diff options
| -rw-r--r-- | src/citra_qt/main.cpp | 5 | ||||
| -rw-r--r-- | src/core/file_sys/archive_backend.h | 7 | ||||
| -rw-r--r-- | src/core/file_sys/disk_archive.cpp | 4 | ||||
| -rw-r--r-- | src/core/file_sys/disk_archive.h | 1 | ||||
| -rw-r--r-- | src/core/file_sys/ivfc_archive.cpp | 6 | ||||
| -rw-r--r-- | src/core/file_sys/ivfc_archive.h | 1 | ||||
| -rw-r--r-- | src/core/hle/service/fs/archive.cpp | 12 | ||||
| -rw-r--r-- | src/core/hle/service/fs/archive.h | 9 | ||||
| -rw-r--r-- | src/core/hle/service/fs/fs_user.cpp | 75 | ||||
| -rw-r--r-- | src/core/hw/gpu.cpp | 573 | ||||
| -rw-r--r-- | src/core/memory.cpp | 3 | ||||
| -rw-r--r-- | src/video_core/rasterizer_interface.h | 7 | ||||
| -rw-r--r-- | src/video_core/renderer_opengl/gl_rasterizer.cpp | 52 | ||||
| -rw-r--r-- | src/video_core/renderer_opengl/gl_rasterizer.h | 4 | ||||
| -rw-r--r-- | src/video_core/renderer_opengl/gl_shader_gen.cpp | 9 |
15 files changed, 468 insertions, 300 deletions
diff --git a/src/citra_qt/main.cpp b/src/citra_qt/main.cpp index 929392b14..8322e2305 100644 --- a/src/citra_qt/main.cpp +++ b/src/citra_qt/main.cpp @@ -295,7 +295,6 @@ bool GMainWindow::LoadROM(const std::string& filename) { Loader::ResultStatus result = app_loader->Load(); if (Loader::ResultStatus::Success != result) { LOG_CRITICAL(Frontend, "Failed to load ROM!"); - System::Shutdown(); switch (result) { case Loader::ResultStatus::ErrorEncrypted: { @@ -335,8 +334,10 @@ void GMainWindow::BootGame(const std::string& filename) { if (!InitializeSystem()) return; - if (!LoadROM(filename)) + if (!LoadROM(filename)) { + System::Shutdown(); return; + } // Create and start the emulation thread emu_thread = std::make_unique<EmuThread>(render_window); diff --git a/src/core/file_sys/archive_backend.h b/src/core/file_sys/archive_backend.h index d69c3c785..06b8f2ed7 100644 --- a/src/core/file_sys/archive_backend.h +++ b/src/core/file_sys/archive_backend.h @@ -112,6 +112,13 @@ public: virtual bool DeleteDirectory(const Path& path) const = 0; /** + * Delete a directory specified by its path and anything under it + * @param path Path relative to the archive + * @return Whether the directory could be deleted + */ + virtual bool DeleteDirectoryRecursively(const Path& path) const = 0; + + /** * Create a file specified by its path * @param path Path relative to the Archive * @param size The size of the new file, filled with zeroes diff --git a/src/core/file_sys/disk_archive.cpp b/src/core/file_sys/disk_archive.cpp index 0f66998e1..2f05af361 100644 --- a/src/core/file_sys/disk_archive.cpp +++ b/src/core/file_sys/disk_archive.cpp @@ -51,6 +51,10 @@ bool DiskArchive::DeleteDirectory(const Path& path) const { return FileUtil::DeleteDir(mount_point + path.AsString()); } +bool DiskArchive::DeleteDirectoryRecursively(const Path& path) const { + return FileUtil::DeleteDirRecursively(mount_point + path.AsString()); +} + ResultCode DiskArchive::CreateFile(const FileSys::Path& path, u64 size) const { std::string full_path = mount_point + path.AsString(); diff --git a/src/core/file_sys/disk_archive.h b/src/core/file_sys/disk_archive.h index 2165f27f9..59ebb2002 100644 --- a/src/core/file_sys/disk_archive.h +++ b/src/core/file_sys/disk_archive.h @@ -38,6 +38,7 @@ public: ResultCode DeleteFile(const Path& path) const override; bool RenameFile(const Path& src_path, const Path& dest_path) const override; bool DeleteDirectory(const Path& path) const override; + bool DeleteDirectoryRecursively(const Path& path) const override; ResultCode CreateFile(const Path& path, u64 size) const override; bool CreateDirectory(const Path& path) const override; bool RenameDirectory(const Path& src_path, const Path& dest_path) const override; diff --git a/src/core/file_sys/ivfc_archive.cpp b/src/core/file_sys/ivfc_archive.cpp index 49cc1de10..af59d296d 100644 --- a/src/core/file_sys/ivfc_archive.cpp +++ b/src/core/file_sys/ivfc_archive.cpp @@ -43,6 +43,12 @@ bool IVFCArchive::DeleteDirectory(const Path& path) const { return false; } +bool IVFCArchive::DeleteDirectoryRecursively(const Path& path) const { + LOG_CRITICAL(Service_FS, "Attempted to delete a directory from an IVFC archive (%s).", + GetName().c_str()); + return false; +} + ResultCode IVFCArchive::CreateFile(const Path& path, u64 size) const { LOG_CRITICAL(Service_FS, "Attempted to create a file in an IVFC archive (%s).", GetName().c_str()); diff --git a/src/core/file_sys/ivfc_archive.h b/src/core/file_sys/ivfc_archive.h index 0df6cf83a..2fbb3a568 100644 --- a/src/core/file_sys/ivfc_archive.h +++ b/src/core/file_sys/ivfc_archive.h @@ -37,6 +37,7 @@ public: ResultCode DeleteFile(const Path& path) const override; bool RenameFile(const Path& src_path, const Path& dest_path) const override; bool DeleteDirectory(const Path& path) const override; + bool DeleteDirectoryRecursively(const Path& path) const override; ResultCode CreateFile(const Path& path, u64 size) const override; bool CreateDirectory(const Path& path) const override; bool RenameDirectory(const Path& src_path, const Path& dest_path) const override; diff --git a/src/core/hle/service/fs/archive.cpp b/src/core/hle/service/fs/archive.cpp index 4dc7e1e3c..7f9696bfb 100644 --- a/src/core/hle/service/fs/archive.cpp +++ b/src/core/hle/service/fs/archive.cpp @@ -362,6 +362,18 @@ ResultCode DeleteDirectoryFromArchive(ArchiveHandle archive_handle, const FileSy ErrorSummary::Canceled, ErrorLevel::Status); } +ResultCode DeleteDirectoryRecursivelyFromArchive(ArchiveHandle archive_handle, + const FileSys::Path& path) { + ArchiveBackend* archive = GetArchive(archive_handle); + if (archive == nullptr) + return ERR_INVALID_ARCHIVE_HANDLE; + + if (archive->DeleteDirectoryRecursively(path)) + return RESULT_SUCCESS; + return ResultCode(ErrorDescription::NoData, ErrorModule::FS, // TODO: verify description + ErrorSummary::Canceled, ErrorLevel::Status); +} + ResultCode CreateFileInArchive(ArchiveHandle archive_handle, const FileSys::Path& path, u64 file_size) { ArchiveBackend* archive = GetArchive(archive_handle); diff --git a/src/core/hle/service/fs/archive.h b/src/core/hle/service/fs/archive.h index 533be34a4..41a76285c 100644 --- a/src/core/hle/service/fs/archive.h +++ b/src/core/hle/service/fs/archive.h @@ -133,6 +133,15 @@ ResultCode RenameFileBetweenArchives(ArchiveHandle src_archive_handle, ResultCode DeleteDirectoryFromArchive(ArchiveHandle archive_handle, const FileSys::Path& path); /** + * Delete a Directory and anything under it from an Archive + * @param archive_handle Handle to an open Archive object + * @param path Path to the Directory inside of the Archive + * @return Whether deletion succeeded + */ +ResultCode DeleteDirectoryRecursivelyFromArchive(ArchiveHandle archive_handle, + const FileSys::Path& path); + +/** * Create a File in an Archive * @param archive_handle Handle to an open Archive object * @param path Path to the File inside of the Archive diff --git a/src/core/hle/service/fs/fs_user.cpp b/src/core/hle/service/fs/fs_user.cpp index 94f053dc2..00edc7622 100644 --- a/src/core/hle/service/fs/fs_user.cpp +++ b/src/core/hle/service/fs/fs_user.cpp @@ -64,7 +64,7 @@ static void OpenFile(Service::Interface* self) { u32 filename_ptr = cmd_buff[9]; FileSys::Path file_path(filename_type, filename_size, filename_ptr); - LOG_DEBUG(Service_FS, "path=%s, mode=%d attrs=%u", file_path.DebugStr().c_str(), mode.hex, + LOG_DEBUG(Service_FS, "path=%s, mode=%u attrs=%u", file_path.DebugStr().c_str(), mode.hex, attributes); ResultVal<SharedPtr<File>> file_res = OpenFileFromArchive(archive_handle, file_path, mode); @@ -112,15 +112,15 @@ static void OpenFileDirectly(Service::Interface* self) { FileSys::Path archive_path(archivename_type, archivename_size, archivename_ptr); FileSys::Path file_path(filename_type, filename_size, filename_ptr); - LOG_DEBUG(Service_FS, "archive_id=0x%08X archive_path=%s file_path=%s, mode=%u attributes=%d", - archive_id, archive_path.DebugStr().c_str(), file_path.DebugStr().c_str(), mode.hex, - attributes); + LOG_DEBUG(Service_FS, "archive_id=0x%08X archive_path=%s file_path=%s, mode=%u attributes=%u", + static_cast<u32>(archive_id), archive_path.DebugStr().c_str(), + file_path.DebugStr().c_str(), mode.hex, attributes); ResultVal<ArchiveHandle> archive_handle = OpenArchive(archive_id, archive_path); if (archive_handle.Failed()) { LOG_ERROR(Service_FS, "failed to get a handle for archive archive_id=0x%08X archive_path=%s", - archive_id, archive_path.DebugStr().c_str()); + static_cast<u32>(archive_id), archive_path.DebugStr().c_str()); cmd_buff[1] = archive_handle.Code().raw; cmd_buff[3] = 0; return; @@ -133,7 +133,7 @@ static void OpenFileDirectly(Service::Interface* self) { cmd_buff[3] = Kernel::g_handle_table.Create(*file_res).MoveFrom(); } else { cmd_buff[3] = 0; - LOG_ERROR(Service_FS, "failed to get a handle for file %s mode=%u attributes=%d", + LOG_ERROR(Service_FS, "failed to get a handle for file %s mode=%u attributes=%u", file_path.DebugStr().c_str(), mode.hex, attributes); } } @@ -159,7 +159,7 @@ static void DeleteFile(Service::Interface* self) { FileSys::Path file_path(filename_type, filename_size, filename_ptr); - LOG_DEBUG(Service_FS, "type=%d size=%d data=%s", filename_type, filename_size, + LOG_DEBUG(Service_FS, "type=%u size=%u data=%s", static_cast<u32>(filename_type), filename_size, file_path.DebugStr().c_str()); cmd_buff[1] = DeleteFileFromArchive(archive_handle, file_path).raw; @@ -198,9 +198,10 @@ static void RenameFile(Service::Interface* self) { FileSys::Path dest_file_path(dest_filename_type, dest_filename_size, dest_filename_ptr); LOG_DEBUG(Service_FS, - "src_type=%d src_size=%d src_data=%s dest_type=%d dest_size=%d dest_data=%s", - src_filename_type, src_filename_size, src_file_path.DebugStr().c_str(), - dest_filename_type, dest_filename_size, dest_file_path.DebugStr().c_str()); + "src_type=%u src_size=%u src_data=%s dest_type=%u dest_size=%u dest_data=%s", + static_cast<u32>(src_filename_type), src_filename_size, + src_file_path.DebugStr().c_str(), static_cast<u32>(dest_filename_type), + dest_filename_size, dest_file_path.DebugStr().c_str()); cmd_buff[1] = RenameFileBetweenArchives(src_archive_handle, src_file_path, dest_archive_handle, dest_file_path) @@ -228,13 +229,42 @@ static void DeleteDirectory(Service::Interface* self) { FileSys::Path dir_path(dirname_type, dirname_size, dirname_ptr); - LOG_DEBUG(Service_FS, "type=%d size=%d data=%s", dirname_type, dirname_size, + LOG_DEBUG(Service_FS, "type=%u size=%u data=%s", static_cast<u32>(dirname_type), dirname_size, dir_path.DebugStr().c_str()); cmd_buff[1] = DeleteDirectoryFromArchive(archive_handle, dir_path).raw; } /* + * FS_User::DeleteDirectoryRecursively service function + * Inputs: + * 0 : Command header 0x08070142 + * 1 : Transaction + * 2 : Archive handle lower word + * 3 : Archive handle upper word + * 4 : Directory path string type + * 5 : Directory path string size + * 7 : Directory path string data + * Outputs: + * 1 : Result of function, 0 on success, otherwise error code + */ +static void DeleteDirectoryRecursively(Service::Interface* self) { + u32* cmd_buff = Kernel::GetCommandBuffer(); + + ArchiveHandle archive_handle = MakeArchiveHandle(cmd_buff[2], cmd_buff[3]); + auto dirname_type = static_cast<FileSys::LowPathType>(cmd_buff[4]); + u32 dirname_size = cmd_buff[5]; + u32 dirname_ptr = cmd_buff[7]; + + FileSys::Path dir_path(dirname_type, dirname_size, dirname_ptr); + + LOG_DEBUG(Service_FS, "type=%u size=%u data=%s", static_cast<u32>(dirname_type), dirname_size, + dir_path.DebugStr().c_str()); + + cmd_buff[1] = DeleteDirectoryRecursivelyFromArchive(archive_handle, dir_path).raw; +} + +/* * FS_User::CreateFile service function * Inputs: * 0 : Command header 0x08080202 @@ -258,7 +288,7 @@ static void CreateFile(Service::Interface* self) { FileSys::Path file_path(filename_type, filename_size, filename_ptr); - LOG_DEBUG(Service_FS, "type=%d size=%llu data=%s", filename_type, file_size, + LOG_DEBUG(Service_FS, "type=%u size=%llu data=%s", static_cast<u32>(filename_type), file_size, file_path.DebugStr().c_str()); cmd_buff[1] = CreateFileInArchive(archive_handle, file_path, file_size).raw; @@ -285,7 +315,7 @@ static void CreateDirectory(Service::Interface* self) { FileSys::Path dir_path(dirname_type, dirname_size, dirname_ptr); - LOG_DEBUG(Service_FS, "type=%d size=%d data=%s", dirname_type, dirname_size, + LOG_DEBUG(Service_FS, "type=%u size=%d data=%s", static_cast<u32>(dirname_type), dirname_size, dir_path.DebugStr().c_str()); cmd_buff[1] = CreateDirectoryFromArchive(archive_handle, dir_path).raw; @@ -322,10 +352,10 @@ static void RenameDirectory(Service::Interface* self) { FileSys::Path src_dir_path(src_dirname_type, src_dirname_size, src_dirname_ptr); FileSys::Path dest_dir_path(dest_dirname_type, dest_dirname_size, dest_dirname_ptr); - LOG_DEBUG(Service_FS, - "src_type=%d src_size=%d src_data=%s dest_type=%d dest_size=%d dest_data=%s", - src_dirname_type, src_dirname_size, src_dir_path.DebugStr().c_str(), - dest_dirname_type, dest_dirname_size, dest_dir_path.DebugStr().c_str()); + LOG_DEBUG( + Service_FS, "src_type=%u src_size=%u src_data=%s dest_type=%u dest_size=%u dest_data=%s", + static_cast<u32>(src_dirname_type), src_dirname_size, src_dir_path.DebugStr().c_str(), + static_cast<u32>(dest_dirname_type), dest_dirname_size, dest_dir_path.DebugStr().c_str()); cmd_buff[1] = RenameDirectoryBetweenArchives(src_archive_handle, src_dir_path, dest_archive_handle, dest_dir_path) @@ -355,7 +385,7 @@ static void OpenDirectory(Service::Interface* self) { FileSys::Path dir_path(dirname_type, dirname_size, dirname_ptr); - LOG_DEBUG(Service_FS, "type=%d size=%d data=%s", dirname_type, dirname_size, + LOG_DEBUG(Service_FS, "type=%u size=%u data=%s", static_cast<u32>(dirname_type), dirname_size, dir_path.DebugStr().c_str()); ResultVal<SharedPtr<Directory>> dir_res = OpenDirectoryFromArchive(archive_handle, dir_path); @@ -390,7 +420,7 @@ static void OpenArchive(Service::Interface* self) { u32 archivename_ptr = cmd_buff[5]; FileSys::Path archive_path(archivename_type, archivename_size, archivename_ptr); - LOG_DEBUG(Service_FS, "archive_id=0x%08X archive_path=%s", archive_id, + LOG_DEBUG(Service_FS, "archive_id=0x%08X archive_path=%s", static_cast<u32>(archive_id), archive_path.DebugStr().c_str()); ResultVal<ArchiveHandle> handle = OpenArchive(archive_id, archive_path); @@ -402,7 +432,7 @@ static void OpenArchive(Service::Interface* self) { cmd_buff[2] = cmd_buff[3] = 0; LOG_ERROR(Service_FS, "failed to get a handle for archive archive_id=0x%08X archive_path=%s", - archive_id, archive_path.DebugStr().c_str()); + static_cast<u32>(archive_id), archive_path.DebugStr().c_str()); } } @@ -485,7 +515,8 @@ static void FormatSaveData(Service::Interface* self) { LOG_DEBUG(Service_FS, "archive_path=%s", archive_path.DebugStr().c_str()); if (archive_id != FS::ArchiveIdCode::SaveData) { - LOG_ERROR(Service_FS, "tried to format an archive different than SaveData, %u", archive_id); + LOG_ERROR(Service_FS, "tried to format an archive different than SaveData, %u", + static_cast<u32>(archive_id)); cmd_buff[1] = ResultCode(ErrorDescription::FS_InvalidPath, ErrorModule::FS, ErrorSummary::InvalidArgument, ErrorLevel::Usage) .raw; @@ -868,7 +899,7 @@ const Interface::FunctionInfo FunctionTable[] = { {0x08040142, DeleteFile, "DeleteFile"}, {0x08050244, RenameFile, "RenameFile"}, {0x08060142, DeleteDirectory, "DeleteDirectory"}, - {0x08070142, nullptr, "DeleteDirectoryRecursively"}, + {0x08070142, DeleteDirectoryRecursively, "DeleteDirectoryRecursively"}, {0x08080202, CreateFile, "CreateFile"}, {0x08090182, CreateDirectory, "CreateDirectory"}, {0x080A0244, RenameDirectory, "RenameDirectory"}, diff --git a/src/core/hw/gpu.cpp b/src/core/hw/gpu.cpp index 0e6b91e3a..28cb97d8e 100644 --- a/src/core/hw/gpu.cpp +++ b/src/core/hw/gpu.cpp @@ -80,6 +80,319 @@ static Math::Vec4<u8> DecodePixel(Regs::PixelFormat input_format, const u8* src_ MICROPROFILE_DEFINE(GPU_DisplayTransfer, "GPU", "DisplayTransfer", MP_RGB(100, 100, 255)); MICROPROFILE_DEFINE(GPU_CmdlistProcessing, "GPU", "Cmdlist Processing", MP_RGB(100, 255, 100)); +static void MemoryFill(const Regs::MemoryFillConfig& config) { + const PAddr start_addr = config.GetStartAddress(); + const PAddr end_addr = config.GetEndAddress(); + + // TODO: do hwtest with these cases + if (!Memory::IsValidPhysicalAddress(start_addr)) { + LOG_CRITICAL(HW_GPU, "invalid start address 0x%08X", start_addr); + return; + } + + if (!Memory::IsValidPhysicalAddress(end_addr)) { + LOG_CRITICAL(HW_GPU, "invalid end address 0x%08X", end_addr); + return; + } + + if (end_addr <= start_addr) { + LOG_CRITICAL(HW_GPU, "invalid memory range from 0x%08X to 0x%08X", start_addr, end_addr); + return; + } + + u8* start = Memory::GetPhysicalPointer(start_addr); + u8* end = Memory::GetPhysicalPointer(end_addr); + + // TODO: Consider always accelerating and returning vector of + // regions that the accelerated fill did not cover to + // reduce/eliminate the fill that the cpu has to do. + // This would also mean that the flush below is not needed. + // Fill should first flush all surfaces that touch but are + // not completely within the fill range. + // Then fill all completely covered surfaces, and return the + // regions that were between surfaces or within the touching + // ones for cpu to manually fill here. + if (VideoCore::g_renderer->Rasterizer()->AccelerateFill(config)) + return; + + Memory::RasterizerFlushAndInvalidateRegion(config.GetStartAddress(), + config.GetEndAddress() - config.GetStartAddress()); + + if (config.fill_24bit) { + // fill with 24-bit values + for (u8* ptr = start; ptr < end; ptr += 3) { + ptr[0] = config.value_24bit_r; + ptr[1] = config.value_24bit_g; + ptr[2] = config.value_24bit_b; + } + } else if (config.fill_32bit) { + // fill with 32-bit values + if (end > start) { + u32 value = config.value_32bit; + size_t len = (end - start) / sizeof(u32); + for (size_t i = 0; i < len; ++i) + memcpy(&start[i * sizeof(u32)], &value, sizeof(u32)); + } + } else { + // fill with 16-bit values + u16 value_16bit = config.value_16bit.Value(); + for (u8* ptr = start; ptr < end; ptr += sizeof(u16)) + memcpy(ptr, &value_16bit, sizeof(u16)); + } +} + +static void DisplayTransfer(const Regs::DisplayTransferConfig& config) { + const PAddr src_addr = config.GetPhysicalInputAddress(); + const PAddr dst_addr = config.GetPhysicalOutputAddress(); + + // TODO: do hwtest with these cases + if (!Memory::IsValidPhysicalAddress(src_addr)) { + LOG_CRITICAL(HW_GPU, "invalid input address 0x%08X", src_addr); + return; + } + + if (!Memory::IsValidPhysicalAddress(dst_addr)) { + LOG_CRITICAL(HW_GPU, "invalid output address 0x%08X", dst_addr); + return; + } + + if (config.input_width == 0) { + LOG_CRITICAL(HW_GPU, "zero input width"); + return; + } + + if (config.input_height == 0) { + LOG_CRITICAL(HW_GPU, "zero input height"); + return; + } + + if (config.output_width == 0) { + LOG_CRITICAL(HW_GPU, "zero output width"); + return; + } + + if (config.output_height == 0) { + LOG_CRITICAL(HW_GPU, "zero output height"); + return; + } + + if (VideoCore::g_renderer->Rasterizer()->AccelerateDisplayTransfer(config)) + return; + + u8* src_pointer = Memory::GetPhysicalPointer(src_addr); + u8* dst_pointer = Memory::GetPhysicalPointer(dst_addr); + + if (config.scaling > config.ScaleXY) { + LOG_CRITICAL(HW_GPU, "Unimplemented display transfer scaling mode %u", + config.scaling.Value()); + UNIMPLEMENTED(); + return; + } + + if (config.input_linear && config.scaling != config.NoScale) { + LOG_CRITICAL(HW_GPU, "Scaling is only implemented on tiled input"); + UNIMPLEMENTED(); + return; + } + + int horizontal_scale = config.scaling != config.NoScale ? 1 : 0; + int vertical_scale = config.scaling == config.ScaleXY ? 1 : 0; + + u32 output_width = config.output_width >> horizontal_scale; + u32 output_height = config.output_height >> vertical_scale; + + u32 input_size = + config.input_width * config.input_height * GPU::Regs::BytesPerPixel(config.input_format); + u32 output_size = output_width * output_height * GPU::Regs::BytesPerPixel(config.output_format); + + Memory::RasterizerFlushRegion(config.GetPhysicalInputAddress(), input_size); + Memory::RasterizerFlushAndInvalidateRegion(config.GetPhysicalOutputAddress(), output_size); + + for (u32 y = 0; y < output_height; ++y) { + for (u32 x = 0; x < output_width; ++x) { + Math::Vec4<u8> src_color; + + // Calculate the [x,y] position of the input image + // based on the current output position and the scale + u32 input_x = x << horizontal_scale; + u32 input_y = y << vertical_scale; + + u32 output_y; + if (config.flip_vertically) { + // Flip the y value of the output data, + // we do this after calculating the [x,y] position of the input image + // to account for the scaling options. + output_y = output_height - y - 1; + } else { + output_y = y; + } + + u32 dst_bytes_per_pixel = GPU::Regs::BytesPerPixel(config.output_format); + u32 src_bytes_per_pixel = GPU::Regs::BytesPerPixel(config.input_format); + u32 src_offset; + u32 dst_offset; + + if (config.input_linear) { + if (!config.dont_swizzle) { + // Interpret the input as linear and the output as tiled + u32 coarse_y = output_y & ~7; + u32 stride = output_width * dst_bytes_per_pixel; + + src_offset = (input_x + input_y * config.input_width) * src_bytes_per_pixel; + dst_offset = VideoCore::GetMortonOffset(x, output_y, dst_bytes_per_pixel) + + coarse_y * stride; + } else { + // Both input and output are linear + src_offset = (input_x + input_y * config.input_width) * src_bytes_per_pixel; + dst_offset = (x + output_y * output_width) * dst_bytes_per_pixel; + } + } else { + if (!config.dont_swizzle) { + // Interpret the input as tiled and the output as linear + u32 coarse_y = input_y & ~7; + u32 stride = config.input_width * src_bytes_per_pixel; + + src_offset = VideoCore::GetMortonOffset(input_x, input_y, src_bytes_per_pixel) + + coarse_y * stride; + dst_offset = (x + output_y * output_width) * dst_bytes_per_pixel; + } else { + // Both input and output are tiled + u32 out_coarse_y = output_y & ~7; + u32 out_stride = output_width * dst_bytes_per_pixel; + + u32 in_coarse_y = input_y & ~7; + u32 in_stride = config.input_width * src_bytes_per_pixel; + + src_offset = VideoCore::GetMortonOffset(input_x, input_y, src_bytes_per_pixel) + + in_coarse_y * in_stride; + dst_offset = VideoCore::GetMortonOffset(x, output_y, dst_bytes_per_pixel) + + out_coarse_y * out_stride; + } + } + + const u8* src_pixel = src_pointer + src_offset; + src_color = DecodePixel(config.input_format, src_pixel); + if (config.scaling == config.ScaleX) { + Math::Vec4<u8> pixel = + DecodePixel(config.input_format, src_pixel + src_bytes_per_pixel); + src_color = ((src_color + pixel) / 2).Cast<u8>(); + } else if (config.scaling == config.ScaleXY) { + Math::Vec4<u8> pixel1 = + DecodePixel(config.input_format, src_pixel + 1 * src_bytes_per_pixel); + Math::Vec4<u8> pixel2 = + DecodePixel(config.input_format, src_pixel + 2 * src_bytes_per_pixel); + Math::Vec4<u8> pixel3 = + DecodePixel(config.input_format, src_pixel + 3 * src_bytes_per_pixel); + src_color = (((src_color + pixel1) + (pixel2 + pixel3)) / 4).Cast<u8>(); + } + + u8* dst_pixel = dst_pointer + dst_offset; + switch (config.output_format) { + case Regs::PixelFormat::RGBA8: + Color::EncodeRGBA8(src_color, dst_pixel); + break; + + case Regs::PixelFormat::RGB8: + Color::EncodeRGB8(src_color, dst_pixel); + break; + + case Regs::PixelFormat::RGB565: + Color::EncodeRGB565(src_color, dst_pixel); + break; + + case Regs::PixelFormat::RGB5A1: + Color::EncodeRGB5A1(src_color, dst_pixel); + break; + + case Regs::PixelFormat::RGBA4: + Color::EncodeRGBA4(src_color, dst_pixel); + break; + + default: + LOG_ERROR(HW_GPU, "Unknown destination framebuffer format %x", + config.output_format.Value()); + break; + } + } + } +} + +static void TextureCopy(const Regs::DisplayTransferConfig& config) { + const PAddr src_addr = config.GetPhysicalInputAddress(); + const PAddr dst_addr = config.GetPhysicalOutputAddress(); + + // TODO: do hwtest with these cases + if (!Memory::IsValidPhysicalAddress(src_addr)) { + LOG_CRITICAL(HW_GPU, "invalid input address 0x%08X", src_addr); + return; + } + + if (!Memory::IsValidPhysicalAddress(dst_addr)) { + LOG_CRITICAL(HW_GPU, "invalid output address 0x%08X", dst_addr); + return; + } + + if (config.texture_copy.input_width == 0) { + LOG_CRITICAL(HW_GPU, "zero input width"); + return; + } + + if (config.texture_copy.output_width == 0) { + LOG_CRITICAL(HW_GPU, "zero output width"); + return; + } + + if (config.texture_copy.size == 0) { + LOG_CRITICAL(HW_GPU, "zero size"); + return; + } + + if (VideoCore::g_renderer->Rasterizer()->AccelerateTextureCopy(config)) + return; + + u8* src_pointer = Memory::GetPhysicalPointer(src_addr); + u8* dst_pointer = Memory::GetPhysicalPointer(dst_addr); + + u32 input_width = config.texture_copy.input_width * 16; + u32 input_gap = config.texture_copy.input_gap * 16; + u32 output_width = config.texture_copy.output_width * 16; + u32 output_gap = config.texture_copy.output_gap * 16; + + size_t contiguous_input_size = + config.texture_copy.size / input_width * (input_width + input_gap); + Memory::RasterizerFlushRegion(config.GetPhysicalInputAddress(), + static_cast<u32>(contiguous_input_size)); + + size_t contiguous_output_size = + config.texture_copy.size / output_width * (output_width + output_gap); + Memory::RasterizerFlushAndInvalidateRegion(config.GetPhysicalOutputAddress(), + static_cast<u32>(contiguous_output_size)); + + u32 remaining_size = config.texture_copy.size; + u32 remaining_input = input_width; + u32 remaining_output = output_width; + while (remaining_size > 0) { + u32 copy_size = std::min({remaining_input, remaining_output, remaining_size}); + + std::memcpy(dst_pointer, src_pointer, copy_size); + src_pointer += copy_size; + dst_pointer += copy_size; + + remaining_input -= copy_size; + remaining_output -= copy_size; + remaining_size -= copy_size; + + if (remaining_input == 0) { + remaining_input = input_width; + src_pointer += input_gap; + } + if (remaining_output == 0) { + remaining_output = output_width; + dst_pointer += output_gap; + } + } +} + template <typename T> inline void Write(u32 addr, const T data) { addr -= HW::VADDR_GPU; @@ -102,50 +415,13 @@ inline void Write(u32 addr, const T data) { auto& config = g_regs.memory_fill_config[is_second_filler]; if (config.trigger) { - if (config.address_start) { // Some games pass invalid values here - u8* start = Memory::GetPhysicalPointer(config.GetStartAddress()); - u8* end = Memory::GetPhysicalPointer(config.GetEndAddress()); - - // TODO: Consider always accelerating and returning vector of - // regions that the accelerated fill did not cover to - // reduce/eliminate the fill that the cpu has to do. - // This would also mean that the flush below is not needed. - // Fill should first flush all surfaces that touch but are - // not completely within the fill range. - // Then fill all completely covered surfaces, and return the - // regions that were between surfaces or within the touching - // ones for cpu to manually fill here. - if (!VideoCore::g_renderer->Rasterizer()->AccelerateFill(config)) { - Memory::RasterizerFlushAndInvalidateRegion(config.GetStartAddress(), - config.GetEndAddress() - - config.GetStartAddress()); - - if (config.fill_24bit) { - // fill with 24-bit values - for (u8* ptr = start; ptr < end; ptr += 3) { - ptr[0] = config.value_24bit_r; - ptr[1] = config.value_24bit_g; - ptr[2] = config.value_24bit_b; - } - } else if (config.fill_32bit) { - // fill with 32-bit values - if (end > start) { - u32 value = config.value_32bit; - size_t len = (end - start) / sizeof(u32); - for (size_t i = 0; i < len; ++i) - memcpy(&start[i * sizeof(u32)], &value, sizeof(u32)); - } - } else { - // fill with 16-bit values - u16 value_16bit = config.value_16bit.Value(); - for (u8* ptr = start; ptr < end; ptr += sizeof(u16)) - memcpy(ptr, &value_16bit, sizeof(u16)); - } - } - - LOG_TRACE(HW_GPU, "MemoryFill from 0x%08x to 0x%08x", config.GetStartAddress(), - config.GetEndAddress()); + MemoryFill(config); + LOG_TRACE(HW_GPU, "MemoryFill from 0x%08x to 0x%08x", config.GetStartAddress(), + config.GetEndAddress()); + // It seems that it won't signal interrupt if "address_start" is zero. + // TODO: hwtest this + if (config.GetStartAddress() != 0) { if (!is_second_filler) { GSP_GPU::SignalInterrupt(GSP_GPU::InterruptId::PSC0); } else { @@ -171,207 +447,22 @@ inline void Write(u32 addr, const T data) { Pica::g_debug_context->OnEvent(Pica::DebugContext::Event::IncomingDisplayTransfer, nullptr); - if (!VideoCore::g_renderer->Rasterizer()->AccelerateDisplayTransfer(config)) { - u8* src_pointer = Memory::GetPhysicalPointer(config.GetPhysicalInputAddress()); - u8* dst_pointer = Memory::GetPhysicalPointer(config.GetPhysicalOutputAddress()); - - if (config.is_texture_copy) { - u32 input_width = config.texture_copy.input_width * 16; - u32 input_gap = config.texture_copy.input_gap * 16; - u32 output_width = config.texture_copy.output_width * 16; - u32 output_gap = config.texture_copy.output_gap * 16; - - size_t contiguous_input_size = - config.texture_copy.size / input_width * (input_width + input_gap); - Memory::RasterizerFlushRegion(config.GetPhysicalInputAddress(), - static_cast<u32>(contiguous_input_size)); - - size_t contiguous_output_size = - config.texture_copy.size / output_width * (output_width + output_gap); - Memory::RasterizerFlushAndInvalidateRegion( - config.GetPhysicalOutputAddress(), - static_cast<u32>(contiguous_output_size)); - - u32 remaining_size = config.texture_copy.size; - u32 remaining_input = input_width; - u32 remaining_output = output_width; - while (remaining_size > 0) { - u32 copy_size = - std::min({remaining_input, remaining_output, remaining_size}); - - std::memcpy(dst_pointer, src_pointer, copy_size); - src_pointer += copy_size; - dst_pointer += copy_size; - - remaining_input -= copy_size; - remaining_output -= copy_size; - remaining_size -= copy_size; - - if (remaining_input == 0) { - remaining_input = input_width; - src_pointer += input_gap; - } - if (remaining_output == 0) { - remaining_output = output_width; - dst_pointer += output_gap; - } - } - - LOG_TRACE( - HW_GPU, - "TextureCopy: 0x%X bytes from 0x%08X(%u+%u)-> 0x%08X(%u+%u), flags 0x%08X", - config.texture_copy.size, config.GetPhysicalInputAddress(), input_width, - input_gap, config.GetPhysicalOutputAddress(), output_width, output_gap, - config.flags); - - GSP_GPU::SignalInterrupt(GSP_GPU::InterruptId::PPF); - break; - } - - if (config.scaling > config.ScaleXY) { - LOG_CRITICAL(HW_GPU, "Unimplemented display transfer scaling mode %u", - config.scaling.Value()); - UNIMPLEMENTED(); - break; - } - - if (config.input_linear && config.scaling != config.NoScale) { - LOG_CRITICAL(HW_GPU, "Scaling is only implemented on tiled input"); - UNIMPLEMENTED(); - break; - } - - int horizontal_scale = config.scaling != config.NoScale ? 1 : 0; - int vertical_scale = config.scaling == config.ScaleXY ? 1 : 0; - - u32 output_width = config.output_width >> horizontal_scale; - u32 output_height = config.output_height >> vertical_scale; - - u32 input_size = config.input_width * config.input_height * - GPU::Regs::BytesPerPixel(config.input_format); - u32 output_size = - output_width * output_height * GPU::Regs::BytesPerPixel(config.output_format); - - Memory::RasterizerFlushRegion(config.GetPhysicalInputAddress(), input_size); - Memory::RasterizerFlushAndInvalidateRegion(config.GetPhysicalOutputAddress(), - output_size); - - for (u32 y = 0; y < output_height; ++y) { - for (u32 x = 0; x < output_width; ++x) { - Math::Vec4<u8> src_color; - - // Calculate the [x,y] position of the input image - // based on the current output position and the scale - u32 input_x = x << horizontal_scale; - u32 input_y = y << vertical_scale; - - if (config.flip_vertically) { - // Flip the y value of the output data, - // we do this after calculating the [x,y] position of the input image - // to account for the scaling options. - y = output_height - y - 1; - } - - u32 dst_bytes_per_pixel = GPU::Regs::BytesPerPixel(config.output_format); - u32 src_bytes_per_pixel = GPU::Regs::BytesPerPixel(config.input_format); - u32 src_offset; - u32 dst_offset; - - if (config.input_linear) { - if (!config.dont_swizzle) { - // Interpret the input as linear and the output as tiled - u32 coarse_y = y & ~7; - u32 stride = output_width * dst_bytes_per_pixel; - - src_offset = - (input_x + input_y * config.input_width) * src_bytes_per_pixel; - dst_offset = VideoCore::GetMortonOffset(x, y, dst_bytes_per_pixel) + - coarse_y * stride; - } else { - // Both input and output are linear - src_offset = - (input_x + input_y * config.input_width) * src_bytes_per_pixel; - dst_offset = (x + y * output_width) * dst_bytes_per_pixel; - } - } else { - if (!config.dont_swizzle) { - // Interpret the input as tiled and the output as linear - u32 coarse_y = input_y & ~7; - u32 stride = config.input_width * src_bytes_per_pixel; - - src_offset = VideoCore::GetMortonOffset(input_x, input_y, - src_bytes_per_pixel) + - coarse_y * stride; - dst_offset = (x + y * output_width) * dst_bytes_per_pixel; - } else { - // Both input and output are tiled - u32 out_coarse_y = y & ~7; - u32 out_stride = output_width * dst_bytes_per_pixel; - - u32 in_coarse_y = input_y & ~7; - u32 in_stride = config.input_width * src_bytes_per_pixel; - - src_offset = VideoCore::GetMortonOffset(input_x, input_y, - src_bytes_per_pixel) + - in_coarse_y * in_stride; - dst_offset = VideoCore::GetMortonOffset(x, y, dst_bytes_per_pixel) + - out_coarse_y * out_stride; - } - } - - const u8* src_pixel = src_pointer + src_offset; - src_color = DecodePixel(config.input_format, src_pixel); - if (config.scaling == config.ScaleX) { - Math::Vec4<u8> pixel = - DecodePixel(config.input_format, src_pixel + src_bytes_per_pixel); - src_color = ((src_color + pixel) / 2).Cast<u8>(); - } else if (config.scaling == config.ScaleXY) { - Math::Vec4<u8> pixel1 = DecodePixel( - config.input_format, src_pixel + 1 * src_bytes_per_pixel); - Math::Vec4<u8> pixel2 = DecodePixel( - config.input_format, src_pixel + 2 * src_bytes_per_pixel); - Math::Vec4<u8> pixel3 = DecodePixel( - config.input_format, src_pixel + 3 * src_bytes_per_pixel); - src_color = (((src_color + pixel1) + (pixel2 + pixel3)) / 4).Cast<u8>(); - } - - u8* dst_pixel = dst_pointer + dst_offset; - switch (config.output_format) { - case Regs::PixelFormat::RGBA8: - Color::EncodeRGBA8(src_color, dst_pixel); - break; - - case Regs::PixelFormat::RGB8: - Color::EncodeRGB8(src_color, dst_pixel); - break; - - case Regs::PixelFormat::RGB565: - Color::EncodeRGB565(src_color, dst_pixel); - break; - - case Regs::PixelFormat::RGB5A1: - Color::EncodeRGB5A1(src_color, dst_pixel); - break; - - case Regs::PixelFormat::RGBA4: - Color::EncodeRGBA4(src_color, dst_pixel); - break; - - default: - LOG_ERROR(HW_GPU, "Unknown destination framebuffer format %x", - config.output_format.Value()); - break; - } - } - } - - LOG_TRACE(HW_GPU, "DisplayTriggerTransfer: 0x%08x bytes from 0x%08x(%ux%u)-> " + if (config.is_texture_copy) { + TextureCopy(config); + LOG_TRACE(HW_GPU, "TextureCopy: 0x%X bytes from 0x%08X(%u+%u)-> " + "0x%08X(%u+%u), flags 0x%08X", + config.texture_copy.size, config.GetPhysicalInputAddress(), + config.texture_copy.input_width * 16, config.texture_copy.input_gap * 16, + config.GetPhysicalOutputAddress(), config.texture_copy.output_width * 16, + config.texture_copy.output_gap * 16, config.flags); + } else { + DisplayTransfer(config); + LOG_TRACE(HW_GPU, "DisplayTransfer: 0x%08x(%ux%u)-> " "0x%08x(%ux%u), dst format %x, flags 0x%08X", - config.output_height * output_width * - GPU::Regs::BytesPerPixel(config.output_format), config.GetPhysicalInputAddress(), config.input_width.Value(), config.input_height.Value(), config.GetPhysicalOutputAddress(), - output_width, output_height, config.output_format.Value(), config.flags); + config.output_width.Value(), config.output_height.Value(), + config.output_format.Value(), config.flags); } g_regs.display_transfer_config.trigger = 0; diff --git a/src/core/memory.cpp b/src/core/memory.cpp index df029d655..64c388374 100644 --- a/src/core/memory.cpp +++ b/src/core/memory.cpp @@ -251,6 +251,9 @@ bool IsValidVirtualAddress(const VAddr vaddr) { if (page_pointer) return true; + if (current_page_table->attributes[vaddr >> PAGE_BITS] == PageType::RasterizerCachedMemory) + return true; + if (current_page_table->attributes[vaddr >> PAGE_BITS] != PageType::Special) return false; diff --git a/src/video_core/rasterizer_interface.h b/src/video_core/rasterizer_interface.h index 71df233b5..8ef7e74c7 100644 --- a/src/video_core/rasterizer_interface.h +++ b/src/video_core/rasterizer_interface.h @@ -42,11 +42,16 @@ public: /// and invalidated virtual void FlushAndInvalidateRegion(PAddr addr, u32 size) = 0; - /// Attempt to use a faster method to perform a display transfer + /// Attempt to use a faster method to perform a display transfer with is_texture_copy = 0 virtual bool AccelerateDisplayTransfer(const GPU::Regs::DisplayTransferConfig& config) { return false; } + /// Attempt to use a faster method to perform a display transfer with is_texture_copy = 1 + virtual bool AccelerateTextureCopy(const GPU::Regs::DisplayTransferConfig& config) { + return false; + } + /// Attempt to use a faster method to fill a region virtual bool AccelerateFill(const GPU::Regs::MemoryFillConfig& config) { return false; diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp index 60c9d9180..7cc3b407a 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer.cpp +++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp @@ -211,6 +211,27 @@ void RasterizerOpenGL::DrawTriangles() { uniform_block_data.dirty = true; } + // Scissor checks are window-, not viewport-relative, which means that if the cached texture + // sub-rect changes, the scissor bounds also need to be updated. + GLint scissor_x1 = rect.left + regs.scissor_test.x1 * color_surface->res_scale_width; + GLint scissor_y1 = rect.bottom + regs.scissor_test.y1 * color_surface->res_scale_height; + // x2, y2 have +1 added to cover the entire pixel area, otherwise you might get cracks when + // scaling or doing multisampling. + GLint scissor_x2 = rect.left + (regs.scissor_test.x2 + 1) * color_surface->res_scale_width; + GLint scissor_y2 = rect.bottom + (regs.scissor_test.y2 + 1) * color_surface->res_scale_height; + + if (uniform_block_data.data.scissor_x1 != scissor_x1 || + uniform_block_data.data.scissor_x2 != scissor_x2 || + uniform_block_data.data.scissor_y1 != scissor_y1 || + uniform_block_data.data.scissor_y2 != scissor_y2) { + + uniform_block_data.data.scissor_x1 = scissor_x1; + uniform_block_data.data.scissor_x2 = scissor_x2; + uniform_block_data.data.scissor_y1 = scissor_y1; + uniform_block_data.data.scissor_y2 = scissor_y2; + uniform_block_data.dirty = true; + } + // Sync and bind the texture surfaces const auto pica_textures = regs.GetTextures(); for (unsigned texture_index = 0; texture_index < pica_textures.size(); ++texture_index) { @@ -374,10 +395,6 @@ void RasterizerOpenGL::NotifyPicaRegisterChanged(u32 id) { case PICA_REG_INDEX(scissor_test.mode): shader_dirty = true; break; - case PICA_REG_INDEX(scissor_test.x1): // and y1 - case PICA_REG_INDEX(scissor_test.x2): // and y2 - SyncScissorTest(); - break; // Logic op case PICA_REG_INDEX(output_merger.logic_op): @@ -692,11 +709,6 @@ bool RasterizerOpenGL::AccelerateDisplayTransfer(const GPU::Regs::DisplayTransfe using PixelFormat = CachedSurface::PixelFormat; using SurfaceType = CachedSurface::SurfaceType; - if (config.is_texture_copy) { - // TODO(tfarley): Try to hardware accelerate this - return false; - } - CachedSurface src_params; src_params.addr = config.GetPhysicalInputAddress(); src_params.width = config.output_width; @@ -751,6 +763,11 @@ bool RasterizerOpenGL::AccelerateDisplayTransfer(const GPU::Regs::DisplayTransfe return true; } +bool RasterizerOpenGL::AccelerateTextureCopy(const GPU::Regs::DisplayTransferConfig& config) { + // TODO(tfarley): Try to hardware accelerate this + return false; +} + bool RasterizerOpenGL::AccelerateFill(const GPU::Regs::MemoryFillConfig& config) { using PixelFormat = CachedSurface::PixelFormat; using SurfaceType = CachedSurface::SurfaceType; @@ -1061,7 +1078,6 @@ void RasterizerOpenGL::SetShader() { SyncDepthOffset(); SyncAlphaTest(); SyncCombinerColor(); - SyncScissorTest(); auto& tev_stages = Pica::g_state.regs.GetTevStages(); for (int index = 0; index < tev_stages.size(); ++index) SyncTevConstColor(index, tev_stages[index]); @@ -1236,22 +1252,6 @@ void RasterizerOpenGL::SyncDepthTest() { : GL_ALWAYS; } -void RasterizerOpenGL::SyncScissorTest() { - const auto& regs = Pica::g_state.regs; - - if (uniform_block_data.data.scissor_x1 != regs.scissor_test.x1 || - uniform_block_data.data.scissor_y1 != regs.scissor_test.y1 || - uniform_block_data.data.scissor_x2 != regs.scissor_test.x2 || - uniform_block_data.data.scissor_y2 != regs.scissor_test.y2) { - - uniform_block_data.data.scissor_x1 = regs.scissor_test.x1; - uniform_block_data.data.scissor_y1 = regs.scissor_test.y1; - uniform_block_data.data.scissor_x2 = regs.scissor_test.x2; - uniform_block_data.data.scissor_y2 = regs.scissor_test.y2; - uniform_block_data.dirty = true; - } -} - void RasterizerOpenGL::SyncCombinerColor() { auto combiner_color = PicaToGL::ColorRGBA8(Pica::g_state.regs.tev_combiner_buffer_color.raw); if (combiner_color != uniform_block_data.data.tev_combiner_buffer_color) { diff --git a/src/video_core/renderer_opengl/gl_rasterizer.h b/src/video_core/renderer_opengl/gl_rasterizer.h index 24fefed1b..e1a9cb361 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer.h +++ b/src/video_core/renderer_opengl/gl_rasterizer.h @@ -238,6 +238,7 @@ public: void FlushRegion(PAddr addr, u32 size) override; void FlushAndInvalidateRegion(PAddr addr, u32 size) override; bool AccelerateDisplayTransfer(const GPU::Regs::DisplayTransferConfig& config) override; + bool AccelerateTextureCopy(const GPU::Regs::DisplayTransferConfig& config) override; bool AccelerateFill(const GPU::Regs::MemoryFillConfig& config) override; bool AccelerateDisplay(const GPU::Regs::FramebufferConfig& config, PAddr framebuffer_addr, u32 pixel_stride, ScreenInfo& screen_info) override; @@ -393,9 +394,6 @@ private: /// Syncs the depth test states to match the PICA register void SyncDepthTest(); - /// Syncs the scissor test state to match the PICA register - void SyncScissorTest(); - /// Syncs the TEV combiner color buffer to match the PICA register void SyncCombinerColor(); diff --git a/src/video_core/renderer_opengl/gl_shader_gen.cpp b/src/video_core/renderer_opengl/gl_shader_gen.cpp index 1808ee0a9..8f278722d 100644 --- a/src/video_core/renderer_opengl/gl_shader_gen.cpp +++ b/src/video_core/renderer_opengl/gl_shader_gen.cpp @@ -645,11 +645,10 @@ vec4 secondary_fragment_color = vec4(0.0); // Negate the condition if we have to keep only the pixels outside the scissor box if (state.scissor_test_mode == Regs::ScissorMode::Include) out += "!"; - // x2,y2 have +1 added to cover the entire pixel area - out += "(gl_FragCoord.x >= scissor_x1 * framebuffer_scale.x && " - "gl_FragCoord.y >= scissor_y1 * framebuffer_scale.y && " - "gl_FragCoord.x < (scissor_x2 + 1) * framebuffer_scale.x && " - "gl_FragCoord.y < (scissor_y2 + 1) * framebuffer_scale.y)) discard;\n"; + out += "(gl_FragCoord.x >= scissor_x1 && " + "gl_FragCoord.y >= scissor_y1 && " + "gl_FragCoord.x < scissor_x2 && " + "gl_FragCoord.y < scissor_y2)) discard;\n"; } out += "float z_over_w = 1.0 - gl_FragCoord.z * 2.0;\n"; |