideoCommon::Shader::ConstBuffer{max_offset, is_indirect}, index{index} {}
+ explicit ConstBufferEntry(u32 max_offset_, bool is_indirect_, u32 index_)
+ : ConstBuffer{max_offset_, is_indirect_}, index{index_} {}
u32 GetIndex() const {
return index;
@@ -37,10 +37,10 @@ private:
};
struct GlobalMemoryEntry {
- constexpr explicit GlobalMemoryEntry(u32 cbuf_index, u32 cbuf_offset, bool is_read,
- bool is_written)
- : cbuf_index{cbuf_index}, cbuf_offset{cbuf_offset}, is_read{is_read}, is_written{
- is_written} {}
+ constexpr explicit GlobalMemoryEntry(u32 cbuf_index_, u32 cbuf_offset_, bool is_read_,
+ bool is_written_)
+ : cbuf_index{cbuf_index_}, cbuf_offset{cbuf_offset_}, is_read{is_read_}, is_written{
+ is_written_} {}
u32 cbuf_index = 0;
u32 cbuf_offset = 0;
diff --git a/src/video_core/renderer_opengl/gl_shader_disk_cache.cpp b/src/video_core/renderer_opengl/gl_shader_disk_cache.cpp
index 70dd0c3c6..955b2abc4 100644
--- a/src/video_core/renderer_opengl/gl_shader_disk_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_disk_cache.cpp
@@ -343,7 +343,7 @@ void ShaderDiskCacheOpenGL::SaveEntry(const ShaderDiskCacheEntry& entry) {
}
const u64 id = entry.unique_identifier;
- if (stored_transferable.find(id) != stored_transferable.end()) {
+ if (stored_transferable.contains(id)) {
// The shader already exists
return;
}
diff --git a/src/video_core/renderer_opengl/gl_shader_manager.cpp b/src/video_core/renderer_opengl/gl_shader_manager.cpp
index 691c6c79b..553e6e8d6 100644
--- a/src/video_core/renderer_opengl/gl_shader_manager.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_manager.cpp
@@ -83,6 +83,21 @@ void ProgramManager::RestoreGuestPipeline() {
}
}
+void ProgramManager::BindHostCompute(GLuint program) {
+ if (use_assembly_programs) {
+ glDisable(GL_COMPUTE_PROGRAM_NV);
+ }
+ glUseProgram(program);
+ is_graphics_bound = false;
+}
+
+void ProgramManager::RestoreGuestCompute() {
+ if (use_assembly_programs) {
+ glEnable(GL_COMPUTE_PROGRAM_NV);
+ glUseProgram(0);
+ }
+}
+
void ProgramManager::UseVertexShader(GLuint program) {
if (use_assembly_programs) {
BindProgram(GL_VERTEX_PROGRAM_NV, program, current_state.vertex, vertex_enabled);
diff --git a/src/video_core/renderer_opengl/gl_shader_manager.h b/src/video_core/renderer_opengl/gl_shader_manager.h
index 950e0dfcb..ad42cce74 100644
--- a/src/video_core/renderer_opengl/gl_shader_manager.h
+++ b/src/video_core/renderer_opengl/gl_shader_manager.h
@@ -45,6 +45,12 @@ public:
/// Rewinds BindHostPipeline state changes.
void RestoreGuestPipeline();
+ /// Binds an OpenGL GLSL program object unsynchronized with the guest state.
+ void BindHostCompute(GLuint program);
+
+ /// Rewinds BindHostCompute state changes.
+ void RestoreGuestCompute();
+
void UseVertexShader(GLuint program);
void UseGeometryShader(GLuint program);
void UseFragmentShader(GLuint program);
diff --git a/src/video_core/renderer_opengl/gl_state_tracker.cpp b/src/video_core/renderer_opengl/gl_state_tracker.cpp
index 6bcf831f2..60e6fa39f 100644
--- a/src/video_core/renderer_opengl/gl_state_tracker.cpp
+++ b/src/video_core/renderer_opengl/gl_state_tracker.cpp
@@ -13,7 +13,7 @@
#include "video_core/renderer_opengl/gl_state_tracker.h"
#define OFF(field_name) MAXWELL3D_REG_INDEX(field_name)
-#define NUM(field_name) (sizeof(Maxwell3D::Regs::field_name) / sizeof(u32))
+#define NUM(field_name) (sizeof(Maxwell3D::Regs::field_name) / (sizeof(u32)))
namespace OpenGL {
@@ -249,4 +249,11 @@ StateTracker::StateTracker(Tegra::GPU& gpu) : flags{gpu.Maxwell3D().dirty.flags}
}
}
+void StateTracker::InvalidateStreamBuffer() {
+ flags[Dirty::VertexBuffers] = true;
+ for (int index = Dirty::VertexBuffer0; index <= Dirty::VertexBuffer31; ++index) {
+ flags[index] = true;
+ }
+}
+
} // namespace OpenGL
diff --git a/src/video_core/renderer_opengl/gl_state_tracker.h b/src/video_core/renderer_opengl/gl_state_tracker.h
index 9d127548f..574615d3c 100644
--- a/src/video_core/renderer_opengl/gl_state_tracker.h
+++ b/src/video_core/renderer_opengl/gl_state_tracker.h
@@ -92,6 +92,8 @@ class StateTracker {
public:
explicit StateTracker(Tegra::GPU& gpu);
+ void InvalidateStreamBuffer();
+
void BindIndexBuffer(GLuint new_index_buffer) {
if (index_buffer == new_index_buffer) {
return;
@@ -100,6 +102,14 @@ public:
glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, new_index_buffer);
}
+ void BindFramebuffer(GLuint new_framebuffer) {
+ if (framebuffer == new_framebuffer) {
+ return;
+ }
+ framebuffer = new_framebuffer;
+ glBindFramebuffer(GL_DRAW_FRAMEBUFFER, framebuffer);
+ }
+
void NotifyScreenDrawVertexArray() {
flags[OpenGL::Dirty::VertexFormats] = true;
flags[OpenGL::Dirty::VertexFormat0 + 0] = true;
@@ -129,9 +139,9 @@ public:
flags[OpenGL::Dirty::Scissor0] = true;
}
- void NotifyColorMask0() {
+ void NotifyColorMask(size_t index) {
flags[OpenGL::Dirty::ColorMasks] = true;
- flags[OpenGL::Dirty::ColorMask0] = true;
+ flags[OpenGL::Dirty::ColorMask0 + index] = true;
}
void NotifyBlend0() {
@@ -190,6 +200,7 @@ public:
private:
Tegra::Engines::Maxwell3D::DirtyState::Flags& flags;
+ GLuint framebuffer = 0;
GLuint index_buffer = 0;
};
diff --git a/src/video_core/renderer_opengl/gl_stream_buffer.cpp b/src/video_core/renderer_opengl/gl_stream_buffer.cpp
index 887995cf4..e0819cdf2 100644
--- a/src/video_core/renderer_opengl/gl_stream_buffer.cpp
+++ b/src/video_core/renderer_opengl/gl_stream_buffer.cpp
@@ -9,6 +9,7 @@
#include "common/assert.h"
#include "common/microprofile.h"
#include "video_core/renderer_opengl/gl_device.h"
+#include "video_core/renderer_opengl/gl_state_tracker.h"
#include "video_core/renderer_opengl/gl_stream_buffer.h"
MICROPROFILE_DEFINE(OpenGL_StreamBuffer, "OpenGL", "Stream Buffer Orphaning",
@@ -16,24 +17,14 @@ MICROPROFILE_DEFINE(OpenGL_StreamBuffer, "OpenGL", "Stream Buffer Orphaning",
namespace OpenGL {
-OGLStreamBuffer::OGLStreamBuffer(const Device& device, GLsizeiptr size, bool vertex_data_usage)
- : buffer_size(size) {
+OGLStreamBuffer::OGLStreamBuffer(const Device& device, StateTracker& state_tracker_)
+ : state_tracker{state_tracker_} {
gl_buffer.Create();
- GLsizeiptr allocate_size = size;
- if (vertex_data_usage) {
- // On AMD GPU there is a strange crash in indexed drawing. The crash happens when the buffer
- // read position is near the end and is an out-of-bound access to the vertex buffer. This is
- // probably a bug in the driver and is related to the usage of vec3<byte> attributes in the
- // vertex array. Doubling the allocation size for the vertex buffer seems to avoid the
- // crash.
- allocate_size *= 2;
- }
-
static constexpr GLbitfield flags = GL_MAP_WRITE_BIT | GL_MAP_PERSISTENT_BIT;
- glNamedBufferStorage(gl_buffer.handle, allocate_size, nullptr, flags);
+ glNamedBufferStorage(gl_buffer.handle, BUFFER_SIZE, nullptr, flags);
mapped_ptr = static_cast<u8*>(
- glMapNamedBufferRange(gl_buffer.handle, 0, buffer_size, flags | GL_MAP_FLUSH_EXPLICIT_BIT));
+ glMapNamedBufferRange(gl_buffer.handle, 0, BUFFER_SIZE, flags | GL_MAP_FLUSH_EXPLICIT_BIT));
if (device.UseAssemblyShaders() || device.HasVertexBufferUnifiedMemory()) {
glMakeNamedBufferResidentNV(gl_buffer.handle, GL_READ_ONLY);
@@ -46,25 +37,24 @@ OGLStreamBuffer::~OGLStreamBuffer() {
gl_buffer.Release();
}
-std::tuple<u8*, GLintptr, bool> OGLStreamBuffer::Map(GLsizeiptr size, GLintptr alignment) {
- ASSERT(size <= buffer_size);
- ASSERT(alignment <= buffer_size);
+std::pair<u8*, GLintptr> OGLStreamBuffer::Map(GLsizeiptr size, GLintptr alignment) {
+ ASSERT(size <= BUFFER_SIZE);
+ ASSERT(alignment <= BUFFER_SIZE);
mapped_size = size;
if (alignment > 0) {
buffer_pos = Common::AlignUp<std::size_t>(buffer_pos, alignment);
}
- bool invalidate = false;
- if (buffer_pos + size > buffer_size) {
+ if (buffer_pos + size > BUFFER_SIZE) {
MICROPROFILE_SCOPE(OpenGL_StreamBuffer);
glInvalidateBufferData(gl_buffer.handle);
+ state_tracker.InvalidateStreamBuffer();
buffer_pos = 0;
- invalidate = true;
}
- return std::make_tuple(mapped_ptr + buffer_pos, buffer_pos, invalidate);
+ return std::make_pair(mapped_ptr + buffer_pos, buffer_pos);
}
void OGLStreamBuffer::Unmap(GLsizeiptr size) {
diff --git a/src/video_core/renderer_opengl/gl_stream_buffer.h b/src/video_core/renderer_opengl/gl_stream_buffer.h
index 307a67113..dd9cf67eb 100644
--- a/src/video_core/renderer_opengl/gl_stream_buffer.h
+++ b/src/video_core/renderer_opengl/gl_stream_buffer.h
@@ -4,29 +4,31 @@
#pragma once
-#include <tuple>
+#include <utility>
+
#include <glad/glad.h>
+
#include "common/common_types.h"
#include "video_core/renderer_opengl/gl_resource_manager.h"
namespace OpenGL {
class Device;
+class StateTracker;
class OGLStreamBuffer : private NonCopyable {
public:
- explicit OGLStreamBuffer(const Device& device, GLsizeiptr size, bool vertex_data_usage);
+ explicit OGLStreamBuffer(const Device& device, StateTracker& state_tracker_);
~OGLStreamBuffer();
/*
* Allocates a linear chunk of memory in the GPU buffer with at least "size" bytes
* and the optional alignment requirement.
* If the buffer is full, the whole buffer is reallocated which invalidates old chunks.
- * The return values are the pointer to the new chunk, the offset within the buffer,
- * and the invalidation flag for previous chunks.
+ * The return values are the pointer to the new chunk, and the offset within the buffer.
* The actual used size must be specified on unmapping the chunk.
*/
- std::tuple<u8*, GLintptr, bool> Map(GLsizeiptr size, GLintptr alignment = 0);
+ std::pair<u8*, GLintptr> Map(GLsizeiptr size, GLintptr alignment = 0);
void Unmap(GLsizeiptr size);
@@ -39,15 +41,18 @@ public:
}
GLsizeiptr Size() const noexcept {
- return buffer_size;
+ return BUFFER_SIZE;
}
private:
+ static constexpr GLsizeiptr BUFFER_SIZE = 256 * 1024 * 1024;
+
+ StateTracker& state_tracker;
+
OGLBuffer gl_buffer;
GLuint64EXT gpu_address = 0;
GLintptr buffer_pos = 0;
- GLsizeiptr buffer_size = 0;
GLsizeiptr mapped_size = 0;
u8* mapped_ptr = nullptr;
};
diff --git a/src/video_core/renderer_opengl/gl_texture_cache.cpp b/src/video_core/renderer_opengl/gl_texture_cache.cpp
index a863ef218..4c690418c 100644
--- a/src/video_core/renderer_opengl/gl_texture_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_texture_cache.cpp
@@ -2,173 +2,238 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
-#include "common/assert.h"
-#include "common/bit_util.h"
-#include "common/common_types.h"
-#include "common/microprofile.h"
-#include "common/scope_exit.h"
-#include "core/core.h"
-#include "video_core/morton.h"
-#include "video_core/renderer_opengl/gl_resource_manager.h"
+#include <algorithm>
+#include <array>
+#include <bit>
+#include <string>
+
+#include <glad/glad.h>
+
+#include "video_core/renderer_opengl/gl_device.h"
+#include "video_core/renderer_opengl/gl_shader_manager.h"
#include "video_core/renderer_opengl/gl_state_tracker.h"
#include "video_core/renderer_opengl/gl_texture_cache.h"
-#include "video_core/renderer_opengl/utils.h"
-#include "video_core/texture_cache/surface_base.h"
+#include "video_core/renderer_opengl/maxwell_to_gl.h"
+#include "video_core/renderer_opengl/util_shaders.h"
+#include "video_core/surface.h"
+#include "video_core/texture_cache/format_lookup_table.h"
+#include "video_core/texture_cache/samples_helper.h"
#include "video_core/texture_cache/texture_cache.h"
-#include "video_core/textures/convert.h"
-#include "video_core/textures/texture.h"
+#include "video_core/textures/decoders.h"
namespace OpenGL {
-using Tegra::Texture::SwizzleSource;
-using VideoCore::MortonSwizzleMode;
+namespace {
+using Tegra::Texture::SwizzleSource;
+using Tegra::Texture::TextureMipmapFilter;
+using Tegra::Texture::TextureType;
+using Tegra::Texture::TICEntry;
+using Tegra::Texture::TSCEntry;
+using VideoCommon::CalculateLevelStrideAlignment;
+using VideoCommon::ImageCopy;
+using VideoCommon::ImageFlagBits;
+using VideoCommon::ImageType;
+using VideoCommon::NUM_RT;
+using VideoCommon::SamplesLog2;
+using VideoCommon::SwizzleParameters;
+using VideoCore::Surface::BytesPerBlock;
+using VideoCore::Surface::IsPixelFormatASTC;
+using VideoCore::Surface::IsPixelFormatSRGB;
+using VideoCore::Surface::MaxPixelFormat;
using VideoCore::Surface::PixelFormat;
-using VideoCore::Surface::SurfaceTarget;
using VideoCore::Surface::SurfaceType;
-MICROPROFILE_DEFINE(OpenGL_Texture_Upload, "OpenGL", "Texture Upload", MP_RGB(128, 192, 128));
-MICROPROFILE_DEFINE(OpenGL_Texture_Download, "OpenGL", "Texture Download", MP_RGB(128, 192, 128));
-MICROPROFILE_DEFINE(OpenGL_Texture_Buffer_Copy, "OpenGL", "Texture Buffer Copy",
- MP_RGB(128, 192, 128));
+struct CopyOrigin {
+ GLint level;
+ GLint x;
+ GLint y;
+ GLint z;
+};
-namespace {
+struct CopyRegion {
+ GLsizei width;
+ GLsizei height;
+ GLsizei depth;
+};
struct FormatTuple {
GLenum internal_format;
GLenum format = GL_NONE;
GLenum type = GL_NONE;
+ GLenum store_format = internal_format;
};
-constexpr std::array<FormatTuple, VideoCore::Surface::MaxPixelFormat> tex_format_tuples = {{
- {GL_RGBA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV}, // A8B8G8R8_UNORM
- {GL_RGBA8_SNORM, GL_RGBA, GL_BYTE}, // A8B8G8R8_SNORM
- {GL_RGBA8I, GL_RGBA_INTEGER, GL_BYTE}, // A8B8G8R8_SINT
- {GL_RGBA8UI, GL_RGBA_INTEGER, GL_UNSIGNED_BYTE}, // A8B8G8R8_UINT
- {GL_RGB565, GL_RGB, GL_UNSIGNED_SHORT_5_6_5}, // R5G6B5_UNORM
- {GL_RGB565, GL_RGB, GL_UNSIGNED_SHORT_5_6_5_REV}, // B5G6R5_UNORM
- {GL_RGB5_A1, GL_BGRA, GL_UNSIGNED_SHORT_1_5_5_5_REV}, // A1R5G5B5_UNORM
- {GL_RGB10_A2, GL_RGBA, GL_UNSIGNED_INT_2_10_10_10_REV}, // A2B10G10R10_UNORM
- {GL_RGB10_A2UI, GL_RGBA_INTEGER, GL_UNSIGNED_INT_2_10_10_10_REV}, // A2B10G10R10_UINT
- {GL_RGB5_A1, GL_RGBA, GL_UNSIGNED_SHORT_1_5_5_5_REV}, // A1B5G5R5_UNORM
- {GL_R8, GL_RED, GL_UNSIGNED_BYTE}, // R8_UNORM
- {GL_R8_SNORM, GL_RED, GL_BYTE}, // R8_SNORM
- {GL_R8I, GL_RED_INTEGER, GL_BYTE}, // R8_SINT
- {GL_R8UI, GL_RED_INTEGER, GL_UNSIGNED_BYTE}, // R8_UINT
- {GL_RGBA16F, GL_RGBA, GL_HALF_FLOAT}, // R16G16B16A16_FLOAT
- {GL_RGBA16, GL_RGBA, GL_UNSIGNED_SHORT}, // R16G16B16A16_UNORM
- {GL_RGBA16_SNORM, GL_RGBA, GL_SHORT}, // R16G16B16A16_SNORM
- {GL_RGBA16I, GL_RGBA_INTEGER, GL_SHORT}, // R16G16B16A16_SINT
- {GL_RGBA16UI, GL_RGBA_INTEGER, GL_UNSIGNED_SHORT}, // R16G16B16A16_UINT
- {GL_R11F_G11F_B10F, GL_RGB, GL_UNSIGNED_INT_10F_11F_11F_REV}, // B10G11R11_FLOAT
- {GL_RGBA32UI, GL_RGBA_INTEGER, GL_UNSIGNED_INT}, // R32G32B32A32_UINT
- {GL_COMPRESSED_RGBA_S3TC_DXT1_EXT}, // BC1_RGBA_UNORM
- {GL_COMPRESSED_RGBA_S3TC_DXT3_EXT}, // BC2_UNORM
- {GL_COMPRESSED_RGBA_S3TC_DXT5_EXT}, // BC3_UNORM
- {GL_COMPRESSED_RED_RGTC1}, // BC4_UNORM
- {GL_COMPRESSED_SIGNED_RED_RGTC1}, // BC4_SNORM
- {GL_COMPRESSED_RG_RGTC2}, // BC5_UNORM
- {GL_COMPRESSED_SIGNED_RG_RGTC2}, // BC5_SNORM
- {GL_COMPRESSED_RGBA_BPTC_UNORM}, // BC7_UNORM
- {GL_COMPRESSED_RGB_BPTC_UNSIGNED_FLOAT}, // BC6H_UFLOAT
- {GL_COMPRESSED_RGB_BPTC_SIGNED_FLOAT}, // BC6H_SFLOAT
- {GL_COMPRESSED_RGBA_ASTC_4x4_KHR}, // ASTC_2D_4X4_UNORM
- {GL_RGBA8, GL_BGRA, GL_UNSIGNED_BYTE}, // B8G8R8A8_UNORM
- {GL_RGBA32F, GL_RGBA, GL_FLOAT}, // R32G32B32A32_FLOAT
- {GL_RGBA32I, GL_RGBA_INTEGER, GL_INT}, // R32G32B32A32_SINT
- {GL_RG32F, GL_RG, GL_FLOAT}, // R32G32_FLOAT
- {GL_RG32I, GL_RG_INTEGER, GL_INT}, // R32G32_SINT
- {GL_R32F, GL_RED, GL_FLOAT}, // R32_FLOAT
- {GL_R16F, GL_RED, GL_HALF_FLOAT}, // R16_FLOAT
- {GL_R16, GL_RED, GL_UNSIGNED_SHORT}, // R16_UNORM
- {GL_R16_SNORM, GL_RED, GL_SHORT}, // R16_SNORM
- {GL_R16UI, GL_RED_INTEGER, GL_UNSIGNED_SHORT}, // R16_UINT
- {GL_R16I, GL_RED_INTEGER, GL_SHORT}, // R16_SINT
- {GL_RG16, GL_RG, GL_UNSIGNED_SHORT}, // R16G16_UNORM
- {GL_RG16F, GL_RG, GL_HALF_FLOAT}, // R16G16_FLOAT
- {GL_RG16UI, GL_RG_INTEGER, GL_UNSIGNED_SHORT}, // R16G16_UINT
- {GL_RG16I, GL_RG_INTEGER, GL_SHORT}, // R16G16_SINT
- {GL_RG16_SNORM, GL_RG, GL_SHORT}, // R16G16_SNORM
- {GL_RGB32F, GL_RGB, GL_FLOAT}, // R32G32B32_FLOAT
- {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV}, // A8B8G8R8_SRGB
- {GL_RG8, GL_RG, GL_UNSIGNED_BYTE}, // R8G8_UNORM
- {GL_RG8_SNORM, GL_RG, GL_BYTE}, // R8G8_SNORM
- {GL_RG8I, GL_RG_INTEGER, GL_BYTE}, // R8G8_SINT
- {GL_RG8UI, GL_RG_INTEGER, GL_UNSIGNED_BYTE}, // R8G8_UINT
- {GL_RG32UI, GL_RG_INTEGER, GL_UNSIGNED_INT}, // R32G32_UINT
- {GL_RGB16F, GL_RGBA, GL_HALF_FLOAT}, // R16G16B16X16_FLOAT
- {GL_R32UI, GL_RED_INTEGER, GL_UNSIGNED_INT}, // R32_UINT
- {GL_R32I, GL_RED_INTEGER, GL_INT}, // R32_SINT
- {GL_COMPRESSED_RGBA_ASTC_8x8_KHR}, // ASTC_2D_8X8_UNORM
- {GL_COMPRESSED_RGBA_ASTC_8x5_KHR}, // ASTC_2D_8X5_UNORM
- {GL_COMPRESSED_RGBA_ASTC_5x4_KHR}, // ASTC_2D_5X4_UNORM
- {GL_SRGB8_ALPHA8, GL_BGRA, GL_UNSIGNED_BYTE}, // B8G8R8A8_UNORM
- // Compressed sRGB formats
- {GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT1_EXT}, // BC1_RGBA_SRGB
- {GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT3_EXT}, // BC2_SRGB
- {GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT}, // BC3_SRGB
- {GL_COMPRESSED_SRGB_ALPHA_BPTC_UNORM}, // BC7_SRGB
- {GL_RGBA4, GL_RGBA, GL_UNSIGNED_SHORT_4_4_4_4_REV}, // A4B4G4R4_UNORM
- {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_4x4_KHR}, // ASTC_2D_4X4_SRGB
- {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_8x8_KHR}, // ASTC_2D_8X8_SRGB
- {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_8x5_KHR}, // ASTC_2D_8X5_SRGB
- {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_5x4_KHR}, // ASTC_2D_5X4_SRGB
- {GL_COMPRESSED_RGBA_ASTC_5x5_KHR}, // ASTC_2D_5X5_UNORM
- {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_5x5_KHR}, // ASTC_2D_5X5_SRGB
- {GL_COMPRESSED_RGBA_ASTC_10x8_KHR}, // ASTC_2D_10X8_UNORM
- {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_10x8_KHR}, // ASTC_2D_10X8_SRGB
- {GL_COMPRESSED_RGBA_ASTC_6x6_KHR}, // ASTC_2D_6X6_UNORM
- {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_6x6_KHR}, // ASTC_2D_6X6_SRGB
- {GL_COMPRESSED_RGBA_ASTC_10x10_KHR}, // ASTC_2D_10X10_UNORM
- {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_10x10_KHR}, // ASTC_2D_10X10_SRGB
- {GL_COMPRESSED_RGBA_ASTC_12x12_KHR}, // ASTC_2D_12X12_UNORM
- {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_12x12_KHR}, // ASTC_2D_12X12_SRGB
- {GL_COMPRESSED_RGBA_ASTC_8x6_KHR}, // ASTC_2D_8X6_UNORM
- {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_8x6_KHR}, // ASTC_2D_8X6_SRGB
- {GL_COMPRESSED_RGBA_ASTC_6x5_KHR}, // ASTC_2D_6X5_UNORM
- {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_6x5_KHR}, // ASTC_2D_6X5_SRGB
- {GL_RGB9_E5, GL_RGB, GL_UNSIGNED_INT_5_9_9_9_REV}, // E5B9G9R9_FLOAT
-
- // Depth formats
- {GL_DEPTH_COMPONENT32F, GL_DEPTH_COMPONENT, GL_FLOAT}, // D32_FLOAT
- {GL_DEPTH_COMPONENT16, GL_DEPTH_COMPONENT, GL_UNSIGNED_SHORT}, // D16_UNORM
-
- // DepthStencil formats
- {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8}, // D24_UNORM_S8_UINT
- {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8}, // S8_UINT_D24_UNORM
+constexpr std::array<FormatTuple, MaxPixelFormat> FORMAT_TABLE = {{
+ {GL_RGBA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV}, // A8B8G8R8_UNORM
+ {GL_RGBA8_SNORM, GL_RGBA, GL_BYTE}, // A8B8G8R8_SNORM
+ {GL_RGBA8I, GL_RGBA_INTEGER, GL_BYTE}, // A8B8G8R8_SINT
+ {GL_RGBA8UI, GL_RGBA_INTEGER, GL_UNSIGNED_BYTE}, // A8B8G8R8_UINT
+ {GL_RGB565, GL_RGB, GL_UNSIGNED_SHORT_5_6_5}, // R5G6B5_UNORM
+ {GL_RGB565, GL_RGB, GL_UNSIGNED_SHORT_5_6_5_REV}, // B5G6R5_UNORM
+ {GL_RGB5_A1, GL_BGRA, GL_UNSIGNED_SHORT_1_5_5_5_REV}, // A1R5G5B5_UNORM
+ {GL_RGB10_A2, GL_RGBA, GL_UNSIGNED_INT_2_10_10_10_REV}, // A2B10G10R10_UNORM
+ {GL_RGB10_A2UI, GL_RGBA_INTEGER, GL_UNSIGNED_INT_2_10_10_10_REV}, // A2B10G10R10_UINT
+ {GL_RGB5_A1, GL_RGBA, GL_UNSIGNED_SHORT_1_5_5_5_REV}, // A1B5G5R5_UNORM
+ {GL_R8, GL_RED, GL_UNSIGNED_BYTE}, // R8_UNORM
+ {GL_R8_SNORM, GL_RED, GL_BYTE}, // R8_SNORM
+ {GL_R8I, GL_RED_INTEGER, GL_BYTE}, // R8_SINT
+ {GL_R8UI, GL_RED_INTEGER, GL_UNSIGNED_BYTE}, // R8_UINT
+ {GL_RGBA16F, GL_RGBA, GL_HALF_FLOAT}, // R16G16B16A16_FLOAT
+ {GL_RGBA16, GL_RGBA, GL_UNSIGNED_SHORT}, // R16G16B16A16_UNORM
+ {GL_RGBA16_SNORM, GL_RGBA, GL_SHORT}, // R16G16B16A16_SNORM
+ {GL_RGBA16I, GL_RGBA_INTEGER, GL_SHORT}, // R16G16B16A16_SINT
+ {GL_RGBA16UI, GL_RGBA_INTEGER, GL_UNSIGNED_SHORT}, // R16G16B16A16_UINT
+ {GL_R11F_G11F_B10F, GL_RGB, GL_UNSIGNED_INT_10F_11F_11F_REV}, // B10G11R11_FLOAT
+ {GL_RGBA32UI, GL_RGBA_INTEGER, GL_UNSIGNED_INT}, // R32G32B32A32_UINT
+ {GL_COMPRESSED_RGBA_S3TC_DXT1_EXT}, // BC1_RGBA_UNORM
+ {GL_COMPRESSED_RGBA_S3TC_DXT3_EXT}, // BC2_UNORM
+ {GL_COMPRESSED_RGBA_S3TC_DXT5_EXT}, // BC3_UNORM
+ {GL_COMPRESSED_RED_RGTC1}, // BC4_UNORM
+ {GL_COMPRESSED_SIGNED_RED_RGTC1}, // BC4_SNORM
+ {GL_COMPRESSED_RG_RGTC2}, // BC5_UNORM
+ {GL_COMPRESSED_SIGNED_RG_RGTC2}, // BC5_SNORM
+ {GL_COMPRESSED_RGBA_BPTC_UNORM}, // BC7_UNORM
+ {GL_COMPRESSED_RGB_BPTC_UNSIGNED_FLOAT}, // BC6H_UFLOAT
+ {GL_COMPRESSED_RGB_BPTC_SIGNED_FLOAT}, // BC6H_SFLOAT
+ {GL_COMPRESSED_RGBA_ASTC_4x4_KHR}, // ASTC_2D_4X4_UNORM
+ {GL_RGBA8, GL_BGRA, GL_UNSIGNED_BYTE}, // B8G8R8A8_UNORM
+ {GL_RGBA32F, GL_RGBA, GL_FLOAT}, // R32G32B32A32_FLOAT
+ {GL_RGBA32I, GL_RGBA_INTEGER, GL_INT}, // R32G32B32A32_SINT
+ {GL_RG32F, GL_RG, GL_FLOAT}, // R32G32_FLOAT
+ {GL_RG32I, GL_RG_INTEGER, GL_INT}, // R32G32_SINT
+ {GL_R32F, GL_RED, GL_FLOAT}, // R32_FLOAT
+ {GL_R16F, GL_RED, GL_HALF_FLOAT}, // R16_FLOAT
+ {GL_R16, GL_RED, GL_UNSIGNED_SHORT}, // R16_UNORM
+ {GL_R16_SNORM, GL_RED, GL_SHORT}, // R16_SNORM
+ {GL_R16UI, GL_RED_INTEGER, GL_UNSIGNED_SHORT}, // R16_UINT
+ {GL_R16I, GL_RED_INTEGER, GL_SHORT}, // R16_SINT
+ {GL_RG16, GL_RG, GL_UNSIGNED_SHORT}, // R16G16_UNORM
+ {GL_RG16F, GL_RG, GL_HALF_FLOAT}, // R16G16_FLOAT
+ {GL_RG16UI, GL_RG_INTEGER, GL_UNSIGNED_SHORT}, // R16G16_UINT
+ {GL_RG16I, GL_RG_INTEGER, GL_SHORT}, // R16G16_SINT
+ {GL_RG16_SNORM, GL_RG, GL_SHORT}, // R16G16_SNORM
+ {GL_RGB32F, GL_RGB, GL_FLOAT}, // R32G32B32_FLOAT
+ {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV, GL_RGBA8}, // A8B8G8R8_SRGB
+ {GL_RG8, GL_RG, GL_UNSIGNED_BYTE}, // R8G8_UNORM
+ {GL_RG8_SNORM, GL_RG, GL_BYTE}, // R8G8_SNORM
+ {GL_RG8I, GL_RG_INTEGER, GL_BYTE}, // R8G8_SINT
+ {GL_RG8UI, GL_RG_INTEGER, GL_UNSIGNED_BYTE}, // R8G8_UINT
+ {GL_RG32UI, GL_RG_INTEGER, GL_UNSIGNED_INT}, // R32G32_UINT
+ {GL_RGB16F, GL_RGBA, GL_HALF_FLOAT}, // R16G16B16X16_FLOAT
+ {GL_R32UI, GL_RED_INTEGER, GL_UNSIGNED_INT}, // R32_UINT
+ {GL_R32I, GL_RED_INTEGER, GL_INT}, // R32_SINT
+ {GL_COMPRESSED_RGBA_ASTC_8x8_KHR}, // ASTC_2D_8X8_UNORM
+ {GL_COMPRESSED_RGBA_ASTC_8x5_KHR}, // ASTC_2D_8X5_UNORM
+ {GL_COMPRESSED_RGBA_ASTC_5x4_KHR}, // ASTC_2D_5X4_UNORM
+ {GL_SRGB8_ALPHA8, GL_BGRA, GL_UNSIGNED_BYTE, GL_RGBA8}, // B8G8R8A8_UNORM
+ {GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT1_EXT}, // BC1_RGBA_SRGB
+ {GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT3_EXT}, // BC2_SRGB
+ {GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT}, // BC3_SRGB
+ {GL_COMPRESSED_SRGB_ALPHA_BPTC_UNORM}, // BC7_SRGB
+ {GL_RGBA4, GL_RGBA, GL_UNSIGNED_SHORT_4_4_4_4_REV}, // A4B4G4R4_UNORM
+ {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_4x4_KHR}, // ASTC_2D_4X4_SRGB
+ {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_8x8_KHR}, // ASTC_2D_8X8_SRGB
+ {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_8x5_KHR}, // ASTC_2D_8X5_SRGB
+ {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_5x4_KHR}, // ASTC_2D_5X4_SRGB
+ {GL_COMPRESSED_RGBA_ASTC_5x5_KHR}, // ASTC_2D_5X5_UNORM
+ {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_5x5_KHR}, // ASTC_2D_5X5_SRGB
+ {GL_COMPRESSED_RGBA_ASTC_10x8_KHR}, // ASTC_2D_10X8_UNORM
+ {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_10x8_KHR}, // ASTC_2D_10X8_SRGB
+ {GL_COMPRESSED_RGBA_ASTC_6x6_KHR}, // ASTC_2D_6X6_UNORM
+ {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_6x6_KHR}, // ASTC_2D_6X6_SRGB
+ {GL_COMPRESSED_RGBA_ASTC_10x10_KHR}, // ASTC_2D_10X10_UNORM
+ {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_10x10_KHR}, // ASTC_2D_10X10_SRGB
+ {GL_COMPRESSED_RGBA_ASTC_12x12_KHR}, // ASTC_2D_12X12_UNORM
+ {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_12x12_KHR}, // ASTC_2D_12X12_SRGB
+ {GL_COMPRESSED_RGBA_ASTC_8x6_KHR}, // ASTC_2D_8X6_UNORM
+ {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_8x6_KHR}, // ASTC_2D_8X6_SRGB
+ {GL_COMPRESSED_RGBA_ASTC_6x5_KHR}, // ASTC_2D_6X5_UNORM
+ {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_6x5_KHR}, // ASTC_2D_6X5_SRGB
+ {GL_RGB9_E5, GL_RGB, GL_UNSIGNED_INT_5_9_9_9_REV}, // E5B9G9R9_FLOAT
+ {GL_DEPTH_COMPONENT32F, GL_DEPTH_COMPONENT, GL_FLOAT}, // D32_FLOAT
+ {GL_DEPTH_COMPONENT16, GL_DEPTH_COMPONENT, GL_UNSIGNED_SHORT}, // D16_UNORM
+ {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8}, // D24_UNORM_S8_UINT
+ {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8}, // S8_UINT_D24_UNORM
{GL_DEPTH32F_STENCIL8, GL_DEPTH_STENCIL,
GL_FLOAT_32_UNSIGNED_INT_24_8_REV}, // D32_FLOAT_S8_UINT
}};
+constexpr std::array ACCELERATED_FORMATS{
+ GL_RGBA32F, GL_RGBA16F, GL_RG32F, GL_RG16F, GL_R11F_G11F_B10F, GL_R32F,
+ GL_R16F, GL_RGBA32UI, GL_RGBA16UI, GL_RGB10_A2UI, GL_RGBA8UI, GL_RG32UI,
+ GL_RG16UI, GL_RG8UI, GL_R32UI, GL_R16UI, GL_R8UI, GL_RGBA32I,
+ GL_RGBA16I, GL_RGBA8I, GL_RG32I, GL_RG16I, GL_RG8I, GL_R32I,
+ GL_R16I, GL_R8I, GL_RGBA16, GL_RGB10_A2, GL_RGBA8, GL_RG16,
+ GL_RG8, GL_R16, GL_R8, GL_RGBA16_SNORM, GL_RGBA8_SNORM, GL_RG16_SNORM,
+ GL_RG8_SNORM, GL_R16_SNORM, GL_R8_SNORM,
+};
+
const FormatTuple& GetFormatTuple(PixelFormat pixel_format) {
- ASSERT(static_cast<std::size_t>(pixel_format) < tex_format_tuples.size());
- return tex_format_tuples[static_cast<std::size_t>(pixel_format)];
+ ASSERT(static_cast<size_t>(pixel_format) < FORMAT_TABLE.size());
+ return FORMAT_TABLE[static_cast<size_t>(pixel_format)];
}
-GLenum GetTextureTarget(const SurfaceTarget& target) {
- switch (target) {
- case SurfaceTarget::TextureBuffer:
+GLenum ImageTarget(const VideoCommon::ImageInfo& info) {
+ switch (info.type) {
+ case ImageType::e1D:
+ return GL_TEXTURE_1D_ARRAY;
+ case ImageType::e2D:
+ if (info.num_samples > 1) {
+ return GL_TEXTURE_2D_MULTISAMPLE_ARRAY;
+ }
+ return GL_TEXTURE_2D_ARRAY;
+ case ImageType::e3D:
+ return GL_TEXTURE_3D;
+ case ImageType::Linear:
+ return GL_TEXTURE_2D_ARRAY;
+ case ImageType::Buffer:
return GL_TEXTURE_BUFFER;
- case SurfaceTarget::Texture1D:
+ }
+ UNREACHABLE_MSG("Invalid image type={}", info.type);
+ return GL_NONE;
+}
+
+GLenum ImageTarget(ImageViewType type, int num_samples = 1) {
+ const bool is_multisampled = num_samples > 1;
+ switch (type) {
+ case ImageViewType::e1D:
return GL_TEXTURE_1D;
- case SurfaceTarget::Texture2D:
- return GL_TEXTURE_2D;
- case SurfaceTarget::Texture3D:
+ case ImageViewType::e2D:
+ return is_multisampled ? GL_TEXTURE_2D_MULTISAMPLE : GL_TEXTURE_2D;
+ case ImageViewType::Cube:
+ return GL_TEXTURE_CUBE_MAP;
+ case ImageViewType::e3D:
return GL_TEXTURE_3D;
- case SurfaceTarget::Texture1DArray:
+ case ImageViewType::e1DArray:
return GL_TEXTURE_1D_ARRAY;
- case SurfaceTarget::Texture2DArray:
- return GL_TEXTURE_2D_ARRAY;
- case SurfaceTarget::TextureCubemap:
- return GL_TEXTURE_CUBE_MAP;
- case SurfaceTarget::TextureCubeArray:
+ case ImageViewType::e2DArray:
+ return is_multisampled ? GL_TEXTURE_2D_MULTISAMPLE_ARRAY : GL_TEXTURE_2D_ARRAY;
+ case ImageViewType::CubeArray:
return GL_TEXTURE_CUBE_MAP_ARRAY;
+ case ImageViewType::Rect:
+ return GL_TEXTURE_RECTANGLE;
+ case ImageViewType::Buffer:
+ return GL_TEXTURE_BUFFER;
}
- UNREACHABLE();
- return {};
+ UNREACHABLE_MSG("Invalid image view type={}", type);
+ return GL_NONE;
}
-GLint GetSwizzleSource(SwizzleSource source) {
+GLenum TextureMode(PixelFormat format, bool is_first) {
+ switch (format) {
+ case PixelFormat::D24_UNORM_S8_UINT:
+ case PixelFormat::D32_FLOAT_S8_UINT:
+ return is_first ? GL_DEPTH_COMPONENT : GL_STENCIL_INDEX;
+ case PixelFormat::S8_UINT_D24_UNORM:
+ return is_first ? GL_STENCIL_INDEX : GL_DEPTH_COMPONENT;
+ default:
+ UNREACHABLE();
+ return GL_DEPTH_COMPONENT;
+ }
+}
+
+GLint Swizzle(SwizzleSource source) {
switch (source) {
case SwizzleSource::Zero:
return GL_ZERO;
@@ -184,531 +249,813 @@ GLint GetSwizzleSource(SwizzleSource source) {
case SwizzleSource::OneFloat:
return GL_ONE;
}
- UNREACHABLE();
+ UNREACHABLE_MSG("Invalid swizzle source={}", source);
return GL_NONE;
}
-GLenum GetComponent(PixelFormat format, bool is_first) {
- switch (format) {
- case PixelFormat::D24_UNORM_S8_UINT:
- case PixelFormat::D32_FLOAT_S8_UINT:
- return is_first ? GL_DEPTH_COMPONENT : GL_STENCIL_INDEX;
- case PixelFormat::S8_UINT_D24_UNORM:
- return is_first ? GL_STENCIL_INDEX : GL_DEPTH_COMPONENT;
+GLenum AttachmentType(PixelFormat format) {
+ switch (const SurfaceType type = VideoCore::Surface::GetFormatType(format); type) {
+ case SurfaceType::Depth:
+ return GL_DEPTH_ATTACHMENT;
+ case SurfaceType::DepthStencil:
+ return GL_DEPTH_STENCIL_ATTACHMENT;
default:
- UNREACHABLE();
- return GL_DEPTH_COMPONENT;
+ UNIMPLEMENTED_MSG("Unimplemented type={}", type);
+ return GL_NONE;
}
}
-void ApplyTextureDefaults(const SurfaceParams& params, GLuint texture) {
- if (params.IsBuffer()) {
- return;
+[[nodiscard]] bool IsConverted(const Device& device, PixelFormat format, ImageType type) {
+ if (!device.HasASTC() && IsPixelFormatASTC(format)) {
+ return true;
}
- glTextureParameteri(texture, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
- glTextureParameteri(texture, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
- glTextureParameteri(texture, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
- glTextureParameteri(texture, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
- glTextureParameteri(texture, GL_TEXTURE_MAX_LEVEL, static_cast<GLint>(params.num_levels - 1));
- if (params.num_levels == 1) {
- glTextureParameterf(texture, GL_TEXTURE_LOD_BIAS, 1000.0f);
+ switch (format) {
+ case PixelFormat::BC4_UNORM:
+ case PixelFormat::BC5_UNORM:
+ return type == ImageType::e3D;
+ default:
+ break;
}
+ return false;
}
-OGLTexture CreateTexture(const SurfaceParams& params, GLenum target, GLenum internal_format,
- OGLBuffer& texture_buffer) {
- OGLTexture texture;
- texture.Create(target);
+[[nodiscard]] constexpr SwizzleSource ConvertGreenRed(SwizzleSource value) {
+ switch (value) {
+ case SwizzleSource::G:
+ return SwizzleSource::R;
+ default:
+ return value;
+ }
+}
- switch (params.target) {
- case SurfaceTarget::Texture1D:
- glTextureStorage1D(texture.handle, params.emulated_levels, internal_format, params.width);
- break;
- case SurfaceTarget::TextureBuffer:
- texture_buffer.Create();
- glNamedBufferStorage(texture_buffer.handle, params.width * params.GetBytesPerPixel(),
- nullptr, GL_DYNAMIC_STORAGE_BIT);
- glTextureBuffer(texture.handle, internal_format, texture_buffer.handle);
+void ApplySwizzle(GLuint handle, PixelFormat format, std::array<SwizzleSource, 4> swizzle) {
+ switch (format) {
+ case PixelFormat::D24_UNORM_S8_UINT:
+ case PixelFormat::D32_FLOAT_S8_UINT:
+ case PixelFormat::S8_UINT_D24_UNORM:
+ UNIMPLEMENTED_IF(swizzle[0] != SwizzleSource::R && swizzle[0] != SwizzleSource::G);
+ glTextureParameteri(handle, GL_DEPTH_STENCIL_TEXTURE_MODE,
+ TextureMode(format, swizzle[0] == SwizzleSource::R));
+ std::ranges::transform(swizzle, swizzle.begin(), ConvertGreenRed);
break;
- case SurfaceTarget::Texture2D:
- case SurfaceTarget::TextureCubemap:
- glTextureStorage2D(texture.handle, params.emulated_levels, internal_format, params.width,
- params.height);
+ default:
break;
- case SurfaceTarget::Texture3D:
- case SurfaceTarget::Texture2DArray:
- case SurfaceTarget::TextureCubeArray:
- glTextureStorage3D(texture.handle, params.emulated_levels, internal_format, params.width,
- params.height, params.depth);
+ }
+ std::array<GLint, 4> gl_swizzle;
+ std::ranges::transform(swizzle, gl_swizzle.begin(), Swizzle);
+ glTextureParameteriv(handle, GL_TEXTURE_SWIZZLE_RGBA, gl_swizzle.data());
+}
+
+[[nodiscard]] bool CanBeAccelerated(const TextureCacheRuntime& runtime,
+ const VideoCommon::ImageInfo& info) {
+ // Disable accelerated uploads for now as they don't implement swizzled uploads
+ return false;
+ switch (info.type) {
+ case ImageType::e2D:
+ case ImageType::e3D:
+ case ImageType::Linear:
break;
default:
- UNREACHABLE();
+ return false;
+ }
+ const GLenum internal_format = GetFormatTuple(info.format).internal_format;
+ const auto& format_info = runtime.FormatInfo(info.type, internal_format);
+ if (format_info.is_compressed) {
+ return false;
+ }
+ if (std::ranges::find(ACCELERATED_FORMATS, internal_format) == ACCELERATED_FORMATS.end()) {
+ return false;
}
+ if (format_info.compatibility_by_size) {
+ return true;
+ }
+ const GLenum store_format = StoreFormat(BytesPerBlock(info.format));
+ const GLenum store_class = runtime.FormatInfo(info.type, store_format).compatibility_class;
+ return format_info.compatibility_class == store_class;
+}
- ApplyTextureDefaults(params, texture.handle);
+[[nodiscard]] CopyOrigin MakeCopyOrigin(VideoCommon::Offset3D offset,
+ VideoCommon::SubresourceLayers subresource, GLenum target) {
+ switch (target) {
+ case GL_TEXTURE_2D_ARRAY:
+ case GL_TEXTURE_2D_MULTISAMPLE_ARRAY:
+ return CopyOrigin{
+ .level = static_cast<GLint>(subresource.base_level),
+ .x = static_cast<GLint>(offset.x),
+ .y = static_cast<GLint>(offset.y),
+ .z = static_cast<GLint>(subresource.base_layer),
+ };
+ case GL_TEXTURE_3D:
+ return CopyOrigin{
+ .level = static_cast<GLint>(subresource.base_level),
+ .x = static_cast<GLint>(offset.x),
+ .y = static_cast<GLint>(offset.y),
+ .z = static_cast<GLint>(offset.z),
+ };
+ default:
+ UNIMPLEMENTED_MSG("Unimplemented copy target={}", target);
+ return CopyOrigin{.level = 0, .x = 0, .y = 0, .z = 0};
+ }
+}
- return texture;
+[[nodiscard]] CopyRegion MakeCopyRegion(VideoCommon::Extent3D extent,
+ VideoCommon::SubresourceLayers dst_subresource,
+ GLenum target) {
+ switch (target) {
+ case GL_TEXTURE_2D_ARRAY:
+ case GL_TEXTURE_2D_MULTISAMPLE_ARRAY:
+ return CopyRegion{
+ .width = static_cast<GLsizei>(extent.width),
+ .height = static_cast<GLsizei>(extent.height),
+ .depth = static_cast<GLsizei>(dst_subresource.num_layers),
+ };
+ case GL_TEXTURE_3D:
+ return CopyRegion{
+ .width = static_cast<GLsizei>(extent.width),
+ .height = static_cast<GLsizei>(extent.height),
+ .depth = static_cast<GLsizei>(extent.depth),
+ };
+ default:
+ UNIMPLEMENTED_MSG("Unimplemented copy target={}", target);
+ return CopyRegion{.width = 0, .height = 0, .depth = 0};
+ }
}
-constexpr u32 EncodeSwizzle(SwizzleSource x_source, SwizzleSource y_source, SwizzleSource z_source,
- SwizzleSource w_source) {
- return (static_cast<u32>(x_source) << 24) | (static_cast<u32>(y_source) << 16) |
- (static_cast<u32>(z_source) << 8) | static_cast<u32>(w_source);
+void AttachTexture(GLuint fbo, GLenum attachment, const ImageView* image_view) {
+ if (False(image_view->flags & VideoCommon::ImageViewFlagBits::Slice)) {
+ const GLuint texture = image_view->DefaultHandle();
+ glNamedFramebufferTexture(fbo, attachment, texture, 0);
+ return;
+ }
+ const GLuint texture = image_view->Handle(ImageViewType::e3D);
+ if (image_view->range.extent.layers > 1) {
+ // TODO: OpenGL doesn't support rendering to a fixed number of slices
+ glNamedFramebufferTexture(fbo, attachment, texture, 0);
+ } else {
+ const u32 slice = image_view->range.base.layer;
+ glNamedFramebufferTextureLayer(fbo, attachment, texture, 0, slice);
+ }
}
} // Anonymous namespace
-CachedSurface::CachedSurface(const GPUVAddr gpu_addr, const SurfaceParams& params,
- bool is_astc_supported)
- : VideoCommon::SurfaceBase<View>(gpu_addr, params, is_astc_supported) {
- if (is_converted) {
- internal_format = params.srgb_conversion ? GL_SRGB8_ALPHA8 : GL_RGBA8;
- format = GL_RGBA;
- type = GL_UNSIGNED_BYTE;
- } else {
- const auto& tuple{GetFormatTuple(params.pixel_format)};
- internal_format = tuple.internal_format;
- format = tuple.format;
- type = tuple.type;
- is_compressed = params.IsCompressed();
- }
- target = GetTextureTarget(params.target);
- texture = CreateTexture(params, target, internal_format, texture_buffer);
- DecorateSurfaceName();
+ImageBufferMap::ImageBufferMap(GLuint handle_, u8* map, size_t size, OGLSync* sync_)
+ : span(map, size), sync{sync_}, handle{handle_} {}
- u32 num_layers = 1;
- if (params.is_layered || params.target == SurfaceTarget::Texture3D) {
- num_layers = params.depth;
+ImageBufferMap::~ImageBufferMap() {
+ if (sync) {
+ sync->Create();
}
-
- main_view =
- CreateViewInner(ViewParams(params.target, 0, num_layers, 0, params.num_levels), true);
}
-CachedSurface::~CachedSurface() = default;
+TextureCacheRuntime::TextureCacheRuntime(const Device& device_, ProgramManager& program_manager,
+ StateTracker& state_tracker_)
+ : device{device_}, state_tracker{state_tracker_}, util_shaders(program_manager) {
+ static constexpr std::array TARGETS{GL_TEXTURE_1D_ARRAY, GL_TEXTURE_2D_ARRAY, GL_TEXTURE_3D};
+ for (size_t i = 0; i < TARGETS.size(); ++i) {
+ const GLenum target = TARGETS[i];
+ for (const FormatTuple& tuple : FORMAT_TABLE) {
+ const GLenum format = tuple.internal_format;
+ GLint compat_class;
+ GLint compat_type;
+ GLint is_compressed;
+ glGetInternalformativ(target, format, GL_IMAGE_COMPATIBILITY_CLASS, 1, &compat_class);
+ glGetInternalformativ(target, format, GL_IMAGE_FORMAT_COMPATIBILITY_TYPE, 1,
+ &compat_type);
+ glGetInternalformativ(target, format, GL_TEXTURE_COMPRESSED, 1, &is_compressed);
+ const FormatProperties properties{
+ .compatibility_class = static_cast<GLenum>(compat_class),
+ .compatibility_by_size = compat_type == GL_IMAGE_FORMAT_COMPATIBILITY_BY_SIZE,
+ .is_compressed = is_compressed == GL_TRUE,
+ };
+ format_properties[i].emplace(format, properties);
+ }
+ }
+ null_image_1d_array.Create(GL_TEXTURE_1D_ARRAY);
+ null_image_cube_array.Create(GL_TEXTURE_CUBE_MAP_ARRAY);
+ null_image_3d.Create(GL_TEXTURE_3D);
+ null_image_rect.Create(GL_TEXTURE_RECTANGLE);
+ glTextureStorage2D(null_image_1d_array.handle, 1, GL_R8, 1, 1);
+ glTextureStorage3D(null_image_cube_array.handle, 1, GL_R8, 1, 1, 6);
+ glTextureStorage3D(null_image_3d.handle, 1, GL_R8, 1, 1, 1);
+ glTextureStorage2D(null_image_rect.handle, 1, GL_R8, 1, 1);
+
+ std::array<GLuint, 4> new_handles;
+ glGenTextures(static_cast<GLsizei>(new_handles.size()), new_handles.data());
+ null_image_view_1d.handle = new_handles[0];
+ null_image_view_2d.handle = new_handles[1];
+ null_image_view_2d_array.handle = new_handles[2];
+ null_image_view_cube.handle = new_handles[3];
+ glTextureView(null_image_view_1d.handle, GL_TEXTURE_1D, null_image_1d_array.handle, GL_R8, 0, 1,
+ 0, 1);
+ glTextureView(null_image_view_2d.handle, GL_TEXTURE_2D, null_image_cube_array.handle, GL_R8, 0,
+ 1, 0, 1);
+ glTextureView(null_image_view_2d_array.handle, GL_TEXTURE_2D_ARRAY,
+ null_image_cube_array.handle, GL_R8, 0, 1, 0, 1);
+ glTextureView(null_image_view_cube.handle, GL_TEXTURE_CUBE_MAP, null_image_cube_array.handle,
+ GL_R8, 0, 1, 0, 6);
+ const std::array texture_handles{
+ null_image_1d_array.handle, null_image_cube_array.handle, null_image_3d.handle,
+ null_image_rect.handle, null_image_view_1d.handle, null_image_view_2d.handle,
+ null_image_view_2d_array.handle, null_image_view_cube.handle,
+ };
+ for (const GLuint handle : texture_handles) {
+ static constexpr std::array NULL_SWIZZLE{GL_ZERO, GL_ZERO, GL_ZERO, GL_ZERO};
+ glTextureParameteriv(handle, GL_TEXTURE_SWIZZLE_RGBA, NULL_SWIZZLE.data());
+ }
+ const auto set_view = [this](ImageViewType type, GLuint handle) {
+ if (device.HasDebuggingToolAttached()) {
+ const std::string name = fmt::format("NullImage {}", type);
+ glObjectLabel(GL_TEXTURE, handle, static_cast<GLsizei>(name.size()), name.data());
+ }
+ null_image_views[static_cast<size_t>(type)] = handle;
+ };
+ set_view(ImageViewType::e1D, null_image_view_1d.handle);
+ set_view(ImageViewType::e2D, null_image_view_2d.handle);
+ set_view(ImageViewType::Cube, null_image_view_cube.handle);
+ set_view(ImageViewType::e3D, null_image_3d.handle);
+ set_view(ImageViewType::e1DArray, null_image_1d_array.handle);
+ set_view(ImageViewType::e2DArray, null_image_view_2d_array.handle);
+ set_view(ImageViewType::CubeArray, null_image_cube_array.handle);
+ set_view(ImageViewType::Rect, null_image_rect.handle);
+}
-void CachedSurface::DownloadTexture(std::vector<u8>& staging_buffer) {
- MICROPROFILE_SCOPE(OpenGL_Texture_Download);
+TextureCacheRuntime::~TextureCacheRuntime() = default;
- if (params.IsBuffer()) {
- glGetNamedBufferSubData(texture_buffer.handle, 0,
- static_cast<GLsizeiptr>(params.GetHostSizeInBytes(false)),
- staging_buffer.data());
- return;
- }
+void TextureCacheRuntime::Finish() {
+ glFinish();
+}
- SCOPE_EXIT({ glPixelStorei(GL_PACK_ROW_LENGTH, 0); });
+ImageBufferMap TextureCacheRuntime::MapUploadBuffer(size_t size) {
+ return upload_buffers.RequestMap(size, true);
+}
- for (u32 level = 0; level < params.emulated_levels; ++level) {
- glPixelStorei(GL_PACK_ALIGNMENT, std::min(8U, params.GetRowAlignment(level, is_converted)));
- glPixelStorei(GL_PACK_ROW_LENGTH, static_cast<GLint>(params.GetMipWidth(level)));
- const std::size_t mip_offset = params.GetHostMipmapLevelOffset(level, is_converted);
+ImageBufferMap TextureCacheRuntime::MapDownloadBuffer(size_t size) {
+ return download_buffers.RequestMap(size, false);
+}
- u8* const mip_data = staging_buffer.data() + mip_offset;
- const GLsizei size = static_cast<GLsizei>(params.GetHostMipmapSize(level));
- if (is_compressed) {
- glGetCompressedTextureImage(texture.handle, level, size, mip_data);
- } else {
- glGetTextureImage(texture.handle, level, format, type, size, mip_data);
- }
+void TextureCacheRuntime::CopyImage(Image& dst_image, Image& src_image,
+ std::span<const ImageCopy> copies) {
+ const GLuint dst_name = dst_image.Handle();
+ const GLuint src_name = src_image.Handle();
+ const GLenum dst_target = ImageTarget(dst_image.info);
+ const GLenum src_target = ImageTarget(src_image.info);
+ for (const ImageCopy& copy : copies) {
+ const auto src_origin = MakeCopyOrigin(copy.src_offset, copy.src_subresource, src_target);
+ const auto dst_origin = MakeCopyOrigin(copy.dst_offset, copy.dst_subresource, dst_target);
+ const auto region = MakeCopyRegion(copy.extent, copy.dst_subresource, dst_target);
+ glCopyImageSubData(src_name, src_target, src_origin.level, src_origin.x, src_origin.y,
+ src_origin.z, dst_name, dst_target, dst_origin.level, dst_origin.x,
+ dst_origin.y, dst_origin.z, region.width, region.height, region.depth);
}
}
-void CachedSurface::UploadTexture(const std::vector<u8>& staging_buffer) {
- MICROPROFILE_SCOPE(OpenGL_Texture_Upload);
- SCOPE_EXIT({ glPixelStorei(GL_UNPACK_ROW_LENGTH, 0); });
- for (u32 level = 0; level < params.emulated_levels; ++level) {
- UploadTextureMipmap(level, staging_buffer);
+bool TextureCacheRuntime::CanImageBeCopied(const Image& dst, const Image& src) {
+ if (dst.info.type == ImageType::e3D && dst.info.format == PixelFormat::BC4_UNORM) {
+ return false;
}
+ return true;
}
-void CachedSurface::UploadTextureMipmap(u32 level, const std::vector<u8>& staging_buffer) {
- glPixelStorei(GL_UNPACK_ALIGNMENT, std::min(8U, params.GetRowAlignment(level, is_converted)));
- glPixelStorei(GL_UNPACK_ROW_LENGTH, static_cast<GLint>(params.GetMipWidth(level)));
-
- const std::size_t mip_offset = params.GetHostMipmapLevelOffset(level, is_converted);
- const u8* buffer{staging_buffer.data() + mip_offset};
- if (is_compressed) {
- const auto image_size{static_cast<GLsizei>(params.GetHostMipmapSize(level))};
- switch (params.target) {
- case SurfaceTarget::Texture2D:
- glCompressedTextureSubImage2D(texture.handle, level, 0, 0,
- static_cast<GLsizei>(params.GetMipWidth(level)),
- static_cast<GLsizei>(params.GetMipHeight(level)),
- internal_format, image_size, buffer);
- break;
- case SurfaceTarget::Texture3D:
- case SurfaceTarget::Texture2DArray:
- case SurfaceTarget::TextureCubeArray:
- glCompressedTextureSubImage3D(texture.handle, level, 0, 0, 0,
- static_cast<GLsizei>(params.GetMipWidth(level)),
- static_cast<GLsizei>(params.GetMipHeight(level)),
- static_cast<GLsizei>(params.GetMipDepth(level)),
- internal_format, image_size, buffer);
- break;
- case SurfaceTarget::TextureCubemap: {
- const std::size_t layer_size{params.GetHostLayerSize(level)};
- for (std::size_t face = 0; face < params.depth; ++face) {
- glCompressedTextureSubImage3D(texture.handle, level, 0, 0, static_cast<GLint>(face),
- static_cast<GLsizei>(params.GetMipWidth(level)),
- static_cast<GLsizei>(params.GetMipHeight(level)), 1,
- internal_format, static_cast<GLsizei>(layer_size),
- buffer);
- buffer += layer_size;
- }
- break;
- }
- default:
- UNREACHABLE();
- }
+void TextureCacheRuntime::EmulateCopyImage(Image& dst, Image& src,
+ std::span<const ImageCopy> copies) {
+ if (dst.info.type == ImageType::e3D && dst.info.format == PixelFormat::BC4_UNORM) {
+ ASSERT(src.info.type == ImageType::e3D);
+ util_shaders.CopyBC4(dst, src, copies);
} else {
- switch (params.target) {
- case SurfaceTarget::Texture1D:
- glTextureSubImage1D(texture.handle, level, 0, params.GetMipWidth(level), format, type,
- buffer);
- break;
- case SurfaceTarget::TextureBuffer:
- ASSERT(level == 0);
- glNamedBufferSubData(texture_buffer.handle, 0,
- params.GetMipWidth(level) * params.GetBytesPerPixel(), buffer);
- break;
- case SurfaceTarget::Texture1DArray:
- case SurfaceTarget::Texture2D:
- glTextureSubImage2D(texture.handle, level, 0, 0, params.GetMipWidth(level),
- params.GetMipHeight(level), format, type, buffer);
- break;
- case SurfaceTarget::Texture3D:
- case SurfaceTarget::Texture2DArray:
- case SurfaceTarget::TextureCubeArray:
- glTextureSubImage3D(
- texture.handle, level, 0, 0, 0, static_cast<GLsizei>(params.GetMipWidth(level)),
- static_cast<GLsizei>(params.GetMipHeight(level)),
- static_cast<GLsizei>(params.GetMipDepth(level)), format, type, buffer);
- break;
- case SurfaceTarget::TextureCubemap:
- for (std::size_t face = 0; face < params.depth; ++face) {
- glTextureSubImage3D(texture.handle, level, 0, 0, static_cast<GLint>(face),
- params.GetMipWidth(level), params.GetMipHeight(level), 1,
- format, type, buffer);
- buffer += params.GetHostLayerSize(level);
- }
- break;
- default:
- UNREACHABLE();
- }
+ UNREACHABLE();
}
}
-void CachedSurface::DecorateSurfaceName() {
- LabelGLObject(GL_TEXTURE, texture.handle, GetGpuAddr(), params.TargetName());
-}
+void TextureCacheRuntime::BlitFramebuffer(Framebuffer* dst, Framebuffer* src,
+ const std::array<Offset2D, 2>& dst_region,
+ const std::array<Offset2D, 2>& src_region,
+ Tegra::Engines::Fermi2D::Filter filter,
+ Tegra::Engines::Fermi2D::Operation operation) {
+ state_tracker.NotifyScissor0();
+ state_tracker.NotifyRasterizeEnable();
+ state_tracker.NotifyFramebufferSRGB();
-void CachedSurfaceView::DecorateViewName(GPUVAddr gpu_addr, const std::string& prefix) {
- LabelGLObject(GL_TEXTURE, main_view.handle, gpu_addr, prefix);
+ ASSERT(dst->BufferBits() == src->BufferBits());
+
+ glEnable(GL_FRAMEBUFFER_SRGB);
+ glDisable(GL_RASTERIZER_DISCARD);
+ glDisablei(GL_SCISSOR_TEST, 0);
+
+ const GLbitfield buffer_bits = dst->BufferBits();
+ const bool has_depth = (buffer_bits & ~GL_COLOR_BUFFER_BIT) != 0;
+ const bool is_linear = !has_depth && filter == Tegra::Engines::Fermi2D::Filter::Bilinear;
+ glBlitNamedFramebuffer(src->Handle(), dst->Handle(), src_region[0].x, src_region[0].y,
+ src_region[1].x, src_region[1].y, dst_region[0].x, dst_region[0].y,
+ dst_region[1].x, dst_region[1].y, buffer_bits,
+ is_linear ? GL_LINEAR : GL_NEAREST);
}
-View CachedSurface::CreateView(const ViewParams& view_key) {
- return CreateViewInner(view_key, false);
+void TextureCacheRuntime::AccelerateImageUpload(Image& image, const ImageBufferMap& map,
+ size_t buffer_offset,
+ std::span<const SwizzleParameters> swizzles) {
+ switch (image.info.type) {
+ case ImageType::e2D:
+ return util_shaders.BlockLinearUpload2D(image, map, buffer_offset, swizzles);
+ case ImageType::e3D:
+ return util_shaders.BlockLinearUpload3D(image, map, buffer_offset, swizzles);
+ case ImageType::Linear:
+ return util_shaders.PitchUpload(image, map, buffer_offset, swizzles);
+ default:
+ UNREACHABLE();
+ break;
+ }
}
-View CachedSurface::CreateViewInner(const ViewParams& view_key, const bool is_proxy) {
- auto view = std::make_shared<CachedSurfaceView>(*this, view_key, is_proxy);
- views[view_key] = view;
- if (!is_proxy)
- view->DecorateViewName(gpu_addr, params.TargetName() + "V:" + std::to_string(view_count++));
- return view;
+void TextureCacheRuntime::InsertUploadMemoryBarrier() {
+ glMemoryBarrier(GL_TEXTURE_FETCH_BARRIER_BIT | GL_SHADER_IMAGE_ACCESS_BARRIER_BIT);
}
-CachedSurfaceView::CachedSurfaceView(CachedSurface& surface, const ViewParams& params,
- bool is_proxy)
- : VideoCommon::ViewBase(params), surface{surface}, format{surface.internal_format},
- target{GetTextureTarget(params.target)}, is_proxy{is_proxy} {
- if (!is_proxy) {
- main_view = CreateTextureView();
+FormatProperties TextureCacheRuntime::FormatInfo(ImageType type, GLenum internal_format) const {
+ switch (type) {
+ case ImageType::e1D:
+ return format_properties[0].at(internal_format);
+ case ImageType::e2D:
+ case ImageType::Linear:
+ return format_properties[1].at(internal_format);
+ case ImageType::e3D:
+ return format_properties[2].at(internal_format);
+ default:
+ UNREACHABLE();
+ return FormatProperties{};
}
}
-CachedSurfaceView::~CachedSurfaceView() = default;
+TextureCacheRuntime::StagingBuffers::StagingBuffers(GLenum storage_flags_, GLenum map_flags_)
+ : storage_flags{storage_flags_}, map_flags{map_flags_} {}
-void CachedSurfaceView::Attach(GLenum attachment, GLenum fb_target) const {
- ASSERT(params.num_levels == 1);
+TextureCacheRuntime::StagingBuffers::~StagingBuffers() = default;
- if (params.target == SurfaceTarget::Texture3D) {
- if (params.num_layers > 1) {
- ASSERT(params.base_layer == 0);
- glFramebufferTexture(fb_target, attachment, surface.texture.handle, params.base_level);
- } else {
- glFramebufferTexture3D(fb_target, attachment, target, surface.texture.handle,
- params.base_level, params.base_layer);
- }
- return;
+ImageBufferMap TextureCacheRuntime::StagingBuffers::RequestMap(size_t requested_size,
+ bool insert_fence) {
+ const size_t index = RequestBuffer(requested_size);
+ OGLSync* const sync = insert_fence ? &syncs[index] : nullptr;
+ return ImageBufferMap(buffers[index].handle, maps[index], requested_size, sync);
+}
+
+size_t TextureCacheRuntime::StagingBuffers::RequestBuffer(size_t requested_size) {
+ if (const std::optional<size_t> index = FindBuffer(requested_size); index) {
+ return *index;
}
- if (params.num_layers > 1) {
- UNIMPLEMENTED_IF(params.base_layer != 0);
- glFramebufferTexture(fb_target, attachment, GetTexture(), 0);
- return;
+ OGLBuffer& buffer = buffers.emplace_back();
+ buffer.Create();
+ glNamedBufferStorage(buffer.handle, requested_size, nullptr,
+ storage_flags | GL_MAP_PERSISTENT_BIT);
+ maps.push_back(static_cast<u8*>(glMapNamedBufferRange(buffer.handle, 0, requested_size,
+ map_flags | GL_MAP_PERSISTENT_BIT)));
+
+ syncs.emplace_back();
+ sizes.push_back(requested_size);
+
+ ASSERT(syncs.size() == buffers.size() && buffers.size() == maps.size() &&
+ maps.size() == sizes.size());
+
+ return buffers.size() - 1;
+}
+
+std::optional<size_t> TextureCacheRuntime::StagingBuffers::FindBuffer(size_t requested_size) {
+ size_t smallest_buffer = std::numeric_limits<size_t>::max();
+ std::optional<size_t> found;
+ const size_t num_buffers = sizes.size();
+ for (size_t index = 0; index < num_buffers; ++index) {
+ const size_t buffer_size = sizes[index];
+ if (buffer_size < requested_size || buffer_size >= smallest_buffer) {
+ continue;
+ }
+ if (syncs[index].handle != 0) {
+ GLint status;
+ glGetSynciv(syncs[index].handle, GL_SYNC_STATUS, 1, nullptr, &status);
+ if (status != GL_SIGNALED) {
+ continue;
+ }
+ syncs[index].Release();
+ }
+ smallest_buffer = buffer_size;
+ found = index;
}
+ return found;
+}
- const GLenum view_target = surface.GetTarget();
- const GLuint texture = surface.GetTexture();
- switch (surface.GetSurfaceParams().target) {
- case SurfaceTarget::Texture1D:
- glFramebufferTexture1D(fb_target, attachment, view_target, texture, params.base_level);
+Image::Image(TextureCacheRuntime& runtime, const VideoCommon::ImageInfo& info_, GPUVAddr gpu_addr_,
+ VAddr cpu_addr_)
+ : VideoCommon::ImageBase(info_, gpu_addr_, cpu_addr_) {
+ if (CanBeAccelerated(runtime, info)) {
+ flags |= ImageFlagBits::AcceleratedUpload;
+ }
+ if (IsConverted(runtime.device, info.format, info.type)) {
+ flags |= ImageFlagBits::Converted;
+ gl_internal_format = IsPixelFormatSRGB(info.format) ? GL_SRGB8_ALPHA8 : GL_RGBA8;
+ gl_store_format = GL_RGBA8;
+ gl_format = GL_RGBA;
+ gl_type = GL_UNSIGNED_INT_8_8_8_8_REV;
+ } else {
+ const auto& tuple = GetFormatTuple(info.format);
+ gl_internal_format = tuple.internal_format;
+ gl_store_format = tuple.store_format;
+ gl_format = tuple.format;
+ gl_type = tuple.type;
+ }
+ const GLenum target = ImageTarget(info);
+ const GLsizei width = info.size.width;
+ const GLsizei height = info.size.height;
+ const GLsizei depth = info.size.depth;
+ const int max_host_mip_levels = std::bit_width(info.size.width);
+ const GLsizei num_levels = std::min(info.resources.levels, max_host_mip_levels);
+ const GLsizei num_layers = info.resources.layers;
+ const GLsizei num_samples = info.num_samples;
+
+ GLuint handle = 0;
+ if (target != GL_TEXTURE_BUFFER) {
+ texture.Create(target);
+ handle = texture.handle;
+ }
+ switch (target) {
+ case GL_TEXTURE_1D_ARRAY:
+ glTextureStorage2D(handle, num_levels, gl_store_format, width, num_layers);
break;
- case SurfaceTarget::Texture2D:
- glFramebufferTexture2D(fb_target, attachment, view_target, texture, params.base_level);
+ case GL_TEXTURE_2D_ARRAY:
+ glTextureStorage3D(handle, num_levels, gl_store_format, width, height, num_layers);
break;
- case SurfaceTarget::Texture1DArray:
- case SurfaceTarget::Texture2DArray:
- case SurfaceTarget::TextureCubemap:
- case SurfaceTarget::TextureCubeArray:
- glFramebufferTextureLayer(fb_target, attachment, texture, params.base_level,
- params.base_layer);
+ case GL_TEXTURE_2D_MULTISAMPLE_ARRAY: {
+ // TODO: Where should 'fixedsamplelocations' come from?
+ const auto [samples_x, samples_y] = SamplesLog2(info.num_samples);
+ glTextureStorage3DMultisample(handle, num_samples, gl_store_format, width >> samples_x,
+ height >> samples_y, num_layers, GL_FALSE);
+ break;
+ }
+ case GL_TEXTURE_RECTANGLE:
+ glTextureStorage2D(handle, num_levels, gl_store_format, width, height);
+ break;
+ case GL_TEXTURE_3D:
+ glTextureStorage3D(handle, num_levels, gl_store_format, width, height, depth);
+ break;
+ case GL_TEXTURE_BUFFER:
+ buffer.Create();
+ glNamedBufferStorage(buffer.handle, guest_size_bytes, nullptr, 0);
break;
default:
- UNIMPLEMENTED();
+ UNREACHABLE_MSG("Invalid target=0x{:x}", target);
+ break;
+ }
+ if (runtime.device.HasDebuggingToolAttached()) {
+ const std::string name = VideoCommon::Name(*this);
+ glObjectLabel(target == GL_TEXTURE_BUFFER ? GL_BUFFER : GL_TEXTURE, handle,
+ static_cast<GLsizei>(name.size()), name.data());
}
}
-GLuint CachedSurfaceView::GetTexture(SwizzleSource x_source, SwizzleSource y_source,
- SwizzleSource z_source, SwizzleSource w_source) {
- if (GetSurfaceParams().IsBuffer()) {
- return GetTexture();
- }
- const u32 new_swizzle = EncodeSwizzle(x_source, y_source, z_source, w_source);
- if (current_swizzle == new_swizzle) {
- return current_view;
- }
- current_swizzle = new_swizzle;
+void Image::UploadMemory(const ImageBufferMap& map, size_t buffer_offset,
+ std::span<const VideoCommon::BufferImageCopy> copies) {
+ glBindBuffer(GL_PIXEL_UNPACK_BUFFER, map.Handle());
+ glFlushMappedBufferRange(GL_PIXEL_UNPACK_BUFFER, buffer_offset, unswizzled_size_bytes);
- const auto [entry, is_cache_miss] = view_cache.try_emplace(new_swizzle);
- OGLTextureView& view = entry->second;
- if (!is_cache_miss) {
- current_view = view.handle;
- return view.handle;
- }
- view = CreateTextureView();
- current_view = view.handle;
+ glPixelStorei(GL_UNPACK_ALIGNMENT, 1);
- std::array swizzle{x_source, y_source, z_source, w_source};
+ u32 current_row_length = std::numeric_limits<u32>::max();
+ u32 current_image_height = std::numeric_limits<u32>::max();
- switch (const PixelFormat format = GetSurfaceParams().pixel_format) {
- case PixelFormat::D24_UNORM_S8_UINT:
- case PixelFormat::D32_FLOAT_S8_UINT:
- case PixelFormat::S8_UINT_D24_UNORM:
- UNIMPLEMENTED_IF(x_source != SwizzleSource::R && x_source != SwizzleSource::G);
- glTextureParameteri(view.handle, GL_DEPTH_STENCIL_TEXTURE_MODE,
- GetComponent(format, x_source == SwizzleSource::R));
-
- // Make sure we sample the first component
- std::transform(swizzle.begin(), swizzle.end(), swizzle.begin(), [](SwizzleSource value) {
- return value == SwizzleSource::G ? SwizzleSource::R : value;
- });
- [[fallthrough]];
- default: {
- const std::array gl_swizzle = {GetSwizzleSource(swizzle[0]), GetSwizzleSource(swizzle[1]),
- GetSwizzleSource(swizzle[2]), GetSwizzleSource(swizzle[3])};
- glTextureParameteriv(view.handle, GL_TEXTURE_SWIZZLE_RGBA, gl_swizzle.data());
- break;
- }
+ for (const VideoCommon::BufferImageCopy& copy : copies) {
+ if (current_row_length != copy.buffer_row_length) {
+ current_row_length = copy.buffer_row_length;
+ glPixelStorei(GL_UNPACK_ROW_LENGTH, current_row_length);
+ }
+ if (current_image_height != copy.buffer_image_height) {
+ current_image_height = copy.buffer_image_height;
+ glPixelStorei(GL_UNPACK_IMAGE_HEIGHT, current_image_height);
+ }
+ CopyBufferToImage(copy, buffer_offset);
}
- return view.handle;
}
-OGLTextureView CachedSurfaceView::CreateTextureView() const {
- OGLTextureView texture_view;
- texture_view.Create();
-
- if (target == GL_TEXTURE_3D) {
- glTextureView(texture_view.handle, target, surface.texture.handle, format,
- params.base_level, params.num_levels, 0, 1);
- } else {
- glTextureView(texture_view.handle, target, surface.texture.handle, format,
- params.base_level, params.num_levels, params.base_layer, params.num_layers);
+void Image::UploadMemory(const ImageBufferMap& map, size_t buffer_offset,
+ std::span<const VideoCommon::BufferCopy> copies) {
+ for (const VideoCommon::BufferCopy& copy : copies) {
+ glCopyNamedBufferSubData(map.Handle(), buffer.handle, copy.src_offset + buffer_offset,
+ copy.dst_offset, copy.size);
}
- ApplyTextureDefaults(surface.GetSurfaceParams(), texture_view.handle);
-
- return texture_view;
}
-TextureCacheOpenGL::TextureCacheOpenGL(VideoCore::RasterizerInterface& rasterizer,
- Tegra::Engines::Maxwell3D& maxwell3d,
- Tegra::MemoryManager& gpu_memory, const Device& device,
- StateTracker& state_tracker_)
- : TextureCacheBase{rasterizer, maxwell3d, gpu_memory, device.HasASTC()}, state_tracker{
- state_tracker_} {
- src_framebuffer.Create();
- dst_framebuffer.Create();
-}
+void Image::DownloadMemory(ImageBufferMap& map, size_t buffer_offset,
+ std::span<const VideoCommon::BufferImageCopy> copies) {
+ glMemoryBarrier(GL_PIXEL_BUFFER_BARRIER_BIT); // TODO: Move this to its own API
-TextureCacheOpenGL::~TextureCacheOpenGL() = default;
+ glBindBuffer(GL_PIXEL_PACK_BUFFER, map.Handle());
+ glPixelStorei(GL_PACK_ALIGNMENT, 1);
-Surface TextureCacheOpenGL::CreateSurface(GPUVAddr gpu_addr, const SurfaceParams& params) {
- return std::make_shared<CachedSurface>(gpu_addr, params, is_astc_supported);
-}
+ u32 current_row_length = std::numeric_limits<u32>::max();
+ u32 current_image_height = std::numeric_limits<u32>::max();
-void TextureCacheOpenGL::ImageCopy(Surface& src_surface, Surface& dst_surface,
- const VideoCommon::CopyParams& copy_params) {
- const auto& src_params = src_surface->GetSurfaceParams();
- const auto& dst_params = dst_surface->GetSurfaceParams();
- if (src_params.type != dst_params.type) {
- // A fallback is needed
- return;
+ for (const VideoCommon::BufferImageCopy& copy : copies) {
+ if (current_row_length != copy.buffer_row_length) {
+ current_row_length = copy.buffer_row_length;
+ glPixelStorei(GL_PACK_ROW_LENGTH, current_row_length);
+ }
+ if (current_image_height != copy.buffer_image_height) {
+ current_image_height = copy.buffer_image_height;
+ glPixelStorei(GL_PACK_IMAGE_HEIGHT, current_image_height);
+ }
+ CopyImageToBuffer(copy, buffer_offset);
}
- const auto src_handle = src_surface->GetTexture();
- const auto src_target = src_surface->GetTarget();
- const auto dst_handle = dst_surface->GetTexture();
- const auto dst_target = dst_surface->GetTarget();
- glCopyImageSubData(src_handle, src_target, copy_params.source_level, copy_params.source_x,
- copy_params.source_y, copy_params.source_z, dst_handle, dst_target,
- copy_params.dest_level, copy_params.dest_x, copy_params.dest_y,
- copy_params.dest_z, copy_params.width, copy_params.height,
- copy_params.depth);
}
-void TextureCacheOpenGL::ImageBlit(View& src_view, View& dst_view,
- const Tegra::Engines::Fermi2D::Config& copy_config) {
- const auto& src_params{src_view->GetSurfaceParams()};
- const auto& dst_params{dst_view->GetSurfaceParams()};
- UNIMPLEMENTED_IF(src_params.depth != 1);
- UNIMPLEMENTED_IF(dst_params.depth != 1);
-
- state_tracker.NotifyScissor0();
- state_tracker.NotifyFramebuffer();
- state_tracker.NotifyRasterizeEnable();
- state_tracker.NotifyFramebufferSRGB();
+void Image::CopyBufferToImage(const VideoCommon::BufferImageCopy& copy, size_t buffer_offset) {
+ // Compressed formats don't have a pixel format or type
+ const bool is_compressed = gl_format == GL_NONE;
+ const void* const offset = reinterpret_cast<const void*>(copy.buffer_offset + buffer_offset);
- if (dst_params.srgb_conversion) {
- glEnable(GL_FRAMEBUFFER_SRGB);
- } else {
- glDisable(GL_FRAMEBUFFER_SRGB);
+ switch (info.type) {
+ case ImageType::e1D:
+ if (is_compressed) {
+ glCompressedTextureSubImage2D(texture.handle, copy.image_subresource.base_level,
+ copy.image_offset.x, copy.image_subresource.base_layer,
+ copy.image_extent.width,
+ copy.image_subresource.num_layers, gl_internal_format,
+ static_cast<GLsizei>(copy.buffer_size), offset);
+ } else {
+ glTextureSubImage2D(texture.handle, copy.image_subresource.base_level,
+ copy.image_offset.x, copy.image_subresource.base_layer,
+ copy.image_extent.width, copy.image_subresource.num_layers,
+ gl_format, gl_type, offset);
+ }
+ break;
+ case ImageType::e2D:
+ case ImageType::Linear:
+ if (is_compressed) {
+ glCompressedTextureSubImage3D(
+ texture.handle, copy.image_subresource.base_level, copy.image_offset.x,
+ copy.image_offset.y, copy.image_subresource.base_layer, copy.image_extent.width,
+ copy.image_extent.height, copy.image_subresource.num_layers, gl_internal_format,
+ static_cast<GLsizei>(copy.buffer_size), offset);
+ } else {
+ glTextureSubImage3D(texture.handle, copy.image_subresource.base_level,
+ copy.image_offset.x, copy.image_offset.y,
+ copy.image_subresource.base_layer, copy.image_extent.width,
+ copy.image_extent.height, copy.image_subresource.num_layers,
+ gl_format, gl_type, offset);
+ }
+ break;
+ case ImageType::e3D:
+ if (is_compressed) {
+ glCompressedTextureSubImage3D(
+ texture.handle, copy.image_subresource.base_level, copy.image_offset.x,
+ copy.image_offset.y, copy.image_offset.z, copy.image_extent.width,
+ copy.image_extent.height, copy.image_extent.depth, gl_internal_format,
+ static_cast<GLsizei>(copy.buffer_size), offset);
+ } else {
+ glTextureSubImage3D(texture.handle, copy.image_subresource.base_level,
+ copy.image_offset.x, copy.image_offset.y, copy.image_offset.z,
+ copy.image_extent.width, copy.image_extent.height,
+ copy.image_extent.depth, gl_format, gl_type, offset);
+ }
+ break;
+ default:
+ UNREACHABLE();
}
- glDisable(GL_RASTERIZER_DISCARD);
- glDisablei(GL_SCISSOR_TEST, 0);
-
- glBindFramebuffer(GL_READ_FRAMEBUFFER, src_framebuffer.handle);
- glBindFramebuffer(GL_DRAW_FRAMEBUFFER, dst_framebuffer.handle);
-
- GLenum buffers = 0;
- if (src_params.type == SurfaceType::ColorTexture) {
- src_view->Attach(GL_COLOR_ATTACHMENT0, GL_READ_FRAMEBUFFER);
- glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0,
- 0);
-
- dst_view->Attach(GL_COLOR_ATTACHMENT0, GL_DRAW_FRAMEBUFFER);
- glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0,
- 0);
-
- buffers = GL_COLOR_BUFFER_BIT;
- } else if (src_params.type == SurfaceType::Depth) {
- glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, 0, 0);
- src_view->Attach(GL_DEPTH_ATTACHMENT, GL_READ_FRAMEBUFFER);
- glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0, 0);
+}
- glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, 0, 0);
- dst_view->Attach(GL_DEPTH_ATTACHMENT, GL_DRAW_FRAMEBUFFER);
- glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0, 0);
+void Image::CopyImageToBuffer(const VideoCommon::BufferImageCopy& copy, size_t buffer_offset) {
+ const GLint x_offset = copy.image_offset.x;
+ const GLsizei width = copy.image_extent.width;
- buffers = GL_DEPTH_BUFFER_BIT;
- } else if (src_params.type == SurfaceType::DepthStencil) {
- glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, 0, 0);
- src_view->Attach(GL_DEPTH_STENCIL_ATTACHMENT, GL_READ_FRAMEBUFFER);
+ const GLint level = copy.image_subresource.base_level;
+ const GLsizei buffer_size = static_cast<GLsizei>(copy.buffer_size);
+ void* const offset = reinterpret_cast<void*>(copy.buffer_offset + buffer_offset);
- glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, 0, 0);
- dst_view->Attach(GL_DEPTH_STENCIL_ATTACHMENT, GL_DRAW_FRAMEBUFFER);
+ GLint y_offset = 0;
+ GLint z_offset = 0;
+ GLsizei height = 1;
+ GLsizei depth = 1;
- buffers = GL_DEPTH_BUFFER_BIT | GL_STENCIL_BUFFER_BIT;
+ switch (info.type) {
+ case ImageType::e1D:
+ y_offset = copy.image_subresource.base_layer;
+ height = copy.image_subresource.num_layers;
+ break;
+ case ImageType::e2D:
+ case ImageType::Linear:
+ y_offset = copy.image_offset.y;
+ z_offset = copy.image_subresource.base_layer;
+ height = copy.image_extent.height;
+ depth = copy.image_subresource.num_layers;
+ break;
+ case ImageType::e3D:
+ y_offset = copy.image_offset.y;
+ z_offset = copy.image_offset.z;
+ height = copy.image_extent.height;
+ depth = copy.image_extent.depth;
+ break;
+ default:
+ UNREACHABLE();
+ }
+ // Compressed formats don't have a pixel format or type
+ const bool is_compressed = gl_format == GL_NONE;
+ if (is_compressed) {
+ glGetCompressedTextureSubImage(texture.handle, level, x_offset, y_offset, z_offset, width,
+ height, depth, buffer_size, offset);
+ } else {
+ glGetTextureSubImage(texture.handle, level, x_offset, y_offset, z_offset, width, height,
+ depth, gl_format, gl_type, buffer_size, offset);
}
-
- const Common::Rectangle<u32>& src_rect = copy_config.src_rect;
- const Common::Rectangle<u32>& dst_rect = copy_config.dst_rect;
- const bool is_linear = copy_config.filter == Tegra::Engines::Fermi2D::Filter::Linear;
-
- glBlitFramebuffer(static_cast<GLint>(src_rect.left), static_cast<GLint>(src_rect.top),
- static_cast<GLint>(src_rect.right), static_cast<GLint>(src_rect.bottom),
- static_cast<GLint>(dst_rect.left), static_cast<GLint>(dst_rect.top),
- static_cast<GLint>(dst_rect.right), static_cast<GLint>(dst_rect.bottom),
- buffers,
- is_linear && (buffers == GL_COLOR_BUFFER_BIT) ? GL_LINEAR : GL_NEAREST);
}
-void TextureCacheOpenGL::BufferCopy(Surface& src_surface, Surface& dst_surface) {
- MICROPROFILE_SCOPE(OpenGL_Texture_Buffer_Copy);
- const auto& src_params = src_surface->GetSurfaceParams();
- const auto& dst_params = dst_surface->GetSurfaceParams();
- UNIMPLEMENTED_IF(src_params.num_levels > 1 || dst_params.num_levels > 1);
+ImageView::ImageView(TextureCacheRuntime& runtime, const VideoCommon::ImageViewInfo& info,
+ ImageId image_id_, Image& image)
+ : VideoCommon::ImageViewBase{info, image.info, image_id_}, views{runtime.null_image_views} {
+ const Device& device = runtime.device;
+ if (True(image.flags & ImageFlagBits::Converted)) {
+ internal_format = IsPixelFormatSRGB(info.format) ? GL_SRGB8_ALPHA8 : GL_RGBA8;
+ } else {
+ internal_format = GetFormatTuple(format).internal_format;
+ }
+ VideoCommon::SubresourceRange flatten_range = info.range;
+ std::array<GLuint, 2> handles;
+ stored_views.reserve(2);
- const auto source_format = GetFormatTuple(src_params.pixel_format);
- const auto dest_format = GetFormatTuple(dst_params.pixel_format);
+ switch (info.type) {
+ case ImageViewType::e1DArray:
+ flatten_range.extent.layers = 1;
+ [[fallthrough]];
+ case ImageViewType::e1D:
+ glGenTextures(2, handles.data());
+ SetupView(device, image, ImageViewType::e1D, handles[0], info, flatten_range);
+ SetupView(device, image, ImageViewType::e1DArray, handles[1], info, info.range);
+ break;
+ case ImageViewType::e2DArray:
+ flatten_range.extent.layers = 1;
+ [[fallthrough]];
+ case ImageViewType::e2D:
+ if (True(flags & VideoCommon::ImageViewFlagBits::Slice)) {
+ // 2D and 2D array views on a 3D textures are used exclusively for render targets
+ ASSERT(info.range.extent.levels == 1);
+ const VideoCommon::SubresourceRange slice_range{
+ .base = {.level = info.range.base.level, .layer = 0},
+ .extent = {.levels = 1, .layers = 1},
+ };
+ glGenTextures(1, handles.data());
+ SetupView(device, image, ImageViewType::e3D, handles[0], info, slice_range);
+ break;
+ }
+ glGenTextures(2, handles.data());
+ SetupView(device, image, ImageViewType::e2D, handles[0], info, flatten_range);
+ SetupView(device, image, ImageViewType::e2DArray, handles[1], info, info.range);
+ break;
+ case ImageViewType::e3D:
+ glGenTextures(1, handles.data());
+ SetupView(device, image, ImageViewType::e3D, handles[0], info, info.range);
+ break;
+ case ImageViewType::CubeArray:
+ flatten_range.extent.layers = 6;
+ [[fallthrough]];
+ case ImageViewType::Cube:
+ glGenTextures(2, handles.data());
+ SetupView(device, image, ImageViewType::Cube, handles[0], info, flatten_range);
+ SetupView(device, image, ImageViewType::CubeArray, handles[1], info, info.range);
+ break;
+ case ImageViewType::Rect:
+ glGenTextures(1, handles.data());
+ SetupView(device, image, ImageViewType::Rect, handles[0], info, info.range);
+ break;
+ case ImageViewType::Buffer:
+ glCreateTextures(GL_TEXTURE_BUFFER, 1, handles.data());
+ SetupView(device, image, ImageViewType::Buffer, handles[0], info, info.range);
+ break;
+ }
+ default_handle = Handle(info.type);
+}
- const std::size_t source_size = src_surface->GetHostSizeInBytes();
- const std::size_t dest_size = dst_surface->GetHostSizeInBytes();
+ImageView::ImageView(TextureCacheRuntime& runtime, const VideoCommon::NullImageParams& params)
+ : VideoCommon::ImageViewBase{params}, views{runtime.null_image_views} {}
- const std::size_t buffer_size = std::max(source_size, dest_size);
+void ImageView::SetupView(const Device& device, Image& image, ImageViewType view_type,
+ GLuint handle, const VideoCommon::ImageViewInfo& info,
+ VideoCommon::SubresourceRange view_range) {
+ if (info.type == ImageViewType::Buffer) {
+ // TODO: Take offset from buffer cache
+ glTextureBufferRange(handle, internal_format, image.buffer.handle, 0,
+ image.guest_size_bytes);
+ } else {
+ const GLuint parent = image.texture.handle;
+ const GLenum target = ImageTarget(view_type, image.info.num_samples);
+ glTextureView(handle, target, parent, internal_format, view_range.base.level,
+ view_range.extent.levels, view_range.base.layer, view_range.extent.layers);
+ if (!info.IsRenderTarget()) {
+ ApplySwizzle(handle, format, info.Swizzle());
+ }
+ }
+ if (device.HasDebuggingToolAttached()) {
+ const std::string name = VideoCommon::Name(*this, view_type);
+ glObjectLabel(GL_TEXTURE, handle, static_cast<GLsizei>(name.size()), name.data());
+ }
+ stored_views.emplace_back().handle = handle;
+ views[static_cast<size_t>(view_type)] = handle;
+}
- GLuint copy_pbo_handle = FetchPBO(buffer_size);
+Sampler::Sampler(TextureCacheRuntime& runtime, const TSCEntry& config) {
+ const GLenum compare_mode = config.depth_compare_enabled ? GL_COMPARE_REF_TO_TEXTURE : GL_NONE;
+ const GLenum compare_func = MaxwellToGL::DepthCompareFunc(config.depth_compare_func);
+ const GLenum mag = MaxwellToGL::TextureFilterMode(config.mag_filter, TextureMipmapFilter::None);
+ const GLenum min = MaxwellToGL::TextureFilterMode(config.min_filter, config.mipmap_filter);
+ const GLenum reduction_filter = MaxwellToGL::ReductionFilter(config.reduction_filter);
+ const GLint seamless = config.cubemap_interface_filtering ? GL_TRUE : GL_FALSE;
+
+ UNIMPLEMENTED_IF(config.cubemap_anisotropy != 1);
+ UNIMPLEMENTED_IF(config.float_coord_normalization != 0);
+
+ sampler.Create();
+ const GLuint handle = sampler.handle;
+ glSamplerParameteri(handle, GL_TEXTURE_WRAP_S, MaxwellToGL::WrapMode(config.wrap_u));
+ glSamplerParameteri(handle, GL_TEXTURE_WRAP_T, MaxwellToGL::WrapMode(config.wrap_v));
+ glSamplerParameteri(handle, GL_TEXTURE_WRAP_R, MaxwellToGL::WrapMode(config.wrap_p));
+ glSamplerParameteri(handle, GL_TEXTURE_COMPARE_MODE, compare_mode);
+ glSamplerParameteri(handle, GL_TEXTURE_COMPARE_FUNC, compare_func);
+ glSamplerParameteri(handle, GL_TEXTURE_MAG_FILTER, mag);
+ glSamplerParameteri(handle, GL_TEXTURE_MIN_FILTER, min);
+ glSamplerParameterf(handle, GL_TEXTURE_LOD_BIAS, config.LodBias());
+ glSamplerParameterf(handle, GL_TEXTURE_MIN_LOD, config.MinLod());
+ glSamplerParameterf(handle, GL_TEXTURE_MAX_LOD, config.MaxLod());
+ glSamplerParameterfv(handle, GL_TEXTURE_BORDER_COLOR, config.BorderColor().data());
+
+ if (GLAD_GL_ARB_texture_filter_anisotropic || GLAD_GL_EXT_texture_filter_anisotropic) {
+ glSamplerParameterf(handle, GL_TEXTURE_MAX_ANISOTROPY, config.MaxAnisotropy());
+ } else {
+ LOG_WARNING(Render_OpenGL, "GL_ARB_texture_filter_anisotropic is required");
+ }
+ if (GLAD_GL_ARB_texture_filter_minmax || GLAD_GL_EXT_texture_filter_minmax) {
+ glSamplerParameteri(handle, GL_TEXTURE_REDUCTION_MODE_ARB, reduction_filter);
+ } else if (reduction_filter != GL_WEIGHTED_AVERAGE_ARB) {
+ LOG_WARNING(Render_OpenGL, "GL_ARB_texture_filter_minmax is required");
+ }
+ if (GLAD_GL_ARB_seamless_cubemap_per_texture || GLAD_GL_AMD_seamless_cubemap_per_texture) {
+ glSamplerParameteri(handle, GL_TEXTURE_CUBE_MAP_SEAMLESS, seamless);
+ } else if (seamless == GL_FALSE) {
+ // We default to false because it's more common
+ LOG_WARNING(Render_OpenGL, "GL_ARB_seamless_cubemap_per_texture is required");
+ }
+}
- glBindBuffer(GL_PIXEL_PACK_BUFFER, copy_pbo_handle);
+Framebuffer::Framebuffer(TextureCacheRuntime& runtime, std::span<ImageView*, NUM_RT> color_buffers,
+ ImageView* depth_buffer, const VideoCommon::RenderTargets& key) {
+ // Bind to READ_FRAMEBUFFER to stop Nvidia's driver from creating an EXT_framebuffer instead of
+ // a core framebuffer. EXT framebuffer attachments have to match in size and can be shared
+ // across contexts. yuzu doesn't share framebuffers across contexts and we need attachments with
+ // mismatching size, this is why core framebuffers are preferred.
+ GLuint handle;
+ glGenFramebuffers(1, &handle);
+ glBindFramebuffer(GL_READ_FRAMEBUFFER, handle);
+
+ GLsizei num_buffers = 0;
+ std::array<GLenum, NUM_RT> gl_draw_buffers;
+ gl_draw_buffers.fill(GL_NONE);
+
+ for (size_t index = 0; index < color_buffers.size(); ++index) {
+ const ImageView* const image_view = color_buffers[index];
+ if (!image_view) {
+ continue;
+ }
+ buffer_bits |= GL_COLOR_BUFFER_BIT;
+ gl_draw_buffers[index] = GL_COLOR_ATTACHMENT0 + key.draw_buffers[index];
+ num_buffers = static_cast<GLsizei>(index + 1);
- if (src_surface->IsCompressed()) {
- glGetCompressedTextureImage(src_surface->GetTexture(), 0, static_cast<GLsizei>(source_size),
- nullptr);
- } else {
- glGetTextureImage(src_surface->GetTexture(), 0, source_format.format, source_format.type,
- static_cast<GLsizei>(source_size), nullptr);
+ const GLenum attachment = static_cast<GLenum>(GL_COLOR_ATTACHMENT0 + index);
+ AttachTexture(handle, attachment, image_view);
}
- glBindBuffer(GL_PIXEL_PACK_BUFFER, 0);
- glBindBuffer(GL_PIXEL_UNPACK_BUFFER, copy_pbo_handle);
+ if (const ImageView* const image_view = depth_buffer; image_view) {
+ if (GetFormatType(image_view->format) == SurfaceType::DepthStencil) {
+ buffer_bits |= GL_DEPTH_BUFFER_BIT | GL_STENCIL_BUFFER_BIT;
+ } else {
+ buffer_bits |= GL_DEPTH_BUFFER_BIT;
+ }
+ const GLenum attachment = AttachmentType(image_view->format);
+ AttachTexture(handle, attachment, image_view);
+ }
- const GLsizei width = static_cast<GLsizei>(dst_params.width);
- const GLsizei height = static_cast<GLsizei>(dst_params.height);
- const GLsizei depth = static_cast<GLsizei>(dst_params.depth);
- if (dst_surface->IsCompressed()) {
- LOG_CRITICAL(HW_GPU, "Compressed buffer copy is unimplemented!");
- UNREACHABLE();
+ if (num_buffers > 1) {
+ glNamedFramebufferDrawBuffers(handle, num_buffers, gl_draw_buffers.data());
+ } else if (num_buffers > 0) {
+ glNamedFramebufferDrawBuffer(handle, gl_draw_buffers[0]);
} else {
- switch (dst_params.target) {
- case SurfaceTarget::Texture1D:
- glTextureSubImage1D(dst_surface->GetTexture(), 0, 0, width, dest_format.format,
- dest_format.type, nullptr);
- break;
- case SurfaceTarget::Texture2D:
- glTextureSubImage2D(dst_surface->GetTexture(), 0, 0, 0, width, height,
- dest_format.format, dest_format.type, nullptr);
- break;
- case SurfaceTarget::Texture3D:
- case SurfaceTarget::Texture2DArray:
- case SurfaceTarget::TextureCubeArray:
- glTextureSubImage3D(dst_surface->GetTexture(), 0, 0, 0, 0, width, height, depth,
- dest_format.format, dest_format.type, nullptr);
- break;
- case SurfaceTarget::TextureCubemap:
- glTextureSubImage3D(dst_surface->GetTexture(), 0, 0, 0, 0, width, height, depth,
- dest_format.format, dest_format.type, nullptr);
- break;
- default:
- LOG_CRITICAL(Render_OpenGL, "Unimplemented surface target={}",
- static_cast<u32>(dst_params.target));
- UNREACHABLE();
- }
+ glNamedFramebufferDrawBuffer(handle, GL_NONE);
}
- glBindBuffer(GL_PIXEL_UNPACK_BUFFER, 0);
- glTextureBarrier();
-}
+ glNamedFramebufferParameteri(handle, GL_FRAMEBUFFER_DEFAULT_WIDTH, key.size.width);
+ glNamedFramebufferParameteri(handle, GL_FRAMEBUFFER_DEFAULT_HEIGHT, key.size.height);
+ // TODO
+ // glNamedFramebufferParameteri(handle, GL_FRAMEBUFFER_DEFAULT_LAYERS, ...);
+ // glNamedFramebufferParameteri(handle, GL_FRAMEBUFFER_DEFAULT_SAMPLES, ...);
+ // glNamedFramebufferParameteri(handle, GL_FRAMEBUFFER_DEFAULT_FIXED_SAMPLE_LOCATIONS, ...);
-GLuint TextureCacheOpenGL::FetchPBO(std::size_t buffer_size) {
- ASSERT_OR_EXECUTE(buffer_size > 0, { return 0; });
- const u32 l2 = Common::Log2Ceil64(static_cast<u64>(buffer_size));
- OGLBuffer& cp = copy_pbo_cache[l2];
- if (cp.handle == 0) {
- const std::size_t ceil_size = 1ULL << l2;
- cp.Create();
- cp.MakeStreamCopy(ceil_size);
+ if (runtime.device.HasDebuggingToolAttached()) {
+ const std::string name = VideoCommon::Name(key);
+ glObjectLabel(GL_FRAMEBUFFER, handle, static_cast<GLsizei>(name.size()), name.data());
}
- return cp.handle;
+ framebuffer.handle = handle;
}
} // namespace OpenGL
diff --git a/src/video_core/renderer_opengl/gl_texture_cache.h b/src/video_core/renderer_opengl/gl_texture_cache.h
index 7787134fc..04193e31e 100644
--- a/src/video_core/renderer_opengl/gl_texture_cache.h
+++ b/src/video_core/renderer_opengl/gl_texture_cache.h
@@ -4,156 +4,247 @@
#pragma once
-#include <array>
-#include <functional>
#include <memory>
-#include <unordered_map>
-#include <utility>
-#include <vector>
+#include <span>
#include <glad/glad.h>
-#include "common/common_types.h"
-#include "video_core/engines/shader_bytecode.h"
-#include "video_core/renderer_opengl/gl_device.h"
#include "video_core/renderer_opengl/gl_resource_manager.h"
+#include "video_core/renderer_opengl/util_shaders.h"
#include "video_core/texture_cache/texture_cache.h"
namespace OpenGL {
-using VideoCommon::SurfaceParams;
-using VideoCommon::ViewParams;
-
-class CachedSurfaceView;
-class CachedSurface;
-class TextureCacheOpenGL;
+class Device;
+class ProgramManager;
class StateTracker;
-using Surface = std::shared_ptr<CachedSurface>;
-using View = std::shared_ptr<CachedSurfaceView>;
-using TextureCacheBase = VideoCommon::TextureCache<Surface, View>;
+class Framebuffer;
+class Image;
+class ImageView;
+class Sampler;
-class CachedSurface final : public VideoCommon::SurfaceBase<View> {
- friend CachedSurfaceView;
+using VideoCommon::ImageId;
+using VideoCommon::ImageViewId;
+using VideoCommon::ImageViewType;
+using VideoCommon::NUM_RT;
+using VideoCommon::Offset2D;
+using VideoCommon::RenderTargets;
+class ImageBufferMap {
public:
- explicit CachedSurface(GPUVAddr gpu_addr, const SurfaceParams& params, bool is_astc_supported);
- ~CachedSurface();
-
- void UploadTexture(const std::vector<u8>& staging_buffer) override;
- void DownloadTexture(std::vector<u8>& staging_buffer) override;
+ explicit ImageBufferMap(GLuint handle, u8* map, size_t size, OGLSync* sync);
+ ~ImageBufferMap();
- GLenum GetTarget() const {
- return target;
+ GLuint Handle() const noexcept {
+ return handle;
}
- GLuint GetTexture() const {
- return texture.handle;
+ std::span<u8> Span() const noexcept {
+ return span;
}
- bool IsCompressed() const {
- return is_compressed;
+private:
+ std::span<u8> span;
+ OGLSync* sync;
+ GLuint handle;
+};
+
+struct FormatProperties {
+ GLenum compatibility_class;
+ bool compatibility_by_size;
+ bool is_compressed;
+};
+
+class TextureCacheRuntime {
+ friend Framebuffer;
+ friend Image;
+ friend ImageView;
+ friend Sampler;
+
+public:
+ explicit TextureCacheRuntime(const Device& device, ProgramManager& program_manager,
+ StateTracker& state_tracker);
+ ~TextureCacheRuntime();
+
+ void Finish();
+
+ ImageBufferMap MapUploadBuffer(size_t size);
+
+ ImageBufferMap MapDownloadBuffer(size_t size);
+
+ void CopyImage(Image& dst, Image& src, std::span<const VideoCommon::ImageCopy> copies);
+
+ void ConvertImage(Framebuffer* dst, ImageView& dst_view, ImageView& src_view) {
+ UNIMPLEMENTED();
}
-protected:
- void DecorateSurfaceName() override;
+ bool CanImageBeCopied(const Image& dst, const Image& src);
+
+ void EmulateCopyImage(Image& dst, Image& src, std::span<const VideoCommon::ImageCopy> copies);
+
+ void BlitFramebuffer(Framebuffer* dst, Framebuffer* src,
+ const std::array<Offset2D, 2>& dst_region,
+ const std::array<Offset2D, 2>& src_region,
+ Tegra::Engines::Fermi2D::Filter filter,
+ Tegra::Engines::Fermi2D::Operation operation);
- View CreateView(const ViewParams& view_key) override;
- View CreateViewInner(const ViewParams& view_key, bool is_proxy);
+ void AccelerateImageUpload(Image& image, const ImageBufferMap& map, size_t buffer_offset,
+ std::span<const VideoCommon::SwizzleParameters> swizzles);
+
+ void InsertUploadMemoryBarrier();
+
+ FormatProperties FormatInfo(VideoCommon::ImageType type, GLenum internal_format) const;
private:
- void UploadTextureMipmap(u32 level, const std::vector<u8>& staging_buffer);
+ struct StagingBuffers {
+ explicit StagingBuffers(GLenum storage_flags_, GLenum map_flags_);
+ ~StagingBuffers();
- GLenum internal_format{};
- GLenum format{};
- GLenum type{};
- bool is_compressed{};
- GLenum target{};
- u32 view_count{};
+ ImageBufferMap RequestMap(size_t requested_size, bool insert_fence);
- OGLTexture texture;
- OGLBuffer texture_buffer;
+ size_t RequestBuffer(size_t requested_size);
+
+ std::optional<size_t> FindBuffer(size_t requested_size);
+
+ std::vector<OGLSync> syncs;
+ std::vector<OGLBuffer> buffers;
+ std::vector<u8*> maps;
+ std::vector<size_t> sizes;
+ GLenum storage_flags;
+ GLenum map_flags;
+ };
+
+ const Device& device;
+ StateTracker& state_tracker;
+ UtilShaders util_shaders;
+
+ std::array<std::unordered_map<GLenum, FormatProperties>, 3> format_properties;
+
+ StagingBuffers upload_buffers{GL_MAP_WRITE_BIT, GL_MAP_WRITE_BIT | GL_MAP_FLUSH_EXPLICIT_BIT};
+ StagingBuffers download_buffers{GL_MAP_READ_BIT, GL_MAP_READ_BIT};
+
+ OGLTexture null_image_1d_array;
+ OGLTexture null_image_cube_array;
+ OGLTexture null_image_3d;
+ OGLTexture null_image_rect;
+ OGLTextureView null_image_view_1d;
+ OGLTextureView null_image_view_2d;
+ OGLTextureView null_image_view_2d_array;
+ OGLTextureView null_image_view_cube;
+
+ std::array<GLuint, VideoCommon::NUM_IMAGE_VIEW_TYPES> null_image_views;
};
-class CachedSurfaceView final : public VideoCommon::ViewBase {
+class Image : public VideoCommon::ImageBase {
+ friend ImageView;
+
public:
- explicit CachedSurfaceView(CachedSurface& surface, const ViewParams& params, bool is_proxy);
- ~CachedSurfaceView();
+ explicit Image(TextureCacheRuntime&, const VideoCommon::ImageInfo& info, GPUVAddr gpu_addr,
+ VAddr cpu_addr);
- /// @brief Attaches this texture view to the currently bound fb_target framebuffer
- /// @param attachment Attachment to bind textures to
- /// @param fb_target Framebuffer target to attach to (e.g. DRAW_FRAMEBUFFER)
- void Attach(GLenum attachment, GLenum fb_target) const;
+ void UploadMemory(const ImageBufferMap& map, size_t buffer_offset,
+ std::span<const VideoCommon::BufferImageCopy> copies);
- GLuint GetTexture(Tegra::Texture::SwizzleSource x_source,
- Tegra::Texture::SwizzleSource y_source,
- Tegra::Texture::SwizzleSource z_source,
- Tegra::Texture::SwizzleSource w_source);
+ void UploadMemory(const ImageBufferMap& map, size_t buffer_offset,
+ std::span<const VideoCommon::BufferCopy> copies);
- void DecorateViewName(GPUVAddr gpu_addr, const std::string& prefix);
+ void DownloadMemory(ImageBufferMap& map, size_t buffer_offset,
+ std::span<const VideoCommon::BufferImageCopy> copies);
- void MarkAsModified(u64 tick) {
- surface.MarkAsModified(true, tick);
+ GLuint Handle() const noexcept {
+ return texture.handle;
}
- GLuint GetTexture() const {
- if (is_proxy) {
- return surface.GetTexture();
- }
- return main_view.handle;
+private:
+ void CopyBufferToImage(const VideoCommon::BufferImageCopy& copy, size_t buffer_offset);
+
+ void CopyImageToBuffer(const VideoCommon::BufferImageCopy& copy, size_t buffer_offset);
+
+ OGLTexture texture;
+ OGLTextureView store_view;
+ OGLBuffer buffer;
+ GLenum gl_internal_format = GL_NONE;
+ GLenum gl_store_format = GL_NONE;
+ GLenum gl_format = GL_NONE;
+ GLenum gl_type = GL_NONE;
+};
+
+class ImageView : public VideoCommon::ImageViewBase {
+ friend Image;
+
+public:
+ explicit ImageView(TextureCacheRuntime&, const VideoCommon::ImageViewInfo&, ImageId, Image&);
+ explicit ImageView(TextureCacheRuntime&, const VideoCommon::NullImageParams&);
+
+ [[nodiscard]] GLuint Handle(ImageViewType query_type) const noexcept {
+ return views[static_cast<size_t>(query_type)];
}
- GLenum GetFormat() const {
- return format;
+ [[nodiscard]] GLuint DefaultHandle() const noexcept {
+ return default_handle;
}
- const SurfaceParams& GetSurfaceParams() const {
- return surface.GetSurfaceParams();
+ [[nodiscard]] GLenum Format() const noexcept {
+ return internal_format;
}
private:
- OGLTextureView CreateTextureView() const;
+ void SetupView(const Device& device, Image& image, ImageViewType view_type, GLuint handle,
+ const VideoCommon::ImageViewInfo& info,
+ VideoCommon::SubresourceRange view_range);
+
+ std::array<GLuint, VideoCommon::NUM_IMAGE_VIEW_TYPES> views{};
+ std::vector<OGLTextureView> stored_views;
+ GLuint default_handle = 0;
+ GLenum internal_format = GL_NONE;
+};
+
+class ImageAlloc : public VideoCommon::ImageAllocBase {};
- CachedSurface& surface;
- const GLenum format;
- const GLenum target;
- const bool is_proxy;
+class Sampler {
+public:
+ explicit Sampler(TextureCacheRuntime&, const Tegra::Texture::TSCEntry&);
- std::unordered_map<u32, OGLTextureView> view_cache;
- OGLTextureView main_view;
+ GLuint Handle() const noexcept {
+ return sampler.handle;
+ }
- // Use an invalid default so it always fails the comparison test
- u32 current_swizzle = 0xffffffff;
- GLuint current_view = 0;
+private:
+ OGLSampler sampler;
};
-class TextureCacheOpenGL final : public TextureCacheBase {
+class Framebuffer {
public:
- explicit TextureCacheOpenGL(VideoCore::RasterizerInterface& rasterizer,
- Tegra::Engines::Maxwell3D& maxwell3d,
- Tegra::MemoryManager& gpu_memory, const Device& device,
- StateTracker& state_tracker);
- ~TextureCacheOpenGL();
-
-protected:
- Surface CreateSurface(GPUVAddr gpu_addr, const SurfaceParams& params) override;
-
- void ImageCopy(Surface& src_surface, Surface& dst_surface,
- const VideoCommon::CopyParams& copy_params) override;
+ explicit Framebuffer(TextureCacheRuntime&, std::span<ImageView*, NUM_RT> color_buffers,
+ ImageView* depth_buffer, const VideoCommon::RenderTargets& key);
- void ImageBlit(View& src_view, View& dst_view,
- const Tegra::Engines::Fermi2D::Config& copy_config) override;
+ [[nodiscard]] GLuint Handle() const noexcept {
+ return framebuffer.handle;
+ }
- void BufferCopy(Surface& src_surface, Surface& dst_surface) override;
+ [[nodiscard]] GLbitfield BufferBits() const noexcept {
+ return buffer_bits;
+ }
private:
- GLuint FetchPBO(std::size_t buffer_size);
-
- StateTracker& state_tracker;
+ OGLFramebuffer framebuffer;
+ GLbitfield buffer_bits = GL_NONE;
+};
- OGLFramebuffer src_framebuffer;
- OGLFramebuffer dst_framebuffer;
- std::unordered_map<u32, OGLBuffer> copy_pbo_cache;
+struct TextureCacheParams {
+ static constexpr bool ENABLE_VALIDATION = true;
+ static constexpr bool FRAMEBUFFER_BLITS = true;
+ static constexpr bool HAS_EMULATED_COPIES = true;
+
+ using Runtime = OpenGL::TextureCacheRuntime;
+ using Image = OpenGL::Image;
+ using ImageAlloc = OpenGL::ImageAlloc;
+ using ImageView = OpenGL::ImageView;
+ using Sampler = OpenGL::Sampler;
+ using Framebuffer = OpenGL::Framebuffer;
};
+using TextureCache = VideoCommon::TextureCache<TextureCacheParams>;
+
} // namespace OpenGL
diff --git a/src/video_core/renderer_opengl/maxwell_to_gl.h b/src/video_core/renderer_opengl/maxwell_to_gl.h
index a8be2aa37..cbccfdeb4 100644
--- a/src/video_core/renderer_opengl/maxwell_to_gl.h
+++ b/src/video_core/renderer_opengl/maxwell_to_gl.h
@@ -107,7 +107,7 @@ inline GLenum IndexFormat(Maxwell::IndexFormat index_format) {
case Maxwell::IndexFormat::UnsignedInt:
return GL_UNSIGNED_INT;
}
- UNREACHABLE_MSG("Invalid index_format={}", static_cast<u32>(index_format));
+ UNREACHABLE_MSG("Invalid index_format={}", index_format);
return {};
}
@@ -144,7 +144,7 @@ inline GLenum PrimitiveTopology(Maxwell::PrimitiveTopology topology) {
case Maxwell::PrimitiveTopology::Patches:
return GL_PATCHES;
}
- UNREACHABLE_MSG("Invalid topology={}", static_cast<int>(topology));
+ UNREACHABLE_MSG("Invalid topology={}", topology);
return GL_POINTS;
}
@@ -172,8 +172,8 @@ inline GLenum TextureFilterMode(Tegra::Texture::TextureFilter filter_mode,
}
break;
}
- UNREACHABLE_MSG("Invalid texture filter mode={} and mipmap filter mode={}",
- static_cast<u32>(filter_mode), static_cast<u32>(mipmap_filter_mode));
+ UNREACHABLE_MSG("Invalid texture filter mode={} and mipmap filter mode={}", filter_mode,
+ mipmap_filter_mode);
return GL_NEAREST;
}
@@ -204,7 +204,7 @@ inline GLenum WrapMode(Tegra::Texture::WrapMode wrap_mode) {
return GL_MIRROR_CLAMP_TO_EDGE;
}
}
- UNIMPLEMENTED_MSG("Unimplemented texture wrap mode={}", static_cast<u32>(wrap_mode));
+ UNIMPLEMENTED_MSG("Unimplemented texture wrap mode={}", wrap_mode);
return GL_REPEAT;
}
@@ -227,7 +227,7 @@ inline GLenum DepthCompareFunc(Tegra::Texture::DepthCompareFunc func) {
case Tegra::Texture::DepthCompareFunc::Always:
return GL_ALWAYS;
}
- UNIMPLEMENTED_MSG("Unimplemented texture depth compare function={}", static_cast<u32>(func));
+ UNIMPLEMENTED_MSG("Unimplemented texture depth compare function={}", func);
return GL_GREATER;
}
@@ -249,7 +249,7 @@ inline GLenum BlendEquation(Maxwell::Blend::Equation equation) {
case Maxwell::Blend::Equation::MaxGL:
return GL_MAX;
}
- UNIMPLEMENTED_MSG("Unimplemented blend equation={}", static_cast<u32>(equation));
+ UNIMPLEMENTED_MSG("Unimplemented blend equation={}", equation);
return GL_FUNC_ADD;
}
@@ -313,7 +313,7 @@ inline GLenum BlendFunc(Maxwell::Blend::Factor factor) {
case Maxwell::Blend::Factor::OneMinusConstantAlphaGL:
return GL_ONE_MINUS_CONSTANT_ALPHA;
}
- UNIMPLEMENTED_MSG("Unimplemented blend factor={}", static_cast<u32>(factor));
+ UNIMPLEMENTED_MSG("Unimplemented blend factor={}", factor);
return GL_ZERO;
}
@@ -333,7 +333,7 @@ inline GLenum SwizzleSource(Tegra::Texture::SwizzleSource source) {
case Tegra::Texture::SwizzleSource::OneFloat:
return GL_ONE;
}
- UNIMPLEMENTED_MSG("Unimplemented swizzle source={}", static_cast<u32>(source));
+ UNIMPLEMENTED_MSG("Unimplemented swizzle source={}", source);
return GL_ZERO;
}
@@ -364,7 +364,7 @@ inline GLenum ComparisonOp(Maxwell::ComparisonOp comparison) {
case Maxwell::ComparisonOp::AlwaysOld:
return GL_ALWAYS;
}
- UNIMPLEMENTED_MSG("Unimplemented comparison op={}", static_cast<u32>(comparison));
+ UNIMPLEMENTED_MSG("Unimplemented comparison op={}", comparison);
return GL_ALWAYS;
}
@@ -395,7 +395,7 @@ inline GLenum StencilOp(Maxwell::StencilOp stencil) {
case Maxwell::StencilOp::DecrWrapOGL:
return GL_DECR_WRAP;
}
- UNIMPLEMENTED_MSG("Unimplemented stencil op={}", static_cast<u32>(stencil));
+ UNIMPLEMENTED_MSG("Unimplemented stencil op={}", stencil);
return GL_KEEP;
}
@@ -406,7 +406,7 @@ inline GLenum FrontFace(Maxwell::FrontFace front_face) {
case Maxwell::FrontFace::CounterClockWise:
return GL_CCW;
}
- UNIMPLEMENTED_MSG("Unimplemented front face cull={}", static_cast<u32>(front_face));
+ UNIMPLEMENTED_MSG("Unimplemented front face cull={}", front_face);
return GL_CCW;
}
@@ -419,7 +419,7 @@ inline GLenum CullFace(Maxwell::CullFace cull_face) {
case Maxwell::CullFace::FrontAndBack:
return GL_FRONT_AND_BACK;
}
- UNIMPLEMENTED_MSG("Unimplemented cull face={}", static_cast<u32>(cull_face));
+ UNIMPLEMENTED_MSG("Unimplemented cull face={}", cull_face);
return GL_BACK;
}
@@ -458,7 +458,7 @@ inline GLenum LogicOp(Maxwell::LogicOperation operation) {
case Maxwell::LogicOperation::Set:
return GL_SET;
}
- UNIMPLEMENTED_MSG("Unimplemented logic operation={}", static_cast<u32>(operation));
+ UNIMPLEMENTED_MSG("Unimplemented logic operation={}", operation);
return GL_COPY;
}
@@ -471,10 +471,23 @@ inline GLenum PolygonMode(Maxwell::PolygonMode polygon_mode) {
case Maxwell::PolygonMode::Fill:
return GL_FILL;
}
- UNREACHABLE_MSG("Invalid polygon mode={}", static_cast<int>(polygon_mode));
+ UNREACHABLE_MSG("Invalid polygon mode={}", polygon_mode);
return GL_FILL;
}
+inline GLenum ReductionFilter(Tegra::Texture::SamplerReduction filter) {
+ switch (filter) {
+ case Tegra::Texture::SamplerReduction::WeightedAverage:
+ return GL_WEIGHTED_AVERAGE_ARB;
+ case Tegra::Texture::SamplerReduction::Min:
+ return GL_MIN;
+ case Tegra::Texture::SamplerReduction::Max:
+ return GL_MAX;
+ }
+ UNREACHABLE_MSG("Invalid reduction filter={}", static_cast<int>(filter));
+ return GL_WEIGHTED_AVERAGE_ARB;
+}
+
inline GLenum ViewportSwizzle(Maxwell::ViewportSwizzle swizzle) {
// Enumeration order matches register order. We can convert it arithmetically.
return GL_VIEWPORT_SWIZZLE_POSITIVE_X_NV + static_cast<GLenum>(swizzle);
diff --git a/src/video_core/renderer_opengl/renderer_opengl.cpp b/src/video_core/renderer_opengl/renderer_opengl.cpp
index c869bb0e2..dd77a543c 100644
--- a/src/video_core/renderer_opengl/renderer_opengl.cpp
+++ b/src/video_core/renderer_opengl/renderer_opengl.cpp
@@ -23,10 +23,10 @@
#include "core/telemetry_session.h"
#include "video_core/host_shaders/opengl_present_frag.h"
#include "video_core/host_shaders/opengl_present_vert.h"
-#include "video_core/morton.h"
#include "video_core/renderer_opengl/gl_rasterizer.h"
#include "video_core/renderer_opengl/gl_shader_manager.h"
#include "video_core/renderer_opengl/renderer_opengl.h"
+#include "video_core/textures/decoders.h"
namespace OpenGL {
@@ -130,8 +130,8 @@ void APIENTRY DebugHandler(GLenum source, GLenum type, GLuint id, GLenum severit
RendererOpenGL::RendererOpenGL(Core::TelemetrySession& telemetry_session_,
Core::Frontend::EmuWindow& emu_window_,
Core::Memory::Memory& cpu_memory_, Tegra::GPU& gpu_,
- std::unique_ptr<Core::Frontend::GraphicsContext> context)
- : RendererBase{emu_window_, std::move(context)}, telemetry_session{telemetry_session_},
+ std::unique_ptr<Core::Frontend::GraphicsContext> context_)
+ : RendererBase{emu_window_, std::move(context_)}, telemetry_session{telemetry_session_},
emu_window{emu_window_}, cpu_memory{cpu_memory_}, gpu{gpu_}, program_manager{device} {}
RendererOpenGL::~RendererOpenGL() = default;
@@ -140,11 +140,10 @@ void RendererOpenGL::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
if (!framebuffer) {
return;
}
-
PrepareRendertarget(framebuffer);
RenderScreenshot();
- glBindFramebuffer(GL_DRAW_FRAMEBUFFER, 0);
+ state_tracker.BindFramebuffer(0);
DrawScreen(emu_window.GetFramebufferLayout());
++m_current_frame;
@@ -187,19 +186,20 @@ void RendererOpenGL::LoadFBToScreenInfo(const Tegra::FramebufferConfig& framebuf
// Reset the screen info's display texture to its own permanent texture
screen_info.display_texture = screen_info.texture.resource.handle;
- const auto pixel_format{
- VideoCore::Surface::PixelFormatFromGPUPixelFormat(framebuffer.pixel_format)};
- const u32 bytes_per_pixel{VideoCore::Surface::GetBytesPerPixel(pixel_format)};
- const u64 size_in_bytes{framebuffer.stride * framebuffer.height * bytes_per_pixel};
- u8* const host_ptr{cpu_memory.GetPointer(framebuffer_addr)};
- rasterizer->FlushRegion(ToCacheAddr(host_ptr), size_in_bytes);
-
// TODO(Rodrigo): Read this from HLE
constexpr u32 block_height_log2 = 4;
- VideoCore::MortonSwizzle(VideoCore::MortonSwizzleMode::MortonToLinear, pixel_format,
- framebuffer.stride, block_height_log2, framebuffer.height, 0, 1, 1,
- gl_framebuffer_data.data(), host_ptr);
-
+ const auto pixel_format{
+ VideoCore::Surface::PixelFormatFromGPUPixelFormat(framebuffer.pixel_format)};
+ const u32 bytes_per_pixel{VideoCore::Surface::BytesPerBlock(pixel_format)};
+ const u64 size_in_bytes{Tegra::Texture::CalculateSize(
+ true, bytes_per_pixel, framebuffer.stride, framebuffer.height, 1, block_height_log2, 0)};
+ const u8* const host_ptr{cpu_memory.GetPointer(framebuffer_addr)};
+ const std::span<const u8> input_data(host_ptr, size_in_bytes);
+ Tegra::Texture::UnswizzleTexture(gl_framebuffer_data, input_data, bytes_per_pixel,
+ framebuffer.width, framebuffer.height, 1, block_height_log2,
+ 0);
+
+ glBindBuffer(GL_PIXEL_UNPACK_BUFFER, 0);
glPixelStorei(GL_UNPACK_ROW_LENGTH, static_cast<GLint>(framebuffer.stride));
// Update existing texture
@@ -238,6 +238,10 @@ void RendererOpenGL::InitOpenGLObjects() {
glUseProgramStages(pipeline.handle, GL_VERTEX_SHADER_BIT, vertex_program.handle);
glUseProgramStages(pipeline.handle, GL_FRAGMENT_SHADER_BIT, fragment_program.handle);
+ // Generate presentation sampler
+ present_sampler.Create();
+ glSamplerParameteri(present_sampler.handle, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
+
// Generate VBO handle for drawing
vertex_buffer.Create();
@@ -255,6 +259,11 @@ void RendererOpenGL::InitOpenGLObjects() {
// Clear screen to black
LoadColorToActiveGLTexture(0, 0, 0, 0, screen_info.texture);
+ // Enable seamless cubemaps when per texture parameters are not available
+ if (!GLAD_GL_ARB_seamless_cubemap_per_texture && !GLAD_GL_AMD_seamless_cubemap_per_texture) {
+ glEnable(GL_TEXTURE_CUBE_MAP_SEAMLESS);
+ }
+
// Enable unified vertex attributes and query vertex buffer address when the driver supports it
if (device.HasVertexBufferUnifiedMemory()) {
glEnableClientState(GL_VERTEX_ATTRIB_ARRAY_UNIFIED_NV);
@@ -275,9 +284,9 @@ void RendererOpenGL::AddTelemetryFields() {
LOG_INFO(Render_OpenGL, "GL_RENDERER: {}", gpu_model);
constexpr auto user_system = Common::Telemetry::FieldType::UserSystem;
- telemetry_session.AddField(user_system, "GPU_Vendor", gpu_vendor);
- telemetry_session.AddField(user_system, "GPU_Model", gpu_model);
- telemetry_session.AddField(user_system, "GPU_OpenGL_Version", gl_version);
+ telemetry_session.AddField(user_system, "GPU_Vendor", std::string(gpu_vendor));
+ telemetry_session.AddField(user_system, "GPU_Model", std::string(gpu_model));
+ telemetry_session.AddField(user_system, "GPU_OpenGL_Version", std::string(gl_version));
}
void RendererOpenGL::CreateRasterizer() {
@@ -296,7 +305,7 @@ void RendererOpenGL::ConfigureFramebufferTexture(TextureInfo& texture,
const auto pixel_format{
VideoCore::Surface::PixelFormatFromGPUPixelFormat(framebuffer.pixel_format)};
- const u32 bytes_per_pixel{VideoCore::Surface::GetBytesPerPixel(pixel_format)};
+ const u32 bytes_per_pixel{VideoCore::Surface::BytesPerBlock(pixel_format)};
gl_framebuffer_data.resize(texture.width * texture.height * bytes_per_pixel);
GLint internal_format;
@@ -315,8 +324,8 @@ void RendererOpenGL::ConfigureFramebufferTexture(TextureInfo& texture,
internal_format = GL_RGBA8;
texture.gl_format = GL_RGBA;
texture.gl_type = GL_UNSIGNED_INT_8_8_8_8_REV;
- UNIMPLEMENTED_MSG("Unknown framebuffer pixel format: {}",
- static_cast<u32>(framebuffer.pixel_format));
+ // UNIMPLEMENTED_MSG("Unknown framebuffer pixel format: {}",
+ // static_cast<u32>(framebuffer.pixel_format));
}
texture.resource.Release();
@@ -348,7 +357,7 @@ void RendererOpenGL::DrawScreen(const Layout::FramebufferLayout& layout) {
} else {
// Other transformations are unsupported
LOG_CRITICAL(Render_OpenGL, "Unsupported framebuffer_transform_flags={}",
- static_cast<u32>(framebuffer_transform_flags));
+ framebuffer_transform_flags);
UNIMPLEMENTED();
}
}
@@ -382,7 +391,7 @@ void RendererOpenGL::DrawScreen(const Layout::FramebufferLayout& layout) {
state_tracker.NotifyPolygonModes();
state_tracker.NotifyViewport0();
state_tracker.NotifyScissor0();
- state_tracker.NotifyColorMask0();
+ state_tracker.NotifyColorMask(0);
state_tracker.NotifyBlend0();
state_tracker.NotifyFramebuffer();
state_tracker.NotifyFrontFace();
@@ -440,7 +449,7 @@ void RendererOpenGL::DrawScreen(const Layout::FramebufferLayout& layout) {
}
glBindTextureUnit(0, screen_info.display_texture);
- glBindSampler(0, 0);
+ glBindSampler(0, present_sampler.handle);
glClear(GL_COLOR_BUFFER_BIT);
glDrawArrays(GL_TRIANGLE_STRIP, 0, 4);
@@ -473,6 +482,8 @@ void RendererOpenGL::RenderScreenshot() {
DrawScreen(layout);
+ glBindBuffer(GL_PIXEL_PACK_BUFFER, 0);
+ glPixelStorei(GL_PACK_ROW_LENGTH, 0);
glReadPixels(0, 0, layout.width, layout.height, GL_BGRA, GL_UNSIGNED_INT_8_8_8_8_REV,
renderer_settings.screenshot_bits);
diff --git a/src/video_core/renderer_opengl/renderer_opengl.h b/src/video_core/renderer_opengl/renderer_opengl.h
index 9ef181f95..44e109794 100644
--- a/src/video_core/renderer_opengl/renderer_opengl.h
+++ b/src/video_core/renderer_opengl/renderer_opengl.h
@@ -57,10 +57,10 @@ struct ScreenInfo {
class RendererOpenGL final : public VideoCore::RendererBase {
public:
- explicit RendererOpenGL(Core::TelemetrySession& telemetry_session,
- Core::Frontend::EmuWindow& emu_window, Core::Memory::Memory& cpu_memory,
- Tegra::GPU& gpu,
- std::unique_ptr<Core::Frontend::GraphicsContext> context);
+ explicit RendererOpenGL(Core::TelemetrySession& telemetry_session_,
+ Core::Frontend::EmuWindow& emu_window_,
+ Core::Memory::Memory& cpu_memory_, Tegra::GPU& gpu_,
+ std::unique_ptr<Core::Frontend::GraphicsContext> context_);
~RendererOpenGL() override;
bool Init() override;
@@ -102,6 +102,7 @@ private:
StateTracker state_tracker{gpu};
// OpenGL object IDs
+ OGLSampler present_sampler;
OGLBuffer vertex_buffer;
OGLProgram vertex_program;
OGLProgram fragment_program;
diff --git a/src/video_core/renderer_opengl/util_shaders.cpp b/src/video_core/renderer_opengl/util_shaders.cpp
new file mode 100644
index 000000000..eb849cbf2
--- /dev/null
+++ b/src/video_core/renderer_opengl/util_shaders.cpp
@@ -0,0 +1,224 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <bit>
+#include <span>
+#include <string_view>
+
+#include <glad/glad.h>
+
+#include "common/assert.h"
+#include "common/common_types.h"
+#include "common/div_ceil.h"
+#include "video_core/host_shaders/block_linear_unswizzle_2d_comp.h"
+#include "video_core/host_shaders/block_linear_unswizzle_3d_comp.h"
+#include "video_core/host_shaders/opengl_copy_bc4_comp.h"
+#include "video_core/host_shaders/pitch_unswizzle_comp.h"
+#include "video_core/renderer_opengl/gl_resource_manager.h"
+#include "video_core/renderer_opengl/gl_shader_manager.h"
+#include "video_core/renderer_opengl/gl_texture_cache.h"
+#include "video_core/renderer_opengl/util_shaders.h"
+#include "video_core/surface.h"
+#include "video_core/texture_cache/accelerated_swizzle.h"
+#include "video_core/texture_cache/types.h"
+#include "video_core/texture_cache/util.h"
+#include "video_core/textures/decoders.h"
+
+namespace OpenGL {
+
+using namespace HostShaders;
+
+using VideoCommon::Extent3D;
+using VideoCommon::ImageCopy;
+using VideoCommon::ImageType;
+using VideoCommon::SwizzleParameters;
+using VideoCommon::Accelerated::MakeBlockLinearSwizzle2DParams;
+using VideoCommon::Accelerated::MakeBlockLinearSwizzle3DParams;
+using VideoCore::Surface::BytesPerBlock;
+
+namespace {
+
+OGLProgram MakeProgram(std::string_view source) {
+ OGLShader shader;
+ shader.Create(source, GL_COMPUTE_SHADER);
+
+ OGLProgram program;
+ program.Create(true, false, shader.handle);
+ return program;
+}
+
+} // Anonymous namespace
+
+UtilShaders::UtilShaders(ProgramManager& program_manager_)
+ : program_manager{program_manager_},
+ block_linear_unswizzle_2d_program(MakeProgram(BLOCK_LINEAR_UNSWIZZLE_2D_COMP)),
+ block_linear_unswizzle_3d_program(MakeProgram(BLOCK_LINEAR_UNSWIZZLE_3D_COMP)),
+ pitch_unswizzle_program(MakeProgram(PITCH_UNSWIZZLE_COMP)),
+ copy_bc4_program(MakeProgram(OPENGL_COPY_BC4_COMP)) {
+ const auto swizzle_table = Tegra::Texture::MakeSwizzleTable();
+ swizzle_table_buffer.Create();
+ glNamedBufferStorage(swizzle_table_buffer.handle, sizeof(swizzle_table), &swizzle_table, 0);
+}
+
+UtilShaders::~UtilShaders() = default;
+
+void UtilShaders::BlockLinearUpload2D(Image& image, const ImageBufferMap& map, size_t buffer_offset,
+ std::span<const SwizzleParameters> swizzles) {
+ static constexpr Extent3D WORKGROUP_SIZE{32, 32, 1};
+ static constexpr GLuint BINDING_SWIZZLE_BUFFER = 0;
+ static constexpr GLuint BINDING_INPUT_BUFFER = 1;
+ static constexpr GLuint BINDING_OUTPUT_IMAGE = 0;
+
+ program_manager.BindHostCompute(block_linear_unswizzle_2d_program.handle);
+ glFlushMappedNamedBufferRange(map.Handle(), buffer_offset, image.guest_size_bytes);
+ glBindBufferBase(GL_SHADER_STORAGE_BUFFER, BINDING_SWIZZLE_BUFFER, swizzle_table_buffer.handle);
+
+ const GLenum store_format = StoreFormat(BytesPerBlock(image.info.format));
+ for (const SwizzleParameters& swizzle : swizzles) {
+ const Extent3D num_tiles = swizzle.num_tiles;
+ const size_t input_offset = swizzle.buffer_offset + buffer_offset;
+
+ const u32 num_dispatches_x = Common::DivCeil(num_tiles.width, WORKGROUP_SIZE.width);
+ const u32 num_dispatches_y = Common::DivCeil(num_tiles.height, WORKGROUP_SIZE.height);
+
+ const auto params = MakeBlockLinearSwizzle2DParams(swizzle, image.info);
+ glUniform3uiv(0, 1, params.origin.data());
+ glUniform3iv(1, 1, params.destination.data());
+ glUniform1ui(2, params.bytes_per_block_log2);
+ glUniform1ui(3, params.layer_stride);
+ glUniform1ui(4, params.block_size);
+ glUniform1ui(5, params.x_shift);
+ glUniform1ui(6, params.block_height);
+ glUniform1ui(7, params.block_height_mask);
+ glBindBufferRange(GL_SHADER_STORAGE_BUFFER, BINDING_INPUT_BUFFER, map.Handle(),
+ input_offset, image.guest_size_bytes - swizzle.buffer_offset);
+ glBindImageTexture(BINDING_OUTPUT_IMAGE, image.Handle(), swizzle.level, GL_TRUE, 0,
+ GL_WRITE_ONLY, store_format);
+ glDispatchCompute(num_dispatches_x, num_dispatches_y, image.info.resources.layers);
+ }
+ program_manager.RestoreGuestCompute();
+}
+
+void UtilShaders::BlockLinearUpload3D(Image& image, const ImageBufferMap& map, size_t buffer_offset,
+ std::span<const SwizzleParameters> swizzles) {
+ static constexpr Extent3D WORKGROUP_SIZE{16, 8, 8};
+
+ static constexpr GLuint BINDING_SWIZZLE_BUFFER = 0;
+ static constexpr GLuint BINDING_INPUT_BUFFER = 1;
+ static constexpr GLuint BINDING_OUTPUT_IMAGE = 0;
+
+ glFlushMappedNamedBufferRange(map.Handle(), buffer_offset, image.guest_size_bytes);
+ program_manager.BindHostCompute(block_linear_unswizzle_3d_program.handle);
+ glBindBufferBase(GL_SHADER_STORAGE_BUFFER, BINDING_SWIZZLE_BUFFER, swizzle_table_buffer.handle);
+
+ const GLenum store_format = StoreFormat(BytesPerBlock(image.info.format));
+ for (const SwizzleParameters& swizzle : swizzles) {
+ const Extent3D num_tiles = swizzle.num_tiles;
+ const size_t input_offset = swizzle.buffer_offset + buffer_offset;
+
+ const u32 num_dispatches_x = Common::DivCeil(num_tiles.width, WORKGROUP_SIZE.width);
+ const u32 num_dispatches_y = Common::DivCeil(num_tiles.height, WORKGROUP_SIZE.height);
+ const u32 num_dispatches_z = Common::DivCeil(num_tiles.depth, WORKGROUP_SIZE.depth);
+
+ const auto params = MakeBlockLinearSwizzle3DParams(swizzle, image.info);
+ glUniform3uiv(0, 1, params.origin.data());
+ glUniform3iv(1, 1, params.destination.data());
+ glUniform1ui(2, params.bytes_per_block_log2);
+ glUniform1ui(3, params.slice_size);
+ glUniform1ui(4, params.block_size);
+ glUniform1ui(5, params.x_shift);
+ glUniform1ui(6, params.block_height);
+ glUniform1ui(7, params.block_height_mask);
+ glUniform1ui(8, params.block_depth);
+ glUniform1ui(9, params.block_depth_mask);
+ glBindBufferRange(GL_SHADER_STORAGE_BUFFER, BINDING_INPUT_BUFFER, map.Handle(),
+ input_offset, image.guest_size_bytes - swizzle.buffer_offset);
+ glBindImageTexture(BINDING_OUTPUT_IMAGE, image.Handle(), swizzle.level, GL_TRUE, 0,
+ GL_WRITE_ONLY, store_format);
+ glDispatchCompute(num_dispatches_x, num_dispatches_y, num_dispatches_z);
+ }
+ program_manager.RestoreGuestCompute();
+}
+
+void UtilShaders::PitchUpload(Image& image, const ImageBufferMap& map, size_t buffer_offset,
+ std::span<const SwizzleParameters> swizzles) {
+ static constexpr Extent3D WORKGROUP_SIZE{32, 32, 1};
+ static constexpr GLuint BINDING_INPUT_BUFFER = 0;
+ static constexpr GLuint BINDING_OUTPUT_IMAGE = 0;
+ static constexpr GLuint LOC_ORIGIN = 0;
+ static constexpr GLuint LOC_DESTINATION = 1;
+ static constexpr GLuint LOC_BYTES_PER_BLOCK = 2;
+ static constexpr GLuint LOC_PITCH = 3;
+
+ const u32 bytes_per_block = BytesPerBlock(image.info.format);
+ const GLenum format = StoreFormat(bytes_per_block);
+ const u32 pitch = image.info.pitch;
+
+ UNIMPLEMENTED_IF_MSG(!std::has_single_bit(bytes_per_block),
+ "Non-power of two images are not implemented");
+
+ program_manager.BindHostCompute(pitch_unswizzle_program.handle);
+ glFlushMappedNamedBufferRange(map.Handle(), buffer_offset, image.guest_size_bytes);
+ glUniform2ui(LOC_ORIGIN, 0, 0);
+ glUniform2i(LOC_DESTINATION, 0, 0);
+ glUniform1ui(LOC_BYTES_PER_BLOCK, bytes_per_block);
+ glUniform1ui(LOC_PITCH, pitch);
+ glBindImageTexture(BINDING_OUTPUT_IMAGE, image.Handle(), 0, GL_FALSE, 0, GL_WRITE_ONLY, format);
+ for (const SwizzleParameters& swizzle : swizzles) {
+ const Extent3D num_tiles = swizzle.num_tiles;
+ const size_t input_offset = swizzle.buffer_offset + buffer_offset;
+
+ const u32 num_dispatches_x = Common::DivCeil(num_tiles.width, WORKGROUP_SIZE.width);
+ const u32 num_dispatches_y = Common::DivCeil(num_tiles.height, WORKGROUP_SIZE.height);
+
+ glBindBufferRange(GL_SHADER_STORAGE_BUFFER, BINDING_INPUT_BUFFER, map.Handle(),
+ input_offset, image.guest_size_bytes - swizzle.buffer_offset);
+ glDispatchCompute(num_dispatches_x, num_dispatches_y, 1);
+ }
+ program_manager.RestoreGuestCompute();
+}
+
+void UtilShaders::CopyBC4(Image& dst_image, Image& src_image, std::span<const ImageCopy> copies) {
+ static constexpr GLuint BINDING_INPUT_IMAGE = 0;
+ static constexpr GLuint BINDING_OUTPUT_IMAGE = 1;
+ static constexpr GLuint LOC_SRC_OFFSET = 0;
+ static constexpr GLuint LOC_DST_OFFSET = 1;
+
+ program_manager.BindHostCompute(copy_bc4_program.handle);
+
+ for (const ImageCopy& copy : copies) {
+ ASSERT(copy.src_subresource.base_layer == 0);
+ ASSERT(copy.src_subresource.num_layers == 1);
+ ASSERT(copy.dst_subresource.base_layer == 0);
+ ASSERT(copy.dst_subresource.num_layers == 1);
+
+ glUniform3ui(LOC_SRC_OFFSET, copy.src_offset.x, copy.src_offset.y, copy.src_offset.z);
+ glUniform3ui(LOC_DST_OFFSET, copy.dst_offset.x, copy.dst_offset.y, copy.dst_offset.z);
+ glBindImageTexture(BINDING_INPUT_IMAGE, src_image.Handle(), copy.src_subresource.base_level,
+ GL_FALSE, 0, GL_READ_ONLY, GL_RG32UI);
+ glBindImageTexture(BINDING_OUTPUT_IMAGE, dst_image.Handle(),
+ copy.dst_subresource.base_level, GL_FALSE, 0, GL_WRITE_ONLY, GL_RGBA8UI);
+ glDispatchCompute(copy.extent.width, copy.extent.height, copy.extent.depth);
+ }
+ program_manager.RestoreGuestCompute();
+}
+
+GLenum StoreFormat(u32 bytes_per_block) {
+ switch (bytes_per_block) {
+ case 1:
+ return GL_R8UI;
+ case 2:
+ return GL_R16UI;
+ case 4:
+ return GL_R32UI;
+ case 8:
+ return GL_RG32UI;
+ case 16:
+ return GL_RGBA32UI;
+ }
+ UNREACHABLE();
+ return GL_R8UI;
+}
+
+} // namespace OpenGL
diff --git a/src/video_core/renderer_opengl/util_shaders.h b/src/video_core/renderer_opengl/util_shaders.h
new file mode 100644
index 000000000..359997255
--- /dev/null
+++ b/src/video_core/renderer_opengl/util_shaders.h
@@ -0,0 +1,51 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <span>
+
+#include <glad/glad.h>
+
+#include "common/common_types.h"
+#include "video_core/renderer_opengl/gl_resource_manager.h"
+#include "video_core/texture_cache/types.h"
+
+namespace OpenGL {
+
+class Image;
+class ImageBufferMap;
+class ProgramManager;
+
+class UtilShaders {
+public:
+ explicit UtilShaders(ProgramManager& program_manager);
+ ~UtilShaders();
+
+ void BlockLinearUpload2D(Image& image, const ImageBufferMap& map, size_t buffer_offset,
+ std::span<const VideoCommon::SwizzleParameters> swizzles);
+
+ void BlockLinearUpload3D(Image& image, const ImageBufferMap& map, size_t buffer_offset,
+ std::span<const VideoCommon::SwizzleParameters> swizzles);
+
+ void PitchUpload(Image& image, const ImageBufferMap& map, size_t buffer_offset,
+ std::span<const VideoCommon::SwizzleParameters> swizzles);
+
+ void CopyBC4(Image& dst_image, Image& src_image,
+ std::span<const VideoCommon::ImageCopy> copies);
+
+private:
+ ProgramManager& program_manager;
+
+ OGLBuffer swizzle_table_buffer;
+
+ OGLProgram block_linear_unswizzle_2d_program;
+ OGLProgram block_linear_unswizzle_3d_program;
+ OGLProgram pitch_unswizzle_program;
+ OGLProgram copy_bc4_program;
+};
+
+GLenum StoreFormat(u32 bytes_per_block);
+
+} // namespace OpenGL
diff --git a/src/video_core/renderer_opengl/utils.cpp b/src/video_core/renderer_opengl/utils.cpp
deleted file mode 100644
index 6d7bb16b2..000000000
--- a/src/video_core/renderer_opengl/utils.cpp
+++ /dev/null
@@ -1,42 +0,0 @@
-// Copyright 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include <string>
-#include <vector>
-
-#include <fmt/format.h>
-#include <glad/glad.h>
-
-#include "common/common_types.h"
-#include "video_core/renderer_opengl/gl_state_tracker.h"
-#include "video_core/renderer_opengl/utils.h"
-
-namespace OpenGL {
-
-void LabelGLObject(GLenum identifier, GLuint handle, VAddr addr, std::string_view extra_info) {
- if (!GLAD_GL_KHR_debug) {
- // We don't need to throw an error as this is just for debugging
- return;
- }
-
- std::string object_label;
- if (extra_info.empty()) {
- switch (identifier) {
- case GL_TEXTURE:
- object_label = fmt::format("Texture@0x{:016X}", addr);
- break;
- case GL_PROGRAM:
- object_label = fmt::format("Shader@0x{:016X}", addr);
- break;
- default:
- object_label = fmt::format("Object(0x{:X})@0x{:016X}", identifier, addr);
- break;
- }
- } else {
- object_label = fmt::format("{}@0x{:016X}", extra_info, addr);
- }
- glObjectLabel(identifier, handle, -1, static_cast<const GLchar*>(object_label.c_str()));
-}
-
-} // namespace OpenGL
diff --git a/src/video_core/renderer_opengl/utils.h b/src/video_core/renderer_opengl/utils.h
deleted file mode 100644
index 9c09ee12c..000000000
--- a/src/video_core/renderer_opengl/utils.h
+++ /dev/null
@@ -1,16 +0,0 @@
-// Copyright 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <string_view>
-#include <vector>
-#include <glad/glad.h>
-#include "common/common_types.h"
-
-namespace OpenGL {
-
-void LabelGLObject(GLenum identifier, GLuint handle, VAddr addr, std::string_view extra_info = {});
-
-} // namespace OpenGL
diff --git a/src/video_core/renderer_vulkan/blit_image.cpp b/src/video_core/renderer_vulkan/blit_image.cpp
new file mode 100644
index 000000000..87c8e5693
--- /dev/null
+++ b/src/video_core/renderer_vulkan/blit_image.cpp
@@ -0,0 +1,624 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <algorithm>
+
+#include "video_core/host_shaders/convert_depth_to_float_frag_spv.h"
+#include "video_core/host_shaders/convert_float_to_depth_frag_spv.h"
+#include "video_core/host_shaders/full_screen_triangle_vert_spv.h"
+#include "video_core/host_shaders/vulkan_blit_color_float_frag_spv.h"
+#include "video_core/host_shaders/vulkan_blit_depth_stencil_frag_spv.h"
+#include "video_core/renderer_vulkan/blit_image.h"
+#include "video_core/renderer_vulkan/maxwell_to_vk.h"
+#include "video_core/renderer_vulkan/vk_device.h"
+#include "video_core/renderer_vulkan/vk_scheduler.h"
+#include "video_core/renderer_vulkan/vk_shader_util.h"
+#include "video_core/renderer_vulkan/vk_state_tracker.h"
+#include "video_core/renderer_vulkan/vk_texture_cache.h"
+#include "video_core/renderer_vulkan/vk_update_descriptor.h"
+#include "video_core/renderer_vulkan/wrapper.h"
+#include "video_core/surface.h"
+
+namespace Vulkan {
+
+using VideoCommon::ImageViewType;
+
+namespace {
+struct PushConstants {
+ std::array<float, 2> tex_scale;
+ std::array<float, 2> tex_offset;
+};
+
+template <u32 binding>
+inline constexpr VkDescriptorSetLayoutBinding TEXTURE_DESCRIPTOR_SET_LAYOUT_BINDING{
+ .binding = binding,
+ .descriptorType = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER,
+ .descriptorCount = 1,
+ .stageFlags = VK_SHADER_STAGE_FRAGMENT_BIT,
+ .pImmutableSamplers = nullptr,
+};
+constexpr std::array TWO_TEXTURES_DESCRIPTOR_SET_LAYOUT_BINDINGS{
+ TEXTURE_DESCRIPTOR_SET_LAYOUT_BINDING<0>,
+ TEXTURE_DESCRIPTOR_SET_LAYOUT_BINDING<1>,
+};
+constexpr VkDescriptorSetLayoutCreateInfo ONE_TEXTURE_DESCRIPTOR_SET_LAYOUT_CREATE_INFO{
+ .sType = VK_STRUCTURE_TYPE_DESCRIPTOR_SET_LAYOUT_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .bindingCount = 1,
+ .pBindings = &TEXTURE_DESCRIPTOR_SET_LAYOUT_BINDING<0>,
+};
+constexpr VkDescriptorSetLayoutCreateInfo TWO_TEXTURES_DESCRIPTOR_SET_LAYOUT_CREATE_INFO{
+ .sType = VK_STRUCTURE_TYPE_DESCRIPTOR_SET_LAYOUT_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .bindingCount = static_cast<u32>(TWO_TEXTURES_DESCRIPTOR_SET_LAYOUT_BINDINGS.size()),
+ .pBindings = TWO_TEXTURES_DESCRIPTOR_SET_LAYOUT_BINDINGS.data(),
+};
+constexpr VkPushConstantRange PUSH_CONSTANT_RANGE{
+ .stageFlags = VK_SHADER_STAGE_VERTEX_BIT,
+ .offset = 0,
+ .size = sizeof(PushConstants),
+};
+constexpr VkPipelineVertexInputStateCreateInfo PIPELINE_VERTEX_INPUT_STATE_CREATE_INFO{
+ .sType = VK_STRUCTURE_TYPE_PIPELINE_VERTEX_INPUT_STATE_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .vertexBindingDescriptionCount = 0,
+ .pVertexBindingDescriptions = nullptr,
+ .vertexAttributeDescriptionCount = 0,
+ .pVertexAttributeDescriptions = nullptr,
+};
+constexpr VkPipelineInputAssemblyStateCreateInfo PIPELINE_INPUT_ASSEMBLY_STATE_CREATE_INFO{
+ .sType = VK_STRUCTURE_TYPE_PIPELINE_INPUT_ASSEMBLY_STATE_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .topology = VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST,
+ .primitiveRestartEnable = VK_FALSE,
+};
+constexpr VkPipelineViewportStateCreateInfo PIPELINE_VIEWPORT_STATE_CREATE_INFO{
+ .sType = VK_STRUCTURE_TYPE_PIPELINE_VIEWPORT_STATE_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .viewportCount = 1,
+ .pViewports = nullptr,
+ .scissorCount = 1,
+ .pScissors = nullptr,
+};
+constexpr VkPipelineRasterizationStateCreateInfo PIPELINE_RASTERIZATION_STATE_CREATE_INFO{
+ .sType = VK_STRUCTURE_TYPE_PIPELINE_RASTERIZATION_STATE_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .depthClampEnable = VK_FALSE,
+ .rasterizerDiscardEnable = VK_FALSE,
+ .polygonMode = VK_POLYGON_MODE_FILL,
+ .cullMode = VK_CULL_MODE_BACK_BIT,
+ .frontFace = VK_FRONT_FACE_CLOCKWISE,
+ .depthBiasEnable = VK_FALSE,
+ .depthBiasConstantFactor = 0.0f,
+ .depthBiasClamp = 0.0f,
+ .depthBiasSlopeFactor = 0.0f,
+ .lineWidth = 1.0f,
+};
+constexpr VkPipelineMultisampleStateCreateInfo PIPELINE_MULTISAMPLE_STATE_CREATE_INFO{
+ .sType = VK_STRUCTURE_TYPE_PIPELINE_MULTISAMPLE_STATE_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .rasterizationSamples = VK_SAMPLE_COUNT_1_BIT,
+ .sampleShadingEnable = VK_FALSE,
+ .minSampleShading = 0.0f,
+ .pSampleMask = nullptr,
+ .alphaToCoverageEnable = VK_FALSE,
+ .alphaToOneEnable = VK_FALSE,
+};
+constexpr std::array DYNAMIC_STATES{
+ VK_DYNAMIC_STATE_VIEWPORT,
+ VK_DYNAMIC_STATE_SCISSOR,
+};
+constexpr VkPipelineDynamicStateCreateInfo PIPELINE_DYNAMIC_STATE_CREATE_INFO{
+ .sType = VK_STRUCTURE_TYPE_PIPELINE_DYNAMIC_STATE_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .dynamicStateCount = static_cast<u32>(DYNAMIC_STATES.size()),
+ .pDynamicStates = DYNAMIC_STATES.data(),
+};
+constexpr VkPipelineColorBlendStateCreateInfo PIPELINE_COLOR_BLEND_STATE_EMPTY_CREATE_INFO{
+ .sType = VK_STRUCTURE_TYPE_PIPELINE_COLOR_BLEND_STATE_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .logicOpEnable = VK_FALSE,
+ .logicOp = VK_LOGIC_OP_CLEAR,
+ .attachmentCount = 0,
+ .pAttachments = nullptr,
+ .blendConstants = {0.0f, 0.0f, 0.0f, 0.0f},
+};
+constexpr VkPipelineColorBlendAttachmentState PIPELINE_COLOR_BLEND_ATTACHMENT_STATE{
+ .blendEnable = VK_FALSE,
+ .srcColorBlendFactor = VK_BLEND_FACTOR_ZERO,
+ .dstColorBlendFactor = VK_BLEND_FACTOR_ZERO,
+ .colorBlendOp = VK_BLEND_OP_ADD,
+ .srcAlphaBlendFactor = VK_BLEND_FACTOR_ZERO,
+ .dstAlphaBlendFactor = VK_BLEND_FACTOR_ZERO,
+ .alphaBlendOp = VK_BLEND_OP_ADD,
+ .colorWriteMask = VK_COLOR_COMPONENT_R_BIT | VK_COLOR_COMPONENT_G_BIT |
+ VK_COLOR_COMPONENT_B_BIT | VK_COLOR_COMPONENT_A_BIT,
+};
+constexpr VkPipelineColorBlendStateCreateInfo PIPELINE_COLOR_BLEND_STATE_GENERIC_CREATE_INFO{
+ .sType = VK_STRUCTURE_TYPE_PIPELINE_COLOR_BLEND_STATE_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .logicOpEnable = VK_FALSE,
+ .logicOp = VK_LOGIC_OP_CLEAR,
+ .attachmentCount = 1,
+ .pAttachments = &PIPELINE_COLOR_BLEND_ATTACHMENT_STATE,
+ .blendConstants = {0.0f, 0.0f, 0.0f, 0.0f},
+};
+constexpr VkPipelineDepthStencilStateCreateInfo PIPELINE_DEPTH_STENCIL_STATE_CREATE_INFO{
+ .sType = VK_STRUCTURE_TYPE_PIPELINE_DEPTH_STENCIL_STATE_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .depthTestEnable = VK_TRUE,
+ .depthWriteEnable = VK_TRUE,
+ .depthCompareOp = VK_COMPARE_OP_ALWAYS,
+ .depthBoundsTestEnable = VK_FALSE,
+ .stencilTestEnable = VK_FALSE,
+ .front = VkStencilOpState{},
+ .back = VkStencilOpState{},
+ .minDepthBounds = 0.0f,
+ .maxDepthBounds = 0.0f,
+};
+
+template <VkFilter filter>
+inline constexpr VkSamplerCreateInfo SAMPLER_CREATE_INFO{
+ .sType = VK_STRUCTURE_TYPE_SAMPLER_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .magFilter = filter,
+ .minFilter = filter,
+ .mipmapMode = VK_SAMPLER_MIPMAP_MODE_NEAREST,
+ .addressModeU = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_BORDER,
+ .addressModeV = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_BORDER,
+ .addressModeW = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_BORDER,
+ .mipLodBias = 0.0f,
+ .anisotropyEnable = VK_FALSE,
+ .maxAnisotropy = 0.0f,
+ .compareEnable = VK_FALSE,
+ .compareOp = VK_COMPARE_OP_NEVER,
+ .minLod = 0.0f,
+ .maxLod = 0.0f,
+ .borderColor = VK_BORDER_COLOR_FLOAT_OPAQUE_WHITE,
+ .unnormalizedCoordinates = VK_TRUE,
+};
+
+constexpr VkPipelineLayoutCreateInfo PipelineLayoutCreateInfo(
+ const VkDescriptorSetLayout* set_layout) {
+ return VkPipelineLayoutCreateInfo{
+ .sType = VK_STRUCTURE_TYPE_PIPELINE_LAYOUT_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .setLayoutCount = 1,
+ .pSetLayouts = set_layout,
+ .pushConstantRangeCount = 1,
+ .pPushConstantRanges = &PUSH_CONSTANT_RANGE,
+ };
+}
+
+constexpr VkPipelineShaderStageCreateInfo PipelineShaderStageCreateInfo(VkShaderStageFlagBits stage,
+ VkShaderModule shader) {
+ return VkPipelineShaderStageCreateInfo{
+ .sType = VK_STRUCTURE_TYPE_PIPELINE_SHADER_STAGE_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .stage = stage,
+ .module = shader,
+ .pName = "main",
+ .pSpecializationInfo = nullptr,
+ };
+}
+
+constexpr std::array<VkPipelineShaderStageCreateInfo, 2> MakeStages(
+ VkShaderModule vertex_shader, VkShaderModule fragment_shader) {
+ return std::array{
+ PipelineShaderStageCreateInfo(VK_SHADER_STAGE_VERTEX_BIT, vertex_shader),
+ PipelineShaderStageCreateInfo(VK_SHADER_STAGE_FRAGMENT_BIT, fragment_shader),
+ };
+}
+
+void UpdateOneTextureDescriptorSet(const VKDevice& device, VkDescriptorSet descriptor_set,
+ VkSampler sampler, VkImageView image_view) {
+ const VkDescriptorImageInfo image_info{
+ .sampler = sampler,
+ .imageView = image_view,
+ .imageLayout = VK_IMAGE_LAYOUT_GENERAL,
+ };
+ const VkWriteDescriptorSet write_descriptor_set{
+ .sType = VK_STRUCTURE_TYPE_WRITE_DESCRIPTOR_SET,
+ .pNext = nullptr,
+ .dstSet = descriptor_set,
+ .dstBinding = 0,
+ .dstArrayElement = 0,
+ .descriptorCount = 1,
+ .descriptorType = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER,
+ .pImageInfo = &image_info,
+ .pBufferInfo = nullptr,
+ .pTexelBufferView = nullptr,
+ };
+ device.GetLogical().UpdateDescriptorSets(write_descriptor_set, nullptr);
+}
+
+void UpdateTwoTexturesDescriptorSet(const VKDevice& device, VkDescriptorSet descriptor_set,
+ VkSampler sampler, VkImageView image_view_0,
+ VkImageView image_view_1) {
+ const VkDescriptorImageInfo image_info_0{
+ .sampler = sampler,
+ .imageView = image_view_0,
+ .imageLayout = VK_IMAGE_LAYOUT_GENERAL,
+ };
+ const VkDescriptorImageInfo image_info_1{
+ .sampler = sampler,
+ .imageView = image_view_1,
+ .imageLayout = VK_IMAGE_LAYOUT_GENERAL,
+ };
+ const std::array write_descriptor_sets{
+ VkWriteDescriptorSet{
+ .sType = VK_STRUCTURE_TYPE_WRITE_DESCRIPTOR_SET,
+ .pNext = nullptr,
+ .dstSet = descriptor_set,
+ .dstBinding = 0,
+ .dstArrayElement = 0,
+ .descriptorCount = 1,
+ .descriptorType = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER,
+ .pImageInfo = &image_info_0,
+ .pBufferInfo = nullptr,
+ .pTexelBufferView = nullptr,
+ },
+ VkWriteDescriptorSet{
+ .sType = VK_STRUCTURE_TYPE_WRITE_DESCRIPTOR_SET,
+ .pNext = nullptr,
+ .dstSet = descriptor_set,
+ .dstBinding = 1,
+ .dstArrayElement = 0,
+ .descriptorCount = 1,
+ .descriptorType = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER,
+ .pImageInfo = &image_info_1,
+ .pBufferInfo = nullptr,
+ .pTexelBufferView = nullptr,
+ },
+ };
+ device.GetLogical().UpdateDescriptorSets(write_descriptor_sets, nullptr);
+}
+
+void BindBlitState(vk::CommandBuffer cmdbuf, VkPipelineLayout layout,
+ const std::array<Offset2D, 2>& dst_region,
+ const std::array<Offset2D, 2>& src_region) {
+ const VkOffset2D offset{
+ .x = std::min(dst_region[0].x, dst_region[1].x),
+ .y = std::min(dst_region[0].y, dst_region[1].y),
+ };
+ const VkExtent2D extent{
+ .width = static_cast<u32>(std::abs(dst_region[1].x - dst_region[0].x)),
+ .height = static_cast<u32>(std::abs(dst_region[1].y - dst_region[0].y)),
+ };
+ const VkViewport viewport{
+ .x = static_cast<float>(offset.x),
+ .y = static_cast<float>(offset.y),
+ .width = static_cast<float>(extent.width),
+ .height = static_cast<float>(extent.height),
+ .minDepth = 0.0f,
+ .maxDepth = 1.0f,
+ };
+ // TODO: Support scissored blits
+ const VkRect2D scissor{
+ .offset = offset,
+ .extent = extent,
+ };
+ const float scale_x = static_cast<float>(src_region[1].x - src_region[0].x);
+ const float scale_y = static_cast<float>(src_region[1].y - src_region[0].y);
+ const PushConstants push_constants{
+ .tex_scale = {scale_x, scale_y},
+ .tex_offset = {static_cast<float>(src_region[0].x), static_cast<float>(src_region[0].y)},
+ };
+ cmdbuf.SetViewport(0, viewport);
+ cmdbuf.SetScissor(0, scissor);
+ cmdbuf.PushConstants(layout, VK_SHADER_STAGE_VERTEX_BIT, push_constants);
+}
+
+} // Anonymous namespace
+
+BlitImageHelper::BlitImageHelper(const VKDevice& device_, VKScheduler& scheduler_,
+ StateTracker& state_tracker_, VKDescriptorPool& descriptor_pool)
+ : device{device_}, scheduler{scheduler_}, state_tracker{state_tracker_},
+ one_texture_set_layout(device.GetLogical().CreateDescriptorSetLayout(
+ ONE_TEXTURE_DESCRIPTOR_SET_LAYOUT_CREATE_INFO)),
+ two_textures_set_layout(device.GetLogical().CreateDescriptorSetLayout(
+ TWO_TEXTURES_DESCRIPTOR_SET_LAYOUT_CREATE_INFO)),
+ one_texture_descriptor_allocator(descriptor_pool, *one_texture_set_layout),
+ two_textures_descriptor_allocator(descriptor_pool, *two_textures_set_layout),
+ one_texture_pipeline_layout(device.GetLogical().CreatePipelineLayout(
+ PipelineLayoutCreateInfo(one_texture_set_layout.address()))),
+ two_textures_pipeline_layout(device.GetLogical().CreatePipelineLayout(
+ PipelineLayoutCreateInfo(two_textures_set_layout.address()))),
+ full_screen_vert(BuildShader(device, FULL_SCREEN_TRIANGLE_VERT_SPV)),
+ blit_color_to_color_frag(BuildShader(device, VULKAN_BLIT_COLOR_FLOAT_FRAG_SPV)),
+ convert_depth_to_float_frag(BuildShader(device, CONVERT_DEPTH_TO_FLOAT_FRAG_SPV)),
+ convert_float_to_depth_frag(BuildShader(device, CONVERT_FLOAT_TO_DEPTH_FRAG_SPV)),
+ linear_sampler(device.GetLogical().CreateSampler(SAMPLER_CREATE_INFO<VK_FILTER_LINEAR>)),
+ nearest_sampler(device.GetLogical().CreateSampler(SAMPLER_CREATE_INFO<VK_FILTER_NEAREST>)) {
+ if (device.IsExtShaderStencilExportSupported()) {
+ blit_depth_stencil_frag = BuildShader(device, VULKAN_BLIT_DEPTH_STENCIL_FRAG_SPV);
+ }
+}
+
+BlitImageHelper::~BlitImageHelper() = default;
+
+void BlitImageHelper::BlitColor(const Framebuffer* dst_framebuffer, const ImageView& src_image_view,
+ const std::array<Offset2D, 2>& dst_region,
+ const std::array<Offset2D, 2>& src_region,
+ Tegra::Engines::Fermi2D::Filter filter,
+ Tegra::Engines::Fermi2D::Operation operation) {
+ const bool is_linear = filter == Tegra::Engines::Fermi2D::Filter::Bilinear;
+ const BlitImagePipelineKey key{
+ .renderpass = dst_framebuffer->RenderPass(),
+ .operation = operation,
+ };
+ const VkPipelineLayout layout = *one_texture_pipeline_layout;
+ const VkImageView src_view = src_image_view.Handle(ImageViewType::e2D);
+ const VkSampler sampler = is_linear ? *linear_sampler : *nearest_sampler;
+ const VkPipeline pipeline = FindOrEmplacePipeline(key);
+ const VkDescriptorSet descriptor_set = one_texture_descriptor_allocator.Commit();
+ scheduler.RequestRenderpass(dst_framebuffer);
+ scheduler.Record([dst_region, src_region, pipeline, layout, sampler, src_view, descriptor_set,
+ &device = device](vk::CommandBuffer cmdbuf) {
+ // TODO: Barriers
+ UpdateOneTextureDescriptorSet(device, descriptor_set, sampler, src_view);
+ cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, pipeline);
+ cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_GRAPHICS, layout, 0, descriptor_set,
+ nullptr);
+ BindBlitState(cmdbuf, layout, dst_region, src_region);
+ cmdbuf.Draw(3, 1, 0, 0);
+ });
+ scheduler.InvalidateState();
+}
+
+void BlitImageHelper::BlitDepthStencil(const Framebuffer* dst_framebuffer,
+ VkImageView src_depth_view, VkImageView src_stencil_view,
+ const std::array<Offset2D, 2>& dst_region,
+ const std::array<Offset2D, 2>& src_region,
+ Tegra::Engines::Fermi2D::Filter filter,
+ Tegra::Engines::Fermi2D::Operation operation) {
+ ASSERT(filter == Tegra::Engines::Fermi2D::Filter::Point);
+ ASSERT(operation == Tegra::Engines::Fermi2D::Operation::SrcCopy);
+
+ const VkPipelineLayout layout = *two_textures_pipeline_layout;
+ const VkSampler sampler = *nearest_sampler;
+ const VkPipeline pipeline = BlitDepthStencilPipeline(dst_framebuffer->RenderPass());
+ const VkDescriptorSet descriptor_set = two_textures_descriptor_allocator.Commit();
+ scheduler.RequestRenderpass(dst_framebuffer);
+ scheduler.Record([dst_region, src_region, pipeline, layout, sampler, src_depth_view,
+ src_stencil_view, descriptor_set,
+ &device = device](vk::CommandBuffer cmdbuf) {
+ // TODO: Barriers
+ UpdateTwoTexturesDescriptorSet(device, descriptor_set, sampler, src_depth_view,
+ src_stencil_view);
+ cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, pipeline);
+ cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_GRAPHICS, layout, 0, descriptor_set,
+ nullptr);
+ BindBlitState(cmdbuf, layout, dst_region, src_region);
+ cmdbuf.Draw(3, 1, 0, 0);
+ });
+ scheduler.InvalidateState();
+}
+
+void BlitImageHelper::ConvertD32ToR32(const Framebuffer* dst_framebuffer,
+ const ImageView& src_image_view) {
+ ConvertDepthToColorPipeline(convert_d32_to_r32_pipeline, dst_framebuffer->RenderPass());
+ Convert(*convert_d32_to_r32_pipeline, dst_framebuffer, src_image_view);
+}
+
+void BlitImageHelper::ConvertR32ToD32(const Framebuffer* dst_framebuffer,
+ const ImageView& src_image_view) {
+
+ ConvertColorToDepthPipeline(convert_r32_to_d32_pipeline, dst_framebuffer->RenderPass());
+ Convert(*convert_r32_to_d32_pipeline, dst_framebuffer, src_image_view);
+}
+
+void BlitImageHelper::ConvertD16ToR16(const Framebuffer* dst_framebuffer,
+ const ImageView& src_image_view) {
+ ConvertDepthToColorPipeline(convert_d16_to_r16_pipeline, dst_framebuffer->RenderPass());
+ Convert(*convert_d16_to_r16_pipeline, dst_framebuffer, src_image_view);
+}
+
+void BlitImageHelper::ConvertR16ToD16(const Framebuffer* dst_framebuffer,
+ const ImageView& src_image_view) {
+ ConvertColorToDepthPipeline(convert_r16_to_d16_pipeline, dst_framebuffer->RenderPass());
+ Convert(*convert_r16_to_d16_pipeline, dst_framebuffer, src_image_view);
+}
+
+void BlitImageHelper::Convert(VkPipeline pipeline, const Framebuffer* dst_framebuffer,
+ const ImageView& src_image_view) {
+ const VkPipelineLayout layout = *one_texture_pipeline_layout;
+ const VkImageView src_view = src_image_view.Handle(ImageViewType::e2D);
+ const VkSampler sampler = *nearest_sampler;
+ const VkDescriptorSet descriptor_set = one_texture_descriptor_allocator.Commit();
+ const VkExtent2D extent{
+ .width = src_image_view.size.width,
+ .height = src_image_view.size.height,
+ };
+ scheduler.RequestRenderpass(dst_framebuffer);
+ scheduler.Record([pipeline, layout, sampler, src_view, descriptor_set, extent,
+ &device = device](vk::CommandBuffer cmdbuf) {
+ const VkOffset2D offset{
+ .x = 0,
+ .y = 0,
+ };
+ const VkViewport viewport{
+ .x = 0.0f,
+ .y = 0.0f,
+ .width = static_cast<float>(extent.width),
+ .height = static_cast<float>(extent.height),
+ .minDepth = 0.0f,
+ .maxDepth = 0.0f,
+ };
+ const VkRect2D scissor{
+ .offset = offset,
+ .extent = extent,
+ };
+ const PushConstants push_constants{
+ .tex_scale = {viewport.width, viewport.height},
+ .tex_offset = {0.0f, 0.0f},
+ };
+ UpdateOneTextureDescriptorSet(device, descriptor_set, sampler, src_view);
+
+ // TODO: Barriers
+ cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, pipeline);
+ cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_GRAPHICS, layout, 0, descriptor_set,
+ nullptr);
+ cmdbuf.SetViewport(0, viewport);
+ cmdbuf.SetScissor(0, scissor);
+ cmdbuf.PushConstants(layout, VK_SHADER_STAGE_VERTEX_BIT, push_constants);
+ cmdbuf.Draw(3, 1, 0, 0);
+ });
+ scheduler.InvalidateState();
+}
+
+VkPipeline BlitImageHelper::FindOrEmplacePipeline(const BlitImagePipelineKey& key) {
+ const auto it = std::ranges::find(blit_color_keys, key);
+ if (it != blit_color_keys.end()) {
+ return *blit_color_pipelines[std::distance(blit_color_keys.begin(), it)];
+ }
+ blit_color_keys.push_back(key);
+
+ const std::array stages = MakeStages(*full_screen_vert, *blit_color_to_color_frag);
+ const VkPipelineColorBlendAttachmentState blend_attachment{
+ .blendEnable = VK_FALSE,
+ .srcColorBlendFactor = VK_BLEND_FACTOR_ZERO,
+ .dstColorBlendFactor = VK_BLEND_FACTOR_ZERO,
+ .colorBlendOp = VK_BLEND_OP_ADD,
+ .srcAlphaBlendFactor = VK_BLEND_FACTOR_ZERO,
+ .dstAlphaBlendFactor = VK_BLEND_FACTOR_ZERO,
+ .alphaBlendOp = VK_BLEND_OP_ADD,
+ .colorWriteMask = VK_COLOR_COMPONENT_R_BIT | VK_COLOR_COMPONENT_G_BIT |
+ VK_COLOR_COMPONENT_B_BIT | VK_COLOR_COMPONENT_A_BIT,
+ };
+ // TODO: programmable blending
+ const VkPipelineColorBlendStateCreateInfo color_blend_create_info{
+ .sType = VK_STRUCTURE_TYPE_PIPELINE_COLOR_BLEND_STATE_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .logicOpEnable = VK_FALSE,
+ .logicOp = VK_LOGIC_OP_CLEAR,
+ .attachmentCount = 1,
+ .pAttachments = &blend_attachment,
+ .blendConstants = {0.0f, 0.0f, 0.0f, 0.0f},
+ };
+ blit_color_pipelines.push_back(device.GetLogical().CreateGraphicsPipeline({
+ .sType = VK_STRUCTURE_TYPE_GRAPHICS_PIPELINE_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .stageCount = static_cast<u32>(stages.size()),
+ .pStages = stages.data(),
+ .pVertexInputState = &PIPELINE_VERTEX_INPUT_STATE_CREATE_INFO,
+ .pInputAssemblyState = &PIPELINE_INPUT_ASSEMBLY_STATE_CREATE_INFO,
+ .pTessellationState = nullptr,
+ .pViewportState = &PIPELINE_VIEWPORT_STATE_CREATE_INFO,
+ .pRasterizationState = &PIPELINE_RASTERIZATION_STATE_CREATE_INFO,
+ .pMultisampleState = &PIPELINE_MULTISAMPLE_STATE_CREATE_INFO,
+ .pDepthStencilState = nullptr,
+ .pColorBlendState = &color_blend_create_info,
+ .pDynamicState = &PIPELINE_DYNAMIC_STATE_CREATE_INFO,
+ .layout = *one_texture_pipeline_layout,
+ .renderPass = key.renderpass,
+ .subpass = 0,
+ .basePipelineHandle = VK_NULL_HANDLE,
+ .basePipelineIndex = 0,
+ }));
+ return *blit_color_pipelines.back();
+}
+
+VkPipeline BlitImageHelper::BlitDepthStencilPipeline(VkRenderPass renderpass) {
+ if (blit_depth_stencil_pipeline) {
+ return *blit_depth_stencil_pipeline;
+ }
+ const std::array stages = MakeStages(*full_screen_vert, *blit_depth_stencil_frag);
+ blit_depth_stencil_pipeline = device.GetLogical().CreateGraphicsPipeline({
+ .sType = VK_STRUCTURE_TYPE_GRAPHICS_PIPELINE_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .stageCount = static_cast<u32>(stages.size()),
+ .pStages = stages.data(),
+ .pVertexInputState = &PIPELINE_VERTEX_INPUT_STATE_CREATE_INFO,
+ .pInputAssemblyState = &PIPELINE_INPUT_ASSEMBLY_STATE_CREATE_INFO,
+ .pTessellationState = nullptr,
+ .pViewportState = &PIPELINE_VIEWPORT_STATE_CREATE_INFO,
+ .pRasterizationState = &PIPELINE_RASTERIZATION_STATE_CREATE_INFO,
+ .pMultisampleState = &PIPELINE_MULTISAMPLE_STATE_CREATE_INFO,
+ .pDepthStencilState = &PIPELINE_DEPTH_STENCIL_STATE_CREATE_INFO,
+ .pColorBlendState = &PIPELINE_COLOR_BLEND_STATE_EMPTY_CREATE_INFO,
+ .pDynamicState = &PIPELINE_DYNAMIC_STATE_CREATE_INFO,
+ .layout = *two_textures_pipeline_layout,
+ .renderPass = renderpass,
+ .subpass = 0,
+ .basePipelineHandle = VK_NULL_HANDLE,
+ .basePipelineIndex = 0,
+ });
+ return *blit_depth_stencil_pipeline;
+}
+
+void BlitImageHelper::ConvertDepthToColorPipeline(vk::Pipeline& pipeline, VkRenderPass renderpass) {
+ if (pipeline) {
+ return;
+ }
+ const std::array stages = MakeStages(*full_screen_vert, *convert_depth_to_float_frag);
+ pipeline = device.GetLogical().CreateGraphicsPipeline({
+ .sType = VK_STRUCTURE_TYPE_GRAPHICS_PIPELINE_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .stageCount = static_cast<u32>(stages.size()),
+ .pStages = stages.data(),
+ .pVertexInputState = &PIPELINE_VERTEX_INPUT_STATE_CREATE_INFO,
+ .pInputAssemblyState = &PIPELINE_INPUT_ASSEMBLY_STATE_CREATE_INFO,
+ .pTessellationState = nullptr,
+ .pViewportState = &PIPELINE_VIEWPORT_STATE_CREATE_INFO,
+ .pRasterizationState = &PIPELINE_RASTERIZATION_STATE_CREATE_INFO,
+ .pMultisampleState = &PIPELINE_MULTISAMPLE_STATE_CREATE_INFO,
+ .pDepthStencilState = nullptr,
+ .pColorBlendState = &PIPELINE_COLOR_BLEND_STATE_GENERIC_CREATE_INFO,
+ .pDynamicState = &PIPELINE_DYNAMIC_STATE_CREATE_INFO,
+ .layout = *one_texture_pipeline_layout,
+ .renderPass = renderpass,
+ .subpass = 0,
+ .basePipelineHandle = VK_NULL_HANDLE,
+ .basePipelineIndex = 0,
+ });
+}
+
+void BlitImageHelper::ConvertColorToDepthPipeline(vk::Pipeline& pipeline, VkRenderPass renderpass) {
+ if (pipeline) {
+ return;
+ }
+ const std::array stages = MakeStages(*full_screen_vert, *convert_float_to_depth_frag);
+ pipeline = device.GetLogical().CreateGraphicsPipeline({
+ .sType = VK_STRUCTURE_TYPE_GRAPHICS_PIPELINE_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .stageCount = static_cast<u32>(stages.size()),
+ .pStages = stages.data(),
+ .pVertexInputState = &PIPELINE_VERTEX_INPUT_STATE_CREATE_INFO,
+ .pInputAssemblyState = &PIPELINE_INPUT_ASSEMBLY_STATE_CREATE_INFO,
+ .pTessellationState = nullptr,
+ .pViewportState = &PIPELINE_VIEWPORT_STATE_CREATE_INFO,
+ .pRasterizationState = &PIPELINE_RASTERIZATION_STATE_CREATE_INFO,
+ .pMultisampleState = &PIPELINE_MULTISAMPLE_STATE_CREATE_INFO,
+ .pDepthStencilState = &PIPELINE_DEPTH_STENCIL_STATE_CREATE_INFO,
+ .pColorBlendState = &PIPELINE_COLOR_BLEND_STATE_EMPTY_CREATE_INFO,
+ .pDynamicState = &PIPELINE_DYNAMIC_STATE_CREATE_INFO,
+ .layout = *one_texture_pipeline_layout,
+ .renderPass = renderpass,
+ .subpass = 0,
+ .basePipelineHandle = VK_NULL_HANDLE,
+ .basePipelineIndex = 0,
+ });
+}
+
+} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/blit_image.h b/src/video_core/renderer_vulkan/blit_image.h
new file mode 100644
index 000000000..2c2790bf9
--- /dev/null
+++ b/src/video_core/renderer_vulkan/blit_image.h
@@ -0,0 +1,97 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <compare>
+
+#include "video_core/engines/fermi_2d.h"
+#include "video_core/renderer_vulkan/vk_descriptor_pool.h"
+#include "video_core/renderer_vulkan/wrapper.h"
+#include "video_core/texture_cache/types.h"
+
+namespace Vulkan {
+
+using VideoCommon::Offset2D;
+
+class VKDevice;
+class VKScheduler;
+class StateTracker;
+
+class Framebuffer;
+class ImageView;
+
+struct BlitImagePipelineKey {
+ constexpr auto operator<=>(const BlitImagePipelineKey&) const noexcept = default;
+
+ VkRenderPass renderpass;
+ Tegra::Engines::Fermi2D::Operation operation;
+};
+
+class BlitImageHelper {
+public:
+ explicit BlitImageHelper(const VKDevice& device, VKScheduler& scheduler,
+ StateTracker& state_tracker, VKDescriptorPool& descriptor_pool);
+ ~BlitImageHelper();
+
+ void BlitColor(const Framebuffer* dst_framebuffer, const ImageView& src_image_view,
+ const std::array<Offset2D, 2>& dst_region,
+ const std::array<Offset2D, 2>& src_region,
+ Tegra::Engines::Fermi2D::Filter filter,
+ Tegra::Engines::Fermi2D::Operation operation);
+
+ void BlitDepthStencil(const Framebuffer* dst_framebuffer, VkImageView src_depth_view,
+ VkImageView src_stencil_view, const std::array<Offset2D, 2>& dst_region,
+ const std::array<Offset2D, 2>& src_region,
+ Tegra::Engines::Fermi2D::Filter filter,
+ Tegra::Engines::Fermi2D::Operation operation);
+
+ void ConvertD32ToR32(const Framebuffer* dst_framebuffer, const ImageView& src_image_view);
+
+ void ConvertR32ToD32(const Framebuffer* dst_framebuffer, const ImageView& src_image_view);
+
+ void ConvertD16ToR16(const Framebuffer* dst_framebuffer, const ImageView& src_image_view);
+
+ void ConvertR16ToD16(const Framebuffer* dst_framebuffer, const ImageView& src_image_view);
+
+private:
+ void Convert(VkPipeline pipeline, const Framebuffer* dst_framebuffer,
+ const ImageView& src_image_view);
+
+ [[nodiscard]] VkPipeline FindOrEmplacePipeline(const BlitImagePipelineKey& key);
+
+ [[nodiscard]] VkPipeline BlitDepthStencilPipeline(VkRenderPass renderpass);
+
+ void ConvertDepthToColorPipeline(vk::Pipeline& pipeline, VkRenderPass renderpass);
+
+ void ConvertColorToDepthPipeline(vk::Pipeline& pipeline, VkRenderPass renderpass);
+
+ const VKDevice& device;
+ VKScheduler& scheduler;
+ StateTracker& state_tracker;
+
+ vk::DescriptorSetLayout one_texture_set_layout;
+ vk::DescriptorSetLayout two_textures_set_layout;
+ DescriptorAllocator one_texture_descriptor_allocator;
+ DescriptorAllocator two_textures_descriptor_allocator;
+ vk::PipelineLayout one_texture_pipeline_layout;
+ vk::PipelineLayout two_textures_pipeline_layout;
+ vk::ShaderModule full_screen_vert;
+ vk::ShaderModule blit_color_to_color_frag;
+ vk::ShaderModule blit_depth_stencil_frag;
+ vk::ShaderModule convert_depth_to_float_frag;
+ vk::ShaderModule convert_float_to_depth_frag;
+ vk::Sampler linear_sampler;
+ vk::Sampler nearest_sampler;
+
+ std::vector<BlitImagePipelineKey> blit_color_keys;
+ std::vector<vk::Pipeline> blit_color_pipelines;
+ vk::Pipeline blit_depth_stencil_pipeline;
+ vk::Pipeline convert_d32_to_r32_pipeline;
+ vk::Pipeline convert_r32_to_d32_pipeline;
+ vk::Pipeline convert_d16_to_r16_pipeline;
+ vk::Pipeline convert_r16_to_d16_pipeline;
+};
+
+} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/fixed_pipeline_state.cpp b/src/video_core/renderer_vulkan/fixed_pipeline_state.cpp
index 5ec43db11..67dd10500 100644
--- a/src/video_core/renderer_vulkan/fixed_pipeline_state.cpp
+++ b/src/video_core/renderer_vulkan/fixed_pipeline_state.cpp
@@ -60,6 +60,7 @@ void FixedPipelineState::Fill(const Maxwell& regs, bool has_extended_dynamic_sta
logic_op.Assign(PackLogicOp(regs.logic_op.operation));
rasterize_enable.Assign(regs.rasterize_enable != 0 ? 1 : 0);
topology.Assign(regs.draw.topology);
+ msaa_mode.Assign(regs.multisample_mode);
raw2 = 0;
const auto test_func =
diff --git a/src/video_core/renderer_vulkan/fixed_pipeline_state.h b/src/video_core/renderer_vulkan/fixed_pipeline_state.h
index c26b77790..7e95e6fce 100644
--- a/src/video_core/renderer_vulkan/fixed_pipeline_state.h
+++ b/src/video_core/renderer_vulkan/fixed_pipeline_state.h
@@ -186,6 +186,7 @@ struct FixedPipelineState {
BitField<19, 4, u32> logic_op;
BitField<23, 1, u32> rasterize_enable;
BitField<24, 4, Maxwell::PrimitiveTopology> topology;
+ BitField<28, 4, Tegra::Texture::MsaaMode> msaa_mode;
};
union {
u32 raw2;
diff --git a/src/video_core/renderer_vulkan/maxwell_to_vk.cpp b/src/video_core/renderer_vulkan/maxwell_to_vk.cpp
index d22de1d81..4c988429f 100644
--- a/src/video_core/renderer_vulkan/maxwell_to_vk.cpp
+++ b/src/video_core/renderer_vulkan/maxwell_to_vk.cpp
@@ -26,7 +26,7 @@ VkFilter Filter(Tegra::Texture::TextureFilter filter) {
case Tegra::Texture::TextureFilter::Linear:
return VK_FILTER_LINEAR;
}
- UNREACHABLE_MSG("Invalid sampler filter={}", static_cast<u32>(filter));
+ UNREACHABLE_MSG("Invalid sampler filter={}", filter);
return {};
}
@@ -43,7 +43,7 @@ VkSamplerMipmapMode MipmapMode(Tegra::Texture::TextureMipmapFilter mipmap_filter
case Tegra::Texture::TextureMipmapFilter::Linear:
return VK_SAMPLER_MIPMAP_MODE_LINEAR;
}
- UNREACHABLE_MSG("Invalid sampler mipmap mode={}", static_cast<u32>(mipmap_filter));
+ UNREACHABLE_MSG("Invalid sampler mipmap mode={}", mipmap_filter);
return {};
}
@@ -79,7 +79,7 @@ VkSamplerAddressMode WrapMode(const VKDevice& device, Tegra::Texture::WrapMode w
UNIMPLEMENTED();
return VK_SAMPLER_ADDRESS_MODE_MIRROR_CLAMP_TO_EDGE;
default:
- UNIMPLEMENTED_MSG("Unimplemented wrap mode={}", static_cast<u32>(wrap_mode));
+ UNIMPLEMENTED_MSG("Unimplemented wrap mode={}", wrap_mode);
return {};
}
}
@@ -103,8 +103,7 @@ VkCompareOp DepthCompareFunction(Tegra::Texture::DepthCompareFunc depth_compare_
case Tegra::Texture::DepthCompareFunc::Always:
return VK_COMPARE_OP_ALWAYS;
}
- UNIMPLEMENTED_MSG("Unimplemented sampler depth compare function={}",
- static_cast<u32>(depth_compare_func));
+ UNIMPLEMENTED_MSG("Unimplemented sampler depth compare function={}", depth_compare_func);
return {};
}
@@ -123,7 +122,7 @@ struct FormatTuple {
{VK_FORMAT_A8B8G8R8_SINT_PACK32, Attachable | Storage}, // A8B8G8R8_SINT
{VK_FORMAT_A8B8G8R8_UINT_PACK32, Attachable | Storage}, // A8B8G8R8_UINT
{VK_FORMAT_R5G6B5_UNORM_PACK16, Attachable}, // R5G6B5_UNORM
- {VK_FORMAT_B5G6R5_UNORM_PACK16, Attachable}, // B5G6R5_UNORM
+ {VK_FORMAT_B5G6R5_UNORM_PACK16}, // B5G6R5_UNORM
{VK_FORMAT_A1R5G5B5_UNORM_PACK16, Attachable}, // A1R5G5B5_UNORM
{VK_FORMAT_A2B10G10R10_UNORM_PACK32, Attachable | Storage}, // A2B10G10R10_UNORM
{VK_FORMAT_A2B10G10R10_UINT_PACK32, Attachable | Storage}, // A2B10G10R10_UINT
@@ -164,7 +163,7 @@ struct FormatTuple {
{VK_FORMAT_R16G16_UNORM, Attachable | Storage}, // R16G16_UNORM
{VK_FORMAT_R16G16_SFLOAT, Attachable | Storage}, // R16G16_FLOAT
{VK_FORMAT_UNDEFINED}, // R16G16_UINT
- {VK_FORMAT_UNDEFINED}, // R16G16_SINT
+ {VK_FORMAT_R16G16_SINT, Attachable | Storage}, // R16G16_SINT
{VK_FORMAT_R16G16_SNORM, Attachable | Storage}, // R16G16_SNORM
{VK_FORMAT_UNDEFINED}, // R32G32B32_FLOAT
{VK_FORMAT_R8G8B8A8_SRGB, Attachable}, // A8B8G8R8_SRGB
@@ -228,25 +227,26 @@ FormatInfo SurfaceFormat(const VKDevice& device, FormatType format_type, PixelFo
auto tuple = tex_format_tuples[static_cast<std::size_t>(pixel_format)];
if (tuple.format == VK_FORMAT_UNDEFINED) {
- UNIMPLEMENTED_MSG("Unimplemented texture format with pixel format={}",
- static_cast<u32>(pixel_format));
+ UNIMPLEMENTED_MSG("Unimplemented texture format with pixel format={}", pixel_format);
return {VK_FORMAT_A8B8G8R8_UNORM_PACK32, true, true};
}
// Use A8B8G8R8_UNORM on hardware that doesn't support ASTC natively
if (!device.IsOptimalAstcSupported() && VideoCore::Surface::IsPixelFormatASTC(pixel_format)) {
- tuple.format = VideoCore::Surface::IsPixelFormatSRGB(pixel_format)
- ? VK_FORMAT_A8B8G8R8_SRGB_PACK32
- : VK_FORMAT_A8B8G8R8_UNORM_PACK32;
+ const bool is_srgb = VideoCore::Surface::IsPixelFormatSRGB(pixel_format);
+ tuple.format = is_srgb ? VK_FORMAT_A8B8G8R8_SRGB_PACK32 : VK_FORMAT_A8B8G8R8_UNORM_PACK32;
}
const bool attachable = tuple.usage & Attachable;
const bool storage = tuple.usage & Storage;
- VkFormatFeatureFlags usage;
- if (format_type == FormatType::Buffer) {
+ VkFormatFeatureFlags usage{};
+ switch (format_type) {
+ case FormatType::Buffer:
usage =
VK_FORMAT_FEATURE_STORAGE_TEXEL_BUFFER_BIT | VK_FORMAT_FEATURE_UNIFORM_TEXEL_BUFFER_BIT;
- } else {
+ break;
+ case FormatType::Linear:
+ case FormatType::Optimal:
usage = VK_FORMAT_FEATURE_SAMPLED_IMAGE_BIT | VK_FORMAT_FEATURE_TRANSFER_DST_BIT |
VK_FORMAT_FEATURE_TRANSFER_SRC_BIT;
if (attachable) {
@@ -256,6 +256,7 @@ FormatInfo SurfaceFormat(const VKDevice& device, FormatType format_type, PixelFo
if (storage) {
usage |= VK_FORMAT_FEATURE_STORAGE_IMAGE_BIT;
}
+ break;
}
return {device.GetSupportedFormat(tuple.format, usage, format_type), attachable, storage};
}
@@ -275,7 +276,7 @@ VkShaderStageFlagBits ShaderStage(Tegra::Engines::ShaderType stage) {
case Tegra::Engines::ShaderType::Compute:
return VK_SHADER_STAGE_COMPUTE_BIT;
}
- UNIMPLEMENTED_MSG("Unimplemented shader stage={}", static_cast<u32>(stage));
+ UNIMPLEMENTED_MSG("Unimplemented shader stage={}", stage);
return {};
}
@@ -300,7 +301,7 @@ VkPrimitiveTopology PrimitiveTopology([[maybe_unused]] const VKDevice& device,
case Maxwell::PrimitiveTopology::Patches:
return VK_PRIMITIVE_TOPOLOGY_PATCH_LIST;
default:
- UNIMPLEMENTED_MSG("Unimplemented topology={}", static_cast<u32>(topology));
+ UNIMPLEMENTED_MSG("Unimplemented topology={}", topology);
return {};
}
}
@@ -490,8 +491,7 @@ VkFormat VertexFormat(Maxwell::VertexAttribute::Type type, Maxwell::VertexAttrib
}
break;
}
- UNIMPLEMENTED_MSG("Unimplemented vertex format of type={} and size={}", static_cast<u32>(type),
- static_cast<u32>(size));
+ UNIMPLEMENTED_MSG("Unimplemented vertex format of type={} and size={}", type, size);
return {};
}
@@ -522,7 +522,7 @@ VkCompareOp ComparisonOp(Maxwell::ComparisonOp comparison) {
case Maxwell::ComparisonOp::AlwaysOld:
return VK_COMPARE_OP_ALWAYS;
}
- UNIMPLEMENTED_MSG("Unimplemented comparison op={}", static_cast<u32>(comparison));
+ UNIMPLEMENTED_MSG("Unimplemented comparison op={}", comparison);
return {};
}
@@ -539,7 +539,7 @@ VkIndexType IndexFormat(const VKDevice& device, Maxwell::IndexFormat index_forma
case Maxwell::IndexFormat::UnsignedInt:
return VK_INDEX_TYPE_UINT32;
}
- UNIMPLEMENTED_MSG("Unimplemented index_format={}", static_cast<u32>(index_format));
+ UNIMPLEMENTED_MSG("Unimplemented index_format={}", index_format);
return {};
}
@@ -570,7 +570,7 @@ VkStencilOp StencilOp(Maxwell::StencilOp stencil_op) {
case Maxwell::StencilOp::DecrWrapOGL:
return VK_STENCIL_OP_DECREMENT_AND_WRAP;
}
- UNIMPLEMENTED_MSG("Unimplemented stencil op={}", static_cast<u32>(stencil_op));
+ UNIMPLEMENTED_MSG("Unimplemented stencil op={}", stencil_op);
return {};
}
@@ -592,7 +592,7 @@ VkBlendOp BlendEquation(Maxwell::Blend::Equation equation) {
case Maxwell::Blend::Equation::MaxGL:
return VK_BLEND_OP_MAX;
}
- UNIMPLEMENTED_MSG("Unimplemented blend equation={}", static_cast<u32>(equation));
+ UNIMPLEMENTED_MSG("Unimplemented blend equation={}", equation);
return {};
}
@@ -656,7 +656,7 @@ VkBlendFactor BlendFactor(Maxwell::Blend::Factor factor) {
case Maxwell::Blend::Factor::OneMinusConstantAlphaGL:
return VK_BLEND_FACTOR_ONE_MINUS_CONSTANT_ALPHA;
}
- UNIMPLEMENTED_MSG("Unimplemented blend factor={}", static_cast<u32>(factor));
+ UNIMPLEMENTED_MSG("Unimplemented blend factor={}", factor);
return {};
}
@@ -667,7 +667,7 @@ VkFrontFace FrontFace(Maxwell::FrontFace front_face) {
case Maxwell::FrontFace::CounterClockWise:
return VK_FRONT_FACE_COUNTER_CLOCKWISE;
}
- UNIMPLEMENTED_MSG("Unimplemented front face={}", static_cast<u32>(front_face));
+ UNIMPLEMENTED_MSG("Unimplemented front face={}", front_face);
return {};
}
@@ -680,7 +680,7 @@ VkCullModeFlags CullFace(Maxwell::CullFace cull_face) {
case Maxwell::CullFace::FrontAndBack:
return VK_CULL_MODE_FRONT_AND_BACK;
}
- UNIMPLEMENTED_MSG("Unimplemented cull face={}", static_cast<u32>(cull_face));
+ UNIMPLEMENTED_MSG("Unimplemented cull face={}", cull_face);
return {};
}
@@ -700,7 +700,7 @@ VkComponentSwizzle SwizzleSource(Tegra::Texture::SwizzleSource swizzle) {
case Tegra::Texture::SwizzleSource::OneFloat:
return VK_COMPONENT_SWIZZLE_ONE;
}
- UNIMPLEMENTED_MSG("Unimplemented swizzle source={}", static_cast<u32>(swizzle));
+ UNIMPLEMENTED_MSG("Unimplemented swizzle source={}", swizzle);
return {};
}
@@ -723,8 +723,21 @@ VkViewportCoordinateSwizzleNV ViewportSwizzle(Maxwell::ViewportSwizzle swizzle)
case Maxwell::ViewportSwizzle::NegativeW:
return VK_VIEWPORT_COORDINATE_SWIZZLE_NEGATIVE_W_NV;
}
- UNREACHABLE_MSG("Invalid swizzle={}", static_cast<int>(swizzle));
+ UNREACHABLE_MSG("Invalid swizzle={}", swizzle);
return {};
}
+VkSamplerReductionMode SamplerReduction(Tegra::Texture::SamplerReduction reduction) {
+ switch (reduction) {
+ case Tegra::Texture::SamplerReduction::WeightedAverage:
+ return VK_SAMPLER_REDUCTION_MODE_WEIGHTED_AVERAGE_EXT;
+ case Tegra::Texture::SamplerReduction::Min:
+ return VK_SAMPLER_REDUCTION_MODE_MIN_EXT;
+ case Tegra::Texture::SamplerReduction::Max:
+ return VK_SAMPLER_REDUCTION_MODE_MAX_EXT;
+ }
+ UNREACHABLE_MSG("Invalid sampler mode={}", static_cast<int>(reduction));
+ return VK_SAMPLER_REDUCTION_MODE_WEIGHTED_AVERAGE_EXT;
+}
+
} // namespace Vulkan::MaxwellToVK
diff --git a/src/video_core/renderer_vulkan/maxwell_to_vk.h b/src/video_core/renderer_vulkan/maxwell_to_vk.h
index 7e213452f..1a90f192e 100644
--- a/src/video_core/renderer_vulkan/maxwell_to_vk.h
+++ b/src/video_core/renderer_vulkan/maxwell_to_vk.h
@@ -61,4 +61,6 @@ VkComponentSwizzle SwizzleSource(Tegra::Texture::SwizzleSource swizzle);
VkViewportCoordinateSwizzleNV ViewportSwizzle(Maxwell::ViewportSwizzle swizzle);
+VkSamplerReductionMode SamplerReduction(Tegra::Texture::SamplerReduction reduction);
+
} // namespace Vulkan::MaxwellToVK
diff --git a/src/video_core/renderer_vulkan/renderer_vulkan.cpp b/src/video_core/renderer_vulkan/renderer_vulkan.cpp
index a2173edd2..7f521cb9b 100644
--- a/src/video_core/renderer_vulkan/renderer_vulkan.cpp
+++ b/src/video_core/renderer_vulkan/renderer_vulkan.cpp
@@ -92,9 +92,9 @@ Common::DynamicLibrary OpenVulkanLibrary() {
return library;
}
-std::pair<vk::Instance, u32> CreateInstance(
- Common::DynamicLibrary& library, vk::InstanceDispatch& dld,
- WindowSystemType window_type = WindowSystemType::Headless, bool enable_layers = false) {
+std::pair<vk::Instance, u32> CreateInstance(Common::DynamicLibrary& library,
+ vk::InstanceDispatch& dld, WindowSystemType window_type,
+ bool enable_debug_utils, bool enable_layers) {
if (!library.IsOpen()) {
LOG_ERROR(Render_Vulkan, "Vulkan library not available");
return {};
@@ -133,7 +133,7 @@ std::pair<vk::Instance, u32> CreateInstance(
if (window_type != Core::Frontend::WindowSystemType::Headless) {
extensions.push_back(VK_KHR_SURFACE_EXTENSION_NAME);
}
- if (enable_layers) {
+ if (enable_debug_utils) {
extensions.push_back(VK_EXT_DEBUG_UTILS_EXTENSION_NAME);
}
extensions.push_back(VK_KHR_GET_PHYSICAL_DEVICE_PROPERTIES_2_EXTENSION_NAME);
@@ -243,8 +243,8 @@ std::string BuildCommaSeparatedExtensions(std::vector<std::string> available_ext
RendererVulkan::RendererVulkan(Core::TelemetrySession& telemetry_session_,
Core::Frontend::EmuWindow& emu_window,
Core::Memory::Memory& cpu_memory_, Tegra::GPU& gpu_,
- std::unique_ptr<Core::Frontend::GraphicsContext> context)
- : RendererBase{emu_window, std::move(context)}, telemetry_session{telemetry_session_},
+ std::unique_ptr<Core::Frontend::GraphicsContext> context_)
+ : RendererBase{emu_window, std::move(context_)}, telemetry_session{telemetry_session_},
cpu_memory{cpu_memory_}, gpu{gpu_} {}
RendererVulkan::~RendererVulkan() {
@@ -287,7 +287,7 @@ void RendererVulkan::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
bool RendererVulkan::Init() {
library = OpenVulkanLibrary();
std::tie(instance, instance_version) = CreateInstance(
- library, dld, render_window.GetWindowInfo().type, Settings::values.renderer_debug);
+ library, dld, render_window.GetWindowInfo().type, true, Settings::values.renderer_debug);
if (!instance || !CreateDebugCallback() || !CreateSurface() || !PickDevices()) {
return false;
}
@@ -447,7 +447,8 @@ void RendererVulkan::Report() const {
std::vector<std::string> RendererVulkan::EnumerateDevices() {
vk::InstanceDispatch dld;
Common::DynamicLibrary library = OpenVulkanLibrary();
- vk::Instance instance = CreateInstance(library, dld).first;
+ vk::Instance instance =
+ CreateInstance(library, dld, WindowSystemType::Headless, false, false).first;
if (!instance) {
return {};
}
diff --git a/src/video_core/renderer_vulkan/renderer_vulkan.h b/src/video_core/renderer_vulkan/renderer_vulkan.h
index 1044ca124..74642fba4 100644
--- a/src/video_core/renderer_vulkan/renderer_vulkan.h
+++ b/src/video_core/renderer_vulkan/renderer_vulkan.h
@@ -33,10 +33,9 @@ class VKDevice;
class VKMemoryManager;
class VKSwapchain;
class VKScheduler;
-class VKImage;
struct VKScreenInfo {
- VKImage* image{};
+ VkImageView image_view{};
u32 width{};
u32 height{};
bool is_srgb{};
@@ -45,9 +44,9 @@ struct VKScreenInfo {
class RendererVulkan final : public VideoCore::RendererBase {
public:
explicit RendererVulkan(Core::TelemetrySession& telemtry_session,
- Core::Frontend::EmuWindow& emu_window, Core::Memory::Memory& cpu_memory,
- Tegra::GPU& gpu,
- std::unique_ptr<Core::Frontend::GraphicsContext> context);
+ Core::Frontend::EmuWindow& emu_window,
+ Core::Memory::Memory& cpu_memory_, Tegra::GPU& gpu_,
+ std::unique_ptr<Core::Frontend::GraphicsContext> context_);
~RendererVulkan() override;
bool Init() override;
diff --git a/src/video_core/renderer_vulkan/vk_blit_screen.cpp b/src/video_core/renderer_vulkan/vk_blit_screen.cpp
index b5b60309e..d3a83f22f 100644
--- a/src/video_core/renderer_vulkan/vk_blit_screen.cpp
+++ b/src/video_core/renderer_vulkan/vk_blit_screen.cpp
@@ -16,12 +16,12 @@
#include "core/frontend/emu_window.h"
#include "core/memory.h"
#include "video_core/gpu.h"
-#include "video_core/morton.h"
+#include "video_core/host_shaders/vulkan_present_frag_spv.h"
+#include "video_core/host_shaders/vulkan_present_vert_spv.h"
#include "video_core/rasterizer_interface.h"
#include "video_core/renderer_vulkan/renderer_vulkan.h"
#include "video_core/renderer_vulkan/vk_blit_screen.h"
#include "video_core/renderer_vulkan/vk_device.h"
-#include "video_core/renderer_vulkan/vk_image.h"
#include "video_core/renderer_vulkan/vk_master_semaphore.h"
#include "video_core/renderer_vulkan/vk_memory_manager.h"
#include "video_core/renderer_vulkan/vk_scheduler.h"
@@ -29,108 +29,12 @@
#include "video_core/renderer_vulkan/vk_swapchain.h"
#include "video_core/renderer_vulkan/wrapper.h"
#include "video_core/surface.h"
+#include "video_core/textures/decoders.h"
namespace Vulkan {
namespace {
-// Generated from the "shaders/" directory, read the instructions there.
-constexpr u8 blit_vertex_code[] = {
- 0x03, 0x02, 0x23, 0x07, 0x00, 0x00, 0x01, 0x00, 0x07, 0x00, 0x08, 0x00, 0x27, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x11, 0x00, 0x02, 0x00, 0x01, 0x00, 0x00, 0x00, 0x0b, 0x00, 0x06, 0x00,
- 0x01, 0x00, 0x00, 0x00, 0x47, 0x4c, 0x53, 0x4c, 0x2e, 0x73, 0x74, 0x64, 0x2e, 0x34, 0x35, 0x30,
- 0x00, 0x00, 0x00, 0x00, 0x0e, 0x00, 0x03, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00,
- 0x0f, 0x00, 0x09, 0x00, 0x00, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x00, 0x6d, 0x61, 0x69, 0x6e,
- 0x00, 0x00, 0x00, 0x00, 0x0d, 0x00, 0x00, 0x00, 0x19, 0x00, 0x00, 0x00, 0x24, 0x00, 0x00, 0x00,
- 0x25, 0x00, 0x00, 0x00, 0x48, 0x00, 0x05, 0x00, 0x0b, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0x0b, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x48, 0x00, 0x05, 0x00, 0x0b, 0x00, 0x00, 0x00,
- 0x01, 0x00, 0x00, 0x00, 0x0b, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x48, 0x00, 0x05, 0x00,
- 0x0b, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00, 0x0b, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00,
- 0x48, 0x00, 0x05, 0x00, 0x0b, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00, 0x0b, 0x00, 0x00, 0x00,
- 0x04, 0x00, 0x00, 0x00, 0x47, 0x00, 0x03, 0x00, 0x0b, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00,
- 0x48, 0x00, 0x04, 0x00, 0x11, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x05, 0x00, 0x00, 0x00,
- 0x48, 0x00, 0x05, 0x00, 0x11, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x23, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x48, 0x00, 0x05, 0x00, 0x11, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0x07, 0x00, 0x00, 0x00, 0x10, 0x00, 0x00, 0x00, 0x47, 0x00, 0x03, 0x00, 0x11, 0x00, 0x00, 0x00,
- 0x02, 0x00, 0x00, 0x00, 0x47, 0x00, 0x04, 0x00, 0x13, 0x00, 0x00, 0x00, 0x22, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x47, 0x00, 0x04, 0x00, 0x13, 0x00, 0x00, 0x00, 0x21, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x47, 0x00, 0x04, 0x00, 0x19, 0x00, 0x00, 0x00, 0x1e, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x47, 0x00, 0x04, 0x00, 0x24, 0x00, 0x00, 0x00, 0x1e, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x47, 0x00, 0x04, 0x00, 0x25, 0x00, 0x00, 0x00, 0x1e, 0x00, 0x00, 0x00,
- 0x01, 0x00, 0x00, 0x00, 0x13, 0x00, 0x02, 0x00, 0x02, 0x00, 0x00, 0x00, 0x21, 0x00, 0x03, 0x00,
- 0x03, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00, 0x16, 0x00, 0x03, 0x00, 0x06, 0x00, 0x00, 0x00,
- 0x20, 0x00, 0x00, 0x00, 0x17, 0x00, 0x04, 0x00, 0x07, 0x00, 0x00, 0x00, 0x06, 0x00, 0x00, 0x00,
- 0x04, 0x00, 0x00, 0x00, 0x15, 0x00, 0x04, 0x00, 0x08, 0x00, 0x00, 0x00, 0x20, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x2b, 0x00, 0x04, 0x00, 0x08, 0x00, 0x00, 0x00, 0x09, 0x00, 0x00, 0x00,
- 0x01, 0x00, 0x00, 0x00, 0x1c, 0x00, 0x04, 0x00, 0x0a, 0x00, 0x00, 0x00, 0x06, 0x00, 0x00, 0x00,
- 0x09, 0x00, 0x00, 0x00, 0x1e, 0x00, 0x06, 0x00, 0x0b, 0x00, 0x00, 0x00, 0x07, 0x00, 0x00, 0x00,
- 0x06, 0x00, 0x00, 0x00, 0x0a, 0x00, 0x00, 0x00, 0x0a, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00,
- 0x0c, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00, 0x0b, 0x00, 0x00, 0x00, 0x3b, 0x00, 0x04, 0x00,
- 0x0c, 0x00, 0x00, 0x00, 0x0d, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00, 0x15, 0x00, 0x04, 0x00,
- 0x0e, 0x00, 0x00, 0x00, 0x20, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x2b, 0x00, 0x04, 0x00,
- 0x0e, 0x00, 0x00, 0x00, 0x0f, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x18, 0x00, 0x04, 0x00,
- 0x10, 0x00, 0x00, 0x00, 0x07, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x00, 0x1e, 0x00, 0x03, 0x00,
- 0x11, 0x00, 0x00, 0x00, 0x10, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00, 0x12, 0x00, 0x00, 0x00,
- 0x02, 0x00, 0x00, 0x00, 0x11, 0x00, 0x00, 0x00, 0x3b, 0x00, 0x04, 0x00, 0x12, 0x00, 0x00, 0x00,
- 0x13, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00, 0x14, 0x00, 0x00, 0x00,
- 0x02, 0x00, 0x00, 0x00, 0x10, 0x00, 0x00, 0x00, 0x17, 0x00, 0x04, 0x00, 0x17, 0x00, 0x00, 0x00,
- 0x06, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00, 0x18, 0x00, 0x00, 0x00,
- 0x01, 0x00, 0x00, 0x00, 0x17, 0x00, 0x00, 0x00, 0x3b, 0x00, 0x04, 0x00, 0x18, 0x00, 0x00, 0x00,
- 0x19, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x2b, 0x00, 0x04, 0x00, 0x06, 0x00, 0x00, 0x00,
- 0x1b, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x2b, 0x00, 0x04, 0x00, 0x06, 0x00, 0x00, 0x00,
- 0x1c, 0x00, 0x00, 0x00, 0x00, 0x00, 0x80, 0x3f, 0x20, 0x00, 0x04, 0x00, 0x21, 0x00, 0x00, 0x00,
- 0x03, 0x00, 0x00, 0x00, 0x07, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00, 0x23, 0x00, 0x00, 0x00,
- 0x03, 0x00, 0x00, 0x00, 0x17, 0x00, 0x00, 0x00, 0x3b, 0x00, 0x04, 0x00, 0x23, 0x00, 0x00, 0x00,
- 0x24, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00, 0x3b, 0x00, 0x04, 0x00, 0x18, 0x00, 0x00, 0x00,
- 0x25, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x36, 0x00, 0x05, 0x00, 0x02, 0x00, 0x00, 0x00,
- 0x04, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00, 0xf8, 0x00, 0x02, 0x00,
- 0x05, 0x00, 0x00, 0x00, 0x41, 0x00, 0x05, 0x00, 0x14, 0x00, 0x00, 0x00, 0x15, 0x00, 0x00, 0x00,
- 0x13, 0x00, 0x00, 0x00, 0x0f, 0x00, 0x00, 0x00, 0x3d, 0x00, 0x04, 0x00, 0x10, 0x00, 0x00, 0x00,
- 0x16, 0x00, 0x00, 0x00, 0x15, 0x00, 0x00, 0x00, 0x3d, 0x00, 0x04, 0x00, 0x17, 0x00, 0x00, 0x00,
- 0x1a, 0x00, 0x00, 0x00, 0x19, 0x00, 0x00, 0x00, 0x51, 0x00, 0x05, 0x00, 0x06, 0x00, 0x00, 0x00,
- 0x1d, 0x00, 0x00, 0x00, 0x1a, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x51, 0x00, 0x05, 0x00,
- 0x06, 0x00, 0x00, 0x00, 0x1e, 0x00, 0x00, 0x00, 0x1a, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00,
- 0x50, 0x00, 0x07, 0x00, 0x07, 0x00, 0x00, 0x00, 0x1f, 0x00, 0x00, 0x00, 0x1d, 0x00, 0x00, 0x00,
- 0x1e, 0x00, 0x00, 0x00, 0x1b, 0x00, 0x00, 0x00, 0x1c, 0x00, 0x00, 0x00, 0x91, 0x00, 0x05, 0x00,
- 0x07, 0x00, 0x00, 0x00, 0x20, 0x00, 0x00, 0x00, 0x16, 0x00, 0x00, 0x00, 0x1f, 0x00, 0x00, 0x00,
- 0x41, 0x00, 0x05, 0x00, 0x21, 0x00, 0x00, 0x00, 0x22, 0x00, 0x00, 0x00, 0x0d, 0x00, 0x00, 0x00,
- 0x0f, 0x00, 0x00, 0x00, 0x3e, 0x00, 0x03, 0x00, 0x22, 0x00, 0x00, 0x00, 0x20, 0x00, 0x00, 0x00,
- 0x3d, 0x00, 0x04, 0x00, 0x17, 0x00, 0x00, 0x00, 0x26, 0x00, 0x00, 0x00, 0x25, 0x00, 0x00, 0x00,
- 0x3e, 0x00, 0x03, 0x00, 0x24, 0x00, 0x00, 0x00, 0x26, 0x00, 0x00, 0x00, 0xfd, 0x00, 0x01, 0x00,
- 0x38, 0x00, 0x01, 0x00};
-
-constexpr u8 blit_fragment_code[] = {
- 0x03, 0x02, 0x23, 0x07, 0x00, 0x00, 0x01, 0x00, 0x07, 0x00, 0x08, 0x00, 0x14, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x11, 0x00, 0x02, 0x00, 0x01, 0x00, 0x00, 0x00, 0x0b, 0x00, 0x06, 0x00,
- 0x01, 0x00, 0x00, 0x00, 0x47, 0x4c, 0x53, 0x4c, 0x2e, 0x73, 0x74, 0x64, 0x2e, 0x34, 0x35, 0x30,
- 0x00, 0x00, 0x00, 0x00, 0x0e, 0x00, 0x03, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00,
- 0x0f, 0x00, 0x07, 0x00, 0x04, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x00, 0x6d, 0x61, 0x69, 0x6e,
- 0x00, 0x00, 0x00, 0x00, 0x09, 0x00, 0x00, 0x00, 0x11, 0x00, 0x00, 0x00, 0x10, 0x00, 0x03, 0x00,
- 0x04, 0x00, 0x00, 0x00, 0x07, 0x00, 0x00, 0x00, 0x47, 0x00, 0x04, 0x00, 0x09, 0x00, 0x00, 0x00,
- 0x1e, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x47, 0x00, 0x04, 0x00, 0x0d, 0x00, 0x00, 0x00,
- 0x22, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x47, 0x00, 0x04, 0x00, 0x0d, 0x00, 0x00, 0x00,
- 0x21, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x47, 0x00, 0x04, 0x00, 0x11, 0x00, 0x00, 0x00,
- 0x1e, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x13, 0x00, 0x02, 0x00, 0x02, 0x00, 0x00, 0x00,
- 0x21, 0x00, 0x03, 0x00, 0x03, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00, 0x16, 0x00, 0x03, 0x00,
- 0x06, 0x00, 0x00, 0x00, 0x20, 0x00, 0x00, 0x00, 0x17, 0x00, 0x04, 0x00, 0x07, 0x00, 0x00, 0x00,
- 0x06, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00, 0x08, 0x00, 0x00, 0x00,
- 0x03, 0x00, 0x00, 0x00, 0x07, 0x00, 0x00, 0x00, 0x3b, 0x00, 0x04, 0x00, 0x08, 0x00, 0x00, 0x00,
- 0x09, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00, 0x19, 0x00, 0x09, 0x00, 0x0a, 0x00, 0x00, 0x00,
- 0x06, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x1b, 0x00, 0x03, 0x00,
- 0x0b, 0x00, 0x00, 0x00, 0x0a, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00, 0x0c, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x0b, 0x00, 0x00, 0x00, 0x3b, 0x00, 0x04, 0x00, 0x0c, 0x00, 0x00, 0x00,
- 0x0d, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x17, 0x00, 0x04, 0x00, 0x0f, 0x00, 0x00, 0x00,
- 0x06, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00, 0x10, 0x00, 0x00, 0x00,
- 0x01, 0x00, 0x00, 0x00, 0x0f, 0x00, 0x00, 0x00, 0x3b, 0x00, 0x04, 0x00, 0x10, 0x00, 0x00, 0x00,
- 0x11, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x36, 0x00, 0x05, 0x00, 0x02, 0x00, 0x00, 0x00,
- 0x04, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00, 0xf8, 0x00, 0x02, 0x00,
- 0x05, 0x00, 0x00, 0x00, 0x3d, 0x00, 0x04, 0x00, 0x0b, 0x00, 0x00, 0x00, 0x0e, 0x00, 0x00, 0x00,
- 0x0d, 0x00, 0x00, 0x00, 0x3d, 0x00, 0x04, 0x00, 0x0f, 0x00, 0x00, 0x00, 0x12, 0x00, 0x00, 0x00,
- 0x11, 0x00, 0x00, 0x00, 0x57, 0x00, 0x05, 0x00, 0x07, 0x00, 0x00, 0x00, 0x13, 0x00, 0x00, 0x00,
- 0x0e, 0x00, 0x00, 0x00, 0x12, 0x00, 0x00, 0x00, 0x3e, 0x00, 0x03, 0x00, 0x09, 0x00, 0x00, 0x00,
- 0x13, 0x00, 0x00, 0x00, 0xfd, 0x00, 0x01, 0x00, 0x38, 0x00, 0x01, 0x00};
-
struct ScreenRectVertex {
ScreenRectVertex() = default;
explicit ScreenRectVertex(f32 x, f32 y, f32 u, f32 v) : position{{x, y}}, tex_coord{{u, v}} {}
@@ -173,9 +77,9 @@ constexpr std::array<f32, 4 * 4> MakeOrthographicMatrix(f32 width, f32 height) {
// clang-format on
}
-std::size_t GetBytesPerPixel(const Tegra::FramebufferConfig& framebuffer) {
+u32 GetBytesPerPixel(const Tegra::FramebufferConfig& framebuffer) {
using namespace VideoCore::Surface;
- return GetBytesPerPixel(PixelFormatFromGPUPixelFormat(framebuffer.pixel_format));
+ return BytesPerBlock(PixelFormatFromGPUPixelFormat(framebuffer.pixel_format));
}
std::size_t GetSizeInBytes(const Tegra::FramebufferConfig& framebuffer) {
@@ -239,34 +143,30 @@ VkSemaphore VKBlitScreen::Draw(const Tegra::FramebufferConfig& framebuffer, bool
scheduler.Wait(resource_ticks[image_index]);
resource_ticks[image_index] = scheduler.CurrentTick();
- VKImage* blit_image = use_accelerated ? screen_info.image : raw_images[image_index].get();
-
- UpdateDescriptorSet(image_index, blit_image->GetPresentView());
+ UpdateDescriptorSet(image_index,
+ use_accelerated ? screen_info.image_view : *raw_image_views[image_index]);
BufferData data;
SetUniformData(data, framebuffer);
SetVertexData(data, framebuffer);
auto map = buffer_commit->Map();
- std::memcpy(map.GetAddress(), &data, sizeof(data));
+ std::memcpy(map.Address(), &data, sizeof(data));
if (!use_accelerated) {
const u64 image_offset = GetRawImageOffset(framebuffer, image_index);
- const auto pixel_format =
- VideoCore::Surface::PixelFormatFromGPUPixelFormat(framebuffer.pixel_format);
const VAddr framebuffer_addr = framebuffer.address + framebuffer.offset;
- const auto host_ptr = cpu_memory.GetPointer(framebuffer_addr);
- rasterizer.FlushRegion(ToCacheAddr(host_ptr), GetSizeInBytes(framebuffer));
+ const u8* const host_ptr = cpu_memory.GetPointer(framebuffer_addr);
+ const size_t size_bytes = GetSizeInBytes(framebuffer);
+ rasterizer.FlushRegion(ToCacheAddr(host_ptr), size_bytes);
// TODO(Rodrigo): Read this from HLE
constexpr u32 block_height_log2 = 4;
- VideoCore::MortonSwizzle(VideoCore::MortonSwizzleMode::MortonToLinear, pixel_format,
- framebuffer.stride, block_height_log2, framebuffer.height, 0, 1, 1,
- map.GetAddress() + image_offset, host_ptr);
-
- blit_image->Transition(0, 1, 0, 1, VK_PIPELINE_STAGE_TRANSFER_BIT,
- VK_ACCESS_TRANSFER_WRITE_BIT, VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL);
+ const u32 bytes_per_pixel = GetBytesPerPixel(framebuffer);
+ Tegra::Texture::UnswizzleTexture(
+ std::span(map.Address() + image_offset, size_bytes), std::span(host_ptr, size_bytes),
+ bytes_per_pixel, framebuffer.width, framebuffer.height, 1, block_height_log2, 0);
const VkBufferImageCopy copy{
.bufferOffset = image_offset,
@@ -288,15 +188,44 @@ VkSemaphore VKBlitScreen::Draw(const Tegra::FramebufferConfig& framebuffer, bool
},
};
scheduler.Record(
- [buffer = *buffer, image = *blit_image->GetHandle(), copy](vk::CommandBuffer cmdbuf) {
- cmdbuf.CopyBufferToImage(buffer, image, VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL, copy);
+ [buffer = *buffer, image = *raw_images[image_index], copy](vk::CommandBuffer cmdbuf) {
+ const VkImageMemoryBarrier base_barrier{
+ .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
+ .pNext = nullptr,
+ .srcAccessMask = 0,
+ .dstAccessMask = 0,
+ .oldLayout = VK_IMAGE_LAYOUT_GENERAL,
+ .newLayout = VK_IMAGE_LAYOUT_GENERAL,
+ .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+ .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+ .image = image,
+ .subresourceRange =
+ {
+ .aspectMask = VK_IMAGE_ASPECT_COLOR_BIT,
+ .baseMipLevel = 0,
+ .levelCount = 1,
+ .baseArrayLayer = 0,
+ .layerCount = 1,
+ },
+ };
+ VkImageMemoryBarrier read_barrier = base_barrier;
+ read_barrier.srcAccessMask = VK_ACCESS_HOST_WRITE_BIT;
+ read_barrier.dstAccessMask = VK_ACCESS_TRANSFER_WRITE_BIT;
+ read_barrier.oldLayout = VK_IMAGE_LAYOUT_UNDEFINED;
+
+ VkImageMemoryBarrier write_barrier = base_barrier;
+ write_barrier.srcAccessMask = VK_ACCESS_TRANSFER_WRITE_BIT;
+ write_barrier.dstAccessMask = VK_ACCESS_SHADER_READ_BIT;
+
+ cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_HOST_BIT, VK_PIPELINE_STAGE_TRANSFER_BIT,
+ 0, read_barrier);
+ cmdbuf.CopyBufferToImage(buffer, image, VK_IMAGE_LAYOUT_GENERAL, copy);
+ cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_TRANSFER_BIT,
+ VK_PIPELINE_STAGE_FRAGMENT_SHADER_BIT, 0, write_barrier);
});
}
map.Release();
- blit_image->Transition(0, 1, 0, 1, VK_PIPELINE_STAGE_FRAGMENT_SHADER_BIT,
- VK_ACCESS_SHADER_READ_BIT, VK_IMAGE_LAYOUT_SHADER_READ_ONLY_OPTIMAL);
-
scheduler.Record([renderpass = *renderpass, framebuffer = *framebuffers[image_index],
descriptor_set = descriptor_sets[image_index], buffer = *buffer,
size = swapchain.GetSize(), pipeline = *pipeline,
@@ -304,31 +233,31 @@ VkSemaphore VKBlitScreen::Draw(const Tegra::FramebufferConfig& framebuffer, bool
const VkClearValue clear_color{
.color = {.float32 = {0.0f, 0.0f, 0.0f, 0.0f}},
};
-
- VkRenderPassBeginInfo renderpass_bi;
- renderpass_bi.sType = VK_STRUCTURE_TYPE_RENDER_PASS_BEGIN_INFO;
- renderpass_bi.pNext = nullptr;
- renderpass_bi.renderPass = renderpass;
- renderpass_bi.framebuffer = framebuffer;
- renderpass_bi.renderArea.offset.x = 0;
- renderpass_bi.renderArea.offset.y = 0;
- renderpass_bi.renderArea.extent = size;
- renderpass_bi.clearValueCount = 1;
- renderpass_bi.pClearValues = &clear_color;
-
- VkViewport viewport;
- viewport.x = 0.0f;
- viewport.y = 0.0f;
- viewport.width = static_cast<float>(size.width);
- viewport.height = static_cast<float>(size.height);
- viewport.minDepth = 0.0f;
- viewport.maxDepth = 1.0f;
-
- VkRect2D scissor;
- scissor.offset.x = 0;
- scissor.offset.y = 0;
- scissor.extent = size;
-
+ const VkRenderPassBeginInfo renderpass_bi{
+ .sType = VK_STRUCTURE_TYPE_RENDER_PASS_BEGIN_INFO,
+ .pNext = nullptr,
+ .renderPass = renderpass,
+ .framebuffer = framebuffer,
+ .renderArea =
+ {
+ .offset = {0, 0},
+ .extent = size,
+ },
+ .clearValueCount = 1,
+ .pClearValues = &clear_color,
+ };
+ const VkViewport viewport{
+ .x = 0.0f,
+ .y = 0.0f,
+ .width = static_cast<float>(size.width),
+ .height = static_cast<float>(size.height),
+ .minDepth = 0.0f,
+ .maxDepth = 1.0f,
+ };
+ const VkRect2D scissor{
+ .offset = {0, 0},
+ .extent = size,
+ };
cmdbuf.BeginRenderPass(renderpass_bi, VK_SUBPASS_CONTENTS_INLINE);
cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, pipeline);
cmdbuf.SetViewport(0, viewport);
@@ -372,8 +301,8 @@ void VKBlitScreen::RefreshResources(const Tegra::FramebufferConfig& framebuffer)
}
void VKBlitScreen::CreateShaders() {
- vertex_shader = BuildShader(device, sizeof(blit_vertex_code), blit_vertex_code);
- fragment_shader = BuildShader(device, sizeof(blit_fragment_code), blit_fragment_code);
+ vertex_shader = BuildShader(device, VULKAN_PRESENT_VERT_SPV);
+ fragment_shader = BuildShader(device, VULKAN_PRESENT_FRAG_SPV);
}
void VKBlitScreen::CreateSemaphores() {
@@ -420,7 +349,7 @@ void VKBlitScreen::CreateRenderPass() {
const VkAttachmentReference color_attachment_ref{
.attachment = 0,
- .layout = VK_IMAGE_LAYOUT_COLOR_ATTACHMENT_OPTIMAL,
+ .layout = VK_IMAGE_LAYOUT_GENERAL,
};
const VkSubpassDescription subpass_description{
@@ -735,34 +664,56 @@ void VKBlitScreen::CreateStagingBuffer(const Tegra::FramebufferConfig& framebuff
void VKBlitScreen::CreateRawImages(const Tegra::FramebufferConfig& framebuffer) {
raw_images.resize(image_count);
+ raw_image_views.resize(image_count);
raw_buffer_commits.resize(image_count);
- const VkImageCreateInfo ci{
- .sType = VK_STRUCTURE_TYPE_IMAGE_CREATE_INFO,
- .pNext = nullptr,
- .flags = 0,
- .imageType = VK_IMAGE_TYPE_2D,
- .format = GetFormat(framebuffer),
- .extent =
- {
- .width = framebuffer.width,
- .height = framebuffer.height,
- .depth = 1,
- },
- .mipLevels = 1,
- .arrayLayers = 1,
- .samples = VK_SAMPLE_COUNT_1_BIT,
- .tiling = VK_IMAGE_TILING_LINEAR,
- .usage = VK_IMAGE_USAGE_TRANSFER_DST_BIT | VK_IMAGE_USAGE_SAMPLED_BIT,
- .sharingMode = VK_SHARING_MODE_EXCLUSIVE,
- .queueFamilyIndexCount = 0,
- .pQueueFamilyIndices = nullptr,
- .initialLayout = VK_IMAGE_LAYOUT_UNDEFINED,
- };
-
- for (std::size_t i = 0; i < image_count; ++i) {
- raw_images[i] = std::make_unique<VKImage>(device, scheduler, ci, VK_IMAGE_ASPECT_COLOR_BIT);
- raw_buffer_commits[i] = memory_manager.Commit(raw_images[i]->GetHandle(), false);
+ for (size_t i = 0; i < image_count; ++i) {
+ raw_images[i] = device.GetLogical().CreateImage(VkImageCreateInfo{
+ .sType = VK_STRUCTURE_TYPE_IMAGE_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .imageType = VK_IMAGE_TYPE_2D,
+ .format = GetFormat(framebuffer),
+ .extent =
+ {
+ .width = framebuffer.width,
+ .height = framebuffer.height,
+ .depth = 1,
+ },
+ .mipLevels = 1,
+ .arrayLayers = 1,
+ .samples = VK_SAMPLE_COUNT_1_BIT,
+ .tiling = VK_IMAGE_TILING_LINEAR,
+ .usage = VK_IMAGE_USAGE_TRANSFER_DST_BIT | VK_IMAGE_USAGE_SAMPLED_BIT,
+ .sharingMode = VK_SHARING_MODE_EXCLUSIVE,
+ .queueFamilyIndexCount = 0,
+ .pQueueFamilyIndices = nullptr,
+ .initialLayout = VK_IMAGE_LAYOUT_UNDEFINED,
+ });
+ raw_buffer_commits[i] = memory_manager.Commit(raw_images[i], false);
+ raw_image_views[i] = device.GetLogical().CreateImageView(VkImageViewCreateInfo{
+ .sType = VK_STRUCTURE_TYPE_IMAGE_VIEW_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .image = *raw_images[i],
+ .viewType = VK_IMAGE_VIEW_TYPE_2D,
+ .format = GetFormat(framebuffer),
+ .components =
+ {
+ .r = VK_COMPONENT_SWIZZLE_IDENTITY,
+ .g = VK_COMPONENT_SWIZZLE_IDENTITY,
+ .b = VK_COMPONENT_SWIZZLE_IDENTITY,
+ .a = VK_COMPONENT_SWIZZLE_IDENTITY,
+ },
+ .subresourceRange =
+ {
+ .aspectMask = VK_IMAGE_ASPECT_COLOR_BIT,
+ .baseMipLevel = 0,
+ .levelCount = 1,
+ .baseArrayLayer = 0,
+ .layerCount = 1,
+ },
+ });
}
}
@@ -789,7 +740,7 @@ void VKBlitScreen::UpdateDescriptorSet(std::size_t image_index, VkImageView imag
const VkDescriptorImageInfo image_info{
.sampler = *sampler,
.imageView = image_view,
- .imageLayout = VK_IMAGE_LAYOUT_SHADER_READ_ONLY_OPTIMAL,
+ .imageLayout = VK_IMAGE_LAYOUT_GENERAL,
};
const VkWriteDescriptorSet sampler_write{
diff --git a/src/video_core/renderer_vulkan/vk_blit_screen.h b/src/video_core/renderer_vulkan/vk_blit_screen.h
index 8f2839214..2ee374247 100644
--- a/src/video_core/renderer_vulkan/vk_blit_screen.h
+++ b/src/video_core/renderer_vulkan/vk_blit_screen.h
@@ -35,7 +35,6 @@ struct ScreenInfo;
class RasterizerVulkan;
class VKDevice;
-class VKImage;
class VKScheduler;
class VKSwapchain;
@@ -110,7 +109,8 @@ private:
std::vector<u64> resource_ticks;
std::vector<vk::Semaphore> semaphores;
- std::vector<std::unique_ptr<VKImage>> raw_images;
+ std::vector<vk::Image> raw_images;
+ std::vector<vk::ImageView> raw_image_views;
std::vector<VKMemoryCommit> raw_buffer_commits;
u32 raw_width = 0;
u32 raw_height = 0;
diff --git a/src/video_core/renderer_vulkan/vk_buffer_cache.cpp b/src/video_core/renderer_vulkan/vk_buffer_cache.cpp
index d9d3da9ea..10d296c2f 100644
--- a/src/video_core/renderer_vulkan/vk_buffer_cache.cpp
+++ b/src/video_core/renderer_vulkan/vk_buffer_cache.cpp
@@ -31,20 +31,24 @@ constexpr VkAccessFlags UPLOAD_ACCESS_BARRIERS =
VK_ACCESS_TRANSFER_READ_BIT | VK_ACCESS_SHADER_READ_BIT | VK_ACCESS_UNIFORM_READ_BIT |
VK_ACCESS_VERTEX_ATTRIBUTE_READ_BIT | VK_ACCESS_INDEX_READ_BIT;
+constexpr VkAccessFlags TRANSFORM_FEEDBACK_WRITE_ACCESS =
+ VK_ACCESS_TRANSFORM_FEEDBACK_WRITE_BIT_EXT | VK_ACCESS_TRANSFORM_FEEDBACK_COUNTER_WRITE_BIT_EXT;
+
std::unique_ptr<VKStreamBuffer> CreateStreamBuffer(const VKDevice& device, VKScheduler& scheduler) {
- return std::make_unique<VKStreamBuffer>(device, scheduler, BUFFER_USAGE);
+ return std::make_unique<VKStreamBuffer>(device, scheduler);
}
} // Anonymous namespace
-Buffer::Buffer(const VKDevice& device, VKMemoryManager& memory_manager, VKScheduler& scheduler_,
- VKStagingBufferPool& staging_pool_, VAddr cpu_addr, std::size_t size)
- : BufferBlock{cpu_addr, size}, scheduler{scheduler_}, staging_pool{staging_pool_} {
+Buffer::Buffer(const VKDevice& device_, VKMemoryManager& memory_manager, VKScheduler& scheduler_,
+ VKStagingBufferPool& staging_pool_, VAddr cpu_addr_, std::size_t size_)
+ : BufferBlock{cpu_addr_, size_}, device{device_}, scheduler{scheduler_}, staging_pool{
+ staging_pool_} {
const VkBufferCreateInfo ci{
.sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO,
.pNext = nullptr,
.flags = 0,
- .size = static_cast<VkDeviceSize>(size),
+ .size = static_cast<VkDeviceSize>(size_),
.usage = BUFFER_USAGE | VK_BUFFER_USAGE_TRANSFER_SRC_BIT | VK_BUFFER_USAGE_TRANSFER_DST_BIT,
.sharingMode = VK_SHARING_MODE_EXCLUSIVE,
.queueFamilyIndexCount = 0,
@@ -57,69 +61,86 @@ Buffer::Buffer(const VKDevice& device, VKMemoryManager& memory_manager, VKSchedu
Buffer::~Buffer() = default;
-void Buffer::Upload(std::size_t offset, std::size_t size, const u8* data) {
- const auto& staging = staging_pool.GetUnusedBuffer(size, true);
- std::memcpy(staging.commit->Map(size), data, size);
+void Buffer::Upload(std::size_t offset, std::size_t data_size, const u8* data) {
+ const auto& staging = staging_pool.GetUnusedBuffer(data_size, true);
+ std::memcpy(staging.commit->Map(data_size), data, data_size);
scheduler.RequestOutsideRenderPassOperationContext();
const VkBuffer handle = Handle();
- scheduler.Record([staging = *staging.handle, handle, offset, size](vk::CommandBuffer cmdbuf) {
- cmdbuf.CopyBuffer(staging, handle, VkBufferCopy{0, offset, size});
-
- const VkBufferMemoryBarrier barrier{
+ scheduler.Record([staging = *staging.handle, handle, offset, data_size,
+ &device = device](vk::CommandBuffer cmdbuf) {
+ const VkBufferMemoryBarrier read_barrier{
.sType = VK_STRUCTURE_TYPE_BUFFER_MEMORY_BARRIER,
.pNext = nullptr,
- .srcAccessMask = VK_ACCESS_TRANSFER_WRITE_BIT,
- .dstAccessMask = UPLOAD_ACCESS_BARRIERS,
+ .srcAccessMask =
+ VK_ACCESS_SHADER_WRITE_BIT | VK_ACCESS_TRANSFER_WRITE_BIT |
+ VK_ACCESS_HOST_WRITE_BIT |
+ (device.IsExtTransformFeedbackSupported() ? TRANSFORM_FEEDBACK_WRITE_ACCESS : 0),
+ .dstAccessMask = VK_ACCESS_TRANSFER_READ_BIT,
.srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
.dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
.buffer = handle,
.offset = offset,
- .size = size,
+ .size = data_size,
};
- cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_TRANSFER_BIT, UPLOAD_PIPELINE_STAGE, 0, {},
- barrier, {});
- });
-}
-
-void Buffer::Download(std::size_t offset, std::size_t size, u8* data) {
- const auto& staging = staging_pool.GetUnusedBuffer(size, true);
- scheduler.RequestOutsideRenderPassOperationContext();
-
- const VkBuffer handle = Handle();
- scheduler.Record([staging = *staging.handle, handle, offset, size](vk::CommandBuffer cmdbuf) {
- const VkBufferMemoryBarrier barrier{
+ const VkBufferMemoryBarrier write_barrier{
.sType = VK_STRUCTURE_TYPE_BUFFER_MEMORY_BARRIER,
.pNext = nullptr,
- .srcAccessMask = VK_ACCESS_SHADER_WRITE_BIT,
- .dstAccessMask = VK_ACCESS_TRANSFER_READ_BIT,
+ .srcAccessMask = VK_ACCESS_TRANSFER_WRITE_BIT,
+ .dstAccessMask = UPLOAD_ACCESS_BARRIERS,
.srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
.dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
.buffer = handle,
.offset = offset,
- .size = size,
+ .size = data_size,
};
-
- cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_VERTEX_SHADER_BIT |
- VK_PIPELINE_STAGE_FRAGMENT_SHADER_BIT |
- VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT,
- VK_PIPELINE_STAGE_TRANSFER_BIT, 0, {}, barrier, {});
- cmdbuf.CopyBuffer(handle, staging, VkBufferCopy{offset, 0, size});
+ cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_ALL_COMMANDS_BIT, VK_PIPELINE_STAGE_TRANSFER_BIT,
+ 0, read_barrier);
+ cmdbuf.CopyBuffer(staging, handle, VkBufferCopy{0, offset, data_size});
+ cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_TRANSFER_BIT, UPLOAD_PIPELINE_STAGE, 0,
+ write_barrier);
});
+}
+
+void Buffer::Download(std::size_t offset, std::size_t data_size, u8* data) {
+ const auto& staging = staging_pool.GetUnusedBuffer(data_size, true);
+ scheduler.RequestOutsideRenderPassOperationContext();
+
+ const VkBuffer handle = Handle();
+ scheduler.Record(
+ [staging = *staging.handle, handle, offset, data_size](vk::CommandBuffer cmdbuf) {
+ const VkBufferMemoryBarrier barrier{
+ .sType = VK_STRUCTURE_TYPE_BUFFER_MEMORY_BARRIER,
+ .pNext = nullptr,
+ .srcAccessMask = VK_ACCESS_SHADER_WRITE_BIT,
+ .dstAccessMask = VK_ACCESS_TRANSFER_READ_BIT,
+ .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+ .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+ .buffer = handle,
+ .offset = offset,
+ .size = data_size,
+ };
+
+ cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_VERTEX_SHADER_BIT |
+ VK_PIPELINE_STAGE_FRAGMENT_SHADER_BIT |
+ VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT,
+ VK_PIPELINE_STAGE_TRANSFER_BIT, 0, {}, barrier, {});
+ cmdbuf.CopyBuffer(handle, staging, VkBufferCopy{offset, 0, data_size});
+ });
scheduler.Finish();
- std::memcpy(data, staging.commit->Map(size), size);
+ std::memcpy(data, staging.commit->Map(data_size), data_size);
}
void Buffer::CopyFrom(const Buffer& src, std::size_t src_offset, std::size_t dst_offset,
- std::size_t size) {
+ std::size_t copy_size) {
scheduler.RequestOutsideRenderPassOperationContext();
const VkBuffer dst_buffer = Handle();
scheduler.Record([src_buffer = src.Handle(), dst_buffer, src_offset, dst_offset,
- size](vk::CommandBuffer cmdbuf) {
- cmdbuf.CopyBuffer(src_buffer, dst_buffer, VkBufferCopy{src_offset, dst_offset, size});
+ copy_size](vk::CommandBuffer cmdbuf) {
+ cmdbuf.CopyBuffer(src_buffer, dst_buffer, VkBufferCopy{src_offset, dst_offset, copy_size});
std::array<VkBufferMemoryBarrier, 2> barriers;
barriers[0].sType = VK_STRUCTURE_TYPE_BUFFER_MEMORY_BARRIER;
@@ -130,7 +151,7 @@ void Buffer::CopyFrom(const Buffer& src, std::size_t src_offset, std::size_t dst
barriers[0].dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED;
barriers[0].buffer = src_buffer;
barriers[0].offset = src_offset;
- barriers[0].size = size;
+ barriers[0].size = copy_size;
barriers[1].sType = VK_STRUCTURE_TYPE_BUFFER_MEMORY_BARRIER;
barriers[1].pNext = nullptr;
barriers[1].srcAccessMask = VK_ACCESS_TRANSFER_WRITE_BIT;
@@ -139,19 +160,19 @@ void Buffer::CopyFrom(const Buffer& src, std::size_t src_offset, std::size_t dst
barriers[1].dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED;
barriers[1].buffer = dst_buffer;
barriers[1].offset = dst_offset;
- barriers[1].size = size;
+ barriers[1].size = copy_size;
cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_TRANSFER_BIT, UPLOAD_PIPELINE_STAGE, 0, {},
barriers, {});
});
}
-VKBufferCache::VKBufferCache(VideoCore::RasterizerInterface& rasterizer,
- Tegra::MemoryManager& gpu_memory, Core::Memory::Memory& cpu_memory,
+VKBufferCache::VKBufferCache(VideoCore::RasterizerInterface& rasterizer_,
+ Tegra::MemoryManager& gpu_memory_, Core::Memory::Memory& cpu_memory_,
const VKDevice& device_, VKMemoryManager& memory_manager_,
- VKScheduler& scheduler_, VKStagingBufferPool& staging_pool_)
- : VideoCommon::BufferCache<Buffer, VkBuffer, VKStreamBuffer>{rasterizer, gpu_memory, cpu_memory,
- CreateStreamBuffer(device_,
- scheduler_)},
+ VKScheduler& scheduler_, VKStreamBuffer& stream_buffer_,
+ VKStagingBufferPool& staging_pool_)
+ : VideoCommon::BufferCache<Buffer, VkBuffer, VKStreamBuffer>{rasterizer_, gpu_memory_,
+ cpu_memory_, stream_buffer_},
device{device_}, memory_manager{memory_manager_}, scheduler{scheduler_}, staging_pool{
staging_pool_} {}
diff --git a/src/video_core/renderer_vulkan/vk_buffer_cache.h b/src/video_core/renderer_vulkan/vk_buffer_cache.h
index 7fb5ceedf..daf498222 100644
--- a/src/video_core/renderer_vulkan/vk_buffer_cache.h
+++ b/src/video_core/renderer_vulkan/vk_buffer_cache.h
@@ -22,15 +22,15 @@ class VKScheduler;
class Buffer final : public VideoCommon::BufferBlock {
public:
explicit Buffer(const VKDevice& device, VKMemoryManager& memory_manager, VKScheduler& scheduler,
- VKStagingBufferPool& staging_pool, VAddr cpu_addr, std::size_t size);
+ VKStagingBufferPool& staging_pool, VAddr cpu_addr_, std::size_t size_);
~Buffer();
- void Upload(std::size_t offset, std::size_t size, const u8* data);
+ void Upload(std::size_t offset, std::size_t data_size, const u8* data);
- void Download(std::size_t offset, std::size_t size, u8* data);
+ void Download(std::size_t offset, std::size_t data_size, u8* data);
void CopyFrom(const Buffer& src, std::size_t src_offset, std::size_t dst_offset,
- std::size_t size);
+ std::size_t copy_size);
VkBuffer Handle() const {
return *buffer.handle;
@@ -41,6 +41,7 @@ public:
}
private:
+ const VKDevice& device;
VKScheduler& scheduler;
VKStagingBufferPool& staging_pool;
@@ -52,7 +53,8 @@ public:
explicit VKBufferCache(VideoCore::RasterizerInterface& rasterizer,
Tegra::MemoryManager& gpu_memory, Core::Memory::Memory& cpu_memory,
const VKDevice& device, VKMemoryManager& memory_manager,
- VKScheduler& scheduler, VKStagingBufferPool& staging_pool);
+ VKScheduler& scheduler, VKStreamBuffer& stream_buffer,
+ VKStagingBufferPool& staging_pool);
~VKBufferCache();
BufferInfo GetEmptyBuffer(std::size_t size) override;
diff --git a/src/video_core/renderer_vulkan/vk_command_pool.cpp b/src/video_core/renderer_vulkan/vk_command_pool.cpp
index 6339f4fe0..8f7d6410e 100644
--- a/src/video_core/renderer_vulkan/vk_command_pool.cpp
+++ b/src/video_core/renderer_vulkan/vk_command_pool.cpp
@@ -17,8 +17,8 @@ struct CommandPool::Pool {
vk::CommandBuffers cmdbufs;
};
-CommandPool::CommandPool(MasterSemaphore& master_semaphore, const VKDevice& device)
- : ResourcePool(master_semaphore, COMMAND_BUFFER_POOL_SIZE), device{device} {}
+CommandPool::CommandPool(MasterSemaphore& master_semaphore_, const VKDevice& device_)
+ : ResourcePool(master_semaphore_, COMMAND_BUFFER_POOL_SIZE), device{device_} {}
CommandPool::~CommandPool() = default;
diff --git a/src/video_core/renderer_vulkan/vk_command_pool.h b/src/video_core/renderer_vulkan/vk_command_pool.h
index b9cb3fb5d..62a7ce3f1 100644
--- a/src/video_core/renderer_vulkan/vk_command_pool.h
+++ b/src/video_core/renderer_vulkan/vk_command_pool.h
@@ -17,7 +17,7 @@ class VKDevice;
class CommandPool final : public ResourcePool {
public:
- explicit CommandPool(MasterSemaphore& master_semaphore, const VKDevice& device);
+ explicit CommandPool(MasterSemaphore& master_semaphore_, const VKDevice& device_);
~CommandPool() override;
void Allocate(size_t begin, size_t end) override;
diff --git a/src/video_core/renderer_vulkan/vk_compute_pass.cpp b/src/video_core/renderer_vulkan/vk_compute_pass.cpp
index 9637c6059..2c030e910 100644
--- a/src/video_core/renderer_vulkan/vk_compute_pass.cpp
+++ b/src/video_core/renderer_vulkan/vk_compute_pass.cpp
@@ -10,6 +10,9 @@
#include "common/alignment.h"
#include "common/assert.h"
#include "common/common_types.h"
+#include "video_core/host_shaders/vulkan_quad_array_comp_spv.h"
+#include "video_core/host_shaders/vulkan_quad_indexed_comp_spv.h"
+#include "video_core/host_shaders/vulkan_uint8_comp_spv.h"
#include "video_core/renderer_vulkan/vk_compute_pass.h"
#include "video_core/renderer_vulkan/vk_descriptor_pool.h"
#include "video_core/renderer_vulkan/vk_device.h"
@@ -22,99 +25,6 @@ namespace Vulkan {
namespace {
-// Quad array SPIR-V module. Generated from the "shaders/" directory, read the instructions there.
-constexpr u8 quad_array[] = {
- 0x03, 0x02, 0x23, 0x07, 0x00, 0x00, 0x01, 0x00, 0x07, 0x00, 0x08, 0x00, 0x54, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x11, 0x00, 0x02, 0x00, 0x01, 0x00, 0x00, 0x00, 0x0b, 0x00, 0x06, 0x00,
- 0x01, 0x00, 0x00, 0x00, 0x47, 0x4c, 0x53, 0x4c, 0x2e, 0x73, 0x74, 0x64, 0x2e, 0x34, 0x35, 0x30,
- 0x00, 0x00, 0x00, 0x00, 0x0e, 0x00, 0x03, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00,
- 0x0f, 0x00, 0x06, 0x00, 0x05, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x00, 0x6d, 0x61, 0x69, 0x6e,
- 0x00, 0x00, 0x00, 0x00, 0x0b, 0x00, 0x00, 0x00, 0x10, 0x00, 0x06, 0x00, 0x04, 0x00, 0x00, 0x00,
- 0x11, 0x00, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00,
- 0x47, 0x00, 0x04, 0x00, 0x0b, 0x00, 0x00, 0x00, 0x0b, 0x00, 0x00, 0x00, 0x1c, 0x00, 0x00, 0x00,
- 0x47, 0x00, 0x04, 0x00, 0x13, 0x00, 0x00, 0x00, 0x06, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x00,
- 0x48, 0x00, 0x05, 0x00, 0x14, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x23, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x47, 0x00, 0x03, 0x00, 0x14, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00,
- 0x47, 0x00, 0x04, 0x00, 0x16, 0x00, 0x00, 0x00, 0x22, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0x47, 0x00, 0x04, 0x00, 0x16, 0x00, 0x00, 0x00, 0x21, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0x48, 0x00, 0x05, 0x00, 0x29, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x23, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x47, 0x00, 0x03, 0x00, 0x29, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00,
- 0x47, 0x00, 0x04, 0x00, 0x4a, 0x00, 0x00, 0x00, 0x0b, 0x00, 0x00, 0x00, 0x19, 0x00, 0x00, 0x00,
- 0x13, 0x00, 0x02, 0x00, 0x02, 0x00, 0x00, 0x00, 0x21, 0x00, 0x03, 0x00, 0x03, 0x00, 0x00, 0x00,
- 0x02, 0x00, 0x00, 0x00, 0x15, 0x00, 0x04, 0x00, 0x06, 0x00, 0x00, 0x00, 0x20, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00, 0x07, 0x00, 0x00, 0x00, 0x07, 0x00, 0x00, 0x00,
- 0x06, 0x00, 0x00, 0x00, 0x17, 0x00, 0x04, 0x00, 0x09, 0x00, 0x00, 0x00, 0x06, 0x00, 0x00, 0x00,
- 0x03, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00, 0x0a, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00,
- 0x09, 0x00, 0x00, 0x00, 0x3b, 0x00, 0x04, 0x00, 0x0a, 0x00, 0x00, 0x00, 0x0b, 0x00, 0x00, 0x00,
- 0x01, 0x00, 0x00, 0x00, 0x2b, 0x00, 0x04, 0x00, 0x06, 0x00, 0x00, 0x00, 0x0c, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00, 0x0d, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00,
- 0x06, 0x00, 0x00, 0x00, 0x2b, 0x00, 0x04, 0x00, 0x06, 0x00, 0x00, 0x00, 0x11, 0x00, 0x00, 0x00,
- 0x06, 0x00, 0x00, 0x00, 0x1d, 0x00, 0x03, 0x00, 0x13, 0x00, 0x00, 0x00, 0x06, 0x00, 0x00, 0x00,
- 0x1e, 0x00, 0x03, 0x00, 0x14, 0x00, 0x00, 0x00, 0x13, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00,
- 0x15, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00, 0x14, 0x00, 0x00, 0x00, 0x3b, 0x00, 0x04, 0x00,
- 0x15, 0x00, 0x00, 0x00, 0x16, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00, 0x15, 0x00, 0x04, 0x00,
- 0x18, 0x00, 0x00, 0x00, 0x20, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x14, 0x00, 0x02, 0x00,
- 0x1b, 0x00, 0x00, 0x00, 0x1e, 0x00, 0x03, 0x00, 0x29, 0x00, 0x00, 0x00, 0x06, 0x00, 0x00, 0x00,
- 0x20, 0x00, 0x04, 0x00, 0x2a, 0x00, 0x00, 0x00, 0x09, 0x00, 0x00, 0x00, 0x29, 0x00, 0x00, 0x00,
- 0x3b, 0x00, 0x04, 0x00, 0x2a, 0x00, 0x00, 0x00, 0x2b, 0x00, 0x00, 0x00, 0x09, 0x00, 0x00, 0x00,
- 0x2b, 0x00, 0x04, 0x00, 0x18, 0x00, 0x00, 0x00, 0x2c, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0x20, 0x00, 0x04, 0x00, 0x2d, 0x00, 0x00, 0x00, 0x09, 0x00, 0x00, 0x00, 0x06, 0x00, 0x00, 0x00,
- 0x2b, 0x00, 0x04, 0x00, 0x06, 0x00, 0x00, 0x00, 0x31, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x00,
- 0x1c, 0x00, 0x04, 0x00, 0x34, 0x00, 0x00, 0x00, 0x06, 0x00, 0x00, 0x00, 0x11, 0x00, 0x00, 0x00,
- 0x2b, 0x00, 0x04, 0x00, 0x06, 0x00, 0x00, 0x00, 0x35, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00,
- 0x2b, 0x00, 0x04, 0x00, 0x06, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00,
- 0x2b, 0x00, 0x04, 0x00, 0x06, 0x00, 0x00, 0x00, 0x37, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00,
- 0x2c, 0x00, 0x09, 0x00, 0x34, 0x00, 0x00, 0x00, 0x38, 0x00, 0x00, 0x00, 0x0c, 0x00, 0x00, 0x00,
- 0x35, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x00, 0x0c, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x00,
- 0x37, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00, 0x3a, 0x00, 0x00, 0x00, 0x07, 0x00, 0x00, 0x00,
- 0x34, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00, 0x44, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00,
- 0x06, 0x00, 0x00, 0x00, 0x2b, 0x00, 0x04, 0x00, 0x18, 0x00, 0x00, 0x00, 0x47, 0x00, 0x00, 0x00,
- 0x01, 0x00, 0x00, 0x00, 0x2b, 0x00, 0x04, 0x00, 0x06, 0x00, 0x00, 0x00, 0x49, 0x00, 0x00, 0x00,
- 0x00, 0x04, 0x00, 0x00, 0x2c, 0x00, 0x06, 0x00, 0x09, 0x00, 0x00, 0x00, 0x4a, 0x00, 0x00, 0x00,
- 0x49, 0x00, 0x00, 0x00, 0x35, 0x00, 0x00, 0x00, 0x35, 0x00, 0x00, 0x00, 0x36, 0x00, 0x05, 0x00,
- 0x02, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00,
- 0xf8, 0x00, 0x02, 0x00, 0x05, 0x00, 0x00, 0x00, 0x3b, 0x00, 0x04, 0x00, 0x3a, 0x00, 0x00, 0x00,
- 0x3b, 0x00, 0x00, 0x00, 0x07, 0x00, 0x00, 0x00, 0xf9, 0x00, 0x02, 0x00, 0x4c, 0x00, 0x00, 0x00,
- 0xf8, 0x00, 0x02, 0x00, 0x4c, 0x00, 0x00, 0x00, 0xf6, 0x00, 0x04, 0x00, 0x4b, 0x00, 0x00, 0x00,
- 0x4e, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xf9, 0x00, 0x02, 0x00, 0x4d, 0x00, 0x00, 0x00,
- 0xf8, 0x00, 0x02, 0x00, 0x4d, 0x00, 0x00, 0x00, 0x41, 0x00, 0x05, 0x00, 0x0d, 0x00, 0x00, 0x00,
- 0x0e, 0x00, 0x00, 0x00, 0x0b, 0x00, 0x00, 0x00, 0x0c, 0x00, 0x00, 0x00, 0x3d, 0x00, 0x04, 0x00,
- 0x06, 0x00, 0x00, 0x00, 0x0f, 0x00, 0x00, 0x00, 0x0e, 0x00, 0x00, 0x00, 0x84, 0x00, 0x05, 0x00,
- 0x06, 0x00, 0x00, 0x00, 0x12, 0x00, 0x00, 0x00, 0x0f, 0x00, 0x00, 0x00, 0x11, 0x00, 0x00, 0x00,
- 0x44, 0x00, 0x05, 0x00, 0x06, 0x00, 0x00, 0x00, 0x17, 0x00, 0x00, 0x00, 0x16, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x7c, 0x00, 0x04, 0x00, 0x18, 0x00, 0x00, 0x00, 0x19, 0x00, 0x00, 0x00,
- 0x17, 0x00, 0x00, 0x00, 0x7c, 0x00, 0x04, 0x00, 0x06, 0x00, 0x00, 0x00, 0x1a, 0x00, 0x00, 0x00,
- 0x19, 0x00, 0x00, 0x00, 0xae, 0x00, 0x05, 0x00, 0x1b, 0x00, 0x00, 0x00, 0x1c, 0x00, 0x00, 0x00,
- 0x12, 0x00, 0x00, 0x00, 0x1a, 0x00, 0x00, 0x00, 0xf7, 0x00, 0x03, 0x00, 0x1e, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0xfa, 0x00, 0x04, 0x00, 0x1c, 0x00, 0x00, 0x00, 0x1d, 0x00, 0x00, 0x00,
- 0x1e, 0x00, 0x00, 0x00, 0xf8, 0x00, 0x02, 0x00, 0x1d, 0x00, 0x00, 0x00, 0xf9, 0x00, 0x02, 0x00,
- 0x4b, 0x00, 0x00, 0x00, 0xf8, 0x00, 0x02, 0x00, 0x1e, 0x00, 0x00, 0x00, 0xf9, 0x00, 0x02, 0x00,
- 0x21, 0x00, 0x00, 0x00, 0xf8, 0x00, 0x02, 0x00, 0x21, 0x00, 0x00, 0x00, 0xf5, 0x00, 0x07, 0x00,
- 0x06, 0x00, 0x00, 0x00, 0x53, 0x00, 0x00, 0x00, 0x0c, 0x00, 0x00, 0x00, 0x1e, 0x00, 0x00, 0x00,
- 0x48, 0x00, 0x00, 0x00, 0x22, 0x00, 0x00, 0x00, 0xb0, 0x00, 0x05, 0x00, 0x1b, 0x00, 0x00, 0x00,
- 0x27, 0x00, 0x00, 0x00, 0x53, 0x00, 0x00, 0x00, 0x11, 0x00, 0x00, 0x00, 0xf6, 0x00, 0x04, 0x00,
- 0x23, 0x00, 0x00, 0x00, 0x22, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xfa, 0x00, 0x04, 0x00,
- 0x27, 0x00, 0x00, 0x00, 0x22, 0x00, 0x00, 0x00, 0x23, 0x00, 0x00, 0x00, 0xf8, 0x00, 0x02, 0x00,
- 0x22, 0x00, 0x00, 0x00, 0x41, 0x00, 0x05, 0x00, 0x2d, 0x00, 0x00, 0x00, 0x2e, 0x00, 0x00, 0x00,
- 0x2b, 0x00, 0x00, 0x00, 0x2c, 0x00, 0x00, 0x00, 0x3d, 0x00, 0x04, 0x00, 0x06, 0x00, 0x00, 0x00,
- 0x2f, 0x00, 0x00, 0x00, 0x2e, 0x00, 0x00, 0x00, 0x84, 0x00, 0x05, 0x00, 0x06, 0x00, 0x00, 0x00,
- 0x32, 0x00, 0x00, 0x00, 0x0f, 0x00, 0x00, 0x00, 0x31, 0x00, 0x00, 0x00, 0x80, 0x00, 0x05, 0x00,
- 0x06, 0x00, 0x00, 0x00, 0x33, 0x00, 0x00, 0x00, 0x2f, 0x00, 0x00, 0x00, 0x32, 0x00, 0x00, 0x00,
- 0x3e, 0x00, 0x03, 0x00, 0x3b, 0x00, 0x00, 0x00, 0x38, 0x00, 0x00, 0x00, 0x41, 0x00, 0x05, 0x00,
- 0x07, 0x00, 0x00, 0x00, 0x3c, 0x00, 0x00, 0x00, 0x3b, 0x00, 0x00, 0x00, 0x53, 0x00, 0x00, 0x00,
- 0x3d, 0x00, 0x04, 0x00, 0x06, 0x00, 0x00, 0x00, 0x3d, 0x00, 0x00, 0x00, 0x3c, 0x00, 0x00, 0x00,
- 0x80, 0x00, 0x05, 0x00, 0x06, 0x00, 0x00, 0x00, 0x3e, 0x00, 0x00, 0x00, 0x33, 0x00, 0x00, 0x00,
- 0x3d, 0x00, 0x00, 0x00, 0x80, 0x00, 0x05, 0x00, 0x06, 0x00, 0x00, 0x00, 0x42, 0x00, 0x00, 0x00,
- 0x12, 0x00, 0x00, 0x00, 0x53, 0x00, 0x00, 0x00, 0x41, 0x00, 0x06, 0x00, 0x44, 0x00, 0x00, 0x00,
- 0x45, 0x00, 0x00, 0x00, 0x16, 0x00, 0x00, 0x00, 0x2c, 0x00, 0x00, 0x00, 0x42, 0x00, 0x00, 0x00,
- 0x3e, 0x00, 0x03, 0x00, 0x45, 0x00, 0x00, 0x00, 0x3e, 0x00, 0x00, 0x00, 0x80, 0x00, 0x05, 0x00,
- 0x06, 0x00, 0x00, 0x00, 0x48, 0x00, 0x00, 0x00, 0x53, 0x00, 0x00, 0x00, 0x47, 0x00, 0x00, 0x00,
- 0xf9, 0x00, 0x02, 0x00, 0x21, 0x00, 0x00, 0x00, 0xf8, 0x00, 0x02, 0x00, 0x23, 0x00, 0x00, 0x00,
- 0xf9, 0x00, 0x02, 0x00, 0x4b, 0x00, 0x00, 0x00, 0xf8, 0x00, 0x02, 0x00, 0x4e, 0x00, 0x00, 0x00,
- 0xf9, 0x00, 0x02, 0x00, 0x4c, 0x00, 0x00, 0x00, 0xf8, 0x00, 0x02, 0x00, 0x4b, 0x00, 0x00, 0x00,
- 0xfd, 0x00, 0x01, 0x00, 0x38, 0x00, 0x01, 0x00,
-};
-
VkDescriptorSetLayoutBinding BuildQuadArrayPassDescriptorSetLayoutBinding() {
return {
.binding = 0,
@@ -144,208 +54,6 @@ VkPushConstantRange BuildComputePushConstantRange(std::size_t size) {
};
}
-// Uint8 SPIR-V module. Generated from the "shaders/" directory.
-constexpr u8 uint8_pass[] = {
- 0x03, 0x02, 0x23, 0x07, 0x00, 0x00, 0x01, 0x00, 0x07, 0x00, 0x08, 0x00, 0x2f, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x11, 0x00, 0x02, 0x00, 0x01, 0x00, 0x00, 0x00, 0x11, 0x00, 0x02, 0x00,
- 0x51, 0x11, 0x00, 0x00, 0x11, 0x00, 0x02, 0x00, 0x61, 0x11, 0x00, 0x00, 0x0a, 0x00, 0x07, 0x00,
- 0x53, 0x50, 0x56, 0x5f, 0x4b, 0x48, 0x52, 0x5f, 0x31, 0x36, 0x62, 0x69, 0x74, 0x5f, 0x73, 0x74,
- 0x6f, 0x72, 0x61, 0x67, 0x65, 0x00, 0x00, 0x00, 0x0a, 0x00, 0x07, 0x00, 0x53, 0x50, 0x56, 0x5f,
- 0x4b, 0x48, 0x52, 0x5f, 0x38, 0x62, 0x69, 0x74, 0x5f, 0x73, 0x74, 0x6f, 0x72, 0x61, 0x67, 0x65,
- 0x00, 0x00, 0x00, 0x00, 0x0b, 0x00, 0x06, 0x00, 0x01, 0x00, 0x00, 0x00, 0x47, 0x4c, 0x53, 0x4c,
- 0x2e, 0x73, 0x74, 0x64, 0x2e, 0x34, 0x35, 0x30, 0x00, 0x00, 0x00, 0x00, 0x0e, 0x00, 0x03, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x0f, 0x00, 0x06, 0x00, 0x05, 0x00, 0x00, 0x00,
- 0x04, 0x00, 0x00, 0x00, 0x6d, 0x61, 0x69, 0x6e, 0x00, 0x00, 0x00, 0x00, 0x0b, 0x00, 0x00, 0x00,
- 0x10, 0x00, 0x06, 0x00, 0x04, 0x00, 0x00, 0x00, 0x11, 0x00, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00,
- 0x01, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x47, 0x00, 0x04, 0x00, 0x0b, 0x00, 0x00, 0x00,
- 0x0b, 0x00, 0x00, 0x00, 0x1c, 0x00, 0x00, 0x00, 0x47, 0x00, 0x04, 0x00, 0x12, 0x00, 0x00, 0x00,
- 0x06, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x48, 0x00, 0x04, 0x00, 0x13, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x18, 0x00, 0x00, 0x00, 0x48, 0x00, 0x05, 0x00, 0x13, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x23, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x47, 0x00, 0x03, 0x00,
- 0x13, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00, 0x47, 0x00, 0x04, 0x00, 0x15, 0x00, 0x00, 0x00,
- 0x22, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x47, 0x00, 0x04, 0x00, 0x15, 0x00, 0x00, 0x00,
- 0x21, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x47, 0x00, 0x04, 0x00, 0x1f, 0x00, 0x00, 0x00,
- 0x06, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00, 0x48, 0x00, 0x04, 0x00, 0x20, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x19, 0x00, 0x00, 0x00, 0x48, 0x00, 0x05, 0x00, 0x20, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x23, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x47, 0x00, 0x03, 0x00,
- 0x20, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00, 0x47, 0x00, 0x04, 0x00, 0x22, 0x00, 0x00, 0x00,
- 0x22, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x47, 0x00, 0x04, 0x00, 0x22, 0x00, 0x00, 0x00,
- 0x21, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x47, 0x00, 0x04, 0x00, 0x2e, 0x00, 0x00, 0x00,
- 0x0b, 0x00, 0x00, 0x00, 0x19, 0x00, 0x00, 0x00, 0x13, 0x00, 0x02, 0x00, 0x02, 0x00, 0x00, 0x00,
- 0x21, 0x00, 0x03, 0x00, 0x03, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00, 0x15, 0x00, 0x04, 0x00,
- 0x06, 0x00, 0x00, 0x00, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00,
- 0x07, 0x00, 0x00, 0x00, 0x07, 0x00, 0x00, 0x00, 0x06, 0x00, 0x00, 0x00, 0x17, 0x00, 0x04, 0x00,
- 0x09, 0x00, 0x00, 0x00, 0x06, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00,
- 0x0a, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x09, 0x00, 0x00, 0x00, 0x3b, 0x00, 0x04, 0x00,
- 0x0a, 0x00, 0x00, 0x00, 0x0b, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x2b, 0x00, 0x04, 0x00,
- 0x06, 0x00, 0x00, 0x00, 0x0c, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00,
- 0x0d, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x06, 0x00, 0x00, 0x00, 0x15, 0x00, 0x04, 0x00,
- 0x11, 0x00, 0x00, 0x00, 0x08, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x1d, 0x00, 0x03, 0x00,
- 0x12, 0x00, 0x00, 0x00, 0x11, 0x00, 0x00, 0x00, 0x1e, 0x00, 0x03, 0x00, 0x13, 0x00, 0x00, 0x00,
- 0x12, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00, 0x14, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00,
- 0x13, 0x00, 0x00, 0x00, 0x3b, 0x00, 0x04, 0x00, 0x14, 0x00, 0x00, 0x00, 0x15, 0x00, 0x00, 0x00,
- 0x02, 0x00, 0x00, 0x00, 0x15, 0x00, 0x04, 0x00, 0x17, 0x00, 0x00, 0x00, 0x20, 0x00, 0x00, 0x00,
- 0x01, 0x00, 0x00, 0x00, 0x14, 0x00, 0x02, 0x00, 0x1a, 0x00, 0x00, 0x00, 0x15, 0x00, 0x04, 0x00,
- 0x1e, 0x00, 0x00, 0x00, 0x10, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x1d, 0x00, 0x03, 0x00,
- 0x1f, 0x00, 0x00, 0x00, 0x1e, 0x00, 0x00, 0x00, 0x1e, 0x00, 0x03, 0x00, 0x20, 0x00, 0x00, 0x00,
- 0x1f, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00, 0x21, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00,
- 0x20, 0x00, 0x00, 0x00, 0x3b, 0x00, 0x04, 0x00, 0x21, 0x00, 0x00, 0x00, 0x22, 0x00, 0x00, 0x00,
- 0x02, 0x00, 0x00, 0x00, 0x2b, 0x00, 0x04, 0x00, 0x17, 0x00, 0x00, 0x00, 0x23, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00, 0x26, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00,
- 0x11, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00, 0x2a, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00,
- 0x1e, 0x00, 0x00, 0x00, 0x2b, 0x00, 0x04, 0x00, 0x06, 0x00, 0x00, 0x00, 0x2c, 0x00, 0x00, 0x00,
- 0x00, 0x04, 0x00, 0x00, 0x2b, 0x00, 0x04, 0x00, 0x06, 0x00, 0x00, 0x00, 0x2d, 0x00, 0x00, 0x00,
- 0x01, 0x00, 0x00, 0x00, 0x2c, 0x00, 0x06, 0x00, 0x09, 0x00, 0x00, 0x00, 0x2e, 0x00, 0x00, 0x00,
- 0x2c, 0x00, 0x00, 0x00, 0x2d, 0x00, 0x00, 0x00, 0x2d, 0x00, 0x00, 0x00, 0x36, 0x00, 0x05, 0x00,
- 0x02, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00,
- 0xf8, 0x00, 0x02, 0x00, 0x05, 0x00, 0x00, 0x00, 0x3b, 0x00, 0x04, 0x00, 0x07, 0x00, 0x00, 0x00,
- 0x08, 0x00, 0x00, 0x00, 0x07, 0x00, 0x00, 0x00, 0x41, 0x00, 0x05, 0x00, 0x0d, 0x00, 0x00, 0x00,
- 0x0e, 0x00, 0x00, 0x00, 0x0b, 0x00, 0x00, 0x00, 0x0c, 0x00, 0x00, 0x00, 0x3d, 0x00, 0x04, 0x00,
- 0x06, 0x00, 0x00, 0x00, 0x0f, 0x00, 0x00, 0x00, 0x0e, 0x00, 0x00, 0x00, 0x3e, 0x00, 0x03, 0x00,
- 0x08, 0x00, 0x00, 0x00, 0x0f, 0x00, 0x00, 0x00, 0x3d, 0x00, 0x04, 0x00, 0x06, 0x00, 0x00, 0x00,
- 0x10, 0x00, 0x00, 0x00, 0x08, 0x00, 0x00, 0x00, 0x44, 0x00, 0x05, 0x00, 0x06, 0x00, 0x00, 0x00,
- 0x16, 0x00, 0x00, 0x00, 0x15, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x7c, 0x00, 0x04, 0x00,
- 0x17, 0x00, 0x00, 0x00, 0x18, 0x00, 0x00, 0x00, 0x16, 0x00, 0x00, 0x00, 0x7c, 0x00, 0x04, 0x00,
- 0x06, 0x00, 0x00, 0x00, 0x19, 0x00, 0x00, 0x00, 0x18, 0x00, 0x00, 0x00, 0xb0, 0x00, 0x05, 0x00,
- 0x1a, 0x00, 0x00, 0x00, 0x1b, 0x00, 0x00, 0x00, 0x10, 0x00, 0x00, 0x00, 0x19, 0x00, 0x00, 0x00,
- 0xf7, 0x00, 0x03, 0x00, 0x1d, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xfa, 0x00, 0x04, 0x00,
- 0x1b, 0x00, 0x00, 0x00, 0x1c, 0x00, 0x00, 0x00, 0x1d, 0x00, 0x00, 0x00, 0xf8, 0x00, 0x02, 0x00,
- 0x1c, 0x00, 0x00, 0x00, 0x3d, 0x00, 0x04, 0x00, 0x06, 0x00, 0x00, 0x00, 0x24, 0x00, 0x00, 0x00,
- 0x08, 0x00, 0x00, 0x00, 0x3d, 0x00, 0x04, 0x00, 0x06, 0x00, 0x00, 0x00, 0x25, 0x00, 0x00, 0x00,
- 0x08, 0x00, 0x00, 0x00, 0x41, 0x00, 0x06, 0x00, 0x26, 0x00, 0x00, 0x00, 0x27, 0x00, 0x00, 0x00,
- 0x15, 0x00, 0x00, 0x00, 0x23, 0x00, 0x00, 0x00, 0x25, 0x00, 0x00, 0x00, 0x3d, 0x00, 0x04, 0x00,
- 0x11, 0x00, 0x00, 0x00, 0x28, 0x00, 0x00, 0x00, 0x27, 0x00, 0x00, 0x00, 0x71, 0x00, 0x04, 0x00,
- 0x1e, 0x00, 0x00, 0x00, 0x29, 0x00, 0x00, 0x00, 0x28, 0x00, 0x00, 0x00, 0x41, 0x00, 0x06, 0x00,
- 0x2a, 0x00, 0x00, 0x00, 0x2b, 0x00, 0x00, 0x00, 0x22, 0x00, 0x00, 0x00, 0x23, 0x00, 0x00, 0x00,
- 0x24, 0x00, 0x00, 0x00, 0x3e, 0x00, 0x03, 0x00, 0x2b, 0x00, 0x00, 0x00, 0x29, 0x00, 0x00, 0x00,
- 0xf9, 0x00, 0x02, 0x00, 0x1d, 0x00, 0x00, 0x00, 0xf8, 0x00, 0x02, 0x00, 0x1d, 0x00, 0x00, 0x00,
- 0xfd, 0x00, 0x01, 0x00, 0x38, 0x00, 0x01, 0x00,
-};
-
-// Quad indexed SPIR-V module. Generated from the "shaders/" directory.
-constexpr u8 QUAD_INDEXED_SPV[] = {
- 0x03, 0x02, 0x23, 0x07, 0x00, 0x00, 0x01, 0x00, 0x07, 0x00, 0x08, 0x00, 0x7c, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x11, 0x00, 0x02, 0x00, 0x01, 0x00, 0x00, 0x00, 0x0b, 0x00, 0x06, 0x00,
- 0x01, 0x00, 0x00, 0x00, 0x47, 0x4c, 0x53, 0x4c, 0x2e, 0x73, 0x74, 0x64, 0x2e, 0x34, 0x35, 0x30,
- 0x00, 0x00, 0x00, 0x00, 0x0e, 0x00, 0x03, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00,
- 0x0f, 0x00, 0x06, 0x00, 0x05, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x00, 0x6d, 0x61, 0x69, 0x6e,
- 0x00, 0x00, 0x00, 0x00, 0x0c, 0x00, 0x00, 0x00, 0x10, 0x00, 0x06, 0x00, 0x04, 0x00, 0x00, 0x00,
- 0x11, 0x00, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00,
- 0x47, 0x00, 0x04, 0x00, 0x0c, 0x00, 0x00, 0x00, 0x0b, 0x00, 0x00, 0x00, 0x1c, 0x00, 0x00, 0x00,
- 0x47, 0x00, 0x04, 0x00, 0x15, 0x00, 0x00, 0x00, 0x06, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x00,
- 0x48, 0x00, 0x04, 0x00, 0x16, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x19, 0x00, 0x00, 0x00,
- 0x48, 0x00, 0x05, 0x00, 0x16, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x23, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x47, 0x00, 0x03, 0x00, 0x16, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00,
- 0x47, 0x00, 0x04, 0x00, 0x18, 0x00, 0x00, 0x00, 0x22, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0x47, 0x00, 0x04, 0x00, 0x18, 0x00, 0x00, 0x00, 0x21, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00,
- 0x48, 0x00, 0x05, 0x00, 0x22, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x23, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x48, 0x00, 0x05, 0x00, 0x22, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00,
- 0x23, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x00, 0x47, 0x00, 0x03, 0x00, 0x22, 0x00, 0x00, 0x00,
- 0x02, 0x00, 0x00, 0x00, 0x47, 0x00, 0x04, 0x00, 0x56, 0x00, 0x00, 0x00, 0x06, 0x00, 0x00, 0x00,
- 0x04, 0x00, 0x00, 0x00, 0x48, 0x00, 0x04, 0x00, 0x57, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0x18, 0x00, 0x00, 0x00, 0x48, 0x00, 0x05, 0x00, 0x57, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0x23, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x47, 0x00, 0x03, 0x00, 0x57, 0x00, 0x00, 0x00,
- 0x03, 0x00, 0x00, 0x00, 0x47, 0x00, 0x04, 0x00, 0x59, 0x00, 0x00, 0x00, 0x22, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x47, 0x00, 0x04, 0x00, 0x59, 0x00, 0x00, 0x00, 0x21, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x47, 0x00, 0x04, 0x00, 0x72, 0x00, 0x00, 0x00, 0x0b, 0x00, 0x00, 0x00,
- 0x19, 0x00, 0x00, 0x00, 0x13, 0x00, 0x02, 0x00, 0x02, 0x00, 0x00, 0x00, 0x21, 0x00, 0x03, 0x00,
- 0x03, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00, 0x15, 0x00, 0x04, 0x00, 0x06, 0x00, 0x00, 0x00,
- 0x20, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00, 0x07, 0x00, 0x00, 0x00,
- 0x07, 0x00, 0x00, 0x00, 0x06, 0x00, 0x00, 0x00, 0x15, 0x00, 0x04, 0x00, 0x09, 0x00, 0x00, 0x00,
- 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x17, 0x00, 0x04, 0x00, 0x0a, 0x00, 0x00, 0x00,
- 0x09, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00, 0x0b, 0x00, 0x00, 0x00,
- 0x01, 0x00, 0x00, 0x00, 0x0a, 0x00, 0x00, 0x00, 0x3b, 0x00, 0x04, 0x00, 0x0b, 0x00, 0x00, 0x00,
- 0x0c, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x2b, 0x00, 0x04, 0x00, 0x09, 0x00, 0x00, 0x00,
- 0x0d, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00, 0x0e, 0x00, 0x00, 0x00,
- 0x01, 0x00, 0x00, 0x00, 0x09, 0x00, 0x00, 0x00, 0x2b, 0x00, 0x04, 0x00, 0x06, 0x00, 0x00, 0x00,
- 0x13, 0x00, 0x00, 0x00, 0x06, 0x00, 0x00, 0x00, 0x1d, 0x00, 0x03, 0x00, 0x15, 0x00, 0x00, 0x00,
- 0x09, 0x00, 0x00, 0x00, 0x1e, 0x00, 0x03, 0x00, 0x16, 0x00, 0x00, 0x00, 0x15, 0x00, 0x00, 0x00,
- 0x20, 0x00, 0x04, 0x00, 0x17, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00, 0x16, 0x00, 0x00, 0x00,
- 0x3b, 0x00, 0x04, 0x00, 0x17, 0x00, 0x00, 0x00, 0x18, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00,
- 0x14, 0x00, 0x02, 0x00, 0x1b, 0x00, 0x00, 0x00, 0x2b, 0x00, 0x04, 0x00, 0x06, 0x00, 0x00, 0x00,
- 0x21, 0x00, 0x00, 0x00, 0x08, 0x00, 0x00, 0x00, 0x1e, 0x00, 0x04, 0x00, 0x22, 0x00, 0x00, 0x00,
- 0x09, 0x00, 0x00, 0x00, 0x06, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00, 0x23, 0x00, 0x00, 0x00,
- 0x09, 0x00, 0x00, 0x00, 0x22, 0x00, 0x00, 0x00, 0x3b, 0x00, 0x04, 0x00, 0x23, 0x00, 0x00, 0x00,
- 0x24, 0x00, 0x00, 0x00, 0x09, 0x00, 0x00, 0x00, 0x2b, 0x00, 0x04, 0x00, 0x06, 0x00, 0x00, 0x00,
- 0x25, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00, 0x26, 0x00, 0x00, 0x00,
- 0x09, 0x00, 0x00, 0x00, 0x06, 0x00, 0x00, 0x00, 0x2b, 0x00, 0x04, 0x00, 0x06, 0x00, 0x00, 0x00,
- 0x2b, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00, 0x2b, 0x00, 0x04, 0x00, 0x09, 0x00, 0x00, 0x00,
- 0x3b, 0x00, 0x00, 0x00, 0x06, 0x00, 0x00, 0x00, 0x2b, 0x00, 0x04, 0x00, 0x06, 0x00, 0x00, 0x00,
- 0x3f, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x00, 0x1c, 0x00, 0x04, 0x00, 0x41, 0x00, 0x00, 0x00,
- 0x06, 0x00, 0x00, 0x00, 0x3b, 0x00, 0x00, 0x00, 0x2b, 0x00, 0x04, 0x00, 0x06, 0x00, 0x00, 0x00,
- 0x42, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x2b, 0x00, 0x04, 0x00, 0x06, 0x00, 0x00, 0x00,
- 0x43, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00, 0x2c, 0x00, 0x09, 0x00, 0x41, 0x00, 0x00, 0x00,
- 0x44, 0x00, 0x00, 0x00, 0x42, 0x00, 0x00, 0x00, 0x25, 0x00, 0x00, 0x00, 0x2b, 0x00, 0x00, 0x00,
- 0x42, 0x00, 0x00, 0x00, 0x2b, 0x00, 0x00, 0x00, 0x43, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00,
- 0x46, 0x00, 0x00, 0x00, 0x07, 0x00, 0x00, 0x00, 0x41, 0x00, 0x00, 0x00, 0x1d, 0x00, 0x03, 0x00,
- 0x56, 0x00, 0x00, 0x00, 0x09, 0x00, 0x00, 0x00, 0x1e, 0x00, 0x03, 0x00, 0x57, 0x00, 0x00, 0x00,
- 0x56, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00, 0x58, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00,
- 0x57, 0x00, 0x00, 0x00, 0x3b, 0x00, 0x04, 0x00, 0x58, 0x00, 0x00, 0x00, 0x59, 0x00, 0x00, 0x00,
- 0x02, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00, 0x5b, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00,
- 0x09, 0x00, 0x00, 0x00, 0x20, 0x00, 0x04, 0x00, 0x69, 0x00, 0x00, 0x00, 0x09, 0x00, 0x00, 0x00,
- 0x09, 0x00, 0x00, 0x00, 0x2b, 0x00, 0x04, 0x00, 0x09, 0x00, 0x00, 0x00, 0x70, 0x00, 0x00, 0x00,
- 0x00, 0x04, 0x00, 0x00, 0x2b, 0x00, 0x04, 0x00, 0x09, 0x00, 0x00, 0x00, 0x71, 0x00, 0x00, 0x00,
- 0x01, 0x00, 0x00, 0x00, 0x2c, 0x00, 0x06, 0x00, 0x0a, 0x00, 0x00, 0x00, 0x72, 0x00, 0x00, 0x00,
- 0x70, 0x00, 0x00, 0x00, 0x71, 0x00, 0x00, 0x00, 0x71, 0x00, 0x00, 0x00, 0x36, 0x00, 0x05, 0x00,
- 0x02, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00,
- 0xf8, 0x00, 0x02, 0x00, 0x05, 0x00, 0x00, 0x00, 0x3b, 0x00, 0x04, 0x00, 0x46, 0x00, 0x00, 0x00,
- 0x47, 0x00, 0x00, 0x00, 0x07, 0x00, 0x00, 0x00, 0xf9, 0x00, 0x02, 0x00, 0x74, 0x00, 0x00, 0x00,
- 0xf8, 0x00, 0x02, 0x00, 0x74, 0x00, 0x00, 0x00, 0xf6, 0x00, 0x04, 0x00, 0x73, 0x00, 0x00, 0x00,
- 0x76, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xf9, 0x00, 0x02, 0x00, 0x75, 0x00, 0x00, 0x00,
- 0xf8, 0x00, 0x02, 0x00, 0x75, 0x00, 0x00, 0x00, 0x41, 0x00, 0x05, 0x00, 0x0e, 0x00, 0x00, 0x00,
- 0x0f, 0x00, 0x00, 0x00, 0x0c, 0x00, 0x00, 0x00, 0x0d, 0x00, 0x00, 0x00, 0x3d, 0x00, 0x04, 0x00,
- 0x09, 0x00, 0x00, 0x00, 0x10, 0x00, 0x00, 0x00, 0x0f, 0x00, 0x00, 0x00, 0x7c, 0x00, 0x04, 0x00,
- 0x06, 0x00, 0x00, 0x00, 0x11, 0x00, 0x00, 0x00, 0x10, 0x00, 0x00, 0x00, 0x84, 0x00, 0x05, 0x00,
- 0x06, 0x00, 0x00, 0x00, 0x14, 0x00, 0x00, 0x00, 0x11, 0x00, 0x00, 0x00, 0x13, 0x00, 0x00, 0x00,
- 0x44, 0x00, 0x05, 0x00, 0x09, 0x00, 0x00, 0x00, 0x19, 0x00, 0x00, 0x00, 0x18, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x7c, 0x00, 0x04, 0x00, 0x06, 0x00, 0x00, 0x00, 0x1a, 0x00, 0x00, 0x00,
- 0x19, 0x00, 0x00, 0x00, 0xaf, 0x00, 0x05, 0x00, 0x1b, 0x00, 0x00, 0x00, 0x1c, 0x00, 0x00, 0x00,
- 0x14, 0x00, 0x00, 0x00, 0x1a, 0x00, 0x00, 0x00, 0xf7, 0x00, 0x03, 0x00, 0x1e, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0xfa, 0x00, 0x04, 0x00, 0x1c, 0x00, 0x00, 0x00, 0x1d, 0x00, 0x00, 0x00,
- 0x1e, 0x00, 0x00, 0x00, 0xf8, 0x00, 0x02, 0x00, 0x1d, 0x00, 0x00, 0x00, 0xf9, 0x00, 0x02, 0x00,
- 0x73, 0x00, 0x00, 0x00, 0xf8, 0x00, 0x02, 0x00, 0x1e, 0x00, 0x00, 0x00, 0x41, 0x00, 0x05, 0x00,
- 0x26, 0x00, 0x00, 0x00, 0x27, 0x00, 0x00, 0x00, 0x24, 0x00, 0x00, 0x00, 0x25, 0x00, 0x00, 0x00,
- 0x3d, 0x00, 0x04, 0x00, 0x06, 0x00, 0x00, 0x00, 0x28, 0x00, 0x00, 0x00, 0x27, 0x00, 0x00, 0x00,
- 0xc4, 0x00, 0x05, 0x00, 0x06, 0x00, 0x00, 0x00, 0x29, 0x00, 0x00, 0x00, 0x21, 0x00, 0x00, 0x00,
- 0x28, 0x00, 0x00, 0x00, 0x82, 0x00, 0x05, 0x00, 0x06, 0x00, 0x00, 0x00, 0x2e, 0x00, 0x00, 0x00,
- 0x2b, 0x00, 0x00, 0x00, 0x28, 0x00, 0x00, 0x00, 0xc4, 0x00, 0x05, 0x00, 0x06, 0x00, 0x00, 0x00,
- 0x31, 0x00, 0x00, 0x00, 0x25, 0x00, 0x00, 0x00, 0x2e, 0x00, 0x00, 0x00, 0x82, 0x00, 0x05, 0x00,
- 0x06, 0x00, 0x00, 0x00, 0x32, 0x00, 0x00, 0x00, 0x31, 0x00, 0x00, 0x00, 0x25, 0x00, 0x00, 0x00,
- 0xf9, 0x00, 0x02, 0x00, 0x35, 0x00, 0x00, 0x00, 0xf8, 0x00, 0x02, 0x00, 0x35, 0x00, 0x00, 0x00,
- 0xf5, 0x00, 0x07, 0x00, 0x09, 0x00, 0x00, 0x00, 0x7b, 0x00, 0x00, 0x00, 0x0d, 0x00, 0x00, 0x00,
- 0x1e, 0x00, 0x00, 0x00, 0x6f, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x00, 0xb0, 0x00, 0x05, 0x00,
- 0x1b, 0x00, 0x00, 0x00, 0x3c, 0x00, 0x00, 0x00, 0x7b, 0x00, 0x00, 0x00, 0x3b, 0x00, 0x00, 0x00,
- 0xf6, 0x00, 0x04, 0x00, 0x37, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0xfa, 0x00, 0x04, 0x00, 0x3c, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x00, 0x37, 0x00, 0x00, 0x00,
- 0xf8, 0x00, 0x02, 0x00, 0x36, 0x00, 0x00, 0x00, 0x84, 0x00, 0x05, 0x00, 0x06, 0x00, 0x00, 0x00,
- 0x40, 0x00, 0x00, 0x00, 0x11, 0x00, 0x00, 0x00, 0x3f, 0x00, 0x00, 0x00, 0x3e, 0x00, 0x03, 0x00,
- 0x47, 0x00, 0x00, 0x00, 0x44, 0x00, 0x00, 0x00, 0x41, 0x00, 0x05, 0x00, 0x07, 0x00, 0x00, 0x00,
- 0x48, 0x00, 0x00, 0x00, 0x47, 0x00, 0x00, 0x00, 0x7b, 0x00, 0x00, 0x00, 0x3d, 0x00, 0x04, 0x00,
- 0x06, 0x00, 0x00, 0x00, 0x49, 0x00, 0x00, 0x00, 0x48, 0x00, 0x00, 0x00, 0x80, 0x00, 0x05, 0x00,
- 0x06, 0x00, 0x00, 0x00, 0x4a, 0x00, 0x00, 0x00, 0x40, 0x00, 0x00, 0x00, 0x49, 0x00, 0x00, 0x00,
- 0xc3, 0x00, 0x05, 0x00, 0x06, 0x00, 0x00, 0x00, 0x4e, 0x00, 0x00, 0x00, 0x4a, 0x00, 0x00, 0x00,
- 0x2e, 0x00, 0x00, 0x00, 0xc7, 0x00, 0x05, 0x00, 0x06, 0x00, 0x00, 0x00, 0x52, 0x00, 0x00, 0x00,
- 0x4a, 0x00, 0x00, 0x00, 0x32, 0x00, 0x00, 0x00, 0x84, 0x00, 0x05, 0x00, 0x06, 0x00, 0x00, 0x00,
- 0x54, 0x00, 0x00, 0x00, 0x52, 0x00, 0x00, 0x00, 0x29, 0x00, 0x00, 0x00, 0x41, 0x00, 0x06, 0x00,
- 0x5b, 0x00, 0x00, 0x00, 0x5c, 0x00, 0x00, 0x00, 0x59, 0x00, 0x00, 0x00, 0x42, 0x00, 0x00, 0x00,
- 0x4e, 0x00, 0x00, 0x00, 0x3d, 0x00, 0x04, 0x00, 0x09, 0x00, 0x00, 0x00, 0x5d, 0x00, 0x00, 0x00,
- 0x5c, 0x00, 0x00, 0x00, 0xcb, 0x00, 0x06, 0x00, 0x09, 0x00, 0x00, 0x00, 0x62, 0x00, 0x00, 0x00,
- 0x5d, 0x00, 0x00, 0x00, 0x54, 0x00, 0x00, 0x00, 0x29, 0x00, 0x00, 0x00, 0x7c, 0x00, 0x04, 0x00,
- 0x09, 0x00, 0x00, 0x00, 0x65, 0x00, 0x00, 0x00, 0x14, 0x00, 0x00, 0x00, 0x80, 0x00, 0x05, 0x00,
- 0x09, 0x00, 0x00, 0x00, 0x67, 0x00, 0x00, 0x00, 0x65, 0x00, 0x00, 0x00, 0x7b, 0x00, 0x00, 0x00,
- 0x41, 0x00, 0x05, 0x00, 0x69, 0x00, 0x00, 0x00, 0x6a, 0x00, 0x00, 0x00, 0x24, 0x00, 0x00, 0x00,
- 0x42, 0x00, 0x00, 0x00, 0x3d, 0x00, 0x04, 0x00, 0x09, 0x00, 0x00, 0x00, 0x6b, 0x00, 0x00, 0x00,
- 0x6a, 0x00, 0x00, 0x00, 0x80, 0x00, 0x05, 0x00, 0x09, 0x00, 0x00, 0x00, 0x6c, 0x00, 0x00, 0x00,
- 0x62, 0x00, 0x00, 0x00, 0x6b, 0x00, 0x00, 0x00, 0x41, 0x00, 0x06, 0x00, 0x5b, 0x00, 0x00, 0x00,
- 0x6d, 0x00, 0x00, 0x00, 0x18, 0x00, 0x00, 0x00, 0x42, 0x00, 0x00, 0x00, 0x67, 0x00, 0x00, 0x00,
- 0x3e, 0x00, 0x03, 0x00, 0x6d, 0x00, 0x00, 0x00, 0x6c, 0x00, 0x00, 0x00, 0x80, 0x00, 0x05, 0x00,
- 0x09, 0x00, 0x00, 0x00, 0x6f, 0x00, 0x00, 0x00, 0x7b, 0x00, 0x00, 0x00, 0x25, 0x00, 0x00, 0x00,
- 0xf9, 0x00, 0x02, 0x00, 0x35, 0x00, 0x00, 0x00, 0xf8, 0x00, 0x02, 0x00, 0x37, 0x00, 0x00, 0x00,
- 0xf9, 0x00, 0x02, 0x00, 0x73, 0x00, 0x00, 0x00, 0xf8, 0x00, 0x02, 0x00, 0x76, 0x00, 0x00, 0x00,
- 0xf9, 0x00, 0x02, 0x00, 0x74, 0x00, 0x00, 0x00, 0xf8, 0x00, 0x02, 0x00, 0x73, 0x00, 0x00, 0x00,
- 0xfd, 0x00, 0x01, 0x00, 0x38, 0x00, 0x01, 0x00,
-};
-
std::array<VkDescriptorSetLayoutBinding, 2> BuildInputOutputDescriptorSetBindings() {
return {{
{
@@ -381,8 +89,8 @@ VkDescriptorUpdateTemplateEntryKHR BuildInputOutputDescriptorUpdateTemplate() {
VKComputePass::VKComputePass(const VKDevice& device, VKDescriptorPool& descriptor_pool,
vk::Span<VkDescriptorSetLayoutBinding> bindings,
vk::Span<VkDescriptorUpdateTemplateEntryKHR> templates,
- vk::Span<VkPushConstantRange> push_constants, std::size_t code_size,
- const u8* code) {
+ vk::Span<VkPushConstantRange> push_constants,
+ std::span<const u32> code) {
descriptor_set_layout = device.GetLogical().CreateDescriptorSetLayout({
.sType = VK_STRUCTURE_TYPE_DESCRIPTOR_SET_LAYOUT_CREATE_INFO,
.pNext = nullptr,
@@ -390,7 +98,6 @@ VKComputePass::VKComputePass(const VKDevice& device, VKDescriptorPool& descripto
.bindingCount = bindings.size(),
.pBindings = bindings.data(),
});
-
layout = device.GetLogical().CreatePipelineLayout({
.sType = VK_STRUCTURE_TYPE_PIPELINE_LAYOUT_CREATE_INFO,
.pNext = nullptr,
@@ -400,7 +107,6 @@ VKComputePass::VKComputePass(const VKDevice& device, VKDescriptorPool& descripto
.pushConstantRangeCount = push_constants.size(),
.pPushConstantRanges = push_constants.data(),
});
-
if (!templates.empty()) {
descriptor_template = device.GetLogical().CreateDescriptorUpdateTemplateKHR({
.sType = VK_STRUCTURE_TYPE_DESCRIPTOR_UPDATE_TEMPLATE_CREATE_INFO_KHR,
@@ -417,18 +123,13 @@ VKComputePass::VKComputePass(const VKDevice& device, VKDescriptorPool& descripto
descriptor_allocator.emplace(descriptor_pool, *descriptor_set_layout);
}
-
- auto code_copy = std::make_unique<u32[]>(code_size / sizeof(u32) + 1);
- std::memcpy(code_copy.get(), code, code_size);
-
module = device.GetLogical().CreateShaderModule({
.sType = VK_STRUCTURE_TYPE_SHADER_MODULE_CREATE_INFO,
.pNext = nullptr,
.flags = 0,
- .codeSize = code_size,
- .pCode = code_copy.get(),
+ .codeSize = static_cast<u32>(code.size_bytes()),
+ .pCode = code.data(),
});
-
pipeline = device.GetLogical().CreateComputePipeline({
.sType = VK_STRUCTURE_TYPE_COMPUTE_PIPELINE_CREATE_INFO,
.pNext = nullptr,
@@ -461,15 +162,15 @@ VkDescriptorSet VKComputePass::CommitDescriptorSet(
return set;
}
-QuadArrayPass::QuadArrayPass(const VKDevice& device, VKScheduler& scheduler,
- VKDescriptorPool& descriptor_pool,
- VKStagingBufferPool& staging_buffer_pool,
- VKUpdateDescriptorQueue& update_descriptor_queue)
- : VKComputePass(device, descriptor_pool, BuildQuadArrayPassDescriptorSetLayoutBinding(),
+QuadArrayPass::QuadArrayPass(const VKDevice& device_, VKScheduler& scheduler_,
+ VKDescriptorPool& descriptor_pool_,
+ VKStagingBufferPool& staging_buffer_pool_,
+ VKUpdateDescriptorQueue& update_descriptor_queue_)
+ : VKComputePass(device_, descriptor_pool_, BuildQuadArrayPassDescriptorSetLayoutBinding(),
BuildQuadArrayPassDescriptorUpdateTemplateEntry(),
- BuildComputePushConstantRange(sizeof(u32)), std::size(quad_array), quad_array),
- scheduler{scheduler}, staging_buffer_pool{staging_buffer_pool},
- update_descriptor_queue{update_descriptor_queue} {}
+ BuildComputePushConstantRange(sizeof(u32)), VULKAN_QUAD_ARRAY_COMP_SPV),
+ scheduler{scheduler_}, staging_buffer_pool{staging_buffer_pool_},
+ update_descriptor_queue{update_descriptor_queue_} {}
QuadArrayPass::~QuadArrayPass() = default;
@@ -510,14 +211,13 @@ std::pair<VkBuffer, VkDeviceSize> QuadArrayPass::Assemble(u32 num_vertices, u32
return {*buffer.handle, 0};
}
-Uint8Pass::Uint8Pass(const VKDevice& device, VKScheduler& scheduler,
- VKDescriptorPool& descriptor_pool, VKStagingBufferPool& staging_buffer_pool,
- VKUpdateDescriptorQueue& update_descriptor_queue)
+Uint8Pass::Uint8Pass(const VKDevice& device, VKScheduler& scheduler_,
+ VKDescriptorPool& descriptor_pool, VKStagingBufferPool& staging_buffer_pool_,
+ VKUpdateDescriptorQueue& update_descriptor_queue_)
: VKComputePass(device, descriptor_pool, BuildInputOutputDescriptorSetBindings(),
- BuildInputOutputDescriptorUpdateTemplate(), {}, std::size(uint8_pass),
- uint8_pass),
- scheduler{scheduler}, staging_buffer_pool{staging_buffer_pool},
- update_descriptor_queue{update_descriptor_queue} {}
+ BuildInputOutputDescriptorUpdateTemplate(), {}, VULKAN_UINT8_COMP_SPV),
+ scheduler{scheduler_}, staging_buffer_pool{staging_buffer_pool_},
+ update_descriptor_queue{update_descriptor_queue_} {}
Uint8Pass::~Uint8Pass() = default;
@@ -555,16 +255,15 @@ std::pair<VkBuffer, u64> Uint8Pass::Assemble(u32 num_vertices, VkBuffer src_buff
return {*buffer.handle, 0};
}
-QuadIndexedPass::QuadIndexedPass(const VKDevice& device, VKScheduler& scheduler,
- VKDescriptorPool& descriptor_pool,
- VKStagingBufferPool& staging_buffer_pool,
- VKUpdateDescriptorQueue& update_descriptor_queue)
- : VKComputePass(device, descriptor_pool, BuildInputOutputDescriptorSetBindings(),
+QuadIndexedPass::QuadIndexedPass(const VKDevice& device_, VKScheduler& scheduler_,
+ VKDescriptorPool& descriptor_pool_,
+ VKStagingBufferPool& staging_buffer_pool_,
+ VKUpdateDescriptorQueue& update_descriptor_queue_)
+ : VKComputePass(device_, descriptor_pool_, BuildInputOutputDescriptorSetBindings(),
BuildInputOutputDescriptorUpdateTemplate(),
- BuildComputePushConstantRange(sizeof(u32) * 2), std::size(QUAD_INDEXED_SPV),
- QUAD_INDEXED_SPV),
- scheduler{scheduler}, staging_buffer_pool{staging_buffer_pool},
- update_descriptor_queue{update_descriptor_queue} {}
+ BuildComputePushConstantRange(sizeof(u32) * 2), VULKAN_QUAD_INDEXED_COMP_SPV),
+ scheduler{scheduler_}, staging_buffer_pool{staging_buffer_pool_},
+ update_descriptor_queue{update_descriptor_queue_} {}
QuadIndexedPass::~QuadIndexedPass() = default;
diff --git a/src/video_core/renderer_vulkan/vk_compute_pass.h b/src/video_core/renderer_vulkan/vk_compute_pass.h
index acc94f27e..abdf61e2c 100644
--- a/src/video_core/renderer_vulkan/vk_compute_pass.h
+++ b/src/video_core/renderer_vulkan/vk_compute_pass.h
@@ -5,6 +5,7 @@
#pragma once
#include <optional>
+#include <span>
#include <utility>
#include "common/common_types.h"
@@ -24,8 +25,7 @@ public:
explicit VKComputePass(const VKDevice& device, VKDescriptorPool& descriptor_pool,
vk::Span<VkDescriptorSetLayoutBinding> bindings,
vk::Span<VkDescriptorUpdateTemplateEntryKHR> templates,
- vk::Span<VkPushConstantRange> push_constants, std::size_t code_size,
- const u8* code);
+ vk::Span<VkPushConstantRange> push_constants, std::span<const u32> code);
~VKComputePass();
protected:
@@ -43,10 +43,10 @@ private:
class QuadArrayPass final : public VKComputePass {
public:
- explicit QuadArrayPass(const VKDevice& device, VKScheduler& scheduler,
- VKDescriptorPool& descriptor_pool,
- VKStagingBufferPool& staging_buffer_pool,
- VKUpdateDescriptorQueue& update_descriptor_queue);
+ explicit QuadArrayPass(const VKDevice& device_, VKScheduler& scheduler_,
+ VKDescriptorPool& descriptor_pool_,
+ VKStagingBufferPool& staging_buffer_pool_,
+ VKUpdateDescriptorQueue& update_descriptor_queue_);
~QuadArrayPass();
std::pair<VkBuffer, VkDeviceSize> Assemble(u32 num_vertices, u32 first);
@@ -59,9 +59,10 @@ private:
class Uint8Pass final : public VKComputePass {
public:
- explicit Uint8Pass(const VKDevice& device, VKScheduler& scheduler,
- VKDescriptorPool& descriptor_pool, VKStagingBufferPool& staging_buffer_pool,
- VKUpdateDescriptorQueue& update_descriptor_queue);
+ explicit Uint8Pass(const VKDevice& device_, VKScheduler& scheduler_,
+ VKDescriptorPool& descriptor_pool_,
+ VKStagingBufferPool& staging_buffer_pool_,
+ VKUpdateDescriptorQueue& update_descriptor_queue_);
~Uint8Pass();
std::pair<VkBuffer, u64> Assemble(u32 num_vertices, VkBuffer src_buffer, u64 src_offset);
@@ -74,10 +75,10 @@ private:
class QuadIndexedPass final : public VKComputePass {
public:
- explicit QuadIndexedPass(const VKDevice& device, VKScheduler& scheduler,
- VKDescriptorPool& descriptor_pool,
- VKStagingBufferPool& staging_buffer_pool,
- VKUpdateDescriptorQueue& update_descriptor_queue);
+ explicit QuadIndexedPass(const VKDevice& device_, VKScheduler& scheduler_,
+ VKDescriptorPool& descriptor_pool_,
+ VKStagingBufferPool& staging_buffer_pool_,
+ VKUpdateDescriptorQueue& update_descriptor_queue_);
~QuadIndexedPass();
std::pair<VkBuffer, u64> Assemble(Tegra::Engines::Maxwell3D::Regs::IndexFormat index_format,
diff --git a/src/video_core/renderer_vulkan/vk_compute_pipeline.cpp b/src/video_core/renderer_vulkan/vk_compute_pipeline.cpp
index 9be72dc9b..62f44d6da 100644
--- a/src/video_core/renderer_vulkan/vk_compute_pipeline.cpp
+++ b/src/video_core/renderer_vulkan/vk_compute_pipeline.cpp
@@ -15,16 +15,16 @@
namespace Vulkan {
-VKComputePipeline::VKComputePipeline(const VKDevice& device, VKScheduler& scheduler,
- VKDescriptorPool& descriptor_pool,
- VKUpdateDescriptorQueue& update_descriptor_queue,
- const SPIRVShader& shader)
- : device{device}, scheduler{scheduler}, entries{shader.entries},
+VKComputePipeline::VKComputePipeline(const VKDevice& device_, VKScheduler& scheduler_,
+ VKDescriptorPool& descriptor_pool_,
+ VKUpdateDescriptorQueue& update_descriptor_queue_,
+ const SPIRVShader& shader_)
+ : device{device_}, scheduler{scheduler_}, entries{shader_.entries},
descriptor_set_layout{CreateDescriptorSetLayout()},
- descriptor_allocator{descriptor_pool, *descriptor_set_layout},
- update_descriptor_queue{update_descriptor_queue}, layout{CreatePipelineLayout()},
+ descriptor_allocator{descriptor_pool_, *descriptor_set_layout},
+ update_descriptor_queue{update_descriptor_queue_}, layout{CreatePipelineLayout()},
descriptor_template{CreateDescriptorUpdateTemplate()},
- shader_module{CreateShaderModule(shader.code)}, pipeline{CreatePipeline()} {}
+ shader_module{CreateShaderModule(shader_.code)}, pipeline{CreatePipeline()} {}
VKComputePipeline::~VKComputePipeline() = default;
diff --git a/src/video_core/renderer_vulkan/vk_compute_pipeline.h b/src/video_core/renderer_vulkan/vk_compute_pipeline.h
index 6e2f22a4a..49e2113a2 100644
--- a/src/video_core/renderer_vulkan/vk_compute_pipeline.h
+++ b/src/video_core/renderer_vulkan/vk_compute_pipeline.h
@@ -17,10 +17,10 @@ class VKUpdateDescriptorQueue;
class VKComputePipeline final {
public:
- explicit VKComputePipeline(const VKDevice& device, VKScheduler& scheduler,
- VKDescriptorPool& descriptor_pool,
- VKUpdateDescriptorQueue& update_descriptor_queue,
- const SPIRVShader& shader);
+ explicit VKComputePipeline(const VKDevice& device_, VKScheduler& scheduler_,
+ VKDescriptorPool& descriptor_pool_,
+ VKUpdateDescriptorQueue& update_descriptor_queue_,
+ const SPIRVShader& shader_);
~VKComputePipeline();
VkDescriptorSet CommitDescriptorSet();
diff --git a/src/video_core/renderer_vulkan/vk_device.cpp b/src/video_core/renderer_vulkan/vk_device.cpp
index f34ed6735..370a63f74 100644
--- a/src/video_core/renderer_vulkan/vk_device.cpp
+++ b/src/video_core/renderer_vulkan/vk_device.cpp
@@ -46,6 +46,7 @@ constexpr std::array REQUIRED_EXTENSIONS{
VK_KHR_DRIVER_PROPERTIES_EXTENSION_NAME,
VK_KHR_DESCRIPTOR_UPDATE_TEMPLATE_EXTENSION_NAME,
VK_KHR_TIMELINE_SEMAPHORE_EXTENSION_NAME,
+ VK_KHR_SAMPLER_MIRROR_CLAMP_TO_EDGE_EXTENSION_NAME,
VK_EXT_VERTEX_ATTRIBUTE_DIVISOR_EXTENSION_NAME,
VK_EXT_SHADER_SUBGROUP_BALLOT_EXTENSION_NAME,
VK_EXT_SHADER_SUBGROUP_VOTE_EXTENSION_NAME,
@@ -122,6 +123,7 @@ std::unordered_map<VkFormat, VkFormatProperties> GetFormatProperties(
VK_FORMAT_R16G16_UNORM,
VK_FORMAT_R16G16_SNORM,
VK_FORMAT_R16G16_SFLOAT,
+ VK_FORMAT_R16G16_SINT,
VK_FORMAT_R16_UNORM,
VK_FORMAT_R16_UINT,
VK_FORMAT_R8G8B8A8_SRGB,
@@ -161,18 +163,32 @@ std::unordered_map<VkFormat, VkFormatProperties> GetFormatProperties(
VK_FORMAT_BC2_SRGB_BLOCK,
VK_FORMAT_BC3_SRGB_BLOCK,
VK_FORMAT_BC7_SRGB_BLOCK,
+ VK_FORMAT_ASTC_4x4_UNORM_BLOCK,
VK_FORMAT_ASTC_4x4_SRGB_BLOCK,
- VK_FORMAT_ASTC_8x8_SRGB_BLOCK,
- VK_FORMAT_ASTC_8x5_SRGB_BLOCK,
+ VK_FORMAT_ASTC_5x4_UNORM_BLOCK,
VK_FORMAT_ASTC_5x4_SRGB_BLOCK,
VK_FORMAT_ASTC_5x5_UNORM_BLOCK,
VK_FORMAT_ASTC_5x5_SRGB_BLOCK,
- VK_FORMAT_ASTC_10x8_UNORM_BLOCK,
- VK_FORMAT_ASTC_10x8_SRGB_BLOCK,
+ VK_FORMAT_ASTC_6x5_UNORM_BLOCK,
+ VK_FORMAT_ASTC_6x5_SRGB_BLOCK,
VK_FORMAT_ASTC_6x6_UNORM_BLOCK,
VK_FORMAT_ASTC_6x6_SRGB_BLOCK,
+ VK_FORMAT_ASTC_8x5_UNORM_BLOCK,
+ VK_FORMAT_ASTC_8x5_SRGB_BLOCK,
+ VK_FORMAT_ASTC_8x6_UNORM_BLOCK,
+ VK_FORMAT_ASTC_8x6_SRGB_BLOCK,
+ VK_FORMAT_ASTC_8x8_UNORM_BLOCK,
+ VK_FORMAT_ASTC_8x8_SRGB_BLOCK,
+ VK_FORMAT_ASTC_10x5_UNORM_BLOCK,
+ VK_FORMAT_ASTC_10x5_SRGB_BLOCK,
+ VK_FORMAT_ASTC_10x6_UNORM_BLOCK,
+ VK_FORMAT_ASTC_10x6_SRGB_BLOCK,
+ VK_FORMAT_ASTC_10x8_UNORM_BLOCK,
+ VK_FORMAT_ASTC_10x8_SRGB_BLOCK,
VK_FORMAT_ASTC_10x10_UNORM_BLOCK,
VK_FORMAT_ASTC_10x10_SRGB_BLOCK,
+ VK_FORMAT_ASTC_12x10_UNORM_BLOCK,
+ VK_FORMAT_ASTC_12x10_SRGB_BLOCK,
VK_FORMAT_ASTC_12x12_UNORM_BLOCK,
VK_FORMAT_ASTC_12x12_SRGB_BLOCK,
VK_FORMAT_ASTC_8x6_UNORM_BLOCK,
@@ -192,7 +208,7 @@ std::unordered_map<VkFormat, VkFormatProperties> GetFormatProperties(
VKDevice::VKDevice(VkInstance instance_, u32 instance_version_, vk::PhysicalDevice physical_,
VkSurfaceKHR surface, const vk::InstanceDispatch& dld_)
- : dld{dld_}, physical{physical_}, properties{physical.GetProperties()},
+ : instance{instance_}, dld{dld_}, physical{physical_}, properties{physical.GetProperties()},
instance_version{instance_version_}, format_properties{GetFormatProperties(physical, dld)} {
SetupFamilies(surface);
SetupFeatures();
@@ -214,7 +230,7 @@ bool VKDevice::Create() {
features2.features = {
.robustBufferAccess = false,
.fullDrawIndexUint32 = false,
- .imageCubeArray = false,
+ .imageCubeArray = true,
.independentBlend = true,
.geometryShader = true,
.tessellationShader = true,
@@ -242,7 +258,7 @@ bool VKDevice::Create() {
.shaderTessellationAndGeometryPointSize = false,
.shaderImageGatherExtended = true,
.shaderStorageImageExtendedFormats = false,
- .shaderStorageImageMultisample = false,
+ .shaderStorageImageMultisample = true,
.shaderStorageImageReadWithoutFormat = is_formatless_image_load_supported,
.shaderStorageImageWriteWithoutFormat = true,
.shaderUniformBufferArrayDynamicIndexing = false,
@@ -268,7 +284,6 @@ bool VKDevice::Create() {
.variableMultisampleRate = false,
.inheritedQueries = false,
};
-
VkPhysicalDeviceTimelineSemaphoreFeaturesKHR timeline_semaphore{
.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_TIMELINE_SEMAPHORE_FEATURES_KHR,
.pNext = nullptr,
@@ -380,6 +395,20 @@ bool VKDevice::Create() {
LOG_INFO(Render_Vulkan, "Device doesn't support extended dynamic state");
}
+ VkPhysicalDeviceRobustness2FeaturesEXT robustness2;
+ if (ext_robustness2) {
+ robustness2 = {
+ .sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_ROBUSTNESS_2_FEATURES_EXT,
+ .pNext = nullptr,
+ .robustBufferAccess2 = false,
+ .robustImageAccess2 = true,
+ .nullDescriptor = true,
+ };
+ SetNext(next, robustness2);
+ } else {
+ LOG_INFO(Render_Vulkan, "Device doesn't support robustness2");
+ }
+
if (!ext_depth_range_unrestricted) {
LOG_INFO(Render_Vulkan, "Device doesn't support depth range unrestricted");
}
@@ -405,7 +434,14 @@ bool VKDevice::Create() {
}
CollectTelemetryParameters();
+ CollectToolingInfo();
+ if (ext_extended_dynamic_state && driver_id == VK_DRIVER_ID_MESA_RADV) {
+ LOG_WARNING(
+ Render_Vulkan,
+ "Blacklisting RADV for VK_EXT_extended_dynamic state, likely due to a bug in yuzu");
+ ext_extended_dynamic_state = false;
+ }
if (ext_extended_dynamic_state && IsRDNA(properties.deviceName, driver_id)) {
// AMD's proprietary driver supports VK_EXT_extended_dynamic_state but on RDNA devices it
// seems to cause stability issues
@@ -458,7 +494,7 @@ void VKDevice::ReportLoss() const {
LOG_CRITICAL(Render_Vulkan, "Device loss occured!");
// Wait for the log to flush and for Nsight Aftermath to dump the results
- std::this_thread::sleep_for(std::chrono::seconds{3});
+ std::this_thread::sleep_for(std::chrono::seconds{15});
}
void VKDevice::SaveShader(const std::vector<u32>& spirv) const {
@@ -491,14 +527,24 @@ bool VKDevice::IsOptimalAstcSupported(const VkPhysicalDeviceFeatures& features)
VK_FORMAT_FEATURE_BLIT_DST_BIT | VK_FORMAT_FEATURE_TRANSFER_SRC_BIT |
VK_FORMAT_FEATURE_TRANSFER_DST_BIT};
for (const auto format : astc_formats) {
- const auto format_properties{physical.GetFormatProperties(format)};
- if (!(format_properties.optimalTilingFeatures & format_feature_usage)) {
+ const auto physical_format_properties{physical.GetFormatProperties(format)};
+ if ((physical_format_properties.optimalTilingFeatures & format_feature_usage) == 0) {
return false;
}
}
return true;
}
+bool VKDevice::TestDepthStencilBlits() const {
+ static constexpr VkFormatFeatureFlags required_features =
+ VK_FORMAT_FEATURE_BLIT_SRC_BIT | VK_FORMAT_FEATURE_BLIT_DST_BIT;
+ const auto test_features = [](VkFormatProperties props) {
+ return (props.optimalTilingFeatures & required_features) == required_features;
+ };
+ return test_features(format_properties.at(VK_FORMAT_D32_SFLOAT_S8_UINT)) &&
+ test_features(format_properties.at(VK_FORMAT_D24_UNORM_S8_UINT));
+}
+
bool VKDevice::IsFormatSupported(VkFormat wanted_format, VkFormatFeatureFlags wanted_usage,
FormatType format_type) const {
const auto it = format_properties.find(wanted_format);
@@ -569,6 +615,7 @@ bool VKDevice::IsSuitable(vk::PhysicalDevice physical, VkSurfaceKHR surface) {
const auto features{physical.GetFeatures()};
const std::array feature_report = {
std::make_pair(features.vertexPipelineStoresAndAtomics, "vertexPipelineStoresAndAtomics"),
+ std::make_pair(features.imageCubeArray, "imageCubeArray"),
std::make_pair(features.independentBlend, "independentBlend"),
std::make_pair(features.depthClamp, "depthClamp"),
std::make_pair(features.samplerAnisotropy, "samplerAnisotropy"),
@@ -580,6 +627,7 @@ bool VKDevice::IsSuitable(vk::PhysicalDevice physical, VkSurfaceKHR surface) {
std::make_pair(features.occlusionQueryPrecise, "occlusionQueryPrecise"),
std::make_pair(features.fragmentStoresAndAtomics, "fragmentStoresAndAtomics"),
std::make_pair(features.shaderImageGatherExtended, "shaderImageGatherExtended"),
+ std::make_pair(features.shaderStorageImageMultisample, "shaderStorageImageMultisample"),
std::make_pair(features.shaderStorageImageWriteWithoutFormat,
"shaderStorageImageWriteWithoutFormat"),
};
@@ -608,6 +656,7 @@ std::vector<const char*> VKDevice::LoadExtensions() {
bool has_ext_transform_feedback{};
bool has_ext_custom_border_color{};
bool has_ext_extended_dynamic_state{};
+ bool has_ext_robustness2{};
for (const VkExtensionProperties& extension : physical.EnumerateDeviceExtensionProperties()) {
const auto test = [&](std::optional<std::reference_wrapper<bool>> status, const char* name,
bool push) {
@@ -627,11 +676,15 @@ std::vector<const char*> VKDevice::LoadExtensions() {
test(has_khr_shader_float16_int8, VK_KHR_SHADER_FLOAT16_INT8_EXTENSION_NAME, false);
test(ext_depth_range_unrestricted, VK_EXT_DEPTH_RANGE_UNRESTRICTED_EXTENSION_NAME, true);
test(ext_index_type_uint8, VK_EXT_INDEX_TYPE_UINT8_EXTENSION_NAME, true);
+ test(ext_sampler_filter_minmax, VK_EXT_SAMPLER_FILTER_MINMAX_EXTENSION_NAME, true);
test(ext_shader_viewport_index_layer, VK_EXT_SHADER_VIEWPORT_INDEX_LAYER_EXTENSION_NAME,
true);
+ test(ext_tooling_info, VK_EXT_TOOLING_INFO_EXTENSION_NAME, true);
+ test(ext_shader_stencil_export, VK_EXT_SHADER_STENCIL_EXPORT_EXTENSION_NAME, true);
test(has_ext_transform_feedback, VK_EXT_TRANSFORM_FEEDBACK_EXTENSION_NAME, false);
test(has_ext_custom_border_color, VK_EXT_CUSTOM_BORDER_COLOR_EXTENSION_NAME, false);
test(has_ext_extended_dynamic_state, VK_EXT_EXTENDED_DYNAMIC_STATE_EXTENSION_NAME, false);
+ test(has_ext_robustness2, VK_EXT_ROBUSTNESS_2_EXTENSION_NAME, false);
if (instance_version >= VK_API_VERSION_1_1) {
test(has_ext_subgroup_size_control, VK_EXT_SUBGROUP_SIZE_CONTROL_EXTENSION_NAME, false);
}
@@ -644,8 +697,8 @@ std::vector<const char*> VKDevice::LoadExtensions() {
VkPhysicalDeviceFeatures2KHR features;
features.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_FEATURES_2_KHR;
- VkPhysicalDeviceProperties2KHR properties;
- properties.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PROPERTIES_2_KHR;
+ VkPhysicalDeviceProperties2KHR physical_properties;
+ physical_properties.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PROPERTIES_2_KHR;
if (has_khr_shader_float16_int8) {
VkPhysicalDeviceFloat16Int8FeaturesKHR float16_int8_features;
@@ -670,8 +723,8 @@ std::vector<const char*> VKDevice::LoadExtensions() {
subgroup_properties.sType =
VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SUBGROUP_SIZE_CONTROL_PROPERTIES_EXT;
subgroup_properties.pNext = nullptr;
- properties.pNext = &subgroup_properties;
- physical.GetProperties2KHR(properties);
+ physical_properties.pNext = &subgroup_properties;
+ physical.GetProperties2KHR(physical_properties);
is_warp_potentially_bigger = subgroup_properties.maxSubgroupSize > GuestWarpSize;
@@ -695,8 +748,8 @@ std::vector<const char*> VKDevice::LoadExtensions() {
VkPhysicalDeviceTransformFeedbackPropertiesEXT tfb_properties;
tfb_properties.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_TRANSFORM_FEEDBACK_PROPERTIES_EXT;
tfb_properties.pNext = nullptr;
- properties.pNext = &tfb_properties;
- physical.GetProperties2KHR(properties);
+ physical_properties.pNext = &tfb_properties;
+ physical.GetProperties2KHR(physical_properties);
if (tfb_features.transformFeedback && tfb_features.geometryStreams &&
tfb_properties.maxTransformFeedbackStreams >= 4 &&
@@ -733,6 +786,18 @@ std::vector<const char*> VKDevice::LoadExtensions() {
}
}
+ if (has_ext_robustness2) {
+ VkPhysicalDeviceRobustness2FeaturesEXT robustness2;
+ robustness2.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_ROBUSTNESS_2_FEATURES_EXT;
+ robustness2.pNext = nullptr;
+ features.pNext = &robustness2;
+ physical.GetFeatures2KHR(features);
+ if (robustness2.nullDescriptor && robustness2.robustImageAccess2) {
+ extensions.push_back(VK_EXT_ROBUSTNESS_2_EXTENSION_NAME);
+ ext_robustness2 = true;
+ }
+ }
+
return extensions;
}
@@ -764,6 +829,7 @@ void VKDevice::SetupFamilies(VkSurfaceKHR surface) {
void VKDevice::SetupFeatures() {
const auto supported_features{physical.GetFeatures()};
is_formatless_image_load_supported = supported_features.shaderStorageImageReadWithoutFormat;
+ is_blit_depth_stencil_supported = TestDepthStencilBlits();
is_optimal_astc_supported = IsOptimalAstcSupported(supported_features);
}
@@ -794,6 +860,32 @@ void VKDevice::CollectTelemetryParameters() {
}
}
+void VKDevice::CollectToolingInfo() {
+ if (!ext_tooling_info) {
+ return;
+ }
+ const auto vkGetPhysicalDeviceToolPropertiesEXT =
+ reinterpret_cast<PFN_vkGetPhysicalDeviceToolPropertiesEXT>(
+ dld.vkGetInstanceProcAddr(instance, "vkGetPhysicalDeviceToolPropertiesEXT"));
+ if (!vkGetPhysicalDeviceToolPropertiesEXT) {
+ return;
+ }
+ u32 tool_count = 0;
+ if (vkGetPhysicalDeviceToolPropertiesEXT(physical, &tool_count, nullptr) != VK_SUCCESS) {
+ return;
+ }
+ std::vector<VkPhysicalDeviceToolPropertiesEXT> tools(tool_count);
+ if (vkGetPhysicalDeviceToolPropertiesEXT(physical, &tool_count, tools.data()) != VK_SUCCESS) {
+ return;
+ }
+ for (const VkPhysicalDeviceToolPropertiesEXT& tool : tools) {
+ const std::string_view name = tool.name;
+ LOG_INFO(Render_Vulkan, "{}", name);
+ has_renderdoc = has_renderdoc || name == "RenderDoc";
+ has_nsight_graphics = has_nsight_graphics || name == "NVIDIA Nsight Graphics";
+ }
+}
+
std::vector<VkDeviceQueueCreateInfo> VKDevice::GetDeviceQueueCreateInfos() const {
static constexpr float QUEUE_PRIORITY = 1.0f;
diff --git a/src/video_core/renderer_vulkan/vk_device.h b/src/video_core/renderer_vulkan/vk_device.h
index 4286673d9..995dcfc0f 100644
--- a/src/video_core/renderer_vulkan/vk_device.h
+++ b/src/video_core/renderer_vulkan/vk_device.h
@@ -157,6 +157,11 @@ public:
return is_formatless_image_load_supported;
}
+ /// Returns true when blitting from and to depth stencil images is supported.
+ bool IsBlitDepthStencilSupported() const {
+ return is_blit_depth_stencil_supported;
+ }
+
/// Returns true if the device supports VK_NV_viewport_swizzle.
bool IsNvViewportSwizzleSupported() const {
return nv_viewport_swizzle;
@@ -172,6 +177,11 @@ public:
return ext_index_type_uint8;
}
+ /// Returns true if the device supports VK_EXT_sampler_filter_minmax.
+ bool IsExtSamplerFilterMinmaxSupported() const {
+ return ext_sampler_filter_minmax;
+ }
+
/// Returns true if the device supports VK_EXT_depth_range_unrestricted.
bool IsExtDepthRangeUnrestrictedSupported() const {
return ext_depth_range_unrestricted;
@@ -197,6 +207,16 @@ public:
return ext_extended_dynamic_state;
}
+ /// Returns true if the device supports VK_EXT_shader_stencil_export.
+ bool IsExtShaderStencilExportSupported() const {
+ return ext_shader_stencil_export;
+ }
+
+ /// Returns true when a known debugging tool is attached.
+ bool HasDebuggingToolAttached() const {
+ return has_renderdoc || has_nsight_graphics;
+ }
+
/// Returns the vendor name reported from Vulkan.
std::string_view GetVendorName() const {
return vendor_name;
@@ -228,16 +248,23 @@ private:
/// Collects telemetry information from the device.
void CollectTelemetryParameters();
+ /// Collects information about attached tools.
+ void CollectToolingInfo();
+
/// Returns a list of queue initialization descriptors.
std::vector<VkDeviceQueueCreateInfo> GetDeviceQueueCreateInfos() const;
/// Returns true if ASTC textures are natively supported.
bool IsOptimalAstcSupported(const VkPhysicalDeviceFeatures& features) const;
+ /// Returns true if the device natively supports blitting depth stencil images.
+ bool TestDepthStencilBlits() const;
+
/// Returns true if a format is supported.
bool IsFormatSupported(VkFormat wanted_format, VkFormatFeatureFlags wanted_usage,
FormatType format_type) const;
+ VkInstance instance; ///< Vulkan instance.
vk::DeviceDispatch dld; ///< Device function pointers.
vk::PhysicalDevice physical; ///< Physical device.
VkPhysicalDeviceProperties properties; ///< Device properties.
@@ -253,15 +280,22 @@ private:
bool is_float16_supported{}; ///< Support for float16 arithmetics.
bool is_warp_potentially_bigger{}; ///< Host warp size can be bigger than guest.
bool is_formatless_image_load_supported{}; ///< Support for shader image read without format.
+ bool is_blit_depth_stencil_supported{}; ///< Support for blitting from and to depth stencil.
bool nv_viewport_swizzle{}; ///< Support for VK_NV_viewport_swizzle.
bool khr_uniform_buffer_standard_layout{}; ///< Support for std430 on UBOs.
bool ext_index_type_uint8{}; ///< Support for VK_EXT_index_type_uint8.
+ bool ext_sampler_filter_minmax{}; ///< Support for VK_EXT_sampler_filter_minmax.
bool ext_depth_range_unrestricted{}; ///< Support for VK_EXT_depth_range_unrestricted.
bool ext_shader_viewport_index_layer{}; ///< Support for VK_EXT_shader_viewport_index_layer.
+ bool ext_tooling_info{}; ///< Support for VK_EXT_tooling_info.
bool ext_transform_feedback{}; ///< Support for VK_EXT_transform_feedback.
bool ext_custom_border_color{}; ///< Support for VK_EXT_custom_border_color.
bool ext_extended_dynamic_state{}; ///< Support for VK_EXT_extended_dynamic_state.
+ bool ext_robustness2{}; ///< Support for VK_EXT_robustness2.
+ bool ext_shader_stencil_export{}; ///< Support for VK_EXT_shader_stencil_export.
bool nv_device_diagnostics_config{}; ///< Support for VK_NV_device_diagnostics_config.
+ bool has_renderdoc{}; ///< Has RenderDoc attached
+ bool has_nsight_graphics{}; ///< Has Nsight Graphics attached
// Asynchronous Graphics Pipeline setting
bool use_asynchronous_shaders{}; ///< Setting to use asynchronous shaders/graphics pipeline
diff --git a/src/video_core/renderer_vulkan/vk_fence_manager.cpp b/src/video_core/renderer_vulkan/vk_fence_manager.cpp
index 5babbdd0b..774a12a53 100644
--- a/src/video_core/renderer_vulkan/vk_fence_manager.cpp
+++ b/src/video_core/renderer_vulkan/vk_fence_manager.cpp
@@ -14,12 +14,13 @@
namespace Vulkan {
-InnerFence::InnerFence(const VKDevice& device, VKScheduler& scheduler, u32 payload, bool is_stubbed)
- : VideoCommon::FenceBase(payload, is_stubbed), device{device}, scheduler{scheduler} {}
+InnerFence::InnerFence(const VKDevice& device_, VKScheduler& scheduler_, u32 payload_,
+ bool is_stubbed_)
+ : FenceBase{payload_, is_stubbed_}, device{device_}, scheduler{scheduler_} {}
-InnerFence::InnerFence(const VKDevice& device, VKScheduler& scheduler, GPUVAddr address,
- u32 payload, bool is_stubbed)
- : VideoCommon::FenceBase(address, payload, is_stubbed), device{device}, scheduler{scheduler} {}
+InnerFence::InnerFence(const VKDevice& device_, VKScheduler& scheduler_, GPUVAddr address_,
+ u32 payload_, bool is_stubbed_)
+ : FenceBase{address_, payload_, is_stubbed_}, device{device_}, scheduler{scheduler_} {}
InnerFence::~InnerFence() = default;
@@ -71,11 +72,11 @@ bool InnerFence::IsEventSignalled() const {
}
}
-VKFenceManager::VKFenceManager(VideoCore::RasterizerInterface& rasterizer, Tegra::GPU& gpu,
- Tegra::MemoryManager& memory_manager, VKTextureCache& texture_cache,
- VKBufferCache& buffer_cache, VKQueryCache& query_cache,
+VKFenceManager::VKFenceManager(VideoCore::RasterizerInterface& rasterizer_, Tegra::GPU& gpu_,
+ Tegra::MemoryManager& memory_manager_, TextureCache& texture_cache_,
+ VKBufferCache& buffer_cache_, VKQueryCache& query_cache_,
const VKDevice& device_, VKScheduler& scheduler_)
- : GenericFenceManager(rasterizer, gpu, texture_cache, buffer_cache, query_cache),
+ : GenericFenceManager{rasterizer_, gpu_, texture_cache_, buffer_cache_, query_cache_},
device{device_}, scheduler{scheduler_} {}
Fence VKFenceManager::CreateFence(u32 value, bool is_stubbed) {
diff --git a/src/video_core/renderer_vulkan/vk_fence_manager.h b/src/video_core/renderer_vulkan/vk_fence_manager.h
index 1547d6d30..c2869e8e3 100644
--- a/src/video_core/renderer_vulkan/vk_fence_manager.h
+++ b/src/video_core/renderer_vulkan/vk_fence_manager.h
@@ -8,6 +8,7 @@
#include "video_core/fence_manager.h"
#include "video_core/renderer_vulkan/vk_buffer_cache.h"
+#include "video_core/renderer_vulkan/vk_texture_cache.h"
#include "video_core/renderer_vulkan/wrapper.h"
namespace Core {
@@ -24,14 +25,13 @@ class VKBufferCache;
class VKDevice;
class VKQueryCache;
class VKScheduler;
-class VKTextureCache;
class InnerFence : public VideoCommon::FenceBase {
public:
- explicit InnerFence(const VKDevice& device, VKScheduler& scheduler, u32 payload,
- bool is_stubbed);
- explicit InnerFence(const VKDevice& device, VKScheduler& scheduler, GPUVAddr address,
- u32 payload, bool is_stubbed);
+ explicit InnerFence(const VKDevice& device_, VKScheduler& scheduler_, u32 payload_,
+ bool is_stubbed_);
+ explicit InnerFence(const VKDevice& device_, VKScheduler& scheduler_, GPUVAddr address_,
+ u32 payload_, bool is_stubbed_);
~InnerFence();
void Queue();
@@ -51,14 +51,14 @@ private:
using Fence = std::shared_ptr<InnerFence>;
using GenericFenceManager =
- VideoCommon::FenceManager<Fence, VKTextureCache, VKBufferCache, VKQueryCache>;
+ VideoCommon::FenceManager<Fence, TextureCache, VKBufferCache, VKQueryCache>;
class VKFenceManager final : public GenericFenceManager {
public:
- explicit VKFenceManager(VideoCore::RasterizerInterface& rasterizer, Tegra::GPU& gpu,
- Tegra::MemoryManager& memory_manager, VKTextureCache& texture_cache,
- VKBufferCache& buffer_cache, VKQueryCache& query_cache,
- const VKDevice& device, VKScheduler& scheduler);
+ explicit VKFenceManager(VideoCore::RasterizerInterface& rasterizer_, Tegra::GPU& gpu_,
+ Tegra::MemoryManager& memory_manager_, TextureCache& texture_cache_,
+ VKBufferCache& buffer_cache_, VKQueryCache& query_cache_,
+ const VKDevice& device_, VKScheduler& scheduler_);
protected:
Fence CreateFence(u32 value, bool is_stubbed) override;
diff --git a/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp b/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp
index 0e8f9c352..7979df3a8 100644
--- a/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp
+++ b/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp
@@ -15,7 +15,6 @@
#include "video_core/renderer_vulkan/vk_device.h"
#include "video_core/renderer_vulkan/vk_graphics_pipeline.h"
#include "video_core/renderer_vulkan/vk_pipeline_cache.h"
-#include "video_core/renderer_vulkan/vk_renderpass_cache.h"
#include "video_core/renderer_vulkan/vk_scheduler.h"
#include "video_core/renderer_vulkan/vk_update_descriptor.h"
#include "video_core/renderer_vulkan/wrapper.h"
@@ -69,23 +68,45 @@ VkViewportSwizzleNV UnpackViewportSwizzle(u16 swizzle) {
};
}
+VkSampleCountFlagBits ConvertMsaaMode(Tegra::Texture::MsaaMode msaa_mode) {
+ switch (msaa_mode) {
+ case Tegra::Texture::MsaaMode::Msaa1x1:
+ return VK_SAMPLE_COUNT_1_BIT;
+ case Tegra::Texture::MsaaMode::Msaa2x1:
+ case Tegra::Texture::MsaaMode::Msaa2x1_D3D:
+ return VK_SAMPLE_COUNT_2_BIT;
+ case Tegra::Texture::MsaaMode::Msaa2x2:
+ case Tegra::Texture::MsaaMode::Msaa2x2_VC4:
+ case Tegra::Texture::MsaaMode::Msaa2x2_VC12:
+ return VK_SAMPLE_COUNT_4_BIT;
+ case Tegra::Texture::MsaaMode::Msaa4x2:
+ case Tegra::Texture::MsaaMode::Msaa4x2_D3D:
+ case Tegra::Texture::MsaaMode::Msaa4x2_VC8:
+ case Tegra::Texture::MsaaMode::Msaa4x2_VC24:
+ return VK_SAMPLE_COUNT_8_BIT;
+ case Tegra::Texture::MsaaMode::Msaa4x4:
+ return VK_SAMPLE_COUNT_16_BIT;
+ default:
+ UNREACHABLE_MSG("Invalid msaa_mode={}", static_cast<int>(msaa_mode));
+ return VK_SAMPLE_COUNT_1_BIT;
+ }
+}
+
} // Anonymous namespace
-VKGraphicsPipeline::VKGraphicsPipeline(const VKDevice& device, VKScheduler& scheduler,
- VKDescriptorPool& descriptor_pool,
- VKUpdateDescriptorQueue& update_descriptor_queue,
- VKRenderPassCache& renderpass_cache,
+VKGraphicsPipeline::VKGraphicsPipeline(const VKDevice& device_, VKScheduler& scheduler_,
+ VKDescriptorPool& descriptor_pool_,
+ VKUpdateDescriptorQueue& update_descriptor_queue_,
const GraphicsPipelineCacheKey& key,
vk::Span<VkDescriptorSetLayoutBinding> bindings,
- const SPIRVProgram& program)
- : device{device}, scheduler{scheduler}, cache_key{key}, hash{cache_key.Hash()},
+ const SPIRVProgram& program, u32 num_color_buffers)
+ : device{device_}, scheduler{scheduler_}, cache_key{key}, hash{cache_key.Hash()},
descriptor_set_layout{CreateDescriptorSetLayout(bindings)},
- descriptor_allocator{descriptor_pool, *descriptor_set_layout},
- update_descriptor_queue{update_descriptor_queue}, layout{CreatePipelineLayout()},
- descriptor_template{CreateDescriptorUpdateTemplate(program)}, modules{CreateShaderModules(
- program)},
- renderpass{renderpass_cache.GetRenderPass(cache_key.renderpass_params)},
- pipeline{CreatePipeline(cache_key.renderpass_params, program)} {}
+ descriptor_allocator{descriptor_pool_, *descriptor_set_layout},
+ update_descriptor_queue{update_descriptor_queue_}, layout{CreatePipelineLayout()},
+ descriptor_template{CreateDescriptorUpdateTemplate(program)},
+ modules(CreateShaderModules(program)),
+ pipeline(CreatePipeline(program, cache_key.renderpass, num_color_buffers)) {}
VKGraphicsPipeline::~VKGraphicsPipeline() = default;
@@ -162,8 +183,8 @@ std::vector<vk::ShaderModule> VKGraphicsPipeline::CreateShaderModules(
.codeSize = 0,
};
- std::vector<vk::ShaderModule> modules;
- modules.reserve(Maxwell::MaxShaderStage);
+ std::vector<vk::ShaderModule> shader_modules;
+ shader_modules.reserve(Maxwell::MaxShaderStage);
for (std::size_t i = 0; i < Maxwell::MaxShaderStage; ++i) {
const auto& stage = program[i];
if (!stage) {
@@ -174,13 +195,14 @@ std::vector<vk::ShaderModule> VKGraphicsPipeline::CreateShaderModules(
ci.codeSize = stage->code.size() * sizeof(u32);
ci.pCode = stage->code.data();
- modules.push_back(device.GetLogical().CreateShaderModule(ci));
+ shader_modules.push_back(device.GetLogical().CreateShaderModule(ci));
}
- return modules;
+ return shader_modules;
}
-vk::Pipeline VKGraphicsPipeline::CreatePipeline(const RenderPassParams& renderpass_params,
- const SPIRVProgram& program) const {
+vk::Pipeline VKGraphicsPipeline::CreatePipeline(const SPIRVProgram& program,
+ VkRenderPass renderpass,
+ u32 num_color_buffers) const {
const auto& state = cache_key.fixed_state;
const auto& viewport_swizzles = state.viewport_swizzles;
@@ -230,7 +252,7 @@ vk::Pipeline VKGraphicsPipeline::CreatePipeline(const RenderPassParams& renderpa
if (!attribute.enabled) {
continue;
}
- if (input_attributes.find(static_cast<u32>(index)) == input_attributes.end()) {
+ if (!input_attributes.contains(static_cast<u32>(index))) {
// Skip attributes not used by the vertex shaders.
continue;
}
@@ -290,8 +312,7 @@ vk::Pipeline VKGraphicsPipeline::CreatePipeline(const RenderPassParams& renderpa
};
std::array<VkViewportSwizzleNV, Maxwell::NumViewports> swizzles;
- std::transform(viewport_swizzles.begin(), viewport_swizzles.end(), swizzles.begin(),
- UnpackViewportSwizzle);
+ std::ranges::transform(viewport_swizzles, swizzles.begin(), UnpackViewportSwizzle);
VkPipelineViewportSwizzleStateCreateInfoNV swizzle_ci{
.sType = VK_STRUCTURE_TYPE_PIPELINE_VIEWPORT_SWIZZLE_STATE_CREATE_INFO_NV,
.pNext = nullptr,
@@ -326,7 +347,7 @@ vk::Pipeline VKGraphicsPipeline::CreatePipeline(const RenderPassParams& renderpa
.sType = VK_STRUCTURE_TYPE_PIPELINE_MULTISAMPLE_STATE_CREATE_INFO,
.pNext = nullptr,
.flags = 0,
- .rasterizationSamples = VK_SAMPLE_COUNT_1_BIT,
+ .rasterizationSamples = ConvertMsaaMode(state.msaa_mode),
.sampleShadingEnable = VK_FALSE,
.minSampleShading = 0.0f,
.pSampleMask = nullptr,
@@ -352,8 +373,7 @@ vk::Pipeline VKGraphicsPipeline::CreatePipeline(const RenderPassParams& renderpa
};
std::array<VkPipelineColorBlendAttachmentState, Maxwell::NumRenderTargets> cb_attachments;
- const auto num_attachments = static_cast<std::size_t>(renderpass_params.num_color_attachments);
- for (std::size_t index = 0; index < num_attachments; ++index) {
+ for (std::size_t index = 0; index < num_color_buffers; ++index) {
static constexpr std::array COMPONENT_TABLE{
VK_COLOR_COMPONENT_R_BIT,
VK_COLOR_COMPONENT_G_BIT,
@@ -387,7 +407,7 @@ vk::Pipeline VKGraphicsPipeline::CreatePipeline(const RenderPassParams& renderpa
.flags = 0,
.logicOpEnable = VK_FALSE,
.logicOp = VK_LOGIC_OP_COPY,
- .attachmentCount = static_cast<u32>(num_attachments),
+ .attachmentCount = num_color_buffers,
.pAttachments = cb_attachments.data(),
.blendConstants = {},
};
@@ -447,8 +467,7 @@ vk::Pipeline VKGraphicsPipeline::CreatePipeline(const RenderPassParams& renderpa
stage_ci.pNext = &subgroup_size_ci;
}
}
-
- const VkGraphicsPipelineCreateInfo ci{
+ return device.GetLogical().CreateGraphicsPipeline(VkGraphicsPipelineCreateInfo{
.sType = VK_STRUCTURE_TYPE_GRAPHICS_PIPELINE_CREATE_INFO,
.pNext = nullptr,
.flags = 0,
@@ -468,8 +487,7 @@ vk::Pipeline VKGraphicsPipeline::CreatePipeline(const RenderPassParams& renderpa
.subpass = 0,
.basePipelineHandle = nullptr,
.basePipelineIndex = 0,
- };
- return device.GetLogical().CreateGraphicsPipeline(ci);
+ });
}
} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/vk_graphics_pipeline.h b/src/video_core/renderer_vulkan/vk_graphics_pipeline.h
index 58aa35efd..214d06b4c 100644
--- a/src/video_core/renderer_vulkan/vk_graphics_pipeline.h
+++ b/src/video_core/renderer_vulkan/vk_graphics_pipeline.h
@@ -8,10 +8,10 @@
#include <optional>
#include <vector>
+#include "common/common_types.h"
#include "video_core/engines/maxwell_3d.h"
#include "video_core/renderer_vulkan/fixed_pipeline_state.h"
#include "video_core/renderer_vulkan/vk_descriptor_pool.h"
-#include "video_core/renderer_vulkan/vk_renderpass_cache.h"
#include "video_core/renderer_vulkan/vk_shader_decompiler.h"
#include "video_core/renderer_vulkan/wrapper.h"
@@ -20,8 +20,7 @@ namespace Vulkan {
using Maxwell = Tegra::Engines::Maxwell3D::Regs;
struct GraphicsPipelineCacheKey {
- RenderPassParams renderpass_params;
- u32 padding;
+ VkRenderPass renderpass;
std::array<GPUVAddr, Maxwell::MaxShaderProgram> shaders;
FixedPipelineState fixed_state;
@@ -34,7 +33,7 @@ struct GraphicsPipelineCacheKey {
}
std::size_t Size() const noexcept {
- return sizeof(renderpass_params) + sizeof(padding) + sizeof(shaders) + fixed_state.Size();
+ return sizeof(renderpass) + sizeof(shaders) + fixed_state.Size();
}
};
static_assert(std::has_unique_object_representations_v<GraphicsPipelineCacheKey>);
@@ -43,7 +42,6 @@ static_assert(std::is_trivially_constructible_v<GraphicsPipelineCacheKey>);
class VKDescriptorPool;
class VKDevice;
-class VKRenderPassCache;
class VKScheduler;
class VKUpdateDescriptorQueue;
@@ -51,13 +49,12 @@ using SPIRVProgram = std::array<std::optional<SPIRVShader>, Maxwell::MaxShaderSt
class VKGraphicsPipeline final {
public:
- explicit VKGraphicsPipeline(const VKDevice& device, VKScheduler& scheduler,
+ explicit VKGraphicsPipeline(const VKDevice& device_, VKScheduler& scheduler_,
VKDescriptorPool& descriptor_pool,
- VKUpdateDescriptorQueue& update_descriptor_queue,
- VKRenderPassCache& renderpass_cache,
+ VKUpdateDescriptorQueue& update_descriptor_queue_,
const GraphicsPipelineCacheKey& key,
vk::Span<VkDescriptorSetLayoutBinding> bindings,
- const SPIRVProgram& program);
+ const SPIRVProgram& program, u32 num_color_buffers);
~VKGraphicsPipeline();
VkDescriptorSet CommitDescriptorSet();
@@ -70,10 +67,6 @@ public:
return *layout;
}
- VkRenderPass GetRenderPass() const {
- return renderpass;
- }
-
GraphicsPipelineCacheKey GetCacheKey() const {
return cache_key;
}
@@ -89,8 +82,8 @@ private:
std::vector<vk::ShaderModule> CreateShaderModules(const SPIRVProgram& program) const;
- vk::Pipeline CreatePipeline(const RenderPassParams& renderpass_params,
- const SPIRVProgram& program) const;
+ vk::Pipeline CreatePipeline(const SPIRVProgram& program, VkRenderPass renderpass,
+ u32 num_color_buffers) const;
const VKDevice& device;
VKScheduler& scheduler;
@@ -104,7 +97,6 @@ private:
vk::DescriptorUpdateTemplateKHR descriptor_template;
std::vector<vk::ShaderModule> modules;
- VkRenderPass renderpass;
vk::Pipeline pipeline;
};
diff --git a/src/video_core/renderer_vulkan/vk_image.cpp b/src/video_core/renderer_vulkan/vk_image.cpp
deleted file mode 100644
index 1c418ea17..000000000
--- a/src/video_core/renderer_vulkan/vk_image.cpp
+++ /dev/null
@@ -1,135 +0,0 @@
-// Copyright 2018 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include <memory>
-#include <vector>
-
-#include "common/assert.h"
-#include "video_core/renderer_vulkan/vk_device.h"
-#include "video_core/renderer_vulkan/vk_image.h"
-#include "video_core/renderer_vulkan/vk_scheduler.h"
-#include "video_core/renderer_vulkan/wrapper.h"
-
-namespace Vulkan {
-
-VKImage::VKImage(const VKDevice& device, VKScheduler& scheduler, const VkImageCreateInfo& image_ci,
- VkImageAspectFlags aspect_mask)
- : device{device}, scheduler{scheduler}, format{image_ci.format}, aspect_mask{aspect_mask},
- image_num_layers{image_ci.arrayLayers}, image_num_levels{image_ci.mipLevels} {
- UNIMPLEMENTED_IF_MSG(image_ci.queueFamilyIndexCount != 0,
- "Queue family tracking is not implemented");
-
- image = device.GetLogical().CreateImage(image_ci);
-
- const u32 num_ranges = image_num_layers * image_num_levels;
- barriers.resize(num_ranges);
- subrange_states.resize(num_ranges, {{}, image_ci.initialLayout});
-}
-
-VKImage::~VKImage() = default;
-
-void VKImage::Transition(u32 base_layer, u32 num_layers, u32 base_level, u32 num_levels,
- VkPipelineStageFlags new_stage_mask, VkAccessFlags new_access,
- VkImageLayout new_layout) {
- if (!HasChanged(base_layer, num_layers, base_level, num_levels, new_access, new_layout)) {
- return;
- }
-
- std::size_t cursor = 0;
- for (u32 layer_it = 0; layer_it < num_layers; ++layer_it) {
- for (u32 level_it = 0; level_it < num_levels; ++level_it, ++cursor) {
- const u32 layer = base_layer + layer_it;
- const u32 level = base_level + level_it;
- auto& state = GetSubrangeState(layer, level);
- auto& barrier = barriers[cursor];
- barrier.sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER;
- barrier.pNext = nullptr;
- barrier.srcAccessMask = state.access;
- barrier.dstAccessMask = new_access;
- barrier.oldLayout = state.layout;
- barrier.newLayout = new_layout;
- barrier.srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED;
- barrier.dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED;
- barrier.image = *image;
- barrier.subresourceRange.aspectMask = aspect_mask;
- barrier.subresourceRange.baseMipLevel = level;
- barrier.subresourceRange.levelCount = 1;
- barrier.subresourceRange.baseArrayLayer = layer;
- barrier.subresourceRange.layerCount = 1;
- state.access = new_access;
- state.layout = new_layout;
- }
- }
-
- scheduler.RequestOutsideRenderPassOperationContext();
-
- scheduler.Record([barriers = barriers, cursor](vk::CommandBuffer cmdbuf) {
- // TODO(Rodrigo): Implement a way to use the latest stage across subresources.
- cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_ALL_COMMANDS_BIT,
- VK_PIPELINE_STAGE_ALL_COMMANDS_BIT, 0, {}, {},
- vk::Span(barriers.data(), cursor));
- });
-}
-
-bool VKImage::HasChanged(u32 base_layer, u32 num_layers, u32 base_level, u32 num_levels,
- VkAccessFlags new_access, VkImageLayout new_layout) noexcept {
- const bool is_full_range = base_layer == 0 && num_layers == image_num_layers &&
- base_level == 0 && num_levels == image_num_levels;
- if (!is_full_range) {
- state_diverged = true;
- }
-
- if (!state_diverged) {
- auto& state = GetSubrangeState(0, 0);
- if (state.access != new_access || state.layout != new_layout) {
- return true;
- }
- }
-
- for (u32 layer_it = 0; layer_it < num_layers; ++layer_it) {
- for (u32 level_it = 0; level_it < num_levels; ++level_it) {
- const u32 layer = base_layer + layer_it;
- const u32 level = base_level + level_it;
- auto& state = GetSubrangeState(layer, level);
- if (state.access != new_access || state.layout != new_layout) {
- return true;
- }
- }
- }
- return false;
-}
-
-void VKImage::CreatePresentView() {
- // Image type has to be 2D to be presented.
- present_view = device.GetLogical().CreateImageView({
- .sType = VK_STRUCTURE_TYPE_IMAGE_VIEW_CREATE_INFO,
- .pNext = nullptr,
- .flags = 0,
- .image = *image,
- .viewType = VK_IMAGE_VIEW_TYPE_2D,
- .format = format,
- .components =
- {
- .r = VK_COMPONENT_SWIZZLE_IDENTITY,
- .g = VK_COMPONENT_SWIZZLE_IDENTITY,
- .b = VK_COMPONENT_SWIZZLE_IDENTITY,
- .a = VK_COMPONENT_SWIZZLE_IDENTITY,
- },
- .subresourceRange =
- {
- .aspectMask = aspect_mask,
- .baseMipLevel = 0,
- .levelCount = 1,
- .baseArrayLayer = 0,
- .layerCount = 1,
- },
- });
-}
-
-VKImage::SubrangeState& VKImage::GetSubrangeState(u32 layer, u32 level) noexcept {
- return subrange_states[static_cast<std::size_t>(layer * image_num_levels) +
- static_cast<std::size_t>(level)];
-}
-
-} // namespace Vulkan \ No newline at end of file
diff --git a/src/video_core/renderer_vulkan/vk_image.h b/src/video_core/renderer_vulkan/vk_image.h
deleted file mode 100644
index b4d7229e5..000000000
--- a/src/video_core/renderer_vulkan/vk_image.h
+++ /dev/null
@@ -1,84 +0,0 @@
-// Copyright 2018 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <memory>
-#include <vector>
-
-#include "common/common_types.h"
-#include "video_core/renderer_vulkan/wrapper.h"
-
-namespace Vulkan {
-
-class VKDevice;
-class VKScheduler;
-
-class VKImage {
-public:
- explicit VKImage(const VKDevice& device, VKScheduler& scheduler,
- const VkImageCreateInfo& image_ci, VkImageAspectFlags aspect_mask);
- ~VKImage();
-
- /// Records in the passed command buffer an image transition and updates the state of the image.
- void Transition(u32 base_layer, u32 num_layers, u32 base_level, u32 num_levels,
- VkPipelineStageFlags new_stage_mask, VkAccessFlags new_access,
- VkImageLayout new_layout);
-
- /// Returns a view compatible with presentation, the image has to be 2D.
- VkImageView GetPresentView() {
- if (!present_view) {
- CreatePresentView();
- }
- return *present_view;
- }
-
- /// Returns the Vulkan image handler.
- const vk::Image& GetHandle() const {
- return image;
- }
-
- /// Returns the Vulkan format for this image.
- VkFormat GetFormat() const {
- return format;
- }
-
- /// Returns the Vulkan aspect mask.
- VkImageAspectFlags GetAspectMask() const {
- return aspect_mask;
- }
-
-private:
- struct SubrangeState final {
- VkAccessFlags access = 0; ///< Current access bits.
- VkImageLayout layout = VK_IMAGE_LAYOUT_UNDEFINED; ///< Current image layout.
- };
-
- bool HasChanged(u32 base_layer, u32 num_layers, u32 base_level, u32 num_levels,
- VkAccessFlags new_access, VkImageLayout new_layout) noexcept;
-
- /// Creates a presentation view.
- void CreatePresentView();
-
- /// Returns the subrange state for a layer and layer.
- SubrangeState& GetSubrangeState(u32 layer, u32 level) noexcept;
-
- const VKDevice& device; ///< Device handler.
- VKScheduler& scheduler; ///< Device scheduler.
-
- const VkFormat format; ///< Vulkan format.
- const VkImageAspectFlags aspect_mask; ///< Vulkan aspect mask.
- const u32 image_num_layers; ///< Number of layers.
- const u32 image_num_levels; ///< Number of mipmap levels.
-
- vk::Image image; ///< Image handle.
- vk::ImageView present_view; ///< Image view compatible with presentation.
-
- std::vector<VkImageMemoryBarrier> barriers; ///< Pool of barriers.
- std::vector<SubrangeState> subrange_states; ///< Current subrange state.
-
- bool state_diverged = false; ///< True when subresources mismatch in layout.
-};
-
-} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/vk_memory_manager.cpp b/src/video_core/renderer_vulkan/vk_memory_manager.cpp
index 24c8960ac..56b24b70f 100644
--- a/src/video_core/renderer_vulkan/vk_memory_manager.cpp
+++ b/src/video_core/renderer_vulkan/vk_memory_manager.cpp
@@ -29,10 +29,10 @@ u64 GetAllocationChunkSize(u64 required_size) {
class VKMemoryAllocation final {
public:
- explicit VKMemoryAllocation(const VKDevice& device, vk::DeviceMemory memory,
- VkMemoryPropertyFlags properties, u64 allocation_size, u32 type)
- : device{device}, memory{std::move(memory)}, properties{properties},
- allocation_size{allocation_size}, shifted_type{ShiftType(type)} {}
+ explicit VKMemoryAllocation(const VKDevice& device_, vk::DeviceMemory memory_,
+ VkMemoryPropertyFlags properties_, u64 allocation_size_, u32 type_)
+ : device{device_}, memory{std::move(memory_)}, properties{properties_},
+ allocation_size{allocation_size_}, shifted_type{ShiftType(type_)} {}
VKMemoryCommit Commit(VkDeviceSize commit_size, VkDeviceSize alignment) {
auto found = TryFindFreeSection(free_iterator, allocation_size,
@@ -117,8 +117,8 @@ private:
std::vector<const VKMemoryCommitImpl*> commits;
};
-VKMemoryManager::VKMemoryManager(const VKDevice& device)
- : device{device}, properties{device.GetPhysical().GetMemoryProperties()} {}
+VKMemoryManager::VKMemoryManager(const VKDevice& device_)
+ : device{device_}, properties{device_.GetPhysical().GetMemoryProperties()} {}
VKMemoryManager::~VKMemoryManager() = default;
@@ -207,16 +207,16 @@ VKMemoryCommit VKMemoryManager::TryAllocCommit(const VkMemoryRequirements& requi
return {};
}
-VKMemoryCommitImpl::VKMemoryCommitImpl(const VKDevice& device, VKMemoryAllocation* allocation,
- const vk::DeviceMemory& memory, u64 begin, u64 end)
- : device{device}, memory{memory}, interval{begin, end}, allocation{allocation} {}
+VKMemoryCommitImpl::VKMemoryCommitImpl(const VKDevice& device_, VKMemoryAllocation* allocation_,
+ const vk::DeviceMemory& memory_, u64 begin_, u64 end_)
+ : device{device_}, memory{memory_}, interval{begin_, end_}, allocation{allocation_} {}
VKMemoryCommitImpl::~VKMemoryCommitImpl() {
allocation->Free(this);
}
MemoryMap VKMemoryCommitImpl::Map(u64 size, u64 offset_) const {
- return MemoryMap{this, memory.Map(interval.first + offset_, size)};
+ return MemoryMap(this, std::span<u8>(memory.Map(interval.first + offset_, size), size));
}
void VKMemoryCommitImpl::Unmap() const {
diff --git a/src/video_core/renderer_vulkan/vk_memory_manager.h b/src/video_core/renderer_vulkan/vk_memory_manager.h
index 1af88e3d4..318f8b43e 100644
--- a/src/video_core/renderer_vulkan/vk_memory_manager.h
+++ b/src/video_core/renderer_vulkan/vk_memory_manager.h
@@ -5,6 +5,7 @@
#pragma once
#include <memory>
+#include <span>
#include <utility>
#include <vector>
#include "common/common_types.h"
@@ -21,7 +22,7 @@ using VKMemoryCommit = std::unique_ptr<VKMemoryCommitImpl>;
class VKMemoryManager final {
public:
- explicit VKMemoryManager(const VKDevice& device);
+ explicit VKMemoryManager(const VKDevice& device_);
VKMemoryManager(const VKMemoryManager&) = delete;
~VKMemoryManager();
@@ -58,8 +59,8 @@ class VKMemoryCommitImpl final {
friend MemoryMap;
public:
- explicit VKMemoryCommitImpl(const VKDevice& device, VKMemoryAllocation* allocation,
- const vk::DeviceMemory& memory, u64 begin, u64 end);
+ explicit VKMemoryCommitImpl(const VKDevice& device_, VKMemoryAllocation* allocation_,
+ const vk::DeviceMemory& memory_, u64 begin_, u64 end_);
~VKMemoryCommitImpl();
/// Maps a memory region and returns a pointer to it.
@@ -93,8 +94,8 @@ private:
/// Holds ownership of a memory map.
class MemoryMap final {
public:
- explicit MemoryMap(const VKMemoryCommitImpl* commit, u8* address)
- : commit{commit}, address{address} {}
+ explicit MemoryMap(const VKMemoryCommitImpl* commit_, std::span<u8> span_)
+ : commit{commit_}, span{span_} {}
~MemoryMap() {
if (commit) {
@@ -108,19 +109,24 @@ public:
commit = nullptr;
}
+ /// Returns a span to the memory map.
+ [[nodiscard]] std::span<u8> Span() const noexcept {
+ return span;
+ }
+
/// Returns the address of the memory map.
- u8* GetAddress() const {
- return address;
+ [[nodiscard]] u8* Address() const noexcept {
+ return span.data();
}
/// Returns the address of the memory map;
- operator u8*() const {
- return address;
+ [[nodiscard]] operator u8*() const noexcept {
+ return span.data();
}
private:
const VKMemoryCommitImpl* commit{}; ///< Mapped memory commit.
- u8* address{}; ///< Address to the mapped memory.
+ std::span<u8> span; ///< Address to the mapped memory.
};
} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp b/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp
index df7e8c864..083796d05 100644
--- a/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp
+++ b/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp
@@ -8,6 +8,7 @@
#include <vector>
#include "common/bit_cast.h"
+#include "common/cityhash.h"
#include "common/microprofile.h"
#include "core/core.h"
#include "core/memory.h"
@@ -22,7 +23,6 @@
#include "video_core/renderer_vulkan/vk_graphics_pipeline.h"
#include "video_core/renderer_vulkan/vk_pipeline_cache.h"
#include "video_core/renderer_vulkan/vk_rasterizer.h"
-#include "video_core/renderer_vulkan/vk_renderpass_cache.h"
#include "video_core/renderer_vulkan/vk_scheduler.h"
#include "video_core/renderer_vulkan/vk_update_descriptor.h"
#include "video_core/renderer_vulkan/wrapper.h"
@@ -52,7 +52,9 @@ constexpr VkDescriptorType STORAGE_TEXEL_BUFFER = VK_DESCRIPTOR_TYPE_STORAGE_TEX
constexpr VkDescriptorType STORAGE_IMAGE = VK_DESCRIPTOR_TYPE_STORAGE_IMAGE;
constexpr VideoCommon::Shader::CompilerSettings compiler_settings{
- VideoCommon::Shader::CompileDepth::FullDecompile};
+ .depth = VideoCommon::Shader::CompileDepth::FullDecompile,
+ .disable_else_derivation = true,
+};
constexpr std::size_t GetStageFromProgram(std::size_t program) {
return program == 0 ? 0 : program - 1;
@@ -75,7 +77,7 @@ ShaderType GetShaderType(Maxwell::ShaderProgram program) {
case Maxwell::ShaderProgram::Fragment:
return ShaderType::Fragment;
default:
- UNIMPLEMENTED_MSG("program={}", static_cast<u32>(program));
+ UNIMPLEMENTED_MSG("program={}", program);
return ShaderType::Vertex;
}
}
@@ -136,26 +138,24 @@ bool ComputePipelineCacheKey::operator==(const ComputePipelineCacheKey& rhs) con
return std::memcmp(&rhs, this, sizeof *this) == 0;
}
-Shader::Shader(Tegra::Engines::ConstBufferEngineInterface& engine, Tegra::Engines::ShaderType stage,
- GPUVAddr gpu_addr_, VAddr cpu_addr, VideoCommon::Shader::ProgramCode program_code_,
- u32 main_offset)
- : gpu_addr(gpu_addr_), program_code(std::move(program_code_)), registry(stage, engine),
- shader_ir(program_code, main_offset, compiler_settings, registry),
+Shader::Shader(Tegra::Engines::ConstBufferEngineInterface& engine_, ShaderType stage_,
+ GPUVAddr gpu_addr_, VAddr cpu_addr_, ProgramCode program_code_, u32 main_offset_)
+ : gpu_addr(gpu_addr_), program_code(std::move(program_code_)), registry(stage_, engine_),
+ shader_ir(program_code, main_offset_, compiler_settings, registry),
entries(GenerateShaderEntries(shader_ir)) {}
Shader::~Shader() = default;
-VKPipelineCache::VKPipelineCache(RasterizerVulkan& rasterizer, Tegra::GPU& gpu_,
+VKPipelineCache::VKPipelineCache(RasterizerVulkan& rasterizer_, Tegra::GPU& gpu_,
Tegra::Engines::Maxwell3D& maxwell3d_,
Tegra::Engines::KeplerCompute& kepler_compute_,
Tegra::MemoryManager& gpu_memory_, const VKDevice& device_,
VKScheduler& scheduler_, VKDescriptorPool& descriptor_pool_,
- VKUpdateDescriptorQueue& update_descriptor_queue_,
- VKRenderPassCache& renderpass_cache_)
- : VideoCommon::ShaderCache<Shader>{rasterizer}, gpu{gpu_}, maxwell3d{maxwell3d_},
+ VKUpdateDescriptorQueue& update_descriptor_queue_)
+ : VideoCommon::ShaderCache<Shader>{rasterizer_}, gpu{gpu_}, maxwell3d{maxwell3d_},
kepler_compute{kepler_compute_}, gpu_memory{gpu_memory_}, device{device_},
- scheduler{scheduler_}, descriptor_pool{descriptor_pool_},
- update_descriptor_queue{update_descriptor_queue_}, renderpass_cache{renderpass_cache_} {}
+ scheduler{scheduler_}, descriptor_pool{descriptor_pool_}, update_descriptor_queue{
+ update_descriptor_queue_} {}
VKPipelineCache::~VKPipelineCache() = default;
@@ -200,7 +200,8 @@ std::array<Shader*, Maxwell::MaxShaderProgram> VKPipelineCache::GetShaders() {
}
VKGraphicsPipeline* VKPipelineCache::GetGraphicsPipeline(
- const GraphicsPipelineCacheKey& key, VideoCommon::Shader::AsyncShaders& async_shaders) {
+ const GraphicsPipelineCacheKey& key, u32 num_color_buffers,
+ VideoCommon::Shader::AsyncShaders& async_shaders) {
MICROPROFILE_SCOPE(Vulkan_PipelineCache);
if (last_graphics_pipeline && last_graphics_key == key) {
@@ -216,8 +217,8 @@ VKGraphicsPipeline* VKPipelineCache::GetGraphicsPipeline(
LOG_INFO(Render_Vulkan, "Compile 0x{:016X}", key.Hash());
const auto [program, bindings] = DecompileShaders(key.fixed_state);
async_shaders.QueueVulkanShader(this, device, scheduler, descriptor_pool,
- update_descriptor_queue, renderpass_cache, bindings,
- program, key);
+ update_descriptor_queue, bindings, program, key,
+ num_color_buffers);
}
last_graphics_pipeline = pair->second.get();
return last_graphics_pipeline;
@@ -230,8 +231,8 @@ VKGraphicsPipeline* VKPipelineCache::GetGraphicsPipeline(
LOG_INFO(Render_Vulkan, "Compile 0x{:016X}", key.Hash());
const auto [program, bindings] = DecompileShaders(key.fixed_state);
entry = std::make_unique<VKGraphicsPipeline>(device, scheduler, descriptor_pool,
- update_descriptor_queue, renderpass_cache, key,
- bindings, program);
+ update_descriptor_queue, key, bindings,
+ program, num_color_buffers);
gpu.ShaderNotify().MarkShaderComplete();
}
last_graphics_pipeline = entry.get();
diff --git a/src/video_core/renderer_vulkan/vk_pipeline_cache.h b/src/video_core/renderer_vulkan/vk_pipeline_cache.h
index e558e6658..fbaa8257c 100644
--- a/src/video_core/renderer_vulkan/vk_pipeline_cache.h
+++ b/src/video_core/renderer_vulkan/vk_pipeline_cache.h
@@ -19,7 +19,6 @@
#include "video_core/engines/maxwell_3d.h"
#include "video_core/renderer_vulkan/fixed_pipeline_state.h"
#include "video_core/renderer_vulkan/vk_graphics_pipeline.h"
-#include "video_core/renderer_vulkan/vk_renderpass_cache.h"
#include "video_core/renderer_vulkan/vk_shader_decompiler.h"
#include "video_core/renderer_vulkan/wrapper.h"
#include "video_core/shader/async_shaders.h"
@@ -84,9 +83,9 @@ namespace Vulkan {
class Shader {
public:
- explicit Shader(Tegra::Engines::ConstBufferEngineInterface& engine,
- Tegra::Engines::ShaderType stage, GPUVAddr gpu_addr, VAddr cpu_addr,
- VideoCommon::Shader::ProgramCode program_code, u32 main_offset);
+ explicit Shader(Tegra::Engines::ConstBufferEngineInterface& engine_,
+ Tegra::Engines::ShaderType stage_, GPUVAddr gpu_addr, VAddr cpu_addr_,
+ VideoCommon::Shader::ProgramCode program_code, u32 main_offset_);
~Shader();
GPUVAddr GetGpuAddr() const {
@@ -124,13 +123,13 @@ public:
Tegra::Engines::KeplerCompute& kepler_compute,
Tegra::MemoryManager& gpu_memory, const VKDevice& device,
VKScheduler& scheduler, VKDescriptorPool& descriptor_pool,
- VKUpdateDescriptorQueue& update_descriptor_queue,
- VKRenderPassCache& renderpass_cache);
+ VKUpdateDescriptorQueue& update_descriptor_queue);
~VKPipelineCache() override;
std::array<Shader*, Maxwell::MaxShaderProgram> GetShaders();
VKGraphicsPipeline* GetGraphicsPipeline(const GraphicsPipelineCacheKey& key,
+ u32 num_color_buffers,
VideoCommon::Shader::AsyncShaders& async_shaders);
VKComputePipeline& GetComputePipeline(const ComputePipelineCacheKey& key);
@@ -153,7 +152,6 @@ private:
VKScheduler& scheduler;
VKDescriptorPool& descriptor_pool;
VKUpdateDescriptorQueue& update_descriptor_queue;
- VKRenderPassCache& renderpass_cache;
std::unique_ptr<Shader> null_shader;
std::unique_ptr<Shader> null_kernel;
diff --git a/src/video_core/renderer_vulkan/vk_query_cache.cpp b/src/video_core/renderer_vulkan/vk_query_cache.cpp
index ee2d871e3..038760de3 100644
--- a/src/video_core/renderer_vulkan/vk_query_cache.cpp
+++ b/src/video_core/renderer_vulkan/vk_query_cache.cpp
@@ -66,15 +66,13 @@ void QueryPool::Reserve(std::pair<VkQueryPool, u32> query) {
usage[pool_index * GROW_STEP + static_cast<std::ptrdiff_t>(query.second)] = false;
}
-VKQueryCache::VKQueryCache(VideoCore::RasterizerInterface& rasterizer,
- Tegra::Engines::Maxwell3D& maxwell3d, Tegra::MemoryManager& gpu_memory,
- const VKDevice& device, VKScheduler& scheduler)
- : VideoCommon::QueryCacheBase<VKQueryCache, CachedQuery, CounterStream,
- HostCounter>{rasterizer, maxwell3d, gpu_memory},
- device{device}, scheduler{scheduler}, query_pools{
- QueryPool{device, scheduler,
- QueryType::SamplesPassed},
- } {}
+VKQueryCache::VKQueryCache(VideoCore::RasterizerInterface& rasterizer_,
+ Tegra::Engines::Maxwell3D& maxwell3d_, Tegra::MemoryManager& gpu_memory_,
+ const VKDevice& device_, VKScheduler& scheduler_)
+ : QueryCacheBase{rasterizer_, maxwell3d_, gpu_memory_}, device{device_}, scheduler{scheduler_},
+ query_pools{
+ QueryPool{device_, scheduler_, QueryType::SamplesPassed},
+ } {}
VKQueryCache::~VKQueryCache() {
// TODO(Rodrigo): This is a hack to destroy all HostCounter instances before the base class
@@ -95,12 +93,12 @@ void VKQueryCache::Reserve(QueryType type, std::pair<VkQueryPool, u32> query) {
query_pools[static_cast<std::size_t>(type)].Reserve(query);
}
-HostCounter::HostCounter(VKQueryCache& cache, std::shared_ptr<HostCounter> dependency,
- QueryType type)
- : VideoCommon::HostCounterBase<VKQueryCache, HostCounter>{std::move(dependency)}, cache{cache},
- type{type}, query{cache.AllocateQuery(type)}, tick{cache.Scheduler().CurrentTick()} {
- const vk::Device* logical = &cache.Device().GetLogical();
- cache.Scheduler().Record([logical, query = query](vk::CommandBuffer cmdbuf) {
+HostCounter::HostCounter(VKQueryCache& cache_, std::shared_ptr<HostCounter> dependency_,
+ QueryType type_)
+ : HostCounterBase{std::move(dependency_)}, cache{cache_}, type{type_},
+ query{cache_.AllocateQuery(type_)}, tick{cache_.Scheduler().CurrentTick()} {
+ const vk::Device* logical = &cache_.Device().GetLogical();
+ cache_.Scheduler().Record([logical, query = query](vk::CommandBuffer cmdbuf) {
logical->ResetQueryPoolEXT(query.first, query.second, 1);
cmdbuf.BeginQuery(query.first, query.second, VK_QUERY_CONTROL_PRECISE_BIT);
});
@@ -119,18 +117,20 @@ u64 HostCounter::BlockingQuery() const {
if (tick >= cache.Scheduler().CurrentTick()) {
cache.Scheduler().Flush();
}
+
u64 data;
- const VkResult result = cache.Device().GetLogical().GetQueryResults(
+ const VkResult query_result = cache.Device().GetLogical().GetQueryResults(
query.first, query.second, 1, sizeof(data), &data, sizeof(data),
VK_QUERY_RESULT_64_BIT | VK_QUERY_RESULT_WAIT_BIT);
- switch (result) {
+
+ switch (query_result) {
case VK_SUCCESS:
return data;
case VK_ERROR_DEVICE_LOST:
cache.Device().ReportLoss();
[[fallthrough]];
default:
- throw vk::Exception(result);
+ throw vk::Exception(query_result);
}
}
diff --git a/src/video_core/renderer_vulkan/vk_query_cache.h b/src/video_core/renderer_vulkan/vk_query_cache.h
index 2e57fb75d..837fe9ebf 100644
--- a/src/video_core/renderer_vulkan/vk_query_cache.h
+++ b/src/video_core/renderer_vulkan/vk_query_cache.h
@@ -53,9 +53,9 @@ private:
class VKQueryCache final
: public VideoCommon::QueryCacheBase<VKQueryCache, CachedQuery, CounterStream, HostCounter> {
public:
- explicit VKQueryCache(VideoCore::RasterizerInterface& rasterizer,
- Tegra::Engines::Maxwell3D& maxwell3d, Tegra::MemoryManager& gpu_memory,
- const VKDevice& device, VKScheduler& scheduler);
+ explicit VKQueryCache(VideoCore::RasterizerInterface& rasterizer_,
+ Tegra::Engines::Maxwell3D& maxwell3d_, Tegra::MemoryManager& gpu_memory_,
+ const VKDevice& device_, VKScheduler& scheduler_);
~VKQueryCache();
std::pair<VkQueryPool, u32> AllocateQuery(VideoCore::QueryType type);
@@ -78,8 +78,8 @@ private:
class HostCounter final : public VideoCommon::HostCounterBase<VKQueryCache, HostCounter> {
public:
- explicit HostCounter(VKQueryCache& cache, std::shared_ptr<HostCounter> dependency,
- VideoCore::QueryType type);
+ explicit HostCounter(VKQueryCache& cache_, std::shared_ptr<HostCounter> dependency_,
+ VideoCore::QueryType type_);
~HostCounter();
void EndQuery();
@@ -95,8 +95,8 @@ private:
class CachedQuery : public VideoCommon::CachedQueryBase<HostCounter> {
public:
- explicit CachedQuery(VKQueryCache&, VideoCore::QueryType, VAddr cpu_addr, u8* host_ptr)
- : VideoCommon::CachedQueryBase<HostCounter>{cpu_addr, host_ptr} {}
+ explicit CachedQuery(VKQueryCache&, VideoCore::QueryType, VAddr cpu_addr_, u8* host_ptr_)
+ : CachedQueryBase{cpu_addr_, host_ptr_} {}
};
} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.cpp b/src/video_core/renderer_vulkan/vk_rasterizer.cpp
index e0fb8693f..04c5c859c 100644
--- a/src/video_core/renderer_vulkan/vk_rasterizer.cpp
+++ b/src/video_core/renderer_vulkan/vk_rasterizer.cpp
@@ -19,6 +19,7 @@
#include "core/settings.h"
#include "video_core/engines/kepler_compute.h"
#include "video_core/engines/maxwell_3d.h"
+#include "video_core/renderer_vulkan/blit_image.h"
#include "video_core/renderer_vulkan/fixed_pipeline_state.h"
#include "video_core/renderer_vulkan/maxwell_to_vk.h"
#include "video_core/renderer_vulkan/renderer_vulkan.h"
@@ -30,8 +31,6 @@
#include "video_core/renderer_vulkan/vk_graphics_pipeline.h"
#include "video_core/renderer_vulkan/vk_pipeline_cache.h"
#include "video_core/renderer_vulkan/vk_rasterizer.h"
-#include "video_core/renderer_vulkan/vk_renderpass_cache.h"
-#include "video_core/renderer_vulkan/vk_sampler_cache.h"
#include "video_core/renderer_vulkan/vk_scheduler.h"
#include "video_core/renderer_vulkan/vk_staging_buffer_pool.h"
#include "video_core/renderer_vulkan/vk_state_tracker.h"
@@ -39,10 +38,13 @@
#include "video_core/renderer_vulkan/vk_update_descriptor.h"
#include "video_core/renderer_vulkan/wrapper.h"
#include "video_core/shader_cache.h"
+#include "video_core/texture_cache/texture_cache.h"
namespace Vulkan {
using Maxwell = Tegra::Engines::Maxwell3D::Regs;
+using VideoCommon::ImageViewId;
+using VideoCommon::ImageViewType;
MICROPROFILE_DEFINE(Vulkan_WaitForWorker, "Vulkan", "Wait for worker", MP_RGB(255, 192, 192));
MICROPROFILE_DEFINE(Vulkan_Drawing, "Vulkan", "Record drawing", MP_RGB(192, 128, 128));
@@ -58,9 +60,9 @@ MICROPROFILE_DEFINE(Vulkan_PipelineCache, "Vulkan", "Pipeline cache", MP_RGB(192
namespace {
-constexpr auto ComputeShaderIndex = static_cast<std::size_t>(Tegra::Engines::ShaderType::Compute);
+constexpr auto COMPUTE_SHADER_INDEX = static_cast<size_t>(Tegra::Engines::ShaderType::Compute);
-VkViewport GetViewportState(const VKDevice& device, const Maxwell& regs, std::size_t index) {
+VkViewport GetViewportState(const VKDevice& device, const Maxwell& regs, size_t index) {
const auto& src = regs.viewport_transform[index];
const float width = src.scale_x * 2.0f;
const float height = src.scale_y * 2.0f;
@@ -83,7 +85,7 @@ VkViewport GetViewportState(const VKDevice& device, const Maxwell& regs, std::si
return viewport;
}
-VkRect2D GetScissorState(const Maxwell& regs, std::size_t index) {
+VkRect2D GetScissorState(const Maxwell& regs, size_t index) {
const auto& src = regs.scissor_test[index];
VkRect2D scissor;
if (src.enable) {
@@ -103,98 +105,122 @@ VkRect2D GetScissorState(const Maxwell& regs, std::size_t index) {
std::array<GPUVAddr, Maxwell::MaxShaderProgram> GetShaderAddresses(
const std::array<Shader*, Maxwell::MaxShaderProgram>& shaders) {
std::array<GPUVAddr, Maxwell::MaxShaderProgram> addresses;
- for (std::size_t i = 0; i < std::size(addresses); ++i) {
+ for (size_t i = 0; i < std::size(addresses); ++i) {
addresses[i] = shaders[i] ? shaders[i]->GetGpuAddr() : 0;
}
return addresses;
}
-void TransitionImages(const std::vector<ImageView>& views, VkPipelineStageFlags pipeline_stage,
- VkAccessFlags access) {
- for (auto& [view, layout] : views) {
- view->Transition(*layout, pipeline_stage, access);
+struct TextureHandle {
+ constexpr TextureHandle(u32 data, bool via_header_index) {
+ const Tegra::Texture::TextureHandle handle{data};
+ image = handle.tic_id;
+ sampler = via_header_index ? image : handle.tsc_id.Value();
}
-}
+
+ u32 image;
+ u32 sampler;
+};
template <typename Engine, typename Entry>
-Tegra::Texture::FullTextureInfo GetTextureInfo(const Engine& engine, const Entry& entry,
- std::size_t stage, std::size_t index = 0) {
- const auto stage_type = static_cast<Tegra::Engines::ShaderType>(stage);
+TextureHandle GetTextureInfo(const Engine& engine, bool via_header_index, const Entry& entry,
+ size_t stage, size_t index = 0) {
+ const auto shader_type = static_cast<Tegra::Engines::ShaderType>(stage);
if constexpr (std::is_same_v<Entry, SamplerEntry>) {
if (entry.is_separated) {
const u32 buffer_1 = entry.buffer;
const u32 buffer_2 = entry.secondary_buffer;
const u32 offset_1 = entry.offset;
const u32 offset_2 = entry.secondary_offset;
- const u32 handle_1 = engine.AccessConstBuffer32(stage_type, buffer_1, offset_1);
- const u32 handle_2 = engine.AccessConstBuffer32(stage_type, buffer_2, offset_2);
- return engine.GetTextureInfo(handle_1 | handle_2);
+ const u32 handle_1 = engine.AccessConstBuffer32(shader_type, buffer_1, offset_1);
+ const u32 handle_2 = engine.AccessConstBuffer32(shader_type, buffer_2, offset_2);
+ return TextureHandle(handle_1 | handle_2, via_header_index);
}
}
if (entry.is_bindless) {
- const auto tex_handle = engine.AccessConstBuffer32(stage_type, entry.buffer, entry.offset);
- return engine.GetTextureInfo(tex_handle);
- }
- const auto& gpu_profile = engine.AccessGuestDriverProfile();
- const u32 entry_offset = static_cast<u32>(index * gpu_profile.GetTextureHandlerSize());
- const u32 offset = entry.offset + entry_offset;
- if constexpr (std::is_same_v<Engine, Tegra::Engines::Maxwell3D>) {
- return engine.GetStageTexture(stage_type, offset);
- } else {
- return engine.GetTexture(offset);
- }
-}
-
-/// @brief Determine if an attachment to be updated has to preserve contents
-/// @param is_clear True when a clear is being executed
-/// @param regs 3D registers
-/// @return True when the contents have to be preserved
-bool HasToPreserveColorContents(bool is_clear, const Maxwell& regs) {
- if (!is_clear) {
- return true;
- }
- // First we have to make sure all clear masks are enabled.
- if (!regs.clear_buffers.R || !regs.clear_buffers.G || !regs.clear_buffers.B ||
- !regs.clear_buffers.A) {
- return true;
- }
- // If scissors are disabled, the whole screen is cleared
- if (!regs.clear_flags.scissor) {
- return false;
+ const u32 raw = engine.AccessConstBuffer32(shader_type, entry.buffer, entry.offset);
+ return TextureHandle(raw, via_header_index);
}
- // Then we have to confirm scissor testing clears the whole image
- const std::size_t index = regs.clear_buffers.RT;
- const auto& scissor = regs.scissor_test[0];
- return scissor.min_x > 0 || scissor.min_y > 0 || scissor.max_x < regs.rt[index].width ||
- scissor.max_y < regs.rt[index].height;
+ const u32 buffer = engine.GetBoundBuffer();
+ const u64 offset = (entry.offset + index) * sizeof(u32);
+ return TextureHandle(engine.AccessConstBuffer32(shader_type, buffer, offset), via_header_index);
}
-/// @brief Determine if an attachment to be updated has to preserve contents
-/// @param is_clear True when a clear is being executed
-/// @param regs 3D registers
-/// @return True when the contents have to be preserved
-bool HasToPreserveDepthContents(bool is_clear, const Maxwell& regs) {
- // If we are not clearing, the contents have to be preserved
- if (!is_clear) {
- return true;
- }
- // For depth stencil clears we only have to confirm scissor test covers the whole image
- if (!regs.clear_flags.scissor) {
- return false;
- }
- // Make sure the clear cover the whole image
- const auto& scissor = regs.scissor_test[0];
- return scissor.min_x > 0 || scissor.min_y > 0 || scissor.max_x < regs.zeta_width ||
- scissor.max_y < regs.zeta_height;
-}
-
-template <std::size_t N>
+template <size_t N>
std::array<VkDeviceSize, N> ExpandStrides(const std::array<u16, N>& strides) {
std::array<VkDeviceSize, N> expanded;
std::copy(strides.begin(), strides.end(), expanded.begin());
return expanded;
}
+ImageViewType ImageViewTypeFromEntry(const SamplerEntry& entry) {
+ if (entry.is_buffer) {
+ return ImageViewType::e2D;
+ }
+ switch (entry.type) {
+ case Tegra::Shader::TextureType::Texture1D:
+ return entry.is_array ? ImageViewType::e1DArray : ImageViewType::e1D;
+ case Tegra::Shader::TextureType::Texture2D:
+ return entry.is_array ? ImageViewType::e2DArray : ImageViewType::e2D;
+ case Tegra::Shader::TextureType::Texture3D:
+ return ImageViewType::e3D;
+ case Tegra::Shader::TextureType::TextureCube:
+ return entry.is_array ? ImageViewType::CubeArray : ImageViewType::Cube;
+ }
+ UNREACHABLE();
+ return ImageViewType::e2D;
+}
+
+ImageViewType ImageViewTypeFromEntry(const ImageEntry& entry) {
+ switch (entry.type) {
+ case Tegra::Shader::ImageType::Texture1D:
+ return ImageViewType::e1D;
+ case Tegra::Shader::ImageType::Texture1DArray:
+ return ImageViewType::e1DArray;
+ case Tegra::Shader::ImageType::Texture2D:
+ return ImageViewType::e2D;
+ case Tegra::Shader::ImageType::Texture2DArray:
+ return ImageViewType::e2DArray;
+ case Tegra::Shader::ImageType::Texture3D:
+ return ImageViewType::e3D;
+ case Tegra::Shader::ImageType::TextureBuffer:
+ return ImageViewType::Buffer;
+ }
+ UNREACHABLE();
+ return ImageViewType::e2D;
+}
+
+void PushImageDescriptors(const ShaderEntries& entries, TextureCache& texture_cache,
+ VKUpdateDescriptorQueue& update_descriptor_queue,
+ ImageViewId*& image_view_id_ptr, VkSampler*& sampler_ptr) {
+ for ([[maybe_unused]] const auto& entry : entries.uniform_texels) {
+ const ImageViewId image_view_id = *image_view_id_ptr++;
+ const ImageView& image_view = texture_cache.GetImageView(image_view_id);
+ update_descriptor_queue.AddTexelBuffer(image_view.BufferView());
+ }
+ for (const auto& entry : entries.samplers) {
+ for (size_t i = 0; i < entry.size; ++i) {
+ const VkSampler sampler = *sampler_ptr++;
+ const ImageViewId image_view_id = *image_view_id_ptr++;
+ const ImageView& image_view = texture_cache.GetImageView(image_view_id);
+ const VkImageView handle = image_view.Handle(ImageViewTypeFromEntry(entry));
+ update_descriptor_queue.AddSampledImage(handle, sampler);
+ }
+ }
+ for ([[maybe_unused]] const auto& entry : entries.storage_texels) {
+ const ImageViewId image_view_id = *image_view_id_ptr++;
+ const ImageView& image_view = texture_cache.GetImageView(image_view_id);
+ update_descriptor_queue.AddTexelBuffer(image_view.BufferView());
+ }
+ for (const auto& entry : entries.images) {
+ // TODO: Mark as modified
+ const ImageViewId image_view_id = *image_view_id_ptr++;
+ const ImageView& image_view = texture_cache.GetImageView(image_view_id);
+ const VkImageView handle = image_view.Handle(ImageViewTypeFromEntry(entry));
+ update_descriptor_queue.AddImage(handle);
+ }
+}
+
} // Anonymous namespace
class BufferBindings final {
@@ -290,7 +316,7 @@ public:
private:
// Some of these fields are intentionally left uninitialized to avoid initializing them twice.
struct {
- std::size_t num_buffers = 0;
+ size_t num_buffers = 0;
std::array<VkBuffer, Maxwell::NumVertexArrays> buffers;
std::array<VkDeviceSize, Maxwell::NumVertexArrays> offsets;
std::array<VkDeviceSize, Maxwell::NumVertexArrays> sizes;
@@ -303,7 +329,7 @@ private:
VkIndexType type;
} index;
- template <std::size_t N>
+ template <size_t N>
void BindStatic(const VKDevice& device, VKScheduler& scheduler) const {
if (device.IsExtExtendedDynamicStateSupported()) {
if (index.buffer) {
@@ -320,7 +346,7 @@ private:
}
}
- template <std::size_t N, bool is_indexed, bool has_extended_dynamic_state>
+ template <size_t N, bool is_indexed, bool has_extended_dynamic_state>
void BindStatic(VKScheduler& scheduler) const {
static_assert(N <= Maxwell::NumVertexArrays);
if constexpr (N == 0) {
@@ -380,28 +406,31 @@ void RasterizerVulkan::DrawParameters::Draw(vk::CommandBuffer cmdbuf) const {
}
}
-RasterizerVulkan::RasterizerVulkan(Core::Frontend::EmuWindow& emu_window, Tegra::GPU& gpu_,
+RasterizerVulkan::RasterizerVulkan(Core::Frontend::EmuWindow& emu_window_, Tegra::GPU& gpu_,
Tegra::MemoryManager& gpu_memory_,
- Core::Memory::Memory& cpu_memory, VKScreenInfo& screen_info_,
+ Core::Memory::Memory& cpu_memory_, VKScreenInfo& screen_info_,
const VKDevice& device_, VKMemoryManager& memory_manager_,
StateTracker& state_tracker_, VKScheduler& scheduler_)
- : RasterizerAccelerated(cpu_memory), gpu(gpu_), gpu_memory(gpu_memory_),
- maxwell3d(gpu.Maxwell3D()), kepler_compute(gpu.KeplerCompute()), screen_info(screen_info_),
- device(device_), memory_manager(memory_manager_), state_tracker(state_tracker_),
- scheduler(scheduler_), staging_pool(device, memory_manager, scheduler),
- descriptor_pool(device, scheduler_), update_descriptor_queue(device, scheduler),
- renderpass_cache(device),
+ : RasterizerAccelerated{cpu_memory_}, gpu{gpu_},
+ gpu_memory{gpu_memory_}, maxwell3d{gpu.Maxwell3D()}, kepler_compute{gpu.KeplerCompute()},
+ screen_info{screen_info_}, device{device_}, memory_manager{memory_manager_},
+ state_tracker{state_tracker_}, scheduler{scheduler_}, stream_buffer(device, scheduler),
+ staging_pool(device, memory_manager, scheduler), descriptor_pool(device, scheduler),
+ update_descriptor_queue(device, scheduler),
+ blit_image(device, scheduler, state_tracker, descriptor_pool),
quad_array_pass(device, scheduler, descriptor_pool, staging_pool, update_descriptor_queue),
quad_indexed_pass(device, scheduler, descriptor_pool, staging_pool, update_descriptor_queue),
uint8_pass(device, scheduler, descriptor_pool, staging_pool, update_descriptor_queue),
- texture_cache(*this, maxwell3d, gpu_memory, device, memory_manager, scheduler, staging_pool),
+ texture_cache_runtime{device, scheduler, memory_manager, staging_pool, blit_image},
+ texture_cache(texture_cache_runtime, *this, maxwell3d, kepler_compute, gpu_memory),
pipeline_cache(*this, gpu, maxwell3d, kepler_compute, gpu_memory, device, scheduler,
- descriptor_pool, update_descriptor_queue, renderpass_cache),
- buffer_cache(*this, gpu_memory, cpu_memory, device, memory_manager, scheduler, staging_pool),
- sampler_cache(device), query_cache(*this, maxwell3d, gpu_memory, device, scheduler),
+ descriptor_pool, update_descriptor_queue),
+ buffer_cache(*this, gpu_memory, cpu_memory_, device, memory_manager, scheduler, stream_buffer,
+ staging_pool),
+ query_cache{*this, maxwell3d, gpu_memory, device, scheduler},
fence_manager(*this, gpu, gpu_memory, texture_cache, buffer_cache, query_cache, device,
scheduler),
- wfi_event(device.GetLogical().CreateEvent()), async_shaders(emu_window) {
+ wfi_event(device.GetLogical().CreateEvent()), async_shaders(emu_window_) {
scheduler.SetQueryCache(query_cache);
if (device.UseAsynchronousShaders()) {
async_shaders.AllocateWorkers();
@@ -427,9 +456,10 @@ void RasterizerVulkan::Draw(bool is_indexed, bool is_instanced) {
const DrawParameters draw_params =
SetupGeometry(key.fixed_state, buffer_bindings, is_indexed, is_instanced);
- update_descriptor_queue.Acquire();
- sampled_views.clear();
- image_views.clear();
+ auto lock = texture_cache.AcquireLock();
+ texture_cache.SynchronizeGraphicsDescriptors();
+
+ texture_cache.UpdateRenderTargets(false);
const auto shaders = pipeline_cache.GetShaders();
key.shaders = GetShaderAddresses(shaders);
@@ -437,30 +467,24 @@ void RasterizerVulkan::Draw(bool is_indexed, bool is_instanced) {
buffer_cache.Unmap();
- const Texceptions texceptions = UpdateAttachments(false);
- SetupImageTransitions(texceptions, color_attachments, zeta_attachment);
-
- key.renderpass_params = GetRenderPassParams(texceptions);
- key.padding = 0;
+ const Framebuffer* const framebuffer = texture_cache.GetFramebuffer();
+ key.renderpass = framebuffer->RenderPass();
- auto* pipeline = pipeline_cache.GetGraphicsPipeline(key, async_shaders);
+ auto* const pipeline =
+ pipeline_cache.GetGraphicsPipeline(key, framebuffer->NumColorBuffers(), async_shaders);
if (pipeline == nullptr || pipeline->GetHandle() == VK_NULL_HANDLE) {
// Async graphics pipeline was not ready.
return;
}
- scheduler.BindGraphicsPipeline(pipeline->GetHandle());
-
- const auto renderpass = pipeline->GetRenderPass();
- const auto [framebuffer, render_area] = ConfigureFramebuffers(renderpass);
- scheduler.RequestRenderpass(renderpass, framebuffer, render_area);
-
- UpdateDynamicStates();
-
buffer_bindings.Bind(device, scheduler);
BeginTransformFeedback();
+ scheduler.RequestRenderpass(framebuffer);
+ scheduler.BindGraphicsPipeline(pipeline->GetHandle());
+ UpdateDynamicStates();
+
const auto pipeline_layout = pipeline->GetLayout();
const auto descriptor_set = pipeline->CommitDescriptorSet();
scheduler.Record([pipeline_layout, descriptor_set, draw_params](vk::CommandBuffer cmdbuf) {
@@ -481,9 +505,6 @@ void RasterizerVulkan::Clear() {
return;
}
- sampled_views.clear();
- image_views.clear();
-
query_cache.UpdateCounters();
const auto& regs = maxwell3d.regs;
@@ -495,20 +516,24 @@ void RasterizerVulkan::Clear() {
return;
}
- [[maybe_unused]] const auto texceptions = UpdateAttachments(true);
- DEBUG_ASSERT(texceptions.none());
- SetupImageTransitions(0, color_attachments, zeta_attachment);
+ auto lock = texture_cache.AcquireLock();
+ texture_cache.UpdateRenderTargets(true);
+ const Framebuffer* const framebuffer = texture_cache.GetFramebuffer();
+ const VkExtent2D render_area = framebuffer->RenderArea();
+ scheduler.RequestRenderpass(framebuffer);
- const VkRenderPass renderpass = renderpass_cache.GetRenderPass(GetRenderPassParams(0));
- const auto [framebuffer, render_area] = ConfigureFramebuffers(renderpass);
- scheduler.RequestRenderpass(renderpass, framebuffer, render_area);
-
- VkClearRect clear_rect;
- clear_rect.baseArrayLayer = regs.clear_buffers.layer;
- clear_rect.layerCount = 1;
- clear_rect.rect = GetScissorState(regs, 0);
- clear_rect.rect.extent.width = std::min(clear_rect.rect.extent.width, render_area.width);
- clear_rect.rect.extent.height = std::min(clear_rect.rect.extent.height, render_area.height);
+ VkClearRect clear_rect{
+ .rect = GetScissorState(regs, 0),
+ .baseArrayLayer = regs.clear_buffers.layer,
+ .layerCount = 1,
+ };
+ if (clear_rect.rect.extent.width == 0 || clear_rect.rect.extent.height == 0) {
+ return;
+ }
+ clear_rect.rect.extent = VkExtent2D{
+ .width = std::min(clear_rect.rect.extent.width, render_area.width),
+ .height = std::min(clear_rect.rect.extent.height, render_area.height),
+ };
if (use_color) {
VkClearValue clear_value;
@@ -549,9 +574,6 @@ void RasterizerVulkan::Clear() {
void RasterizerVulkan::DispatchCompute(GPUVAddr code_addr) {
MICROPROFILE_SCOPE(Vulkan_Compute);
- update_descriptor_queue.Acquire();
- sampled_views.clear();
- image_views.clear();
query_cache.UpdateCounters();
@@ -570,29 +592,43 @@ void RasterizerVulkan::DispatchCompute(GPUVAddr code_addr) {
// Compute dispatches can't be executed inside a renderpass
scheduler.RequestOutsideRenderPassOperationContext();
- buffer_cache.Map(CalculateComputeStreamBufferSize());
+ image_view_indices.clear();
+ sampler_handles.clear();
+
+ auto lock = texture_cache.AcquireLock();
+ texture_cache.SynchronizeComputeDescriptors();
const auto& entries = pipeline.GetEntries();
- SetupComputeConstBuffers(entries);
- SetupComputeGlobalBuffers(entries);
SetupComputeUniformTexels(entries);
SetupComputeTextures(entries);
SetupComputeStorageTexels(entries);
SetupComputeImages(entries);
- buffer_cache.Unmap();
+ const std::span indices_span(image_view_indices.data(), image_view_indices.size());
+ texture_cache.FillComputeImageViews(indices_span, image_view_ids);
- TransitionImages(sampled_views, VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT,
- VK_ACCESS_SHADER_READ_BIT);
- TransitionImages(image_views, VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT,
- VK_ACCESS_SHADER_READ_BIT | VK_ACCESS_SHADER_WRITE_BIT);
+ buffer_cache.Map(CalculateComputeStreamBufferSize());
+ update_descriptor_queue.Acquire();
+
+ SetupComputeConstBuffers(entries);
+ SetupComputeGlobalBuffers(entries);
+
+ ImageViewId* image_view_id_ptr = image_view_ids.data();
+ VkSampler* sampler_ptr = sampler_handles.data();
+ PushImageDescriptors(entries, texture_cache, update_descriptor_queue, image_view_id_ptr,
+ sampler_ptr);
+
+ buffer_cache.Unmap();
+
+ const VkPipeline pipeline_handle = pipeline.GetHandle();
+ const VkPipelineLayout pipeline_layout = pipeline.GetLayout();
+ const VkDescriptorSet descriptor_set = pipeline.CommitDescriptorSet();
scheduler.Record([grid_x = launch_desc.grid_dim_x, grid_y = launch_desc.grid_dim_y,
- grid_z = launch_desc.grid_dim_z, pipeline_handle = pipeline.GetHandle(),
- layout = pipeline.GetLayout(),
- descriptor_set = pipeline.CommitDescriptorSet()](vk::CommandBuffer cmdbuf) {
+ grid_z = launch_desc.grid_dim_z, pipeline_handle, pipeline_layout,
+ descriptor_set](vk::CommandBuffer cmdbuf) {
cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_COMPUTE, pipeline_handle);
- cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_COMPUTE, layout, DESCRIPTOR_SET,
+ cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_COMPUTE, pipeline_layout, DESCRIPTOR_SET,
descriptor_set, {});
cmdbuf.Dispatch(grid_x, grid_y, grid_z);
});
@@ -613,7 +649,10 @@ void RasterizerVulkan::FlushRegion(VAddr addr, u64 size) {
if (addr == 0 || size == 0) {
return;
}
- texture_cache.FlushRegion(addr, size);
+ {
+ auto lock = texture_cache.AcquireLock();
+ texture_cache.DownloadMemory(addr, size);
+ }
buffer_cache.FlushRegion(addr, size);
query_cache.FlushRegion(addr, size);
}
@@ -622,14 +661,18 @@ bool RasterizerVulkan::MustFlushRegion(VAddr addr, u64 size) {
if (!Settings::IsGPULevelHigh()) {
return buffer_cache.MustFlushRegion(addr, size);
}
- return texture_cache.MustFlushRegion(addr, size) || buffer_cache.MustFlushRegion(addr, size);
+ return texture_cache.IsRegionGpuModified(addr, size) ||
+ buffer_cache.MustFlushRegion(addr, size);
}
void RasterizerVulkan::InvalidateRegion(VAddr addr, u64 size) {
if (addr == 0 || size == 0) {
return;
}
- texture_cache.InvalidateRegion(addr, size);
+ {
+ auto lock = texture_cache.AcquireLock();
+ texture_cache.WriteMemory(addr, size);
+ }
pipeline_cache.InvalidateRegion(addr, size);
buffer_cache.InvalidateRegion(addr, size);
query_cache.InvalidateRegion(addr, size);
@@ -639,17 +682,28 @@ void RasterizerVulkan::OnCPUWrite(VAddr addr, u64 size) {
if (addr == 0 || size == 0) {
return;
}
- texture_cache.OnCPUWrite(addr, size);
+ {
+ auto lock = texture_cache.AcquireLock();
+ texture_cache.WriteMemory(addr, size);
+ }
pipeline_cache.OnCPUWrite(addr, size);
buffer_cache.OnCPUWrite(addr, size);
}
void RasterizerVulkan::SyncGuestHost() {
- texture_cache.SyncGuestHost();
buffer_cache.SyncGuestHost();
pipeline_cache.SyncGuestHost();
}
+void RasterizerVulkan::UnmapMemory(VAddr addr, u64 size) {
+ {
+ auto lock = texture_cache.AcquireLock();
+ texture_cache.UnmapMemory(addr, size);
+ }
+ buffer_cache.OnCPUWrite(addr, size);
+ pipeline_cache.OnCPUWrite(addr, size);
+}
+
void RasterizerVulkan::SignalSemaphore(GPUVAddr addr, u32 value) {
if (!gpu.IsAsync()) {
gpu_memory.Write<u32>(addr, value);
@@ -700,6 +754,14 @@ void RasterizerVulkan::WaitForIdle() {
});
}
+void RasterizerVulkan::FragmentBarrier() {
+ // We already put barriers when a render pass finishes
+}
+
+void RasterizerVulkan::TiledCacheBarrier() {
+ // TODO: Implementing tiled barriers requires rewriting a good chunk of the Vulkan backend
+}
+
void RasterizerVulkan::FlushCommands() {
if (draw_counter > 0) {
draw_counter = 0;
@@ -710,14 +772,20 @@ void RasterizerVulkan::FlushCommands() {
void RasterizerVulkan::TickFrame() {
draw_counter = 0;
update_descriptor_queue.TickFrame();
+ fence_manager.TickFrame();
buffer_cache.TickFrame();
staging_pool.TickFrame();
+ {
+ auto lock = texture_cache.AcquireLock();
+ texture_cache.TickFrame();
+ }
}
-bool RasterizerVulkan::AccelerateSurfaceCopy(const Tegra::Engines::Fermi2D::Regs::Surface& src,
- const Tegra::Engines::Fermi2D::Regs::Surface& dst,
+bool RasterizerVulkan::AccelerateSurfaceCopy(const Tegra::Engines::Fermi2D::Surface& src,
+ const Tegra::Engines::Fermi2D::Surface& dst,
const Tegra::Engines::Fermi2D::Config& copy_config) {
- texture_cache.DoFermiCopy(src, dst, copy_config);
+ auto lock = texture_cache.AcquireLock();
+ texture_cache.BlitImage(dst, src, copy_config);
return true;
}
@@ -727,20 +795,16 @@ bool RasterizerVulkan::AccelerateDisplay(const Tegra::FramebufferConfig& config,
return false;
}
- const auto surface{texture_cache.TryFindFramebufferSurface(framebuffer_addr)};
- if (!surface) {
+ auto lock = texture_cache.AcquireLock();
+ ImageView* const image_view = texture_cache.TryFindFramebufferImageView(framebuffer_addr);
+ if (!image_view) {
return false;
}
- // Verify that the cached surface is the same size and format as the requested framebuffer
- const auto& params{surface->GetSurfaceParams()};
- ASSERT_MSG(params.width == config.width, "Framebuffer width is different");
- ASSERT_MSG(params.height == config.height, "Framebuffer height is different");
-
- screen_info.image = &surface->GetImage();
- screen_info.width = params.width;
- screen_info.height = params.height;
- screen_info.is_srgb = surface->GetSurfaceParams().srgb_conversion;
+ screen_info.image_view = image_view->Handle(VideoCommon::ImageViewType::e2D);
+ screen_info.width = image_view->size.width;
+ screen_info.height = image_view->size.height;
+ screen_info.is_srgb = VideoCore::Surface::IsPixelFormatSRGB(image_view->format);
return true;
}
@@ -765,103 +829,6 @@ void RasterizerVulkan::FlushWork() {
draw_counter = 0;
}
-RasterizerVulkan::Texceptions RasterizerVulkan::UpdateAttachments(bool is_clear) {
- MICROPROFILE_SCOPE(Vulkan_RenderTargets);
-
- const auto& regs = maxwell3d.regs;
- auto& dirty = maxwell3d.dirty.flags;
- const bool update_rendertargets = dirty[VideoCommon::Dirty::RenderTargets];
- dirty[VideoCommon::Dirty::RenderTargets] = false;
-
- texture_cache.GuardRenderTargets(true);
-
- Texceptions texceptions;
- for (std::size_t rt = 0; rt < Maxwell::NumRenderTargets; ++rt) {
- if (update_rendertargets) {
- const bool preserve_contents = HasToPreserveColorContents(is_clear, regs);
- color_attachments[rt] = texture_cache.GetColorBufferSurface(rt, preserve_contents);
- }
- if (color_attachments[rt] && WalkAttachmentOverlaps(*color_attachments[rt])) {
- texceptions[rt] = true;
- }
- }
-
- if (update_rendertargets) {
- const bool preserve_contents = HasToPreserveDepthContents(is_clear, regs);
- zeta_attachment = texture_cache.GetDepthBufferSurface(preserve_contents);
- }
- if (zeta_attachment && WalkAttachmentOverlaps(*zeta_attachment)) {
- texceptions[ZETA_TEXCEPTION_INDEX] = true;
- }
-
- texture_cache.GuardRenderTargets(false);
-
- return texceptions;
-}
-
-bool RasterizerVulkan::WalkAttachmentOverlaps(const CachedSurfaceView& attachment) {
- bool overlap = false;
- for (auto& [view, layout] : sampled_views) {
- if (!attachment.IsSameSurface(*view)) {
- continue;
- }
- overlap = true;
- *layout = VK_IMAGE_LAYOUT_GENERAL;
- }
- return overlap;
-}
-
-std::tuple<VkFramebuffer, VkExtent2D> RasterizerVulkan::ConfigureFramebuffers(
- VkRenderPass renderpass) {
- FramebufferCacheKey key{
- .renderpass = renderpass,
- .width = std::numeric_limits<u32>::max(),
- .height = std::numeric_limits<u32>::max(),
- .layers = std::numeric_limits<u32>::max(),
- .views = {},
- };
-
- const auto try_push = [&key](const View& view) {
- if (!view) {
- return false;
- }
- key.views.push_back(view->GetAttachment());
- key.width = std::min(key.width, view->GetWidth());
- key.height = std::min(key.height, view->GetHeight());
- key.layers = std::min(key.layers, view->GetNumLayers());
- return true;
- };
-
- const auto& regs = maxwell3d.regs;
- const std::size_t num_attachments = static_cast<std::size_t>(regs.rt_control.count);
- for (std::size_t index = 0; index < num_attachments; ++index) {
- if (try_push(color_attachments[index])) {
- texture_cache.MarkColorBufferInUse(index);
- }
- }
- if (try_push(zeta_attachment)) {
- texture_cache.MarkDepthBufferInUse();
- }
-
- const auto [fbentry, is_cache_miss] = framebuffer_cache.try_emplace(key);
- auto& framebuffer = fbentry->second;
- if (is_cache_miss) {
- framebuffer = device.GetLogical().CreateFramebuffer({
- .sType = VK_STRUCTURE_TYPE_FRAMEBUFFER_CREATE_INFO,
- .pNext = nullptr,
- .flags = 0,
- .renderPass = key.renderpass,
- .attachmentCount = static_cast<u32>(key.views.size()),
- .pAttachments = key.views.data(),
- .width = key.width,
- .height = key.height,
- .layers = key.layers,
- });
- }
-
- return {*framebuffer, VkExtent2D{key.width, key.height}};
-}
-
RasterizerVulkan::DrawParameters RasterizerVulkan::SetupGeometry(FixedPipelineState& fixed_state,
BufferBindings& buffer_bindings,
bool is_indexed,
@@ -885,51 +852,37 @@ RasterizerVulkan::DrawParameters RasterizerVulkan::SetupGeometry(FixedPipelineSt
void RasterizerVulkan::SetupShaderDescriptors(
const std::array<Shader*, Maxwell::MaxShaderProgram>& shaders) {
- texture_cache.GuardSamplers(true);
-
- for (std::size_t stage = 0; stage < Maxwell::MaxShaderStage; ++stage) {
- // Skip VertexA stage
+ image_view_indices.clear();
+ sampler_handles.clear();
+ for (size_t stage = 0; stage < Maxwell::MaxShaderStage; ++stage) {
Shader* const shader = shaders[stage + 1];
if (!shader) {
continue;
}
const auto& entries = shader->GetEntries();
- SetupGraphicsConstBuffers(entries, stage);
- SetupGraphicsGlobalBuffers(entries, stage);
SetupGraphicsUniformTexels(entries, stage);
SetupGraphicsTextures(entries, stage);
SetupGraphicsStorageTexels(entries, stage);
SetupGraphicsImages(entries, stage);
}
- texture_cache.GuardSamplers(false);
-}
+ const std::span indices_span(image_view_indices.data(), image_view_indices.size());
+ texture_cache.FillGraphicsImageViews(indices_span, image_view_ids);
-void RasterizerVulkan::SetupImageTransitions(
- Texceptions texceptions, const std::array<View, Maxwell::NumRenderTargets>& color_attachments,
- const View& zeta_attachment) {
- TransitionImages(sampled_views, VK_PIPELINE_STAGE_ALL_GRAPHICS_BIT, VK_ACCESS_SHADER_READ_BIT);
- TransitionImages(image_views, VK_PIPELINE_STAGE_ALL_GRAPHICS_BIT,
- VK_ACCESS_SHADER_READ_BIT | VK_ACCESS_SHADER_WRITE_BIT);
+ update_descriptor_queue.Acquire();
- for (std::size_t rt = 0; rt < std::size(color_attachments); ++rt) {
- const auto color_attachment = color_attachments[rt];
- if (color_attachment == nullptr) {
+ ImageViewId* image_view_id_ptr = image_view_ids.data();
+ VkSampler* sampler_ptr = sampler_handles.data();
+ for (size_t stage = 0; stage < Maxwell::MaxShaderStage; ++stage) {
+ // Skip VertexA stage
+ Shader* const shader = shaders[stage + 1];
+ if (!shader) {
continue;
}
- const auto image_layout =
- texceptions[rt] ? VK_IMAGE_LAYOUT_GENERAL : VK_IMAGE_LAYOUT_COLOR_ATTACHMENT_OPTIMAL;
- color_attachment->Transition(image_layout, VK_PIPELINE_STAGE_COLOR_ATTACHMENT_OUTPUT_BIT,
- VK_ACCESS_COLOR_ATTACHMENT_READ_BIT |
- VK_ACCESS_COLOR_ATTACHMENT_WRITE_BIT);
- }
-
- if (zeta_attachment != nullptr) {
- const auto image_layout = texceptions[ZETA_TEXCEPTION_INDEX]
- ? VK_IMAGE_LAYOUT_GENERAL
- : VK_IMAGE_LAYOUT_DEPTH_STENCIL_ATTACHMENT_OPTIMAL;
- zeta_attachment->Transition(image_layout, VK_PIPELINE_STAGE_LATE_FRAGMENT_TESTS_BIT,
- VK_ACCESS_DEPTH_STENCIL_ATTACHMENT_READ_BIT |
- VK_ACCESS_DEPTH_STENCIL_ATTACHMENT_WRITE_BIT);
+ const auto& entries = shader->GetEntries();
+ SetupGraphicsConstBuffers(entries, stage);
+ SetupGraphicsGlobalBuffers(entries, stage);
+ PushImageDescriptors(entries, texture_cache, update_descriptor_queue, image_view_id_ptr,
+ sampler_ptr);
}
}
@@ -1001,7 +954,7 @@ void RasterizerVulkan::EndTransformFeedback() {
void RasterizerVulkan::SetupVertexArrays(BufferBindings& buffer_bindings) {
const auto& regs = maxwell3d.regs;
- for (std::size_t index = 0; index < Maxwell::NumVertexArrays; ++index) {
+ for (size_t index = 0; index < Maxwell::NumVertexArrays; ++index) {
const auto& vertex_array = regs.vertex_array[index];
if (!vertex_array.IsEnabled()) {
continue;
@@ -1010,7 +963,7 @@ void RasterizerVulkan::SetupVertexArrays(BufferBindings& buffer_bindings) {
const GPUVAddr end{regs.vertex_array_limit[index].LimitAddress()};
ASSERT(end >= start);
- const std::size_t size = end - start;
+ const size_t size = end - start;
if (size == 0) {
buffer_bindings.AddVertexBinding(DefaultBuffer(), 0, DEFAULT_BUFFER_SIZE, 0);
continue;
@@ -1071,7 +1024,7 @@ void RasterizerVulkan::SetupIndexBuffer(BufferBindings& buffer_bindings, DrawPar
}
}
-void RasterizerVulkan::SetupGraphicsConstBuffers(const ShaderEntries& entries, std::size_t stage) {
+void RasterizerVulkan::SetupGraphicsConstBuffers(const ShaderEntries& entries, size_t stage) {
MICROPROFILE_SCOPE(Vulkan_ConstBuffers);
const auto& shader_stage = maxwell3d.state.shader_stages[stage];
for (const auto& entry : entries.const_buffers) {
@@ -1079,7 +1032,7 @@ void RasterizerVulkan::SetupGraphicsConstBuffers(const ShaderEntries& entries, s
}
}
-void RasterizerVulkan::SetupGraphicsGlobalBuffers(const ShaderEntries& entries, std::size_t stage) {
+void RasterizerVulkan::SetupGraphicsGlobalBuffers(const ShaderEntries& entries, size_t stage) {
MICROPROFILE_SCOPE(Vulkan_GlobalBuffers);
const auto& cbufs{maxwell3d.state.shader_stages[stage]};
@@ -1089,37 +1042,49 @@ void RasterizerVulkan::SetupGraphicsGlobalBuffers(const ShaderEntries& entries,
}
}
-void RasterizerVulkan::SetupGraphicsUniformTexels(const ShaderEntries& entries, std::size_t stage) {
+void RasterizerVulkan::SetupGraphicsUniformTexels(const ShaderEntries& entries, size_t stage) {
MICROPROFILE_SCOPE(Vulkan_Textures);
+ const auto& regs = maxwell3d.regs;
+ const bool via_header_index = regs.sampler_index == Maxwell::SamplerIndex::ViaHeaderIndex;
for (const auto& entry : entries.uniform_texels) {
- const auto image = GetTextureInfo(maxwell3d, entry, stage).tic;
- SetupUniformTexels(image, entry);
+ const TextureHandle handle = GetTextureInfo(maxwell3d, via_header_index, entry, stage);
+ image_view_indices.push_back(handle.image);
}
}
-void RasterizerVulkan::SetupGraphicsTextures(const ShaderEntries& entries, std::size_t stage) {
+void RasterizerVulkan::SetupGraphicsTextures(const ShaderEntries& entries, size_t stage) {
MICROPROFILE_SCOPE(Vulkan_Textures);
+ const auto& regs = maxwell3d.regs;
+ const bool via_header_index = regs.sampler_index == Maxwell::SamplerIndex::ViaHeaderIndex;
for (const auto& entry : entries.samplers) {
- for (std::size_t i = 0; i < entry.size; ++i) {
- const auto texture = GetTextureInfo(maxwell3d, entry, stage, i);
- SetupTexture(texture, entry);
+ for (size_t index = 0; index < entry.size; ++index) {
+ const TextureHandle handle =
+ GetTextureInfo(maxwell3d, via_header_index, entry, stage, index);
+ image_view_indices.push_back(handle.image);
+
+ Sampler* const sampler = texture_cache.GetGraphicsSampler(handle.sampler);
+ sampler_handles.push_back(sampler->Handle());
}
}
}
-void RasterizerVulkan::SetupGraphicsStorageTexels(const ShaderEntries& entries, std::size_t stage) {
+void RasterizerVulkan::SetupGraphicsStorageTexels(const ShaderEntries& entries, size_t stage) {
MICROPROFILE_SCOPE(Vulkan_Textures);
+ const auto& regs = maxwell3d.regs;
+ const bool via_header_index = regs.sampler_index == Maxwell::SamplerIndex::ViaHeaderIndex;
for (const auto& entry : entries.storage_texels) {
- const auto image = GetTextureInfo(maxwell3d, entry, stage).tic;
- SetupStorageTexel(image, entry);
+ const TextureHandle handle = GetTextureInfo(maxwell3d, via_header_index, entry, stage);
+ image_view_indices.push_back(handle.image);
}
}
-void RasterizerVulkan::SetupGraphicsImages(const ShaderEntries& entries, std::size_t stage) {
+void RasterizerVulkan::SetupGraphicsImages(const ShaderEntries& entries, size_t stage) {
MICROPROFILE_SCOPE(Vulkan_Images);
+ const auto& regs = maxwell3d.regs;
+ const bool via_header_index = regs.sampler_index == Maxwell::SamplerIndex::ViaHeaderIndex;
for (const auto& entry : entries.images) {
- const auto tic = GetTextureInfo(maxwell3d, entry, stage).tic;
- SetupImage(tic, entry);
+ const TextureHandle handle = GetTextureInfo(maxwell3d, via_header_index, entry, stage);
+ image_view_indices.push_back(handle.image);
}
}
@@ -1129,11 +1094,12 @@ void RasterizerVulkan::SetupComputeConstBuffers(const ShaderEntries& entries) {
for (const auto& entry : entries.const_buffers) {
const auto& config = launch_desc.const_buffer_config[entry.GetIndex()];
const std::bitset<8> mask = launch_desc.const_buffer_enable_mask.Value();
- Tegra::Engines::ConstBufferInfo buffer;
- buffer.address = config.Address();
- buffer.size = config.size;
- buffer.enabled = mask[entry.GetIndex()];
- SetupConstBuffer(entry, buffer);
+ const Tegra::Engines::ConstBufferInfo info{
+ .address = config.Address(),
+ .size = config.size,
+ .enabled = mask[entry.GetIndex()],
+ };
+ SetupConstBuffer(entry, info);
}
}
@@ -1148,35 +1114,46 @@ void RasterizerVulkan::SetupComputeGlobalBuffers(const ShaderEntries& entries) {
void RasterizerVulkan::SetupComputeUniformTexels(const ShaderEntries& entries) {
MICROPROFILE_SCOPE(Vulkan_Textures);
+ const bool via_header_index = kepler_compute.launch_description.linked_tsc;
for (const auto& entry : entries.uniform_texels) {
- const auto image = GetTextureInfo(kepler_compute, entry, ComputeShaderIndex).tic;
- SetupUniformTexels(image, entry);
+ const TextureHandle handle =
+ GetTextureInfo(kepler_compute, via_header_index, entry, COMPUTE_SHADER_INDEX);
+ image_view_indices.push_back(handle.image);
}
}
void RasterizerVulkan::SetupComputeTextures(const ShaderEntries& entries) {
MICROPROFILE_SCOPE(Vulkan_Textures);
+ const bool via_header_index = kepler_compute.launch_description.linked_tsc;
for (const auto& entry : entries.samplers) {
- for (std::size_t i = 0; i < entry.size; ++i) {
- const auto texture = GetTextureInfo(kepler_compute, entry, ComputeShaderIndex, i);
- SetupTexture(texture, entry);
+ for (size_t index = 0; index < entry.size; ++index) {
+ const TextureHandle handle = GetTextureInfo(kepler_compute, via_header_index, entry,
+ COMPUTE_SHADER_INDEX, index);
+ image_view_indices.push_back(handle.image);
+
+ Sampler* const sampler = texture_cache.GetComputeSampler(handle.sampler);
+ sampler_handles.push_back(sampler->Handle());
}
}
}
void RasterizerVulkan::SetupComputeStorageTexels(const ShaderEntries& entries) {
MICROPROFILE_SCOPE(Vulkan_Textures);
+ const bool via_header_index = kepler_compute.launch_description.linked_tsc;
for (const auto& entry : entries.storage_texels) {
- const auto image = GetTextureInfo(kepler_compute, entry, ComputeShaderIndex).tic;
- SetupStorageTexel(image, entry);
+ const TextureHandle handle =
+ GetTextureInfo(kepler_compute, via_header_index, entry, COMPUTE_SHADER_INDEX);
+ image_view_indices.push_back(handle.image);
}
}
void RasterizerVulkan::SetupComputeImages(const ShaderEntries& entries) {
MICROPROFILE_SCOPE(Vulkan_Images);
+ const bool via_header_index = kepler_compute.launch_description.linked_tsc;
for (const auto& entry : entries.images) {
- const auto tic = GetTextureInfo(kepler_compute, entry, ComputeShaderIndex).tic;
- SetupImage(tic, entry);
+ const TextureHandle handle =
+ GetTextureInfo(kepler_compute, via_header_index, entry, COMPUTE_SHADER_INDEX);
+ image_view_indices.push_back(handle.image);
}
}
@@ -1187,14 +1164,12 @@ void RasterizerVulkan::SetupConstBuffer(const ConstBufferEntry& entry,
update_descriptor_queue.AddBuffer(DefaultBuffer(), 0, DEFAULT_BUFFER_SIZE);
return;
}
-
// Align the size to avoid bad std140 interactions
- const std::size_t size =
- Common::AlignUp(CalculateConstBufferSize(entry, buffer), 4 * sizeof(float));
+ const size_t size = Common::AlignUp(CalculateConstBufferSize(entry, buffer), 4 * sizeof(float));
ASSERT(size <= MaxConstbufferSize);
- const auto info =
- buffer_cache.UploadMemory(buffer.address, size, device.GetUniformBufferAlignment());
+ const u64 alignment = device.GetUniformBufferAlignment();
+ const auto info = buffer_cache.UploadMemory(buffer.address, size, alignment);
update_descriptor_queue.AddBuffer(info.handle, info.offset, size);
}
@@ -1207,7 +1182,7 @@ void RasterizerVulkan::SetupGlobalBuffer(const GlobalBufferEntry& entry, GPUVAdd
// because Vulkan doesn't like empty buffers.
// Note: Do *not* use DefaultBuffer() here, storage buffers can be written breaking the
// default buffer.
- static constexpr std::size_t dummy_size = 4;
+ static constexpr size_t dummy_size = 4;
const auto info = buffer_cache.GetEmptyBuffer(dummy_size);
update_descriptor_queue.AddBuffer(info.handle, info.offset, dummy_size);
return;
@@ -1218,55 +1193,6 @@ void RasterizerVulkan::SetupGlobalBuffer(const GlobalBufferEntry& entry, GPUVAdd
update_descriptor_queue.AddBuffer(info.handle, info.offset, size);
}
-void RasterizerVulkan::SetupUniformTexels(const Tegra::Texture::TICEntry& tic,
- const UniformTexelEntry& entry) {
- const auto view = texture_cache.GetTextureSurface(tic, entry);
- ASSERT(view->IsBufferView());
-
- update_descriptor_queue.AddTexelBuffer(view->GetBufferView());
-}
-
-void RasterizerVulkan::SetupTexture(const Tegra::Texture::FullTextureInfo& texture,
- const SamplerEntry& entry) {
- auto view = texture_cache.GetTextureSurface(texture.tic, entry);
- ASSERT(!view->IsBufferView());
-
- const VkImageView image_view = view->GetImageView(texture.tic.x_source, texture.tic.y_source,
- texture.tic.z_source, texture.tic.w_source);
- const auto sampler = sampler_cache.GetSampler(texture.tsc);
- update_descriptor_queue.AddSampledImage(sampler, image_view);
-
- VkImageLayout* const image_layout = update_descriptor_queue.LastImageLayout();
- *image_layout = VK_IMAGE_LAYOUT_SHADER_READ_ONLY_OPTIMAL;
- sampled_views.push_back(ImageView{std::move(view), image_layout});
-}
-
-void RasterizerVulkan::SetupStorageTexel(const Tegra::Texture::TICEntry& tic,
- const StorageTexelEntry& entry) {
- const auto view = texture_cache.GetImageSurface(tic, entry);
- ASSERT(view->IsBufferView());
-
- update_descriptor_queue.AddTexelBuffer(view->GetBufferView());
-}
-
-void RasterizerVulkan::SetupImage(const Tegra::Texture::TICEntry& tic, const ImageEntry& entry) {
- auto view = texture_cache.GetImageSurface(tic, entry);
-
- if (entry.is_written) {
- view->MarkAsModified(texture_cache.Tick());
- }
-
- UNIMPLEMENTED_IF(tic.IsBuffer());
-
- const VkImageView image_view =
- view->GetImageView(tic.x_source, tic.y_source, tic.z_source, tic.w_source);
- update_descriptor_queue.AddImage(image_view);
-
- VkImageLayout* const image_layout = update_descriptor_queue.LastImageLayout();
- *image_layout = VK_IMAGE_LAYOUT_GENERAL;
- image_views.push_back(ImageView{std::move(view), image_layout});
-}
-
void RasterizerVulkan::UpdateViewportsState(Tegra::Engines::Maxwell3D::Regs& regs) {
if (!state_tracker.TouchViewports()) {
return;
@@ -1458,8 +1384,8 @@ void RasterizerVulkan::UpdateStencilTestEnable(Tegra::Engines::Maxwell3D::Regs&
});
}
-std::size_t RasterizerVulkan::CalculateGraphicsStreamBufferSize(bool is_indexed) const {
- std::size_t size = CalculateVertexArraysSize();
+size_t RasterizerVulkan::CalculateGraphicsStreamBufferSize(bool is_indexed) const {
+ size_t size = CalculateVertexArraysSize();
if (is_indexed) {
size = Common::AlignUp(size, 4) + CalculateIndexBufferSize();
}
@@ -1467,15 +1393,15 @@ std::size_t RasterizerVulkan::CalculateGraphicsStreamBufferSize(bool is_indexed)
return size;
}
-std::size_t RasterizerVulkan::CalculateComputeStreamBufferSize() const {
+size_t RasterizerVulkan::CalculateComputeStreamBufferSize() const {
return Tegra::Engines::KeplerCompute::NumConstBuffers *
(Maxwell::MaxConstBufferSize + device.GetUniformBufferAlignment());
}
-std::size_t RasterizerVulkan::CalculateVertexArraysSize() const {
+size_t RasterizerVulkan::CalculateVertexArraysSize() const {
const auto& regs = maxwell3d.regs;
- std::size_t size = 0;
+ size_t size = 0;
for (u32 index = 0; index < Maxwell::NumVertexArrays; ++index) {
// This implementation assumes that all attributes are used in the shader.
const GPUVAddr start{regs.vertex_array[index].StartAddress()};
@@ -1487,12 +1413,12 @@ std::size_t RasterizerVulkan::CalculateVertexArraysSize() const {
return size;
}
-std::size_t RasterizerVulkan::CalculateIndexBufferSize() const {
- return static_cast<std::size_t>(maxwell3d.regs.index_array.count) *
- static_cast<std::size_t>(maxwell3d.regs.index_array.FormatSizeInBytes());
+size_t RasterizerVulkan::CalculateIndexBufferSize() const {
+ return static_cast<size_t>(maxwell3d.regs.index_array.count) *
+ static_cast<size_t>(maxwell3d.regs.index_array.FormatSizeInBytes());
}
-std::size_t RasterizerVulkan::CalculateConstBufferSize(
+size_t RasterizerVulkan::CalculateConstBufferSize(
const ConstBufferEntry& entry, const Tegra::Engines::ConstBufferInfo& buffer) const {
if (entry.IsIndirect()) {
// Buffer is accessed indirectly, so upload the entire thing
@@ -1503,37 +1429,10 @@ std::size_t RasterizerVulkan::CalculateConstBufferSize(
}
}
-RenderPassParams RasterizerVulkan::GetRenderPassParams(Texceptions texceptions) const {
- const auto& regs = maxwell3d.regs;
- const std::size_t num_attachments = static_cast<std::size_t>(regs.rt_control.count);
-
- RenderPassParams params;
- params.color_formats = {};
- std::size_t color_texceptions = 0;
-
- std::size_t index = 0;
- for (std::size_t rt = 0; rt < num_attachments; ++rt) {
- const auto& rendertarget = regs.rt[rt];
- if (rendertarget.Address() == 0 || rendertarget.format == Tegra::RenderTargetFormat::NONE) {
- continue;
- }
- params.color_formats[index] = static_cast<u8>(rendertarget.format);
- color_texceptions |= (texceptions[rt] ? 1ULL : 0ULL) << index;
- ++index;
- }
- params.num_color_attachments = static_cast<u8>(index);
- params.texceptions = static_cast<u8>(color_texceptions);
-
- params.zeta_format = regs.zeta_enable ? static_cast<u8>(regs.zeta.format) : 0;
- params.zeta_texception = texceptions[ZETA_TEXCEPTION_INDEX];
- return params;
-}
-
VkBuffer RasterizerVulkan::DefaultBuffer() {
if (default_buffer) {
return *default_buffer;
}
-
default_buffer = device.GetLogical().CreateBuffer({
.sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO,
.pNext = nullptr,
diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.h b/src/video_core/renderer_vulkan/vk_rasterizer.h
index 237e51fa4..990f9e031 100644
--- a/src/video_core/renderer_vulkan/vk_rasterizer.h
+++ b/src/video_core/renderer_vulkan/vk_rasterizer.h
@@ -11,11 +11,11 @@
#include <vector>
#include <boost/container/static_vector.hpp>
-#include <boost/functional/hash.hpp>
#include "common/common_types.h"
#include "video_core/rasterizer_accelerated.h"
#include "video_core/rasterizer_interface.h"
+#include "video_core/renderer_vulkan/blit_image.h"
#include "video_core/renderer_vulkan/fixed_pipeline_state.h"
#include "video_core/renderer_vulkan/vk_buffer_cache.h"
#include "video_core/renderer_vulkan/vk_compute_pass.h"
@@ -24,10 +24,9 @@
#include "video_core/renderer_vulkan/vk_memory_manager.h"
#include "video_core/renderer_vulkan/vk_pipeline_cache.h"
#include "video_core/renderer_vulkan/vk_query_cache.h"
-#include "video_core/renderer_vulkan/vk_renderpass_cache.h"
-#include "video_core/renderer_vulkan/vk_sampler_cache.h"
#include "video_core/renderer_vulkan/vk_scheduler.h"
#include "video_core/renderer_vulkan/vk_staging_buffer_pool.h"
+#include "video_core/renderer_vulkan/vk_stream_buffer.h"
#include "video_core/renderer_vulkan/vk_texture_cache.h"
#include "video_core/renderer_vulkan/vk_update_descriptor.h"
#include "video_core/renderer_vulkan/wrapper.h"
@@ -49,67 +48,16 @@ namespace Vulkan {
struct VKScreenInfo;
-using ImageViewsPack = boost::container::static_vector<VkImageView, Maxwell::NumRenderTargets + 1>;
-
-struct FramebufferCacheKey {
- VkRenderPass renderpass{};
- u32 width = 0;
- u32 height = 0;
- u32 layers = 0;
- ImageViewsPack views;
-
- std::size_t Hash() const noexcept {
- std::size_t hash = 0;
- boost::hash_combine(hash, static_cast<VkRenderPass>(renderpass));
- for (const auto& view : views) {
- boost::hash_combine(hash, static_cast<VkImageView>(view));
- }
- boost::hash_combine(hash, width);
- boost::hash_combine(hash, height);
- boost::hash_combine(hash, layers);
- return hash;
- }
-
- bool operator==(const FramebufferCacheKey& rhs) const noexcept {
- return std::tie(renderpass, views, width, height, layers) ==
- std::tie(rhs.renderpass, rhs.views, rhs.width, rhs.height, rhs.layers);
- }
-
- bool operator!=(const FramebufferCacheKey& rhs) const noexcept {
- return !operator==(rhs);
- }
-};
-
-} // namespace Vulkan
-
-namespace std {
-
-template <>
-struct hash<Vulkan::FramebufferCacheKey> {
- std::size_t operator()(const Vulkan::FramebufferCacheKey& k) const noexcept {
- return k.Hash();
- }
-};
-
-} // namespace std
-
-namespace Vulkan {
-
class StateTracker;
class BufferBindings;
-struct ImageView {
- View view;
- VkImageLayout* layout = nullptr;
-};
-
class RasterizerVulkan final : public VideoCore::RasterizerAccelerated {
public:
- explicit RasterizerVulkan(Core::Frontend::EmuWindow& emu_window, Tegra::GPU& gpu,
- Tegra::MemoryManager& gpu_memory, Core::Memory::Memory& cpu_memory,
- VKScreenInfo& screen_info, const VKDevice& device,
- VKMemoryManager& memory_manager, StateTracker& state_tracker,
- VKScheduler& scheduler);
+ explicit RasterizerVulkan(Core::Frontend::EmuWindow& emu_window_, Tegra::GPU& gpu_,
+ Tegra::MemoryManager& gpu_memory_, Core::Memory::Memory& cpu_memory_,
+ VKScreenInfo& screen_info_, const VKDevice& device_,
+ VKMemoryManager& memory_manager_, StateTracker& state_tracker_,
+ VKScheduler& scheduler_);
~RasterizerVulkan() override;
void Draw(bool is_indexed, bool is_instanced) override;
@@ -123,15 +71,18 @@ public:
void InvalidateRegion(VAddr addr, u64 size) override;
void OnCPUWrite(VAddr addr, u64 size) override;
void SyncGuestHost() override;
+ void UnmapMemory(VAddr addr, u64 size) override;
void SignalSemaphore(GPUVAddr addr, u32 value) override;
void SignalSyncPoint(u32 value) override;
void ReleaseFences() override;
void FlushAndInvalidateRegion(VAddr addr, u64 size) override;
void WaitForIdle() override;
+ void FragmentBarrier() override;
+ void TiledCacheBarrier() override;
void FlushCommands() override;
void TickFrame() override;
- bool AccelerateSurfaceCopy(const Tegra::Engines::Fermi2D::Regs::Surface& src,
- const Tegra::Engines::Fermi2D::Regs::Surface& dst,
+ bool AccelerateSurfaceCopy(const Tegra::Engines::Fermi2D::Surface& src,
+ const Tegra::Engines::Fermi2D::Surface& dst,
const Tegra::Engines::Fermi2D::Config& copy_config) override;
bool AccelerateDisplay(const Tegra::FramebufferConfig& config, VAddr framebuffer_addr,
u32 pixel_stride) override;
@@ -145,11 +96,17 @@ public:
}
/// Maximum supported size that a constbuffer can have in bytes.
- static constexpr std::size_t MaxConstbufferSize = 0x10000;
+ static constexpr size_t MaxConstbufferSize = 0x10000;
static_assert(MaxConstbufferSize % (4 * sizeof(float)) == 0,
"The maximum size of a constbuffer must be a multiple of the size of GLvec4");
private:
+ static constexpr size_t MAX_TEXTURES = 192;
+ static constexpr size_t MAX_IMAGES = 48;
+ static constexpr size_t MAX_IMAGE_VIEWS = MAX_TEXTURES + MAX_IMAGES;
+
+ static constexpr VkDeviceSize DEFAULT_BUFFER_SIZE = 4 * sizeof(float);
+
struct DrawParameters {
void Draw(vk::CommandBuffer cmdbuf) const;
@@ -160,20 +117,8 @@ private:
bool is_indexed = 0;
};
- using Texceptions = std::bitset<Maxwell::NumRenderTargets + 1>;
-
- static constexpr std::size_t ZETA_TEXCEPTION_INDEX = 8;
- static constexpr VkDeviceSize DEFAULT_BUFFER_SIZE = 4 * sizeof(float);
-
void FlushWork();
- /// @brief Updates the currently bound attachments
- /// @param is_clear True when the framebuffer is updated as a clear
- /// @return Bitfield of attachments being used as sampled textures
- Texceptions UpdateAttachments(bool is_clear);
-
- std::tuple<VkFramebuffer, VkExtent2D> ConfigureFramebuffers(VkRenderPass renderpass);
-
/// Setups geometry buffers and state.
DrawParameters SetupGeometry(FixedPipelineState& fixed_state, BufferBindings& buffer_bindings,
bool is_indexed, bool is_instanced);
@@ -181,18 +126,12 @@ private:
/// Setup descriptors in the graphics pipeline.
void SetupShaderDescriptors(const std::array<Shader*, Maxwell::MaxShaderProgram>& shaders);
- void SetupImageTransitions(Texceptions texceptions,
- const std::array<View, Maxwell::NumRenderTargets>& color_attachments,
- const View& zeta_attachment);
-
void UpdateDynamicStates();
void BeginTransformFeedback();
void EndTransformFeedback();
- bool WalkAttachmentOverlaps(const CachedSurfaceView& attachment);
-
void SetupVertexArrays(BufferBindings& buffer_bindings);
void SetupIndexBuffer(BufferBindings& buffer_bindings, DrawParameters& params, bool is_indexed);
@@ -238,14 +177,6 @@ private:
void SetupGlobalBuffer(const GlobalBufferEntry& entry, GPUVAddr address);
- void SetupUniformTexels(const Tegra::Texture::TICEntry& image, const UniformTexelEntry& entry);
-
- void SetupTexture(const Tegra::Texture::FullTextureInfo& texture, const SamplerEntry& entry);
-
- void SetupStorageTexel(const Tegra::Texture::TICEntry& tic, const StorageTexelEntry& entry);
-
- void SetupImage(const Tegra::Texture::TICEntry& tic, const ImageEntry& entry);
-
void UpdateViewportsState(Tegra::Engines::Maxwell3D::Regs& regs);
void UpdateScissorsState(Tegra::Engines::Maxwell3D::Regs& regs);
void UpdateDepthBias(Tegra::Engines::Maxwell3D::Regs& regs);
@@ -262,18 +193,16 @@ private:
void UpdateStencilOp(Tegra::Engines::Maxwell3D::Regs& regs);
void UpdateStencilTestEnable(Tegra::Engines::Maxwell3D::Regs& regs);
- std::size_t CalculateGraphicsStreamBufferSize(bool is_indexed) const;
-
- std::size_t CalculateComputeStreamBufferSize() const;
+ size_t CalculateGraphicsStreamBufferSize(bool is_indexed) const;
- std::size_t CalculateVertexArraysSize() const;
+ size_t CalculateComputeStreamBufferSize() const;
- std::size_t CalculateIndexBufferSize() const;
+ size_t CalculateVertexArraysSize() const;
- std::size_t CalculateConstBufferSize(const ConstBufferEntry& entry,
- const Tegra::Engines::ConstBufferInfo& buffer) const;
+ size_t CalculateIndexBufferSize() const;
- RenderPassParams GetRenderPassParams(Texceptions texceptions) const;
+ size_t CalculateConstBufferSize(const ConstBufferEntry& entry,
+ const Tegra::Engines::ConstBufferInfo& buffer) const;
VkBuffer DefaultBuffer();
@@ -288,18 +217,19 @@ private:
StateTracker& state_tracker;
VKScheduler& scheduler;
+ VKStreamBuffer stream_buffer;
VKStagingBufferPool staging_pool;
VKDescriptorPool descriptor_pool;
VKUpdateDescriptorQueue update_descriptor_queue;
- VKRenderPassCache renderpass_cache;
+ BlitImageHelper blit_image;
QuadArrayPass quad_array_pass;
QuadIndexedPass quad_indexed_pass;
Uint8Pass uint8_pass;
- VKTextureCache texture_cache;
+ TextureCacheRuntime texture_cache_runtime;
+ TextureCache texture_cache;
VKPipelineCache pipeline_cache;
VKBufferCache buffer_cache;
- VKSamplerCache sampler_cache;
VKQueryCache query_cache;
VKFenceManager fence_manager;
@@ -308,16 +238,11 @@ private:
vk::Event wfi_event;
VideoCommon::Shader::AsyncShaders async_shaders;
- std::array<View, Maxwell::NumRenderTargets> color_attachments;
- View zeta_attachment;
-
- std::vector<ImageView> sampled_views;
- std::vector<ImageView> image_views;
+ boost::container::static_vector<u32, MAX_IMAGE_VIEWS> image_view_indices;
+ std::array<VideoCommon::ImageViewId, MAX_IMAGE_VIEWS> image_view_ids;
+ boost::container::static_vector<VkSampler, MAX_TEXTURES> sampler_handles;
u32 draw_counter = 0;
-
- // TODO(Rodrigo): Invalidate on image destruction
- std::unordered_map<FramebufferCacheKey, vk::Framebuffer> framebuffer_cache;
};
} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/vk_renderpass_cache.cpp b/src/video_core/renderer_vulkan/vk_renderpass_cache.cpp
deleted file mode 100644
index 80284cf92..000000000
--- a/src/video_core/renderer_vulkan/vk_renderpass_cache.cpp
+++ /dev/null
@@ -1,158 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include <cstring>
-#include <memory>
-#include <vector>
-
-#include "common/cityhash.h"
-#include "video_core/engines/maxwell_3d.h"
-#include "video_core/renderer_vulkan/maxwell_to_vk.h"
-#include "video_core/renderer_vulkan/vk_device.h"
-#include "video_core/renderer_vulkan/vk_renderpass_cache.h"
-#include "video_core/renderer_vulkan/wrapper.h"
-
-namespace Vulkan {
-
-std::size_t RenderPassParams::Hash() const noexcept {
- const u64 hash = Common::CityHash64(reinterpret_cast<const char*>(this), sizeof *this);
- return static_cast<std::size_t>(hash);
-}
-
-bool RenderPassParams::operator==(const RenderPassParams& rhs) const noexcept {
- return std::memcmp(&rhs, this, sizeof *this) == 0;
-}
-
-VKRenderPassCache::VKRenderPassCache(const VKDevice& device) : device{device} {}
-
-VKRenderPassCache::~VKRenderPassCache() = default;
-
-VkRenderPass VKRenderPassCache::GetRenderPass(const RenderPassParams& params) {
- const auto [pair, is_cache_miss] = cache.try_emplace(params);
- auto& entry = pair->second;
- if (is_cache_miss) {
- entry = CreateRenderPass(params);
- }
- return *entry;
-}
-
-vk::RenderPass VKRenderPassCache::CreateRenderPass(const RenderPassParams& params) const {
- using namespace VideoCore::Surface;
- const std::size_t num_attachments = static_cast<std::size_t>(params.num_color_attachments);
-
- std::vector<VkAttachmentDescription> descriptors;
- descriptors.reserve(num_attachments);
-
- std::vector<VkAttachmentReference> color_references;
- color_references.reserve(num_attachments);
-
- for (std::size_t rt = 0; rt < num_attachments; ++rt) {
- const auto guest_format = static_cast<Tegra::RenderTargetFormat>(params.color_formats[rt]);
- const PixelFormat pixel_format = PixelFormatFromRenderTargetFormat(guest_format);
- const auto format = MaxwellToVK::SurfaceFormat(device, FormatType::Optimal, pixel_format);
- ASSERT_MSG(format.attachable, "Trying to attach a non-attachable format with format={}",
- static_cast<int>(pixel_format));
-
- // TODO(Rodrigo): Add MAY_ALIAS_BIT when it's needed.
- const VkImageLayout color_layout = ((params.texceptions >> rt) & 1) != 0
- ? VK_IMAGE_LAYOUT_GENERAL
- : VK_IMAGE_LAYOUT_COLOR_ATTACHMENT_OPTIMAL;
- descriptors.push_back({
- .flags = VK_ATTACHMENT_DESCRIPTION_MAY_ALIAS_BIT,
- .format = format.format,
- .samples = VK_SAMPLE_COUNT_1_BIT,
- .loadOp = VK_ATTACHMENT_LOAD_OP_LOAD,
- .storeOp = VK_ATTACHMENT_STORE_OP_STORE,
- .stencilLoadOp = VK_ATTACHMENT_LOAD_OP_DONT_CARE,
- .stencilStoreOp = VK_ATTACHMENT_STORE_OP_DONT_CARE,
- .initialLayout = color_layout,
- .finalLayout = color_layout,
- });
-
- color_references.push_back({
- .attachment = static_cast<u32>(rt),
- .layout = color_layout,
- });
- }
-
- VkAttachmentReference zeta_attachment_ref;
- const bool has_zeta = params.zeta_format != 0;
- if (has_zeta) {
- const auto guest_format = static_cast<Tegra::DepthFormat>(params.zeta_format);
- const PixelFormat pixel_format = PixelFormatFromDepthFormat(guest_format);
- const auto format = MaxwellToVK::SurfaceFormat(device, FormatType::Optimal, pixel_format);
- ASSERT_MSG(format.attachable, "Trying to attach a non-attachable format with format={}",
- static_cast<int>(pixel_format));
-
- const VkImageLayout zeta_layout = params.zeta_texception != 0
- ? VK_IMAGE_LAYOUT_GENERAL
- : VK_IMAGE_LAYOUT_DEPTH_STENCIL_ATTACHMENT_OPTIMAL;
- descriptors.push_back({
- .flags = 0,
- .format = format.format,
- .samples = VK_SAMPLE_COUNT_1_BIT,
- .loadOp = VK_ATTACHMENT_LOAD_OP_LOAD,
- .storeOp = VK_ATTACHMENT_STORE_OP_STORE,
- .stencilLoadOp = VK_ATTACHMENT_LOAD_OP_LOAD,
- .stencilStoreOp = VK_ATTACHMENT_STORE_OP_STORE,
- .initialLayout = zeta_layout,
- .finalLayout = zeta_layout,
- });
-
- zeta_attachment_ref = {
- .attachment = static_cast<u32>(num_attachments),
- .layout = zeta_layout,
- };
- }
-
- const VkSubpassDescription subpass_description{
- .flags = 0,
- .pipelineBindPoint = VK_PIPELINE_BIND_POINT_GRAPHICS,
- .inputAttachmentCount = 0,
- .pInputAttachments = nullptr,
- .colorAttachmentCount = static_cast<u32>(color_references.size()),
- .pColorAttachments = color_references.data(),
- .pResolveAttachments = nullptr,
- .pDepthStencilAttachment = has_zeta ? &zeta_attachment_ref : nullptr,
- .preserveAttachmentCount = 0,
- .pPreserveAttachments = nullptr,
- };
-
- VkAccessFlags access = 0;
- VkPipelineStageFlags stage = 0;
- if (!color_references.empty()) {
- access |= VK_ACCESS_COLOR_ATTACHMENT_READ_BIT | VK_ACCESS_COLOR_ATTACHMENT_WRITE_BIT;
- stage |= VK_PIPELINE_STAGE_COLOR_ATTACHMENT_OUTPUT_BIT;
- }
-
- if (has_zeta) {
- access |= VK_ACCESS_DEPTH_STENCIL_ATTACHMENT_READ_BIT |
- VK_ACCESS_DEPTH_STENCIL_ATTACHMENT_WRITE_BIT;
- stage |= VK_PIPELINE_STAGE_LATE_FRAGMENT_TESTS_BIT;
- }
-
- const VkSubpassDependency subpass_dependency{
- .srcSubpass = VK_SUBPASS_EXTERNAL,
- .dstSubpass = 0,
- .srcStageMask = stage,
- .dstStageMask = stage,
- .srcAccessMask = 0,
- .dstAccessMask = access,
- .dependencyFlags = 0,
- };
-
- return device.GetLogical().CreateRenderPass({
- .sType = VK_STRUCTURE_TYPE_RENDER_PASS_CREATE_INFO,
- .pNext = nullptr,
- .flags = 0,
- .attachmentCount = static_cast<u32>(descriptors.size()),
- .pAttachments = descriptors.data(),
- .subpassCount = 1,
- .pSubpasses = &subpass_description,
- .dependencyCount = 1,
- .pDependencies = &subpass_dependency,
- });
-}
-
-} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/vk_renderpass_cache.h b/src/video_core/renderer_vulkan/vk_renderpass_cache.h
deleted file mode 100644
index 8b0fec720..000000000
--- a/src/video_core/renderer_vulkan/vk_renderpass_cache.h
+++ /dev/null
@@ -1,70 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <type_traits>
-#include <unordered_map>
-
-#include <boost/container/static_vector.hpp>
-#include <boost/functional/hash.hpp>
-
-#include "video_core/engines/maxwell_3d.h"
-#include "video_core/renderer_vulkan/wrapper.h"
-#include "video_core/surface.h"
-
-namespace Vulkan {
-
-class VKDevice;
-
-struct RenderPassParams {
- std::array<u8, Tegra::Engines::Maxwell3D::Regs::NumRenderTargets> color_formats;
- u8 num_color_attachments;
- u8 texceptions;
-
- u8 zeta_format;
- u8 zeta_texception;
-
- std::size_t Hash() const noexcept;
-
- bool operator==(const RenderPassParams& rhs) const noexcept;
-
- bool operator!=(const RenderPassParams& rhs) const noexcept {
- return !operator==(rhs);
- }
-};
-static_assert(std::has_unique_object_representations_v<RenderPassParams>);
-static_assert(std::is_trivially_copyable_v<RenderPassParams>);
-static_assert(std::is_trivially_constructible_v<RenderPassParams>);
-
-} // namespace Vulkan
-
-namespace std {
-
-template <>
-struct hash<Vulkan::RenderPassParams> {
- std::size_t operator()(const Vulkan::RenderPassParams& k) const noexcept {
- return k.Hash();
- }
-};
-
-} // namespace std
-
-namespace Vulkan {
-
-class VKRenderPassCache final {
-public:
- explicit VKRenderPassCache(const VKDevice& device);
- ~VKRenderPassCache();
-
- VkRenderPass GetRenderPass(const RenderPassParams& params);
-
-private:
- vk::RenderPass CreateRenderPass(const RenderPassParams& params) const;
-
- const VKDevice& device;
- std::unordered_map<RenderPassParams, vk::RenderPass> cache;
-};
-
-} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/vk_sampler_cache.cpp b/src/video_core/renderer_vulkan/vk_sampler_cache.cpp
deleted file mode 100644
index b068888f9..000000000
--- a/src/video_core/renderer_vulkan/vk_sampler_cache.cpp
+++ /dev/null
@@ -1,83 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include <unordered_map>
-
-#include "video_core/renderer_vulkan/maxwell_to_vk.h"
-#include "video_core/renderer_vulkan/vk_sampler_cache.h"
-#include "video_core/renderer_vulkan/wrapper.h"
-#include "video_core/textures/texture.h"
-
-using Tegra::Texture::TextureMipmapFilter;
-
-namespace Vulkan {
-
-namespace {
-
-VkBorderColor ConvertBorderColor(std::array<float, 4> color) {
- // TODO(Rodrigo): Manage integer border colors
- if (color == std::array<float, 4>{0, 0, 0, 0}) {
- return VK_BORDER_COLOR_FLOAT_TRANSPARENT_BLACK;
- } else if (color == std::array<float, 4>{0, 0, 0, 1}) {
- return VK_BORDER_COLOR_FLOAT_OPAQUE_BLACK;
- } else if (color == std::array<float, 4>{1, 1, 1, 1}) {
- return VK_BORDER_COLOR_FLOAT_OPAQUE_WHITE;
- }
- if (color[0] + color[1] + color[2] > 1.35f) {
- // If color elements are brighter than roughly 0.5 average, use white border
- return VK_BORDER_COLOR_FLOAT_OPAQUE_WHITE;
- } else if (color[3] > 0.5f) {
- return VK_BORDER_COLOR_FLOAT_OPAQUE_BLACK;
- } else {
- return VK_BORDER_COLOR_FLOAT_TRANSPARENT_BLACK;
- }
-}
-
-} // Anonymous namespace
-
-VKSamplerCache::VKSamplerCache(const VKDevice& device) : device{device} {}
-
-VKSamplerCache::~VKSamplerCache() = default;
-
-vk::Sampler VKSamplerCache::CreateSampler(const Tegra::Texture::TSCEntry& tsc) const {
- const bool arbitrary_borders = device.IsExtCustomBorderColorSupported();
- const std::array color = tsc.GetBorderColor();
-
- VkSamplerCustomBorderColorCreateInfoEXT border{
- .sType = VK_STRUCTURE_TYPE_SAMPLER_CUSTOM_BORDER_COLOR_CREATE_INFO_EXT,
- .pNext = nullptr,
- .customBorderColor = {},
- .format = VK_FORMAT_UNDEFINED,
- };
- std::memcpy(&border.customBorderColor, color.data(), sizeof(color));
-
- return device.GetLogical().CreateSampler({
- .sType = VK_STRUCTURE_TYPE_SAMPLER_CREATE_INFO,
- .pNext = arbitrary_borders ? &border : nullptr,
- .flags = 0,
- .magFilter = MaxwellToVK::Sampler::Filter(tsc.mag_filter),
- .minFilter = MaxwellToVK::Sampler::Filter(tsc.min_filter),
- .mipmapMode = MaxwellToVK::Sampler::MipmapMode(tsc.mipmap_filter),
- .addressModeU = MaxwellToVK::Sampler::WrapMode(device, tsc.wrap_u, tsc.mag_filter),
- .addressModeV = MaxwellToVK::Sampler::WrapMode(device, tsc.wrap_v, tsc.mag_filter),
- .addressModeW = MaxwellToVK::Sampler::WrapMode(device, tsc.wrap_p, tsc.mag_filter),
- .mipLodBias = tsc.GetLodBias(),
- .anisotropyEnable =
- static_cast<VkBool32>(tsc.GetMaxAnisotropy() > 1.0f ? VK_TRUE : VK_FALSE),
- .maxAnisotropy = tsc.GetMaxAnisotropy(),
- .compareEnable = tsc.depth_compare_enabled,
- .compareOp = MaxwellToVK::Sampler::DepthCompareFunction(tsc.depth_compare_func),
- .minLod = tsc.mipmap_filter == TextureMipmapFilter::None ? 0.0f : tsc.GetMinLod(),
- .maxLod = tsc.mipmap_filter == TextureMipmapFilter::None ? 0.25f : tsc.GetMaxLod(),
- .borderColor =
- arbitrary_borders ? VK_BORDER_COLOR_INT_CUSTOM_EXT : ConvertBorderColor(color),
- .unnormalizedCoordinates = VK_FALSE,
- });
-}
-
-VkSampler VKSamplerCache::ToSamplerType(const vk::Sampler& sampler) const {
- return *sampler;
-}
-
-} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/vk_sampler_cache.h b/src/video_core/renderer_vulkan/vk_sampler_cache.h
deleted file mode 100644
index a33d1c0ee..000000000
--- a/src/video_core/renderer_vulkan/vk_sampler_cache.h
+++ /dev/null
@@ -1,29 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include "video_core/renderer_vulkan/wrapper.h"
-#include "video_core/sampler_cache.h"
-#include "video_core/textures/texture.h"
-
-namespace Vulkan {
-
-class VKDevice;
-
-class VKSamplerCache final : public VideoCommon::SamplerCache<VkSampler, vk::Sampler> {
-public:
- explicit VKSamplerCache(const VKDevice& device);
- ~VKSamplerCache();
-
-protected:
- vk::Sampler CreateSampler(const Tegra::Texture::TSCEntry& tsc) const override;
-
- VkSampler ToSamplerType(const vk::Sampler& sampler) const override;
-
-private:
- const VKDevice& device;
-};
-
-} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/vk_scheduler.cpp b/src/video_core/renderer_vulkan/vk_scheduler.cpp
index 1a483dc71..c104c6fe3 100644
--- a/src/video_core/renderer_vulkan/vk_scheduler.cpp
+++ b/src/video_core/renderer_vulkan/vk_scheduler.cpp
@@ -16,6 +16,7 @@
#include "video_core/renderer_vulkan/vk_query_cache.h"
#include "video_core/renderer_vulkan/vk_scheduler.h"
#include "video_core/renderer_vulkan/vk_state_tracker.h"
+#include "video_core/renderer_vulkan/vk_texture_cache.h"
#include "video_core/renderer_vulkan/wrapper.h"
namespace Vulkan {
@@ -96,38 +97,39 @@ void VKScheduler::DispatchWork() {
AcquireNewChunk();
}
-void VKScheduler::RequestRenderpass(VkRenderPass renderpass, VkFramebuffer framebuffer,
- VkExtent2D render_area) {
- if (renderpass == state.renderpass && framebuffer == state.framebuffer &&
+void VKScheduler::RequestRenderpass(const Framebuffer* framebuffer) {
+ const VkRenderPass renderpass = framebuffer->RenderPass();
+ const VkFramebuffer framebuffer_handle = framebuffer->Handle();
+ const VkExtent2D render_area = framebuffer->RenderArea();
+ if (renderpass == state.renderpass && framebuffer_handle == state.framebuffer &&
render_area.width == state.render_area.width &&
render_area.height == state.render_area.height) {
return;
}
- const bool end_renderpass = state.renderpass != nullptr;
+ EndRenderPass();
state.renderpass = renderpass;
- state.framebuffer = framebuffer;
+ state.framebuffer = framebuffer_handle;
state.render_area = render_area;
- const VkRenderPassBeginInfo renderpass_bi{
- .sType = VK_STRUCTURE_TYPE_RENDER_PASS_BEGIN_INFO,
- .pNext = nullptr,
- .renderPass = renderpass,
- .framebuffer = framebuffer,
- .renderArea =
- {
- .offset = {.x = 0, .y = 0},
- .extent = render_area,
- },
- .clearValueCount = 0,
- .pClearValues = nullptr,
- };
-
- Record([renderpass_bi, end_renderpass](vk::CommandBuffer cmdbuf) {
- if (end_renderpass) {
- cmdbuf.EndRenderPass();
- }
+ Record([renderpass, framebuffer_handle, render_area](vk::CommandBuffer cmdbuf) {
+ const VkRenderPassBeginInfo renderpass_bi{
+ .sType = VK_STRUCTURE_TYPE_RENDER_PASS_BEGIN_INFO,
+ .pNext = nullptr,
+ .renderPass = renderpass,
+ .framebuffer = framebuffer_handle,
+ .renderArea =
+ {
+ .offset = {.x = 0, .y = 0},
+ .extent = render_area,
+ },
+ .clearValueCount = 0,
+ .pClearValues = nullptr,
+ };
cmdbuf.BeginRenderPass(renderpass_bi, VK_SUBPASS_CONTENTS_INLINE);
});
+ num_renderpass_images = framebuffer->NumImages();
+ renderpass_images = framebuffer->Images();
+ renderpass_image_ranges = framebuffer->ImageRanges();
}
void VKScheduler::RequestOutsideRenderPassOperationContext() {
@@ -241,8 +243,37 @@ void VKScheduler::EndRenderPass() {
if (!state.renderpass) {
return;
}
+ Record([num_images = num_renderpass_images, images = renderpass_images,
+ ranges = renderpass_image_ranges](vk::CommandBuffer cmdbuf) {
+ std::array<VkImageMemoryBarrier, 9> barriers;
+ for (size_t i = 0; i < num_images; ++i) {
+ barriers[i] = VkImageMemoryBarrier{
+ .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
+ .pNext = nullptr,
+ .srcAccessMask = VK_ACCESS_COLOR_ATTACHMENT_WRITE_BIT |
+ VK_ACCESS_DEPTH_STENCIL_ATTACHMENT_WRITE_BIT,
+ .dstAccessMask = VK_ACCESS_SHADER_READ_BIT | VK_ACCESS_SHADER_WRITE_BIT |
+ VK_ACCESS_COLOR_ATTACHMENT_READ_BIT |
+ VK_ACCESS_COLOR_ATTACHMENT_WRITE_BIT |
+ VK_ACCESS_DEPTH_STENCIL_ATTACHMENT_READ_BIT |
+ VK_ACCESS_DEPTH_STENCIL_ATTACHMENT_WRITE_BIT,
+ .oldLayout = VK_IMAGE_LAYOUT_GENERAL,
+ .newLayout = VK_IMAGE_LAYOUT_GENERAL,
+ .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+ .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+ .image = images[i],
+ .subresourceRange = ranges[i],
+ };
+ }
+ cmdbuf.EndRenderPass();
+ cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_EARLY_FRAGMENT_TESTS_BIT |
+ VK_PIPELINE_STAGE_LATE_FRAGMENT_TESTS_BIT |
+ VK_PIPELINE_STAGE_COLOR_ATTACHMENT_OUTPUT_BIT,
+ VK_PIPELINE_STAGE_ALL_GRAPHICS_BIT, 0, nullptr, nullptr,
+ vk::Span(barriers.data(), num_images));
+ });
state.renderpass = nullptr;
- Record([](vk::CommandBuffer cmdbuf) { cmdbuf.EndRenderPass(); });
+ num_renderpass_images = 0;
}
void VKScheduler::AcquireNewChunk() {
diff --git a/src/video_core/renderer_vulkan/vk_scheduler.h b/src/video_core/renderer_vulkan/vk_scheduler.h
index 7be8a19f0..0a36c8fad 100644
--- a/src/video_core/renderer_vulkan/vk_scheduler.h
+++ b/src/video_core/renderer_vulkan/vk_scheduler.h
@@ -17,6 +17,7 @@
namespace Vulkan {
class CommandPool;
+class Framebuffer;
class MasterSemaphore;
class StateTracker;
class VKDevice;
@@ -52,8 +53,7 @@ public:
void DispatchWork();
/// Requests to begin a renderpass.
- void RequestRenderpass(VkRenderPass renderpass, VkFramebuffer framebuffer,
- VkExtent2D render_area);
+ void RequestRenderpass(const Framebuffer* framebuffer);
/// Requests the current executino context to be able to execute operations only allowed outside
/// of a renderpass.
@@ -62,6 +62,9 @@ public:
/// Binds a pipeline to the current execution context.
void BindGraphicsPipeline(VkPipeline pipeline);
+ /// Invalidates current command buffer state except for render passes
+ void InvalidateState();
+
/// Assigns the query cache.
void SetQueryCache(VKQueryCache& query_cache_) {
query_cache = &query_cache_;
@@ -104,7 +107,7 @@ private:
template <typename T>
class TypedCommand final : public Command {
public:
- explicit TypedCommand(T&& command) : command{std::move(command)} {}
+ explicit TypedCommand(T&& command_) : command{std::move(command_)} {}
~TypedCommand() override = default;
TypedCommand(TypedCommand&&) = delete;
@@ -170,8 +173,6 @@ private:
void AllocateNewContext();
- void InvalidateState();
-
void EndPendingOperations();
void EndRenderPass();
@@ -192,6 +193,11 @@ private:
std::thread worker_thread;
State state;
+
+ u32 num_renderpass_images = 0;
+ std::array<VkImage, 9> renderpass_images{};
+ std::array<VkImageSubresourceRange, 9> renderpass_image_ranges{};
+
Common::SPSCQueue<std::unique_ptr<CommandChunk>> chunk_queue;
Common::SPSCQueue<std::unique_ptr<CommandChunk>> chunk_reserve;
std::mutex mutex;
diff --git a/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp b/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp
index fed9ebecd..09d6f9f35 100644
--- a/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp
+++ b/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp
@@ -55,8 +55,8 @@ enum class Type { Void, Bool, Bool2, Float, Int, Uint, HalfFloat };
class Expression final {
public:
- Expression(Id id, Type type) : id{id}, type{type} {
- ASSERT(type != Type::Void);
+ Expression(Id id_, Type type_) : id{id_}, type{type_} {
+ ASSERT(type_ != Type::Void);
}
Expression() : type{Type::Void} {}
@@ -102,7 +102,7 @@ struct GenericVaryingDescription {
bool is_scalar = false;
};
-spv::Dim GetSamplerDim(const Sampler& sampler) {
+spv::Dim GetSamplerDim(const SamplerEntry& sampler) {
ASSERT(!sampler.is_buffer);
switch (sampler.type) {
case Tegra::Shader::TextureType::Texture1D:
@@ -114,12 +114,12 @@ spv::Dim GetSamplerDim(const Sampler& sampler) {
case Tegra::Shader::TextureType::TextureCube:
return spv::Dim::Cube;
default:
- UNIMPLEMENTED_MSG("Unimplemented sampler type={}", static_cast<int>(sampler.type));
+ UNIMPLEMENTED_MSG("Unimplemented sampler type={}", sampler.type);
return spv::Dim::Dim2D;
}
}
-std::pair<spv::Dim, bool> GetImageDim(const Image& image) {
+std::pair<spv::Dim, bool> GetImageDim(const ImageEntry& image) {
switch (image.type) {
case Tegra::Shader::ImageType::Texture1D:
return {spv::Dim::Dim1D, false};
@@ -134,7 +134,7 @@ std::pair<spv::Dim, bool> GetImageDim(const Image& image) {
case Tegra::Shader::ImageType::Texture3D:
return {spv::Dim::Dim3D, false};
default:
- UNIMPLEMENTED_MSG("Unimplemented image type={}", static_cast<int>(image.type));
+ UNIMPLEMENTED_MSG("Unimplemented image type={}", image.type);
return {spv::Dim::Dim2D, false};
}
}
@@ -281,12 +281,12 @@ u32 ShaderVersion(const VKDevice& device) {
class SPIRVDecompiler final : public Sirit::Module {
public:
- explicit SPIRVDecompiler(const VKDevice& device, const ShaderIR& ir, ShaderType stage,
- const Registry& registry, const Specialization& specialization)
- : Module(ShaderVersion(device)), device{device}, ir{ir}, stage{stage},
- header{ir.GetHeader()}, registry{registry}, specialization{specialization} {
- if (stage != ShaderType::Compute) {
- transform_feedback = BuildTransformFeedback(registry.GetGraphicsInfo());
+ explicit SPIRVDecompiler(const VKDevice& device_, const ShaderIR& ir_, ShaderType stage_,
+ const Registry& registry_, const Specialization& specialization_)
+ : Module(ShaderVersion(device_)), device{device_}, ir{ir_}, stage{stage_},
+ header{ir_.GetHeader()}, registry{registry_}, specialization{specialization_} {
+ if (stage_ != ShaderType::Compute) {
+ transform_feedback = BuildTransformFeedback(registry_.GetGraphicsInfo());
}
AddCapability(spv::Capability::Shader);
@@ -330,7 +330,7 @@ public:
if (device.IsFloat16Supported()) {
AddCapability(spv::Capability::Float16);
}
- t_scalar_half = Name(TypeFloat(device.IsFloat16Supported() ? 16 : 32), "scalar_half");
+ t_scalar_half = Name(TypeFloat(device_.IsFloat16Supported() ? 16 : 32), "scalar_half");
t_half = Name(TypeVector(t_scalar_half, 2), "half");
const Id main = Decompile();
@@ -980,7 +980,7 @@ private:
return binding;
}
- void DeclareImage(const Image& image, u32& binding) {
+ void DeclareImage(const ImageEntry& image, u32& binding) {
const auto [dim, arrayed] = GetImageDim(image);
constexpr int depth = 0;
constexpr bool ms = false;
@@ -1088,9 +1088,9 @@ private:
indices.point_size = AddBuiltIn(t_float, spv::BuiltIn::PointSize, "point_size");
}
- const auto& output_attributes = ir.GetOutputAttributes();
- const bool declare_clip_distances =
- std::any_of(output_attributes.begin(), output_attributes.end(), [](const auto& index) {
+ const auto& ir_output_attributes = ir.GetOutputAttributes();
+ const bool declare_clip_distances = std::any_of(
+ ir_output_attributes.begin(), ir_output_attributes.end(), [](const auto& index) {
return index == Attribute::Index::ClipDistances0123 ||
index == Attribute::Index::ClipDistances4567;
});
@@ -1254,7 +1254,7 @@ private:
const Id pointer = ArrayPass(type_descriptor.scalar, attribute_id, elements);
return {OpLoad(GetTypeDefinition(type), pointer), type};
}
- UNIMPLEMENTED_MSG("Unhandled input attribute: {}", static_cast<u32>(attribute));
+ UNIMPLEMENTED_MSG("Unhandled input attribute: {}", attribute);
return {v_float_zero, Type::Float};
}
@@ -1890,7 +1890,7 @@ private:
case Tegra::Shader::TextureType::Texture3D:
return 3;
default:
- UNREACHABLE_MSG("Invalid texture type={}", static_cast<int>(type));
+ UNREACHABLE_MSG("Invalid texture type={}", type);
return 2;
}
}();
@@ -2094,6 +2094,7 @@ private:
return OpFOrdGreaterThanEqual(t_bool, operand_1, operand_2);
default:
UNREACHABLE();
+ return v_true;
}
}
@@ -2125,8 +2126,7 @@ private:
OpStore(z_pointer, depth);
}
if (stage == ShaderType::Fragment) {
- const auto SafeGetRegister = [&](u32 reg) {
- // TODO(Rodrigo): Replace with contains once C++20 releases
+ const auto SafeGetRegister = [this](u32 reg) {
if (const auto it = registers.find(reg); it != registers.end()) {
return OpLoad(t_float, it->second);
}
@@ -2891,7 +2891,7 @@ private:
class ExprDecompiler {
public:
- explicit ExprDecompiler(SPIRVDecompiler& decomp) : decomp{decomp} {}
+ explicit ExprDecompiler(SPIRVDecompiler& decomp_) : decomp{decomp_} {}
Id operator()(const ExprAnd& expr) {
const Id type_def = decomp.GetTypeDefinition(Type::Bool);
@@ -2947,7 +2947,7 @@ private:
class ASTDecompiler {
public:
- explicit ASTDecompiler(SPIRVDecompiler& decomp) : decomp{decomp} {}
+ explicit ASTDecompiler(SPIRVDecompiler& decomp_) : decomp{decomp_} {}
void operator()(const ASTProgram& ast) {
ASTNode current = ast.nodes.GetFirst();
diff --git a/src/video_core/renderer_vulkan/vk_shader_decompiler.h b/src/video_core/renderer_vulkan/vk_shader_decompiler.h
index 110848922..ad91ad5de 100644
--- a/src/video_core/renderer_vulkan/vk_shader_decompiler.h
+++ b/src/video_core/renderer_vulkan/vk_shader_decompiler.h
@@ -21,17 +21,17 @@ class VKDevice;
namespace Vulkan {
using Maxwell = Tegra::Engines::Maxwell3D::Regs;
-using UniformTexelEntry = VideoCommon::Shader::Sampler;
-using SamplerEntry = VideoCommon::Shader::Sampler;
-using StorageTexelEntry = VideoCommon::Shader::Image;
-using ImageEntry = VideoCommon::Shader::Image;
+using UniformTexelEntry = VideoCommon::Shader::SamplerEntry;
+using SamplerEntry = VideoCommon::Shader::SamplerEntry;
+using StorageTexelEntry = VideoCommon::Shader::ImageEntry;
+using ImageEntry = VideoCommon::Shader::ImageEntry;
constexpr u32 DESCRIPTOR_SET = 0;
class ConstBufferEntry : public VideoCommon::Shader::ConstBuffer {
public:
- explicit constexpr ConstBufferEntry(const VideoCommon::Shader::ConstBuffer& entry, u32 index)
- : VideoCommon::Shader::ConstBuffer{entry}, index{index} {}
+ explicit constexpr ConstBufferEntry(const ConstBuffer& entry_, u32 index_)
+ : ConstBuffer{entry_}, index{index_} {}
constexpr u32 GetIndex() const {
return index;
@@ -43,8 +43,8 @@ private:
class GlobalBufferEntry {
public:
- constexpr explicit GlobalBufferEntry(u32 cbuf_index, u32 cbuf_offset, bool is_written)
- : cbuf_index{cbuf_index}, cbuf_offset{cbuf_offset}, is_written{is_written} {}
+ constexpr explicit GlobalBufferEntry(u32 cbuf_index_, u32 cbuf_offset_, bool is_written_)
+ : cbuf_index{cbuf_index_}, cbuf_offset{cbuf_offset_}, is_written{is_written_} {}
constexpr u32 GetCbufIndex() const {
return cbuf_index;
diff --git a/src/video_core/renderer_vulkan/vk_shader_util.cpp b/src/video_core/renderer_vulkan/vk_shader_util.cpp
index c1a218d76..38a0be7f2 100644
--- a/src/video_core/renderer_vulkan/vk_shader_util.cpp
+++ b/src/video_core/renderer_vulkan/vk_shader_util.cpp
@@ -13,18 +13,13 @@
namespace Vulkan {
-vk::ShaderModule BuildShader(const VKDevice& device, std::size_t code_size, const u8* code_data) {
- // Avoid undefined behavior by copying to a staging allocation
- ASSERT(code_size % sizeof(u32) == 0);
- const auto data = std::make_unique<u32[]>(code_size / sizeof(u32));
- std::memcpy(data.get(), code_data, code_size);
-
+vk::ShaderModule BuildShader(const VKDevice& device, std::span<const u32> code) {
return device.GetLogical().CreateShaderModule({
.sType = VK_STRUCTURE_TYPE_SHADER_MODULE_CREATE_INFO,
.pNext = nullptr,
.flags = 0,
- .codeSize = code_size,
- .pCode = data.get(),
+ .codeSize = static_cast<u32>(code.size_bytes()),
+ .pCode = code.data(),
});
}
diff --git a/src/video_core/renderer_vulkan/vk_shader_util.h b/src/video_core/renderer_vulkan/vk_shader_util.h
index d1d3f3cae..dce34a140 100644
--- a/src/video_core/renderer_vulkan/vk_shader_util.h
+++ b/src/video_core/renderer_vulkan/vk_shader_util.h
@@ -4,6 +4,8 @@
#pragma once
+#include <span>
+
#include "common/common_types.h"
#include "video_core/renderer_vulkan/wrapper.h"
@@ -11,6 +13,6 @@ namespace Vulkan {
class VKDevice;
-vk::ShaderModule BuildShader(const VKDevice& device, std::size_t code_size, const u8* code_data);
+vk::ShaderModule BuildShader(const VKDevice& device, std::span<const u32> code);
} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/vk_state_tracker.cpp b/src/video_core/renderer_vulkan/vk_state_tracker.cpp
index 5d2c4a796..1779a2e30 100644
--- a/src/video_core/renderer_vulkan/vk_state_tracker.cpp
+++ b/src/video_core/renderer_vulkan/vk_state_tracker.cpp
@@ -3,6 +3,7 @@
// Refer to the license.txt file included.
#include <algorithm>
+#include <array>
#include <cstddef>
#include <iterator>
@@ -14,7 +15,7 @@
#include "video_core/renderer_vulkan/vk_state_tracker.h"
#define OFF(field_name) MAXWELL3D_REG_INDEX(field_name)
-#define NUM(field_name) (sizeof(Maxwell3D::Regs::field_name) / sizeof(u32))
+#define NUM(field_name) (sizeof(Maxwell3D::Regs::field_name) / (sizeof(u32)))
namespace Vulkan {
@@ -29,21 +30,15 @@ using Table = Maxwell3D::DirtyState::Table;
using Flags = Maxwell3D::DirtyState::Flags;
Flags MakeInvalidationFlags() {
+ static constexpr std::array INVALIDATION_FLAGS{
+ Viewports, Scissors, DepthBias, BlendConstants, DepthBounds,
+ StencilProperties, CullMode, DepthBoundsEnable, DepthTestEnable, DepthWriteEnable,
+ DepthCompareOp, FrontFace, StencilOp, StencilTestEnable,
+ };
Flags flags{};
- flags[Viewports] = true;
- flags[Scissors] = true;
- flags[DepthBias] = true;
- flags[BlendConstants] = true;
- flags[DepthBounds] = true;
- flags[StencilProperties] = true;
- flags[CullMode] = true;
- flags[DepthBoundsEnable] = true;
- flags[DepthTestEnable] = true;
- flags[DepthWriteEnable] = true;
- flags[DepthCompareOp] = true;
- flags[FrontFace] = true;
- flags[StencilOp] = true;
- flags[StencilTestEnable] = true;
+ for (const int flag : INVALIDATION_FLAGS) {
+ flags[flag] = true;
+ }
return flags;
}
diff --git a/src/video_core/renderer_vulkan/vk_state_tracker.h b/src/video_core/renderer_vulkan/vk_state_tracker.h
index 1de789e57..c335d2bdf 100644
--- a/src/video_core/renderer_vulkan/vk_state_tracker.h
+++ b/src/video_core/renderer_vulkan/vk_state_tracker.h
@@ -52,6 +52,14 @@ public:
current_topology = INVALID_TOPOLOGY;
}
+ void InvalidateViewports() {
+ flags[Dirty::Viewports] = true;
+ }
+
+ void InvalidateScissors() {
+ flags[Dirty::Scissors] = true;
+ }
+
bool TouchViewports() {
return Exchange(Dirty::Viewports, false);
}
diff --git a/src/video_core/renderer_vulkan/vk_stream_buffer.cpp b/src/video_core/renderer_vulkan/vk_stream_buffer.cpp
index 1b59612b9..419cb154d 100644
--- a/src/video_core/renderer_vulkan/vk_stream_buffer.cpp
+++ b/src/video_core/renderer_vulkan/vk_stream_buffer.cpp
@@ -19,6 +19,10 @@ namespace Vulkan {
namespace {
+constexpr VkBufferUsageFlags BUFFER_USAGE =
+ VK_BUFFER_USAGE_VERTEX_BUFFER_BIT | VK_BUFFER_USAGE_INDEX_BUFFER_BIT |
+ VK_BUFFER_USAGE_UNIFORM_BUFFER_BIT | VK_BUFFER_USAGE_STORAGE_BUFFER_BIT;
+
constexpr u64 WATCHES_INITIAL_RESERVE = 0x4000;
constexpr u64 WATCHES_RESERVE_CHUNK = 0x1000;
@@ -56,17 +60,16 @@ u32 GetMemoryType(const VkPhysicalDeviceMemoryProperties& properties,
} // Anonymous namespace
-VKStreamBuffer::VKStreamBuffer(const VKDevice& device_, VKScheduler& scheduler_,
- VkBufferUsageFlags usage)
+VKStreamBuffer::VKStreamBuffer(const VKDevice& device_, VKScheduler& scheduler_)
: device{device_}, scheduler{scheduler_} {
- CreateBuffers(usage);
+ CreateBuffers();
ReserveWatches(current_watches, WATCHES_INITIAL_RESERVE);
ReserveWatches(previous_watches, WATCHES_INITIAL_RESERVE);
}
VKStreamBuffer::~VKStreamBuffer() = default;
-std::tuple<u8*, u64, bool> VKStreamBuffer::Map(u64 size, u64 alignment) {
+std::pair<u8*, u64> VKStreamBuffer::Map(u64 size, u64 alignment) {
ASSERT(size <= stream_buffer_size);
mapped_size = size;
@@ -76,7 +79,6 @@ std::tuple<u8*, u64, bool> VKStreamBuffer::Map(u64 size, u64 alignment) {
WaitPendingOperations(offset);
- bool invalidated = false;
if (offset + size > stream_buffer_size) {
// The buffer would overflow, save the amount of used watches and reset the state.
invalidation_mark = current_watch_cursor;
@@ -90,11 +92,9 @@ std::tuple<u8*, u64, bool> VKStreamBuffer::Map(u64 size, u64 alignment) {
// Ensure that we don't wait for uncommitted fences.
scheduler.Flush();
-
- invalidated = true;
}
- return {memory.Map(offset, size), offset, invalidated};
+ return std::make_pair(memory.Map(offset, size), offset);
}
void VKStreamBuffer::Unmap(u64 size) {
@@ -113,7 +113,7 @@ void VKStreamBuffer::Unmap(u64 size) {
watch.tick = scheduler.CurrentTick();
}
-void VKStreamBuffer::CreateBuffers(VkBufferUsageFlags usage) {
+void VKStreamBuffer::CreateBuffers() {
const auto memory_properties = device.GetPhysical().GetMemoryProperties();
const u32 preferred_type = GetMemoryType(memory_properties);
const u32 preferred_heap = memory_properties.memoryTypes[preferred_type].heapIndex;
@@ -127,7 +127,7 @@ void VKStreamBuffer::CreateBuffers(VkBufferUsageFlags usage) {
.pNext = nullptr,
.flags = 0,
.size = std::min(PREFERRED_STREAM_BUFFER_SIZE, allocable_size),
- .usage = usage,
+ .usage = BUFFER_USAGE,
.sharingMode = VK_SHARING_MODE_EXCLUSIVE,
.queueFamilyIndexCount = 0,
.pQueueFamilyIndices = nullptr,
diff --git a/src/video_core/renderer_vulkan/vk_stream_buffer.h b/src/video_core/renderer_vulkan/vk_stream_buffer.h
index 5e15ad78f..1428f77bf 100644
--- a/src/video_core/renderer_vulkan/vk_stream_buffer.h
+++ b/src/video_core/renderer_vulkan/vk_stream_buffer.h
@@ -5,7 +5,7 @@
#pragma once
#include <optional>
-#include <tuple>
+#include <utility>
#include <vector>
#include "common/common_types.h"
@@ -19,17 +19,15 @@ class VKScheduler;
class VKStreamBuffer final {
public:
- explicit VKStreamBuffer(const VKDevice& device, VKScheduler& scheduler,
- VkBufferUsageFlags usage);
+ explicit VKStreamBuffer(const VKDevice& device, VKScheduler& scheduler);
~VKStreamBuffer();
/**
* Reserves a region of memory from the stream buffer.
* @param size Size to reserve.
- * @returns A tuple in the following order: Raw memory pointer (with offset added), buffer
- * offset and a boolean that's true when buffer has been invalidated.
+ * @returns A pair of a raw memory pointer (with offset added), and the buffer offset
*/
- std::tuple<u8*, u64, bool> Map(u64 size, u64 alignment);
+ std::pair<u8*, u64> Map(u64 size, u64 alignment);
/// Ensures that "size" bytes of memory are available to the GPU, potentially recording a copy.
void Unmap(u64 size);
@@ -49,7 +47,7 @@ private:
};
/// Creates Vulkan buffer handles committing the required the required memory.
- void CreateBuffers(VkBufferUsageFlags usage);
+ void CreateBuffers();
/// Increases the amount of watches available.
void ReserveWatches(std::vector<Watch>& watches, std::size_t grow_size);
diff --git a/src/video_core/renderer_vulkan/vk_texture_cache.cpp b/src/video_core/renderer_vulkan/vk_texture_cache.cpp
index f2c8f2ae1..261808391 100644
--- a/src/video_core/renderer_vulkan/vk_texture_cache.cpp
+++ b/src/video_core/renderer_vulkan/vk_texture_cache.cpp
@@ -4,613 +4,1103 @@
#include <algorithm>
#include <array>
-#include <cstddef>
-#include <cstring>
-#include <memory>
-#include <variant>
+#include <span>
#include <vector>
-#include "common/assert.h"
-#include "common/common_types.h"
-#include "core/core.h"
-#include "video_core/engines/maxwell_3d.h"
-#include "video_core/morton.h"
+#include "video_core/engines/fermi_2d.h"
+#include "video_core/renderer_vulkan/blit_image.h"
#include "video_core/renderer_vulkan/maxwell_to_vk.h"
#include "video_core/renderer_vulkan/vk_device.h"
-#include "video_core/renderer_vulkan/vk_memory_manager.h"
-#include "video_core/renderer_vulkan/vk_rasterizer.h"
#include "video_core/renderer_vulkan/vk_scheduler.h"
#include "video_core/renderer_vulkan/vk_staging_buffer_pool.h"
#include "video_core/renderer_vulkan/vk_texture_cache.h"
#include "video_core/renderer_vulkan/wrapper.h"
-#include "video_core/surface.h"
namespace Vulkan {
-using VideoCore::MortonSwizzle;
-using VideoCore::MortonSwizzleMode;
-
+using Tegra::Engines::Fermi2D;
using Tegra::Texture::SwizzleSource;
-using VideoCore::Surface::PixelFormat;
-using VideoCore::Surface::SurfaceTarget;
+using Tegra::Texture::TextureMipmapFilter;
+using VideoCommon::BufferImageCopy;
+using VideoCommon::ImageInfo;
+using VideoCommon::ImageType;
+using VideoCommon::SubresourceRange;
+using VideoCore::Surface::IsPixelFormatASTC;
namespace {
-VkImageType SurfaceTargetToImage(SurfaceTarget target) {
- switch (target) {
- case SurfaceTarget::Texture1D:
- case SurfaceTarget::Texture1DArray:
+constexpr std::array ATTACHMENT_REFERENCES{
+ VkAttachmentReference{0, VK_IMAGE_LAYOUT_GENERAL},
+ VkAttachmentReference{1, VK_IMAGE_LAYOUT_GENERAL},
+ VkAttachmentReference{2, VK_IMAGE_LAYOUT_GENERAL},
+ VkAttachmentReference{3, VK_IMAGE_LAYOUT_GENERAL},
+ VkAttachmentReference{4, VK_IMAGE_LAYOUT_GENERAL},
+ VkAttachmentReference{5, VK_IMAGE_LAYOUT_GENERAL},
+ VkAttachmentReference{6, VK_IMAGE_LAYOUT_GENERAL},
+ VkAttachmentReference{7, VK_IMAGE_LAYOUT_GENERAL},
+ VkAttachmentReference{8, VK_IMAGE_LAYOUT_GENERAL},
+};
+
+constexpr VkBorderColor ConvertBorderColor(const std::array<float, 4>& color) {
+ if (color == std::array<float, 4>{0, 0, 0, 0}) {
+ return VK_BORDER_COLOR_FLOAT_TRANSPARENT_BLACK;
+ } else if (color == std::array<float, 4>{0, 0, 0, 1}) {
+ return VK_BORDER_COLOR_FLOAT_OPAQUE_BLACK;
+ } else if (color == std::array<float, 4>{1, 1, 1, 1}) {
+ return VK_BORDER_COLOR_FLOAT_OPAQUE_WHITE;
+ }
+ if (color[0] + color[1] + color[2] > 1.35f) {
+ // If color elements are brighter than roughly 0.5 average, use white border
+ return VK_BORDER_COLOR_FLOAT_OPAQUE_WHITE;
+ } else if (color[3] > 0.5f) {
+ return VK_BORDER_COLOR_FLOAT_OPAQUE_BLACK;
+ } else {
+ return VK_BORDER_COLOR_FLOAT_TRANSPARENT_BLACK;
+ }
+}
+
+[[nodiscard]] VkImageType ConvertImageType(const ImageType type) {
+ switch (type) {
+ case ImageType::e1D:
return VK_IMAGE_TYPE_1D;
- case SurfaceTarget::Texture2D:
- case SurfaceTarget::Texture2DArray:
- case SurfaceTarget::TextureCubemap:
- case SurfaceTarget::TextureCubeArray:
+ case ImageType::e2D:
+ case ImageType::Linear:
return VK_IMAGE_TYPE_2D;
- case SurfaceTarget::Texture3D:
+ case ImageType::e3D:
return VK_IMAGE_TYPE_3D;
- case SurfaceTarget::TextureBuffer:
- UNREACHABLE();
- return {};
+ case ImageType::Buffer:
+ break;
}
- UNREACHABLE_MSG("Unknown texture target={}", static_cast<u32>(target));
+ UNREACHABLE_MSG("Invalid image type={}", type);
return {};
}
-VkImageAspectFlags PixelFormatToImageAspect(PixelFormat pixel_format) {
- if (pixel_format < PixelFormat::MaxColorFormat) {
- return VK_IMAGE_ASPECT_COLOR_BIT;
- } else if (pixel_format < PixelFormat::MaxDepthFormat) {
- return VK_IMAGE_ASPECT_DEPTH_BIT;
- } else if (pixel_format < PixelFormat::MaxDepthStencilFormat) {
- return VK_IMAGE_ASPECT_DEPTH_BIT | VK_IMAGE_ASPECT_STENCIL_BIT;
- } else {
- UNREACHABLE_MSG("Invalid pixel format={}", static_cast<int>(pixel_format));
- return VK_IMAGE_ASPECT_COLOR_BIT;
+[[nodiscard]] VkSampleCountFlagBits ConvertSampleCount(u32 num_samples) {
+ switch (num_samples) {
+ case 1:
+ return VK_SAMPLE_COUNT_1_BIT;
+ case 2:
+ return VK_SAMPLE_COUNT_2_BIT;
+ case 4:
+ return VK_SAMPLE_COUNT_4_BIT;
+ case 8:
+ return VK_SAMPLE_COUNT_8_BIT;
+ case 16:
+ return VK_SAMPLE_COUNT_16_BIT;
+ default:
+ UNREACHABLE_MSG("Invalid number of samples={}", num_samples);
+ return VK_SAMPLE_COUNT_1_BIT;
}
}
-VkImageViewType GetImageViewType(SurfaceTarget target) {
- switch (target) {
- case SurfaceTarget::Texture1D:
- return VK_IMAGE_VIEW_TYPE_1D;
- case SurfaceTarget::Texture2D:
- return VK_IMAGE_VIEW_TYPE_2D;
- case SurfaceTarget::Texture3D:
- return VK_IMAGE_VIEW_TYPE_3D;
- case SurfaceTarget::Texture1DArray:
- return VK_IMAGE_VIEW_TYPE_1D_ARRAY;
- case SurfaceTarget::Texture2DArray:
- return VK_IMAGE_VIEW_TYPE_2D_ARRAY;
- case SurfaceTarget::TextureCubemap:
- return VK_IMAGE_VIEW_TYPE_CUBE;
- case SurfaceTarget::TextureCubeArray:
- return VK_IMAGE_VIEW_TYPE_CUBE_ARRAY;
- case SurfaceTarget::TextureBuffer:
- break;
+[[nodiscard]] VkImageCreateInfo MakeImageCreateInfo(const VKDevice& device, const ImageInfo& info) {
+ const auto format_info = MaxwellToVK::SurfaceFormat(device, FormatType::Optimal, info.format);
+ VkImageCreateFlags flags = VK_IMAGE_CREATE_MUTABLE_FORMAT_BIT;
+ if (info.type == ImageType::e2D && info.resources.layers >= 6 &&
+ info.size.width == info.size.height) {
+ flags |= VK_IMAGE_CREATE_CUBE_COMPATIBLE_BIT;
}
- UNREACHABLE();
- return {};
-}
-
-vk::Buffer CreateBuffer(const VKDevice& device, const SurfaceParams& params,
- std::size_t host_memory_size) {
- // TODO(Rodrigo): Move texture buffer creation to the buffer cache
- return device.GetLogical().CreateBuffer({
- .sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO,
+ if (info.type == ImageType::e3D) {
+ flags |= VK_IMAGE_CREATE_2D_ARRAY_COMPATIBLE_BIT;
+ }
+ VkImageUsageFlags usage = VK_IMAGE_USAGE_TRANSFER_SRC_BIT | VK_IMAGE_USAGE_TRANSFER_DST_BIT |
+ VK_IMAGE_USAGE_SAMPLED_BIT;
+ if (format_info.attachable) {
+ switch (VideoCore::Surface::GetFormatType(info.format)) {
+ case VideoCore::Surface::SurfaceType::ColorTexture:
+ usage |= VK_IMAGE_USAGE_COLOR_ATTACHMENT_BIT;
+ break;
+ case VideoCore::Surface::SurfaceType::Depth:
+ case VideoCore::Surface::SurfaceType::DepthStencil:
+ usage |= VK_IMAGE_USAGE_DEPTH_STENCIL_ATTACHMENT_BIT;
+ break;
+ default:
+ UNREACHABLE_MSG("Invalid surface type");
+ }
+ }
+ if (format_info.storage) {
+ usage |= VK_IMAGE_USAGE_STORAGE_BIT;
+ }
+ const auto [samples_x, samples_y] = VideoCommon::SamplesLog2(info.num_samples);
+ return VkImageCreateInfo{
+ .sType = VK_STRUCTURE_TYPE_IMAGE_CREATE_INFO,
.pNext = nullptr,
- .flags = 0,
- .size = static_cast<VkDeviceSize>(host_memory_size),
- .usage = VK_BUFFER_USAGE_UNIFORM_TEXEL_BUFFER_BIT |
- VK_BUFFER_USAGE_STORAGE_TEXEL_BUFFER_BIT | VK_BUFFER_USAGE_TRANSFER_SRC_BIT |
- VK_BUFFER_USAGE_TRANSFER_DST_BIT,
+ .flags = flags,
+ .imageType = ConvertImageType(info.type),
+ .format = format_info.format,
+ .extent =
+ {
+ .width = info.size.width >> samples_x,
+ .height = info.size.height >> samples_y,
+ .depth = info.size.depth,
+ },
+ .mipLevels = static_cast<u32>(info.resources.levels),
+ .arrayLayers = static_cast<u32>(info.resources.layers),
+ .samples = ConvertSampleCount(info.num_samples),
+ .tiling = VK_IMAGE_TILING_OPTIMAL,
+ .usage = usage,
.sharingMode = VK_SHARING_MODE_EXCLUSIVE,
.queueFamilyIndexCount = 0,
.pQueueFamilyIndices = nullptr,
- });
-}
-
-VkBufferViewCreateInfo GenerateBufferViewCreateInfo(const VKDevice& device,
- const SurfaceParams& params, VkBuffer buffer,
- std::size_t host_memory_size) {
- ASSERT(params.IsBuffer());
-
- return {
- .sType = VK_STRUCTURE_TYPE_BUFFER_VIEW_CREATE_INFO,
- .pNext = nullptr,
- .flags = 0,
- .buffer = buffer,
- .format =
- MaxwellToVK::SurfaceFormat(device, FormatType::Buffer, params.pixel_format).format,
- .offset = 0,
- .range = static_cast<VkDeviceSize>(host_memory_size),
+ .initialLayout = VK_IMAGE_LAYOUT_UNDEFINED,
};
}
-VkImageCreateInfo GenerateImageCreateInfo(const VKDevice& device, const SurfaceParams& params) {
- ASSERT(!params.IsBuffer());
-
- const auto [format, attachable, storage] =
- MaxwellToVK::SurfaceFormat(device, FormatType::Optimal, params.pixel_format);
+[[nodiscard]] vk::Image MakeImage(const VKDevice& device, const ImageInfo& info) {
+ if (info.type == ImageType::Buffer) {
+ return vk::Image{};
+ }
+ return device.GetLogical().CreateImage(MakeImageCreateInfo(device, info));
+}
- VkImageCreateInfo ci{
- .sType = VK_STRUCTURE_TYPE_IMAGE_CREATE_INFO,
+[[nodiscard]] vk::Buffer MakeBuffer(const VKDevice& device, const ImageInfo& info) {
+ if (info.type != ImageType::Buffer) {
+ return vk::Buffer{};
+ }
+ const size_t bytes_per_block = VideoCore::Surface::BytesPerBlock(info.format);
+ return device.GetLogical().CreateBuffer(VkBufferCreateInfo{
+ .sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO,
.pNext = nullptr,
.flags = 0,
- .imageType = SurfaceTargetToImage(params.target),
- .format = format,
- .extent = {},
- .mipLevels = params.num_levels,
- .arrayLayers = static_cast<u32>(params.GetNumLayers()),
- .samples = VK_SAMPLE_COUNT_1_BIT,
- .tiling = VK_IMAGE_TILING_OPTIMAL,
- .usage = VK_IMAGE_USAGE_SAMPLED_BIT | VK_IMAGE_USAGE_TRANSFER_DST_BIT |
- VK_IMAGE_USAGE_TRANSFER_SRC_BIT,
+ .size = info.size.width * bytes_per_block,
+ .usage = VK_BUFFER_USAGE_TRANSFER_SRC_BIT | VK_BUFFER_USAGE_TRANSFER_DST_BIT |
+ VK_BUFFER_USAGE_UNIFORM_TEXEL_BUFFER_BIT |
+ VK_BUFFER_USAGE_STORAGE_TEXEL_BUFFER_BIT,
.sharingMode = VK_SHARING_MODE_EXCLUSIVE,
.queueFamilyIndexCount = 0,
.pQueueFamilyIndices = nullptr,
- .initialLayout = VK_IMAGE_LAYOUT_UNDEFINED,
- };
- if (attachable) {
- ci.usage |= params.IsPixelFormatZeta() ? VK_IMAGE_USAGE_DEPTH_STENCIL_ATTACHMENT_BIT
- : VK_IMAGE_USAGE_COLOR_ATTACHMENT_BIT;
- }
- if (storage) {
- ci.usage |= VK_IMAGE_USAGE_STORAGE_BIT;
- }
-
- switch (params.target) {
- case SurfaceTarget::TextureCubemap:
- case SurfaceTarget::TextureCubeArray:
- ci.flags |= VK_IMAGE_CREATE_CUBE_COMPATIBLE_BIT;
- [[fallthrough]];
- case SurfaceTarget::Texture1D:
- case SurfaceTarget::Texture1DArray:
- case SurfaceTarget::Texture2D:
- case SurfaceTarget::Texture2DArray:
- ci.extent = {params.width, params.height, 1};
- break;
- case SurfaceTarget::Texture3D:
- ci.flags |= VK_IMAGE_CREATE_2D_ARRAY_COMPATIBLE_BIT;
- ci.extent = {params.width, params.height, params.depth};
- break;
- case SurfaceTarget::TextureBuffer:
- UNREACHABLE();
- }
-
- return ci;
+ });
}
-u32 EncodeSwizzle(Tegra::Texture::SwizzleSource x_source, Tegra::Texture::SwizzleSource y_source,
- Tegra::Texture::SwizzleSource z_source, Tegra::Texture::SwizzleSource w_source) {
- return (static_cast<u32>(x_source) << 24) | (static_cast<u32>(y_source) << 16) |
- (static_cast<u32>(z_source) << 8) | static_cast<u32>(w_source);
+[[nodiscard]] VkImageAspectFlags ImageAspectMask(PixelFormat format) {
+ switch (VideoCore::Surface::GetFormatType(format)) {
+ case VideoCore::Surface::SurfaceType::ColorTexture:
+ return VK_IMAGE_ASPECT_COLOR_BIT;
+ case VideoCore::Surface::SurfaceType::Depth:
+ return VK_IMAGE_ASPECT_DEPTH_BIT;
+ case VideoCore::Surface::SurfaceType::DepthStencil:
+ return VK_IMAGE_ASPECT_DEPTH_BIT | VK_IMAGE_ASPECT_STENCIL_BIT;
+ default:
+ UNREACHABLE_MSG("Invalid surface type");
+ return VkImageAspectFlags{};
+ }
}
-} // Anonymous namespace
-
-CachedSurface::CachedSurface(const VKDevice& device, VKMemoryManager& memory_manager,
- VKScheduler& scheduler, VKStagingBufferPool& staging_pool,
- GPUVAddr gpu_addr, const SurfaceParams& params)
- : SurfaceBase<View>{gpu_addr, params, device.IsOptimalAstcSupported()}, device{device},
- memory_manager{memory_manager}, scheduler{scheduler}, staging_pool{staging_pool} {
- if (params.IsBuffer()) {
- buffer = CreateBuffer(device, params, host_memory_size);
- commit = memory_manager.Commit(buffer, false);
-
- const auto buffer_view_ci =
- GenerateBufferViewCreateInfo(device, params, *buffer, host_memory_size);
- format = buffer_view_ci.format;
-
- buffer_view = device.GetLogical().CreateBufferView(buffer_view_ci);
- } else {
- const auto image_ci = GenerateImageCreateInfo(device, params);
- format = image_ci.format;
-
- image.emplace(device, scheduler, image_ci, PixelFormatToImageAspect(params.pixel_format));
- commit = memory_manager.Commit(image->GetHandle(), false);
+[[nodiscard]] VkImageAspectFlags ImageViewAspectMask(const VideoCommon::ImageViewInfo& info) {
+ if (info.IsRenderTarget()) {
+ return ImageAspectMask(info.format);
}
-
- // TODO(Rodrigo): Move this to a virtual function.
- u32 num_layers = 1;
- if (params.is_layered || params.target == SurfaceTarget::Texture3D) {
- num_layers = params.depth;
+ const bool is_first = info.Swizzle()[0] == SwizzleSource::R;
+ switch (info.format) {
+ case PixelFormat::D24_UNORM_S8_UINT:
+ case PixelFormat::D32_FLOAT_S8_UINT:
+ return is_first ? VK_IMAGE_ASPECT_DEPTH_BIT : VK_IMAGE_ASPECT_STENCIL_BIT;
+ case PixelFormat::S8_UINT_D24_UNORM:
+ return is_first ? VK_IMAGE_ASPECT_STENCIL_BIT : VK_IMAGE_ASPECT_DEPTH_BIT;
+ case PixelFormat::D16_UNORM:
+ case PixelFormat::D32_FLOAT:
+ return VK_IMAGE_ASPECT_DEPTH_BIT;
+ default:
+ return VK_IMAGE_ASPECT_COLOR_BIT;
}
- main_view = CreateView(ViewParams(params.target, 0, num_layers, 0, params.num_levels));
}
-CachedSurface::~CachedSurface() = default;
-
-void CachedSurface::UploadTexture(const std::vector<u8>& staging_buffer) {
- // To upload data we have to be outside of a renderpass
- scheduler.RequestOutsideRenderPassOperationContext();
+[[nodiscard]] VkAttachmentDescription AttachmentDescription(const VKDevice& device,
+ const ImageView* image_view) {
+ const auto pixel_format = image_view->format;
+ return VkAttachmentDescription{
+ .flags = VK_ATTACHMENT_DESCRIPTION_MAY_ALIAS_BIT,
+ .format = MaxwellToVK::SurfaceFormat(device, FormatType::Optimal, pixel_format).format,
+ .samples = image_view->Samples(),
+ .loadOp = VK_ATTACHMENT_LOAD_OP_LOAD,
+ .storeOp = VK_ATTACHMENT_STORE_OP_STORE,
+ .stencilLoadOp = VK_ATTACHMENT_LOAD_OP_LOAD,
+ .stencilStoreOp = VK_ATTACHMENT_STORE_OP_STORE,
+ .initialLayout = VK_IMAGE_LAYOUT_GENERAL,
+ .finalLayout = VK_IMAGE_LAYOUT_GENERAL,
+ };
+}
- if (params.IsBuffer()) {
- UploadBuffer(staging_buffer);
- } else {
- UploadImage(staging_buffer);
+[[nodiscard]] VkComponentSwizzle ComponentSwizzle(SwizzleSource swizzle) {
+ switch (swizzle) {
+ case SwizzleSource::Zero:
+ return VK_COMPONENT_SWIZZLE_ZERO;
+ case SwizzleSource::R:
+ return VK_COMPONENT_SWIZZLE_R;
+ case SwizzleSource::G:
+ return VK_COMPONENT_SWIZZLE_G;
+ case SwizzleSource::B:
+ return VK_COMPONENT_SWIZZLE_B;
+ case SwizzleSource::A:
+ return VK_COMPONENT_SWIZZLE_A;
+ case SwizzleSource::OneFloat:
+ case SwizzleSource::OneInt:
+ return VK_COMPONENT_SWIZZLE_ONE;
}
+ UNREACHABLE_MSG("Invalid swizzle={}", swizzle);
+ return VK_COMPONENT_SWIZZLE_ZERO;
}
-void CachedSurface::DownloadTexture(std::vector<u8>& staging_buffer) {
- UNIMPLEMENTED_IF(params.IsBuffer());
-
- if (params.pixel_format == VideoCore::Surface::PixelFormat::A1B5G5R5_UNORM) {
- LOG_WARNING(Render_Vulkan, "A1B5G5R5 flushing is stubbed");
+[[nodiscard]] VkImageViewType ImageViewType(VideoCommon::ImageViewType type) {
+ switch (type) {
+ case VideoCommon::ImageViewType::e1D:
+ return VK_IMAGE_VIEW_TYPE_1D;
+ case VideoCommon::ImageViewType::e2D:
+ return VK_IMAGE_VIEW_TYPE_2D;
+ case VideoCommon::ImageViewType::Cube:
+ return VK_IMAGE_VIEW_TYPE_CUBE;
+ case VideoCommon::ImageViewType::e3D:
+ return VK_IMAGE_VIEW_TYPE_3D;
+ case VideoCommon::ImageViewType::e1DArray:
+ return VK_IMAGE_VIEW_TYPE_1D_ARRAY;
+ case VideoCommon::ImageViewType::e2DArray:
+ return VK_IMAGE_VIEW_TYPE_2D_ARRAY;
+ case VideoCommon::ImageViewType::CubeArray:
+ return VK_IMAGE_VIEW_TYPE_CUBE_ARRAY;
+ case VideoCommon::ImageViewType::Rect:
+ LOG_WARNING(Render_Vulkan, "Unnormalized image view type not supported");
+ return VK_IMAGE_VIEW_TYPE_2D;
+ case VideoCommon::ImageViewType::Buffer:
+ UNREACHABLE_MSG("Texture buffers can't be image views");
+ return VK_IMAGE_VIEW_TYPE_1D;
}
+ UNREACHABLE_MSG("Invalid image view type={}", type);
+ return VK_IMAGE_VIEW_TYPE_2D;
+}
- // We can't copy images to buffers inside a renderpass
- scheduler.RequestOutsideRenderPassOperationContext();
+[[nodiscard]] VkImageSubresourceLayers MakeImageSubresourceLayers(
+ VideoCommon::SubresourceLayers subresource, VkImageAspectFlags aspect_mask) {
+ return VkImageSubresourceLayers{
+ .aspectMask = aspect_mask,
+ .mipLevel = static_cast<u32>(subresource.base_level),
+ .baseArrayLayer = static_cast<u32>(subresource.base_layer),
+ .layerCount = static_cast<u32>(subresource.num_layers),
+ };
+}
- FullTransition(VK_PIPELINE_STAGE_TRANSFER_BIT, VK_ACCESS_TRANSFER_READ_BIT,
- VK_IMAGE_LAYOUT_TRANSFER_SRC_OPTIMAL);
+[[nodiscard]] VkOffset3D MakeOffset3D(VideoCommon::Offset3D offset3d) {
+ return VkOffset3D{
+ .x = offset3d.x,
+ .y = offset3d.y,
+ .z = offset3d.z,
+ };
+}
- const auto& buffer = staging_pool.GetUnusedBuffer(host_memory_size, true);
- // TODO(Rodrigo): Do this in a single copy
- for (u32 level = 0; level < params.num_levels; ++level) {
- scheduler.Record([image = *image->GetHandle(), buffer = *buffer.handle,
- copy = GetBufferImageCopy(level)](vk::CommandBuffer cmdbuf) {
- cmdbuf.CopyImageToBuffer(image, VK_IMAGE_LAYOUT_TRANSFER_SRC_OPTIMAL, buffer, copy);
- });
- }
- scheduler.Finish();
+[[nodiscard]] VkExtent3D MakeExtent3D(VideoCommon::Extent3D extent3d) {
+ return VkExtent3D{
+ .width = static_cast<u32>(extent3d.width),
+ .height = static_cast<u32>(extent3d.height),
+ .depth = static_cast<u32>(extent3d.depth),
+ };
+}
- // TODO(Rodrigo): Use an intern buffer for staging buffers and avoid this unnecessary memcpy.
- std::memcpy(staging_buffer.data(), buffer.commit->Map(host_memory_size), host_memory_size);
+[[nodiscard]] VkImageCopy MakeImageCopy(const VideoCommon::ImageCopy& copy,
+ VkImageAspectFlags aspect_mask) noexcept {
+ return VkImageCopy{
+ .srcSubresource = MakeImageSubresourceLayers(copy.src_subresource, aspect_mask),
+ .srcOffset = MakeOffset3D(copy.src_offset),
+ .dstSubresource = MakeImageSubresourceLayers(copy.dst_subresource, aspect_mask),
+ .dstOffset = MakeOffset3D(copy.dst_offset),
+ .extent = MakeExtent3D(copy.extent),
+ };
}
-void CachedSurface::DecorateSurfaceName() {
- // TODO(Rodrigo): Add name decorations
+[[nodiscard]] std::vector<VkBufferCopy> TransformBufferCopies(
+ std::span<const VideoCommon::BufferCopy> copies, size_t buffer_offset) {
+ std::vector<VkBufferCopy> result(copies.size());
+ std::ranges::transform(
+ copies, result.begin(), [buffer_offset](const VideoCommon::BufferCopy& copy) {
+ return VkBufferCopy{
+ .srcOffset = static_cast<VkDeviceSize>(copy.src_offset + buffer_offset),
+ .dstOffset = static_cast<VkDeviceSize>(copy.dst_offset),
+ .size = static_cast<VkDeviceSize>(copy.size),
+ };
+ });
+ return result;
}
-View CachedSurface::CreateView(const ViewParams& params) {
- // TODO(Rodrigo): Add name decorations
- return views[params] = std::make_shared<CachedSurfaceView>(device, *this, params);
+[[nodiscard]] std::vector<VkBufferImageCopy> TransformBufferImageCopies(
+ std::span<const BufferImageCopy> copies, size_t buffer_offset, VkImageAspectFlags aspect_mask) {
+ struct Maker {
+ VkBufferImageCopy operator()(const BufferImageCopy& copy) const {
+ return VkBufferImageCopy{
+ .bufferOffset = copy.buffer_offset + buffer_offset,
+ .bufferRowLength = copy.buffer_row_length,
+ .bufferImageHeight = copy.buffer_image_height,
+ .imageSubresource =
+ {
+ .aspectMask = aspect_mask,
+ .mipLevel = static_cast<u32>(copy.image_subresource.base_level),
+ .baseArrayLayer = static_cast<u32>(copy.image_subresource.base_layer),
+ .layerCount = static_cast<u32>(copy.image_subresource.num_layers),
+ },
+ .imageOffset =
+ {
+ .x = copy.image_offset.x,
+ .y = copy.image_offset.y,
+ .z = copy.image_offset.z,
+ },
+ .imageExtent =
+ {
+ .width = copy.image_extent.width,
+ .height = copy.image_extent.height,
+ .depth = copy.image_extent.depth,
+ },
+ };
+ }
+ size_t buffer_offset;
+ VkImageAspectFlags aspect_mask;
+ };
+ if (aspect_mask == (VK_IMAGE_ASPECT_DEPTH_BIT | VK_IMAGE_ASPECT_STENCIL_BIT)) {
+ std::vector<VkBufferImageCopy> result(copies.size() * 2);
+ std::ranges::transform(copies, result.begin(),
+ Maker{buffer_offset, VK_IMAGE_ASPECT_DEPTH_BIT});
+ std::ranges::transform(copies, result.begin() + copies.size(),
+ Maker{buffer_offset, VK_IMAGE_ASPECT_STENCIL_BIT});
+ return result;
+ } else {
+ std::vector<VkBufferImageCopy> result(copies.size());
+ std::ranges::transform(copies, result.begin(), Maker{buffer_offset, aspect_mask});
+ return result;
+ }
}
-void CachedSurface::UploadBuffer(const std::vector<u8>& staging_buffer) {
- const auto& src_buffer = staging_pool.GetUnusedBuffer(host_memory_size, true);
- std::memcpy(src_buffer.commit->Map(host_memory_size), staging_buffer.data(), host_memory_size);
+[[nodiscard]] VkImageSubresourceRange MakeSubresourceRange(VkImageAspectFlags aspect_mask,
+ const SubresourceRange& range) {
+ return VkImageSubresourceRange{
+ .aspectMask = aspect_mask,
+ .baseMipLevel = static_cast<u32>(range.base.level),
+ .levelCount = static_cast<u32>(range.extent.levels),
+ .baseArrayLayer = static_cast<u32>(range.base.layer),
+ .layerCount = static_cast<u32>(range.extent.layers),
+ };
+}
- scheduler.Record([src_buffer = *src_buffer.handle, dst_buffer = *buffer,
- size = host_memory_size](vk::CommandBuffer cmdbuf) {
- VkBufferCopy copy;
- copy.srcOffset = 0;
- copy.dstOffset = 0;
- copy.size = size;
- cmdbuf.CopyBuffer(src_buffer, dst_buffer, copy);
+[[nodiscard]] VkImageSubresourceRange MakeSubresourceRange(const ImageView* image_view) {
+ SubresourceRange range = image_view->range;
+ if (True(image_view->flags & VideoCommon::ImageViewFlagBits::Slice)) {
+ // Slice image views always affect a single layer, but their subresource range corresponds
+ // to the slice. Override the value to affect a single layer.
+ range.base.layer = 0;
+ range.extent.layers = 1;
+ }
+ return MakeSubresourceRange(ImageAspectMask(image_view->format), range);
+}
- VkBufferMemoryBarrier barrier;
- barrier.sType = VK_STRUCTURE_TYPE_BUFFER_MEMORY_BARRIER;
- barrier.pNext = nullptr;
- barrier.srcAccessMask = VK_ACCESS_TRANSFER_WRITE_BIT;
- barrier.dstAccessMask = VK_ACCESS_SHADER_READ_BIT;
- barrier.srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED; // They'll be ignored anyway
- barrier.dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED;
- barrier.buffer = dst_buffer;
- barrier.offset = 0;
- barrier.size = size;
- cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_TRANSFER_BIT, VK_PIPELINE_STAGE_VERTEX_SHADER_BIT,
- 0, {}, barrier, {});
- });
+[[nodiscard]] VkImageSubresourceLayers MakeSubresourceLayers(const ImageView* image_view) {
+ return VkImageSubresourceLayers{
+ .aspectMask = ImageAspectMask(image_view->format),
+ .mipLevel = static_cast<u32>(image_view->range.base.level),
+ .baseArrayLayer = static_cast<u32>(image_view->range.base.layer),
+ .layerCount = static_cast<u32>(image_view->range.extent.layers),
+ };
}
-void CachedSurface::UploadImage(const std::vector<u8>& staging_buffer) {
- const auto& src_buffer = staging_pool.GetUnusedBuffer(host_memory_size, true);
- std::memcpy(src_buffer.commit->Map(host_memory_size), staging_buffer.data(), host_memory_size);
-
- FullTransition(VK_PIPELINE_STAGE_TRANSFER_BIT, VK_ACCESS_TRANSFER_WRITE_BIT,
- VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL);
-
- for (u32 level = 0; level < params.num_levels; ++level) {
- const VkBufferImageCopy copy = GetBufferImageCopy(level);
- if (image->GetAspectMask() == (VK_IMAGE_ASPECT_DEPTH_BIT | VK_IMAGE_ASPECT_STENCIL_BIT)) {
- scheduler.Record([buffer = *src_buffer.handle, image = *image->GetHandle(),
- copy](vk::CommandBuffer cmdbuf) {
- std::array<VkBufferImageCopy, 2> copies = {copy, copy};
- copies[0].imageSubresource.aspectMask = VK_IMAGE_ASPECT_DEPTH_BIT;
- copies[1].imageSubresource.aspectMask = VK_IMAGE_ASPECT_STENCIL_BIT;
- cmdbuf.CopyBufferToImage(buffer, image, VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL,
- copies);
- });
- } else {
- scheduler.Record([buffer = *src_buffer.handle, image = *image->GetHandle(),
- copy](vk::CommandBuffer cmdbuf) {
- cmdbuf.CopyBufferToImage(buffer, image, VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL, copy);
- });
- }
+[[nodiscard]] constexpr SwizzleSource ConvertGreenRed(SwizzleSource value) {
+ switch (value) {
+ case SwizzleSource::G:
+ return SwizzleSource::R;
+ default:
+ return value;
}
}
-VkBufferImageCopy CachedSurface::GetBufferImageCopy(u32 level) const {
- return {
- .bufferOffset = params.GetHostMipmapLevelOffset(level, is_converted),
- .bufferRowLength = 0,
- .bufferImageHeight = 0,
- .imageSubresource =
+void CopyBufferToImage(vk::CommandBuffer cmdbuf, VkBuffer src_buffer, VkImage image,
+ VkImageAspectFlags aspect_mask, bool is_initialized,
+ std::span<const VkBufferImageCopy> copies) {
+ static constexpr VkAccessFlags ACCESS_FLAGS = VK_ACCESS_SHADER_WRITE_BIT |
+ VK_ACCESS_COLOR_ATTACHMENT_WRITE_BIT |
+ VK_ACCESS_DEPTH_STENCIL_ATTACHMENT_WRITE_BIT;
+ const VkImageMemoryBarrier read_barrier{
+ .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
+ .pNext = nullptr,
+ .srcAccessMask = ACCESS_FLAGS,
+ .dstAccessMask = VK_ACCESS_TRANSFER_WRITE_BIT,
+ .oldLayout = is_initialized ? VK_IMAGE_LAYOUT_GENERAL : VK_IMAGE_LAYOUT_UNDEFINED,
+ .newLayout = VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL,
+ .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+ .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+ .image = image,
+ .subresourceRange =
{
- .aspectMask = image->GetAspectMask(),
- .mipLevel = level,
+ .aspectMask = aspect_mask,
+ .baseMipLevel = 0,
+ .levelCount = VK_REMAINING_MIP_LEVELS,
.baseArrayLayer = 0,
- .layerCount = static_cast<u32>(params.GetNumLayers()),
+ .layerCount = VK_REMAINING_ARRAY_LAYERS,
},
- .imageOffset = {.x = 0, .y = 0, .z = 0},
- .imageExtent =
+ };
+ const VkImageMemoryBarrier write_barrier{
+ .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
+ .pNext = nullptr,
+ .srcAccessMask = VK_ACCESS_TRANSFER_WRITE_BIT,
+ .dstAccessMask = ACCESS_FLAGS,
+ .oldLayout = VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL,
+ .newLayout = VK_IMAGE_LAYOUT_GENERAL,
+ .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+ .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+ .image = image,
+ .subresourceRange =
{
- .width = params.GetMipWidth(level),
- .height = params.GetMipHeight(level),
- .depth = params.target == SurfaceTarget::Texture3D ? params.GetMipDepth(level) : 1U,
+ .aspectMask = aspect_mask,
+ .baseMipLevel = 0,
+ .levelCount = VK_REMAINING_MIP_LEVELS,
+ .baseArrayLayer = 0,
+ .layerCount = VK_REMAINING_ARRAY_LAYERS,
},
};
+ cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_ALL_COMMANDS_BIT, VK_PIPELINE_STAGE_TRANSFER_BIT, 0,
+ read_barrier);
+ cmdbuf.CopyBufferToImage(src_buffer, image, VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL, copies);
+ // TODO: Move this to another API
+ cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_TRANSFER_BIT, VK_PIPELINE_STAGE_ALL_COMMANDS_BIT, 0,
+ write_barrier);
}
-VkImageSubresourceRange CachedSurface::GetImageSubresourceRange() const {
- return {image->GetAspectMask(), 0, params.num_levels, 0,
- static_cast<u32>(params.GetNumLayers())};
+[[nodiscard]] VkImageBlit MakeImageBlit(const std::array<Offset2D, 2>& dst_region,
+ const std::array<Offset2D, 2>& src_region,
+ const VkImageSubresourceLayers& dst_layers,
+ const VkImageSubresourceLayers& src_layers) {
+ return VkImageBlit{
+ .srcSubresource = src_layers,
+ .srcOffsets =
+ {
+ {
+ .x = src_region[0].x,
+ .y = src_region[0].y,
+ .z = 0,
+ },
+ {
+ .x = src_region[1].x,
+ .y = src_region[1].y,
+ .z = 1,
+ },
+ },
+ .dstSubresource = dst_layers,
+ .dstOffsets =
+ {
+ {
+ .x = dst_region[0].x,
+ .y = dst_region[0].y,
+ .z = 0,
+ },
+ {
+ .x = dst_region[1].x,
+ .y = dst_region[1].y,
+ .z = 1,
+ },
+ },
+ };
}
-CachedSurfaceView::CachedSurfaceView(const VKDevice& device, CachedSurface& surface,
- const ViewParams& params)
- : VideoCommon::ViewBase{params}, params{surface.GetSurfaceParams()},
- image{surface.GetImageHandle()}, buffer_view{surface.GetBufferViewHandle()},
- aspect_mask{surface.GetAspectMask()}, device{device}, surface{surface},
- base_level{params.base_level}, num_levels{params.num_levels},
- image_view_type{image ? GetImageViewType(params.target) : VK_IMAGE_VIEW_TYPE_1D} {
- if (image_view_type == VK_IMAGE_VIEW_TYPE_3D) {
- base_layer = 0;
- num_layers = 1;
- base_slice = params.base_layer;
- num_slices = params.num_layers;
- } else {
- base_layer = params.base_layer;
- num_layers = params.num_layers;
- }
+[[nodiscard]] VkImageResolve MakeImageResolve(const std::array<Offset2D, 2>& dst_region,
+ const std::array<Offset2D, 2>& src_region,
+ const VkImageSubresourceLayers& dst_layers,
+ const VkImageSubresourceLayers& src_layers) {
+ return VkImageResolve{
+ .srcSubresource = src_layers,
+ .srcOffset =
+ {
+ .x = src_region[0].x,
+ .y = src_region[0].y,
+ .z = 0,
+ },
+ .dstSubresource = dst_layers,
+ .dstOffset =
+ {
+ .x = dst_region[0].x,
+ .y = dst_region[0].y,
+ .z = 0,
+ },
+ .extent =
+ {
+ .width = static_cast<u32>(dst_region[1].x - dst_region[0].x),
+ .height = static_cast<u32>(dst_region[1].y - dst_region[0].y),
+ .depth = 1,
+ },
+ };
}
-CachedSurfaceView::~CachedSurfaceView() = default;
-
-VkImageView CachedSurfaceView::GetImageView(SwizzleSource x_source, SwizzleSource y_source,
- SwizzleSource z_source, SwizzleSource w_source) {
- const u32 new_swizzle = EncodeSwizzle(x_source, y_source, z_source, w_source);
- if (last_image_view && last_swizzle == new_swizzle) {
- return last_image_view;
+struct RangedBarrierRange {
+ u32 min_mip = std::numeric_limits<u32>::max();
+ u32 max_mip = std::numeric_limits<u32>::min();
+ u32 min_layer = std::numeric_limits<u32>::max();
+ u32 max_layer = std::numeric_limits<u32>::min();
+
+ void AddLayers(const VkImageSubresourceLayers& layers) {
+ min_mip = std::min(min_mip, layers.mipLevel);
+ max_mip = std::max(max_mip, layers.mipLevel + 1);
+ min_layer = std::min(min_layer, layers.baseArrayLayer);
+ max_layer = std::max(max_layer, layers.baseArrayLayer + layers.layerCount);
}
- last_swizzle = new_swizzle;
- const auto [entry, is_cache_miss] = view_cache.try_emplace(new_swizzle);
- auto& image_view = entry->second;
- if (!is_cache_miss) {
- return last_image_view = *image_view;
+ VkImageSubresourceRange SubresourceRange(VkImageAspectFlags aspect_mask) const noexcept {
+ return VkImageSubresourceRange{
+ .aspectMask = aspect_mask,
+ .baseMipLevel = min_mip,
+ .levelCount = max_mip - min_mip,
+ .baseArrayLayer = min_layer,
+ .layerCount = max_layer - min_layer,
+ };
}
+};
- std::array swizzle{MaxwellToVK::SwizzleSource(x_source), MaxwellToVK::SwizzleSource(y_source),
- MaxwellToVK::SwizzleSource(z_source), MaxwellToVK::SwizzleSource(w_source)};
- if (params.pixel_format == VideoCore::Surface::PixelFormat::A1B5G5R5_UNORM) {
- // A1B5G5R5 is implemented as A1R5G5B5, we have to change the swizzle here.
- std::swap(swizzle[0], swizzle[2]);
- }
+} // Anonymous namespace
- // Games can sample depth or stencil values on textures. This is decided by the swizzle value on
- // hardware. To emulate this on Vulkan we specify it in the aspect.
- VkImageAspectFlags aspect = aspect_mask;
- if (aspect == (VK_IMAGE_ASPECT_DEPTH_BIT | VK_IMAGE_ASPECT_STENCIL_BIT)) {
- UNIMPLEMENTED_IF(x_source != SwizzleSource::R && x_source != SwizzleSource::G);
- const bool is_first = x_source == SwizzleSource::R;
- switch (params.pixel_format) {
- case VideoCore::Surface::PixelFormat::D24_UNORM_S8_UINT:
- case VideoCore::Surface::PixelFormat::D32_FLOAT_S8_UINT:
- aspect = is_first ? VK_IMAGE_ASPECT_DEPTH_BIT : VK_IMAGE_ASPECT_STENCIL_BIT;
- break;
- case VideoCore::Surface::PixelFormat::S8_UINT_D24_UNORM:
- aspect = is_first ? VK_IMAGE_ASPECT_STENCIL_BIT : VK_IMAGE_ASPECT_DEPTH_BIT;
- break;
- default:
- aspect = VK_IMAGE_ASPECT_DEPTH_BIT;
- UNIMPLEMENTED();
- }
+void TextureCacheRuntime::Finish() {
+ scheduler.Finish();
+}
- // Make sure we sample the first component
- std::transform(
- swizzle.begin(), swizzle.end(), swizzle.begin(), [](VkComponentSwizzle component) {
- return component == VK_COMPONENT_SWIZZLE_G ? VK_COMPONENT_SWIZZLE_R : component;
- });
- }
+ImageBufferMap TextureCacheRuntime::MapUploadBuffer(size_t size) {
+ const auto& buffer = staging_buffer_pool.GetUnusedBuffer(size, true);
+ return ImageBufferMap{
+ .handle = *buffer.handle,
+ .map = buffer.commit->Map(size),
+ };
+}
- if (image_view_type == VK_IMAGE_VIEW_TYPE_3D) {
- ASSERT(base_slice == 0);
- ASSERT(num_slices == params.depth);
+void TextureCacheRuntime::BlitImage(Framebuffer* dst_framebuffer, ImageView& dst, ImageView& src,
+ const std::array<Offset2D, 2>& dst_region,
+ const std::array<Offset2D, 2>& src_region,
+ Tegra::Engines::Fermi2D::Filter filter,
+ Tegra::Engines::Fermi2D::Operation operation) {
+ const VkImageAspectFlags aspect_mask = ImageAspectMask(src.format);
+ const bool is_dst_msaa = dst.Samples() != VK_SAMPLE_COUNT_1_BIT;
+ const bool is_src_msaa = src.Samples() != VK_SAMPLE_COUNT_1_BIT;
+ ASSERT(aspect_mask == ImageAspectMask(dst.format));
+ if (aspect_mask == VK_IMAGE_ASPECT_COLOR_BIT && !is_src_msaa && !is_dst_msaa) {
+ blit_image_helper.BlitColor(dst_framebuffer, src, dst_region, src_region, filter,
+ operation);
+ return;
}
-
- image_view = device.GetLogical().CreateImageView({
- .sType = VK_STRUCTURE_TYPE_IMAGE_VIEW_CREATE_INFO,
- .pNext = nullptr,
- .flags = 0,
- .image = surface.GetImageHandle(),
- .viewType = image_view_type,
- .format = surface.GetImage().GetFormat(),
- .components =
- {
- .r = swizzle[0],
- .g = swizzle[1],
- .b = swizzle[2],
- .a = swizzle[3],
+ if (aspect_mask == (VK_IMAGE_ASPECT_DEPTH_BIT | VK_IMAGE_ASPECT_STENCIL_BIT)) {
+ if (!device.IsBlitDepthStencilSupported()) {
+ UNIMPLEMENTED_IF(is_src_msaa || is_dst_msaa);
+ blit_image_helper.BlitDepthStencil(dst_framebuffer, src.DepthView(), src.StencilView(),
+ dst_region, src_region, filter, operation);
+ return;
+ }
+ }
+ ASSERT(src.ImageFormat() == dst.ImageFormat());
+ ASSERT(!(is_dst_msaa && !is_src_msaa));
+ ASSERT(operation == Fermi2D::Operation::SrcCopy);
+
+ const VkImage dst_image = dst.ImageHandle();
+ const VkImage src_image = src.ImageHandle();
+ const VkImageSubresourceLayers dst_layers = MakeSubresourceLayers(&dst);
+ const VkImageSubresourceLayers src_layers = MakeSubresourceLayers(&src);
+ const bool is_resolve = is_src_msaa && !is_dst_msaa;
+ scheduler.RequestOutsideRenderPassOperationContext();
+ scheduler.Record([filter, dst_region, src_region, dst_image, src_image, dst_layers, src_layers,
+ aspect_mask, is_resolve](vk::CommandBuffer cmdbuf) {
+ const std::array read_barriers{
+ VkImageMemoryBarrier{
+ .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
+ .pNext = nullptr,
+ .srcAccessMask = VK_ACCESS_SHADER_WRITE_BIT |
+ VK_ACCESS_DEPTH_STENCIL_ATTACHMENT_WRITE_BIT |
+ VK_ACCESS_TRANSFER_WRITE_BIT,
+ .dstAccessMask = VK_ACCESS_TRANSFER_READ_BIT,
+ .oldLayout = VK_IMAGE_LAYOUT_GENERAL,
+ .newLayout = VK_IMAGE_LAYOUT_GENERAL,
+ .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+ .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+ .image = src_image,
+ .subresourceRange{
+ .aspectMask = aspect_mask,
+ .baseMipLevel = 0,
+ .levelCount = VK_REMAINING_MIP_LEVELS,
+ .baseArrayLayer = 0,
+ .layerCount = VK_REMAINING_ARRAY_LAYERS,
+ },
},
- .subresourceRange =
- {
- .aspectMask = aspect,
- .baseMipLevel = base_level,
- .levelCount = num_levels,
- .baseArrayLayer = base_layer,
- .layerCount = num_layers,
+ VkImageMemoryBarrier{
+ .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
+ .pNext = nullptr,
+ .srcAccessMask = VK_ACCESS_SHADER_WRITE_BIT |
+ VK_ACCESS_DEPTH_STENCIL_ATTACHMENT_WRITE_BIT |
+ VK_ACCESS_TRANSFER_WRITE_BIT,
+ .dstAccessMask = VK_ACCESS_TRANSFER_WRITE_BIT,
+ .oldLayout = VK_IMAGE_LAYOUT_GENERAL,
+ .newLayout = VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL,
+ .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+ .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+ .image = dst_image,
+ .subresourceRange{
+ .aspectMask = aspect_mask,
+ .baseMipLevel = 0,
+ .levelCount = VK_REMAINING_MIP_LEVELS,
+ .baseArrayLayer = 0,
+ .layerCount = VK_REMAINING_ARRAY_LAYERS,
+ },
+ },
+ };
+ VkImageMemoryBarrier write_barrier{
+ .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
+ .pNext = nullptr,
+ .srcAccessMask = VK_ACCESS_TRANSFER_WRITE_BIT,
+ .dstAccessMask = VK_ACCESS_SHADER_READ_BIT | VK_ACCESS_SHADER_WRITE_BIT |
+ VK_ACCESS_DEPTH_STENCIL_ATTACHMENT_READ_BIT |
+ VK_ACCESS_DEPTH_STENCIL_ATTACHMENT_WRITE_BIT |
+ VK_ACCESS_TRANSFER_READ_BIT | VK_ACCESS_TRANSFER_WRITE_BIT,
+ .oldLayout = VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL,
+ .newLayout = VK_IMAGE_LAYOUT_GENERAL,
+ .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+ .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+ .image = dst_image,
+ .subresourceRange{
+ .aspectMask = aspect_mask,
+ .baseMipLevel = 0,
+ .levelCount = VK_REMAINING_MIP_LEVELS,
+ .baseArrayLayer = 0,
+ .layerCount = VK_REMAINING_ARRAY_LAYERS,
},
+ };
+ cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_ALL_COMMANDS_BIT, VK_PIPELINE_STAGE_TRANSFER_BIT,
+ 0, nullptr, nullptr, read_barriers);
+ if (is_resolve) {
+ cmdbuf.ResolveImage(src_image, VK_IMAGE_LAYOUT_GENERAL, dst_image,
+ VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL,
+ MakeImageResolve(dst_region, src_region, dst_layers, src_layers));
+ } else {
+ const bool is_linear = filter == Fermi2D::Filter::Bilinear;
+ const VkFilter vk_filter = is_linear ? VK_FILTER_LINEAR : VK_FILTER_NEAREST;
+ cmdbuf.BlitImage(
+ src_image, VK_IMAGE_LAYOUT_GENERAL, dst_image, VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL,
+ MakeImageBlit(dst_region, src_region, dst_layers, src_layers), vk_filter);
+ }
+ cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_TRANSFER_BIT, VK_PIPELINE_STAGE_ALL_COMMANDS_BIT,
+ 0, write_barrier);
});
-
- return last_image_view = *image_view;
}
-VkImageView CachedSurfaceView::GetAttachment() {
- if (render_target) {
- return *render_target;
+void TextureCacheRuntime::ConvertImage(Framebuffer* dst, ImageView& dst_view, ImageView& src_view) {
+ switch (dst_view.format) {
+ case PixelFormat::R16_UNORM:
+ if (src_view.format == PixelFormat::D16_UNORM) {
+ return blit_image_helper.ConvertD16ToR16(dst, src_view);
+ }
+ break;
+ case PixelFormat::R32_FLOAT:
+ if (src_view.format == PixelFormat::D32_FLOAT) {
+ return blit_image_helper.ConvertD32ToR32(dst, src_view);
+ }
+ break;
+ case PixelFormat::D16_UNORM:
+ if (src_view.format == PixelFormat::R16_UNORM) {
+ return blit_image_helper.ConvertR16ToD16(dst, src_view);
+ }
+ break;
+ case PixelFormat::D32_FLOAT:
+ if (src_view.format == PixelFormat::R32_FLOAT) {
+ return blit_image_helper.ConvertR32ToD32(dst, src_view);
+ }
+ break;
+ default:
+ break;
}
+ UNIMPLEMENTED_MSG("Unimplemented format copy from {} to {}", src_view.format, dst_view.format);
+}
- VkImageViewCreateInfo ci{
- .sType = VK_STRUCTURE_TYPE_IMAGE_VIEW_CREATE_INFO,
- .pNext = nullptr,
- .flags = 0,
- .image = surface.GetImageHandle(),
- .viewType = VK_IMAGE_VIEW_TYPE_1D,
- .format = surface.GetImage().GetFormat(),
- .components =
- {
- .r = VK_COMPONENT_SWIZZLE_IDENTITY,
- .g = VK_COMPONENT_SWIZZLE_IDENTITY,
- .b = VK_COMPONENT_SWIZZLE_IDENTITY,
- .a = VK_COMPONENT_SWIZZLE_IDENTITY,
+void TextureCacheRuntime::CopyImage(Image& dst, Image& src,
+ std::span<const VideoCommon::ImageCopy> copies) {
+ std::vector<VkImageCopy> vk_copies(copies.size());
+ const VkImageAspectFlags aspect_mask = dst.AspectMask();
+ ASSERT(aspect_mask == src.AspectMask());
+
+ std::ranges::transform(copies, vk_copies.begin(), [aspect_mask](const auto& copy) {
+ return MakeImageCopy(copy, aspect_mask);
+ });
+ const VkImage dst_image = dst.Handle();
+ const VkImage src_image = src.Handle();
+ scheduler.RequestOutsideRenderPassOperationContext();
+ scheduler.Record([dst_image, src_image, aspect_mask, vk_copies](vk::CommandBuffer cmdbuf) {
+ RangedBarrierRange dst_range;
+ RangedBarrierRange src_range;
+ for (const VkImageCopy& copy : vk_copies) {
+ dst_range.AddLayers(copy.dstSubresource);
+ src_range.AddLayers(copy.srcSubresource);
+ }
+ const std::array read_barriers{
+ VkImageMemoryBarrier{
+ .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
+ .pNext = nullptr,
+ .srcAccessMask = VK_ACCESS_SHADER_WRITE_BIT | VK_ACCESS_COLOR_ATTACHMENT_WRITE_BIT |
+ VK_ACCESS_DEPTH_STENCIL_ATTACHMENT_WRITE_BIT |
+ VK_ACCESS_TRANSFER_WRITE_BIT,
+ .dstAccessMask = VK_ACCESS_TRANSFER_WRITE_BIT,
+ .oldLayout = VK_IMAGE_LAYOUT_GENERAL,
+ .newLayout = VK_IMAGE_LAYOUT_GENERAL,
+ .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+ .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+ .image = src_image,
+ .subresourceRange = src_range.SubresourceRange(aspect_mask),
},
- .subresourceRange =
- {
- .aspectMask = aspect_mask,
- .baseMipLevel = base_level,
- .levelCount = num_levels,
- .baseArrayLayer = 0,
- .layerCount = 0,
+ VkImageMemoryBarrier{
+ .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
+ .pNext = nullptr,
+ .srcAccessMask = VK_ACCESS_SHADER_READ_BIT | VK_ACCESS_SHADER_WRITE_BIT |
+ VK_ACCESS_COLOR_ATTACHMENT_READ_BIT |
+ VK_ACCESS_COLOR_ATTACHMENT_WRITE_BIT |
+ VK_ACCESS_DEPTH_STENCIL_ATTACHMENT_READ_BIT |
+ VK_ACCESS_DEPTH_STENCIL_ATTACHMENT_WRITE_BIT |
+ VK_ACCESS_TRANSFER_READ_BIT | VK_ACCESS_TRANSFER_WRITE_BIT,
+ .dstAccessMask = VK_ACCESS_TRANSFER_WRITE_BIT,
+ .oldLayout = VK_IMAGE_LAYOUT_GENERAL,
+ .newLayout = VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL,
+ .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+ .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+ .image = dst_image,
+ .subresourceRange = dst_range.SubresourceRange(aspect_mask),
},
- };
- if (image_view_type == VK_IMAGE_VIEW_TYPE_3D) {
- ci.viewType = num_slices > 1 ? VK_IMAGE_VIEW_TYPE_2D_ARRAY : VK_IMAGE_VIEW_TYPE_2D;
- ci.subresourceRange.baseArrayLayer = base_slice;
- ci.subresourceRange.layerCount = num_slices;
+ };
+ const VkImageMemoryBarrier write_barrier{
+ .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
+ .pNext = nullptr,
+ .srcAccessMask = VK_ACCESS_TRANSFER_WRITE_BIT,
+ .dstAccessMask = VK_ACCESS_SHADER_READ_BIT | VK_ACCESS_SHADER_WRITE_BIT |
+ VK_ACCESS_COLOR_ATTACHMENT_READ_BIT |
+ VK_ACCESS_COLOR_ATTACHMENT_WRITE_BIT |
+ VK_ACCESS_DEPTH_STENCIL_ATTACHMENT_READ_BIT |
+ VK_ACCESS_DEPTH_STENCIL_ATTACHMENT_WRITE_BIT |
+ VK_ACCESS_TRANSFER_READ_BIT | VK_ACCESS_TRANSFER_WRITE_BIT,
+ .oldLayout = VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL,
+ .newLayout = VK_IMAGE_LAYOUT_GENERAL,
+ .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+ .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+ .image = dst_image,
+ .subresourceRange = dst_range.SubresourceRange(aspect_mask),
+ };
+ cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_ALL_COMMANDS_BIT, VK_PIPELINE_STAGE_TRANSFER_BIT,
+ 0, {}, {}, read_barriers);
+ cmdbuf.CopyImage(src_image, VK_IMAGE_LAYOUT_GENERAL, dst_image,
+ VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL, vk_copies);
+ cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_TRANSFER_BIT, VK_PIPELINE_STAGE_ALL_COMMANDS_BIT,
+ 0, write_barrier);
+ });
+}
+
+Image::Image(TextureCacheRuntime& runtime, const ImageInfo& info_, GPUVAddr gpu_addr_,
+ VAddr cpu_addr_)
+ : VideoCommon::ImageBase(info_, gpu_addr_, cpu_addr_), scheduler{&runtime.scheduler},
+ image(MakeImage(runtime.device, info)), buffer(MakeBuffer(runtime.device, info)),
+ aspect_mask(ImageAspectMask(info.format)) {
+ if (image) {
+ commit = runtime.memory_manager.Commit(image, false);
} else {
- ci.viewType = image_view_type;
- ci.subresourceRange.baseArrayLayer = base_layer;
- ci.subresourceRange.layerCount = num_layers;
+ commit = runtime.memory_manager.Commit(buffer, false);
+ }
+ if (IsPixelFormatASTC(info.format) && !runtime.device.IsOptimalAstcSupported()) {
+ flags |= VideoCommon::ImageFlagBits::Converted;
+ }
+ if (runtime.device.HasDebuggingToolAttached()) {
+ if (image) {
+ image.SetObjectNameEXT(VideoCommon::Name(*this).c_str());
+ } else {
+ buffer.SetObjectNameEXT(VideoCommon::Name(*this).c_str());
+ }
}
- render_target = device.GetLogical().CreateImageView(ci);
- return *render_target;
}
-VKTextureCache::VKTextureCache(VideoCore::RasterizerInterface& rasterizer,
- Tegra::Engines::Maxwell3D& maxwell3d,
- Tegra::MemoryManager& gpu_memory, const VKDevice& device_,
- VKMemoryManager& memory_manager_, VKScheduler& scheduler_,
- VKStagingBufferPool& staging_pool_)
- : TextureCache(rasterizer, maxwell3d, gpu_memory, device_.IsOptimalAstcSupported()),
- device{device_}, memory_manager{memory_manager_}, scheduler{scheduler_}, staging_pool{
- staging_pool_} {}
-
-VKTextureCache::~VKTextureCache() = default;
-
-Surface VKTextureCache::CreateSurface(GPUVAddr gpu_addr, const SurfaceParams& params) {
- return std::make_shared<CachedSurface>(device, memory_manager, scheduler, staging_pool,
- gpu_addr, params);
+void Image::UploadMemory(const ImageBufferMap& map, size_t buffer_offset,
+ std::span<const BufferImageCopy> copies) {
+ // TODO: Move this to another API
+ scheduler->RequestOutsideRenderPassOperationContext();
+ std::vector vk_copies = TransformBufferImageCopies(copies, buffer_offset, aspect_mask);
+ const VkBuffer src_buffer = map.handle;
+ const VkImage vk_image = *image;
+ const VkImageAspectFlags vk_aspect_mask = aspect_mask;
+ const bool is_initialized = std::exchange(initialized, true);
+ scheduler->Record([src_buffer, vk_image, vk_aspect_mask, is_initialized,
+ vk_copies](vk::CommandBuffer cmdbuf) {
+ CopyBufferToImage(cmdbuf, src_buffer, vk_image, vk_aspect_mask, is_initialized, vk_copies);
+ });
}
-void VKTextureCache::ImageCopy(Surface& src_surface, Surface& dst_surface,
- const VideoCommon::CopyParams& copy_params) {
- const bool src_3d = src_surface->GetSurfaceParams().target == SurfaceTarget::Texture3D;
- const bool dst_3d = dst_surface->GetSurfaceParams().target == SurfaceTarget::Texture3D;
- UNIMPLEMENTED_IF(src_3d);
+void Image::UploadMemory(const ImageBufferMap& map, size_t buffer_offset,
+ std::span<const VideoCommon::BufferCopy> copies) {
+ // TODO: Move this to another API
+ scheduler->RequestOutsideRenderPassOperationContext();
+ std::vector vk_copies = TransformBufferCopies(copies, buffer_offset);
+ const VkBuffer src_buffer = map.handle;
+ const VkBuffer dst_buffer = *buffer;
+ scheduler->Record([src_buffer, dst_buffer, vk_copies](vk::CommandBuffer cmdbuf) {
+ // TODO: Barriers
+ cmdbuf.CopyBuffer(src_buffer, dst_buffer, vk_copies);
+ });
+}
- // The texture cache handles depth in OpenGL terms, we have to handle it as subresource and
- // dimension respectively.
- const u32 dst_base_layer = dst_3d ? 0 : copy_params.dest_z;
- const u32 dst_offset_z = dst_3d ? copy_params.dest_z : 0;
+void Image::DownloadMemory(const ImageBufferMap& map, size_t buffer_offset,
+ std::span<const BufferImageCopy> copies) {
+ std::vector vk_copies = TransformBufferImageCopies(copies, buffer_offset, aspect_mask);
+ scheduler->Record([buffer = map.handle, image = *image, aspect_mask = aspect_mask,
+ vk_copies](vk::CommandBuffer cmdbuf) {
+ // TODO: Barriers
+ cmdbuf.CopyImageToBuffer(image, VK_IMAGE_LAYOUT_GENERAL, buffer, vk_copies);
+ });
+}
- const u32 extent_z = dst_3d ? copy_params.depth : 1;
- const u32 num_layers = dst_3d ? 1 : copy_params.depth;
+ImageView::ImageView(TextureCacheRuntime& runtime, const VideoCommon::ImageViewInfo& info,
+ ImageId image_id_, Image& image)
+ : VideoCommon::ImageViewBase{info, image.info, image_id_}, device{&runtime.device},
+ image_handle{image.Handle()}, image_format{image.info.format}, samples{ConvertSampleCount(
+ image.info.num_samples)} {
+ const VkImageAspectFlags aspect_mask = ImageViewAspectMask(info);
+ std::array<SwizzleSource, 4> swizzle{
+ SwizzleSource::R,
+ SwizzleSource::G,
+ SwizzleSource::B,
+ SwizzleSource::A,
+ };
+ if (!info.IsRenderTarget()) {
+ swizzle = info.Swizzle();
+ if ((aspect_mask & (VK_IMAGE_ASPECT_DEPTH_BIT | VK_IMAGE_ASPECT_STENCIL_BIT)) != 0) {
+ std::ranges::transform(swizzle, swizzle.begin(), ConvertGreenRed);
+ }
+ }
+ const VkFormat vk_format =
+ MaxwellToVK::SurfaceFormat(*device, FormatType::Optimal, format).format;
+ const VkImageViewCreateInfo create_info{
+ .sType = VK_STRUCTURE_TYPE_IMAGE_VIEW_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .image = image.Handle(),
+ .viewType = VkImageViewType{},
+ .format = vk_format,
+ .components{
+ .r = ComponentSwizzle(swizzle[0]),
+ .g = ComponentSwizzle(swizzle[1]),
+ .b = ComponentSwizzle(swizzle[2]),
+ .a = ComponentSwizzle(swizzle[3]),
+ },
+ .subresourceRange = MakeSubresourceRange(aspect_mask, info.range),
+ };
+ const auto create = [&](VideoCommon::ImageViewType view_type, std::optional<u32> num_layers) {
+ VkImageViewCreateInfo ci{create_info};
+ ci.viewType = ImageViewType(view_type);
+ if (num_layers) {
+ ci.subresourceRange.layerCount = *num_layers;
+ }
+ vk::ImageView handle = device->GetLogical().CreateImageView(ci);
+ if (device->HasDebuggingToolAttached()) {
+ handle.SetObjectNameEXT(VideoCommon::Name(*this, view_type).c_str());
+ }
+ image_views[static_cast<size_t>(view_type)] = std::move(handle);
+ };
+ switch (info.type) {
+ case VideoCommon::ImageViewType::e1D:
+ case VideoCommon::ImageViewType::e1DArray:
+ create(VideoCommon::ImageViewType::e1D, 1);
+ create(VideoCommon::ImageViewType::e1DArray, std::nullopt);
+ render_target = Handle(VideoCommon::ImageViewType::e1DArray);
+ break;
+ case VideoCommon::ImageViewType::e2D:
+ case VideoCommon::ImageViewType::e2DArray:
+ create(VideoCommon::ImageViewType::e2D, 1);
+ create(VideoCommon::ImageViewType::e2DArray, std::nullopt);
+ render_target = Handle(VideoCommon::ImageViewType::e2DArray);
+ break;
+ case VideoCommon::ImageViewType::e3D:
+ create(VideoCommon::ImageViewType::e3D, std::nullopt);
+ render_target = Handle(VideoCommon::ImageViewType::e3D);
+ break;
+ case VideoCommon::ImageViewType::Cube:
+ case VideoCommon::ImageViewType::CubeArray:
+ create(VideoCommon::ImageViewType::Cube, 6);
+ create(VideoCommon::ImageViewType::CubeArray, std::nullopt);
+ break;
+ case VideoCommon::ImageViewType::Rect:
+ UNIMPLEMENTED();
+ break;
+ case VideoCommon::ImageViewType::Buffer:
+ buffer_view = device->GetLogical().CreateBufferView(VkBufferViewCreateInfo{
+ .sType = VK_STRUCTURE_TYPE_BUFFER_VIEW_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .buffer = image.Buffer(),
+ .format = vk_format,
+ .offset = 0, // TODO: Redesign buffer cache to support this
+ .range = image.guest_size_bytes,
+ });
+ break;
+ }
+}
- // We can't copy inside a renderpass
- scheduler.RequestOutsideRenderPassOperationContext();
+ImageView::ImageView(TextureCacheRuntime&, const VideoCommon::NullImageParams& params)
+ : VideoCommon::ImageViewBase{params} {}
- src_surface->Transition(copy_params.source_z, copy_params.depth, copy_params.source_level, 1,
- VK_PIPELINE_STAGE_TRANSFER_BIT, VK_ACCESS_TRANSFER_READ_BIT,
- VK_IMAGE_LAYOUT_TRANSFER_SRC_OPTIMAL);
- dst_surface->Transition(dst_base_layer, num_layers, copy_params.dest_level, 1,
- VK_PIPELINE_STAGE_TRANSFER_BIT, VK_ACCESS_TRANSFER_WRITE_BIT,
- VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL);
+VkImageView ImageView::DepthView() {
+ if (depth_view) {
+ return *depth_view;
+ }
+ depth_view = MakeDepthStencilView(VK_IMAGE_ASPECT_DEPTH_BIT);
+ return *depth_view;
+}
- const VkImageCopy copy{
- .srcSubresource =
- {
- .aspectMask = src_surface->GetAspectMask(),
- .mipLevel = copy_params.source_level,
- .baseArrayLayer = copy_params.source_z,
- .layerCount = num_layers,
- },
- .srcOffset =
- {
- .x = static_cast<s32>(copy_params.source_x),
- .y = static_cast<s32>(copy_params.source_y),
- .z = 0,
- },
- .dstSubresource =
- {
- .aspectMask = dst_surface->GetAspectMask(),
- .mipLevel = copy_params.dest_level,
- .baseArrayLayer = dst_base_layer,
- .layerCount = num_layers,
- },
- .dstOffset =
- {
- .x = static_cast<s32>(copy_params.dest_x),
- .y = static_cast<s32>(copy_params.dest_y),
- .z = static_cast<s32>(dst_offset_z),
- },
- .extent =
- {
- .width = copy_params.width,
- .height = copy_params.height,
- .depth = extent_z,
- },
- };
+VkImageView ImageView::StencilView() {
+ if (stencil_view) {
+ return *stencil_view;
+ }
+ stencil_view = MakeDepthStencilView(VK_IMAGE_ASPECT_STENCIL_BIT);
+ return *stencil_view;
+}
- const VkImage src_image = src_surface->GetImageHandle();
- const VkImage dst_image = dst_surface->GetImageHandle();
- scheduler.Record([src_image, dst_image, copy](vk::CommandBuffer cmdbuf) {
- cmdbuf.CopyImage(src_image, VK_IMAGE_LAYOUT_TRANSFER_SRC_OPTIMAL, dst_image,
- VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL, copy);
+vk::ImageView ImageView::MakeDepthStencilView(VkImageAspectFlags aspect_mask) {
+ return device->GetLogical().CreateImageView({
+ .sType = VK_STRUCTURE_TYPE_IMAGE_VIEW_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .image = image_handle,
+ .viewType = ImageViewType(type),
+ .format = MaxwellToVK::SurfaceFormat(*device, FormatType::Optimal, format).format,
+ .components{
+ .r = VK_COMPONENT_SWIZZLE_IDENTITY,
+ .g = VK_COMPONENT_SWIZZLE_IDENTITY,
+ .b = VK_COMPONENT_SWIZZLE_IDENTITY,
+ .a = VK_COMPONENT_SWIZZLE_IDENTITY,
+ },
+ .subresourceRange = MakeSubresourceRange(aspect_mask, range),
});
}
-void VKTextureCache::ImageBlit(View& src_view, View& dst_view,
- const Tegra::Engines::Fermi2D::Config& copy_config) {
- // We can't blit inside a renderpass
- scheduler.RequestOutsideRenderPassOperationContext();
-
- src_view->Transition(VK_IMAGE_LAYOUT_TRANSFER_SRC_OPTIMAL, VK_PIPELINE_STAGE_TRANSFER_BIT,
- VK_ACCESS_TRANSFER_READ_BIT);
- dst_view->Transition(VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL, VK_PIPELINE_STAGE_TRANSFER_BIT,
- VK_ACCESS_TRANSFER_WRITE_BIT);
-
- VkImageBlit blit;
- blit.srcSubresource = src_view->GetImageSubresourceLayers();
- blit.srcOffsets[0].x = copy_config.src_rect.left;
- blit.srcOffsets[0].y = copy_config.src_rect.top;
- blit.srcOffsets[0].z = 0;
- blit.srcOffsets[1].x = copy_config.src_rect.right;
- blit.srcOffsets[1].y = copy_config.src_rect.bottom;
- blit.srcOffsets[1].z = 1;
- blit.dstSubresource = dst_view->GetImageSubresourceLayers();
- blit.dstOffsets[0].x = copy_config.dst_rect.left;
- blit.dstOffsets[0].y = copy_config.dst_rect.top;
- blit.dstOffsets[0].z = 0;
- blit.dstOffsets[1].x = copy_config.dst_rect.right;
- blit.dstOffsets[1].y = copy_config.dst_rect.bottom;
- blit.dstOffsets[1].z = 1;
-
- const bool is_linear = copy_config.filter == Tegra::Engines::Fermi2D::Filter::Linear;
-
- scheduler.Record([src_image = src_view->GetImage(), dst_image = dst_view->GetImage(), blit,
- is_linear](vk::CommandBuffer cmdbuf) {
- cmdbuf.BlitImage(src_image, VK_IMAGE_LAYOUT_TRANSFER_SRC_OPTIMAL, dst_image,
- VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL, blit,
- is_linear ? VK_FILTER_LINEAR : VK_FILTER_NEAREST);
+Sampler::Sampler(TextureCacheRuntime& runtime, const Tegra::Texture::TSCEntry& tsc) {
+ const auto& device = runtime.device;
+ const bool arbitrary_borders = runtime.device.IsExtCustomBorderColorSupported();
+ const std::array<float, 4> color = tsc.BorderColor();
+ // C++20 bit_cast
+ VkClearColorValue border_color;
+ std::memcpy(&border_color, &color, sizeof(color));
+ const VkSamplerCustomBorderColorCreateInfoEXT border_ci{
+ .sType = VK_STRUCTURE_TYPE_SAMPLER_CUSTOM_BORDER_COLOR_CREATE_INFO_EXT,
+ .pNext = nullptr,
+ .customBorderColor = border_color,
+ .format = VK_FORMAT_UNDEFINED,
+ };
+ const void* pnext = nullptr;
+ if (arbitrary_borders) {
+ pnext = &border_ci;
+ }
+ const VkSamplerReductionModeCreateInfoEXT reduction_ci{
+ .sType = VK_STRUCTURE_TYPE_SAMPLER_REDUCTION_MODE_CREATE_INFO_EXT,
+ .pNext = pnext,
+ .reductionMode = MaxwellToVK::SamplerReduction(tsc.reduction_filter),
+ };
+ if (runtime.device.IsExtSamplerFilterMinmaxSupported()) {
+ pnext = &reduction_ci;
+ } else if (reduction_ci.reductionMode != VK_SAMPLER_REDUCTION_MODE_WEIGHTED_AVERAGE_EXT) {
+ LOG_WARNING(Render_Vulkan, "VK_EXT_sampler_filter_minmax is required");
+ }
+ // Some games have samplers with garbage. Sanitize them here.
+ const float max_anisotropy = std::clamp(tsc.MaxAnisotropy(), 1.0f, 16.0f);
+ sampler = device.GetLogical().CreateSampler(VkSamplerCreateInfo{
+ .sType = VK_STRUCTURE_TYPE_SAMPLER_CREATE_INFO,
+ .pNext = pnext,
+ .flags = 0,
+ .magFilter = MaxwellToVK::Sampler::Filter(tsc.mag_filter),
+ .minFilter = MaxwellToVK::Sampler::Filter(tsc.min_filter),
+ .mipmapMode = MaxwellToVK::Sampler::MipmapMode(tsc.mipmap_filter),
+ .addressModeU = MaxwellToVK::Sampler::WrapMode(device, tsc.wrap_u, tsc.mag_filter),
+ .addressModeV = MaxwellToVK::Sampler::WrapMode(device, tsc.wrap_v, tsc.mag_filter),
+ .addressModeW = MaxwellToVK::Sampler::WrapMode(device, tsc.wrap_p, tsc.mag_filter),
+ .mipLodBias = tsc.LodBias(),
+ .anisotropyEnable = static_cast<VkBool32>(max_anisotropy > 1.0f ? VK_TRUE : VK_FALSE),
+ .maxAnisotropy = max_anisotropy,
+ .compareEnable = tsc.depth_compare_enabled,
+ .compareOp = MaxwellToVK::Sampler::DepthCompareFunction(tsc.depth_compare_func),
+ .minLod = tsc.mipmap_filter == TextureMipmapFilter::None ? 0.0f : tsc.MinLod(),
+ .maxLod = tsc.mipmap_filter == TextureMipmapFilter::None ? 0.25f : tsc.MaxLod(),
+ .borderColor =
+ arbitrary_borders ? VK_BORDER_COLOR_INT_CUSTOM_EXT : ConvertBorderColor(color),
+ .unnormalizedCoordinates = VK_FALSE,
});
}
-void VKTextureCache::BufferCopy(Surface& src_surface, Surface& dst_surface) {
- // Currently unimplemented. PBO copies should be dropped and we should use a render pass to
- // convert from color to depth and viceversa.
- LOG_WARNING(Render_Vulkan, "Unimplemented");
+Framebuffer::Framebuffer(TextureCacheRuntime& runtime, std::span<ImageView*, NUM_RT> color_buffers,
+ ImageView* depth_buffer, const VideoCommon::RenderTargets& key) {
+ std::vector<VkAttachmentDescription> descriptions;
+ std::vector<VkImageView> attachments;
+ RenderPassKey renderpass_key{};
+ s32 num_layers = 1;
+
+ for (size_t index = 0; index < NUM_RT; ++index) {
+ const ImageView* const color_buffer = color_buffers[index];
+ if (!color_buffer) {
+ renderpass_key.color_formats[index] = PixelFormat::Invalid;
+ continue;
+ }
+ descriptions.push_back(AttachmentDescription(runtime.device, color_buffer));
+ attachments.push_back(color_buffer->RenderTarget());
+ renderpass_key.color_formats[index] = color_buffer->format;
+ num_layers = std::max(num_layers, color_buffer->range.extent.layers);
+ images[num_images] = color_buffer->ImageHandle();
+ image_ranges[num_images] = MakeSubresourceRange(color_buffer);
+ samples = color_buffer->Samples();
+ ++num_images;
+ }
+ const size_t num_colors = attachments.size();
+ const VkAttachmentReference* depth_attachment =
+ depth_buffer ? &ATTACHMENT_REFERENCES[num_colors] : nullptr;
+ if (depth_buffer) {
+ descriptions.push_back(AttachmentDescription(runtime.device, depth_buffer));
+ attachments.push_back(depth_buffer->RenderTarget());
+ renderpass_key.depth_format = depth_buffer->format;
+ num_layers = std::max(num_layers, depth_buffer->range.extent.layers);
+ images[num_images] = depth_buffer->ImageHandle();
+ image_ranges[num_images] = MakeSubresourceRange(depth_buffer);
+ samples = depth_buffer->Samples();
+ ++num_images;
+ } else {
+ renderpass_key.depth_format = PixelFormat::Invalid;
+ }
+ renderpass_key.samples = samples;
+
+ const auto& device = runtime.device.GetLogical();
+ const auto [cache_pair, is_new] = runtime.renderpass_cache.try_emplace(renderpass_key);
+ if (is_new) {
+ const VkSubpassDescription subpass{
+ .flags = 0,
+ .pipelineBindPoint = VK_PIPELINE_BIND_POINT_GRAPHICS,
+ .inputAttachmentCount = 0,
+ .pInputAttachments = nullptr,
+ .colorAttachmentCount = static_cast<u32>(num_colors),
+ .pColorAttachments = num_colors != 0 ? ATTACHMENT_REFERENCES.data() : nullptr,
+ .pResolveAttachments = nullptr,
+ .pDepthStencilAttachment = depth_attachment,
+ .preserveAttachmentCount = 0,
+ .pPreserveAttachments = nullptr,
+ };
+ cache_pair->second = device.CreateRenderPass(VkRenderPassCreateInfo{
+ .sType = VK_STRUCTURE_TYPE_RENDER_PASS_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .attachmentCount = static_cast<u32>(descriptions.size()),
+ .pAttachments = descriptions.data(),
+ .subpassCount = 1,
+ .pSubpasses = &subpass,
+ .dependencyCount = 0,
+ .pDependencies = nullptr,
+ });
+ }
+ renderpass = *cache_pair->second;
+ render_area = VkExtent2D{
+ .width = key.size.width,
+ .height = key.size.height,
+ };
+ num_color_buffers = static_cast<u32>(num_colors);
+ framebuffer = device.CreateFramebuffer(VkFramebufferCreateInfo{
+ .sType = VK_STRUCTURE_TYPE_FRAMEBUFFER_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .renderPass = renderpass,
+ .attachmentCount = static_cast<u32>(attachments.size()),
+ .pAttachments = attachments.data(),
+ .width = key.size.width,
+ .height = key.size.height,
+ .layers = static_cast<u32>(num_layers),
+ });
+ if (runtime.device.HasDebuggingToolAttached()) {
+ framebuffer.SetObjectNameEXT(VideoCommon::Name(key).c_str());
+ }
}
} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/vk_texture_cache.h b/src/video_core/renderer_vulkan/vk_texture_cache.h
index 39202feba..edc3d80c0 100644
--- a/src/video_core/renderer_vulkan/vk_texture_cache.h
+++ b/src/video_core/renderer_vulkan/vk_texture_cache.h
@@ -4,216 +4,265 @@
#pragma once
-#include <memory>
-#include <unordered_map>
+#include <compare>
+#include <span>
-#include "common/common_types.h"
-#include "video_core/renderer_vulkan/vk_image.h"
#include "video_core/renderer_vulkan/vk_memory_manager.h"
-#include "video_core/renderer_vulkan/vk_scheduler.h"
#include "video_core/renderer_vulkan/wrapper.h"
-#include "video_core/texture_cache/surface_base.h"
#include "video_core/texture_cache/texture_cache.h"
-namespace VideoCore {
-class RasterizerInterface;
-}
-
namespace Vulkan {
-class RasterizerVulkan;
+using VideoCommon::ImageId;
+using VideoCommon::NUM_RT;
+using VideoCommon::Offset2D;
+using VideoCommon::RenderTargets;
+using VideoCore::Surface::PixelFormat;
+
class VKDevice;
class VKScheduler;
class VKStagingBufferPool;
-class CachedSurfaceView;
-class CachedSurface;
+class BlitImageHelper;
+class Image;
+class ImageView;
+class Framebuffer;
-using Surface = std::shared_ptr<CachedSurface>;
-using View = std::shared_ptr<CachedSurfaceView>;
-using TextureCacheBase = VideoCommon::TextureCache<Surface, View>;
+struct RenderPassKey {
+ constexpr auto operator<=>(const RenderPassKey&) const noexcept = default;
-using VideoCommon::SurfaceParams;
-using VideoCommon::ViewParams;
+ std::array<PixelFormat, NUM_RT> color_formats;
+ PixelFormat depth_format;
+ VkSampleCountFlagBits samples;
+};
-class CachedSurface final : public VideoCommon::SurfaceBase<View> {
- friend CachedSurfaceView;
+} // namespace Vulkan
-public:
- explicit CachedSurface(const VKDevice& device, VKMemoryManager& memory_manager,
- VKScheduler& scheduler, VKStagingBufferPool& staging_pool,
- GPUVAddr gpu_addr, const SurfaceParams& params);
- ~CachedSurface();
+namespace std {
+template <>
+struct hash<Vulkan::RenderPassKey> {
+ [[nodiscard]] constexpr size_t operator()(const Vulkan::RenderPassKey& key) const noexcept {
+ size_t value = static_cast<size_t>(key.depth_format) << 48;
+ value ^= static_cast<size_t>(key.samples) << 52;
+ for (size_t i = 0; i < key.color_formats.size(); ++i) {
+ value ^= static_cast<size_t>(key.color_formats[i]) << (i * 6);
+ }
+ return value;
+ }
+};
+} // namespace std
- void UploadTexture(const std::vector<u8>& staging_buffer) override;
- void DownloadTexture(std::vector<u8>& staging_buffer) override;
+namespace Vulkan {
- void FullTransition(VkPipelineStageFlags new_stage_mask, VkAccessFlags new_access,
- VkImageLayout new_layout) {
- image->Transition(0, static_cast<u32>(params.GetNumLayers()), 0, params.num_levels,
- new_stage_mask, new_access, new_layout);
+struct ImageBufferMap {
+ [[nodiscard]] VkBuffer Handle() const noexcept {
+ return handle;
}
- void Transition(u32 base_layer, u32 num_layers, u32 base_level, u32 num_levels,
- VkPipelineStageFlags new_stage_mask, VkAccessFlags new_access,
- VkImageLayout new_layout) {
- image->Transition(base_layer, num_layers, base_level, num_levels, new_stage_mask,
- new_access, new_layout);
+ [[nodiscard]] std::span<u8> Span() const noexcept {
+ return map.Span();
}
- VKImage& GetImage() {
- return *image;
- }
+ VkBuffer handle;
+ MemoryMap map;
+};
- const VKImage& GetImage() const {
- return *image;
- }
+struct TextureCacheRuntime {
+ const VKDevice& device;
+ VKScheduler& scheduler;
+ VKMemoryManager& memory_manager;
+ VKStagingBufferPool& staging_buffer_pool;
+ BlitImageHelper& blit_image_helper;
+ std::unordered_map<RenderPassKey, vk::RenderPass> renderpass_cache;
+
+ void Finish();
- VkImage GetImageHandle() const {
- return *image->GetHandle();
+ [[nodiscard]] ImageBufferMap MapUploadBuffer(size_t size);
+
+ [[nodiscard]] ImageBufferMap MapDownloadBuffer(size_t size) {
+ // TODO: Have a special function for this
+ return MapUploadBuffer(size);
}
- VkImageAspectFlags GetAspectMask() const {
- return image->GetAspectMask();
+ void BlitImage(Framebuffer* dst_framebuffer, ImageView& dst, ImageView& src,
+ const std::array<Offset2D, 2>& dst_region,
+ const std::array<Offset2D, 2>& src_region,
+ Tegra::Engines::Fermi2D::Filter filter,
+ Tegra::Engines::Fermi2D::Operation operation);
+
+ void CopyImage(Image& dst, Image& src, std::span<const VideoCommon::ImageCopy> copies);
+
+ void ConvertImage(Framebuffer* dst, ImageView& dst_view, ImageView& src_view);
+
+ [[nodiscard]] bool CanAccelerateImageUpload(Image&) const noexcept {
+ return false;
}
- VkBufferView GetBufferViewHandle() const {
- return *buffer_view;
+ void AccelerateImageUpload(Image&, const ImageBufferMap&, size_t,
+ std::span<const VideoCommon::SwizzleParameters>) {
+ UNREACHABLE();
}
-protected:
- void DecorateSurfaceName();
+ void InsertUploadMemoryBarrier() {}
+};
- View CreateView(const ViewParams& params) override;
+class Image : public VideoCommon::ImageBase {
+public:
+ explicit Image(TextureCacheRuntime&, const VideoCommon::ImageInfo& info, GPUVAddr gpu_addr,
+ VAddr cpu_addr);
-private:
- void UploadBuffer(const std::vector<u8>& staging_buffer);
+ void UploadMemory(const ImageBufferMap& map, size_t buffer_offset,
+ std::span<const VideoCommon::BufferImageCopy> copies);
- void UploadImage(const std::vector<u8>& staging_buffer);
+ void UploadMemory(const ImageBufferMap& map, size_t buffer_offset,
+ std::span<const VideoCommon::BufferCopy> copies);
- VkBufferImageCopy GetBufferImageCopy(u32 level) const;
+ void DownloadMemory(const ImageBufferMap& map, size_t buffer_offset,
+ std::span<const VideoCommon::BufferImageCopy> copies);
- VkImageSubresourceRange GetImageSubresourceRange() const;
+ [[nodiscard]] VkImage Handle() const noexcept {
+ return *image;
+ }
- const VKDevice& device;
- VKMemoryManager& memory_manager;
- VKScheduler& scheduler;
- VKStagingBufferPool& staging_pool;
+ [[nodiscard]] VkBuffer Buffer() const noexcept {
+ return *buffer;
+ }
+
+ [[nodiscard]] VkImageCreateFlags AspectMask() const noexcept {
+ return aspect_mask;
+ }
- std::optional<VKImage> image;
+private:
+ VKScheduler* scheduler;
+ vk::Image image;
vk::Buffer buffer;
- vk::BufferView buffer_view;
VKMemoryCommit commit;
-
- VkFormat format = VK_FORMAT_UNDEFINED;
+ VkImageAspectFlags aspect_mask = 0;
+ bool initialized = false;
};
-class CachedSurfaceView final : public VideoCommon::ViewBase {
+class ImageView : public VideoCommon::ImageViewBase {
public:
- explicit CachedSurfaceView(const VKDevice& device, CachedSurface& surface,
- const ViewParams& params);
- ~CachedSurfaceView();
+ explicit ImageView(TextureCacheRuntime&, const VideoCommon::ImageViewInfo&, ImageId, Image&);
+ explicit ImageView(TextureCacheRuntime&, const VideoCommon::NullImageParams&);
- VkImageView GetImageView(Tegra::Texture::SwizzleSource x_source,
- Tegra::Texture::SwizzleSource y_source,
- Tegra::Texture::SwizzleSource z_source,
- Tegra::Texture::SwizzleSource w_source);
+ [[nodiscard]] VkImageView DepthView();
- VkImageView GetAttachment();
+ [[nodiscard]] VkImageView StencilView();
- bool IsSameSurface(const CachedSurfaceView& rhs) const {
- return &surface == &rhs.surface;
+ [[nodiscard]] VkImageView Handle(VideoCommon::ImageViewType query_type) const noexcept {
+ return *image_views[static_cast<size_t>(query_type)];
}
- u32 GetWidth() const {
- return params.GetMipWidth(base_level);
+ [[nodiscard]] VkBufferView BufferView() const noexcept {
+ return *buffer_view;
}
- u32 GetHeight() const {
- return params.GetMipHeight(base_level);
+ [[nodiscard]] VkImage ImageHandle() const noexcept {
+ return image_handle;
}
- u32 GetNumLayers() const {
- return num_layers;
+ [[nodiscard]] VkImageView RenderTarget() const noexcept {
+ return render_target;
}
- bool IsBufferView() const {
- return buffer_view;
+ [[nodiscard]] PixelFormat ImageFormat() const noexcept {
+ return image_format;
}
- VkImage GetImage() const {
- return image;
+ [[nodiscard]] VkSampleCountFlagBits Samples() const noexcept {
+ return samples;
}
- VkBufferView GetBufferView() const {
- return buffer_view;
- }
+private:
+ [[nodiscard]] vk::ImageView MakeDepthStencilView(VkImageAspectFlags aspect_mask);
- VkImageSubresourceRange GetImageSubresourceRange() const {
- return {aspect_mask, base_level, num_levels, base_layer, num_layers};
- }
+ const VKDevice* device = nullptr;
+ std::array<vk::ImageView, VideoCommon::NUM_IMAGE_VIEW_TYPES> image_views;
+ vk::ImageView depth_view;
+ vk::ImageView stencil_view;
+ vk::BufferView buffer_view;
+ VkImage image_handle = VK_NULL_HANDLE;
+ VkImageView render_target = VK_NULL_HANDLE;
+ PixelFormat image_format = PixelFormat::Invalid;
+ VkSampleCountFlagBits samples = VK_SAMPLE_COUNT_1_BIT;
+};
- VkImageSubresourceLayers GetImageSubresourceLayers() const {
- return {surface.GetAspectMask(), base_level, base_layer, num_layers};
- }
+class ImageAlloc : public VideoCommon::ImageAllocBase {};
- void Transition(VkImageLayout new_layout, VkPipelineStageFlags new_stage_mask,
- VkAccessFlags new_access) const {
- surface.Transition(base_layer, num_layers, base_level, num_levels, new_stage_mask,
- new_access, new_layout);
- }
+class Sampler {
+public:
+ explicit Sampler(TextureCacheRuntime&, const Tegra::Texture::TSCEntry&);
- void MarkAsModified(u64 tick) {
- surface.MarkAsModified(true, tick);
+ [[nodiscard]] VkSampler Handle() const noexcept {
+ return *sampler;
}
private:
- // Store a copy of these values to avoid double dereference when reading them
- const SurfaceParams params;
- const VkImage image;
- const VkBufferView buffer_view;
- const VkImageAspectFlags aspect_mask;
-
- const VKDevice& device;
- CachedSurface& surface;
- const u32 base_level;
- const u32 num_levels;
- const VkImageViewType image_view_type;
- u32 base_layer = 0;
- u32 num_layers = 0;
- u32 base_slice = 0;
- u32 num_slices = 0;
-
- VkImageView last_image_view = nullptr;
- u32 last_swizzle = 0;
-
- vk::ImageView render_target;
- std::unordered_map<u32, vk::ImageView> view_cache;
+ vk::Sampler sampler;
};
-class VKTextureCache final : public TextureCacheBase {
+class Framebuffer {
public:
- explicit VKTextureCache(VideoCore::RasterizerInterface& rasterizer,
- Tegra::Engines::Maxwell3D& maxwell3d, Tegra::MemoryManager& gpu_memory,
- const VKDevice& device, VKMemoryManager& memory_manager,
- VKScheduler& scheduler, VKStagingBufferPool& staging_pool);
- ~VKTextureCache();
+ explicit Framebuffer(TextureCacheRuntime&, std::span<ImageView*, NUM_RT> color_buffers,
+ ImageView* depth_buffer, const VideoCommon::RenderTargets& key);
-private:
- Surface CreateSurface(GPUVAddr gpu_addr, const SurfaceParams& params) override;
+ [[nodiscard]] VkFramebuffer Handle() const noexcept {
+ return *framebuffer;
+ }
- void ImageCopy(Surface& src_surface, Surface& dst_surface,
- const VideoCommon::CopyParams& copy_params) override;
+ [[nodiscard]] VkRenderPass RenderPass() const noexcept {
+ return renderpass;
+ }
- void ImageBlit(View& src_view, View& dst_view,
- const Tegra::Engines::Fermi2D::Config& copy_config) override;
+ [[nodiscard]] VkExtent2D RenderArea() const noexcept {
+ return render_area;
+ }
- void BufferCopy(Surface& src_surface, Surface& dst_surface) override;
+ [[nodiscard]] VkSampleCountFlagBits Samples() const noexcept {
+ return samples;
+ }
- const VKDevice& device;
- VKMemoryManager& memory_manager;
- VKScheduler& scheduler;
- VKStagingBufferPool& staging_pool;
+ [[nodiscard]] u32 NumColorBuffers() const noexcept {
+ return num_color_buffers;
+ }
+
+ [[nodiscard]] u32 NumImages() const noexcept {
+ return num_images;
+ }
+
+ [[nodiscard]] const std::array<VkImage, 9>& Images() const noexcept {
+ return images;
+ }
+
+ [[nodiscard]] const std::array<VkImageSubresourceRange, 9>& ImageRanges() const noexcept {
+ return image_ranges;
+ }
+
+private:
+ vk::Framebuffer framebuffer;
+ VkRenderPass renderpass{};
+ VkExtent2D render_area{};
+ VkSampleCountFlagBits samples = VK_SAMPLE_COUNT_1_BIT;
+ u32 num_color_buffers = 0;
+ u32 num_images = 0;
+ std::array<VkImage, 9> images{};
+ std::array<VkImageSubresourceRange, 9> image_ranges{};
+};
+
+struct TextureCacheParams {
+ static constexpr bool ENABLE_VALIDATION = true;
+ static constexpr bool FRAMEBUFFER_BLITS = false;
+ static constexpr bool HAS_EMULATED_COPIES = false;
+
+ using Runtime = Vulkan::TextureCacheRuntime;
+ using Image = Vulkan::Image;
+ using ImageAlloc = Vulkan::ImageAlloc;
+ using ImageView = Vulkan::ImageView;
+ using Sampler = Vulkan::Sampler;
+ using Framebuffer = Vulkan::Framebuffer;
};
+using TextureCache = VideoCommon::TextureCache<TextureCacheParams>;
+
} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/vk_update_descriptor.cpp b/src/video_core/renderer_vulkan/vk_update_descriptor.cpp
index 351c048d2..8826da325 100644
--- a/src/video_core/renderer_vulkan/vk_update_descriptor.cpp
+++ b/src/video_core/renderer_vulkan/vk_update_descriptor.cpp
@@ -14,8 +14,8 @@
namespace Vulkan {
-VKUpdateDescriptorQueue::VKUpdateDescriptorQueue(const VKDevice& device, VKScheduler& scheduler)
- : device{device}, scheduler{scheduler} {}
+VKUpdateDescriptorQueue::VKUpdateDescriptorQueue(const VKDevice& device_, VKScheduler& scheduler_)
+ : device{device_}, scheduler{scheduler_} {}
VKUpdateDescriptorQueue::~VKUpdateDescriptorQueue() = default;
diff --git a/src/video_core/renderer_vulkan/vk_update_descriptor.h b/src/video_core/renderer_vulkan/vk_update_descriptor.h
index 945320c72..f098a8540 100644
--- a/src/video_core/renderer_vulkan/vk_update_descriptor.h
+++ b/src/video_core/renderer_vulkan/vk_update_descriptor.h
@@ -31,7 +31,7 @@ struct DescriptorUpdateEntry {
class VKUpdateDescriptorQueue final {
public:
- explicit VKUpdateDescriptorQueue(const VKDevice& device, VKScheduler& scheduler);
+ explicit VKUpdateDescriptorQueue(const VKDevice& device_, VKScheduler& scheduler_);
~VKUpdateDescriptorQueue();
void TickFrame();
@@ -40,30 +40,34 @@ public:
void Send(VkDescriptorUpdateTemplateKHR update_template, VkDescriptorSet set);
- void AddSampledImage(VkSampler sampler, VkImageView image_view) {
- payload.emplace_back(VkDescriptorImageInfo{sampler, image_view, {}});
+ void AddSampledImage(VkImageView image_view, VkSampler sampler) {
+ payload.emplace_back(VkDescriptorImageInfo{
+ .sampler = sampler,
+ .imageView = image_view,
+ .imageLayout = VK_IMAGE_LAYOUT_GENERAL,
+ });
}
void AddImage(VkImageView image_view) {
- payload.emplace_back(VkDescriptorImageInfo{{}, image_view, {}});
+ payload.emplace_back(VkDescriptorImageInfo{
+ .sampler = VK_NULL_HANDLE,
+ .imageView = image_view,
+ .imageLayout = VK_IMAGE_LAYOUT_GENERAL,
+ });
}
- void AddBuffer(VkBuffer buffer, u64 offset, std::size_t size) {
- payload.emplace_back(VkDescriptorBufferInfo{buffer, offset, size});
+ void AddBuffer(VkBuffer buffer, u64 offset, size_t size) {
+ payload.emplace_back(VkDescriptorBufferInfo{
+ .buffer = buffer,
+ .offset = offset,
+ .range = size,
+ });
}
void AddTexelBuffer(VkBufferView texel_buffer) {
payload.emplace_back(texel_buffer);
}
- VkImageLayout* LastImageLayout() {
- return &payload.back().image.imageLayout;
- }
-
- const VkImageLayout* LastImageLayout() const {
- return &payload.back().image.imageLayout;
- }
-
private:
const VKDevice& device;
VKScheduler& scheduler;
diff --git a/src/video_core/renderer_vulkan/wrapper.cpp b/src/video_core/renderer_vulkan/wrapper.cpp
index 4e83303d8..2a21e850d 100644
--- a/src/video_core/renderer_vulkan/wrapper.cpp
+++ b/src/video_core/renderer_vulkan/wrapper.cpp
@@ -81,6 +81,7 @@ void Load(VkDevice device, DeviceDispatch& dld) noexcept {
X(vkCmdBeginQuery);
X(vkCmdBeginRenderPass);
X(vkCmdBeginTransformFeedbackEXT);
+ X(vkCmdBeginDebugUtilsLabelEXT);
X(vkCmdBindDescriptorSets);
X(vkCmdBindIndexBuffer);
X(vkCmdBindPipeline);
@@ -98,6 +99,7 @@ void Load(VkDevice device, DeviceDispatch& dld) noexcept {
X(vkCmdEndQuery);
X(vkCmdEndRenderPass);
X(vkCmdEndTransformFeedbackEXT);
+ X(vkCmdEndDebugUtilsLabelEXT);
X(vkCmdFillBuffer);
X(vkCmdPipelineBarrier);
X(vkCmdPushConstants);
@@ -121,6 +123,7 @@ void Load(VkDevice device, DeviceDispatch& dld) noexcept {
X(vkCmdSetPrimitiveTopologyEXT);
X(vkCmdSetStencilOpEXT);
X(vkCmdSetStencilTestEnableEXT);
+ X(vkCmdResolveImage);
X(vkCreateBuffer);
X(vkCreateBufferView);
X(vkCreateCommandPool);
@@ -176,6 +179,8 @@ void Load(VkDevice device, DeviceDispatch& dld) noexcept {
X(vkQueueSubmit);
X(vkResetFences);
X(vkResetQueryPoolEXT);
+ X(vkSetDebugUtilsObjectNameEXT);
+ X(vkSetDebugUtilsObjectTagEXT);
X(vkUnmapMemory);
X(vkUpdateDescriptorSetWithTemplateKHR);
X(vkUpdateDescriptorSets);
@@ -184,6 +189,19 @@ void Load(VkDevice device, DeviceDispatch& dld) noexcept {
#undef X
}
+template <typename T>
+void SetObjectName(const DeviceDispatch* dld, VkDevice device, T handle, VkObjectType type,
+ const char* name) {
+ const VkDebugUtilsObjectNameInfoEXT name_info{
+ .sType = VK_STRUCTURE_TYPE_DEBUG_UTILS_OBJECT_NAME_INFO_EXT,
+ .pNext = nullptr,
+ .objectType = VK_OBJECT_TYPE_IMAGE,
+ .objectHandle = reinterpret_cast<u64>(handle),
+ .pObjectName = name,
+ };
+ Check(dld->vkSetDebugUtilsObjectNameEXT(device, &name_info));
+}
+
} // Anonymous namespace
bool Load(InstanceDispatch& dld) noexcept {
@@ -417,7 +435,7 @@ VkResult Free(VkDevice device, VkCommandPool handle, Span<VkCommandBuffer> buffe
}
Instance Instance::Create(u32 version, Span<const char*> layers, Span<const char*> extensions,
- InstanceDispatch& dld) noexcept {
+ InstanceDispatch& dispatch) noexcept {
const VkApplicationInfo application_info{
.sType = VK_STRUCTURE_TYPE_APPLICATION_INFO,
.pNext = nullptr,
@@ -439,17 +457,17 @@ Instance Instance::Create(u32 version, Span<const char*> layers, Span<const char
};
VkInstance instance;
- if (dld.vkCreateInstance(&ci, nullptr, &instance) != VK_SUCCESS) {
+ if (dispatch.vkCreateInstance(&ci, nullptr, &instance) != VK_SUCCESS) {
// Failed to create the instance.
return {};
}
- if (!Proc(dld.vkDestroyInstance, dld, "vkDestroyInstance", instance)) {
+ if (!Proc(dispatch.vkDestroyInstance, dispatch, "vkDestroyInstance", instance)) {
// We successfully created an instance but the destroy function couldn't be loaded.
// This is a good moment to panic.
return {};
}
- return Instance(instance, dld);
+ return Instance(instance, dispatch);
}
std::optional<std::vector<VkPhysicalDevice>> Instance::EnumeratePhysicalDevices() {
@@ -476,8 +494,7 @@ DebugCallback Instance::TryCreateDebugCallback(
VK_DEBUG_UTILS_MESSAGE_SEVERITY_INFO_BIT_EXT |
VK_DEBUG_UTILS_MESSAGE_SEVERITY_VERBOSE_BIT_EXT,
.messageType = VK_DEBUG_UTILS_MESSAGE_TYPE_GENERAL_BIT_EXT |
- VK_DEBUG_UTILS_MESSAGE_TYPE_VALIDATION_BIT_EXT |
- VK_DEBUG_UTILS_MESSAGE_TYPE_PERFORMANCE_BIT_EXT,
+ VK_DEBUG_UTILS_MESSAGE_TYPE_VALIDATION_BIT_EXT,
.pfnUserCallback = callback,
.pUserData = nullptr,
};
@@ -493,10 +510,38 @@ void Buffer::BindMemory(VkDeviceMemory memory, VkDeviceSize offset) const {
Check(dld->vkBindBufferMemory(owner, handle, memory, offset));
}
+void Buffer::SetObjectNameEXT(const char* name) const {
+ SetObjectName(dld, owner, handle, VK_OBJECT_TYPE_BUFFER, name);
+}
+
+void BufferView::SetObjectNameEXT(const char* name) const {
+ SetObjectName(dld, owner, handle, VK_OBJECT_TYPE_BUFFER_VIEW, name);
+}
+
void Image::BindMemory(VkDeviceMemory memory, VkDeviceSize offset) const {
Check(dld->vkBindImageMemory(owner, handle, memory, offset));
}
+void Image::SetObjectNameEXT(const char* name) const {
+ SetObjectName(dld, owner, handle, VK_OBJECT_TYPE_IMAGE, name);
+}
+
+void ImageView::SetObjectNameEXT(const char* name) const {
+ SetObjectName(dld, owner, handle, VK_OBJECT_TYPE_IMAGE_VIEW, name);
+}
+
+void DeviceMemory::SetObjectNameEXT(const char* name) const {
+ SetObjectName(dld, owner, handle, VK_OBJECT_TYPE_DEVICE_MEMORY, name);
+}
+
+void Fence::SetObjectNameEXT(const char* name) const {
+ SetObjectName(dld, owner, handle, VK_OBJECT_TYPE_FENCE, name);
+}
+
+void Framebuffer::SetObjectNameEXT(const char* name) const {
+ SetObjectName(dld, owner, handle, VK_OBJECT_TYPE_FRAMEBUFFER, name);
+}
+
DescriptorSets DescriptorPool::Allocate(const VkDescriptorSetAllocateInfo& ai) const {
const std::size_t num = ai.descriptorSetCount;
std::unique_ptr sets = std::make_unique<VkDescriptorSet[]>(num);
@@ -510,6 +555,10 @@ DescriptorSets DescriptorPool::Allocate(const VkDescriptorSetAllocateInfo& ai) c
}
}
+void DescriptorPool::SetObjectNameEXT(const char* name) const {
+ SetObjectName(dld, owner, handle, VK_OBJECT_TYPE_DESCRIPTOR_POOL, name);
+}
+
CommandBuffers CommandPool::Allocate(std::size_t num_buffers, VkCommandBufferLevel level) const {
const VkCommandBufferAllocateInfo ai{
.sType = VK_STRUCTURE_TYPE_COMMAND_BUFFER_ALLOCATE_INFO,
@@ -530,6 +579,10 @@ CommandBuffers CommandPool::Allocate(std::size_t num_buffers, VkCommandBufferLev
}
}
+void CommandPool::SetObjectNameEXT(const char* name) const {
+ SetObjectName(dld, owner, handle, VK_OBJECT_TYPE_COMMAND_POOL, name);
+}
+
std::vector<VkImage> SwapchainKHR::GetImages() const {
u32 num;
Check(dld->vkGetSwapchainImagesKHR(owner, handle, &num, nullptr));
@@ -538,9 +591,21 @@ std::vector<VkImage> SwapchainKHR::GetImages() const {
return images;
}
+void Event::SetObjectNameEXT(const char* name) const {
+ SetObjectName(dld, owner, handle, VK_OBJECT_TYPE_EVENT, name);
+}
+
+void ShaderModule::SetObjectNameEXT(const char* name) const {
+ SetObjectName(dld, owner, handle, VK_OBJECT_TYPE_SHADER_MODULE, name);
+}
+
+void Semaphore::SetObjectNameEXT(const char* name) const {
+ SetObjectName(dld, owner, handle, VK_OBJECT_TYPE_SEMAPHORE, name);
+}
+
Device Device::Create(VkPhysicalDevice physical_device, Span<VkDeviceQueueCreateInfo> queues_ci,
Span<const char*> enabled_extensions, const void* next,
- DeviceDispatch& dld) noexcept {
+ DeviceDispatch& dispatch) noexcept {
const VkDeviceCreateInfo ci{
.sType = VK_STRUCTURE_TYPE_DEVICE_CREATE_INFO,
.pNext = next,
@@ -555,11 +620,11 @@ Device Device::Create(VkPhysicalDevice physical_device, Span<VkDeviceQueueCreate
};
VkDevice device;
- if (dld.vkCreateDevice(physical_device, &ci, nullptr, &device) != VK_SUCCESS) {
+ if (dispatch.vkCreateDevice(physical_device, &ci, nullptr, &device) != VK_SUCCESS) {
return {};
}
- Load(device, dld);
- return Device(device, dld);
+ Load(device, dispatch);
+ return Device(device, dispatch);
}
Queue Device::GetQueue(u32 family_index) const noexcept {
diff --git a/src/video_core/renderer_vulkan/wrapper.h b/src/video_core/renderer_vulkan/wrapper.h
index f64919623..f9a184e00 100644
--- a/src/video_core/renderer_vulkan/wrapper.h
+++ b/src/video_core/renderer_vulkan/wrapper.h
@@ -9,6 +9,7 @@
#include <limits>
#include <memory>
#include <optional>
+#include <span>
#include <type_traits>
#include <utility>
#include <vector>
@@ -18,6 +19,10 @@
#include "common/common_types.h"
+#ifdef _MSC_VER
+#pragma warning(disable : 26812) // Disable prefer enum class over enum
+#endif
+
namespace Vulkan::vk {
/**
@@ -41,6 +46,9 @@ public:
/// Construct an empty span.
constexpr Span() noexcept = default;
+ /// Construct an empty span
+ constexpr Span(std::nullptr_t) noexcept {}
+
/// Construct a span from a single element.
constexpr Span(const T& value) noexcept : ptr{&value}, num{1} {}
@@ -52,7 +60,7 @@ public:
/// Construct a span from a pointer and a size.
/// This is inteded for subranges.
- constexpr Span(const T* ptr, std::size_t num) noexcept : ptr{ptr}, num{num} {}
+ constexpr Span(const T* ptr_, std::size_t num_) noexcept : ptr{ptr_}, num{num_} {}
/// Returns the data pointer by the span.
constexpr const T* data() const noexcept {
@@ -177,6 +185,7 @@ struct DeviceDispatch : public InstanceDispatch {
PFN_vkCmdBeginQuery vkCmdBeginQuery;
PFN_vkCmdBeginRenderPass vkCmdBeginRenderPass;
PFN_vkCmdBeginTransformFeedbackEXT vkCmdBeginTransformFeedbackEXT;
+ PFN_vkCmdBeginDebugUtilsLabelEXT vkCmdBeginDebugUtilsLabelEXT;
PFN_vkCmdBindDescriptorSets vkCmdBindDescriptorSets;
PFN_vkCmdBindIndexBuffer vkCmdBindIndexBuffer;
PFN_vkCmdBindPipeline vkCmdBindPipeline;
@@ -194,6 +203,7 @@ struct DeviceDispatch : public InstanceDispatch {
PFN_vkCmdEndQuery vkCmdEndQuery;
PFN_vkCmdEndRenderPass vkCmdEndRenderPass;
PFN_vkCmdEndTransformFeedbackEXT vkCmdEndTransformFeedbackEXT;
+ PFN_vkCmdEndDebugUtilsLabelEXT vkCmdEndDebugUtilsLabelEXT;
PFN_vkCmdFillBuffer vkCmdFillBuffer;
PFN_vkCmdPipelineBarrier vkCmdPipelineBarrier;
PFN_vkCmdPushConstants vkCmdPushConstants;
@@ -217,6 +227,7 @@ struct DeviceDispatch : public InstanceDispatch {
PFN_vkCmdSetPrimitiveTopologyEXT vkCmdSetPrimitiveTopologyEXT;
PFN_vkCmdSetStencilOpEXT vkCmdSetStencilOpEXT;
PFN_vkCmdSetStencilTestEnableEXT vkCmdSetStencilTestEnableEXT;
+ PFN_vkCmdResolveImage vkCmdResolveImage;
PFN_vkCreateBuffer vkCreateBuffer;
PFN_vkCreateBufferView vkCreateBufferView;
PFN_vkCreateCommandPool vkCreateCommandPool;
@@ -272,6 +283,8 @@ struct DeviceDispatch : public InstanceDispatch {
PFN_vkQueueSubmit vkQueueSubmit;
PFN_vkResetFences vkResetFences;
PFN_vkResetQueryPoolEXT vkResetQueryPoolEXT;
+ PFN_vkSetDebugUtilsObjectNameEXT vkSetDebugUtilsObjectNameEXT;
+ PFN_vkSetDebugUtilsObjectTagEXT vkSetDebugUtilsObjectTagEXT;
PFN_vkUnmapMemory vkUnmapMemory;
PFN_vkUpdateDescriptorSetWithTemplateKHR vkUpdateDescriptorSetWithTemplateKHR;
PFN_vkUpdateDescriptorSets vkUpdateDescriptorSets;
@@ -469,9 +482,10 @@ public:
PoolAllocations() = default;
/// Construct an allocation. Errors are reported through IsOutOfPoolMemory().
- explicit PoolAllocations(std::unique_ptr<AllocationType[]> allocations, std::size_t num,
- VkDevice device, PoolType pool, const DeviceDispatch& dld) noexcept
- : allocations{std::move(allocations)}, num{num}, device{device}, pool{pool}, dld{&dld} {}
+ explicit PoolAllocations(std::unique_ptr<AllocationType[]> allocations_, std::size_t num_,
+ VkDevice device_, PoolType pool_, const DeviceDispatch& dld_) noexcept
+ : allocations{std::move(allocations_)}, num{num_}, device{device_}, pool{pool_},
+ dld{&dld_} {}
/// Copying Vulkan allocations is not supported and will never be.
PoolAllocations(const PoolAllocations&) = delete;
@@ -541,18 +555,14 @@ private:
const DeviceDispatch* dld = nullptr;
};
-using BufferView = Handle<VkBufferView, VkDevice, DeviceDispatch>;
using DebugCallback = Handle<VkDebugUtilsMessengerEXT, VkInstance, InstanceDispatch>;
using DescriptorSetLayout = Handle<VkDescriptorSetLayout, VkDevice, DeviceDispatch>;
using DescriptorUpdateTemplateKHR = Handle<VkDescriptorUpdateTemplateKHR, VkDevice, DeviceDispatch>;
-using Framebuffer = Handle<VkFramebuffer, VkDevice, DeviceDispatch>;
-using ImageView = Handle<VkImageView, VkDevice, DeviceDispatch>;
using Pipeline = Handle<VkPipeline, VkDevice, DeviceDispatch>;
using PipelineLayout = Handle<VkPipelineLayout, VkDevice, DeviceDispatch>;
using QueryPool = Handle<VkQueryPool, VkDevice, DeviceDispatch>;
using RenderPass = Handle<VkRenderPass, VkDevice, DeviceDispatch>;
using Sampler = Handle<VkSampler, VkDevice, DeviceDispatch>;
-using ShaderModule = Handle<VkShaderModule, VkDevice, DeviceDispatch>;
using SurfaceKHR = Handle<VkSurfaceKHR, VkInstance, InstanceDispatch>;
using DescriptorSets = PoolAllocations<VkDescriptorSet, VkDescriptorPool>;
@@ -565,7 +575,7 @@ class Instance : public Handle<VkInstance, NoOwner, InstanceDispatch> {
public:
/// Creates a Vulkan instance. Use "operator bool" for error handling.
static Instance Create(u32 version, Span<const char*> layers, Span<const char*> extensions,
- InstanceDispatch& dld) noexcept;
+ InstanceDispatch& dispatch) noexcept;
/// Enumerates physical devices.
/// @return Physical devices and an empty handle on failure.
@@ -581,7 +591,8 @@ public:
constexpr Queue() noexcept = default;
/// Construct a queue handle.
- constexpr Queue(VkQueue queue, const DeviceDispatch& dld) noexcept : queue{queue}, dld{&dld} {}
+ constexpr Queue(VkQueue queue_, const DeviceDispatch& dld_) noexcept
+ : queue{queue_}, dld{&dld_} {}
VkResult Submit(Span<VkSubmitInfo> submit_infos,
VkFence fence = VK_NULL_HANDLE) const noexcept {
@@ -603,6 +614,17 @@ class Buffer : public Handle<VkBuffer, VkDevice, DeviceDispatch> {
public:
/// Attaches a memory allocation.
void BindMemory(VkDeviceMemory memory, VkDeviceSize offset) const;
+
+ /// Set object name.
+ void SetObjectNameEXT(const char* name) const;
+};
+
+class BufferView : public Handle<VkBufferView, VkDevice, DeviceDispatch> {
+ using Handle<VkBufferView, VkDevice, DeviceDispatch>::Handle;
+
+public:
+ /// Set object name.
+ void SetObjectNameEXT(const char* name) const;
};
class Image : public Handle<VkImage, VkDevice, DeviceDispatch> {
@@ -611,12 +633,26 @@ class Image : public Handle<VkImage, VkDevice, DeviceDispatch> {
public:
/// Attaches a memory allocation.
void BindMemory(VkDeviceMemory memory, VkDeviceSize offset) const;
+
+ /// Set object name.
+ void SetObjectNameEXT(const char* name) const;
+};
+
+class ImageView : public Handle<VkImageView, VkDevice, DeviceDispatch> {
+ using Handle<VkImageView, VkDevice, DeviceDispatch>::Handle;
+
+public:
+ /// Set object name.
+ void SetObjectNameEXT(const char* name) const;
};
class DeviceMemory : public Handle<VkDeviceMemory, VkDevice, DeviceDispatch> {
using Handle<VkDeviceMemory, VkDevice, DeviceDispatch>::Handle;
public:
+ /// Set object name.
+ void SetObjectNameEXT(const char* name) const;
+
u8* Map(VkDeviceSize offset, VkDeviceSize size) const {
void* data;
Check(dld->vkMapMemory(owner, handle, offset, size, 0, &data));
@@ -632,6 +668,9 @@ class Fence : public Handle<VkFence, VkDevice, DeviceDispatch> {
using Handle<VkFence, VkDevice, DeviceDispatch>::Handle;
public:
+ /// Set object name.
+ void SetObjectNameEXT(const char* name) const;
+
VkResult Wait(u64 timeout = std::numeric_limits<u64>::max()) const noexcept {
return dld->vkWaitForFences(owner, 1, &handle, true, timeout);
}
@@ -645,11 +684,22 @@ public:
}
};
+class Framebuffer : public Handle<VkFramebuffer, VkDevice, DeviceDispatch> {
+ using Handle<VkFramebuffer, VkDevice, DeviceDispatch>::Handle;
+
+public:
+ /// Set object name.
+ void SetObjectNameEXT(const char* name) const;
+};
+
class DescriptorPool : public Handle<VkDescriptorPool, VkDevice, DeviceDispatch> {
using Handle<VkDescriptorPool, VkDevice, DeviceDispatch>::Handle;
public:
DescriptorSets Allocate(const VkDescriptorSetAllocateInfo& ai) const;
+
+ /// Set object name.
+ void SetObjectNameEXT(const char* name) const;
};
class CommandPool : public Handle<VkCommandPool, VkDevice, DeviceDispatch> {
@@ -658,6 +708,9 @@ class CommandPool : public Handle<VkCommandPool, VkDevice, DeviceDispatch> {
public:
CommandBuffers Allocate(std::size_t num_buffers,
VkCommandBufferLevel level = VK_COMMAND_BUFFER_LEVEL_PRIMARY) const;
+
+ /// Set object name.
+ void SetObjectNameEXT(const char* name) const;
};
class SwapchainKHR : public Handle<VkSwapchainKHR, VkDevice, DeviceDispatch> {
@@ -671,15 +724,29 @@ class Event : public Handle<VkEvent, VkDevice, DeviceDispatch> {
using Handle<VkEvent, VkDevice, DeviceDispatch>::Handle;
public:
+ /// Set object name.
+ void SetObjectNameEXT(const char* name) const;
+
VkResult GetStatus() const noexcept {
return dld->vkGetEventStatus(owner, handle);
}
};
+class ShaderModule : public Handle<VkShaderModule, VkDevice, DeviceDispatch> {
+ using Handle<VkShaderModule, VkDevice, DeviceDispatch>::Handle;
+
+public:
+ /// Set object name.
+ void SetObjectNameEXT(const char* name) const;
+};
+
class Semaphore : public Handle<VkSemaphore, VkDevice, DeviceDispatch> {
using Handle<VkSemaphore, VkDevice, DeviceDispatch>::Handle;
public:
+ /// Set object name.
+ void SetObjectNameEXT(const char* name) const;
+
[[nodiscard]] u64 GetCounter() const {
u64 value;
Check(dld->vkGetSemaphoreCounterValueKHR(owner, handle, &value));
@@ -720,7 +787,7 @@ class Device : public Handle<VkDevice, NoOwner, DeviceDispatch> {
public:
static Device Create(VkPhysicalDevice physical_device, Span<VkDeviceQueueCreateInfo> queues_ci,
Span<const char*> enabled_extensions, const void* next,
- DeviceDispatch& dld) noexcept;
+ DeviceDispatch& dispatch) noexcept;
Queue GetQueue(u32 family_index) const noexcept;
@@ -809,8 +876,9 @@ class PhysicalDevice {
public:
constexpr PhysicalDevice() noexcept = default;
- constexpr PhysicalDevice(VkPhysicalDevice physical_device, const InstanceDispatch& dld) noexcept
- : physical_device{physical_device}, dld{&dld} {}
+ constexpr PhysicalDevice(VkPhysicalDevice physical_device_,
+ const InstanceDispatch& dld_) noexcept
+ : physical_device{physical_device_}, dld{&dld_} {}
constexpr operator VkPhysicalDevice() const noexcept {
return physical_device;
@@ -849,8 +917,8 @@ class CommandBuffer {
public:
CommandBuffer() noexcept = default;
- explicit CommandBuffer(VkCommandBuffer handle, const DeviceDispatch& dld) noexcept
- : handle{handle}, dld{&dld} {}
+ explicit CommandBuffer(VkCommandBuffer handle_, const DeviceDispatch& dld_) noexcept
+ : handle{handle_}, dld{&dld_} {}
const VkCommandBuffer* address() const noexcept {
return &handle;
@@ -929,6 +997,12 @@ public:
regions.data(), filter);
}
+ void ResolveImage(VkImage src_image, VkImageLayout src_layout, VkImage dst_image,
+ VkImageLayout dst_layout, Span<VkImageResolve> regions) {
+ dld->vkCmdResolveImage(handle, src_image, src_layout, dst_image, dst_layout, regions.size(),
+ regions.data());
+ }
+
void Dispatch(u32 x, u32 y, u32 z) const noexcept {
dld->vkCmdDispatch(handle, x, y, z);
}
@@ -943,6 +1017,23 @@ public:
image_barriers.size(), image_barriers.data());
}
+ void PipelineBarrier(VkPipelineStageFlags src_stage_mask, VkPipelineStageFlags dst_stage_mask,
+ VkDependencyFlags dependency_flags = 0) const noexcept {
+ PipelineBarrier(src_stage_mask, dst_stage_mask, dependency_flags, {}, {}, {});
+ }
+
+ void PipelineBarrier(VkPipelineStageFlags src_stage_mask, VkPipelineStageFlags dst_stage_mask,
+ VkDependencyFlags dependency_flags,
+ const VkBufferMemoryBarrier& buffer_barrier) const noexcept {
+ PipelineBarrier(src_stage_mask, dst_stage_mask, dependency_flags, {}, buffer_barrier, {});
+ }
+
+ void PipelineBarrier(VkPipelineStageFlags src_stage_mask, VkPipelineStageFlags dst_stage_mask,
+ VkDependencyFlags dependency_flags,
+ const VkImageMemoryBarrier& image_barrier) const noexcept {
+ PipelineBarrier(src_stage_mask, dst_stage_mask, dependency_flags, {}, {}, image_barrier);
+ }
+
void CopyBufferToImage(VkBuffer src_buffer, VkImage dst_image, VkImageLayout dst_image_layout,
Span<VkBufferImageCopy> regions) const noexcept {
dld->vkCmdCopyBufferToImage(handle, src_buffer, dst_image, dst_image_layout, regions.size(),
@@ -976,6 +1067,13 @@ public:
dld->vkCmdPushConstants(handle, layout, flags, offset, size, values);
}
+ template <typename T>
+ void PushConstants(VkPipelineLayout layout, VkShaderStageFlags flags,
+ const T& data) const noexcept {
+ static_assert(std::is_trivially_copyable_v<T>, "<data> is not trivially copyable");
+ dld->vkCmdPushConstants(handle, layout, flags, 0, static_cast<u32>(sizeof(T)), &data);
+ }
+
void SetViewport(u32 first, Span<VkViewport> viewports) const noexcept {
dld->vkCmdSetViewport(handle, first, viewports.size(), viewports.data());
}
@@ -1085,6 +1183,20 @@ public:
counter_buffers, counter_buffer_offsets);
}
+ void BeginDebugUtilsLabelEXT(const char* label, std::span<float, 4> color) const noexcept {
+ const VkDebugUtilsLabelEXT label_info{
+ .sType = VK_STRUCTURE_TYPE_DEBUG_UTILS_LABEL_EXT,
+ .pNext = nullptr,
+ .pLabelName = label,
+ .color{color[0], color[1], color[2], color[3]},
+ };
+ dld->vkCmdBeginDebugUtilsLabelEXT(handle, &label_info);
+ }
+
+ void EndDebugUtilsLabelEXT() const noexcept {
+ dld->vkCmdEndDebugUtilsLabelEXT(handle);
+ }
+
private:
VkCommandBuffer handle;
const DeviceDispatch* dld;
diff --git a/src/video_core/sampler_cache.cpp b/src/video_core/sampler_cache.cpp
deleted file mode 100644
index 53c7ef12d..000000000
--- a/src/video_core/sampler_cache.cpp
+++ /dev/null
@@ -1,21 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include "common/cityhash.h"
-#include "common/common_types.h"
-#include "video_core/sampler_cache.h"
-
-namespace VideoCommon {
-
-std::size_t SamplerCacheKey::Hash() const {
- static_assert(sizeof(raw) % sizeof(u64) == 0);
- return static_cast<std::size_t>(
- Common::CityHash64(reinterpret_cast<const char*>(raw.data()), sizeof(raw) / sizeof(u64)));
-}
-
-bool SamplerCacheKey::operator==(const SamplerCacheKey& rhs) const {
- return raw == rhs.raw;
-}
-
-} // namespace VideoCommon
diff --git a/src/video_core/sampler_cache.h b/src/video_core/sampler_cache.h
deleted file mode 100644
index cbe3ad071..000000000
--- a/src/video_core/sampler_cache.h
+++ /dev/null
@@ -1,60 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <cstddef>
-#include <unordered_map>
-
-#include "video_core/textures/texture.h"
-
-namespace VideoCommon {
-
-struct SamplerCacheKey final : public Tegra::Texture::TSCEntry {
- std::size_t Hash() const;
-
- bool operator==(const SamplerCacheKey& rhs) const;
-
- bool operator!=(const SamplerCacheKey& rhs) const {
- return !operator==(rhs);
- }
-};
-
-} // namespace VideoCommon
-
-namespace std {
-
-template <>
-struct hash<VideoCommon::SamplerCacheKey> {
- std::size_t operator()(const VideoCommon::SamplerCacheKey& k) const noexcept {
- return k.Hash();
- }
-};
-
-} // namespace std
-
-namespace VideoCommon {
-
-template <typename SamplerType, typename SamplerStorageType>
-class SamplerCache {
-public:
- SamplerType GetSampler(const Tegra::Texture::TSCEntry& tsc) {
- const auto [entry, is_cache_miss] = cache.try_emplace(SamplerCacheKey{tsc});
- auto& sampler = entry->second;
- if (is_cache_miss) {
- sampler = CreateSampler(tsc);
- }
- return ToSamplerType(sampler);
- }
-
-protected:
- virtual SamplerStorageType CreateSampler(const Tegra::Texture::TSCEntry& tsc) const = 0;
-
- virtual SamplerType ToSamplerType(const SamplerStorageType& sampler) const = 0;
-
-private:
- std::unordered_map<SamplerCacheKey, SamplerStorageType> cache;
-};
-
-} // namespace VideoCommon \ No newline at end of file
diff --git a/src/video_core/shader/ast.cpp b/src/video_core/shader/ast.cpp
index 3f96d9076..db11144c7 100644
--- a/src/video_core/shader/ast.cpp
+++ b/src/video_core/shader/ast.cpp
@@ -212,16 +212,15 @@ public:
}
void operator()(const ExprPredicate& expr) {
- inner += "P" + std::to_string(expr.predicate);
+ inner += fmt::format("P{}", expr.predicate);
}
void operator()(const ExprCondCode& expr) {
- u32 cc = static_cast<u32>(expr.cc);
- inner += "CC" + std::to_string(cc);
+ inner += fmt::format("CC{}", expr.cc);
}
void operator()(const ExprVar& expr) {
- inner += "V" + std::to_string(expr.var_index);
+ inner += fmt::format("V{}", expr.var_index);
}
void operator()(const ExprBoolean& expr) {
@@ -229,7 +228,7 @@ public:
}
void operator()(const ExprGprEqual& expr) {
- inner += "( gpr_" + std::to_string(expr.gpr) + " == " + std::to_string(expr.value) + ')';
+ inner += fmt::format("(gpr_{} == {})", expr.gpr, expr.value);
}
const std::string& GetResult() const {
@@ -374,8 +373,8 @@ std::string ASTManager::Print() const {
return printer.GetResult();
}
-ASTManager::ASTManager(bool full_decompile, bool disable_else_derivation)
- : full_decompile{full_decompile}, disable_else_derivation{disable_else_derivation} {};
+ASTManager::ASTManager(bool do_full_decompile, bool disable_else_derivation_)
+ : full_decompile{do_full_decompile}, disable_else_derivation{disable_else_derivation_} {}
ASTManager::~ASTManager() {
Clear();
diff --git a/src/video_core/shader/ast.h b/src/video_core/shader/ast.h
index 8e5a22ab3..dc49b369e 100644
--- a/src/video_core/shader/ast.h
+++ b/src/video_core/shader/ast.h
@@ -76,7 +76,7 @@ public:
class ASTIfThen {
public:
- explicit ASTIfThen(Expr condition) : condition{std::move(condition)} {}
+ explicit ASTIfThen(Expr condition_) : condition{std::move(condition_)} {}
Expr condition;
ASTZipper nodes{};
};
@@ -88,63 +88,68 @@ public:
class ASTBlockEncoded {
public:
- explicit ASTBlockEncoded(u32 start, u32 end) : start{start}, end{end} {}
+ explicit ASTBlockEncoded(u32 start_, u32 _) : start{start_}, end{_} {}
u32 start;
u32 end;
};
class ASTBlockDecoded {
public:
- explicit ASTBlockDecoded(NodeBlock&& new_nodes) : nodes(std::move(new_nodes)) {}
+ explicit ASTBlockDecoded(NodeBlock&& new_nodes_) : nodes(std::move(new_nodes_)) {}
NodeBlock nodes;
};
class ASTVarSet {
public:
- explicit ASTVarSet(u32 index, Expr condition) : index{index}, condition{std::move(condition)} {}
+ explicit ASTVarSet(u32 index_, Expr condition_)
+ : index{index_}, condition{std::move(condition_)} {}
+
u32 index;
Expr condition;
};
class ASTLabel {
public:
- explicit ASTLabel(u32 index) : index{index} {}
+ explicit ASTLabel(u32 index_) : index{index_} {}
u32 index;
bool unused{};
};
class ASTGoto {
public:
- explicit ASTGoto(Expr condition, u32 label) : condition{std::move(condition)}, label{label} {}
+ explicit ASTGoto(Expr condition_, u32 label_)
+ : condition{std::move(condition_)}, label{label_} {}
+
Expr condition;
u32 label;
};
class ASTDoWhile {
public:
- explicit ASTDoWhile(Expr condition) : condition{std::move(condition)} {}
+ explicit ASTDoWhile(Expr condition_) : condition{std::move(condition_)} {}
Expr condition;
ASTZipper nodes{};
};
class ASTReturn {
public:
- explicit ASTReturn(Expr condition, bool kills)
- : condition{std::move(condition)}, kills{kills} {}
+ explicit ASTReturn(Expr condition_, bool kills_)
+ : condition{std::move(condition_)}, kills{kills_} {}
+
Expr condition;
bool kills;
};
class ASTBreak {
public:
- explicit ASTBreak(Expr condition) : condition{std::move(condition)} {}
+ explicit ASTBreak(Expr condition_) : condition{std::move(condition_)} {}
Expr condition;
};
class ASTBase {
public:
- explicit ASTBase(ASTNode parent, ASTData data)
- : data{std::move(data)}, parent{std::move(parent)} {}
+ explicit ASTBase(ASTNode parent_, ASTData data_)
+ : data{std::move(data_)}, parent{std::move(parent_)} {}
template <class U, class... Args>
static ASTNode Make(ASTNode parent, Args&&... args) {
@@ -300,7 +305,7 @@ private:
class ASTManager final {
public:
- ASTManager(bool full_decompile, bool disable_else_derivation);
+ explicit ASTManager(bool do_full_decompile, bool disable_else_derivation_);
~ASTManager();
ASTManager(const ASTManager& o) = delete;
diff --git a/src/video_core/shader/async_shaders.cpp b/src/video_core/shader/async_shaders.cpp
index 6920afdf2..09f93463b 100644
--- a/src/video_core/shader/async_shaders.cpp
+++ b/src/video_core/shader/async_shaders.cpp
@@ -13,7 +13,7 @@
namespace VideoCommon::Shader {
-AsyncShaders::AsyncShaders(Core::Frontend::EmuWindow& emu_window) : emu_window(emu_window) {}
+AsyncShaders::AsyncShaders(Core::Frontend::EmuWindow& emu_window_) : emu_window(emu_window_) {}
AsyncShaders::~AsyncShaders() {
KillWorkers();
@@ -137,10 +137,9 @@ void AsyncShaders::QueueVulkanShader(Vulkan::VKPipelineCache* pp_cache,
const Vulkan::VKDevice& device, Vulkan::VKScheduler& scheduler,
Vulkan::VKDescriptorPool& descriptor_pool,
Vulkan::VKUpdateDescriptorQueue& update_descriptor_queue,
- Vulkan::VKRenderPassCache& renderpass_cache,
std::vector<VkDescriptorSetLayoutBinding> bindings,
Vulkan::SPIRVProgram program,
- Vulkan::GraphicsPipelineCacheKey key) {
+ Vulkan::GraphicsPipelineCacheKey key, u32 num_color_buffers) {
std::unique_lock lock(queue_mutex);
pending_queue.push({
.backend = Backend::Vulkan,
@@ -149,10 +148,10 @@ void AsyncShaders::QueueVulkanShader(Vulkan::VKPipelineCache* pp_cache,
.scheduler = &scheduler,
.descriptor_pool = &descriptor_pool,
.update_descriptor_queue = &update_descriptor_queue,
- .renderpass_cache = &renderpass_cache,
.bindings = std::move(bindings),
.program = std::move(program),
.key = key,
+ .num_color_buffers = num_color_buffers,
});
cv.notify_one();
}
@@ -205,8 +204,8 @@ void AsyncShaders::ShaderCompilerThread(Core::Frontend::GraphicsContext* context
} else if (work.backend == Backend::Vulkan) {
auto pipeline = std::make_unique<Vulkan::VKGraphicsPipeline>(
*work.vk_device, *work.scheduler, *work.descriptor_pool,
- *work.update_descriptor_queue, *work.renderpass_cache, work.key, work.bindings,
- work.program);
+ *work.update_descriptor_queue, work.key, work.bindings, work.program,
+ work.num_color_buffers);
work.pp_cache->EmplacePipeline(std::move(pipeline));
}
diff --git a/src/video_core/shader/async_shaders.h b/src/video_core/shader/async_shaders.h
index 7a99e1dc5..004e214a8 100644
--- a/src/video_core/shader/async_shaders.h
+++ b/src/video_core/shader/async_shaders.h
@@ -66,7 +66,7 @@ public:
Tegra::Engines::ShaderType shader_type;
};
- explicit AsyncShaders(Core::Frontend::EmuWindow& emu_window);
+ explicit AsyncShaders(Core::Frontend::EmuWindow& emu_window_);
~AsyncShaders();
/// Start up shader worker threads
@@ -98,9 +98,9 @@ public:
Vulkan::VKScheduler& scheduler,
Vulkan::VKDescriptorPool& descriptor_pool,
Vulkan::VKUpdateDescriptorQueue& update_descriptor_queue,
- Vulkan::VKRenderPassCache& renderpass_cache,
std::vector<VkDescriptorSetLayoutBinding> bindings,
- Vulkan::SPIRVProgram program, Vulkan::GraphicsPipelineCacheKey key);
+ Vulkan::SPIRVProgram program, Vulkan::GraphicsPipelineCacheKey key,
+ u32 num_color_buffers);
private:
void ShaderCompilerThread(Core::Frontend::GraphicsContext* context);
@@ -127,10 +127,10 @@ private:
Vulkan::VKScheduler* scheduler;
Vulkan::VKDescriptorPool* descriptor_pool;
Vulkan::VKUpdateDescriptorQueue* update_descriptor_queue;
- Vulkan::VKRenderPassCache* renderpass_cache;
std::vector<VkDescriptorSetLayoutBinding> bindings;
Vulkan::SPIRVProgram program;
Vulkan::GraphicsPipelineCacheKey key;
+ u32 num_color_buffers;
};
std::condition_variable cv;
diff --git a/src/video_core/shader/control_flow.cpp b/src/video_core/shader/control_flow.cpp
index 4c8971615..43d965f2f 100644
--- a/src/video_core/shader/control_flow.cpp
+++ b/src/video_core/shader/control_flow.cpp
@@ -66,8 +66,8 @@ struct BlockInfo {
};
struct CFGRebuildState {
- explicit CFGRebuildState(const ProgramCode& program_code, u32 start, Registry& registry)
- : program_code{program_code}, registry{registry}, start{start} {}
+ explicit CFGRebuildState(const ProgramCode& program_code_, u32 start_, Registry& registry_)
+ : program_code{program_code_}, registry{registry_}, start{start_} {}
const ProgramCode& program_code;
Registry& registry;
@@ -241,10 +241,10 @@ std::pair<ParseResult, ParseInfo> ParseCode(CFGRebuildState& state, u32 address)
ParseInfo parse_info{};
SingleBranch single_branch{};
- const auto insert_label = [](CFGRebuildState& state, u32 address) {
- const auto pair = state.labels.emplace(address);
+ const auto insert_label = [](CFGRebuildState& rebuild_state, u32 label_address) {
+ const auto pair = rebuild_state.labels.emplace(label_address);
if (pair.second) {
- state.inspect_queries.push_back(address);
+ rebuild_state.inspect_queries.push_back(label_address);
}
};
@@ -257,7 +257,7 @@ std::pair<ParseResult, ParseInfo> ParseCode(CFGRebuildState& state, u32 address)
single_branch.ignore = false;
break;
}
- if (state.registered.count(offset) != 0) {
+ if (state.registered.contains(offset)) {
single_branch.address = offset;
single_branch.ignore = true;
break;
@@ -632,12 +632,12 @@ void DecompileShader(CFGRebuildState& state) {
for (auto label : state.labels) {
state.manager->DeclareLabel(label);
}
- for (auto& block : state.block_info) {
- if (state.labels.count(block.start) != 0) {
+ for (const auto& block : state.block_info) {
+ if (state.labels.contains(block.start)) {
state.manager->InsertLabel(block.start);
}
const bool ignore = BlockBranchIsIgnored(block.branch);
- u32 end = ignore ? block.end + 1 : block.end;
+ const u32 end = ignore ? block.end + 1 : block.end;
state.manager->InsertBlock(block.start, end);
if (!ignore) {
InsertBranch(*state.manager, block.branch);
@@ -737,7 +737,7 @@ std::unique_ptr<ShaderCharacteristics> ScanFlow(const ProgramCode& program_code,
auto back = result_out->blocks.begin();
auto next = std::next(back);
while (next != result_out->blocks.end()) {
- if (state.labels.count(next->start) == 0 && next->start == back->end + 1) {
+ if (!state.labels.contains(next->start) && next->start == back->end + 1) {
back->end = next->end;
next = result_out->blocks.erase(next);
continue;
diff --git a/src/video_core/shader/control_flow.h b/src/video_core/shader/control_flow.h
index 62a3510d8..37bf96492 100644
--- a/src/video_core/shader/control_flow.h
+++ b/src/video_core/shader/control_flow.h
@@ -42,10 +42,10 @@ struct Condition {
class SingleBranch {
public:
SingleBranch() = default;
- SingleBranch(Condition condition, s32 address, bool kill, bool is_sync, bool is_brk,
- bool ignore)
- : condition{condition}, address{address}, kill{kill}, is_sync{is_sync}, is_brk{is_brk},
- ignore{ignore} {}
+ explicit SingleBranch(Condition condition_, s32 address_, bool kill_, bool is_sync_,
+ bool is_brk_, bool ignore_)
+ : condition{condition_}, address{address_}, kill{kill_}, is_sync{is_sync_}, is_brk{is_brk_},
+ ignore{ignore_} {}
bool operator==(const SingleBranch& b) const {
return std::tie(condition, address, kill, is_sync, is_brk, ignore) ==
@@ -65,15 +65,15 @@ public:
};
struct CaseBranch {
- CaseBranch(u32 cmp_value, u32 address) : cmp_value{cmp_value}, address{address} {}
+ explicit CaseBranch(u32 cmp_value_, u32 address_) : cmp_value{cmp_value_}, address{address_} {}
u32 cmp_value;
u32 address;
};
class MultiBranch {
public:
- MultiBranch(u32 gpr, std::vector<CaseBranch>&& branches)
- : gpr{gpr}, branches{std::move(branches)} {}
+ explicit MultiBranch(u32 gpr_, std::vector<CaseBranch>&& branches_)
+ : gpr{gpr_}, branches{std::move(branches_)} {}
u32 gpr{};
std::vector<CaseBranch> branches{};
diff --git a/src/video_core/shader/decode.cpp b/src/video_core/shader/decode.cpp
index eeac328a6..6576d1208 100644
--- a/src/video_core/shader/decode.cpp
+++ b/src/video_core/shader/decode.cpp
@@ -25,7 +25,7 @@ using Tegra::Shader::OpCode;
namespace {
void DeduceTextureHandlerSize(VideoCore::GuestDriverProfile& gpu_driver,
- const std::list<Sampler>& used_samplers) {
+ const std::list<SamplerEntry>& used_samplers) {
if (gpu_driver.IsTextureHandlerSizeKnown() || used_samplers.size() <= 1) {
return;
}
@@ -43,9 +43,9 @@ void DeduceTextureHandlerSize(VideoCore::GuestDriverProfile& gpu_driver,
}
}
-std::optional<u32> TryDeduceSamplerSize(const Sampler& sampler_to_deduce,
+std::optional<u32> TryDeduceSamplerSize(const SamplerEntry& sampler_to_deduce,
VideoCore::GuestDriverProfile& gpu_driver,
- const std::list<Sampler>& used_samplers) {
+ const std::list<SamplerEntry>& used_samplers) {
const u32 base_offset = sampler_to_deduce.offset;
u32 max_offset{std::numeric_limits<u32>::max()};
for (const auto& sampler : used_samplers) {
@@ -66,7 +66,7 @@ std::optional<u32> TryDeduceSamplerSize(const Sampler& sampler_to_deduce,
class ASTDecoder {
public:
- ASTDecoder(ShaderIR& ir) : ir(ir) {}
+ explicit ASTDecoder(ShaderIR& ir_) : ir(ir_) {}
void operator()(ASTProgram& ast) {
ASTNode current = ast.nodes.GetFirst();
@@ -153,8 +153,8 @@ void ShaderIR::Decode() {
const auto& blocks = shader_info.blocks;
NodeBlock current_block;
u32 current_label = static_cast<u32>(exit_branch);
- for (auto& block : blocks) {
- if (shader_info.labels.count(block.start) != 0) {
+ for (const auto& block : blocks) {
+ if (shader_info.labels.contains(block.start)) {
insert_block(current_block, current_label);
current_block.clear();
current_label = block.start;
diff --git a/src/video_core/shader/decode/arithmetic.cpp b/src/video_core/shader/decode/arithmetic.cpp
index afef5948d..15eb700e7 100644
--- a/src/video_core/shader/decode/arithmetic.cpp
+++ b/src/video_core/shader/decode/arithmetic.cpp
@@ -110,8 +110,7 @@ u32 ShaderIR::DecodeArithmetic(NodeBlock& bb, u32 pc) {
case SubOp::Sqrt:
return Operation(OperationCode::FSqrt, PRECISE, op_a);
default:
- UNIMPLEMENTED_MSG("Unhandled MUFU sub op={0:x}",
- static_cast<unsigned>(instr.sub_op.Value()));
+ UNIMPLEMENTED_MSG("Unhandled MUFU sub op={0:x}", instr.sub_op.Value());
return Immediate(0);
}
}();
diff --git a/src/video_core/shader/decode/arithmetic_integer.cpp b/src/video_core/shader/decode/arithmetic_integer.cpp
index 73155966f..7b5bb7003 100644
--- a/src/video_core/shader/decode/arithmetic_integer.cpp
+++ b/src/video_core/shader/decode/arithmetic_integer.cpp
@@ -83,7 +83,7 @@ u32 ShaderIR::DecodeArithmeticInteger(NodeBlock& bb, u32 pc) {
case IAdd3Height::UpperHalfWord:
return BitfieldExtract(value, 16, 16);
default:
- UNIMPLEMENTED_MSG("Unhandled IADD3 height: {}", static_cast<u32>(height));
+ UNIMPLEMENTED_MSG("Unhandled IADD3 height: {}", height);
return Immediate(0);
}
};
@@ -258,7 +258,7 @@ u32 ShaderIR::DecodeArithmeticInteger(NodeBlock& bb, u32 pc) {
case OpCode::Id::LEA_IMM:
case OpCode::Id::LEA_RZ:
case OpCode::Id::LEA_HI: {
- auto [op_a, op_b, op_c] = [&]() -> std::tuple<Node, Node, Node> {
+ auto [op_a_, op_b_, op_c_] = [&]() -> std::tuple<Node, Node, Node> {
switch (opcode->get().GetId()) {
case OpCode::Id::LEA_R2: {
return {GetRegister(instr.gpr20), GetRegister(instr.gpr39),
@@ -294,8 +294,9 @@ u32 ShaderIR::DecodeArithmeticInteger(NodeBlock& bb, u32 pc) {
UNIMPLEMENTED_IF_MSG(instr.lea.pred48 != static_cast<u64>(Pred::UnusedIndex),
"Unhandled LEA Predicate");
- Node value = Operation(OperationCode::ILogicalShiftLeft, std::move(op_a), std::move(op_c));
- value = Operation(OperationCode::IAdd, std::move(op_b), std::move(value));
+ Node value =
+ Operation(OperationCode::ILogicalShiftLeft, std::move(op_a_), std::move(op_c_));
+ value = Operation(OperationCode::IAdd, std::move(op_b_), std::move(value));
SetRegister(bb, instr.gpr0, std::move(value));
break;
diff --git a/src/video_core/shader/decode/arithmetic_integer_immediate.cpp b/src/video_core/shader/decode/arithmetic_integer_immediate.cpp
index 2a30aab2b..73580277a 100644
--- a/src/video_core/shader/decode/arithmetic_integer_immediate.cpp
+++ b/src/video_core/shader/decode/arithmetic_integer_immediate.cpp
@@ -72,7 +72,7 @@ void ShaderIR::WriteLogicOperation(NodeBlock& bb, Register dest, LogicOperation
case LogicOperation::PassB:
return op_b;
default:
- UNIMPLEMENTED_MSG("Unimplemented logic operation={}", static_cast<u32>(logic_op));
+ UNIMPLEMENTED_MSG("Unimplemented logic operation={}", logic_op);
return Immediate(0);
}
}();
@@ -92,8 +92,7 @@ void ShaderIR::WriteLogicOperation(NodeBlock& bb, Register dest, LogicOperation
break;
}
default:
- UNIMPLEMENTED_MSG("Unimplemented predicate result mode: {}",
- static_cast<u32>(predicate_mode));
+ UNIMPLEMENTED_MSG("Unimplemented predicate result mode: {}", predicate_mode);
}
}
diff --git a/src/video_core/shader/decode/conversion.cpp b/src/video_core/shader/decode/conversion.cpp
index b9989c88c..fea7a54df 100644
--- a/src/video_core/shader/decode/conversion.cpp
+++ b/src/video_core/shader/decode/conversion.cpp
@@ -244,7 +244,7 @@ u32 ShaderIR::DecodeConversion(NodeBlock& bb, u32 pc) {
return Operation(OperationCode::FTrunc, value);
default:
UNIMPLEMENTED_MSG("Unimplemented F2F rounding mode {}",
- static_cast<u32>(instr.conversion.f2f.rounding.Value()));
+ instr.conversion.f2f.rounding.Value());
return value;
}
}();
@@ -300,7 +300,7 @@ u32 ShaderIR::DecodeConversion(NodeBlock& bb, u32 pc) {
return Operation(OperationCode::FTrunc, PRECISE, value);
default:
UNIMPLEMENTED_MSG("Unimplemented F2I rounding mode {}",
- static_cast<u32>(instr.conversion.f2i.rounding.Value()));
+ instr.conversion.f2i.rounding.Value());
return Immediate(0);
}
}();
diff --git a/src/video_core/shader/decode/half_set.cpp b/src/video_core/shader/decode/half_set.cpp
index b2e88fa20..fa83108cd 100644
--- a/src/video_core/shader/decode/half_set.cpp
+++ b/src/video_core/shader/decode/half_set.cpp
@@ -22,13 +22,13 @@ u32 ShaderIR::DecodeHalfSet(NodeBlock& bb, u32 pc) {
const Instruction instr = {program_code[pc]};
const auto opcode = OpCode::Decode(instr);
- PredCondition cond;
- bool bf;
- bool ftz;
- bool neg_a;
- bool abs_a;
- bool neg_b;
- bool abs_b;
+ PredCondition cond{};
+ bool bf = false;
+ bool ftz = false;
+ bool neg_a = false;
+ bool abs_a = false;
+ bool neg_b = false;
+ bool abs_b = false;
switch (opcode->get().GetId()) {
case OpCode::Id::HSET2_C:
case OpCode::Id::HSET2_IMM:
diff --git a/src/video_core/shader/decode/image.cpp b/src/video_core/shader/decode/image.cpp
index 1ed4212ee..5470e8cf4 100644
--- a/src/video_core/shader/decode/image.cpp
+++ b/src/video_core/shader/decode/image.cpp
@@ -358,9 +358,9 @@ u32 ShaderIR::DecodeImage(NodeBlock& bb, u32 pc) {
instr.suldst.GetStoreDataLayout() != StoreType::Bits64);
auto descriptor = [this, instr] {
- std::optional<Tegra::Engines::SamplerDescriptor> descriptor;
+ std::optional<Tegra::Engines::SamplerDescriptor> sampler_descriptor;
if (instr.suldst.is_immediate) {
- descriptor =
+ sampler_descriptor =
registry.ObtainBoundSampler(static_cast<u32>(instr.image.index.Value()));
} else {
const Node image_register = GetRegister(instr.gpr39);
@@ -368,12 +368,12 @@ u32 ShaderIR::DecodeImage(NodeBlock& bb, u32 pc) {
static_cast<s64>(global_code.size()));
const auto buffer = std::get<1>(result);
const auto offset = std::get<2>(result);
- descriptor = registry.ObtainBindlessSampler(buffer, offset);
+ sampler_descriptor = registry.ObtainBindlessSampler(buffer, offset);
}
- if (!descriptor) {
+ if (!sampler_descriptor) {
UNREACHABLE_MSG("Failed to obtain image descriptor");
}
- return *descriptor;
+ return *sampler_descriptor;
}();
const auto comp_mask = GetImageComponentMask(descriptor.format);
@@ -497,11 +497,12 @@ u32 ShaderIR::DecodeImage(NodeBlock& bb, u32 pc) {
return pc;
}
-Image& ShaderIR::GetImage(Tegra::Shader::Image image, Tegra::Shader::ImageType type) {
+ImageEntry& ShaderIR::GetImage(Tegra::Shader::Image image, Tegra::Shader::ImageType type) {
const auto offset = static_cast<u32>(image.index.Value());
- const auto it = std::find_if(std::begin(used_images), std::end(used_images),
- [offset](const Image& entry) { return entry.offset == offset; });
+ const auto it =
+ std::find_if(std::begin(used_images), std::end(used_images),
+ [offset](const ImageEntry& entry) { return entry.offset == offset; });
if (it != std::end(used_images)) {
ASSERT(!it->is_bindless && it->type == type);
return *it;
@@ -511,7 +512,7 @@ Image& ShaderIR::GetImage(Tegra::Shader::Image image, Tegra::Shader::ImageType t
return used_images.emplace_back(next_index, offset, type);
}
-Image& ShaderIR::GetBindlessImage(Tegra::Shader::Register reg, Tegra::Shader::ImageType type) {
+ImageEntry& ShaderIR::GetBindlessImage(Tegra::Shader::Register reg, Tegra::Shader::ImageType type) {
const Node image_register = GetRegister(reg);
const auto result =
TrackCbuf(image_register, global_code, static_cast<s64>(global_code.size()));
@@ -520,7 +521,7 @@ Image& ShaderIR::GetBindlessImage(Tegra::Shader::Register reg, Tegra::Shader::Im
const auto offset = std::get<2>(result);
const auto it = std::find_if(std::begin(used_images), std::end(used_images),
- [buffer, offset](const Image& entry) {
+ [buffer, offset](const ImageEntry& entry) {
return entry.buffer == buffer && entry.offset == offset;
});
if (it != std::end(used_images)) {
diff --git a/src/video_core/shader/decode/memory.cpp b/src/video_core/shader/decode/memory.cpp
index e2bba88dd..50f4e7d35 100644
--- a/src/video_core/shader/decode/memory.cpp
+++ b/src/video_core/shader/decode/memory.cpp
@@ -47,7 +47,7 @@ OperationCode GetAtomOperation(AtomicOp op) {
case AtomicOp::Exch:
return OperationCode::AtomicIExchange;
default:
- UNIMPLEMENTED_MSG("op={}", static_cast<int>(op));
+ UNIMPLEMENTED_MSG("op={}", op);
return OperationCode::AtomicIAdd;
}
}
@@ -83,7 +83,7 @@ u32 GetMemorySize(Tegra::Shader::UniformType uniform_type) {
case Tegra::Shader::UniformType::UnsignedQuad:
return 128;
default:
- UNIMPLEMENTED_MSG("Unimplemented size={}!", static_cast<u32>(uniform_type));
+ UNIMPLEMENTED_MSG("Unimplemented size={}!", uniform_type);
return 32;
}
}
@@ -175,12 +175,12 @@ u32 ShaderIR::DecodeMemory(NodeBlock& bb, u32 pc) {
break;
}
default:
- UNIMPLEMENTED_MSG("Unhandled type: {}", static_cast<unsigned>(instr.ld_c.type.Value()));
+ UNIMPLEMENTED_MSG("Unhandled type: {}", instr.ld_c.type.Value());
}
break;
}
case OpCode::Id::LD_L:
- LOG_DEBUG(HW_GPU, "LD_L cache management mode: {}", static_cast<u64>(instr.ld_l.unknown));
+ LOG_DEBUG(HW_GPU, "LD_L cache management mode: {}", instr.ld_l.unknown);
[[fallthrough]];
case OpCode::Id::LD_S: {
const auto GetAddress = [&](s32 offset) {
@@ -224,7 +224,7 @@ u32 ShaderIR::DecodeMemory(NodeBlock& bb, u32 pc) {
}
default:
UNIMPLEMENTED_MSG("{} Unhandled type: {}", opcode->get().GetName(),
- static_cast<u32>(instr.ldst_sl.type.Value()));
+ instr.ldst_sl.type.Value());
}
break;
}
@@ -306,8 +306,7 @@ u32 ShaderIR::DecodeMemory(NodeBlock& bb, u32 pc) {
break;
}
case OpCode::Id::ST_L:
- LOG_DEBUG(HW_GPU, "ST_L cache management mode: {}",
- static_cast<u64>(instr.st_l.cache_management.Value()));
+ LOG_DEBUG(HW_GPU, "ST_L cache management mode: {}", instr.st_l.cache_management.Value());
[[fallthrough]];
case OpCode::Id::ST_S: {
const auto GetAddress = [&](s32 offset) {
@@ -340,7 +339,7 @@ u32 ShaderIR::DecodeMemory(NodeBlock& bb, u32 pc) {
}
default:
UNIMPLEMENTED_MSG("{} unhandled type: {}", opcode->get().GetName(),
- static_cast<u32>(instr.ldst_sl.type.Value()));
+ instr.ldst_sl.type.Value());
}
break;
}
@@ -387,7 +386,7 @@ u32 ShaderIR::DecodeMemory(NodeBlock& bb, u32 pc) {
}
case OpCode::Id::RED: {
UNIMPLEMENTED_IF_MSG(instr.red.type != GlobalAtomicType::U32, "type={}",
- static_cast<int>(instr.red.type.Value()));
+ instr.red.type.Value());
const auto [real_address, base_address, descriptor] =
TrackGlobalMemory(bb, instr, true, true);
if (!real_address || !base_address) {
@@ -403,12 +402,12 @@ u32 ShaderIR::DecodeMemory(NodeBlock& bb, u32 pc) {
UNIMPLEMENTED_IF_MSG(instr.atom.operation == AtomicOp::Inc ||
instr.atom.operation == AtomicOp::Dec ||
instr.atom.operation == AtomicOp::SafeAdd,
- "operation={}", static_cast<int>(instr.atom.operation.Value()));
+ "operation={}", instr.atom.operation.Value());
UNIMPLEMENTED_IF_MSG(instr.atom.type == GlobalAtomicType::S64 ||
instr.atom.type == GlobalAtomicType::U64 ||
instr.atom.type == GlobalAtomicType::F16x2_FTZ_RN ||
instr.atom.type == GlobalAtomicType::F32_FTZ_RN,
- "type={}", static_cast<int>(instr.atom.type.Value()));
+ "type={}", instr.atom.type.Value());
const auto [real_address, base_address, descriptor] =
TrackGlobalMemory(bb, instr, true, true);
@@ -428,10 +427,10 @@ u32 ShaderIR::DecodeMemory(NodeBlock& bb, u32 pc) {
case OpCode::Id::ATOMS: {
UNIMPLEMENTED_IF_MSG(instr.atoms.operation == AtomicOp::Inc ||
instr.atoms.operation == AtomicOp::Dec,
- "operation={}", static_cast<int>(instr.atoms.operation.Value()));
+ "operation={}", instr.atoms.operation.Value());
UNIMPLEMENTED_IF_MSG(instr.atoms.type == AtomicType::S64 ||
instr.atoms.type == AtomicType::U64,
- "type={}", static_cast<int>(instr.atoms.type.Value()));
+ "type={}", instr.atoms.type.Value());
const bool is_signed =
instr.atoms.type == AtomicType::S32 || instr.atoms.type == AtomicType::S64;
const s32 offset = instr.atoms.GetImmediateOffset();
diff --git a/src/video_core/shader/decode/other.cpp b/src/video_core/shader/decode/other.cpp
index 29a7cfbfe..d3ea07aac 100644
--- a/src/video_core/shader/decode/other.cpp
+++ b/src/video_core/shader/decode/other.cpp
@@ -34,14 +34,13 @@ u32 ShaderIR::DecodeOther(NodeBlock& bb, u32 pc) {
break;
}
case OpCode::Id::EXIT: {
- const Tegra::Shader::ConditionCode cc = instr.flow_condition_code;
- UNIMPLEMENTED_IF_MSG(cc != Tegra::Shader::ConditionCode::T, "EXIT condition code used: {}",
- static_cast<u32>(cc));
+ const ConditionCode cc = instr.flow_condition_code;
+ UNIMPLEMENTED_IF_MSG(cc != ConditionCode::T, "EXIT condition code used: {}", cc);
switch (instr.flow.cond) {
case Tegra::Shader::FlowCondition::Always:
bb.push_back(Operation(OperationCode::Exit));
- if (instr.pred.pred_index == static_cast<u64>(Tegra::Shader::Pred::UnusedIndex)) {
+ if (instr.pred.pred_index == static_cast<u64>(Pred::UnusedIndex)) {
// If this is an unconditional exit then just end processing here,
// otherwise we have to account for the possibility of the condition
// not being met, so continue processing the next instruction.
@@ -56,17 +55,15 @@ u32 ShaderIR::DecodeOther(NodeBlock& bb, u32 pc) {
break;
default:
- UNIMPLEMENTED_MSG("Unhandled flow condition: {}",
- static_cast<u32>(instr.flow.cond.Value()));
+ UNIMPLEMENTED_MSG("Unhandled flow condition: {}", instr.flow.cond.Value());
}
break;
}
case OpCode::Id::KIL: {
UNIMPLEMENTED_IF(instr.flow.cond != Tegra::Shader::FlowCondition::Always);
- const Tegra::Shader::ConditionCode cc = instr.flow_condition_code;
- UNIMPLEMENTED_IF_MSG(cc != Tegra::Shader::ConditionCode::T, "KIL condition code used: {}",
- static_cast<u32>(cc));
+ const ConditionCode cc = instr.flow_condition_code;
+ UNIMPLEMENTED_IF_MSG(cc != ConditionCode::T, "KIL condition code used: {}", cc);
bb.push_back(Operation(OperationCode::Discard));
break;
@@ -90,11 +87,11 @@ u32 ShaderIR::DecodeOther(NodeBlock& bb, u32 pc) {
UNIMPLEMENTED_MSG("S2R WscaleFactorZ is not implemented");
return Immediate(0U);
case SystemVariable::Tid: {
- Node value = Immediate(0);
- value = BitfieldInsert(value, Operation(OperationCode::LocalInvocationIdX), 0, 9);
- value = BitfieldInsert(value, Operation(OperationCode::LocalInvocationIdY), 16, 9);
- value = BitfieldInsert(value, Operation(OperationCode::LocalInvocationIdZ), 26, 5);
- return value;
+ Node val = Immediate(0);
+ val = BitfieldInsert(val, Operation(OperationCode::LocalInvocationIdX), 0, 9);
+ val = BitfieldInsert(val, Operation(OperationCode::LocalInvocationIdY), 16, 9);
+ val = BitfieldInsert(val, Operation(OperationCode::LocalInvocationIdZ), 26, 5);
+ return val;
}
case SystemVariable::TidX:
return Operation(OperationCode::LocalInvocationIdX);
@@ -130,8 +127,7 @@ u32 ShaderIR::DecodeOther(NodeBlock& bb, u32 pc) {
return Immediate(0u);
}
default:
- UNIMPLEMENTED_MSG("Unhandled system move: {}",
- static_cast<u32>(instr.sys20.Value()));
+ UNIMPLEMENTED_MSG("Unhandled system move: {}", instr.sys20.Value());
return Immediate(0u);
}
}();
@@ -181,8 +177,8 @@ u32 ShaderIR::DecodeOther(NodeBlock& bb, u32 pc) {
}
const Node branch = Operation(OperationCode::BranchIndirect, operand);
- const Tegra::Shader::ConditionCode cc = instr.flow_condition_code;
- if (cc != Tegra::Shader::ConditionCode::T) {
+ const ConditionCode cc = instr.flow_condition_code;
+ if (cc != ConditionCode::T) {
bb.push_back(Conditional(GetConditionCode(cc), {branch}));
} else {
bb.push_back(branch);
@@ -218,9 +214,8 @@ u32 ShaderIR::DecodeOther(NodeBlock& bb, u32 pc) {
break;
}
case OpCode::Id::SYNC: {
- const Tegra::Shader::ConditionCode cc = instr.flow_condition_code;
- UNIMPLEMENTED_IF_MSG(cc != Tegra::Shader::ConditionCode::T, "SYNC condition code used: {}",
- static_cast<u32>(cc));
+ const ConditionCode cc = instr.flow_condition_code;
+ UNIMPLEMENTED_IF_MSG(cc != ConditionCode::T, "SYNC condition code used: {}", cc);
if (decompiled) {
break;
@@ -231,9 +226,8 @@ u32 ShaderIR::DecodeOther(NodeBlock& bb, u32 pc) {
break;
}
case OpCode::Id::BRK: {
- const Tegra::Shader::ConditionCode cc = instr.flow_condition_code;
- UNIMPLEMENTED_IF_MSG(cc != Tegra::Shader::ConditionCode::T, "BRK condition code used: {}",
- static_cast<u32>(cc));
+ const ConditionCode cc = instr.flow_condition_code;
+ UNIMPLEMENTED_IF_MSG(cc != ConditionCode::T, "BRK condition code used: {}", cc);
if (decompiled) {
break;
}
@@ -306,7 +300,7 @@ u32 ShaderIR::DecodeOther(NodeBlock& bb, u32 pc) {
case Tegra::Shader::MembarType::GL:
return OperationCode::MemoryBarrierGlobal;
default:
- UNIMPLEMENTED_MSG("MEMBAR type={}", static_cast<int>(instr.membar.type.Value()));
+ UNIMPLEMENTED_MSG("MEMBAR type={}", instr.membar.type.Value());
return OperationCode::MemoryBarrierGlobal;
}
}();
diff --git a/src/video_core/shader/decode/shift.cpp b/src/video_core/shader/decode/shift.cpp
index d4ffa8014..a53819c15 100644
--- a/src/video_core/shader/decode/shift.cpp
+++ b/src/video_core/shader/decode/shift.cpp
@@ -125,7 +125,7 @@ u32 ShaderIR::DecodeShift(NodeBlock& bb, u32 pc) {
case OpCode::Id::SHF_LEFT_IMM: {
UNIMPLEMENTED_IF(instr.generates_cc);
UNIMPLEMENTED_IF_MSG(instr.shf.xmode != ShfXmode::None, "xmode={}",
- static_cast<int>(instr.shf.xmode.Value()));
+ instr.shf.xmode.Value());
if (instr.is_b_imm) {
op_b = Immediate(static_cast<u32>(instr.shf.immediate));
diff --git a/src/video_core/shader/decode/texture.cpp b/src/video_core/shader/decode/texture.cpp
index 02fdccd86..833fa2a39 100644
--- a/src/video_core/shader/decode/texture.cpp
+++ b/src/video_core/shader/decode/texture.cpp
@@ -34,7 +34,7 @@ static std::size_t GetCoordCount(TextureType texture_type) {
case TextureType::TextureCube:
return 3;
default:
- UNIMPLEMENTED_MSG("Unhandled texture type: {}", static_cast<u32>(texture_type));
+ UNIMPLEMENTED_MSG("Unhandled texture type: {}", texture_type);
return 0;
}
}
@@ -141,7 +141,7 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
SamplerInfo info;
info.is_shadow = is_depth_compare;
- const std::optional<Sampler> sampler = GetSampler(instr.sampler, info);
+ const std::optional<SamplerEntry> sampler = GetSampler(instr.sampler, info);
Node4 values;
for (u32 element = 0; element < values.size(); ++element) {
@@ -173,9 +173,9 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
SamplerInfo info;
info.type = texture_type;
info.is_array = is_array;
- const std::optional<Sampler> sampler = is_bindless
- ? GetBindlessSampler(base_reg, info, index_var)
- : GetSampler(instr.sampler, info);
+ const std::optional<SamplerEntry> sampler =
+ is_bindless ? GetBindlessSampler(base_reg, info, index_var)
+ : GetSampler(instr.sampler, info);
Node4 values;
if (!sampler) {
std::generate(values.begin(), values.end(), [this] { return Immediate(0); });
@@ -217,9 +217,9 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
[[fallthrough]];
case OpCode::Id::TXQ: {
Node index_var;
- const std::optional<Sampler> sampler = is_bindless
- ? GetBindlessSampler(instr.gpr8, {}, index_var)
- : GetSampler(instr.sampler, {});
+ const std::optional<SamplerEntry> sampler =
+ is_bindless ? GetBindlessSampler(instr.gpr8, {}, index_var)
+ : GetSampler(instr.sampler, {});
if (!sampler) {
u32 indexer = 0;
@@ -255,8 +255,7 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
break;
}
default:
- UNIMPLEMENTED_MSG("Unhandled texture query type: {}",
- static_cast<u32>(instr.txq.query_type.Value()));
+ UNIMPLEMENTED_MSG("Unhandled texture query type: {}", instr.txq.query_type.Value());
}
break;
}
@@ -273,7 +272,7 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
info.type = texture_type;
info.is_array = is_array;
Node index_var;
- const std::optional<Sampler> sampler =
+ const std::optional<SamplerEntry> sampler =
is_bindless ? GetBindlessSampler(instr.gpr20, info, index_var)
: GetSampler(instr.sampler, info);
@@ -302,7 +301,7 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
case TextureType::TextureCube:
return 3;
default:
- UNIMPLEMENTED_MSG("Unhandled texture type {}", static_cast<int>(texture_type));
+ UNIMPLEMENTED_MSG("Unhandled texture type {}", texture_type);
return 2;
}
}();
@@ -380,14 +379,15 @@ ShaderIR::SamplerInfo ShaderIR::GetSamplerInfo(
return info;
}
-std::optional<Sampler> ShaderIR::GetSampler(Tegra::Shader::Sampler sampler,
- SamplerInfo sampler_info) {
+std::optional<SamplerEntry> ShaderIR::GetSampler(Tegra::Shader::Sampler sampler,
+ SamplerInfo sampler_info) {
const u32 offset = static_cast<u32>(sampler.index.Value());
const auto info = GetSamplerInfo(sampler_info, registry.ObtainBoundSampler(offset));
// If this sampler has already been used, return the existing mapping.
- const auto it = std::find_if(used_samplers.begin(), used_samplers.end(),
- [offset](const Sampler& entry) { return entry.offset == offset; });
+ const auto it =
+ std::find_if(used_samplers.begin(), used_samplers.end(),
+ [offset](const SamplerEntry& entry) { return entry.offset == offset; });
if (it != used_samplers.end()) {
ASSERT(!it->is_bindless && it->type == info.type && it->is_array == info.is_array &&
it->is_shadow == info.is_shadow && it->is_buffer == info.is_buffer);
@@ -400,8 +400,8 @@ std::optional<Sampler> ShaderIR::GetSampler(Tegra::Shader::Sampler sampler,
*info.is_shadow, *info.is_buffer, false);
}
-std::optional<Sampler> ShaderIR::GetBindlessSampler(Tegra::Shader::Register reg, SamplerInfo info,
- Node& index_var) {
+std::optional<SamplerEntry> ShaderIR::GetBindlessSampler(Tegra::Shader::Register reg,
+ SamplerInfo info, Node& index_var) {
const Node sampler_register = GetRegister(reg);
const auto [base_node, tracked_sampler_info] =
TrackBindlessSampler(sampler_register, global_code, static_cast<s64>(global_code.size()));
@@ -417,7 +417,7 @@ std::optional<Sampler> ShaderIR::GetBindlessSampler(Tegra::Shader::Register reg,
// If this sampler has already been used, return the existing mapping.
const auto it = std::find_if(used_samplers.begin(), used_samplers.end(),
- [buffer, offset](const Sampler& entry) {
+ [buffer, offset](const SamplerEntry& entry) {
return entry.buffer == buffer && entry.offset == offset;
});
if (it != used_samplers.end()) {
@@ -437,11 +437,12 @@ std::optional<Sampler> ShaderIR::GetBindlessSampler(Tegra::Shader::Register reg,
info = GetSamplerInfo(info, registry.ObtainSeparateSampler(indices, offsets));
// Try to use an already created sampler if it exists
- const auto it = std::find_if(
- used_samplers.begin(), used_samplers.end(), [indices, offsets](const Sampler& entry) {
- return offsets == std::pair{entry.offset, entry.secondary_offset} &&
- indices == std::pair{entry.buffer, entry.secondary_buffer};
- });
+ const auto it =
+ std::find_if(used_samplers.begin(), used_samplers.end(),
+ [indices, offsets](const SamplerEntry& entry) {
+ return offsets == std::pair{entry.offset, entry.secondary_offset} &&
+ indices == std::pair{entry.buffer, entry.secondary_buffer};
+ });
if (it != used_samplers.end()) {
ASSERT(it->is_separated && it->type == info.type && it->is_array == info.is_array &&
it->is_shadow == info.is_shadow && it->is_buffer == info.is_buffer);
@@ -461,7 +462,7 @@ std::optional<Sampler> ShaderIR::GetBindlessSampler(Tegra::Shader::Register reg,
// If this sampler has already been used, return the existing mapping.
const auto it = std::find_if(
used_samplers.begin(), used_samplers.end(),
- [base_offset](const Sampler& entry) { return entry.offset == base_offset; });
+ [base_offset](const SamplerEntry& entry) { return entry.offset == base_offset; });
if (it != used_samplers.end()) {
ASSERT(!it->is_bindless && it->type == info.type && it->is_array == info.is_array &&
it->is_shadow == info.is_shadow && it->is_buffer == info.is_buffer &&
@@ -566,9 +567,9 @@ Node4 ShaderIR::GetTextureCode(Instruction instr, TextureType texture_type,
info.is_buffer = false;
Node index_var;
- const std::optional<Sampler> sampler = is_bindless
- ? GetBindlessSampler(*bindless_reg, info, index_var)
- : GetSampler(instr.sampler, info);
+ const std::optional<SamplerEntry> sampler =
+ is_bindless ? GetBindlessSampler(*bindless_reg, info, index_var)
+ : GetSampler(instr.sampler, info);
if (!sampler) {
return {Immediate(0), Immediate(0), Immediate(0), Immediate(0)};
}
@@ -595,7 +596,7 @@ Node4 ShaderIR::GetTextureCode(Instruction instr, TextureType texture_type,
lod = GetRegister(instr.gpr20.Value() + bias_offset);
break;
default:
- UNIMPLEMENTED_MSG("Unimplemented process mode={}", static_cast<u32>(process_mode));
+ UNIMPLEMENTED_MSG("Unimplemented process mode={}", process_mode);
break;
}
@@ -725,7 +726,7 @@ Node4 ShaderIR::GetTld4Code(Instruction instr, TextureType texture_type, bool de
info.is_shadow = depth_compare;
Node index_var;
- const std::optional<Sampler> sampler =
+ const std::optional<SamplerEntry> sampler =
is_bindless ? GetBindlessSampler(parameter_register++, info, index_var)
: GetSampler(instr.sampler, info);
Node4 values;
@@ -784,7 +785,7 @@ Node4 ShaderIR::GetTldCode(Tegra::Shader::Instruction instr) {
// const Node aoffi_register{is_aoffi ? GetRegister(gpr20_cursor++) : nullptr};
// const Node multisample{is_multisample ? GetRegister(gpr20_cursor++) : nullptr};
- const std::optional<Sampler> sampler = GetSampler(instr.sampler, {});
+ const std::optional<SamplerEntry> sampler = GetSampler(instr.sampler, {});
Node4 values;
for (u32 element = 0; element < values.size(); ++element) {
@@ -801,7 +802,7 @@ Node4 ShaderIR::GetTldsCode(Instruction instr, TextureType texture_type, bool is
info.type = texture_type;
info.is_array = is_array;
info.is_shadow = false;
- const std::optional<Sampler> sampler = GetSampler(instr.sampler, info);
+ const std::optional<SamplerEntry> sampler = GetSampler(instr.sampler, info);
const std::size_t type_coord_count = GetCoordCount(texture_type);
const bool lod_enabled = instr.tlds.GetTextureProcessMode() == TextureProcessMode::LL;
diff --git a/src/video_core/shader/decode/warp.cpp b/src/video_core/shader/decode/warp.cpp
index 11b77f795..37433d783 100644
--- a/src/video_core/shader/decode/warp.cpp
+++ b/src/video_core/shader/decode/warp.cpp
@@ -27,7 +27,7 @@ OperationCode GetOperationCode(VoteOperation vote_op) {
case VoteOperation::Eq:
return OperationCode::VoteEqual;
default:
- UNREACHABLE_MSG("Invalid vote operation={}", static_cast<u64>(vote_op));
+ UNREACHABLE_MSG("Invalid vote operation={}", vote_op);
return OperationCode::VoteAll;
}
}
diff --git a/src/video_core/shader/expr.h b/src/video_core/shader/expr.h
index 4e8264367..cda284c72 100644
--- a/src/video_core/shader/expr.h
+++ b/src/video_core/shader/expr.h
@@ -76,7 +76,7 @@ public:
class ExprPredicate final {
public:
- explicit ExprPredicate(u32 predicate) : predicate{predicate} {}
+ explicit ExprPredicate(u32 predicate_) : predicate{predicate_} {}
bool operator==(const ExprPredicate& b) const {
return predicate == b.predicate;
@@ -91,7 +91,7 @@ public:
class ExprCondCode final {
public:
- explicit ExprCondCode(ConditionCode cc) : cc{cc} {}
+ explicit ExprCondCode(ConditionCode condition_code) : cc{condition_code} {}
bool operator==(const ExprCondCode& b) const {
return cc == b.cc;
@@ -121,7 +121,7 @@ public:
class ExprGprEqual final {
public:
- ExprGprEqual(u32 gpr, u32 value) : gpr{gpr}, value{value} {}
+ explicit ExprGprEqual(u32 gpr_, u32 value_) : gpr{gpr_}, value{value_} {}
bool operator==(const ExprGprEqual& b) const {
return gpr == b.gpr && value == b.value;
diff --git a/src/video_core/shader/node.h b/src/video_core/shader/node.h
index a1e2c4d8e..b54d33763 100644
--- a/src/video_core/shader/node.h
+++ b/src/video_core/shader/node.h
@@ -282,26 +282,25 @@ struct SeparateSamplerNode;
using TrackSamplerData = std::variant<BindlessSamplerNode, SeparateSamplerNode, ArraySamplerNode>;
using TrackSampler = std::shared_ptr<TrackSamplerData>;
-struct Sampler {
+struct SamplerEntry {
/// Bound samplers constructor
- constexpr explicit Sampler(u32 index_, u32 offset_, Tegra::Shader::TextureType type_,
- bool is_array_, bool is_shadow_, bool is_buffer_, bool is_indexed_)
+ explicit SamplerEntry(u32 index_, u32 offset_, Tegra::Shader::TextureType type_, bool is_array_,
+ bool is_shadow_, bool is_buffer_, bool is_indexed_)
: index{index_}, offset{offset_}, type{type_}, is_array{is_array_}, is_shadow{is_shadow_},
is_buffer{is_buffer_}, is_indexed{is_indexed_} {}
/// Separate sampler constructor
- constexpr explicit Sampler(u32 index_, std::pair<u32, u32> offsets, std::pair<u32, u32> buffers,
- Tegra::Shader::TextureType type, bool is_array_, bool is_shadow_,
- bool is_buffer_)
+ explicit SamplerEntry(u32 index_, std::pair<u32, u32> offsets, std::pair<u32, u32> buffers,
+ Tegra::Shader::TextureType type_, bool is_array_, bool is_shadow_,
+ bool is_buffer_)
: index{index_}, offset{offsets.first}, secondary_offset{offsets.second},
- buffer{buffers.first}, secondary_buffer{buffers.second}, type{type}, is_array{is_array_},
+ buffer{buffers.first}, secondary_buffer{buffers.second}, type{type_}, is_array{is_array_},
is_shadow{is_shadow_}, is_buffer{is_buffer_}, is_separated{true} {}
/// Bindless samplers constructor
- constexpr explicit Sampler(u32 index_, u32 offset_, u32 buffer_,
- Tegra::Shader::TextureType type, bool is_array_, bool is_shadow_,
- bool is_buffer_, bool is_indexed_)
- : index{index_}, offset{offset_}, buffer{buffer_}, type{type}, is_array{is_array_},
+ explicit SamplerEntry(u32 index_, u32 offset_, u32 buffer_, Tegra::Shader::TextureType type_,
+ bool is_array_, bool is_shadow_, bool is_buffer_, bool is_indexed_)
+ : index{index_}, offset{offset_}, buffer{buffer_}, type{type_}, is_array{is_array_},
is_shadow{is_shadow_}, is_buffer{is_buffer_}, is_bindless{true}, is_indexed{is_indexed_} {
}
@@ -340,14 +339,14 @@ struct BindlessSamplerNode {
u32 offset;
};
-struct Image {
+struct ImageEntry {
public:
/// Bound images constructor
- constexpr explicit Image(u32 index_, u32 offset_, Tegra::Shader::ImageType type_)
+ explicit ImageEntry(u32 index_, u32 offset_, Tegra::Shader::ImageType type_)
: index{index_}, offset{offset_}, type{type_} {}
/// Bindless samplers constructor
- constexpr explicit Image(u32 index_, u32 offset_, u32 buffer_, Tegra::Shader::ImageType type_)
+ explicit ImageEntry(u32 index_, u32 offset_, u32 buffer_, Tegra::Shader::ImageType type_)
: index{index_}, offset{offset_}, buffer{buffer_}, type{type_}, is_bindless{true} {}
void MarkWrite() {
@@ -391,7 +390,7 @@ struct MetaArithmetic {
/// Parameters describing a texture sampler
struct MetaTexture {
- Sampler sampler;
+ SamplerEntry sampler;
Node array;
Node depth_compare;
std::vector<Node> aoffi;
@@ -405,7 +404,7 @@ struct MetaTexture {
};
struct MetaImage {
- const Image& image;
+ const ImageEntry& image;
std::vector<Node> values;
u32 element{};
};
diff --git a/src/video_core/shader/node_helper.cpp b/src/video_core/shader/node_helper.cpp
index 7bf4ff387..6a5b6940d 100644
--- a/src/video_core/shader/node_helper.cpp
+++ b/src/video_core/shader/node_helper.cpp
@@ -107,7 +107,7 @@ OperationCode SignedToUnsignedCode(OperationCode operation_code, bool is_signed)
UNREACHABLE_MSG("Can't apply absolute to an unsigned integer");
return {};
default:
- UNREACHABLE_MSG("Unknown signed operation with code={}", static_cast<u32>(operation_code));
+ UNREACHABLE_MSG("Unknown signed operation with code={}", operation_code);
return {};
}
}
diff --git a/src/video_core/shader/shader_ir.cpp b/src/video_core/shader/shader_ir.cpp
index 29d794b34..a4987ffc6 100644
--- a/src/video_core/shader/shader_ir.cpp
+++ b/src/video_core/shader/shader_ir.cpp
@@ -25,9 +25,10 @@ using Tegra::Shader::PredCondition;
using Tegra::Shader::PredOperation;
using Tegra::Shader::Register;
-ShaderIR::ShaderIR(const ProgramCode& program_code, u32 main_offset, CompilerSettings settings,
- Registry& registry)
- : program_code{program_code}, main_offset{main_offset}, settings{settings}, registry{registry} {
+ShaderIR::ShaderIR(const ProgramCode& program_code_, u32 main_offset_, CompilerSettings settings_,
+ Registry& registry_)
+ : program_code{program_code_}, main_offset{main_offset_}, settings{settings_}, registry{
+ registry_} {
Decode();
PostDecode();
}
@@ -170,7 +171,7 @@ Node ShaderIR::ConvertIntegerSize(Node value, Register::Size size, bool is_signe
// Default - do nothing
return value;
default:
- UNREACHABLE_MSG("Unimplemented conversion size: {}", static_cast<u32>(size));
+ UNREACHABLE_MSG("Unimplemented conversion size: {}", size);
return value;
}
}
@@ -335,15 +336,15 @@ OperationCode ShaderIR::GetPredicateCombiner(PredOperation operation) {
return operation_table[index];
}
-Node ShaderIR::GetConditionCode(Tegra::Shader::ConditionCode cc) const {
+Node ShaderIR::GetConditionCode(ConditionCode cc) const {
switch (cc) {
- case Tegra::Shader::ConditionCode::NEU:
+ case ConditionCode::NEU:
return GetInternalFlag(InternalFlag::Zero, true);
- case Tegra::Shader::ConditionCode::FCSM_TR:
+ case ConditionCode::FCSM_TR:
UNIMPLEMENTED_MSG("EXIT.FCSM_TR is not implemented");
return MakeNode<PredicateNode>(Pred::NeverExecute, false);
default:
- UNIMPLEMENTED_MSG("Unimplemented condition code: {}", static_cast<u32>(cc));
+ UNIMPLEMENTED_MSG("Unimplemented condition code: {}", cc);
return MakeNode<PredicateNode>(Pred::NeverExecute, false);
}
}
@@ -451,8 +452,8 @@ void ShaderIR::MarkAttributeUsage(Attribute::Index index, u64 element) {
}
std::size_t ShaderIR::DeclareAmend(Node new_amend) {
- const std::size_t id = amend_code.size();
- amend_code.push_back(new_amend);
+ const auto id = amend_code.size();
+ amend_code.push_back(std::move(new_amend));
return id;
}
diff --git a/src/video_core/shader/shader_ir.h b/src/video_core/shader/shader_ir.h
index 3a98b2104..0c6ab0f07 100644
--- a/src/video_core/shader/shader_ir.h
+++ b/src/video_core/shader/shader_ir.h
@@ -29,8 +29,8 @@ struct ShaderBlock;
constexpr u32 MAX_PROGRAM_LENGTH = 0x1000;
struct ConstBuffer {
- constexpr explicit ConstBuffer(u32 max_offset, bool is_indirect)
- : max_offset{max_offset}, is_indirect{is_indirect} {}
+ constexpr explicit ConstBuffer(u32 max_offset_, bool is_indirect_)
+ : max_offset{max_offset_}, is_indirect{is_indirect_} {}
constexpr ConstBuffer() = default;
@@ -66,8 +66,8 @@ struct GlobalMemoryUsage {
class ShaderIR final {
public:
- explicit ShaderIR(const ProgramCode& program_code, u32 main_offset, CompilerSettings settings,
- Registry& registry);
+ explicit ShaderIR(const ProgramCode& program_code_, u32 main_offset_,
+ CompilerSettings settings_, Registry& registry_);
~ShaderIR();
const std::map<u32, NodeBlock>& GetBasicBlocks() const {
@@ -94,11 +94,11 @@ public:
return used_cbufs;
}
- const std::list<Sampler>& GetSamplers() const {
+ const std::list<SamplerEntry>& GetSamplers() const {
return used_samplers;
}
- const std::list<Image>& GetImages() const {
+ const std::list<ImageEntry>& GetImages() const {
return used_images;
}
@@ -334,17 +334,17 @@ private:
std::optional<Tegra::Engines::SamplerDescriptor> sampler);
/// Accesses a texture sampler.
- std::optional<Sampler> GetSampler(Tegra::Shader::Sampler sampler, SamplerInfo info);
+ std::optional<SamplerEntry> GetSampler(Tegra::Shader::Sampler sampler, SamplerInfo info);
/// Accesses a texture sampler for a bindless texture.
- std::optional<Sampler> GetBindlessSampler(Tegra::Shader::Register reg, SamplerInfo info,
- Node& index_var);
+ std::optional<SamplerEntry> GetBindlessSampler(Tegra::Shader::Register reg, SamplerInfo info,
+ Node& index_var);
/// Accesses an image.
- Image& GetImage(Tegra::Shader::Image image, Tegra::Shader::ImageType type);
+ ImageEntry& GetImage(Tegra::Shader::Image image, Tegra::Shader::ImageType type);
/// Access a bindless image sampler.
- Image& GetBindlessImage(Tegra::Shader::Register reg, Tegra::Shader::ImageType type);
+ ImageEntry& GetBindlessImage(Tegra::Shader::Register reg, Tegra::Shader::ImageType type);
/// Extracts a sequence of bits from a node
Node BitfieldExtract(Node value, u32 offset, u32 bits);
@@ -454,8 +454,8 @@ private:
std::set<Tegra::Shader::Attribute::Index> used_input_attributes;
std::set<Tegra::Shader::Attribute::Index> used_output_attributes;
std::map<u32, ConstBuffer> used_cbufs;
- std::list<Sampler> used_samplers;
- std::list<Image> used_images;
+ std::list<SamplerEntry> used_samplers;
+ std::list<ImageEntry> used_images;
std::array<bool, Tegra::Engines::Maxwell3D::Regs::NumClipDistances> used_clip_distances{};
std::map<GlobalMemoryBase, GlobalMemoryUsage> used_global_memory;
bool uses_layer{};
diff --git a/src/video_core/surface.cpp b/src/video_core/surface.cpp
index 1688267bb..6308aef94 100644
--- a/src/video_core/surface.cpp
+++ b/src/video_core/surface.cpp
@@ -28,7 +28,7 @@ SurfaceTarget SurfaceTargetFromTextureType(Tegra::Texture::TextureType texture_t
case Tegra::Texture::TextureType::Texture2DArray:
return SurfaceTarget::Texture2DArray;
default:
- LOG_CRITICAL(HW_GPU, "Unimplemented texture_type={}", static_cast<u32>(texture_type));
+ LOG_CRITICAL(HW_GPU, "Unimplemented texture_type={}", texture_type);
UNREACHABLE();
return SurfaceTarget::Texture2D;
}
@@ -47,7 +47,7 @@ bool SurfaceTargetIsLayered(SurfaceTarget target) {
case SurfaceTarget::TextureCubeArray:
return true;
default:
- LOG_CRITICAL(HW_GPU, "Unimplemented surface_target={}", static_cast<u32>(target));
+ LOG_CRITICAL(HW_GPU, "Unimplemented surface_target={}", target);
UNREACHABLE();
return false;
}
@@ -66,7 +66,7 @@ bool SurfaceTargetIsArray(SurfaceTarget target) {
case SurfaceTarget::TextureCubeArray:
return true;
default:
- LOG_CRITICAL(HW_GPU, "Unimplemented surface_target={}", static_cast<u32>(target));
+ LOG_CRITICAL(HW_GPU, "Unimplemented surface_target={}", target);
UNREACHABLE();
return false;
}
@@ -85,7 +85,7 @@ PixelFormat PixelFormatFromDepthFormat(Tegra::DepthFormat format) {
case Tegra::DepthFormat::D32_FLOAT_S8X24_UINT:
return PixelFormat::D32_FLOAT_S8_UINT;
default:
- UNIMPLEMENTED_MSG("Unimplemented format={}", static_cast<u32>(format));
+ UNIMPLEMENTED_MSG("Unimplemented format={}", format);
return PixelFormat::S8_UINT_D24_UNORM;
}
}
@@ -183,7 +183,7 @@ PixelFormat PixelFormatFromRenderTargetFormat(Tegra::RenderTargetFormat format)
case Tegra::RenderTargetFormat::R8_UINT:
return PixelFormat::R8_UINT;
default:
- UNIMPLEMENTED_MSG("Unimplemented format={}", static_cast<int>(format));
+ UNIMPLEMENTED_MSG("Unimplemented format={}", format);
return PixelFormat::A8B8G8R8_UNORM;
}
}
@@ -197,7 +197,7 @@ PixelFormat PixelFormatFromGPUPixelFormat(Tegra::FramebufferConfig::PixelFormat
case Tegra::FramebufferConfig::PixelFormat::B8G8R8A8_UNORM:
return PixelFormat::B8G8R8A8_UNORM;
default:
- UNIMPLEMENTED_MSG("Unimplemented format={}", static_cast<u32>(format));
+ UNIMPLEMENTED_MSG("Unimplemented format={}", format);
return PixelFormat::A8B8G8R8_UNORM;
}
}
@@ -280,7 +280,7 @@ bool IsPixelFormatSRGB(PixelFormat format) {
}
std::pair<u32, u32> GetASTCBlockSize(PixelFormat format) {
- return {GetDefaultBlockWidth(format), GetDefaultBlockHeight(format)};
+ return {DefaultBlockWidth(format), DefaultBlockHeight(format)};
}
} // namespace VideoCore::Surface
diff --git a/src/video_core/surface.h b/src/video_core/surface.h
index cfd12fa61..c40ab89d0 100644
--- a/src/video_core/surface.h
+++ b/src/video_core/surface.h
@@ -120,7 +120,7 @@ enum class PixelFormat {
Max = MaxDepthStencilFormat,
Invalid = 255,
};
-static constexpr std::size_t MaxPixelFormat = static_cast<std::size_t>(PixelFormat::Max);
+constexpr std::size_t MaxPixelFormat = static_cast<std::size_t>(PixelFormat::Max);
enum class SurfaceType {
ColorTexture = 0,
@@ -140,117 +140,7 @@ enum class SurfaceTarget {
TextureCubeArray,
};
-constexpr std::array<u32, MaxPixelFormat> compression_factor_shift_table = {{
- 0, // A8B8G8R8_UNORM
- 0, // A8B8G8R8_SNORM
- 0, // A8B8G8R8_SINT
- 0, // A8B8G8R8_UINT
- 0, // R5G6B5_UNORM
- 0, // B5G6R5_UNORM
- 0, // A1R5G5B5_UNORM
- 0, // A2B10G10R10_UNORM
- 0, // A2B10G10R10_UINT
- 0, // A1B5G5R5_UNORM
- 0, // R8_UNORM
- 0, // R8_SNORM
- 0, // R8_SINT
- 0, // R8_UINT
- 0, // R16G16B16A16_FLOAT
- 0, // R16G16B16A16_UNORM
- 0, // R16G16B16A16_SNORM
- 0, // R16G16B16A16_SINT
- 0, // R16G16B16A16_UINT
- 0, // B10G11R11_FLOAT
- 0, // R32G32B32A32_UINT
- 2, // BC1_RGBA_UNORM
- 2, // BC2_UNORM
- 2, // BC3_UNORM
- 2, // BC4_UNORM
- 2, // BC4_SNORM
- 2, // BC5_UNORM
- 2, // BC5_SNORM
- 2, // BC7_UNORM
- 2, // BC6H_UFLOAT
- 2, // BC6H_SFLOAT
- 2, // ASTC_2D_4X4_UNORM
- 0, // B8G8R8A8_UNORM
- 0, // R32G32B32A32_FLOAT
- 0, // R32G32B32A32_SINT
- 0, // R32G32_FLOAT
- 0, // R32G32_SINT
- 0, // R32_FLOAT
- 0, // R16_FLOAT
- 0, // R16_UNORM
- 0, // R16_SNORM
- 0, // R16_UINT
- 0, // R16_SINT
- 0, // R16G16_UNORM
- 0, // R16G16_FLOAT
- 0, // R16G16_UINT
- 0, // R16G16_SINT
- 0, // R16G16_SNORM
- 0, // R32G32B32_FLOAT
- 0, // A8B8G8R8_SRGB
- 0, // R8G8_UNORM
- 0, // R8G8_SNORM
- 0, // R8G8_SINT
- 0, // R8G8_UINT
- 0, // R32G32_UINT
- 0, // R16G16B16X16_FLOAT
- 0, // R32_UINT
- 0, // R32_SINT
- 2, // ASTC_2D_8X8_UNORM
- 2, // ASTC_2D_8X5_UNORM
- 2, // ASTC_2D_5X4_UNORM
- 0, // B8G8R8A8_SRGB
- 2, // BC1_RGBA_SRGB
- 2, // BC2_SRGB
- 2, // BC3_SRGB
- 2, // BC7_SRGB
- 0, // A4B4G4R4_UNORM
- 2, // ASTC_2D_4X4_SRGB
- 2, // ASTC_2D_8X8_SRGB
- 2, // ASTC_2D_8X5_SRGB
- 2, // ASTC_2D_5X4_SRGB
- 2, // ASTC_2D_5X5_UNORM
- 2, // ASTC_2D_5X5_SRGB
- 2, // ASTC_2D_10X8_UNORM
- 2, // ASTC_2D_10X8_SRGB
- 2, // ASTC_2D_6X6_UNORM
- 2, // ASTC_2D_6X6_SRGB
- 2, // ASTC_2D_10X10_UNORM
- 2, // ASTC_2D_10X10_SRGB
- 2, // ASTC_2D_12X12_UNORM
- 2, // ASTC_2D_12X12_SRGB
- 2, // ASTC_2D_8X6_UNORM
- 2, // ASTC_2D_8X6_SRGB
- 2, // ASTC_2D_6X5_UNORM
- 2, // ASTC_2D_6X5_SRGB
- 0, // E5B9G9R9_FLOAT
- 0, // D32_FLOAT
- 0, // D16_UNORM
- 0, // D24_UNORM_S8_UINT
- 0, // S8_UINT_D24_UNORM
- 0, // D32_FLOAT_S8_UINT
-}};
-
-/**
- * Gets the compression factor for the specified PixelFormat. This applies to just the
- * "compressed width" and "compressed height", not the overall compression factor of a
- * compressed image. This is used for maintaining proper surface sizes for compressed
- * texture formats.
- */
-inline constexpr u32 GetCompressionFactorShift(PixelFormat format) {
- DEBUG_ASSERT(format != PixelFormat::Invalid);
- DEBUG_ASSERT(static_cast<std::size_t>(format) < compression_factor_shift_table.size());
- return compression_factor_shift_table[static_cast<std::size_t>(format)];
-}
-
-inline constexpr u32 GetCompressionFactor(PixelFormat format) {
- return 1U << GetCompressionFactorShift(format);
-}
-
-constexpr std::array<u32, MaxPixelFormat> block_width_table = {{
+constexpr std::array<u32, MaxPixelFormat> BLOCK_WIDTH_TABLE = {{
1, // A8B8G8R8_UNORM
1, // A8B8G8R8_SNORM
1, // A8B8G8R8_SINT
@@ -344,15 +234,12 @@ constexpr std::array<u32, MaxPixelFormat> block_width_table = {{
1, // D32_FLOAT_S8_UINT
}};
-static constexpr u32 GetDefaultBlockWidth(PixelFormat format) {
- if (format == PixelFormat::Invalid)
- return 0;
-
- ASSERT(static_cast<std::size_t>(format) < block_width_table.size());
- return block_width_table[static_cast<std::size_t>(format)];
+constexpr u32 DefaultBlockWidth(PixelFormat format) {
+ ASSERT(static_cast<std::size_t>(format) < BLOCK_WIDTH_TABLE.size());
+ return BLOCK_WIDTH_TABLE[static_cast<std::size_t>(format)];
}
-constexpr std::array<u32, MaxPixelFormat> block_height_table = {{
+constexpr std::array<u32, MaxPixelFormat> BLOCK_HEIGHT_TABLE = {{
1, // A8B8G8R8_UNORM
1, // A8B8G8R8_SNORM
1, // A8B8G8R8_SINT
@@ -446,15 +333,12 @@ constexpr std::array<u32, MaxPixelFormat> block_height_table = {{
1, // D32_FLOAT_S8_UINT
}};
-static constexpr u32 GetDefaultBlockHeight(PixelFormat format) {
- if (format == PixelFormat::Invalid)
- return 0;
-
- ASSERT(static_cast<std::size_t>(format) < block_height_table.size());
- return block_height_table[static_cast<std::size_t>(format)];
+constexpr u32 DefaultBlockHeight(PixelFormat format) {
+ ASSERT(static_cast<std::size_t>(format) < BLOCK_HEIGHT_TABLE.size());
+ return BLOCK_HEIGHT_TABLE[static_cast<std::size_t>(format)];
}
-constexpr std::array<u32, MaxPixelFormat> bpp_table = {{
+constexpr std::array<u32, MaxPixelFormat> BITS_PER_BLOCK_TABLE = {{
32, // A8B8G8R8_UNORM
32, // A8B8G8R8_SNORM
32, // A8B8G8R8_SINT
@@ -548,20 +432,14 @@ constexpr std::array<u32, MaxPixelFormat> bpp_table = {{
64, // D32_FLOAT_S8_UINT
}};
-static constexpr u32 GetFormatBpp(PixelFormat format) {
- if (format == PixelFormat::Invalid)
- return 0;
-
- ASSERT(static_cast<std::size_t>(format) < bpp_table.size());
- return bpp_table[static_cast<std::size_t>(format)];
+constexpr u32 BitsPerBlock(PixelFormat format) {
+ ASSERT(static_cast<std::size_t>(format) < BITS_PER_BLOCK_TABLE.size());
+ return BITS_PER_BLOCK_TABLE[static_cast<std::size_t>(format)];
}
/// Returns the sizer in bytes of the specified pixel format
-static constexpr u32 GetBytesPerPixel(PixelFormat pixel_format) {
- if (pixel_format == PixelFormat::Invalid) {
- return 0;
- }
- return GetFormatBpp(pixel_format) / CHAR_BIT;
+constexpr u32 BytesPerBlock(PixelFormat pixel_format) {
+ return BitsPerBlock(pixel_format) / CHAR_BIT;
}
SurfaceTarget SurfaceTargetFromTextureType(Tegra::Texture::TextureType texture_type);
diff --git a/src/video_core/texture_cache/accelerated_swizzle.cpp b/src/video_core/texture_cache/accelerated_swizzle.cpp
new file mode 100644
index 000000000..a4fc1184b
--- /dev/null
+++ b/src/video_core/texture_cache/accelerated_swizzle.cpp
@@ -0,0 +1,70 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <array>
+#include <bit>
+
+#include "common/alignment.h"
+#include "common/common_types.h"
+#include "common/div_ceil.h"
+#include "video_core/surface.h"
+#include "video_core/texture_cache/accelerated_swizzle.h"
+#include "video_core/texture_cache/util.h"
+#include "video_core/textures/decoders.h"
+
+namespace VideoCommon::Accelerated {
+
+using Tegra::Texture::GOB_SIZE_SHIFT;
+using Tegra::Texture::GOB_SIZE_X;
+using Tegra::Texture::GOB_SIZE_X_SHIFT;
+using Tegra::Texture::GOB_SIZE_Y_SHIFT;
+using VideoCore::Surface::BytesPerBlock;
+
+BlockLinearSwizzle2DParams MakeBlockLinearSwizzle2DParams(const SwizzleParameters& swizzle,
+ const ImageInfo& info) {
+ const Extent3D block = swizzle.block;
+ const Extent3D num_tiles = swizzle.num_tiles;
+ const u32 bytes_per_block = BytesPerBlock(info.format);
+ const u32 stride_alignment = CalculateLevelStrideAlignment(info, swizzle.level);
+ const u32 stride = Common::AlignBits(num_tiles.width, stride_alignment) * bytes_per_block;
+ const u32 gobs_in_x = Common::DivCeilLog2(stride, GOB_SIZE_X_SHIFT);
+ return BlockLinearSwizzle2DParams{
+ .origin{0, 0, 0},
+ .destination{0, 0, 0},
+ .bytes_per_block_log2 = static_cast<u32>(std::countr_zero(bytes_per_block)),
+ .layer_stride = info.layer_stride,
+ .block_size = gobs_in_x << (GOB_SIZE_SHIFT + block.height + block.depth),
+ .x_shift = GOB_SIZE_SHIFT + block.height + block.depth,
+ .block_height = block.height,
+ .block_height_mask = (1U << block.height) - 1,
+ };
+}
+
+BlockLinearSwizzle3DParams MakeBlockLinearSwizzle3DParams(const SwizzleParameters& swizzle,
+ const ImageInfo& info) {
+ const Extent3D block = swizzle.block;
+ const Extent3D num_tiles = swizzle.num_tiles;
+ const u32 bytes_per_block = BytesPerBlock(info.format);
+ const u32 stride_alignment = CalculateLevelStrideAlignment(info, swizzle.level);
+ const u32 stride = Common::AlignBits(num_tiles.width, stride_alignment) * bytes_per_block;
+
+ const u32 gobs_in_x = (stride + GOB_SIZE_X - 1) >> GOB_SIZE_X_SHIFT;
+ const u32 block_size = gobs_in_x << (GOB_SIZE_SHIFT + block.height + block.depth);
+ const u32 slice_size =
+ Common::DivCeilLog2(num_tiles.height, block.height + GOB_SIZE_Y_SHIFT) * block_size;
+ return BlockLinearSwizzle3DParams{
+ .origin{0, 0, 0},
+ .destination{0, 0, 0},
+ .bytes_per_block_log2 = static_cast<u32>(std::countr_zero(bytes_per_block)),
+ .slice_size = slice_size,
+ .block_size = block_size,
+ .x_shift = GOB_SIZE_SHIFT + block.height + block.depth,
+ .block_height = block.height,
+ .block_height_mask = (1U << block.height) - 1,
+ .block_depth = block.depth,
+ .block_depth_mask = (1U << block.depth) - 1,
+ };
+}
+
+} // namespace VideoCommon::Accelerated \ No newline at end of file
diff --git a/src/video_core/texture_cache/accelerated_swizzle.h b/src/video_core/texture_cache/accelerated_swizzle.h
new file mode 100644
index 000000000..6ec5c78c4
--- /dev/null
+++ b/src/video_core/texture_cache/accelerated_swizzle.h
@@ -0,0 +1,45 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+
+#include "common/common_types.h"
+#include "video_core/texture_cache/image_info.h"
+#include "video_core/texture_cache/types.h"
+
+namespace VideoCommon::Accelerated {
+
+struct BlockLinearSwizzle2DParams {
+ std::array<u32, 3> origin;
+ std::array<s32, 3> destination;
+ u32 bytes_per_block_log2;
+ u32 layer_stride;
+ u32 block_size;
+ u32 x_shift;
+ u32 block_height;
+ u32 block_height_mask;
+};
+
+struct BlockLinearSwizzle3DParams {
+ std::array<u32, 3> origin;
+ std::array<s32, 3> destination;
+ u32 bytes_per_block_log2;
+ u32 slice_size;
+ u32 block_size;
+ u32 x_shift;
+ u32 block_height;
+ u32 block_height_mask;
+ u32 block_depth;
+ u32 block_depth_mask;
+};
+
+[[nodiscard]] BlockLinearSwizzle2DParams MakeBlockLinearSwizzle2DParams(
+ const SwizzleParameters& swizzle, const ImageInfo& info);
+
+[[nodiscard]] BlockLinearSwizzle3DParams MakeBlockLinearSwizzle3DParams(
+ const SwizzleParameters& swizzle, const ImageInfo& info);
+
+} // namespace VideoCommon::Accelerated
diff --git a/src/video_core/texture_cache/copy_params.h b/src/video_core/texture_cache/copy_params.h
deleted file mode 100644
index 9c21a0649..000000000
--- a/src/video_core/texture_cache/copy_params.h
+++ /dev/null
@@ -1,36 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include "common/common_types.h"
-
-namespace VideoCommon {
-
-struct CopyParams {
- constexpr CopyParams(u32 source_x, u32 source_y, u32 source_z, u32 dest_x, u32 dest_y,
- u32 dest_z, u32 source_level, u32 dest_level, u32 width, u32 height,
- u32 depth)
- : source_x{source_x}, source_y{source_y}, source_z{source_z}, dest_x{dest_x},
- dest_y{dest_y}, dest_z{dest_z}, source_level{source_level},
- dest_level{dest_level}, width{width}, height{height}, depth{depth} {}
-
- constexpr CopyParams(u32 width, u32 height, u32 depth, u32 level)
- : source_x{}, source_y{}, source_z{}, dest_x{}, dest_y{}, dest_z{}, source_level{level},
- dest_level{level}, width{width}, height{height}, depth{depth} {}
-
- u32 source_x;
- u32 source_y;
- u32 source_z;
- u32 dest_x;
- u32 dest_y;
- u32 dest_z;
- u32 source_level;
- u32 dest_level;
- u32 width;
- u32 height;
- u32 depth;
-};
-
-} // namespace VideoCommon
diff --git a/src/video_core/texture_cache/decode_bc4.cpp b/src/video_core/texture_cache/decode_bc4.cpp
new file mode 100644
index 000000000..017327975
--- /dev/null
+++ b/src/video_core/texture_cache/decode_bc4.cpp
@@ -0,0 +1,97 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <algorithm>
+#include <array>
+#include <span>
+
+#include "common/assert.h"
+#include "common/common_types.h"
+#include "video_core/texture_cache/decode_bc4.h"
+#include "video_core/texture_cache/types.h"
+
+namespace VideoCommon {
+
+// https://www.khronos.org/registry/OpenGL/extensions/ARB/ARB_texture_compression_rgtc.txt
+[[nodiscard]] constexpr u32 DecompressBlock(u64 bits, u32 x, u32 y) {
+ const u32 code_offset = 16 + 3 * (4 * y + x);
+ const u32 code = (bits >> code_offset) & 7;
+ const u32 red0 = (bits >> 0) & 0xff;
+ const u32 red1 = (bits >> 8) & 0xff;
+ if (red0 > red1) {
+ switch (code) {
+ case 0:
+ return red0;
+ case 1:
+ return red1;
+ case 2:
+ return (6 * red0 + 1 * red1) / 7;
+ case 3:
+ return (5 * red0 + 2 * red1) / 7;
+ case 4:
+ return (4 * red0 + 3 * red1) / 7;
+ case 5:
+ return (3 * red0 + 4 * red1) / 7;
+ case 6:
+ return (2 * red0 + 5 * red1) / 7;
+ case 7:
+ return (1 * red0 + 6 * red1) / 7;
+ }
+ } else {
+ switch (code) {
+ case 0:
+ return red0;
+ case 1:
+ return red1;
+ case 2:
+ return (4 * red0 + 1 * red1) / 5;
+ case 3:
+ return (3 * red0 + 2 * red1) / 5;
+ case 4:
+ return (2 * red0 + 3 * red1) / 5;
+ case 5:
+ return (1 * red0 + 4 * red1) / 5;
+ case 6:
+ return 0;
+ case 7:
+ return 0xff;
+ }
+ }
+ return 0;
+}
+
+void DecompressBC4(std::span<const u8> input, Extent3D extent, std::span<u8> output) {
+ UNIMPLEMENTED_IF_MSG(extent.width % 4 != 0, "Unaligned width={}", extent.width);
+ UNIMPLEMENTED_IF_MSG(extent.height % 4 != 0, "Unaligned height={}", extent.height);
+ static constexpr u32 BLOCK_SIZE = 4;
+ size_t input_offset = 0;
+ for (u32 slice = 0; slice < extent.depth; ++slice) {
+ for (u32 block_y = 0; block_y < extent.height / 4; ++block_y) {
+ for (u32 block_x = 0; block_x < extent.width / 4; ++block_x) {
+ u64 bits;
+ std::memcpy(&bits, &input[input_offset], sizeof(bits));
+ input_offset += sizeof(bits);
+
+ for (u32 y = 0; y < BLOCK_SIZE; ++y) {
+ for (u32 x = 0; x < BLOCK_SIZE; ++x) {
+ const u32 linear_z = slice;
+ const u32 linear_y = block_y * BLOCK_SIZE + y;
+ const u32 linear_x = block_x * BLOCK_SIZE + x;
+ const u32 offset_z = linear_z * extent.width * extent.height;
+ const u32 offset_y = linear_y * extent.width;
+ const u32 offset_x = linear_x;
+ const u32 output_offset = (offset_z + offset_y + offset_x) * 4ULL;
+ const u32 color = DecompressBlock(bits, x, y);
+ output[output_offset + 0] = static_cast<u8>(color);
+ output[output_offset + 1] = 0;
+ output[output_offset + 2] = 0;
+ output[output_offset + 3] = 0xff;
+ }
+ }
+ }
+ }
+ }
+}
+
+} // namespace VideoCommon
diff --git a/src/video_core/texture_cache/decode_bc4.h b/src/video_core/texture_cache/decode_bc4.h
new file mode 100644
index 000000000..63fb23508
--- /dev/null
+++ b/src/video_core/texture_cache/decode_bc4.h
@@ -0,0 +1,16 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <span>
+
+#include "common/common_types.h"
+#include "video_core/texture_cache/types.h"
+
+namespace VideoCommon {
+
+void DecompressBC4(std::span<const u8> data, Extent3D extent, std::span<u8> output);
+
+} // namespace VideoCommon
diff --git a/src/video_core/texture_cache/descriptor_table.h b/src/video_core/texture_cache/descriptor_table.h
new file mode 100644
index 000000000..3a03b786f
--- /dev/null
+++ b/src/video_core/texture_cache/descriptor_table.h
@@ -0,0 +1,82 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <algorithm>
+#include <vector>
+
+#include "common/common_types.h"
+#include "common/div_ceil.h"
+#include "common/logging/log.h"
+#include "video_core/memory_manager.h"
+#include "video_core/rasterizer_interface.h"
+
+namespace VideoCommon {
+
+template <typename Descriptor>
+class DescriptorTable {
+public:
+ explicit DescriptorTable(Tegra::MemoryManager& gpu_memory_) : gpu_memory{gpu_memory_} {}
+
+ [[nodiscard]] bool Synchornize(GPUVAddr gpu_addr, u32 limit) {
+ [[likely]] if (current_gpu_addr == gpu_addr && current_limit == limit) {
+ return false;
+ }
+ Refresh(gpu_addr, limit);
+ return true;
+ }
+
+ void Invalidate() noexcept {
+ std::ranges::fill(read_descriptors, 0);
+ }
+
+ [[nodiscard]] std::pair<Descriptor, bool> Read(u32 index) {
+ DEBUG_ASSERT(index <= current_limit);
+ const GPUVAddr gpu_addr = current_gpu_addr + index * sizeof(Descriptor);
+ std::pair<Descriptor, bool> result;
+ gpu_memory.ReadBlockUnsafe(gpu_addr, &result.first, sizeof(Descriptor));
+ if (IsDescriptorRead(index)) {
+ result.second = result.first != descriptors[index];
+ } else {
+ MarkDescriptorAsRead(index);
+ result.second = true;
+ }
+ if (result.second) {
+ descriptors[index] = result.first;
+ }
+ return result;
+ }
+
+ [[nodiscard]] u32 Limit() const noexcept {
+ return current_limit;
+ }
+
+private:
+ void Refresh(GPUVAddr gpu_addr, u32 limit) {
+ current_gpu_addr = gpu_addr;
+ current_limit = limit;
+
+ const size_t num_descriptors = static_cast<size_t>(limit) + 1;
+ read_descriptors.clear();
+ read_descriptors.resize(Common::DivCeil(num_descriptors, 64U), 0);
+ descriptors.resize(num_descriptors);
+ }
+
+ void MarkDescriptorAsRead(u32 index) noexcept {
+ read_descriptors[index / 64] |= 1ULL << (index % 64);
+ }
+
+ [[nodiscard]] bool IsDescriptorRead(u32 index) const noexcept {
+ return (read_descriptors[index / 64] & (1ULL << (index % 64))) != 0;
+ }
+
+ Tegra::MemoryManager& gpu_memory;
+ GPUVAddr current_gpu_addr{};
+ u32 current_limit{};
+ std::vector<u64> read_descriptors;
+ std::vector<Descriptor> descriptors;
+};
+
+} // namespace VideoCommon
diff --git a/src/video_core/texture_cache/format_lookup_table.cpp b/src/video_core/texture_cache/format_lookup_table.cpp
index 7d5a75648..ddfb726fe 100644
--- a/src/video_core/texture_cache/format_lookup_table.cpp
+++ b/src/video_core/texture_cache/format_lookup_table.cpp
@@ -2,7 +2,6 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
-#include <array>
#include "common/common_types.h"
#include "common/logging/log.h"
#include "video_core/texture_cache/format_lookup_table.h"
@@ -20,198 +19,207 @@ constexpr auto UNORM = ComponentType::UNORM;
constexpr auto SINT = ComponentType::SINT;
constexpr auto UINT = ComponentType::UINT;
constexpr auto FLOAT = ComponentType::FLOAT;
-constexpr bool C = false; // Normal color
-constexpr bool S = true; // Srgb
-
-struct Table {
- constexpr Table(TextureFormat texture_format, bool is_srgb, ComponentType red_component,
- ComponentType green_component, ComponentType blue_component,
- ComponentType alpha_component, PixelFormat pixel_format)
- : texture_format{texture_format}, pixel_format{pixel_format}, red_component{red_component},
- green_component{green_component}, blue_component{blue_component},
- alpha_component{alpha_component}, is_srgb{is_srgb} {}
-
- TextureFormat texture_format;
- PixelFormat pixel_format;
- ComponentType red_component;
- ComponentType green_component;
- ComponentType blue_component;
- ComponentType alpha_component;
- bool is_srgb;
-};
-constexpr std::array<Table, 86> DefinitionTable = {{
- {TextureFormat::A8R8G8B8, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::A8B8G8R8_UNORM},
- {TextureFormat::A8R8G8B8, C, SNORM, SNORM, SNORM, SNORM, PixelFormat::A8B8G8R8_SNORM},
- {TextureFormat::A8R8G8B8, C, UINT, UINT, UINT, UINT, PixelFormat::A8B8G8R8_UINT},
- {TextureFormat::A8R8G8B8, C, SINT, SINT, SINT, SINT, PixelFormat::A8B8G8R8_SINT},
- {TextureFormat::A8R8G8B8, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::A8B8G8R8_SRGB},
-
- {TextureFormat::B5G6R5, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::B5G6R5_UNORM},
-
- {TextureFormat::A2B10G10R10, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::A2B10G10R10_UNORM},
- {TextureFormat::A2B10G10R10, C, UINT, UINT, UINT, UINT, PixelFormat::A2B10G10R10_UINT},
-
- {TextureFormat::A1B5G5R5, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::A1B5G5R5_UNORM},
-
- {TextureFormat::A4B4G4R4, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::A4B4G4R4_UNORM},
-
- {TextureFormat::R8, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::R8_UNORM},
- {TextureFormat::R8, C, SNORM, SNORM, SNORM, SNORM, PixelFormat::R8_SNORM},
- {TextureFormat::R8, C, UINT, UINT, UINT, UINT, PixelFormat::R8_UINT},
- {TextureFormat::R8, C, SINT, SINT, SINT, SINT, PixelFormat::R8_SINT},
-
- {TextureFormat::R8G8, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::R8G8_UNORM},
- {TextureFormat::R8G8, C, SNORM, SNORM, SNORM, SNORM, PixelFormat::R8G8_SNORM},
- {TextureFormat::R8G8, C, UINT, UINT, UINT, UINT, PixelFormat::R8G8_UINT},
- {TextureFormat::R8G8, C, SINT, SINT, SINT, SINT, PixelFormat::R8G8_SINT},
-
- {TextureFormat::R16G16B16A16, C, SNORM, SNORM, SNORM, SNORM, PixelFormat::R16G16B16A16_SNORM},
- {TextureFormat::R16G16B16A16, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::R16G16B16A16_UNORM},
- {TextureFormat::R16G16B16A16, C, FLOAT, FLOAT, FLOAT, FLOAT, PixelFormat::R16G16B16A16_FLOAT},
- {TextureFormat::R16G16B16A16, C, UINT, UINT, UINT, UINT, PixelFormat::R16G16B16A16_UINT},
- {TextureFormat::R16G16B16A16, C, SINT, SINT, SINT, SINT, PixelFormat::R16G16B16A16_SINT},
-
- {TextureFormat::R16G16, C, FLOAT, FLOAT, FLOAT, FLOAT, PixelFormat::R16G16_FLOAT},
- {TextureFormat::R16G16, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::R16G16_UNORM},
- {TextureFormat::R16G16, C, SNORM, SNORM, SNORM, SNORM, PixelFormat::R16G16_SNORM},
- {TextureFormat::R16G16, C, UINT, UINT, UINT, UINT, PixelFormat::R16G16_UINT},
- {TextureFormat::R16G16, C, SINT, SINT, SINT, SINT, PixelFormat::R16G16_SINT},
-
- {TextureFormat::R16, C, FLOAT, FLOAT, FLOAT, FLOAT, PixelFormat::R16_FLOAT},
- {TextureFormat::R16, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::R16_UNORM},
- {TextureFormat::R16, C, SNORM, SNORM, SNORM, SNORM, PixelFormat::R16_SNORM},
- {TextureFormat::R16, C, UINT, UINT, UINT, UINT, PixelFormat::R16_UINT},
- {TextureFormat::R16, C, SINT, SINT, SINT, SINT, PixelFormat::R16_SINT},
-
- {TextureFormat::B10G11R11, C, FLOAT, FLOAT, FLOAT, FLOAT, PixelFormat::B10G11R11_FLOAT},
-
- {TextureFormat::R32G32B32A32, C, FLOAT, FLOAT, FLOAT, FLOAT, PixelFormat::R32G32B32A32_FLOAT},
- {TextureFormat::R32G32B32A32, C, UINT, UINT, UINT, UINT, PixelFormat::R32G32B32A32_UINT},
- {TextureFormat::R32G32B32A32, C, SINT, SINT, SINT, SINT, PixelFormat::R32G32B32A32_SINT},
-
- {TextureFormat::R32G32B32, C, FLOAT, FLOAT, FLOAT, FLOAT, PixelFormat::R32G32B32_FLOAT},
-
- {TextureFormat::R32G32, C, FLOAT, FLOAT, FLOAT, FLOAT, PixelFormat::R32G32_FLOAT},
- {TextureFormat::R32G32, C, UINT, UINT, UINT, UINT, PixelFormat::R32G32_UINT},
- {TextureFormat::R32G32, C, SINT, SINT, SINT, SINT, PixelFormat::R32G32_SINT},
-
- {TextureFormat::R32, C, FLOAT, FLOAT, FLOAT, FLOAT, PixelFormat::R32_FLOAT},
- {TextureFormat::R32, C, UINT, UINT, UINT, UINT, PixelFormat::R32_UINT},
- {TextureFormat::R32, C, SINT, SINT, SINT, SINT, PixelFormat::R32_SINT},
-
- {TextureFormat::E5B9G9R9, C, FLOAT, FLOAT, FLOAT, FLOAT, PixelFormat::E5B9G9R9_FLOAT},
-
- {TextureFormat::D32, C, FLOAT, FLOAT, FLOAT, FLOAT, PixelFormat::D32_FLOAT},
- {TextureFormat::D16, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::D16_UNORM},
- {TextureFormat::S8D24, C, UINT, UNORM, UNORM, UNORM, PixelFormat::S8_UINT_D24_UNORM},
- {TextureFormat::R8G24, C, UINT, UNORM, UNORM, UNORM, PixelFormat::S8_UINT_D24_UNORM},
- {TextureFormat::D32S8, C, FLOAT, UINT, UNORM, UNORM, PixelFormat::D32_FLOAT_S8_UINT},
-
- {TextureFormat::BC1_RGBA, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::BC1_RGBA_UNORM},
- {TextureFormat::BC1_RGBA, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::BC1_RGBA_SRGB},
-
- {TextureFormat::BC2, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::BC2_UNORM},
- {TextureFormat::BC2, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::BC2_SRGB},
-
- {TextureFormat::BC3, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::BC3_UNORM},
- {TextureFormat::BC3, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::BC3_SRGB},
-
- {TextureFormat::BC4, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::BC4_UNORM},
- {TextureFormat::BC4, C, SNORM, SNORM, SNORM, SNORM, PixelFormat::BC4_SNORM},
-
- {TextureFormat::BC5, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::BC5_UNORM},
- {TextureFormat::BC5, C, SNORM, SNORM, SNORM, SNORM, PixelFormat::BC5_SNORM},
-
- {TextureFormat::BC7, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::BC7_UNORM},
- {TextureFormat::BC7, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::BC7_SRGB},
-
- {TextureFormat::BC6H_SFLOAT, C, FLOAT, FLOAT, FLOAT, FLOAT, PixelFormat::BC6H_SFLOAT},
- {TextureFormat::BC6H_UFLOAT, C, FLOAT, FLOAT, FLOAT, FLOAT, PixelFormat::BC6H_UFLOAT},
-
- {TextureFormat::ASTC_2D_4X4, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_4X4_UNORM},
- {TextureFormat::ASTC_2D_4X4, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_4X4_SRGB},
-
- {TextureFormat::ASTC_2D_5X4, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_5X4_UNORM},
- {TextureFormat::ASTC_2D_5X4, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_5X4_SRGB},
-
- {TextureFormat::ASTC_2D_5X5, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_5X5_UNORM},
- {TextureFormat::ASTC_2D_5X5, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_5X5_SRGB},
-
- {TextureFormat::ASTC_2D_8X8, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_8X8_UNORM},
- {TextureFormat::ASTC_2D_8X8, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_8X8_SRGB},
-
- {TextureFormat::ASTC_2D_8X5, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_8X5_UNORM},
- {TextureFormat::ASTC_2D_8X5, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_8X5_SRGB},
-
- {TextureFormat::ASTC_2D_10X8, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_10X8_UNORM},
- {TextureFormat::ASTC_2D_10X8, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_10X8_SRGB},
-
- {TextureFormat::ASTC_2D_6X6, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_6X6_UNORM},
- {TextureFormat::ASTC_2D_6X6, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_6X6_SRGB},
-
- {TextureFormat::ASTC_2D_10X10, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_10X10_UNORM},
- {TextureFormat::ASTC_2D_10X10, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_10X10_SRGB},
-
- {TextureFormat::ASTC_2D_12X12, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_12X12_UNORM},
- {TextureFormat::ASTC_2D_12X12, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_12X12_SRGB},
-
- {TextureFormat::ASTC_2D_8X6, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_8X6_UNORM},
- {TextureFormat::ASTC_2D_8X6, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_8X6_SRGB},
+constexpr bool LINEAR = false;
+constexpr bool SRGB = true;
+
+constexpr u32 Hash(TextureFormat format, ComponentType red_component, ComponentType green_component,
+ ComponentType blue_component, ComponentType alpha_component, bool is_srgb) {
+ u32 hash = is_srgb ? 1 : 0;
+ hash |= static_cast<u32>(red_component) << 1;
+ hash |= static_cast<u32>(green_component) << 4;
+ hash |= static_cast<u32>(blue_component) << 7;
+ hash |= static_cast<u32>(alpha_component) << 10;
+ hash |= static_cast<u32>(format) << 13;
+ return hash;
+}
- {TextureFormat::ASTC_2D_6X5, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_6X5_UNORM},
- {TextureFormat::ASTC_2D_6X5, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_6X5_SRGB},
-}};
+constexpr u32 Hash(TextureFormat format, ComponentType component, bool is_srgb = LINEAR) {
+ return Hash(format, component, component, component, component, is_srgb);
+}
} // Anonymous namespace
-FormatLookupTable::FormatLookupTable() {
- table.fill(static_cast<u8>(PixelFormat::Invalid));
-
- for (const auto& entry : DefinitionTable) {
- table[CalculateIndex(entry.texture_format, entry.is_srgb != 0, entry.red_component,
- entry.green_component, entry.blue_component, entry.alpha_component)] =
- static_cast<u8>(entry.pixel_format);
- }
-}
-
-PixelFormat FormatLookupTable::GetPixelFormat(TextureFormat format, bool is_srgb,
- ComponentType red_component,
- ComponentType green_component,
- ComponentType blue_component,
- ComponentType alpha_component) const noexcept {
- const auto pixel_format = static_cast<PixelFormat>(table[CalculateIndex(
- format, is_srgb, red_component, green_component, blue_component, alpha_component)]);
- // [[likely]]
- if (pixel_format != PixelFormat::Invalid) {
- return pixel_format;
+PixelFormat PixelFormatFromTextureInfo(TextureFormat format, ComponentType red, ComponentType green,
+ ComponentType blue, ComponentType alpha,
+ bool is_srgb) noexcept {
+ switch (Hash(format, red, green, blue, alpha, is_srgb)) {
+ case Hash(TextureFormat::A8R8G8B8, UNORM):
+ return PixelFormat::A8B8G8R8_UNORM;
+ case Hash(TextureFormat::A8R8G8B8, SNORM):
+ return PixelFormat::A8B8G8R8_SNORM;
+ case Hash(TextureFormat::A8R8G8B8, UINT):
+ return PixelFormat::A8B8G8R8_UINT;
+ case Hash(TextureFormat::A8R8G8B8, SINT):
+ return PixelFormat::A8B8G8R8_SINT;
+ case Hash(TextureFormat::A8R8G8B8, UNORM, SRGB):
+ return PixelFormat::A8B8G8R8_SRGB;
+ case Hash(TextureFormat::B5G6R5, UNORM):
+ return PixelFormat::B5G6R5_UNORM;
+ case Hash(TextureFormat::A2B10G10R10, UNORM):
+ return PixelFormat::A2B10G10R10_UNORM;
+ case Hash(TextureFormat::A2B10G10R10, UINT):
+ return PixelFormat::A2B10G10R10_UINT;
+ case Hash(TextureFormat::A1B5G5R5, UNORM):
+ return PixelFormat::A1B5G5R5_UNORM;
+ case Hash(TextureFormat::A4B4G4R4, UNORM):
+ return PixelFormat::A4B4G4R4_UNORM;
+ case Hash(TextureFormat::R8, UNORM):
+ return PixelFormat::R8_UNORM;
+ case Hash(TextureFormat::R8, SNORM):
+ return PixelFormat::R8_SNORM;
+ case Hash(TextureFormat::R8, UINT):
+ return PixelFormat::R8_UINT;
+ case Hash(TextureFormat::R8, SINT):
+ return PixelFormat::R8_SINT;
+ case Hash(TextureFormat::R8G8, UNORM):
+ return PixelFormat::R8G8_UNORM;
+ case Hash(TextureFormat::R8G8, SNORM):
+ return PixelFormat::R8G8_SNORM;
+ case Hash(TextureFormat::R8G8, UINT):
+ return PixelFormat::R8G8_UINT;
+ case Hash(TextureFormat::R8G8, SINT):
+ return PixelFormat::R8G8_SINT;
+ case Hash(TextureFormat::R16G16B16A16, FLOAT):
+ return PixelFormat::R16G16B16A16_FLOAT;
+ case Hash(TextureFormat::R16G16B16A16, UNORM):
+ return PixelFormat::R16G16B16A16_UNORM;
+ case Hash(TextureFormat::R16G16B16A16, SNORM):
+ return PixelFormat::R16G16B16A16_SNORM;
+ case Hash(TextureFormat::R16G16B16A16, UINT):
+ return PixelFormat::R16G16B16A16_UINT;
+ case Hash(TextureFormat::R16G16B16A16, SINT):
+ return PixelFormat::R16G16B16A16_SINT;
+ case Hash(TextureFormat::R16G16, FLOAT):
+ return PixelFormat::R16G16_FLOAT;
+ case Hash(TextureFormat::R16G16, UNORM):
+ return PixelFormat::R16G16_UNORM;
+ case Hash(TextureFormat::R16G16, SNORM):
+ return PixelFormat::R16G16_SNORM;
+ case Hash(TextureFormat::R16G16, UINT):
+ return PixelFormat::R16G16_UINT;
+ case Hash(TextureFormat::R16G16, SINT):
+ return PixelFormat::R16G16_SINT;
+ case Hash(TextureFormat::R16, FLOAT):
+ return PixelFormat::R16_FLOAT;
+ case Hash(TextureFormat::R16, UNORM):
+ return PixelFormat::R16_UNORM;
+ case Hash(TextureFormat::R16, SNORM):
+ return PixelFormat::R16_SNORM;
+ case Hash(TextureFormat::R16, UINT):
+ return PixelFormat::R16_UINT;
+ case Hash(TextureFormat::R16, SINT):
+ return PixelFormat::R16_SINT;
+ case Hash(TextureFormat::B10G11R11, FLOAT):
+ return PixelFormat::B10G11R11_FLOAT;
+ case Hash(TextureFormat::R32G32B32A32, FLOAT):
+ return PixelFormat::R32G32B32A32_FLOAT;
+ case Hash(TextureFormat::R32G32B32A32, UINT):
+ return PixelFormat::R32G32B32A32_UINT;
+ case Hash(TextureFormat::R32G32B32A32, SINT):
+ return PixelFormat::R32G32B32A32_SINT;
+ case Hash(TextureFormat::R32G32B32, FLOAT):
+ return PixelFormat::R32G32B32_FLOAT;
+ case Hash(TextureFormat::R32G32, FLOAT):
+ return PixelFormat::R32G32_FLOAT;
+ case Hash(TextureFormat::R32G32, UINT):
+ return PixelFormat::R32G32_UINT;
+ case Hash(TextureFormat::R32G32, SINT):
+ return PixelFormat::R32G32_SINT;
+ case Hash(TextureFormat::R32, FLOAT):
+ return PixelFormat::R32_FLOAT;
+ case Hash(TextureFormat::R32, UINT):
+ return PixelFormat::R32_UINT;
+ case Hash(TextureFormat::R32, SINT):
+ return PixelFormat::R32_SINT;
+ case Hash(TextureFormat::E5B9G9R9, FLOAT):
+ return PixelFormat::E5B9G9R9_FLOAT;
+ case Hash(TextureFormat::D32, FLOAT):
+ return PixelFormat::D32_FLOAT;
+ case Hash(TextureFormat::D16, UNORM):
+ return PixelFormat::D16_UNORM;
+ case Hash(TextureFormat::S8D24, UINT, UNORM, UNORM, UNORM, LINEAR):
+ return PixelFormat::S8_UINT_D24_UNORM;
+ case Hash(TextureFormat::R8G24, UINT, UNORM, UNORM, UNORM, LINEAR):
+ return PixelFormat::S8_UINT_D24_UNORM;
+ case Hash(TextureFormat::D32S8, FLOAT, UINT, UNORM, UNORM, LINEAR):
+ return PixelFormat::D32_FLOAT_S8_UINT;
+ case Hash(TextureFormat::BC1_RGBA, UNORM, LINEAR):
+ return PixelFormat::BC1_RGBA_UNORM;
+ case Hash(TextureFormat::BC1_RGBA, UNORM, SRGB):
+ return PixelFormat::BC1_RGBA_SRGB;
+ case Hash(TextureFormat::BC2, UNORM, LINEAR):
+ return PixelFormat::BC2_UNORM;
+ case Hash(TextureFormat::BC2, UNORM, SRGB):
+ return PixelFormat::BC2_SRGB;
+ case Hash(TextureFormat::BC3, UNORM, LINEAR):
+ return PixelFormat::BC3_UNORM;
+ case Hash(TextureFormat::BC3, UNORM, SRGB):
+ return PixelFormat::BC3_SRGB;
+ case Hash(TextureFormat::BC4, UNORM):
+ return PixelFormat::BC4_UNORM;
+ case Hash(TextureFormat::BC4, SNORM):
+ return PixelFormat::BC4_SNORM;
+ case Hash(TextureFormat::BC5, UNORM):
+ return PixelFormat::BC5_UNORM;
+ case Hash(TextureFormat::BC5, SNORM):
+ return PixelFormat::BC5_SNORM;
+ case Hash(TextureFormat::BC7, UNORM, LINEAR):
+ return PixelFormat::BC7_UNORM;
+ case Hash(TextureFormat::BC7, UNORM, SRGB):
+ return PixelFormat::BC7_SRGB;
+ case Hash(TextureFormat::BC6H_SFLOAT, FLOAT):
+ return PixelFormat::BC6H_SFLOAT;
+ case Hash(TextureFormat::BC6H_UFLOAT, FLOAT):
+ return PixelFormat::BC6H_UFLOAT;
+ case Hash(TextureFormat::ASTC_2D_4X4, UNORM, LINEAR):
+ return PixelFormat::ASTC_2D_4X4_UNORM;
+ case Hash(TextureFormat::ASTC_2D_4X4, UNORM, SRGB):
+ return PixelFormat::ASTC_2D_4X4_SRGB;
+ case Hash(TextureFormat::ASTC_2D_5X4, UNORM, LINEAR):
+ return PixelFormat::ASTC_2D_5X4_UNORM;
+ case Hash(TextureFormat::ASTC_2D_5X4, UNORM, SRGB):
+ return PixelFormat::ASTC_2D_5X4_SRGB;
+ case Hash(TextureFormat::ASTC_2D_5X5, UNORM, LINEAR):
+ return PixelFormat::ASTC_2D_5X5_UNORM;
+ case Hash(TextureFormat::ASTC_2D_5X5, UNORM, SRGB):
+ return PixelFormat::ASTC_2D_5X5_SRGB;
+ case Hash(TextureFormat::ASTC_2D_8X8, UNORM, LINEAR):
+ return PixelFormat::ASTC_2D_8X8_UNORM;
+ case Hash(TextureFormat::ASTC_2D_8X8, UNORM, SRGB):
+ return PixelFormat::ASTC_2D_8X8_SRGB;
+ case Hash(TextureFormat::ASTC_2D_8X5, UNORM, LINEAR):
+ return PixelFormat::ASTC_2D_8X5_UNORM;
+ case Hash(TextureFormat::ASTC_2D_8X5, UNORM, SRGB):
+ return PixelFormat::ASTC_2D_8X5_SRGB;
+ case Hash(TextureFormat::ASTC_2D_10X8, UNORM, LINEAR):
+ return PixelFormat::ASTC_2D_10X8_UNORM;
+ case Hash(TextureFormat::ASTC_2D_10X8, UNORM, SRGB):
+ return PixelFormat::ASTC_2D_10X8_SRGB;
+ case Hash(TextureFormat::ASTC_2D_6X6, UNORM, LINEAR):
+ return PixelFormat::ASTC_2D_6X6_UNORM;
+ case Hash(TextureFormat::ASTC_2D_6X6, UNORM, SRGB):
+ return PixelFormat::ASTC_2D_6X6_SRGB;
+ case Hash(TextureFormat::ASTC_2D_10X10, UNORM, LINEAR):
+ return PixelFormat::ASTC_2D_10X10_UNORM;
+ case Hash(TextureFormat::ASTC_2D_10X10, UNORM, SRGB):
+ return PixelFormat::ASTC_2D_10X10_SRGB;
+ case Hash(TextureFormat::ASTC_2D_12X12, UNORM, LINEAR):
+ return PixelFormat::ASTC_2D_12X12_UNORM;
+ case Hash(TextureFormat::ASTC_2D_12X12, UNORM, SRGB):
+ return PixelFormat::ASTC_2D_12X12_SRGB;
+ case Hash(TextureFormat::ASTC_2D_8X6, UNORM, LINEAR):
+ return PixelFormat::ASTC_2D_8X6_UNORM;
+ case Hash(TextureFormat::ASTC_2D_8X6, UNORM, SRGB):
+ return PixelFormat::ASTC_2D_8X6_SRGB;
+ case Hash(TextureFormat::ASTC_2D_6X5, UNORM, LINEAR):
+ return PixelFormat::ASTC_2D_6X5_UNORM;
+ case Hash(TextureFormat::ASTC_2D_6X5, UNORM, SRGB):
+ return PixelFormat::ASTC_2D_6X5_SRGB;
}
UNIMPLEMENTED_MSG("texture format={} srgb={} components={{{} {} {} {}}}",
- static_cast<int>(format), is_srgb, static_cast<int>(red_component),
- static_cast<int>(green_component), static_cast<int>(blue_component),
- static_cast<int>(alpha_component));
+ static_cast<int>(format), is_srgb, static_cast<int>(red),
+ static_cast<int>(green), static_cast<int>(blue), static_cast<int>(alpha));
return PixelFormat::A8B8G8R8_UNORM;
}
-void FormatLookupTable::Set(TextureFormat format, bool is_srgb, ComponentType red_component,
- ComponentType green_component, ComponentType blue_component,
- ComponentType alpha_component, PixelFormat pixel_format) {}
-
-std::size_t FormatLookupTable::CalculateIndex(TextureFormat format, bool is_srgb,
- ComponentType red_component,
- ComponentType green_component,
- ComponentType blue_component,
- ComponentType alpha_component) noexcept {
- const auto format_index = static_cast<std::size_t>(format);
- const auto red_index = static_cast<std::size_t>(red_component);
- const auto green_index = static_cast<std::size_t>(green_component);
- const auto blue_index = static_cast<std::size_t>(blue_component);
- const auto alpha_index = static_cast<std::size_t>(alpha_component);
- const std::size_t srgb_index = is_srgb ? 1 : 0;
-
- return format_index * PerFormat +
- srgb_index * PerComponent * PerComponent * PerComponent * PerComponent +
- alpha_index * PerComponent * PerComponent * PerComponent +
- blue_index * PerComponent * PerComponent + green_index * PerComponent + red_index;
-}
-
} // namespace VideoCommon
diff --git a/src/video_core/texture_cache/format_lookup_table.h b/src/video_core/texture_cache/format_lookup_table.h
index aa77e0a5a..729533999 100644
--- a/src/video_core/texture_cache/format_lookup_table.h
+++ b/src/video_core/texture_cache/format_lookup_table.h
@@ -4,48 +4,14 @@
#pragma once
-#include <array>
-#include <limits>
#include "video_core/surface.h"
#include "video_core/textures/texture.h"
namespace VideoCommon {
-class FormatLookupTable {
-public:
- explicit FormatLookupTable();
-
- VideoCore::Surface::PixelFormat GetPixelFormat(
- Tegra::Texture::TextureFormat format, bool is_srgb,
- Tegra::Texture::ComponentType red_component, Tegra::Texture::ComponentType green_component,
- Tegra::Texture::ComponentType blue_component,
- Tegra::Texture::ComponentType alpha_component) const noexcept;
-
-private:
- static_assert(VideoCore::Surface::MaxPixelFormat <= std::numeric_limits<u8>::max());
-
- static constexpr std::size_t NumTextureFormats = 128;
-
- static constexpr std::size_t PerComponent = 8;
- static constexpr std::size_t PerComponents2 = PerComponent * PerComponent;
- static constexpr std::size_t PerComponents3 = PerComponents2 * PerComponent;
- static constexpr std::size_t PerComponents4 = PerComponents3 * PerComponent;
- static constexpr std::size_t PerFormat = PerComponents4 * 2;
-
- static std::size_t CalculateIndex(Tegra::Texture::TextureFormat format, bool is_srgb,
- Tegra::Texture::ComponentType red_component,
- Tegra::Texture::ComponentType green_component,
- Tegra::Texture::ComponentType blue_component,
- Tegra::Texture::ComponentType alpha_component) noexcept;
-
- void Set(Tegra::Texture::TextureFormat format, bool is_srgb,
- Tegra::Texture::ComponentType red_component,
- Tegra::Texture::ComponentType green_component,
- Tegra::Texture::ComponentType blue_component,
- Tegra::Texture::ComponentType alpha_component,
- VideoCore::Surface::PixelFormat pixel_format);
-
- std::array<u8, NumTextureFormats * PerFormat> table;
-};
+VideoCore::Surface::PixelFormat PixelFormatFromTextureInfo(
+ Tegra::Texture::TextureFormat format, Tegra::Texture::ComponentType red_component,
+ Tegra::Texture::ComponentType green_component, Tegra::Texture::ComponentType blue_component,
+ Tegra::Texture::ComponentType alpha_component, bool is_srgb) noexcept;
} // namespace VideoCommon
diff --git a/src/video_core/texture_cache/formatter.cpp b/src/video_core/texture_cache/formatter.cpp
new file mode 100644
index 000000000..d10ba4ccd
--- /dev/null
+++ b/src/video_core/texture_cache/formatter.cpp
@@ -0,0 +1,95 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <algorithm>
+#include <string>
+
+#include "video_core/texture_cache/formatter.h"
+#include "video_core/texture_cache/image_base.h"
+#include "video_core/texture_cache/image_info.h"
+#include "video_core/texture_cache/image_view_base.h"
+#include "video_core/texture_cache/render_targets.h"
+
+namespace VideoCommon {
+
+std::string Name(const ImageBase& image) {
+ const GPUVAddr gpu_addr = image.gpu_addr;
+ const ImageInfo& info = image.info;
+ const u32 width = info.size.width;
+ const u32 height = info.size.height;
+ const u32 depth = info.size.depth;
+ const u32 num_layers = image.info.resources.layers;
+ const u32 num_levels = image.info.resources.levels;
+ std::string resource;
+ if (num_layers > 1) {
+ resource += fmt::format(":L{}", num_layers);
+ }
+ if (num_levels > 1) {
+ resource += fmt::format(":M{}", num_levels);
+ }
+ switch (image.info.type) {
+ case ImageType::e1D:
+ return fmt::format("Image 1D 0x{:x} {}{}", gpu_addr, width, resource);
+ case ImageType::e2D:
+ return fmt::format("Image 2D 0x{:x} {}x{}{}", gpu_addr, width, height, resource);
+ case ImageType::e3D:
+ return fmt::format("Image 2D 0x{:x} {}x{}x{}{}", gpu_addr, width, height, depth, resource);
+ case ImageType::Linear:
+ return fmt::format("Image Linear 0x{:x} {}x{}", gpu_addr, width, height);
+ case ImageType::Buffer:
+ return fmt::format("Buffer 0x{:x} {}", image.gpu_addr, image.info.size.width);
+ }
+ return "Invalid";
+}
+
+std::string Name(const ImageViewBase& image_view, std::optional<ImageViewType> type) {
+ const u32 width = image_view.size.width;
+ const u32 height = image_view.size.height;
+ const u32 depth = image_view.size.depth;
+ const u32 num_levels = image_view.range.extent.levels;
+ const u32 num_layers = image_view.range.extent.layers;
+
+ const std::string level = num_levels > 1 ? fmt::format(":{}", num_levels) : "";
+ switch (type.value_or(image_view.type)) {
+ case ImageViewType::e1D:
+ return fmt::format("ImageView 1D {}{}", width, level);
+ case ImageViewType::e2D:
+ return fmt::format("ImageView 2D {}x{}{}", width, height, level);
+ case ImageViewType::Cube:
+ return fmt::format("ImageView Cube {}x{}{}", width, height, level);
+ case ImageViewType::e3D:
+ return fmt::format("ImageView 3D {}x{}x{}{}", width, height, depth, level);
+ case ImageViewType::e1DArray:
+ return fmt::format("ImageView 1DArray {}{}|{}", width, level, num_layers);
+ case ImageViewType::e2DArray:
+ return fmt::format("ImageView 2DArray {}x{}{}|{}", width, height, level, num_layers);
+ case ImageViewType::CubeArray:
+ return fmt::format("ImageView CubeArray {}x{}{}|{}", width, height, level, num_layers);
+ case ImageViewType::Rect:
+ return fmt::format("ImageView Rect {}x{}{}", width, height, level);
+ case ImageViewType::Buffer:
+ return fmt::format("BufferView {}", width);
+ }
+ return "Invalid";
+}
+
+std::string Name(const RenderTargets& render_targets) {
+ std::string_view debug_prefix;
+ const auto num_color = std::ranges::count_if(
+ render_targets.color_buffer_ids, [](ImageViewId id) { return static_cast<bool>(id); });
+ if (render_targets.depth_buffer_id) {
+ debug_prefix = num_color > 0 ? "R" : "Z";
+ } else {
+ debug_prefix = num_color > 0 ? "C" : "X";
+ }
+ const Extent2D size = render_targets.size;
+ if (num_color > 0) {
+ return fmt::format("Framebuffer {}{} {}x{}", debug_prefix, num_color, size.width,
+ size.height);
+ } else {
+ return fmt::format("Framebuffer {} {}x{}", debug_prefix, size.width, size.height);
+ }
+}
+
+} // namespace VideoCommon
diff --git a/src/video_core/texture_cache/formatter.h b/src/video_core/texture_cache/formatter.h
new file mode 100644
index 000000000..a48413983
--- /dev/null
+++ b/src/video_core/texture_cache/formatter.h
@@ -0,0 +1,263 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <string>
+
+#include <fmt/format.h>
+
+#include "video_core/surface.h"
+#include "video_core/texture_cache/types.h"
+
+template <>
+struct fmt::formatter<VideoCore::Surface::PixelFormat> : fmt::formatter<fmt::string_view> {
+ template <typename FormatContext>
+ auto format(VideoCore::Surface::PixelFormat format, FormatContext& ctx) {
+ using VideoCore::Surface::PixelFormat;
+ const string_view name = [format] {
+ switch (format) {
+ case PixelFormat::A8B8G8R8_UNORM:
+ return "A8B8G8R8_UNORM";
+ case PixelFormat::A8B8G8R8_SNORM:
+ return "A8B8G8R8_SNORM";
+ case PixelFormat::A8B8G8R8_SINT:
+ return "A8B8G8R8_SINT";
+ case PixelFormat::A8B8G8R8_UINT:
+ return "A8B8G8R8_UINT";
+ case PixelFormat::R5G6B5_UNORM:
+ return "R5G6B5_UNORM";
+ case PixelFormat::B5G6R5_UNORM:
+ return "B5G6R5_UNORM";
+ case PixelFormat::A1R5G5B5_UNORM:
+ return "A1R5G5B5_UNORM";
+ case PixelFormat::A2B10G10R10_UNORM:
+ return "A2B10G10R10_UNORM";
+ case PixelFormat::A2B10G10R10_UINT:
+ return "A2B10G10R10_UINT";
+ case PixelFormat::A1B5G5R5_UNORM:
+ return "A1B5G5R5_UNORM";
+ case PixelFormat::R8_UNORM:
+ return "R8_UNORM";
+ case PixelFormat::R8_SNORM:
+ return "R8_SNORM";
+ case PixelFormat::R8_SINT:
+ return "R8_SINT";
+ case PixelFormat::R8_UINT:
+ return "R8_UINT";
+ case PixelFormat::R16G16B16A16_FLOAT:
+ return "R16G16B16A16_FLOAT";
+ case PixelFormat::R16G16B16A16_UNORM:
+ return "R16G16B16A16_UNORM";
+ case PixelFormat::R16G16B16A16_SNORM:
+ return "R16G16B16A16_SNORM";
+ case PixelFormat::R16G16B16A16_SINT:
+ return "R16G16B16A16_SINT";
+ case PixelFormat::R16G16B16A16_UINT:
+ return "R16G16B16A16_UINT";
+ case PixelFormat::B10G11R11_FLOAT:
+ return "B10G11R11_FLOAT";
+ case PixelFormat::R32G32B32A32_UINT:
+ return "R32G32B32A32_UINT";
+ case PixelFormat::BC1_RGBA_UNORM:
+ return "BC1_RGBA_UNORM";
+ case PixelFormat::BC2_UNORM:
+ return "BC2_UNORM";
+ case PixelFormat::BC3_UNORM:
+ return "BC3_UNORM";
+ case PixelFormat::BC4_UNORM:
+ return "BC4_UNORM";
+ case PixelFormat::BC4_SNORM:
+ return "BC4_SNORM";
+ case PixelFormat::BC5_UNORM:
+ return "BC5_UNORM";
+ case PixelFormat::BC5_SNORM:
+ return "BC5_SNORM";
+ case PixelFormat::BC7_UNORM:
+ return "BC7_UNORM";
+ case PixelFormat::BC6H_UFLOAT:
+ return "BC6H_UFLOAT";
+ case PixelFormat::BC6H_SFLOAT:
+ return "BC6H_SFLOAT";
+ case PixelFormat::ASTC_2D_4X4_UNORM:
+ return "ASTC_2D_4X4_UNORM";
+ case PixelFormat::B8G8R8A8_UNORM:
+ return "B8G8R8A8_UNORM";
+ case PixelFormat::R32G32B32A32_FLOAT:
+ return "R32G32B32A32_FLOAT";
+ case PixelFormat::R32G32B32A32_SINT:
+ return "R32G32B32A32_SINT";
+ case PixelFormat::R32G32_FLOAT:
+ return "R32G32_FLOAT";
+ case PixelFormat::R32G32_SINT:
+ return "R32G32_SINT";
+ case PixelFormat::R32_FLOAT:
+ return "R32_FLOAT";
+ case PixelFormat::R16_FLOAT:
+ return "R16_FLOAT";
+ case PixelFormat::R16_UNORM:
+ return "R16_UNORM";
+ case PixelFormat::R16_SNORM:
+ return "R16_SNORM";
+ case PixelFormat::R16_UINT:
+ return "R16_UINT";
+ case PixelFormat::R16_SINT:
+ return "R16_SINT";
+ case PixelFormat::R16G16_UNORM:
+ return "R16G16_UNORM";
+ case PixelFormat::R16G16_FLOAT:
+ return "R16G16_FLOAT";
+ case PixelFormat::R16G16_UINT:
+ return "R16G16_UINT";
+ case PixelFormat::R16G16_SINT:
+ return "R16G16_SINT";
+ case PixelFormat::R16G16_SNORM:
+ return "R16G16_SNORM";
+ case PixelFormat::R32G32B32_FLOAT:
+ return "R32G32B32_FLOAT";
+ case PixelFormat::A8B8G8R8_SRGB:
+ return "A8B8G8R8_SRGB";
+ case PixelFormat::R8G8_UNORM:
+ return "R8G8_UNORM";
+ case PixelFormat::R8G8_SNORM:
+ return "R8G8_SNORM";
+ case PixelFormat::R8G8_SINT:
+ return "R8G8_SINT";
+ case PixelFormat::R8G8_UINT:
+ return "R8G8_UINT";
+ case PixelFormat::R32G32_UINT:
+ return "R32G32_UINT";
+ case PixelFormat::R16G16B16X16_FLOAT:
+ return "R16G16B16X16_FLOAT";
+ case PixelFormat::R32_UINT:
+ return "R32_UINT";
+ case PixelFormat::R32_SINT:
+ return "R32_SINT";
+ case PixelFormat::ASTC_2D_8X8_UNORM:
+ return "ASTC_2D_8X8_UNORM";
+ case PixelFormat::ASTC_2D_8X5_UNORM:
+ return "ASTC_2D_8X5_UNORM";
+ case PixelFormat::ASTC_2D_5X4_UNORM:
+ return "ASTC_2D_5X4_UNORM";
+ case PixelFormat::B8G8R8A8_SRGB:
+ return "B8G8R8A8_SRGB";
+ case PixelFormat::BC1_RGBA_SRGB:
+ return "BC1_RGBA_SRGB";
+ case PixelFormat::BC2_SRGB:
+ return "BC2_SRGB";
+ case PixelFormat::BC3_SRGB:
+ return "BC3_SRGB";
+ case PixelFormat::BC7_SRGB:
+ return "BC7_SRGB";
+ case PixelFormat::A4B4G4R4_UNORM:
+ return "A4B4G4R4_UNORM";
+ case PixelFormat::ASTC_2D_4X4_SRGB:
+ return "ASTC_2D_4X4_SRGB";
+ case PixelFormat::ASTC_2D_8X8_SRGB:
+ return "ASTC_2D_8X8_SRGB";
+ case PixelFormat::ASTC_2D_8X5_SRGB:
+ return "ASTC_2D_8X5_SRGB";
+ case PixelFormat::ASTC_2D_5X4_SRGB:
+ return "ASTC_2D_5X4_SRGB";
+ case PixelFormat::ASTC_2D_5X5_UNORM:
+ return "ASTC_2D_5X5_UNORM";
+ case PixelFormat::ASTC_2D_5X5_SRGB:
+ return "ASTC_2D_5X5_SRGB";
+ case PixelFormat::ASTC_2D_10X8_UNORM:
+ return "ASTC_2D_10X8_UNORM";
+ case PixelFormat::ASTC_2D_10X8_SRGB:
+ return "ASTC_2D_10X8_SRGB";
+ case PixelFormat::ASTC_2D_6X6_UNORM:
+ return "ASTC_2D_6X6_UNORM";
+ case PixelFormat::ASTC_2D_6X6_SRGB:
+ return "ASTC_2D_6X6_SRGB";
+ case PixelFormat::ASTC_2D_10X10_UNORM:
+ return "ASTC_2D_10X10_UNORM";
+ case PixelFormat::ASTC_2D_10X10_SRGB:
+ return "ASTC_2D_10X10_SRGB";
+ case PixelFormat::ASTC_2D_12X12_UNORM:
+ return "ASTC_2D_12X12_UNORM";
+ case PixelFormat::ASTC_2D_12X12_SRGB:
+ return "ASTC_2D_12X12_SRGB";
+ case PixelFormat::ASTC_2D_8X6_UNORM:
+ return "ASTC_2D_8X6_UNORM";
+ case PixelFormat::ASTC_2D_8X6_SRGB:
+ return "ASTC_2D_8X6_SRGB";
+ case PixelFormat::ASTC_2D_6X5_UNORM:
+ return "ASTC_2D_6X5_UNORM";
+ case PixelFormat::ASTC_2D_6X5_SRGB:
+ return "ASTC_2D_6X5_SRGB";
+ case PixelFormat::E5B9G9R9_FLOAT:
+ return "E5B9G9R9_FLOAT";
+ case PixelFormat::D32_FLOAT:
+ return "D32_FLOAT";
+ case PixelFormat::D16_UNORM:
+ return "D16_UNORM";
+ case PixelFormat::D24_UNORM_S8_UINT:
+ return "D24_UNORM_S8_UINT";
+ case PixelFormat::S8_UINT_D24_UNORM:
+ return "S8_UINT_D24_UNORM";
+ case PixelFormat::D32_FLOAT_S8_UINT:
+ return "D32_FLOAT_S8_UINT";
+ case PixelFormat::MaxDepthStencilFormat:
+ case PixelFormat::Invalid:
+ return "Invalid";
+ }
+ return "Invalid";
+ }();
+ return formatter<string_view>::format(name, ctx);
+ }
+};
+
+template <>
+struct fmt::formatter<VideoCommon::ImageType> : fmt::formatter<fmt::string_view> {
+ template <typename FormatContext>
+ auto format(VideoCommon::ImageType type, FormatContext& ctx) {
+ const string_view name = [type] {
+ using VideoCommon::ImageType;
+ switch (type) {
+ case ImageType::e1D:
+ return "1D";
+ case ImageType::e2D:
+ return "2D";
+ case ImageType::e3D:
+ return "3D";
+ case ImageType::Linear:
+ return "Linear";
+ case ImageType::Buffer:
+ return "Buffer";
+ }
+ return "Invalid";
+ }();
+ return formatter<string_view>::format(name, ctx);
+ }
+};
+
+template <>
+struct fmt::formatter<VideoCommon::Extent3D> {
+ constexpr auto parse(fmt::format_parse_context& ctx) {
+ return ctx.begin();
+ }
+
+ template <typename FormatContext>
+ auto format(const VideoCommon::Extent3D& extent, FormatContext& ctx) {
+ return fmt::format_to(ctx.out(), "{{{}, {}, {}}}", extent.width, extent.height,
+ extent.depth);
+ }
+};
+
+namespace VideoCommon {
+
+struct ImageBase;
+struct ImageViewBase;
+struct RenderTargets;
+
+[[nodiscard]] std::string Name(const ImageBase& image);
+
+[[nodiscard]] std::string Name(const ImageViewBase& image_view,
+ std::optional<ImageViewType> type = std::nullopt);
+
+[[nodiscard]] std::string Name(const RenderTargets& render_targets);
+
+} // namespace VideoCommon
diff --git a/src/video_core/texture_cache/image_base.cpp b/src/video_core/texture_cache/image_base.cpp
new file mode 100644
index 000000000..448a05fcc
--- /dev/null
+++ b/src/video_core/texture_cache/image_base.cpp
@@ -0,0 +1,216 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <algorithm>
+#include <optional>
+#include <utility>
+#include <vector>
+
+#include "common/common_types.h"
+#include "video_core/surface.h"
+#include "video_core/texture_cache/formatter.h"
+#include "video_core/texture_cache/image_base.h"
+#include "video_core/texture_cache/image_view_info.h"
+#include "video_core/texture_cache/util.h"
+
+namespace VideoCommon {
+
+using VideoCore::Surface::DefaultBlockHeight;
+using VideoCore::Surface::DefaultBlockWidth;
+
+namespace {
+/// Returns the base layer and mip level offset
+[[nodiscard]] std::pair<s32, s32> LayerMipOffset(s32 diff, u32 layer_stride) {
+ if (layer_stride == 0) {
+ return {0, diff};
+ } else {
+ return {diff / layer_stride, diff % layer_stride};
+ }
+}
+
+[[nodiscard]] bool ValidateLayers(const SubresourceLayers& layers, const ImageInfo& info) {
+ return layers.base_level < info.resources.levels &&
+ layers.base_layer + layers.num_layers <= info.resources.layers;
+}
+
+[[nodiscard]] bool ValidateCopy(const ImageCopy& copy, const ImageInfo& dst, const ImageInfo& src) {
+ const Extent3D src_size = MipSize(src.size, copy.src_subresource.base_level);
+ const Extent3D dst_size = MipSize(dst.size, copy.dst_subresource.base_level);
+ if (!ValidateLayers(copy.src_subresource, src)) {
+ return false;
+ }
+ if (!ValidateLayers(copy.dst_subresource, dst)) {
+ return false;
+ }
+ if (copy.src_offset.x + copy.extent.width > src_size.width ||
+ copy.src_offset.y + copy.extent.height > src_size.height ||
+ copy.src_offset.z + copy.extent.depth > src_size.depth) {
+ return false;
+ }
+ if (copy.dst_offset.x + copy.extent.width > dst_size.width ||
+ copy.dst_offset.y + copy.extent.height > dst_size.height ||
+ copy.dst_offset.z + copy.extent.depth > dst_size.depth) {
+ return false;
+ }
+ return true;
+}
+} // Anonymous namespace
+
+ImageBase::ImageBase(const ImageInfo& info_, GPUVAddr gpu_addr_, VAddr cpu_addr_)
+ : info{info_}, guest_size_bytes{CalculateGuestSizeInBytes(info)},
+ unswizzled_size_bytes{CalculateUnswizzledSizeBytes(info)},
+ converted_size_bytes{CalculateConvertedSizeBytes(info)}, gpu_addr{gpu_addr_},
+ cpu_addr{cpu_addr_}, cpu_addr_end{cpu_addr + guest_size_bytes},
+ mip_level_offsets{CalculateMipLevelOffsets(info)} {
+ if (info.type == ImageType::e3D) {
+ slice_offsets = CalculateSliceOffsets(info);
+ slice_subresources = CalculateSliceSubresources(info);
+ }
+}
+
+std::optional<SubresourceBase> ImageBase::TryFindBase(GPUVAddr other_addr) const noexcept {
+ if (other_addr < gpu_addr) {
+ // Subresource address can't be lower than the base
+ return std::nullopt;
+ }
+ const u32 diff = static_cast<u32>(other_addr - gpu_addr);
+ if (diff > guest_size_bytes) {
+ // This can happen when two CPU addresses are used for different GPU addresses
+ return std::nullopt;
+ }
+ if (info.type != ImageType::e3D) {
+ const auto [layer, mip_offset] = LayerMipOffset(diff, info.layer_stride);
+ const auto end = mip_level_offsets.begin() + info.resources.levels;
+ const auto it = std::find(mip_level_offsets.begin(), end, mip_offset);
+ if (layer > info.resources.layers || it == end) {
+ return std::nullopt;
+ }
+ return SubresourceBase{
+ .level = static_cast<s32>(std::distance(mip_level_offsets.begin(), it)),
+ .layer = layer,
+ };
+ } else {
+ // TODO: Consider using binary_search after a threshold
+ const auto it = std::ranges::find(slice_offsets, diff);
+ if (it == slice_offsets.cend()) {
+ return std::nullopt;
+ }
+ return slice_subresources[std::distance(slice_offsets.begin(), it)];
+ }
+}
+
+ImageViewId ImageBase::FindView(const ImageViewInfo& view_info) const noexcept {
+ const auto it = std::ranges::find(image_view_infos, view_info);
+ if (it == image_view_infos.end()) {
+ return ImageViewId{};
+ }
+ return image_view_ids[std::distance(image_view_infos.begin(), it)];
+}
+
+void ImageBase::InsertView(const ImageViewInfo& view_info, ImageViewId image_view_id) {
+ image_view_infos.push_back(view_info);
+ image_view_ids.push_back(image_view_id);
+}
+
+void AddImageAlias(ImageBase& lhs, ImageBase& rhs, ImageId lhs_id, ImageId rhs_id) {
+ static constexpr auto OPTIONS = RelaxedOptions::Size | RelaxedOptions::Format;
+ ASSERT(lhs.info.type == rhs.info.type);
+ std::optional<SubresourceBase> base;
+ if (lhs.info.type == ImageType::Linear) {
+ base = SubresourceBase{.level = 0, .layer = 0};
+ } else {
+ base = FindSubresource(rhs.info, lhs, rhs.gpu_addr, OPTIONS);
+ }
+ if (!base) {
+ LOG_ERROR(HW_GPU, "Image alias should have been flipped");
+ return;
+ }
+ const PixelFormat lhs_format = lhs.info.format;
+ const PixelFormat rhs_format = rhs.info.format;
+ const Extent2D lhs_block{
+ .width = DefaultBlockWidth(lhs_format),
+ .height = DefaultBlockHeight(lhs_format),
+ };
+ const Extent2D rhs_block{
+ .width = DefaultBlockWidth(rhs_format),
+ .height = DefaultBlockHeight(rhs_format),
+ };
+ const bool is_lhs_compressed = lhs_block.width > 1 || lhs_block.height > 1;
+ const bool is_rhs_compressed = rhs_block.width > 1 || rhs_block.height > 1;
+ if (is_lhs_compressed && is_rhs_compressed) {
+ LOG_ERROR(HW_GPU, "Compressed to compressed image aliasing is not implemented");
+ return;
+ }
+ const s32 lhs_mips = lhs.info.resources.levels;
+ const s32 rhs_mips = rhs.info.resources.levels;
+ const s32 num_mips = std::min(lhs_mips - base->level, rhs_mips);
+ AliasedImage lhs_alias;
+ AliasedImage rhs_alias;
+ lhs_alias.id = rhs_id;
+ rhs_alias.id = lhs_id;
+ lhs_alias.copies.reserve(num_mips);
+ rhs_alias.copies.reserve(num_mips);
+ for (s32 mip_level = 0; mip_level < num_mips; ++mip_level) {
+ Extent3D lhs_size = MipSize(lhs.info.size, base->level + mip_level);
+ Extent3D rhs_size = MipSize(rhs.info.size, mip_level);
+ if (is_lhs_compressed) {
+ lhs_size.width /= lhs_block.width;
+ lhs_size.height /= lhs_block.height;
+ }
+ if (is_rhs_compressed) {
+ rhs_size.width /= rhs_block.width;
+ rhs_size.height /= rhs_block.height;
+ }
+ const Extent3D copy_size{
+ .width = std::min(lhs_size.width, rhs_size.width),
+ .height = std::min(lhs_size.height, rhs_size.height),
+ .depth = std::min(lhs_size.depth, rhs_size.depth),
+ };
+ if (copy_size.width == 0 || copy_size.height == 0) {
+ LOG_WARNING(HW_GPU, "Copy size is smaller than block size. Mip cannot be aliased.");
+ continue;
+ }
+ const bool is_lhs_3d = lhs.info.type == ImageType::e3D;
+ const bool is_rhs_3d = rhs.info.type == ImageType::e3D;
+ const Offset3D lhs_offset{0, 0, 0};
+ const Offset3D rhs_offset{0, 0, is_rhs_3d ? base->layer : 0};
+ const s32 lhs_layers = is_lhs_3d ? 1 : lhs.info.resources.layers - base->layer;
+ const s32 rhs_layers = is_rhs_3d ? 1 : rhs.info.resources.layers;
+ const s32 num_layers = std::min(lhs_layers, rhs_layers);
+ const SubresourceLayers lhs_subresource{
+ .base_level = mip_level,
+ .base_layer = 0,
+ .num_layers = num_layers,
+ };
+ const SubresourceLayers rhs_subresource{
+ .base_level = base->level + mip_level,
+ .base_layer = is_rhs_3d ? 0 : base->layer,
+ .num_layers = num_layers,
+ };
+ [[maybe_unused]] const ImageCopy& to_lhs_copy = lhs_alias.copies.emplace_back(ImageCopy{
+ .src_subresource = lhs_subresource,
+ .dst_subresource = rhs_subresource,
+ .src_offset = lhs_offset,
+ .dst_offset = rhs_offset,
+ .extent = copy_size,
+ });
+ [[maybe_unused]] const ImageCopy& to_rhs_copy = rhs_alias.copies.emplace_back(ImageCopy{
+ .src_subresource = rhs_subresource,
+ .dst_subresource = lhs_subresource,
+ .src_offset = rhs_offset,
+ .dst_offset = lhs_offset,
+ .extent = copy_size,
+ });
+ ASSERT_MSG(ValidateCopy(to_lhs_copy, lhs.info, rhs.info), "Invalid RHS to LHS copy");
+ ASSERT_MSG(ValidateCopy(to_rhs_copy, rhs.info, lhs.info), "Invalid LHS to RHS copy");
+ }
+ ASSERT(lhs_alias.copies.empty() == rhs_alias.copies.empty());
+ if (lhs_alias.copies.empty()) {
+ return;
+ }
+ lhs.aliased_images.push_back(std::move(lhs_alias));
+ rhs.aliased_images.push_back(std::move(rhs_alias));
+}
+
+} // namespace VideoCommon
diff --git a/src/video_core/texture_cache/image_base.h b/src/video_core/texture_cache/image_base.h
new file mode 100644
index 000000000..b7f3b7e43
--- /dev/null
+++ b/src/video_core/texture_cache/image_base.h
@@ -0,0 +1,83 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include <optional>
+#include <vector>
+
+#include "common/common_funcs.h"
+#include "common/common_types.h"
+#include "video_core/texture_cache/image_info.h"
+#include "video_core/texture_cache/image_view_info.h"
+#include "video_core/texture_cache/types.h"
+
+namespace VideoCommon {
+
+enum class ImageFlagBits : u32 {
+ AcceleratedUpload = 1 << 0, ///< Upload can be accelerated in the GPU
+ Converted = 1 << 1, ///< Guest format is not supported natively and it has to be converted
+ CpuModified = 1 << 2, ///< Contents have been modified from the CPU
+ GpuModified = 1 << 3, ///< Contents have been modified from the GPU
+ Tracked = 1 << 4, ///< Writes and reads are being hooked from the CPU JIT
+ Strong = 1 << 5, ///< Exists in the image table, the dimensions are can be trusted
+ Registered = 1 << 6, ///< True when the image is registered
+ Picked = 1 << 7, ///< Temporary flag to mark the image as picked
+};
+DECLARE_ENUM_FLAG_OPERATORS(ImageFlagBits)
+
+struct ImageViewInfo;
+
+struct AliasedImage {
+ std::vector<ImageCopy> copies;
+ ImageId id;
+};
+
+struct ImageBase {
+ explicit ImageBase(const ImageInfo& info, GPUVAddr gpu_addr, VAddr cpu_addr);
+
+ [[nodiscard]] std::optional<SubresourceBase> TryFindBase(GPUVAddr other_addr) const noexcept;
+
+ [[nodiscard]] ImageViewId FindView(const ImageViewInfo& view_info) const noexcept;
+
+ void InsertView(const ImageViewInfo& view_info, ImageViewId image_view_id);
+
+ [[nodiscard]] bool Overlaps(VAddr overlap_cpu_addr, size_t overlap_size) const noexcept {
+ const VAddr overlap_end = overlap_cpu_addr + overlap_size;
+ return cpu_addr < overlap_end && overlap_cpu_addr < cpu_addr_end;
+ }
+
+ ImageInfo info;
+
+ u32 guest_size_bytes = 0;
+ u32 unswizzled_size_bytes = 0;
+ u32 converted_size_bytes = 0;
+ ImageFlagBits flags = ImageFlagBits::CpuModified;
+
+ GPUVAddr gpu_addr = 0;
+ VAddr cpu_addr = 0;
+ VAddr cpu_addr_end = 0;
+
+ u64 modification_tick = 0;
+ u64 frame_tick = 0;
+
+ std::array<u32, MAX_MIP_LEVELS> mip_level_offsets{};
+
+ std::vector<ImageViewInfo> image_view_infos;
+ std::vector<ImageViewId> image_view_ids;
+
+ std::vector<u32> slice_offsets;
+ std::vector<SubresourceBase> slice_subresources;
+
+ std::vector<AliasedImage> aliased_images;
+};
+
+struct ImageAllocBase {
+ std::vector<ImageId> images;
+};
+
+void AddImageAlias(ImageBase& lhs, ImageBase& rhs, ImageId lhs_id, ImageId rhs_id);
+
+} // namespace VideoCommon
diff --git a/src/video_core/texture_cache/image_info.cpp b/src/video_core/texture_cache/image_info.cpp
new file mode 100644
index 000000000..64fd7010a
--- /dev/null
+++ b/src/video_core/texture_cache/image_info.cpp
@@ -0,0 +1,189 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "common/assert.h"
+#include "video_core/surface.h"
+#include "video_core/texture_cache/format_lookup_table.h"
+#include "video_core/texture_cache/image_info.h"
+#include "video_core/texture_cache/samples_helper.h"
+#include "video_core/texture_cache/types.h"
+#include "video_core/texture_cache/util.h"
+#include "video_core/textures/texture.h"
+
+namespace VideoCommon {
+
+using Tegra::Texture::TextureType;
+using Tegra::Texture::TICEntry;
+using VideoCore::Surface::PixelFormat;
+
+ImageInfo::ImageInfo(const TICEntry& config) noexcept {
+ format = PixelFormatFromTextureInfo(config.format, config.r_type, config.g_type, config.b_type,
+ config.a_type, config.srgb_conversion);
+ num_samples = NumSamples(config.msaa_mode);
+ resources.levels = config.max_mip_level + 1;
+ if (config.IsPitchLinear()) {
+ pitch = config.Pitch();
+ } else if (config.IsBlockLinear()) {
+ block = Extent3D{
+ .width = config.block_width,
+ .height = config.block_height,
+ .depth = config.block_depth,
+ };
+ }
+ tile_width_spacing = config.tile_width_spacing;
+ if (config.texture_type != TextureType::Texture2D &&
+ config.texture_type != TextureType::Texture2DNoMipmap) {
+ ASSERT(!config.IsPitchLinear());
+ }
+ switch (config.texture_type) {
+ case TextureType::Texture1D:
+ ASSERT(config.BaseLayer() == 0);
+ type = ImageType::e1D;
+ size.width = config.Width();
+ break;
+ case TextureType::Texture1DArray:
+ UNIMPLEMENTED_IF(config.BaseLayer() != 0);
+ type = ImageType::e1D;
+ size.width = config.Width();
+ resources.layers = config.Depth();
+ break;
+ case TextureType::Texture2D:
+ case TextureType::Texture2DNoMipmap:
+ ASSERT(config.Depth() == 1);
+ type = config.IsPitchLinear() ? ImageType::Linear : ImageType::e2D;
+ size.width = config.Width();
+ size.height = config.Height();
+ resources.layers = config.BaseLayer() + 1;
+ break;
+ case TextureType::Texture2DArray:
+ type = ImageType::e2D;
+ size.width = config.Width();
+ size.height = config.Height();
+ resources.layers = config.BaseLayer() + config.Depth();
+ break;
+ case TextureType::TextureCubemap:
+ ASSERT(config.Depth() == 1);
+ type = ImageType::e2D;
+ size.width = config.Width();
+ size.height = config.Height();
+ resources.layers = config.BaseLayer() + 6;
+ break;
+ case TextureType::TextureCubeArray:
+ UNIMPLEMENTED_IF(config.load_store_hint != 0);
+ type = ImageType::e2D;
+ size.width = config.Width();
+ size.height = config.Height();
+ resources.layers = config.BaseLayer() + config.Depth() * 6;
+ break;
+ case TextureType::Texture3D:
+ ASSERT(config.BaseLayer() == 0);
+ type = ImageType::e3D;
+ size.width = config.Width();
+ size.height = config.Height();
+ size.depth = config.Depth();
+ break;
+ case TextureType::Texture1DBuffer:
+ type = ImageType::Buffer;
+ size.width = config.Width();
+ break;
+ default:
+ UNREACHABLE_MSG("Invalid texture_type={}", static_cast<int>(config.texture_type.Value()));
+ break;
+ }
+ if (type != ImageType::Linear) {
+ // FIXME: Call this without passing *this
+ layer_stride = CalculateLayerStride(*this);
+ maybe_unaligned_layer_stride = CalculateLayerSize(*this);
+ }
+}
+
+ImageInfo::ImageInfo(const Tegra::Engines::Maxwell3D::Regs& regs, size_t index) noexcept {
+ const auto& rt = regs.rt[index];
+ format = VideoCore::Surface::PixelFormatFromRenderTargetFormat(rt.format);
+ if (rt.tile_mode.is_pitch_linear) {
+ ASSERT(rt.tile_mode.is_3d == 0);
+ type = ImageType::Linear;
+ pitch = rt.width;
+ size = Extent3D{
+ .width = pitch / BytesPerBlock(format),
+ .height = rt.height,
+ .depth = 1,
+ };
+ return;
+ }
+ size.width = rt.width;
+ size.height = rt.height;
+ layer_stride = rt.layer_stride * 4;
+ maybe_unaligned_layer_stride = layer_stride;
+ num_samples = NumSamples(regs.multisample_mode);
+ block = Extent3D{
+ .width = rt.tile_mode.block_width,
+ .height = rt.tile_mode.block_height,
+ .depth = rt.tile_mode.block_depth,
+ };
+ if (rt.tile_mode.is_3d) {
+ type = ImageType::e3D;
+ size.depth = rt.depth;
+ } else {
+ type = ImageType::e2D;
+ resources.layers = rt.depth;
+ }
+}
+
+ImageInfo::ImageInfo(const Tegra::Engines::Maxwell3D::Regs& regs) noexcept {
+ format = VideoCore::Surface::PixelFormatFromDepthFormat(regs.zeta.format);
+ size.width = regs.zeta_width;
+ size.height = regs.zeta_height;
+ resources.levels = 1;
+ layer_stride = regs.zeta.layer_stride * 4;
+ maybe_unaligned_layer_stride = layer_stride;
+ num_samples = NumSamples(regs.multisample_mode);
+ block = Extent3D{
+ .width = regs.zeta.tile_mode.block_width,
+ .height = regs.zeta.tile_mode.block_height,
+ .depth = regs.zeta.tile_mode.block_depth,
+ };
+ if (regs.zeta.tile_mode.is_pitch_linear) {
+ ASSERT(regs.zeta.tile_mode.is_3d == 0);
+ type = ImageType::Linear;
+ pitch = size.width * BytesPerBlock(format);
+ } else if (regs.zeta.tile_mode.is_3d) {
+ ASSERT(regs.zeta.tile_mode.is_pitch_linear == 0);
+ type = ImageType::e3D;
+ size.depth = regs.zeta_depth;
+ } else {
+ type = ImageType::e2D;
+ resources.layers = regs.zeta_depth;
+ }
+}
+
+ImageInfo::ImageInfo(const Tegra::Engines::Fermi2D::Surface& config) noexcept {
+ UNIMPLEMENTED_IF_MSG(config.layer != 0, "Surface layer is not zero");
+ format = VideoCore::Surface::PixelFormatFromRenderTargetFormat(config.format);
+ if (config.linear == Tegra::Engines::Fermi2D::MemoryLayout::Pitch) {
+ type = ImageType::Linear;
+ size = Extent3D{
+ .width = config.pitch / VideoCore::Surface::BytesPerBlock(format),
+ .height = config.height,
+ .depth = 1,
+ };
+ pitch = config.pitch;
+ } else {
+ type = config.block_depth > 0 ? ImageType::e3D : ImageType::e2D;
+ block = Extent3D{
+ .width = config.block_width,
+ .height = config.block_height,
+ .depth = config.block_depth,
+ };
+ // 3D blits with more than once slice are not implemented for now
+ // Render to individual slices
+ size = Extent3D{
+ .width = config.width,
+ .height = config.height,
+ .depth = 1,
+ };
+ }
+}
+
+} // namespace VideoCommon
diff --git a/src/video_core/texture_cache/image_info.h b/src/video_core/texture_cache/image_info.h
new file mode 100644
index 000000000..5049fc36e
--- /dev/null
+++ b/src/video_core/texture_cache/image_info.h
@@ -0,0 +1,38 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "video_core/engines/fermi_2d.h"
+#include "video_core/engines/maxwell_3d.h"
+#include "video_core/surface.h"
+#include "video_core/texture_cache/types.h"
+
+namespace VideoCommon {
+
+using Tegra::Texture::TICEntry;
+using VideoCore::Surface::PixelFormat;
+
+struct ImageInfo {
+ explicit ImageInfo() = default;
+ explicit ImageInfo(const TICEntry& config) noexcept;
+ explicit ImageInfo(const Tegra::Engines::Maxwell3D::Regs& regs, size_t index) noexcept;
+ explicit ImageInfo(const Tegra::Engines::Maxwell3D::Regs& regs) noexcept;
+ explicit ImageInfo(const Tegra::Engines::Fermi2D::Surface& config) noexcept;
+
+ PixelFormat format = PixelFormat::Invalid;
+ ImageType type = ImageType::e1D;
+ SubresourceExtent resources;
+ Extent3D size{1, 1, 1};
+ union {
+ Extent3D block{0, 0, 0};
+ u32 pitch;
+ };
+ u32 layer_stride = 0;
+ u32 maybe_unaligned_layer_stride = 0;
+ u32 num_samples = 1;
+ u32 tile_width_spacing = 0;
+};
+
+} // namespace VideoCommon
diff --git a/src/video_core/texture_cache/image_view_base.cpp b/src/video_core/texture_cache/image_view_base.cpp
new file mode 100644
index 000000000..076a4bcfd
--- /dev/null
+++ b/src/video_core/texture_cache/image_view_base.cpp
@@ -0,0 +1,41 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <algorithm>
+
+#include "common/assert.h"
+#include "core/settings.h"
+#include "video_core/compatible_formats.h"
+#include "video_core/surface.h"
+#include "video_core/texture_cache/formatter.h"
+#include "video_core/texture_cache/image_info.h"
+#include "video_core/texture_cache/image_view_base.h"
+#include "video_core/texture_cache/image_view_info.h"
+#include "video_core/texture_cache/types.h"
+
+namespace VideoCommon {
+
+ImageViewBase::ImageViewBase(const ImageViewInfo& info, const ImageInfo& image_info,
+ ImageId image_id_)
+ : image_id{image_id_}, format{info.format}, type{info.type}, range{info.range},
+ size{
+ .width = std::max(image_info.size.width >> range.base.level, 1u),
+ .height = std::max(image_info.size.height >> range.base.level, 1u),
+ .depth = std::max(image_info.size.depth >> range.base.level, 1u),
+ } {
+ ASSERT_MSG(VideoCore::Surface::IsViewCompatible(image_info.format, info.format),
+ "Image view format {} is incompatible with image format {}", info.format,
+ image_info.format);
+ const bool is_async = Settings::values.use_asynchronous_gpu_emulation.GetValue();
+ if (image_info.type == ImageType::Linear && is_async) {
+ flags |= ImageViewFlagBits::PreemtiveDownload;
+ }
+ if (image_info.type == ImageType::e3D && info.type != ImageViewType::e3D) {
+ flags |= ImageViewFlagBits::Slice;
+ }
+}
+
+ImageViewBase::ImageViewBase(const NullImageParams&) {}
+
+} // namespace VideoCommon
diff --git a/src/video_core/texture_cache/image_view_base.h b/src/video_core/texture_cache/image_view_base.h
new file mode 100644
index 000000000..73954167e
--- /dev/null
+++ b/src/video_core/texture_cache/image_view_base.h
@@ -0,0 +1,47 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "common/common_funcs.h"
+#include "video_core/surface.h"
+#include "video_core/texture_cache/types.h"
+
+namespace VideoCommon {
+
+using VideoCore::Surface::PixelFormat;
+
+struct ImageViewInfo;
+struct ImageInfo;
+
+struct NullImageParams {};
+
+enum class ImageViewFlagBits : u16 {
+ PreemtiveDownload = 1 << 0,
+ Strong = 1 << 1,
+ Slice = 1 << 2,
+};
+DECLARE_ENUM_FLAG_OPERATORS(ImageViewFlagBits)
+
+struct ImageViewBase {
+ explicit ImageViewBase(const ImageViewInfo& info, const ImageInfo& image_info,
+ ImageId image_id);
+ explicit ImageViewBase(const NullImageParams&);
+
+ [[nodiscard]] bool IsBuffer() const noexcept {
+ return type == ImageViewType::Buffer;
+ }
+
+ ImageId image_id{};
+ PixelFormat format{};
+ ImageViewType type{};
+ SubresourceRange range;
+ Extent3D size{0, 0, 0};
+ ImageViewFlagBits flags{};
+
+ u64 invalidation_tick = 0;
+ u64 modification_tick = 0;
+};
+
+} // namespace VideoCommon
diff --git a/src/video_core/texture_cache/image_view_info.cpp b/src/video_core/texture_cache/image_view_info.cpp
new file mode 100644
index 000000000..faf5b151f
--- /dev/null
+++ b/src/video_core/texture_cache/image_view_info.cpp
@@ -0,0 +1,88 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <limits>
+
+#include "common/assert.h"
+#include "video_core/texture_cache/image_view_info.h"
+#include "video_core/texture_cache/texture_cache.h"
+#include "video_core/texture_cache/types.h"
+#include "video_core/textures/texture.h"
+
+namespace VideoCommon {
+
+namespace {
+
+constexpr u8 RENDER_TARGET_SWIZZLE = std::numeric_limits<u8>::max();
+
+[[nodiscard]] u8 CastSwizzle(SwizzleSource source) {
+ const u8 casted = static_cast<u8>(source);
+ ASSERT(static_cast<SwizzleSource>(casted) == source);
+ return casted;
+}
+
+} // Anonymous namespace
+
+ImageViewInfo::ImageViewInfo(const TICEntry& config, s32 base_layer) noexcept
+ : format{PixelFormatFromTIC(config)}, x_source{CastSwizzle(config.x_source)},
+ y_source{CastSwizzle(config.y_source)}, z_source{CastSwizzle(config.z_source)},
+ w_source{CastSwizzle(config.w_source)} {
+ range.base = SubresourceBase{
+ .level = static_cast<s32>(config.res_min_mip_level),
+ .layer = base_layer,
+ };
+ range.extent.levels = config.res_max_mip_level - config.res_min_mip_level + 1;
+
+ switch (config.texture_type) {
+ case TextureType::Texture1D:
+ ASSERT(config.Height() == 1);
+ ASSERT(config.Depth() == 1);
+ type = ImageViewType::e1D;
+ break;
+ case TextureType::Texture2D:
+ case TextureType::Texture2DNoMipmap:
+ ASSERT(config.Depth() == 1);
+ type = config.normalized_coords ? ImageViewType::e2D : ImageViewType::Rect;
+ break;
+ case TextureType::Texture3D:
+ type = ImageViewType::e3D;
+ break;
+ case TextureType::TextureCubemap:
+ ASSERT(config.Depth() == 1);
+ type = ImageViewType::Cube;
+ range.extent.layers = 6;
+ break;
+ case TextureType::Texture1DArray:
+ type = ImageViewType::e1DArray;
+ range.extent.layers = config.Depth();
+ break;
+ case TextureType::Texture2DArray:
+ type = ImageViewType::e2DArray;
+ range.extent.layers = config.Depth();
+ break;
+ case TextureType::Texture1DBuffer:
+ type = ImageViewType::Buffer;
+ break;
+ case TextureType::TextureCubeArray:
+ type = ImageViewType::CubeArray;
+ range.extent.layers = config.Depth() * 6;
+ break;
+ default:
+ UNREACHABLE_MSG("Invalid texture_type={}", static_cast<int>(config.texture_type.Value()));
+ break;
+ }
+}
+
+ImageViewInfo::ImageViewInfo(ImageViewType type_, PixelFormat format_,
+ SubresourceRange range_) noexcept
+ : type{type_}, format{format_}, range{range_}, x_source{RENDER_TARGET_SWIZZLE},
+ y_source{RENDER_TARGET_SWIZZLE}, z_source{RENDER_TARGET_SWIZZLE},
+ w_source{RENDER_TARGET_SWIZZLE} {}
+
+bool ImageViewInfo::IsRenderTarget() const noexcept {
+ return x_source == RENDER_TARGET_SWIZZLE && y_source == RENDER_TARGET_SWIZZLE &&
+ z_source == RENDER_TARGET_SWIZZLE && w_source == RENDER_TARGET_SWIZZLE;
+}
+
+} // namespace VideoCommon
diff --git a/src/video_core/texture_cache/image_view_info.h b/src/video_core/texture_cache/image_view_info.h
new file mode 100644
index 000000000..0c1f99117
--- /dev/null
+++ b/src/video_core/texture_cache/image_view_info.h
@@ -0,0 +1,50 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include <type_traits>
+
+#include "video_core/surface.h"
+#include "video_core/texture_cache/types.h"
+#include "video_core/textures/texture.h"
+
+namespace VideoCommon {
+
+using Tegra::Texture::SwizzleSource;
+using Tegra::Texture::TICEntry;
+using VideoCore::Surface::PixelFormat;
+
+/// Properties used to determine a image view
+struct ImageViewInfo {
+ explicit ImageViewInfo() noexcept = default;
+ explicit ImageViewInfo(const TICEntry& config, s32 base_layer) noexcept;
+ explicit ImageViewInfo(ImageViewType type, PixelFormat format,
+ SubresourceRange range = {}) noexcept;
+
+ auto operator<=>(const ImageViewInfo&) const noexcept = default;
+
+ [[nodiscard]] bool IsRenderTarget() const noexcept;
+
+ [[nodiscard]] std::array<SwizzleSource, 4> Swizzle() const noexcept {
+ return std::array{
+ static_cast<SwizzleSource>(x_source),
+ static_cast<SwizzleSource>(y_source),
+ static_cast<SwizzleSource>(z_source),
+ static_cast<SwizzleSource>(w_source),
+ };
+ }
+
+ ImageViewType type{};
+ PixelFormat format{};
+ SubresourceRange range;
+ u8 x_source = static_cast<u8>(SwizzleSource::R);
+ u8 y_source = static_cast<u8>(SwizzleSource::G);
+ u8 z_source = static_cast<u8>(SwizzleSource::B);
+ u8 w_source = static_cast<u8>(SwizzleSource::A);
+};
+static_assert(std::has_unique_object_representations_v<ImageViewInfo>);
+
+} // namespace VideoCommon
diff --git a/src/video_core/texture_cache/render_targets.h b/src/video_core/texture_cache/render_targets.h
new file mode 100644
index 000000000..9b9544b07
--- /dev/null
+++ b/src/video_core/texture_cache/render_targets.h
@@ -0,0 +1,51 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <algorithm>
+#include <span>
+#include <utility>
+
+#include "common/bit_cast.h"
+#include "video_core/texture_cache/types.h"
+
+namespace VideoCommon {
+
+/// Framebuffer properties used to lookup a framebuffer
+struct RenderTargets {
+ constexpr auto operator<=>(const RenderTargets&) const noexcept = default;
+
+ constexpr bool Contains(std::span<const ImageViewId> elements) const noexcept {
+ const auto contains = [elements](ImageViewId item) {
+ return std::ranges::find(elements, item) != elements.end();
+ };
+ return std::ranges::any_of(color_buffer_ids, contains) || contains(depth_buffer_id);
+ }
+
+ std::array<ImageViewId, NUM_RT> color_buffer_ids;
+ ImageViewId depth_buffer_id;
+ std::array<u8, NUM_RT> draw_buffers{};
+ Extent2D size;
+};
+
+} // namespace VideoCommon
+
+namespace std {
+
+template <>
+struct hash<VideoCommon::RenderTargets> {
+ size_t operator()(const VideoCommon::RenderTargets& rt) const noexcept {
+ using VideoCommon::ImageViewId;
+ size_t value = std::hash<ImageViewId>{}(rt.depth_buffer_id);
+ for (const ImageViewId color_buffer_id : rt.color_buffer_ids) {
+ value ^= std::hash<ImageViewId>{}(color_buffer_id);
+ }
+ value ^= Common::BitCast<u64>(rt.draw_buffers);
+ value ^= Common::BitCast<u64>(rt.size);
+ return value;
+ }
+};
+
+} // namespace std
diff --git a/src/video_core/texture_cache/samples_helper.h b/src/video_core/texture_cache/samples_helper.h
new file mode 100644
index 000000000..04539a43c
--- /dev/null
+++ b/src/video_core/texture_cache/samples_helper.h
@@ -0,0 +1,55 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <utility>
+
+#include "common/assert.h"
+#include "video_core/textures/texture.h"
+
+namespace VideoCommon {
+
+[[nodiscard]] inline std::pair<int, int> SamplesLog2(int num_samples) {
+ switch (num_samples) {
+ case 1:
+ return {0, 0};
+ case 2:
+ return {1, 0};
+ case 4:
+ return {1, 1};
+ case 8:
+ return {2, 1};
+ case 16:
+ return {2, 2};
+ }
+ UNREACHABLE_MSG("Invalid number of samples={}", num_samples);
+ return {1, 1};
+}
+
+[[nodiscard]] inline int NumSamples(Tegra::Texture::MsaaMode msaa_mode) {
+ using Tegra::Texture::MsaaMode;
+ switch (msaa_mode) {
+ case MsaaMode::Msaa1x1:
+ return 1;
+ case MsaaMode::Msaa2x1:
+ case MsaaMode::Msaa2x1_D3D:
+ return 2;
+ case MsaaMode::Msaa2x2:
+ case MsaaMode::Msaa2x2_VC4:
+ case MsaaMode::Msaa2x2_VC12:
+ return 4;
+ case MsaaMode::Msaa4x2:
+ case MsaaMode::Msaa4x2_D3D:
+ case MsaaMode::Msaa4x2_VC8:
+ case MsaaMode::Msaa4x2_VC24:
+ return 8;
+ case MsaaMode::Msaa4x4:
+ return 16;
+ }
+ UNREACHABLE_MSG("Invalid MSAA mode={}", static_cast<int>(msaa_mode));
+ return 1;
+}
+
+} // namespace VideoCommon
diff --git a/src/video_core/texture_cache/slot_vector.h b/src/video_core/texture_cache/slot_vector.h
new file mode 100644
index 000000000..eae3be6ea
--- /dev/null
+++ b/src/video_core/texture_cache/slot_vector.h
@@ -0,0 +1,156 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include <concepts>
+#include <numeric>
+#include <type_traits>
+#include <utility>
+#include <vector>
+
+#include "common/assert.h"
+#include "common/common_types.h"
+
+namespace VideoCommon {
+
+struct SlotId {
+ static constexpr u32 INVALID_INDEX = std::numeric_limits<u32>::max();
+
+ constexpr auto operator<=>(const SlotId&) const noexcept = default;
+
+ constexpr explicit operator bool() const noexcept {
+ return index != INVALID_INDEX;
+ }
+
+ u32 index = INVALID_INDEX;
+};
+
+template <class T>
+requires std::is_nothrow_move_assignable_v<T>&&
+ std::is_nothrow_move_constructible_v<T> class SlotVector {
+public:
+ ~SlotVector() noexcept {
+ size_t index = 0;
+ for (u64 bits : stored_bitset) {
+ for (size_t bit = 0; bits; ++bit, bits >>= 1) {
+ if ((bits & 1) != 0) {
+ values[index + bit].object.~T();
+ }
+ }
+ index += 64;
+ }
+ delete[] values;
+ }
+
+ [[nodiscard]] T& operator[](SlotId id) noexcept {
+ ValidateIndex(id);
+ return values[id.index].object;
+ }
+
+ [[nodiscard]] const T& operator[](SlotId id) const noexcept {
+ ValidateIndex(id);
+ return values[id.index].object;
+ }
+
+ template <typename... Args>
+ [[nodiscard]] SlotId insert(Args&&... args) noexcept {
+ const u32 index = FreeValueIndex();
+ new (&values[index].object) T(std::forward<Args>(args)...);
+ SetStorageBit(index);
+
+ return SlotId{index};
+ }
+
+ void erase(SlotId id) noexcept {
+ values[id.index].object.~T();
+ free_list.push_back(id.index);
+ ResetStorageBit(id.index);
+ }
+
+private:
+ struct NonTrivialDummy {
+ NonTrivialDummy() noexcept {}
+ };
+
+ union Entry {
+ Entry() noexcept : dummy{} {}
+ ~Entry() noexcept {}
+
+ NonTrivialDummy dummy;
+ T object;
+ };
+
+ void SetStorageBit(u32 index) noexcept {
+ stored_bitset[index / 64] |= u64(1) << (index % 64);
+ }
+
+ void ResetStorageBit(u32 index) noexcept {
+ stored_bitset[index / 64] &= ~(u64(1) << (index % 64));
+ }
+
+ bool ReadStorageBit(u32 index) noexcept {
+ return ((stored_bitset[index / 64] >> (index % 64)) & 1) != 0;
+ }
+
+ void ValidateIndex(SlotId id) const noexcept {
+ DEBUG_ASSERT(id);
+ DEBUG_ASSERT(id.index / 64 < stored_bitset.size());
+ DEBUG_ASSERT(((stored_bitset[id.index / 64] >> (id.index % 64)) & 1) != 0);
+ }
+
+ [[nodiscard]] u32 FreeValueIndex() noexcept {
+ if (free_list.empty()) {
+ Reserve(values_capacity ? (values_capacity << 1) : 1);
+ }
+ const u32 free_index = free_list.back();
+ free_list.pop_back();
+ return free_index;
+ }
+
+ void Reserve(size_t new_capacity) noexcept {
+ Entry* const new_values = new Entry[new_capacity];
+ size_t index = 0;
+ for (u64 bits : stored_bitset) {
+ for (size_t bit = 0; bits; ++bit, bits >>= 1) {
+ const size_t i = index + bit;
+ if ((bits & 1) == 0) {
+ continue;
+ }
+ T& old_value = values[i].object;
+ new (&new_values[i].object) T(std::move(old_value));
+ old_value.~T();
+ }
+ index += 64;
+ }
+
+ stored_bitset.resize((new_capacity + 63) / 64);
+
+ const size_t old_free_size = free_list.size();
+ free_list.resize(old_free_size + (new_capacity - values_capacity));
+ std::iota(free_list.begin() + old_free_size, free_list.end(),
+ static_cast<u32>(values_capacity));
+
+ delete[] values;
+ values = new_values;
+ values_capacity = new_capacity;
+ }
+
+ Entry* values = nullptr;
+ size_t values_capacity = 0;
+ size_t values_size = 0;
+
+ std::vector<u64> stored_bitset;
+ std::vector<u32> free_list;
+};
+
+} // namespace VideoCommon
+
+template <>
+struct std::hash<VideoCommon::SlotId> {
+ size_t operator()(const VideoCommon::SlotId& id) const noexcept {
+ return std::hash<u32>{}(id.index);
+ }
+};
diff --git a/src/video_core/texture_cache/surface_base.cpp b/src/video_core/texture_cache/surface_base.cpp
deleted file mode 100644
index b44c09d71..000000000
--- a/src/video_core/texture_cache/surface_base.cpp
+++ /dev/null
@@ -1,298 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include "common/algorithm.h"
-#include "common/assert.h"
-#include "common/common_types.h"
-#include "common/microprofile.h"
-#include "video_core/memory_manager.h"
-#include "video_core/texture_cache/surface_base.h"
-#include "video_core/texture_cache/surface_params.h"
-#include "video_core/textures/convert.h"
-
-namespace VideoCommon {
-
-MICROPROFILE_DEFINE(GPU_Load_Texture, "GPU", "Texture Load", MP_RGB(128, 192, 128));
-MICROPROFILE_DEFINE(GPU_Flush_Texture, "GPU", "Texture Flush", MP_RGB(128, 192, 128));
-
-using Tegra::Texture::ConvertFromGuestToHost;
-using VideoCore::MortonSwizzleMode;
-using VideoCore::Surface::IsPixelFormatASTC;
-using VideoCore::Surface::PixelFormat;
-
-StagingCache::StagingCache() = default;
-
-StagingCache::~StagingCache() = default;
-
-SurfaceBaseImpl::SurfaceBaseImpl(GPUVAddr gpu_addr, const SurfaceParams& params,
- bool is_astc_supported)
- : params{params}, gpu_addr{gpu_addr}, mipmap_sizes(params.num_levels),
- mipmap_offsets(params.num_levels) {
- is_converted = IsPixelFormatASTC(params.pixel_format) && !is_astc_supported;
- host_memory_size = params.GetHostSizeInBytes(is_converted);
-
- std::size_t offset = 0;
- for (u32 level = 0; level < params.num_levels; ++level) {
- const std::size_t mipmap_size{params.GetGuestMipmapSize(level)};
- mipmap_sizes[level] = mipmap_size;
- mipmap_offsets[level] = offset;
- offset += mipmap_size;
- }
- layer_size = offset;
- if (params.is_layered) {
- if (params.is_tiled) {
- layer_size =
- SurfaceParams::AlignLayered(layer_size, params.block_height, params.block_depth);
- }
- guest_memory_size = layer_size * params.depth;
- } else {
- guest_memory_size = layer_size;
- }
-}
-
-MatchTopologyResult SurfaceBaseImpl::MatchesTopology(const SurfaceParams& rhs) const {
- const u32 src_bpp{params.GetBytesPerPixel()};
- const u32 dst_bpp{rhs.GetBytesPerPixel()};
- const bool ib1 = params.IsBuffer();
- const bool ib2 = rhs.IsBuffer();
- if (std::tie(src_bpp, params.is_tiled, ib1) == std::tie(dst_bpp, rhs.is_tiled, ib2)) {
- const bool cb1 = params.IsCompressed();
- const bool cb2 = rhs.IsCompressed();
- if (cb1 == cb2) {
- return MatchTopologyResult::FullMatch;
- }
- return MatchTopologyResult::CompressUnmatch;
- }
- return MatchTopologyResult::None;
-}
-
-MatchStructureResult SurfaceBaseImpl::MatchesStructure(const SurfaceParams& rhs) const {
- // Buffer surface Check
- if (params.IsBuffer()) {
- const std::size_t wd1 = params.width * params.GetBytesPerPixel();
- const std::size_t wd2 = rhs.width * rhs.GetBytesPerPixel();
- if (wd1 == wd2) {
- return MatchStructureResult::FullMatch;
- }
- return MatchStructureResult::None;
- }
-
- // Linear Surface check
- if (!params.is_tiled) {
- if (std::tie(params.height, params.pitch) == std::tie(rhs.height, rhs.pitch)) {
- if (params.width == rhs.width) {
- return MatchStructureResult::FullMatch;
- } else {
- return MatchStructureResult::SemiMatch;
- }
- }
- return MatchStructureResult::None;
- }
-
- // Tiled Surface check
- if (std::tie(params.depth, params.block_width, params.block_height, params.block_depth,
- params.tile_width_spacing, params.num_levels) ==
- std::tie(rhs.depth, rhs.block_width, rhs.block_height, rhs.block_depth,
- rhs.tile_width_spacing, rhs.num_levels)) {
- if (std::tie(params.width, params.height) == std::tie(rhs.width, rhs.height)) {
- return MatchStructureResult::FullMatch;
- }
- const u32 ws = SurfaceParams::ConvertWidth(rhs.GetBlockAlignedWidth(), params.pixel_format,
- rhs.pixel_format);
- const u32 hs =
- SurfaceParams::ConvertHeight(rhs.height, params.pixel_format, rhs.pixel_format);
- const u32 w1 = params.GetBlockAlignedWidth();
- if (std::tie(w1, params.height) == std::tie(ws, hs)) {
- return MatchStructureResult::SemiMatch;
- }
- }
- return MatchStructureResult::None;
-}
-
-std::optional<std::pair<u32, u32>> SurfaceBaseImpl::GetLayerMipmap(
- const GPUVAddr candidate_gpu_addr) const {
- if (gpu_addr == candidate_gpu_addr) {
- return {{0, 0}};
- }
-
- if (candidate_gpu_addr < gpu_addr) {
- return std::nullopt;
- }
-
- const auto relative_address{static_cast<GPUVAddr>(candidate_gpu_addr - gpu_addr)};
- const auto layer{static_cast<u32>(relative_address / layer_size)};
- if (layer >= params.depth) {
- return std::nullopt;
- }
-
- const GPUVAddr mipmap_address = relative_address - layer_size * layer;
- const auto mipmap_it =
- Common::BinaryFind(mipmap_offsets.begin(), mipmap_offsets.end(), mipmap_address);
- if (mipmap_it == mipmap_offsets.end()) {
- return std::nullopt;
- }
-
- const auto level{static_cast<u32>(std::distance(mipmap_offsets.begin(), mipmap_it))};
- return std::make_pair(layer, level);
-}
-
-std::vector<CopyParams> SurfaceBaseImpl::BreakDownLayered(const SurfaceParams& in_params) const {
- const u32 layers{params.depth};
- const u32 mipmaps{params.num_levels};
- std::vector<CopyParams> result;
- result.reserve(static_cast<std::size_t>(layers) * static_cast<std::size_t>(mipmaps));
-
- for (u32 layer = 0; layer < layers; layer++) {
- for (u32 level = 0; level < mipmaps; level++) {
- const u32 width = SurfaceParams::IntersectWidth(params, in_params, level, level);
- const u32 height = SurfaceParams::IntersectHeight(params, in_params, level, level);
- result.emplace_back(0, 0, layer, 0, 0, layer, level, level, width, height, 1);
- }
- }
- return result;
-}
-
-std::vector<CopyParams> SurfaceBaseImpl::BreakDownNonLayered(const SurfaceParams& in_params) const {
- const u32 mipmaps{params.num_levels};
- std::vector<CopyParams> result;
- result.reserve(mipmaps);
-
- for (u32 level = 0; level < mipmaps; level++) {
- const u32 width = SurfaceParams::IntersectWidth(params, in_params, level, level);
- const u32 height = SurfaceParams::IntersectHeight(params, in_params, level, level);
- const u32 depth{std::min(params.GetMipDepth(level), in_params.GetMipDepth(level))};
- result.emplace_back(width, height, depth, level);
- }
- return result;
-}
-
-void SurfaceBaseImpl::SwizzleFunc(MortonSwizzleMode mode, u8* memory, const SurfaceParams& params,
- u8* buffer, u32 level) {
- const u32 width{params.GetMipWidth(level)};
- const u32 height{params.GetMipHeight(level)};
- const u32 block_height{params.GetMipBlockHeight(level)};
- const u32 block_depth{params.GetMipBlockDepth(level)};
-
- std::size_t guest_offset{mipmap_offsets[level]};
- if (params.is_layered) {
- std::size_t host_offset = 0;
- const std::size_t guest_stride = layer_size;
- const std::size_t host_stride = params.GetHostLayerSize(level);
- for (u32 layer = 0; layer < params.depth; ++layer) {
- MortonSwizzle(mode, params.pixel_format, width, block_height, height, block_depth, 1,
- params.tile_width_spacing, buffer + host_offset, memory + guest_offset);
- guest_offset += guest_stride;
- host_offset += host_stride;
- }
- } else {
- MortonSwizzle(mode, params.pixel_format, width, block_height, height, block_depth,
- params.GetMipDepth(level), params.tile_width_spacing, buffer,
- memory + guest_offset);
- }
-}
-
-void SurfaceBaseImpl::LoadBuffer(Tegra::MemoryManager& memory_manager,
- StagingCache& staging_cache) {
- MICROPROFILE_SCOPE(GPU_Load_Texture);
- auto& staging_buffer = staging_cache.GetBuffer(0);
- u8* host_ptr;
- // Use an extra temporal buffer
- auto& tmp_buffer = staging_cache.GetBuffer(1);
- tmp_buffer.resize(guest_memory_size);
- host_ptr = tmp_buffer.data();
- memory_manager.ReadBlockUnsafe(gpu_addr, host_ptr, guest_memory_size);
-
- if (params.is_tiled) {
- ASSERT_MSG(params.block_width == 0, "Block width is defined as {} on texture target {}",
- params.block_width, static_cast<u32>(params.target));
- for (u32 level = 0; level < params.num_levels; ++level) {
- const std::size_t host_offset{params.GetHostMipmapLevelOffset(level, false)};
- SwizzleFunc(MortonSwizzleMode::MortonToLinear, host_ptr, params,
- staging_buffer.data() + host_offset, level);
- }
- } else {
- ASSERT_MSG(params.num_levels == 1, "Linear mipmap loading is not implemented");
- const u32 bpp{params.GetBytesPerPixel()};
- const u32 block_width{params.GetDefaultBlockWidth()};
- const u32 block_height{params.GetDefaultBlockHeight()};
- const u32 width{(params.width + block_width - 1) / block_width};
- const u32 height{(params.height + block_height - 1) / block_height};
- const u32 copy_size{width * bpp};
- if (params.pitch == copy_size) {
- std::memcpy(staging_buffer.data(), host_ptr, params.GetHostSizeInBytes(false));
- } else {
- const u8* start{host_ptr};
- u8* write_to{staging_buffer.data()};
- for (u32 h = height; h > 0; --h) {
- std::memcpy(write_to, start, copy_size);
- start += params.pitch;
- write_to += copy_size;
- }
- }
- }
-
- if (!is_converted && params.pixel_format != PixelFormat::S8_UINT_D24_UNORM) {
- return;
- }
-
- for (u32 level = params.num_levels; level--;) {
- const std::size_t in_host_offset{params.GetHostMipmapLevelOffset(level, false)};
- const std::size_t out_host_offset{params.GetHostMipmapLevelOffset(level, is_converted)};
- u8* const in_buffer = staging_buffer.data() + in_host_offset;
- u8* const out_buffer = staging_buffer.data() + out_host_offset;
- ConvertFromGuestToHost(in_buffer, out_buffer, params.pixel_format,
- params.GetMipWidth(level), params.GetMipHeight(level),
- params.GetMipDepth(level), true, true);
- }
-}
-
-void SurfaceBaseImpl::FlushBuffer(Tegra::MemoryManager& memory_manager,
- StagingCache& staging_cache) {
- MICROPROFILE_SCOPE(GPU_Flush_Texture);
- auto& staging_buffer = staging_cache.GetBuffer(0);
- u8* host_ptr;
-
- // Use an extra temporal buffer
- auto& tmp_buffer = staging_cache.GetBuffer(1);
- tmp_buffer.resize(guest_memory_size);
- host_ptr = tmp_buffer.data();
-
- if (params.target == SurfaceTarget::Texture3D) {
- // Special case for 3D texture segments
- memory_manager.ReadBlockUnsafe(gpu_addr, host_ptr, guest_memory_size);
- }
-
- if (params.is_tiled) {
- ASSERT_MSG(params.block_width == 0, "Block width is defined as {}", params.block_width);
- for (u32 level = 0; level < params.num_levels; ++level) {
- const std::size_t host_offset{params.GetHostMipmapLevelOffset(level, false)};
- SwizzleFunc(MortonSwizzleMode::LinearToMorton, host_ptr, params,
- staging_buffer.data() + host_offset, level);
- }
- } else if (params.IsBuffer()) {
- // Buffers don't have pitch or any fancy layout property. We can just memcpy them to guest
- // memory.
- std::memcpy(host_ptr, staging_buffer.data(), guest_memory_size);
- } else {
- ASSERT(params.target == SurfaceTarget::Texture2D);
- ASSERT(params.num_levels == 1);
-
- const u32 bpp{params.GetBytesPerPixel()};
- const u32 copy_size{params.width * bpp};
- if (params.pitch == copy_size) {
- std::memcpy(host_ptr, staging_buffer.data(), guest_memory_size);
- } else {
- u8* start{host_ptr};
- const u8* read_to{staging_buffer.data()};
- for (u32 h = params.height; h > 0; --h) {
- std::memcpy(start, read_to, copy_size);
- start += params.pitch;
- read_to += copy_size;
- }
- }
- }
- memory_manager.WriteBlockUnsafe(gpu_addr, host_ptr, guest_memory_size);
-}
-
-} // namespace VideoCommon
diff --git a/src/video_core/texture_cache/surface_base.h b/src/video_core/texture_cache/surface_base.h
deleted file mode 100644
index 173f2edba..000000000
--- a/src/video_core/texture_cache/surface_base.h
+++ /dev/null
@@ -1,333 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <optional>
-#include <tuple>
-#include <unordered_map>
-#include <vector>
-
-#include "common/common_types.h"
-#include "video_core/gpu.h"
-#include "video_core/morton.h"
-#include "video_core/texture_cache/copy_params.h"
-#include "video_core/texture_cache/surface_params.h"
-#include "video_core/texture_cache/surface_view.h"
-
-namespace Tegra {
-class MemoryManager;
-}
-
-namespace VideoCommon {
-
-using VideoCore::MortonSwizzleMode;
-using VideoCore::Surface::SurfaceTarget;
-
-enum class MatchStructureResult : u32 {
- FullMatch = 0,
- SemiMatch = 1,
- None = 2,
-};
-
-enum class MatchTopologyResult : u32 {
- FullMatch = 0,
- CompressUnmatch = 1,
- None = 2,
-};
-
-class StagingCache {
-public:
- explicit StagingCache();
- ~StagingCache();
-
- std::vector<u8>& GetBuffer(std::size_t index) {
- return staging_buffer[index];
- }
-
- const std::vector<u8>& GetBuffer(std::size_t index) const {
- return staging_buffer[index];
- }
-
- void SetSize(std::size_t size) {
- staging_buffer.resize(size);
- }
-
-private:
- std::vector<std::vector<u8>> staging_buffer;
-};
-
-class SurfaceBaseImpl {
-public:
- void LoadBuffer(Tegra::MemoryManager& memory_manager, StagingCache& staging_cache);
-
- void FlushBuffer(Tegra::MemoryManager& memory_manager, StagingCache& staging_cache);
-
- GPUVAddr GetGpuAddr() const {
- return gpu_addr;
- }
-
- bool Overlaps(const VAddr start, const VAddr end) const {
- return (cpu_addr < end) && (cpu_addr_end > start);
- }
-
- bool IsInside(const GPUVAddr other_start, const GPUVAddr other_end) const {
- const GPUVAddr gpu_addr_end = gpu_addr + guest_memory_size;
- return gpu_addr <= other_start && other_end <= gpu_addr_end;
- }
-
- // Use only when recycling a surface
- void SetGpuAddr(const GPUVAddr new_addr) {
- gpu_addr = new_addr;
- }
-
- VAddr GetCpuAddr() const {
- return cpu_addr;
- }
-
- VAddr GetCpuAddrEnd() const {
- return cpu_addr_end;
- }
-
- void SetCpuAddr(const VAddr new_addr) {
- cpu_addr = new_addr;
- cpu_addr_end = new_addr + guest_memory_size;
- }
-
- const SurfaceParams& GetSurfaceParams() const {
- return params;
- }
-
- std::size_t GetSizeInBytes() const {
- return guest_memory_size;
- }
-
- std::size_t GetHostSizeInBytes() const {
- return host_memory_size;
- }
-
- std::size_t GetMipmapSize(const u32 level) const {
- return mipmap_sizes[level];
- }
-
- bool IsLinear() const {
- return !params.is_tiled;
- }
-
- bool IsConverted() const {
- return is_converted;
- }
-
- bool MatchFormat(VideoCore::Surface::PixelFormat pixel_format) const {
- return params.pixel_format == pixel_format;
- }
-
- VideoCore::Surface::PixelFormat GetFormat() const {
- return params.pixel_format;
- }
-
- bool MatchTarget(VideoCore::Surface::SurfaceTarget target) const {
- return params.target == target;
- }
-
- MatchTopologyResult MatchesTopology(const SurfaceParams& rhs) const;
-
- MatchStructureResult MatchesStructure(const SurfaceParams& rhs) const;
-
- bool MatchesSubTexture(const SurfaceParams& rhs, const GPUVAddr other_gpu_addr) const {
- return std::tie(gpu_addr, params.target, params.num_levels) ==
- std::tie(other_gpu_addr, rhs.target, rhs.num_levels) &&
- params.target == SurfaceTarget::Texture2D && params.num_levels == 1;
- }
-
- std::optional<std::pair<u32, u32>> GetLayerMipmap(const GPUVAddr candidate_gpu_addr) const;
-
- std::vector<CopyParams> BreakDown(const SurfaceParams& in_params) const {
- return params.is_layered ? BreakDownLayered(in_params) : BreakDownNonLayered(in_params);
- }
-
-protected:
- explicit SurfaceBaseImpl(GPUVAddr gpu_addr, const SurfaceParams& params,
- bool is_astc_supported);
- ~SurfaceBaseImpl() = default;
-
- virtual void DecorateSurfaceName() = 0;
-
- const SurfaceParams params;
- std::size_t layer_size;
- std::size_t guest_memory_size;
- std::size_t host_memory_size;
- GPUVAddr gpu_addr{};
- VAddr cpu_addr{};
- VAddr cpu_addr_end{};
- bool is_converted{};
-
- std::vector<std::size_t> mipmap_sizes;
- std::vector<std::size_t> mipmap_offsets;
-
-private:
- void SwizzleFunc(MortonSwizzleMode mode, u8* memory, const SurfaceParams& params, u8* buffer,
- u32 level);
-
- std::vector<CopyParams> BreakDownLayered(const SurfaceParams& in_params) const;
-
- std::vector<CopyParams> BreakDownNonLayered(const SurfaceParams& in_params) const;
-};
-
-template <typename TView>
-class SurfaceBase : public SurfaceBaseImpl {
-public:
- virtual void UploadTexture(const std::vector<u8>& staging_buffer) = 0;
-
- virtual void DownloadTexture(std::vector<u8>& staging_buffer) = 0;
-
- void MarkAsModified(bool is_modified_, u64 tick) {
- is_modified = is_modified_ || is_target;
- modification_tick = tick;
- }
-
- void MarkAsRenderTarget(bool is_target_, u32 index_) {
- is_target = is_target_;
- index = index_;
- }
-
- void SetMemoryMarked(bool is_memory_marked_) {
- is_memory_marked = is_memory_marked_;
- }
-
- bool IsMemoryMarked() const {
- return is_memory_marked;
- }
-
- void SetSyncPending(bool is_sync_pending_) {
- is_sync_pending = is_sync_pending_;
- }
-
- bool IsSyncPending() const {
- return is_sync_pending;
- }
-
- void MarkAsPicked(bool is_picked_) {
- is_picked = is_picked_;
- }
-
- bool IsModified() const {
- return is_modified;
- }
-
- bool IsProtected() const {
- // Only 3D slices are to be protected
- return is_target && params.target == SurfaceTarget::Texture3D;
- }
-
- bool IsRenderTarget() const {
- return is_target;
- }
-
- u32 GetRenderTarget() const {
- return index;
- }
-
- bool IsRegistered() const {
- return is_registered;
- }
-
- bool IsPicked() const {
- return is_picked;
- }
-
- void MarkAsRegistered(bool is_reg) {
- is_registered = is_reg;
- }
-
- u64 GetModificationTick() const {
- return modification_tick;
- }
-
- TView EmplaceOverview(const SurfaceParams& overview_params) {
- const u32 num_layers{(params.is_layered && !overview_params.is_layered) ? 1 : params.depth};
- return GetView(ViewParams(overview_params.target, 0, num_layers, 0, params.num_levels));
- }
-
- TView Emplace3DView(u32 slice, u32 depth, u32 base_level, u32 num_levels) {
- return GetView(ViewParams(VideoCore::Surface::SurfaceTarget::Texture3D, slice, depth,
- base_level, num_levels));
- }
-
- std::optional<TView> EmplaceIrregularView(const SurfaceParams& view_params,
- const GPUVAddr view_addr,
- const std::size_t candidate_size, const u32 mipmap,
- const u32 layer) {
- const auto layer_mipmap{GetLayerMipmap(view_addr + candidate_size)};
- if (!layer_mipmap) {
- return {};
- }
- const auto [end_layer, end_mipmap] = *layer_mipmap;
- if (layer != end_layer) {
- if (mipmap == 0 && end_mipmap == 0) {
- return GetView(ViewParams(view_params.target, layer, end_layer - layer, 0, 1));
- }
- return {};
- } else {
- return GetView(ViewParams(view_params.target, layer, 1, mipmap, end_mipmap - mipmap));
- }
- }
-
- std::optional<TView> EmplaceView(const SurfaceParams& view_params, const GPUVAddr view_addr,
- const std::size_t candidate_size) {
- if (params.target == SurfaceTarget::Texture3D ||
- view_params.target == SurfaceTarget::Texture3D ||
- (params.num_levels == 1 && !params.is_layered)) {
- return {};
- }
- const auto layer_mipmap{GetLayerMipmap(view_addr)};
- if (!layer_mipmap) {
- return {};
- }
- const auto [layer, mipmap] = *layer_mipmap;
- if (GetMipmapSize(mipmap) != candidate_size) {
- return EmplaceIrregularView(view_params, view_addr, candidate_size, mipmap, layer);
- }
- return GetView(ViewParams(view_params.target, layer, 1, mipmap, 1));
- }
-
- TView GetMainView() const {
- return main_view;
- }
-
-protected:
- explicit SurfaceBase(const GPUVAddr gpu_addr, const SurfaceParams& params,
- bool is_astc_supported)
- : SurfaceBaseImpl(gpu_addr, params, is_astc_supported) {}
-
- ~SurfaceBase() = default;
-
- virtual TView CreateView(const ViewParams& view_key) = 0;
-
- TView main_view;
- std::unordered_map<ViewParams, TView> views;
-
-private:
- TView GetView(const ViewParams& key) {
- const auto [entry, is_cache_miss] = views.try_emplace(key);
- auto& view{entry->second};
- if (is_cache_miss) {
- view = CreateView(key);
- }
- return view;
- }
-
- static constexpr u32 NO_RT = 0xFFFFFFFF;
-
- bool is_modified{};
- bool is_target{};
- bool is_registered{};
- bool is_picked{};
- bool is_memory_marked{};
- bool is_sync_pending{};
- u32 index{NO_RT};
- u64 modification_tick{};
-};
-
-} // namespace VideoCommon
diff --git a/src/video_core/texture_cache/surface_params.cpp b/src/video_core/texture_cache/surface_params.cpp
deleted file mode 100644
index 13dd16356..000000000
--- a/src/video_core/texture_cache/surface_params.cpp
+++ /dev/null
@@ -1,445 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include <algorithm>
-#include <string>
-#include <tuple>
-
-#include "common/alignment.h"
-#include "common/bit_util.h"
-#include "core/core.h"
-#include "video_core/engines/shader_bytecode.h"
-#include "video_core/surface.h"
-#include "video_core/texture_cache/format_lookup_table.h"
-#include "video_core/texture_cache/surface_params.h"
-
-namespace VideoCommon {
-
-using VideoCore::Surface::PixelFormat;
-using VideoCore::Surface::PixelFormatFromDepthFormat;
-using VideoCore::Surface::PixelFormatFromRenderTargetFormat;
-using VideoCore::Surface::SurfaceTarget;
-using VideoCore::Surface::SurfaceTargetFromTextureType;
-using VideoCore::Surface::SurfaceType;
-
-namespace {
-
-SurfaceTarget TextureTypeToSurfaceTarget(Tegra::Shader::TextureType type, bool is_array) {
- switch (type) {
- case Tegra::Shader::TextureType::Texture1D:
- return is_array ? SurfaceTarget::Texture1DArray : SurfaceTarget::Texture1D;
- case Tegra::Shader::TextureType::Texture2D:
- return is_array ? SurfaceTarget::Texture2DArray : SurfaceTarget::Texture2D;
- case Tegra::Shader::TextureType::Texture3D:
- ASSERT(!is_array);
- return SurfaceTarget::Texture3D;
- case Tegra::Shader::TextureType::TextureCube:
- return is_array ? SurfaceTarget::TextureCubeArray : SurfaceTarget::TextureCubemap;
- default:
- UNREACHABLE();
- return SurfaceTarget::Texture2D;
- }
-}
-
-SurfaceTarget ImageTypeToSurfaceTarget(Tegra::Shader::ImageType type) {
- switch (type) {
- case Tegra::Shader::ImageType::Texture1D:
- return SurfaceTarget::Texture1D;
- case Tegra::Shader::ImageType::TextureBuffer:
- return SurfaceTarget::TextureBuffer;
- case Tegra::Shader::ImageType::Texture1DArray:
- return SurfaceTarget::Texture1DArray;
- case Tegra::Shader::ImageType::Texture2D:
- return SurfaceTarget::Texture2D;
- case Tegra::Shader::ImageType::Texture2DArray:
- return SurfaceTarget::Texture2DArray;
- case Tegra::Shader::ImageType::Texture3D:
- return SurfaceTarget::Texture3D;
- default:
- UNREACHABLE();
- return SurfaceTarget::Texture2D;
- }
-}
-
-constexpr u32 GetMipmapSize(bool uncompressed, u32 mip_size, u32 tile) {
- return uncompressed ? mip_size : std::max(1U, (mip_size + tile - 1) / tile);
-}
-
-} // Anonymous namespace
-
-SurfaceParams SurfaceParams::CreateForTexture(const FormatLookupTable& lookup_table,
- const Tegra::Texture::TICEntry& tic,
- const VideoCommon::Shader::Sampler& entry) {
- SurfaceParams params;
- params.is_tiled = tic.IsTiled();
- params.srgb_conversion = tic.IsSrgbConversionEnabled();
- params.block_width = params.is_tiled ? tic.BlockWidth() : 0;
- params.block_height = params.is_tiled ? tic.BlockHeight() : 0;
- params.block_depth = params.is_tiled ? tic.BlockDepth() : 0;
- params.tile_width_spacing = params.is_tiled ? (1 << tic.tile_width_spacing.Value()) : 1;
- params.pixel_format = lookup_table.GetPixelFormat(
- tic.format, params.srgb_conversion, tic.r_type, tic.g_type, tic.b_type, tic.a_type);
- params.type = GetFormatType(params.pixel_format);
- if (entry.is_shadow && params.type == SurfaceType::ColorTexture) {
- switch (params.pixel_format) {
- case PixelFormat::R16_UNORM:
- case PixelFormat::R16_FLOAT:
- params.pixel_format = PixelFormat::D16_UNORM;
- break;
- case PixelFormat::R32_FLOAT:
- params.pixel_format = PixelFormat::D32_FLOAT;
- break;
- default:
- UNIMPLEMENTED_MSG("Unimplemented shadow convert format: {}",
- static_cast<u32>(params.pixel_format));
- }
- params.type = GetFormatType(params.pixel_format);
- }
- // TODO: on 1DBuffer we should use the tic info.
- if (tic.IsBuffer()) {
- params.target = SurfaceTarget::TextureBuffer;
- params.width = tic.Width();
- params.pitch = params.width * params.GetBytesPerPixel();
- params.height = 1;
- params.depth = 1;
- params.num_levels = 1;
- params.emulated_levels = 1;
- params.is_layered = false;
- } else {
- params.target = TextureTypeToSurfaceTarget(entry.type, entry.is_array);
- params.width = tic.Width();
- params.height = tic.Height();
- params.depth = tic.Depth();
- params.pitch = params.is_tiled ? 0 : tic.Pitch();
- if (params.target == SurfaceTarget::TextureCubemap ||
- params.target == SurfaceTarget::TextureCubeArray) {
- params.depth *= 6;
- }
- params.num_levels = tic.max_mip_level + 1;
- params.emulated_levels = std::min(params.num_levels, params.MaxPossibleMipmap());
- params.is_layered = params.IsLayered();
- }
- return params;
-}
-
-SurfaceParams SurfaceParams::CreateForImage(const FormatLookupTable& lookup_table,
- const Tegra::Texture::TICEntry& tic,
- const VideoCommon::Shader::Image& entry) {
- SurfaceParams params;
- params.is_tiled = tic.IsTiled();
- params.srgb_conversion = tic.IsSrgbConversionEnabled();
- params.block_width = params.is_tiled ? tic.BlockWidth() : 0;
- params.block_height = params.is_tiled ? tic.BlockHeight() : 0;
- params.block_depth = params.is_tiled ? tic.BlockDepth() : 0;
- params.tile_width_spacing = params.is_tiled ? (1 << tic.tile_width_spacing.Value()) : 1;
- params.pixel_format = lookup_table.GetPixelFormat(
- tic.format, params.srgb_conversion, tic.r_type, tic.g_type, tic.b_type, tic.a_type);
- params.type = GetFormatType(params.pixel_format);
- params.target = ImageTypeToSurfaceTarget(entry.type);
- // TODO: on 1DBuffer we should use the tic info.
- if (tic.IsBuffer()) {
- params.target = SurfaceTarget::TextureBuffer;
- params.width = tic.Width();
- params.pitch = params.width * params.GetBytesPerPixel();
- params.height = 1;
- params.depth = 1;
- params.num_levels = 1;
- params.emulated_levels = 1;
- params.is_layered = false;
- } else {
- params.width = tic.Width();
- params.height = tic.Height();
- params.depth = tic.Depth();
- params.pitch = params.is_tiled ? 0 : tic.Pitch();
- if (params.target == SurfaceTarget::TextureCubemap ||
- params.target == SurfaceTarget::TextureCubeArray) {
- params.depth *= 6;
- }
- params.num_levels = tic.max_mip_level + 1;
- params.emulated_levels = std::min(params.num_levels, params.MaxPossibleMipmap());
- params.is_layered = params.IsLayered();
- }
- return params;
-}
-
-SurfaceParams SurfaceParams::CreateForDepthBuffer(Tegra::Engines::Maxwell3D& maxwell3d) {
- const auto& regs = maxwell3d.regs;
- const auto block_depth = std::min(regs.zeta.memory_layout.block_depth.Value(), 5U);
- const bool is_layered = regs.zeta_layers > 1 && block_depth == 0;
- const auto pixel_format = PixelFormatFromDepthFormat(regs.zeta.format);
- return {
- .is_tiled = regs.zeta.memory_layout.type ==
- Tegra::Engines::Maxwell3D::Regs::InvMemoryLayout::BlockLinear,
- .srgb_conversion = false,
- .is_layered = is_layered,
- .block_width = std::min(regs.zeta.memory_layout.block_width.Value(), 5U),
- .block_height = std::min(regs.zeta.memory_layout.block_height.Value(), 5U),
- .block_depth = block_depth,
- .tile_width_spacing = 1,
- .width = regs.zeta_width,
- .height = regs.zeta_height,
- .depth = is_layered ? regs.zeta_layers.Value() : 1U,
- .pitch = 0,
- .num_levels = 1,
- .emulated_levels = 1,
- .pixel_format = pixel_format,
- .type = GetFormatType(pixel_format),
- .target = is_layered ? SurfaceTarget::Texture2DArray : SurfaceTarget::Texture2D,
- };
-}
-
-SurfaceParams SurfaceParams::CreateForFramebuffer(Tegra::Engines::Maxwell3D& maxwell3d,
- std::size_t index) {
- const auto& config{maxwell3d.regs.rt[index]};
- SurfaceParams params;
- params.is_tiled =
- config.memory_layout.type == Tegra::Engines::Maxwell3D::Regs::InvMemoryLayout::BlockLinear;
- params.srgb_conversion = config.format == Tegra::RenderTargetFormat::B8G8R8A8_SRGB ||
- config.format == Tegra::RenderTargetFormat::A8B8G8R8_SRGB;
- params.block_width = config.memory_layout.block_width;
- params.block_height = config.memory_layout.block_height;
- params.block_depth = config.memory_layout.block_depth;
- params.tile_width_spacing = 1;
- params.pixel_format = PixelFormatFromRenderTargetFormat(config.format);
- params.type = GetFormatType(params.pixel_format);
- if (params.is_tiled) {
- params.pitch = 0;
- params.width = config.width;
- } else {
- const u32 bpp = GetFormatBpp(params.pixel_format) / CHAR_BIT;
- params.pitch = config.width;
- params.width = params.pitch / bpp;
- }
- params.height = config.height;
- params.num_levels = 1;
- params.emulated_levels = 1;
-
- if (config.memory_layout.is_3d != 0) {
- params.depth = config.layers.Value();
- params.is_layered = false;
- params.target = SurfaceTarget::Texture3D;
- } else if (config.layers > 1) {
- params.depth = config.layers.Value();
- params.is_layered = true;
- params.target = SurfaceTarget::Texture2DArray;
- } else {
- params.depth = 1;
- params.is_layered = false;
- params.target = SurfaceTarget::Texture2D;
- }
- return params;
-}
-
-SurfaceParams SurfaceParams::CreateForFermiCopySurface(
- const Tegra::Engines::Fermi2D::Regs::Surface& config) {
- const bool is_tiled = !config.linear;
- const auto pixel_format = PixelFormatFromRenderTargetFormat(config.format);
-
- SurfaceParams params{
- .is_tiled = is_tiled,
- .srgb_conversion = config.format == Tegra::RenderTargetFormat::B8G8R8A8_SRGB ||
- config.format == Tegra::RenderTargetFormat::A8B8G8R8_SRGB,
- .is_layered = false,
- .block_width = is_tiled ? std::min(config.BlockWidth(), 5U) : 0U,
- .block_height = is_tiled ? std::min(config.BlockHeight(), 5U) : 0U,
- .block_depth = is_tiled ? std::min(config.BlockDepth(), 5U) : 0U,
- .tile_width_spacing = 1,
- .width = config.width,
- .height = config.height,
- .depth = 1,
- .pitch = config.pitch,
- .num_levels = 1,
- .emulated_levels = 1,
- .pixel_format = pixel_format,
- .type = GetFormatType(pixel_format),
- // TODO(Rodrigo): Try to guess texture arrays from parameters
- .target = SurfaceTarget::Texture2D,
- };
-
- params.is_layered = params.IsLayered();
- return params;
-}
-
-VideoCore::Surface::SurfaceTarget SurfaceParams::ExpectedTarget(
- const VideoCommon::Shader::Sampler& entry) {
- return TextureTypeToSurfaceTarget(entry.type, entry.is_array);
-}
-
-VideoCore::Surface::SurfaceTarget SurfaceParams::ExpectedTarget(
- const VideoCommon::Shader::Image& entry) {
- return ImageTypeToSurfaceTarget(entry.type);
-}
-
-bool SurfaceParams::IsLayered() const {
- switch (target) {
- case SurfaceTarget::Texture1DArray:
- case SurfaceTarget::Texture2DArray:
- case SurfaceTarget::TextureCubemap:
- case SurfaceTarget::TextureCubeArray:
- return true;
- default:
- return false;
- }
-}
-
-// Auto block resizing algorithm from:
-// https://cgit.freedesktop.org/mesa/mesa/tree/src/gallium/drivers/nouveau/nv50/nv50_miptree.c
-u32 SurfaceParams::GetMipBlockHeight(u32 level) const {
- if (level == 0) {
- return this->block_height;
- }
-
- const u32 height_new{GetMipHeight(level)};
- const u32 default_block_height{GetDefaultBlockHeight()};
- const u32 blocks_in_y{(height_new + default_block_height - 1) / default_block_height};
- const u32 block_height_new = Common::Log2Ceil32(blocks_in_y);
- return std::clamp(block_height_new, 3U, 7U) - 3U;
-}
-
-u32 SurfaceParams::GetMipBlockDepth(u32 level) const {
- if (level == 0) {
- return this->block_depth;
- }
- if (is_layered) {
- return 0;
- }
-
- const u32 depth_new{GetMipDepth(level)};
- const u32 block_depth_new = Common::Log2Ceil32(depth_new);
- if (block_depth_new > 4) {
- return 5 - (GetMipBlockHeight(level) >= 2);
- }
- return block_depth_new;
-}
-
-std::size_t SurfaceParams::GetGuestMipmapLevelOffset(u32 level) const {
- std::size_t offset = 0;
- for (u32 i = 0; i < level; i++) {
- offset += GetInnerMipmapMemorySize(i, false, false);
- }
- return offset;
-}
-
-std::size_t SurfaceParams::GetHostMipmapLevelOffset(u32 level, bool is_converted) const {
- std::size_t offset = 0;
- if (is_converted) {
- for (u32 i = 0; i < level; ++i) {
- offset += GetConvertedMipmapSize(i) * GetNumLayers();
- }
- } else {
- for (u32 i = 0; i < level; ++i) {
- offset += GetInnerMipmapMemorySize(i, true, false) * GetNumLayers();
- }
- }
- return offset;
-}
-
-std::size_t SurfaceParams::GetConvertedMipmapSize(u32 level) const {
- constexpr std::size_t rgba8_bpp = 4ULL;
- const std::size_t mip_width = GetMipWidth(level);
- const std::size_t mip_height = GetMipHeight(level);
- const std::size_t mip_depth = is_layered ? 1 : GetMipDepth(level);
- return mip_width * mip_height * mip_depth * rgba8_bpp;
-}
-
-std::size_t SurfaceParams::GetLayerSize(bool as_host_size, bool uncompressed) const {
- std::size_t size = 0;
- for (u32 level = 0; level < num_levels; ++level) {
- size += GetInnerMipmapMemorySize(level, as_host_size, uncompressed);
- }
- if (is_tiled && is_layered) {
- return Common::AlignBits(size, Tegra::Texture::GOB_SIZE_SHIFT + block_height + block_depth);
- }
- return size;
-}
-
-std::size_t SurfaceParams::GetInnerMipmapMemorySize(u32 level, bool as_host_size,
- bool uncompressed) const {
- const u32 width{GetMipmapSize(uncompressed, GetMipWidth(level), GetDefaultBlockWidth())};
- const u32 height{GetMipmapSize(uncompressed, GetMipHeight(level), GetDefaultBlockHeight())};
- const u32 depth{is_layered ? 1U : GetMipDepth(level)};
- if (is_tiled) {
- return Tegra::Texture::CalculateSize(!as_host_size, GetBytesPerPixel(), width, height,
- depth, GetMipBlockHeight(level),
- GetMipBlockDepth(level));
- } else if (as_host_size || IsBuffer()) {
- return GetBytesPerPixel() * width * height * depth;
- } else {
- // Linear Texture Case
- return pitch * height * depth;
- }
-}
-
-bool SurfaceParams::operator==(const SurfaceParams& rhs) const {
- return std::tie(is_tiled, block_width, block_height, block_depth, tile_width_spacing, width,
- height, depth, pitch, num_levels, pixel_format, type, target) ==
- std::tie(rhs.is_tiled, rhs.block_width, rhs.block_height, rhs.block_depth,
- rhs.tile_width_spacing, rhs.width, rhs.height, rhs.depth, rhs.pitch,
- rhs.num_levels, rhs.pixel_format, rhs.type, rhs.target);
-}
-
-std::string SurfaceParams::TargetName() const {
- switch (target) {
- case SurfaceTarget::Texture1D:
- return "1D";
- case SurfaceTarget::TextureBuffer:
- return "TexBuffer";
- case SurfaceTarget::Texture2D:
- return "2D";
- case SurfaceTarget::Texture3D:
- return "3D";
- case SurfaceTarget::Texture1DArray:
- return "1DArray";
- case SurfaceTarget::Texture2DArray:
- return "2DArray";
- case SurfaceTarget::TextureCubemap:
- return "Cube";
- case SurfaceTarget::TextureCubeArray:
- return "CubeArray";
- default:
- LOG_CRITICAL(HW_GPU, "Unimplemented surface_target={}", static_cast<u32>(target));
- UNREACHABLE();
- return fmt::format("TUK({})", static_cast<u32>(target));
- }
-}
-
-u32 SurfaceParams::GetBlockSize() const {
- const u32 x = 64U << block_width;
- const u32 y = 8U << block_height;
- const u32 z = 1U << block_depth;
- return x * y * z;
-}
-
-std::pair<u32, u32> SurfaceParams::GetBlockXY() const {
- const u32 x_pixels = 64U / GetBytesPerPixel();
- const u32 x = x_pixels << block_width;
- const u32 y = 8U << block_height;
- return {x, y};
-}
-
-std::tuple<u32, u32, u32> SurfaceParams::GetBlockOffsetXYZ(u32 offset) const {
- const auto div_ceil = [](const u32 x, const u32 y) { return ((x + y - 1) / y); };
- const u32 block_size = GetBlockSize();
- const u32 block_index = offset / block_size;
- const u32 gob_offset = offset % block_size;
- const u32 gob_index = gob_offset / static_cast<u32>(Tegra::Texture::GOB_SIZE);
- const u32 x_gob_pixels = 64U / GetBytesPerPixel();
- const u32 x_block_pixels = x_gob_pixels << block_width;
- const u32 y_block_pixels = 8U << block_height;
- const u32 z_block_pixels = 1U << block_depth;
- const u32 x_blocks = div_ceil(width, x_block_pixels);
- const u32 y_blocks = div_ceil(height, y_block_pixels);
- const u32 z_blocks = div_ceil(depth, z_block_pixels);
- const u32 base_x = block_index % x_blocks;
- const u32 base_y = (block_index / x_blocks) % y_blocks;
- const u32 base_z = (block_index / (x_blocks * y_blocks)) % z_blocks;
- u32 x = base_x * x_block_pixels;
- u32 y = base_y * y_block_pixels;
- u32 z = base_z * z_block_pixels;
- z += gob_index >> block_height;
- y += (gob_index * 8U) % y_block_pixels;
- return {x, y, z};
-}
-
-} // namespace VideoCommon
diff --git a/src/video_core/texture_cache/surface_params.h b/src/video_core/texture_cache/surface_params.h
deleted file mode 100644
index 4466c3c34..000000000
--- a/src/video_core/texture_cache/surface_params.h
+++ /dev/null
@@ -1,294 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <utility>
-
-#include "common/alignment.h"
-#include "common/bit_util.h"
-#include "common/cityhash.h"
-#include "common/common_types.h"
-#include "video_core/engines/fermi_2d.h"
-#include "video_core/engines/maxwell_3d.h"
-#include "video_core/shader/shader_ir.h"
-#include "video_core/surface.h"
-#include "video_core/textures/decoders.h"
-
-namespace VideoCommon {
-
-class FormatLookupTable;
-
-class SurfaceParams {
-public:
- /// Creates SurfaceCachedParams from a texture configuration.
- static SurfaceParams CreateForTexture(const FormatLookupTable& lookup_table,
- const Tegra::Texture::TICEntry& tic,
- const VideoCommon::Shader::Sampler& entry);
-
- /// Creates SurfaceCachedParams from an image configuration.
- static SurfaceParams CreateForImage(const FormatLookupTable& lookup_table,
- const Tegra::Texture::TICEntry& tic,
- const VideoCommon::Shader::Image& entry);
-
- /// Creates SurfaceCachedParams for a depth buffer configuration.
- static SurfaceParams CreateForDepthBuffer(Tegra::Engines::Maxwell3D& maxwell3d);
-
- /// Creates SurfaceCachedParams from a framebuffer configuration.
- static SurfaceParams CreateForFramebuffer(Tegra::Engines::Maxwell3D& maxwell3d,
- std::size_t index);
-
- /// Creates SurfaceCachedParams from a Fermi2D surface configuration.
- static SurfaceParams CreateForFermiCopySurface(
- const Tegra::Engines::Fermi2D::Regs::Surface& config);
-
- /// Obtains the texture target from a shader's sampler entry.
- static VideoCore::Surface::SurfaceTarget ExpectedTarget(
- const VideoCommon::Shader::Sampler& entry);
-
- /// Obtains the texture target from a shader's sampler entry.
- static VideoCore::Surface::SurfaceTarget ExpectedTarget(
- const VideoCommon::Shader::Image& entry);
-
- std::size_t Hash() const {
- return static_cast<std::size_t>(
- Common::CityHash64(reinterpret_cast<const char*>(this), sizeof(*this)));
- }
-
- bool operator==(const SurfaceParams& rhs) const;
-
- bool operator!=(const SurfaceParams& rhs) const {
- return !operator==(rhs);
- }
-
- std::size_t GetGuestSizeInBytes() const {
- return GetInnerMemorySize(false, false, false);
- }
-
- std::size_t GetHostSizeInBytes(bool is_converted) const {
- if (!is_converted) {
- return GetInnerMemorySize(true, false, false);
- }
- // ASTC is uncompressed in software, in emulated as RGBA8
- std::size_t host_size_in_bytes = 0;
- for (u32 level = 0; level < num_levels; ++level) {
- host_size_in_bytes += GetConvertedMipmapSize(level) * GetNumLayers();
- }
- return host_size_in_bytes;
- }
-
- u32 GetBlockAlignedWidth() const {
- return Common::AlignUp(width, 64 / GetBytesPerPixel());
- }
-
- /// Returns the width of a given mipmap level.
- u32 GetMipWidth(u32 level) const {
- return std::max(1U, width >> level);
- }
-
- /// Returns the height of a given mipmap level.
- u32 GetMipHeight(u32 level) const {
- return std::max(1U, height >> level);
- }
-
- /// Returns the depth of a given mipmap level.
- u32 GetMipDepth(u32 level) const {
- return is_layered ? depth : std::max(1U, depth >> level);
- }
-
- /// Returns the block height of a given mipmap level.
- u32 GetMipBlockHeight(u32 level) const;
-
- /// Returns the block depth of a given mipmap level.
- u32 GetMipBlockDepth(u32 level) const;
-
- /// Returns the best possible row/pitch alignment for the surface.
- u32 GetRowAlignment(u32 level, bool is_converted) const {
- const u32 bpp = is_converted ? 4 : GetBytesPerPixel();
- return 1U << Common::CountTrailingZeroes32(GetMipWidth(level) * bpp);
- }
-
- /// Returns the offset in bytes in guest memory of a given mipmap level.
- std::size_t GetGuestMipmapLevelOffset(u32 level) const;
-
- /// Returns the offset in bytes in host memory (linear) of a given mipmap level.
- std::size_t GetHostMipmapLevelOffset(u32 level, bool is_converted) const;
-
- /// Returns the size in bytes in guest memory of a given mipmap level.
- std::size_t GetGuestMipmapSize(u32 level) const {
- return GetInnerMipmapMemorySize(level, false, false);
- }
-
- /// Returns the size in bytes in host memory (linear) of a given mipmap level.
- std::size_t GetHostMipmapSize(u32 level) const {
- return GetInnerMipmapMemorySize(level, true, false) * GetNumLayers();
- }
-
- std::size_t GetConvertedMipmapSize(u32 level) const;
-
- /// Get this texture Tegra Block size in guest memory layout
- u32 GetBlockSize() const;
-
- /// Get X, Y coordinates max sizes of a single block.
- std::pair<u32, u32> GetBlockXY() const;
-
- /// Get the offset in x, y, z coordinates from a memory offset
- std::tuple<u32, u32, u32> GetBlockOffsetXYZ(u32 offset) const;
-
- /// Returns the size of a layer in bytes in guest memory.
- std::size_t GetGuestLayerSize() const {
- return GetLayerSize(false, false);
- }
-
- /// Returns the size of a layer in bytes in host memory for a given mipmap level.
- std::size_t GetHostLayerSize(u32 level) const {
- ASSERT(target != VideoCore::Surface::SurfaceTarget::Texture3D);
- return GetInnerMipmapMemorySize(level, true, false);
- }
-
- /// Returns the max possible mipmap that the texture can have in host gpu
- u32 MaxPossibleMipmap() const {
- const u32 max_mipmap_w = Common::Log2Ceil32(width) + 1U;
- const u32 max_mipmap_h = Common::Log2Ceil32(height) + 1U;
- const u32 max_mipmap = std::max(max_mipmap_w, max_mipmap_h);
- if (target != VideoCore::Surface::SurfaceTarget::Texture3D)
- return max_mipmap;
- return std::max(max_mipmap, Common::Log2Ceil32(depth) + 1U);
- }
-
- /// Returns if the guest surface is a compressed surface.
- bool IsCompressed() const {
- return GetDefaultBlockHeight() > 1 || GetDefaultBlockWidth() > 1;
- }
-
- /// Returns the default block width.
- u32 GetDefaultBlockWidth() const {
- return VideoCore::Surface::GetDefaultBlockWidth(pixel_format);
- }
-
- /// Returns the default block height.
- u32 GetDefaultBlockHeight() const {
- return VideoCore::Surface::GetDefaultBlockHeight(pixel_format);
- }
-
- /// Returns the bits per pixel.
- u32 GetBitsPerPixel() const {
- return VideoCore::Surface::GetFormatBpp(pixel_format);
- }
-
- /// Returns the bytes per pixel.
- u32 GetBytesPerPixel() const {
- return VideoCore::Surface::GetBytesPerPixel(pixel_format);
- }
-
- /// Returns true if the pixel format is a depth and/or stencil format.
- bool IsPixelFormatZeta() const {
- return pixel_format >= VideoCore::Surface::PixelFormat::MaxColorFormat &&
- pixel_format < VideoCore::Surface::PixelFormat::MaxDepthStencilFormat;
- }
-
- /// Returns is the surface is a TextureBuffer type of surface.
- bool IsBuffer() const {
- return target == VideoCore::Surface::SurfaceTarget::TextureBuffer;
- }
-
- /// Returns the number of layers in the surface.
- std::size_t GetNumLayers() const {
- return is_layered ? depth : 1;
- }
-
- /// Returns the debug name of the texture for use in graphic debuggers.
- std::string TargetName() const;
-
- // Helper used for out of class size calculations
- static std::size_t AlignLayered(const std::size_t out_size, const u32 block_height,
- const u32 block_depth) {
- return Common::AlignBits(out_size,
- Tegra::Texture::GOB_SIZE_SHIFT + block_height + block_depth);
- }
-
- /// Converts a width from a type of surface into another. This helps represent the
- /// equivalent value between compressed/non-compressed textures.
- static u32 ConvertWidth(u32 width, VideoCore::Surface::PixelFormat pixel_format_from,
- VideoCore::Surface::PixelFormat pixel_format_to) {
- const u32 bw1 = VideoCore::Surface::GetDefaultBlockWidth(pixel_format_from);
- const u32 bw2 = VideoCore::Surface::GetDefaultBlockWidth(pixel_format_to);
- return (width * bw2 + bw1 - 1) / bw1;
- }
-
- /// Converts a height from a type of surface into another. This helps represent the
- /// equivalent value between compressed/non-compressed textures.
- static u32 ConvertHeight(u32 height, VideoCore::Surface::PixelFormat pixel_format_from,
- VideoCore::Surface::PixelFormat pixel_format_to) {
- const u32 bh1 = VideoCore::Surface::GetDefaultBlockHeight(pixel_format_from);
- const u32 bh2 = VideoCore::Surface::GetDefaultBlockHeight(pixel_format_to);
- return (height * bh2 + bh1 - 1) / bh1;
- }
-
- // Finds the maximun possible width between 2 2D layers of different formats
- static u32 IntersectWidth(const SurfaceParams& src_params, const SurfaceParams& dst_params,
- const u32 src_level, const u32 dst_level) {
- const u32 bw1 = src_params.GetDefaultBlockWidth();
- const u32 bw2 = dst_params.GetDefaultBlockWidth();
- const u32 t_src_width = (src_params.GetMipWidth(src_level) * bw2 + bw1 - 1) / bw1;
- const u32 t_dst_width = (dst_params.GetMipWidth(dst_level) * bw1 + bw2 - 1) / bw2;
- return std::min(t_src_width, t_dst_width);
- }
-
- // Finds the maximun possible height between 2 2D layers of different formats
- static u32 IntersectHeight(const SurfaceParams& src_params, const SurfaceParams& dst_params,
- const u32 src_level, const u32 dst_level) {
- const u32 bh1 = src_params.GetDefaultBlockHeight();
- const u32 bh2 = dst_params.GetDefaultBlockHeight();
- const u32 t_src_height = (src_params.GetMipHeight(src_level) * bh2 + bh1 - 1) / bh1;
- const u32 t_dst_height = (dst_params.GetMipHeight(dst_level) * bh1 + bh2 - 1) / bh2;
- return std::min(t_src_height, t_dst_height);
- }
-
- bool is_tiled;
- bool srgb_conversion;
- bool is_layered;
- u32 block_width;
- u32 block_height;
- u32 block_depth;
- u32 tile_width_spacing;
- u32 width;
- u32 height;
- u32 depth;
- u32 pitch;
- u32 num_levels;
- u32 emulated_levels;
- VideoCore::Surface::PixelFormat pixel_format;
- VideoCore::Surface::SurfaceType type;
- VideoCore::Surface::SurfaceTarget target;
-
-private:
- /// Returns the size of a given mipmap level inside a layer.
- std::size_t GetInnerMipmapMemorySize(u32 level, bool as_host_size, bool uncompressed) const;
-
- /// Returns the size of all mipmap levels and aligns as needed.
- std::size_t GetInnerMemorySize(bool as_host_size, bool layer_only, bool uncompressed) const {
- return GetLayerSize(as_host_size, uncompressed) *
- (layer_only ? 1U : (is_layered ? depth : 1U));
- }
-
- /// Returns the size of a layer
- std::size_t GetLayerSize(bool as_host_size, bool uncompressed) const;
-
- /// Returns true if these parameters are from a layered surface.
- bool IsLayered() const;
-};
-
-} // namespace VideoCommon
-
-namespace std {
-
-template <>
-struct hash<VideoCommon::SurfaceParams> {
- std::size_t operator()(const VideoCommon::SurfaceParams& k) const noexcept {
- return k.Hash();
- }
-};
-
-} // namespace std
diff --git a/src/video_core/texture_cache/surface_view.cpp b/src/video_core/texture_cache/surface_view.cpp
deleted file mode 100644
index 6b5f5984b..000000000
--- a/src/video_core/texture_cache/surface_view.cpp
+++ /dev/null
@@ -1,27 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include <tuple>
-
-#include "common/common_types.h"
-#include "video_core/texture_cache/surface_view.h"
-
-namespace VideoCommon {
-
-std::size_t ViewParams::Hash() const {
- return static_cast<std::size_t>(base_layer) ^ (static_cast<std::size_t>(num_layers) << 16) ^
- (static_cast<std::size_t>(base_level) << 24) ^
- (static_cast<std::size_t>(num_levels) << 32) ^ (static_cast<std::size_t>(target) << 36);
-}
-
-bool ViewParams::operator==(const ViewParams& rhs) const {
- return std::tie(base_layer, num_layers, base_level, num_levels, target) ==
- std::tie(rhs.base_layer, rhs.num_layers, rhs.base_level, rhs.num_levels, rhs.target);
-}
-
-bool ViewParams::operator!=(const ViewParams& rhs) const {
- return !operator==(rhs);
-}
-
-} // namespace VideoCommon
diff --git a/src/video_core/texture_cache/surface_view.h b/src/video_core/texture_cache/surface_view.h
deleted file mode 100644
index 90a8bb0ae..000000000
--- a/src/video_core/texture_cache/surface_view.h
+++ /dev/null
@@ -1,68 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <functional>
-
-#include "common/common_types.h"
-#include "video_core/surface.h"
-#include "video_core/texture_cache/surface_params.h"
-
-namespace VideoCommon {
-
-struct ViewParams {
- constexpr explicit ViewParams(VideoCore::Surface::SurfaceTarget target, u32 base_layer,
- u32 num_layers, u32 base_level, u32 num_levels)
- : target{target}, base_layer{base_layer}, num_layers{num_layers}, base_level{base_level},
- num_levels{num_levels} {}
-
- std::size_t Hash() const;
-
- bool operator==(const ViewParams& rhs) const;
- bool operator!=(const ViewParams& rhs) const;
-
- bool IsLayered() const {
- switch (target) {
- case VideoCore::Surface::SurfaceTarget::Texture1DArray:
- case VideoCore::Surface::SurfaceTarget::Texture2DArray:
- case VideoCore::Surface::SurfaceTarget::TextureCubemap:
- case VideoCore::Surface::SurfaceTarget::TextureCubeArray:
- return true;
- default:
- return false;
- }
- }
-
- VideoCore::Surface::SurfaceTarget target{};
- u32 base_layer{};
- u32 num_layers{};
- u32 base_level{};
- u32 num_levels{};
-};
-
-class ViewBase {
-public:
- constexpr explicit ViewBase(const ViewParams& params) : params{params} {}
-
- constexpr const ViewParams& GetViewParams() const {
- return params;
- }
-
-protected:
- ViewParams params;
-};
-
-} // namespace VideoCommon
-
-namespace std {
-
-template <>
-struct hash<VideoCommon::ViewParams> {
- std::size_t operator()(const VideoCommon::ViewParams& k) const noexcept {
- return k.Hash();
- }
-};
-
-} // namespace std
diff --git a/src/video_core/texture_cache/texture_cache.h b/src/video_core/texture_cache/texture_cache.h
index ea835c59f..968059842 100644
--- a/src/video_core/texture_cache/texture_cache.h
+++ b/src/video_core/texture_cache/texture_cache.h
@@ -6,1299 +6,1449 @@
#include <algorithm>
#include <array>
-#include <list>
+#include <bit>
#include <memory>
#include <mutex>
-#include <set>
-#include <tuple>
+#include <optional>
+#include <span>
+#include <type_traits>
#include <unordered_map>
+#include <utility>
#include <vector>
#include <boost/container/small_vector.hpp>
-#include <boost/icl/interval_map.hpp>
-#include <boost/range/iterator_range.hpp>
-#include "common/assert.h"
+#include "common/alignment.h"
+#include "common/common_funcs.h"
#include "common/common_types.h"
-#include "common/math_util.h"
-#include "core/core.h"
-#include "core/memory.h"
-#include "core/settings.h"
+#include "common/logging/log.h"
#include "video_core/compatible_formats.h"
+#include "video_core/delayed_destruction_ring.h"
#include "video_core/dirty_flags.h"
#include "video_core/engines/fermi_2d.h"
+#include "video_core/engines/kepler_compute.h"
#include "video_core/engines/maxwell_3d.h"
-#include "video_core/gpu.h"
#include "video_core/memory_manager.h"
#include "video_core/rasterizer_interface.h"
#include "video_core/surface.h"
-#include "video_core/texture_cache/copy_params.h"
+#include "video_core/texture_cache/descriptor_table.h"
#include "video_core/texture_cache/format_lookup_table.h"
-#include "video_core/texture_cache/surface_base.h"
-#include "video_core/texture_cache/surface_params.h"
-#include "video_core/texture_cache/surface_view.h"
-
-namespace Tegra::Texture {
-struct FullTextureInfo;
-}
-
-namespace VideoCore {
-class RasterizerInterface;
-}
+#include "video_core/texture_cache/formatter.h"
+#include "video_core/texture_cache/image_base.h"
+#include "video_core/texture_cache/image_info.h"
+#include "video_core/texture_cache/image_view_base.h"
+#include "video_core/texture_cache/image_view_info.h"
+#include "video_core/texture_cache/render_targets.h"
+#include "video_core/texture_cache/samples_helper.h"
+#include "video_core/texture_cache/slot_vector.h"
+#include "video_core/texture_cache/types.h"
+#include "video_core/texture_cache/util.h"
+#include "video_core/textures/texture.h"
namespace VideoCommon {
-using VideoCore::Surface::FormatCompatibility;
+using Tegra::Texture::SwizzleSource;
+using Tegra::Texture::TextureType;
+using Tegra::Texture::TICEntry;
+using Tegra::Texture::TSCEntry;
+using VideoCore::Surface::GetFormatType;
+using VideoCore::Surface::IsCopyCompatible;
using VideoCore::Surface::PixelFormat;
-using VideoCore::Surface::SurfaceTarget;
-using RenderTargetConfig = Tegra::Engines::Maxwell3D::Regs::RenderTargetConfig;
+using VideoCore::Surface::PixelFormatFromDepthFormat;
+using VideoCore::Surface::PixelFormatFromRenderTargetFormat;
+using VideoCore::Surface::SurfaceType;
-template <typename TSurface, typename TView>
+template <class P>
class TextureCache {
- using VectorSurface = boost::container::small_vector<TSurface, 1>;
+ /// Address shift for caching images into a hash table
+ static constexpr u64 PAGE_SHIFT = 20;
+
+ /// Enables debugging features to the texture cache
+ static constexpr bool ENABLE_VALIDATION = P::ENABLE_VALIDATION;
+ /// Implement blits as copies between framebuffers
+ static constexpr bool FRAMEBUFFER_BLITS = P::FRAMEBUFFER_BLITS;
+ /// True when some copies have to be emulated
+ static constexpr bool HAS_EMULATED_COPIES = P::HAS_EMULATED_COPIES;
+
+ /// Image view ID for null descriptors
+ static constexpr ImageViewId NULL_IMAGE_VIEW_ID{0};
+ /// Sampler ID for bugged sampler ids
+ static constexpr SamplerId NULL_SAMPLER_ID{0};
+
+ using Runtime = typename P::Runtime;
+ using Image = typename P::Image;
+ using ImageAlloc = typename P::ImageAlloc;
+ using ImageView = typename P::ImageView;
+ using Sampler = typename P::Sampler;
+ using Framebuffer = typename P::Framebuffer;
+
+ struct BlitImages {
+ ImageId dst_id;
+ ImageId src_id;
+ PixelFormat dst_format;
+ PixelFormat src_format;
+ };
+
+ template <typename T>
+ struct IdentityHash {
+ [[nodiscard]] size_t operator()(T value) const noexcept {
+ return static_cast<size_t>(value);
+ }
+ };
public:
- void InvalidateRegion(VAddr addr, std::size_t size) {
- std::lock_guard lock{mutex};
+ explicit TextureCache(Runtime&, VideoCore::RasterizerInterface&, Tegra::Engines::Maxwell3D&,
+ Tegra::Engines::KeplerCompute&, Tegra::MemoryManager&);
- for (const auto& surface : GetSurfacesInRegion(addr, size)) {
- Unregister(surface);
- }
- }
+ /// Notify the cache that a new frame has been queued
+ void TickFrame();
- void OnCPUWrite(VAddr addr, std::size_t size) {
- std::lock_guard lock{mutex};
+ /// Return an unique mutually exclusive lock for the cache
+ [[nodiscard]] std::unique_lock<std::mutex> AcquireLock();
- for (const auto& surface : GetSurfacesInRegion(addr, size)) {
- if (surface->IsMemoryMarked()) {
- UnmarkMemory(surface);
- surface->SetSyncPending(true);
- marked_for_unregister.emplace_back(surface);
- }
- }
- }
+ /// Return a constant reference to the given image view id
+ [[nodiscard]] const ImageView& GetImageView(ImageViewId id) const noexcept;
- void SyncGuestHost() {
- std::lock_guard lock{mutex};
+ /// Return a reference to the given image view id
+ [[nodiscard]] ImageView& GetImageView(ImageViewId id) noexcept;
- for (const auto& surface : marked_for_unregister) {
- if (surface->IsRegistered()) {
- surface->SetSyncPending(false);
- Unregister(surface);
- }
- }
- marked_for_unregister.clear();
- }
+ /// Fill image_view_ids with the graphics images in indices
+ void FillGraphicsImageViews(std::span<const u32> indices,
+ std::span<ImageViewId> image_view_ids);
- /**
- * Guarantees that rendertargets don't unregister themselves if the
- * collide. Protection is currently only done on 3D slices.
- */
- void GuardRenderTargets(bool new_guard) {
- guard_render_targets = new_guard;
- }
+ /// Fill image_view_ids with the compute images in indices
+ void FillComputeImageViews(std::span<const u32> indices, std::span<ImageViewId> image_view_ids);
- void GuardSamplers(bool new_guard) {
- guard_samplers = new_guard;
- }
+ /// Get the sampler from the graphics descriptor table in the specified index
+ Sampler* GetGraphicsSampler(u32 index);
- void FlushRegion(VAddr addr, std::size_t size) {
- std::lock_guard lock{mutex};
+ /// Get the sampler from the compute descriptor table in the specified index
+ Sampler* GetComputeSampler(u32 index);
- auto surfaces = GetSurfacesInRegion(addr, size);
- if (surfaces.empty()) {
- return;
- }
- std::sort(surfaces.begin(), surfaces.end(), [](const TSurface& a, const TSurface& b) {
- return a->GetModificationTick() < b->GetModificationTick();
- });
- for (const auto& surface : surfaces) {
- mutex.unlock();
- FlushSurface(surface);
- mutex.lock();
- }
- }
+ /// Refresh the state for graphics image view and sampler descriptors
+ void SynchronizeGraphicsDescriptors();
- bool MustFlushRegion(VAddr addr, std::size_t size) {
- std::lock_guard lock{mutex};
+ /// Refresh the state for compute image view and sampler descriptors
+ void SynchronizeComputeDescriptors();
- const auto surfaces = GetSurfacesInRegion(addr, size);
- return std::any_of(surfaces.cbegin(), surfaces.cend(),
- [](const TSurface& surface) { return surface->IsModified(); });
- }
+ /// Update bound render targets and upload memory if necessary
+ /// @param is_clear True when the render targets are being used for clears
+ void UpdateRenderTargets(bool is_clear);
- TView GetTextureSurface(const Tegra::Texture::TICEntry& tic,
- const VideoCommon::Shader::Sampler& entry) {
- std::lock_guard lock{mutex};
- const auto gpu_addr{tic.Address()};
- if (!gpu_addr) {
- return GetNullSurface(SurfaceParams::ExpectedTarget(entry));
- }
+ /// Find a framebuffer with the currently bound render targets
+ /// UpdateRenderTargets should be called before this
+ Framebuffer* GetFramebuffer();
- const std::optional<VAddr> cpu_addr = gpu_memory.GpuToCpuAddress(gpu_addr);
- if (!cpu_addr) {
- return GetNullSurface(SurfaceParams::ExpectedTarget(entry));
- }
+ /// Mark images in a range as modified from the CPU
+ void WriteMemory(VAddr cpu_addr, size_t size);
- if (!IsTypeCompatible(tic.texture_type, entry)) {
- return GetNullSurface(SurfaceParams::ExpectedTarget(entry));
- }
+ /// Download contents of host images to guest memory in a region
+ void DownloadMemory(VAddr cpu_addr, size_t size);
- const auto params{SurfaceParams::CreateForTexture(format_lookup_table, tic, entry)};
- const auto [surface, view] = GetSurface(gpu_addr, *cpu_addr, params, true, false);
- if (guard_samplers) {
- sampled_textures.push_back(surface);
- }
- return view;
- }
+ /// Remove images in a region
+ void UnmapMemory(VAddr cpu_addr, size_t size);
- TView GetImageSurface(const Tegra::Texture::TICEntry& tic,
- const VideoCommon::Shader::Image& entry) {
- std::lock_guard lock{mutex};
- const auto gpu_addr{tic.Address()};
- if (!gpu_addr) {
- return GetNullSurface(SurfaceParams::ExpectedTarget(entry));
- }
- const std::optional<VAddr> cpu_addr = gpu_memory.GpuToCpuAddress(gpu_addr);
- if (!cpu_addr) {
- return GetNullSurface(SurfaceParams::ExpectedTarget(entry));
- }
- const auto params{SurfaceParams::CreateForImage(format_lookup_table, tic, entry)};
- const auto [surface, view] = GetSurface(gpu_addr, *cpu_addr, params, true, false);
- if (guard_samplers) {
- sampled_textures.push_back(surface);
- }
- return view;
- }
+ /// Blit an image with the given parameters
+ void BlitImage(const Tegra::Engines::Fermi2D::Surface& dst,
+ const Tegra::Engines::Fermi2D::Surface& src,
+ const Tegra::Engines::Fermi2D::Config& copy);
- bool TextureBarrier() {
- const bool any_rt =
- std::any_of(sampled_textures.begin(), sampled_textures.end(),
- [](const auto& surface) { return surface->IsRenderTarget(); });
- sampled_textures.clear();
- return any_rt;
- }
+ /// Invalidate the contents of the color buffer index
+ /// These contents become unspecified, the cache can assume aggressive optimizations.
+ void InvalidateColorBuffer(size_t index);
- TView GetDepthBufferSurface(bool preserve_contents) {
- std::lock_guard lock{mutex};
- auto& dirty = maxwell3d.dirty;
- if (!dirty.flags[VideoCommon::Dirty::ZetaBuffer]) {
- return depth_buffer.view;
- }
- dirty.flags[VideoCommon::Dirty::ZetaBuffer] = false;
+ /// Invalidate the contents of the depth buffer
+ /// These contents become unspecified, the cache can assume aggressive optimizations.
+ void InvalidateDepthBuffer();
- const auto& regs{maxwell3d.regs};
- const auto gpu_addr{regs.zeta.Address()};
- if (!gpu_addr || !regs.zeta_enable) {
- SetEmptyDepthBuffer();
- return {};
- }
- const std::optional<VAddr> cpu_addr = gpu_memory.GpuToCpuAddress(gpu_addr);
- if (!cpu_addr) {
- SetEmptyDepthBuffer();
- return {};
- }
- const auto depth_params{SurfaceParams::CreateForDepthBuffer(maxwell3d)};
- auto surface_view = GetSurface(gpu_addr, *cpu_addr, depth_params, preserve_contents, true);
- if (depth_buffer.target)
- depth_buffer.target->MarkAsRenderTarget(false, NO_RT);
- depth_buffer.target = surface_view.first;
- depth_buffer.view = surface_view.second;
- if (depth_buffer.target)
- depth_buffer.target->MarkAsRenderTarget(true, DEPTH_RT);
- return surface_view.second;
- }
-
- TView GetColorBufferSurface(std::size_t index, bool preserve_contents) {
- std::lock_guard lock{mutex};
- ASSERT(index < Tegra::Engines::Maxwell3D::Regs::NumRenderTargets);
- if (!maxwell3d.dirty.flags[VideoCommon::Dirty::ColorBuffer0 + index]) {
- return render_targets[index].view;
- }
- maxwell3d.dirty.flags[VideoCommon::Dirty::ColorBuffer0 + index] = false;
+ /// Try to find a cached image view in the given CPU address
+ [[nodiscard]] ImageView* TryFindFramebufferImageView(VAddr cpu_addr);
- const auto& regs{maxwell3d.regs};
- if (index >= regs.rt_control.count || regs.rt[index].Address() == 0 ||
- regs.rt[index].format == Tegra::RenderTargetFormat::NONE) {
- SetEmptyColorBuffer(index);
- return {};
- }
+ /// Return true when there are uncommitted images to be downloaded
+ [[nodiscard]] bool HasUncommittedFlushes() const noexcept;
- const auto& config{regs.rt[index]};
- const auto gpu_addr{config.Address()};
- if (!gpu_addr) {
- SetEmptyColorBuffer(index);
- return {};
- }
+ /// Return true when the caller should wait for async downloads
+ [[nodiscard]] bool ShouldWaitAsyncFlushes() const noexcept;
- const std::optional<VAddr> cpu_addr = gpu_memory.GpuToCpuAddress(gpu_addr);
- if (!cpu_addr) {
- SetEmptyColorBuffer(index);
- return {};
- }
+ /// Commit asynchronous downloads
+ void CommitAsyncFlushes();
+
+ /// Pop asynchronous downloads
+ void PopAsyncFlushes();
+
+ /// Return true when a CPU region is modified from the GPU
+ [[nodiscard]] bool IsRegionGpuModified(VAddr addr, size_t size);
- auto surface_view =
- GetSurface(gpu_addr, *cpu_addr, SurfaceParams::CreateForFramebuffer(maxwell3d, index),
- preserve_contents, true);
- if (render_targets[index].target) {
- auto& surface = render_targets[index].target;
- surface->MarkAsRenderTarget(false, NO_RT);
- const auto& cr_params = surface->GetSurfaceParams();
- if (!cr_params.is_tiled && Settings::values.use_asynchronous_gpu_emulation.GetValue()) {
- AsyncFlushSurface(surface);
+private:
+ /// Iterate over all page indices in a range
+ template <typename Func>
+ static void ForEachPage(VAddr addr, size_t size, Func&& func) {
+ static constexpr bool RETURNS_BOOL = std::is_same_v<std::invoke_result<Func, u64>, bool>;
+ const u64 page_end = (addr + size - 1) >> PAGE_SHIFT;
+ for (u64 page = addr >> PAGE_SHIFT; page <= page_end; ++page) {
+ if constexpr (RETURNS_BOOL) {
+ if (func(page)) {
+ break;
+ }
+ } else {
+ func(page);
}
}
- render_targets[index].target = surface_view.first;
- render_targets[index].view = surface_view.second;
- if (render_targets[index].target)
- render_targets[index].target->MarkAsRenderTarget(true, static_cast<u32>(index));
- return surface_view.second;
}
- void MarkColorBufferInUse(std::size_t index) {
- if (auto& render_target = render_targets[index].target) {
- render_target->MarkAsModified(true, Tick());
- }
- }
+ /// Fills image_view_ids in the image views in indices
+ void FillImageViews(DescriptorTable<TICEntry>& table,
+ std::span<ImageViewId> cached_image_view_ids, std::span<const u32> indices,
+ std::span<ImageViewId> image_view_ids);
- void MarkDepthBufferInUse() {
- if (depth_buffer.target) {
- depth_buffer.target->MarkAsModified(true, Tick());
- }
- }
+ /// Find or create an image view in the guest descriptor table
+ ImageViewId VisitImageView(DescriptorTable<TICEntry>& table,
+ std::span<ImageViewId> cached_image_view_ids, u32 index);
- void SetEmptyDepthBuffer() {
- if (depth_buffer.target == nullptr) {
- return;
- }
- depth_buffer.target->MarkAsRenderTarget(false, NO_RT);
- depth_buffer.target = nullptr;
- depth_buffer.view = nullptr;
- }
+ /// Find or create a framebuffer with the given render target parameters
+ FramebufferId GetFramebufferId(const RenderTargets& key);
- void SetEmptyColorBuffer(std::size_t index) {
- if (render_targets[index].target == nullptr) {
- return;
- }
- render_targets[index].target->MarkAsRenderTarget(false, NO_RT);
- render_targets[index].target = nullptr;
- render_targets[index].view = nullptr;
- }
-
- void DoFermiCopy(const Tegra::Engines::Fermi2D::Regs::Surface& src_config,
- const Tegra::Engines::Fermi2D::Regs::Surface& dst_config,
- const Tegra::Engines::Fermi2D::Config& copy_config) {
- std::lock_guard lock{mutex};
- SurfaceParams src_params = SurfaceParams::CreateForFermiCopySurface(src_config);
- SurfaceParams dst_params = SurfaceParams::CreateForFermiCopySurface(dst_config);
- const GPUVAddr src_gpu_addr = src_config.Address();
- const GPUVAddr dst_gpu_addr = dst_config.Address();
- DeduceBestBlit(src_params, dst_params, src_gpu_addr, dst_gpu_addr);
-
- const std::optional<VAddr> dst_cpu_addr = gpu_memory.GpuToCpuAddress(dst_gpu_addr);
- const std::optional<VAddr> src_cpu_addr = gpu_memory.GpuToCpuAddress(src_gpu_addr);
- std::pair dst_surface = GetSurface(dst_gpu_addr, *dst_cpu_addr, dst_params, true, false);
- TView src_surface = GetSurface(src_gpu_addr, *src_cpu_addr, src_params, true, false).second;
- ImageBlit(src_surface, dst_surface.second, copy_config);
- dst_surface.first->MarkAsModified(true, Tick());
- }
-
- TSurface TryFindFramebufferSurface(VAddr addr) const {
- if (!addr) {
- return nullptr;
- }
- const VAddr page = addr >> registry_page_bits;
- const auto it = registry.find(page);
- if (it == registry.end()) {
- return nullptr;
- }
- const auto& list = it->second;
- const auto found = std::find_if(list.begin(), list.end(), [addr](const auto& surface) {
- return surface->GetCpuAddr() == addr;
- });
- return found != list.end() ? *found : nullptr;
- }
+ /// Refresh the contents (pixel data) of an image
+ void RefreshContents(Image& image);
- u64 Tick() {
- return ++ticks;
- }
+ /// Upload data from guest to an image
+ template <typename MapBuffer>
+ void UploadImageContents(Image& image, MapBuffer& map, size_t buffer_offset);
- void CommitAsyncFlushes() {
- committed_flushes.push_back(uncommitted_flushes);
- uncommitted_flushes.reset();
- }
+ /// Find or create an image view from a guest descriptor
+ [[nodiscard]] ImageViewId FindImageView(const TICEntry& config);
- bool HasUncommittedFlushes() const {
- return uncommitted_flushes != nullptr;
- }
+ /// Create a new image view from a guest descriptor
+ [[nodiscard]] ImageViewId CreateImageView(const TICEntry& config);
- bool ShouldWaitAsyncFlushes() const {
- return !committed_flushes.empty() && committed_flushes.front() != nullptr;
- }
+ /// Find or create an image from the given parameters
+ [[nodiscard]] ImageId FindOrInsertImage(const ImageInfo& info, GPUVAddr gpu_addr,
+ RelaxedOptions options = RelaxedOptions{});
- void PopAsyncFlushes() {
- if (committed_flushes.empty()) {
- return;
- }
- auto& flush_list = committed_flushes.front();
- if (!flush_list) {
- committed_flushes.pop_front();
- return;
- }
- for (TSurface& surface : *flush_list) {
- FlushSurface(surface);
- }
- committed_flushes.pop_front();
- }
+ /// Find an image from the given parameters
+ [[nodiscard]] ImageId FindImage(const ImageInfo& info, GPUVAddr gpu_addr,
+ RelaxedOptions options);
-protected:
- explicit TextureCache(VideoCore::RasterizerInterface& rasterizer_,
- Tegra::Engines::Maxwell3D& maxwell3d_, Tegra::MemoryManager& gpu_memory_,
- bool is_astc_supported_)
- : is_astc_supported{is_astc_supported_}, rasterizer{rasterizer_}, maxwell3d{maxwell3d_},
- gpu_memory{gpu_memory_} {
- for (std::size_t i = 0; i < Tegra::Engines::Maxwell3D::Regs::NumRenderTargets; i++) {
- SetEmptyColorBuffer(i);
- }
+ /// Create an image from the given parameters
+ [[nodiscard]] ImageId InsertImage(const ImageInfo& info, GPUVAddr gpu_addr,
+ RelaxedOptions options);
- SetEmptyDepthBuffer();
- staging_cache.SetSize(2);
+ /// Create a new image and join perfectly matching existing images
+ /// Remove joined images from the cache
+ [[nodiscard]] ImageId JoinImages(const ImageInfo& info, GPUVAddr gpu_addr, VAddr cpu_addr);
- const auto make_siblings = [this](PixelFormat a, PixelFormat b) {
- siblings_table[static_cast<std::size_t>(a)] = b;
- siblings_table[static_cast<std::size_t>(b)] = a;
- };
- std::fill(siblings_table.begin(), siblings_table.end(), PixelFormat::Invalid);
- make_siblings(PixelFormat::D16_UNORM, PixelFormat::R16_UNORM);
- make_siblings(PixelFormat::D32_FLOAT, PixelFormat::R32_FLOAT);
- make_siblings(PixelFormat::D32_FLOAT_S8_UINT, PixelFormat::R32G32_FLOAT);
+ /// Return a blit image pair from the given guest blit parameters
+ [[nodiscard]] BlitImages GetBlitImages(const Tegra::Engines::Fermi2D::Surface& dst,
+ const Tegra::Engines::Fermi2D::Surface& src);
- sampled_textures.reserve(64);
- }
+ /// Find or create a sampler from a guest descriptor sampler
+ [[nodiscard]] SamplerId FindSampler(const TSCEntry& config);
- ~TextureCache() = default;
+ /// Find or create an image view for the given color buffer index
+ [[nodiscard]] ImageViewId FindColorBuffer(size_t index, bool is_clear);
- virtual TSurface CreateSurface(GPUVAddr gpu_addr, const SurfaceParams& params) = 0;
+ /// Find or create an image view for the depth buffer
+ [[nodiscard]] ImageViewId FindDepthBuffer(bool is_clear);
- virtual void ImageCopy(TSurface& src_surface, TSurface& dst_surface,
- const CopyParams& copy_params) = 0;
+ /// Find or create a view for a render target with the given image parameters
+ [[nodiscard]] ImageViewId FindRenderTargetView(const ImageInfo& info, GPUVAddr gpu_addr,
+ bool is_clear);
- virtual void ImageBlit(TView& src_view, TView& dst_view,
- const Tegra::Engines::Fermi2D::Config& copy_config) = 0;
+ /// Iterates over all the images in a region calling func
+ template <typename Func>
+ void ForEachImageInRegion(VAddr cpu_addr, size_t size, Func&& func);
- // Depending on the backend, a buffer copy can be slow as it means deoptimizing the texture
- // and reading it from a separate buffer.
- virtual void BufferCopy(TSurface& src_surface, TSurface& dst_surface) = 0;
+ /// Find or create an image view in the given image with the passed parameters
+ [[nodiscard]] ImageViewId FindOrEmplaceImageView(ImageId image_id, const ImageViewInfo& info);
- void ManageRenderTargetUnregister(TSurface& surface) {
- auto& dirty = maxwell3d.dirty;
- const u32 index = surface->GetRenderTarget();
- if (index == DEPTH_RT) {
- dirty.flags[VideoCommon::Dirty::ZetaBuffer] = true;
- } else {
- dirty.flags[VideoCommon::Dirty::ColorBuffer0 + index] = true;
- }
- dirty.flags[VideoCommon::Dirty::RenderTargets] = true;
+ /// Register image in the page table
+ void RegisterImage(ImageId image);
+
+ /// Unregister image from the page table
+ void UnregisterImage(ImageId image);
+
+ /// Track CPU reads and writes for image
+ void TrackImage(ImageBase& image);
+
+ /// Stop tracking CPU reads and writes for image
+ void UntrackImage(ImageBase& image);
+
+ /// Delete image from the cache
+ void DeleteImage(ImageId image);
+
+ /// Remove image views references from the cache
+ void RemoveImageViewReferences(std::span<const ImageViewId> removed_views);
+
+ /// Remove framebuffers using the given image views from the cache
+ void RemoveFramebuffers(std::span<const ImageViewId> removed_views);
+
+ /// Mark an image as modified from the GPU
+ void MarkModification(ImageBase& image) noexcept;
+
+ /// Synchronize image aliases, copying data if needed
+ void SynchronizeAliases(ImageId image_id);
+
+ /// Prepare an image to be used
+ void PrepareImage(ImageId image_id, bool is_modification, bool invalidate);
+
+ /// Prepare an image view to be used
+ void PrepareImageView(ImageViewId image_view_id, bool is_modification, bool invalidate);
+
+ /// Execute copies from one image to the other, even if they are incompatible
+ void CopyImage(ImageId dst_id, ImageId src_id, std::span<const ImageCopy> copies);
+
+ /// Bind an image view as render target, downloading resources preemtively if needed
+ void BindRenderTarget(ImageViewId* old_id, ImageViewId new_id);
+
+ /// Create a render target from a given image and image view parameters
+ [[nodiscard]] std::pair<FramebufferId, ImageViewId> RenderTargetFromImage(
+ ImageId, const ImageViewInfo& view_info);
+
+ /// Returns true if the current clear parameters clear the whole image of a given image view
+ [[nodiscard]] bool IsFullClear(ImageViewId id);
+
+ Runtime& runtime;
+ VideoCore::RasterizerInterface& rasterizer;
+ Tegra::Engines::Maxwell3D& maxwell3d;
+ Tegra::Engines::KeplerCompute& kepler_compute;
+ Tegra::MemoryManager& gpu_memory;
+
+ DescriptorTable<TICEntry> graphics_image_table{gpu_memory};
+ DescriptorTable<TSCEntry> graphics_sampler_table{gpu_memory};
+ std::vector<SamplerId> graphics_sampler_ids;
+ std::vector<ImageViewId> graphics_image_view_ids;
+
+ DescriptorTable<TICEntry> compute_image_table{gpu_memory};
+ DescriptorTable<TSCEntry> compute_sampler_table{gpu_memory};
+ std::vector<SamplerId> compute_sampler_ids;
+ std::vector<ImageViewId> compute_image_view_ids;
+
+ RenderTargets render_targets;
+
+ std::mutex mutex;
+
+ std::unordered_map<TICEntry, ImageViewId> image_views;
+ std::unordered_map<TSCEntry, SamplerId> samplers;
+ std::unordered_map<RenderTargets, FramebufferId> framebuffers;
+
+ std::unordered_map<u64, std::vector<ImageId>, IdentityHash<u64>> page_table;
+
+ bool has_deleted_images = false;
+
+ SlotVector<Image> slot_images;
+ SlotVector<ImageView> slot_image_views;
+ SlotVector<ImageAlloc> slot_image_allocs;
+ SlotVector<Sampler> slot_samplers;
+ SlotVector<Framebuffer> slot_framebuffers;
+
+ // TODO: This data structure is not optimal and it should be reworked
+ std::vector<ImageId> uncommitted_downloads;
+ std::queue<std::vector<ImageId>> committed_downloads;
+
+ static constexpr size_t TICKS_TO_DESTROY = 6;
+ DelayedDestructionRing<Image, TICKS_TO_DESTROY> sentenced_images;
+ DelayedDestructionRing<ImageView, TICKS_TO_DESTROY> sentenced_image_view;
+ DelayedDestructionRing<Framebuffer, TICKS_TO_DESTROY> sentenced_framebuffers;
+
+ std::unordered_map<GPUVAddr, ImageAllocId> image_allocs_table;
+
+ u64 modification_tick = 0;
+ u64 frame_tick = 0;
+};
+
+template <class P>
+TextureCache<P>::TextureCache(Runtime& runtime_, VideoCore::RasterizerInterface& rasterizer_,
+ Tegra::Engines::Maxwell3D& maxwell3d_,
+ Tegra::Engines::KeplerCompute& kepler_compute_,
+ Tegra::MemoryManager& gpu_memory_)
+ : runtime{runtime_}, rasterizer{rasterizer_}, maxwell3d{maxwell3d_},
+ kepler_compute{kepler_compute_}, gpu_memory{gpu_memory_} {
+ // Configure null sampler
+ TSCEntry sampler_descriptor{};
+ sampler_descriptor.min_filter.Assign(Tegra::Texture::TextureFilter::Linear);
+ sampler_descriptor.mag_filter.Assign(Tegra::Texture::TextureFilter::Linear);
+ sampler_descriptor.mipmap_filter.Assign(Tegra::Texture::TextureMipmapFilter::Linear);
+ sampler_descriptor.cubemap_anisotropy.Assign(1);
+
+ // Make sure the first index is reserved for the null resources
+ // This way the null resource becomes a compile time constant
+ void(slot_image_views.insert(runtime, NullImageParams{}));
+ void(slot_samplers.insert(runtime, sampler_descriptor));
+}
+
+template <class P>
+void TextureCache<P>::TickFrame() {
+ // Tick sentenced resources in this order to ensure they are destroyed in the right order
+ sentenced_images.Tick();
+ sentenced_framebuffers.Tick();
+ sentenced_image_view.Tick();
+ ++frame_tick;
+}
+
+template <class P>
+std::unique_lock<std::mutex> TextureCache<P>::AcquireLock() {
+ return std::unique_lock{mutex};
+}
+
+template <class P>
+const typename P::ImageView& TextureCache<P>::GetImageView(ImageViewId id) const noexcept {
+ return slot_image_views[id];
+}
+
+template <class P>
+typename P::ImageView& TextureCache<P>::GetImageView(ImageViewId id) noexcept {
+ return slot_image_views[id];
+}
+
+template <class P>
+void TextureCache<P>::FillGraphicsImageViews(std::span<const u32> indices,
+ std::span<ImageViewId> image_view_ids) {
+ FillImageViews(graphics_image_table, graphics_image_view_ids, indices, image_view_ids);
+}
+
+template <class P>
+void TextureCache<P>::FillComputeImageViews(std::span<const u32> indices,
+ std::span<ImageViewId> image_view_ids) {
+ FillImageViews(compute_image_table, compute_image_view_ids, indices, image_view_ids);
+}
+
+template <class P>
+typename P::Sampler* TextureCache<P>::GetGraphicsSampler(u32 index) {
+ [[unlikely]] if (index > graphics_sampler_table.Limit()) {
+ LOG_ERROR(HW_GPU, "Invalid sampler index={}", index);
+ return &slot_samplers[NULL_SAMPLER_ID];
+ }
+ const auto [descriptor, is_new] = graphics_sampler_table.Read(index);
+ SamplerId& id = graphics_sampler_ids[index];
+ [[unlikely]] if (is_new) {
+ id = FindSampler(descriptor);
}
+ return &slot_samplers[id];
+}
- void Register(TSurface surface) {
- const GPUVAddr gpu_addr = surface->GetGpuAddr();
- const std::size_t size = surface->GetSizeInBytes();
- const std::optional<VAddr> cpu_addr = gpu_memory.GpuToCpuAddress(gpu_addr);
- if (!cpu_addr) {
- LOG_CRITICAL(HW_GPU, "Failed to register surface with unmapped gpu_address 0x{:016x}",
- gpu_addr);
- return;
- }
- surface->SetCpuAddr(*cpu_addr);
- RegisterInnerCache(surface);
- surface->MarkAsRegistered(true);
- surface->SetMemoryMarked(true);
- rasterizer.UpdatePagesCachedCount(*cpu_addr, size, 1);
+template <class P>
+typename P::Sampler* TextureCache<P>::GetComputeSampler(u32 index) {
+ [[unlikely]] if (index > compute_sampler_table.Limit()) {
+ LOG_ERROR(HW_GPU, "Invalid sampler index={}", index);
+ return &slot_samplers[NULL_SAMPLER_ID];
+ }
+ const auto [descriptor, is_new] = compute_sampler_table.Read(index);
+ SamplerId& id = compute_sampler_ids[index];
+ [[unlikely]] if (is_new) {
+ id = FindSampler(descriptor);
}
+ return &slot_samplers[id];
+}
- void UnmarkMemory(TSurface surface) {
- if (!surface->IsMemoryMarked()) {
- return;
- }
- const std::size_t size = surface->GetSizeInBytes();
- const VAddr cpu_addr = surface->GetCpuAddr();
- rasterizer.UpdatePagesCachedCount(cpu_addr, size, -1);
- surface->SetMemoryMarked(false);
+template <class P>
+void TextureCache<P>::SynchronizeGraphicsDescriptors() {
+ using SamplerIndex = Tegra::Engines::Maxwell3D::Regs::SamplerIndex;
+ const bool linked_tsc = maxwell3d.regs.sampler_index == SamplerIndex::ViaHeaderIndex;
+ const u32 tic_limit = maxwell3d.regs.tic.limit;
+ const u32 tsc_limit = linked_tsc ? tic_limit : maxwell3d.regs.tsc.limit;
+ if (graphics_sampler_table.Synchornize(maxwell3d.regs.tsc.Address(), tsc_limit)) {
+ graphics_sampler_ids.resize(tsc_limit + 1, CORRUPT_ID);
}
+ if (graphics_image_table.Synchornize(maxwell3d.regs.tic.Address(), tic_limit)) {
+ graphics_image_view_ids.resize(tic_limit + 1, CORRUPT_ID);
+ }
+}
- void Unregister(TSurface surface) {
- if (guard_render_targets && surface->IsProtected()) {
- return;
- }
- if (!guard_render_targets && surface->IsRenderTarget()) {
- ManageRenderTargetUnregister(surface);
- }
- UnmarkMemory(surface);
- if (surface->IsSyncPending()) {
- marked_for_unregister.remove(surface);
- surface->SetSyncPending(false);
- }
- UnregisterInnerCache(surface);
- surface->MarkAsRegistered(false);
- ReserveSurface(surface->GetSurfaceParams(), surface);
+template <class P>
+void TextureCache<P>::SynchronizeComputeDescriptors() {
+ const bool linked_tsc = kepler_compute.launch_description.linked_tsc;
+ const u32 tic_limit = kepler_compute.regs.tic.limit;
+ const u32 tsc_limit = linked_tsc ? tic_limit : kepler_compute.regs.tsc.limit;
+ const GPUVAddr tsc_gpu_addr = kepler_compute.regs.tsc.Address();
+ if (compute_sampler_table.Synchornize(tsc_gpu_addr, tsc_limit)) {
+ compute_sampler_ids.resize(tsc_limit + 1, CORRUPT_ID);
}
+ if (compute_image_table.Synchornize(kepler_compute.regs.tic.Address(), tic_limit)) {
+ compute_image_view_ids.resize(tic_limit + 1, CORRUPT_ID);
+ }
+}
- TSurface GetUncachedSurface(const GPUVAddr gpu_addr, const SurfaceParams& params) {
- if (const auto surface = TryGetReservedSurface(params); surface) {
- surface->SetGpuAddr(gpu_addr);
- return surface;
- }
- // No reserved surface available, create a new one and reserve it
- auto new_surface{CreateSurface(gpu_addr, params)};
- return new_surface;
+template <class P>
+void TextureCache<P>::UpdateRenderTargets(bool is_clear) {
+ using namespace VideoCommon::Dirty;
+ auto& flags = maxwell3d.dirty.flags;
+ if (!flags[Dirty::RenderTargets]) {
+ return;
}
+ flags[Dirty::RenderTargets] = false;
- const bool is_astc_supported;
+ // Render target control is used on all render targets, so force look ups when this one is up
+ const bool force = flags[Dirty::RenderTargetControl];
+ flags[Dirty::RenderTargetControl] = false;
-private:
- enum class RecycleStrategy : u32 {
- Ignore = 0,
- Flush = 1,
- BufferCopy = 3,
- };
+ for (size_t index = 0; index < NUM_RT; ++index) {
+ ImageViewId& color_buffer_id = render_targets.color_buffer_ids[index];
+ if (flags[Dirty::ColorBuffer0 + index] || force) {
+ flags[Dirty::ColorBuffer0 + index] = false;
+ BindRenderTarget(&color_buffer_id, FindColorBuffer(index, is_clear));
+ }
+ PrepareImageView(color_buffer_id, true, is_clear && IsFullClear(color_buffer_id));
+ }
+ if (flags[Dirty::ZetaBuffer] || force) {
+ flags[Dirty::ZetaBuffer] = false;
+ BindRenderTarget(&render_targets.depth_buffer_id, FindDepthBuffer(is_clear));
+ }
+ const ImageViewId depth_buffer_id = render_targets.depth_buffer_id;
+ PrepareImageView(depth_buffer_id, true, is_clear && IsFullClear(depth_buffer_id));
- enum class DeductionType : u32 {
- DeductionComplete,
- DeductionIncomplete,
- DeductionFailed,
+ for (size_t index = 0; index < NUM_RT; ++index) {
+ render_targets.draw_buffers[index] = static_cast<u8>(maxwell3d.regs.rt_control.Map(index));
+ }
+ render_targets.size = Extent2D{
+ maxwell3d.regs.render_area.width,
+ maxwell3d.regs.render_area.height,
};
+}
- struct Deduction {
- DeductionType type{DeductionType::DeductionFailed};
- TSurface surface{};
+template <class P>
+typename P::Framebuffer* TextureCache<P>::GetFramebuffer() {
+ return &slot_framebuffers[GetFramebufferId(render_targets)];
+}
- bool Failed() const {
- return type == DeductionType::DeductionFailed;
- }
+template <class P>
+void TextureCache<P>::FillImageViews(DescriptorTable<TICEntry>& table,
+ std::span<ImageViewId> cached_image_view_ids,
+ std::span<const u32> indices,
+ std::span<ImageViewId> image_view_ids) {
+ ASSERT(indices.size() <= image_view_ids.size());
+ do {
+ has_deleted_images = false;
+ std::ranges::transform(indices, image_view_ids.begin(), [&](u32 index) {
+ return VisitImageView(table, cached_image_view_ids, index);
+ });
+ } while (has_deleted_images);
+}
- bool Incomplete() const {
- return type == DeductionType::DeductionIncomplete;
- }
+template <class P>
+ImageViewId TextureCache<P>::VisitImageView(DescriptorTable<TICEntry>& table,
+ std::span<ImageViewId> cached_image_view_ids,
+ u32 index) {
+ if (index > table.Limit()) {
+ LOG_ERROR(HW_GPU, "Invalid image view index={}", index);
+ return NULL_IMAGE_VIEW_ID;
+ }
+ const auto [descriptor, is_new] = table.Read(index);
+ ImageViewId& image_view_id = cached_image_view_ids[index];
+ if (is_new) {
+ image_view_id = FindImageView(descriptor);
+ }
+ if (image_view_id != NULL_IMAGE_VIEW_ID) {
+ PrepareImageView(image_view_id, false, false);
+ }
+ return image_view_id;
+}
- bool IsDepth() const {
- return surface->GetSurfaceParams().IsPixelFormatZeta();
- }
- };
+template <class P>
+FramebufferId TextureCache<P>::GetFramebufferId(const RenderTargets& key) {
+ const auto [pair, is_new] = framebuffers.try_emplace(key);
+ FramebufferId& framebuffer_id = pair->second;
+ if (!is_new) {
+ return framebuffer_id;
+ }
+ std::array<ImageView*, NUM_RT> color_buffers;
+ std::ranges::transform(key.color_buffer_ids, color_buffers.begin(),
+ [this](ImageViewId id) { return id ? &slot_image_views[id] : nullptr; });
+ ImageView* const depth_buffer =
+ key.depth_buffer_id ? &slot_image_views[key.depth_buffer_id] : nullptr;
+ framebuffer_id = slot_framebuffers.insert(runtime, color_buffers, depth_buffer, key);
+ return framebuffer_id;
+}
- /**
- * Takes care of selecting a proper strategy to deal with a texture recycle.
- *
- * @param overlaps The overlapping surfaces registered in the cache.
- * @param params The parameters on the new surface.
- * @param gpu_addr The starting address of the new surface.
- * @param untopological Indicates to the recycler that the texture has no way
- * to match the overlaps due to topological reasons.
- **/
- RecycleStrategy PickStrategy(VectorSurface& overlaps, const SurfaceParams& params,
- const GPUVAddr gpu_addr, const MatchTopologyResult untopological) {
- if (Settings::IsGPULevelExtreme()) {
- return RecycleStrategy::Flush;
- }
- // 3D Textures decision
- if (params.target == SurfaceTarget::Texture3D) {
- return RecycleStrategy::Flush;
- }
- for (const auto& s : overlaps) {
- const auto& s_params = s->GetSurfaceParams();
- if (s_params.target == SurfaceTarget::Texture3D) {
- return RecycleStrategy::Flush;
- }
- }
- // Untopological decision
- if (untopological == MatchTopologyResult::CompressUnmatch) {
- return RecycleStrategy::Flush;
- }
- if (untopological == MatchTopologyResult::FullMatch && !params.is_tiled) {
- return RecycleStrategy::Flush;
- }
- return RecycleStrategy::Ignore;
- }
-
- /**
- * Used to decide what to do with textures we can't resolve in the cache It has 2 implemented
- * strategies: Ignore and Flush.
- *
- * - Ignore: Just unregisters all the overlaps and loads the new texture.
- * - Flush: Flushes all the overlaps into memory and loads the new surface from that data.
- *
- * @param overlaps The overlapping surfaces registered in the cache.
- * @param params The parameters for the new surface.
- * @param gpu_addr The starting address of the new surface.
- * @param preserve_contents Indicates that the new surface should be loaded from memory or left
- * blank.
- * @param untopological Indicates to the recycler that the texture has no way to match the
- * overlaps due to topological reasons.
- **/
- std::pair<TSurface, TView> RecycleSurface(VectorSurface& overlaps, const SurfaceParams& params,
- const GPUVAddr gpu_addr, const bool preserve_contents,
- const MatchTopologyResult untopological) {
- const bool do_load = preserve_contents && Settings::IsGPULevelExtreme();
- for (auto& surface : overlaps) {
- Unregister(surface);
- }
- switch (PickStrategy(overlaps, params, gpu_addr, untopological)) {
- case RecycleStrategy::Ignore: {
- return InitializeSurface(gpu_addr, params, do_load);
- }
- case RecycleStrategy::Flush: {
- std::sort(overlaps.begin(), overlaps.end(),
- [](const TSurface& a, const TSurface& b) -> bool {
- return a->GetModificationTick() < b->GetModificationTick();
- });
- for (auto& surface : overlaps) {
- FlushSurface(surface);
- }
- return InitializeSurface(gpu_addr, params, preserve_contents);
+template <class P>
+void TextureCache<P>::WriteMemory(VAddr cpu_addr, size_t size) {
+ ForEachImageInRegion(cpu_addr, size, [this](ImageId image_id, Image& image) {
+ if (True(image.flags & ImageFlagBits::CpuModified)) {
+ return;
}
- case RecycleStrategy::BufferCopy: {
- auto new_surface = GetUncachedSurface(gpu_addr, params);
- BufferCopy(overlaps[0], new_surface);
- return {new_surface, new_surface->GetMainView()};
+ image.flags |= ImageFlagBits::CpuModified;
+ UntrackImage(image);
+ });
+}
+
+template <class P>
+void TextureCache<P>::DownloadMemory(VAddr cpu_addr, size_t size) {
+ std::vector<ImageId> images;
+ ForEachImageInRegion(cpu_addr, size, [this, &images](ImageId image_id, ImageBase& image) {
+ // Skip images that were not modified from the GPU
+ if (False(image.flags & ImageFlagBits::GpuModified)) {
+ return;
}
- default: {
- UNIMPLEMENTED_MSG("Unimplemented Texture Cache Recycling Strategy!");
- return InitializeSurface(gpu_addr, params, do_load);
+ // Skip images that .are. modified from the CPU
+ // We don't want to write sensitive data from the guest
+ if (True(image.flags & ImageFlagBits::CpuModified)) {
+ return;
}
+ if (image.info.num_samples > 1) {
+ LOG_WARNING(HW_GPU, "MSAA image downloads are not implemented");
+ return;
}
+ image.flags &= ~ImageFlagBits::GpuModified;
+ images.push_back(image_id);
+ });
+ if (images.empty()) {
+ return;
+ }
+ std::ranges::sort(images, [this](ImageId lhs, ImageId rhs) {
+ return slot_images[lhs].modification_tick < slot_images[rhs].modification_tick;
+ });
+ for (const ImageId image_id : images) {
+ Image& image = slot_images[image_id];
+ auto map = runtime.MapDownloadBuffer(image.unswizzled_size_bytes);
+ const auto copies = FullDownloadCopies(image.info);
+ image.DownloadMemory(map, 0, copies);
+ runtime.Finish();
+ SwizzleImage(gpu_memory, image.gpu_addr, image.info, copies, map.Span());
}
+}
- /**
- * Takes a single surface and recreates into another that may differ in
- * format, target or width alignment.
- *
- * @param current_surface The registered surface in the cache which we want to convert.
- * @param params The new surface params which we'll use to recreate the surface.
- * @param is_render Whether or not the surface is a render target.
- **/
- std::pair<TSurface, TView> RebuildSurface(TSurface current_surface, const SurfaceParams& params,
- bool is_render) {
- const auto gpu_addr = current_surface->GetGpuAddr();
- const auto& cr_params = current_surface->GetSurfaceParams();
- TSurface new_surface;
- if (cr_params.pixel_format != params.pixel_format && !is_render &&
- GetSiblingFormat(cr_params.pixel_format) == params.pixel_format) {
- SurfaceParams new_params = params;
- new_params.pixel_format = cr_params.pixel_format;
- new_params.type = cr_params.type;
- new_surface = GetUncachedSurface(gpu_addr, new_params);
- } else {
- new_surface = GetUncachedSurface(gpu_addr, params);
- }
- const SurfaceParams& final_params = new_surface->GetSurfaceParams();
- if (cr_params.type != final_params.type) {
- if (Settings::IsGPULevelExtreme()) {
- BufferCopy(current_surface, new_surface);
- }
- } else {
- std::vector<CopyParams> bricks = current_surface->BreakDown(final_params);
- for (auto& brick : bricks) {
- TryCopyImage(current_surface, new_surface, brick);
- }
- }
- Unregister(current_surface);
- Register(new_surface);
- new_surface->MarkAsModified(current_surface->IsModified(), Tick());
- return {new_surface, new_surface->GetMainView()};
- }
-
- /**
- * Takes a single surface and checks with the new surface's params if it's an exact
- * match, we return the main view of the registered surface. If its formats don't
- * match, we rebuild the surface. We call this last method a `Mirage`. If formats
- * match but the targets don't, we create an overview View of the registered surface.
- *
- * @param current_surface The registered surface in the cache which we want to convert.
- * @param params The new surface params which we want to check.
- * @param is_render Whether or not the surface is a render target.
- **/
- std::pair<TSurface, TView> ManageStructuralMatch(TSurface current_surface,
- const SurfaceParams& params, bool is_render) {
- const bool is_mirage = !current_surface->MatchFormat(params.pixel_format);
- const bool matches_target = current_surface->MatchTarget(params.target);
- const auto match_check = [&]() -> std::pair<TSurface, TView> {
- if (matches_target) {
- return {current_surface, current_surface->GetMainView()};
- }
- return {current_surface, current_surface->EmplaceOverview(params)};
- };
- if (!is_mirage) {
- return match_check();
- }
- if (!is_render && GetSiblingFormat(current_surface->GetFormat()) == params.pixel_format) {
- return match_check();
- }
- return RebuildSurface(current_surface, params, is_render);
- }
-
- /**
- * Unlike RebuildSurface where we know whether or not registered surfaces match the candidate
- * in some way, we have no guarantees here. We try to see if the overlaps are sublayers/mipmaps
- * of the new surface, if they all match we end up recreating a surface for them,
- * else we return nothing.
- *
- * @param overlaps The overlapping surfaces registered in the cache.
- * @param params The parameters on the new surface.
- * @param gpu_addr The starting address of the new surface.
- **/
- std::optional<std::pair<TSurface, TView>> TryReconstructSurface(VectorSurface& overlaps,
- const SurfaceParams& params,
- GPUVAddr gpu_addr) {
- if (params.target == SurfaceTarget::Texture3D) {
- return std::nullopt;
- }
- const auto test_modified = [](TSurface& surface) { return surface->IsModified(); };
- TSurface new_surface = GetUncachedSurface(gpu_addr, params);
+template <class P>
+void TextureCache<P>::UnmapMemory(VAddr cpu_addr, size_t size) {
+ std::vector<ImageId> deleted_images;
+ ForEachImageInRegion(cpu_addr, size, [&](ImageId id, Image&) { deleted_images.push_back(id); });
+ for (const ImageId id : deleted_images) {
+ Image& image = slot_images[id];
+ if (True(image.flags & ImageFlagBits::Tracked)) {
+ UntrackImage(image);
+ }
+ UnregisterImage(id);
+ DeleteImage(id);
+ }
+}
- if (std::none_of(overlaps.begin(), overlaps.end(), test_modified)) {
- LoadSurface(new_surface);
- for (const auto& surface : overlaps) {
- Unregister(surface);
- }
- Register(new_surface);
- return {{new_surface, new_surface->GetMainView()}};
- }
+template <class P>
+void TextureCache<P>::BlitImage(const Tegra::Engines::Fermi2D::Surface& dst,
+ const Tegra::Engines::Fermi2D::Surface& src,
+ const Tegra::Engines::Fermi2D::Config& copy) {
+ const BlitImages images = GetBlitImages(dst, src);
+ const ImageId dst_id = images.dst_id;
+ const ImageId src_id = images.src_id;
+ PrepareImage(src_id, false, false);
+ PrepareImage(dst_id, true, false);
+
+ ImageBase& dst_image = slot_images[dst_id];
+ const ImageBase& src_image = slot_images[src_id];
+
+ // TODO: Deduplicate
+ const std::optional dst_base = dst_image.TryFindBase(dst.Address());
+ const SubresourceRange dst_range{.base = dst_base.value(), .extent = {1, 1}};
+ const ImageViewInfo dst_view_info(ImageViewType::e2D, images.dst_format, dst_range);
+ const auto [dst_framebuffer_id, dst_view_id] = RenderTargetFromImage(dst_id, dst_view_info);
+ const auto [src_samples_x, src_samples_y] = SamplesLog2(src_image.info.num_samples);
+ const std::array src_region{
+ Offset2D{.x = copy.src_x0 >> src_samples_x, .y = copy.src_y0 >> src_samples_y},
+ Offset2D{.x = copy.src_x1 >> src_samples_x, .y = copy.src_y1 >> src_samples_y},
+ };
- std::size_t passed_tests = 0;
- for (auto& surface : overlaps) {
- const SurfaceParams& src_params = surface->GetSurfaceParams();
- const auto mipmap_layer{new_surface->GetLayerMipmap(surface->GetGpuAddr())};
- if (!mipmap_layer) {
- continue;
- }
- const auto [base_layer, base_mipmap] = *mipmap_layer;
- if (new_surface->GetMipmapSize(base_mipmap) != surface->GetMipmapSize(0)) {
- continue;
- }
- ++passed_tests;
-
- // Copy all mipmaps and layers
- const u32 block_width = params.GetDefaultBlockWidth();
- const u32 block_height = params.GetDefaultBlockHeight();
- for (u32 mipmap = base_mipmap; mipmap < base_mipmap + src_params.num_levels; ++mipmap) {
- const u32 width = SurfaceParams::IntersectWidth(src_params, params, 0, mipmap);
- const u32 height = SurfaceParams::IntersectHeight(src_params, params, 0, mipmap);
- if (width < block_width || height < block_height) {
- // Current APIs forbid copying small compressed textures, avoid errors
- break;
- }
- const CopyParams copy_params(0, 0, 0, 0, 0, base_layer, 0, mipmap, width, height,
- src_params.depth);
- TryCopyImage(surface, new_surface, copy_params);
- }
- }
- if (passed_tests == 0) {
- return std::nullopt;
- }
- if (Settings::IsGPULevelExtreme() && passed_tests != overlaps.size()) {
- // In Accurate GPU all tests should pass, else we recycle
- return std::nullopt;
- }
+ const std::optional src_base = src_image.TryFindBase(src.Address());
+ const SubresourceRange src_range{.base = src_base.value(), .extent = {1, 1}};
+ const ImageViewInfo src_view_info(ImageViewType::e2D, images.src_format, src_range);
+ const auto [src_framebuffer_id, src_view_id] = RenderTargetFromImage(src_id, src_view_info);
+ const auto [dst_samples_x, dst_samples_y] = SamplesLog2(dst_image.info.num_samples);
+ const std::array dst_region{
+ Offset2D{.x = copy.dst_x0 >> dst_samples_x, .y = copy.dst_y0 >> dst_samples_y},
+ Offset2D{.x = copy.dst_x1 >> dst_samples_x, .y = copy.dst_y1 >> dst_samples_y},
+ };
- const bool modified = std::any_of(overlaps.begin(), overlaps.end(), test_modified);
- for (const auto& surface : overlaps) {
- Unregister(surface);
- }
+ // Always call this after src_framebuffer_id was queried, as the address might be invalidated.
+ Framebuffer* const dst_framebuffer = &slot_framebuffers[dst_framebuffer_id];
+ if constexpr (FRAMEBUFFER_BLITS) {
+ // OpenGL blits from framebuffers, not images
+ Framebuffer* const src_framebuffer = &slot_framebuffers[src_framebuffer_id];
+ runtime.BlitFramebuffer(dst_framebuffer, src_framebuffer, dst_region, src_region,
+ copy.filter, copy.operation);
+ } else {
+ // Vulkan can blit images, but it lacks format reinterpretations
+ // Provide a framebuffer in case it's necessary
+ ImageView& dst_view = slot_image_views[dst_view_id];
+ ImageView& src_view = slot_image_views[src_view_id];
+ runtime.BlitImage(dst_framebuffer, dst_view, src_view, dst_region, src_region, copy.filter,
+ copy.operation);
+ }
+}
- new_surface->MarkAsModified(modified, Tick());
- Register(new_surface);
- return {{new_surface, new_surface->GetMainView()}};
- }
-
- /**
- * Takes care of managing 3D textures and its slices. Does HLE methods for reconstructing the 3D
- * textures within the GPU if possible. Falls back to LLE when it isn't possible to use any of
- * the HLE methods.
- *
- * @param overlaps The overlapping surfaces registered in the cache.
- * @param params The parameters on the new surface.
- * @param gpu_addr The starting address of the new surface.
- * @param cpu_addr The starting address of the new surface on physical memory.
- * @param preserve_contents Indicates that the new surface should be loaded from memory or
- * left blank.
- */
- std::optional<std::pair<TSurface, TView>> Manage3DSurfaces(VectorSurface& overlaps,
- const SurfaceParams& params,
- GPUVAddr gpu_addr, VAddr cpu_addr,
- bool preserve_contents) {
- if (params.target != SurfaceTarget::Texture3D) {
- for (const auto& surface : overlaps) {
- if (!surface->MatchTarget(params.target)) {
- if (overlaps.size() == 1 && surface->GetCpuAddr() == cpu_addr) {
- if (Settings::IsGPULevelExtreme()) {
- return std::nullopt;
- }
- Unregister(surface);
- return InitializeSurface(gpu_addr, params, preserve_contents);
- }
- return std::nullopt;
- }
- if (surface->GetCpuAddr() != cpu_addr) {
- continue;
- }
- if (surface->MatchesStructure(params) == MatchStructureResult::FullMatch) {
- return std::make_pair(surface, surface->GetMainView());
- }
- }
- return InitializeSurface(gpu_addr, params, preserve_contents);
- }
+template <class P>
+void TextureCache<P>::InvalidateColorBuffer(size_t index) {
+ ImageViewId& color_buffer_id = render_targets.color_buffer_ids[index];
+ color_buffer_id = FindColorBuffer(index, false);
+ if (!color_buffer_id) {
+ LOG_ERROR(HW_GPU, "Invalidating invalid color buffer in index={}", index);
+ return;
+ }
+ // When invalidating a color buffer, the old contents are no longer relevant
+ ImageView& color_buffer = slot_image_views[color_buffer_id];
+ Image& image = slot_images[color_buffer.image_id];
+ image.flags &= ~ImageFlagBits::CpuModified;
+ image.flags &= ~ImageFlagBits::GpuModified;
- if (params.num_levels > 1) {
- // We can't handle mipmaps in 3D textures yet, better fallback to LLE approach
- return std::nullopt;
- }
+ runtime.InvalidateColorBuffer(color_buffer, index);
+}
- if (overlaps.size() == 1) {
- const auto& surface = overlaps[0];
- const SurfaceParams& overlap_params = surface->GetSurfaceParams();
- // Don't attempt to render to textures with more than one level for now
- // The texture has to be to the right or the sample address if we want to render to it
- if (overlap_params.num_levels == 1 && cpu_addr >= surface->GetCpuAddr()) {
- const u32 offset = static_cast<u32>(cpu_addr - surface->GetCpuAddr());
- const u32 slice = std::get<2>(params.GetBlockOffsetXYZ(offset));
- if (slice < overlap_params.depth) {
- auto view = surface->Emplace3DView(slice, params.depth, 0, 1);
- return std::make_pair(std::move(surface), std::move(view));
- }
- }
- }
+template <class P>
+void TextureCache<P>::InvalidateDepthBuffer() {
+ ImageViewId& depth_buffer_id = render_targets.depth_buffer_id;
+ depth_buffer_id = FindDepthBuffer(false);
+ if (!depth_buffer_id) {
+ LOG_ERROR(HW_GPU, "Invalidating invalid depth buffer");
+ return;
+ }
+ // When invalidating the depth buffer, the old contents are no longer relevant
+ ImageBase& image = slot_images[slot_image_views[depth_buffer_id].image_id];
+ image.flags &= ~ImageFlagBits::CpuModified;
+ image.flags &= ~ImageFlagBits::GpuModified;
- TSurface new_surface = GetUncachedSurface(gpu_addr, params);
- bool modified = false;
+ ImageView& depth_buffer = slot_image_views[depth_buffer_id];
+ runtime.InvalidateDepthBuffer(depth_buffer);
+}
- for (auto& surface : overlaps) {
- const SurfaceParams& src_params = surface->GetSurfaceParams();
- if (src_params.target != SurfaceTarget::Texture2D ||
- src_params.height != params.height ||
- src_params.block_depth != params.block_depth ||
- src_params.block_height != params.block_height) {
- return std::nullopt;
- }
- modified |= surface->IsModified();
-
- const u32 offset = static_cast<u32>(surface->GetCpuAddr() - cpu_addr);
- const u32 slice = std::get<2>(params.GetBlockOffsetXYZ(offset));
- const u32 width = params.width;
- const u32 height = params.height;
- const CopyParams copy_params(0, 0, 0, 0, 0, slice, 0, 0, width, height, 1);
- TryCopyImage(surface, new_surface, copy_params);
+template <class P>
+typename P::ImageView* TextureCache<P>::TryFindFramebufferImageView(VAddr cpu_addr) {
+ // TODO: Properly implement this
+ const auto it = page_table.find(cpu_addr >> PAGE_SHIFT);
+ if (it == page_table.end()) {
+ return nullptr;
+ }
+ const auto& image_ids = it->second;
+ for (const ImageId image_id : image_ids) {
+ const ImageBase& image = slot_images[image_id];
+ if (image.cpu_addr != cpu_addr) {
+ continue;
}
- for (const auto& surface : overlaps) {
- Unregister(surface);
+ if (image.image_view_ids.empty()) {
+ continue;
}
- new_surface->MarkAsModified(modified, Tick());
- Register(new_surface);
-
- TView view = new_surface->GetMainView();
- return std::make_pair(std::move(new_surface), std::move(view));
- }
-
- /**
- * Gets the starting address and parameters of a candidate surface and tries
- * to find a matching surface within the cache. This is done in 3 big steps:
- *
- * 1. Check the 1st Level Cache in order to find an exact match, if we fail, we move to step 2.
- *
- * 2. Check if there are any overlaps at all, if there are none, we just load the texture from
- * memory else we move to step 3.
- *
- * 3. Consists of figuring out the relationship between the candidate texture and the
- * overlaps. We divide the scenarios depending if there's 1 or many overlaps. If
- * there's many, we just try to reconstruct a new surface out of them based on the
- * candidate's parameters, if we fail, we recycle. When there's only 1 overlap then we
- * have to check if the candidate is a view (layer/mipmap) of the overlap or if the
- * registered surface is a mipmap/layer of the candidate. In this last case we reconstruct
- * a new surface.
- *
- * @param gpu_addr The starting address of the candidate surface.
- * @param params The parameters on the candidate surface.
- * @param preserve_contents Indicates that the new surface should be loaded from memory or
- * left blank.
- * @param is_render Whether or not the surface is a render target.
- **/
- std::pair<TSurface, TView> GetSurface(const GPUVAddr gpu_addr, const VAddr cpu_addr,
- const SurfaceParams& params, bool preserve_contents,
- bool is_render) {
- // Step 1
- // Check Level 1 Cache for a fast structural match. If candidate surface
- // matches at certain level we are pretty much done.
- if (const auto iter = l1_cache.find(cpu_addr); iter != l1_cache.end()) {
- TSurface& current_surface = iter->second;
- const auto topological_result = current_surface->MatchesTopology(params);
- if (topological_result != MatchTopologyResult::FullMatch) {
- VectorSurface overlaps{current_surface};
- return RecycleSurface(overlaps, params, gpu_addr, preserve_contents,
- topological_result);
- }
+ return &slot_image_views[image.image_view_ids.at(0)];
+ }
+ return nullptr;
+}
- const auto struct_result = current_surface->MatchesStructure(params);
- if (struct_result != MatchStructureResult::None) {
- const auto& old_params = current_surface->GetSurfaceParams();
- const bool not_3d = params.target != SurfaceTarget::Texture3D &&
- old_params.target != SurfaceTarget::Texture3D;
- if (not_3d || current_surface->MatchTarget(params.target)) {
- if (struct_result == MatchStructureResult::FullMatch) {
- return ManageStructuralMatch(current_surface, params, is_render);
- } else {
- return RebuildSurface(current_surface, params, is_render);
- }
- }
- }
- }
+template <class P>
+bool TextureCache<P>::HasUncommittedFlushes() const noexcept {
+ return !uncommitted_downloads.empty();
+}
- // Step 2
- // Obtain all possible overlaps in the memory region
- const std::size_t candidate_size = params.GetGuestSizeInBytes();
- auto overlaps{GetSurfacesInRegion(cpu_addr, candidate_size)};
+template <class P>
+bool TextureCache<P>::ShouldWaitAsyncFlushes() const noexcept {
+ return !committed_downloads.empty() && !committed_downloads.front().empty();
+}
- // If none are found, we are done. we just load the surface and create it.
- if (overlaps.empty()) {
- return InitializeSurface(gpu_addr, params, preserve_contents);
- }
+template <class P>
+void TextureCache<P>::CommitAsyncFlushes() {
+ // This is intentionally passing the value by copy
+ committed_downloads.push(uncommitted_downloads);
+ uncommitted_downloads.clear();
+}
- // Step 3
- // Now we need to figure the relationship between the texture and its overlaps
- // we do a topological test to ensure we can find some relationship. If it fails
- // immediately recycle the texture
- for (const auto& surface : overlaps) {
- const auto topological_result = surface->MatchesTopology(params);
- if (topological_result != MatchTopologyResult::FullMatch) {
- return RecycleSurface(overlaps, params, gpu_addr, preserve_contents,
- topological_result);
- }
- }
+template <class P>
+void TextureCache<P>::PopAsyncFlushes() {
+ if (committed_downloads.empty()) {
+ return;
+ }
+ const std::span<const ImageId> download_ids = committed_downloads.front();
+ if (download_ids.empty()) {
+ committed_downloads.pop();
+ return;
+ }
+ size_t total_size_bytes = 0;
+ for (const ImageId image_id : download_ids) {
+ total_size_bytes += slot_images[image_id].unswizzled_size_bytes;
+ }
+ auto download_map = runtime.MapDownloadBuffer(total_size_bytes);
+ size_t buffer_offset = 0;
+ for (const ImageId image_id : download_ids) {
+ Image& image = slot_images[image_id];
+ const auto copies = FullDownloadCopies(image.info);
+ image.DownloadMemory(download_map, buffer_offset, copies);
+ buffer_offset += image.unswizzled_size_bytes;
+ }
+ // Wait for downloads to finish
+ runtime.Finish();
+
+ buffer_offset = 0;
+ const std::span<u8> download_span = download_map.Span();
+ for (const ImageId image_id : download_ids) {
+ const ImageBase& image = slot_images[image_id];
+ const auto copies = FullDownloadCopies(image.info);
+ const std::span<u8> image_download_span = download_span.subspan(buffer_offset);
+ SwizzleImage(gpu_memory, image.gpu_addr, image.info, copies, image_download_span);
+ buffer_offset += image.unswizzled_size_bytes;
+ }
+ committed_downloads.pop();
+}
- // Manage 3D textures
- if (params.block_depth > 0) {
- auto surface =
- Manage3DSurfaces(overlaps, params, gpu_addr, cpu_addr, preserve_contents);
- if (surface) {
- return *surface;
- }
+template <class P>
+bool TextureCache<P>::IsRegionGpuModified(VAddr addr, size_t size) {
+ bool is_modified = false;
+ ForEachImageInRegion(addr, size, [&is_modified](ImageId, ImageBase& image) {
+ if (False(image.flags & ImageFlagBits::GpuModified)) {
+ return false;
}
+ is_modified = true;
+ return true;
+ });
+ return is_modified;
+}
- // Split cases between 1 overlap or many.
- if (overlaps.size() == 1) {
- TSurface current_surface = overlaps[0];
- // First check if the surface is within the overlap. If not, it means
- // two things either the candidate surface is a supertexture of the overlap
- // or they don't match in any known way.
- if (!current_surface->IsInside(gpu_addr, gpu_addr + candidate_size)) {
- const std::optional view = TryReconstructSurface(overlaps, params, gpu_addr);
- if (view) {
- return *view;
- }
- return RecycleSurface(overlaps, params, gpu_addr, preserve_contents,
- MatchTopologyResult::FullMatch);
- }
- // Now we check if the candidate is a mipmap/layer of the overlap
- std::optional<TView> view =
- current_surface->EmplaceView(params, gpu_addr, candidate_size);
- if (view) {
- const bool is_mirage = !current_surface->MatchFormat(params.pixel_format);
- if (is_mirage) {
- // On a mirage view, we need to recreate the surface under this new view
- // and then obtain a view again.
- SurfaceParams new_params = current_surface->GetSurfaceParams();
- const u32 wh = SurfaceParams::ConvertWidth(
- new_params.width, new_params.pixel_format, params.pixel_format);
- const u32 hh = SurfaceParams::ConvertHeight(
- new_params.height, new_params.pixel_format, params.pixel_format);
- new_params.width = wh;
- new_params.height = hh;
- new_params.pixel_format = params.pixel_format;
- std::pair<TSurface, TView> pair =
- RebuildSurface(current_surface, new_params, is_render);
- std::optional<TView> mirage_view =
- pair.first->EmplaceView(params, gpu_addr, candidate_size);
- if (mirage_view)
- return {pair.first, *mirage_view};
- return RecycleSurface(overlaps, params, gpu_addr, preserve_contents,
- MatchTopologyResult::FullMatch);
- }
- return {current_surface, *view};
- }
- } else {
- // If there are many overlaps, odds are they are subtextures of the candidate
- // surface. We try to construct a new surface based on the candidate parameters,
- // using the overlaps. If a single overlap fails, this will fail.
- std::optional<std::pair<TSurface, TView>> view =
- TryReconstructSurface(overlaps, params, gpu_addr);
- if (view) {
- return *view;
- }
- }
- // We failed all the tests, recycle the overlaps into a new texture.
- return RecycleSurface(overlaps, params, gpu_addr, preserve_contents,
- MatchTopologyResult::FullMatch);
- }
-
- /**
- * Gets the starting address and parameters of a candidate surface and tries to find a
- * matching surface within the cache that's similar to it. If there are many textures
- * or the texture found if entirely incompatible, it will fail. If no texture is found, the
- * blit will be unsuccessful.
- *
- * @param gpu_addr The starting address of the candidate surface.
- * @param params The parameters on the candidate surface.
- **/
- Deduction DeduceSurface(const GPUVAddr gpu_addr, const SurfaceParams& params) {
- const std::optional<VAddr> cpu_addr = gpu_memory.GpuToCpuAddress(gpu_addr);
-
- if (!cpu_addr) {
- Deduction result{};
- result.type = DeductionType::DeductionFailed;
- return result;
- }
+template <class P>
+void TextureCache<P>::RefreshContents(Image& image) {
+ if (False(image.flags & ImageFlagBits::CpuModified)) {
+ // Only upload modified images
+ return;
+ }
+ image.flags &= ~ImageFlagBits::CpuModified;
+ TrackImage(image);
- if (const auto iter = l1_cache.find(*cpu_addr); iter != l1_cache.end()) {
- TSurface& current_surface = iter->second;
- const auto topological_result = current_surface->MatchesTopology(params);
- if (topological_result != MatchTopologyResult::FullMatch) {
- Deduction result{};
- result.type = DeductionType::DeductionFailed;
- return result;
- }
- const auto struct_result = current_surface->MatchesStructure(params);
- if (struct_result != MatchStructureResult::None &&
- current_surface->MatchTarget(params.target)) {
- Deduction result{};
- result.type = DeductionType::DeductionComplete;
- result.surface = current_surface;
- return result;
- }
- }
+ if (image.info.num_samples > 1) {
+ LOG_WARNING(HW_GPU, "MSAA image uploads are not implemented");
+ return;
+ }
+ auto map = runtime.MapUploadBuffer(MapSizeBytes(image));
+ UploadImageContents(image, map, 0);
+ runtime.InsertUploadMemoryBarrier();
+}
- const std::size_t candidate_size = params.GetGuestSizeInBytes();
- auto overlaps{GetSurfacesInRegion(*cpu_addr, candidate_size)};
+template <class P>
+template <typename MapBuffer>
+void TextureCache<P>::UploadImageContents(Image& image, MapBuffer& map, size_t buffer_offset) {
+ const std::span<u8> mapped_span = map.Span().subspan(buffer_offset);
+ const GPUVAddr gpu_addr = image.gpu_addr;
+
+ if (True(image.flags & ImageFlagBits::AcceleratedUpload)) {
+ gpu_memory.ReadBlockUnsafe(gpu_addr, mapped_span.data(), mapped_span.size_bytes());
+ const auto uploads = FullUploadSwizzles(image.info);
+ runtime.AccelerateImageUpload(image, map, buffer_offset, uploads);
+ } else if (True(image.flags & ImageFlagBits::Converted)) {
+ std::vector<u8> unswizzled_data(image.unswizzled_size_bytes);
+ auto copies = UnswizzleImage(gpu_memory, gpu_addr, image.info, unswizzled_data);
+ ConvertImage(unswizzled_data, image.info, mapped_span, copies);
+ image.UploadMemory(map, buffer_offset, copies);
+ } else if (image.info.type == ImageType::Buffer) {
+ const std::array copies{UploadBufferCopy(gpu_memory, gpu_addr, image, mapped_span)};
+ image.UploadMemory(map, buffer_offset, copies);
+ } else {
+ const auto copies = UnswizzleImage(gpu_memory, gpu_addr, image.info, mapped_span);
+ image.UploadMemory(map, buffer_offset, copies);
+ }
+}
- if (overlaps.empty()) {
- Deduction result{};
- result.type = DeductionType::DeductionIncomplete;
- return result;
- }
+template <class P>
+ImageViewId TextureCache<P>::FindImageView(const TICEntry& config) {
+ if (!IsValidAddress(gpu_memory, config)) {
+ return NULL_IMAGE_VIEW_ID;
+ }
+ const auto [pair, is_new] = image_views.try_emplace(config);
+ ImageViewId& image_view_id = pair->second;
+ if (is_new) {
+ image_view_id = CreateImageView(config);
+ }
+ return image_view_id;
+}
- if (overlaps.size() > 1) {
- Deduction result{};
- result.type = DeductionType::DeductionFailed;
- return result;
- } else {
- Deduction result{};
- result.type = DeductionType::DeductionComplete;
- result.surface = overlaps[0];
- return result;
- }
+template <class P>
+ImageViewId TextureCache<P>::CreateImageView(const TICEntry& config) {
+ const ImageInfo info(config);
+ const GPUVAddr image_gpu_addr = config.Address() - config.BaseLayer() * info.layer_stride;
+ const ImageId image_id = FindOrInsertImage(info, image_gpu_addr);
+ if (!image_id) {
+ return NULL_IMAGE_VIEW_ID;
}
+ ImageBase& image = slot_images[image_id];
+ const SubresourceBase base = image.TryFindBase(config.Address()).value();
+ ASSERT(base.level == 0);
+ const ImageViewInfo view_info(config, base.layer);
+ const ImageViewId image_view_id = FindOrEmplaceImageView(image_id, view_info);
+ ImageViewBase& image_view = slot_image_views[image_view_id];
+ image_view.flags |= ImageViewFlagBits::Strong;
+ image.flags |= ImageFlagBits::Strong;
+ return image_view_id;
+}
- /**
- * Gets a null surface based on a target texture.
- * @param target The target of the null surface.
- */
- TView GetNullSurface(SurfaceTarget target) {
- const u32 i_target = static_cast<u32>(target);
- if (const auto it = invalid_cache.find(i_target); it != invalid_cache.end()) {
- return it->second->GetMainView();
- }
- SurfaceParams params{};
- params.target = target;
- params.is_tiled = false;
- params.srgb_conversion = false;
- params.is_layered =
- target == SurfaceTarget::Texture1DArray || target == SurfaceTarget::Texture2DArray ||
- target == SurfaceTarget::TextureCubemap || target == SurfaceTarget::TextureCubeArray;
- params.block_width = 0;
- params.block_height = 0;
- params.block_depth = 0;
- params.tile_width_spacing = 1;
- params.width = 1;
- params.height = 1;
- params.depth = 1;
- if (target == SurfaceTarget::TextureCubemap || target == SurfaceTarget::TextureCubeArray) {
- params.depth = 6;
- }
- params.pitch = 4;
- params.num_levels = 1;
- params.emulated_levels = 1;
- params.pixel_format = VideoCore::Surface::PixelFormat::R8_UNORM;
- params.type = VideoCore::Surface::SurfaceType::ColorTexture;
- auto surface = CreateSurface(0ULL, params);
- invalid_memory.resize(surface->GetHostSizeInBytes(), 0U);
- surface->UploadTexture(invalid_memory);
- surface->MarkAsModified(false, Tick());
- invalid_cache.emplace(i_target, surface);
- return surface->GetMainView();
- }
-
- /**
- * Gets the a source and destination starting address and parameters,
- * and tries to deduce if they are supposed to be depth textures. If so, their
- * parameters are modified and fixed into so.
- *
- * @param src_params The parameters of the candidate surface.
- * @param dst_params The parameters of the destination surface.
- * @param src_gpu_addr The starting address of the candidate surface.
- * @param dst_gpu_addr The starting address of the destination surface.
- **/
- void DeduceBestBlit(SurfaceParams& src_params, SurfaceParams& dst_params,
- const GPUVAddr src_gpu_addr, const GPUVAddr dst_gpu_addr) {
- auto deduced_src = DeduceSurface(src_gpu_addr, src_params);
- auto deduced_dst = DeduceSurface(dst_gpu_addr, dst_params);
- if (deduced_src.Failed() || deduced_dst.Failed()) {
- return;
+template <class P>
+ImageId TextureCache<P>::FindOrInsertImage(const ImageInfo& info, GPUVAddr gpu_addr,
+ RelaxedOptions options) {
+ if (const ImageId image_id = FindImage(info, gpu_addr, options); image_id) {
+ return image_id;
+ }
+ return InsertImage(info, gpu_addr, options);
+}
+
+template <class P>
+ImageId TextureCache<P>::FindImage(const ImageInfo& info, GPUVAddr gpu_addr,
+ RelaxedOptions options) {
+ const std::optional<VAddr> cpu_addr = gpu_memory.GpuToCpuAddress(gpu_addr);
+ if (!cpu_addr) {
+ return ImageId{};
+ }
+ ImageId image_id;
+ const auto lambda = [&](ImageId existing_image_id, ImageBase& existing_image) {
+ if (info.type == ImageType::Linear || existing_image.info.type == ImageType::Linear) {
+ const bool strict_size = False(options & RelaxedOptions::Size) &&
+ True(existing_image.flags & ImageFlagBits::Strong);
+ const ImageInfo& existing = existing_image.info;
+ if (existing_image.gpu_addr == gpu_addr && existing.type == info.type &&
+ existing.pitch == info.pitch &&
+ IsPitchLinearSameSize(existing, info, strict_size) &&
+ IsViewCompatible(existing.format, info.format)) {
+ image_id = existing_image_id;
+ return true;
+ }
+ } else if (IsSubresource(info, existing_image, gpu_addr, options)) {
+ image_id = existing_image_id;
+ return true;
}
+ return false;
+ };
+ ForEachImageInRegion(*cpu_addr, CalculateGuestSizeInBytes(info), lambda);
+ return image_id;
+}
- const bool incomplete_src = deduced_src.Incomplete();
- const bool incomplete_dst = deduced_dst.Incomplete();
+template <class P>
+ImageId TextureCache<P>::InsertImage(const ImageInfo& info, GPUVAddr gpu_addr,
+ RelaxedOptions options) {
+ const std::optional<VAddr> cpu_addr = gpu_memory.GpuToCpuAddress(gpu_addr);
+ ASSERT_MSG(cpu_addr, "Tried to insert an image to an invalid gpu_addr=0x{:x}", gpu_addr);
+ const ImageId image_id = JoinImages(info, gpu_addr, *cpu_addr);
+ const Image& image = slot_images[image_id];
+ // Using "image.gpu_addr" instead of "gpu_addr" is important because it might be different
+ const auto [it, is_new] = image_allocs_table.try_emplace(image.gpu_addr);
+ if (is_new) {
+ it->second = slot_image_allocs.insert();
+ }
+ slot_image_allocs[it->second].images.push_back(image_id);
+ return image_id;
+}
- if (incomplete_src && incomplete_dst) {
+template <class P>
+ImageId TextureCache<P>::JoinImages(const ImageInfo& info, GPUVAddr gpu_addr, VAddr cpu_addr) {
+ ImageInfo new_info = info;
+ const size_t size_bytes = CalculateGuestSizeInBytes(new_info);
+ std::vector<ImageId> overlap_ids;
+ std::vector<ImageId> left_aliased_ids;
+ std::vector<ImageId> right_aliased_ids;
+ ForEachImageInRegion(cpu_addr, size_bytes, [&](ImageId overlap_id, ImageBase& overlap) {
+ if (info.type != overlap.info.type) {
return;
}
-
- const bool any_incomplete = incomplete_src || incomplete_dst;
-
- if (!any_incomplete) {
- if (!(deduced_src.IsDepth() && deduced_dst.IsDepth())) {
- return;
- }
- } else {
- if (incomplete_src && !(deduced_dst.IsDepth())) {
- return;
- }
-
- if (incomplete_dst && !(deduced_src.IsDepth())) {
- return;
+ if (info.type == ImageType::Linear) {
+ if (info.pitch == overlap.info.pitch && gpu_addr == overlap.gpu_addr) {
+ // Alias linear images with the same pitch
+ left_aliased_ids.push_back(overlap_id);
}
+ return;
+ }
+ const auto solution = ResolveOverlap(new_info, gpu_addr, cpu_addr, overlap, true);
+ if (solution) {
+ gpu_addr = solution->gpu_addr;
+ cpu_addr = solution->cpu_addr;
+ new_info.resources = solution->resources;
+ overlap_ids.push_back(overlap_id);
+ return;
+ }
+ static constexpr auto options = RelaxedOptions::Size | RelaxedOptions::Format;
+ const ImageBase new_image_base(new_info, gpu_addr, cpu_addr);
+ if (IsSubresource(new_info, overlap, gpu_addr, options)) {
+ left_aliased_ids.push_back(overlap_id);
+ } else if (IsSubresource(overlap.info, new_image_base, overlap.gpu_addr, options)) {
+ right_aliased_ids.push_back(overlap_id);
}
+ });
+ const ImageId new_image_id = slot_images.insert(runtime, new_info, gpu_addr, cpu_addr);
+ Image& new_image = slot_images[new_image_id];
- const auto inherit_format = [](SurfaceParams& to, TSurface from) {
- const SurfaceParams& params = from->GetSurfaceParams();
- to.pixel_format = params.pixel_format;
- to.type = params.type;
- };
- // Now we got the cases where one or both is Depth and the other is not known
- if (!incomplete_src) {
- inherit_format(src_params, deduced_src.surface);
+ // TODO: Only upload what we need
+ RefreshContents(new_image);
+
+ for (const ImageId overlap_id : overlap_ids) {
+ Image& overlap = slot_images[overlap_id];
+ if (overlap.info.num_samples != new_image.info.num_samples) {
+ LOG_WARNING(HW_GPU, "Copying between images with different samples is not implemented");
} else {
- inherit_format(src_params, deduced_dst.surface);
+ const SubresourceBase base = new_image.TryFindBase(overlap.gpu_addr).value();
+ const auto copies = MakeShrinkImageCopies(new_info, overlap.info, base);
+ runtime.CopyImage(new_image, overlap, copies);
}
- if (!incomplete_dst) {
- inherit_format(dst_params, deduced_dst.surface);
- } else {
- inherit_format(dst_params, deduced_src.surface);
+ if (True(overlap.flags & ImageFlagBits::Tracked)) {
+ UntrackImage(overlap);
}
+ UnregisterImage(overlap_id);
+ DeleteImage(overlap_id);
+ }
+ ImageBase& new_image_base = new_image;
+ for (const ImageId aliased_id : right_aliased_ids) {
+ ImageBase& aliased = slot_images[aliased_id];
+ AddImageAlias(new_image_base, aliased, new_image_id, aliased_id);
+ }
+ for (const ImageId aliased_id : left_aliased_ids) {
+ ImageBase& aliased = slot_images[aliased_id];
+ AddImageAlias(aliased, new_image_base, aliased_id, new_image_id);
}
+ RegisterImage(new_image_id);
+ return new_image_id;
+}
- std::pair<TSurface, TView> InitializeSurface(GPUVAddr gpu_addr, const SurfaceParams& params,
- bool preserve_contents) {
- auto new_surface{GetUncachedSurface(gpu_addr, params)};
- Register(new_surface);
- if (preserve_contents) {
- LoadSurface(new_surface);
- }
- return {new_surface, new_surface->GetMainView()};
+template <class P>
+typename TextureCache<P>::BlitImages TextureCache<P>::GetBlitImages(
+ const Tegra::Engines::Fermi2D::Surface& dst, const Tegra::Engines::Fermi2D::Surface& src) {
+ static constexpr auto FIND_OPTIONS = RelaxedOptions::Format | RelaxedOptions::Samples;
+ const GPUVAddr dst_addr = dst.Address();
+ const GPUVAddr src_addr = src.Address();
+ ImageInfo dst_info(dst);
+ ImageInfo src_info(src);
+ ImageId dst_id;
+ ImageId src_id;
+ do {
+ has_deleted_images = false;
+ dst_id = FindImage(dst_info, dst_addr, FIND_OPTIONS);
+ src_id = FindImage(src_info, src_addr, FIND_OPTIONS);
+ const ImageBase* const dst_image = dst_id ? &slot_images[dst_id] : nullptr;
+ const ImageBase* const src_image = src_id ? &slot_images[src_id] : nullptr;
+ DeduceBlitImages(dst_info, src_info, dst_image, src_image);
+ if (GetFormatType(dst_info.format) != GetFormatType(src_info.format)) {
+ continue;
+ }
+ if (!dst_id) {
+ dst_id = InsertImage(dst_info, dst_addr, RelaxedOptions{});
+ }
+ if (!src_id) {
+ src_id = InsertImage(src_info, src_addr, RelaxedOptions{});
+ }
+ } while (has_deleted_images);
+ return BlitImages{
+ .dst_id = dst_id,
+ .src_id = src_id,
+ .dst_format = dst_info.format,
+ .src_format = src_info.format,
+ };
+}
+
+template <class P>
+SamplerId TextureCache<P>::FindSampler(const TSCEntry& config) {
+ if (std::ranges::all_of(config.raw, [](u64 value) { return value == 0; })) {
+ return NULL_SAMPLER_ID;
+ }
+ const auto [pair, is_new] = samplers.try_emplace(config);
+ if (is_new) {
+ pair->second = slot_samplers.insert(runtime, config);
}
+ return pair->second;
+}
- void LoadSurface(const TSurface& surface) {
- staging_cache.GetBuffer(0).resize(surface->GetHostSizeInBytes());
- surface->LoadBuffer(gpu_memory, staging_cache);
- surface->UploadTexture(staging_cache.GetBuffer(0));
- surface->MarkAsModified(false, Tick());
+template <class P>
+ImageViewId TextureCache<P>::FindColorBuffer(size_t index, bool is_clear) {
+ const auto& regs = maxwell3d.regs;
+ if (index >= regs.rt_control.count) {
+ return ImageViewId{};
+ }
+ const auto& rt = regs.rt[index];
+ const GPUVAddr gpu_addr = rt.Address();
+ if (gpu_addr == 0) {
+ return ImageViewId{};
+ }
+ if (rt.format == Tegra::RenderTargetFormat::NONE) {
+ return ImageViewId{};
}
+ const ImageInfo info(regs, index);
+ return FindRenderTargetView(info, gpu_addr, is_clear);
+}
- void FlushSurface(const TSurface& surface) {
- if (!surface->IsModified()) {
- return;
- }
- staging_cache.GetBuffer(0).resize(surface->GetHostSizeInBytes());
- surface->DownloadTexture(staging_cache.GetBuffer(0));
- surface->FlushBuffer(gpu_memory, staging_cache);
- surface->MarkAsModified(false, Tick());
- }
-
- void RegisterInnerCache(TSurface& surface) {
- const VAddr cpu_addr = surface->GetCpuAddr();
- VAddr start = cpu_addr >> registry_page_bits;
- const VAddr end = (surface->GetCpuAddrEnd() - 1) >> registry_page_bits;
- l1_cache[cpu_addr] = surface;
- while (start <= end) {
- registry[start].push_back(surface);
- start++;
- }
+template <class P>
+ImageViewId TextureCache<P>::FindDepthBuffer(bool is_clear) {
+ const auto& regs = maxwell3d.regs;
+ if (!regs.zeta_enable) {
+ return ImageViewId{};
+ }
+ const GPUVAddr gpu_addr = regs.zeta.Address();
+ if (gpu_addr == 0) {
+ return ImageViewId{};
}
+ const ImageInfo info(regs);
+ return FindRenderTargetView(info, gpu_addr, is_clear);
+}
- void UnregisterInnerCache(TSurface& surface) {
- const VAddr cpu_addr = surface->GetCpuAddr();
- VAddr start = cpu_addr >> registry_page_bits;
- const VAddr end = (surface->GetCpuAddrEnd() - 1) >> registry_page_bits;
- l1_cache.erase(cpu_addr);
- while (start <= end) {
- auto& reg{registry[start]};
- reg.erase(std::find(reg.begin(), reg.end(), surface));
- start++;
- }
+template <class P>
+ImageViewId TextureCache<P>::FindRenderTargetView(const ImageInfo& info, GPUVAddr gpu_addr,
+ bool is_clear) {
+ const auto options = is_clear ? RelaxedOptions::Samples : RelaxedOptions{};
+ const ImageId image_id = FindOrInsertImage(info, gpu_addr, options);
+ if (!image_id) {
+ return NULL_IMAGE_VIEW_ID;
+ }
+ Image& image = slot_images[image_id];
+ const ImageViewType view_type = RenderTargetImageViewType(info);
+ SubresourceBase base;
+ if (image.info.type == ImageType::Linear) {
+ base = SubresourceBase{.level = 0, .layer = 0};
+ } else {
+ base = image.TryFindBase(gpu_addr).value();
}
+ const s32 layers = image.info.type == ImageType::e3D ? info.size.depth : info.resources.layers;
+ const SubresourceRange range{
+ .base = base,
+ .extent = {.levels = 1, .layers = layers},
+ };
+ return FindOrEmplaceImageView(image_id, ImageViewInfo(view_type, info.format, range));
+}
- VectorSurface GetSurfacesInRegion(const VAddr cpu_addr, const std::size_t size) {
- if (size == 0) {
- return {};
+template <class P>
+template <typename Func>
+void TextureCache<P>::ForEachImageInRegion(VAddr cpu_addr, size_t size, Func&& func) {
+ using FuncReturn = typename std::invoke_result<Func, ImageId, Image&>::type;
+ static constexpr bool BOOL_BREAK = std::is_same_v<FuncReturn, bool>;
+ boost::container::small_vector<ImageId, 32> images;
+ ForEachPage(cpu_addr, size, [this, &images, cpu_addr, size, func](u64 page) {
+ const auto it = page_table.find(page);
+ if (it == page_table.end()) {
+ if constexpr (BOOL_BREAK) {
+ return false;
+ } else {
+ return;
+ }
}
- const VAddr cpu_addr_end = cpu_addr + size;
- const VAddr end = (cpu_addr_end - 1) >> registry_page_bits;
- VectorSurface surfaces;
- for (VAddr start = cpu_addr >> registry_page_bits; start <= end; ++start) {
- const auto it = registry.find(start);
- if (it == registry.end()) {
+ for (const ImageId image_id : it->second) {
+ Image& image = slot_images[image_id];
+ if (True(image.flags & ImageFlagBits::Picked)) {
continue;
}
- for (auto& surface : it->second) {
- if (surface->IsPicked() || !surface->Overlaps(cpu_addr, cpu_addr_end)) {
- continue;
+ if (!image.Overlaps(cpu_addr, size)) {
+ continue;
+ }
+ image.flags |= ImageFlagBits::Picked;
+ images.push_back(image_id);
+ if constexpr (BOOL_BREAK) {
+ if (func(image_id, image)) {
+ return true;
}
- surface->MarkAsPicked(true);
- surfaces.push_back(surface);
+ } else {
+ func(image_id, image);
}
}
- for (auto& surface : surfaces) {
- surface->MarkAsPicked(false);
+ if constexpr (BOOL_BREAK) {
+ return false;
}
- return surfaces;
+ });
+ for (const ImageId image_id : images) {
+ slot_images[image_id].flags &= ~ImageFlagBits::Picked;
}
+}
- void ReserveSurface(const SurfaceParams& params, TSurface surface) {
- surface_reserve[params].push_back(std::move(surface));
+template <class P>
+ImageViewId TextureCache<P>::FindOrEmplaceImageView(ImageId image_id, const ImageViewInfo& info) {
+ Image& image = slot_images[image_id];
+ if (const ImageViewId image_view_id = image.FindView(info); image_view_id) {
+ return image_view_id;
}
+ const ImageViewId image_view_id = slot_image_views.insert(runtime, info, image_id, image);
+ image.InsertView(info, image_view_id);
+ return image_view_id;
+}
+
+template <class P>
+void TextureCache<P>::RegisterImage(ImageId image_id) {
+ ImageBase& image = slot_images[image_id];
+ ASSERT_MSG(False(image.flags & ImageFlagBits::Registered),
+ "Trying to register an already registered image");
+ image.flags |= ImageFlagBits::Registered;
+ ForEachPage(image.cpu_addr, image.guest_size_bytes,
+ [this, image_id](u64 page) { page_table[page].push_back(image_id); });
+}
- TSurface TryGetReservedSurface(const SurfaceParams& params) {
- auto search{surface_reserve.find(params)};
- if (search == surface_reserve.end()) {
- return {};
+template <class P>
+void TextureCache<P>::UnregisterImage(ImageId image_id) {
+ Image& image = slot_images[image_id];
+ ASSERT_MSG(True(image.flags & ImageFlagBits::Registered),
+ "Trying to unregister an already registered image");
+ image.flags &= ~ImageFlagBits::Registered;
+ ForEachPage(image.cpu_addr, image.guest_size_bytes, [this, image_id](u64 page) {
+ const auto page_it = page_table.find(page);
+ if (page_it == page_table.end()) {
+ UNREACHABLE_MSG("Unregistering unregistered page=0x{:x}", page << PAGE_SHIFT);
+ return;
}
- for (auto& surface : search->second) {
- if (!surface->IsRegistered()) {
- return surface;
- }
+ std::vector<ImageId>& image_ids = page_it->second;
+ const auto vector_it = std::ranges::find(image_ids, image_id);
+ if (vector_it == image_ids.end()) {
+ UNREACHABLE_MSG("Unregistering unregistered image in page=0x{:x}", page << PAGE_SHIFT);
+ return;
}
- return {};
- }
+ image_ids.erase(vector_it);
+ });
+}
- /// Try to do an image copy logging when formats are incompatible.
- void TryCopyImage(TSurface& src, TSurface& dst, const CopyParams& copy) {
- const SurfaceParams& src_params = src->GetSurfaceParams();
- const SurfaceParams& dst_params = dst->GetSurfaceParams();
- if (!format_compatibility.TestCopy(src_params.pixel_format, dst_params.pixel_format)) {
- LOG_ERROR(HW_GPU, "Illegal copy between formats={{{}, {}}}",
- static_cast<int>(dst_params.pixel_format),
- static_cast<int>(src_params.pixel_format));
- return;
+template <class P>
+void TextureCache<P>::TrackImage(ImageBase& image) {
+ ASSERT(False(image.flags & ImageFlagBits::Tracked));
+ image.flags |= ImageFlagBits::Tracked;
+ rasterizer.UpdatePagesCachedCount(image.cpu_addr, image.guest_size_bytes, 1);
+}
+
+template <class P>
+void TextureCache<P>::UntrackImage(ImageBase& image) {
+ ASSERT(True(image.flags & ImageFlagBits::Tracked));
+ image.flags &= ~ImageFlagBits::Tracked;
+ rasterizer.UpdatePagesCachedCount(image.cpu_addr, image.guest_size_bytes, -1);
+}
+
+template <class P>
+void TextureCache<P>::DeleteImage(ImageId image_id) {
+ ImageBase& image = slot_images[image_id];
+ const GPUVAddr gpu_addr = image.gpu_addr;
+ const auto alloc_it = image_allocs_table.find(gpu_addr);
+ if (alloc_it == image_allocs_table.end()) {
+ UNREACHABLE_MSG("Trying to delete an image alloc that does not exist in address 0x{:x}",
+ gpu_addr);
+ return;
+ }
+ const ImageAllocId alloc_id = alloc_it->second;
+ std::vector<ImageId>& alloc_images = slot_image_allocs[alloc_id].images;
+ const auto alloc_image_it = std::ranges::find(alloc_images, image_id);
+ if (alloc_image_it == alloc_images.end()) {
+ UNREACHABLE_MSG("Trying to delete an image that does not exist");
+ return;
+ }
+ ASSERT_MSG(False(image.flags & ImageFlagBits::Tracked), "Image was not untracked");
+ ASSERT_MSG(False(image.flags & ImageFlagBits::Registered), "Image was not unregistered");
+
+ // Mark render targets as dirty
+ auto& dirty = maxwell3d.dirty.flags;
+ dirty[Dirty::RenderTargets] = true;
+ dirty[Dirty::ZetaBuffer] = true;
+ for (size_t rt = 0; rt < NUM_RT; ++rt) {
+ dirty[Dirty::ColorBuffer0 + rt] = true;
+ }
+ const std::span<const ImageViewId> image_view_ids = image.image_view_ids;
+ for (const ImageViewId image_view_id : image_view_ids) {
+ std::ranges::replace(render_targets.color_buffer_ids, image_view_id, ImageViewId{});
+ if (render_targets.depth_buffer_id == image_view_id) {
+ render_targets.depth_buffer_id = ImageViewId{};
}
- ImageCopy(src, dst, copy);
}
+ RemoveImageViewReferences(image_view_ids);
+ RemoveFramebuffers(image_view_ids);
+
+ for (const AliasedImage& alias : image.aliased_images) {
+ ImageBase& other_image = slot_images[alias.id];
+ [[maybe_unused]] const size_t num_removed_aliases =
+ std::erase_if(other_image.aliased_images, [image_id](const AliasedImage& other_alias) {
+ return other_alias.id == image_id;
+ });
+ ASSERT_MSG(num_removed_aliases == 1, "Invalid number of removed aliases: {}",
+ num_removed_aliases);
+ }
+ for (const ImageViewId image_view_id : image_view_ids) {
+ sentenced_image_view.Push(std::move(slot_image_views[image_view_id]));
+ slot_image_views.erase(image_view_id);
+ }
+ sentenced_images.Push(std::move(slot_images[image_id]));
+ slot_images.erase(image_id);
- constexpr PixelFormat GetSiblingFormat(PixelFormat format) const {
- return siblings_table[static_cast<std::size_t>(format)];
+ alloc_images.erase(alloc_image_it);
+ if (alloc_images.empty()) {
+ image_allocs_table.erase(alloc_it);
}
+ if constexpr (ENABLE_VALIDATION) {
+ std::ranges::fill(graphics_image_view_ids, CORRUPT_ID);
+ std::ranges::fill(compute_image_view_ids, CORRUPT_ID);
+ }
+ graphics_image_table.Invalidate();
+ compute_image_table.Invalidate();
+ has_deleted_images = true;
+}
- /// Returns true the shader sampler entry is compatible with the TIC texture type.
- static bool IsTypeCompatible(Tegra::Texture::TextureType tic_type,
- const VideoCommon::Shader::Sampler& entry) {
- const auto shader_type = entry.type;
- switch (tic_type) {
- case Tegra::Texture::TextureType::Texture1D:
- case Tegra::Texture::TextureType::Texture1DArray:
- return shader_type == Tegra::Shader::TextureType::Texture1D;
- case Tegra::Texture::TextureType::Texture1DBuffer:
- // TODO(Rodrigo): Assume as valid for now
- return true;
- case Tegra::Texture::TextureType::Texture2D:
- case Tegra::Texture::TextureType::Texture2DNoMipmap:
- return shader_type == Tegra::Shader::TextureType::Texture2D;
- case Tegra::Texture::TextureType::Texture2DArray:
- return shader_type == Tegra::Shader::TextureType::Texture2D ||
- shader_type == Tegra::Shader::TextureType::TextureCube;
- case Tegra::Texture::TextureType::Texture3D:
- return shader_type == Tegra::Shader::TextureType::Texture3D;
- case Tegra::Texture::TextureType::TextureCubeArray:
- case Tegra::Texture::TextureType::TextureCubemap:
- if (shader_type == Tegra::Shader::TextureType::TextureCube) {
- return true;
- }
- return shader_type == Tegra::Shader::TextureType::Texture2D && entry.is_array;
+template <class P>
+void TextureCache<P>::RemoveImageViewReferences(std::span<const ImageViewId> removed_views) {
+ auto it = image_views.begin();
+ while (it != image_views.end()) {
+ const auto found = std::ranges::find(removed_views, it->second);
+ if (found != removed_views.end()) {
+ it = image_views.erase(it);
+ } else {
+ ++it;
}
- UNREACHABLE();
- return true;
}
+}
- struct FramebufferTargetInfo {
- TSurface target;
- TView view;
- };
-
- void AsyncFlushSurface(TSurface& surface) {
- if (!uncommitted_flushes) {
- uncommitted_flushes = std::make_shared<std::list<TSurface>>();
+template <class P>
+void TextureCache<P>::RemoveFramebuffers(std::span<const ImageViewId> removed_views) {
+ auto it = framebuffers.begin();
+ while (it != framebuffers.end()) {
+ if (it->first.Contains(removed_views)) {
+ it = framebuffers.erase(it);
+ } else {
+ ++it;
}
- uncommitted_flushes->push_back(surface);
}
+}
- VideoCore::RasterizerInterface& rasterizer;
- Tegra::Engines::Maxwell3D& maxwell3d;
- Tegra::MemoryManager& gpu_memory;
-
- FormatLookupTable format_lookup_table;
- FormatCompatibility format_compatibility;
-
- u64 ticks{};
-
- // Guards the cache for protection conflicts.
- bool guard_render_targets{};
- bool guard_samplers{};
-
- // The siblings table is for formats that can inter exchange with one another
- // without causing issues. This is only valid when a conflict occurs on a non
- // rendering use.
- std::array<PixelFormat, static_cast<std::size_t>(PixelFormat::Max)> siblings_table;
-
- // The internal Cache is different for the Texture Cache. It's based on buckets
- // of 1MB. This fits better for the purpose of this cache as textures are normaly
- // large in size.
- static constexpr u64 registry_page_bits{20};
- static constexpr u64 registry_page_size{1 << registry_page_bits};
- std::unordered_map<VAddr, std::vector<TSurface>> registry;
+template <class P>
+void TextureCache<P>::MarkModification(ImageBase& image) noexcept {
+ image.flags |= ImageFlagBits::GpuModified;
+ image.modification_tick = ++modification_tick;
+}
- static constexpr u32 DEPTH_RT = 8;
- static constexpr u32 NO_RT = 0xFFFFFFFF;
+template <class P>
+void TextureCache<P>::SynchronizeAliases(ImageId image_id) {
+ boost::container::small_vector<const AliasedImage*, 1> aliased_images;
+ ImageBase& image = slot_images[image_id];
+ u64 most_recent_tick = image.modification_tick;
+ for (const AliasedImage& aliased : image.aliased_images) {
+ ImageBase& aliased_image = slot_images[aliased.id];
+ if (image.modification_tick < aliased_image.modification_tick) {
+ most_recent_tick = std::max(most_recent_tick, aliased_image.modification_tick);
+ aliased_images.push_back(&aliased);
+ }
+ }
+ if (aliased_images.empty()) {
+ return;
+ }
+ image.modification_tick = most_recent_tick;
+ std::ranges::sort(aliased_images, [this](const AliasedImage* lhs, const AliasedImage* rhs) {
+ const ImageBase& lhs_image = slot_images[lhs->id];
+ const ImageBase& rhs_image = slot_images[rhs->id];
+ return lhs_image.modification_tick < rhs_image.modification_tick;
+ });
+ for (const AliasedImage* const aliased : aliased_images) {
+ CopyImage(image_id, aliased->id, aliased->copies);
+ }
+}
- // The L1 Cache is used for fast texture lookup before checking the overlaps
- // This avoids calculating size and other stuffs.
- std::unordered_map<VAddr, TSurface> l1_cache;
+template <class P>
+void TextureCache<P>::PrepareImage(ImageId image_id, bool is_modification, bool invalidate) {
+ Image& image = slot_images[image_id];
+ if (invalidate) {
+ image.flags &= ~(ImageFlagBits::CpuModified | ImageFlagBits::GpuModified);
+ if (False(image.flags & ImageFlagBits::Tracked)) {
+ TrackImage(image);
+ }
+ } else {
+ RefreshContents(image);
+ SynchronizeAliases(image_id);
+ }
+ if (is_modification) {
+ MarkModification(image);
+ }
+ image.frame_tick = frame_tick;
+}
- /// The surface reserve is a "backup" cache, this is where we put unique surfaces that have
- /// previously been used. This is to prevent surfaces from being constantly created and
- /// destroyed when used with different surface parameters.
- std::unordered_map<SurfaceParams, std::vector<TSurface>> surface_reserve;
- std::array<FramebufferTargetInfo, Tegra::Engines::Maxwell3D::Regs::NumRenderTargets>
- render_targets;
- FramebufferTargetInfo depth_buffer;
+template <class P>
+void TextureCache<P>::PrepareImageView(ImageViewId image_view_id, bool is_modification,
+ bool invalidate) {
+ if (!image_view_id) {
+ return;
+ }
+ const ImageViewBase& image_view = slot_image_views[image_view_id];
+ PrepareImage(image_view.image_id, is_modification, invalidate);
+}
- std::vector<TSurface> sampled_textures;
+template <class P>
+void TextureCache<P>::CopyImage(ImageId dst_id, ImageId src_id, std::span<const ImageCopy> copies) {
+ Image& dst = slot_images[dst_id];
+ Image& src = slot_images[src_id];
+ const auto dst_format_type = GetFormatType(dst.info.format);
+ const auto src_format_type = GetFormatType(src.info.format);
+ if (src_format_type == dst_format_type) {
+ if constexpr (HAS_EMULATED_COPIES) {
+ if (!runtime.CanImageBeCopied(dst, src)) {
+ return runtime.EmulateCopyImage(dst, src, copies);
+ }
+ }
+ return runtime.CopyImage(dst, src, copies);
+ }
+ UNIMPLEMENTED_IF(dst.info.type != ImageType::e2D);
+ UNIMPLEMENTED_IF(src.info.type != ImageType::e2D);
+ for (const ImageCopy& copy : copies) {
+ UNIMPLEMENTED_IF(copy.dst_subresource.num_layers != 1);
+ UNIMPLEMENTED_IF(copy.src_subresource.num_layers != 1);
+ UNIMPLEMENTED_IF(copy.src_offset != Offset3D{});
+ UNIMPLEMENTED_IF(copy.dst_offset != Offset3D{});
+
+ const SubresourceBase dst_base{
+ .level = copy.dst_subresource.base_level,
+ .layer = copy.dst_subresource.base_layer,
+ };
+ const SubresourceBase src_base{
+ .level = copy.src_subresource.base_level,
+ .layer = copy.src_subresource.base_layer,
+ };
+ const SubresourceExtent dst_extent{.levels = 1, .layers = 1};
+ const SubresourceExtent src_extent{.levels = 1, .layers = 1};
+ const SubresourceRange dst_range{.base = dst_base, .extent = dst_extent};
+ const SubresourceRange src_range{.base = src_base, .extent = src_extent};
+ const ImageViewInfo dst_view_info(ImageViewType::e2D, dst.info.format, dst_range);
+ const ImageViewInfo src_view_info(ImageViewType::e2D, src.info.format, src_range);
+ const auto [dst_framebuffer_id, dst_view_id] = RenderTargetFromImage(dst_id, dst_view_info);
+ Framebuffer* const dst_framebuffer = &slot_framebuffers[dst_framebuffer_id];
+ const ImageViewId src_view_id = FindOrEmplaceImageView(src_id, src_view_info);
+ ImageView& dst_view = slot_image_views[dst_view_id];
+ ImageView& src_view = slot_image_views[src_view_id];
+ [[maybe_unused]] const Extent3D expected_size{
+ .width = std::min(dst_view.size.width, src_view.size.width),
+ .height = std::min(dst_view.size.height, src_view.size.height),
+ .depth = std::min(dst_view.size.depth, src_view.size.depth),
+ };
+ UNIMPLEMENTED_IF(copy.extent != expected_size);
- /// This cache stores null surfaces in order to be used as a placeholder
- /// for invalid texture calls.
- std::unordered_map<u32, TSurface> invalid_cache;
- std::vector<u8> invalid_memory;
+ runtime.ConvertImage(dst_framebuffer, dst_view, src_view);
+ }
+}
- std::list<TSurface> marked_for_unregister;
+template <class P>
+void TextureCache<P>::BindRenderTarget(ImageViewId* old_id, ImageViewId new_id) {
+ if (*old_id == new_id) {
+ return;
+ }
+ if (*old_id) {
+ const ImageViewBase& old_view = slot_image_views[*old_id];
+ if (True(old_view.flags & ImageViewFlagBits::PreemtiveDownload)) {
+ uncommitted_downloads.push_back(old_view.image_id);
+ }
+ }
+ *old_id = new_id;
+}
- std::shared_ptr<std::list<TSurface>> uncommitted_flushes{};
- std::list<std::shared_ptr<std::list<TSurface>>> committed_flushes;
+template <class P>
+std::pair<FramebufferId, ImageViewId> TextureCache<P>::RenderTargetFromImage(
+ ImageId image_id, const ImageViewInfo& view_info) {
+ const ImageViewId view_id = FindOrEmplaceImageView(image_id, view_info);
+ const ImageBase& image = slot_images[image_id];
+ const bool is_color = GetFormatType(image.info.format) == SurfaceType::ColorTexture;
+ const ImageViewId color_view_id = is_color ? view_id : ImageViewId{};
+ const ImageViewId depth_view_id = is_color ? ImageViewId{} : view_id;
+ const Extent3D extent = MipSize(image.info.size, view_info.range.base.level);
+ const u32 num_samples = image.info.num_samples;
+ const auto [samples_x, samples_y] = SamplesLog2(num_samples);
+ const FramebufferId framebuffer_id = GetFramebufferId(RenderTargets{
+ .color_buffer_ids = {color_view_id},
+ .depth_buffer_id = depth_view_id,
+ .size = {extent.width >> samples_x, extent.height >> samples_y},
+ });
+ return {framebuffer_id, view_id};
+}
- StagingCache staging_cache;
- std::recursive_mutex mutex;
-};
+template <class P>
+bool TextureCache<P>::IsFullClear(ImageViewId id) {
+ if (!id) {
+ return true;
+ }
+ const ImageViewBase& image_view = slot_image_views[id];
+ const ImageBase& image = slot_images[image_view.image_id];
+ const Extent3D size = image_view.size;
+ const auto& regs = maxwell3d.regs;
+ const auto& scissor = regs.scissor_test[0];
+ if (image.info.resources.levels > 1 || image.info.resources.layers > 1) {
+ // Images with multiple resources can't be cleared in a single call
+ return false;
+ }
+ if (regs.clear_flags.scissor == 0) {
+ // If scissor testing is disabled, the clear is always full
+ return true;
+ }
+ // Make sure the clear covers all texels in the subresource
+ return scissor.min_x == 0 && scissor.min_y == 0 && scissor.max_x >= size.width &&
+ scissor.max_y >= size.height;
+}
} // namespace VideoCommon
diff --git a/src/video_core/texture_cache/types.h b/src/video_core/texture_cache/types.h
new file mode 100644
index 000000000..2ad2d72a6
--- /dev/null
+++ b/src/video_core/texture_cache/types.h
@@ -0,0 +1,140 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "common/common_funcs.h"
+#include "common/common_types.h"
+#include "video_core/texture_cache/slot_vector.h"
+
+namespace VideoCommon {
+
+constexpr size_t NUM_RT = 8;
+constexpr size_t MAX_MIP_LEVELS = 14;
+
+constexpr SlotId CORRUPT_ID{0xfffffffe};
+
+using ImageId = SlotId;
+using ImageViewId = SlotId;
+using ImageAllocId = SlotId;
+using SamplerId = SlotId;
+using FramebufferId = SlotId;
+
+enum class ImageType : u32 {
+ e1D,
+ e2D,
+ e3D,
+ Linear,
+ Buffer,
+};
+
+enum class ImageViewType : u32 {
+ e1D,
+ e2D,
+ Cube,
+ e3D,
+ e1DArray,
+ e2DArray,
+ CubeArray,
+ Rect,
+ Buffer,
+};
+constexpr size_t NUM_IMAGE_VIEW_TYPES = 9;
+
+enum class RelaxedOptions : u32 {
+ Size = 1 << 0,
+ Format = 1 << 1,
+ Samples = 1 << 2,
+};
+DECLARE_ENUM_FLAG_OPERATORS(RelaxedOptions)
+
+struct Offset2D {
+ constexpr auto operator<=>(const Offset2D&) const noexcept = default;
+
+ s32 x;
+ s32 y;
+};
+
+struct Offset3D {
+ constexpr auto operator<=>(const Offset3D&) const noexcept = default;
+
+ s32 x;
+ s32 y;
+ s32 z;
+};
+
+struct Extent2D {
+ constexpr auto operator<=>(const Extent2D&) const noexcept = default;
+
+ u32 width;
+ u32 height;
+};
+
+struct Extent3D {
+ constexpr auto operator<=>(const Extent3D&) const noexcept = default;
+
+ u32 width;
+ u32 height;
+ u32 depth;
+};
+
+struct SubresourceLayers {
+ s32 base_level = 0;
+ s32 base_layer = 0;
+ s32 num_layers = 1;
+};
+
+struct SubresourceBase {
+ constexpr auto operator<=>(const SubresourceBase&) const noexcept = default;
+
+ s32 level = 0;
+ s32 layer = 0;
+};
+
+struct SubresourceExtent {
+ constexpr auto operator<=>(const SubresourceExtent&) const noexcept = default;
+
+ s32 levels = 1;
+ s32 layers = 1;
+};
+
+struct SubresourceRange {
+ constexpr auto operator<=>(const SubresourceRange&) const noexcept = default;
+
+ SubresourceBase base;
+ SubresourceExtent extent;
+};
+
+struct ImageCopy {
+ SubresourceLayers src_subresource;
+ SubresourceLayers dst_subresource;
+ Offset3D src_offset;
+ Offset3D dst_offset;
+ Extent3D extent;
+};
+
+struct BufferImageCopy {
+ size_t buffer_offset;
+ size_t buffer_size;
+ u32 buffer_row_length;
+ u32 buffer_image_height;
+ SubresourceLayers image_subresource;
+ Offset3D image_offset;
+ Extent3D image_extent;
+};
+
+struct BufferCopy {
+ size_t src_offset;
+ size_t dst_offset;
+ size_t size;
+};
+
+struct SwizzleParameters {
+ Extent3D num_tiles;
+ Extent3D block;
+ size_t buffer_offset;
+ s32 level;
+};
+
+} // namespace VideoCommon
diff --git a/src/video_core/texture_cache/util.cpp b/src/video_core/texture_cache/util.cpp
new file mode 100644
index 000000000..9ed1fc007
--- /dev/null
+++ b/src/video_core/texture_cache/util.cpp
@@ -0,0 +1,1232 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+// This files contains code from Ryujinx
+// A copy of the code can be obtained from https://github.com/Ryujinx/Ryujinx
+// The sections using code from Ryujinx are marked with a link to the original version
+
+// MIT License
+//
+// Copyright (c) Ryujinx Team and Contributors
+//
+// Permission is hereby granted, free of charge, to any person obtaining a copy of this software and
+// associated documentation files (the "Software"), to deal in the Software without restriction,
+// including without limitation the rights to use, copy, modify, merge, publish, distribute,
+// sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is
+// furnished to do so, subject to the following conditions:
+//
+// The above copyright notice and this permission notice shall be included in all copies or
+// substantial portions of the Software.
+//
+// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT
+// NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+// NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM,
+// DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+//
+
+#include <algorithm>
+#include <array>
+#include <numeric>
+#include <optional>
+#include <span>
+#include <vector>
+
+#include "common/alignment.h"
+#include "common/assert.h"
+#include "common/bit_util.h"
+#include "common/common_types.h"
+#include "common/div_ceil.h"
+#include "video_core/compatible_formats.h"
+#include "video_core/engines/maxwell_3d.h"
+#include "video_core/memory_manager.h"
+#include "video_core/surface.h"
+#include "video_core/texture_cache/decode_bc4.h"
+#include "video_core/texture_cache/format_lookup_table.h"
+#include "video_core/texture_cache/formatter.h"
+#include "video_core/texture_cache/samples_helper.h"
+#include "video_core/texture_cache/util.h"
+#include "video_core/textures/astc.h"
+#include "video_core/textures/decoders.h"
+
+namespace VideoCommon {
+
+namespace {
+
+using Tegra::Texture::GOB_SIZE;
+using Tegra::Texture::GOB_SIZE_SHIFT;
+using Tegra::Texture::GOB_SIZE_X;
+using Tegra::Texture::GOB_SIZE_X_SHIFT;
+using Tegra::Texture::GOB_SIZE_Y;
+using Tegra::Texture::GOB_SIZE_Y_SHIFT;
+using Tegra::Texture::GOB_SIZE_Z;
+using Tegra::Texture::GOB_SIZE_Z_SHIFT;
+using Tegra::Texture::MsaaMode;
+using Tegra::Texture::SwizzleTexture;
+using Tegra::Texture::TextureFormat;
+using Tegra::Texture::TextureType;
+using Tegra::Texture::TICEntry;
+using Tegra::Texture::UnswizzleTexture;
+using VideoCore::Surface::BytesPerBlock;
+using VideoCore::Surface::DefaultBlockHeight;
+using VideoCore::Surface::DefaultBlockWidth;
+using VideoCore::Surface::IsCopyCompatible;
+using VideoCore::Surface::IsPixelFormatASTC;
+using VideoCore::Surface::IsViewCompatible;
+using VideoCore::Surface::PixelFormatFromDepthFormat;
+using VideoCore::Surface::PixelFormatFromRenderTargetFormat;
+using VideoCore::Surface::SurfaceType;
+
+constexpr u32 CONVERTED_BYTES_PER_BLOCK = BytesPerBlock(PixelFormat::A8B8G8R8_UNORM);
+
+struct LevelInfo {
+ Extent3D size;
+ Extent3D block;
+ Extent2D tile_size;
+ u32 bpp_log2;
+ u32 tile_width_spacing;
+};
+
+[[nodiscard]] constexpr u32 AdjustTileSize(u32 shift, u32 unit_factor, u32 dimension) {
+ if (shift == 0) {
+ return 0;
+ }
+ u32 x = unit_factor << (shift - 1);
+ if (x >= dimension) {
+ while (--shift) {
+ x >>= 1;
+ if (x < dimension) {
+ break;
+ }
+ }
+ }
+ return shift;
+}
+
+[[nodiscard]] constexpr u32 AdjustMipSize(u32 size, u32 level) {
+ return std::max<u32>(size >> level, 1);
+}
+
+[[nodiscard]] constexpr Extent3D AdjustMipSize(Extent3D size, s32 level) {
+ return Extent3D{
+ .width = AdjustMipSize(size.width, level),
+ .height = AdjustMipSize(size.height, level),
+ .depth = AdjustMipSize(size.depth, level),
+ };
+}
+
+[[nodiscard]] Extent3D AdjustSamplesSize(Extent3D size, s32 num_samples) {
+ const auto [samples_x, samples_y] = SamplesLog2(num_samples);
+ return Extent3D{
+ .width = size.width >> samples_x,
+ .height = size.height >> samples_y,
+ .depth = size.depth,
+ };
+}
+
+template <u32 GOB_EXTENT>
+[[nodiscard]] constexpr u32 AdjustMipBlockSize(u32 num_tiles, u32 block_size, u32 level) {
+ do {
+ while (block_size > 0 && num_tiles <= (1U << (block_size - 1)) * GOB_EXTENT) {
+ --block_size;
+ }
+ } while (level--);
+ return block_size;
+}
+
+[[nodiscard]] constexpr Extent3D AdjustMipBlockSize(Extent3D num_tiles, Extent3D block_size,
+ u32 level) {
+ return {
+ .width = AdjustMipBlockSize<GOB_SIZE_X>(num_tiles.width, block_size.width, level),
+ .height = AdjustMipBlockSize<GOB_SIZE_Y>(num_tiles.height, block_size.height, level),
+ .depth = AdjustMipBlockSize<GOB_SIZE_Z>(num_tiles.depth, block_size.depth, level),
+ };
+}
+
+[[nodiscard]] constexpr Extent3D AdjustTileSize(Extent3D size, Extent2D tile_size) {
+ return {
+ .width = Common::DivCeil(size.width, tile_size.width),
+ .height = Common::DivCeil(size.height, tile_size.height),
+ .depth = size.depth,
+ };
+}
+
+[[nodiscard]] constexpr u32 BytesPerBlockLog2(u32 bytes_per_block) {
+ return std::countl_zero(bytes_per_block) ^ 0x1F;
+}
+
+[[nodiscard]] constexpr u32 BytesPerBlockLog2(PixelFormat format) {
+ return BytesPerBlockLog2(BytesPerBlock(format));
+}
+
+[[nodiscard]] constexpr u32 NumBlocks(Extent3D size, Extent2D tile_size) {
+ const Extent3D num_blocks = AdjustTileSize(size, tile_size);
+ return num_blocks.width * num_blocks.height * num_blocks.depth;
+}
+
+[[nodiscard]] constexpr u32 AdjustSize(u32 size, u32 level, u32 block_size) {
+ return Common::DivCeil(AdjustMipSize(size, level), block_size);
+}
+
+[[nodiscard]] constexpr u32 LayerSize(const TICEntry& config, PixelFormat format) {
+ return config.Width() * config.Height() * BytesPerBlock(format);
+}
+
+[[nodiscard]] constexpr bool HasTwoDimsPerLayer(TextureType type) {
+ switch (type) {
+ case TextureType::Texture2D:
+ case TextureType::Texture2DArray:
+ case TextureType::Texture2DNoMipmap:
+ case TextureType::Texture3D:
+ case TextureType::TextureCubeArray:
+ case TextureType::TextureCubemap:
+ return true;
+ case TextureType::Texture1D:
+ case TextureType::Texture1DArray:
+ case TextureType::Texture1DBuffer:
+ return false;
+ }
+ return false;
+}
+
+[[nodiscard]] constexpr bool HasTwoDimsPerLayer(ImageType type) {
+ switch (type) {
+ case ImageType::e2D:
+ case ImageType::e3D:
+ case ImageType::Linear:
+ return true;
+ case ImageType::e1D:
+ case ImageType::Buffer:
+ return false;
+ }
+ UNREACHABLE_MSG("Invalid image type={}", static_cast<int>(type));
+}
+
+[[nodiscard]] constexpr std::pair<int, int> Samples(int num_samples) {
+ switch (num_samples) {
+ case 1:
+ return {1, 1};
+ case 2:
+ return {2, 1};
+ case 4:
+ return {2, 2};
+ case 8:
+ return {4, 2};
+ case 16:
+ return {4, 4};
+ }
+ UNREACHABLE_MSG("Invalid number of samples={}", num_samples);
+ return {1, 1};
+}
+
+[[nodiscard]] constexpr Extent2D DefaultBlockSize(PixelFormat format) {
+ return {DefaultBlockWidth(format), DefaultBlockHeight(format)};
+}
+
+[[nodiscard]] constexpr Extent3D NumLevelBlocks(const LevelInfo& info, u32 level) {
+ return Extent3D{
+ .width = AdjustSize(info.size.width, level, info.tile_size.width) << info.bpp_log2,
+ .height = AdjustSize(info.size.height, level, info.tile_size.height),
+ .depth = AdjustMipSize(info.size.depth, level),
+ };
+}
+
+[[nodiscard]] constexpr Extent3D TileShift(const LevelInfo& info, u32 level) {
+ const Extent3D blocks = NumLevelBlocks(info, level);
+ return Extent3D{
+ .width = AdjustTileSize(info.block.width, GOB_SIZE_X, blocks.width),
+ .height = AdjustTileSize(info.block.height, GOB_SIZE_Y, blocks.height),
+ .depth = AdjustTileSize(info.block.depth, GOB_SIZE_Z, blocks.depth),
+ };
+}
+
+[[nodiscard]] constexpr Extent2D GobSize(u32 bpp_log2, u32 block_height, u32 tile_width_spacing) {
+ return Extent2D{
+ .width = GOB_SIZE_X_SHIFT - bpp_log2 + tile_width_spacing,
+ .height = GOB_SIZE_Y_SHIFT + block_height,
+ };
+}
+
+[[nodiscard]] constexpr bool IsSmallerThanGobSize(Extent3D num_tiles, Extent2D gob,
+ u32 block_depth) {
+ return num_tiles.width <= (1U << gob.width) || num_tiles.height <= (1U << gob.height) ||
+ num_tiles.depth < (1U << block_depth);
+}
+
+[[nodiscard]] constexpr u32 StrideAlignment(Extent3D num_tiles, Extent3D block, Extent2D gob,
+ u32 bpp_log2) {
+ if (IsSmallerThanGobSize(num_tiles, gob, block.depth)) {
+ return GOB_SIZE_X_SHIFT - bpp_log2;
+ } else {
+ return gob.width;
+ }
+}
+
+[[nodiscard]] constexpr u32 StrideAlignment(Extent3D num_tiles, Extent3D block, u32 bpp_log2,
+ u32 tile_width_spacing) {
+ const Extent2D gob = GobSize(bpp_log2, block.height, tile_width_spacing);
+ return StrideAlignment(num_tiles, block, gob, bpp_log2);
+}
+
+[[nodiscard]] constexpr Extent2D NumGobs(const LevelInfo& info, u32 level) {
+ const Extent3D blocks = NumLevelBlocks(info, level);
+ const Extent2D gobs{
+ .width = Common::DivCeilLog2(blocks.width, GOB_SIZE_X_SHIFT),
+ .height = Common::DivCeilLog2(blocks.height, GOB_SIZE_Y_SHIFT),
+ };
+ const Extent2D gob = GobSize(info.bpp_log2, info.block.height, info.tile_width_spacing);
+ const bool is_small = IsSmallerThanGobSize(blocks, gob, info.block.depth);
+ const u32 alignment = is_small ? 0 : info.tile_width_spacing;
+ return Extent2D{
+ .width = Common::AlignBits(gobs.width, alignment),
+ .height = gobs.height,
+ };
+}
+
+[[nodiscard]] constexpr Extent3D LevelTiles(const LevelInfo& info, u32 level) {
+ const Extent3D blocks = NumLevelBlocks(info, level);
+ const Extent3D tile_shift = TileShift(info, level);
+ const Extent2D gobs = NumGobs(info, level);
+ return Extent3D{
+ .width = Common::DivCeilLog2(gobs.width, tile_shift.width),
+ .height = Common::DivCeilLog2(gobs.height, tile_shift.height),
+ .depth = Common::DivCeilLog2(blocks.depth, tile_shift.depth),
+ };
+}
+
+[[nodiscard]] constexpr u32 CalculateLevelSize(const LevelInfo& info, u32 level) {
+ const Extent3D tile_shift = TileShift(info, level);
+ const Extent3D tiles = LevelTiles(info, level);
+ const u32 num_tiles = tiles.width * tiles.height * tiles.depth;
+ const u32 shift = GOB_SIZE_SHIFT + tile_shift.width + tile_shift.height + tile_shift.depth;
+ return num_tiles << shift;
+}
+
+[[nodiscard]] constexpr std::array<u32, MAX_MIP_LEVELS> CalculateLevelSizes(const LevelInfo& info,
+ u32 num_levels) {
+ ASSERT(num_levels <= MAX_MIP_LEVELS);
+ std::array<u32, MAX_MIP_LEVELS> sizes{};
+ for (u32 level = 0; level < num_levels; ++level) {
+ sizes[level] = CalculateLevelSize(info, level);
+ }
+ return sizes;
+}
+
+[[nodiscard]] constexpr LevelInfo MakeLevelInfo(PixelFormat format, Extent3D size, Extent3D block,
+ u32 num_samples, u32 tile_width_spacing) {
+ const auto [samples_x, samples_y] = Samples(num_samples);
+ const u32 bytes_per_block = BytesPerBlock(format);
+ return {
+ .size =
+ {
+ .width = size.width * samples_x,
+ .height = size.height * samples_y,
+ .depth = size.depth,
+ },
+ .block = block,
+ .tile_size = DefaultBlockSize(format),
+ .bpp_log2 = BytesPerBlockLog2(bytes_per_block),
+ .tile_width_spacing = tile_width_spacing,
+ };
+}
+
+[[nodiscard]] constexpr LevelInfo MakeLevelInfo(const ImageInfo& info) {
+ return MakeLevelInfo(info.format, info.size, info.block, info.num_samples,
+ info.tile_width_spacing);
+}
+
+[[nodiscard]] constexpr u32 CalculateLevelOffset(PixelFormat format, Extent3D size, Extent3D block,
+ u32 num_samples, u32 tile_width_spacing,
+ u32 level) {
+ const LevelInfo info = MakeLevelInfo(format, size, block, num_samples, tile_width_spacing);
+ u32 offset = 0;
+ for (u32 current_level = 0; current_level < level; ++current_level) {
+ offset += CalculateLevelSize(info, current_level);
+ }
+ return offset;
+}
+
+[[nodiscard]] constexpr u32 AlignLayerSize(u32 size_bytes, Extent3D size, Extent3D block,
+ u32 tile_size_y, u32 tile_width_spacing) {
+ // https://github.com/Ryujinx/Ryujinx/blob/1c9aba6de1520aea5480c032e0ff5664ac1bb36f/Ryujinx.Graphics.Texture/SizeCalculator.cs#L134
+ if (tile_width_spacing > 0) {
+ const u32 alignment_log2 = GOB_SIZE_SHIFT + tile_width_spacing + block.height + block.depth;
+ return Common::AlignBits(size_bytes, alignment_log2);
+ }
+ const u32 aligned_height = Common::AlignUp(size.height, tile_size_y);
+ while (block.height != 0 && aligned_height <= (1U << (block.height - 1)) * GOB_SIZE_Y) {
+ --block.height;
+ }
+ while (block.depth != 0 && size.depth <= (1U << (block.depth - 1))) {
+ --block.depth;
+ }
+ const u32 block_shift = GOB_SIZE_SHIFT + block.height + block.depth;
+ const u32 num_blocks = size_bytes >> block_shift;
+ if (size_bytes != num_blocks << block_shift) {
+ return (num_blocks + 1) << block_shift;
+ }
+ return size_bytes;
+}
+
+[[nodiscard]] std::optional<SubresourceExtent> ResolveOverlapEqualAddress(const ImageInfo& new_info,
+ const ImageBase& overlap,
+ bool strict_size) {
+ const ImageInfo& info = overlap.info;
+ if (!IsBlockLinearSizeCompatible(new_info, info, 0, 0, strict_size)) {
+ return std::nullopt;
+ }
+ if (new_info.block != info.block) {
+ return std::nullopt;
+ }
+ const SubresourceExtent resources = new_info.resources;
+ return SubresourceExtent{
+ .levels = std::max(resources.levels, info.resources.levels),
+ .layers = std::max(resources.layers, info.resources.layers),
+ };
+}
+
+[[nodiscard]] std::optional<SubresourceExtent> ResolveOverlapRightAddress3D(
+ const ImageInfo& new_info, GPUVAddr gpu_addr, const ImageBase& overlap, bool strict_size) {
+ const std::vector<u32> slice_offsets = CalculateSliceOffsets(new_info);
+ const u32 diff = static_cast<u32>(overlap.gpu_addr - gpu_addr);
+ const auto it = std::ranges::find(slice_offsets, diff);
+ if (it == slice_offsets.end()) {
+ return std::nullopt;
+ }
+ const std::vector subresources = CalculateSliceSubresources(new_info);
+ const SubresourceBase base = subresources[std::distance(slice_offsets.begin(), it)];
+ const ImageInfo& info = overlap.info;
+ if (!IsBlockLinearSizeCompatible(new_info, info, base.level, 0, strict_size)) {
+ return std::nullopt;
+ }
+ const u32 mip_depth = std::max(1U, new_info.size.depth << base.level);
+ if (mip_depth < info.size.depth + base.layer) {
+ return std::nullopt;
+ }
+ if (MipBlockSize(new_info, base.level) != info.block) {
+ return std::nullopt;
+ }
+ return SubresourceExtent{
+ .levels = std::max(new_info.resources.levels, info.resources.levels + base.level),
+ .layers = 1,
+ };
+}
+
+[[nodiscard]] std::optional<SubresourceExtent> ResolveOverlapRightAddress2D(
+ const ImageInfo& new_info, GPUVAddr gpu_addr, const ImageBase& overlap, bool strict_size) {
+ const u32 layer_stride = new_info.layer_stride;
+ const s32 new_size = layer_stride * new_info.resources.layers;
+ const s32 diff = static_cast<s32>(overlap.gpu_addr - gpu_addr);
+ if (diff > new_size) {
+ return std::nullopt;
+ }
+ const s32 base_layer = diff / layer_stride;
+ const s32 mip_offset = diff % layer_stride;
+ const std::array offsets = CalculateMipLevelOffsets(new_info);
+ const auto end = offsets.begin() + new_info.resources.levels;
+ const auto it = std::find(offsets.begin(), end, mip_offset);
+ if (it == end) {
+ // Mipmap is not aligned to any valid size
+ return std::nullopt;
+ }
+ const SubresourceBase base{
+ .level = static_cast<s32>(std::distance(offsets.begin(), it)),
+ .layer = base_layer,
+ };
+ const ImageInfo& info = overlap.info;
+ if (!IsBlockLinearSizeCompatible(new_info, info, base.level, 0, strict_size)) {
+ return std::nullopt;
+ }
+ if (MipBlockSize(new_info, base.level) != info.block) {
+ return std::nullopt;
+ }
+ return SubresourceExtent{
+ .levels = std::max(new_info.resources.levels, info.resources.levels + base.level),
+ .layers = std::max(new_info.resources.layers, info.resources.layers + base.layer),
+ };
+}
+
+[[nodiscard]] std::optional<OverlapResult> ResolveOverlapRightAddress(const ImageInfo& new_info,
+ GPUVAddr gpu_addr,
+ VAddr cpu_addr,
+ const ImageBase& overlap,
+ bool strict_size) {
+ std::optional<SubresourceExtent> resources;
+ if (new_info.type != ImageType::e3D) {
+ resources = ResolveOverlapRightAddress2D(new_info, gpu_addr, overlap, strict_size);
+ } else {
+ resources = ResolveOverlapRightAddress3D(new_info, gpu_addr, overlap, strict_size);
+ }
+ if (!resources) {
+ return std::nullopt;
+ }
+ return OverlapResult{
+ .gpu_addr = gpu_addr,
+ .cpu_addr = cpu_addr,
+ .resources = *resources,
+ };
+}
+
+[[nodiscard]] std::optional<OverlapResult> ResolveOverlapLeftAddress(const ImageInfo& new_info,
+ GPUVAddr gpu_addr,
+ VAddr cpu_addr,
+ const ImageBase& overlap,
+ bool strict_size) {
+ const std::optional<SubresourceBase> base = overlap.TryFindBase(gpu_addr);
+ if (!base) {
+ return std::nullopt;
+ }
+ const ImageInfo& info = overlap.info;
+ if (!IsBlockLinearSizeCompatible(new_info, info, base->level, 0, strict_size)) {
+ return std::nullopt;
+ }
+ if (new_info.block != MipBlockSize(info, base->level)) {
+ return std::nullopt;
+ }
+ const SubresourceExtent resources = new_info.resources;
+ s32 layers = 1;
+ if (info.type != ImageType::e3D) {
+ layers = std::max(resources.layers, info.resources.layers + base->layer);
+ }
+ return OverlapResult{
+ .gpu_addr = overlap.gpu_addr,
+ .cpu_addr = overlap.cpu_addr,
+ .resources =
+ {
+ .levels = std::max(resources.levels + base->level, info.resources.levels),
+ .layers = layers,
+ },
+ };
+}
+
+[[nodiscard]] Extent2D PitchLinearAlignedSize(const ImageInfo& info) {
+ // https://github.com/Ryujinx/Ryujinx/blob/1c9aba6de1520aea5480c032e0ff5664ac1bb36f/Ryujinx.Graphics.Texture/SizeCalculator.cs#L212
+ static constexpr u32 STRIDE_ALIGNMENT = 32;
+ ASSERT(info.type == ImageType::Linear);
+ const Extent2D num_tiles{
+ .width = Common::DivCeil(info.size.width, DefaultBlockWidth(info.format)),
+ .height = Common::DivCeil(info.size.height, DefaultBlockHeight(info.format)),
+ };
+ const u32 width_alignment = STRIDE_ALIGNMENT / BytesPerBlock(info.format);
+ return Extent2D{
+ .width = Common::AlignUp(num_tiles.width, width_alignment),
+ .height = num_tiles.height,
+ };
+}
+
+[[nodiscard]] Extent3D BlockLinearAlignedSize(const ImageInfo& info, u32 level) {
+ // https://github.com/Ryujinx/Ryujinx/blob/1c9aba6de1520aea5480c032e0ff5664ac1bb36f/Ryujinx.Graphics.Texture/SizeCalculator.cs#L176
+ ASSERT(info.type != ImageType::Linear);
+ const Extent3D size = AdjustMipSize(info.size, level);
+ const Extent3D num_tiles{
+ .width = Common::DivCeil(size.width, DefaultBlockWidth(info.format)),
+ .height = Common::DivCeil(size.height, DefaultBlockHeight(info.format)),
+ .depth = size.depth,
+ };
+ const u32 bpp_log2 = BytesPerBlockLog2(info.format);
+ const u32 alignment = StrideAlignment(num_tiles, info.block, bpp_log2, info.tile_width_spacing);
+ const Extent3D mip_block = AdjustMipBlockSize(num_tiles, info.block, 0);
+ return Extent3D{
+ .width = Common::AlignBits(num_tiles.width, alignment),
+ .height = Common::AlignBits(num_tiles.height, GOB_SIZE_Y_SHIFT + mip_block.height),
+ .depth = Common::AlignBits(num_tiles.depth, GOB_SIZE_Z_SHIFT + mip_block.depth),
+ };
+}
+
+[[nodiscard]] constexpr u32 NumBlocksPerLayer(const ImageInfo& info, Extent2D tile_size) noexcept {
+ u32 num_blocks = 0;
+ for (s32 level = 0; level < info.resources.levels; ++level) {
+ const Extent3D mip_size = AdjustMipSize(info.size, level);
+ num_blocks += NumBlocks(mip_size, tile_size);
+ }
+ return num_blocks;
+}
+
+[[nodiscard]] u32 NumSlices(const ImageInfo& info) noexcept {
+ ASSERT(info.type == ImageType::e3D);
+ u32 num_slices = 0;
+ for (s32 level = 0; level < info.resources.levels; ++level) {
+ num_slices += AdjustMipSize(info.size.depth, level);
+ }
+ return num_slices;
+}
+
+void SwizzlePitchLinearImage(Tegra::MemoryManager& gpu_memory, GPUVAddr gpu_addr,
+ const ImageInfo& info, const BufferImageCopy& copy,
+ std::span<const u8> memory) {
+ ASSERT(copy.image_offset.z == 0);
+ ASSERT(copy.image_extent.depth == 1);
+ ASSERT(copy.image_subresource.base_level == 0);
+ ASSERT(copy.image_subresource.base_layer == 0);
+ ASSERT(copy.image_subresource.num_layers == 1);
+
+ const u32 bytes_per_block = BytesPerBlock(info.format);
+ const u32 row_length = copy.image_extent.width * bytes_per_block;
+ const u32 guest_offset_x = copy.image_offset.x * bytes_per_block;
+
+ for (u32 line = 0; line < copy.image_extent.height; ++line) {
+ const u32 host_offset_y = line * info.pitch;
+ const u32 guest_offset_y = (copy.image_offset.y + line) * info.pitch;
+ const u32 guest_offset = guest_offset_x + guest_offset_y;
+ gpu_memory.WriteBlockUnsafe(gpu_addr + guest_offset, memory.data() + host_offset_y,
+ row_length);
+ }
+}
+
+void SwizzleBlockLinearImage(Tegra::MemoryManager& gpu_memory, GPUVAddr gpu_addr,
+ const ImageInfo& info, const BufferImageCopy& copy,
+ std::span<const u8> input) {
+ const Extent3D size = info.size;
+ const LevelInfo level_info = MakeLevelInfo(info);
+ const Extent2D tile_size = DefaultBlockSize(info.format);
+ const u32 bytes_per_block = BytesPerBlock(info.format);
+
+ const s32 level = copy.image_subresource.base_level;
+ const Extent3D level_size = AdjustMipSize(size, level);
+ const u32 num_blocks_per_layer = NumBlocks(level_size, tile_size);
+ const u32 host_bytes_per_layer = num_blocks_per_layer * bytes_per_block;
+
+ UNIMPLEMENTED_IF(info.tile_width_spacing > 0);
+
+ UNIMPLEMENTED_IF(copy.image_offset.x != 0);
+ UNIMPLEMENTED_IF(copy.image_offset.y != 0);
+ UNIMPLEMENTED_IF(copy.image_offset.z != 0);
+ UNIMPLEMENTED_IF(copy.image_extent != level_size);
+
+ const Extent3D num_tiles = AdjustTileSize(level_size, tile_size);
+ const Extent3D block = AdjustMipBlockSize(num_tiles, level_info.block, level);
+
+ size_t host_offset = copy.buffer_offset;
+
+ const u32 num_levels = info.resources.levels;
+ const std::array sizes = CalculateLevelSizes(level_info, num_levels);
+ size_t guest_offset = std::reduce(sizes.begin(), sizes.begin() + level, 0);
+ const size_t layer_stride =
+ AlignLayerSize(std::reduce(sizes.begin(), sizes.begin() + num_levels, 0), size,
+ level_info.block, tile_size.height, info.tile_width_spacing);
+ const size_t subresource_size = sizes[level];
+
+ const auto dst_data = std::make_unique<u8[]>(subresource_size);
+ const std::span<u8> dst(dst_data.get(), subresource_size);
+
+ for (s32 layer = 0; layer < info.resources.layers; ++layer) {
+ const std::span<const u8> src = input.subspan(host_offset);
+ SwizzleTexture(dst, src, bytes_per_block, num_tiles.width, num_tiles.height,
+ num_tiles.depth, block.height, block.depth);
+
+ gpu_memory.WriteBlockUnsafe(gpu_addr + guest_offset, dst.data(), dst.size_bytes());
+
+ host_offset += host_bytes_per_layer;
+ guest_offset += layer_stride;
+ }
+ ASSERT(host_offset - copy.buffer_offset == copy.buffer_size);
+}
+
+} // Anonymous namespace
+
+u32 CalculateGuestSizeInBytes(const ImageInfo& info) noexcept {
+ if (info.type == ImageType::Buffer) {
+ return info.size.width * BytesPerBlock(info.format);
+ }
+ if (info.type == ImageType::Linear) {
+ return info.pitch * Common::DivCeil(info.size.height, DefaultBlockHeight(info.format));
+ }
+ if (info.resources.layers > 1) {
+ ASSERT(info.layer_stride != 0);
+ return info.layer_stride * info.resources.layers;
+ } else {
+ return CalculateLayerSize(info);
+ }
+}
+
+u32 CalculateUnswizzledSizeBytes(const ImageInfo& info) noexcept {
+ if (info.type == ImageType::Buffer) {
+ return info.size.width * BytesPerBlock(info.format);
+ }
+ if (info.num_samples > 1) {
+ // Multisample images can't be uploaded or downloaded to the host
+ return 0;
+ }
+ if (info.type == ImageType::Linear) {
+ return info.pitch * Common::DivCeil(info.size.height, DefaultBlockHeight(info.format));
+ }
+ const Extent2D tile_size = DefaultBlockSize(info.format);
+ return NumBlocksPerLayer(info, tile_size) * info.resources.layers * BytesPerBlock(info.format);
+}
+
+u32 CalculateConvertedSizeBytes(const ImageInfo& info) noexcept {
+ if (info.type == ImageType::Buffer) {
+ return info.size.width * BytesPerBlock(info.format);
+ }
+ static constexpr Extent2D TILE_SIZE{1, 1};
+ return NumBlocksPerLayer(info, TILE_SIZE) * info.resources.layers * CONVERTED_BYTES_PER_BLOCK;
+}
+
+u32 CalculateLayerStride(const ImageInfo& info) noexcept {
+ ASSERT(info.type != ImageType::Linear);
+ const u32 layer_size = CalculateLayerSize(info);
+ const Extent3D size = info.size;
+ const Extent3D block = info.block;
+ const u32 tile_size_y = DefaultBlockHeight(info.format);
+ return AlignLayerSize(layer_size, size, block, tile_size_y, info.tile_width_spacing);
+}
+
+u32 CalculateLayerSize(const ImageInfo& info) noexcept {
+ ASSERT(info.type != ImageType::Linear);
+ return CalculateLevelOffset(info.format, info.size, info.block, info.num_samples,
+ info.tile_width_spacing, info.resources.levels);
+}
+
+std::array<u32, MAX_MIP_LEVELS> CalculateMipLevelOffsets(const ImageInfo& info) noexcept {
+ ASSERT(info.resources.levels <= MAX_MIP_LEVELS);
+ const LevelInfo level_info = MakeLevelInfo(info);
+ std::array<u32, MAX_MIP_LEVELS> offsets{};
+ u32 offset = 0;
+ for (s32 level = 0; level < info.resources.levels; ++level) {
+ offsets[level] = offset;
+ offset += CalculateLevelSize(level_info, level);
+ }
+ return offsets;
+}
+
+std::vector<u32> CalculateSliceOffsets(const ImageInfo& info) {
+ ASSERT(info.type == ImageType::e3D);
+ std::vector<u32> offsets;
+ offsets.reserve(NumSlices(info));
+
+ const LevelInfo level_info = MakeLevelInfo(info);
+ u32 mip_offset = 0;
+ for (s32 level = 0; level < info.resources.levels; ++level) {
+ const Extent3D tile_shift = TileShift(level_info, level);
+ const Extent3D tiles = LevelTiles(level_info, level);
+ const u32 gob_size_shift = tile_shift.height + GOB_SIZE_SHIFT;
+ const u32 slice_size = (tiles.width * tiles.height) << gob_size_shift;
+ const u32 z_mask = (1U << tile_shift.depth) - 1;
+ const u32 depth = AdjustMipSize(info.size.depth, level);
+ for (u32 slice = 0; slice < depth; ++slice) {
+ const u32 z_low = slice & z_mask;
+ const u32 z_high = slice & ~z_mask;
+ offsets.push_back(mip_offset + (z_low << gob_size_shift) + (z_high * slice_size));
+ }
+ mip_offset += CalculateLevelSize(level_info, level);
+ }
+ return offsets;
+}
+
+std::vector<SubresourceBase> CalculateSliceSubresources(const ImageInfo& info) {
+ ASSERT(info.type == ImageType::e3D);
+ std::vector<SubresourceBase> subresources;
+ subresources.reserve(NumSlices(info));
+ for (s32 level = 0; level < info.resources.levels; ++level) {
+ const s32 depth = AdjustMipSize(info.size.depth, level);
+ for (s32 slice = 0; slice < depth; ++slice) {
+ subresources.emplace_back(SubresourceBase{
+ .level = level,
+ .layer = slice,
+ });
+ }
+ }
+ return subresources;
+}
+
+u32 CalculateLevelStrideAlignment(const ImageInfo& info, u32 level) {
+ const Extent2D tile_size = DefaultBlockSize(info.format);
+ const Extent3D level_size = AdjustMipSize(info.size, level);
+ const Extent3D num_tiles = AdjustTileSize(level_size, tile_size);
+ const Extent3D block = AdjustMipBlockSize(num_tiles, info.block, level);
+ const u32 bpp_log2 = BytesPerBlockLog2(info.format);
+ return StrideAlignment(num_tiles, block, bpp_log2, info.tile_width_spacing);
+}
+
+PixelFormat PixelFormatFromTIC(const TICEntry& config) noexcept {
+ return PixelFormatFromTextureInfo(config.format, config.r_type, config.g_type, config.b_type,
+ config.a_type, config.srgb_conversion);
+}
+
+ImageViewType RenderTargetImageViewType(const ImageInfo& info) noexcept {
+ switch (info.type) {
+ case ImageType::e2D:
+ return info.resources.layers > 1 ? ImageViewType::e2DArray : ImageViewType::e2D;
+ case ImageType::e3D:
+ return ImageViewType::e2DArray;
+ case ImageType::Linear:
+ return ImageViewType::e2D;
+ default:
+ UNIMPLEMENTED_MSG("Unimplemented image type={}", static_cast<int>(info.type));
+ return ImageViewType{};
+ }
+}
+
+std::vector<ImageCopy> MakeShrinkImageCopies(const ImageInfo& dst, const ImageInfo& src,
+ SubresourceBase base) {
+ ASSERT(dst.resources.levels >= src.resources.levels);
+ ASSERT(dst.num_samples == src.num_samples);
+
+ const bool is_dst_3d = dst.type == ImageType::e3D;
+ if (is_dst_3d) {
+ ASSERT(src.type == ImageType::e3D);
+ ASSERT(src.resources.levels == 1);
+ }
+
+ std::vector<ImageCopy> copies;
+ copies.reserve(src.resources.levels);
+ for (s32 level = 0; level < src.resources.levels; ++level) {
+ ImageCopy& copy = copies.emplace_back();
+ copy.src_subresource = SubresourceLayers{
+ .base_level = level,
+ .base_layer = 0,
+ .num_layers = src.resources.layers,
+ };
+ copy.dst_subresource = SubresourceLayers{
+ .base_level = base.level + level,
+ .base_layer = is_dst_3d ? 0 : base.layer,
+ .num_layers = is_dst_3d ? 1 : src.resources.layers,
+ };
+ copy.src_offset = Offset3D{
+ .x = 0,
+ .y = 0,
+ .z = 0,
+ };
+ copy.dst_offset = Offset3D{
+ .x = 0,
+ .y = 0,
+ .z = is_dst_3d ? base.layer : 0,
+ };
+ const Extent3D mip_size = AdjustMipSize(dst.size, base.level + level);
+ copy.extent = AdjustSamplesSize(mip_size, dst.num_samples);
+ if (is_dst_3d) {
+ copy.extent.depth = src.size.depth;
+ }
+ }
+ return copies;
+}
+
+bool IsValidAddress(const Tegra::MemoryManager& gpu_memory, const TICEntry& config) {
+ if (config.Address() == 0) {
+ return false;
+ }
+ if (config.Address() > (u64(1) << 48)) {
+ return false;
+ }
+ return gpu_memory.GpuToCpuAddress(config.Address()).has_value();
+}
+
+std::vector<BufferImageCopy> UnswizzleImage(Tegra::MemoryManager& gpu_memory, GPUVAddr gpu_addr,
+ const ImageInfo& info, std::span<u8> output) {
+ const size_t guest_size_bytes = CalculateGuestSizeInBytes(info);
+ const u32 bpp_log2 = BytesPerBlockLog2(info.format);
+ const Extent3D size = info.size;
+
+ if (info.type == ImageType::Linear) {
+ gpu_memory.ReadBlockUnsafe(gpu_addr, output.data(), guest_size_bytes);
+
+ ASSERT((info.pitch >> bpp_log2) << bpp_log2 == info.pitch);
+ return {{
+ .buffer_offset = 0,
+ .buffer_size = guest_size_bytes,
+ .buffer_row_length = info.pitch >> bpp_log2,
+ .buffer_image_height = size.height,
+ .image_subresource =
+ {
+ .base_level = 0,
+ .base_layer = 0,
+ .num_layers = 1,
+ },
+ .image_offset = {0, 0, 0},
+ .image_extent = size,
+ }};
+ }
+ const auto input_data = std::make_unique<u8[]>(guest_size_bytes);
+ gpu_memory.ReadBlockUnsafe(gpu_addr, input_data.get(), guest_size_bytes);
+ const std::span<const u8> input(input_data.get(), guest_size_bytes);
+
+ const LevelInfo level_info = MakeLevelInfo(info);
+ const s32 num_layers = info.resources.layers;
+ const s32 num_levels = info.resources.levels;
+ const Extent2D tile_size = DefaultBlockSize(info.format);
+ const std::array level_sizes = CalculateLevelSizes(level_info, num_levels);
+ const Extent2D gob = GobSize(bpp_log2, info.block.height, info.tile_width_spacing);
+ const u32 layer_size = std::reduce(level_sizes.begin(), level_sizes.begin() + num_levels, 0);
+ const u32 layer_stride = AlignLayerSize(layer_size, size, level_info.block, tile_size.height,
+ info.tile_width_spacing);
+ size_t guest_offset = 0;
+ u32 host_offset = 0;
+ std::vector<BufferImageCopy> copies(num_levels);
+
+ for (s32 level = 0; level < num_levels; ++level) {
+ const Extent3D level_size = AdjustMipSize(size, level);
+ const u32 num_blocks_per_layer = NumBlocks(level_size, tile_size);
+ const u32 host_bytes_per_layer = num_blocks_per_layer << bpp_log2;
+ copies[level] = BufferImageCopy{
+ .buffer_offset = host_offset,
+ .buffer_size = static_cast<size_t>(host_bytes_per_layer) * num_layers,
+ .buffer_row_length = Common::AlignUp(level_size.width, tile_size.width),
+ .buffer_image_height = Common::AlignUp(level_size.height, tile_size.height),
+ .image_subresource =
+ {
+ .base_level = level,
+ .base_layer = 0,
+ .num_layers = info.resources.layers,
+ },
+ .image_offset = {0, 0, 0},
+ .image_extent = level_size,
+ };
+ const Extent3D num_tiles = AdjustTileSize(level_size, tile_size);
+ const Extent3D block = AdjustMipBlockSize(num_tiles, level_info.block, level);
+ const u32 stride_alignment = StrideAlignment(num_tiles, info.block, gob, bpp_log2);
+ size_t guest_layer_offset = 0;
+
+ for (s32 layer = 0; layer < info.resources.layers; ++layer) {
+ const std::span<u8> dst = output.subspan(host_offset);
+ const std::span<const u8> src = input.subspan(guest_offset + guest_layer_offset);
+ UnswizzleTexture(dst, src, 1U << bpp_log2, num_tiles.width, num_tiles.height,
+ num_tiles.depth, block.height, block.depth, stride_alignment);
+ guest_layer_offset += layer_stride;
+ host_offset += host_bytes_per_layer;
+ }
+ guest_offset += level_sizes[level];
+ }
+ return copies;
+}
+
+BufferCopy UploadBufferCopy(Tegra::MemoryManager& gpu_memory, GPUVAddr gpu_addr,
+ const ImageBase& image, std::span<u8> output) {
+ gpu_memory.ReadBlockUnsafe(gpu_addr, output.data(), image.guest_size_bytes);
+ return BufferCopy{
+ .src_offset = 0,
+ .dst_offset = 0,
+ .size = image.guest_size_bytes,
+ };
+}
+
+void ConvertImage(std::span<const u8> input, const ImageInfo& info, std::span<u8> output,
+ std::span<BufferImageCopy> copies) {
+ u32 output_offset = 0;
+
+ const Extent2D tile_size = DefaultBlockSize(info.format);
+ for (BufferImageCopy& copy : copies) {
+ const u32 level = copy.image_subresource.base_level;
+ const Extent3D mip_size = AdjustMipSize(info.size, level);
+ ASSERT(copy.image_offset == Offset3D{});
+ ASSERT(copy.image_subresource.base_layer == 0);
+ ASSERT(copy.image_extent == mip_size);
+ ASSERT(copy.buffer_row_length == Common::AlignUp(mip_size.width, tile_size.width));
+ ASSERT(copy.buffer_image_height == Common::AlignUp(mip_size.height, tile_size.height));
+
+ if (IsPixelFormatASTC(info.format)) {
+ ASSERT(copy.image_extent.depth == 1);
+ Tegra::Texture::ASTC::Decompress(input.subspan(copy.buffer_offset),
+ copy.image_extent.width, copy.image_extent.height,
+ copy.image_subresource.num_layers, tile_size.width,
+ tile_size.height, output.subspan(output_offset));
+ } else {
+ DecompressBC4(input.subspan(copy.buffer_offset), copy.image_extent,
+ output.subspan(output_offset));
+ }
+ copy.buffer_offset = output_offset;
+ copy.buffer_row_length = mip_size.width;
+ copy.buffer_image_height = mip_size.height;
+
+ output_offset += copy.image_extent.width * copy.image_extent.height *
+ copy.image_subresource.num_layers * CONVERTED_BYTES_PER_BLOCK;
+ }
+}
+
+std::vector<BufferImageCopy> FullDownloadCopies(const ImageInfo& info) {
+ const Extent3D size = info.size;
+ const u32 bytes_per_block = BytesPerBlock(info.format);
+ if (info.type == ImageType::Linear) {
+ ASSERT(info.pitch % bytes_per_block == 0);
+ return {{
+ .buffer_offset = 0,
+ .buffer_size = static_cast<size_t>(info.pitch) * size.height,
+ .buffer_row_length = info.pitch / bytes_per_block,
+ .buffer_image_height = size.height,
+ .image_subresource =
+ {
+ .base_level = 0,
+ .base_layer = 0,
+ .num_layers = 1,
+ },
+ .image_offset = {0, 0, 0},
+ .image_extent = size,
+ }};
+ }
+ UNIMPLEMENTED_IF(info.tile_width_spacing > 0);
+
+ const s32 num_layers = info.resources.layers;
+ const s32 num_levels = info.resources.levels;
+ const Extent2D tile_size = DefaultBlockSize(info.format);
+
+ u32 host_offset = 0;
+
+ std::vector<BufferImageCopy> copies(num_levels);
+ for (s32 level = 0; level < num_levels; ++level) {
+ const Extent3D level_size = AdjustMipSize(size, level);
+ const u32 num_blocks_per_layer = NumBlocks(level_size, tile_size);
+ const u32 host_bytes_per_level = num_blocks_per_layer * bytes_per_block * num_layers;
+ copies[level] = BufferImageCopy{
+ .buffer_offset = host_offset,
+ .buffer_size = host_bytes_per_level,
+ .buffer_row_length = level_size.width,
+ .buffer_image_height = level_size.height,
+ .image_subresource =
+ {
+ .base_level = level,
+ .base_layer = 0,
+ .num_layers = info.resources.layers,
+ },
+ .image_offset = {0, 0, 0},
+ .image_extent = level_size,
+ };
+ host_offset += host_bytes_per_level;
+ }
+ return copies;
+}
+
+Extent3D MipSize(Extent3D size, u32 level) {
+ return AdjustMipSize(size, level);
+}
+
+Extent3D MipBlockSize(const ImageInfo& info, u32 level) {
+ const LevelInfo level_info = MakeLevelInfo(info);
+ const Extent2D tile_size = DefaultBlockSize(info.format);
+ const Extent3D level_size = AdjustMipSize(info.size, level);
+ const Extent3D num_tiles = AdjustTileSize(level_size, tile_size);
+ return AdjustMipBlockSize(num_tiles, level_info.block, level);
+}
+
+std::vector<SwizzleParameters> FullUploadSwizzles(const ImageInfo& info) {
+ const Extent2D tile_size = DefaultBlockSize(info.format);
+ if (info.type == ImageType::Linear) {
+ return std::vector{SwizzleParameters{
+ .num_tiles = AdjustTileSize(info.size, tile_size),
+ .block = {},
+ .buffer_offset = 0,
+ .level = 0,
+ }};
+ }
+ const LevelInfo level_info = MakeLevelInfo(info);
+ const Extent3D size = info.size;
+ const s32 num_levels = info.resources.levels;
+
+ u32 guest_offset = 0;
+ std::vector<SwizzleParameters> params(num_levels);
+ for (s32 level = 0; level < num_levels; ++level) {
+ const Extent3D level_size = AdjustMipSize(size, level);
+ const Extent3D num_tiles = AdjustTileSize(level_size, tile_size);
+ const Extent3D block = AdjustMipBlockSize(num_tiles, level_info.block, level);
+ params[level] = SwizzleParameters{
+ .num_tiles = num_tiles,
+ .block = block,
+ .buffer_offset = guest_offset,
+ .level = level,
+ };
+ guest_offset += CalculateLevelSize(level_info, level);
+ }
+ return params;
+}
+
+void SwizzleImage(Tegra::MemoryManager& gpu_memory, GPUVAddr gpu_addr, const ImageInfo& info,
+ std::span<const BufferImageCopy> copies, std::span<const u8> memory) {
+ const bool is_pitch_linear = info.type == ImageType::Linear;
+ for (const BufferImageCopy& copy : copies) {
+ if (is_pitch_linear) {
+ SwizzlePitchLinearImage(gpu_memory, gpu_addr, info, copy, memory);
+ } else {
+ SwizzleBlockLinearImage(gpu_memory, gpu_addr, info, copy, memory);
+ }
+ }
+}
+
+bool IsBlockLinearSizeCompatible(const ImageInfo& lhs, const ImageInfo& rhs, u32 lhs_level,
+ u32 rhs_level, bool strict_size) noexcept {
+ ASSERT(lhs.type != ImageType::Linear);
+ ASSERT(rhs.type != ImageType::Linear);
+ if (strict_size) {
+ const Extent3D lhs_size = AdjustMipSize(lhs.size, lhs_level);
+ const Extent3D rhs_size = AdjustMipSize(rhs.size, rhs_level);
+ return lhs_size.width == rhs_size.width && lhs_size.height == rhs_size.height;
+ } else {
+ const Extent3D lhs_size = BlockLinearAlignedSize(lhs, lhs_level);
+ const Extent3D rhs_size = BlockLinearAlignedSize(rhs, rhs_level);
+ return lhs_size.width == rhs_size.width && lhs_size.height == rhs_size.height;
+ }
+}
+
+bool IsPitchLinearSameSize(const ImageInfo& lhs, const ImageInfo& rhs, bool strict_size) noexcept {
+ ASSERT(lhs.type == ImageType::Linear);
+ ASSERT(rhs.type == ImageType::Linear);
+ if (strict_size) {
+ return lhs.size.width == rhs.size.width && lhs.size.height == rhs.size.height;
+ } else {
+ const Extent2D lhs_size = PitchLinearAlignedSize(lhs);
+ const Extent2D rhs_size = PitchLinearAlignedSize(rhs);
+ return lhs_size == rhs_size;
+ }
+}
+
+std::optional<OverlapResult> ResolveOverlap(const ImageInfo& new_info, GPUVAddr gpu_addr,
+ VAddr cpu_addr, const ImageBase& overlap,
+ bool strict_size) {
+ ASSERT(new_info.type != ImageType::Linear);
+ ASSERT(overlap.info.type != ImageType::Linear);
+ if (!IsLayerStrideCompatible(new_info, overlap.info)) {
+ return std::nullopt;
+ }
+ if (!IsViewCompatible(overlap.info.format, new_info.format)) {
+ return std::nullopt;
+ }
+ if (gpu_addr == overlap.gpu_addr) {
+ const std::optional solution = ResolveOverlapEqualAddress(new_info, overlap, strict_size);
+ if (!solution) {
+ return std::nullopt;
+ }
+ return OverlapResult{
+ .gpu_addr = gpu_addr,
+ .cpu_addr = cpu_addr,
+ .resources = *solution,
+ };
+ }
+ if (overlap.gpu_addr > gpu_addr) {
+ return ResolveOverlapRightAddress(new_info, gpu_addr, cpu_addr, overlap, strict_size);
+ }
+ // if overlap.gpu_addr < gpu_addr
+ return ResolveOverlapLeftAddress(new_info, gpu_addr, cpu_addr, overlap, strict_size);
+}
+
+bool IsLayerStrideCompatible(const ImageInfo& lhs, const ImageInfo& rhs) {
+ // If either of the layer strides is zero, we can assume they are compatible
+ // These images generally come from rendertargets
+ if (lhs.layer_stride == 0) {
+ return true;
+ }
+ if (rhs.layer_stride == 0) {
+ return true;
+ }
+ // It's definitely compatible if the layer stride matches
+ if (lhs.layer_stride == rhs.layer_stride) {
+ return true;
+ }
+ // Although we also have to compare for cases where it can be unaligned
+ // This can happen if the image doesn't have layers, so the stride is not aligned
+ if (lhs.maybe_unaligned_layer_stride == rhs.maybe_unaligned_layer_stride) {
+ return true;
+ }
+ return false;
+}
+
+std::optional<SubresourceBase> FindSubresource(const ImageInfo& candidate, const ImageBase& image,
+ GPUVAddr candidate_addr, RelaxedOptions options) {
+ const std::optional<SubresourceBase> base = image.TryFindBase(candidate_addr);
+ if (!base) {
+ return std::nullopt;
+ }
+ const ImageInfo& existing = image.info;
+ if (False(options & RelaxedOptions::Format)) {
+ if (!IsViewCompatible(existing.format, candidate.format)) {
+ return std::nullopt;
+ }
+ }
+ if (!IsLayerStrideCompatible(existing, candidate)) {
+ return std::nullopt;
+ }
+ if (existing.type != candidate.type) {
+ return std::nullopt;
+ }
+ if (False(options & RelaxedOptions::Samples)) {
+ if (existing.num_samples != candidate.num_samples) {
+ return std::nullopt;
+ }
+ }
+ if (existing.resources.levels < candidate.resources.levels + base->level) {
+ return std::nullopt;
+ }
+ if (existing.type == ImageType::e3D) {
+ const u32 mip_depth = std::max(1U, existing.size.depth << base->level);
+ if (mip_depth < candidate.size.depth + base->layer) {
+ return std::nullopt;
+ }
+ } else {
+ if (existing.resources.layers < candidate.resources.layers + base->layer) {
+ return std::nullopt;
+ }
+ }
+ const bool strict_size = False(options & RelaxedOptions::Size);
+ if (!IsBlockLinearSizeCompatible(existing, candidate, base->level, 0, strict_size)) {
+ return std::nullopt;
+ }
+ // TODO: compare block sizes
+ return base;
+}
+
+bool IsSubresource(const ImageInfo& candidate, const ImageBase& image, GPUVAddr candidate_addr,
+ RelaxedOptions options) {
+ return FindSubresource(candidate, image, candidate_addr, options).has_value();
+}
+
+void DeduceBlitImages(ImageInfo& dst_info, ImageInfo& src_info, const ImageBase* dst,
+ const ImageBase* src) {
+ if (src && GetFormatType(src->info.format) != SurfaceType::ColorTexture) {
+ src_info.format = src->info.format;
+ }
+ if (dst && GetFormatType(dst->info.format) != SurfaceType::ColorTexture) {
+ dst_info.format = dst->info.format;
+ }
+ if (!dst && src && GetFormatType(src->info.format) != SurfaceType::ColorTexture) {
+ dst_info.format = src->info.format;
+ }
+ if (!src && dst && GetFormatType(dst->info.format) != SurfaceType::ColorTexture) {
+ src_info.format = src->info.format;
+ }
+}
+
+u32 MapSizeBytes(const ImageBase& image) {
+ if (True(image.flags & ImageFlagBits::AcceleratedUpload)) {
+ return image.guest_size_bytes;
+ } else if (True(image.flags & ImageFlagBits::Converted)) {
+ return image.converted_size_bytes;
+ } else {
+ return image.unswizzled_size_bytes;
+ }
+}
+
+using P = PixelFormat;
+
+static_assert(CalculateLevelSize(LevelInfo{{1920, 1080}, {0, 2, 0}, {1, 1}, 2, 0}, 0) == 0x7f8000);
+static_assert(CalculateLevelSize(LevelInfo{{32, 32}, {0, 0, 4}, {1, 1}, 4, 0}, 0) == 0x4000);
+
+static_assert(CalculateLevelOffset(P::R8_SINT, {1920, 1080}, {0, 2}, 1, 0, 7) == 0x2afc00);
+static_assert(CalculateLevelOffset(P::ASTC_2D_12X12_UNORM, {8192, 4096}, {0, 2}, 1, 0, 12) ==
+ 0x50d200);
+
+static_assert(CalculateLevelOffset(P::A8B8G8R8_UNORM, {1024, 1024}, {0, 4}, 1, 0, 0) == 0);
+static_assert(CalculateLevelOffset(P::A8B8G8R8_UNORM, {1024, 1024}, {0, 4}, 1, 0, 1) == 0x400000);
+static_assert(CalculateLevelOffset(P::A8B8G8R8_UNORM, {1024, 1024}, {0, 4}, 1, 0, 2) == 0x500000);
+static_assert(CalculateLevelOffset(P::A8B8G8R8_UNORM, {1024, 1024}, {0, 4}, 1, 0, 3) == 0x540000);
+static_assert(CalculateLevelOffset(P::A8B8G8R8_UNORM, {1024, 1024}, {0, 4}, 1, 0, 4) == 0x550000);
+static_assert(CalculateLevelOffset(P::A8B8G8R8_UNORM, {1024, 1024}, {0, 4}, 1, 0, 5) == 0x554000);
+static_assert(CalculateLevelOffset(P::A8B8G8R8_UNORM, {1024, 1024}, {0, 4}, 1, 0, 6) == 0x555000);
+static_assert(CalculateLevelOffset(P::A8B8G8R8_UNORM, {1024, 1024}, {0, 4}, 1, 0, 7) == 0x555400);
+static_assert(CalculateLevelOffset(P::A8B8G8R8_UNORM, {1024, 1024}, {0, 4}, 1, 0, 8) == 0x555600);
+static_assert(CalculateLevelOffset(P::A8B8G8R8_UNORM, {1024, 1024}, {0, 4}, 1, 0, 9) == 0x555800);
+
+constexpr u32 ValidateLayerSize(PixelFormat format, u32 width, u32 height, u32 block_height,
+ u32 tile_width_spacing, u32 level) {
+ const Extent3D size{width, height, 1};
+ const Extent3D block{0, block_height, 0};
+ const u32 offset = CalculateLevelOffset(format, size, block, 1, tile_width_spacing, level);
+ return AlignLayerSize(offset, size, block, DefaultBlockHeight(format), tile_width_spacing);
+}
+
+static_assert(ValidateLayerSize(P::ASTC_2D_12X12_UNORM, 8192, 4096, 2, 0, 12) == 0x50d800);
+static_assert(ValidateLayerSize(P::A8B8G8R8_UNORM, 1024, 1024, 2, 0, 10) == 0x556000);
+static_assert(ValidateLayerSize(P::BC3_UNORM, 128, 128, 2, 0, 8) == 0x6000);
+
+static_assert(ValidateLayerSize(P::A8B8G8R8_UNORM, 518, 572, 4, 3, 1) == 0x190000,
+ "Tile width spacing is not working");
+static_assert(ValidateLayerSize(P::BC5_UNORM, 1024, 1024, 3, 4, 11) == 0x160000,
+ "Compressed tile width spacing is not working");
+
+} // namespace VideoCommon
diff --git a/src/video_core/texture_cache/util.h b/src/video_core/texture_cache/util.h
new file mode 100644
index 000000000..dbbbd33cd
--- /dev/null
+++ b/src/video_core/texture_cache/util.h
@@ -0,0 +1,107 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <optional>
+#include <span>
+
+#include "common/common_types.h"
+
+#include "video_core/engines/maxwell_3d.h"
+#include "video_core/surface.h"
+#include "video_core/texture_cache/image_base.h"
+#include "video_core/texture_cache/image_view_base.h"
+#include "video_core/texture_cache/types.h"
+#include "video_core/textures/texture.h"
+
+namespace VideoCommon {
+
+using Tegra::Texture::TICEntry;
+
+struct OverlapResult {
+ GPUVAddr gpu_addr;
+ VAddr cpu_addr;
+ SubresourceExtent resources;
+};
+
+[[nodiscard]] u32 CalculateGuestSizeInBytes(const ImageInfo& info) noexcept;
+
+[[nodiscard]] u32 CalculateUnswizzledSizeBytes(const ImageInfo& info) noexcept;
+
+[[nodiscard]] u32 CalculateConvertedSizeBytes(const ImageInfo& info) noexcept;
+
+[[nodiscard]] u32 CalculateLayerStride(const ImageInfo& info) noexcept;
+
+[[nodiscard]] u32 CalculateLayerSize(const ImageInfo& info) noexcept;
+
+[[nodiscard]] std::array<u32, MAX_MIP_LEVELS> CalculateMipLevelOffsets(
+ const ImageInfo& info) noexcept;
+
+[[nodiscard]] std::vector<u32> CalculateSliceOffsets(const ImageInfo& info);
+
+[[nodiscard]] std::vector<SubresourceBase> CalculateSliceSubresources(const ImageInfo& info);
+
+[[nodiscard]] u32 CalculateLevelStrideAlignment(const ImageInfo& info, u32 level);
+
+[[nodiscard]] VideoCore::Surface::PixelFormat PixelFormatFromTIC(
+ const Tegra::Texture::TICEntry& config) noexcept;
+
+[[nodiscard]] ImageViewType RenderTargetImageViewType(const ImageInfo& info) noexcept;
+
+[[nodiscard]] std::vector<ImageCopy> MakeShrinkImageCopies(const ImageInfo& dst,
+ const ImageInfo& src,
+ SubresourceBase base);
+
+[[nodiscard]] bool IsValidAddress(const Tegra::MemoryManager& gpu_memory, const TICEntry& config);
+
+[[nodiscard]] std::vector<BufferImageCopy> UnswizzleImage(Tegra::MemoryManager& gpu_memory,
+ GPUVAddr gpu_addr, const ImageInfo& info,
+ std::span<u8> output);
+
+[[nodiscard]] BufferCopy UploadBufferCopy(Tegra::MemoryManager& gpu_memory, GPUVAddr gpu_addr,
+ const ImageBase& image, std::span<u8> output);
+
+void ConvertImage(std::span<const u8> input, const ImageInfo& info, std::span<u8> output,
+ std::span<BufferImageCopy> copies);
+
+[[nodiscard]] std::vector<BufferImageCopy> FullDownloadCopies(const ImageInfo& info);
+
+[[nodiscard]] Extent3D MipSize(Extent3D size, u32 level);
+
+[[nodiscard]] Extent3D MipBlockSize(const ImageInfo& info, u32 level);
+
+[[nodiscard]] std::vector<SwizzleParameters> FullUploadSwizzles(const ImageInfo& info);
+
+void SwizzleImage(Tegra::MemoryManager& gpu_memory, GPUVAddr gpu_addr, const ImageInfo& info,
+ std::span<const BufferImageCopy> copies, std::span<const u8> memory);
+
+[[nodiscard]] bool IsBlockLinearSizeCompatible(const ImageInfo& new_info,
+ const ImageInfo& overlap_info, u32 new_level,
+ u32 overlap_level, bool strict_size) noexcept;
+
+[[nodiscard]] bool IsPitchLinearSameSize(const ImageInfo& lhs, const ImageInfo& rhs,
+ bool strict_size) noexcept;
+
+[[nodiscard]] std::optional<OverlapResult> ResolveOverlap(const ImageInfo& new_info,
+ GPUVAddr gpu_addr, VAddr cpu_addr,
+ const ImageBase& overlap,
+ bool strict_size);
+
+[[nodiscard]] bool IsLayerStrideCompatible(const ImageInfo& lhs, const ImageInfo& rhs);
+
+[[nodiscard]] std::optional<SubresourceBase> FindSubresource(const ImageInfo& candidate,
+ const ImageBase& image,
+ GPUVAddr candidate_addr,
+ RelaxedOptions options);
+
+[[nodiscard]] bool IsSubresource(const ImageInfo& candidate, const ImageBase& image,
+ GPUVAddr candidate_addr, RelaxedOptions options);
+
+void DeduceBlitImages(ImageInfo& dst_info, ImageInfo& src_info, const ImageBase* dst,
+ const ImageBase* src);
+
+[[nodiscard]] u32 MapSizeBytes(const ImageBase& image);
+
+} // namespace VideoCommon
diff --git a/src/video_core/textures/astc.cpp b/src/video_core/textures/astc.cpp
index 365bde2f1..acd5bdd78 100644
--- a/src/video_core/textures/astc.cpp
+++ b/src/video_core/textures/astc.cpp
@@ -18,6 +18,7 @@
#include <algorithm>
#include <cassert>
#include <cstring>
+#include <span>
#include <vector>
#include <boost/container/static_vector.hpp>
@@ -600,7 +601,7 @@ static TexelWeightParams DecodeBlockInfo(InputBitStream& strm) {
return params;
}
-static void FillVoidExtentLDR(InputBitStream& strm, u32* const outBuf, u32 blockWidth,
+static void FillVoidExtentLDR(InputBitStream& strm, std::span<u32> outBuf, u32 blockWidth,
u32 blockHeight) {
// Don't actually care about the void extent, just read the bits...
for (s32 i = 0; i < 4; ++i) {
@@ -623,7 +624,7 @@ static void FillVoidExtentLDR(InputBitStream& strm, u32* const outBuf, u32 block
}
}
-static void FillError(u32* outBuf, u32 blockWidth, u32 blockHeight) {
+static void FillError(std::span<u32> outBuf, u32 blockWidth, u32 blockHeight) {
for (u32 j = 0; j < blockHeight; j++) {
for (u32 i = 0; i < blockWidth; i++) {
outBuf[j * blockWidth + i] = 0xFFFF00FF;
@@ -1438,9 +1439,9 @@ static void ComputeEndpos32s(Pixel& ep1, Pixel& ep2, const u32*& colorValues,
#undef READ_INT_VALUES
}
-static void DecompressBlock(const u8 inBuf[16], const u32 blockWidth, const u32 blockHeight,
- u32* outBuf) {
- InputBitStream strm(inBuf);
+static void DecompressBlock(std::span<const u8, 16> inBuf, const u32 blockWidth,
+ const u32 blockHeight, std::span<u32, 12 * 12> outBuf) {
+ InputBitStream strm(inBuf.data());
TexelWeightParams weightParams = DecodeBlockInfo(strm);
// Was there an error?
@@ -1601,8 +1602,8 @@ static void DecompressBlock(const u8 inBuf[16], const u32 blockWidth, const u32
}
// Read the texel weight data..
- u8 texelWeightData[16];
- memcpy(texelWeightData, inBuf, sizeof(texelWeightData));
+ std::array<u8, 16> texelWeightData;
+ std::ranges::copy(inBuf, texelWeightData.begin());
// Reverse everything
for (u32 i = 0; i < 8; i++) {
@@ -1618,14 +1619,15 @@ static void DecompressBlock(const u8 inBuf[16], const u32 blockWidth, const u32
// Make sure that higher non-texel bits are set to zero
const u32 clearByteStart = (weightParams.GetPackedBitSize() >> 3) + 1;
- texelWeightData[clearByteStart - 1] =
- texelWeightData[clearByteStart - 1] &
- static_cast<u8>((1 << (weightParams.GetPackedBitSize() % 8)) - 1);
- memset(texelWeightData + clearByteStart, 0, 16 - clearByteStart);
+ if (clearByteStart > 0) {
+ texelWeightData[clearByteStart - 1] &=
+ static_cast<u8>((1 << (weightParams.GetPackedBitSize() % 8)) - 1);
+ }
+ std::memset(texelWeightData.data() + clearByteStart, 0, std::min(16U - clearByteStart, 16U));
IntegerEncodedVector texelWeightValues;
- InputBitStream weightStream(texelWeightData);
+ InputBitStream weightStream(texelWeightData.data());
DecodeIntegerSequence(texelWeightValues, weightStream, weightParams.m_MaxWeight,
weightParams.GetNumWeightValues());
@@ -1672,36 +1674,32 @@ static void DecompressBlock(const u8 inBuf[16], const u32 blockWidth, const u32
namespace Tegra::Texture::ASTC {
-std::vector<u8> Decompress(const u8* data, u32 width, u32 height, u32 depth, u32 block_width,
- u32 block_height) {
- u32 blockIdx = 0;
+void Decompress(std::span<const uint8_t> data, uint32_t width, uint32_t height, uint32_t depth,
+ uint32_t block_width, uint32_t block_height, std::span<uint8_t> output) {
+ u32 block_index = 0;
std::size_t depth_offset = 0;
- std::vector<u8> outData(height * width * depth * 4);
- for (u32 k = 0; k < depth; k++) {
- for (u32 j = 0; j < height; j += block_height) {
- for (u32 i = 0; i < width; i += block_width) {
-
- const u8* blockPtr = data + blockIdx * 16;
+ for (u32 z = 0; z < depth; z++) {
+ for (u32 y = 0; y < height; y += block_height) {
+ for (u32 x = 0; x < width; x += block_width) {
+ const std::span<const u8, 16> blockPtr{data.subspan(block_index * 16, 16)};
// Blocks can be at most 12x12
- u32 uncompData[144];
+ std::array<u32, 12 * 12> uncompData;
ASTCC::DecompressBlock(blockPtr, block_width, block_height, uncompData);
- u32 decompWidth = std::min(block_width, width - i);
- u32 decompHeight = std::min(block_height, height - j);
+ u32 decompWidth = std::min(block_width, width - x);
+ u32 decompHeight = std::min(block_height, height - y);
- u8* outRow = depth_offset + outData.data() + (j * width + i) * 4;
+ const std::span<u8> outRow = output.subspan(depth_offset + (y * width + x) * 4);
for (u32 jj = 0; jj < decompHeight; jj++) {
- memcpy(outRow + jj * width * 4, uncompData + jj * block_width, decompWidth * 4);
+ std::memcpy(outRow.data() + jj * width * 4,
+ uncompData.data() + jj * block_width, decompWidth * 4);
}
-
- blockIdx++;
+ ++block_index;
}
}
depth_offset += height * width * 4;
}
-
- return outData;
}
} // namespace Tegra::Texture::ASTC
diff --git a/src/video_core/textures/astc.h b/src/video_core/textures/astc.h
index 991cdba72..9105119bc 100644
--- a/src/video_core/textures/astc.h
+++ b/src/video_core/textures/astc.h
@@ -5,11 +5,10 @@
#pragma once
#include <cstdint>
-#include <vector>
namespace Tegra::Texture::ASTC {
-std::vector<uint8_t> Decompress(const uint8_t* data, uint32_t width, uint32_t height,
- uint32_t depth, uint32_t block_width, uint32_t block_height);
+void Decompress(std::span<const uint8_t> data, uint32_t width, uint32_t height, uint32_t depth,
+ uint32_t block_width, uint32_t block_height, std::span<uint8_t> output);
} // namespace Tegra::Texture::ASTC
diff --git a/src/video_core/textures/convert.cpp b/src/video_core/textures/convert.cpp
deleted file mode 100644
index 962921483..000000000
--- a/src/video_core/textures/convert.cpp
+++ /dev/null
@@ -1,93 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include <algorithm>
-#include <cstring>
-#include <tuple>
-#include <vector>
-
-#include "common/assert.h"
-#include "common/common_types.h"
-#include "common/logging/log.h"
-#include "video_core/surface.h"
-#include "video_core/textures/astc.h"
-#include "video_core/textures/convert.h"
-
-namespace Tegra::Texture {
-
-using VideoCore::Surface::PixelFormat;
-
-template <bool reverse>
-void SwapS8Z24ToZ24S8(u8* data, u32 width, u32 height) {
- union S8Z24 {
- BitField<0, 24, u32> z24;
- BitField<24, 8, u32> s8;
- };
- static_assert(sizeof(S8Z24) == 4, "S8Z24 is incorrect size");
-
- union Z24S8 {
- BitField<0, 8, u32> s8;
- BitField<8, 24, u32> z24;
- };
- static_assert(sizeof(Z24S8) == 4, "Z24S8 is incorrect size");
-
- S8Z24 s8z24_pixel{};
- Z24S8 z24s8_pixel{};
- constexpr auto bpp{
- VideoCore::Surface::GetBytesPerPixel(VideoCore::Surface::PixelFormat::S8_UINT_D24_UNORM)};
- for (std::size_t y = 0; y < height; ++y) {
- for (std::size_t x = 0; x < width; ++x) {
- const std::size_t offset{bpp * (y * width + x)};
- if constexpr (reverse) {
- std::memcpy(&z24s8_pixel, &data[offset], sizeof(Z24S8));
- s8z24_pixel.s8.Assign(z24s8_pixel.s8);
- s8z24_pixel.z24.Assign(z24s8_pixel.z24);
- std::memcpy(&data[offset], &s8z24_pixel, sizeof(S8Z24));
- } else {
- std::memcpy(&s8z24_pixel, &data[offset], sizeof(S8Z24));
- z24s8_pixel.s8.Assign(s8z24_pixel.s8);
- z24s8_pixel.z24.Assign(s8z24_pixel.z24);
- std::memcpy(&data[offset], &z24s8_pixel, sizeof(Z24S8));
- }
- }
- }
-}
-
-static void ConvertS8Z24ToZ24S8(u8* data, u32 width, u32 height) {
- SwapS8Z24ToZ24S8<false>(data, width, height);
-}
-
-static void ConvertZ24S8ToS8Z24(u8* data, u32 width, u32 height) {
- SwapS8Z24ToZ24S8<true>(data, width, height);
-}
-
-void ConvertFromGuestToHost(u8* in_data, u8* out_data, PixelFormat pixel_format, u32 width,
- u32 height, u32 depth, bool convert_astc, bool convert_s8z24) {
- if (convert_astc && IsPixelFormatASTC(pixel_format)) {
- // Convert ASTC pixel formats to RGBA8, as most desktop GPUs do not support ASTC.
- u32 block_width{};
- u32 block_height{};
- std::tie(block_width, block_height) = GetASTCBlockSize(pixel_format);
- const std::vector<u8> rgba8_data = Tegra::Texture::ASTC::Decompress(
- in_data, width, height, depth, block_width, block_height);
- std::copy(rgba8_data.begin(), rgba8_data.end(), out_data);
-
- } else if (convert_s8z24 && pixel_format == PixelFormat::S8_UINT_D24_UNORM) {
- Tegra::Texture::ConvertS8Z24ToZ24S8(in_data, width, height);
- }
-}
-
-void ConvertFromHostToGuest(u8* data, PixelFormat pixel_format, u32 width, u32 height, u32 depth,
- bool convert_astc, bool convert_s8z24) {
- if (convert_astc && IsPixelFormatASTC(pixel_format)) {
- LOG_CRITICAL(HW_GPU, "Conversion of format {} after texture flushing is not implemented",
- static_cast<u32>(pixel_format));
- UNREACHABLE();
-
- } else if (convert_s8z24 && pixel_format == PixelFormat::S8_UINT_D24_UNORM) {
- Tegra::Texture::ConvertZ24S8ToS8Z24(data, width, height);
- }
-}
-
-} // namespace Tegra::Texture
diff --git a/src/video_core/textures/convert.h b/src/video_core/textures/convert.h
deleted file mode 100644
index d5d6c77bb..000000000
--- a/src/video_core/textures/convert.h
+++ /dev/null
@@ -1,22 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include "common/common_types.h"
-
-namespace VideoCore::Surface {
-enum class PixelFormat;
-}
-
-namespace Tegra::Texture {
-
-void ConvertFromGuestToHost(u8* in_data, u8* out_data, VideoCore::Surface::PixelFormat pixel_format,
- u32 width, u32 height, u32 depth, bool convert_astc,
- bool convert_s8z24);
-
-void ConvertFromHostToGuest(u8* data, VideoCore::Surface::PixelFormat pixel_format, u32 width,
- u32 height, u32 depth, bool convert_astc, bool convert_s8z24);
-
-} // namespace Tegra::Texture
diff --git a/src/video_core/textures/decoders.cpp b/src/video_core/textures/decoders.cpp
index 16d46a018..9f5181318 100644
--- a/src/video_core/textures/decoders.cpp
+++ b/src/video_core/textures/decoders.cpp
@@ -2,204 +2,111 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
+#include <array>
#include <cmath>
#include <cstring>
+#include <span>
+#include <utility>
+
#include "common/alignment.h"
#include "common/assert.h"
#include "common/bit_util.h"
+#include "common/div_ceil.h"
#include "video_core/gpu.h"
#include "video_core/textures/decoders.h"
#include "video_core/textures/texture.h"
namespace Tegra::Texture {
-namespace {
+namespace {
/**
- * This table represents the internal swizzle of a gob,
- * in format 16 bytes x 2 sector packing.
+ * This table represents the internal swizzle of a gob, in format 16 bytes x 2 sector packing.
* Calculates the offset of an (x, y) position within a swizzled texture.
* Taken from the Tegra X1 Technical Reference Manual. pages 1187-1188
*/
-template <std::size_t N, std::size_t M, u32 Align>
-struct alignas(64) SwizzleTable {
- static_assert(M * Align == 64, "Swizzle Table does not align to GOB");
- constexpr SwizzleTable() {
- for (u32 y = 0; y < N; ++y) {
- for (u32 x = 0; x < M; ++x) {
- const u32 x2 = x * Align;
- values[y][x] = static_cast<u16>(((x2 % 64) / 32) * 256 + ((y % 8) / 2) * 64 +
- ((x2 % 32) / 16) * 32 + (y % 2) * 16 + (x2 % 16));
- }
+constexpr SwizzleTable MakeSwizzleTableConst() {
+ SwizzleTable table{};
+ for (u32 y = 0; y < table.size(); ++y) {
+ for (u32 x = 0; x < table[0].size(); ++x) {
+ table[y][x] = ((x % 64) / 32) * 256 + ((y % 8) / 2) * 64 + ((x % 32) / 16) * 32 +
+ (y % 2) * 16 + (x % 16);
}
}
- const std::array<u16, M>& operator[](std::size_t index) const {
- return values[index];
- }
- std::array<std::array<u16, M>, N> values{};
-};
+ return table;
+}
-constexpr u32 FAST_SWIZZLE_ALIGN = 16;
+constexpr SwizzleTable SWIZZLE_TABLE = MakeSwizzleTableConst();
-constexpr auto LEGACY_SWIZZLE_TABLE = SwizzleTable<GOB_SIZE_X, GOB_SIZE_X, GOB_SIZE_Z>();
-constexpr auto FAST_SWIZZLE_TABLE = SwizzleTable<GOB_SIZE_Y, 4, FAST_SWIZZLE_ALIGN>();
+template <bool TO_LINEAR>
+void Swizzle(std::span<u8> output, std::span<const u8> input, u32 bytes_per_pixel, u32 width,
+ u32 height, u32 depth, u32 block_height, u32 block_depth, u32 stride_alignment) {
+ // The origin of the transformation can be configured here, leave it as zero as the current API
+ // doesn't expose it.
+ static constexpr u32 origin_x = 0;
+ static constexpr u32 origin_y = 0;
+ static constexpr u32 origin_z = 0;
-/**
- * This function manages ALL the GOBs(Group of Bytes) Inside a single block.
- * Instead of going gob by gob, we map the coordinates inside a block and manage from
- * those. Block_Width is assumed to be 1.
- */
-void PreciseProcessBlock(u8* const swizzled_data, u8* const unswizzled_data, const bool unswizzle,
- const u32 x_start, const u32 y_start, const u32 z_start, const u32 x_end,
- const u32 y_end, const u32 z_end, const u32 tile_offset,
- const u32 xy_block_size, const u32 layer_z, const u32 stride_x,
- const u32 bytes_per_pixel, const u32 out_bytes_per_pixel) {
- std::array<u8*, 2> data_ptrs;
- u32 z_address = tile_offset;
-
- for (u32 z = z_start; z < z_end; z++) {
- u32 y_address = z_address;
- u32 pixel_base = layer_z * z + y_start * stride_x;
- for (u32 y = y_start; y < y_end; y++) {
- const auto& table = LEGACY_SWIZZLE_TABLE[y % GOB_SIZE_Y];
- for (u32 x = x_start; x < x_end; x++) {
- const u32 swizzle_offset{y_address + table[x * bytes_per_pixel % GOB_SIZE_X]};
- const u32 pixel_index{x * out_bytes_per_pixel + pixel_base};
- data_ptrs[unswizzle] = swizzled_data + swizzle_offset;
- data_ptrs[!unswizzle] = unswizzled_data + pixel_index;
- std::memcpy(data_ptrs[0], data_ptrs[1], bytes_per_pixel);
- }
- pixel_base += stride_x;
- if ((y + 1) % GOB_SIZE_Y == 0)
- y_address += GOB_SIZE;
- }
- z_address += xy_block_size;
- }
-}
+ // We can configure here a custom pitch
+ // As it's not exposed 'width * bpp' will be the expected pitch.
+ const u32 pitch = width * bytes_per_pixel;
+ const u32 stride = Common::AlignBits(width, stride_alignment) * bytes_per_pixel;
-/**
- * This function manages ALL the GOBs(Group of Bytes) Inside a single block.
- * Instead of going gob by gob, we map the coordinates inside a block and manage from
- * those. Block_Width is assumed to be 1.
- */
-void FastProcessBlock(u8* const swizzled_data, u8* const unswizzled_data, const bool unswizzle,
- const u32 x_start, const u32 y_start, const u32 z_start, const u32 x_end,
- const u32 y_end, const u32 z_end, const u32 tile_offset,
- const u32 xy_block_size, const u32 layer_z, const u32 stride_x,
- const u32 bytes_per_pixel, const u32 out_bytes_per_pixel) {
- std::array<u8*, 2> data_ptrs;
- u32 z_address = tile_offset;
- const u32 x_startb = x_start * bytes_per_pixel;
- const u32 x_endb = x_end * bytes_per_pixel;
-
- for (u32 z = z_start; z < z_end; z++) {
- u32 y_address = z_address;
- u32 pixel_base = layer_z * z + y_start * stride_x;
- for (u32 y = y_start; y < y_end; y++) {
- const auto& table = FAST_SWIZZLE_TABLE[y % GOB_SIZE_Y];
- for (u32 xb = x_startb; xb < x_endb; xb += FAST_SWIZZLE_ALIGN) {
- const u32 swizzle_offset{y_address + table[(xb / FAST_SWIZZLE_ALIGN) % 4]};
- const u32 out_x = xb * out_bytes_per_pixel / bytes_per_pixel;
- const u32 pixel_index{out_x + pixel_base};
- data_ptrs[unswizzle ? 1 : 0] = swizzled_data + swizzle_offset;
- data_ptrs[unswizzle ? 0 : 1] = unswizzled_data + pixel_index;
- std::memcpy(data_ptrs[0], data_ptrs[1], FAST_SWIZZLE_ALIGN);
- }
- pixel_base += stride_x;
- if ((y + 1) % GOB_SIZE_Y == 0)
- y_address += GOB_SIZE;
- }
- z_address += xy_block_size;
- }
-}
+ const u32 gobs_in_x = Common::DivCeilLog2(stride, GOB_SIZE_X_SHIFT);
+ const u32 block_size = gobs_in_x << (GOB_SIZE_SHIFT + block_height + block_depth);
+ const u32 slice_size =
+ Common::DivCeilLog2(height, block_height + GOB_SIZE_Y_SHIFT) * block_size;
-/**
- * This function unswizzles or swizzles a texture by mapping Linear to BlockLinear Textue.
- * The body of this function takes care of splitting the swizzled texture into blocks,
- * and managing the extents of it. Once all the parameters of a single block are obtained,
- * the function calls 'ProcessBlock' to process that particular Block.
- *
- * Documentation for the memory layout and decoding can be found at:
- * https://envytools.readthedocs.io/en/latest/hw/memory/g80-surface.html#blocklinear-surfaces
- */
-template <bool fast>
-void SwizzledData(u8* const swizzled_data, u8* const unswizzled_data, const bool unswizzle,
- const u32 width, const u32 height, const u32 depth, const u32 bytes_per_pixel,
- const u32 out_bytes_per_pixel, const u32 block_height, const u32 block_depth,
- const u32 width_spacing) {
- auto div_ceil = [](const u32 x, const u32 y) { return ((x + y - 1) / y); };
- const u32 stride_x = width * out_bytes_per_pixel;
- const u32 layer_z = height * stride_x;
- const u32 gob_elements_x = GOB_SIZE_X / bytes_per_pixel;
- constexpr u32 gob_elements_y = GOB_SIZE_Y;
- constexpr u32 gob_elements_z = GOB_SIZE_Z;
- const u32 block_x_elements = gob_elements_x;
- const u32 block_y_elements = gob_elements_y * block_height;
- const u32 block_z_elements = gob_elements_z * block_depth;
- const u32 aligned_width = Common::AlignUp(width, gob_elements_x * width_spacing);
- const u32 blocks_on_x = div_ceil(aligned_width, block_x_elements);
- const u32 blocks_on_y = div_ceil(height, block_y_elements);
- const u32 blocks_on_z = div_ceil(depth, block_z_elements);
- const u32 xy_block_size = GOB_SIZE * block_height;
- const u32 block_size = xy_block_size * block_depth;
- u32 tile_offset = 0;
- for (u32 zb = 0; zb < blocks_on_z; zb++) {
- const u32 z_start = zb * block_z_elements;
- const u32 z_end = std::min(depth, z_start + block_z_elements);
- for (u32 yb = 0; yb < blocks_on_y; yb++) {
- const u32 y_start = yb * block_y_elements;
- const u32 y_end = std::min(height, y_start + block_y_elements);
- for (u32 xb = 0; xb < blocks_on_x; xb++) {
- const u32 x_start = xb * block_x_elements;
- const u32 x_end = std::min(width, x_start + block_x_elements);
- if constexpr (fast) {
- FastProcessBlock(swizzled_data, unswizzled_data, unswizzle, x_start, y_start,
- z_start, x_end, y_end, z_end, tile_offset, xy_block_size,
- layer_z, stride_x, bytes_per_pixel, out_bytes_per_pixel);
- } else {
- PreciseProcessBlock(swizzled_data, unswizzled_data, unswizzle, x_start, y_start,
- z_start, x_end, y_end, z_end, tile_offset, xy_block_size,
- layer_z, stride_x, bytes_per_pixel, out_bytes_per_pixel);
- }
- tile_offset += block_size;
+ const u32 block_height_mask = (1U << block_height) - 1;
+ const u32 block_depth_mask = (1U << block_depth) - 1;
+ const u32 x_shift = GOB_SIZE_SHIFT + block_height + block_depth;
+
+ for (u32 slice = 0; slice < depth; ++slice) {
+ const u32 z = slice + origin_z;
+ const u32 offset_z = (z >> block_depth) * slice_size +
+ ((z & block_depth_mask) << (GOB_SIZE_SHIFT + block_height));
+ for (u32 line = 0; line < height; ++line) {
+ const u32 y = line + origin_y;
+ const auto& table = SWIZZLE_TABLE[y % GOB_SIZE_Y];
+
+ const u32 block_y = y >> GOB_SIZE_Y_SHIFT;
+ const u32 offset_y = (block_y >> block_height) * block_size +
+ ((block_y & block_height_mask) << GOB_SIZE_SHIFT);
+
+ for (u32 column = 0; column < width; ++column) {
+ const u32 x = (column + origin_x) * bytes_per_pixel;
+ const u32 offset_x = (x >> GOB_SIZE_X_SHIFT) << x_shift;
+
+ const u32 base_swizzled_offset = offset_z + offset_y + offset_x;
+ const u32 swizzled_offset = base_swizzled_offset + table[x % GOB_SIZE_X];
+
+ const u32 unswizzled_offset =
+ slice * pitch * height + line * pitch + column * bytes_per_pixel;
+
+ u8* const dst = &output[TO_LINEAR ? swizzled_offset : unswizzled_offset];
+ const u8* const src = &input[TO_LINEAR ? unswizzled_offset : swizzled_offset];
+ std::memcpy(dst, src, bytes_per_pixel);
}
}
}
}
-
} // Anonymous namespace
-void CopySwizzledData(u32 width, u32 height, u32 depth, u32 bytes_per_pixel,
- u32 out_bytes_per_pixel, u8* const swizzled_data, u8* const unswizzled_data,
- bool unswizzle, u32 block_height, u32 block_depth, u32 width_spacing) {
- const u32 block_height_size{1U << block_height};
- const u32 block_depth_size{1U << block_depth};
- if (bytes_per_pixel % 3 != 0 && (width * bytes_per_pixel) % FAST_SWIZZLE_ALIGN == 0) {
- SwizzledData<true>(swizzled_data, unswizzled_data, unswizzle, width, height, depth,
- bytes_per_pixel, out_bytes_per_pixel, block_height_size,
- block_depth_size, width_spacing);
- } else {
- SwizzledData<false>(swizzled_data, unswizzled_data, unswizzle, width, height, depth,
- bytes_per_pixel, out_bytes_per_pixel, block_height_size,
- block_depth_size, width_spacing);
- }
+SwizzleTable MakeSwizzleTable() {
+ return SWIZZLE_TABLE;
}
-void UnswizzleTexture(u8* const unswizzled_data, u8* address, u32 tile_size_x, u32 tile_size_y,
- u32 bytes_per_pixel, u32 width, u32 height, u32 depth, u32 block_height,
- u32 block_depth, u32 width_spacing) {
- CopySwizzledData((width + tile_size_x - 1) / tile_size_x,
- (height + tile_size_y - 1) / tile_size_y, depth, bytes_per_pixel,
- bytes_per_pixel, address, unswizzled_data, true, block_height, block_depth,
- width_spacing);
+void UnswizzleTexture(std::span<u8> output, std::span<const u8> input, u32 bytes_per_pixel,
+ u32 width, u32 height, u32 depth, u32 block_height, u32 block_depth,
+ u32 stride_alignment) {
+ Swizzle<false>(output, input, bytes_per_pixel, width, height, depth, block_height, block_depth,
+ stride_alignment);
}
-std::vector<u8> UnswizzleTexture(u8* address, u32 tile_size_x, u32 tile_size_y, u32 bytes_per_pixel,
- u32 width, u32 height, u32 depth, u32 block_height,
- u32 block_depth, u32 width_spacing) {
- std::vector<u8> unswizzled_data(width * height * depth * bytes_per_pixel);
- UnswizzleTexture(unswizzled_data.data(), address, tile_size_x, tile_size_y, bytes_per_pixel,
- width, height, depth, block_height, block_depth, width_spacing);
- return unswizzled_data;
+void SwizzleTexture(std::span<u8> output, std::span<const u8> input, u32 bytes_per_pixel, u32 width,
+ u32 height, u32 depth, u32 block_height, u32 block_depth,
+ u32 stride_alignment) {
+ Swizzle<true>(output, input, bytes_per_pixel, width, height, depth, block_height, block_depth,
+ stride_alignment);
}
void SwizzleSubrect(u32 subrect_width, u32 subrect_height, u32 source_pitch, u32 swizzled_width,
@@ -213,7 +120,7 @@ void SwizzleSubrect(u32 subrect_width, u32 subrect_height, u32 source_pitch, u32
const u32 gob_address_y =
(dst_y / (GOB_SIZE_Y * block_height)) * GOB_SIZE * block_height * image_width_in_gobs +
((dst_y % (GOB_SIZE_Y * block_height)) / GOB_SIZE_Y) * GOB_SIZE;
- const auto& table = LEGACY_SWIZZLE_TABLE[dst_y % GOB_SIZE_Y];
+ const auto& table = SWIZZLE_TABLE[dst_y % GOB_SIZE_Y];
for (u32 x = 0; x < subrect_width; ++x) {
const u32 dst_x = x + offset_x;
const u32 gob_address =
@@ -235,11 +142,11 @@ void UnswizzleSubrect(u32 line_length_in, u32 line_count, u32 pitch, u32 width,
const u32 block_size = gobs_in_x << (GOB_SIZE_SHIFT + block_height);
const u32 block_height_mask = (1U << block_height) - 1;
- const u32 x_shift = static_cast<u32>(GOB_SIZE_SHIFT) + block_height;
+ const u32 x_shift = GOB_SIZE_SHIFT + block_height;
for (u32 line = 0; line < line_count; ++line) {
const u32 src_y = line + origin_y;
- const auto& table = LEGACY_SWIZZLE_TABLE[src_y % GOB_SIZE_Y];
+ const auto& table = SWIZZLE_TABLE[src_y % GOB_SIZE_Y];
const u32 block_y = src_y >> GOB_SIZE_Y_SHIFT;
const u32 src_offset_y = (block_y >> block_height) * block_size +
@@ -270,7 +177,7 @@ void SwizzleSliceToVoxel(u32 line_length_in, u32 line_count, u32 pitch, u32 widt
const u32 x_shift = static_cast<u32>(GOB_SIZE_SHIFT) + block_height + block_depth;
for (u32 line = 0; line < line_count; ++line) {
- const auto& table = LEGACY_SWIZZLE_TABLE[line % GOB_SIZE_Y];
+ const auto& table = SWIZZLE_TABLE[line % GOB_SIZE_Y];
const u32 block_y = line / GOB_SIZE_Y;
const u32 dst_offset_y =
(block_y >> block_height) * block_size + (block_y & block_height_mask) * GOB_SIZE;
@@ -293,7 +200,7 @@ void SwizzleKepler(const u32 width, const u32 height, const u32 dst_x, const u32
const std::size_t gob_address_y =
(y / (GOB_SIZE_Y * block_height)) * GOB_SIZE * block_height * image_width_in_gobs +
((y % (GOB_SIZE_Y * block_height)) / GOB_SIZE_Y) * GOB_SIZE;
- const auto& table = LEGACY_SWIZZLE_TABLE[y % GOB_SIZE_Y];
+ const auto& table = SWIZZLE_TABLE[y % GOB_SIZE_Y];
for (std::size_t x = dst_x; x < width && count < copy_size; ++x) {
const std::size_t gob_address =
gob_address_y + (x / GOB_SIZE_X) * GOB_SIZE * block_height;
diff --git a/src/video_core/textures/decoders.h b/src/video_core/textures/decoders.h
index 01e156bc8..d7cdc81e8 100644
--- a/src/video_core/textures/decoders.h
+++ b/src/video_core/textures/decoders.h
@@ -4,7 +4,8 @@
#pragma once
-#include <vector>
+#include <span>
+
#include "common/common_types.h"
#include "video_core/textures/texture.h"
@@ -15,28 +16,25 @@ constexpr u32 GOB_SIZE_Y = 8;
constexpr u32 GOB_SIZE_Z = 1;
constexpr u32 GOB_SIZE = GOB_SIZE_X * GOB_SIZE_Y * GOB_SIZE_Z;
-constexpr std::size_t GOB_SIZE_X_SHIFT = 6;
-constexpr std::size_t GOB_SIZE_Y_SHIFT = 3;
-constexpr std::size_t GOB_SIZE_Z_SHIFT = 0;
-constexpr std::size_t GOB_SIZE_SHIFT = GOB_SIZE_X_SHIFT + GOB_SIZE_Y_SHIFT + GOB_SIZE_Z_SHIFT;
-
-/// Unswizzles a swizzled texture without changing its format.
-void UnswizzleTexture(u8* unswizzled_data, u8* address, u32 tile_size_x, u32 tile_size_y,
- u32 bytes_per_pixel, u32 width, u32 height, u32 depth,
- u32 block_height = TICEntry::DefaultBlockHeight,
- u32 block_depth = TICEntry::DefaultBlockHeight, u32 width_spacing = 0);
-
-/// Unswizzles a swizzled texture without changing its format.
-std::vector<u8> UnswizzleTexture(u8* address, u32 tile_size_x, u32 tile_size_y, u32 bytes_per_pixel,
- u32 width, u32 height, u32 depth,
- u32 block_height = TICEntry::DefaultBlockHeight,
- u32 block_depth = TICEntry::DefaultBlockHeight,
- u32 width_spacing = 0);
-
-/// Copies texture data from a buffer and performs swizzling/unswizzling as necessary.
-void CopySwizzledData(u32 width, u32 height, u32 depth, u32 bytes_per_pixel,
- u32 out_bytes_per_pixel, u8* swizzled_data, u8* unswizzled_data,
- bool unswizzle, u32 block_height, u32 block_depth, u32 width_spacing);
+constexpr u32 GOB_SIZE_X_SHIFT = 6;
+constexpr u32 GOB_SIZE_Y_SHIFT = 3;
+constexpr u32 GOB_SIZE_Z_SHIFT = 0;
+constexpr u32 GOB_SIZE_SHIFT = GOB_SIZE_X_SHIFT + GOB_SIZE_Y_SHIFT + GOB_SIZE_Z_SHIFT;
+
+using SwizzleTable = std::array<std::array<u32, GOB_SIZE_X>, GOB_SIZE_Y>;
+
+/// Returns a z-order swizzle table
+SwizzleTable MakeSwizzleTable();
+
+/// Unswizzles a block linear texture into linear memory.
+void UnswizzleTexture(std::span<u8> output, std::span<const u8> input, u32 bytes_per_pixel,
+ u32 width, u32 height, u32 depth, u32 block_height, u32 block_depth,
+ u32 stride_alignment = 1);
+
+/// Swizzles linear memory into a block linear texture.
+void SwizzleTexture(std::span<u8> output, std::span<const u8> input, u32 bytes_per_pixel, u32 width,
+ u32 height, u32 depth, u32 block_height, u32 block_depth,
+ u32 stride_alignment = 1);
/// This function calculates the correct size of a texture depending if it's tiled or not.
std::size_t CalculateSize(bool tiled, u32 bytes_per_pixel, u32 width, u32 height, u32 depth,
diff --git a/src/video_core/textures/texture.cpp b/src/video_core/textures/texture.cpp
index 4171e3ef2..ae5621a7d 100644
--- a/src/video_core/textures/texture.cpp
+++ b/src/video_core/textures/texture.cpp
@@ -5,9 +5,13 @@
#include <algorithm>
#include <array>
+#include "common/cityhash.h"
#include "core/settings.h"
#include "video_core/textures/texture.h"
+using Tegra::Texture::TICEntry;
+using Tegra::Texture::TSCEntry;
+
namespace Tegra::Texture {
namespace {
@@ -65,7 +69,7 @@ unsigned SettingsMinimumAnisotropy() noexcept {
} // Anonymous namespace
-std::array<float, 4> TSCEntry::GetBorderColor() const noexcept {
+std::array<float, 4> TSCEntry::BorderColor() const noexcept {
if (!srgb_conversion) {
return border_color;
}
@@ -73,8 +77,16 @@ std::array<float, 4> TSCEntry::GetBorderColor() const noexcept {
SRGB_CONVERSION_LUT[srgb_border_color_b], border_color[3]};
}
-float TSCEntry::GetMaxAnisotropy() const noexcept {
+float TSCEntry::MaxAnisotropy() const noexcept {
return static_cast<float>(std::max(1U << max_anisotropy, SettingsMinimumAnisotropy()));
}
} // namespace Tegra::Texture
+
+size_t std::hash<TICEntry>::operator()(const TICEntry& tic) const noexcept {
+ return Common::CityHash64(reinterpret_cast<const char*>(&tic), sizeof tic);
+}
+
+size_t std::hash<TSCEntry>::operator()(const TSCEntry& tsc) const noexcept {
+ return Common::CityHash64(reinterpret_cast<const char*>(&tsc), sizeof tsc);
+}
diff --git a/src/video_core/textures/texture.h b/src/video_core/textures/texture.h
index 0574fef12..c1d14335e 100644
--- a/src/video_core/textures/texture.h
+++ b/src/video_core/textures/texture.h
@@ -53,27 +53,27 @@ enum class TextureFormat : u32 {
BC4 = 0x27,
BC5 = 0x28,
S8D24 = 0x29,
- X8Z24 = 0x2a,
+ X8D24 = 0x2a,
D24S8 = 0x2b,
- X4V4Z24__COV4R4V = 0x2c,
- X4V4Z24__COV8R8V = 0x2d,
- V8Z24__COV4R12V = 0x2e,
+ X4V4D24__COV4R4V = 0x2c,
+ X4V4D24__COV8R8V = 0x2d,
+ V8D24__COV4R12V = 0x2e,
D32 = 0x2f,
D32S8 = 0x30,
- X8Z24_X20V4S8__COV4R4V = 0x31,
- X8Z24_X20V4S8__COV8R8V = 0x32,
- ZF32_X20V4X8__COV4R4V = 0x33,
- ZF32_X20V4X8__COV8R8V = 0x34,
- ZF32_X20V4S8__COV4R4V = 0x35,
- ZF32_X20V4S8__COV8R8V = 0x36,
- X8Z24_X16V8S8__COV4R12V = 0x37,
- ZF32_X16V8X8__COV4R12V = 0x38,
- ZF32_X16V8S8__COV4R12V = 0x39,
+ X8D24_X20V4S8__COV4R4V = 0x31,
+ X8D24_X20V4S8__COV8R8V = 0x32,
+ D32_X20V4X8__COV4R4V = 0x33,
+ D32_X20V4X8__COV8R8V = 0x34,
+ D32_X20V4S8__COV4R4V = 0x35,
+ D32_X20V4S8__COV8R8V = 0x36,
+ X8D24_X16V8S8__COV4R12V = 0x37,
+ D32_X16V8X8__COV4R12V = 0x38,
+ D32_X16V8S8__COV4R12V = 0x39,
D16 = 0x3a,
- V8Z24__COV8R24V = 0x3b,
- X8Z24_X16V8S8__COV8R24V = 0x3c,
- ZF32_X16V8X8__COV8R24V = 0x3d,
- ZF32_X16V8S8__COV8R24V = 0x3e,
+ V8D24__COV8R24V = 0x3b,
+ X8D24_X16V8S8__COV8R24V = 0x3c,
+ D32_X16V8X8__COV8R24V = 0x3d,
+ D32_X16V8S8__COV8R24V = 0x3e,
ASTC_2D_4X4 = 0x40,
ASTC_2D_5X5 = 0x41,
ASTC_2D_6X6 = 0x42,
@@ -146,7 +146,7 @@ enum class MsaaMode : u32 {
};
union TextureHandle {
- TextureHandle(u32 raw) : raw{raw} {}
+ /* implicit */ constexpr TextureHandle(u32 raw_) : raw{raw_} {}
u32 raw;
BitField<0, 20, u32> tic_id;
@@ -155,124 +155,124 @@ union TextureHandle {
static_assert(sizeof(TextureHandle) == 4, "TextureHandle has wrong size");
struct TICEntry {
- static constexpr u32 DefaultBlockHeight = 16;
- static constexpr u32 DefaultBlockDepth = 1;
-
- union {
- u32 raw;
- BitField<0, 7, TextureFormat> format;
- BitField<7, 3, ComponentType> r_type;
- BitField<10, 3, ComponentType> g_type;
- BitField<13, 3, ComponentType> b_type;
- BitField<16, 3, ComponentType> a_type;
-
- BitField<19, 3, SwizzleSource> x_source;
- BitField<22, 3, SwizzleSource> y_source;
- BitField<25, 3, SwizzleSource> z_source;
- BitField<28, 3, SwizzleSource> w_source;
- };
- u32 address_low;
union {
- BitField<0, 16, u32> address_high;
- BitField<21, 3, TICHeaderVersion> header_version;
- };
- union {
- BitField<0, 3, u32> block_width;
- BitField<3, 3, u32> block_height;
- BitField<6, 3, u32> block_depth;
+ struct {
+ union {
+ BitField<0, 7, TextureFormat> format;
+ BitField<7, 3, ComponentType> r_type;
+ BitField<10, 3, ComponentType> g_type;
+ BitField<13, 3, ComponentType> b_type;
+ BitField<16, 3, ComponentType> a_type;
+
+ BitField<19, 3, SwizzleSource> x_source;
+ BitField<22, 3, SwizzleSource> y_source;
+ BitField<25, 3, SwizzleSource> z_source;
+ BitField<28, 3, SwizzleSource> w_source;
+ };
+ u32 address_low;
+ union {
+ BitField<0, 16, u32> address_high;
+ BitField<16, 5, u32> layer_base_3_7;
+ BitField<21, 3, TICHeaderVersion> header_version;
+ BitField<24, 1, u32> load_store_hint;
+ BitField<25, 4, u32> view_coherency_hash;
+ BitField<29, 3, u32> layer_base_8_10;
+ };
+ union {
+ BitField<0, 3, u32> block_width;
+ BitField<3, 3, u32> block_height;
+ BitField<6, 3, u32> block_depth;
- BitField<10, 3, u32> tile_width_spacing;
+ BitField<10, 3, u32> tile_width_spacing;
- // High 16 bits of the pitch value
- BitField<0, 16, u32> pitch_high;
- BitField<26, 1, u32> use_header_opt_control;
- BitField<27, 1, u32> depth_texture;
- BitField<28, 4, u32> max_mip_level;
+ // High 16 bits of the pitch value
+ BitField<0, 16, u32> pitch_high;
+ BitField<26, 1, u32> use_header_opt_control;
+ BitField<27, 1, u32> depth_texture;
+ BitField<28, 4, u32> max_mip_level;
- BitField<0, 16, u32> buffer_high_width_minus_one;
- };
- union {
- BitField<0, 16, u32> width_minus_1;
- BitField<22, 1, u32> srgb_conversion;
- BitField<23, 4, TextureType> texture_type;
- BitField<29, 3, u32> border_size;
+ BitField<0, 16, u32> buffer_high_width_minus_one;
+ };
+ union {
+ BitField<0, 16, u32> width_minus_one;
+ BitField<16, 3, u32> layer_base_0_2;
+ BitField<22, 1, u32> srgb_conversion;
+ BitField<23, 4, TextureType> texture_type;
+ BitField<29, 3, u32> border_size;
- BitField<0, 16, u32> buffer_low_width_minus_one;
- };
- union {
- BitField<0, 16, u32> height_minus_1;
- BitField<16, 14, u32> depth_minus_1;
- };
- union {
- BitField<6, 13, u32> mip_lod_bias;
- BitField<27, 3, u32> max_anisotropy;
+ BitField<0, 16, u32> buffer_low_width_minus_one;
+ };
+ union {
+ BitField<0, 16, u32> height_minus_1;
+ BitField<16, 14, u32> depth_minus_1;
+ BitField<30, 1, u32> is_sparse;
+ BitField<31, 1, u32> normalized_coords;
+ };
+ union {
+ BitField<6, 13, u32> mip_lod_bias;
+ BitField<27, 3, u32> max_anisotropy;
+ };
+ union {
+ BitField<0, 4, u32> res_min_mip_level;
+ BitField<4, 4, u32> res_max_mip_level;
+ BitField<8, 4, MsaaMode> msaa_mode;
+ BitField<12, 12, u32> min_lod_clamp;
+ };
+ };
+ std::array<u64, 4> raw;
};
- union {
- BitField<0, 4, u32> res_min_mip_level;
- BitField<4, 4, u32> res_max_mip_level;
- BitField<8, 4, MsaaMode> msaa_mode;
- BitField<12, 12, u32> min_lod_clamp;
- };
+ constexpr bool operator==(const TICEntry& rhs) const noexcept {
+ return raw == rhs.raw;
+ }
- GPUVAddr Address() const {
+ constexpr bool operator!=(const TICEntry& rhs) const noexcept {
+ return raw != rhs.raw;
+ }
+
+ constexpr GPUVAddr Address() const {
return static_cast<GPUVAddr>((static_cast<GPUVAddr>(address_high) << 32) | address_low);
}
- u32 Pitch() const {
+ constexpr u32 Pitch() const {
ASSERT(header_version == TICHeaderVersion::Pitch ||
header_version == TICHeaderVersion::PitchColorKey);
// The pitch value is 21 bits, and is 32B aligned.
return pitch_high << 5;
}
- u32 Width() const {
+ constexpr u32 Width() const {
if (header_version != TICHeaderVersion::OneDBuffer) {
- return width_minus_1 + 1;
+ return width_minus_one + 1;
}
- return ((buffer_high_width_minus_one << 16) | buffer_low_width_minus_one) + 1;
+ return (buffer_high_width_minus_one << 16 | buffer_low_width_minus_one) + 1;
}
- u32 Height() const {
+ constexpr u32 Height() const {
return height_minus_1 + 1;
}
- u32 Depth() const {
+ constexpr u32 Depth() const {
return depth_minus_1 + 1;
}
- u32 BlockWidth() const {
- ASSERT(IsTiled());
- return block_width;
- }
-
- u32 BlockHeight() const {
- ASSERT(IsTiled());
- return block_height;
- }
-
- u32 BlockDepth() const {
- ASSERT(IsTiled());
- return block_depth;
+ constexpr u32 BaseLayer() const {
+ return layer_base_0_2 | layer_base_3_7 << 3 | layer_base_8_10 << 8;
}
- bool IsTiled() const {
+ constexpr bool IsBlockLinear() const {
return header_version == TICHeaderVersion::BlockLinear ||
header_version == TICHeaderVersion::BlockLinearColorKey;
}
- bool IsLineal() const {
+ constexpr bool IsPitchLinear() const {
return header_version == TICHeaderVersion::Pitch ||
header_version == TICHeaderVersion::PitchColorKey;
}
- bool IsBuffer() const {
+ constexpr bool IsBuffer() const {
return header_version == TICHeaderVersion::OneDBuffer;
}
-
- bool IsSrgbConversionEnabled() const {
- return srgb_conversion != 0;
- }
};
static_assert(sizeof(TICEntry) == 0x20, "TICEntry has wrong size");
@@ -309,6 +309,12 @@ enum class TextureMipmapFilter : u32 {
Linear = 3,
};
+enum class SamplerReduction : u32 {
+ WeightedAverage = 0,
+ Min = 1,
+ Max = 2,
+};
+
enum class Anisotropy {
Default,
Filter2x,
@@ -333,8 +339,12 @@ struct TSCEntry {
BitField<0, 2, TextureFilter> mag_filter;
BitField<4, 2, TextureFilter> min_filter;
BitField<6, 2, TextureMipmapFilter> mipmap_filter;
+ BitField<8, 1, u32> cubemap_anisotropy;
BitField<9, 1, u32> cubemap_interface_filtering;
+ BitField<10, 2, SamplerReduction> reduction_filter;
BitField<12, 13, u32> mip_lod_bias;
+ BitField<25, 1, u32> float_coord_normalization;
+ BitField<26, 5, u32> trilin_opt;
};
union {
BitField<0, 12, u32> min_lod_clamp;
@@ -347,32 +357,45 @@ struct TSCEntry {
};
std::array<f32, 4> border_color;
};
- std::array<u8, 0x20> raw;
+ std::array<u64, 4> raw;
};
- std::array<float, 4> GetBorderColor() const noexcept;
+ constexpr bool operator==(const TSCEntry& rhs) const noexcept {
+ return raw == rhs.raw;
+ }
+
+ constexpr bool operator!=(const TSCEntry& rhs) const noexcept {
+ return raw != rhs.raw;
+ }
+
+ std::array<float, 4> BorderColor() const noexcept;
- float GetMaxAnisotropy() const noexcept;
+ float MaxAnisotropy() const noexcept;
- float GetMinLod() const {
+ float MinLod() const {
return static_cast<float>(min_lod_clamp) / 256.0f;
}
- float GetMaxLod() const {
+ float MaxLod() const {
return static_cast<float>(max_lod_clamp) / 256.0f;
}
- float GetLodBias() const {
+ float LodBias() const {
// Sign extend the 13-bit value.
- constexpr u32 mask = 1U << (13 - 1);
+ static constexpr u32 mask = 1U << (13 - 1);
return static_cast<float>(static_cast<s32>((mip_lod_bias ^ mask) - mask)) / 256.0f;
}
};
static_assert(sizeof(TSCEntry) == 0x20, "TSCEntry has wrong size");
-struct FullTextureInfo {
- TICEntry tic;
- TSCEntry tsc;
+} // namespace Tegra::Texture
+
+template <>
+struct std::hash<Tegra::Texture::TICEntry> {
+ size_t operator()(const Tegra::Texture::TICEntry& tic) const noexcept;
};
-} // namespace Tegra::Texture
+template <>
+struct std::hash<Tegra::Texture::TSCEntry> {
+ size_t operator()(const Tegra::Texture::TSCEntry& tsc) const noexcept;
+};
diff --git a/src/video_core/video_core.cpp b/src/video_core/video_core.cpp
index dd5cee4a1..53444e945 100644
--- a/src/video_core/video_core.cpp
+++ b/src/video_core/video_core.cpp
@@ -7,13 +7,9 @@
#include "common/logging/log.h"
#include "core/core.h"
#include "core/settings.h"
-#include "video_core/gpu_asynch.h"
-#include "video_core/gpu_synch.h"
#include "video_core/renderer_base.h"
#include "video_core/renderer_opengl/renderer_opengl.h"
-#ifdef HAS_VULKAN
#include "video_core/renderer_vulkan/renderer_vulkan.h"
-#endif
#include "video_core/video_core.h"
namespace {
@@ -28,11 +24,9 @@ std::unique_ptr<VideoCore::RendererBase> CreateRenderer(
case Settings::RendererBackend::OpenGL:
return std::make_unique<OpenGL::RendererOpenGL>(telemetry_session, emu_window, cpu_memory,
gpu, std::move(context));
-#ifdef HAS_VULKAN
case Settings::RendererBackend::Vulkan:
return std::make_unique<Vulkan::RendererVulkan>(telemetry_session, emu_window, cpu_memory,
gpu, std::move(context));
-#endif
default:
return nullptr;
}
@@ -43,13 +37,9 @@ std::unique_ptr<VideoCore::RendererBase> CreateRenderer(
namespace VideoCore {
std::unique_ptr<Tegra::GPU> CreateGPU(Core::Frontend::EmuWindow& emu_window, Core::System& system) {
- std::unique_ptr<Tegra::GPU> gpu;
const bool use_nvdec = Settings::values.use_nvdec_emulation.GetValue();
- if (Settings::values.use_asynchronous_gpu_emulation.GetValue()) {
- gpu = std::make_unique<VideoCommon::GPUAsynch>(system, use_nvdec);
- } else {
- gpu = std::make_unique<VideoCommon::GPUSynch>(system, use_nvdec);
- }
+ std::unique_ptr<Tegra::GPU> gpu = std::make_unique<Tegra::GPU>(
+ system, Settings::values.use_asynchronous_gpu_emulation.GetValue(), use_nvdec);
auto context = emu_window.CreateSharedContext();
const auto scope = context->Acquire();
diff --git a/src/yuzu/CMakeLists.txt b/src/yuzu/CMakeLists.txt
index b16b54032..e1bab2112 100644
--- a/src/yuzu/CMakeLists.txt
+++ b/src/yuzu/CMakeLists.txt
@@ -141,6 +141,8 @@ add_executable(yuzu
util/limitable_input_dialog.h
util/sequence_dialog/sequence_dialog.cpp
util/sequence_dialog/sequence_dialog.h
+ util/url_request_interceptor.cpp
+ util/url_request_interceptor.h
util/util.cpp
util/util.h
compatdb.cpp
@@ -217,7 +219,8 @@ target_link_libraries(yuzu PRIVATE common core input_common video_core)
target_link_libraries(yuzu PRIVATE Boost::boost glad Qt5::Widgets)
target_link_libraries(yuzu PRIVATE ${PLATFORM_LIBRARIES} Threads::Threads)
-if (ENABLE_VULKAN AND NOT WIN32)
+target_include_directories(yuzu PRIVATE ../../externals/Vulkan-Headers/include)
+if (NOT WIN32)
target_include_directories(yuzu PRIVATE ${Qt5Gui_PRIVATE_INCLUDE_DIRS})
endif()
@@ -278,8 +281,3 @@ endif()
if (NOT APPLE)
target_compile_definitions(yuzu PRIVATE HAS_OPENGL)
endif()
-
-if (ENABLE_VULKAN)
- target_include_directories(yuzu PRIVATE ../../externals/Vulkan-Headers/include)
- target_compile_definitions(yuzu PRIVATE HAS_VULKAN)
-endif()
diff --git a/src/yuzu/applets/controller.cpp b/src/yuzu/applets/controller.cpp
index 6944478f3..a15e8ca2a 100644
--- a/src/yuzu/applets/controller.cpp
+++ b/src/yuzu/applets/controller.cpp
@@ -660,8 +660,8 @@ QtControllerSelector::QtControllerSelector(GMainWindow& parent) {
QtControllerSelector::~QtControllerSelector() = default;
void QtControllerSelector::ReconfigureControllers(
- std::function<void()> callback, const Core::Frontend::ControllerParameters& parameters) const {
- this->callback = std::move(callback);
+ std::function<void()> callback_, const Core::Frontend::ControllerParameters& parameters) const {
+ callback = std::move(callback_);
emit MainWindowReconfigureControllers(parameters);
}
diff --git a/src/yuzu/applets/controller.h b/src/yuzu/applets/controller.h
index 7a421d856..3518eed56 100644
--- a/src/yuzu/applets/controller.h
+++ b/src/yuzu/applets/controller.h
@@ -147,7 +147,7 @@ public:
~QtControllerSelector() override;
void ReconfigureControllers(
- std::function<void()> callback,
+ std::function<void()> callback_,
const Core::Frontend::ControllerParameters& parameters) const override;
signals:
diff --git a/src/yuzu/applets/error.cpp b/src/yuzu/applets/error.cpp
index 08ed57355..53a993cf6 100644
--- a/src/yuzu/applets/error.cpp
+++ b/src/yuzu/applets/error.cpp
@@ -17,7 +17,7 @@ QtErrorDisplay::QtErrorDisplay(GMainWindow& parent) {
QtErrorDisplay::~QtErrorDisplay() = default;
void QtErrorDisplay::ShowError(ResultCode error, std::function<void()> finished) const {
- this->callback = std::move(finished);
+ callback = std::move(finished);
emit MainWindowDisplayError(
tr("An error has occured.\nPlease try again or contact the developer of the "
"software.\n\nError Code: %1-%2 (0x%3)")
@@ -28,7 +28,7 @@ void QtErrorDisplay::ShowError(ResultCode error, std::function<void()> finished)
void QtErrorDisplay::ShowErrorWithTimestamp(ResultCode error, std::chrono::seconds time,
std::function<void()> finished) const {
- this->callback = std::move(finished);
+ callback = std::move(finished);
const QDateTime date_time = QDateTime::fromSecsSinceEpoch(time.count());
emit MainWindowDisplayError(
@@ -44,7 +44,7 @@ void QtErrorDisplay::ShowErrorWithTimestamp(ResultCode error, std::chrono::secon
void QtErrorDisplay::ShowCustomErrorText(ResultCode error, std::string dialog_text,
std::string fullscreen_text,
std::function<void()> finished) const {
- this->callback = std::move(finished);
+ callback = std::move(finished);
emit MainWindowDisplayError(
tr("An error has occured.\nError Code: %1-%2 (0x%3)\n\n%4\n\n%5")
.arg(static_cast<u32>(error.module.Value()) + 2000, 4, 10, QChar::fromLatin1('0'))
diff --git a/src/yuzu/applets/profile_select.cpp b/src/yuzu/applets/profile_select.cpp
index c9a2f8601..4bf2bfd40 100644
--- a/src/yuzu/applets/profile_select.cpp
+++ b/src/yuzu/applets/profile_select.cpp
@@ -150,8 +150,8 @@ QtProfileSelector::QtProfileSelector(GMainWindow& parent) {
QtProfileSelector::~QtProfileSelector() = default;
void QtProfileSelector::SelectProfile(
- std::function<void(std::optional<Common::UUID>)> callback) const {
- this->callback = std::move(callback);
+ std::function<void(std::optional<Common::UUID>)> callback_) const {
+ callback = std::move(callback_);
emit MainWindowSelectProfile();
}
diff --git a/src/yuzu/applets/profile_select.h b/src/yuzu/applets/profile_select.h
index 29c33cca0..4e9037488 100644
--- a/src/yuzu/applets/profile_select.h
+++ b/src/yuzu/applets/profile_select.h
@@ -60,7 +60,7 @@ public:
explicit QtProfileSelector(GMainWindow& parent);
~QtProfileSelector() override;
- void SelectProfile(std::function<void(std::optional<Common::UUID>)> callback) const override;
+ void SelectProfile(std::function<void(std::optional<Common::UUID>)> callback_) const override;
signals:
void MainWindowSelectProfile() const;
diff --git a/src/yuzu/applets/software_keyboard.cpp b/src/yuzu/applets/software_keyboard.cpp
index af36f07c6..ab8cfd8ee 100644
--- a/src/yuzu/applets/software_keyboard.cpp
+++ b/src/yuzu/applets/software_keyboard.cpp
@@ -135,8 +135,8 @@ void QtSoftwareKeyboard::RequestText(std::function<void(std::optional<std::u16st
}
void QtSoftwareKeyboard::SendTextCheckDialog(std::u16string error_message,
- std::function<void()> finished_check) const {
- this->finished_check = std::move(finished_check);
+ std::function<void()> finished_check_) const {
+ finished_check = std::move(finished_check_);
emit MainWindowTextCheckDialog(error_message);
}
diff --git a/src/yuzu/applets/software_keyboard.h b/src/yuzu/applets/software_keyboard.h
index 44bcece75..9e1094cce 100644
--- a/src/yuzu/applets/software_keyboard.h
+++ b/src/yuzu/applets/software_keyboard.h
@@ -61,7 +61,7 @@ public:
void RequestText(std::function<void(std::optional<std::u16string>)> out,
Core::Frontend::SoftwareKeyboardParameters parameters) const override;
void SendTextCheckDialog(std::u16string error_message,
- std::function<void()> finished_check) const override;
+ std::function<void()> finished_check_) const override;
signals:
void MainWindowGetText(Core::Frontend::SoftwareKeyboardParameters parameters) const;
diff --git a/src/yuzu/applets/web_browser.cpp b/src/yuzu/applets/web_browser.cpp
index 33f1c385d..e482ba029 100644
--- a/src/yuzu/applets/web_browser.cpp
+++ b/src/yuzu/applets/web_browser.cpp
@@ -1,115 +1,414 @@
-// Copyright 2018 yuzu Emulator Project
+// Copyright 2020 yuzu Emulator Project
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
-#include <mutex>
-
+#ifdef YUZU_USE_QT_WEB_ENGINE
#include <QKeyEvent>
-#include "core/hle/lock.h"
+#include <QWebEngineProfile>
+#include <QWebEngineScript>
+#include <QWebEngineScriptCollection>
+#include <QWebEngineSettings>
+#include <QWebEngineUrlScheme>
+#endif
+
+#include "common/file_util.h"
+#include "core/core.h"
+#include "core/frontend/input_interpreter.h"
+#include "input_common/keyboard.h"
+#include "input_common/main.h"
#include "yuzu/applets/web_browser.h"
+#include "yuzu/applets/web_browser_scripts.h"
#include "yuzu/main.h"
+#include "yuzu/util/url_request_interceptor.h"
#ifdef YUZU_USE_QT_WEB_ENGINE
-constexpr char NX_SHIM_INJECT_SCRIPT[] = R"(
- window.nx = {};
- window.nx.playReport = {};
- window.nx.playReport.setCounterSetIdentifier = function () {
- console.log("nx.playReport.setCounterSetIdentifier called - unimplemented");
- };
+namespace {
- window.nx.playReport.incrementCounter = function () {
- console.log("nx.playReport.incrementCounter called - unimplemented");
- };
+constexpr int HIDButtonToKey(HIDButton button) {
+ switch (button) {
+ case HIDButton::DLeft:
+ case HIDButton::LStickLeft:
+ return Qt::Key_Left;
+ case HIDButton::DUp:
+ case HIDButton::LStickUp:
+ return Qt::Key_Up;
+ case HIDButton::DRight:
+ case HIDButton::LStickRight:
+ return Qt::Key_Right;
+ case HIDButton::DDown:
+ case HIDButton::LStickDown:
+ return Qt::Key_Down;
+ default:
+ return 0;
+ }
+}
+
+} // Anonymous namespace
+
+QtNXWebEngineView::QtNXWebEngineView(QWidget* parent, Core::System& system,
+ InputCommon::InputSubsystem* input_subsystem_)
+ : QWebEngineView(parent), input_subsystem{input_subsystem_},
+ url_interceptor(std::make_unique<UrlRequestInterceptor>()),
+ input_interpreter(std::make_unique<InputInterpreter>(system)),
+ default_profile{QWebEngineProfile::defaultProfile()},
+ global_settings{QWebEngineSettings::globalSettings()} {
+ QWebEngineScript gamepad;
+ QWebEngineScript window_nx;
+
+ gamepad.setName(QStringLiteral("gamepad_script.js"));
+ window_nx.setName(QStringLiteral("window_nx_script.js"));
+
+ gamepad.setSourceCode(QString::fromStdString(GAMEPAD_SCRIPT));
+ window_nx.setSourceCode(QString::fromStdString(WINDOW_NX_SCRIPT));
+
+ gamepad.setInjectionPoint(QWebEngineScript::DocumentCreation);
+ window_nx.setInjectionPoint(QWebEngineScript::DocumentCreation);
+
+ gamepad.setWorldId(QWebEngineScript::MainWorld);
+ window_nx.setWorldId(QWebEngineScript::MainWorld);
+
+ gamepad.setRunsOnSubFrames(true);
+ window_nx.setRunsOnSubFrames(true);
+
+ default_profile->scripts()->insert(gamepad);
+ default_profile->scripts()->insert(window_nx);
+
+ default_profile->setRequestInterceptor(url_interceptor.get());
+
+ global_settings->setAttribute(QWebEngineSettings::LocalContentCanAccessRemoteUrls, true);
+ global_settings->setAttribute(QWebEngineSettings::FullScreenSupportEnabled, true);
+ global_settings->setAttribute(QWebEngineSettings::AllowRunningInsecureContent, true);
+ global_settings->setAttribute(QWebEngineSettings::FocusOnNavigationEnabled, true);
+ global_settings->setAttribute(QWebEngineSettings::AllowWindowActivationFromJavaScript, true);
+ global_settings->setAttribute(QWebEngineSettings::ShowScrollBars, false);
+
+ global_settings->setFontFamily(QWebEngineSettings::StandardFont, QStringLiteral("Roboto"));
+
+ connect(
+ page(), &QWebEnginePage::windowCloseRequested, page(),
+ [this] {
+ if (page()->url() == url_interceptor->GetRequestedURL()) {
+ SetFinished(true);
+ SetExitReason(Service::AM::Applets::WebExitReason::WindowClosed);
+ }
+ },
+ Qt::QueuedConnection);
+}
+
+QtNXWebEngineView::~QtNXWebEngineView() {
+ SetFinished(true);
+ StopInputThread();
+}
+
+void QtNXWebEngineView::LoadLocalWebPage(std::string_view main_url,
+ std::string_view additional_args) {
+ is_local = true;
+
+ LoadExtractedFonts();
+ SetUserAgent(UserAgent::WebApplet);
+ SetFinished(false);
+ SetExitReason(Service::AM::Applets::WebExitReason::EndButtonPressed);
+ SetLastURL("http://localhost/");
+ StartInputThread();
+
+ load(QUrl(QUrl::fromLocalFile(QString::fromStdString(std::string(main_url))).toString() +
+ QString::fromStdString(std::string(additional_args))));
+}
+
+void QtNXWebEngineView::LoadExternalWebPage(std::string_view main_url,
+ std::string_view additional_args) {
+ is_local = false;
+
+ SetUserAgent(UserAgent::WebApplet);
+ SetFinished(false);
+ SetExitReason(Service::AM::Applets::WebExitReason::EndButtonPressed);
+ SetLastURL("http://localhost/");
+ StartInputThread();
+
+ load(QUrl(QString::fromStdString(std::string(main_url)) +
+ QString::fromStdString(std::string(additional_args))));
+}
+
+void QtNXWebEngineView::SetUserAgent(UserAgent user_agent) {
+ const QString user_agent_str = [user_agent] {
+ switch (user_agent) {
+ case UserAgent::WebApplet:
+ default:
+ return QStringLiteral("WebApplet");
+ case UserAgent::ShopN:
+ return QStringLiteral("ShopN");
+ case UserAgent::LoginApplet:
+ return QStringLiteral("LoginApplet");
+ case UserAgent::ShareApplet:
+ return QStringLiteral("ShareApplet");
+ case UserAgent::LobbyApplet:
+ return QStringLiteral("LobbyApplet");
+ case UserAgent::WifiWebAuthApplet:
+ return QStringLiteral("WifiWebAuthApplet");
+ }
+ }();
+
+ QWebEngineProfile::defaultProfile()->setHttpUserAgent(
+ QStringLiteral("Mozilla/5.0 (Nintendo Switch; %1) AppleWebKit/606.4 "
+ "(KHTML, like Gecko) NF/6.0.1.15.4 NintendoBrowser/5.1.0.20389")
+ .arg(user_agent_str));
+}
+
+bool QtNXWebEngineView::IsFinished() const {
+ return finished;
+}
+
+void QtNXWebEngineView::SetFinished(bool finished_) {
+ finished = finished_;
+}
+
+Service::AM::Applets::WebExitReason QtNXWebEngineView::GetExitReason() const {
+ return exit_reason;
+}
+
+void QtNXWebEngineView::SetExitReason(Service::AM::Applets::WebExitReason exit_reason_) {
+ exit_reason = exit_reason_;
+}
+
+const std::string& QtNXWebEngineView::GetLastURL() const {
+ return last_url;
+}
+
+void QtNXWebEngineView::SetLastURL(std::string last_url_) {
+ last_url = std::move(last_url_);
+}
+
+QString QtNXWebEngineView::GetCurrentURL() const {
+ return url_interceptor->GetRequestedURL().toString();
+}
+
+void QtNXWebEngineView::hide() {
+ SetFinished(true);
+ StopInputThread();
- window.nx.footer = {};
- window.nx.footer.unsetAssign = function () {
- console.log("nx.footer.unsetAssign called - unimplemented");
+ QWidget::hide();
+}
+
+void QtNXWebEngineView::keyPressEvent(QKeyEvent* event) {
+ if (is_local) {
+ input_subsystem->GetKeyboard()->PressKey(event->key());
+ }
+}
+
+void QtNXWebEngineView::keyReleaseEvent(QKeyEvent* event) {
+ if (is_local) {
+ input_subsystem->GetKeyboard()->ReleaseKey(event->key());
+ }
+}
+
+template <HIDButton... T>
+void QtNXWebEngineView::HandleWindowFooterButtonPressedOnce() {
+ const auto f = [this](HIDButton button) {
+ if (input_interpreter->IsButtonPressedOnce(button)) {
+ page()->runJavaScript(
+ QStringLiteral("yuzu_key_callbacks[%1] == null;").arg(static_cast<u8>(button)),
+ [&](const QVariant& variant) {
+ if (variant.toBool()) {
+ switch (button) {
+ case HIDButton::A:
+ SendMultipleKeyPressEvents<Qt::Key_A, Qt::Key_Space, Qt::Key_Return>();
+ break;
+ case HIDButton::B:
+ SendKeyPressEvent(Qt::Key_B);
+ break;
+ case HIDButton::X:
+ SendKeyPressEvent(Qt::Key_X);
+ break;
+ case HIDButton::Y:
+ SendKeyPressEvent(Qt::Key_Y);
+ break;
+ default:
+ break;
+ }
+ }
+ });
+
+ page()->runJavaScript(
+ QStringLiteral("if (yuzu_key_callbacks[%1] != null) { yuzu_key_callbacks[%1](); }")
+ .arg(static_cast<u8>(button)));
+ }
};
- var yuzu_key_callbacks = [];
- window.nx.footer.setAssign = function(key, discard1, func, discard2) {
- switch (key) {
- case 'A':
- yuzu_key_callbacks[0] = func;
- break;
- case 'B':
- yuzu_key_callbacks[1] = func;
- break;
- case 'X':
- yuzu_key_callbacks[2] = func;
- break;
- case 'Y':
- yuzu_key_callbacks[3] = func;
- break;
- case 'L':
- yuzu_key_callbacks[6] = func;
- break;
- case 'R':
- yuzu_key_callbacks[7] = func;
- break;
+ (f(T), ...);
+}
+
+template <HIDButton... T>
+void QtNXWebEngineView::HandleWindowKeyButtonPressedOnce() {
+ const auto f = [this](HIDButton button) {
+ if (input_interpreter->IsButtonPressedOnce(button)) {
+ SendKeyPressEvent(HIDButtonToKey(button));
}
};
- var applet_done = false;
- window.nx.endApplet = function() {
- applet_done = true;
+ (f(T), ...);
+}
+
+template <HIDButton... T>
+void QtNXWebEngineView::HandleWindowKeyButtonHold() {
+ const auto f = [this](HIDButton button) {
+ if (input_interpreter->IsButtonHeld(button)) {
+ SendKeyPressEvent(HIDButtonToKey(button));
+ }
};
- window.onkeypress = function(e) { if (e.keyCode === 13) { applet_done = true; } };
-)";
+ (f(T), ...);
+}
+
+void QtNXWebEngineView::SendKeyPressEvent(int key) {
+ if (key == 0) {
+ return;
+ }
+
+ QCoreApplication::postEvent(focusProxy(),
+ new QKeyEvent(QKeyEvent::KeyPress, key, Qt::NoModifier));
+ QCoreApplication::postEvent(focusProxy(),
+ new QKeyEvent(QKeyEvent::KeyRelease, key, Qt::NoModifier));
+}
+
+void QtNXWebEngineView::StartInputThread() {
+ if (input_thread_running) {
+ return;
+ }
+
+ input_thread_running = true;
+ input_thread = std::thread(&QtNXWebEngineView::InputThread, this);
+}
+
+void QtNXWebEngineView::StopInputThread() {
+ if (is_local) {
+ QWidget::releaseKeyboard();
+ }
-QString GetNXShimInjectionScript() {
- return QString::fromStdString(NX_SHIM_INJECT_SCRIPT);
+ input_thread_running = false;
+ if (input_thread.joinable()) {
+ input_thread.join();
+ }
}
-NXInputWebEngineView::NXInputWebEngineView(QWidget* parent) : QWebEngineView(parent) {}
+void QtNXWebEngineView::InputThread() {
+ // Wait for 1 second before allowing any inputs to be processed.
+ std::this_thread::sleep_for(std::chrono::seconds(1));
+
+ if (is_local) {
+ QWidget::grabKeyboard();
+ }
+
+ while (input_thread_running) {
+ input_interpreter->PollInput();
+
+ HandleWindowFooterButtonPressedOnce<HIDButton::A, HIDButton::B, HIDButton::X, HIDButton::Y,
+ HIDButton::L, HIDButton::R>();
+
+ HandleWindowKeyButtonPressedOnce<HIDButton::DLeft, HIDButton::DUp, HIDButton::DRight,
+ HIDButton::DDown, HIDButton::LStickLeft,
+ HIDButton::LStickUp, HIDButton::LStickRight,
+ HIDButton::LStickDown>();
-void NXInputWebEngineView::keyPressEvent(QKeyEvent* event) {
- parent()->event(event);
+ HandleWindowKeyButtonHold<HIDButton::DLeft, HIDButton::DUp, HIDButton::DRight,
+ HIDButton::DDown, HIDButton::LStickLeft, HIDButton::LStickUp,
+ HIDButton::LStickRight, HIDButton::LStickDown>();
+
+ std::this_thread::sleep_for(std::chrono::milliseconds(50));
+ }
}
-void NXInputWebEngineView::keyReleaseEvent(QKeyEvent* event) {
- parent()->event(event);
+void QtNXWebEngineView::LoadExtractedFonts() {
+ QWebEngineScript nx_font_css;
+ QWebEngineScript load_nx_font;
+
+ const QString fonts_dir = QString::fromStdString(Common::FS::SanitizePath(
+ fmt::format("{}/fonts", Common::FS::GetUserPath(Common::FS::UserPath::CacheDir))));
+
+ nx_font_css.setName(QStringLiteral("nx_font_css.js"));
+ load_nx_font.setName(QStringLiteral("load_nx_font.js"));
+
+ nx_font_css.setSourceCode(
+ QString::fromStdString(NX_FONT_CSS)
+ .arg(fonts_dir + QStringLiteral("/FontStandard.ttf"))
+ .arg(fonts_dir + QStringLiteral("/FontChineseSimplified.ttf"))
+ .arg(fonts_dir + QStringLiteral("/FontExtendedChineseSimplified.ttf"))
+ .arg(fonts_dir + QStringLiteral("/FontChineseTraditional.ttf"))
+ .arg(fonts_dir + QStringLiteral("/FontKorean.ttf"))
+ .arg(fonts_dir + QStringLiteral("/FontNintendoExtended.ttf"))
+ .arg(fonts_dir + QStringLiteral("/FontNintendoExtended2.ttf")));
+ load_nx_font.setSourceCode(QString::fromStdString(LOAD_NX_FONT));
+
+ nx_font_css.setInjectionPoint(QWebEngineScript::DocumentReady);
+ load_nx_font.setInjectionPoint(QWebEngineScript::Deferred);
+
+ nx_font_css.setWorldId(QWebEngineScript::MainWorld);
+ load_nx_font.setWorldId(QWebEngineScript::MainWorld);
+
+ nx_font_css.setRunsOnSubFrames(true);
+ load_nx_font.setRunsOnSubFrames(true);
+
+ default_profile->scripts()->insert(nx_font_css);
+ default_profile->scripts()->insert(load_nx_font);
+
+ connect(
+ url_interceptor.get(), &UrlRequestInterceptor::FrameChanged, url_interceptor.get(),
+ [this] {
+ std::this_thread::sleep_for(std::chrono::milliseconds(50));
+ page()->runJavaScript(QString::fromStdString(LOAD_NX_FONT));
+ },
+ Qt::QueuedConnection);
}
#endif
QtWebBrowser::QtWebBrowser(GMainWindow& main_window) {
- connect(this, &QtWebBrowser::MainWindowOpenPage, &main_window, &GMainWindow::WebBrowserOpenPage,
- Qt::QueuedConnection);
- connect(&main_window, &GMainWindow::WebBrowserUnpackRomFS, this,
- &QtWebBrowser::MainWindowUnpackRomFS, Qt::QueuedConnection);
- connect(&main_window, &GMainWindow::WebBrowserFinishedBrowsing, this,
- &QtWebBrowser::MainWindowFinishedBrowsing, Qt::QueuedConnection);
+ connect(this, &QtWebBrowser::MainWindowOpenWebPage, &main_window,
+ &GMainWindow::WebBrowserOpenWebPage, Qt::QueuedConnection);
+ connect(&main_window, &GMainWindow::WebBrowserExtractOfflineRomFS, this,
+ &QtWebBrowser::MainWindowExtractOfflineRomFS, Qt::QueuedConnection);
+ connect(&main_window, &GMainWindow::WebBrowserClosed, this,
+ &QtWebBrowser::MainWindowWebBrowserClosed, Qt::QueuedConnection);
}
QtWebBrowser::~QtWebBrowser() = default;
-void QtWebBrowser::OpenPageLocal(std::string_view url, std::function<void()> unpack_romfs_callback,
- std::function<void()> finished_callback) {
- this->unpack_romfs_callback = std::move(unpack_romfs_callback);
- this->finished_callback = std::move(finished_callback);
+void QtWebBrowser::OpenLocalWebPage(
+ std::string_view local_url, std::function<void()> extract_romfs_callback_,
+ std::function<void(Service::AM::Applets::WebExitReason, std::string)> callback_) const {
+ extract_romfs_callback = std::move(extract_romfs_callback_);
+ callback = std::move(callback_);
+
+ const auto index = local_url.find('?');
+
+ if (index == std::string::npos) {
+ emit MainWindowOpenWebPage(local_url, "", true);
+ } else {
+ emit MainWindowOpenWebPage(local_url.substr(0, index), local_url.substr(index), true);
+ }
+}
+
+void QtWebBrowser::OpenExternalWebPage(
+ std::string_view external_url,
+ std::function<void(Service::AM::Applets::WebExitReason, std::string)> callback_) const {
+ callback = std::move(callback_);
+
+ const auto index = external_url.find('?');
- const auto index = url.find('?');
if (index == std::string::npos) {
- emit MainWindowOpenPage(url, "");
+ emit MainWindowOpenWebPage(external_url, "", false);
} else {
- const auto front = url.substr(0, index);
- const auto back = url.substr(index);
- emit MainWindowOpenPage(front, back);
+ emit MainWindowOpenWebPage(external_url.substr(0, index), external_url.substr(index),
+ false);
}
}
-void QtWebBrowser::MainWindowUnpackRomFS() {
- // Acquire the HLE mutex
- std::lock_guard lock{HLE::g_hle_lock};
- unpack_romfs_callback();
+void QtWebBrowser::MainWindowExtractOfflineRomFS() {
+ extract_romfs_callback();
}
-void QtWebBrowser::MainWindowFinishedBrowsing() {
- // Acquire the HLE mutex
- std::lock_guard lock{HLE::g_hle_lock};
- finished_callback();
+void QtWebBrowser::MainWindowWebBrowserClosed(Service::AM::Applets::WebExitReason exit_reason,
+ std::string last_url) {
+ callback(exit_reason, last_url);
}
diff --git a/src/yuzu/applets/web_browser.h b/src/yuzu/applets/web_browser.h
index b38437e46..47f960d69 100644
--- a/src/yuzu/applets/web_browser.h
+++ b/src/yuzu/applets/web_browser.h
@@ -1,10 +1,13 @@
-// Copyright 2018 yuzu Emulator Project
+// Copyright 2020 yuzu Emulator Project
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
#pragma once
-#include <functional>
+#include <atomic>
+#include <memory>
+#include <thread>
+
#include <QObject>
#ifdef YUZU_USE_QT_WEB_ENGINE
@@ -13,19 +16,172 @@
#include "core/frontend/applets/web_browser.h"
+enum class HIDButton : u8;
+
class GMainWindow;
+class InputInterpreter;
+class UrlRequestInterceptor;
+
+namespace Core {
+class System;
+}
+
+namespace InputCommon {
+class InputSubsystem;
+}
#ifdef YUZU_USE_QT_WEB_ENGINE
-QString GetNXShimInjectionScript();
+enum class UserAgent {
+ WebApplet,
+ ShopN,
+ LoginApplet,
+ ShareApplet,
+ LobbyApplet,
+ WifiWebAuthApplet,
+};
+
+class QWebEngineProfile;
+class QWebEngineSettings;
+
+class QtNXWebEngineView : public QWebEngineView {
+ Q_OBJECT
-class NXInputWebEngineView : public QWebEngineView {
public:
- explicit NXInputWebEngineView(QWidget* parent = nullptr);
+ explicit QtNXWebEngineView(QWidget* parent, Core::System& system,
+ InputCommon::InputSubsystem* input_subsystem_);
+ ~QtNXWebEngineView() override;
+
+ /**
+ * Loads a HTML document that exists locally. Cannot be used to load external websites.
+ *
+ * @param main_url The url to the file.
+ * @param additional_args Additional arguments appended to the main url.
+ */
+ void LoadLocalWebPage(std::string_view main_url, std::string_view additional_args);
+
+ /**
+ * Loads an external website. Cannot be used to load local urls.
+ *
+ * @param main_url The url to the website.
+ * @param additional_args Additional arguments appended to the main url.
+ */
+ void LoadExternalWebPage(std::string_view main_url, std::string_view additional_args);
+
+ /**
+ * Sets the background color of the web page.
+ *
+ * @param color The color to set.
+ */
+ void SetBackgroundColor(QColor color);
+
+ /**
+ * Sets the user agent of the web browser.
+ *
+ * @param user_agent The user agent enum.
+ */
+ void SetUserAgent(UserAgent user_agent);
+
+ [[nodiscard]] bool IsFinished() const;
+ void SetFinished(bool finished_);
+
+ [[nodiscard]] Service::AM::Applets::WebExitReason GetExitReason() const;
+ void SetExitReason(Service::AM::Applets::WebExitReason exit_reason_);
+
+ [[nodiscard]] const std::string& GetLastURL() const;
+ void SetLastURL(std::string last_url_);
+
+ /**
+ * This gets the current URL that has been requested by the webpage.
+ * This only applies to the main frame. Sub frames and other resources are ignored.
+ *
+ * @return Currently requested URL
+ */
+ [[nodiscard]] QString GetCurrentURL() const;
+
+public slots:
+ void hide();
protected:
void keyPressEvent(QKeyEvent* event) override;
void keyReleaseEvent(QKeyEvent* event) override;
+
+private:
+ /**
+ * Handles button presses to execute functions assigned in yuzu_key_callbacks.
+ * yuzu_key_callbacks contains specialized functions for the buttons in the window footer
+ * that can be overriden by games to achieve desired functionality.
+ *
+ * @tparam HIDButton The list of buttons contained in yuzu_key_callbacks
+ */
+ template <HIDButton... T>
+ void HandleWindowFooterButtonPressedOnce();
+
+ /**
+ * Handles button presses and converts them into keyboard input.
+ * This should only be used to convert D-Pad or Analog Stick input into arrow keys.
+ *
+ * @tparam HIDButton The list of buttons that can be converted into keyboard input.
+ */
+ template <HIDButton... T>
+ void HandleWindowKeyButtonPressedOnce();
+
+ /**
+ * Handles button holds and converts them into keyboard input.
+ * This should only be used to convert D-Pad or Analog Stick input into arrow keys.
+ *
+ * @tparam HIDButton The list of buttons that can be converted into keyboard input.
+ */
+ template <HIDButton... T>
+ void HandleWindowKeyButtonHold();
+
+ /**
+ * Sends a key press event to QWebEngineView.
+ *
+ * @param key Qt key code.
+ */
+ void SendKeyPressEvent(int key);
+
+ /**
+ * Sends multiple key press events to QWebEngineView.
+ *
+ * @tparam int Qt key code.
+ */
+ template <int... T>
+ void SendMultipleKeyPressEvents() {
+ (SendKeyPressEvent(T), ...);
+ }
+
+ void StartInputThread();
+ void StopInputThread();
+
+ /// The thread where input is being polled and processed.
+ void InputThread();
+
+ /// Loads the extracted fonts using JavaScript.
+ void LoadExtractedFonts();
+
+ InputCommon::InputSubsystem* input_subsystem;
+
+ std::unique_ptr<UrlRequestInterceptor> url_interceptor;
+
+ std::unique_ptr<InputInterpreter> input_interpreter;
+
+ std::thread input_thread;
+
+ std::atomic<bool> input_thread_running{};
+
+ std::atomic<bool> finished{};
+
+ Service::AM::Applets::WebExitReason exit_reason{
+ Service::AM::Applets::WebExitReason::EndButtonPressed};
+
+ std::string last_url{"http://localhost/"};
+
+ bool is_local{};
+
+ QWebEngineProfile* default_profile;
+ QWebEngineSettings* global_settings;
};
#endif
@@ -34,19 +190,28 @@ class QtWebBrowser final : public QObject, public Core::Frontend::WebBrowserAppl
Q_OBJECT
public:
- explicit QtWebBrowser(GMainWindow& main_window);
+ explicit QtWebBrowser(GMainWindow& parent);
~QtWebBrowser() override;
- void OpenPageLocal(std::string_view url, std::function<void()> unpack_romfs_callback,
- std::function<void()> finished_callback) override;
+ void OpenLocalWebPage(std::string_view local_url, std::function<void()> extract_romfs_callback_,
+ std::function<void(Service::AM::Applets::WebExitReason, std::string)>
+ callback_) const override;
+
+ void OpenExternalWebPage(std::string_view external_url,
+ std::function<void(Service::AM::Applets::WebExitReason, std::string)>
+ callback_) const override;
signals:
- void MainWindowOpenPage(std::string_view filename, std::string_view additional_args) const;
+ void MainWindowOpenWebPage(std::string_view main_url, std::string_view additional_args,
+ bool is_local) const;
private:
- void MainWindowUnpackRomFS();
- void MainWindowFinishedBrowsing();
+ void MainWindowExtractOfflineRomFS();
+
+ void MainWindowWebBrowserClosed(Service::AM::Applets::WebExitReason exit_reason,
+ std::string last_url);
+
+ mutable std::function<void()> extract_romfs_callback;
- std::function<void()> unpack_romfs_callback;
- std::function<void()> finished_callback;
+ mutable std::function<void(Service::AM::Applets::WebExitReason, std::string)> callback;
};
diff --git a/src/yuzu/applets/web_browser_scripts.h b/src/yuzu/applets/web_browser_scripts.h
new file mode 100644
index 000000000..992837a85
--- /dev/null
+++ b/src/yuzu/applets/web_browser_scripts.h
@@ -0,0 +1,193 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+constexpr char NX_FONT_CSS[] = R"(
+(function() {
+ css = document.createElement('style');
+ css.type = 'text/css';
+ css.id = 'nx_font';
+ css.innerText = `
+/* FontStandard */
+@font-face {
+ font-family: 'FontStandard';
+ src: url('%1') format('truetype');
+}
+
+/* FontChineseSimplified */
+@font-face {
+ font-family: 'FontChineseSimplified';
+ src: url('%2') format('truetype');
+}
+
+/* FontExtendedChineseSimplified */
+@font-face {
+ font-family: 'FontExtendedChineseSimplified';
+ src: url('%3') format('truetype');
+}
+
+/* FontChineseTraditional */
+@font-face {
+ font-family: 'FontChineseTraditional';
+ src: url('%4') format('truetype');
+}
+
+/* FontKorean */
+@font-face {
+ font-family: 'FontKorean';
+ src: url('%5') format('truetype');
+}
+
+/* FontNintendoExtended */
+@font-face {
+ font-family: 'NintendoExt003';
+ src: url('%6') format('truetype');
+}
+
+/* FontNintendoExtended2 */
+@font-face {
+ font-family: 'NintendoExt003';
+ src: url('%7') format('truetype');
+}
+`;
+
+ document.head.appendChild(css);
+})();
+)";
+
+constexpr char LOAD_NX_FONT[] = R"(
+(function() {
+ var elements = document.querySelectorAll("*");
+
+ for (var i = 0; i < elements.length; i++) {
+ var style = window.getComputedStyle(elements[i], null);
+ if (style.fontFamily.includes("Arial") || style.fontFamily.includes("Calibri") ||
+ style.fontFamily.includes("Century") || style.fontFamily.includes("Times New Roman")) {
+ elements[i].style.fontFamily = "FontStandard, FontChineseSimplified, FontExtendedChineseSimplified, FontChineseTraditional, FontKorean, NintendoExt003";
+ } else {
+ elements[i].style.fontFamily = style.fontFamily + ", FontStandard, FontChineseSimplified, FontExtendedChineseSimplified, FontChineseTraditional, FontKorean, NintendoExt003";
+ }
+ }
+})();
+)";
+
+constexpr char GAMEPAD_SCRIPT[] = R"(
+window.addEventListener("gamepadconnected", function(e) {
+ console.log("Gamepad connected at index %d: %s. %d buttons, %d axes.",
+ e.gamepad.index, e.gamepad.id, e.gamepad.buttons.length, e.gamepad.axes.length);
+});
+
+window.addEventListener("gamepaddisconnected", function(e) {
+ console.log("Gamepad disconnected from index %d: %s", e.gamepad.index, e.gamepad.id);
+});
+)";
+
+constexpr char WINDOW_NX_SCRIPT[] = R"(
+var end_applet = false;
+var yuzu_key_callbacks = [];
+
+(function() {
+ class WindowNX {
+ constructor() {
+ yuzu_key_callbacks[1] = function() { window.history.back(); };
+ yuzu_key_callbacks[2] = function() { window.nx.endApplet(); };
+ }
+
+ addEventListener(type, listener, options) {
+ console.log("nx.addEventListener called, type=%s", type);
+
+ window.addEventListener(type, listener, options);
+ }
+
+ endApplet() {
+ console.log("nx.endApplet called");
+
+ end_applet = true;
+ }
+
+ playSystemSe(system_se) {
+ console.log("nx.playSystemSe is not implemented, system_se=%s", system_se);
+ }
+
+ sendMessage(message) {
+ console.log("nx.sendMessage is not implemented, message=%s", message);
+ }
+
+ setCursorScrollSpeed(scroll_speed) {
+ console.log("nx.setCursorScrollSpeed is not implemented, scroll_speed=%d", scroll_speed);
+ }
+ }
+
+ class WindowNXFooter {
+ setAssign(key, label, func, option) {
+ console.log("nx.footer.setAssign called, key=%s", key);
+
+ switch (key) {
+ case "A":
+ yuzu_key_callbacks[0] = func;
+ break;
+ case "B":
+ yuzu_key_callbacks[1] = func;
+ break;
+ case "X":
+ yuzu_key_callbacks[2] = func;
+ break;
+ case "Y":
+ yuzu_key_callbacks[3] = func;
+ break;
+ case "L":
+ yuzu_key_callbacks[6] = func;
+ break;
+ case "R":
+ yuzu_key_callbacks[7] = func;
+ break;
+ }
+ }
+
+ setFixed(kind) {
+ console.log("nx.footer.setFixed is not implemented, kind=%s", kind);
+ }
+
+ unsetAssign(key) {
+ console.log("nx.footer.unsetAssign called, key=%s", key);
+
+ switch (key) {
+ case "A":
+ yuzu_key_callbacks[0] = function() {};
+ break;
+ case "B":
+ yuzu_key_callbacks[1] = function() {};
+ break;
+ case "X":
+ yuzu_key_callbacks[2] = function() {};
+ break;
+ case "Y":
+ yuzu_key_callbacks[3] = function() {};
+ break;
+ case "L":
+ yuzu_key_callbacks[6] = function() {};
+ break;
+ case "R":
+ yuzu_key_callbacks[7] = function() {};
+ break;
+ }
+ }
+ }
+
+ class WindowNXPlayReport {
+ incrementCounter(counter_id) {
+ console.log("nx.playReport.incrementCounter is not implemented, counter_id=%d", counter_id);
+ }
+
+ setCounterSetIdentifier(counter_id) {
+ console.log("nx.playReport.setCounterSetIdentifier is not implemented, counter_id=%d", counter_id);
+ }
+ }
+
+ window.nx = new WindowNX();
+ window.nx.footer = new WindowNXFooter();
+ window.nx.playReport = new WindowNXPlayReport();
+})();
+)";
diff --git a/src/yuzu/bootmanager.cpp b/src/yuzu/bootmanager.cpp
index 489104d5f..e124836b5 100644
--- a/src/yuzu/bootmanager.cpp
+++ b/src/yuzu/bootmanager.cpp
@@ -19,7 +19,7 @@
#include <QOpenGLContext>
#endif
-#if !defined(WIN32) && HAS_VULKAN
+#if !defined(WIN32)
#include <qpa/qplatformnativeinterface.h>
#endif
@@ -241,14 +241,12 @@ private:
std::unique_ptr<Core::Frontend::GraphicsContext> context;
};
-#ifdef HAS_VULKAN
class VulkanRenderWidget : public RenderWidget {
public:
explicit VulkanRenderWidget(GRenderWindow* parent) : RenderWidget(parent) {
windowHandle()->setSurfaceType(QWindow::VulkanSurface);
}
};
-#endif
static Core::Frontend::WindowSystemType GetWindowSystemType() {
// Determine WSI type based on Qt platform.
@@ -268,7 +266,6 @@ static Core::Frontend::EmuWindow::WindowSystemInfo GetWindowSystemInfo(QWindow*
Core::Frontend::EmuWindow::WindowSystemInfo wsi;
wsi.type = GetWindowSystemType();
-#ifdef HAS_VULKAN
// Our Win32 Qt external doesn't have the private API.
#if defined(WIN32) || defined(__APPLE__)
wsi.render_surface = window ? reinterpret_cast<void*>(window->winId()) : nullptr;
@@ -281,7 +278,6 @@ static Core::Frontend::EmuWindow::WindowSystemInfo GetWindowSystemInfo(QWindow*
wsi.render_surface = window ? reinterpret_cast<void*>(window->winId()) : nullptr;
#endif
wsi.render_surface_scale = window ? static_cast<float>(window->devicePixelRatio()) : 1.0f;
-#endif
return wsi;
}
@@ -569,6 +565,10 @@ void GRenderWindow::CaptureScreenshot(u32 res_scale, const QString& screenshot_p
layout);
}
+bool GRenderWindow::IsLoadingComplete() const {
+ return first_frame;
+}
+
void GRenderWindow::OnMinimalClientAreaChangeRequest(std::pair<u32, u32> minimal_size) {
setMinimumSize(minimal_size.first, minimal_size.second);
}
@@ -594,18 +594,12 @@ bool GRenderWindow::InitializeOpenGL() {
}
bool GRenderWindow::InitializeVulkan() {
-#ifdef HAS_VULKAN
auto child = new VulkanRenderWidget(this);
child_widget = child;
child_widget->windowHandle()->create();
main_context = std::make_unique<DummyContext>();
return true;
-#else
- QMessageBox::critical(this, tr("Vulkan not available!"),
- tr("yuzu has not been compiled with Vulkan support."));
- return false;
-#endif
}
bool GRenderWindow::LoadOpenGL() {
diff --git a/src/yuzu/bootmanager.h b/src/yuzu/bootmanager.h
index a6d788d40..ebe5cb965 100644
--- a/src/yuzu/bootmanager.h
+++ b/src/yuzu/bootmanager.h
@@ -162,6 +162,8 @@ public:
/// Destroy the previous run's child_widget which should also destroy the child_window
void ReleaseRenderTarget();
+ bool IsLoadingComplete() const;
+
void CaptureScreenshot(u32 res_scale, const QString& screenshot_path);
std::pair<u32, u32> ScaleTouch(const QPointF& pos) const;
diff --git a/src/yuzu/configuration/config.cpp b/src/yuzu/configuration/config.cpp
index fcc38b3af..34c2a5f8b 100644
--- a/src/yuzu/configuration/config.cpp
+++ b/src/yuzu/configuration/config.cpp
@@ -511,6 +511,9 @@ void Config::ReadControlValues() {
ReadTouchscreenValues();
ReadMotionTouchValues();
+ Settings::values.emulate_analog_keyboard =
+ ReadSetting(QStringLiteral("emulate_analog_keyboard"), false).toBool();
+
ReadSettingGlobal(Settings::values.use_docked_mode, QStringLiteral("use_docked_mode"), false);
ReadSettingGlobal(Settings::values.vibration_enabled, QStringLiteral("vibration_enabled"),
true);
@@ -634,8 +637,6 @@ void Config::ReadDebuggingValues() {
// Intentionally not using the QT default setting as this is intended to be changed in the ini
Settings::values.record_frame_times =
qt_config->value(QStringLiteral("record_frame_times"), false).toBool();
- Settings::values.use_gdbstub = ReadSetting(QStringLiteral("use_gdbstub"), false).toBool();
- Settings::values.gdbstub_port = ReadSetting(QStringLiteral("gdbstub_port"), 24689).toInt();
Settings::values.program_args =
ReadSetting(QStringLiteral("program_args"), QString{}).toString().toStdString();
Settings::values.dump_exefs = ReadSetting(QStringLiteral("dump_exefs"), false).toBool();
@@ -1186,6 +1187,8 @@ void Config::SaveControlValues() {
QString::fromStdString(Settings::values.touch_device),
QStringLiteral("engine:emu_window"));
WriteSetting(QStringLiteral("keyboard_enabled"), Settings::values.keyboard_enabled, false);
+ WriteSetting(QStringLiteral("emulate_analog_keyboard"),
+ Settings::values.emulate_analog_keyboard, false);
qt_config->endGroup();
}
@@ -1231,8 +1234,6 @@ void Config::SaveDebuggingValues() {
// Intentionally not using the QT default setting as this is intended to be changed in the ini
qt_config->setValue(QStringLiteral("record_frame_times"), Settings::values.record_frame_times);
- WriteSetting(QStringLiteral("use_gdbstub"), Settings::values.use_gdbstub, false);
- WriteSetting(QStringLiteral("gdbstub_port"), Settings::values.gdbstub_port, 24689);
WriteSetting(QStringLiteral("program_args"),
QString::fromStdString(Settings::values.program_args), QString{});
WriteSetting(QStringLiteral("dump_exefs"), Settings::values.dump_exefs, false);
@@ -1588,14 +1589,12 @@ void Config::WriteSettingGlobal(const QString& name, const QVariant& value, bool
void Config::Reload() {
ReadValues();
- Settings::Sanitize();
// To apply default value changes
SaveValues();
Settings::Apply(Core::System::GetInstance());
}
void Config::Save() {
- Settings::Sanitize();
SaveValues();
}
diff --git a/src/yuzu/configuration/configure_debug.cpp b/src/yuzu/configuration/configure_debug.cpp
index 027099ab7..121873f95 100644
--- a/src/yuzu/configuration/configure_debug.cpp
+++ b/src/yuzu/configuration/configure_debug.cpp
@@ -28,9 +28,6 @@ ConfigureDebug::ConfigureDebug(QWidget* parent) : QWidget(parent), ui(new Ui::Co
ConfigureDebug::~ConfigureDebug() = default;
void ConfigureDebug::SetConfiguration() {
- ui->toggle_gdbstub->setChecked(Settings::values.use_gdbstub);
- ui->gdbport_spinbox->setEnabled(Settings::values.use_gdbstub);
- ui->gdbport_spinbox->setValue(Settings::values.gdbstub_port);
ui->toggle_console->setEnabled(!Core::System::GetInstance().IsPoweredOn());
ui->toggle_console->setChecked(UISettings::values.show_console);
ui->log_filter_edit->setText(QString::fromStdString(Settings::values.log_filter));
@@ -45,8 +42,6 @@ void ConfigureDebug::SetConfiguration() {
}
void ConfigureDebug::ApplyConfiguration() {
- Settings::values.use_gdbstub = ui->toggle_gdbstub->isChecked();
- Settings::values.gdbstub_port = ui->gdbport_spinbox->value();
UISettings::values.show_console = ui->toggle_console->isChecked();
Settings::values.log_filter = ui->log_filter_edit->text().toStdString();
Settings::values.program_args = ui->homebrew_args_edit->text().toStdString();
diff --git a/src/yuzu/configuration/configure_debug.ui b/src/yuzu/configuration/configure_debug.ui
index 6f94fe304..9186aa732 100644
--- a/src/yuzu/configuration/configure_debug.ui
+++ b/src/yuzu/configuration/configure_debug.ui
@@ -7,7 +7,7 @@
<x>0</x>
<y>0</y>
<width>400</width>
- <height>467</height>
+ <height>486</height>
</rect>
</property>
<property name="windowTitle">
@@ -15,57 +15,6 @@
</property>
<layout class="QVBoxLayout" name="verticalLayout_1">
<item>
- <layout class="QVBoxLayout" name="verticalLayout_2">
- <item>
- <widget class="QGroupBox" name="groupBox">
- <property name="title">
- <string>GDB</string>
- </property>
- <layout class="QVBoxLayout" name="verticalLayout_3">
- <item>
- <layout class="QHBoxLayout" name="horizontalLayout_1">
- <item>
- <widget class="QCheckBox" name="toggle_gdbstub">
- <property name="text">
- <string>Enable GDB Stub</string>
- </property>
- </widget>
- </item>
- <item>
- <spacer name="horizontalSpacer">
- <property name="orientation">
- <enum>Qt::Horizontal</enum>
- </property>
- <property name="sizeHint" stdset="0">
- <size>
- <width>40</width>
- <height>20</height>
- </size>
- </property>
- </spacer>
- </item>
- <item>
- <widget class="QLabel" name="label_1">
- <property name="text">
- <string>Port:</string>
- </property>
- </widget>
- </item>
- <item>
- <widget class="QSpinBox" name="gdbport_spinbox">
- <property name="maximum">
- <number>65536</number>
- </property>
- </widget>
- </item>
- </layout>
- </item>
- </layout>
- </widget>
- </item>
- </layout>
- </item>
- <item>
<widget class="QGroupBox" name="groupBox_2">
<property name="title">
<string>Logging</string>
@@ -258,8 +207,6 @@
</layout>
</widget>
<tabstops>
- <tabstop>toggle_gdbstub</tabstop>
- <tabstop>gdbport_spinbox</tabstop>
<tabstop>log_filter_edit</tabstop>
<tabstop>toggle_console</tabstop>
<tabstop>open_log_button</tabstop>
@@ -269,22 +216,5 @@
<tabstop>quest_flag</tabstop>
</tabstops>
<resources/>
- <connections>
- <connection>
- <sender>toggle_gdbstub</sender>
- <signal>toggled(bool)</signal>
- <receiver>gdbport_spinbox</receiver>
- <slot>setEnabled(bool)</slot>
- <hints>
- <hint type="sourcelabel">
- <x>84</x>
- <y>157</y>
- </hint>
- <hint type="destinationlabel">
- <x>342</x>
- <y>158</y>
- </hint>
- </hints>
- </connection>
- </connections>
+ <connections/>
</ui>
diff --git a/src/yuzu/configuration/configure_graphics.cpp b/src/yuzu/configuration/configure_graphics.cpp
index 6fda0ce35..b78a5dff0 100644
--- a/src/yuzu/configuration/configure_graphics.cpp
+++ b/src/yuzu/configuration/configure_graphics.cpp
@@ -4,22 +4,17 @@
#include <QColorDialog>
#include <QComboBox>
-#ifdef HAS_VULKAN
#include <QVulkanInstance>
-#endif
#include "common/common_types.h"
#include "common/logging/log.h"
#include "core/core.h"
#include "core/settings.h"
#include "ui_configure_graphics.h"
+#include "video_core/renderer_vulkan/renderer_vulkan.h"
#include "yuzu/configuration/configuration_shared.h"
#include "yuzu/configuration/configure_graphics.h"
-#ifdef HAS_VULKAN
-#include "video_core/renderer_vulkan/renderer_vulkan.h"
-#endif
-
ConfigureGraphics::ConfigureGraphics(QWidget* parent)
: QWidget(parent), ui(new Ui::ConfigureGraphics) {
vulkan_device = Settings::values.vulkan_device.GetValue();
@@ -218,12 +213,10 @@ void ConfigureGraphics::UpdateDeviceComboBox() {
}
void ConfigureGraphics::RetrieveVulkanDevices() {
-#ifdef HAS_VULKAN
vulkan_devices.clear();
- for (auto& name : Vulkan::RendererVulkan::EnumerateDevices()) {
+ for (const auto& name : Vulkan::RendererVulkan::EnumerateDevices()) {
vulkan_devices.push_back(QString::fromStdString(name));
}
-#endif
}
Settings::RendererBackend ConfigureGraphics::GetCurrentGraphicsBackend() const {
diff --git a/src/yuzu/configuration/configure_input_advanced.cpp b/src/yuzu/configuration/configure_input_advanced.cpp
index abaf03630..4e557bc6f 100644
--- a/src/yuzu/configuration/configure_input_advanced.cpp
+++ b/src/yuzu/configuration/configure_input_advanced.cpp
@@ -121,6 +121,7 @@ void ConfigureInputAdvanced::ApplyConfiguration() {
Settings::values.debug_pad_enabled = ui->debug_enabled->isChecked();
Settings::values.mouse_enabled = ui->mouse_enabled->isChecked();
Settings::values.keyboard_enabled = ui->keyboard_enabled->isChecked();
+ Settings::values.emulate_analog_keyboard = ui->emulate_analog_keyboard->isChecked();
Settings::values.touchscreen.enabled = ui->touchscreen_enabled->isChecked();
}
@@ -147,6 +148,7 @@ void ConfigureInputAdvanced::LoadConfiguration() {
ui->debug_enabled->setChecked(Settings::values.debug_pad_enabled);
ui->mouse_enabled->setChecked(Settings::values.mouse_enabled);
ui->keyboard_enabled->setChecked(Settings::values.keyboard_enabled);
+ ui->emulate_analog_keyboard->setChecked(Settings::values.emulate_analog_keyboard);
ui->touchscreen_enabled->setChecked(Settings::values.touchscreen.enabled);
UpdateUIEnabled();
diff --git a/src/yuzu/configuration/configure_input_advanced.ui b/src/yuzu/configuration/configure_input_advanced.ui
index a880a7c68..f207e5d3b 100644
--- a/src/yuzu/configuration/configure_input_advanced.ui
+++ b/src/yuzu/configuration/configure_input_advanced.ui
@@ -2546,14 +2546,27 @@
</property>
</widget>
</item>
- <item row="4" column="2">
+ <item row="1" column="0">
+ <widget class="QCheckBox" name="emulate_analog_keyboard">
+ <property name="minimumSize">
+ <size>
+ <width>0</width>
+ <height>23</height>
+ </size>
+ </property>
+ <property name="text">
+ <string>Emulate Analog with Keyboard Input</string>
+ </property>
+ </widget>
+ </item>
+ <item row="5" column="2">
<widget class="QPushButton" name="touchscreen_advanced">
<property name="text">
<string>Advanced</string>
</property>
</widget>
</item>
- <item row="1" column="1">
+ <item row="2" column="1">
<spacer name="horizontalSpacer_8">
<property name="orientation">
<enum>Qt::Horizontal</enum>
@@ -2569,21 +2582,21 @@
</property>
</spacer>
</item>
- <item row="1" column="2">
+ <item row="2" column="2">
<widget class="QPushButton" name="mouse_advanced">
<property name="text">
<string>Advanced</string>
</property>
</widget>
</item>
- <item row="4" column="0">
+ <item row="5" column="0">
<widget class="QCheckBox" name="touchscreen_enabled">
<property name="text">
<string>Touchscreen</string>
</property>
</widget>
</item>
- <item row="1" column="0">
+ <item row="2" column="0">
<widget class="QCheckBox" name="mouse_enabled">
<property name="minimumSize">
<size>
@@ -2596,28 +2609,28 @@
</property>
</widget>
</item>
- <item row="6" column="0">
+ <item row="7" column="0">
<widget class="QLabel" name="motion_touch">
<property name="text">
<string>Motion / Touch</string>
</property>
</widget>
</item>
- <item row="6" column="2">
+ <item row="7" column="2">
<widget class="QPushButton" name="buttonMotionTouch">
<property name="text">
<string>Configure</string>
</property>
</widget>
</item>
- <item row="5" column="0">
+ <item row="6" column="0">
<widget class="QCheckBox" name="debug_enabled">
<property name="text">
<string>Debug Controller</string>
</property>
</widget>
</item>
- <item row="5" column="2">
+ <item row="6" column="2">
<widget class="QPushButton" name="debug_configure">
<property name="text">
<string>Configure</string>
diff --git a/src/yuzu/configuration/configure_input_player.cpp b/src/yuzu/configuration/configure_input_player.cpp
index f9915fb7a..3c7500ee3 100644
--- a/src/yuzu/configuration/configure_input_player.cpp
+++ b/src/yuzu/configuration/configure_input_player.cpp
@@ -173,61 +173,31 @@ QString AnalogToText(const Common::ParamPackage& param, const std::string& dir)
return ButtonToText(Common::ParamPackage{param.Get(dir, "")});
}
- if (param.Get("engine", "") == "sdl") {
+ const auto engine_str = param.Get("engine", "");
+ const QString axis_x_str = QString::fromStdString(param.Get("axis_x", ""));
+ const QString axis_y_str = QString::fromStdString(param.Get("axis_y", ""));
+ const bool invert_x = param.Get("invert_x", "+") == "-";
+ const bool invert_y = param.Get("invert_y", "+") == "-";
+ if (engine_str == "sdl" || engine_str == "gcpad" || engine_str == "mouse") {
if (dir == "modifier") {
return QObject::tr("[unused]");
}
- if (dir == "left" || dir == "right") {
- const QString axis_x_str = QString::fromStdString(param.Get("axis_x", ""));
-
- return QObject::tr("Axis %1").arg(axis_x_str);
- }
-
- if (dir == "up" || dir == "down") {
- const QString axis_y_str = QString::fromStdString(param.Get("axis_y", ""));
-
- return QObject::tr("Axis %1").arg(axis_y_str);
+ if (dir == "left") {
+ const QString invert_x_str = QString::fromStdString(invert_x ? "+" : "-");
+ return QObject::tr("Axis %1%2").arg(axis_x_str, invert_x_str);
}
-
- return {};
- }
-
- if (param.Get("engine", "") == "gcpad") {
- if (dir == "modifier") {
- return QObject::tr("[unused]");
+ if (dir == "right") {
+ const QString invert_x_str = QString::fromStdString(invert_x ? "-" : "+");
+ return QObject::tr("Axis %1%2").arg(axis_x_str, invert_x_str);
}
-
- if (dir == "left" || dir == "right") {
- const QString axis_x_str = QString::fromStdString(param.Get("axis_x", ""));
-
- return QObject::tr("GC Axis %1").arg(axis_x_str);
+ if (dir == "up") {
+ const QString invert_y_str = QString::fromStdString(invert_y ? "-" : "+");
+ return QObject::tr("Axis %1%2").arg(axis_y_str, invert_y_str);
}
-
- if (dir == "up" || dir == "down") {
- const QString axis_y_str = QString::fromStdString(param.Get("axis_y", ""));
-
- return QObject::tr("GC Axis %1").arg(axis_y_str);
- }
-
- return {};
- }
-
- if (param.Get("engine", "") == "mouse") {
- if (dir == "modifier") {
- return QObject::tr("[unused]");
- }
-
- if (dir == "left" || dir == "right") {
- const QString axis_x_str = QString::fromStdString(param.Get("axis_x", ""));
-
- return QObject::tr("Mouse %1").arg(axis_x_str);
- }
-
- if (dir == "up" || dir == "down") {
- const QString axis_y_str = QString::fromStdString(param.Get("axis_y", ""));
-
- return QObject::tr("Mouse %1").arg(axis_y_str);
+ if (dir == "down") {
+ const QString invert_y_str = QString::fromStdString(invert_y ? "+" : "-");
+ return QObject::tr("Axis %1%2").arg(axis_y_str, invert_y_str);
}
return {};
@@ -396,6 +366,25 @@ ConfigureInputPlayer::ConfigureInputPlayer(QWidget* parent, std::size_t player_i
analogs_param[analog_id].Clear();
analog_map_buttons[analog_id][sub_button_id]->setText(tr("[not set]"));
});
+ context_menu.addAction(tr("Invert axis"), [&] {
+ if (sub_button_id == 2 || sub_button_id == 3) {
+ const bool invert_value =
+ analogs_param[analog_id].Get("invert_x", "+") == "-";
+ const std::string invert_str = invert_value ? "+" : "-";
+ analogs_param[analog_id].Set("invert_x", invert_str);
+ }
+ if (sub_button_id == 0 || sub_button_id == 1) {
+ const bool invert_value =
+ analogs_param[analog_id].Get("invert_y", "+") == "-";
+ const std::string invert_str = invert_value ? "+" : "-";
+ analogs_param[analog_id].Set("invert_y", invert_str);
+ }
+ for (int sub_button_id = 0; sub_button_id < ANALOG_SUB_BUTTONS_NUM;
+ ++sub_button_id) {
+ analog_map_buttons[analog_id][sub_button_id]->setText(AnalogToText(
+ analogs_param[analog_id], analog_sub_buttons[sub_button_id]));
+ }
+ });
context_menu.exec(analog_map_buttons[analog_id][sub_button_id]->mapToGlobal(
menu_location));
});
diff --git a/src/yuzu/configuration/configure_motion_touch.cpp b/src/yuzu/configuration/configure_motion_touch.cpp
index 2afac591a..c2a7113da 100644
--- a/src/yuzu/configuration/configure_motion_touch.cpp
+++ b/src/yuzu/configuration/configure_motion_touch.cpp
@@ -183,8 +183,8 @@ void ConfigureMotionTouch::ConnectEvents() {
}
void ConfigureMotionTouch::OnUDPAddServer() {
- QRegExp re(tr("^(?:(?:25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)\.){3}(?:25[0-5]|2[0-4][0-9]|[01]?["
- "0-9][0-9]?)$")); // a valid ip address
+ QRegExp re(tr(R"re(^(?:(?:25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)\.){3}(?:25[0-5]|2[0-4]"
+ "[0-9]|[01]?[0-9][0-9]?)$)re")); // a valid ip address
bool ok;
QString port_text = ui->udp_port->text();
QString server_text = ui->udp_server->text();
diff --git a/src/yuzu/debugger/profiler.cpp b/src/yuzu/debugger/profiler.cpp
index 0e26f765b..efdc6aa50 100644
--- a/src/yuzu/debugger/profiler.cpp
+++ b/src/yuzu/debugger/profiler.cpp
@@ -48,7 +48,7 @@ private:
MicroProfileDialog::MicroProfileDialog(QWidget* parent) : QWidget(parent, Qt::Dialog) {
setObjectName(QStringLiteral("MicroProfile"));
- setWindowTitle(tr("MicroProfile"));
+ setWindowTitle(tr("&MicroProfile"));
resize(1000, 600);
// Remove the "?" button from the titlebar and enable the maximize button
setWindowFlags((windowFlags() & ~Qt::WindowContextHelpButtonHint) |
diff --git a/src/yuzu/debugger/wait_tree.cpp b/src/yuzu/debugger/wait_tree.cpp
index a20824719..0925c10b4 100644
--- a/src/yuzu/debugger/wait_tree.cpp
+++ b/src/yuzu/debugger/wait_tree.cpp
@@ -13,10 +13,10 @@
#include "core/arm/arm_interface.h"
#include "core/core.h"
#include "core/hle/kernel/handle_table.h"
+#include "core/hle/kernel/k_scheduler.h"
#include "core/hle/kernel/mutex.h"
#include "core/hle/kernel/process.h"
#include "core/hle/kernel/readable_event.h"
-#include "core/hle/kernel/scheduler.h"
#include "core/hle/kernel/synchronization_object.h"
#include "core/hle/kernel/thread.h"
#include "core/memory.h"
@@ -101,7 +101,7 @@ std::vector<std::unique_ptr<WaitTreeThread>> WaitTreeItem::MakeThreadItemList()
};
const auto& system = Core::System::GetInstance();
- add_threads(system.GlobalScheduler().GetThreadList());
+ add_threads(system.GlobalSchedulerContext().GetThreadList());
return item_list;
}
@@ -349,14 +349,14 @@ std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeThread::GetChildren() const {
list.push_back(std::make_unique<WaitTreeText>(tr("processor = %1").arg(processor)));
list.push_back(
std::make_unique<WaitTreeText>(tr("ideal core = %1").arg(thread.GetIdealCore())));
- list.push_back(
- std::make_unique<WaitTreeText>(tr("affinity mask = %1").arg(thread.GetAffinityMask())));
+ list.push_back(std::make_unique<WaitTreeText>(
+ tr("affinity mask = %1").arg(thread.GetAffinityMask().GetAffinityMask())));
list.push_back(std::make_unique<WaitTreeText>(tr("thread id = %1").arg(thread.GetThreadID())));
list.push_back(std::make_unique<WaitTreeText>(tr("priority = %1(current) / %2(normal)")
.arg(thread.GetPriority())
.arg(thread.GetNominalPriority())));
list.push_back(std::make_unique<WaitTreeText>(
- tr("last running ticks = %1").arg(thread.GetLastRunningTicks())));
+ tr("last running ticks = %1").arg(thread.GetLastScheduledTick())));
const VAddr mutex_wait_address = thread.GetMutexWaitAddress();
if (mutex_wait_address != 0) {
@@ -457,7 +457,7 @@ void WaitTreeModel::InitItems() {
thread_items = WaitTreeItem::MakeThreadItemList();
}
-WaitTreeWidget::WaitTreeWidget(QWidget* parent) : QDockWidget(tr("Wait Tree"), parent) {
+WaitTreeWidget::WaitTreeWidget(QWidget* parent) : QDockWidget(tr("&Wait Tree"), parent) {
setObjectName(QStringLiteral("WaitTreeWidget"));
view = new QTreeView(this);
view->setHeaderHidden(true);
diff --git a/src/yuzu/game_list_p.h b/src/yuzu/game_list_p.h
index 248855aff..df935022d 100644
--- a/src/yuzu/game_list_p.h
+++ b/src/yuzu/game_list_p.h
@@ -174,7 +174,8 @@ public:
}
bool operator<(const QStandardItem& other) const override {
- return data(CompatNumberRole) < other.data(CompatNumberRole);
+ return data(CompatNumberRole).value<QString>() <
+ other.data(CompatNumberRole).value<QString>();
}
};
diff --git a/src/yuzu/main.cpp b/src/yuzu/main.cpp
index 26f5e42ed..ab66d7f93 100644
--- a/src/yuzu/main.cpp
+++ b/src/yuzu/main.cpp
@@ -28,8 +28,6 @@
#include "core/hle/service/am/applet_ae.h"
#include "core/hle/service/am/applet_oe.h"
#include "core/hle/service/am/applets/applets.h"
-#include "core/hle/service/hid/controllers/npad.h"
-#include "core/hle/service/hid/hid.h"
// These are wrappers to avoid the calls to CreateDirectory and CreateFile because of the Windows
// defines.
@@ -83,6 +81,7 @@ static FileSys::VirtualFile VfsDirectoryCreateFileWrapper(const FileSys::Virtual
#include "core/core.h"
#include "core/crypto/key_manager.h"
#include "core/file_sys/card_image.h"
+#include "core/file_sys/common_funcs.h"
#include "core/file_sys/content_archive.h"
#include "core/file_sys/control_metadata.h"
#include "core/file_sys/patch_manager.h"
@@ -124,14 +123,6 @@ static FileSys::VirtualFile VfsDirectoryCreateFileWrapper(const FileSys::Virtual
#include "yuzu/discord_impl.h"
#endif
-#ifdef YUZU_USE_QT_WEB_ENGINE
-#include <QWebEngineProfile>
-#include <QWebEngineScript>
-#include <QWebEngineScriptCollection>
-#include <QWebEngineSettings>
-#include <QWebEngineView>
-#endif
-
#ifdef QT_STATICPLUGIN
Q_IMPORT_PLUGIN(QWindowsIntegrationPlugin);
#endif
@@ -148,8 +139,6 @@ __declspec(dllexport) int AmdPowerXpressRequestHighPerformance = 1;
constexpr int default_mouse_timeout = 2500;
-constexpr u64 DLC_BASE_TITLE_ID_MASK = 0xFFFFFFFFFFFFE000;
-
/**
* "Callouts" are one-time instructional messages shown to the user. In the config settings, there
* is a bitfield "callout_flags" options, used to track if a message has already been shown to the
@@ -191,6 +180,30 @@ static void InitializeLogging() {
#endif
}
+static void RemoveCachedContents() {
+ const auto offline_fonts = Common::FS::SanitizePath(
+ fmt::format("{}/fonts", Common::FS::GetUserPath(Common::FS::UserPath::CacheDir)),
+ Common::FS::DirectorySeparator::PlatformDefault);
+
+ const auto offline_manual = Common::FS::SanitizePath(
+ fmt::format("{}/offline_web_applet_manual",
+ Common::FS::GetUserPath(Common::FS::UserPath::CacheDir)),
+ Common::FS::DirectorySeparator::PlatformDefault);
+ const auto offline_legal_information = Common::FS::SanitizePath(
+ fmt::format("{}/offline_web_applet_legal_information",
+ Common::FS::GetUserPath(Common::FS::UserPath::CacheDir)),
+ Common::FS::DirectorySeparator::PlatformDefault);
+ const auto offline_system_data = Common::FS::SanitizePath(
+ fmt::format("{}/offline_web_applet_system_data",
+ Common::FS::GetUserPath(Common::FS::UserPath::CacheDir)),
+ Common::FS::DirectorySeparator::PlatformDefault);
+
+ Common::FS::DeleteDirRecursively(offline_fonts);
+ Common::FS::DeleteDirRecursively(offline_manual);
+ Common::FS::DeleteDirRecursively(offline_legal_information);
+ Common::FS::DeleteDirRecursively(offline_system_data);
+}
+
GMainWindow::GMainWindow()
: input_subsystem{std::make_shared<InputCommon::InputSubsystem>()},
config{std::make_unique<Config>()}, vfs{std::make_shared<FileSys::RealVfsFilesystem>()},
@@ -259,6 +272,9 @@ GMainWindow::GMainWindow()
FileSys::ContentProviderUnionSlot::FrontendManual, provider.get());
Core::System::GetInstance().GetFileSystemController().CreateFactories(*vfs);
+ // Remove cached contents generated during the previous session
+ RemoveCachedContents();
+
// Gen keys if necessary
OnReinitializeKeys(ReinitializeKeyBehavior::NoWarning);
@@ -350,150 +366,141 @@ void GMainWindow::SoftwareKeyboardInvokeCheckDialog(std::u16string error_message
emit SoftwareKeyboardFinishedCheckDialog();
}
+void GMainWindow::WebBrowserOpenWebPage(std::string_view main_url, std::string_view additional_args,
+ bool is_local) {
#ifdef YUZU_USE_QT_WEB_ENGINE
-void GMainWindow::WebBrowserOpenPage(std::string_view filename, std::string_view additional_args) {
- NXInputWebEngineView web_browser_view(this);
+ if (disable_web_applet) {
+ emit WebBrowserClosed(Service::AM::Applets::WebExitReason::WindowClosed,
+ "http://localhost/");
+ return;
+ }
- // Scope to contain the QProgressDialog for initialization
- {
- QProgressDialog progress(this);
- progress.setMinimumDuration(200);
- progress.setLabelText(tr("Loading Web Applet..."));
- progress.setRange(0, 4);
- progress.setValue(0);
- progress.show();
+ QtNXWebEngineView web_browser_view(this, Core::System::GetInstance(), input_subsystem.get());
- auto future = QtConcurrent::run([this] { emit WebBrowserUnpackRomFS(); });
+ ui.action_Pause->setEnabled(false);
+ ui.action_Restart->setEnabled(false);
+ ui.action_Stop->setEnabled(false);
- while (!future.isFinished())
- QApplication::processEvents();
+ {
+ QProgressDialog loading_progress(this);
+ loading_progress.setLabelText(tr("Loading Web Applet..."));
+ loading_progress.setRange(0, 3);
+ loading_progress.setValue(0);
+
+ if (is_local && !Common::FS::Exists(std::string(main_url))) {
+ loading_progress.show();
- progress.setValue(1);
+ auto future = QtConcurrent::run([this] { emit WebBrowserExtractOfflineRomFS(); });
- // Load the special shim script to handle input and exit.
- QWebEngineScript nx_shim;
- nx_shim.setSourceCode(GetNXShimInjectionScript());
- nx_shim.setWorldId(QWebEngineScript::MainWorld);
- nx_shim.setName(QStringLiteral("nx_inject.js"));
- nx_shim.setInjectionPoint(QWebEngineScript::DocumentCreation);
- nx_shim.setRunsOnSubFrames(true);
- web_browser_view.page()->profile()->scripts()->insert(nx_shim);
+ while (!future.isFinished()) {
+ QCoreApplication::processEvents();
+ }
+ }
- web_browser_view.load(
- QUrl(QUrl::fromLocalFile(QString::fromStdString(std::string(filename))).toString() +
- QString::fromStdString(std::string(additional_args))));
+ loading_progress.setValue(1);
- progress.setValue(2);
+ if (is_local) {
+ web_browser_view.LoadLocalWebPage(main_url, additional_args);
+ } else {
+ web_browser_view.LoadExternalWebPage(main_url, additional_args);
+ }
- render_window->hide();
- web_browser_view.setFocus();
+ if (render_window->IsLoadingComplete()) {
+ render_window->hide();
+ }
const auto& layout = render_window->GetFramebufferLayout();
web_browser_view.resize(layout.screen.GetWidth(), layout.screen.GetHeight());
web_browser_view.move(layout.screen.left, layout.screen.top + menuBar()->height());
web_browser_view.setZoomFactor(static_cast<qreal>(layout.screen.GetWidth()) /
- Layout::ScreenUndocked::Width);
- web_browser_view.settings()->setAttribute(
- QWebEngineSettings::LocalContentCanAccessRemoteUrls, true);
+ static_cast<qreal>(Layout::ScreenUndocked::Width));
+ web_browser_view.setFocus();
web_browser_view.show();
- progress.setValue(3);
+ loading_progress.setValue(2);
- QApplication::processEvents();
+ QCoreApplication::processEvents();
- progress.setValue(4);
+ loading_progress.setValue(3);
}
- bool finished = false;
- QAction* exit_action = new QAction(tr("Exit Web Applet"), this);
- connect(exit_action, &QAction::triggered, this, [&finished] { finished = true; });
- ui.menubar->addAction(exit_action);
+ bool exit_check = false;
- auto& npad =
- Core::System::GetInstance()
- .ServiceManager()
- .GetService<Service::HID::Hid>("hid")
- ->GetAppletResource()
- ->GetController<Service::HID::Controller_NPad>(Service::HID::HidController::NPad);
-
- const auto fire_js_keypress = [&web_browser_view](u32 key_code) {
- web_browser_view.page()->runJavaScript(
- QStringLiteral("document.dispatchEvent(new KeyboardEvent('keydown', {'key': %1}));")
- .arg(key_code));
- };
+ // TODO (Morph): Remove this
+ QAction* exit_action = new QAction(tr("Disable Web Applet"), this);
+ connect(exit_action, &QAction::triggered, this, [this, &web_browser_view] {
+ const auto result = QMessageBox::warning(
+ this, tr("Disable Web Applet"),
+ tr("Disabling the web applet will cause it to not be shown again for the rest of the "
+ "emulated session. This can lead to undefined behavior and should only be used with "
+ "Super Mario 3D All-Stars. Are you sure you want to disable the web applet?"),
+ QMessageBox::Yes | QMessageBox::No);
+ if (result == QMessageBox::Yes) {
+ disable_web_applet = true;
+ web_browser_view.SetFinished(true);
+ }
+ });
+ ui.menubar->addAction(exit_action);
- QMessageBox::information(
- this, tr("Exit"),
- tr("To exit the web application, use the game provided controls to select exit, select the "
- "'Exit Web Applet' option in the menu bar, or press the 'Enter' key."));
-
- bool running_exit_check = false;
- while (!finished) {
- QApplication::processEvents();
-
- if (!running_exit_check) {
- web_browser_view.page()->runJavaScript(QStringLiteral("applet_done;"),
- [&](const QVariant& res) {
- running_exit_check = false;
- if (res.toBool())
- finished = true;
- });
- running_exit_check = true;
+ while (!web_browser_view.IsFinished()) {
+ QCoreApplication::processEvents();
+
+ if (!exit_check) {
+ web_browser_view.page()->runJavaScript(
+ QStringLiteral("end_applet;"), [&](const QVariant& variant) {
+ exit_check = false;
+ if (variant.toBool()) {
+ web_browser_view.SetFinished(true);
+ web_browser_view.SetExitReason(
+ Service::AM::Applets::WebExitReason::EndButtonPressed);
+ }
+ });
+
+ exit_check = true;
}
- const auto input = npad.GetAndResetPressState();
- for (std::size_t i = 0; i < Settings::NativeButton::NumButtons; ++i) {
- if ((input & (1 << i)) != 0) {
- LOG_DEBUG(Frontend, "firing input for button id={:02X}", i);
- web_browser_view.page()->runJavaScript(
- QStringLiteral("yuzu_key_callbacks[%1]();").arg(i));
+ if (web_browser_view.GetCurrentURL().contains(QStringLiteral("localhost"))) {
+ if (!web_browser_view.IsFinished()) {
+ web_browser_view.SetFinished(true);
+ web_browser_view.SetExitReason(Service::AM::Applets::WebExitReason::CallbackURL);
}
+
+ web_browser_view.SetLastURL(web_browser_view.GetCurrentURL().toStdString());
}
- if (input & 0x00888000) // RStick Down | LStick Down | DPad Down
- fire_js_keypress(40); // Down Arrow Key
- else if (input & 0x00444000) // RStick Right | LStick Right | DPad Right
- fire_js_keypress(39); // Right Arrow Key
- else if (input & 0x00222000) // RStick Up | LStick Up | DPad Up
- fire_js_keypress(38); // Up Arrow Key
- else if (input & 0x00111000) // RStick Left | LStick Left | DPad Left
- fire_js_keypress(37); // Left Arrow Key
- else if (input & 0x00000001) // A Button
- fire_js_keypress(13); // Enter Key
+ std::this_thread::sleep_for(std::chrono::milliseconds(1));
}
+ const auto exit_reason = web_browser_view.GetExitReason();
+ const auto last_url = web_browser_view.GetLastURL();
+
web_browser_view.hide();
- render_window->show();
+
render_window->setFocus();
- ui.menubar->removeAction(exit_action);
- // Needed to update render window focus/show and remove menubar action
- QApplication::processEvents();
- emit WebBrowserFinishedBrowsing();
-}
+ if (render_window->IsLoadingComplete()) {
+ render_window->show();
+ }
-#else
+ ui.action_Pause->setEnabled(true);
+ ui.action_Restart->setEnabled(true);
+ ui.action_Stop->setEnabled(true);
-void GMainWindow::WebBrowserOpenPage(std::string_view filename, std::string_view additional_args) {
-#ifndef __linux__
- QMessageBox::warning(
- this, tr("Web Applet"),
- tr("This version of yuzu was built without QtWebEngine support, meaning that yuzu cannot "
- "properly display the game manual or web page requested."),
- QMessageBox::Ok, QMessageBox::Ok);
-#endif
+ ui.menubar->removeAction(exit_action);
- LOG_INFO(Frontend,
- "(STUBBED) called - Missing QtWebEngine dependency needed to open website page at "
- "'{}' with arguments '{}'!",
- filename, additional_args);
+ QCoreApplication::processEvents();
- emit WebBrowserFinishedBrowsing();
-}
+ emit WebBrowserClosed(exit_reason, last_url);
+
+#else
+
+ // Utilize the same fallback as the default web browser applet.
+ emit WebBrowserClosed(Service::AM::Applets::WebExitReason::WindowClosed, "http://localhost/");
#endif
+}
void GMainWindow::InitializeWidgets() {
#ifdef YUZU_ENABLE_COMPATIBILITY_REPORTING
@@ -573,9 +580,8 @@ void GMainWindow::InitializeWidgets() {
if (emulation_running) {
return;
}
- const bool is_async = !Settings::values.use_asynchronous_gpu_emulation.GetValue() ||
- Settings::values.use_multi_core.GetValue();
- Settings::values.use_asynchronous_gpu_emulation.SetValue(is_async);
+ Settings::values.use_asynchronous_gpu_emulation.SetValue(
+ !Settings::values.use_asynchronous_gpu_emulation.GetValue());
async_status_button->setChecked(Settings::values.use_asynchronous_gpu_emulation.GetValue());
Settings::Apply(Core::System::GetInstance());
});
@@ -592,16 +598,13 @@ void GMainWindow::InitializeWidgets() {
return;
}
Settings::values.use_multi_core.SetValue(!Settings::values.use_multi_core.GetValue());
- const bool is_async = Settings::values.use_asynchronous_gpu_emulation.GetValue() ||
- Settings::values.use_multi_core.GetValue();
- Settings::values.use_asynchronous_gpu_emulation.SetValue(is_async);
- async_status_button->setChecked(Settings::values.use_asynchronous_gpu_emulation.GetValue());
multicore_status_button->setChecked(Settings::values.use_multi_core.GetValue());
Settings::Apply(Core::System::GetInstance());
});
multicore_status_button->setText(tr("MULTICORE"));
multicore_status_button->setCheckable(true);
multicore_status_button->setChecked(Settings::values.use_multi_core.GetValue());
+
statusBar()->insertPermanentWidget(0, multicore_status_button);
statusBar()->insertPermanentWidget(0, async_status_button);
@@ -615,11 +618,6 @@ void GMainWindow::InitializeWidgets() {
});
renderer_status_button->toggle();
-#ifndef HAS_VULKAN
- renderer_status_button->setChecked(false);
- renderer_status_button->setCheckable(false);
- renderer_status_button->setDisabled(true);
-#else
renderer_status_button->setChecked(Settings::values.renderer_backend.GetValue() ==
Settings::RendererBackend::Vulkan);
connect(renderer_status_button, &QPushButton::clicked, [this] {
@@ -634,7 +632,6 @@ void GMainWindow::InitializeWidgets() {
Settings::Apply(Core::System::GetInstance());
});
-#endif // HAS_VULKAN
statusBar()->insertPermanentWidget(0, renderer_status_button);
statusBar()->setVisible(true);
@@ -670,7 +667,7 @@ void GMainWindow::InitializeRecentFileMenuActions() {
}
ui.menu_recent_files->addSeparator();
QAction* action_clear_recent_files = new QAction(this);
- action_clear_recent_files->setText(tr("Clear Recent Files"));
+ action_clear_recent_files->setText(tr("&Clear Recent Files"));
connect(action_clear_recent_files, &QAction::triggered, this, [this] {
UISettings::values.recent_files.clear();
UpdateRecentFiles();
@@ -932,7 +929,10 @@ void GMainWindow::ConnectMenuEvents() {
&GMainWindow::OnDisplayTitleBars);
connect(ui.action_Show_Filter_Bar, &QAction::triggered, this, &GMainWindow::OnToggleFilterBar);
connect(ui.action_Show_Status_Bar, &QAction::triggered, statusBar(), &QStatusBar::setVisible);
- connect(ui.action_Reset_Window_Size, &QAction::triggered, this, &GMainWindow::ResetWindowSize);
+ connect(ui.action_Reset_Window_Size_720, &QAction::triggered, this,
+ &GMainWindow::ResetWindowSize720);
+ connect(ui.action_Reset_Window_Size_1080, &QAction::triggered, this,
+ &GMainWindow::ResetWindowSize1080);
// Fullscreen
connect(ui.action_Fullscreen, &QAction::triggered, this, &GMainWindow::ToggleFullscreen);
@@ -994,7 +994,6 @@ bool GMainWindow::LoadROM(const QString& filename, std::size_t program_index) {
system.SetAppletFrontendSet({
std::make_unique<QtControllerSelector>(*this), // Controller Selector
- nullptr, // E-Commerce
std::make_unique<QtErrorDisplay>(*this), // Error Display
nullptr, // Parental Controls
nullptr, // Photo Viewer
@@ -1107,6 +1106,11 @@ void GMainWindow::BootGame(const QString& filename, std::size_t program_index) {
ConfigureVibration::SetAllVibrationDevices();
+ // Save configurations
+ UpdateUISettings();
+ game_list->SaveInterfaceLayout();
+ config->Save();
+
Settings::LogSettings();
if (UISettings::values.select_user_on_boot) {
@@ -1240,9 +1244,7 @@ void GMainWindow::ShutdownGame() {
emu_frametime_label->setVisible(false);
async_status_button->setEnabled(true);
multicore_status_button->setEnabled(true);
-#ifdef HAS_VULKAN
renderer_status_button->setEnabled(true);
-#endif
emulation_running = false;
@@ -1529,7 +1531,7 @@ void GMainWindow::RemoveAddOnContent(u64 program_id, const QString& entry_type)
FileSys::TitleType::AOC, FileSys::ContentRecordType::Data);
for (const auto& entry : dlc_entries) {
- if ((entry.title_id & DLC_BASE_TITLE_ID_MASK) == program_id) {
+ if (FileSys::GetBaseTitleID(entry.title_id) == program_id) {
const auto res =
fs_controller.GetUserNANDContents()->RemoveExistingEntry(entry.title_id) ||
fs_controller.GetSDMCContents()->RemoveExistingEntry(entry.title_id);
@@ -2103,11 +2105,12 @@ void GMainWindow::OnStartGame() {
qRegisterMetaType<std::string>("std::string");
qRegisterMetaType<std::optional<std::u16string>>("std::optional<std::u16string>");
qRegisterMetaType<std::string_view>("std::string_view");
+ qRegisterMetaType<Service::AM::Applets::WebExitReason>("Service::AM::Applets::WebExitReason");
connect(emu_thread.get(), &EmuThread::ErrorThrown, this, &GMainWindow::OnCoreError);
ui.action_Start->setEnabled(false);
- ui.action_Start->setText(tr("Continue"));
+ ui.action_Start->setText(tr("&Continue"));
ui.action_Pause->setEnabled(true);
ui.action_Stop->setEnabled(true);
@@ -2251,7 +2254,7 @@ void GMainWindow::ToggleWindowMode() {
}
}
-void GMainWindow::ResetWindowSize() {
+void GMainWindow::ResetWindowSize720() {
const auto aspect_ratio = Layout::EmulationAspectRatio(
static_cast<Layout::AspectRatio>(Settings::values.aspect_ratio.GetValue()),
static_cast<float>(Layout::ScreenUndocked::Height) / Layout::ScreenUndocked::Width);
@@ -2265,6 +2268,20 @@ void GMainWindow::ResetWindowSize() {
}
}
+void GMainWindow::ResetWindowSize1080() {
+ const auto aspect_ratio = Layout::EmulationAspectRatio(
+ static_cast<Layout::AspectRatio>(Settings::values.aspect_ratio.GetValue()),
+ static_cast<float>(Layout::ScreenDocked::Height) / Layout::ScreenDocked::Width);
+ if (!ui.action_Single_Window_Mode->isChecked()) {
+ render_window->resize(Layout::ScreenDocked::Height / aspect_ratio,
+ Layout::ScreenDocked::Height);
+ } else {
+ resize(Layout::ScreenDocked::Height / aspect_ratio,
+ Layout::ScreenDocked::Height + menuBar()->height() +
+ (ui.action_Show_Status_Bar->isChecked() ? statusBar()->height() : 0));
+ }
+}
+
void GMainWindow::OnConfigure() {
const auto old_theme = UISettings::values.theme;
const bool old_discord_presence = UISettings::values.enable_discord_presence;
@@ -2512,14 +2529,27 @@ void GMainWindow::UpdateStatusBar() {
void GMainWindow::UpdateStatusButtons() {
dock_status_button->setChecked(Settings::values.use_docked_mode.GetValue());
multicore_status_button->setChecked(Settings::values.use_multi_core.GetValue());
- Settings::values.use_asynchronous_gpu_emulation.SetValue(
- Settings::values.use_asynchronous_gpu_emulation.GetValue() ||
- Settings::values.use_multi_core.GetValue());
async_status_button->setChecked(Settings::values.use_asynchronous_gpu_emulation.GetValue());
-#ifdef HAS_VULKAN
renderer_status_button->setChecked(Settings::values.renderer_backend.GetValue() ==
Settings::RendererBackend::Vulkan);
+}
+
+void GMainWindow::UpdateUISettings() {
+ if (!ui.action_Fullscreen->isChecked()) {
+ UISettings::values.geometry = saveGeometry();
+ UISettings::values.renderwindow_geometry = render_window->saveGeometry();
+ }
+ UISettings::values.state = saveState();
+#if MICROPROFILE_ENABLED
+ UISettings::values.microprofile_geometry = microProfileDialog->saveGeometry();
+ UISettings::values.microprofile_visible = microProfileDialog->isVisible();
#endif
+ UISettings::values.single_window_mode = ui.action_Single_Window_Mode->isChecked();
+ UISettings::values.fullscreen = ui.action_Fullscreen->isChecked();
+ UISettings::values.display_titlebar = ui.action_Display_Dock_Widget_Headers->isChecked();
+ UISettings::values.show_filter_bar = ui.action_Show_Filter_Bar->isChecked();
+ UISettings::values.show_status_bar = ui.action_Show_Status_Bar->isChecked();
+ UISettings::values.first_start = false;
}
void GMainWindow::HideMouseCursor() {
@@ -2709,7 +2739,7 @@ std::optional<u64> GMainWindow::SelectRomFSDumpTarget(const FileSys::ContentProv
dlc_match.reserve(dlc_entries.size());
std::copy_if(dlc_entries.begin(), dlc_entries.end(), std::back_inserter(dlc_match),
[&program_id, &installed](const FileSys::ContentProviderEntry& entry) {
- return (entry.title_id & DLC_BASE_TITLE_ID_MASK) == program_id &&
+ return FileSys::GetBaseTitleID(entry.title_id) == program_id &&
installed.GetEntry(entry)->GetStatus() == Loader::ResultStatus::Success;
});
@@ -2755,22 +2785,7 @@ void GMainWindow::closeEvent(QCloseEvent* event) {
return;
}
- if (!ui.action_Fullscreen->isChecked()) {
- UISettings::values.geometry = saveGeometry();
- UISettings::values.renderwindow_geometry = render_window->saveGeometry();
- }
- UISettings::values.state = saveState();
-#if MICROPROFILE_ENABLED
- UISettings::values.microprofile_geometry = microProfileDialog->saveGeometry();
- UISettings::values.microprofile_visible = microProfileDialog->isVisible();
-#endif
- UISettings::values.single_window_mode = ui.action_Single_Window_Mode->isChecked();
- UISettings::values.fullscreen = ui.action_Fullscreen->isChecked();
- UISettings::values.display_titlebar = ui.action_Display_Dock_Widget_Headers->isChecked();
- UISettings::values.show_filter_bar = ui.action_Show_Filter_Bar->isChecked();
- UISettings::values.show_status_bar = ui.action_Show_Status_Bar->isChecked();
- UISettings::values.first_start = false;
-
+ UpdateUISettings();
game_list->SaveInterfaceLayout();
hotkey_registry.SaveHotkeys();
@@ -2946,7 +2961,7 @@ void GMainWindow::OnLanguageChanged(const QString& locale) {
UpdateWindowTitle();
if (emulation_running)
- ui.action_Start->setText(tr("Continue"));
+ ui.action_Start->setText(tr("&Continue"));
}
void GMainWindow::SetDiscordEnabled([[maybe_unused]] bool state) {
diff --git a/src/yuzu/main.h b/src/yuzu/main.h
index 6242341d1..ea6d2c30d 100644
--- a/src/yuzu/main.h
+++ b/src/yuzu/main.h
@@ -55,6 +55,10 @@ namespace InputCommon {
class InputSubsystem;
}
+namespace Service::AM::Applets {
+enum class WebExitReason : u32;
+}
+
enum class EmulatedDirectoryTarget {
NAND,
SDMC,
@@ -126,8 +130,8 @@ signals:
void SoftwareKeyboardFinishedText(std::optional<std::u16string> text);
void SoftwareKeyboardFinishedCheckDialog();
- void WebBrowserUnpackRomFS();
- void WebBrowserFinishedBrowsing();
+ void WebBrowserExtractOfflineRomFS();
+ void WebBrowserClosed(Service::AM::Applets::WebExitReason exit_reason, std::string last_url);
public slots:
void OnLoadComplete();
@@ -138,7 +142,8 @@ public slots:
void ProfileSelectorSelectProfile();
void SoftwareKeyboardGetText(const Core::Frontend::SoftwareKeyboardParameters& parameters);
void SoftwareKeyboardInvokeCheckDialog(std::u16string error_message);
- void WebBrowserOpenPage(std::string_view filename, std::string_view arguments);
+ void WebBrowserOpenWebPage(std::string_view main_url, std::string_view additional_args,
+ bool is_local);
void OnAppFocusStateChanged(Qt::ApplicationState state);
private:
@@ -237,7 +242,8 @@ private slots:
void ShowFullscreen();
void HideFullscreen();
void ToggleWindowMode();
- void ResetWindowSize();
+ void ResetWindowSize720();
+ void ResetWindowSize1080();
void OnCaptureScreenshot();
void OnCoreError(Core::System::ResultStatus, std::string);
void OnReinitializeKeys(ReinitializeKeyBehavior behavior);
@@ -257,6 +263,7 @@ private:
const std::string& title_version = {});
void UpdateStatusBar();
void UpdateStatusButtons();
+ void UpdateUISettings();
void HideMouseCursor();
void ShowMouseCursor();
void OpenURL(const QUrl& url);
@@ -321,6 +328,9 @@ private:
// Last game booted, used for multi-process apps
QString last_filename_booted;
+ // Disables the web applet for the rest of the emulated session
+ bool disable_web_applet{};
+
protected:
void dropEvent(QDropEvent* event) override;
void dragEnterEvent(QDragEnterEvent* event) override;
diff --git a/src/yuzu/main.ui b/src/yuzu/main.ui
index 2f3792247..e2ad5baf6 100644
--- a/src/yuzu/main.ui
+++ b/src/yuzu/main.ui
@@ -25,16 +25,7 @@
</property>
<widget class="QWidget" name="centralwidget">
<layout class="QHBoxLayout" name="horizontalLayout">
- <property name="leftMargin">
- <number>0</number>
- </property>
- <property name="topMargin">
- <number>0</number>
- </property>
- <property name="rightMargin">
- <number>0</number>
- </property>
- <property name="bottomMargin">
+ <property name="margin">
<number>0</number>
</property>
</layout>
@@ -45,7 +36,7 @@
<x>0</x>
<y>0</y>
<width>1280</width>
- <height>21</height>
+ <height>26</height>
</rect>
</property>
<widget class="QMenu" name="menu_File">
@@ -54,7 +45,7 @@
</property>
<widget class="QMenu" name="menu_recent_files">
<property name="title">
- <string>Recent Files</string>
+ <string>&amp;Recent Files</string>
</property>
</widget>
<addaction name="action_Install_File_NAND"/>
@@ -89,7 +80,7 @@
</property>
<widget class="QMenu" name="menu_View_Debugging">
<property name="title">
- <string>Debugging</string>
+ <string>&amp;Debugging</string>
</property>
</widget>
<addaction name="action_Fullscreen"/>
@@ -97,13 +88,14 @@
<addaction name="action_Display_Dock_Widget_Headers"/>
<addaction name="action_Show_Filter_Bar"/>
<addaction name="action_Show_Status_Bar"/>
- <addaction name="action_Reset_Window_Size"/>
+ <addaction name="action_Reset_Window_Size_720"/>
+ <addaction name="action_Reset_Window_Size_1080"/>
<addaction name="separator"/>
<addaction name="menu_View_Debugging"/>
</widget>
<widget class="QMenu" name="menu_Tools">
<property name="title">
- <string>Tools</string>
+ <string>&amp;Tools</string>
</property>
<addaction name="action_Rederive"/>
<addaction name="separator"/>
@@ -131,17 +123,17 @@
<bool>true</bool>
</property>
<property name="text">
- <string>Install Files to NAND...</string>
+ <string>&amp;Install Files to NAND...</string>
</property>
</action>
<action name="action_Load_File">
<property name="text">
- <string>Load File...</string>
+ <string>L&amp;oad File...</string>
</property>
</action>
<action name="action_Load_Folder">
<property name="text">
- <string>Load Folder...</string>
+ <string>Load &amp;Folder...</string>
</property>
</action>
<action name="action_Exit">
@@ -175,12 +167,12 @@
</action>
<action name="action_Rederive">
<property name="text">
- <string>Reinitialize keys...</string>
+ <string>&amp;Reinitialize keys...</string>
</property>
</action>
<action name="action_About">
<property name="text">
- <string>About yuzu</string>
+ <string>&amp;About yuzu</string>
</property>
</action>
<action name="action_Single_Window_Mode">
@@ -188,12 +180,12 @@
<bool>true</bool>
</property>
<property name="text">
- <string>Single Window Mode</string>
+ <string>Single &amp;Window Mode</string>
</property>
</action>
<action name="action_Configure">
<property name="text">
- <string>Configure...</string>
+ <string>Con&amp;figure...</string>
</property>
</action>
<action name="action_Display_Dock_Widget_Headers">
@@ -201,7 +193,7 @@
<bool>true</bool>
</property>
<property name="text">
- <string>Display Dock Widget Headers</string>
+ <string>Display D&amp;ock Widget Headers</string>
</property>
</action>
<action name="action_Show_Filter_Bar">
@@ -209,7 +201,7 @@
<bool>true</bool>
</property>
<property name="text">
- <string>Show Filter Bar</string>
+ <string>Show &amp;Filter Bar</string>
</property>
</action>
<action name="action_Show_Status_Bar">
@@ -217,12 +209,26 @@
<bool>true</bool>
</property>
<property name="text">
+ <string>Show &amp;Status Bar</string>
+ </property>
+ <property name="iconText">
<string>Show Status Bar</string>
</property>
</action>
- <action name="action_Reset_Window_Size">
+ <action name="action_Reset_Window_Size_720">
+ <property name="text">
+ <string>Reset Window Size to &amp;720p</string>
+ </property>
+ <property name="iconText">
+ <string>Reset Window Size to 720p</string>
+ </property>
+ </action>
+ <action name="action_Reset_Window_Size_1080">
<property name="text">
- <string>Reset Window Size</string>
+ <string>Reset Window Size to &amp;1080p</string>
+ </property>
+ <property name="iconText">
+ <string>Reset Window Size to 1080p</string>
</property>
</action>
<action name="action_Fullscreen">
@@ -230,7 +236,7 @@
<bool>true</bool>
</property>
<property name="text">
- <string>Fullscreen</string>
+ <string>F&amp;ullscreen</string>
</property>
</action>
<action name="action_Restart">
@@ -238,7 +244,7 @@
<bool>false</bool>
</property>
<property name="text">
- <string>Restart</string>
+ <string>&amp;Restart</string>
</property>
</action>
<action name="action_Load_Amiibo">
@@ -246,7 +252,7 @@
<bool>false</bool>
</property>
<property name="text">
- <string>Load Amiibo...</string>
+ <string>Load &amp;Amiibo...</string>
</property>
</action>
<action name="action_Report_Compatibility">
@@ -254,7 +260,7 @@
<bool>false</bool>
</property>
<property name="text">
- <string>Report Compatibility</string>
+ <string>&amp;Report Compatibility</string>
</property>
<property name="visible">
<bool>false</bool>
@@ -262,22 +268,22 @@
</action>
<action name="action_Open_Mods_Page">
<property name="text">
- <string>Open Mods Page</string>
+ <string>Open &amp;Mods Page</string>
</property>
</action>
<action name="action_Open_Quickstart_Guide">
<property name="text">
- <string>Open Quickstart Guide</string>
+ <string>Open &amp;Quickstart Guide</string>
</property>
</action>
<action name="action_Open_FAQ">
<property name="text">
- <string>FAQ</string>
+ <string>&amp;FAQ</string>
</property>
</action>
<action name="action_Open_yuzu_Folder">
<property name="text">
- <string>Open yuzu Folder</string>
+ <string>Open &amp;yuzu Folder</string>
</property>
</action>
<action name="action_Capture_Screenshot">
@@ -285,7 +291,7 @@
<bool>false</bool>
</property>
<property name="text">
- <string>Capture Screenshot</string>
+ <string>&amp;Capture Screenshot</string>
</property>
</action>
<action name="action_Configure_Current_Game">
@@ -293,7 +299,7 @@
<bool>false</bool>
</property>
<property name="text">
- <string>Configure Current Game...</string>
+ <string>Configure C&amp;urrent Game...</string>
</property>
</action>
</widget>
diff --git a/src/yuzu/util/url_request_interceptor.cpp b/src/yuzu/util/url_request_interceptor.cpp
new file mode 100644
index 000000000..2d491d8c0
--- /dev/null
+++ b/src/yuzu/util/url_request_interceptor.cpp
@@ -0,0 +1,32 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#ifdef YUZU_USE_QT_WEB_ENGINE
+
+#include "yuzu/util/url_request_interceptor.h"
+
+UrlRequestInterceptor::UrlRequestInterceptor(QObject* p) : QWebEngineUrlRequestInterceptor(p) {}
+
+UrlRequestInterceptor::~UrlRequestInterceptor() = default;
+
+void UrlRequestInterceptor::interceptRequest(QWebEngineUrlRequestInfo& info) {
+ const auto resource_type = info.resourceType();
+
+ switch (resource_type) {
+ case QWebEngineUrlRequestInfo::ResourceTypeMainFrame:
+ requested_url = info.requestUrl();
+ emit FrameChanged();
+ break;
+ case QWebEngineUrlRequestInfo::ResourceTypeSubFrame:
+ case QWebEngineUrlRequestInfo::ResourceTypeXhr:
+ emit FrameChanged();
+ break;
+ }
+}
+
+QUrl UrlRequestInterceptor::GetRequestedURL() const {
+ return requested_url;
+}
+
+#endif
diff --git a/src/yuzu/util/url_request_interceptor.h b/src/yuzu/util/url_request_interceptor.h
new file mode 100644
index 000000000..8a7f7499f
--- /dev/null
+++ b/src/yuzu/util/url_request_interceptor.h
@@ -0,0 +1,30 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#ifdef YUZU_USE_QT_WEB_ENGINE
+
+#include <QObject>
+#include <QWebEngineUrlRequestInterceptor>
+
+class UrlRequestInterceptor : public QWebEngineUrlRequestInterceptor {
+ Q_OBJECT
+
+public:
+ explicit UrlRequestInterceptor(QObject* p = nullptr);
+ ~UrlRequestInterceptor() override;
+
+ void interceptRequest(QWebEngineUrlRequestInfo& info) override;
+
+ QUrl GetRequestedURL() const;
+
+signals:
+ void FrameChanged();
+
+private:
+ QUrl requested_url;
+};
+
+#endif
diff --git a/src/yuzu_cmd/CMakeLists.txt b/src/yuzu_cmd/CMakeLists.txt
index 57f9916f6..0b3f2cb54 100644
--- a/src/yuzu_cmd/CMakeLists.txt
+++ b/src/yuzu_cmd/CMakeLists.txt
@@ -4,26 +4,17 @@ add_executable(yuzu-cmd
config.cpp
config.h
default_ini.h
- emu_window/emu_window_sdl2_gl.cpp
- emu_window/emu_window_sdl2_gl.h
emu_window/emu_window_sdl2.cpp
emu_window/emu_window_sdl2.h
emu_window/emu_window_sdl2_gl.cpp
emu_window/emu_window_sdl2_gl.h
+ emu_window/emu_window_sdl2_vk.cpp
+ emu_window/emu_window_sdl2_vk.h
resource.h
yuzu.cpp
yuzu.rc
)
-if (ENABLE_VULKAN)
- target_sources(yuzu-cmd PRIVATE
- emu_window/emu_window_sdl2_vk.cpp
- emu_window/emu_window_sdl2_vk.h)
-
- target_include_directories(yuzu-cmd PRIVATE ../../externals/Vulkan-Headers/include)
- target_compile_definitions(yuzu-cmd PRIVATE HAS_VULKAN)
-endif()
-
create_target_directory_groups(yuzu-cmd)
target_link_libraries(yuzu-cmd PRIVATE common core input_common)
@@ -33,6 +24,8 @@ if (MSVC)
endif()
target_link_libraries(yuzu-cmd PRIVATE ${PLATFORM_LIBRARIES} SDL2 Threads::Threads)
+target_include_directories(yuzu-cmd PRIVATE ../../externals/Vulkan-Headers/include)
+
if(UNIX AND NOT APPLE)
install(TARGETS yuzu-cmd RUNTIME DESTINATION "${CMAKE_INSTALL_PREFIX}/bin")
endif()
diff --git a/src/yuzu_cmd/config.cpp b/src/yuzu_cmd/config.cpp
index 34c9673bc..38075c345 100644
--- a/src/yuzu_cmd/config.cpp
+++ b/src/yuzu_cmd/config.cpp
@@ -345,7 +345,6 @@ void Config::ReadValues() {
// System
Settings::values.use_docked_mode.SetValue(
sdl2_config->GetBoolean("System", "use_docked_mode", false));
- const auto size = sdl2_config->GetInteger("System", "users_size", 0);
Settings::values.current_user = std::clamp<int>(
sdl2_config->GetInteger("System", "current_user", 0), 0, Service::Account::MAX_USERS - 1);
@@ -430,9 +429,6 @@ void Config::ReadValues() {
// Debugging
Settings::values.record_frame_times =
sdl2_config->GetBoolean("Debugging", "record_frame_times", false);
- Settings::values.use_gdbstub = sdl2_config->GetBoolean("Debugging", "use_gdbstub", false);
- Settings::values.gdbstub_port =
- static_cast<u16>(sdl2_config->GetInteger("Debugging", "gdbstub_port", 24689));
Settings::values.program_args = sdl2_config->Get("Debugging", "program_args", "");
Settings::values.dump_exefs = sdl2_config->GetBoolean("Debugging", "dump_exefs", false);
Settings::values.dump_nso = sdl2_config->GetBoolean("Debugging", "dump_nso", false);
diff --git a/src/yuzu_cmd/default_ini.h b/src/yuzu_cmd/default_ini.h
index bcbbcd4ca..2d4b98d9a 100644
--- a/src/yuzu_cmd/default_ini.h
+++ b/src/yuzu_cmd/default_ini.h
@@ -318,9 +318,6 @@ log_filter = *:Trace
[Debugging]
# Record frame time data, can be found in the log directory. Boolean value
record_frame_times =
-# Port for listening to GDB connections.
-use_gdbstub=false
-gdbstub_port=24689
# Determines whether or not yuzu will dump the ExeFS of all games it attempts to load while loading them
dump_exefs=false
# Determines whether or not yuzu will dump all NSOs it attempts to load while loading them
diff --git a/src/yuzu_cmd/yuzu.cpp b/src/yuzu_cmd/yuzu.cpp
index c2efe1ee6..2497c71ae 100644
--- a/src/yuzu_cmd/yuzu.cpp
+++ b/src/yuzu_cmd/yuzu.cpp
@@ -35,9 +35,7 @@
#include "yuzu_cmd/config.h"
#include "yuzu_cmd/emu_window/emu_window_sdl2.h"
#include "yuzu_cmd/emu_window/emu_window_sdl2_gl.h"
-#ifdef HAS_VULKAN
#include "yuzu_cmd/emu_window/emu_window_sdl2_vk.h"
-#endif
#ifdef _WIN32
// windows.h needs to be included before shellapi.h
@@ -64,7 +62,6 @@ __declspec(dllexport) int AmdPowerXpressRequestHighPerformance = 1;
static void PrintHelp(const char* argv0) {
std::cout << "Usage: " << argv0
<< " [options] <filename>\n"
- "-g, --gdbport=NUMBER Enable gdb stub on port NUMBER\n"
"-f, --fullscreen Start in fullscreen mode\n"
"-h, --help Display this help and exit\n"
"-v, --version Output version information and exit\n"
@@ -96,8 +93,6 @@ int main(int argc, char** argv) {
Config config;
int option_index = 0;
- bool use_gdbstub = Settings::values.use_gdbstub;
- u32 gdb_port = static_cast<u32>(Settings::values.gdbstub_port);
InitializeLogging();
@@ -116,26 +111,17 @@ int main(int argc, char** argv) {
bool fullscreen = false;
static struct option long_options[] = {
- {"gdbport", required_argument, 0, 'g'}, {"fullscreen", no_argument, 0, 'f'},
- {"help", no_argument, 0, 'h'}, {"version", no_argument, 0, 'v'},
- {"program", optional_argument, 0, 'p'}, {0, 0, 0, 0},
+ {"fullscreen", no_argument, 0, 'f'},
+ {"help", no_argument, 0, 'h'},
+ {"version", no_argument, 0, 'v'},
+ {"program", optional_argument, 0, 'p'},
+ {0, 0, 0, 0},
};
while (optind < argc) {
int arg = getopt_long(argc, argv, "g:fhvp::", long_options, &option_index);
if (arg != -1) {
switch (static_cast<char>(arg)) {
- case 'g':
- errno = 0;
- gdb_port = strtoul(optarg, &endarg, 0);
- use_gdbstub = true;
- if (endarg == optarg)
- errno = EINVAL;
- if (errno != 0) {
- perror("--gdbport");
- exit(1);
- }
- break;
case 'f':
fullscreen = true;
LOG_INFO(Frontend, "Starting in fullscreen mode...");
@@ -177,8 +163,6 @@ int main(int argc, char** argv) {
InputCommon::InputSubsystem input_subsystem;
// Apply the command line arguments
- Settings::values.gdbstub_port = gdb_port;
- Settings::values.use_gdbstub = use_gdbstub;
Settings::Apply(system);
std::unique_ptr<EmuWindow_SDL2> emu_window;
@@ -187,13 +171,8 @@ int main(int argc, char** argv) {
emu_window = std::make_unique<EmuWindow_SDL2_GL>(&input_subsystem, fullscreen);
break;
case Settings::RendererBackend::Vulkan:
-#ifdef HAS_VULKAN
emu_window = std::make_unique<EmuWindow_SDL2_VK>(&input_subsystem);
break;
-#else
- LOG_CRITICAL(Frontend, "Vulkan backend has not been compiled!");
- return 1;
-#endif
}
system.SetContentProvider(std::make_unique<FileSys::ContentProviderUnion>());
diff --git a/src/yuzu_tester/config.cpp b/src/yuzu_tester/config.cpp
index b6cdc7c1c..91684e96e 100644
--- a/src/yuzu_tester/config.cpp
+++ b/src/yuzu_tester/config.cpp
@@ -158,7 +158,6 @@ void Config::ReadValues() {
Settings::values.use_dev_keys = sdl2_config->GetBoolean("Miscellaneous", "use_dev_keys", false);
// Debugging
- Settings::values.use_gdbstub = false;
Settings::values.program_args = "";
Settings::values.dump_exefs = sdl2_config->GetBoolean("Debugging", "dump_exefs", false);
Settings::values.dump_nso = sdl2_config->GetBoolean("Debugging", "dump_nso", false);
diff --git a/src/yuzu_tester/yuzu.cpp b/src/yuzu_tester/yuzu.cpp
index 50bd7ae41..6435ffabb 100644
--- a/src/yuzu_tester/yuzu.cpp
+++ b/src/yuzu_tester/yuzu.cpp
@@ -162,7 +162,6 @@ int main(int argc, char** argv) {
Core::System& system{Core::System::GetInstance()};
- Settings::values.use_gdbstub = false;
Settings::Apply(system);
const auto emu_window{std::make_unique<EmuWindow_SDL2_Hide>()};