diff options
Diffstat (limited to '')
24 files changed, 165 insertions, 134 deletions
diff --git a/src/core/arm/arm_interface.h b/src/core/arm/arm_interface.h index 5cffe513c..533067d4f 100644 --- a/src/core/arm/arm_interface.h +++ b/src/core/arm/arm_interface.h @@ -14,10 +14,6 @@ namespace Core { /// Generic ARM11 CPU interface class ARM_Interface : NonCopyable { public: - ARM_Interface() { - num_instructions = 0; - } - virtual ~ARM_Interface() { } @@ -146,11 +142,11 @@ public: virtual void PrepareReschedule() = 0; /// Getter for num_instructions - u64 GetNumInstructions() { + u64 GetNumInstructions() const { return num_instructions; } - s64 down_count; ///< A decreasing counter of remaining cycles before the next event, decreased by the cpu run loop + s64 down_count = 0; ///< A decreasing counter of remaining cycles before the next event, decreased by the cpu run loop protected: @@ -162,6 +158,5 @@ protected: private: - u64 num_instructions; ///< Number of instructions executed - + u64 num_instructions = 0; ///< Number of instructions executed }; diff --git a/src/core/arm/dyncom/arm_dyncom_interpreter.cpp b/src/core/arm/dyncom/arm_dyncom_interpreter.cpp index 96c88c83a..2cff2a26a 100644 --- a/src/core/arm/dyncom/arm_dyncom_interpreter.cpp +++ b/src/core/arm/dyncom/arm_dyncom_interpreter.cpp @@ -51,7 +51,7 @@ enum { typedef unsigned int (*shtop_fp_t)(ARMul_State* cpu, unsigned int sht_oper); -static bool CondPassed(ARMul_State* cpu, unsigned int cond) { +static bool CondPassed(const ARMul_State* cpu, unsigned int cond) { const bool n_flag = cpu->NFlag != 0; const bool z_flag = cpu->ZFlag != 0; const bool c_flag = cpu->CFlag != 0; diff --git a/src/core/arm/dyncom/arm_dyncom_run.h b/src/core/arm/dyncom/arm_dyncom_run.h index 13bef17fc..8eb694fee 100644 --- a/src/core/arm/dyncom/arm_dyncom_run.h +++ b/src/core/arm/dyncom/arm_dyncom_run.h @@ -30,7 +30,7 @@ * @return If the PC is being read, then the word-aligned PC value is returned. * If the PC is not being read, then the value stored in the register is returned. */ -static inline u32 CHECK_READ_REG15_WA(ARMul_State* cpu, int Rn) { +inline u32 CHECK_READ_REG15_WA(const ARMul_State* cpu, int Rn) { return (Rn == 15) ? ((cpu->Reg[15] & ~0x3) + cpu->GetInstructionSize() * 2) : cpu->Reg[Rn]; } @@ -43,6 +43,6 @@ static inline u32 CHECK_READ_REG15_WA(ARMul_State* cpu, int Rn) { * @return If the PC is being read, then the incremented PC value is returned. * If the PC is not being read, then the values stored in the register is returned. */ -static inline u32 CHECK_READ_REG15(ARMul_State* cpu, int Rn) { +inline u32 CHECK_READ_REG15(const ARMul_State* cpu, int Rn) { return (Rn == 15) ? ((cpu->Reg[15] & ~0x1) + cpu->GetInstructionSize() * 2) : cpu->Reg[Rn]; } diff --git a/src/core/arm/dyncom/arm_dyncom_thumb.h b/src/core/arm/dyncom/arm_dyncom_thumb.h index 447974363..c1be3c735 100644 --- a/src/core/arm/dyncom/arm_dyncom_thumb.h +++ b/src/core/arm/dyncom/arm_dyncom_thumb.h @@ -38,7 +38,7 @@ enum class ThumbDecodeStatus { // Translates a Thumb mode instruction into its ARM equivalent. ThumbDecodeStatus TranslateThumbInstruction(u32 addr, u32 instr, u32* ainstr, u32* inst_size); -static inline u32 GetThumbInstruction(u32 instr, u32 address) { +inline u32 GetThumbInstruction(u32 instr, u32 address) { // Normally you would need to handle instruction endianness, // however, it is fixed to little-endian on the MPCore, so // there's no need to check for this beforehand. diff --git a/src/core/arm/skyeye_common/vfp/vfp_helper.h b/src/core/arm/skyeye_common/vfp/vfp_helper.h index 91a8d4d57..210972917 100644 --- a/src/core/arm/skyeye_common/vfp/vfp_helper.h +++ b/src/core/arm/skyeye_common/vfp/vfp_helper.h @@ -85,7 +85,7 @@ enum : u32 { #define vfp_single(inst) (((inst) & 0x0000f00) == 0xa00) -static inline u32 vfp_shiftright32jamming(u32 val, unsigned int shift) +inline u32 vfp_shiftright32jamming(u32 val, unsigned int shift) { if (shift) { if (shift < 32) @@ -96,7 +96,7 @@ static inline u32 vfp_shiftright32jamming(u32 val, unsigned int shift) return val; } -static inline u64 vfp_shiftright64jamming(u64 val, unsigned int shift) +inline u64 vfp_shiftright64jamming(u64 val, unsigned int shift) { if (shift) { if (shift < 64) @@ -107,7 +107,7 @@ static inline u64 vfp_shiftright64jamming(u64 val, unsigned int shift) return val; } -static inline u32 vfp_hi64to32jamming(u64 val) +inline u32 vfp_hi64to32jamming(u64 val) { u32 v; u32 highval = val >> 32; @@ -121,7 +121,7 @@ static inline u32 vfp_hi64to32jamming(u64 val) return v; } -static inline void add128(u64* resh, u64* resl, u64 nh, u64 nl, u64 mh, u64 ml) +inline void add128(u64* resh, u64* resl, u64 nh, u64 nl, u64 mh, u64 ml) { *resl = nl + ml; *resh = nh + mh; @@ -129,7 +129,7 @@ static inline void add128(u64* resh, u64* resl, u64 nh, u64 nl, u64 mh, u64 ml) *resh += 1; } -static inline void sub128(u64* resh, u64* resl, u64 nh, u64 nl, u64 mh, u64 ml) +inline void sub128(u64* resh, u64* resl, u64 nh, u64 nl, u64 mh, u64 ml) { *resl = nl - ml; *resh = nh - mh; @@ -137,7 +137,7 @@ static inline void sub128(u64* resh, u64* resl, u64 nh, u64 nl, u64 mh, u64 ml) *resh -= 1; } -static inline void mul64to128(u64* resh, u64* resl, u64 n, u64 m) +inline void mul64to128(u64* resh, u64* resl, u64 n, u64 m) { u32 nh, nl, mh, ml; u64 rh, rma, rmb, rl; @@ -164,20 +164,20 @@ static inline void mul64to128(u64* resh, u64* resl, u64 n, u64 m) *resh = rh; } -static inline void shift64left(u64* resh, u64* resl, u64 n) +inline void shift64left(u64* resh, u64* resl, u64 n) { *resh = n >> 63; *resl = n << 1; } -static inline u64 vfp_hi64multiply64(u64 n, u64 m) +inline u64 vfp_hi64multiply64(u64 n, u64 m) { u64 rh, rl; mul64to128(&rh, &rl, n, m); return rh | (rl != 0); } -static inline u64 vfp_estimate_div128to64(u64 nh, u64 nl, u64 m) +inline u64 vfp_estimate_div128to64(u64 nh, u64 nl, u64 m) { u64 mh, ml, remh, reml, termh, terml, z; @@ -249,7 +249,7 @@ enum : u32 { VFP_SNAN = (VFP_NAN|VFP_NAN_SIGNAL) }; -static inline int vfp_single_type(vfp_single* s) +inline int vfp_single_type(const vfp_single* s) { int type = VFP_NUMBER; if (s->exponent == 255) { @@ -271,7 +271,7 @@ static inline int vfp_single_type(vfp_single* s) // Unpack a single-precision float. Note that this returns the magnitude // of the single-precision float mantissa with the 1. if necessary, // aligned to bit 30. -static inline void vfp_single_unpack(vfp_single* s, s32 val, u32* fpscr) +inline void vfp_single_unpack(vfp_single* s, s32 val, u32* fpscr) { s->sign = vfp_single_packed_sign(val) >> 16, s->exponent = vfp_single_packed_exponent(val); @@ -293,7 +293,7 @@ static inline void vfp_single_unpack(vfp_single* s, s32 val, u32* fpscr) // Re-pack a single-precision float. This assumes that the float is // already normalised such that the MSB is bit 30, _not_ bit 31. -static inline s32 vfp_single_pack(vfp_single* s) +inline s32 vfp_single_pack(const vfp_single* s) { u32 val = (s->sign << 16) + (s->exponent << VFP_SINGLE_MANTISSA_BITS) + @@ -335,7 +335,7 @@ struct vfp_double { #define vfp_double_packed_exponent(v) (((v) >> VFP_DOUBLE_MANTISSA_BITS) & ((1 << VFP_DOUBLE_EXPONENT_BITS) - 1)) #define vfp_double_packed_mantissa(v) ((v) & ((1ULL << VFP_DOUBLE_MANTISSA_BITS) - 1)) -static inline int vfp_double_type(vfp_double* s) +inline int vfp_double_type(const vfp_double* s) { int type = VFP_NUMBER; if (s->exponent == 2047) { @@ -357,7 +357,7 @@ static inline int vfp_double_type(vfp_double* s) // Unpack a double-precision float. Note that this returns the magnitude // of the double-precision float mantissa with the 1. if necessary, // aligned to bit 62. -static inline void vfp_double_unpack(vfp_double* s, s64 val, u32* fpscr) +inline void vfp_double_unpack(vfp_double* s, s64 val, u32* fpscr) { s->sign = vfp_double_packed_sign(val) >> 48; s->exponent = vfp_double_packed_exponent(val); @@ -379,7 +379,7 @@ static inline void vfp_double_unpack(vfp_double* s, s64 val, u32* fpscr) // Re-pack a double-precision float. This assumes that the float is // already normalised such that the MSB is bit 30, _not_ bit 31. -static inline s64 vfp_double_pack(vfp_double* s) +inline s64 vfp_double_pack(const vfp_double* s) { u64 val = ((u64)s->sign << 48) + ((u64)s->exponent << VFP_DOUBLE_MANTISSA_BITS) + @@ -415,7 +415,7 @@ struct op { u32 flags; }; -static inline u32 fls(u32 x) +inline u32 fls(u32 x) { int r = 32; diff --git a/src/core/hle/service/gsp_gpu.cpp b/src/core/hle/service/gsp_gpu.cpp index 481da0c9f..98b11c798 100644 --- a/src/core/hle/service/gsp_gpu.cpp +++ b/src/core/hle/service/gsp_gpu.cpp @@ -275,7 +275,7 @@ static void FlushDataCache(Service::Interface* self) { u32 size = cmd_buff[2]; u32 process = cmd_buff[4]; - VideoCore::g_renderer->hw_rasterizer->NotifyFlush(Memory::VirtualToPhysicalAddress(address), size); + VideoCore::g_renderer->rasterizer->InvalidateRegion(Memory::VirtualToPhysicalAddress(address), size); // TODO(purpasmart96): Verify return header on HW @@ -365,7 +365,7 @@ static void ExecuteCommand(const Command& command, u32 thread_id) { // GX request DMA - typically used for copying memory from GSP heap to VRAM case CommandId::REQUEST_DMA: - VideoCore::g_renderer->hw_rasterizer->NotifyPreRead(Memory::VirtualToPhysicalAddress(command.dma_request.source_address), + VideoCore::g_renderer->rasterizer->FlushRegion(Memory::VirtualToPhysicalAddress(command.dma_request.source_address), command.dma_request.size); memcpy(Memory::GetPointer(command.dma_request.dest_address), @@ -373,7 +373,7 @@ static void ExecuteCommand(const Command& command, u32 thread_id) { command.dma_request.size); SignalInterrupt(InterruptId::DMA); - VideoCore::g_renderer->hw_rasterizer->NotifyFlush(Memory::VirtualToPhysicalAddress(command.dma_request.dest_address), + VideoCore::g_renderer->rasterizer->InvalidateRegion(Memory::VirtualToPhysicalAddress(command.dma_request.dest_address), command.dma_request.size); break; @@ -467,7 +467,7 @@ static void ExecuteCommand(const Command& command, u32 thread_id) { if (region.size == 0) break; - VideoCore::g_renderer->hw_rasterizer->NotifyFlush( + VideoCore::g_renderer->rasterizer->InvalidateRegion( Memory::VirtualToPhysicalAddress(region.address), region.size); } break; diff --git a/src/core/hle/service/y2r_u.cpp b/src/core/hle/service/y2r_u.cpp index 6b1b71fe4..0429927f2 100644 --- a/src/core/hle/service/y2r_u.cpp +++ b/src/core/hle/service/y2r_u.cpp @@ -267,7 +267,7 @@ static void StartConversion(Service::Interface* self) { // dst_image_size would seem to be perfect for this, but it doesn't include the gap :( u32 total_output_size = conversion.input_lines * (conversion.dst.transfer_unit + conversion.dst.gap); - VideoCore::g_renderer->hw_rasterizer->NotifyFlush( + VideoCore::g_renderer->rasterizer->InvalidateRegion( Memory::VirtualToPhysicalAddress(conversion.dst.address), total_output_size); LOG_DEBUG(Service_Y2R, "called"); diff --git a/src/core/hw/gpu.cpp b/src/core/hw/gpu.cpp index bc7bde903..4bd3a632d 100644 --- a/src/core/hw/gpu.cpp +++ b/src/core/hw/gpu.cpp @@ -26,7 +26,7 @@ #include "core/tracer/recorder.h" #include "video_core/command_processor.h" -#include "video_core/hwrasterizer_base.h" +#include "video_core/rasterizer_interface.h" #include "video_core/renderer_base.h" #include "video_core/utils.h" #include "video_core/video_core.h" @@ -141,7 +141,7 @@ inline void Write(u32 addr, const T data) { GSP_GPU::SignalInterrupt(GSP_GPU::InterruptId::PSC1); } - VideoCore::g_renderer->hw_rasterizer->NotifyFlush(config.GetStartAddress(), config.GetEndAddress() - config.GetStartAddress()); + VideoCore::g_renderer->rasterizer->InvalidateRegion(config.GetStartAddress(), config.GetEndAddress() - config.GetStartAddress()); } // Reset "trigger" flag and set the "finish" flag @@ -172,7 +172,7 @@ inline void Write(u32 addr, const T data) { u32 output_gap = config.texture_copy.output_gap * 16; size_t contiguous_input_size = config.texture_copy.size / input_width * (input_width + input_gap); - VideoCore::g_renderer->hw_rasterizer->NotifyPreRead(config.GetPhysicalInputAddress(), contiguous_input_size); + VideoCore::g_renderer->rasterizer->FlushRegion(config.GetPhysicalInputAddress(), contiguous_input_size); u32 remaining_size = config.texture_copy.size; u32 remaining_input = input_width; @@ -205,7 +205,7 @@ inline void Write(u32 addr, const T data) { config.flags); size_t contiguous_output_size = config.texture_copy.size / output_width * (output_width + output_gap); - VideoCore::g_renderer->hw_rasterizer->NotifyFlush(config.GetPhysicalOutputAddress(), contiguous_output_size); + VideoCore::g_renderer->rasterizer->InvalidateRegion(config.GetPhysicalOutputAddress(), contiguous_output_size); GSP_GPU::SignalInterrupt(GSP_GPU::InterruptId::PPF); break; @@ -232,7 +232,7 @@ inline void Write(u32 addr, const T data) { 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); - VideoCore::g_renderer->hw_rasterizer->NotifyPreRead(config.GetPhysicalInputAddress(), input_size); + VideoCore::g_renderer->rasterizer->FlushRegion(config.GetPhysicalInputAddress(), input_size); for (u32 y = 0; y < output_height; ++y) { for (u32 x = 0; x < output_width; ++x) { @@ -339,7 +339,7 @@ inline void Write(u32 addr, const T data) { g_regs.display_transfer_config.trigger = 0; GSP_GPU::SignalInterrupt(GSP_GPU::InterruptId::PPF); - VideoCore::g_renderer->hw_rasterizer->NotifyFlush(config.GetPhysicalOutputAddress(), output_size); + VideoCore::g_renderer->rasterizer->InvalidateRegion(config.GetPhysicalOutputAddress(), output_size); } break; } diff --git a/src/video_core/CMakeLists.txt b/src/video_core/CMakeLists.txt index 2a924f4ad..c3d7294d5 100644 --- a/src/video_core/CMakeLists.txt +++ b/src/video_core/CMakeLists.txt @@ -11,8 +11,10 @@ set(SRCS pica.cpp primitive_assembly.cpp rasterizer.cpp + renderer_base.cpp shader/shader.cpp shader/shader_interpreter.cpp + swrasterizer.cpp utils.cpp video_core.cpp ) @@ -30,13 +32,14 @@ set(HEADERS clipper.h command_processor.h gpu_debugger.h - hwrasterizer_base.h pica.h primitive_assembly.h rasterizer.h + rasterizer_interface.h renderer_base.h shader/shader.h shader/shader_interpreter.h + swrasterizer.h utils.h video_core.h ) diff --git a/src/video_core/clipper.cpp b/src/video_core/clipper.cpp index ed99c4f13..5d609da06 100644 --- a/src/video_core/clipper.cpp +++ b/src/video_core/clipper.cpp @@ -78,7 +78,7 @@ static void InitScreenCoordinates(OutputVertex& vtx) vtx.screenpos[2] = viewport.offset_z + vtx.pos.z * inv_w * viewport.zscale; } -void ProcessTriangle(OutputVertex &v0, OutputVertex &v1, OutputVertex &v2) { +void ProcessTriangle(const OutputVertex &v0, const OutputVertex &v1, const OutputVertex &v2) { using boost::container::static_vector; // Clipping a planar n-gon against a plane will remove at least 1 vertex and introduces 2 at diff --git a/src/video_core/clipper.h b/src/video_core/clipper.h index 6ed01e877..f85d8d4c9 100644 --- a/src/video_core/clipper.h +++ b/src/video_core/clipper.h @@ -14,7 +14,7 @@ namespace Clipper { using Shader::OutputVertex; -void ProcessTriangle(OutputVertex& v0, OutputVertex& v1, OutputVertex& v2); +void ProcessTriangle(const OutputVertex& v0, const OutputVertex& v1, const OutputVertex& v2); } // namespace diff --git a/src/video_core/command_processor.cpp b/src/video_core/command_processor.cpp index bd1b09a4b..35b976c60 100644 --- a/src/video_core/command_processor.cpp +++ b/src/video_core/command_processor.cpp @@ -336,19 +336,14 @@ static void WritePicaReg(u32 id, u32 value, u32 mask) { } } - if (Settings::values.use_hw_renderer) { - // Send to hardware renderer - static auto AddHWTriangle = [](const Pica::Shader::OutputVertex& v0, - const Pica::Shader::OutputVertex& v1, - const Pica::Shader::OutputVertex& v2) { - VideoCore::g_renderer->hw_rasterizer->AddTriangle(v0, v1, v2); - }; - - primitive_assembler.SubmitVertex(output, AddHWTriangle); - } else { - // Send to triangle clipper - primitive_assembler.SubmitVertex(output, Clipper::ProcessTriangle); - } + // Send to renderer + using Pica::Shader::OutputVertex; + auto AddTriangle = []( + const OutputVertex& v0, const OutputVertex& v1, const OutputVertex& v2) { + VideoCore::g_renderer->rasterizer->AddTriangle(v0, v1, v2); + }; + + primitive_assembler.SubmitVertex(output, AddTriangle); } for (auto& range : memory_accesses.ranges) { @@ -356,9 +351,7 @@ static void WritePicaReg(u32 id, u32 value, u32 mask) { range.second, range.first); } - if (Settings::values.use_hw_renderer) { - VideoCore::g_renderer->hw_rasterizer->DrawTriangles(); - } + VideoCore::g_renderer->rasterizer->DrawTriangles(); #if PICA_DUMP_GEOMETRY geometry_dumper.Dump(); @@ -475,7 +468,7 @@ static void WritePicaReg(u32 id, u32 value, u32 mask) { break; } - VideoCore::g_renderer->hw_rasterizer->NotifyPicaRegisterChanged(id); + VideoCore::g_renderer->rasterizer->NotifyPicaRegisterChanged(id); if (g_debug_context) g_debug_context->OnEvent(DebugContext::Event::PicaCommandProcessed, reinterpret_cast<void*>(&id)); diff --git a/src/video_core/debug_utils/debug_utils.cpp b/src/video_core/debug_utils/debug_utils.cpp index f1cfa9361..4f66dbd65 100644 --- a/src/video_core/debug_utils/debug_utils.cpp +++ b/src/video_core/debug_utils/debug_utils.cpp @@ -46,10 +46,8 @@ void DebugContext::OnEvent(Event event, void* data) { { std::unique_lock<std::mutex> lock(breakpoint_mutex); - if (Settings::values.use_hw_renderer) { - // Commit the hardware renderer's framebuffer so it will show on debug widgets - VideoCore::g_renderer->hw_rasterizer->CommitFramebuffer(); - } + // Commit the hardware renderer's framebuffer so it will show on debug widgets + VideoCore::g_renderer->rasterizer->FlushFramebuffer(); // TODO: Should stop the CPU thread here once we multithread emulation. diff --git a/src/video_core/primitive_assembly.cpp b/src/video_core/primitive_assembly.cpp index 44a8dbfe9..d5a0a96a4 100644 --- a/src/video_core/primitive_assembly.cpp +++ b/src/video_core/primitive_assembly.cpp @@ -39,13 +39,12 @@ void PrimitiveAssembler<VertexType>::SubmitVertex(VertexType& vtx, TriangleHandl buffer[buffer_index] = vtx; - if (topology == Regs::TriangleTopology::Strip) { - strip_ready |= (buffer_index == 1); + strip_ready |= (buffer_index == 1); + + if (topology == Regs::TriangleTopology::Strip) buffer_index = !buffer_index; - } else if (topology == Regs::TriangleTopology::Fan) { + else if (topology == Regs::TriangleTopology::Fan) buffer_index = 1; - strip_ready = true; - } break; default: diff --git a/src/video_core/hwrasterizer_base.h b/src/video_core/rasterizer_interface.h index 54b8892fb..008c5827b 100644 --- a/src/video_core/hwrasterizer_base.h +++ b/src/video_core/rasterizer_interface.h @@ -12,10 +12,11 @@ struct OutputVertex; } } -class HWRasterizer { +namespace VideoCore { + +class RasterizerInterface { public: - virtual ~HWRasterizer() { - } + virtual ~RasterizerInterface() {} /// Initialize API-specific GPU objects virtual void InitObjects() = 0; @@ -32,14 +33,16 @@ public: virtual void DrawTriangles() = 0; /// Commit the rasterizer's framebuffer contents immediately to the current 3DS memory framebuffer - virtual void CommitFramebuffer() = 0; + virtual void FlushFramebuffer() = 0; /// Notify rasterizer that the specified PICA register has been changed virtual void NotifyPicaRegisterChanged(u32 id) = 0; - /// Notify rasterizer that the specified 3DS memory region will be read from after this notification - virtual void NotifyPreRead(PAddr addr, u32 size) = 0; + /// Notify rasterizer that any caches of the specified region should be flushed to 3DS memory. + virtual void FlushRegion(PAddr addr, u32 size) = 0; - /// Notify rasterizer that a 3DS memory region has been changed - virtual void NotifyFlush(PAddr addr, u32 size) = 0; + /// Notify rasterizer that any caches of the specified region should be discraded and reloaded from 3DS memory. + virtual void InvalidateRegion(PAddr addr, u32 size) = 0; }; + +} diff --git a/src/video_core/renderer_base.cpp b/src/video_core/renderer_base.cpp new file mode 100644 index 000000000..93e980216 --- /dev/null +++ b/src/video_core/renderer_base.cpp @@ -0,0 +1,28 @@ +// Copyright 2015 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <memory> + +#include "common/make_unique.h" + +#include "core/settings.h" + +#include "video_core/renderer_base.h" +#include "video_core/video_core.h" +#include "video_core/swrasterizer.h" +#include "video_core/renderer_opengl/gl_rasterizer.h" + +void RendererBase::RefreshRasterizerSetting() { + bool hw_renderer_enabled = VideoCore::g_hw_renderer_enabled; + if (rasterizer == nullptr || opengl_rasterizer_active != hw_renderer_enabled) { + opengl_rasterizer_active = hw_renderer_enabled; + + if (hw_renderer_enabled) { + rasterizer = Common::make_unique<RasterizerOpenGL>(); + } else { + rasterizer = Common::make_unique<VideoCore::SWRasterizer>(); + } + rasterizer->InitObjects(); + } +} diff --git a/src/video_core/renderer_base.h b/src/video_core/renderer_base.h index 6587bcf27..506bff815 100644 --- a/src/video_core/renderer_base.h +++ b/src/video_core/renderer_base.h @@ -8,7 +8,7 @@ #include "common/common_types.h" -#include "video_core/hwrasterizer_base.h" +#include "video_core/rasterizer_interface.h" class EmuWindow; @@ -54,10 +54,14 @@ public: return m_current_frame; } - std::unique_ptr<HWRasterizer> hw_rasterizer; + void RefreshRasterizerSetting(); + + std::unique_ptr<VideoCore::RasterizerInterface> rasterizer; protected: f32 m_current_fps; ///< Current framerate, should be set by the renderer int m_current_frame; ///< Current frame, should be set by the renderer +private: + bool opengl_rasterizer_active = false; }; diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp index 23d9517da..092351dce 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer.cpp +++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp @@ -135,7 +135,7 @@ void RasterizerOpenGL::Reset() { SetShader(); - res_cache.FullFlush(); + res_cache.InvalidateAll(); } void RasterizerOpenGL::AddTriangle(const Pica::Shader::OutputVertex& v0, @@ -176,11 +176,11 @@ void RasterizerOpenGL::DrawTriangles() { u32 cur_fb_depth_size = Pica::Regs::BytesPerDepthPixel(regs.framebuffer.depth_format) * regs.framebuffer.GetWidth() * regs.framebuffer.GetHeight(); - res_cache.NotifyFlush(cur_fb_color_addr, cur_fb_color_size, true); - res_cache.NotifyFlush(cur_fb_depth_addr, cur_fb_depth_size, true); + res_cache.InvalidateInRange(cur_fb_color_addr, cur_fb_color_size, true); + res_cache.InvalidateInRange(cur_fb_depth_addr, cur_fb_depth_size, true); } -void RasterizerOpenGL::CommitFramebuffer() { +void RasterizerOpenGL::FlushFramebuffer() { CommitColorBuffer(); CommitDepthBuffer(); } @@ -188,9 +188,6 @@ void RasterizerOpenGL::CommitFramebuffer() { void RasterizerOpenGL::NotifyPicaRegisterChanged(u32 id) { const auto& regs = Pica::g_state.regs; - if (!Settings::values.use_hw_renderer) - return; - switch(id) { // Culling case PICA_REG_INDEX(cull_mode): @@ -284,12 +281,9 @@ void RasterizerOpenGL::NotifyPicaRegisterChanged(u32 id) { } } -void RasterizerOpenGL::NotifyPreRead(PAddr addr, u32 size) { +void RasterizerOpenGL::FlushRegion(PAddr addr, u32 size) { const auto& regs = Pica::g_state.regs; - if (!Settings::values.use_hw_renderer) - return; - PAddr cur_fb_color_addr = regs.framebuffer.GetColorBufferPhysicalAddress(); u32 cur_fb_color_size = Pica::Regs::BytesPerColorPixel(regs.framebuffer.color_format) * regs.framebuffer.GetWidth() * regs.framebuffer.GetHeight(); @@ -306,12 +300,9 @@ void RasterizerOpenGL::NotifyPreRead(PAddr addr, u32 size) { CommitDepthBuffer(); } -void RasterizerOpenGL::NotifyFlush(PAddr addr, u32 size) { +void RasterizerOpenGL::InvalidateRegion(PAddr addr, u32 size) { const auto& regs = Pica::g_state.regs; - if (!Settings::values.use_hw_renderer) - return; - PAddr cur_fb_color_addr = regs.framebuffer.GetColorBufferPhysicalAddress(); u32 cur_fb_color_size = Pica::Regs::BytesPerColorPixel(regs.framebuffer.color_format) * regs.framebuffer.GetWidth() * regs.framebuffer.GetHeight(); @@ -328,7 +319,7 @@ void RasterizerOpenGL::NotifyFlush(PAddr addr, u32 size) { ReloadDepthBuffer(); // Notify cache of flush in case the region touches a cached resource - res_cache.NotifyFlush(addr, size); + res_cache.InvalidateInRange(addr, size); } void RasterizerOpenGL::SamplerInfo::Create() { diff --git a/src/video_core/renderer_opengl/gl_rasterizer.h b/src/video_core/renderer_opengl/gl_rasterizer.h index 5ba898189..92b1f812e 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer.h +++ b/src/video_core/renderer_opengl/gl_rasterizer.h @@ -14,7 +14,7 @@ #include "common/hash.h" #include "video_core/pica.h" -#include "video_core/hwrasterizer_base.h" +#include "video_core/rasterizer_interface.h" #include "video_core/renderer_opengl/gl_rasterizer_cache.h" #include "video_core/renderer_opengl/gl_state.h" #include "video_core/shader/shader_interpreter.h" @@ -102,37 +102,22 @@ struct hash<PicaShaderConfig> { } // namespace std -class RasterizerOpenGL : public HWRasterizer { +class RasterizerOpenGL : public VideoCore::RasterizerInterface { public: RasterizerOpenGL(); ~RasterizerOpenGL() override; - /// Initialize API-specific GPU objects void InitObjects() override; - - /// Reset the rasterizer, such as flushing all caches and updating all state void Reset() override; - - /// Queues the primitive formed by the given vertices for rendering void AddTriangle(const Pica::Shader::OutputVertex& v0, const Pica::Shader::OutputVertex& v1, const Pica::Shader::OutputVertex& v2) override; - - /// Draw the current batch of triangles void DrawTriangles() override; - - /// Commit the rasterizer's framebuffer contents immediately to the current 3DS memory framebuffer - void CommitFramebuffer() override; - - /// Notify rasterizer that the specified PICA register has been changed + void FlushFramebuffer() override; void NotifyPicaRegisterChanged(u32 id) override; - - /// Notify rasterizer that the specified 3DS memory region will be read from after this notification - void NotifyPreRead(PAddr addr, u32 size) override; - - /// Notify rasterizer that a 3DS memory region has been changed - void NotifyFlush(PAddr addr, u32 size) override; + void FlushRegion(PAddr addr, u32 size) override; + void InvalidateRegion(PAddr addr, u32 size) override; /// OpenGL shader generated for a given Pica register state struct PicaShader { diff --git a/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp b/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp index 10d4ab0b6..a9ad46fe0 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp +++ b/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp @@ -15,7 +15,7 @@ #include "video_core/renderer_opengl/pica_to_gl.h" RasterizerCacheOpenGL::~RasterizerCacheOpenGL() { - FullFlush(); + InvalidateAll(); } MICROPROFILE_DEFINE(OpenGL_TextureUpload, "OpenGL", "Texture Upload", MP_RGB(128, 64, 192)); @@ -58,8 +58,7 @@ void RasterizerCacheOpenGL::LoadAndBindTexture(OpenGLState &state, unsigned text } } -void RasterizerCacheOpenGL::NotifyFlush(PAddr addr, u32 size, bool ignore_hash) { - // Flush any texture that falls in the flushed region +void RasterizerCacheOpenGL::InvalidateInRange(PAddr addr, u32 size, bool ignore_hash) { // TODO: Optimize by also inserting upper bound (addr + size) of each texture into the same map and also narrow using lower_bound auto cache_upper_bound = texture_cache.upper_bound(addr + size); @@ -77,6 +76,6 @@ void RasterizerCacheOpenGL::NotifyFlush(PAddr addr, u32 size, bool ignore_hash) } } -void RasterizerCacheOpenGL::FullFlush() { +void RasterizerCacheOpenGL::InvalidateAll() { texture_cache.clear(); } diff --git a/src/video_core/renderer_opengl/gl_rasterizer_cache.h b/src/video_core/renderer_opengl/gl_rasterizer_cache.h index 98a48ffbe..b69651427 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer_cache.h +++ b/src/video_core/renderer_opengl/gl_rasterizer_cache.h @@ -23,11 +23,11 @@ public: LoadAndBindTexture(state, texture_unit, Pica::DebugUtils::TextureInfo::FromPicaRegister(config.config, config.format)); } - /// Flush any cached resource that touches the flushed region - void NotifyFlush(PAddr addr, u32 size, bool ignore_hash = false); + /// Invalidate any cached resource intersecting the specified region. + void InvalidateInRange(PAddr addr, u32 size, bool ignore_hash = false); - /// Flush all cached OpenGL resources tracked by this cache manager - void FullFlush(); + /// Invalidate all cached OpenGL resources tracked by this cache manager + void InvalidateAll(); private: struct CachedTexture { diff --git a/src/video_core/renderer_opengl/renderer_opengl.cpp b/src/video_core/renderer_opengl/renderer_opengl.cpp index 1420229cc..c14bdb8ab 100644 --- a/src/video_core/renderer_opengl/renderer_opengl.cpp +++ b/src/video_core/renderer_opengl/renderer_opengl.cpp @@ -93,7 +93,6 @@ static std::array<GLfloat, 3*2> MakeOrthographicMatrix(const float width, const /// RendererOpenGL constructor RendererOpenGL::RendererOpenGL() { - hw_rasterizer.reset(new RasterizerOpenGL()); resolution_width = std::max(VideoCore::kScreenTopWidth, VideoCore::kScreenBottomWidth); resolution_height = VideoCore::kScreenTopHeight + VideoCore::kScreenBottomHeight; } @@ -157,15 +156,7 @@ void RendererOpenGL::SwapBuffers() { profiler.BeginFrame(); - bool hw_renderer_enabled = VideoCore::g_hw_renderer_enabled; - if (Settings::values.use_hw_renderer != hw_renderer_enabled) { - // TODO: Save new setting value to config file for next startup - Settings::values.use_hw_renderer = hw_renderer_enabled; - - if (Settings::values.use_hw_renderer) { - hw_rasterizer->Reset(); - } - } + RefreshRasterizerSetting(); if (Pica::g_debug_context && Pica::g_debug_context->recorder) { Pica::g_debug_context->recorder->FrameFinished(); @@ -286,8 +277,6 @@ void RendererOpenGL::InitOpenGLObjects() { state.texture_units[0].texture_2d = 0; state.Apply(); - - hw_rasterizer->InitObjects(); } void RendererOpenGL::ConfigureFramebufferTexture(TextureInfo& texture, @@ -419,6 +408,8 @@ void RendererOpenGL::Init() { LOG_INFO(Render_OpenGL, "GL_VENDOR: %s", glGetString(GL_VENDOR)); LOG_INFO(Render_OpenGL, "GL_RENDERER: %s", glGetString(GL_RENDERER)); InitOpenGLObjects(); + + RefreshRasterizerSetting(); } /// Shutdown the renderer diff --git a/src/video_core/swrasterizer.cpp b/src/video_core/swrasterizer.cpp new file mode 100644 index 000000000..03df15b01 --- /dev/null +++ b/src/video_core/swrasterizer.cpp @@ -0,0 +1,16 @@ +// Copyright 2015 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include "video_core/clipper.h" +#include "video_core/swrasterizer.h" + +namespace VideoCore { + +void SWRasterizer::AddTriangle(const Pica::Shader::OutputVertex& v0, + const Pica::Shader::OutputVertex& v1, + const Pica::Shader::OutputVertex& v2) { + Pica::Clipper::ProcessTriangle(v0, v1, v2); +} + +} diff --git a/src/video_core/swrasterizer.h b/src/video_core/swrasterizer.h new file mode 100644 index 000000000..e9a4e39c6 --- /dev/null +++ b/src/video_core/swrasterizer.h @@ -0,0 +1,26 @@ +// Copyright 2015 Citra 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/rasterizer_interface.h" + +namespace VideoCore { + +class SWRasterizer : public RasterizerInterface { + void InitObjects() override {} + void Reset() override {} + void AddTriangle(const Pica::Shader::OutputVertex& v0, + const Pica::Shader::OutputVertex& v1, + const Pica::Shader::OutputVertex& v2); + void DrawTriangles() override {} + void FlushFramebuffer() override {} + void NotifyPicaRegisterChanged(u32 id) override {} + void FlushRegion(PAddr addr, u32 size) override {} + void InvalidateRegion(PAddr addr, u32 size) override {} +}; + +} |