diff options
Diffstat (limited to 'src/video_core')
48 files changed, 3809 insertions, 3345 deletions
diff --git a/src/video_core/CMakeLists.txt b/src/video_core/CMakeLists.txt index d55b84ce0..5317719e8 100644 --- a/src/video_core/CMakeLists.txt +++ b/src/video_core/CMakeLists.txt @@ -1,26 +1,46 @@ set(SRCS + command_processor.cpp + debug_utils/debug_utils.cpp + pica.cpp + primitive_assembly.cpp + regs.cpp + renderer_base.cpp renderer_opengl/gl_rasterizer.cpp renderer_opengl/gl_rasterizer_cache.cpp renderer_opengl/gl_shader_gen.cpp renderer_opengl/gl_shader_util.cpp renderer_opengl/gl_state.cpp renderer_opengl/renderer_opengl.cpp - debug_utils/debug_utils.cpp - clipper.cpp - command_processor.cpp - pica.cpp - primitive_assembly.cpp - rasterizer.cpp - renderer_base.cpp shader/shader.cpp shader/shader_interpreter.cpp - swrasterizer.cpp + swrasterizer/clipper.cpp + swrasterizer/framebuffer.cpp + swrasterizer/rasterizer.cpp + swrasterizer/swrasterizer.cpp + swrasterizer/texturing.cpp + texture/etc1.cpp + texture/texture_decode.cpp vertex_loader.cpp video_core.cpp ) set(HEADERS + command_processor.h debug_utils/debug_utils.h + gpu_debugger.h + pica.h + pica_state.h + pica_types.h + primitive_assembly.h + rasterizer_interface.h + regs.h + regs_framebuffer.h + regs_lighting.h + regs_pipeline.h + regs_rasterizer.h + regs_shader.h + regs_texturing.h + renderer_base.h renderer_opengl/gl_rasterizer.h renderer_opengl/gl_rasterizer_cache.h renderer_opengl/gl_resource_manager.h @@ -29,20 +49,16 @@ set(HEADERS renderer_opengl/gl_state.h renderer_opengl/pica_to_gl.h renderer_opengl/renderer_opengl.h - clipper.h - command_processor.h - gpu_debugger.h - pica.h - pica_state.h - pica_types.h - primitive_assembly.h - rasterizer.h - rasterizer_interface.h - renderer_base.h shader/debug_data.h shader/shader.h shader/shader_interpreter.h - swrasterizer.h + swrasterizer/clipper.h + swrasterizer/framebuffer.h + swrasterizer/rasterizer.h + swrasterizer/swrasterizer.h + swrasterizer/texturing.h + texture/etc1.h + texture/texture_decode.h utils.h vertex_loader.h video_core.h diff --git a/src/video_core/command_processor.cpp b/src/video_core/command_processor.cpp index eb79974a8..2e32ff905 100644 --- a/src/video_core/command_processor.cpp +++ b/src/video_core/command_processor.cpp @@ -16,11 +16,13 @@ #include "core/tracer/recorder.h" #include "video_core/command_processor.h" #include "video_core/debug_utils/debug_utils.h" -#include "video_core/pica.h" #include "video_core/pica_state.h" #include "video_core/pica_types.h" #include "video_core/primitive_assembly.h" #include "video_core/rasterizer_interface.h" +#include "video_core/regs.h" +#include "video_core/regs_pipeline.h" +#include "video_core/regs_texturing.h" #include "video_core/renderer_base.h" #include "video_core/shader/shader.h" #include "video_core/vertex_loader.h" @@ -49,19 +51,23 @@ MICROPROFILE_DEFINE(GPU_Drawing, "GPU", "Drawing", MP_RGB(50, 50, 240)); static void WritePicaReg(u32 id, u32 value, u32 mask) { auto& regs = g_state.regs; - if (id >= regs.NumIds()) + if (id >= Regs::NUM_REGS) { + LOG_ERROR(HW_GPU, + "Commandlist tried to write to invalid register 0x%03X (value: %08X, mask: %X)", + id, value, mask); return; + } // TODO: Figure out how register masking acts on e.g. vs.uniform_setup.set_value - u32 old_value = regs[id]; + u32 old_value = regs.reg_array[id]; const u32 write_mask = expand_bits_to_bytes[mask]; - regs[id] = (old_value & ~write_mask) | (value & write_mask); + regs.reg_array[id] = (old_value & ~write_mask) | (value & write_mask); // Double check for is_pica_tracing to avoid call overhead if (DebugUtils::IsPicaTracing()) { - DebugUtils::OnPicaRegWrite({(u16)id, (u16)mask, regs[id]}); + DebugUtils::OnPicaRegWrite({(u16)id, (u16)mask, regs.reg_array[id]}); } if (g_debug_context) @@ -74,23 +80,23 @@ static void WritePicaReg(u32 id, u32 value, u32 mask) { Service::GSP::SignalInterrupt(Service::GSP::InterruptId::P3D); break; - case PICA_REG_INDEX_WORKAROUND(triangle_topology, 0x25E): - g_state.primitive_assembler.Reconfigure(regs.triangle_topology); + case PICA_REG_INDEX(pipeline.triangle_topology): + g_state.primitive_assembler.Reconfigure(regs.pipeline.triangle_topology); break; - case PICA_REG_INDEX_WORKAROUND(restart_primitive, 0x25F): + case PICA_REG_INDEX(pipeline.restart_primitive): g_state.primitive_assembler.Reset(); break; - case PICA_REG_INDEX_WORKAROUND(vs_default_attributes_setup.index, 0x232): + case PICA_REG_INDEX(pipeline.vs_default_attributes_setup.index): g_state.immediate.current_attribute = 0; default_attr_counter = 0; break; // Load default vertex input attributes - case PICA_REG_INDEX_WORKAROUND(vs_default_attributes_setup.set_value[0], 0x233): - case PICA_REG_INDEX_WORKAROUND(vs_default_attributes_setup.set_value[1], 0x234): - case PICA_REG_INDEX_WORKAROUND(vs_default_attributes_setup.set_value[2], 0x235): { + case PICA_REG_INDEX_WORKAROUND(pipeline.vs_default_attributes_setup.set_value[0], 0x233): + case PICA_REG_INDEX_WORKAROUND(pipeline.vs_default_attributes_setup.set_value[1], 0x234): + case PICA_REG_INDEX_WORKAROUND(pipeline.vs_default_attributes_setup.set_value[2], 0x235): { // TODO: Does actual hardware indeed keep an intermediate buffer or does // it directly write the values? default_attr_write_buffer[default_attr_counter++] = value; @@ -102,7 +108,7 @@ static void WritePicaReg(u32 id, u32 value, u32 mask) { if (default_attr_counter >= 3) { default_attr_counter = 0; - auto& setup = regs.vs_default_attributes_setup; + auto& setup = regs.pipeline.vs_default_attributes_setup; if (setup.index >= 16) { LOG_ERROR(HW_GPU, "Invalid VS default attribute index %d", (int)setup.index); @@ -125,20 +131,21 @@ static void WritePicaReg(u32 id, u32 value, u32 mask) { // TODO: Verify that this actually modifies the register! if (setup.index < 15) { - g_state.vs_default_attributes[setup.index] = attribute; + g_state.input_default_attributes.attr[setup.index] = attribute; setup.index++; } else { - // Put each attribute into an immediate input buffer. - // When all specified immediate attributes are present, the Vertex Shader is invoked - // and everything is - // sent to the primitive assembler. + // Put each attribute into an immediate input buffer. When all specified immediate + // attributes are present, the Vertex Shader is invoked and everything is sent to + // the primitive assembler. auto& immediate_input = g_state.immediate.input_vertex; auto& immediate_attribute_id = g_state.immediate.current_attribute; - immediate_input.attr[immediate_attribute_id++] = attribute; + immediate_input.attr[immediate_attribute_id] = attribute; - if (immediate_attribute_id >= regs.vs.num_input_attributes + 1) { + if (immediate_attribute_id < regs.pipeline.max_input_attrib_index) { + immediate_attribute_id += 1; + } else { MICROPROFILE_SCOPE(GPU_Drawing); immediate_attribute_id = 0; @@ -150,10 +157,11 @@ static void WritePicaReg(u32 id, u32 value, u32 mask) { g_debug_context->OnEvent(DebugContext::Event::VertexShaderInvocation, static_cast<void*>(&immediate_input)); Shader::UnitState shader_unit; - shader_unit.LoadInputVertex(immediate_input, regs.vs.num_input_attributes + 1); + Shader::AttributeBuffer output{}; + + shader_unit.LoadInput(regs.vs, immediate_input); shader_engine->Run(g_state.vs, shader_unit); - auto output_vertex = Shader::OutputVertex::FromRegisters( - shader_unit.registers.output, regs, regs.vs.output_mask); + shader_unit.WriteOutput(regs.vs, output); // Send to renderer using Pica::Shader::OutputVertex; @@ -162,15 +170,17 @@ static void WritePicaReg(u32 id, u32 value, u32 mask) { VideoCore::g_renderer->Rasterizer()->AddTriangle(v0, v1, v2); }; - g_state.primitive_assembler.SubmitVertex(output_vertex, AddTriangle); + g_state.primitive_assembler.SubmitVertex( + Shader::OutputVertex::FromAttributeBuffer(regs.rasterizer, output), + AddTriangle); } } } break; } - case PICA_REG_INDEX(gpu_mode): - if (regs.gpu_mode == Regs::GPUMode::Configuring) { + case PICA_REG_INDEX(pipeline.gpu_mode): + if (regs.pipeline.gpu_mode == PipelineRegs::GPUMode::Configuring) { MICROPROFILE_SCOPE(GPU_Drawing); // Draw immediate mode triangles when GPU Mode is set to GPUMode::Configuring @@ -182,19 +192,20 @@ static void WritePicaReg(u32 id, u32 value, u32 mask) { } break; - case PICA_REG_INDEX_WORKAROUND(command_buffer.trigger[0], 0x23c): - case PICA_REG_INDEX_WORKAROUND(command_buffer.trigger[1], 0x23d): { - unsigned index = static_cast<unsigned>(id - PICA_REG_INDEX(command_buffer.trigger[0])); - u32* head_ptr = - (u32*)Memory::GetPhysicalPointer(regs.command_buffer.GetPhysicalAddress(index)); + case PICA_REG_INDEX_WORKAROUND(pipeline.command_buffer.trigger[0], 0x23c): + case PICA_REG_INDEX_WORKAROUND(pipeline.command_buffer.trigger[1], 0x23d): { + unsigned index = + static_cast<unsigned>(id - PICA_REG_INDEX(pipeline.command_buffer.trigger[0])); + u32* head_ptr = (u32*)Memory::GetPhysicalPointer( + regs.pipeline.command_buffer.GetPhysicalAddress(index)); g_state.cmd_list.head_ptr = g_state.cmd_list.current_ptr = head_ptr; - g_state.cmd_list.length = regs.command_buffer.GetSize(index) / sizeof(u32); + g_state.cmd_list.length = regs.pipeline.command_buffer.GetSize(index) / sizeof(u32); break; } // It seems like these trigger vertex rendering - case PICA_REG_INDEX(trigger_draw): - case PICA_REG_INDEX(trigger_draw_indexed): { + case PICA_REG_INDEX(pipeline.trigger_draw): + case PICA_REG_INDEX(pipeline.trigger_draw_indexed): { MICROPROFILE_SCOPE(GPU_Drawing); #if PICA_LOG_TEV @@ -206,13 +217,13 @@ static void WritePicaReg(u32 id, u32 value, u32 mask) { // Processes information about internal vertex attributes to figure out how a vertex is // loaded. // Later, these can be compiled and cached. - const u32 base_address = regs.vertex_attributes.GetPhysicalBaseAddress(); - VertexLoader loader(regs); + const u32 base_address = regs.pipeline.vertex_attributes.GetPhysicalBaseAddress(); + VertexLoader loader(regs.pipeline); // Load vertices - bool is_indexed = (id == PICA_REG_INDEX(trigger_draw_indexed)); + bool is_indexed = (id == PICA_REG_INDEX(pipeline.trigger_draw_indexed)); - const auto& index_info = regs.index_array; + const auto& index_info = regs.pipeline.index_array; const u8* index_address_8 = Memory::GetPhysicalPointer(base_address + index_info.offset); const u16* index_address_16 = reinterpret_cast<const u16*>(index_address_8); bool index_u16 = index_info.format != 0; @@ -221,13 +232,13 @@ static void WritePicaReg(u32 id, u32 value, u32 mask) { if (g_debug_context && g_debug_context->recorder) { for (int i = 0; i < 3; ++i) { - const auto texture = regs.GetTextures()[i]; + const auto texture = regs.texturing.GetTextures()[i]; if (!texture.enabled) continue; u8* texture_data = Memory::GetPhysicalPointer(texture.config.GetPhysicalAddress()); g_debug_context->recorder->MemoryAccessed( - texture_data, Pica::Regs::NibblesPerPixel(texture.format) * + texture_data, Pica::TexturingRegs::NibblesPerPixel(texture.format) * texture.config.width / 2 * texture.config.height, texture.config.GetPhysicalAddress()); } @@ -250,11 +261,11 @@ static void WritePicaReg(u32 id, u32 value, u32 mask) { shader_engine->SetupBatch(g_state.vs, regs.vs.main_offset); - for (unsigned int index = 0; index < regs.num_vertices; ++index) { + for (unsigned int index = 0; index < regs.pipeline.num_vertices; ++index) { // Indexed rendering doesn't use the start offset unsigned int vertex = is_indexed ? (index_u16 ? index_address_16[index] : index_address_8[index]) - : (index + regs.vertex_offset); + : (index + regs.pipeline.vertex_offset); // -1 is a common special value used for primitive restart. Since it's unknown if // the PICA supports it, and it would mess up the caching, guard against it here. @@ -280,19 +291,19 @@ static void WritePicaReg(u32 id, u32 value, u32 mask) { if (!vertex_cache_hit) { // Initialize data for the current vertex - Shader::InputVertex input; + Shader::AttributeBuffer input, output{}; loader.LoadVertex(base_address, index, vertex, input, memory_accesses); // Send to vertex shader if (g_debug_context) g_debug_context->OnEvent(DebugContext::Event::VertexShaderInvocation, (void*)&input); - shader_unit.LoadInputVertex(input, loader.GetNumTotalAttributes()); + shader_unit.LoadInput(regs.vs, input); shader_engine->Run(g_state.vs, shader_unit); + shader_unit.WriteOutput(regs.vs, output); // Retrieve vertex from register data - output_vertex = Shader::OutputVertex::FromRegisters(shader_unit.registers.output, - regs, regs.vs.output_mask); + output_vertex = Shader::OutputVertex::FromAttributeBuffer(regs.rasterizer, output); if (is_indexed) { vertex_cache[vertex_cache_pos] = output_vertex; @@ -434,16 +445,16 @@ static void WritePicaReg(u32 id, u32 value, u32 mask) { break; } - case PICA_REG_INDEX_WORKAROUND(fog_lut_data[0], 0xe8): - case PICA_REG_INDEX_WORKAROUND(fog_lut_data[1], 0xe9): - case PICA_REG_INDEX_WORKAROUND(fog_lut_data[2], 0xea): - case PICA_REG_INDEX_WORKAROUND(fog_lut_data[3], 0xeb): - case PICA_REG_INDEX_WORKAROUND(fog_lut_data[4], 0xec): - case PICA_REG_INDEX_WORKAROUND(fog_lut_data[5], 0xed): - case PICA_REG_INDEX_WORKAROUND(fog_lut_data[6], 0xee): - case PICA_REG_INDEX_WORKAROUND(fog_lut_data[7], 0xef): { - g_state.fog.lut[regs.fog_lut_offset % 128].raw = value; - regs.fog_lut_offset.Assign(regs.fog_lut_offset + 1); + case PICA_REG_INDEX_WORKAROUND(texturing.fog_lut_data[0], 0xe8): + case PICA_REG_INDEX_WORKAROUND(texturing.fog_lut_data[1], 0xe9): + case PICA_REG_INDEX_WORKAROUND(texturing.fog_lut_data[2], 0xea): + case PICA_REG_INDEX_WORKAROUND(texturing.fog_lut_data[3], 0xeb): + case PICA_REG_INDEX_WORKAROUND(texturing.fog_lut_data[4], 0xec): + case PICA_REG_INDEX_WORKAROUND(texturing.fog_lut_data[5], 0xed): + case PICA_REG_INDEX_WORKAROUND(texturing.fog_lut_data[6], 0xee): + case PICA_REG_INDEX_WORKAROUND(texturing.fog_lut_data[7], 0xef): { + g_state.fog.lut[regs.texturing.fog_lut_offset % 128].raw = value; + regs.texturing.fog_lut_offset.Assign(regs.texturing.fog_lut_offset + 1); break; } diff --git a/src/video_core/debug_utils/debug_utils.cpp b/src/video_core/debug_utils/debug_utils.cpp index c44b3d95a..47dbc8cc8 100644 --- a/src/video_core/debug_utils/debug_utils.cpp +++ b/src/video_core/debug_utils/debug_utils.cpp @@ -29,12 +29,15 @@ #include "common/math_util.h" #include "common/vector_math.h" #include "video_core/debug_utils/debug_utils.h" -#include "video_core/pica.h" #include "video_core/pica_state.h" #include "video_core/pica_types.h" #include "video_core/rasterizer_interface.h" +#include "video_core/regs_rasterizer.h" +#include "video_core/regs_shader.h" +#include "video_core/regs_texturing.h" #include "video_core/renderer_base.h" #include "video_core/shader/shader.h" +#include "video_core/texture/texture_decode.h" #include "video_core/utils.h" #include "video_core/video_core.h" @@ -87,9 +90,9 @@ std::shared_ptr<DebugContext> g_debug_context; // TODO: Get rid of this global namespace DebugUtils { -void DumpShader(const std::string& filename, const Regs::ShaderConfig& config, +void DumpShader(const std::string& filename, const ShaderRegs& config, const Shader::ShaderSetup& setup, - const Regs::VSOutputAttributes* output_attributes) { + const RasterizerRegs::VSOutputAttributes* output_attributes) { struct StuffToWrite { const u8* pointer; u32 size; @@ -128,7 +131,7 @@ void DumpShader(const std::string& filename, const Regs::ShaderConfig& config, // This is put into a try-catch block to make sure we notice unknown configurations. std::vector<OutputRegisterInfo> output_info_table; for (unsigned i = 0; i < 7; ++i) { - using OutputAttributes = Pica::Regs::VSOutputAttributes; + using OutputAttributes = Pica::RasterizerRegs::VSOutputAttributes; // TODO: It's still unclear how the attribute components map to the register! // Once we know that, this code probably will not make much sense anymore. @@ -315,257 +318,6 @@ std::unique_ptr<PicaTrace> FinishPicaTracing() { return ret; } -const Math::Vec4<u8> LookupTexture(const u8* source, int x, int y, const TextureInfo& info, - bool disable_alpha) { - const unsigned int coarse_x = x & ~7; - const unsigned int coarse_y = y & ~7; - - if (info.format != Regs::TextureFormat::ETC1 && info.format != Regs::TextureFormat::ETC1A4) { - // TODO(neobrain): Fix code design to unify vertical block offsets! - source += coarse_y * info.stride; - } - - // TODO: Assert that width/height are multiples of block dimensions - - switch (info.format) { - case Regs::TextureFormat::RGBA8: { - auto res = Color::DecodeRGBA8(source + VideoCore::GetMortonOffset(x, y, 4)); - return {res.r(), res.g(), res.b(), static_cast<u8>(disable_alpha ? 255 : res.a())}; - } - - case Regs::TextureFormat::RGB8: { - auto res = Color::DecodeRGB8(source + VideoCore::GetMortonOffset(x, y, 3)); - return {res.r(), res.g(), res.b(), 255}; - } - - case Regs::TextureFormat::RGB5A1: { - auto res = Color::DecodeRGB5A1(source + VideoCore::GetMortonOffset(x, y, 2)); - return {res.r(), res.g(), res.b(), static_cast<u8>(disable_alpha ? 255 : res.a())}; - } - - case Regs::TextureFormat::RGB565: { - auto res = Color::DecodeRGB565(source + VideoCore::GetMortonOffset(x, y, 2)); - return {res.r(), res.g(), res.b(), 255}; - } - - case Regs::TextureFormat::RGBA4: { - auto res = Color::DecodeRGBA4(source + VideoCore::GetMortonOffset(x, y, 2)); - return {res.r(), res.g(), res.b(), static_cast<u8>(disable_alpha ? 255 : res.a())}; - } - - case Regs::TextureFormat::IA8: { - const u8* source_ptr = source + VideoCore::GetMortonOffset(x, y, 2); - - if (disable_alpha) { - // Show intensity as red, alpha as green - return {source_ptr[1], source_ptr[0], 0, 255}; - } else { - return {source_ptr[1], source_ptr[1], source_ptr[1], source_ptr[0]}; - } - } - - case Regs::TextureFormat::RG8: { - auto res = Color::DecodeRG8(source + VideoCore::GetMortonOffset(x, y, 2)); - return {res.r(), res.g(), 0, 255}; - } - - case Regs::TextureFormat::I8: { - const u8* source_ptr = source + VideoCore::GetMortonOffset(x, y, 1); - return {*source_ptr, *source_ptr, *source_ptr, 255}; - } - - case Regs::TextureFormat::A8: { - const u8* source_ptr = source + VideoCore::GetMortonOffset(x, y, 1); - - if (disable_alpha) { - return {*source_ptr, *source_ptr, *source_ptr, 255}; - } else { - return {0, 0, 0, *source_ptr}; - } - } - - case Regs::TextureFormat::IA4: { - const u8* source_ptr = source + VideoCore::GetMortonOffset(x, y, 1); - - u8 i = Color::Convert4To8(((*source_ptr) & 0xF0) >> 4); - u8 a = Color::Convert4To8((*source_ptr) & 0xF); - - if (disable_alpha) { - // Show intensity as red, alpha as green - return {i, a, 0, 255}; - } else { - return {i, i, i, a}; - } - } - - case Regs::TextureFormat::I4: { - u32 morton_offset = VideoCore::GetMortonOffset(x, y, 1); - const u8* source_ptr = source + morton_offset / 2; - - u8 i = (morton_offset % 2) ? ((*source_ptr & 0xF0) >> 4) : (*source_ptr & 0xF); - i = Color::Convert4To8(i); - - return {i, i, i, 255}; - } - - case Regs::TextureFormat::A4: { - u32 morton_offset = VideoCore::GetMortonOffset(x, y, 1); - const u8* source_ptr = source + morton_offset / 2; - - u8 a = (morton_offset % 2) ? ((*source_ptr & 0xF0) >> 4) : (*source_ptr & 0xF); - a = Color::Convert4To8(a); - - if (disable_alpha) { - return {a, a, a, 255}; - } else { - return {0, 0, 0, a}; - } - } - - case Regs::TextureFormat::ETC1: - case Regs::TextureFormat::ETC1A4: { - bool has_alpha = (info.format == Regs::TextureFormat::ETC1A4); - - // ETC1 further subdivides each 8x8 tile into four 4x4 subtiles - const int subtile_width = 4; - const int subtile_height = 4; - - int subtile_index = ((x / subtile_width) & 1) + 2 * ((y / subtile_height) & 1); - unsigned subtile_bytes = has_alpha ? 2 : 1; // TODO: Name... - - const u64* source_ptr = (const u64*)(source + coarse_x * subtile_bytes * 4 + - coarse_y * subtile_bytes * 4 * (info.width / 8) + - subtile_index * subtile_bytes * 8); - u64 alpha = 0xFFFFFFFFFFFFFFFF; - if (has_alpha) { - alpha = *source_ptr; - source_ptr++; - } - - union ETC1Tile { - // Each of these two is a collection of 16 bits (one per lookup value) - BitField<0, 16, u64> table_subindexes; - BitField<16, 16, u64> negation_flags; - - unsigned GetTableSubIndex(unsigned index) const { - return (table_subindexes >> index) & 1; - } - - bool GetNegationFlag(unsigned index) const { - return ((negation_flags >> index) & 1) == 1; - } - - BitField<32, 1, u64> flip; - BitField<33, 1, u64> differential_mode; - - BitField<34, 3, u64> table_index_2; - BitField<37, 3, u64> table_index_1; - - union { - // delta value + base value - BitField<40, 3, s64> db; - BitField<43, 5, u64> b; - - BitField<48, 3, s64> dg; - BitField<51, 5, u64> g; - - BitField<56, 3, s64> dr; - BitField<59, 5, u64> r; - } differential; - - union { - BitField<40, 4, u64> b2; - BitField<44, 4, u64> b1; - - BitField<48, 4, u64> g2; - BitField<52, 4, u64> g1; - - BitField<56, 4, u64> r2; - BitField<60, 4, u64> r1; - } separate; - - const Math::Vec3<u8> GetRGB(int x, int y) const { - int texel = 4 * x + y; - - if (flip) - std::swap(x, y); - - // Lookup base value - Math::Vec3<int> ret; - if (differential_mode) { - ret.r() = static_cast<int>(differential.r); - ret.g() = static_cast<int>(differential.g); - ret.b() = static_cast<int>(differential.b); - if (x >= 2) { - ret.r() += static_cast<int>(differential.dr); - ret.g() += static_cast<int>(differential.dg); - ret.b() += static_cast<int>(differential.db); - } - ret.r() = Color::Convert5To8(ret.r()); - ret.g() = Color::Convert5To8(ret.g()); - ret.b() = Color::Convert5To8(ret.b()); - } else { - if (x < 2) { - ret.r() = Color::Convert4To8(static_cast<u8>(separate.r1)); - ret.g() = Color::Convert4To8(static_cast<u8>(separate.g1)); - ret.b() = Color::Convert4To8(static_cast<u8>(separate.b1)); - } else { - ret.r() = Color::Convert4To8(static_cast<u8>(separate.r2)); - ret.g() = Color::Convert4To8(static_cast<u8>(separate.g2)); - ret.b() = Color::Convert4To8(static_cast<u8>(separate.b2)); - } - } - - // Add modifier - unsigned table_index = - static_cast<int>((x < 2) ? table_index_1.Value() : table_index_2.Value()); - - static const std::array<std::array<u8, 2>, 8> etc1_modifier_table = {{ - {{2, 8}}, - {{5, 17}}, - {{9, 29}}, - {{13, 42}}, - {{18, 60}}, - {{24, 80}}, - {{33, 106}}, - {{47, 183}}, - }}; - - int modifier = etc1_modifier_table.at(table_index).at(GetTableSubIndex(texel)); - if (GetNegationFlag(texel)) - modifier *= -1; - - ret.r() = MathUtil::Clamp(ret.r() + modifier, 0, 255); - ret.g() = MathUtil::Clamp(ret.g() + modifier, 0, 255); - ret.b() = MathUtil::Clamp(ret.b() + modifier, 0, 255); - - return ret.Cast<u8>(); - } - } const* etc1_tile = reinterpret_cast<const ETC1Tile*>(source_ptr); - - alpha >>= 4 * ((x & 3) * 4 + (y & 3)); - return Math::MakeVec(etc1_tile->GetRGB(x & 3, y & 3), - disable_alpha ? (u8)255 : Color::Convert4To8(alpha & 0xF)); - } - - default: - LOG_ERROR(HW_GPU, "Unknown texture format: %x", (u32)info.format); - DEBUG_ASSERT(false); - return {}; - } -} - -TextureInfo TextureInfo::FromPicaRegister(const Regs::TextureConfig& config, - const Regs::TextureFormat& format) { - TextureInfo info; - info.physical_address = config.GetPhysicalAddress(); - info.width = config.width; - info.height = config.height; - info.format = format; - info.stride = Pica::Regs::NibblesPerPixel(info.format) * info.width / 2; - return info; -} - #ifdef HAVE_PNG // Adapter functions to libpng to write/flush to File::IOFile instances. static void WriteIOFile(png_structp png_ptr, png_bytep data, png_size_t length) { @@ -581,7 +333,7 @@ static void FlushIOFile(png_structp png_ptr) { } #endif -void DumpTexture(const Pica::Regs::TextureConfig& texture_config, u8* data) { +void DumpTexture(const TexturingRegs::TextureConfig& texture_config, u8* data) { #ifndef HAVE_PNG return; #else @@ -642,12 +394,12 @@ void DumpTexture(const Pica::Regs::TextureConfig& texture_config, u8* data) { buf = new u8[row_stride * texture_config.height]; for (unsigned y = 0; y < texture_config.height; ++y) { for (unsigned x = 0; x < texture_config.width; ++x) { - TextureInfo info; + Pica::Texture::TextureInfo info; info.width = texture_config.width; info.height = texture_config.height; info.stride = row_stride; - info.format = g_state.regs.texture0_format; - Math::Vec4<u8> texture_color = LookupTexture(data, x, y, info); + info.format = g_state.regs.texturing.texture0_format; + Math::Vec4<u8> texture_color = Pica::Texture::LookupTexture(data, x, y, info); buf[3 * x + y * row_stride] = texture_color.r(); buf[3 * x + y * row_stride + 1] = texture_color.g(); buf[3 * x + y * row_stride + 2] = texture_color.b(); @@ -684,8 +436,10 @@ static std::string ReplacePattern(const std::string& input, const std::string& p return ret; } -static std::string GetTevStageConfigSourceString(const Pica::Regs::TevStageConfig::Source& source) { - using Source = Pica::Regs::TevStageConfig::Source; +static std::string GetTevStageConfigSourceString( + const TexturingRegs::TevStageConfig::Source& source) { + + using Source = TexturingRegs::TevStageConfig::Source; static const std::map<Source, std::string> source_map = { {Source::PrimaryColor, "PrimaryColor"}, {Source::PrimaryFragmentColor, "PrimaryFragmentColor"}, @@ -707,9 +461,10 @@ static std::string GetTevStageConfigSourceString(const Pica::Regs::TevStageConfi } static std::string GetTevStageConfigColorSourceString( - const Pica::Regs::TevStageConfig::Source& source, - const Pica::Regs::TevStageConfig::ColorModifier modifier) { - using ColorModifier = Pica::Regs::TevStageConfig::ColorModifier; + const TexturingRegs::TevStageConfig::Source& source, + const TexturingRegs::TevStageConfig::ColorModifier modifier) { + + using ColorModifier = TexturingRegs::TevStageConfig::ColorModifier; static const std::map<ColorModifier, std::string> color_modifier_map = { {ColorModifier::SourceColor, "%source.rgb"}, {ColorModifier::OneMinusSourceColor, "(1.0 - %source.rgb)"}, @@ -733,9 +488,10 @@ static std::string GetTevStageConfigColorSourceString( } static std::string GetTevStageConfigAlphaSourceString( - const Pica::Regs::TevStageConfig::Source& source, - const Pica::Regs::TevStageConfig::AlphaModifier modifier) { - using AlphaModifier = Pica::Regs::TevStageConfig::AlphaModifier; + const TexturingRegs::TevStageConfig::Source& source, + const TexturingRegs::TevStageConfig::AlphaModifier modifier) { + + using AlphaModifier = TexturingRegs::TevStageConfig::AlphaModifier; static const std::map<AlphaModifier, std::string> alpha_modifier_map = { {AlphaModifier::SourceAlpha, "%source.a"}, {AlphaModifier::OneMinusSourceAlpha, "(1.0 - %source.a)"}, @@ -757,8 +513,9 @@ static std::string GetTevStageConfigAlphaSourceString( } static std::string GetTevStageConfigOperationString( - const Pica::Regs::TevStageConfig::Operation& operation) { - using Operation = Pica::Regs::TevStageConfig::Operation; + const TexturingRegs::TevStageConfig::Operation& operation) { + + using Operation = TexturingRegs::TevStageConfig::Operation; static const std::map<Operation, std::string> combiner_map = { {Operation::Replace, "%source1"}, {Operation::Modulate, "(%source1 * %source2)"}, @@ -778,7 +535,7 @@ static std::string GetTevStageConfigOperationString( return op_it->second; } -std::string GetTevStageConfigColorCombinerString(const Pica::Regs::TevStageConfig& tev_stage) { +std::string GetTevStageConfigColorCombinerString(const TexturingRegs::TevStageConfig& tev_stage) { auto op_str = GetTevStageConfigOperationString(tev_stage.color_op); op_str = ReplacePattern( op_str, "%source1", @@ -791,7 +548,7 @@ std::string GetTevStageConfigColorCombinerString(const Pica::Regs::TevStageConfi GetTevStageConfigColorSourceString(tev_stage.color_source3, tev_stage.color_modifier3)); } -std::string GetTevStageConfigAlphaCombinerString(const Pica::Regs::TevStageConfig& tev_stage) { +std::string GetTevStageConfigAlphaCombinerString(const TexturingRegs::TevStageConfig& tev_stage) { auto op_str = GetTevStageConfigOperationString(tev_stage.alpha_op); op_str = ReplacePattern( op_str, "%source1", @@ -804,7 +561,7 @@ std::string GetTevStageConfigAlphaCombinerString(const Pica::Regs::TevStageConfi GetTevStageConfigAlphaSourceString(tev_stage.alpha_source3, tev_stage.alpha_modifier3)); } -void DumpTevStageConfig(const std::array<Pica::Regs::TevStageConfig, 6>& stages) { +void DumpTevStageConfig(const std::array<TexturingRegs::TevStageConfig, 6>& stages) { std::string stage_info = "Tev setup:\n"; for (size_t index = 0; index < stages.size(); ++index) { const auto& tev_stage = stages[index]; diff --git a/src/video_core/debug_utils/debug_utils.h b/src/video_core/debug_utils/debug_utils.h index 46ea8d9c7..c1f29c527 100644 --- a/src/video_core/debug_utils/debug_utils.h +++ b/src/video_core/debug_utils/debug_utils.h @@ -17,7 +17,9 @@ #include <vector> #include "common/common_types.h" #include "common/vector_math.h" -#include "video_core/pica.h" +#include "video_core/regs_rasterizer.h" +#include "video_core/regs_shader.h" +#include "video_core/regs_texturing.h" namespace CiTrace { class Recorder; @@ -85,7 +87,7 @@ public: * @param data Optional data pointer (if unused, this is a nullptr) * @note This function will perform nothing unless it is overridden in the child class. */ - virtual void OnPicaBreakPointHit(Event, void*) {} + virtual void OnPicaBreakPointHit(Event event, void* data) {} /** * Action to perform when emulation is resumed from a breakpoint. @@ -182,9 +184,9 @@ namespace DebugUtils { #define PICA_DUMP_TEXTURES 0 #define PICA_LOG_TEV 0 -void DumpShader(const std::string& filename, const Regs::ShaderConfig& config, +void DumpShader(const std::string& filename, const ShaderRegs& config, const Shader::ShaderSetup& setup, - const Regs::VSOutputAttributes* output_attributes); + const RasterizerRegs::VSOutputAttributes* output_attributes); // Utility class to log Pica commands. struct PicaTrace { @@ -205,38 +207,13 @@ inline bool IsPicaTracing() { void OnPicaRegWrite(PicaTrace::Write write); std::unique_ptr<PicaTrace> FinishPicaTracing(); -struct TextureInfo { - PAddr physical_address; - int width; - int height; - int stride; - Pica::Regs::TextureFormat format; +void DumpTexture(const TexturingRegs::TextureConfig& texture_config, u8* data); - static TextureInfo FromPicaRegister(const Pica::Regs::TextureConfig& config, - const Pica::Regs::TextureFormat& format); -}; - -/** - * Lookup texel located at the given coordinates and return an RGBA vector of its color. - * @param source Source pointer to read data from - * @param s,t Texture coordinates to read from - * @param info TextureInfo object describing the texture setup - * @param disable_alpha This is used for debug widgets which use this method to display textures - * without providing a good way to visualize alpha by themselves. If true, this will return 255 for - * the alpha component, and either drop the information entirely or store it in an "unused" color - * channel. - * @todo Eventually we should get rid of the disable_alpha parameter. - */ -const Math::Vec4<u8> LookupTexture(const u8* source, int s, int t, const TextureInfo& info, - bool disable_alpha = false); - -void DumpTexture(const Pica::Regs::TextureConfig& texture_config, u8* data); - -std::string GetTevStageConfigColorCombinerString(const Pica::Regs::TevStageConfig& tev_stage); -std::string GetTevStageConfigAlphaCombinerString(const Pica::Regs::TevStageConfig& tev_stage); +std::string GetTevStageConfigColorCombinerString(const TexturingRegs::TevStageConfig& tev_stage); +std::string GetTevStageConfigAlphaCombinerString(const TexturingRegs::TevStageConfig& tev_stage); /// Dumps the Tev stage config to log at trace level -void DumpTevStageConfig(const std::array<Pica::Regs::TevStageConfig, 6>& stages); +void DumpTevStageConfig(const std::array<TexturingRegs::TevStageConfig, 6>& stages); /** * Used in the vertex loader to merge access records. TODO: Investigate if actually useful. diff --git a/src/video_core/pica.cpp b/src/video_core/pica.cpp index b4a77c632..b95148a6a 100644 --- a/src/video_core/pica.cpp +++ b/src/video_core/pica.cpp @@ -3,497 +3,14 @@ // Refer to the license.txt file included. #include <cstring> -#include <iterator> -#include <unordered_map> -#include <utility> #include "video_core/pica.h" #include "video_core/pica_state.h" -#include "video_core/primitive_assembly.h" -#include "video_core/shader/shader.h" +#include "video_core/regs_pipeline.h" namespace Pica { State g_state; -static const std::pair<u16, const char*> register_names[] = { - {0x010, "GPUREG_FINALIZE"}, - - {0x040, "GPUREG_FACECULLING_CONFIG"}, - {0x041, "GPUREG_VIEWPORT_WIDTH"}, - {0x042, "GPUREG_VIEWPORT_INVW"}, - {0x043, "GPUREG_VIEWPORT_HEIGHT"}, - {0x044, "GPUREG_VIEWPORT_INVH"}, - - {0x047, "GPUREG_FRAGOP_CLIP"}, - {0x048, "GPUREG_FRAGOP_CLIP_DATA0"}, - {0x049, "GPUREG_FRAGOP_CLIP_DATA1"}, - {0x04A, "GPUREG_FRAGOP_CLIP_DATA2"}, - {0x04B, "GPUREG_FRAGOP_CLIP_DATA3"}, - - {0x04D, "GPUREG_DEPTHMAP_SCALE"}, - {0x04E, "GPUREG_DEPTHMAP_OFFSET"}, - {0x04F, "GPUREG_SH_OUTMAP_TOTAL"}, - {0x050, "GPUREG_SH_OUTMAP_O0"}, - {0x051, "GPUREG_SH_OUTMAP_O1"}, - {0x052, "GPUREG_SH_OUTMAP_O2"}, - {0x053, "GPUREG_SH_OUTMAP_O3"}, - {0x054, "GPUREG_SH_OUTMAP_O4"}, - {0x055, "GPUREG_SH_OUTMAP_O5"}, - {0x056, "GPUREG_SH_OUTMAP_O6"}, - - {0x061, "GPUREG_EARLYDEPTH_FUNC"}, - {0x062, "GPUREG_EARLYDEPTH_TEST1"}, - {0x063, "GPUREG_EARLYDEPTH_CLEAR"}, - {0x064, "GPUREG_SH_OUTATTR_MODE"}, - {0x065, "GPUREG_SCISSORTEST_MODE"}, - {0x066, "GPUREG_SCISSORTEST_POS"}, - {0x067, "GPUREG_SCISSORTEST_DIM"}, - {0x068, "GPUREG_VIEWPORT_XY"}, - - {0x06A, "GPUREG_EARLYDEPTH_DATA"}, - - {0x06D, "GPUREG_DEPTHMAP_ENABLE"}, - {0x06E, "GPUREG_RENDERBUF_DIM"}, - {0x06F, "GPUREG_SH_OUTATTR_CLOCK"}, - - {0x080, "GPUREG_TEXUNIT_CONFIG"}, - {0x081, "GPUREG_TEXUNIT0_BORDER_COLOR"}, - {0x082, "GPUREG_TEXUNIT0_DIM"}, - {0x083, "GPUREG_TEXUNIT0_PARAM"}, - {0x084, "GPUREG_TEXUNIT0_LOD"}, - {0x085, "GPUREG_TEXUNIT0_ADDR1"}, - {0x086, "GPUREG_TEXUNIT0_ADDR2"}, - {0x087, "GPUREG_TEXUNIT0_ADDR3"}, - {0x088, "GPUREG_TEXUNIT0_ADDR4"}, - {0x089, "GPUREG_TEXUNIT0_ADDR5"}, - {0x08A, "GPUREG_TEXUNIT0_ADDR6"}, - {0x08B, "GPUREG_TEXUNIT0_SHADOW"}, - - {0x08E, "GPUREG_TEXUNIT0_TYPE"}, - {0x08F, "GPUREG_LIGHTING_ENABLE0"}, - - {0x091, "GPUREG_TEXUNIT1_BORDER_COLOR"}, - {0x092, "GPUREG_TEXUNIT1_DIM"}, - {0x093, "GPUREG_TEXUNIT1_PARAM"}, - {0x094, "GPUREG_TEXUNIT1_LOD"}, - {0x095, "GPUREG_TEXUNIT1_ADDR"}, - {0x096, "GPUREG_TEXUNIT1_TYPE"}, - - {0x099, "GPUREG_TEXUNIT2_BORDER_COLOR"}, - {0x09A, "GPUREG_TEXUNIT2_DIM"}, - {0x09B, "GPUREG_TEXUNIT2_PARAM"}, - {0x09C, "GPUREG_TEXUNIT2_LOD"}, - {0x09D, "GPUREG_TEXUNIT2_ADDR"}, - {0x09E, "GPUREG_TEXUNIT2_TYPE"}, - - {0x0A8, "GPUREG_TEXUNIT3_PROCTEX0"}, - {0x0A9, "GPUREG_TEXUNIT3_PROCTEX1"}, - {0x0AA, "GPUREG_TEXUNIT3_PROCTEX2"}, - {0x0AB, "GPUREG_TEXUNIT3_PROCTEX3"}, - {0x0AC, "GPUREG_TEXUNIT3_PROCTEX4"}, - {0x0AD, "GPUREG_TEXUNIT3_PROCTEX5"}, - - {0x0AF, "GPUREG_PROCTEX_LUT"}, - {0x0B0, "GPUREG_PROCTEX_LUT_DATA0"}, - {0x0B1, "GPUREG_PROCTEX_LUT_DATA1"}, - {0x0B2, "GPUREG_PROCTEX_LUT_DATA2"}, - {0x0B3, "GPUREG_PROCTEX_LUT_DATA3"}, - {0x0B4, "GPUREG_PROCTEX_LUT_DATA4"}, - {0x0B5, "GPUREG_PROCTEX_LUT_DATA5"}, - {0x0B6, "GPUREG_PROCTEX_LUT_DATA6"}, - {0x0B7, "GPUREG_PROCTEX_LUT_DATA7"}, - - {0x0C0, "GPUREG_TEXENV0_SOURCE"}, - {0x0C1, "GPUREG_TEXENV0_OPERAND"}, - {0x0C2, "GPUREG_TEXENV0_COMBINER"}, - {0x0C3, "GPUREG_TEXENV0_COLOR"}, - {0x0C4, "GPUREG_TEXENV0_SCALE"}, - - {0x0C8, "GPUREG_TEXENV1_SOURCE"}, - {0x0C9, "GPUREG_TEXENV1_OPERAND"}, - {0x0CA, "GPUREG_TEXENV1_COMBINER"}, - {0x0CB, "GPUREG_TEXENV1_COLOR"}, - {0x0CC, "GPUREG_TEXENV1_SCALE"}, - - {0x0D0, "GPUREG_TEXENV2_SOURCE"}, - {0x0D1, "GPUREG_TEXENV2_OPERAND"}, - {0x0D2, "GPUREG_TEXENV2_COMBINER"}, - {0x0D3, "GPUREG_TEXENV2_COLOR"}, - {0x0D4, "GPUREG_TEXENV2_SCALE"}, - - {0x0D8, "GPUREG_TEXENV3_SOURCE"}, - {0x0D9, "GPUREG_TEXENV3_OPERAND"}, - {0x0DA, "GPUREG_TEXENV3_COMBINER"}, - {0x0DB, "GPUREG_TEXENV3_COLOR"}, - {0x0DC, "GPUREG_TEXENV3_SCALE"}, - - {0x0E0, "GPUREG_TEXENV_UPDATE_BUFFER"}, - {0x0E1, "GPUREG_FOG_COLOR"}, - - {0x0E4, "GPUREG_GAS_ATTENUATION"}, - {0x0E5, "GPUREG_GAS_ACCMAX"}, - {0x0E6, "GPUREG_FOG_LUT_INDEX"}, - - {0x0E8, "GPUREG_FOG_LUT_DATA0"}, - {0x0E9, "GPUREG_FOG_LUT_DATA1"}, - {0x0EA, "GPUREG_FOG_LUT_DATA2"}, - {0x0EB, "GPUREG_FOG_LUT_DATA3"}, - {0x0EC, "GPUREG_FOG_LUT_DATA4"}, - {0x0ED, "GPUREG_FOG_LUT_DATA5"}, - {0x0EE, "GPUREG_FOG_LUT_DATA6"}, - {0x0EF, "GPUREG_FOG_LUT_DATA7"}, - {0x0F0, "GPUREG_TEXENV4_SOURCE"}, - {0x0F1, "GPUREG_TEXENV4_OPERAND"}, - {0x0F2, "GPUREG_TEXENV4_COMBINER"}, - {0x0F3, "GPUREG_TEXENV4_COLOR"}, - {0x0F4, "GPUREG_TEXENV4_SCALE"}, - - {0x0F8, "GPUREG_TEXENV5_SOURCE"}, - {0x0F9, "GPUREG_TEXENV5_OPERAND"}, - {0x0FA, "GPUREG_TEXENV5_COMBINER"}, - {0x0FB, "GPUREG_TEXENV5_COLOR"}, - {0x0FC, "GPUREG_TEXENV5_SCALE"}, - {0x0FD, "GPUREG_TEXENV_BUFFER_COLOR"}, - - {0x100, "GPUREG_COLOR_OPERATION"}, - {0x101, "GPUREG_BLEND_FUNC"}, - {0x102, "GPUREG_LOGIC_OP"}, - {0x103, "GPUREG_BLEND_COLOR"}, - {0x104, "GPUREG_FRAGOP_ALPHA_TEST"}, - {0x105, "GPUREG_STENCIL_TEST"}, - {0x106, "GPUREG_STENCIL_OP"}, - {0x107, "GPUREG_DEPTH_COLOR_MASK"}, - - {0x110, "GPUREG_FRAMEBUFFER_INVALIDATE"}, - {0x111, "GPUREG_FRAMEBUFFER_FLUSH"}, - {0x112, "GPUREG_COLORBUFFER_READ"}, - {0x113, "GPUREG_COLORBUFFER_WRITE"}, - {0x114, "GPUREG_DEPTHBUFFER_READ"}, - {0x115, "GPUREG_DEPTHBUFFER_WRITE"}, - {0x116, "GPUREG_DEPTHBUFFER_FORMAT"}, - {0x117, "GPUREG_COLORBUFFER_FORMAT"}, - {0x118, "GPUREG_EARLYDEPTH_TEST2"}, - - {0x11B, "GPUREG_FRAMEBUFFER_BLOCK32"}, - {0x11C, "GPUREG_DEPTHBUFFER_LOC"}, - {0x11D, "GPUREG_COLORBUFFER_LOC"}, - {0x11E, "GPUREG_FRAMEBUFFER_DIM"}, - - {0x120, "GPUREG_GAS_LIGHT_XY"}, - {0x121, "GPUREG_GAS_LIGHT_Z"}, - {0x122, "GPUREG_GAS_LIGHT_Z_COLOR"}, - {0x123, "GPUREG_GAS_LUT_INDEX"}, - {0x124, "GPUREG_GAS_LUT_DATA"}, - - {0x126, "GPUREG_GAS_DELTAZ_DEPTH"}, - - {0x130, "GPUREG_FRAGOP_SHADOW"}, - - {0x140, "GPUREG_LIGHT0_SPECULAR0"}, - {0x141, "GPUREG_LIGHT0_SPECULAR1"}, - {0x142, "GPUREG_LIGHT0_DIFFUSE"}, - {0x143, "GPUREG_LIGHT0_AMBIENT"}, - {0x144, "GPUREG_LIGHT0_XY"}, - {0x145, "GPUREG_LIGHT0_Z"}, - {0x146, "GPUREG_LIGHT0_SPOTDIR_XY"}, - {0x147, "GPUREG_LIGHT0_SPOTDIR_Z"}, - - {0x149, "GPUREG_LIGHT0_CONFIG"}, - {0x14A, "GPUREG_LIGHT0_ATTENUATION_BIAS"}, - {0x14B, "GPUREG_LIGHT0_ATTENUATION_SCALE"}, - - {0x150, "GPUREG_LIGHT1_SPECULAR0"}, - {0x151, "GPUREG_LIGHT1_SPECULAR1"}, - {0x152, "GPUREG_LIGHT1_DIFFUSE"}, - {0x153, "GPUREG_LIGHT1_AMBIENT"}, - {0x154, "GPUREG_LIGHT1_XY"}, - {0x155, "GPUREG_LIGHT1_Z"}, - {0x156, "GPUREG_LIGHT1_SPOTDIR_XY"}, - {0x157, "GPUREG_LIGHT1_SPOTDIR_Z"}, - - {0x159, "GPUREG_LIGHT1_CONFIG"}, - {0x15A, "GPUREG_LIGHT1_ATTENUATION_BIAS"}, - {0x15B, "GPUREG_LIGHT1_ATTENUATION_SCALE"}, - - {0x160, "GPUREG_LIGHT2_SPECULAR0"}, - {0x161, "GPUREG_LIGHT2_SPECULAR1"}, - {0x162, "GPUREG_LIGHT2_DIFFUSE"}, - {0x163, "GPUREG_LIGHT2_AMBIENT"}, - {0x164, "GPUREG_LIGHT2_XY"}, - {0x165, "GPUREG_LIGHT2_Z"}, - {0x166, "GPUREG_LIGHT2_SPOTDIR_XY"}, - {0x167, "GPUREG_LIGHT2_SPOTDIR_Z"}, - - {0x169, "GPUREG_LIGHT2_CONFIG"}, - {0x16A, "GPUREG_LIGHT2_ATTENUATION_BIAS"}, - {0x16B, "GPUREG_LIGHT2_ATTENUATION_SCALE"}, - - {0x170, "GPUREG_LIGHT3_SPECULAR0"}, - {0x171, "GPUREG_LIGHT3_SPECULAR1"}, - {0x172, "GPUREG_LIGHT3_DIFFUSE"}, - {0x173, "GPUREG_LIGHT3_AMBIENT"}, - {0x174, "GPUREG_LIGHT3_XY"}, - {0x175, "GPUREG_LIGHT3_Z"}, - {0x176, "GPUREG_LIGHT3_SPOTDIR_XY"}, - {0x177, "GPUREG_LIGHT3_SPOTDIR_Z"}, - - {0x179, "GPUREG_LIGHT3_CONFIG"}, - {0x17A, "GPUREG_LIGHT3_ATTENUATION_BIAS"}, - {0x17B, "GPUREG_LIGHT3_ATTENUATION_SCALE"}, - - {0x180, "GPUREG_LIGHT4_SPECULAR0"}, - {0x181, "GPUREG_LIGHT4_SPECULAR1"}, - {0x182, "GPUREG_LIGHT4_DIFFUSE"}, - {0x183, "GPUREG_LIGHT4_AMBIENT"}, - {0x184, "GPUREG_LIGHT4_XY"}, - {0x185, "GPUREG_LIGHT4_Z"}, - {0x186, "GPUREG_LIGHT4_SPOTDIR_XY"}, - {0x187, "GPUREG_LIGHT4_SPOTDIR_Z"}, - - {0x189, "GPUREG_LIGHT4_CONFIG"}, - {0x18A, "GPUREG_LIGHT4_ATTENUATION_BIAS"}, - {0x18B, "GPUREG_LIGHT4_ATTENUATION_SCALE"}, - - {0x190, "GPUREG_LIGHT5_SPECULAR0"}, - {0x191, "GPUREG_LIGHT5_SPECULAR1"}, - {0x192, "GPUREG_LIGHT5_DIFFUSE"}, - {0x193, "GPUREG_LIGHT5_AMBIENT"}, - {0x194, "GPUREG_LIGHT5_XY"}, - {0x195, "GPUREG_LIGHT5_Z"}, - {0x196, "GPUREG_LIGHT5_SPOTDIR_XY"}, - {0x197, "GPUREG_LIGHT5_SPOTDIR_Z"}, - - {0x199, "GPUREG_LIGHT5_CONFIG"}, - {0x19A, "GPUREG_LIGHT5_ATTENUATION_BIAS"}, - {0x19B, "GPUREG_LIGHT5_ATTENUATION_SCALE"}, - - {0x1A0, "GPUREG_LIGHT6_SPECULAR0"}, - {0x1A1, "GPUREG_LIGHT6_SPECULAR1"}, - {0x1A2, "GPUREG_LIGHT6_DIFFUSE"}, - {0x1A3, "GPUREG_LIGHT6_AMBIENT"}, - {0x1A4, "GPUREG_LIGHT6_XY"}, - {0x1A5, "GPUREG_LIGHT6_Z"}, - {0x1A6, "GPUREG_LIGHT6_SPOTDIR_XY"}, - {0x1A7, "GPUREG_LIGHT6_SPOTDIR_Z"}, - - {0x1A9, "GPUREG_LIGHT6_CONFIG"}, - {0x1AA, "GPUREG_LIGHT6_ATTENUATION_BIAS"}, - {0x1AB, "GPUREG_LIGHT6_ATTENUATION_SCALE"}, - - {0x1B0, "GPUREG_LIGHT7_SPECULAR0"}, - {0x1B1, "GPUREG_LIGHT7_SPECULAR1"}, - {0x1B2, "GPUREG_LIGHT7_DIFFUSE"}, - {0x1B3, "GPUREG_LIGHT7_AMBIENT"}, - {0x1B4, "GPUREG_LIGHT7_XY"}, - {0x1B5, "GPUREG_LIGHT7_Z"}, - {0x1B6, "GPUREG_LIGHT7_SPOTDIR_XY"}, - {0x1B7, "GPUREG_LIGHT7_SPOTDIR_Z"}, - - {0x1B9, "GPUREG_LIGHT7_CONFIG"}, - {0x1BA, "GPUREG_LIGHT7_ATTENUATION_BIAS"}, - {0x1BB, "GPUREG_LIGHT7_ATTENUATION_SCALE"}, - - {0x1C0, "GPUREG_LIGHTING_AMBIENT"}, - - {0x1C2, "GPUREG_LIGHTING_NUM_LIGHTS"}, - {0x1C3, "GPUREG_LIGHTING_CONFIG0"}, - {0x1C4, "GPUREG_LIGHTING_CONFIG1"}, - {0x1C5, "GPUREG_LIGHTING_LUT_INDEX"}, - {0x1C6, "GPUREG_LIGHTING_ENABLE1"}, - - {0x1C8, "GPUREG_LIGHTING_LUT_DATA0"}, - {0x1C9, "GPUREG_LIGHTING_LUT_DATA1"}, - {0x1CA, "GPUREG_LIGHTING_LUT_DATA2"}, - {0x1CB, "GPUREG_LIGHTING_LUT_DATA3"}, - {0x1CC, "GPUREG_LIGHTING_LUT_DATA4"}, - {0x1CD, "GPUREG_LIGHTING_LUT_DATA5"}, - {0x1CE, "GPUREG_LIGHTING_LUT_DATA6"}, - {0x1CF, "GPUREG_LIGHTING_LUT_DATA7"}, - {0x1D0, "GPUREG_LIGHTING_LUTINPUT_ABS"}, - {0x1D1, "GPUREG_LIGHTING_LUTINPUT_SELECT"}, - {0x1D2, "GPUREG_LIGHTING_LUTINPUT_SCALE"}, - - {0x1D9, "GPUREG_LIGHTING_LIGHT_PERMUTATION"}, - - {0x200, "GPUREG_ATTRIBBUFFERS_LOC"}, - {0x201, "GPUREG_ATTRIBBUFFERS_FORMAT_LOW"}, - {0x202, "GPUREG_ATTRIBBUFFERS_FORMAT_HIGH"}, - {0x203, "GPUREG_ATTRIBBUFFER0_OFFSET"}, - {0x204, "GPUREG_ATTRIBBUFFER0_CONFIG1"}, - {0x205, "GPUREG_ATTRIBBUFFER0_CONFIG2"}, - {0x206, "GPUREG_ATTRIBBUFFER1_OFFSET"}, - {0x207, "GPUREG_ATTRIBBUFFER1_CONFIG1"}, - {0x208, "GPUREG_ATTRIBBUFFER1_CONFIG2"}, - {0x209, "GPUREG_ATTRIBBUFFER2_OFFSET"}, - {0x20A, "GPUREG_ATTRIBBUFFER2_CONFIG1"}, - {0x20B, "GPUREG_ATTRIBBUFFER2_CONFIG2"}, - {0x20C, "GPUREG_ATTRIBBUFFER3_OFFSET"}, - {0x20D, "GPUREG_ATTRIBBUFFER3_CONFIG1"}, - {0x20E, "GPUREG_ATTRIBBUFFER3_CONFIG2"}, - {0x20F, "GPUREG_ATTRIBBUFFER4_OFFSET"}, - {0x210, "GPUREG_ATTRIBBUFFER4_CONFIG1"}, - {0x211, "GPUREG_ATTRIBBUFFER4_CONFIG2"}, - {0x212, "GPUREG_ATTRIBBUFFER5_OFFSET"}, - {0x213, "GPUREG_ATTRIBBUFFER5_CONFIG1"}, - {0x214, "GPUREG_ATTRIBBUFFER5_CONFIG2"}, - {0x215, "GPUREG_ATTRIBBUFFER6_OFFSET"}, - {0x216, "GPUREG_ATTRIBBUFFER6_CONFIG1"}, - {0x217, "GPUREG_ATTRIBBUFFER6_CONFIG2"}, - {0x218, "GPUREG_ATTRIBBUFFER7_OFFSET"}, - {0x219, "GPUREG_ATTRIBBUFFER7_CONFIG1"}, - {0x21A, "GPUREG_ATTRIBBUFFER7_CONFIG2"}, - {0x21B, "GPUREG_ATTRIBBUFFER8_OFFSET"}, - {0x21C, "GPUREG_ATTRIBBUFFER8_CONFIG1"}, - {0x21D, "GPUREG_ATTRIBBUFFER8_CONFIG2"}, - {0x21E, "GPUREG_ATTRIBBUFFER9_OFFSET"}, - {0x21F, "GPUREG_ATTRIBBUFFER9_CONFIG1"}, - {0x220, "GPUREG_ATTRIBBUFFER9_CONFIG2"}, - {0x221, "GPUREG_ATTRIBBUFFER10_OFFSET"}, - {0x222, "GPUREG_ATTRIBBUFFER10_CONFIG1"}, - {0x223, "GPUREG_ATTRIBBUFFER10_CONFIG2"}, - {0x224, "GPUREG_ATTRIBBUFFER11_OFFSET"}, - {0x225, "GPUREG_ATTRIBBUFFER11_CONFIG1"}, - {0x226, "GPUREG_ATTRIBBUFFER11_CONFIG2"}, - {0x227, "GPUREG_INDEXBUFFER_CONFIG"}, - {0x228, "GPUREG_NUMVERTICES"}, - {0x229, "GPUREG_GEOSTAGE_CONFIG"}, - {0x22A, "GPUREG_VERTEX_OFFSET"}, - - {0x22D, "GPUREG_POST_VERTEX_CACHE_NUM"}, - {0x22E, "GPUREG_DRAWARRAYS"}, - {0x22F, "GPUREG_DRAWELEMENTS"}, - - {0x231, "GPUREG_VTX_FUNC"}, - {0x232, "GPUREG_FIXEDATTRIB_INDEX"}, - {0x233, "GPUREG_FIXEDATTRIB_DATA0"}, - {0x234, "GPUREG_FIXEDATTRIB_DATA1"}, - {0x235, "GPUREG_FIXEDATTRIB_DATA2"}, - - {0x238, "GPUREG_CMDBUF_SIZE0"}, - {0x239, "GPUREG_CMDBUF_SIZE1"}, - {0x23A, "GPUREG_CMDBUF_ADDR0"}, - {0x23B, "GPUREG_CMDBUF_ADDR1"}, - {0x23C, "GPUREG_CMDBUF_JUMP0"}, - {0x23D, "GPUREG_CMDBUF_JUMP1"}, - - {0x242, "GPUREG_VSH_NUM_ATTR"}, - - {0x244, "GPUREG_VSH_COM_MODE"}, - {0x245, "GPUREG_START_DRAW_FUNC0"}, - - {0x24A, "GPUREG_VSH_OUTMAP_TOTAL1"}, - - {0x251, "GPUREG_VSH_OUTMAP_TOTAL2"}, - {0x252, "GPUREG_GSH_MISC0"}, - {0x253, "GPUREG_GEOSTAGE_CONFIG2"}, - {0x254, "GPUREG_GSH_MISC1"}, - - {0x25E, "GPUREG_PRIMITIVE_CONFIG"}, - {0x25F, "GPUREG_RESTART_PRIMITIVE"}, - - {0x280, "GPUREG_GSH_BOOLUNIFORM"}, - {0x281, "GPUREG_GSH_INTUNIFORM_I0"}, - {0x282, "GPUREG_GSH_INTUNIFORM_I1"}, - {0x283, "GPUREG_GSH_INTUNIFORM_I2"}, - {0x284, "GPUREG_GSH_INTUNIFORM_I3"}, - - {0x289, "GPUREG_GSH_INPUTBUFFER_CONFIG"}, - {0x28A, "GPUREG_GSH_ENTRYPOINT"}, - {0x28B, "GPUREG_GSH_ATTRIBUTES_PERMUTATION_LOW"}, - {0x28C, "GPUREG_GSH_ATTRIBUTES_PERMUTATION_HIGH"}, - {0x28D, "GPUREG_GSH_OUTMAP_MASK"}, - - {0x28F, "GPUREG_GSH_CODETRANSFER_END"}, - {0x290, "GPUREG_GSH_FLOATUNIFORM_INDEX"}, - {0x291, "GPUREG_GSH_FLOATUNIFORM_DATA0"}, - {0x292, "GPUREG_GSH_FLOATUNIFORM_DATA1"}, - {0x293, "GPUREG_GSH_FLOATUNIFORM_DATA2"}, - {0x294, "GPUREG_GSH_FLOATUNIFORM_DATA3"}, - {0x295, "GPUREG_GSH_FLOATUNIFORM_DATA4"}, - {0x296, "GPUREG_GSH_FLOATUNIFORM_DATA5"}, - {0x297, "GPUREG_GSH_FLOATUNIFORM_DATA6"}, - {0x298, "GPUREG_GSH_FLOATUNIFORM_DATA7"}, - - {0x29B, "GPUREG_GSH_CODETRANSFER_INDEX"}, - {0x29C, "GPUREG_GSH_CODETRANSFER_DATA0"}, - {0x29D, "GPUREG_GSH_CODETRANSFER_DATA1"}, - {0x29E, "GPUREG_GSH_CODETRANSFER_DATA2"}, - {0x29F, "GPUREG_GSH_CODETRANSFER_DATA3"}, - {0x2A0, "GPUREG_GSH_CODETRANSFER_DATA4"}, - {0x2A1, "GPUREG_GSH_CODETRANSFER_DATA5"}, - {0x2A2, "GPUREG_GSH_CODETRANSFER_DATA6"}, - {0x2A3, "GPUREG_GSH_CODETRANSFER_DATA7"}, - - {0x2A5, "GPUREG_GSH_OPDESCS_INDEX"}, - {0x2A6, "GPUREG_GSH_OPDESCS_DATA0"}, - {0x2A7, "GPUREG_GSH_OPDESCS_DATA1"}, - {0x2A8, "GPUREG_GSH_OPDESCS_DATA2"}, - {0x2A9, "GPUREG_GSH_OPDESCS_DATA3"}, - {0x2AA, "GPUREG_GSH_OPDESCS_DATA4"}, - {0x2AB, "GPUREG_GSH_OPDESCS_DATA5"}, - {0x2AC, "GPUREG_GSH_OPDESCS_DATA6"}, - {0x2AD, "GPUREG_GSH_OPDESCS_DATA7"}, - - {0x2B0, "GPUREG_VSH_BOOLUNIFORM"}, - {0x2B1, "GPUREG_VSH_INTUNIFORM_I0"}, - {0x2B2, "GPUREG_VSH_INTUNIFORM_I1"}, - {0x2B3, "GPUREG_VSH_INTUNIFORM_I2"}, - {0x2B4, "GPUREG_VSH_INTUNIFORM_I3"}, - - {0x2B9, "GPUREG_VSH_INPUTBUFFER_CONFIG"}, - {0x2BA, "GPUREG_VSH_ENTRYPOINT"}, - {0x2BB, "GPUREG_VSH_ATTRIBUTES_PERMUTATION_LOW"}, - {0x2BC, "GPUREG_VSH_ATTRIBUTES_PERMUTATION_HIGH"}, - {0x2BD, "GPUREG_VSH_OUTMAP_MASK"}, - - {0x2BF, "GPUREG_VSH_CODETRANSFER_END"}, - {0x2C0, "GPUREG_VSH_FLOATUNIFORM_INDEX"}, - {0x2C1, "GPUREG_VSH_FLOATUNIFORM_DATA0"}, - {0x2C2, "GPUREG_VSH_FLOATUNIFORM_DATA1"}, - {0x2C3, "GPUREG_VSH_FLOATUNIFORM_DATA2"}, - {0x2C4, "GPUREG_VSH_FLOATUNIFORM_DATA3"}, - {0x2C5, "GPUREG_VSH_FLOATUNIFORM_DATA4"}, - {0x2C6, "GPUREG_VSH_FLOATUNIFORM_DATA5"}, - {0x2C7, "GPUREG_VSH_FLOATUNIFORM_DATA6"}, - {0x2C8, "GPUREG_VSH_FLOATUNIFORM_DATA7"}, - - {0x2CB, "GPUREG_VSH_CODETRANSFER_INDEX"}, - {0x2CC, "GPUREG_VSH_CODETRANSFER_DATA0"}, - {0x2CD, "GPUREG_VSH_CODETRANSFER_DATA1"}, - {0x2CE, "GPUREG_VSH_CODETRANSFER_DATA2"}, - {0x2CF, "GPUREG_VSH_CODETRANSFER_DATA3"}, - {0x2D0, "GPUREG_VSH_CODETRANSFER_DATA4"}, - {0x2D1, "GPUREG_VSH_CODETRANSFER_DATA5"}, - {0x2D2, "GPUREG_VSH_CODETRANSFER_DATA6"}, - {0x2D3, "GPUREG_VSH_CODETRANSFER_DATA7"}, - - {0x2D5, "GPUREG_VSH_OPDESCS_INDEX"}, - {0x2D6, "GPUREG_VSH_OPDESCS_DATA0"}, - {0x2D7, "GPUREG_VSH_OPDESCS_DATA1"}, - {0x2D8, "GPUREG_VSH_OPDESCS_DATA2"}, - {0x2D9, "GPUREG_VSH_OPDESCS_DATA3"}, - {0x2DA, "GPUREG_VSH_OPDESCS_DATA4"}, - {0x2DB, "GPUREG_VSH_OPDESCS_DATA5"}, - {0x2DC, "GPUREG_VSH_OPDESCS_DATA6"}, - {0x2DD, "GPUREG_VSH_OPDESCS_DATA7"}, -}; - -std::string Regs::GetCommandName(int index) { - static std::unordered_map<u32, const char*> map; - - if (map.empty()) { - map.insert(std::begin(register_names), std::end(register_names)); - } - - // Return empty string if no match is found - auto it = map.find(index); - if (it != map.end()) { - return it->second; - } else { - return std::string(); - } -} - void Init() { g_state.Reset(); } @@ -513,6 +30,6 @@ void State::Reset() { Zero(gs); Zero(cmd_list); Zero(immediate); - primitive_assembler.Reconfigure(Regs::TriangleTopology::List); + primitive_assembler.Reconfigure(PipelineRegs::TriangleTopology::List); } } diff --git a/src/video_core/pica.h b/src/video_core/pica.h index b2db609ec..dc8aa6670 100644 --- a/src/video_core/pica.h +++ b/src/video_core/pica.h @@ -4,1412 +4,9 @@ #pragma once -#include <array> -#include <cstddef> -#include <string> - -#ifndef _MSC_VER -#include <type_traits> // for std::enable_if -#endif - -#include "common/assert.h" -#include "common/bit_field.h" -#include "common/common_funcs.h" -#include "common/common_types.h" -#include "common/logging/log.h" -#include "common/vector_math.h" - +#include "video_core/regs_texturing.h" namespace Pica { -// Returns index corresponding to the Regs member labeled by field_name -// TODO: Due to Visual studio bug 209229, offsetof does not return constant expressions -// when used with array elements (e.g. PICA_REG_INDEX(vs_uniform_setup.set_value[1])). -// For details cf. -// https://connect.microsoft.com/VisualStudio/feedback/details/209229/offsetof-does-not-produce-a-constant-expression-for-array-members -// Hopefully, this will be fixed sometime in the future. -// For lack of better alternatives, we currently hardcode the offsets when constant -// expressions are needed via PICA_REG_INDEX_WORKAROUND (on sane compilers, static_asserts -// will then make sure the offsets indeed match the automatically calculated ones). -#define PICA_REG_INDEX(field_name) (offsetof(Pica::Regs, field_name) / sizeof(u32)) -#if defined(_MSC_VER) -#define PICA_REG_INDEX_WORKAROUND(field_name, backup_workaround_index) (backup_workaround_index) -#else -// NOTE: Yeah, hacking in a static_assert here just to workaround the lacking MSVC compiler -// really is this annoying. This macro just forwards its first argument to PICA_REG_INDEX -// and then performs a (no-op) cast to size_t iff the second argument matches the expected -// field offset. Otherwise, the compiler will fail to compile this code. -#define PICA_REG_INDEX_WORKAROUND(field_name, backup_workaround_index) \ - ((typename std::enable_if<backup_workaround_index == PICA_REG_INDEX(field_name), \ - size_t>::type)PICA_REG_INDEX(field_name)) -#endif // _MSC_VER - -struct Regs { - - INSERT_PADDING_WORDS(0x10); - - u32 trigger_irq; - - INSERT_PADDING_WORDS(0x2f); - - enum class CullMode : u32 { - // Select which polygons are considered to be "frontfacing". - KeepAll = 0, - KeepClockWise = 1, - KeepCounterClockWise = 2, - // TODO: What does the third value imply? - }; - - union { - BitField<0, 2, CullMode> cull_mode; - }; - - BitField<0, 24, u32> viewport_size_x; - - INSERT_PADDING_WORDS(0x1); - - BitField<0, 24, u32> viewport_size_y; - - INSERT_PADDING_WORDS(0x9); - - BitField<0, 24, u32> viewport_depth_range; // float24 - BitField<0, 24, u32> viewport_depth_near_plane; // float24 - - BitField<0, 3, u32> vs_output_total; - - union VSOutputAttributes { - // Maps components of output vertex attributes to semantics - enum Semantic : u32 { - POSITION_X = 0, - POSITION_Y = 1, - POSITION_Z = 2, - POSITION_W = 3, - - QUATERNION_X = 4, - QUATERNION_Y = 5, - QUATERNION_Z = 6, - QUATERNION_W = 7, - - COLOR_R = 8, - COLOR_G = 9, - COLOR_B = 10, - COLOR_A = 11, - - TEXCOORD0_U = 12, - TEXCOORD0_V = 13, - TEXCOORD1_U = 14, - TEXCOORD1_V = 15, - - // TODO: Not verified - VIEW_X = 18, - VIEW_Y = 19, - VIEW_Z = 20, - - TEXCOORD2_U = 22, - TEXCOORD2_V = 23, - - INVALID = 31, - }; - - BitField<0, 5, Semantic> map_x; - BitField<8, 5, Semantic> map_y; - BitField<16, 5, Semantic> map_z; - BitField<24, 5, Semantic> map_w; - } vs_output_attributes[7]; - - INSERT_PADDING_WORDS(0xe); - - enum class ScissorMode : u32 { - Disabled = 0, - Exclude = 1, // Exclude pixels inside the scissor box - - Include = 3 // Exclude pixels outside the scissor box - }; - - struct { - BitField<0, 2, ScissorMode> mode; - - union { - BitField<0, 16, u32> x1; - BitField<16, 16, u32> y1; - }; - - union { - BitField<0, 16, u32> x2; - BitField<16, 16, u32> y2; - }; - } scissor_test; - - union { - BitField<0, 10, s32> x; - BitField<16, 10, s32> y; - } viewport_corner; - - INSERT_PADDING_WORDS(0x1); - - // TODO: early depth - INSERT_PADDING_WORDS(0x1); - - INSERT_PADDING_WORDS(0x2); - - enum DepthBuffering : u32 { - WBuffering = 0, - ZBuffering = 1, - }; - BitField<0, 1, DepthBuffering> depthmap_enable; - - INSERT_PADDING_WORDS(0x12); - - struct TextureConfig { - enum TextureType : u32 { - Texture2D = 0, - TextureCube = 1, - Shadow2D = 2, - Projection2D = 3, - ShadowCube = 4, - Disabled = 5, - }; - - enum WrapMode : u32 { - ClampToEdge = 0, - ClampToBorder = 1, - Repeat = 2, - MirroredRepeat = 3, - }; - - enum TextureFilter : u32 { - Nearest = 0, - Linear = 1, - }; - - union { - u32 raw; - BitField<0, 8, u32> r; - BitField<8, 8, u32> g; - BitField<16, 8, u32> b; - BitField<24, 8, u32> a; - } border_color; - - union { - BitField<0, 16, u32> height; - BitField<16, 16, u32> width; - }; - - union { - BitField<1, 1, TextureFilter> mag_filter; - BitField<2, 1, TextureFilter> min_filter; - BitField<8, 2, WrapMode> wrap_t; - BitField<12, 2, WrapMode> wrap_s; - BitField<28, 2, TextureType> - type; ///< @note Only valid for texture 0 according to 3DBrew. - }; - - INSERT_PADDING_WORDS(0x1); - - u32 address; - - u32 GetPhysicalAddress() const { - return DecodeAddressRegister(address); - } - - // texture1 and texture2 store the texture format directly after the address - // whereas texture0 inserts some additional flags inbetween. - // Hence, we store the format separately so that all other parameters can be described - // in a single structure. - }; - - enum class TextureFormat : u32 { - RGBA8 = 0, - RGB8 = 1, - RGB5A1 = 2, - RGB565 = 3, - RGBA4 = 4, - IA8 = 5, - RG8 = 6, ///< @note Also called HILO8 in 3DBrew. - I8 = 7, - A8 = 8, - IA4 = 9, - I4 = 10, - A4 = 11, - ETC1 = 12, // compressed - ETC1A4 = 13, // compressed - }; - - enum class LogicOp : u32 { - Clear = 0, - And = 1, - AndReverse = 2, - Copy = 3, - Set = 4, - CopyInverted = 5, - NoOp = 6, - Invert = 7, - Nand = 8, - Or = 9, - Nor = 10, - Xor = 11, - Equiv = 12, - AndInverted = 13, - OrReverse = 14, - OrInverted = 15, - }; - - static unsigned NibblesPerPixel(TextureFormat format) { - switch (format) { - case TextureFormat::RGBA8: - return 8; - - case TextureFormat::RGB8: - return 6; - - case TextureFormat::RGB5A1: - case TextureFormat::RGB565: - case TextureFormat::RGBA4: - case TextureFormat::IA8: - case TextureFormat::RG8: - return 4; - - case TextureFormat::I4: - case TextureFormat::A4: - return 1; - - case TextureFormat::I8: - case TextureFormat::A8: - case TextureFormat::IA4: - default: // placeholder for yet unknown formats - return 2; - } - } - - union { - BitField<0, 1, u32> texture0_enable; - BitField<1, 1, u32> texture1_enable; - BitField<2, 1, u32> texture2_enable; - }; - TextureConfig texture0; - INSERT_PADDING_WORDS(0x8); - BitField<0, 4, TextureFormat> texture0_format; - BitField<0, 1, u32> fragment_lighting_enable; - INSERT_PADDING_WORDS(0x1); - TextureConfig texture1; - BitField<0, 4, TextureFormat> texture1_format; - INSERT_PADDING_WORDS(0x2); - TextureConfig texture2; - BitField<0, 4, TextureFormat> texture2_format; - INSERT_PADDING_WORDS(0x21); - - struct FullTextureConfig { - const bool enabled; - const TextureConfig config; - const TextureFormat format; - }; - const std::array<FullTextureConfig, 3> GetTextures() const { - return {{ - {texture0_enable.ToBool(), texture0, texture0_format}, - {texture1_enable.ToBool(), texture1, texture1_format}, - {texture2_enable.ToBool(), texture2, texture2_format}, - }}; - } - - // 0xc0-0xff: Texture Combiner (akin to glTexEnv) - struct TevStageConfig { - enum class Source : u32 { - PrimaryColor = 0x0, - PrimaryFragmentColor = 0x1, - SecondaryFragmentColor = 0x2, - - Texture0 = 0x3, - Texture1 = 0x4, - Texture2 = 0x5, - Texture3 = 0x6, - - PreviousBuffer = 0xd, - Constant = 0xe, - Previous = 0xf, - }; - - enum class ColorModifier : u32 { - SourceColor = 0x0, - OneMinusSourceColor = 0x1, - SourceAlpha = 0x2, - OneMinusSourceAlpha = 0x3, - SourceRed = 0x4, - OneMinusSourceRed = 0x5, - - SourceGreen = 0x8, - OneMinusSourceGreen = 0x9, - - SourceBlue = 0xc, - OneMinusSourceBlue = 0xd, - }; - - enum class AlphaModifier : u32 { - SourceAlpha = 0x0, - OneMinusSourceAlpha = 0x1, - SourceRed = 0x2, - OneMinusSourceRed = 0x3, - SourceGreen = 0x4, - OneMinusSourceGreen = 0x5, - SourceBlue = 0x6, - OneMinusSourceBlue = 0x7, - }; - - enum class Operation : u32 { - Replace = 0, - Modulate = 1, - Add = 2, - AddSigned = 3, - Lerp = 4, - Subtract = 5, - Dot3_RGB = 6, - - MultiplyThenAdd = 8, - AddThenMultiply = 9, - }; - - union { - u32 sources_raw; - BitField<0, 4, Source> color_source1; - BitField<4, 4, Source> color_source2; - BitField<8, 4, Source> color_source3; - BitField<16, 4, Source> alpha_source1; - BitField<20, 4, Source> alpha_source2; - BitField<24, 4, Source> alpha_source3; - }; - - union { - u32 modifiers_raw; - BitField<0, 4, ColorModifier> color_modifier1; - BitField<4, 4, ColorModifier> color_modifier2; - BitField<8, 4, ColorModifier> color_modifier3; - BitField<12, 3, AlphaModifier> alpha_modifier1; - BitField<16, 3, AlphaModifier> alpha_modifier2; - BitField<20, 3, AlphaModifier> alpha_modifier3; - }; - - union { - u32 ops_raw; - BitField<0, 4, Operation> color_op; - BitField<16, 4, Operation> alpha_op; - }; - - union { - u32 const_color; - BitField<0, 8, u32> const_r; - BitField<8, 8, u32> const_g; - BitField<16, 8, u32> const_b; - BitField<24, 8, u32> const_a; - }; - - union { - u32 scales_raw; - BitField<0, 2, u32> color_scale; - BitField<16, 2, u32> alpha_scale; - }; - - inline unsigned GetColorMultiplier() const { - return (color_scale < 3) ? (1 << color_scale) : 1; - } - - inline unsigned GetAlphaMultiplier() const { - return (alpha_scale < 3) ? (1 << alpha_scale) : 1; - } - }; - - TevStageConfig tev_stage0; - INSERT_PADDING_WORDS(0x3); - TevStageConfig tev_stage1; - INSERT_PADDING_WORDS(0x3); - TevStageConfig tev_stage2; - INSERT_PADDING_WORDS(0x3); - TevStageConfig tev_stage3; - INSERT_PADDING_WORDS(0x3); - - enum class FogMode : u32 { - None = 0, - Fog = 5, - Gas = 7, - }; - - union { - BitField<0, 3, FogMode> fog_mode; - BitField<16, 1, u32> fog_flip; - - union { - // Tev stages 0-3 write their output to the combiner buffer if the corresponding bit in - // these masks are set - BitField<8, 4, u32> update_mask_rgb; - BitField<12, 4, u32> update_mask_a; - - bool TevStageUpdatesCombinerBufferColor(unsigned stage_index) const { - return (stage_index < 4) && (update_mask_rgb & (1 << stage_index)); - } - - bool TevStageUpdatesCombinerBufferAlpha(unsigned stage_index) const { - return (stage_index < 4) && (update_mask_a & (1 << stage_index)); - } - } tev_combiner_buffer_input; - }; - - union { - u32 raw; - BitField<0, 8, u32> r; - BitField<8, 8, u32> g; - BitField<16, 8, u32> b; - } fog_color; - - INSERT_PADDING_WORDS(0x4); - - BitField<0, 16, u32> fog_lut_offset; - - INSERT_PADDING_WORDS(0x1); - - u32 fog_lut_data[8]; - - TevStageConfig tev_stage4; - INSERT_PADDING_WORDS(0x3); - TevStageConfig tev_stage5; - - union { - u32 raw; - BitField<0, 8, u32> r; - BitField<8, 8, u32> g; - BitField<16, 8, u32> b; - BitField<24, 8, u32> a; - } tev_combiner_buffer_color; - - INSERT_PADDING_WORDS(0x2); - - const std::array<Regs::TevStageConfig, 6> GetTevStages() const { - return {{tev_stage0, tev_stage1, tev_stage2, tev_stage3, tev_stage4, tev_stage5}}; - }; - - enum class BlendEquation : u32 { - Add = 0, - Subtract = 1, - ReverseSubtract = 2, - Min = 3, - Max = 4, - }; - - enum class BlendFactor : u32 { - Zero = 0, - One = 1, - SourceColor = 2, - OneMinusSourceColor = 3, - DestColor = 4, - OneMinusDestColor = 5, - SourceAlpha = 6, - OneMinusSourceAlpha = 7, - DestAlpha = 8, - OneMinusDestAlpha = 9, - ConstantColor = 10, - OneMinusConstantColor = 11, - ConstantAlpha = 12, - OneMinusConstantAlpha = 13, - SourceAlphaSaturate = 14, - }; - - enum class CompareFunc : u32 { - Never = 0, - Always = 1, - Equal = 2, - NotEqual = 3, - LessThan = 4, - LessThanOrEqual = 5, - GreaterThan = 6, - GreaterThanOrEqual = 7, - }; - - enum class StencilAction : u32 { - Keep = 0, - Zero = 1, - Replace = 2, - Increment = 3, - Decrement = 4, - Invert = 5, - IncrementWrap = 6, - DecrementWrap = 7, - }; - - struct { - union { - // If false, logic blending is used - BitField<8, 1, u32> alphablend_enable; - }; - - union { - BitField<0, 8, BlendEquation> blend_equation_rgb; - BitField<8, 8, BlendEquation> blend_equation_a; - - BitField<16, 4, BlendFactor> factor_source_rgb; - BitField<20, 4, BlendFactor> factor_dest_rgb; - - BitField<24, 4, BlendFactor> factor_source_a; - BitField<28, 4, BlendFactor> factor_dest_a; - } alpha_blending; - - union { - BitField<0, 4, LogicOp> logic_op; - }; - - union { - u32 raw; - BitField<0, 8, u32> r; - BitField<8, 8, u32> g; - BitField<16, 8, u32> b; - BitField<24, 8, u32> a; - } blend_const; - - union { - BitField<0, 1, u32> enable; - BitField<4, 3, CompareFunc> func; - BitField<8, 8, u32> ref; - } alpha_test; - - struct { - union { - // Raw value of this register - u32 raw_func; - - // If true, enable stencil testing - BitField<0, 1, u32> enable; - - // Comparison operation for stencil testing - BitField<4, 3, CompareFunc> func; - - // Mask used to control writing to the stencil buffer - BitField<8, 8, u32> write_mask; - - // Value to compare against for stencil testing - BitField<16, 8, u32> reference_value; - - // Mask to apply on stencil test inputs - BitField<24, 8, u32> input_mask; - }; - - union { - // Raw value of this register - u32 raw_op; - - // Action to perform when the stencil test fails - BitField<0, 3, StencilAction> action_stencil_fail; - - // Action to perform when stencil testing passed but depth testing fails - BitField<4, 3, StencilAction> action_depth_fail; - - // Action to perform when both stencil and depth testing pass - BitField<8, 3, StencilAction> action_depth_pass; - }; - } stencil_test; - - union { - BitField<0, 1, u32> depth_test_enable; - BitField<4, 3, CompareFunc> depth_test_func; - BitField<8, 1, u32> red_enable; - BitField<9, 1, u32> green_enable; - BitField<10, 1, u32> blue_enable; - BitField<11, 1, u32> alpha_enable; - BitField<12, 1, u32> depth_write_enable; - }; - - INSERT_PADDING_WORDS(0x8); - } output_merger; - - // Components are laid out in reverse byte order, most significant bits first. - enum class ColorFormat : u32 { - RGBA8 = 0, - RGB8 = 1, - RGB5A1 = 2, - RGB565 = 3, - RGBA4 = 4, - }; - - enum class DepthFormat : u32 { - D16 = 0, - D24 = 2, - D24S8 = 3, - }; - - // Returns the number of bytes in the specified color format - static unsigned BytesPerColorPixel(ColorFormat format) { - switch (format) { - case ColorFormat::RGBA8: - return 4; - case ColorFormat::RGB8: - return 3; - case ColorFormat::RGB5A1: - case ColorFormat::RGB565: - case ColorFormat::RGBA4: - return 2; - default: - LOG_CRITICAL(HW_GPU, "Unknown color format %u", format); - UNIMPLEMENTED(); - } - } - - struct FramebufferConfig { - INSERT_PADDING_WORDS(0x3); - - union { - BitField<0, 4, u32> allow_color_write; // 0 = disable, else enable - }; - - INSERT_PADDING_WORDS(0x1); - - union { - BitField<0, 2, u32> allow_depth_stencil_write; // 0 = disable, else enable - }; - - DepthFormat depth_format; // TODO: Should be a BitField! - BitField<16, 3, ColorFormat> color_format; - - INSERT_PADDING_WORDS(0x4); - - u32 depth_buffer_address; - u32 color_buffer_address; - - union { - // Apparently, the framebuffer width is stored as expected, - // while the height is stored as the actual height minus one. - // Hence, don't access these fields directly but use the accessors - // GetWidth() and GetHeight() instead. - BitField<0, 11, u32> width; - BitField<12, 10, u32> height; - }; - - INSERT_PADDING_WORDS(0x1); - - inline u32 GetColorBufferPhysicalAddress() const { - return DecodeAddressRegister(color_buffer_address); - } - inline u32 GetDepthBufferPhysicalAddress() const { - return DecodeAddressRegister(depth_buffer_address); - } - - inline u32 GetWidth() const { - return width; - } - - inline u32 GetHeight() const { - return height + 1; - } - } framebuffer; - - // Returns the number of bytes in the specified depth format - static u32 BytesPerDepthPixel(DepthFormat format) { - switch (format) { - case DepthFormat::D16: - return 2; - case DepthFormat::D24: - return 3; - case DepthFormat::D24S8: - return 4; - default: - LOG_CRITICAL(HW_GPU, "Unknown depth format %u", format); - UNIMPLEMENTED(); - } - } - - // Returns the number of bits per depth component of the specified depth format - static u32 DepthBitsPerPixel(DepthFormat format) { - switch (format) { - case DepthFormat::D16: - return 16; - case DepthFormat::D24: - case DepthFormat::D24S8: - return 24; - default: - LOG_CRITICAL(HW_GPU, "Unknown depth format %u", format); - UNIMPLEMENTED(); - } - } - - INSERT_PADDING_WORDS(0x20); - - enum class LightingSampler { - Distribution0 = 0, - Distribution1 = 1, - Fresnel = 3, - ReflectBlue = 4, - ReflectGreen = 5, - ReflectRed = 6, - SpotlightAttenuation = 8, - DistanceAttenuation = 16, - }; - - /** - * Pica fragment lighting supports using different LUTs for each lighting component: - * Reflectance R, G, and B channels, distribution function for specular components 0 and 1, - * fresnel factor, and spotlight attenuation. Furthermore, which LUTs are used for each channel - * (or whether a channel is enabled at all) is specified by various pre-defined lighting - * configurations. With configurations that require more LUTs, more cycles are required on HW to - * perform lighting computations. - */ - enum class LightingConfig { - Config0 = 0, ///< Reflect Red, Distribution 0, Spotlight - Config1 = 1, ///< Reflect Red, Fresnel, Spotlight - Config2 = 2, ///< Reflect Red, Distribution 0/1 - Config3 = 3, ///< Distribution 0/1, Fresnel - Config4 = 4, ///< Reflect Red/Green/Blue, Distribution 0/1, Spotlight - Config5 = 5, ///< Reflect Red/Green/Blue, Distribution 0, Fresnel, Spotlight - Config6 = 6, ///< Reflect Red, Distribution 0/1, Fresnel, Spotlight - Config7 = 8, ///< Reflect Red/Green/Blue, Distribution 0/1, Fresnel, Spotlight - ///< NOTE: '8' is intentional, '7' does not appear to be a valid configuration - }; - - /// Selects which lighting components are affected by fresnel - enum class LightingFresnelSelector { - None = 0, ///< Fresnel is disabled - PrimaryAlpha = 1, ///< Primary (diffuse) lighting alpha is affected by fresnel - SecondaryAlpha = 2, ///< Secondary (specular) lighting alpha is affected by fresnel - Both = - PrimaryAlpha | - SecondaryAlpha, ///< Both primary and secondary lighting alphas are affected by fresnel - }; - - /// Factor used to scale the output of a lighting LUT - enum class LightingScale { - Scale1 = 0, ///< Scale is 1x - Scale2 = 1, ///< Scale is 2x - Scale4 = 2, ///< Scale is 4x - Scale8 = 3, ///< Scale is 8x - Scale1_4 = 6, ///< Scale is 0.25x - Scale1_2 = 7, ///< Scale is 0.5x - }; - - enum class LightingLutInput { - NH = 0, // Cosine of the angle between the normal and half-angle vectors - VH = 1, // Cosine of the angle between the view and half-angle vectors - NV = 2, // Cosine of the angle between the normal and the view vector - LN = 3, // Cosine of the angle between the light and the normal vectors - }; - - enum class LightingBumpMode : u32 { - None = 0, - NormalMap = 1, - TangentMap = 2, - }; - - union LightColor { - BitField<0, 10, u32> b; - BitField<10, 10, u32> g; - BitField<20, 10, u32> r; - - Math::Vec3f ToVec3f() const { - // These fields are 10 bits wide, however 255 corresponds to 1.0f for each color - // component - return Math::MakeVec((f32)r / 255.f, (f32)g / 255.f, (f32)b / 255.f); - } - }; - - /// Returns true if the specified lighting sampler is supported by the current Pica lighting - /// configuration - static bool IsLightingSamplerSupported(LightingConfig config, LightingSampler sampler) { - switch (sampler) { - case LightingSampler::Distribution0: - return (config != LightingConfig::Config1); - - case LightingSampler::Distribution1: - return (config != LightingConfig::Config0) && (config != LightingConfig::Config1) && - (config != LightingConfig::Config5); - - case LightingSampler::Fresnel: - return (config != LightingConfig::Config0) && (config != LightingConfig::Config2) && - (config != LightingConfig::Config4); - - case LightingSampler::ReflectRed: - return (config != LightingConfig::Config3); - - case LightingSampler::ReflectGreen: - case LightingSampler::ReflectBlue: - return (config == LightingConfig::Config4) || (config == LightingConfig::Config5) || - (config == LightingConfig::Config7); - default: - UNREACHABLE_MSG("Regs::IsLightingSamplerSupported: Reached " - "unreachable section, sampler should be one " - "of Distribution0, Distribution1, Fresnel, " - "ReflectRed, ReflectGreen or ReflectBlue, instead " - "got %i", - static_cast<int>(config)); - } - } - - struct { - struct LightSrc { - LightColor specular_0; // material.specular_0 * light.specular_0 - LightColor specular_1; // material.specular_1 * light.specular_1 - LightColor diffuse; // material.diffuse * light.diffuse - LightColor ambient; // material.ambient * light.ambient - - // Encoded as 16-bit floating point - union { - BitField<0, 16, u32> x; - BitField<16, 16, u32> y; - }; - union { - BitField<0, 16, u32> z; - }; - - INSERT_PADDING_WORDS(0x3); - - union { - BitField<0, 1, u32> directional; - BitField<1, 1, u32> two_sided_diffuse; // When disabled, clamp dot-product to 0 - } config; - - BitField<0, 20, u32> dist_atten_bias; - BitField<0, 20, u32> dist_atten_scale; - - INSERT_PADDING_WORDS(0x4); - }; - static_assert(sizeof(LightSrc) == 0x10 * sizeof(u32), - "LightSrc structure must be 0x10 words"); - - LightSrc light[8]; - LightColor global_ambient; // Emission + (material.ambient * lighting.ambient) - INSERT_PADDING_WORDS(0x1); - BitField<0, 3, u32> num_lights; // Number of enabled lights - 1 - - union { - BitField<2, 2, LightingFresnelSelector> fresnel_selector; - BitField<4, 4, LightingConfig> config; - BitField<22, 2, u32> bump_selector; // 0: Texture 0, 1: Texture 1, 2: Texture 2 - BitField<27, 1, u32> clamp_highlights; - BitField<28, 2, LightingBumpMode> bump_mode; - BitField<30, 1, u32> disable_bump_renorm; - } config0; - - union { - BitField<16, 1, u32> disable_lut_d0; - BitField<17, 1, u32> disable_lut_d1; - BitField<19, 1, u32> disable_lut_fr; - BitField<20, 1, u32> disable_lut_rr; - BitField<21, 1, u32> disable_lut_rg; - BitField<22, 1, u32> disable_lut_rb; - - // Each bit specifies whether distance attenuation should be applied for the - // corresponding light - - BitField<24, 1, u32> disable_dist_atten_light_0; - BitField<25, 1, u32> disable_dist_atten_light_1; - BitField<26, 1, u32> disable_dist_atten_light_2; - BitField<27, 1, u32> disable_dist_atten_light_3; - BitField<28, 1, u32> disable_dist_atten_light_4; - BitField<29, 1, u32> disable_dist_atten_light_5; - BitField<30, 1, u32> disable_dist_atten_light_6; - BitField<31, 1, u32> disable_dist_atten_light_7; - } config1; - - bool IsDistAttenDisabled(unsigned index) const { - const unsigned disable[] = { - config1.disable_dist_atten_light_0, config1.disable_dist_atten_light_1, - config1.disable_dist_atten_light_2, config1.disable_dist_atten_light_3, - config1.disable_dist_atten_light_4, config1.disable_dist_atten_light_5, - config1.disable_dist_atten_light_6, config1.disable_dist_atten_light_7}; - return disable[index] != 0; - } - - union { - BitField<0, 8, u32> index; ///< Index at which to set data in the LUT - BitField<8, 5, u32> type; ///< Type of LUT for which to set data - } lut_config; - - BitField<0, 1, u32> disable; - INSERT_PADDING_WORDS(0x1); - - // When data is written to any of these registers, it gets written to the lookup table of - // the selected type at the selected index, specified above in the `lut_config` register. - // With each write, `lut_config.index` is incremented. It does not matter which of these - // registers is written to, the behavior will be the same. - u32 lut_data[8]; - - // These are used to specify if absolute (abs) value should be used for each LUT index. When - // abs mode is disabled, LUT indexes are in the range of (-1.0, 1.0). Otherwise, they are in - // the range of (0.0, 1.0). - union { - BitField<1, 1, u32> disable_d0; - BitField<5, 1, u32> disable_d1; - BitField<9, 1, u32> disable_sp; - BitField<13, 1, u32> disable_fr; - BitField<17, 1, u32> disable_rb; - BitField<21, 1, u32> disable_rg; - BitField<25, 1, u32> disable_rr; - } abs_lut_input; - - union { - BitField<0, 3, LightingLutInput> d0; - BitField<4, 3, LightingLutInput> d1; - BitField<8, 3, LightingLutInput> sp; - BitField<12, 3, LightingLutInput> fr; - BitField<16, 3, LightingLutInput> rb; - BitField<20, 3, LightingLutInput> rg; - BitField<24, 3, LightingLutInput> rr; - } lut_input; - - union { - BitField<0, 3, LightingScale> d0; - BitField<4, 3, LightingScale> d1; - BitField<8, 3, LightingScale> sp; - BitField<12, 3, LightingScale> fr; - BitField<16, 3, LightingScale> rb; - BitField<20, 3, LightingScale> rg; - BitField<24, 3, LightingScale> rr; - - static float GetScale(LightingScale scale) { - switch (scale) { - case LightingScale::Scale1: - return 1.0f; - case LightingScale::Scale2: - return 2.0f; - case LightingScale::Scale4: - return 4.0f; - case LightingScale::Scale8: - return 8.0f; - case LightingScale::Scale1_4: - return 0.25f; - case LightingScale::Scale1_2: - return 0.5f; - } - return 0.0f; - } - } lut_scale; - - INSERT_PADDING_WORDS(0x6); - - union { - // There are 8 light enable "slots", corresponding to the total number of lights - // supported by Pica. For N enabled lights (specified by register 0x1c2, or 'src_num' - // above), the first N slots below will be set to integers within the range of 0-7, - // corresponding to the actual light that is enabled for each slot. - - BitField<0, 3, u32> slot_0; - BitField<4, 3, u32> slot_1; - BitField<8, 3, u32> slot_2; - BitField<12, 3, u32> slot_3; - BitField<16, 3, u32> slot_4; - BitField<20, 3, u32> slot_5; - BitField<24, 3, u32> slot_6; - BitField<28, 3, u32> slot_7; - - unsigned GetNum(unsigned index) const { - const unsigned enable_slots[] = {slot_0, slot_1, slot_2, slot_3, - slot_4, slot_5, slot_6, slot_7}; - return enable_slots[index]; - } - } light_enable; - } lighting; - - INSERT_PADDING_WORDS(0x26); - - enum class VertexAttributeFormat : u64 { - BYTE = 0, - UBYTE = 1, - SHORT = 2, - FLOAT = 3, - }; - - struct { - BitField<0, 29, u32> base_address; - - u32 GetPhysicalBaseAddress() const { - return DecodeAddressRegister(base_address); - } - - // Descriptor for internal vertex attributes - union { - BitField<0, 2, VertexAttributeFormat> format0; // size of one element - BitField<2, 2, u64> size0; // number of elements minus 1 - BitField<4, 2, VertexAttributeFormat> format1; - BitField<6, 2, u64> size1; - BitField<8, 2, VertexAttributeFormat> format2; - BitField<10, 2, u64> size2; - BitField<12, 2, VertexAttributeFormat> format3; - BitField<14, 2, u64> size3; - BitField<16, 2, VertexAttributeFormat> format4; - BitField<18, 2, u64> size4; - BitField<20, 2, VertexAttributeFormat> format5; - BitField<22, 2, u64> size5; - BitField<24, 2, VertexAttributeFormat> format6; - BitField<26, 2, u64> size6; - BitField<28, 2, VertexAttributeFormat> format7; - BitField<30, 2, u64> size7; - BitField<32, 2, VertexAttributeFormat> format8; - BitField<34, 2, u64> size8; - BitField<36, 2, VertexAttributeFormat> format9; - BitField<38, 2, u64> size9; - BitField<40, 2, VertexAttributeFormat> format10; - BitField<42, 2, u64> size10; - BitField<44, 2, VertexAttributeFormat> format11; - BitField<46, 2, u64> size11; - - BitField<48, 12, u64> attribute_mask; - - // number of total attributes minus 1 - BitField<60, 4, u64> num_extra_attributes; - }; - - inline VertexAttributeFormat GetFormat(int n) const { - VertexAttributeFormat formats[] = {format0, format1, format2, format3, - format4, format5, format6, format7, - format8, format9, format10, format11}; - return formats[n]; - } - - inline int GetNumElements(int n) const { - u64 sizes[] = {size0, size1, size2, size3, size4, size5, - size6, size7, size8, size9, size10, size11}; - return (int)sizes[n] + 1; - } - - inline int GetElementSizeInBytes(int n) const { - return (GetFormat(n) == VertexAttributeFormat::FLOAT) - ? 4 - : (GetFormat(n) == VertexAttributeFormat::SHORT) ? 2 : 1; - } - - inline int GetStride(int n) const { - return GetNumElements(n) * GetElementSizeInBytes(n); - } - - inline bool IsDefaultAttribute(int id) const { - return (id >= 12) || (attribute_mask & (1ULL << id)) != 0; - } - - inline int GetNumTotalAttributes() const { - return (int)num_extra_attributes + 1; - } - - // Attribute loaders map the source vertex data to input attributes - // This e.g. allows to load different attributes from different memory locations - struct { - // Source attribute data offset from the base address - u32 data_offset; - - union { - BitField<0, 4, u64> comp0; - BitField<4, 4, u64> comp1; - BitField<8, 4, u64> comp2; - BitField<12, 4, u64> comp3; - BitField<16, 4, u64> comp4; - BitField<20, 4, u64> comp5; - BitField<24, 4, u64> comp6; - BitField<28, 4, u64> comp7; - BitField<32, 4, u64> comp8; - BitField<36, 4, u64> comp9; - BitField<40, 4, u64> comp10; - BitField<44, 4, u64> comp11; - - // bytes for a single vertex in this loader - BitField<48, 8, u64> byte_count; - - BitField<60, 4, u64> component_count; - }; - - inline int GetComponent(int n) const { - u64 components[] = {comp0, comp1, comp2, comp3, comp4, comp5, - comp6, comp7, comp8, comp9, comp10, comp11}; - return (int)components[n]; - } - } attribute_loaders[12]; - } vertex_attributes; - - struct { - enum IndexFormat : u32 { - BYTE = 0, - SHORT = 1, - }; - - union { - BitField<0, 31, u32> offset; // relative to base attribute address - BitField<31, 1, IndexFormat> format; - }; - } index_array; - - // Number of vertices to render - u32 num_vertices; - - INSERT_PADDING_WORDS(0x1); - - // The index of the first vertex to render - u32 vertex_offset; - - INSERT_PADDING_WORDS(0x3); - - // These two trigger rendering of triangles - u32 trigger_draw; - u32 trigger_draw_indexed; - - INSERT_PADDING_WORDS(0x2); - - // These registers are used to setup the default "fall-back" vertex shader attributes - struct { - // Index of the current default attribute - u32 index; - - // Writing to these registers sets the "current" default attribute. - u32 set_value[3]; - } vs_default_attributes_setup; - - INSERT_PADDING_WORDS(0x2); - - struct { - // There are two channels that can be used to configure the next command buffer, which - // can be then executed by writing to the "trigger" registers. There are two reasons why a - // game might use this feature: - // 1) With this, an arbitrary number of additional command buffers may be executed in - // sequence without requiring any intervention of the CPU after the initial one is - // kicked off. - // 2) Games can configure these registers to provide a command list subroutine mechanism. - - BitField<0, 20, u32> size[2]; ///< Size (in bytes / 8) of each channel's command buffer - BitField<0, 28, u32> addr[2]; ///< Physical address / 8 of each channel's command buffer - u32 trigger[2]; ///< Triggers execution of the channel's command buffer when written to - - unsigned GetSize(unsigned index) const { - ASSERT(index < 2); - return 8 * size[index]; - } - - PAddr GetPhysicalAddress(unsigned index) const { - ASSERT(index < 2); - return (PAddr)(8 * addr[index]); - } - } command_buffer; - - INSERT_PADDING_WORDS(0x07); - - enum class GPUMode : u32 { - Drawing = 0, - Configuring = 1, - }; - - GPUMode gpu_mode; - - INSERT_PADDING_WORDS(0x18); - - enum class TriangleTopology : u32 { - List = 0, - Strip = 1, - Fan = 2, - Shader = 3, // Programmable setup unit implemented in a geometry shader - }; - - BitField<8, 2, TriangleTopology> triangle_topology; - - u32 restart_primitive; - - INSERT_PADDING_WORDS(0x20); - - struct ShaderConfig { - BitField<0, 16, u32> bool_uniforms; - - union { - BitField<0, 8, u32> x; - BitField<8, 8, u32> y; - BitField<16, 8, u32> z; - BitField<24, 8, u32> w; - } int_uniforms[4]; - - INSERT_PADDING_WORDS(0x4); - - union { - // Number of input attributes to shader unit - 1 - BitField<0, 4, u32> num_input_attributes; - }; - - // Offset to shader program entry point (in words) - BitField<0, 16, u32> main_offset; - - union { - BitField<0, 4, u64> attribute0_register; - BitField<4, 4, u64> attribute1_register; - BitField<8, 4, u64> attribute2_register; - BitField<12, 4, u64> attribute3_register; - BitField<16, 4, u64> attribute4_register; - BitField<20, 4, u64> attribute5_register; - BitField<24, 4, u64> attribute6_register; - BitField<28, 4, u64> attribute7_register; - BitField<32, 4, u64> attribute8_register; - BitField<36, 4, u64> attribute9_register; - BitField<40, 4, u64> attribute10_register; - BitField<44, 4, u64> attribute11_register; - BitField<48, 4, u64> attribute12_register; - BitField<52, 4, u64> attribute13_register; - BitField<56, 4, u64> attribute14_register; - BitField<60, 4, u64> attribute15_register; - - int GetRegisterForAttribute(int attribute_index) const { - u64 fields[] = { - attribute0_register, attribute1_register, attribute2_register, - attribute3_register, attribute4_register, attribute5_register, - attribute6_register, attribute7_register, attribute8_register, - attribute9_register, attribute10_register, attribute11_register, - attribute12_register, attribute13_register, attribute14_register, - attribute15_register, - }; - return (int)fields[attribute_index]; - } - } input_register_map; - - BitField<0, 16, u32> output_mask; - - // 0x28E, CODETRANSFER_END - INSERT_PADDING_WORDS(0x2); - - struct { - enum Format : u32 { - FLOAT24 = 0, - FLOAT32 = 1, - }; - - bool IsFloat32() const { - return format == FLOAT32; - } - - union { - // Index of the next uniform to write to - // TODO: ctrulib uses 8 bits for this, however that seems to yield lots of invalid - // indices - // TODO: Maybe the uppermost index is for the geometry shader? Investigate! - BitField<0, 7, u32> index; - - BitField<31, 1, Format> format; - }; - - // Writing to these registers sets the current uniform. - u32 set_value[8]; - - } uniform_setup; - - INSERT_PADDING_WORDS(0x2); - - struct { - // Offset of the next instruction to write code to. - // Incremented with each instruction write. - u32 offset; - - // Writing to these registers sets the "current" word in the shader program. - u32 set_word[8]; - } program; - - INSERT_PADDING_WORDS(0x1); - - // This register group is used to load an internal table of swizzling patterns, - // which are indexed by each shader instruction to specify vector component swizzling. - struct { - // Offset of the next swizzle pattern to write code to. - // Incremented with each instruction write. - u32 offset; - - // Writing to these registers sets the current swizzle pattern in the table. - u32 set_word[8]; - } swizzle_patterns; - - INSERT_PADDING_WORDS(0x2); - }; - - ShaderConfig gs; - ShaderConfig vs; - - INSERT_PADDING_WORDS(0x20); - - // Map register indices to names readable by humans - // Used for debugging purposes, so performance is not an issue here - static std::string GetCommandName(int index); - - static constexpr size_t NumIds() { - return sizeof(Regs) / sizeof(u32); - } - - const u32& operator[](int index) const { - const u32* content = reinterpret_cast<const u32*>(this); - return content[index]; - } - - u32& operator[](int index) { - u32* content = reinterpret_cast<u32*>(this); - return content[index]; - } - -private: - /* - * Most physical addresses which Pica registers refer to are 8-byte aligned. - * This function should be used to get the address from a raw register value. - */ - static inline u32 DecodeAddressRegister(u32 register_value) { - return register_value * 8; - } -}; - -// TODO: MSVC does not support using offsetof() on non-static data members even though this -// is technically allowed since C++11. This macro should be enabled once MSVC adds -// support for that. -#ifndef _MSC_VER -#define ASSERT_REG_POSITION(field_name, position) \ - static_assert(offsetof(Regs, field_name) == position * 4, \ - "Field " #field_name " has invalid position") - -ASSERT_REG_POSITION(trigger_irq, 0x10); -ASSERT_REG_POSITION(cull_mode, 0x40); -ASSERT_REG_POSITION(viewport_size_x, 0x41); -ASSERT_REG_POSITION(viewport_size_y, 0x43); -ASSERT_REG_POSITION(viewport_depth_range, 0x4d); -ASSERT_REG_POSITION(viewport_depth_near_plane, 0x4e); -ASSERT_REG_POSITION(vs_output_attributes[0], 0x50); -ASSERT_REG_POSITION(vs_output_attributes[1], 0x51); -ASSERT_REG_POSITION(scissor_test, 0x65); -ASSERT_REG_POSITION(viewport_corner, 0x68); -ASSERT_REG_POSITION(depthmap_enable, 0x6D); -ASSERT_REG_POSITION(texture0_enable, 0x80); -ASSERT_REG_POSITION(texture0, 0x81); -ASSERT_REG_POSITION(texture0_format, 0x8e); -ASSERT_REG_POSITION(fragment_lighting_enable, 0x8f); -ASSERT_REG_POSITION(texture1, 0x91); -ASSERT_REG_POSITION(texture1_format, 0x96); -ASSERT_REG_POSITION(texture2, 0x99); -ASSERT_REG_POSITION(texture2_format, 0x9e); -ASSERT_REG_POSITION(tev_stage0, 0xc0); -ASSERT_REG_POSITION(tev_stage1, 0xc8); -ASSERT_REG_POSITION(tev_stage2, 0xd0); -ASSERT_REG_POSITION(tev_stage3, 0xd8); -ASSERT_REG_POSITION(tev_combiner_buffer_input, 0xe0); -ASSERT_REG_POSITION(fog_mode, 0xe0); -ASSERT_REG_POSITION(fog_color, 0xe1); -ASSERT_REG_POSITION(fog_lut_offset, 0xe6); -ASSERT_REG_POSITION(fog_lut_data, 0xe8); -ASSERT_REG_POSITION(tev_stage4, 0xf0); -ASSERT_REG_POSITION(tev_stage5, 0xf8); -ASSERT_REG_POSITION(tev_combiner_buffer_color, 0xfd); -ASSERT_REG_POSITION(output_merger, 0x100); -ASSERT_REG_POSITION(framebuffer, 0x110); -ASSERT_REG_POSITION(lighting, 0x140); -ASSERT_REG_POSITION(vertex_attributes, 0x200); -ASSERT_REG_POSITION(index_array, 0x227); -ASSERT_REG_POSITION(num_vertices, 0x228); -ASSERT_REG_POSITION(vertex_offset, 0x22a); -ASSERT_REG_POSITION(trigger_draw, 0x22e); -ASSERT_REG_POSITION(trigger_draw_indexed, 0x22f); -ASSERT_REG_POSITION(vs_default_attributes_setup, 0x232); -ASSERT_REG_POSITION(command_buffer, 0x238); -ASSERT_REG_POSITION(gpu_mode, 0x245); -ASSERT_REG_POSITION(triangle_topology, 0x25e); -ASSERT_REG_POSITION(restart_primitive, 0x25f); -ASSERT_REG_POSITION(gs, 0x280); -ASSERT_REG_POSITION(vs, 0x2b0); - -#undef ASSERT_REG_POSITION -#endif // !defined(_MSC_VER) - -static_assert(sizeof(Regs::ShaderConfig) == 0x30 * sizeof(u32), - "ShaderConfig structure has incorrect size"); - -// The total number of registers is chosen arbitrarily, but let's make sure it's not some odd value -// anyway. -static_assert(sizeof(Regs) <= 0x300 * sizeof(u32), - "Register set structure larger than it should be"); -static_assert(sizeof(Regs) >= 0x300 * sizeof(u32), - "Register set structure smaller than it should be"); - /// Initialize Pica state void Init(); diff --git a/src/video_core/pica_state.h b/src/video_core/pica_state.h index e4f2e6d5d..af7536d11 100644 --- a/src/video_core/pica_state.h +++ b/src/video_core/pica_state.h @@ -7,8 +7,8 @@ #include <array> #include "common/bit_field.h" #include "common/common_types.h" -#include "video_core/pica.h" #include "video_core/primitive_assembly.h" +#include "video_core/regs.h" #include "video_core/shader/shader.h" namespace Pica { @@ -23,7 +23,7 @@ struct State { Shader::ShaderSetup vs; Shader::ShaderSetup gs; - std::array<Math::Vec4<float24>, 16> vs_default_attributes; + Shader::AttributeBuffer input_default_attributes; struct { union LutEntry { @@ -66,7 +66,7 @@ struct State { /// Struct used to describe immediate mode rendering state struct ImmediateModeState { // Used to buffer partial vertices for immediate-mode rendering. - Shader::InputVertex input_vertex; + Shader::AttributeBuffer input_vertex; // Index of the next attribute to be loaded into `input_vertex`. u32 current_attribute = 0; } immediate; diff --git a/src/video_core/primitive_assembly.cpp b/src/video_core/primitive_assembly.cpp index be7377290..acd2ac5e2 100644 --- a/src/video_core/primitive_assembly.cpp +++ b/src/video_core/primitive_assembly.cpp @@ -3,23 +3,23 @@ // Refer to the license.txt file included. #include "common/logging/log.h" -#include "video_core/pica.h" #include "video_core/primitive_assembly.h" +#include "video_core/regs_pipeline.h" #include "video_core/shader/shader.h" namespace Pica { template <typename VertexType> -PrimitiveAssembler<VertexType>::PrimitiveAssembler(Regs::TriangleTopology topology) +PrimitiveAssembler<VertexType>::PrimitiveAssembler(PipelineRegs::TriangleTopology topology) : topology(topology), buffer_index(0) {} template <typename VertexType> -void PrimitiveAssembler<VertexType>::SubmitVertex(VertexType& vtx, +void PrimitiveAssembler<VertexType>::SubmitVertex(const VertexType& vtx, TriangleHandler triangle_handler) { switch (topology) { // TODO: Figure out what's different with TriangleTopology::Shader. - case Regs::TriangleTopology::List: - case Regs::TriangleTopology::Shader: + case PipelineRegs::TriangleTopology::List: + case PipelineRegs::TriangleTopology::Shader: if (buffer_index < 2) { buffer[buffer_index++] = vtx; } else { @@ -29,8 +29,8 @@ void PrimitiveAssembler<VertexType>::SubmitVertex(VertexType& vtx, } break; - case Regs::TriangleTopology::Strip: - case Regs::TriangleTopology::Fan: + case PipelineRegs::TriangleTopology::Strip: + case PipelineRegs::TriangleTopology::Fan: if (strip_ready) triangle_handler(buffer[0], buffer[1], vtx); @@ -38,9 +38,9 @@ void PrimitiveAssembler<VertexType>::SubmitVertex(VertexType& vtx, strip_ready |= (buffer_index == 1); - if (topology == Regs::TriangleTopology::Strip) + if (topology == PipelineRegs::TriangleTopology::Strip) buffer_index = !buffer_index; - else if (topology == Regs::TriangleTopology::Fan) + else if (topology == PipelineRegs::TriangleTopology::Fan) buffer_index = 1; break; @@ -57,7 +57,7 @@ void PrimitiveAssembler<VertexType>::Reset() { } template <typename VertexType> -void PrimitiveAssembler<VertexType>::Reconfigure(Regs::TriangleTopology topology) { +void PrimitiveAssembler<VertexType>::Reconfigure(PipelineRegs::TriangleTopology topology) { Reset(); this->topology = topology; } diff --git a/src/video_core/primitive_assembly.h b/src/video_core/primitive_assembly.h index 0384d5984..e8eccdf27 100644 --- a/src/video_core/primitive_assembly.h +++ b/src/video_core/primitive_assembly.h @@ -5,7 +5,7 @@ #pragma once #include <functional> -#include "video_core/pica.h" +#include "video_core/regs_pipeline.h" namespace Pica { @@ -15,9 +15,11 @@ namespace Pica { */ template <typename VertexType> struct PrimitiveAssembler { - using TriangleHandler = std::function<void(VertexType& v0, VertexType& v1, VertexType& v2)>; + using TriangleHandler = + std::function<void(const VertexType& v0, const VertexType& v1, const VertexType& v2)>; - PrimitiveAssembler(Regs::TriangleTopology topology = Regs::TriangleTopology::List); + PrimitiveAssembler( + PipelineRegs::TriangleTopology topology = PipelineRegs::TriangleTopology::List); /* * Queues a vertex, builds primitives from the vertex queue according to the given @@ -25,7 +27,7 @@ struct PrimitiveAssembler { * NOTE: We could specify the triangle handler in the constructor, but this way we can * keep event and handler code next to each other. */ - void SubmitVertex(VertexType& vtx, TriangleHandler triangle_handler); + void SubmitVertex(const VertexType& vtx, TriangleHandler triangle_handler); /** * Resets the internal state of the PrimitiveAssembler. @@ -35,10 +37,10 @@ struct PrimitiveAssembler { /** * Reconfigures the PrimitiveAssembler to use a different triangle topology. */ - void Reconfigure(Regs::TriangleTopology topology); + void Reconfigure(PipelineRegs::TriangleTopology topology); private: - Regs::TriangleTopology topology; + PipelineRegs::TriangleTopology topology; int buffer_index; VertexType buffer[2]; diff --git a/src/video_core/rasterizer.h b/src/video_core/rasterizer.h deleted file mode 100644 index 6cbda3067..000000000 --- a/src/video_core/rasterizer.h +++ /dev/null @@ -1,20 +0,0 @@ -// Copyright 2014 Citra Emulator Project -// Licensed under GPLv2 or any later version -// Refer to the license.txt file included. - -#pragma once - -namespace Pica { - -namespace Shader { -struct OutputVertex; -} - -namespace Rasterizer { - -void ProcessTriangle(const Shader::OutputVertex& v0, const Shader::OutputVertex& v1, - const Shader::OutputVertex& v2); - -} // namespace Rasterizer - -} // namespace Pica diff --git a/src/video_core/regs.cpp b/src/video_core/regs.cpp new file mode 100644 index 000000000..2699e710a --- /dev/null +++ b/src/video_core/regs.cpp @@ -0,0 +1,488 @@ +// Copyright 2015 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <algorithm> +#include <iterator> +#include <utility> + +#include "common/common_types.h" +#include "video_core/regs.h" + +namespace Pica { + +static const std::pair<u16, const char*> register_names[] = { + {0x010, "GPUREG_FINALIZE"}, + + {0x040, "GPUREG_FACECULLING_CONFIG"}, + {0x041, "GPUREG_VIEWPORT_WIDTH"}, + {0x042, "GPUREG_VIEWPORT_INVW"}, + {0x043, "GPUREG_VIEWPORT_HEIGHT"}, + {0x044, "GPUREG_VIEWPORT_INVH"}, + + {0x047, "GPUREG_FRAGOP_CLIP"}, + {0x048, "GPUREG_FRAGOP_CLIP_DATA0"}, + {0x049, "GPUREG_FRAGOP_CLIP_DATA1"}, + {0x04A, "GPUREG_FRAGOP_CLIP_DATA2"}, + {0x04B, "GPUREG_FRAGOP_CLIP_DATA3"}, + + {0x04D, "GPUREG_DEPTHMAP_SCALE"}, + {0x04E, "GPUREG_DEPTHMAP_OFFSET"}, + {0x04F, "GPUREG_SH_OUTMAP_TOTAL"}, + {0x050, "GPUREG_SH_OUTMAP_O0"}, + {0x051, "GPUREG_SH_OUTMAP_O1"}, + {0x052, "GPUREG_SH_OUTMAP_O2"}, + {0x053, "GPUREG_SH_OUTMAP_O3"}, + {0x054, "GPUREG_SH_OUTMAP_O4"}, + {0x055, "GPUREG_SH_OUTMAP_O5"}, + {0x056, "GPUREG_SH_OUTMAP_O6"}, + + {0x061, "GPUREG_EARLYDEPTH_FUNC"}, + {0x062, "GPUREG_EARLYDEPTH_TEST1"}, + {0x063, "GPUREG_EARLYDEPTH_CLEAR"}, + {0x064, "GPUREG_SH_OUTATTR_MODE"}, + {0x065, "GPUREG_SCISSORTEST_MODE"}, + {0x066, "GPUREG_SCISSORTEST_POS"}, + {0x067, "GPUREG_SCISSORTEST_DIM"}, + {0x068, "GPUREG_VIEWPORT_XY"}, + + {0x06A, "GPUREG_EARLYDEPTH_DATA"}, + + {0x06D, "GPUREG_DEPTHMAP_ENABLE"}, + {0x06E, "GPUREG_RENDERBUF_DIM"}, + {0x06F, "GPUREG_SH_OUTATTR_CLOCK"}, + + {0x080, "GPUREG_TEXUNIT_CONFIG"}, + {0x081, "GPUREG_TEXUNIT0_BORDER_COLOR"}, + {0x082, "GPUREG_TEXUNIT0_DIM"}, + {0x083, "GPUREG_TEXUNIT0_PARAM"}, + {0x084, "GPUREG_TEXUNIT0_LOD"}, + {0x085, "GPUREG_TEXUNIT0_ADDR1"}, + {0x086, "GPUREG_TEXUNIT0_ADDR2"}, + {0x087, "GPUREG_TEXUNIT0_ADDR3"}, + {0x088, "GPUREG_TEXUNIT0_ADDR4"}, + {0x089, "GPUREG_TEXUNIT0_ADDR5"}, + {0x08A, "GPUREG_TEXUNIT0_ADDR6"}, + {0x08B, "GPUREG_TEXUNIT0_SHADOW"}, + + {0x08E, "GPUREG_TEXUNIT0_TYPE"}, + {0x08F, "GPUREG_LIGHTING_ENABLE0"}, + + {0x091, "GPUREG_TEXUNIT1_BORDER_COLOR"}, + {0x092, "GPUREG_TEXUNIT1_DIM"}, + {0x093, "GPUREG_TEXUNIT1_PARAM"}, + {0x094, "GPUREG_TEXUNIT1_LOD"}, + {0x095, "GPUREG_TEXUNIT1_ADDR"}, + {0x096, "GPUREG_TEXUNIT1_TYPE"}, + + {0x099, "GPUREG_TEXUNIT2_BORDER_COLOR"}, + {0x09A, "GPUREG_TEXUNIT2_DIM"}, + {0x09B, "GPUREG_TEXUNIT2_PARAM"}, + {0x09C, "GPUREG_TEXUNIT2_LOD"}, + {0x09D, "GPUREG_TEXUNIT2_ADDR"}, + {0x09E, "GPUREG_TEXUNIT2_TYPE"}, + + {0x0A8, "GPUREG_TEXUNIT3_PROCTEX0"}, + {0x0A9, "GPUREG_TEXUNIT3_PROCTEX1"}, + {0x0AA, "GPUREG_TEXUNIT3_PROCTEX2"}, + {0x0AB, "GPUREG_TEXUNIT3_PROCTEX3"}, + {0x0AC, "GPUREG_TEXUNIT3_PROCTEX4"}, + {0x0AD, "GPUREG_TEXUNIT3_PROCTEX5"}, + + {0x0AF, "GPUREG_PROCTEX_LUT"}, + {0x0B0, "GPUREG_PROCTEX_LUT_DATA0"}, + {0x0B1, "GPUREG_PROCTEX_LUT_DATA1"}, + {0x0B2, "GPUREG_PROCTEX_LUT_DATA2"}, + {0x0B3, "GPUREG_PROCTEX_LUT_DATA3"}, + {0x0B4, "GPUREG_PROCTEX_LUT_DATA4"}, + {0x0B5, "GPUREG_PROCTEX_LUT_DATA5"}, + {0x0B6, "GPUREG_PROCTEX_LUT_DATA6"}, + {0x0B7, "GPUREG_PROCTEX_LUT_DATA7"}, + + {0x0C0, "GPUREG_TEXENV0_SOURCE"}, + {0x0C1, "GPUREG_TEXENV0_OPERAND"}, + {0x0C2, "GPUREG_TEXENV0_COMBINER"}, + {0x0C3, "GPUREG_TEXENV0_COLOR"}, + {0x0C4, "GPUREG_TEXENV0_SCALE"}, + + {0x0C8, "GPUREG_TEXENV1_SOURCE"}, + {0x0C9, "GPUREG_TEXENV1_OPERAND"}, + {0x0CA, "GPUREG_TEXENV1_COMBINER"}, + {0x0CB, "GPUREG_TEXENV1_COLOR"}, + {0x0CC, "GPUREG_TEXENV1_SCALE"}, + + {0x0D0, "GPUREG_TEXENV2_SOURCE"}, + {0x0D1, "GPUREG_TEXENV2_OPERAND"}, + {0x0D2, "GPUREG_TEXENV2_COMBINER"}, + {0x0D3, "GPUREG_TEXENV2_COLOR"}, + {0x0D4, "GPUREG_TEXENV2_SCALE"}, + + {0x0D8, "GPUREG_TEXENV3_SOURCE"}, + {0x0D9, "GPUREG_TEXENV3_OPERAND"}, + {0x0DA, "GPUREG_TEXENV3_COMBINER"}, + {0x0DB, "GPUREG_TEXENV3_COLOR"}, + {0x0DC, "GPUREG_TEXENV3_SCALE"}, + + {0x0E0, "GPUREG_TEXENV_UPDATE_BUFFER"}, + {0x0E1, "GPUREG_FOG_COLOR"}, + + {0x0E4, "GPUREG_GAS_ATTENUATION"}, + {0x0E5, "GPUREG_GAS_ACCMAX"}, + {0x0E6, "GPUREG_FOG_LUT_INDEX"}, + + {0x0E8, "GPUREG_FOG_LUT_DATA0"}, + {0x0E9, "GPUREG_FOG_LUT_DATA1"}, + {0x0EA, "GPUREG_FOG_LUT_DATA2"}, + {0x0EB, "GPUREG_FOG_LUT_DATA3"}, + {0x0EC, "GPUREG_FOG_LUT_DATA4"}, + {0x0ED, "GPUREG_FOG_LUT_DATA5"}, + {0x0EE, "GPUREG_FOG_LUT_DATA6"}, + {0x0EF, "GPUREG_FOG_LUT_DATA7"}, + {0x0F0, "GPUREG_TEXENV4_SOURCE"}, + {0x0F1, "GPUREG_TEXENV4_OPERAND"}, + {0x0F2, "GPUREG_TEXENV4_COMBINER"}, + {0x0F3, "GPUREG_TEXENV4_COLOR"}, + {0x0F4, "GPUREG_TEXENV4_SCALE"}, + + {0x0F8, "GPUREG_TEXENV5_SOURCE"}, + {0x0F9, "GPUREG_TEXENV5_OPERAND"}, + {0x0FA, "GPUREG_TEXENV5_COMBINER"}, + {0x0FB, "GPUREG_TEXENV5_COLOR"}, + {0x0FC, "GPUREG_TEXENV5_SCALE"}, + {0x0FD, "GPUREG_TEXENV_BUFFER_COLOR"}, + + {0x100, "GPUREG_COLOR_OPERATION"}, + {0x101, "GPUREG_BLEND_FUNC"}, + {0x102, "GPUREG_LOGIC_OP"}, + {0x103, "GPUREG_BLEND_COLOR"}, + {0x104, "GPUREG_FRAGOP_ALPHA_TEST"}, + {0x105, "GPUREG_STENCIL_TEST"}, + {0x106, "GPUREG_STENCIL_OP"}, + {0x107, "GPUREG_DEPTH_COLOR_MASK"}, + + {0x110, "GPUREG_FRAMEBUFFER_INVALIDATE"}, + {0x111, "GPUREG_FRAMEBUFFER_FLUSH"}, + {0x112, "GPUREG_COLORBUFFER_READ"}, + {0x113, "GPUREG_COLORBUFFER_WRITE"}, + {0x114, "GPUREG_DEPTHBUFFER_READ"}, + {0x115, "GPUREG_DEPTHBUFFER_WRITE"}, + {0x116, "GPUREG_DEPTHBUFFER_FORMAT"}, + {0x117, "GPUREG_COLORBUFFER_FORMAT"}, + {0x118, "GPUREG_EARLYDEPTH_TEST2"}, + + {0x11B, "GPUREG_FRAMEBUFFER_BLOCK32"}, + {0x11C, "GPUREG_DEPTHBUFFER_LOC"}, + {0x11D, "GPUREG_COLORBUFFER_LOC"}, + {0x11E, "GPUREG_FRAMEBUFFER_DIM"}, + + {0x120, "GPUREG_GAS_LIGHT_XY"}, + {0x121, "GPUREG_GAS_LIGHT_Z"}, + {0x122, "GPUREG_GAS_LIGHT_Z_COLOR"}, + {0x123, "GPUREG_GAS_LUT_INDEX"}, + {0x124, "GPUREG_GAS_LUT_DATA"}, + + {0x126, "GPUREG_GAS_DELTAZ_DEPTH"}, + + {0x130, "GPUREG_FRAGOP_SHADOW"}, + + {0x140, "GPUREG_LIGHT0_SPECULAR0"}, + {0x141, "GPUREG_LIGHT0_SPECULAR1"}, + {0x142, "GPUREG_LIGHT0_DIFFUSE"}, + {0x143, "GPUREG_LIGHT0_AMBIENT"}, + {0x144, "GPUREG_LIGHT0_XY"}, + {0x145, "GPUREG_LIGHT0_Z"}, + {0x146, "GPUREG_LIGHT0_SPOTDIR_XY"}, + {0x147, "GPUREG_LIGHT0_SPOTDIR_Z"}, + + {0x149, "GPUREG_LIGHT0_CONFIG"}, + {0x14A, "GPUREG_LIGHT0_ATTENUATION_BIAS"}, + {0x14B, "GPUREG_LIGHT0_ATTENUATION_SCALE"}, + + {0x150, "GPUREG_LIGHT1_SPECULAR0"}, + {0x151, "GPUREG_LIGHT1_SPECULAR1"}, + {0x152, "GPUREG_LIGHT1_DIFFUSE"}, + {0x153, "GPUREG_LIGHT1_AMBIENT"}, + {0x154, "GPUREG_LIGHT1_XY"}, + {0x155, "GPUREG_LIGHT1_Z"}, + {0x156, "GPUREG_LIGHT1_SPOTDIR_XY"}, + {0x157, "GPUREG_LIGHT1_SPOTDIR_Z"}, + + {0x159, "GPUREG_LIGHT1_CONFIG"}, + {0x15A, "GPUREG_LIGHT1_ATTENUATION_BIAS"}, + {0x15B, "GPUREG_LIGHT1_ATTENUATION_SCALE"}, + + {0x160, "GPUREG_LIGHT2_SPECULAR0"}, + {0x161, "GPUREG_LIGHT2_SPECULAR1"}, + {0x162, "GPUREG_LIGHT2_DIFFUSE"}, + {0x163, "GPUREG_LIGHT2_AMBIENT"}, + {0x164, "GPUREG_LIGHT2_XY"}, + {0x165, "GPUREG_LIGHT2_Z"}, + {0x166, "GPUREG_LIGHT2_SPOTDIR_XY"}, + {0x167, "GPUREG_LIGHT2_SPOTDIR_Z"}, + + {0x169, "GPUREG_LIGHT2_CONFIG"}, + {0x16A, "GPUREG_LIGHT2_ATTENUATION_BIAS"}, + {0x16B, "GPUREG_LIGHT2_ATTENUATION_SCALE"}, + + {0x170, "GPUREG_LIGHT3_SPECULAR0"}, + {0x171, "GPUREG_LIGHT3_SPECULAR1"}, + {0x172, "GPUREG_LIGHT3_DIFFUSE"}, + {0x173, "GPUREG_LIGHT3_AMBIENT"}, + {0x174, "GPUREG_LIGHT3_XY"}, + {0x175, "GPUREG_LIGHT3_Z"}, + {0x176, "GPUREG_LIGHT3_SPOTDIR_XY"}, + {0x177, "GPUREG_LIGHT3_SPOTDIR_Z"}, + + {0x179, "GPUREG_LIGHT3_CONFIG"}, + {0x17A, "GPUREG_LIGHT3_ATTENUATION_BIAS"}, + {0x17B, "GPUREG_LIGHT3_ATTENUATION_SCALE"}, + + {0x180, "GPUREG_LIGHT4_SPECULAR0"}, + {0x181, "GPUREG_LIGHT4_SPECULAR1"}, + {0x182, "GPUREG_LIGHT4_DIFFUSE"}, + {0x183, "GPUREG_LIGHT4_AMBIENT"}, + {0x184, "GPUREG_LIGHT4_XY"}, + {0x185, "GPUREG_LIGHT4_Z"}, + {0x186, "GPUREG_LIGHT4_SPOTDIR_XY"}, + {0x187, "GPUREG_LIGHT4_SPOTDIR_Z"}, + + {0x189, "GPUREG_LIGHT4_CONFIG"}, + {0x18A, "GPUREG_LIGHT4_ATTENUATION_BIAS"}, + {0x18B, "GPUREG_LIGHT4_ATTENUATION_SCALE"}, + + {0x190, "GPUREG_LIGHT5_SPECULAR0"}, + {0x191, "GPUREG_LIGHT5_SPECULAR1"}, + {0x192, "GPUREG_LIGHT5_DIFFUSE"}, + {0x193, "GPUREG_LIGHT5_AMBIENT"}, + {0x194, "GPUREG_LIGHT5_XY"}, + {0x195, "GPUREG_LIGHT5_Z"}, + {0x196, "GPUREG_LIGHT5_SPOTDIR_XY"}, + {0x197, "GPUREG_LIGHT5_SPOTDIR_Z"}, + + {0x199, "GPUREG_LIGHT5_CONFIG"}, + {0x19A, "GPUREG_LIGHT5_ATTENUATION_BIAS"}, + {0x19B, "GPUREG_LIGHT5_ATTENUATION_SCALE"}, + + {0x1A0, "GPUREG_LIGHT6_SPECULAR0"}, + {0x1A1, "GPUREG_LIGHT6_SPECULAR1"}, + {0x1A2, "GPUREG_LIGHT6_DIFFUSE"}, + {0x1A3, "GPUREG_LIGHT6_AMBIENT"}, + {0x1A4, "GPUREG_LIGHT6_XY"}, + {0x1A5, "GPUREG_LIGHT6_Z"}, + {0x1A6, "GPUREG_LIGHT6_SPOTDIR_XY"}, + {0x1A7, "GPUREG_LIGHT6_SPOTDIR_Z"}, + + {0x1A9, "GPUREG_LIGHT6_CONFIG"}, + {0x1AA, "GPUREG_LIGHT6_ATTENUATION_BIAS"}, + {0x1AB, "GPUREG_LIGHT6_ATTENUATION_SCALE"}, + + {0x1B0, "GPUREG_LIGHT7_SPECULAR0"}, + {0x1B1, "GPUREG_LIGHT7_SPECULAR1"}, + {0x1B2, "GPUREG_LIGHT7_DIFFUSE"}, + {0x1B3, "GPUREG_LIGHT7_AMBIENT"}, + {0x1B4, "GPUREG_LIGHT7_XY"}, + {0x1B5, "GPUREG_LIGHT7_Z"}, + {0x1B6, "GPUREG_LIGHT7_SPOTDIR_XY"}, + {0x1B7, "GPUREG_LIGHT7_SPOTDIR_Z"}, + + {0x1B9, "GPUREG_LIGHT7_CONFIG"}, + {0x1BA, "GPUREG_LIGHT7_ATTENUATION_BIAS"}, + {0x1BB, "GPUREG_LIGHT7_ATTENUATION_SCALE"}, + + {0x1C0, "GPUREG_LIGHTING_AMBIENT"}, + + {0x1C2, "GPUREG_LIGHTING_NUM_LIGHTS"}, + {0x1C3, "GPUREG_LIGHTING_CONFIG0"}, + {0x1C4, "GPUREG_LIGHTING_CONFIG1"}, + {0x1C5, "GPUREG_LIGHTING_LUT_INDEX"}, + {0x1C6, "GPUREG_LIGHTING_ENABLE1"}, + + {0x1C8, "GPUREG_LIGHTING_LUT_DATA0"}, + {0x1C9, "GPUREG_LIGHTING_LUT_DATA1"}, + {0x1CA, "GPUREG_LIGHTING_LUT_DATA2"}, + {0x1CB, "GPUREG_LIGHTING_LUT_DATA3"}, + {0x1CC, "GPUREG_LIGHTING_LUT_DATA4"}, + {0x1CD, "GPUREG_LIGHTING_LUT_DATA5"}, + {0x1CE, "GPUREG_LIGHTING_LUT_DATA6"}, + {0x1CF, "GPUREG_LIGHTING_LUT_DATA7"}, + {0x1D0, "GPUREG_LIGHTING_LUTINPUT_ABS"}, + {0x1D1, "GPUREG_LIGHTING_LUTINPUT_SELECT"}, + {0x1D2, "GPUREG_LIGHTING_LUTINPUT_SCALE"}, + + {0x1D9, "GPUREG_LIGHTING_LIGHT_PERMUTATION"}, + + {0x200, "GPUREG_ATTRIBBUFFERS_LOC"}, + {0x201, "GPUREG_ATTRIBBUFFERS_FORMAT_LOW"}, + {0x202, "GPUREG_ATTRIBBUFFERS_FORMAT_HIGH"}, + {0x203, "GPUREG_ATTRIBBUFFER0_OFFSET"}, + {0x204, "GPUREG_ATTRIBBUFFER0_CONFIG1"}, + {0x205, "GPUREG_ATTRIBBUFFER0_CONFIG2"}, + {0x206, "GPUREG_ATTRIBBUFFER1_OFFSET"}, + {0x207, "GPUREG_ATTRIBBUFFER1_CONFIG1"}, + {0x208, "GPUREG_ATTRIBBUFFER1_CONFIG2"}, + {0x209, "GPUREG_ATTRIBBUFFER2_OFFSET"}, + {0x20A, "GPUREG_ATTRIBBUFFER2_CONFIG1"}, + {0x20B, "GPUREG_ATTRIBBUFFER2_CONFIG2"}, + {0x20C, "GPUREG_ATTRIBBUFFER3_OFFSET"}, + {0x20D, "GPUREG_ATTRIBBUFFER3_CONFIG1"}, + {0x20E, "GPUREG_ATTRIBBUFFER3_CONFIG2"}, + {0x20F, "GPUREG_ATTRIBBUFFER4_OFFSET"}, + {0x210, "GPUREG_ATTRIBBUFFER4_CONFIG1"}, + {0x211, "GPUREG_ATTRIBBUFFER4_CONFIG2"}, + {0x212, "GPUREG_ATTRIBBUFFER5_OFFSET"}, + {0x213, "GPUREG_ATTRIBBUFFER5_CONFIG1"}, + {0x214, "GPUREG_ATTRIBBUFFER5_CONFIG2"}, + {0x215, "GPUREG_ATTRIBBUFFER6_OFFSET"}, + {0x216, "GPUREG_ATTRIBBUFFER6_CONFIG1"}, + {0x217, "GPUREG_ATTRIBBUFFER6_CONFIG2"}, + {0x218, "GPUREG_ATTRIBBUFFER7_OFFSET"}, + {0x219, "GPUREG_ATTRIBBUFFER7_CONFIG1"}, + {0x21A, "GPUREG_ATTRIBBUFFER7_CONFIG2"}, + {0x21B, "GPUREG_ATTRIBBUFFER8_OFFSET"}, + {0x21C, "GPUREG_ATTRIBBUFFER8_CONFIG1"}, + {0x21D, "GPUREG_ATTRIBBUFFER8_CONFIG2"}, + {0x21E, "GPUREG_ATTRIBBUFFER9_OFFSET"}, + {0x21F, "GPUREG_ATTRIBBUFFER9_CONFIG1"}, + {0x220, "GPUREG_ATTRIBBUFFER9_CONFIG2"}, + {0x221, "GPUREG_ATTRIBBUFFER10_OFFSET"}, + {0x222, "GPUREG_ATTRIBBUFFER10_CONFIG1"}, + {0x223, "GPUREG_ATTRIBBUFFER10_CONFIG2"}, + {0x224, "GPUREG_ATTRIBBUFFER11_OFFSET"}, + {0x225, "GPUREG_ATTRIBBUFFER11_CONFIG1"}, + {0x226, "GPUREG_ATTRIBBUFFER11_CONFIG2"}, + {0x227, "GPUREG_INDEXBUFFER_CONFIG"}, + {0x228, "GPUREG_NUMVERTICES"}, + {0x229, "GPUREG_GEOSTAGE_CONFIG"}, + {0x22A, "GPUREG_VERTEX_OFFSET"}, + + {0x22D, "GPUREG_POST_VERTEX_CACHE_NUM"}, + {0x22E, "GPUREG_DRAWARRAYS"}, + {0x22F, "GPUREG_DRAWELEMENTS"}, + + {0x231, "GPUREG_VTX_FUNC"}, + {0x232, "GPUREG_FIXEDATTRIB_INDEX"}, + {0x233, "GPUREG_FIXEDATTRIB_DATA0"}, + {0x234, "GPUREG_FIXEDATTRIB_DATA1"}, + {0x235, "GPUREG_FIXEDATTRIB_DATA2"}, + + {0x238, "GPUREG_CMDBUF_SIZE0"}, + {0x239, "GPUREG_CMDBUF_SIZE1"}, + {0x23A, "GPUREG_CMDBUF_ADDR0"}, + {0x23B, "GPUREG_CMDBUF_ADDR1"}, + {0x23C, "GPUREG_CMDBUF_JUMP0"}, + {0x23D, "GPUREG_CMDBUF_JUMP1"}, + + {0x242, "GPUREG_VSH_NUM_ATTR"}, + + {0x244, "GPUREG_VSH_COM_MODE"}, + {0x245, "GPUREG_START_DRAW_FUNC0"}, + + {0x24A, "GPUREG_VSH_OUTMAP_TOTAL1"}, + + {0x251, "GPUREG_VSH_OUTMAP_TOTAL2"}, + {0x252, "GPUREG_GSH_MISC0"}, + {0x253, "GPUREG_GEOSTAGE_CONFIG2"}, + {0x254, "GPUREG_GSH_MISC1"}, + + {0x25E, "GPUREG_PRIMITIVE_CONFIG"}, + {0x25F, "GPUREG_RESTART_PRIMITIVE"}, + + {0x280, "GPUREG_GSH_BOOLUNIFORM"}, + {0x281, "GPUREG_GSH_INTUNIFORM_I0"}, + {0x282, "GPUREG_GSH_INTUNIFORM_I1"}, + {0x283, "GPUREG_GSH_INTUNIFORM_I2"}, + {0x284, "GPUREG_GSH_INTUNIFORM_I3"}, + + {0x289, "GPUREG_GSH_INPUTBUFFER_CONFIG"}, + {0x28A, "GPUREG_GSH_ENTRYPOINT"}, + {0x28B, "GPUREG_GSH_ATTRIBUTES_PERMUTATION_LOW"}, + {0x28C, "GPUREG_GSH_ATTRIBUTES_PERMUTATION_HIGH"}, + {0x28D, "GPUREG_GSH_OUTMAP_MASK"}, + + {0x28F, "GPUREG_GSH_CODETRANSFER_END"}, + {0x290, "GPUREG_GSH_FLOATUNIFORM_INDEX"}, + {0x291, "GPUREG_GSH_FLOATUNIFORM_DATA0"}, + {0x292, "GPUREG_GSH_FLOATUNIFORM_DATA1"}, + {0x293, "GPUREG_GSH_FLOATUNIFORM_DATA2"}, + {0x294, "GPUREG_GSH_FLOATUNIFORM_DATA3"}, + {0x295, "GPUREG_GSH_FLOATUNIFORM_DATA4"}, + {0x296, "GPUREG_GSH_FLOATUNIFORM_DATA5"}, + {0x297, "GPUREG_GSH_FLOATUNIFORM_DATA6"}, + {0x298, "GPUREG_GSH_FLOATUNIFORM_DATA7"}, + + {0x29B, "GPUREG_GSH_CODETRANSFER_INDEX"}, + {0x29C, "GPUREG_GSH_CODETRANSFER_DATA0"}, + {0x29D, "GPUREG_GSH_CODETRANSFER_DATA1"}, + {0x29E, "GPUREG_GSH_CODETRANSFER_DATA2"}, + {0x29F, "GPUREG_GSH_CODETRANSFER_DATA3"}, + {0x2A0, "GPUREG_GSH_CODETRANSFER_DATA4"}, + {0x2A1, "GPUREG_GSH_CODETRANSFER_DATA5"}, + {0x2A2, "GPUREG_GSH_CODETRANSFER_DATA6"}, + {0x2A3, "GPUREG_GSH_CODETRANSFER_DATA7"}, + + {0x2A5, "GPUREG_GSH_OPDESCS_INDEX"}, + {0x2A6, "GPUREG_GSH_OPDESCS_DATA0"}, + {0x2A7, "GPUREG_GSH_OPDESCS_DATA1"}, + {0x2A8, "GPUREG_GSH_OPDESCS_DATA2"}, + {0x2A9, "GPUREG_GSH_OPDESCS_DATA3"}, + {0x2AA, "GPUREG_GSH_OPDESCS_DATA4"}, + {0x2AB, "GPUREG_GSH_OPDESCS_DATA5"}, + {0x2AC, "GPUREG_GSH_OPDESCS_DATA6"}, + {0x2AD, "GPUREG_GSH_OPDESCS_DATA7"}, + + {0x2B0, "GPUREG_VSH_BOOLUNIFORM"}, + {0x2B1, "GPUREG_VSH_INTUNIFORM_I0"}, + {0x2B2, "GPUREG_VSH_INTUNIFORM_I1"}, + {0x2B3, "GPUREG_VSH_INTUNIFORM_I2"}, + {0x2B4, "GPUREG_VSH_INTUNIFORM_I3"}, + + {0x2B9, "GPUREG_VSH_INPUTBUFFER_CONFIG"}, + {0x2BA, "GPUREG_VSH_ENTRYPOINT"}, + {0x2BB, "GPUREG_VSH_ATTRIBUTES_PERMUTATION_LOW"}, + {0x2BC, "GPUREG_VSH_ATTRIBUTES_PERMUTATION_HIGH"}, + {0x2BD, "GPUREG_VSH_OUTMAP_MASK"}, + + {0x2BF, "GPUREG_VSH_CODETRANSFER_END"}, + {0x2C0, "GPUREG_VSH_FLOATUNIFORM_INDEX"}, + {0x2C1, "GPUREG_VSH_FLOATUNIFORM_DATA0"}, + {0x2C2, "GPUREG_VSH_FLOATUNIFORM_DATA1"}, + {0x2C3, "GPUREG_VSH_FLOATUNIFORM_DATA2"}, + {0x2C4, "GPUREG_VSH_FLOATUNIFORM_DATA3"}, + {0x2C5, "GPUREG_VSH_FLOATUNIFORM_DATA4"}, + {0x2C6, "GPUREG_VSH_FLOATUNIFORM_DATA5"}, + {0x2C7, "GPUREG_VSH_FLOATUNIFORM_DATA6"}, + {0x2C8, "GPUREG_VSH_FLOATUNIFORM_DATA7"}, + + {0x2CB, "GPUREG_VSH_CODETRANSFER_INDEX"}, + {0x2CC, "GPUREG_VSH_CODETRANSFER_DATA0"}, + {0x2CD, "GPUREG_VSH_CODETRANSFER_DATA1"}, + {0x2CE, "GPUREG_VSH_CODETRANSFER_DATA2"}, + {0x2CF, "GPUREG_VSH_CODETRANSFER_DATA3"}, + {0x2D0, "GPUREG_VSH_CODETRANSFER_DATA4"}, + {0x2D1, "GPUREG_VSH_CODETRANSFER_DATA5"}, + {0x2D2, "GPUREG_VSH_CODETRANSFER_DATA6"}, + {0x2D3, "GPUREG_VSH_CODETRANSFER_DATA7"}, + + {0x2D5, "GPUREG_VSH_OPDESCS_INDEX"}, + {0x2D6, "GPUREG_VSH_OPDESCS_DATA0"}, + {0x2D7, "GPUREG_VSH_OPDESCS_DATA1"}, + {0x2D8, "GPUREG_VSH_OPDESCS_DATA2"}, + {0x2D9, "GPUREG_VSH_OPDESCS_DATA3"}, + {0x2DA, "GPUREG_VSH_OPDESCS_DATA4"}, + {0x2DB, "GPUREG_VSH_OPDESCS_DATA5"}, + {0x2DC, "GPUREG_VSH_OPDESCS_DATA6"}, + {0x2DD, "GPUREG_VSH_OPDESCS_DATA7"}, +}; + +const char* Regs::GetRegisterName(u16 index) { + auto found = std::lower_bound(std::begin(register_names), std::end(register_names), index, + [](auto p, auto i) { return p.first < i; }); + if (found->first == index) { + return found->second; + } else { + // Return empty string if no match is found + return ""; + } +} + +} // namespace Pica diff --git a/src/video_core/regs.h b/src/video_core/regs.h new file mode 100644 index 000000000..86826088b --- /dev/null +++ b/src/video_core/regs.h @@ -0,0 +1,142 @@ +// Copyright 2017 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <array> +#include <cstddef> +#include <string> +#ifndef _MSC_VER +#include <type_traits> // for std::enable_if +#endif + +#include "common/common_funcs.h" +#include "common/common_types.h" +#include "video_core/regs_framebuffer.h" +#include "video_core/regs_lighting.h" +#include "video_core/regs_pipeline.h" +#include "video_core/regs_rasterizer.h" +#include "video_core/regs_shader.h" +#include "video_core/regs_texturing.h" + +namespace Pica { + +// Returns index corresponding to the Regs member labeled by field_name +// TODO: Due to Visual studio bug 209229, offsetof does not return constant expressions +// when used with array elements (e.g. PICA_REG_INDEX(vs_uniform_setup.set_value[1])). +// For details cf. +// https://connect.microsoft.com/VisualStudio/feedback/details/209229/offsetof-does-not-produce-a-constant-expression-for-array-members +// Hopefully, this will be fixed sometime in the future. +// For lack of better alternatives, we currently hardcode the offsets when constant +// expressions are needed via PICA_REG_INDEX_WORKAROUND (on sane compilers, static_asserts +// will then make sure the offsets indeed match the automatically calculated ones). +#define PICA_REG_INDEX(field_name) (offsetof(Pica::Regs, field_name) / sizeof(u32)) +#if defined(_MSC_VER) +#define PICA_REG_INDEX_WORKAROUND(field_name, backup_workaround_index) (backup_workaround_index) +#else +// NOTE: Yeah, hacking in a static_assert here just to workaround the lacking MSVC compiler +// really is this annoying. This macro just forwards its first argument to PICA_REG_INDEX +// and then performs a (no-op) cast to size_t iff the second argument matches the expected +// field offset. Otherwise, the compiler will fail to compile this code. +#define PICA_REG_INDEX_WORKAROUND(field_name, backup_workaround_index) \ + ((typename std::enable_if<backup_workaround_index == PICA_REG_INDEX(field_name), \ + size_t>::type)PICA_REG_INDEX(field_name)) +#endif // _MSC_VER + +struct Regs { + static constexpr size_t NUM_REGS = 0x300; + + union { + struct { + INSERT_PADDING_WORDS(0x10); + u32 trigger_irq; + INSERT_PADDING_WORDS(0x2f); + RasterizerRegs rasterizer; + TexturingRegs texturing; + FramebufferRegs framebuffer; + LightingRegs lighting; + PipelineRegs pipeline; + ShaderRegs gs; + ShaderRegs vs; + INSERT_PADDING_WORDS(0x20); + }; + std::array<u32, NUM_REGS> reg_array; + }; + + /// Map register indices to names readable by humans + static const char* GetRegisterName(u16 index); +}; + +static_assert(sizeof(Regs) == Regs::NUM_REGS * sizeof(u32), "Regs struct has wrong size"); + +// TODO: MSVC does not support using offsetof() on non-static data members even though this +// is technically allowed since C++11. This macro should be enabled once MSVC adds +// support for that. +#ifndef _MSC_VER +#define ASSERT_REG_POSITION(field_name, position) \ + static_assert(offsetof(Regs, field_name) == position * 4, \ + "Field " #field_name " has invalid position") + +ASSERT_REG_POSITION(trigger_irq, 0x10); + +ASSERT_REG_POSITION(rasterizer, 0x40); +ASSERT_REG_POSITION(rasterizer.cull_mode, 0x40); +ASSERT_REG_POSITION(rasterizer.viewport_size_x, 0x41); +ASSERT_REG_POSITION(rasterizer.viewport_size_y, 0x43); +ASSERT_REG_POSITION(rasterizer.viewport_depth_range, 0x4d); +ASSERT_REG_POSITION(rasterizer.viewport_depth_near_plane, 0x4e); +ASSERT_REG_POSITION(rasterizer.vs_output_attributes[0], 0x50); +ASSERT_REG_POSITION(rasterizer.vs_output_attributes[1], 0x51); +ASSERT_REG_POSITION(rasterizer.scissor_test, 0x65); +ASSERT_REG_POSITION(rasterizer.viewport_corner, 0x68); +ASSERT_REG_POSITION(rasterizer.depthmap_enable, 0x6D); + +ASSERT_REG_POSITION(texturing, 0x80); +ASSERT_REG_POSITION(texturing.texture0_enable, 0x80); +ASSERT_REG_POSITION(texturing.texture0, 0x81); +ASSERT_REG_POSITION(texturing.texture0_format, 0x8e); +ASSERT_REG_POSITION(texturing.fragment_lighting_enable, 0x8f); +ASSERT_REG_POSITION(texturing.texture1, 0x91); +ASSERT_REG_POSITION(texturing.texture1_format, 0x96); +ASSERT_REG_POSITION(texturing.texture2, 0x99); +ASSERT_REG_POSITION(texturing.texture2_format, 0x9e); +ASSERT_REG_POSITION(texturing.tev_stage0, 0xc0); +ASSERT_REG_POSITION(texturing.tev_stage1, 0xc8); +ASSERT_REG_POSITION(texturing.tev_stage2, 0xd0); +ASSERT_REG_POSITION(texturing.tev_stage3, 0xd8); +ASSERT_REG_POSITION(texturing.tev_combiner_buffer_input, 0xe0); +ASSERT_REG_POSITION(texturing.fog_mode, 0xe0); +ASSERT_REG_POSITION(texturing.fog_color, 0xe1); +ASSERT_REG_POSITION(texturing.fog_lut_offset, 0xe6); +ASSERT_REG_POSITION(texturing.fog_lut_data, 0xe8); +ASSERT_REG_POSITION(texturing.tev_stage4, 0xf0); +ASSERT_REG_POSITION(texturing.tev_stage5, 0xf8); +ASSERT_REG_POSITION(texturing.tev_combiner_buffer_color, 0xfd); + +ASSERT_REG_POSITION(framebuffer, 0x100); +ASSERT_REG_POSITION(framebuffer.output_merger, 0x100); +ASSERT_REG_POSITION(framebuffer.framebuffer, 0x110); + +ASSERT_REG_POSITION(lighting, 0x140); + +ASSERT_REG_POSITION(pipeline, 0x200); +ASSERT_REG_POSITION(pipeline.vertex_attributes, 0x200); +ASSERT_REG_POSITION(pipeline.index_array, 0x227); +ASSERT_REG_POSITION(pipeline.num_vertices, 0x228); +ASSERT_REG_POSITION(pipeline.vertex_offset, 0x22a); +ASSERT_REG_POSITION(pipeline.trigger_draw, 0x22e); +ASSERT_REG_POSITION(pipeline.trigger_draw_indexed, 0x22f); +ASSERT_REG_POSITION(pipeline.vs_default_attributes_setup, 0x232); +ASSERT_REG_POSITION(pipeline.command_buffer, 0x238); +ASSERT_REG_POSITION(pipeline.gpu_mode, 0x245); +ASSERT_REG_POSITION(pipeline.triangle_topology, 0x25e); +ASSERT_REG_POSITION(pipeline.restart_primitive, 0x25f); + +ASSERT_REG_POSITION(gs, 0x280); +ASSERT_REG_POSITION(vs, 0x2b0); + +#undef ASSERT_REG_POSITION +#endif // !defined(_MSC_VER) + +} // namespace Pica diff --git a/src/video_core/regs_framebuffer.h b/src/video_core/regs_framebuffer.h new file mode 100644 index 000000000..9ddc79243 --- /dev/null +++ b/src/video_core/regs_framebuffer.h @@ -0,0 +1,284 @@ +// Copyright 2017 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <array> + +#include "common/assert.h" +#include "common/bit_field.h" +#include "common/common_funcs.h" +#include "common/common_types.h" +#include "common/logging/log.h" + +namespace Pica { + +struct FramebufferRegs { + enum class LogicOp : u32 { + Clear = 0, + And = 1, + AndReverse = 2, + Copy = 3, + Set = 4, + CopyInverted = 5, + NoOp = 6, + Invert = 7, + Nand = 8, + Or = 9, + Nor = 10, + Xor = 11, + Equiv = 12, + AndInverted = 13, + OrReverse = 14, + OrInverted = 15, + }; + + enum class BlendEquation : u32 { + Add = 0, + Subtract = 1, + ReverseSubtract = 2, + Min = 3, + Max = 4, + }; + + enum class BlendFactor : u32 { + Zero = 0, + One = 1, + SourceColor = 2, + OneMinusSourceColor = 3, + DestColor = 4, + OneMinusDestColor = 5, + SourceAlpha = 6, + OneMinusSourceAlpha = 7, + DestAlpha = 8, + OneMinusDestAlpha = 9, + ConstantColor = 10, + OneMinusConstantColor = 11, + ConstantAlpha = 12, + OneMinusConstantAlpha = 13, + SourceAlphaSaturate = 14, + }; + + enum class CompareFunc : u32 { + Never = 0, + Always = 1, + Equal = 2, + NotEqual = 3, + LessThan = 4, + LessThanOrEqual = 5, + GreaterThan = 6, + GreaterThanOrEqual = 7, + }; + + enum class StencilAction : u32 { + Keep = 0, + Zero = 1, + Replace = 2, + Increment = 3, + Decrement = 4, + Invert = 5, + IncrementWrap = 6, + DecrementWrap = 7, + }; + + struct { + union { + // If false, logic blending is used + BitField<8, 1, u32> alphablend_enable; + }; + + union { + BitField<0, 3, BlendEquation> blend_equation_rgb; + BitField<8, 3, BlendEquation> blend_equation_a; + + BitField<16, 4, BlendFactor> factor_source_rgb; + BitField<20, 4, BlendFactor> factor_dest_rgb; + + BitField<24, 4, BlendFactor> factor_source_a; + BitField<28, 4, BlendFactor> factor_dest_a; + } alpha_blending; + + union { + BitField<0, 4, LogicOp> logic_op; + }; + + union { + u32 raw; + BitField<0, 8, u32> r; + BitField<8, 8, u32> g; + BitField<16, 8, u32> b; + BitField<24, 8, u32> a; + } blend_const; + + union { + BitField<0, 1, u32> enable; + BitField<4, 3, CompareFunc> func; + BitField<8, 8, u32> ref; + } alpha_test; + + struct { + union { + // Raw value of this register + u32 raw_func; + + // If true, enable stencil testing + BitField<0, 1, u32> enable; + + // Comparison operation for stencil testing + BitField<4, 3, CompareFunc> func; + + // Mask used to control writing to the stencil buffer + BitField<8, 8, u32> write_mask; + + // Value to compare against for stencil testing + BitField<16, 8, u32> reference_value; + + // Mask to apply on stencil test inputs + BitField<24, 8, u32> input_mask; + }; + + union { + // Raw value of this register + u32 raw_op; + + // Action to perform when the stencil test fails + BitField<0, 3, StencilAction> action_stencil_fail; + + // Action to perform when stencil testing passed but depth testing fails + BitField<4, 3, StencilAction> action_depth_fail; + + // Action to perform when both stencil and depth testing pass + BitField<8, 3, StencilAction> action_depth_pass; + }; + } stencil_test; + + union { + BitField<0, 1, u32> depth_test_enable; + BitField<4, 3, CompareFunc> depth_test_func; + BitField<8, 1, u32> red_enable; + BitField<9, 1, u32> green_enable; + BitField<10, 1, u32> blue_enable; + BitField<11, 1, u32> alpha_enable; + BitField<12, 1, u32> depth_write_enable; + }; + + INSERT_PADDING_WORDS(0x8); + } output_merger; + + // Components are laid out in reverse byte order, most significant bits first. + enum class ColorFormat : u32 { + RGBA8 = 0, + RGB8 = 1, + RGB5A1 = 2, + RGB565 = 3, + RGBA4 = 4, + }; + + enum class DepthFormat : u32 { + D16 = 0, + D24 = 2, + D24S8 = 3, + }; + + // Returns the number of bytes in the specified color format + static unsigned BytesPerColorPixel(ColorFormat format) { + switch (format) { + case ColorFormat::RGBA8: + return 4; + case ColorFormat::RGB8: + return 3; + case ColorFormat::RGB5A1: + case ColorFormat::RGB565: + case ColorFormat::RGBA4: + return 2; + default: + LOG_CRITICAL(HW_GPU, "Unknown color format %u", format); + UNIMPLEMENTED(); + } + } + + struct FramebufferConfig { + INSERT_PADDING_WORDS(0x3); + + union { + BitField<0, 4, u32> allow_color_write; // 0 = disable, else enable + }; + + INSERT_PADDING_WORDS(0x1); + + union { + BitField<0, 2, u32> allow_depth_stencil_write; // 0 = disable, else enable + }; + + DepthFormat depth_format; // TODO: Should be a BitField! + BitField<16, 3, ColorFormat> color_format; + + INSERT_PADDING_WORDS(0x4); + + u32 depth_buffer_address; + u32 color_buffer_address; + + union { + // Apparently, the framebuffer width is stored as expected, + // while the height is stored as the actual height minus one. + // Hence, don't access these fields directly but use the accessors + // GetWidth() and GetHeight() instead. + BitField<0, 11, u32> width; + BitField<12, 10, u32> height; + }; + + INSERT_PADDING_WORDS(0x1); + + inline PAddr GetColorBufferPhysicalAddress() const { + return color_buffer_address * 8; + } + inline PAddr GetDepthBufferPhysicalAddress() const { + return depth_buffer_address * 8; + } + + inline u32 GetWidth() const { + return width; + } + + inline u32 GetHeight() const { + return height + 1; + } + } framebuffer; + + // Returns the number of bytes in the specified depth format + static u32 BytesPerDepthPixel(DepthFormat format) { + switch (format) { + case DepthFormat::D16: + return 2; + case DepthFormat::D24: + return 3; + case DepthFormat::D24S8: + return 4; + default: + LOG_CRITICAL(HW_GPU, "Unknown depth format %u", format); + UNIMPLEMENTED(); + } + } + + // Returns the number of bits per depth component of the specified depth format + static u32 DepthBitsPerPixel(DepthFormat format) { + switch (format) { + case DepthFormat::D16: + return 16; + case DepthFormat::D24: + case DepthFormat::D24S8: + return 24; + default: + LOG_CRITICAL(HW_GPU, "Unknown depth format %u", format); + UNIMPLEMENTED(); + } + } + + INSERT_PADDING_WORDS(0x20); +}; + +static_assert(sizeof(FramebufferRegs) == 0x40 * sizeof(u32), + "FramebufferRegs struct has incorrect size"); + +} // namespace Pica diff --git a/src/video_core/regs_lighting.h b/src/video_core/regs_lighting.h new file mode 100644 index 000000000..6793405d9 --- /dev/null +++ b/src/video_core/regs_lighting.h @@ -0,0 +1,294 @@ +// Copyright 2017 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <array> + +#include "common/assert.h" +#include "common/bit_field.h" +#include "common/common_funcs.h" +#include "common/common_types.h" +#include "common/vector_math.h" + +namespace Pica { + +struct LightingRegs { + enum class LightingSampler { + Distribution0 = 0, + Distribution1 = 1, + Fresnel = 3, + ReflectBlue = 4, + ReflectGreen = 5, + ReflectRed = 6, + SpotlightAttenuation = 8, + DistanceAttenuation = 16, + }; + + /** + * Pica fragment lighting supports using different LUTs for each lighting component: Reflectance + * R, G, and B channels, distribution function for specular components 0 and 1, fresnel factor, + * and spotlight attenuation. Furthermore, which LUTs are used for each channel (or whether a + * channel is enabled at all) is specified by various pre-defined lighting configurations. With + * configurations that require more LUTs, more cycles are required on HW to perform lighting + * computations. + */ + enum class LightingConfig : u32 { + Config0 = 0, ///< Reflect Red, Distribution 0, Spotlight + Config1 = 1, ///< Reflect Red, Fresnel, Spotlight + Config2 = 2, ///< Reflect Red, Distribution 0/1 + Config3 = 3, ///< Distribution 0/1, Fresnel + Config4 = 4, ///< Reflect Red/Green/Blue, Distribution 0/1, Spotlight + Config5 = 5, ///< Reflect Red/Green/Blue, Distribution 0, Fresnel, Spotlight + Config6 = 6, ///< Reflect Red, Distribution 0/1, Fresnel, Spotlight + + Config7 = 8, ///< Reflect Red/Green/Blue, Distribution 0/1, Fresnel, Spotlight + ///< NOTE: '8' is intentional, '7' does not appear to be a valid configuration + }; + + /// Selects which lighting components are affected by fresnel + enum class LightingFresnelSelector : u32 { + None = 0, ///< Fresnel is disabled + PrimaryAlpha = 1, ///< Primary (diffuse) lighting alpha is affected by fresnel + SecondaryAlpha = 2, ///< Secondary (specular) lighting alpha is affected by fresnel + Both = + PrimaryAlpha | + SecondaryAlpha, ///< Both primary and secondary lighting alphas are affected by fresnel + }; + + /// Factor used to scale the output of a lighting LUT + enum class LightingScale : u32 { + Scale1 = 0, ///< Scale is 1x + Scale2 = 1, ///< Scale is 2x + Scale4 = 2, ///< Scale is 4x + Scale8 = 3, ///< Scale is 8x + + Scale1_4 = 6, ///< Scale is 0.25x + Scale1_2 = 7, ///< Scale is 0.5x + }; + + enum class LightingLutInput : u32 { + NH = 0, // Cosine of the angle between the normal and half-angle vectors + VH = 1, // Cosine of the angle between the view and half-angle vectors + NV = 2, // Cosine of the angle between the normal and the view vector + LN = 3, // Cosine of the angle between the light and the normal vectors + }; + + enum class LightingBumpMode : u32 { + None = 0, + NormalMap = 1, + TangentMap = 2, + }; + + union LightColor { + BitField<0, 10, u32> b; + BitField<10, 10, u32> g; + BitField<20, 10, u32> r; + + Math::Vec3f ToVec3f() const { + // These fields are 10 bits wide, however 255 corresponds to 1.0f for each color + // component + return Math::MakeVec((f32)r / 255.f, (f32)g / 255.f, (f32)b / 255.f); + } + }; + + /// Returns true if the specified lighting sampler is supported by the current Pica lighting + /// configuration + static bool IsLightingSamplerSupported(LightingConfig config, LightingSampler sampler) { + switch (sampler) { + case LightingSampler::Distribution0: + return (config != LightingConfig::Config1); + + case LightingSampler::Distribution1: + return (config != LightingConfig::Config0) && (config != LightingConfig::Config1) && + (config != LightingConfig::Config5); + + case LightingSampler::Fresnel: + return (config != LightingConfig::Config0) && (config != LightingConfig::Config2) && + (config != LightingConfig::Config4); + + case LightingSampler::ReflectRed: + return (config != LightingConfig::Config3); + + case LightingSampler::ReflectGreen: + case LightingSampler::ReflectBlue: + return (config == LightingConfig::Config4) || (config == LightingConfig::Config5) || + (config == LightingConfig::Config7); + default: + UNREACHABLE_MSG("Regs::IsLightingSamplerSupported: Reached " + "unreachable section, sampler should be one " + "of Distribution0, Distribution1, Fresnel, " + "ReflectRed, ReflectGreen or ReflectBlue, instead " + "got %i", + static_cast<int>(config)); + } + } + + struct LightSrc { + LightColor specular_0; // material.specular_0 * light.specular_0 + LightColor specular_1; // material.specular_1 * light.specular_1 + LightColor diffuse; // material.diffuse * light.diffuse + LightColor ambient; // material.ambient * light.ambient + + // Encoded as 16-bit floating point + union { + BitField<0, 16, u32> x; + BitField<16, 16, u32> y; + }; + union { + BitField<0, 16, u32> z; + }; + + INSERT_PADDING_WORDS(0x3); + + union { + BitField<0, 1, u32> directional; + BitField<1, 1, u32> two_sided_diffuse; // When disabled, clamp dot-product to 0 + } config; + + BitField<0, 20, u32> dist_atten_bias; + BitField<0, 20, u32> dist_atten_scale; + + INSERT_PADDING_WORDS(0x4); + }; + static_assert(sizeof(LightSrc) == 0x10 * sizeof(u32), "LightSrc structure must be 0x10 words"); + + LightSrc light[8]; + LightColor global_ambient; // Emission + (material.ambient * lighting.ambient) + INSERT_PADDING_WORDS(0x1); + BitField<0, 3, u32> max_light_index; // Number of enabled lights - 1 + + union { + BitField<2, 2, LightingFresnelSelector> fresnel_selector; + BitField<4, 4, LightingConfig> config; + BitField<22, 2, u32> bump_selector; // 0: Texture 0, 1: Texture 1, 2: Texture 2 + BitField<27, 1, u32> clamp_highlights; + BitField<28, 2, LightingBumpMode> bump_mode; + BitField<30, 1, u32> disable_bump_renorm; + } config0; + + union { + BitField<16, 1, u32> disable_lut_d0; + BitField<17, 1, u32> disable_lut_d1; + BitField<19, 1, u32> disable_lut_fr; + BitField<20, 1, u32> disable_lut_rr; + BitField<21, 1, u32> disable_lut_rg; + BitField<22, 1, u32> disable_lut_rb; + + // Each bit specifies whether distance attenuation should be applied for the corresponding + // light. + BitField<24, 1, u32> disable_dist_atten_light_0; + BitField<25, 1, u32> disable_dist_atten_light_1; + BitField<26, 1, u32> disable_dist_atten_light_2; + BitField<27, 1, u32> disable_dist_atten_light_3; + BitField<28, 1, u32> disable_dist_atten_light_4; + BitField<29, 1, u32> disable_dist_atten_light_5; + BitField<30, 1, u32> disable_dist_atten_light_6; + BitField<31, 1, u32> disable_dist_atten_light_7; + } config1; + + bool IsDistAttenDisabled(unsigned index) const { + const unsigned disable[] = { + config1.disable_dist_atten_light_0, config1.disable_dist_atten_light_1, + config1.disable_dist_atten_light_2, config1.disable_dist_atten_light_3, + config1.disable_dist_atten_light_4, config1.disable_dist_atten_light_5, + config1.disable_dist_atten_light_6, config1.disable_dist_atten_light_7}; + return disable[index] != 0; + } + + union { + BitField<0, 8, u32> index; ///< Index at which to set data in the LUT + BitField<8, 5, u32> type; ///< Type of LUT for which to set data + } lut_config; + + BitField<0, 1, u32> disable; + INSERT_PADDING_WORDS(0x1); + + // When data is written to any of these registers, it gets written to the lookup table of the + // selected type at the selected index, specified above in the `lut_config` register. With each + // write, `lut_config.index` is incremented. It does not matter which of these registers is + // written to, the behavior will be the same. + u32 lut_data[8]; + + // These are used to specify if absolute (abs) value should be used for each LUT index. When + // abs mode is disabled, LUT indexes are in the range of (-1.0, 1.0). Otherwise, they are in + // the range of (0.0, 1.0). + union { + BitField<1, 1, u32> disable_d0; + BitField<5, 1, u32> disable_d1; + BitField<9, 1, u32> disable_sp; + BitField<13, 1, u32> disable_fr; + BitField<17, 1, u32> disable_rb; + BitField<21, 1, u32> disable_rg; + BitField<25, 1, u32> disable_rr; + } abs_lut_input; + + union { + BitField<0, 3, LightingLutInput> d0; + BitField<4, 3, LightingLutInput> d1; + BitField<8, 3, LightingLutInput> sp; + BitField<12, 3, LightingLutInput> fr; + BitField<16, 3, LightingLutInput> rb; + BitField<20, 3, LightingLutInput> rg; + BitField<24, 3, LightingLutInput> rr; + } lut_input; + + union { + BitField<0, 3, LightingScale> d0; + BitField<4, 3, LightingScale> d1; + BitField<8, 3, LightingScale> sp; + BitField<12, 3, LightingScale> fr; + BitField<16, 3, LightingScale> rb; + BitField<20, 3, LightingScale> rg; + BitField<24, 3, LightingScale> rr; + + static float GetScale(LightingScale scale) { + switch (scale) { + case LightingScale::Scale1: + return 1.0f; + case LightingScale::Scale2: + return 2.0f; + case LightingScale::Scale4: + return 4.0f; + case LightingScale::Scale8: + return 8.0f; + case LightingScale::Scale1_4: + return 0.25f; + case LightingScale::Scale1_2: + return 0.5f; + } + return 0.0f; + } + } lut_scale; + + INSERT_PADDING_WORDS(0x6); + + union { + // There are 8 light enable "slots", corresponding to the total number of lights supported + // by Pica. For N enabled lights (specified by register 0x1c2, or 'src_num' above), the + // first N slots below will be set to integers within the range of 0-7, corresponding to the + // actual light that is enabled for each slot. + + BitField<0, 3, u32> slot_0; + BitField<4, 3, u32> slot_1; + BitField<8, 3, u32> slot_2; + BitField<12, 3, u32> slot_3; + BitField<16, 3, u32> slot_4; + BitField<20, 3, u32> slot_5; + BitField<24, 3, u32> slot_6; + BitField<28, 3, u32> slot_7; + + unsigned GetNum(unsigned index) const { + const unsigned enable_slots[] = {slot_0, slot_1, slot_2, slot_3, + slot_4, slot_5, slot_6, slot_7}; + return enable_slots[index]; + } + } light_enable; + + INSERT_PADDING_WORDS(0x26); +}; + +static_assert(sizeof(LightingRegs) == 0xC0 * sizeof(u32), "LightingRegs struct has incorrect size"); + +} // namespace Pica diff --git a/src/video_core/regs_pipeline.h b/src/video_core/regs_pipeline.h new file mode 100644 index 000000000..0a4ec6e1e --- /dev/null +++ b/src/video_core/regs_pipeline.h @@ -0,0 +1,230 @@ +// Copyright 2017 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <array> + +#include "common/assert.h" +#include "common/bit_field.h" +#include "common/common_funcs.h" +#include "common/common_types.h" + +namespace Pica { + +struct PipelineRegs { + enum class VertexAttributeFormat : u32 { + BYTE = 0, + UBYTE = 1, + SHORT = 2, + FLOAT = 3, + }; + + struct { + BitField<0, 29, u32> base_address; + + PAddr GetPhysicalBaseAddress() const { + return base_address * 8; + } + + // Descriptor for internal vertex attributes + union { + BitField<0, 2, VertexAttributeFormat> format0; // size of one element + BitField<2, 2, u32> size0; // number of elements minus 1 + BitField<4, 2, VertexAttributeFormat> format1; + BitField<6, 2, u32> size1; + BitField<8, 2, VertexAttributeFormat> format2; + BitField<10, 2, u32> size2; + BitField<12, 2, VertexAttributeFormat> format3; + BitField<14, 2, u32> size3; + BitField<16, 2, VertexAttributeFormat> format4; + BitField<18, 2, u32> size4; + BitField<20, 2, VertexAttributeFormat> format5; + BitField<22, 2, u32> size5; + BitField<24, 2, VertexAttributeFormat> format6; + BitField<26, 2, u32> size6; + BitField<28, 2, VertexAttributeFormat> format7; + BitField<30, 2, u32> size7; + }; + + union { + BitField<0, 2, VertexAttributeFormat> format8; + BitField<2, 2, u32> size8; + BitField<4, 2, VertexAttributeFormat> format9; + BitField<6, 2, u32> size9; + BitField<8, 2, VertexAttributeFormat> format10; + BitField<10, 2, u32> size10; + BitField<12, 2, VertexAttributeFormat> format11; + BitField<14, 2, u32> size11; + + BitField<16, 12, u32> attribute_mask; + + // number of total attributes minus 1 + BitField<28, 4, u32> max_attribute_index; + }; + + inline VertexAttributeFormat GetFormat(int n) const { + VertexAttributeFormat formats[] = {format0, format1, format2, format3, + format4, format5, format6, format7, + format8, format9, format10, format11}; + return formats[n]; + } + + inline int GetNumElements(int n) const { + u32 sizes[] = {size0, size1, size2, size3, size4, size5, + size6, size7, size8, size9, size10, size11}; + return (int)sizes[n] + 1; + } + + inline int GetElementSizeInBytes(int n) const { + return (GetFormat(n) == VertexAttributeFormat::FLOAT) + ? 4 + : (GetFormat(n) == VertexAttributeFormat::SHORT) ? 2 : 1; + } + + inline int GetStride(int n) const { + return GetNumElements(n) * GetElementSizeInBytes(n); + } + + inline bool IsDefaultAttribute(int id) const { + return (id >= 12) || (attribute_mask & (1ULL << id)) != 0; + } + + inline int GetNumTotalAttributes() const { + return (int)max_attribute_index + 1; + } + + // Attribute loaders map the source vertex data to input attributes + // This e.g. allows to load different attributes from different memory locations + struct { + // Source attribute data offset from the base address + u32 data_offset; + + union { + BitField<0, 4, u32> comp0; + BitField<4, 4, u32> comp1; + BitField<8, 4, u32> comp2; + BitField<12, 4, u32> comp3; + BitField<16, 4, u32> comp4; + BitField<20, 4, u32> comp5; + BitField<24, 4, u32> comp6; + BitField<28, 4, u32> comp7; + }; + + union { + BitField<0, 4, u32> comp8; + BitField<4, 4, u32> comp9; + BitField<8, 4, u32> comp10; + BitField<12, 4, u32> comp11; + + // bytes for a single vertex in this loader + BitField<16, 8, u32> byte_count; + + BitField<28, 4, u32> component_count; + }; + + inline int GetComponent(int n) const { + u32 components[] = {comp0, comp1, comp2, comp3, comp4, comp5, + comp6, comp7, comp8, comp9, comp10, comp11}; + return (int)components[n]; + } + } attribute_loaders[12]; + } vertex_attributes; + + struct { + enum IndexFormat : u32 { + BYTE = 0, + SHORT = 1, + }; + + union { + BitField<0, 31, u32> offset; // relative to base attribute address + BitField<31, 1, IndexFormat> format; + }; + } index_array; + + // Number of vertices to render + u32 num_vertices; + + INSERT_PADDING_WORDS(0x1); + + // The index of the first vertex to render + u32 vertex_offset; + + INSERT_PADDING_WORDS(0x3); + + // These two trigger rendering of triangles + u32 trigger_draw; + u32 trigger_draw_indexed; + + INSERT_PADDING_WORDS(0x2); + + // These registers are used to setup the default "fall-back" vertex shader attributes + struct { + // Index of the current default attribute + u32 index; + + // Writing to these registers sets the "current" default attribute. + u32 set_value[3]; + } vs_default_attributes_setup; + + INSERT_PADDING_WORDS(0x2); + + struct { + // There are two channels that can be used to configure the next command buffer, which can + // be then executed by writing to the "trigger" registers. There are two reasons why a game + // might use this feature: + // 1) With this, an arbitrary number of additional command buffers may be executed in + // sequence without requiring any intervention of the CPU after the initial one is + // kicked off. + // 2) Games can configure these registers to provide a command list subroutine mechanism. + + BitField<0, 20, u32> size[2]; ///< Size (in bytes / 8) of each channel's command buffer + BitField<0, 28, u32> addr[2]; ///< Physical address / 8 of each channel's command buffer + u32 trigger[2]; ///< Triggers execution of the channel's command buffer when written to + + unsigned GetSize(unsigned index) const { + ASSERT(index < 2); + return 8 * size[index]; + } + + PAddr GetPhysicalAddress(unsigned index) const { + ASSERT(index < 2); + return (PAddr)(8 * addr[index]); + } + } command_buffer; + + INSERT_PADDING_WORDS(4); + + /// Number of input attributes to the vertex shader minus 1 + BitField<0, 4, u32> max_input_attrib_index; + + INSERT_PADDING_WORDS(2); + + enum class GPUMode : u32 { + Drawing = 0, + Configuring = 1, + }; + + GPUMode gpu_mode; + + INSERT_PADDING_WORDS(0x18); + + enum class TriangleTopology : u32 { + List = 0, + Strip = 1, + Fan = 2, + Shader = 3, // Programmable setup unit implemented in a geometry shader + }; + + BitField<8, 2, TriangleTopology> triangle_topology; + + u32 restart_primitive; + + INSERT_PADDING_WORDS(0x20); +}; + +static_assert(sizeof(PipelineRegs) == 0x80 * sizeof(u32), "PipelineRegs struct has incorrect size"); + +} // namespace Pica diff --git a/src/video_core/regs_rasterizer.h b/src/video_core/regs_rasterizer.h new file mode 100644 index 000000000..a471a3b38 --- /dev/null +++ b/src/video_core/regs_rasterizer.h @@ -0,0 +1,129 @@ +// Copyright 2017 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <array> + +#include "common/bit_field.h" +#include "common/common_funcs.h" +#include "common/common_types.h" + +namespace Pica { + +struct RasterizerRegs { + enum class CullMode : u32 { + // Select which polygons are considered to be "frontfacing". + KeepAll = 0, + KeepClockWise = 1, + KeepCounterClockWise = 2, + // TODO: What does the third value imply? + }; + + union { + BitField<0, 2, CullMode> cull_mode; + }; + + BitField<0, 24, u32> viewport_size_x; + + INSERT_PADDING_WORDS(0x1); + + BitField<0, 24, u32> viewport_size_y; + + INSERT_PADDING_WORDS(0x9); + + BitField<0, 24, u32> viewport_depth_range; // float24 + BitField<0, 24, u32> viewport_depth_near_plane; // float24 + + BitField<0, 3, u32> vs_output_total; + + union VSOutputAttributes { + // Maps components of output vertex attributes to semantics + enum Semantic : u32 { + POSITION_X = 0, + POSITION_Y = 1, + POSITION_Z = 2, + POSITION_W = 3, + + QUATERNION_X = 4, + QUATERNION_Y = 5, + QUATERNION_Z = 6, + QUATERNION_W = 7, + + COLOR_R = 8, + COLOR_G = 9, + COLOR_B = 10, + COLOR_A = 11, + + TEXCOORD0_U = 12, + TEXCOORD0_V = 13, + TEXCOORD1_U = 14, + TEXCOORD1_V = 15, + + TEXCOORD0_W = 16, + + VIEW_X = 18, + VIEW_Y = 19, + VIEW_Z = 20, + + TEXCOORD2_U = 22, + TEXCOORD2_V = 23, + + INVALID = 31, + }; + + BitField<0, 5, Semantic> map_x; + BitField<8, 5, Semantic> map_y; + BitField<16, 5, Semantic> map_z; + BitField<24, 5, Semantic> map_w; + } vs_output_attributes[7]; + + INSERT_PADDING_WORDS(0xe); + + enum class ScissorMode : u32 { + Disabled = 0, + Exclude = 1, // Exclude pixels inside the scissor box + + Include = 3 // Exclude pixels outside the scissor box + }; + + struct { + BitField<0, 2, ScissorMode> mode; + + union { + BitField<0, 16, u32> x1; + BitField<16, 16, u32> y1; + }; + + union { + BitField<0, 16, u32> x2; + BitField<16, 16, u32> y2; + }; + } scissor_test; + + union { + BitField<0, 10, s32> x; + BitField<16, 10, s32> y; + } viewport_corner; + + INSERT_PADDING_WORDS(0x1); + + // TODO: early depth + INSERT_PADDING_WORDS(0x1); + + INSERT_PADDING_WORDS(0x2); + + enum DepthBuffering : u32 { + WBuffering = 0, + ZBuffering = 1, + }; + BitField<0, 1, DepthBuffering> depthmap_enable; + + INSERT_PADDING_WORDS(0x12); +}; + +static_assert(sizeof(RasterizerRegs) == 0x40 * sizeof(u32), + "RasterizerRegs struct has incorrect size"); + +} // namespace Pica diff --git a/src/video_core/regs_shader.h b/src/video_core/regs_shader.h new file mode 100644 index 000000000..ddb1ee451 --- /dev/null +++ b/src/video_core/regs_shader.h @@ -0,0 +1,104 @@ +// Copyright 2017 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <array> + +#include "common/bit_field.h" +#include "common/common_funcs.h" +#include "common/common_types.h" + +namespace Pica { + +struct ShaderRegs { + BitField<0, 16, u32> bool_uniforms; + + union { + BitField<0, 8, u32> x; + BitField<8, 8, u32> y; + BitField<16, 8, u32> z; + BitField<24, 8, u32> w; + } int_uniforms[4]; + + INSERT_PADDING_WORDS(0x4); + + union { + // Number of input attributes to shader unit - 1 + BitField<0, 4, u32> max_input_attribute_index; + }; + + // Offset to shader program entry point (in words) + BitField<0, 16, u32> main_offset; + + /// Maps input attributes to registers. 4-bits per attribute, specifying a register index + u32 input_attribute_to_register_map_low; + u32 input_attribute_to_register_map_high; + + unsigned int GetRegisterForAttribute(unsigned int attribute_index) const { + u64 map = ((u64)input_attribute_to_register_map_high << 32) | + (u64)input_attribute_to_register_map_low; + return (map >> (attribute_index * 4)) & 0b1111; + } + + BitField<0, 16, u32> output_mask; + + // 0x28E, CODETRANSFER_END + INSERT_PADDING_WORDS(0x2); + + struct { + enum Format : u32 { + FLOAT24 = 0, + FLOAT32 = 1, + }; + + bool IsFloat32() const { + return format == FLOAT32; + } + + union { + // Index of the next uniform to write to + // TODO: ctrulib uses 8 bits for this, however that seems to yield lots of invalid + // indices + // TODO: Maybe the uppermost index is for the geometry shader? Investigate! + BitField<0, 7, u32> index; + + BitField<31, 1, Format> format; + }; + + // Writing to these registers sets the current uniform. + u32 set_value[8]; + + } uniform_setup; + + INSERT_PADDING_WORDS(0x2); + + struct { + // Offset of the next instruction to write code to. + // Incremented with each instruction write. + u32 offset; + + // Writing to these registers sets the "current" word in the shader program. + u32 set_word[8]; + } program; + + INSERT_PADDING_WORDS(0x1); + + // This register group is used to load an internal table of swizzling patterns, + // which are indexed by each shader instruction to specify vector component swizzling. + struct { + // Offset of the next swizzle pattern to write code to. + // Incremented with each instruction write. + u32 offset; + + // Writing to these registers sets the current swizzle pattern in the table. + u32 set_word[8]; + } swizzle_patterns; + + INSERT_PADDING_WORDS(0x2); +}; + +static_assert(sizeof(ShaderRegs) == 0x30 * sizeof(u32), "ShaderRegs struct has incorrect size"); + +} // namespace Pica diff --git a/src/video_core/regs_texturing.h b/src/video_core/regs_texturing.h new file mode 100644 index 000000000..be8bc6826 --- /dev/null +++ b/src/video_core/regs_texturing.h @@ -0,0 +1,328 @@ +// Copyright 2017 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <array> + +#include "common/assert.h" +#include "common/bit_field.h" +#include "common/common_funcs.h" +#include "common/common_types.h" + +namespace Pica { + +struct TexturingRegs { + struct TextureConfig { + enum TextureType : u32 { + Texture2D = 0, + TextureCube = 1, + Shadow2D = 2, + Projection2D = 3, + ShadowCube = 4, + Disabled = 5, + }; + + enum WrapMode : u32 { + ClampToEdge = 0, + ClampToBorder = 1, + Repeat = 2, + MirroredRepeat = 3, + }; + + enum TextureFilter : u32 { + Nearest = 0, + Linear = 1, + }; + + union { + u32 raw; + BitField<0, 8, u32> r; + BitField<8, 8, u32> g; + BitField<16, 8, u32> b; + BitField<24, 8, u32> a; + } border_color; + + union { + BitField<0, 16, u32> height; + BitField<16, 16, u32> width; + }; + + union { + BitField<1, 1, TextureFilter> mag_filter; + BitField<2, 1, TextureFilter> min_filter; + BitField<8, 2, WrapMode> wrap_t; + BitField<12, 2, WrapMode> wrap_s; + BitField<28, 2, TextureType> + type; ///< @note Only valid for texture 0 according to 3DBrew. + }; + + INSERT_PADDING_WORDS(0x1); + + u32 address; + + PAddr GetPhysicalAddress() const { + return address * 8; + } + + // texture1 and texture2 store the texture format directly after the address + // whereas texture0 inserts some additional flags inbetween. + // Hence, we store the format separately so that all other parameters can be described + // in a single structure. + }; + + enum class TextureFormat : u32 { + RGBA8 = 0, + RGB8 = 1, + RGB5A1 = 2, + RGB565 = 3, + RGBA4 = 4, + IA8 = 5, + RG8 = 6, ///< @note Also called HILO8 in 3DBrew. + I8 = 7, + A8 = 8, + IA4 = 9, + I4 = 10, + A4 = 11, + ETC1 = 12, // compressed + ETC1A4 = 13, // compressed + }; + + static unsigned NibblesPerPixel(TextureFormat format) { + switch (format) { + case TextureFormat::RGBA8: + return 8; + + case TextureFormat::RGB8: + return 6; + + case TextureFormat::RGB5A1: + case TextureFormat::RGB565: + case TextureFormat::RGBA4: + case TextureFormat::IA8: + case TextureFormat::RG8: + return 4; + + case TextureFormat::I4: + case TextureFormat::A4: + return 1; + + case TextureFormat::I8: + case TextureFormat::A8: + case TextureFormat::IA4: + + default: // placeholder for yet unknown formats + UNIMPLEMENTED(); + return 0; + } + } + + union { + BitField<0, 1, u32> texture0_enable; + BitField<1, 1, u32> texture1_enable; + BitField<2, 1, u32> texture2_enable; + }; + TextureConfig texture0; + INSERT_PADDING_WORDS(0x8); + BitField<0, 4, TextureFormat> texture0_format; + BitField<0, 1, u32> fragment_lighting_enable; + INSERT_PADDING_WORDS(0x1); + TextureConfig texture1; + BitField<0, 4, TextureFormat> texture1_format; + INSERT_PADDING_WORDS(0x2); + TextureConfig texture2; + BitField<0, 4, TextureFormat> texture2_format; + INSERT_PADDING_WORDS(0x21); + + struct FullTextureConfig { + const bool enabled; + const TextureConfig config; + const TextureFormat format; + }; + const std::array<FullTextureConfig, 3> GetTextures() const { + return {{ + {texture0_enable.ToBool(), texture0, texture0_format}, + {texture1_enable.ToBool(), texture1, texture1_format}, + {texture2_enable.ToBool(), texture2, texture2_format}, + }}; + } + + // 0xc0-0xff: Texture Combiner (akin to glTexEnv) + struct TevStageConfig { + enum class Source : u32 { + PrimaryColor = 0x0, + PrimaryFragmentColor = 0x1, + SecondaryFragmentColor = 0x2, + + Texture0 = 0x3, + Texture1 = 0x4, + Texture2 = 0x5, + Texture3 = 0x6, + + PreviousBuffer = 0xd, + Constant = 0xe, + Previous = 0xf, + }; + + enum class ColorModifier : u32 { + SourceColor = 0x0, + OneMinusSourceColor = 0x1, + SourceAlpha = 0x2, + OneMinusSourceAlpha = 0x3, + SourceRed = 0x4, + OneMinusSourceRed = 0x5, + + SourceGreen = 0x8, + OneMinusSourceGreen = 0x9, + + SourceBlue = 0xc, + OneMinusSourceBlue = 0xd, + }; + + enum class AlphaModifier : u32 { + SourceAlpha = 0x0, + OneMinusSourceAlpha = 0x1, + SourceRed = 0x2, + OneMinusSourceRed = 0x3, + SourceGreen = 0x4, + OneMinusSourceGreen = 0x5, + SourceBlue = 0x6, + OneMinusSourceBlue = 0x7, + }; + + enum class Operation : u32 { + Replace = 0, + Modulate = 1, + Add = 2, + AddSigned = 3, + Lerp = 4, + Subtract = 5, + Dot3_RGB = 6, + + MultiplyThenAdd = 8, + AddThenMultiply = 9, + }; + + union { + u32 sources_raw; + BitField<0, 4, Source> color_source1; + BitField<4, 4, Source> color_source2; + BitField<8, 4, Source> color_source3; + BitField<16, 4, Source> alpha_source1; + BitField<20, 4, Source> alpha_source2; + BitField<24, 4, Source> alpha_source3; + }; + + union { + u32 modifiers_raw; + BitField<0, 4, ColorModifier> color_modifier1; + BitField<4, 4, ColorModifier> color_modifier2; + BitField<8, 4, ColorModifier> color_modifier3; + BitField<12, 3, AlphaModifier> alpha_modifier1; + BitField<16, 3, AlphaModifier> alpha_modifier2; + BitField<20, 3, AlphaModifier> alpha_modifier3; + }; + + union { + u32 ops_raw; + BitField<0, 4, Operation> color_op; + BitField<16, 4, Operation> alpha_op; + }; + + union { + u32 const_color; + BitField<0, 8, u32> const_r; + BitField<8, 8, u32> const_g; + BitField<16, 8, u32> const_b; + BitField<24, 8, u32> const_a; + }; + + union { + u32 scales_raw; + BitField<0, 2, u32> color_scale; + BitField<16, 2, u32> alpha_scale; + }; + + inline unsigned GetColorMultiplier() const { + return (color_scale < 3) ? (1 << color_scale) : 1; + } + + inline unsigned GetAlphaMultiplier() const { + return (alpha_scale < 3) ? (1 << alpha_scale) : 1; + } + }; + + TevStageConfig tev_stage0; + INSERT_PADDING_WORDS(0x3); + TevStageConfig tev_stage1; + INSERT_PADDING_WORDS(0x3); + TevStageConfig tev_stage2; + INSERT_PADDING_WORDS(0x3); + TevStageConfig tev_stage3; + INSERT_PADDING_WORDS(0x3); + + enum class FogMode : u32 { + None = 0, + Fog = 5, + Gas = 7, + }; + + union { + BitField<0, 3, FogMode> fog_mode; + BitField<16, 1, u32> fog_flip; + + union { + // Tev stages 0-3 write their output to the combiner buffer if the corresponding bit in + // these masks are set + BitField<8, 4, u32> update_mask_rgb; + BitField<12, 4, u32> update_mask_a; + + bool TevStageUpdatesCombinerBufferColor(unsigned stage_index) const { + return (stage_index < 4) && (update_mask_rgb & (1 << stage_index)); + } + + bool TevStageUpdatesCombinerBufferAlpha(unsigned stage_index) const { + return (stage_index < 4) && (update_mask_a & (1 << stage_index)); + } + } tev_combiner_buffer_input; + }; + + union { + u32 raw; + BitField<0, 8, u32> r; + BitField<8, 8, u32> g; + BitField<16, 8, u32> b; + } fog_color; + + INSERT_PADDING_WORDS(0x4); + + BitField<0, 16, u32> fog_lut_offset; + + INSERT_PADDING_WORDS(0x1); + + u32 fog_lut_data[8]; + + TevStageConfig tev_stage4; + INSERT_PADDING_WORDS(0x3); + TevStageConfig tev_stage5; + + union { + u32 raw; + BitField<0, 8, u32> r; + BitField<8, 8, u32> g; + BitField<16, 8, u32> b; + BitField<24, 8, u32> a; + } tev_combiner_buffer_color; + + INSERT_PADDING_WORDS(0x2); + + const std::array<TevStageConfig, 6> GetTevStages() const { + return {{tev_stage0, tev_stage1, tev_stage2, tev_stage3, tev_stage4, tev_stage5}}; + }; +}; + +static_assert(sizeof(TexturingRegs) == 0x80 * sizeof(u32), + "TexturingRegs struct has incorrect size"); + +} // namespace Pica diff --git a/src/video_core/renderer_base.cpp b/src/video_core/renderer_base.cpp index fd38175b3..f6ece5c4b 100644 --- a/src/video_core/renderer_base.cpp +++ b/src/video_core/renderer_base.cpp @@ -6,7 +6,7 @@ #include <memory> #include "video_core/renderer_base.h" #include "video_core/renderer_opengl/gl_rasterizer.h" -#include "video_core/swrasterizer.h" +#include "video_core/swrasterizer/swrasterizer.h" #include "video_core/video_core.h" void RendererBase::RefreshRasterizerSetting() { diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp index f3674e965..de1d5eba7 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer.cpp +++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp @@ -14,8 +14,10 @@ #include "common/microprofile.h" #include "common/vector_math.h" #include "core/hw/gpu.h" -#include "video_core/pica.h" #include "video_core/pica_state.h" +#include "video_core/regs_framebuffer.h" +#include "video_core/regs_rasterizer.h" +#include "video_core/regs_texturing.h" #include "video_core/renderer_opengl/gl_rasterizer.h" #include "video_core/renderer_opengl/gl_shader_gen.h" #include "video_core/renderer_opengl/gl_shader_util.h" @@ -26,16 +28,6 @@ MICROPROFILE_DEFINE(OpenGL_Drawing, "OpenGL", "Drawing", MP_RGB(128, 128, 192)); MICROPROFILE_DEFINE(OpenGL_Blits, "OpenGL", "Blits", MP_RGB(100, 100, 255)); MICROPROFILE_DEFINE(OpenGL_CacheManagement, "OpenGL", "Cache Mgmt", MP_RGB(100, 255, 100)); -static bool IsPassThroughTevStage(const Pica::Regs::TevStageConfig& stage) { - return (stage.color_op == Pica::Regs::TevStageConfig::Operation::Replace && - stage.alpha_op == Pica::Regs::TevStageConfig::Operation::Replace && - stage.color_source1 == Pica::Regs::TevStageConfig::Source::Previous && - stage.alpha_source1 == Pica::Regs::TevStageConfig::Source::Previous && - stage.color_modifier1 == Pica::Regs::TevStageConfig::ColorModifier::SourceColor && - stage.alpha_modifier1 == Pica::Regs::TevStageConfig::AlphaModifier::SourceAlpha && - stage.GetColorMultiplier() == 1 && stage.GetAlphaMultiplier() == 1); -} - RasterizerOpenGL::RasterizerOpenGL() : shader_dirty(true) { // Create sampler objects for (size_t i = 0; i < texture_samplers.size(); ++i) { @@ -181,7 +173,7 @@ void RasterizerOpenGL::DrawTriangles() { CachedSurface* depth_surface; MathUtil::Rectangle<int> rect; std::tie(color_surface, depth_surface, rect) = - res_cache.GetFramebufferSurfaces(regs.framebuffer); + res_cache.GetFramebufferSurfaces(regs.framebuffer.framebuffer); state.draw.draw_framebuffer = framebuffer.handle; state.Apply(); @@ -190,20 +182,24 @@ void RasterizerOpenGL::DrawTriangles() { color_surface != nullptr ? color_surface->texture.handle : 0, 0); glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_ATTACHMENT, GL_TEXTURE_2D, depth_surface != nullptr ? depth_surface->texture.handle : 0, 0); - bool has_stencil = regs.framebuffer.depth_format == Pica::Regs::DepthFormat::D24S8; + bool has_stencil = + regs.framebuffer.framebuffer.depth_format == Pica::FramebufferRegs::DepthFormat::D24S8; glFramebufferTexture2D( GL_DRAW_FRAMEBUFFER, GL_STENCIL_ATTACHMENT, GL_TEXTURE_2D, (has_stencil && depth_surface != nullptr) ? depth_surface->texture.handle : 0, 0); // Sync the viewport // These registers hold half-width and half-height, so must be multiplied by 2 - GLsizei viewport_width = (GLsizei)Pica::float24::FromRaw(regs.viewport_size_x).ToFloat32() * 2; - GLsizei viewport_height = (GLsizei)Pica::float24::FromRaw(regs.viewport_size_y).ToFloat32() * 2; + GLsizei viewport_width = + (GLsizei)Pica::float24::FromRaw(regs.rasterizer.viewport_size_x).ToFloat32() * 2; + GLsizei viewport_height = + (GLsizei)Pica::float24::FromRaw(regs.rasterizer.viewport_size_y).ToFloat32() * 2; - glViewport((GLint)(rect.left + regs.viewport_corner.x * color_surface->res_scale_width), - (GLint)(rect.bottom + regs.viewport_corner.y * color_surface->res_scale_height), - (GLsizei)(viewport_width * color_surface->res_scale_width), - (GLsizei)(viewport_height * color_surface->res_scale_height)); + glViewport( + (GLint)(rect.left + regs.rasterizer.viewport_corner.x * color_surface->res_scale_width), + (GLint)(rect.bottom + regs.rasterizer.viewport_corner.y * color_surface->res_scale_height), + (GLsizei)(viewport_width * color_surface->res_scale_width), + (GLsizei)(viewport_height * color_surface->res_scale_height)); if (uniform_block_data.data.framebuffer_scale[0] != color_surface->res_scale_width || uniform_block_data.data.framebuffer_scale[1] != color_surface->res_scale_height) { @@ -215,16 +211,16 @@ void RasterizerOpenGL::DrawTriangles() { // 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 = - static_cast<GLint>(rect.left + regs.scissor_test.x1 * color_surface->res_scale_width); - GLint scissor_y1 = - static_cast<GLint>(rect.bottom + regs.scissor_test.y1 * color_surface->res_scale_height); + GLint scissor_x1 = static_cast<GLint>( + rect.left + regs.rasterizer.scissor_test.x1 * color_surface->res_scale_width); + GLint scissor_y1 = static_cast<GLint>( + rect.bottom + regs.rasterizer.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 = - static_cast<GLint>(rect.left + (regs.scissor_test.x2 + 1) * color_surface->res_scale_width); + GLint scissor_x2 = static_cast<GLint>( + rect.left + (regs.rasterizer.scissor_test.x2 + 1) * color_surface->res_scale_width); GLint scissor_y2 = static_cast<GLint>( - rect.bottom + (regs.scissor_test.y2 + 1) * color_surface->res_scale_height); + rect.bottom + (regs.rasterizer.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 || @@ -239,7 +235,7 @@ void RasterizerOpenGL::DrawTriangles() { } // Sync and bind the texture surfaces - const auto pica_textures = regs.GetTextures(); + const auto pica_textures = regs.texturing.GetTextures(); for (unsigned texture_index = 0; texture_index < pica_textures.size(); ++texture_index) { const auto& texture = pica_textures[texture_index]; @@ -316,69 +312,69 @@ void RasterizerOpenGL::NotifyPicaRegisterChanged(u32 id) { switch (id) { // Culling - case PICA_REG_INDEX(cull_mode): + case PICA_REG_INDEX(rasterizer.cull_mode): SyncCullMode(); break; // Depth modifiers - case PICA_REG_INDEX(viewport_depth_range): + case PICA_REG_INDEX(rasterizer.viewport_depth_range): SyncDepthScale(); break; - case PICA_REG_INDEX(viewport_depth_near_plane): + case PICA_REG_INDEX(rasterizer.viewport_depth_near_plane): SyncDepthOffset(); break; // Depth buffering - case PICA_REG_INDEX(depthmap_enable): + case PICA_REG_INDEX(rasterizer.depthmap_enable): shader_dirty = true; break; // Blending - case PICA_REG_INDEX(output_merger.alphablend_enable): + case PICA_REG_INDEX(framebuffer.output_merger.alphablend_enable): SyncBlendEnabled(); break; - case PICA_REG_INDEX(output_merger.alpha_blending): + case PICA_REG_INDEX(framebuffer.output_merger.alpha_blending): SyncBlendFuncs(); break; - case PICA_REG_INDEX(output_merger.blend_const): + case PICA_REG_INDEX(framebuffer.output_merger.blend_const): SyncBlendColor(); break; // Fog state - case PICA_REG_INDEX(fog_color): + case PICA_REG_INDEX(texturing.fog_color): SyncFogColor(); break; - case PICA_REG_INDEX_WORKAROUND(fog_lut_data[0], 0xe8): - case PICA_REG_INDEX_WORKAROUND(fog_lut_data[1], 0xe9): - case PICA_REG_INDEX_WORKAROUND(fog_lut_data[2], 0xea): - case PICA_REG_INDEX_WORKAROUND(fog_lut_data[3], 0xeb): - case PICA_REG_INDEX_WORKAROUND(fog_lut_data[4], 0xec): - case PICA_REG_INDEX_WORKAROUND(fog_lut_data[5], 0xed): - case PICA_REG_INDEX_WORKAROUND(fog_lut_data[6], 0xee): - case PICA_REG_INDEX_WORKAROUND(fog_lut_data[7], 0xef): + case PICA_REG_INDEX_WORKAROUND(texturing.fog_lut_data[0], 0xe8): + case PICA_REG_INDEX_WORKAROUND(texturing.fog_lut_data[1], 0xe9): + case PICA_REG_INDEX_WORKAROUND(texturing.fog_lut_data[2], 0xea): + case PICA_REG_INDEX_WORKAROUND(texturing.fog_lut_data[3], 0xeb): + case PICA_REG_INDEX_WORKAROUND(texturing.fog_lut_data[4], 0xec): + case PICA_REG_INDEX_WORKAROUND(texturing.fog_lut_data[5], 0xed): + case PICA_REG_INDEX_WORKAROUND(texturing.fog_lut_data[6], 0xee): + case PICA_REG_INDEX_WORKAROUND(texturing.fog_lut_data[7], 0xef): uniform_block_data.fog_lut_dirty = true; break; // Alpha test - case PICA_REG_INDEX(output_merger.alpha_test): + case PICA_REG_INDEX(framebuffer.output_merger.alpha_test): SyncAlphaTest(); shader_dirty = true; break; // Sync GL stencil test + stencil write mask // (Pica stencil test function register also contains a stencil write mask) - case PICA_REG_INDEX(output_merger.stencil_test.raw_func): + case PICA_REG_INDEX(framebuffer.output_merger.stencil_test.raw_func): SyncStencilTest(); SyncStencilWriteMask(); break; - case PICA_REG_INDEX(output_merger.stencil_test.raw_op): - case PICA_REG_INDEX(framebuffer.depth_format): + case PICA_REG_INDEX(framebuffer.output_merger.stencil_test.raw_op): + case PICA_REG_INDEX(framebuffer.framebuffer.depth_format): SyncStencilTest(); break; // Sync GL depth test + depth and color write mask // (Pica depth test function register also contains a depth and color write mask) - case PICA_REG_INDEX(output_merger.depth_test_enable): + case PICA_REG_INDEX(framebuffer.output_merger.depth_test_enable): SyncDepthTest(); SyncDepthWriteMask(); SyncColorWriteMask(); @@ -386,88 +382,88 @@ void RasterizerOpenGL::NotifyPicaRegisterChanged(u32 id) { // Sync GL depth and stencil write mask // (This is a dedicated combined depth / stencil write-enable register) - case PICA_REG_INDEX(framebuffer.allow_depth_stencil_write): + case PICA_REG_INDEX(framebuffer.framebuffer.allow_depth_stencil_write): SyncDepthWriteMask(); SyncStencilWriteMask(); break; // Sync GL color write mask // (This is a dedicated color write-enable register) - case PICA_REG_INDEX(framebuffer.allow_color_write): + case PICA_REG_INDEX(framebuffer.framebuffer.allow_color_write): SyncColorWriteMask(); break; // Scissor test - case PICA_REG_INDEX(scissor_test.mode): + case PICA_REG_INDEX(rasterizer.scissor_test.mode): shader_dirty = true; break; // Logic op - case PICA_REG_INDEX(output_merger.logic_op): + case PICA_REG_INDEX(framebuffer.output_merger.logic_op): SyncLogicOp(); break; // Texture 0 type - case PICA_REG_INDEX(texture0.type): + case PICA_REG_INDEX(texturing.texture0.type): shader_dirty = true; break; // TEV stages // (This also syncs fog_mode and fog_flip which are part of tev_combiner_buffer_input) - case PICA_REG_INDEX(tev_stage0.color_source1): - case PICA_REG_INDEX(tev_stage0.color_modifier1): - case PICA_REG_INDEX(tev_stage0.color_op): - case PICA_REG_INDEX(tev_stage0.color_scale): - case PICA_REG_INDEX(tev_stage1.color_source1): - case PICA_REG_INDEX(tev_stage1.color_modifier1): - case PICA_REG_INDEX(tev_stage1.color_op): - case PICA_REG_INDEX(tev_stage1.color_scale): - case PICA_REG_INDEX(tev_stage2.color_source1): - case PICA_REG_INDEX(tev_stage2.color_modifier1): - case PICA_REG_INDEX(tev_stage2.color_op): - case PICA_REG_INDEX(tev_stage2.color_scale): - case PICA_REG_INDEX(tev_stage3.color_source1): - case PICA_REG_INDEX(tev_stage3.color_modifier1): - case PICA_REG_INDEX(tev_stage3.color_op): - case PICA_REG_INDEX(tev_stage3.color_scale): - case PICA_REG_INDEX(tev_stage4.color_source1): - case PICA_REG_INDEX(tev_stage4.color_modifier1): - case PICA_REG_INDEX(tev_stage4.color_op): - case PICA_REG_INDEX(tev_stage4.color_scale): - case PICA_REG_INDEX(tev_stage5.color_source1): - case PICA_REG_INDEX(tev_stage5.color_modifier1): - case PICA_REG_INDEX(tev_stage5.color_op): - case PICA_REG_INDEX(tev_stage5.color_scale): - case PICA_REG_INDEX(tev_combiner_buffer_input): + case PICA_REG_INDEX(texturing.tev_stage0.color_source1): + case PICA_REG_INDEX(texturing.tev_stage0.color_modifier1): + case PICA_REG_INDEX(texturing.tev_stage0.color_op): + case PICA_REG_INDEX(texturing.tev_stage0.color_scale): + case PICA_REG_INDEX(texturing.tev_stage1.color_source1): + case PICA_REG_INDEX(texturing.tev_stage1.color_modifier1): + case PICA_REG_INDEX(texturing.tev_stage1.color_op): + case PICA_REG_INDEX(texturing.tev_stage1.color_scale): + case PICA_REG_INDEX(texturing.tev_stage2.color_source1): + case PICA_REG_INDEX(texturing.tev_stage2.color_modifier1): + case PICA_REG_INDEX(texturing.tev_stage2.color_op): + case PICA_REG_INDEX(texturing.tev_stage2.color_scale): + case PICA_REG_INDEX(texturing.tev_stage3.color_source1): + case PICA_REG_INDEX(texturing.tev_stage3.color_modifier1): + case PICA_REG_INDEX(texturing.tev_stage3.color_op): + case PICA_REG_INDEX(texturing.tev_stage3.color_scale): + case PICA_REG_INDEX(texturing.tev_stage4.color_source1): + case PICA_REG_INDEX(texturing.tev_stage4.color_modifier1): + case PICA_REG_INDEX(texturing.tev_stage4.color_op): + case PICA_REG_INDEX(texturing.tev_stage4.color_scale): + case PICA_REG_INDEX(texturing.tev_stage5.color_source1): + case PICA_REG_INDEX(texturing.tev_stage5.color_modifier1): + case PICA_REG_INDEX(texturing.tev_stage5.color_op): + case PICA_REG_INDEX(texturing.tev_stage5.color_scale): + case PICA_REG_INDEX(texturing.tev_combiner_buffer_input): shader_dirty = true; break; - case PICA_REG_INDEX(tev_stage0.const_r): - SyncTevConstColor(0, regs.tev_stage0); + case PICA_REG_INDEX(texturing.tev_stage0.const_r): + SyncTevConstColor(0, regs.texturing.tev_stage0); break; - case PICA_REG_INDEX(tev_stage1.const_r): - SyncTevConstColor(1, regs.tev_stage1); + case PICA_REG_INDEX(texturing.tev_stage1.const_r): + SyncTevConstColor(1, regs.texturing.tev_stage1); break; - case PICA_REG_INDEX(tev_stage2.const_r): - SyncTevConstColor(2, regs.tev_stage2); + case PICA_REG_INDEX(texturing.tev_stage2.const_r): + SyncTevConstColor(2, regs.texturing.tev_stage2); break; - case PICA_REG_INDEX(tev_stage3.const_r): - SyncTevConstColor(3, regs.tev_stage3); + case PICA_REG_INDEX(texturing.tev_stage3.const_r): + SyncTevConstColor(3, regs.texturing.tev_stage3); break; - case PICA_REG_INDEX(tev_stage4.const_r): - SyncTevConstColor(4, regs.tev_stage4); + case PICA_REG_INDEX(texturing.tev_stage4.const_r): + SyncTevConstColor(4, regs.texturing.tev_stage4); break; - case PICA_REG_INDEX(tev_stage5.const_r): - SyncTevConstColor(5, regs.tev_stage5); + case PICA_REG_INDEX(texturing.tev_stage5.const_r): + SyncTevConstColor(5, regs.texturing.tev_stage5); break; // TEV combiner buffer color - case PICA_REG_INDEX(tev_combiner_buffer_color): + case PICA_REG_INDEX(texturing.tev_combiner_buffer_color): SyncCombinerColor(); break; // Fragment lighting switches case PICA_REG_INDEX(lighting.disable): - case PICA_REG_INDEX(lighting.num_lights): + case PICA_REG_INDEX(lighting.max_light_index): case PICA_REG_INDEX(lighting.config0): case PICA_REG_INDEX(lighting.config1): case PICA_REG_INDEX(lighting.abs_lut_input): @@ -976,7 +972,9 @@ void RasterizerOpenGL::SamplerInfo::Create() { // Other attributes have correct defaults } -void RasterizerOpenGL::SamplerInfo::SyncWithConfig(const Pica::Regs::TextureConfig& config) { +void RasterizerOpenGL::SamplerInfo::SyncWithConfig( + const Pica::TexturingRegs::TextureConfig& config) { + GLuint s = sampler.handle; if (mag_filter != config.mag_filter) { @@ -1073,67 +1071,69 @@ void RasterizerOpenGL::SetShader() { current_shader = shader_cache.emplace(config, std::move(shader)).first->second.get(); - unsigned int block_index = - glGetUniformBlockIndex(current_shader->shader.handle, "shader_data"); - GLint block_size; - glGetActiveUniformBlockiv(current_shader->shader.handle, block_index, - GL_UNIFORM_BLOCK_DATA_SIZE, &block_size); - ASSERT_MSG(block_size == sizeof(UniformData), - "Uniform block size did not match! Got %d, expected %zu", - static_cast<int>(block_size), sizeof(UniformData)); - glUniformBlockBinding(current_shader->shader.handle, block_index, 0); - - // Update uniforms - SyncDepthScale(); - SyncDepthOffset(); - SyncAlphaTest(); - SyncCombinerColor(); - auto& tev_stages = Pica::g_state.regs.GetTevStages(); - for (int index = 0; index < tev_stages.size(); ++index) - SyncTevConstColor(index, tev_stages[index]); + GLuint block_index = glGetUniformBlockIndex(current_shader->shader.handle, "shader_data"); + if (block_index != GL_INVALID_INDEX) { + GLint block_size; + glGetActiveUniformBlockiv(current_shader->shader.handle, block_index, + GL_UNIFORM_BLOCK_DATA_SIZE, &block_size); + ASSERT_MSG(block_size == sizeof(UniformData), + "Uniform block size did not match! Got %d, expected %zu", + static_cast<int>(block_size), sizeof(UniformData)); + glUniformBlockBinding(current_shader->shader.handle, block_index, 0); + + // Update uniforms + SyncDepthScale(); + SyncDepthOffset(); + SyncAlphaTest(); + SyncCombinerColor(); + auto& tev_stages = Pica::g_state.regs.texturing.GetTevStages(); + for (int index = 0; index < tev_stages.size(); ++index) + SyncTevConstColor(index, tev_stages[index]); + + SyncGlobalAmbient(); + for (int light_index = 0; light_index < 8; light_index++) { + SyncLightSpecular0(light_index); + SyncLightSpecular1(light_index); + SyncLightDiffuse(light_index); + SyncLightAmbient(light_index); + SyncLightPosition(light_index); + SyncLightDistanceAttenuationBias(light_index); + SyncLightDistanceAttenuationScale(light_index); + } - SyncGlobalAmbient(); - for (int light_index = 0; light_index < 8; light_index++) { - SyncLightSpecular0(light_index); - SyncLightSpecular1(light_index); - SyncLightDiffuse(light_index); - SyncLightAmbient(light_index); - SyncLightPosition(light_index); - SyncLightDistanceAttenuationBias(light_index); - SyncLightDistanceAttenuationScale(light_index); + SyncFogColor(); } - - SyncFogColor(); } } void RasterizerOpenGL::SyncCullMode() { const auto& regs = Pica::g_state.regs; - switch (regs.cull_mode) { - case Pica::Regs::CullMode::KeepAll: + switch (regs.rasterizer.cull_mode) { + case Pica::RasterizerRegs::CullMode::KeepAll: state.cull.enabled = false; break; - case Pica::Regs::CullMode::KeepClockWise: + case Pica::RasterizerRegs::CullMode::KeepClockWise: state.cull.enabled = true; state.cull.front_face = GL_CW; break; - case Pica::Regs::CullMode::KeepCounterClockWise: + case Pica::RasterizerRegs::CullMode::KeepCounterClockWise: state.cull.enabled = true; state.cull.front_face = GL_CCW; break; default: - LOG_CRITICAL(Render_OpenGL, "Unknown cull mode %d", regs.cull_mode.Value()); + LOG_CRITICAL(Render_OpenGL, "Unknown cull mode %d", regs.rasterizer.cull_mode.Value()); UNIMPLEMENTED(); break; } } void RasterizerOpenGL::SyncDepthScale() { - float depth_scale = Pica::float24::FromRaw(Pica::g_state.regs.viewport_depth_range).ToFloat32(); + float depth_scale = + Pica::float24::FromRaw(Pica::g_state.regs.rasterizer.viewport_depth_range).ToFloat32(); if (depth_scale != uniform_block_data.data.depth_scale) { uniform_block_data.data.depth_scale = depth_scale; uniform_block_data.dirty = true; @@ -1142,7 +1142,7 @@ void RasterizerOpenGL::SyncDepthScale() { void RasterizerOpenGL::SyncDepthOffset() { float depth_offset = - Pica::float24::FromRaw(Pica::g_state.regs.viewport_depth_near_plane).ToFloat32(); + Pica::float24::FromRaw(Pica::g_state.regs.rasterizer.viewport_depth_near_plane).ToFloat32(); if (depth_offset != uniform_block_data.data.depth_offset) { uniform_block_data.data.depth_offset = depth_offset; uniform_block_data.dirty = true; @@ -1150,25 +1150,28 @@ void RasterizerOpenGL::SyncDepthOffset() { } void RasterizerOpenGL::SyncBlendEnabled() { - state.blend.enabled = (Pica::g_state.regs.output_merger.alphablend_enable == 1); + state.blend.enabled = (Pica::g_state.regs.framebuffer.output_merger.alphablend_enable == 1); } void RasterizerOpenGL::SyncBlendFuncs() { const auto& regs = Pica::g_state.regs; state.blend.rgb_equation = - PicaToGL::BlendEquation(regs.output_merger.alpha_blending.blend_equation_rgb); + PicaToGL::BlendEquation(regs.framebuffer.output_merger.alpha_blending.blend_equation_rgb); state.blend.a_equation = - PicaToGL::BlendEquation(regs.output_merger.alpha_blending.blend_equation_a); + PicaToGL::BlendEquation(regs.framebuffer.output_merger.alpha_blending.blend_equation_a); state.blend.src_rgb_func = - PicaToGL::BlendFunc(regs.output_merger.alpha_blending.factor_source_rgb); + PicaToGL::BlendFunc(regs.framebuffer.output_merger.alpha_blending.factor_source_rgb); state.blend.dst_rgb_func = - PicaToGL::BlendFunc(regs.output_merger.alpha_blending.factor_dest_rgb); - state.blend.src_a_func = PicaToGL::BlendFunc(regs.output_merger.alpha_blending.factor_source_a); - state.blend.dst_a_func = PicaToGL::BlendFunc(regs.output_merger.alpha_blending.factor_dest_a); + PicaToGL::BlendFunc(regs.framebuffer.output_merger.alpha_blending.factor_dest_rgb); + state.blend.src_a_func = + PicaToGL::BlendFunc(regs.framebuffer.output_merger.alpha_blending.factor_source_a); + state.blend.dst_a_func = + PicaToGL::BlendFunc(regs.framebuffer.output_merger.alpha_blending.factor_dest_a); } void RasterizerOpenGL::SyncBlendColor() { - auto blend_color = PicaToGL::ColorRGBA8(Pica::g_state.regs.output_merger.blend_const.raw); + auto blend_color = + PicaToGL::ColorRGBA8(Pica::g_state.regs.framebuffer.output_merger.blend_const.raw); state.blend.color.red = blend_color[0]; state.blend.color.green = blend_color[1]; state.blend.color.blue = blend_color[2]; @@ -1178,8 +1181,8 @@ void RasterizerOpenGL::SyncBlendColor() { void RasterizerOpenGL::SyncFogColor() { const auto& regs = Pica::g_state.regs; uniform_block_data.data.fog_color = { - regs.fog_color.r.Value() / 255.0f, regs.fog_color.g.Value() / 255.0f, - regs.fog_color.b.Value() / 255.0f, + regs.texturing.fog_color.r.Value() / 255.0f, regs.texturing.fog_color.g.Value() / 255.0f, + regs.texturing.fog_color.b.Value() / 255.0f, }; uniform_block_data.dirty = true; } @@ -1200,70 +1203,78 @@ void RasterizerOpenGL::SyncFogLUT() { void RasterizerOpenGL::SyncAlphaTest() { const auto& regs = Pica::g_state.regs; - if (regs.output_merger.alpha_test.ref != uniform_block_data.data.alphatest_ref) { - uniform_block_data.data.alphatest_ref = regs.output_merger.alpha_test.ref; + if (regs.framebuffer.output_merger.alpha_test.ref != uniform_block_data.data.alphatest_ref) { + uniform_block_data.data.alphatest_ref = regs.framebuffer.output_merger.alpha_test.ref; uniform_block_data.dirty = true; } } void RasterizerOpenGL::SyncLogicOp() { - state.logic_op = PicaToGL::LogicOp(Pica::g_state.regs.output_merger.logic_op); + state.logic_op = PicaToGL::LogicOp(Pica::g_state.regs.framebuffer.output_merger.logic_op); } void RasterizerOpenGL::SyncColorWriteMask() { const auto& regs = Pica::g_state.regs; auto IsColorWriteEnabled = [&](u32 value) { - return (regs.framebuffer.allow_color_write != 0 && value != 0) ? GL_TRUE : GL_FALSE; + return (regs.framebuffer.framebuffer.allow_color_write != 0 && value != 0) ? GL_TRUE + : GL_FALSE; }; - state.color_mask.red_enabled = IsColorWriteEnabled(regs.output_merger.red_enable); - state.color_mask.green_enabled = IsColorWriteEnabled(regs.output_merger.green_enable); - state.color_mask.blue_enabled = IsColorWriteEnabled(regs.output_merger.blue_enable); - state.color_mask.alpha_enabled = IsColorWriteEnabled(regs.output_merger.alpha_enable); + state.color_mask.red_enabled = IsColorWriteEnabled(regs.framebuffer.output_merger.red_enable); + state.color_mask.green_enabled = + IsColorWriteEnabled(regs.framebuffer.output_merger.green_enable); + state.color_mask.blue_enabled = IsColorWriteEnabled(regs.framebuffer.output_merger.blue_enable); + state.color_mask.alpha_enabled = + IsColorWriteEnabled(regs.framebuffer.output_merger.alpha_enable); } void RasterizerOpenGL::SyncStencilWriteMask() { const auto& regs = Pica::g_state.regs; - state.stencil.write_mask = (regs.framebuffer.allow_depth_stencil_write != 0) - ? static_cast<GLuint>(regs.output_merger.stencil_test.write_mask) - : 0; + state.stencil.write_mask = + (regs.framebuffer.framebuffer.allow_depth_stencil_write != 0) + ? static_cast<GLuint>(regs.framebuffer.output_merger.stencil_test.write_mask) + : 0; } void RasterizerOpenGL::SyncDepthWriteMask() { const auto& regs = Pica::g_state.regs; - state.depth.write_mask = - (regs.framebuffer.allow_depth_stencil_write != 0 && regs.output_merger.depth_write_enable) - ? GL_TRUE - : GL_FALSE; + state.depth.write_mask = (regs.framebuffer.framebuffer.allow_depth_stencil_write != 0 && + regs.framebuffer.output_merger.depth_write_enable) + ? GL_TRUE + : GL_FALSE; } void RasterizerOpenGL::SyncStencilTest() { const auto& regs = Pica::g_state.regs; - state.stencil.test_enabled = regs.output_merger.stencil_test.enable && - regs.framebuffer.depth_format == Pica::Regs::DepthFormat::D24S8; - state.stencil.test_func = PicaToGL::CompareFunc(regs.output_merger.stencil_test.func); - state.stencil.test_ref = regs.output_merger.stencil_test.reference_value; - state.stencil.test_mask = regs.output_merger.stencil_test.input_mask; + state.stencil.test_enabled = + regs.framebuffer.output_merger.stencil_test.enable && + regs.framebuffer.framebuffer.depth_format == Pica::FramebufferRegs::DepthFormat::D24S8; + state.stencil.test_func = + PicaToGL::CompareFunc(regs.framebuffer.output_merger.stencil_test.func); + state.stencil.test_ref = regs.framebuffer.output_merger.stencil_test.reference_value; + state.stencil.test_mask = regs.framebuffer.output_merger.stencil_test.input_mask; state.stencil.action_stencil_fail = - PicaToGL::StencilOp(regs.output_merger.stencil_test.action_stencil_fail); + PicaToGL::StencilOp(regs.framebuffer.output_merger.stencil_test.action_stencil_fail); state.stencil.action_depth_fail = - PicaToGL::StencilOp(regs.output_merger.stencil_test.action_depth_fail); + PicaToGL::StencilOp(regs.framebuffer.output_merger.stencil_test.action_depth_fail); state.stencil.action_depth_pass = - PicaToGL::StencilOp(regs.output_merger.stencil_test.action_depth_pass); + PicaToGL::StencilOp(regs.framebuffer.output_merger.stencil_test.action_depth_pass); } void RasterizerOpenGL::SyncDepthTest() { const auto& regs = Pica::g_state.regs; - state.depth.test_enabled = - regs.output_merger.depth_test_enable == 1 || regs.output_merger.depth_write_enable == 1; - state.depth.test_func = regs.output_merger.depth_test_enable == 1 - ? PicaToGL::CompareFunc(regs.output_merger.depth_test_func) - : GL_ALWAYS; + state.depth.test_enabled = regs.framebuffer.output_merger.depth_test_enable == 1 || + regs.framebuffer.output_merger.depth_write_enable == 1; + state.depth.test_func = + regs.framebuffer.output_merger.depth_test_enable == 1 + ? PicaToGL::CompareFunc(regs.framebuffer.output_merger.depth_test_func) + : GL_ALWAYS; } void RasterizerOpenGL::SyncCombinerColor() { - auto combiner_color = PicaToGL::ColorRGBA8(Pica::g_state.regs.tev_combiner_buffer_color.raw); + auto combiner_color = + PicaToGL::ColorRGBA8(Pica::g_state.regs.texturing.tev_combiner_buffer_color.raw); if (combiner_color != uniform_block_data.data.tev_combiner_buffer_color) { uniform_block_data.data.tev_combiner_buffer_color = combiner_color; uniform_block_data.dirty = true; @@ -1271,7 +1282,7 @@ void RasterizerOpenGL::SyncCombinerColor() { } void RasterizerOpenGL::SyncTevConstColor(int stage_index, - const Pica::Regs::TevStageConfig& tev_stage) { + const Pica::TexturingRegs::TevStageConfig& tev_stage) { auto const_color = PicaToGL::ColorRGBA8(tev_stage.const_color); if (const_color != uniform_block_data.data.const_color[stage_index]) { uniform_block_data.data.const_color[stage_index] = const_color; diff --git a/src/video_core/renderer_opengl/gl_rasterizer.h b/src/video_core/renderer_opengl/gl_rasterizer.h index cc3e4bed5..ecf737438 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer.h +++ b/src/video_core/renderer_opengl/gl_rasterizer.h @@ -16,10 +16,13 @@ #include "common/hash.h" #include "common/vector_math.h" #include "core/hw/gpu.h" -#include "video_core/pica.h" #include "video_core/pica_state.h" #include "video_core/pica_types.h" #include "video_core/rasterizer_interface.h" +#include "video_core/regs_framebuffer.h" +#include "video_core/regs_lighting.h" +#include "video_core/regs_rasterizer.h" +#include "video_core/regs_texturing.h" #include "video_core/renderer_opengl/gl_rasterizer_cache.h" #include "video_core/renderer_opengl/gl_resource_manager.h" #include "video_core/renderer_opengl/gl_state.h" @@ -52,20 +55,20 @@ union PicaShaderConfig { const auto& regs = Pica::g_state.regs; - state.scissor_test_mode = regs.scissor_test.mode; + state.scissor_test_mode = regs.rasterizer.scissor_test.mode; - state.depthmap_enable = regs.depthmap_enable; + state.depthmap_enable = regs.rasterizer.depthmap_enable; - state.alpha_test_func = regs.output_merger.alpha_test.enable - ? regs.output_merger.alpha_test.func.Value() - : Pica::Regs::CompareFunc::Always; + state.alpha_test_func = regs.framebuffer.output_merger.alpha_test.enable + ? regs.framebuffer.output_merger.alpha_test.func.Value() + : Pica::FramebufferRegs::CompareFunc::Always; - state.texture0_type = regs.texture0.type; + state.texture0_type = regs.texturing.texture0.type; // Copy relevant tev stages fields. // We don't sync const_color here because of the high variance, it is a // shader uniform instead. - const auto& tev_stages = regs.GetTevStages(); + const auto& tev_stages = regs.texturing.GetTevStages(); DEBUG_ASSERT(state.tev_stages.size() == tev_stages.size()); for (size_t i = 0; i < tev_stages.size(); i++) { const auto& tev_stage = tev_stages[i]; @@ -75,16 +78,17 @@ union PicaShaderConfig { state.tev_stages[i].scales_raw = tev_stage.scales_raw; } - state.fog_mode = regs.fog_mode; - state.fog_flip = regs.fog_flip != 0; + state.fog_mode = regs.texturing.fog_mode; + state.fog_flip = regs.texturing.fog_flip != 0; - state.combiner_buffer_input = regs.tev_combiner_buffer_input.update_mask_rgb.Value() | - regs.tev_combiner_buffer_input.update_mask_a.Value() << 4; + state.combiner_buffer_input = + regs.texturing.tev_combiner_buffer_input.update_mask_rgb.Value() | + regs.texturing.tev_combiner_buffer_input.update_mask_a.Value() << 4; // Fragment lighting state.lighting.enable = !regs.lighting.disable; - state.lighting.src_num = regs.lighting.num_lights + 1; + state.lighting.src_num = regs.lighting.max_light_index + 1; for (unsigned light_index = 0; light_index < state.lighting.src_num; ++light_index) { unsigned num = regs.lighting.light_enable.GetNum(light_index); @@ -159,8 +163,8 @@ union PicaShaderConfig { u32 modifiers_raw; u32 ops_raw; u32 scales_raw; - explicit operator Pica::Regs::TevStageConfig() const noexcept { - Pica::Regs::TevStageConfig stage; + explicit operator Pica::TexturingRegs::TevStageConfig() const noexcept { + Pica::TexturingRegs::TevStageConfig stage; stage.sources_raw = sources_raw; stage.modifiers_raw = modifiers_raw; stage.ops_raw = ops_raw; @@ -171,14 +175,14 @@ union PicaShaderConfig { }; struct State { - Pica::Regs::CompareFunc alpha_test_func; - Pica::Regs::ScissorMode scissor_test_mode; - Pica::Regs::TextureConfig::TextureType texture0_type; + Pica::FramebufferRegs::CompareFunc alpha_test_func; + Pica::RasterizerRegs::ScissorMode scissor_test_mode; + Pica::TexturingRegs::TextureConfig::TextureType texture0_type; std::array<TevStageConfigRaw, 6> tev_stages; u8 combiner_buffer_input; - Pica::Regs::DepthBuffering depthmap_enable; - Pica::Regs::FogMode fog_mode; + Pica::RasterizerRegs::DepthBuffering depthmap_enable; + Pica::TexturingRegs::FogMode fog_mode; bool fog_flip; struct { @@ -191,18 +195,18 @@ union PicaShaderConfig { bool enable; unsigned src_num; - Pica::Regs::LightingBumpMode bump_mode; + Pica::LightingRegs::LightingBumpMode bump_mode; unsigned bump_selector; bool bump_renorm; bool clamp_highlights; - Pica::Regs::LightingConfig config; - Pica::Regs::LightingFresnelSelector fresnel_selector; + Pica::LightingRegs::LightingConfig config; + Pica::LightingRegs::LightingFresnelSelector fresnel_selector; struct { bool enable; bool abs_input; - Pica::Regs::LightingLutInput type; + Pica::LightingRegs::LightingLutInput type; float scale; } lut_d0, lut_d1, lut_fr, lut_rr, lut_rg, lut_rb; } lighting; @@ -251,7 +255,7 @@ public: private: struct SamplerInfo { - using TextureConfig = Pica::Regs::TextureConfig; + using TextureConfig = Pica::TexturingRegs::TextureConfig; OGLSampler sampler; @@ -398,7 +402,7 @@ private: void SyncCombinerColor(); /// Syncs the TEV constant color to match the PICA register - void SyncTevConstColor(int tev_index, const Pica::Regs::TevStageConfig& tev_stage); + void SyncTevConstColor(int tev_index, const Pica::TexturingRegs::TevStageConfig& tev_stage); /// Syncs the lighting global ambient color to match the PICA register void SyncGlobalAmbient(); diff --git a/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp b/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp index 1e7eedecb..456443e86 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp +++ b/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp @@ -17,10 +17,11 @@ #include "common/vector_math.h" #include "core/frontend/emu_window.h" #include "core/memory.h" -#include "video_core/debug_utils/debug_utils.h" +#include "core/settings.h" #include "video_core/pica_state.h" #include "video_core/renderer_opengl/gl_rasterizer_cache.h" #include "video_core/renderer_opengl/gl_state.h" +#include "video_core/texture/texture_decode.h" #include "video_core/utils.h" #include "video_core/video_core.h" @@ -339,17 +340,16 @@ CachedSurface* RasterizerCacheOpenGL::GetSurface(const CachedSurface& params, bo std::vector<Math::Vec4<u8>> tex_buffer(params.width * params.height); - Pica::DebugUtils::TextureInfo tex_info; + Pica::Texture::TextureInfo tex_info; tex_info.width = params.width; tex_info.height = params.height; - tex_info.stride = - params.width * CachedSurface::GetFormatBpp(params.pixel_format) / 8; - tex_info.format = (Pica::Regs::TextureFormat)params.pixel_format; + tex_info.format = (Pica::TexturingRegs::TextureFormat)params.pixel_format; + tex_info.SetDefaultStride(); tex_info.physical_address = params.addr; for (unsigned y = 0; y < params.height; ++y) { for (unsigned x = 0; x < params.width; ++x) { - tex_buffer[x + params.width * y] = Pica::DebugUtils::LookupTexture( + tex_buffer[x + params.width * y] = Pica::Texture::LookupTexture( texture_src_data, x, params.height - 1 - y, tex_info); } } @@ -511,9 +511,10 @@ CachedSurface* RasterizerCacheOpenGL::GetSurfaceRect(const CachedSurface& params } CachedSurface* RasterizerCacheOpenGL::GetTextureSurface( - const Pica::Regs::FullTextureConfig& config) { - Pica::DebugUtils::TextureInfo info = - Pica::DebugUtils::TextureInfo::FromPicaRegister(config.config, config.format); + const Pica::TexturingRegs::FullTextureConfig& config) { + + Pica::Texture::TextureInfo info = + Pica::Texture::TextureInfo::FromPicaRegister(config.config, config.format); CachedSurface params; params.addr = info.physical_address; @@ -525,7 +526,9 @@ CachedSurface* RasterizerCacheOpenGL::GetTextureSurface( } std::tuple<CachedSurface*, CachedSurface*, MathUtil::Rectangle<int>> -RasterizerCacheOpenGL::GetFramebufferSurfaces(const Pica::Regs::FramebufferConfig& config) { +RasterizerCacheOpenGL::GetFramebufferSurfaces( + const Pica::FramebufferRegs::FramebufferConfig& config) { + const auto& regs = Pica::g_state.regs; // Make sur that framebuffers don't overlap if both color and depth are being used @@ -537,11 +540,12 @@ RasterizerCacheOpenGL::GetFramebufferSurfaces(const Pica::Regs::FramebufferConfi config.GetColorBufferPhysicalAddress(), fb_area * GPU::Regs::BytesPerPixel(GPU::Regs::PixelFormat(config.color_format.Value())), config.GetDepthBufferPhysicalAddress(), - fb_area * Pica::Regs::BytesPerDepthPixel(config.depth_format)); + fb_area * Pica::FramebufferRegs::BytesPerDepthPixel(config.depth_format)); bool using_color_fb = config.GetColorBufferPhysicalAddress() != 0; - bool using_depth_fb = config.GetDepthBufferPhysicalAddress() != 0 && - (regs.output_merger.depth_test_enable || - regs.output_merger.depth_write_enable || !framebuffers_overlap); + bool using_depth_fb = + config.GetDepthBufferPhysicalAddress() != 0 && + (regs.framebuffer.output_merger.depth_test_enable || + regs.framebuffer.output_merger.depth_write_enable || !framebuffers_overlap); if (framebuffers_overlap && using_color_fb && using_depth_fb) { LOG_CRITICAL(Render_OpenGL, "Color and depth framebuffer memory regions overlap; " diff --git a/src/video_core/renderer_opengl/gl_rasterizer_cache.h b/src/video_core/renderer_opengl/gl_rasterizer_cache.h index f57fdb3cc..aea20c693 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer_cache.h +++ b/src/video_core/renderer_opengl/gl_rasterizer_cache.h @@ -21,7 +21,8 @@ #include "common/common_funcs.h" #include "common/common_types.h" #include "core/hw/gpu.h" -#include "video_core/pica.h" +#include "video_core/regs_framebuffer.h" +#include "video_core/regs_texturing.h" #include "video_core/renderer_opengl/gl_resource_manager.h" namespace MathUtil { @@ -96,15 +97,15 @@ struct CachedSurface { return bpp_table[(unsigned int)format]; } - static PixelFormat PixelFormatFromTextureFormat(Pica::Regs::TextureFormat format) { + static PixelFormat PixelFormatFromTextureFormat(Pica::TexturingRegs::TextureFormat format) { return ((unsigned int)format < 14) ? (PixelFormat)format : PixelFormat::Invalid; } - static PixelFormat PixelFormatFromColorFormat(Pica::Regs::ColorFormat format) { + static PixelFormat PixelFormatFromColorFormat(Pica::FramebufferRegs::ColorFormat format) { return ((unsigned int)format < 5) ? (PixelFormat)format : PixelFormat::Invalid; } - static PixelFormat PixelFormatFromDepthFormat(Pica::Regs::DepthFormat format) { + static PixelFormat PixelFormatFromDepthFormat(Pica::FramebufferRegs::DepthFormat format) { return ((unsigned int)format < 4) ? (PixelFormat)((unsigned int)format + 14) : PixelFormat::Invalid; } @@ -212,12 +213,12 @@ public: bool load_if_create, MathUtil::Rectangle<int>& out_rect); /// Gets a surface based on the texture configuration - CachedSurface* GetTextureSurface(const Pica::Regs::FullTextureConfig& config); + CachedSurface* GetTextureSurface(const Pica::TexturingRegs::FullTextureConfig& config); /// Gets the color and depth surfaces and rect (resolution scaled) based on the framebuffer /// configuration std::tuple<CachedSurface*, CachedSurface*, MathUtil::Rectangle<int>> GetFramebufferSurfaces( - const Pica::Regs::FramebufferConfig& config); + const Pica::FramebufferRegs::FramebufferConfig& config); /// Attempt to get a surface that exactly matches the fill region and format CachedSurface* TryGetFillSurface(const GPU::Regs::MemoryFillConfig& config); diff --git a/src/video_core/renderer_opengl/gl_shader_gen.cpp b/src/video_core/renderer_opengl/gl_shader_gen.cpp index 4c4f98ac9..7abdeba05 100644 --- a/src/video_core/renderer_opengl/gl_shader_gen.cpp +++ b/src/video_core/renderer_opengl/gl_shader_gen.cpp @@ -7,13 +7,19 @@ #include "common/assert.h" #include "common/bit_field.h" #include "common/logging/log.h" -#include "video_core/pica.h" +#include "video_core/regs_framebuffer.h" +#include "video_core/regs_lighting.h" +#include "video_core/regs_rasterizer.h" +#include "video_core/regs_texturing.h" #include "video_core/renderer_opengl/gl_rasterizer.h" #include "video_core/renderer_opengl/gl_shader_gen.h" #include "video_core/renderer_opengl/gl_shader_util.h" -using Pica::Regs; -using TevStageConfig = Regs::TevStageConfig; +using Pica::FramebufferRegs; +using Pica::LightingRegs; +using Pica::RasterizerRegs; +using Pica::TexturingRegs; +using TevStageConfig = TexturingRegs::TevStageConfig; namespace GLShader { @@ -46,10 +52,10 @@ static void AppendSource(std::string& out, const PicaShaderConfig& config, case Source::Texture0: // Only unit 0 respects the texturing type (according to 3DBrew) switch (state.texture0_type) { - case Pica::Regs::TextureConfig::Texture2D: + case TexturingRegs::TextureConfig::Texture2D: out += "texture(tex[0], texcoord[0])"; break; - case Pica::Regs::TextureConfig::Projection2D: + case TexturingRegs::TextureConfig::Projection2D: out += "textureProj(tex[0], vec3(texcoord[0], texcoord0_w))"; break; default: @@ -276,8 +282,8 @@ static void AppendAlphaCombiner(std::string& out, TevStageConfig::Operation oper } /// Writes the if-statement condition used to evaluate alpha testing -static void AppendAlphaTestCondition(std::string& out, Regs::CompareFunc func) { - using CompareFunc = Regs::CompareFunc; +static void AppendAlphaTestCondition(std::string& out, FramebufferRegs::CompareFunc func) { + using CompareFunc = FramebufferRegs::CompareFunc; switch (func) { case CompareFunc::Never: out += "true"; @@ -307,7 +313,7 @@ static void AppendAlphaTestCondition(std::string& out, Regs::CompareFunc func) { /// Writes the code to emulate the specified TEV stage static void WriteTevStage(std::string& out, const PicaShaderConfig& config, unsigned index) { const auto stage = - static_cast<const Pica::Regs::TevStageConfig>(config.state.tev_stages[index]); + static_cast<const TexturingRegs::TevStageConfig>(config.state.tev_stages[index]); if (!IsPassThroughTevStage(stage)) { std::string index_name = std::to_string(index); @@ -364,7 +370,7 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) { "vec3 refl_value = vec3(0.0);\n"; // Compute fragment normals - if (lighting.bump_mode == Pica::Regs::LightingBumpMode::NormalMap) { + if (lighting.bump_mode == LightingRegs::LightingBumpMode::NormalMap) { // Bump mapping is enabled using a normal map, read perturbation vector from the selected // texture std::string bump_selector = std::to_string(lighting.bump_selector); @@ -378,7 +384,7 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) { "(1.0 - (surface_normal.x*surface_normal.x + surface_normal.y*surface_normal.y))"; out += "surface_normal.z = sqrt(max(" + val + ", 0.0));\n"; } - } else if (lighting.bump_mode == Pica::Regs::LightingBumpMode::TangentMap) { + } else if (lighting.bump_mode == LightingRegs::LightingBumpMode::TangentMap) { // Bump mapping is enabled using a tangent map LOG_CRITICAL(HW_GPU, "unimplemented bump mapping mode (tangent mapping)"); UNIMPLEMENTED(); @@ -392,23 +398,24 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) { out += "vec3 normal = normalize(quaternion_rotate(normquat, surface_normal));\n"; // Gets the index into the specified lookup table for specular lighting - auto GetLutIndex = [&lighting](unsigned light_num, Regs::LightingLutInput input, bool abs) { + auto GetLutIndex = [&lighting](unsigned light_num, LightingRegs::LightingLutInput input, + bool abs) { const std::string half_angle = "normalize(normalize(view) + light_vector)"; std::string index; switch (input) { - case Regs::LightingLutInput::NH: + case LightingRegs::LightingLutInput::NH: index = "dot(normal, " + half_angle + ")"; break; - case Regs::LightingLutInput::VH: + case LightingRegs::LightingLutInput::VH: index = std::string("dot(normalize(view), " + half_angle + ")"); break; - case Regs::LightingLutInput::NV: + case LightingRegs::LightingLutInput::NV: index = std::string("dot(normal, normalize(view))"); break; - case Regs::LightingLutInput::LN: + case LightingRegs::LightingLutInput::LN: index = std::string("dot(light_vector, normal)"); break; @@ -432,7 +439,7 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) { }; // Gets the lighting lookup table value given the specified sampler and index - auto GetLutValue = [](Regs::LightingSampler sampler, std::string lut_index) { + auto GetLutValue = [](LightingRegs::LightingSampler sampler, std::string lut_index) { return std::string("texture(lut[" + std::to_string((unsigned)sampler / 4) + "], " + lut_index + ")[" + std::to_string((unsigned)sampler & 3) + "]"); }; @@ -461,8 +468,8 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) { light_src + ".position) + " + light_src + ".dist_atten_bias)"; index = "(OFFSET_256 + SCALE_256 * clamp(" + index + ", 0.0, 1.0))"; const unsigned lut_num = - ((unsigned)Regs::LightingSampler::DistanceAttenuation + light_config.num); - dist_atten = GetLutValue((Regs::LightingSampler)lut_num, index); + ((unsigned)LightingRegs::LightingSampler::DistanceAttenuation + light_config.num); + dist_atten = GetLutValue((LightingRegs::LightingSampler)lut_num, index); } // If enabled, clamp specular component if lighting result is negative @@ -472,24 +479,24 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) { // Specular 0 component std::string d0_lut_value = "1.0"; if (lighting.lut_d0.enable && - Pica::Regs::IsLightingSamplerSupported(lighting.config, - Pica::Regs::LightingSampler::Distribution0)) { + LightingRegs::IsLightingSamplerSupported( + lighting.config, LightingRegs::LightingSampler::Distribution0)) { // Lookup specular "distribution 0" LUT value std::string index = GetLutIndex(light_config.num, lighting.lut_d0.type, lighting.lut_d0.abs_input); d0_lut_value = "(" + std::to_string(lighting.lut_d0.scale) + " * " + - GetLutValue(Regs::LightingSampler::Distribution0, index) + ")"; + GetLutValue(LightingRegs::LightingSampler::Distribution0, index) + ")"; } std::string specular_0 = "(" + d0_lut_value + " * " + light_src + ".specular_0)"; // If enabled, lookup ReflectRed value, otherwise, 1.0 is used if (lighting.lut_rr.enable && - Pica::Regs::IsLightingSamplerSupported(lighting.config, - Pica::Regs::LightingSampler::ReflectRed)) { + LightingRegs::IsLightingSamplerSupported(lighting.config, + LightingRegs::LightingSampler::ReflectRed)) { std::string index = GetLutIndex(light_config.num, lighting.lut_rr.type, lighting.lut_rr.abs_input); std::string value = "(" + std::to_string(lighting.lut_rr.scale) + " * " + - GetLutValue(Regs::LightingSampler::ReflectRed, index) + ")"; + GetLutValue(LightingRegs::LightingSampler::ReflectRed, index) + ")"; out += "refl_value.r = " + value + ";\n"; } else { out += "refl_value.r = 1.0;\n"; @@ -497,12 +504,13 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) { // If enabled, lookup ReflectGreen value, otherwise, ReflectRed value is used if (lighting.lut_rg.enable && - Pica::Regs::IsLightingSamplerSupported(lighting.config, - Pica::Regs::LightingSampler::ReflectGreen)) { + LightingRegs::IsLightingSamplerSupported(lighting.config, + LightingRegs::LightingSampler::ReflectGreen)) { std::string index = GetLutIndex(light_config.num, lighting.lut_rg.type, lighting.lut_rg.abs_input); std::string value = "(" + std::to_string(lighting.lut_rg.scale) + " * " + - GetLutValue(Regs::LightingSampler::ReflectGreen, index) + ")"; + GetLutValue(LightingRegs::LightingSampler::ReflectGreen, index) + + ")"; out += "refl_value.g = " + value + ";\n"; } else { out += "refl_value.g = refl_value.r;\n"; @@ -510,12 +518,13 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) { // If enabled, lookup ReflectBlue value, otherwise, ReflectRed value is used if (lighting.lut_rb.enable && - Pica::Regs::IsLightingSamplerSupported(lighting.config, - Pica::Regs::LightingSampler::ReflectBlue)) { + LightingRegs::IsLightingSamplerSupported(lighting.config, + LightingRegs::LightingSampler::ReflectBlue)) { std::string index = GetLutIndex(light_config.num, lighting.lut_rb.type, lighting.lut_rb.abs_input); std::string value = "(" + std::to_string(lighting.lut_rb.scale) + " * " + - GetLutValue(Regs::LightingSampler::ReflectBlue, index) + ")"; + GetLutValue(LightingRegs::LightingSampler::ReflectBlue, index) + + ")"; out += "refl_value.b = " + value + ";\n"; } else { out += "refl_value.b = refl_value.r;\n"; @@ -524,35 +533,39 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) { // Specular 1 component std::string d1_lut_value = "1.0"; if (lighting.lut_d1.enable && - Pica::Regs::IsLightingSamplerSupported(lighting.config, - Pica::Regs::LightingSampler::Distribution1)) { + LightingRegs::IsLightingSamplerSupported( + lighting.config, LightingRegs::LightingSampler::Distribution1)) { // Lookup specular "distribution 1" LUT value std::string index = GetLutIndex(light_config.num, lighting.lut_d1.type, lighting.lut_d1.abs_input); d1_lut_value = "(" + std::to_string(lighting.lut_d1.scale) + " * " + - GetLutValue(Regs::LightingSampler::Distribution1, index) + ")"; + GetLutValue(LightingRegs::LightingSampler::Distribution1, index) + ")"; } std::string specular_1 = "(" + d1_lut_value + " * refl_value * " + light_src + ".specular_1)"; // Fresnel - if (lighting.lut_fr.enable && Pica::Regs::IsLightingSamplerSupported( - lighting.config, Pica::Regs::LightingSampler::Fresnel)) { + if (lighting.lut_fr.enable && + LightingRegs::IsLightingSamplerSupported(lighting.config, + LightingRegs::LightingSampler::Fresnel)) { // Lookup fresnel LUT value std::string index = GetLutIndex(light_config.num, lighting.lut_fr.type, lighting.lut_fr.abs_input); std::string value = "(" + std::to_string(lighting.lut_fr.scale) + " * " + - GetLutValue(Regs::LightingSampler::Fresnel, index) + ")"; + GetLutValue(LightingRegs::LightingSampler::Fresnel, index) + ")"; // Enabled for difffuse lighting alpha component - if (lighting.fresnel_selector == Pica::Regs::LightingFresnelSelector::PrimaryAlpha || - lighting.fresnel_selector == Pica::Regs::LightingFresnelSelector::Both) + if (lighting.fresnel_selector == LightingRegs::LightingFresnelSelector::PrimaryAlpha || + lighting.fresnel_selector == LightingRegs::LightingFresnelSelector::Both) { out += "diffuse_sum.a *= " + value + ";\n"; + } // Enabled for the specular lighting alpha component - if (lighting.fresnel_selector == Pica::Regs::LightingFresnelSelector::SecondaryAlpha || - lighting.fresnel_selector == Pica::Regs::LightingFresnelSelector::Both) + if (lighting.fresnel_selector == + LightingRegs::LightingFresnelSelector::SecondaryAlpha || + lighting.fresnel_selector == LightingRegs::LightingFresnelSelector::Both) { out += "specular_sum.a *= " + value + ";\n"; + } } // Compute primary fragment color (diffuse lighting) function @@ -633,16 +646,16 @@ vec4 secondary_fragment_color = vec4(0.0); )"; // Do not do any sort of processing if it's obvious we're not going to pass the alpha test - if (state.alpha_test_func == Regs::CompareFunc::Never) { + if (state.alpha_test_func == FramebufferRegs::CompareFunc::Never) { out += "discard; }"; return out; } // Append the scissor test - if (state.scissor_test_mode != Regs::ScissorMode::Disabled) { + if (state.scissor_test_mode != RasterizerRegs::ScissorMode::Disabled) { out += "if ("; // Negate the condition if we have to keep only the pixels outside the scissor box - if (state.scissor_test_mode == Regs::ScissorMode::Include) + if (state.scissor_test_mode == RasterizerRegs::ScissorMode::Include) out += "!"; out += "(gl_FragCoord.x >= scissor_x1 && " "gl_FragCoord.y >= scissor_y1 && " @@ -652,7 +665,7 @@ vec4 secondary_fragment_color = vec4(0.0); out += "float z_over_w = 1.0 - gl_FragCoord.z * 2.0;\n"; out += "float depth = z_over_w * depth_scale + depth_offset;\n"; - if (state.depthmap_enable == Pica::Regs::DepthBuffering::WBuffering) { + if (state.depthmap_enable == RasterizerRegs::DepthBuffering::WBuffering) { out += "depth /= gl_FragCoord.w;\n"; } @@ -666,14 +679,14 @@ vec4 secondary_fragment_color = vec4(0.0); for (size_t index = 0; index < state.tev_stages.size(); ++index) WriteTevStage(out, config, (unsigned)index); - if (state.alpha_test_func != Regs::CompareFunc::Always) { + if (state.alpha_test_func != FramebufferRegs::CompareFunc::Always) { out += "if ("; AppendAlphaTestCondition(out, state.alpha_test_func); out += ") discard;\n"; } // Append fog combiner - if (state.fog_mode == Regs::FogMode::Fog) { + if (state.fog_mode == TexturingRegs::FogMode::Fog) { // Get index into fog LUT if (state.fog_flip) { out += "float fog_index = (1.0 - depth) * 128.0;\n"; diff --git a/src/video_core/renderer_opengl/pica_to_gl.h b/src/video_core/renderer_opengl/pica_to_gl.h index cc49867c8..93d7b0b71 100644 --- a/src/video_core/renderer_opengl/pica_to_gl.h +++ b/src/video_core/renderer_opengl/pica_to_gl.h @@ -12,7 +12,9 @@ #include "common/common_funcs.h" #include "common/common_types.h" #include "common/logging/log.h" -#include "video_core/pica.h" +#include "video_core/regs_framebuffer.h" +#include "video_core/regs_lighting.h" +#include "video_core/regs_texturing.h" using GLvec2 = std::array<GLfloat, 2>; using GLvec3 = std::array<GLfloat, 3>; @@ -20,7 +22,7 @@ using GLvec4 = std::array<GLfloat, 4>; namespace PicaToGL { -inline GLenum TextureFilterMode(Pica::Regs::TextureConfig::TextureFilter mode) { +inline GLenum TextureFilterMode(Pica::TexturingRegs::TextureConfig::TextureFilter mode) { static const GLenum filter_mode_table[] = { GL_NEAREST, // TextureFilter::Nearest GL_LINEAR, // TextureFilter::Linear @@ -47,7 +49,7 @@ inline GLenum TextureFilterMode(Pica::Regs::TextureConfig::TextureFilter mode) { return gl_mode; } -inline GLenum WrapMode(Pica::Regs::TextureConfig::WrapMode mode) { +inline GLenum WrapMode(Pica::TexturingRegs::TextureConfig::WrapMode mode) { static const GLenum wrap_mode_table[] = { GL_CLAMP_TO_EDGE, // WrapMode::ClampToEdge GL_CLAMP_TO_BORDER, // WrapMode::ClampToBorder @@ -76,7 +78,7 @@ inline GLenum WrapMode(Pica::Regs::TextureConfig::WrapMode mode) { return gl_mode; } -inline GLenum BlendEquation(Pica::Regs::BlendEquation equation) { +inline GLenum BlendEquation(Pica::FramebufferRegs::BlendEquation equation) { static const GLenum blend_equation_table[] = { GL_FUNC_ADD, // BlendEquation::Add GL_FUNC_SUBTRACT, // BlendEquation::Subtract @@ -96,7 +98,7 @@ inline GLenum BlendEquation(Pica::Regs::BlendEquation equation) { return blend_equation_table[(unsigned)equation]; } -inline GLenum BlendFunc(Pica::Regs::BlendFactor factor) { +inline GLenum BlendFunc(Pica::FramebufferRegs::BlendFactor factor) { static const GLenum blend_func_table[] = { GL_ZERO, // BlendFactor::Zero GL_ONE, // BlendFactor::One @@ -126,7 +128,7 @@ inline GLenum BlendFunc(Pica::Regs::BlendFactor factor) { return blend_func_table[(unsigned)factor]; } -inline GLenum LogicOp(Pica::Regs::LogicOp op) { +inline GLenum LogicOp(Pica::FramebufferRegs::LogicOp op) { static const GLenum logic_op_table[] = { GL_CLEAR, // Clear GL_AND, // And @@ -157,7 +159,7 @@ inline GLenum LogicOp(Pica::Regs::LogicOp op) { return logic_op_table[(unsigned)op]; } -inline GLenum CompareFunc(Pica::Regs::CompareFunc func) { +inline GLenum CompareFunc(Pica::FramebufferRegs::CompareFunc func) { static const GLenum compare_func_table[] = { GL_NEVER, // CompareFunc::Never GL_ALWAYS, // CompareFunc::Always @@ -180,7 +182,7 @@ inline GLenum CompareFunc(Pica::Regs::CompareFunc func) { return compare_func_table[(unsigned)func]; } -inline GLenum StencilOp(Pica::Regs::StencilAction action) { +inline GLenum StencilOp(Pica::FramebufferRegs::StencilAction action) { static const GLenum stencil_op_table[] = { GL_KEEP, // StencilAction::Keep GL_ZERO, // StencilAction::Zero @@ -210,7 +212,7 @@ inline GLvec4 ColorRGBA8(const u32 color) { }}; } -inline std::array<GLfloat, 3> LightColor(const Pica::Regs::LightColor& color) { +inline std::array<GLfloat, 3> LightColor(const Pica::LightingRegs::LightColor& color) { return {{ color.r / 255.0f, color.g / 255.0f, color.b / 255.0f, }}; diff --git a/src/video_core/renderer_opengl/renderer_opengl.cpp b/src/video_core/renderer_opengl/renderer_opengl.cpp index 2aa90e5c1..e19375466 100644 --- a/src/video_core/renderer_opengl/renderer_opengl.cpp +++ b/src/video_core/renderer_opengl/renderer_opengl.cpp @@ -10,8 +10,8 @@ #include "common/assert.h" #include "common/bit_field.h" #include "common/logging/log.h" -#include "common/profiler_reporting.h" -#include "common/synchronized_wrapper.h" +#include "core/core.h" +#include "core/core_timing.h" #include "core/frontend/emu_window.h" #include "core/hw/gpu.h" #include "core/hw/hw.h" @@ -145,21 +145,16 @@ void RendererOpenGL::SwapBuffers() { DrawScreens(); - auto& profiler = Common::Profiling::GetProfilingManager(); - profiler.FinishFrame(); - { - auto aggregator = Common::Profiling::GetTimingResultsAggregator(); - aggregator->AddFrame(profiler.GetPreviousFrameResults()); - } + Core::System::GetInstance().perf_stats.EndSystemFrame(); // Swap buffers render_window->PollEvents(); render_window->SwapBuffers(); - prev_state.Apply(); - - profiler.BeginFrame(); + Core::System::GetInstance().frame_limiter.DoFrameLimiting(CoreTiming::GetGlobalTimeUs()); + Core::System::GetInstance().perf_stats.BeginSystemFrame(); + prev_state.Apply(); RefreshRasterizerSetting(); if (Pica::g_debug_context && Pica::g_debug_context->recorder) { diff --git a/src/video_core/shader/shader.cpp b/src/video_core/shader/shader.cpp index 2da50bd62..67ed19ba8 100644 --- a/src/video_core/shader/shader.cpp +++ b/src/video_core/shader/shader.cpp @@ -4,10 +4,12 @@ #include <cmath> #include <cstring> +#include "common/bit_set.h" #include "common/logging/log.h" #include "common/microprofile.h" -#include "video_core/pica.h" #include "video_core/pica_state.h" +#include "video_core/regs_rasterizer.h" +#include "video_core/regs_shader.h" #include "video_core/shader/shader.h" #include "video_core/shader/shader_interpreter.h" #ifdef ARCHITECTURE_x86_64 @@ -19,38 +21,31 @@ namespace Pica { namespace Shader { -OutputVertex OutputVertex::FromRegisters(Math::Vec4<float24> output_regs[16], const Regs& regs, - u32 output_mask) { +OutputVertex OutputVertex::FromAttributeBuffer(const RasterizerRegs& regs, AttributeBuffer& input) { // Setup output data - OutputVertex ret; - // TODO(neobrain): Under some circumstances, up to 16 attributes may be output. We need to - // figure out what those circumstances are and enable the remaining outputs then. - unsigned index = 0; - for (unsigned i = 0; i < 7; ++i) { + union { + OutputVertex ret{}; + std::array<float24, 24> vertex_slots; + }; + static_assert(sizeof(vertex_slots) == sizeof(ret), "Struct and array have different sizes."); - if (index >= regs.vs_output_total) - break; + unsigned int num_attributes = regs.vs_output_total; + ASSERT(num_attributes <= 7); + for (unsigned int i = 0; i < num_attributes; ++i) { + const auto& output_register_map = regs.vs_output_attributes[i]; - if ((output_mask & (1 << i)) == 0) - continue; - - const auto& output_register_map = regs.vs_output_attributes[index]; - - u32 semantics[4] = {output_register_map.map_x, output_register_map.map_y, - output_register_map.map_z, output_register_map.map_w}; + RasterizerRegs::VSOutputAttributes::Semantic semantics[4] = { + output_register_map.map_x, output_register_map.map_y, output_register_map.map_z, + output_register_map.map_w}; for (unsigned comp = 0; comp < 4; ++comp) { - float24* out = ((float24*)&ret) + semantics[comp]; - if (semantics[comp] != Regs::VSOutputAttributes::INVALID) { - *out = output_regs[i][comp]; - } else { - // Zero output so that attributes which aren't output won't have denormals in them, - // which would slow us down later. - memset(out, 0, sizeof(*out)); + RasterizerRegs::VSOutputAttributes::Semantic semantic = semantics[comp]; + if (semantic < vertex_slots.size()) { + vertex_slots[semantic] = input.attr[i][comp]; + } else if (semantic != RasterizerRegs::VSOutputAttributes::INVALID) { + LOG_ERROR(HW_GPU, "Invalid/unknown semantic id: %u", (unsigned int)semantic); } } - - index++; } // The hardware takes the absolute and saturates vertex colors like this, *before* doing @@ -71,12 +66,20 @@ OutputVertex OutputVertex::FromRegisters(Math::Vec4<float24> output_regs[16], co return ret; } -void UnitState::LoadInputVertex(const InputVertex& input, int num_attributes) { - // Setup input register table - const auto& attribute_register_map = g_state.regs.vs.input_register_map; +void UnitState::LoadInput(const ShaderRegs& config, const AttributeBuffer& input) { + const unsigned max_attribute = config.max_input_attribute_index; - for (int i = 0; i < num_attributes; i++) - registers.input[attribute_register_map.GetRegisterForAttribute(i)] = input.attr[i]; + for (unsigned attr = 0; attr <= max_attribute; ++attr) { + unsigned reg = config.GetRegisterForAttribute(attr); + registers.input[reg] = input.attr[attr]; + } +} + +void UnitState::WriteOutput(const ShaderRegs& config, AttributeBuffer& output) { + unsigned int output_i = 0; + for (unsigned int reg : Common::BitSet<u32>(config.output_mask)) { + output.attr[output_i++] = registers.output[reg]; + } } MICROPROFILE_DEFINE(GPU_Shader, "GPU", "Shader", MP_RGB(50, 50, 240)); diff --git a/src/video_core/shader/shader.h b/src/video_core/shader/shader.h index 44d9f76c3..38ea717ab 100644 --- a/src/video_core/shader/shader.h +++ b/src/video_core/shader/shader.h @@ -12,8 +12,9 @@ #include "common/common_funcs.h" #include "common/common_types.h" #include "common/vector_math.h" -#include "video_core/pica.h" #include "video_core/pica_types.h" +#include "video_core/regs_rasterizer.h" +#include "video_core/regs_shader.h" using nihstro::RegisterType; using nihstro::SourceRegister; @@ -23,14 +24,11 @@ namespace Pica { namespace Shader { -struct InputVertex { +struct AttributeBuffer { alignas(16) Math::Vec4<float24> attr[16]; }; struct OutputVertex { - OutputVertex() = default; - - // VS output attributes Math::Vec4<float24> pos; Math::Vec4<float24> quat; Math::Vec4<float24> color; @@ -42,43 +40,22 @@ struct OutputVertex { INSERT_PADDING_WORDS(1); Math::Vec2<float24> tc2; - // Padding for optimal alignment - INSERT_PADDING_WORDS(4); - - // Attributes used to store intermediate results - - // position after perspective divide - Math::Vec3<float24> screenpos; - INSERT_PADDING_WORDS(1); - - // Linear interpolation - // factor: 0=this, 1=vtx - void Lerp(float24 factor, const OutputVertex& vtx) { - pos = pos * factor + vtx.pos * (float24::FromFloat32(1) - factor); - - // TODO: Should perform perspective correct interpolation here... - tc0 = tc0 * factor + vtx.tc0 * (float24::FromFloat32(1) - factor); - tc1 = tc1 * factor + vtx.tc1 * (float24::FromFloat32(1) - factor); - tc2 = tc2 * factor + vtx.tc2 * (float24::FromFloat32(1) - factor); - - screenpos = screenpos * factor + vtx.screenpos * (float24::FromFloat32(1) - factor); - - color = color * factor + vtx.color * (float24::FromFloat32(1) - factor); - } - - // Linear interpolation - // factor: 0=v0, 1=v1 - static OutputVertex Lerp(float24 factor, const OutputVertex& v0, const OutputVertex& v1) { - OutputVertex ret = v0; - ret.Lerp(factor, v1); - return ret; - } - - static OutputVertex FromRegisters(Math::Vec4<float24> output_regs[16], const Regs& regs, - u32 output_mask); + static OutputVertex FromAttributeBuffer(const RasterizerRegs& regs, AttributeBuffer& output); }; +#define ASSERT_POS(var, pos) \ + static_assert(offsetof(OutputVertex, var) == pos * sizeof(float24), "Semantic at wrong " \ + "offset.") +ASSERT_POS(pos, RasterizerRegs::VSOutputAttributes::POSITION_X); +ASSERT_POS(quat, RasterizerRegs::VSOutputAttributes::QUATERNION_X); +ASSERT_POS(color, RasterizerRegs::VSOutputAttributes::COLOR_R); +ASSERT_POS(tc0, RasterizerRegs::VSOutputAttributes::TEXCOORD0_U); +ASSERT_POS(tc1, RasterizerRegs::VSOutputAttributes::TEXCOORD1_U); +ASSERT_POS(tc0_w, RasterizerRegs::VSOutputAttributes::TEXCOORD0_W); +ASSERT_POS(view, RasterizerRegs::VSOutputAttributes::VIEW_X); +ASSERT_POS(tc2, RasterizerRegs::VSOutputAttributes::TEXCOORD2_U); +#undef ASSERT_POS static_assert(std::is_pod<OutputVertex>::value, "Structure is not POD"); -static_assert(sizeof(OutputVertex) == 32 * sizeof(float), "OutputVertex has invalid size"); +static_assert(sizeof(OutputVertex) == 24 * sizeof(float), "OutputVertex has invalid size"); /** * This structure contains the state information that needs to be unique for a shader unit. The 3DS @@ -137,10 +114,12 @@ struct UnitState { /** * Loads the unit state with an input vertex. * - * @param input Input vertex into the shader - * @param num_attributes The number of vertex shader attributes to load + * @param config Shader configuration registers corresponding to the unit. + * @param input Attribute buffer to load into the input registers. */ - void LoadInputVertex(const InputVertex& input, int num_attributes); + void LoadInput(const ShaderRegs& config, const AttributeBuffer& input); + + void WriteOutput(const ShaderRegs& config, AttributeBuffer& output); }; struct ShaderSetup { diff --git a/src/video_core/shader/shader_interpreter.cpp b/src/video_core/shader/shader_interpreter.cpp index c0c89b857..f4d1c46c5 100644 --- a/src/video_core/shader/shader_interpreter.cpp +++ b/src/video_core/shader/shader_interpreter.cpp @@ -668,14 +668,14 @@ void InterpreterEngine::Run(const ShaderSetup& setup, UnitState& state) const { } DebugData<true> InterpreterEngine::ProduceDebugInfo(const ShaderSetup& setup, - const InputVertex& input, - int num_attributes) const { + const AttributeBuffer& input, + const ShaderRegs& config) const { UnitState state; DebugData<true> debug_data; // Setup input register table boost::fill(state.registers.input, Math::Vec4<float24>::AssignToAll(float24::Zero())); - state.LoadInputVertex(input, num_attributes); + state.LoadInput(config, input); RunInterpreter(setup, state, debug_data, setup.engine_data.entry_point); return debug_data; } diff --git a/src/video_core/shader/shader_interpreter.h b/src/video_core/shader/shader_interpreter.h index d6c0e2d8c..50fd7c69d 100644 --- a/src/video_core/shader/shader_interpreter.h +++ b/src/video_core/shader/shader_interpreter.h @@ -18,13 +18,13 @@ public: /** * Produce debug information based on the given shader and input vertex - * @param input Input vertex into the shader - * @param num_attributes The number of vertex shader attributes + * @param setup Shader engine state + * @param input Input vertex into the shader * @param config Configuration object for the shader pipeline * @return Debug information for this shader with regards to the given vertex */ - DebugData<true> ProduceDebugInfo(const ShaderSetup& setup, const InputVertex& input, - int num_attributes) const; + DebugData<true> ProduceDebugInfo(const ShaderSetup& setup, const AttributeBuffer& input, + const ShaderRegs& config) const; }; } // namespace diff --git a/src/video_core/shader/shader_jit_x64_compiler.cpp b/src/video_core/shader/shader_jit_x64_compiler.cpp index 49806e8c9..2dbc8b147 100644 --- a/src/video_core/shader/shader_jit_x64_compiler.cpp +++ b/src/video_core/shader/shader_jit_x64_compiler.cpp @@ -144,6 +144,8 @@ static const BitSet32 persistent_regs = BuildRegSet({ ADDROFFS_REG_0, ADDROFFS_REG_1, LOOPCOUNT_REG, COND0, COND1, // Constants ONE, NEGBIT, + // Loop variables + LOOPCOUNT, LOOPINC, }); /// Raw constant for the source register selector that indicates no swizzling is performed @@ -293,14 +295,22 @@ void JitShader::Compile_DestEnable(Instruction instr, Xmm src) { } void JitShader::Compile_SanitizedMul(Xmm src1, Xmm src2, Xmm scratch) { + // 0 * inf and inf * 0 in the PICA should return 0 instead of NaN. This can be implemented by + // checking for NaNs before and after the multiplication. If the multiplication result is NaN + // where neither source was, this NaN was generated by a 0 * inf multiplication, and so the + // result should be transformed to 0 to match PICA fp rules. + + // Set scratch to mask of (src1 != NaN and src2 != NaN) movaps(scratch, src1); cmpordps(scratch, src2); mulps(src1, src2); + // Set src2 to mask of (result == NaN) movaps(src2, src1); cmpunordps(src2, src2); + // Clear components where scratch != src2 (i.e. if result is NaN where neither source was NaN) xorps(scratch, src2); andps(src1, scratch); } @@ -587,7 +597,7 @@ void JitShader::Compile_RSQ(Instruction instr) { void JitShader::Compile_NOP(Instruction instr) {} void JitShader::Compile_END(Instruction instr) { - ABI_PopRegistersAndAdjustStack(*this, ABI_ALL_CALLEE_SAVED, 8); + ABI_PopRegistersAndAdjustStack(*this, ABI_ALL_CALLEE_SAVED, 8, 16); ret(); } @@ -839,7 +849,10 @@ void JitShader::Compile(const std::array<u32, 1024>* program_code_, FindReturnOffsets(); // The stack pointer is 8 modulo 16 at the entry of a procedure - ABI_PushRegistersAndAdjustStack(*this, ABI_ALL_CALLEE_SAVED, 8); + // We reserve 16 bytes and assign a dummy value to the first 8 bytes, to catch any potential + // return checks (see Compile_Return) that happen in shader main routine. + ABI_PushRegistersAndAdjustStack(*this, ABI_ALL_CALLEE_SAVED, 8, 16); + mov(qword[rsp + 8], 0xFFFFFFFFFFFFFFFFULL); mov(SETUP, ABI_PARAM1); mov(STATE, ABI_PARAM2); diff --git a/src/video_core/shader/shader_jit_x64_compiler.h b/src/video_core/shader/shader_jit_x64_compiler.h index 29e9875ea..f27675560 100644 --- a/src/video_core/shader/shader_jit_x64_compiler.h +++ b/src/video_core/shader/shader_jit_x64_compiler.h @@ -12,7 +12,6 @@ #include <xbyak.h> #include "common/bit_set.h" #include "common/common_types.h" -#include "common/x64/emitter.h" #include "video_core/shader/shader.h" using nihstro::Instruction; @@ -94,7 +93,8 @@ private: /** * Assertion evaluated at compile-time, but only triggered if executed at runtime. - * @param msg Message to be logged if the assertion fails. + * @param condition Condition to be evaluated. + * @param msg Message to be logged if the assertion fails. */ void Compile_Assert(bool condition, const char* msg); diff --git a/src/video_core/clipper.cpp b/src/video_core/swrasterizer/clipper.cpp index 05b5cea73..2d80822d9 100644 --- a/src/video_core/clipper.cpp +++ b/src/video_core/swrasterizer/clipper.cpp @@ -11,12 +11,13 @@ #include "common/common_types.h" #include "common/logging/log.h" #include "common/vector_math.h" -#include "video_core/clipper.h" -#include "video_core/pica.h" #include "video_core/pica_state.h" #include "video_core/pica_types.h" -#include "video_core/rasterizer.h" #include "video_core/shader/shader.h" +#include "video_core/swrasterizer/clipper.h" +#include "video_core/swrasterizer/rasterizer.h" + +using Pica::Rasterizer::Vertex; namespace Pica { @@ -29,20 +30,20 @@ public: float24::FromFloat32(0), float24::FromFloat32(0))) : coeffs(coeffs), bias(bias) {} - bool IsInside(const OutputVertex& vertex) const { + bool IsInside(const Vertex& vertex) const { return Math::Dot(vertex.pos + bias, coeffs) <= float24::FromFloat32(0); } - bool IsOutSide(const OutputVertex& vertex) const { + bool IsOutSide(const Vertex& vertex) const { return !IsInside(vertex); } - OutputVertex GetIntersection(const OutputVertex& v0, const OutputVertex& v1) const { + Vertex GetIntersection(const Vertex& v0, const Vertex& v1) const { float24 dp = Math::Dot(v0.pos + bias, coeffs); float24 dp_prev = Math::Dot(v1.pos + bias, coeffs); float24 factor = dp_prev / (dp_prev - dp); - return OutputVertex::Lerp(factor, v0, v1); + return Vertex::Lerp(factor, v0, v1); } private: @@ -51,7 +52,7 @@ private: Math::Vec4<float24> bias; }; -static void InitScreenCoordinates(OutputVertex& vtx) { +static void InitScreenCoordinates(Vertex& vtx) { struct { float24 halfsize_x; float24 offset_x; @@ -62,10 +63,10 @@ static void InitScreenCoordinates(OutputVertex& vtx) { } viewport; const auto& regs = g_state.regs; - viewport.halfsize_x = float24::FromRaw(regs.viewport_size_x); - viewport.halfsize_y = float24::FromRaw(regs.viewport_size_y); - viewport.offset_x = float24::FromFloat32(static_cast<float>(regs.viewport_corner.x)); - viewport.offset_y = float24::FromFloat32(static_cast<float>(regs.viewport_corner.y)); + viewport.halfsize_x = float24::FromRaw(regs.rasterizer.viewport_size_x); + viewport.halfsize_y = float24::FromRaw(regs.rasterizer.viewport_size_y); + viewport.offset_x = float24::FromFloat32(static_cast<float>(regs.rasterizer.viewport_corner.x)); + viewport.offset_y = float24::FromFloat32(static_cast<float>(regs.rasterizer.viewport_corner.y)); float24 inv_w = float24::FromFloat32(1.f) / vtx.pos.w; vtx.color *= inv_w; @@ -91,8 +92,8 @@ void ProcessTriangle(const OutputVertex& v0, const OutputVertex& v1, const Outpu // introduces at most 1 new vertex to the polygon. Since we start with a triangle and have a // fixed 6 clipping planes, the maximum number of vertices of the clipped polygon is 3 + 6 = 9. static const size_t MAX_VERTICES = 9; - static_vector<OutputVertex, MAX_VERTICES> buffer_a = {v0, v1, v2}; - static_vector<OutputVertex, MAX_VERTICES> buffer_b; + static_vector<Vertex, MAX_VERTICES> buffer_a = {v0, v1, v2}; + static_vector<Vertex, MAX_VERTICES> buffer_b; auto* output_list = &buffer_a; auto* input_list = &buffer_b; @@ -123,7 +124,7 @@ void ProcessTriangle(const OutputVertex& v0, const OutputVertex& v1, const Outpu std::swap(input_list, output_list); output_list->clear(); - const OutputVertex* reference_vertex = &input_list->back(); + const Vertex* reference_vertex = &input_list->back(); for (const auto& vertex : *input_list) { // NOTE: This algorithm changes vertex order in some cases! @@ -148,9 +149,9 @@ void ProcessTriangle(const OutputVertex& v0, const OutputVertex& v1, const Outpu InitScreenCoordinates((*output_list)[1]); for (size_t i = 0; i < output_list->size() - 2; i++) { - OutputVertex& vtx0 = (*output_list)[0]; - OutputVertex& vtx1 = (*output_list)[i + 1]; - OutputVertex& vtx2 = (*output_list)[i + 2]; + Vertex& vtx0 = (*output_list)[0]; + Vertex& vtx1 = (*output_list)[i + 1]; + Vertex& vtx2 = (*output_list)[i + 2]; InitScreenCoordinates(vtx2); diff --git a/src/video_core/clipper.h b/src/video_core/swrasterizer/clipper.h index b51af0af9..b51af0af9 100644 --- a/src/video_core/clipper.h +++ b/src/video_core/swrasterizer/clipper.h diff --git a/src/video_core/swrasterizer/framebuffer.cpp b/src/video_core/swrasterizer/framebuffer.cpp new file mode 100644 index 000000000..7de3aac75 --- /dev/null +++ b/src/video_core/swrasterizer/framebuffer.cpp @@ -0,0 +1,358 @@ +// Copyright 2017 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <algorithm> + +#include "common/assert.h" +#include "common/color.h" +#include "common/common_types.h" +#include "common/logging/log.h" +#include "common/math_util.h" +#include "common/vector_math.h" +#include "core/hw/gpu.h" +#include "core/memory.h" +#include "video_core/pica_state.h" +#include "video_core/regs_framebuffer.h" +#include "video_core/swrasterizer/framebuffer.h" +#include "video_core/utils.h" + +namespace Pica { +namespace Rasterizer { + +void DrawPixel(int x, int y, const Math::Vec4<u8>& color) { + const auto& framebuffer = g_state.regs.framebuffer.framebuffer; + const PAddr addr = framebuffer.GetColorBufferPhysicalAddress(); + + // Similarly to textures, the render framebuffer is laid out from bottom to top, too. + // NOTE: The framebuffer height register contains the actual FB height minus one. + y = framebuffer.height - y; + + const u32 coarse_y = y & ~7; + u32 bytes_per_pixel = + GPU::Regs::BytesPerPixel(GPU::Regs::PixelFormat(framebuffer.color_format.Value())); + u32 dst_offset = VideoCore::GetMortonOffset(x, y, bytes_per_pixel) + + coarse_y * framebuffer.width * bytes_per_pixel; + u8* dst_pixel = Memory::GetPhysicalPointer(addr) + dst_offset; + + switch (framebuffer.color_format) { + case FramebufferRegs::ColorFormat::RGBA8: + Color::EncodeRGBA8(color, dst_pixel); + break; + + case FramebufferRegs::ColorFormat::RGB8: + Color::EncodeRGB8(color, dst_pixel); + break; + + case FramebufferRegs::ColorFormat::RGB5A1: + Color::EncodeRGB5A1(color, dst_pixel); + break; + + case FramebufferRegs::ColorFormat::RGB565: + Color::EncodeRGB565(color, dst_pixel); + break; + + case FramebufferRegs::ColorFormat::RGBA4: + Color::EncodeRGBA4(color, dst_pixel); + break; + + default: + LOG_CRITICAL(Render_Software, "Unknown framebuffer color format %x", + framebuffer.color_format.Value()); + UNIMPLEMENTED(); + } +} + +const Math::Vec4<u8> GetPixel(int x, int y) { + const auto& framebuffer = g_state.regs.framebuffer.framebuffer; + const PAddr addr = framebuffer.GetColorBufferPhysicalAddress(); + + y = framebuffer.height - y; + + const u32 coarse_y = y & ~7; + u32 bytes_per_pixel = + GPU::Regs::BytesPerPixel(GPU::Regs::PixelFormat(framebuffer.color_format.Value())); + u32 src_offset = VideoCore::GetMortonOffset(x, y, bytes_per_pixel) + + coarse_y * framebuffer.width * bytes_per_pixel; + u8* src_pixel = Memory::GetPhysicalPointer(addr) + src_offset; + + switch (framebuffer.color_format) { + case FramebufferRegs::ColorFormat::RGBA8: + return Color::DecodeRGBA8(src_pixel); + + case FramebufferRegs::ColorFormat::RGB8: + return Color::DecodeRGB8(src_pixel); + + case FramebufferRegs::ColorFormat::RGB5A1: + return Color::DecodeRGB5A1(src_pixel); + + case FramebufferRegs::ColorFormat::RGB565: + return Color::DecodeRGB565(src_pixel); + + case FramebufferRegs::ColorFormat::RGBA4: + return Color::DecodeRGBA4(src_pixel); + + default: + LOG_CRITICAL(Render_Software, "Unknown framebuffer color format %x", + framebuffer.color_format.Value()); + UNIMPLEMENTED(); + } + + return {0, 0, 0, 0}; +} + +u32 GetDepth(int x, int y) { + const auto& framebuffer = g_state.regs.framebuffer.framebuffer; + const PAddr addr = framebuffer.GetDepthBufferPhysicalAddress(); + u8* depth_buffer = Memory::GetPhysicalPointer(addr); + + y = framebuffer.height - y; + + const u32 coarse_y = y & ~7; + u32 bytes_per_pixel = FramebufferRegs::BytesPerDepthPixel(framebuffer.depth_format); + u32 stride = framebuffer.width * bytes_per_pixel; + + u32 src_offset = VideoCore::GetMortonOffset(x, y, bytes_per_pixel) + coarse_y * stride; + u8* src_pixel = depth_buffer + src_offset; + + switch (framebuffer.depth_format) { + case FramebufferRegs::DepthFormat::D16: + return Color::DecodeD16(src_pixel); + case FramebufferRegs::DepthFormat::D24: + return Color::DecodeD24(src_pixel); + case FramebufferRegs::DepthFormat::D24S8: + return Color::DecodeD24S8(src_pixel).x; + default: + LOG_CRITICAL(HW_GPU, "Unimplemented depth format %u", framebuffer.depth_format); + UNIMPLEMENTED(); + return 0; + } +} + +u8 GetStencil(int x, int y) { + const auto& framebuffer = g_state.regs.framebuffer.framebuffer; + const PAddr addr = framebuffer.GetDepthBufferPhysicalAddress(); + u8* depth_buffer = Memory::GetPhysicalPointer(addr); + + y = framebuffer.height - y; + + const u32 coarse_y = y & ~7; + u32 bytes_per_pixel = Pica::FramebufferRegs::BytesPerDepthPixel(framebuffer.depth_format); + u32 stride = framebuffer.width * bytes_per_pixel; + + u32 src_offset = VideoCore::GetMortonOffset(x, y, bytes_per_pixel) + coarse_y * stride; + u8* src_pixel = depth_buffer + src_offset; + + switch (framebuffer.depth_format) { + case FramebufferRegs::DepthFormat::D24S8: + return Color::DecodeD24S8(src_pixel).y; + + default: + LOG_WARNING( + HW_GPU, + "GetStencil called for function which doesn't have a stencil component (format %u)", + framebuffer.depth_format); + return 0; + } +} + +void SetDepth(int x, int y, u32 value) { + const auto& framebuffer = g_state.regs.framebuffer.framebuffer; + const PAddr addr = framebuffer.GetDepthBufferPhysicalAddress(); + u8* depth_buffer = Memory::GetPhysicalPointer(addr); + + y = framebuffer.height - y; + + const u32 coarse_y = y & ~7; + u32 bytes_per_pixel = FramebufferRegs::BytesPerDepthPixel(framebuffer.depth_format); + u32 stride = framebuffer.width * bytes_per_pixel; + + u32 dst_offset = VideoCore::GetMortonOffset(x, y, bytes_per_pixel) + coarse_y * stride; + u8* dst_pixel = depth_buffer + dst_offset; + + switch (framebuffer.depth_format) { + case FramebufferRegs::DepthFormat::D16: + Color::EncodeD16(value, dst_pixel); + break; + + case FramebufferRegs::DepthFormat::D24: + Color::EncodeD24(value, dst_pixel); + break; + + case FramebufferRegs::DepthFormat::D24S8: + Color::EncodeD24X8(value, dst_pixel); + break; + + default: + LOG_CRITICAL(HW_GPU, "Unimplemented depth format %u", framebuffer.depth_format); + UNIMPLEMENTED(); + break; + } +} + +void SetStencil(int x, int y, u8 value) { + const auto& framebuffer = g_state.regs.framebuffer.framebuffer; + const PAddr addr = framebuffer.GetDepthBufferPhysicalAddress(); + u8* depth_buffer = Memory::GetPhysicalPointer(addr); + + y = framebuffer.height - y; + + const u32 coarse_y = y & ~7; + u32 bytes_per_pixel = Pica::FramebufferRegs::BytesPerDepthPixel(framebuffer.depth_format); + u32 stride = framebuffer.width * bytes_per_pixel; + + u32 dst_offset = VideoCore::GetMortonOffset(x, y, bytes_per_pixel) + coarse_y * stride; + u8* dst_pixel = depth_buffer + dst_offset; + + switch (framebuffer.depth_format) { + case Pica::FramebufferRegs::DepthFormat::D16: + case Pica::FramebufferRegs::DepthFormat::D24: + // Nothing to do + break; + + case Pica::FramebufferRegs::DepthFormat::D24S8: + Color::EncodeX24S8(value, dst_pixel); + break; + + default: + LOG_CRITICAL(HW_GPU, "Unimplemented depth format %u", framebuffer.depth_format); + UNIMPLEMENTED(); + break; + } +} + +u8 PerformStencilAction(FramebufferRegs::StencilAction action, u8 old_stencil, u8 ref) { + switch (action) { + case FramebufferRegs::StencilAction::Keep: + return old_stencil; + + case FramebufferRegs::StencilAction::Zero: + return 0; + + case FramebufferRegs::StencilAction::Replace: + return ref; + + case FramebufferRegs::StencilAction::Increment: + // Saturated increment + return std::min<u8>(old_stencil, 254) + 1; + + case FramebufferRegs::StencilAction::Decrement: + // Saturated decrement + return std::max<u8>(old_stencil, 1) - 1; + + case FramebufferRegs::StencilAction::Invert: + return ~old_stencil; + + case FramebufferRegs::StencilAction::IncrementWrap: + return old_stencil + 1; + + case FramebufferRegs::StencilAction::DecrementWrap: + return old_stencil - 1; + + default: + LOG_CRITICAL(HW_GPU, "Unknown stencil action %x", (int)action); + UNIMPLEMENTED(); + return 0; + } +} + +Math::Vec4<u8> EvaluateBlendEquation(const Math::Vec4<u8>& src, const Math::Vec4<u8>& srcfactor, + const Math::Vec4<u8>& dest, const Math::Vec4<u8>& destfactor, + FramebufferRegs::BlendEquation equation) { + Math::Vec4<int> result; + + auto src_result = (src * srcfactor).Cast<int>(); + auto dst_result = (dest * destfactor).Cast<int>(); + + switch (equation) { + case FramebufferRegs::BlendEquation::Add: + result = (src_result + dst_result) / 255; + break; + + case FramebufferRegs::BlendEquation::Subtract: + result = (src_result - dst_result) / 255; + break; + + case FramebufferRegs::BlendEquation::ReverseSubtract: + result = (dst_result - src_result) / 255; + break; + + // TODO: How do these two actually work? OpenGL doesn't include the blend factors in the + // min/max computations, but is this what the 3DS actually does? + case FramebufferRegs::BlendEquation::Min: + result.r() = std::min(src.r(), dest.r()); + result.g() = std::min(src.g(), dest.g()); + result.b() = std::min(src.b(), dest.b()); + result.a() = std::min(src.a(), dest.a()); + break; + + case FramebufferRegs::BlendEquation::Max: + result.r() = std::max(src.r(), dest.r()); + result.g() = std::max(src.g(), dest.g()); + result.b() = std::max(src.b(), dest.b()); + result.a() = std::max(src.a(), dest.a()); + break; + + default: + LOG_CRITICAL(HW_GPU, "Unknown RGB blend equation %x", equation); + UNIMPLEMENTED(); + } + + return Math::Vec4<u8>(MathUtil::Clamp(result.r(), 0, 255), MathUtil::Clamp(result.g(), 0, 255), + MathUtil::Clamp(result.b(), 0, 255), MathUtil::Clamp(result.a(), 0, 255)); +}; + +u8 LogicOp(u8 src, u8 dest, FramebufferRegs::LogicOp op) { + switch (op) { + case FramebufferRegs::LogicOp::Clear: + return 0; + + case FramebufferRegs::LogicOp::And: + return src & dest; + + case FramebufferRegs::LogicOp::AndReverse: + return src & ~dest; + + case FramebufferRegs::LogicOp::Copy: + return src; + + case FramebufferRegs::LogicOp::Set: + return 255; + + case FramebufferRegs::LogicOp::CopyInverted: + return ~src; + + case FramebufferRegs::LogicOp::NoOp: + return dest; + + case FramebufferRegs::LogicOp::Invert: + return ~dest; + + case FramebufferRegs::LogicOp::Nand: + return ~(src & dest); + + case FramebufferRegs::LogicOp::Or: + return src | dest; + + case FramebufferRegs::LogicOp::Nor: + return ~(src | dest); + + case FramebufferRegs::LogicOp::Xor: + return src ^ dest; + + case FramebufferRegs::LogicOp::Equiv: + return ~(src ^ dest); + + case FramebufferRegs::LogicOp::AndInverted: + return ~src & dest; + + case FramebufferRegs::LogicOp::OrReverse: + return src | ~dest; + + case FramebufferRegs::LogicOp::OrInverted: + return ~src | dest; + } +}; + +} // namespace Rasterizer +} // namespace Pica diff --git a/src/video_core/swrasterizer/framebuffer.h b/src/video_core/swrasterizer/framebuffer.h new file mode 100644 index 000000000..4a32a4979 --- /dev/null +++ b/src/video_core/swrasterizer/framebuffer.h @@ -0,0 +1,29 @@ +// Copyright 2017 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 "common/vector_math.h" +#include "video_core/regs_framebuffer.h" + +namespace Pica { +namespace Rasterizer { + +void DrawPixel(int x, int y, const Math::Vec4<u8>& color); +const Math::Vec4<u8> GetPixel(int x, int y); +u32 GetDepth(int x, int y); +u8 GetStencil(int x, int y); +void SetDepth(int x, int y, u32 value); +void SetStencil(int x, int y, u8 value); +u8 PerformStencilAction(FramebufferRegs::StencilAction action, u8 old_stencil, u8 ref); + +Math::Vec4<u8> EvaluateBlendEquation(const Math::Vec4<u8>& src, const Math::Vec4<u8>& srcfactor, + const Math::Vec4<u8>& dest, const Math::Vec4<u8>& destfactor, + FramebufferRegs::BlendEquation equation); + +u8 LogicOp(u8 src, u8 dest, FramebufferRegs::LogicOp op); + +} // namespace Rasterizer +} // namespace Pica diff --git a/src/video_core/rasterizer.cpp b/src/video_core/swrasterizer/rasterizer.cpp index b9f5d4533..7557fcb89 100644 --- a/src/video_core/rasterizer.cpp +++ b/src/video_core/swrasterizer/rasterizer.cpp @@ -16,253 +16,21 @@ #include "core/hw/gpu.h" #include "core/memory.h" #include "video_core/debug_utils/debug_utils.h" -#include "video_core/pica.h" #include "video_core/pica_state.h" #include "video_core/pica_types.h" -#include "video_core/rasterizer.h" +#include "video_core/regs_framebuffer.h" +#include "video_core/regs_rasterizer.h" +#include "video_core/regs_texturing.h" #include "video_core/shader/shader.h" +#include "video_core/swrasterizer/framebuffer.h" +#include "video_core/swrasterizer/rasterizer.h" +#include "video_core/swrasterizer/texturing.h" +#include "video_core/texture/texture_decode.h" #include "video_core/utils.h" namespace Pica { - namespace Rasterizer { -static void DrawPixel(int x, int y, const Math::Vec4<u8>& color) { - const auto& framebuffer = g_state.regs.framebuffer; - const PAddr addr = framebuffer.GetColorBufferPhysicalAddress(); - - // Similarly to textures, the render framebuffer is laid out from bottom to top, too. - // NOTE: The framebuffer height register contains the actual FB height minus one. - y = framebuffer.height - y; - - const u32 coarse_y = y & ~7; - u32 bytes_per_pixel = - GPU::Regs::BytesPerPixel(GPU::Regs::PixelFormat(framebuffer.color_format.Value())); - u32 dst_offset = VideoCore::GetMortonOffset(x, y, bytes_per_pixel) + - coarse_y * framebuffer.width * bytes_per_pixel; - u8* dst_pixel = Memory::GetPhysicalPointer(addr) + dst_offset; - - switch (framebuffer.color_format) { - case Regs::ColorFormat::RGBA8: - Color::EncodeRGBA8(color, dst_pixel); - break; - - case Regs::ColorFormat::RGB8: - Color::EncodeRGB8(color, dst_pixel); - break; - - case Regs::ColorFormat::RGB5A1: - Color::EncodeRGB5A1(color, dst_pixel); - break; - - case Regs::ColorFormat::RGB565: - Color::EncodeRGB565(color, dst_pixel); - break; - - case Regs::ColorFormat::RGBA4: - Color::EncodeRGBA4(color, dst_pixel); - break; - - default: - LOG_CRITICAL(Render_Software, "Unknown framebuffer color format %x", - framebuffer.color_format.Value()); - UNIMPLEMENTED(); - } -} - -static const Math::Vec4<u8> GetPixel(int x, int y) { - const auto& framebuffer = g_state.regs.framebuffer; - const PAddr addr = framebuffer.GetColorBufferPhysicalAddress(); - - y = framebuffer.height - y; - - const u32 coarse_y = y & ~7; - u32 bytes_per_pixel = - GPU::Regs::BytesPerPixel(GPU::Regs::PixelFormat(framebuffer.color_format.Value())); - u32 src_offset = VideoCore::GetMortonOffset(x, y, bytes_per_pixel) + - coarse_y * framebuffer.width * bytes_per_pixel; - u8* src_pixel = Memory::GetPhysicalPointer(addr) + src_offset; - - switch (framebuffer.color_format) { - case Regs::ColorFormat::RGBA8: - return Color::DecodeRGBA8(src_pixel); - - case Regs::ColorFormat::RGB8: - return Color::DecodeRGB8(src_pixel); - - case Regs::ColorFormat::RGB5A1: - return Color::DecodeRGB5A1(src_pixel); - - case Regs::ColorFormat::RGB565: - return Color::DecodeRGB565(src_pixel); - - case Regs::ColorFormat::RGBA4: - return Color::DecodeRGBA4(src_pixel); - - default: - LOG_CRITICAL(Render_Software, "Unknown framebuffer color format %x", - framebuffer.color_format.Value()); - UNIMPLEMENTED(); - } - - return {0, 0, 0, 0}; -} - -static u32 GetDepth(int x, int y) { - const auto& framebuffer = g_state.regs.framebuffer; - const PAddr addr = framebuffer.GetDepthBufferPhysicalAddress(); - u8* depth_buffer = Memory::GetPhysicalPointer(addr); - - y = framebuffer.height - y; - - const u32 coarse_y = y & ~7; - u32 bytes_per_pixel = Regs::BytesPerDepthPixel(framebuffer.depth_format); - u32 stride = framebuffer.width * bytes_per_pixel; - - u32 src_offset = VideoCore::GetMortonOffset(x, y, bytes_per_pixel) + coarse_y * stride; - u8* src_pixel = depth_buffer + src_offset; - - switch (framebuffer.depth_format) { - case Regs::DepthFormat::D16: - return Color::DecodeD16(src_pixel); - case Regs::DepthFormat::D24: - return Color::DecodeD24(src_pixel); - case Regs::DepthFormat::D24S8: - return Color::DecodeD24S8(src_pixel).x; - default: - LOG_CRITICAL(HW_GPU, "Unimplemented depth format %u", framebuffer.depth_format); - UNIMPLEMENTED(); - return 0; - } -} - -static u8 GetStencil(int x, int y) { - const auto& framebuffer = g_state.regs.framebuffer; - const PAddr addr = framebuffer.GetDepthBufferPhysicalAddress(); - u8* depth_buffer = Memory::GetPhysicalPointer(addr); - - y = framebuffer.height - y; - - const u32 coarse_y = y & ~7; - u32 bytes_per_pixel = Pica::Regs::BytesPerDepthPixel(framebuffer.depth_format); - u32 stride = framebuffer.width * bytes_per_pixel; - - u32 src_offset = VideoCore::GetMortonOffset(x, y, bytes_per_pixel) + coarse_y * stride; - u8* src_pixel = depth_buffer + src_offset; - - switch (framebuffer.depth_format) { - case Regs::DepthFormat::D24S8: - return Color::DecodeD24S8(src_pixel).y; - - default: - LOG_WARNING( - HW_GPU, - "GetStencil called for function which doesn't have a stencil component (format %u)", - framebuffer.depth_format); - return 0; - } -} - -static void SetDepth(int x, int y, u32 value) { - const auto& framebuffer = g_state.regs.framebuffer; - const PAddr addr = framebuffer.GetDepthBufferPhysicalAddress(); - u8* depth_buffer = Memory::GetPhysicalPointer(addr); - - y = framebuffer.height - y; - - const u32 coarse_y = y & ~7; - u32 bytes_per_pixel = Regs::BytesPerDepthPixel(framebuffer.depth_format); - u32 stride = framebuffer.width * bytes_per_pixel; - - u32 dst_offset = VideoCore::GetMortonOffset(x, y, bytes_per_pixel) + coarse_y * stride; - u8* dst_pixel = depth_buffer + dst_offset; - - switch (framebuffer.depth_format) { - case Regs::DepthFormat::D16: - Color::EncodeD16(value, dst_pixel); - break; - - case Regs::DepthFormat::D24: - Color::EncodeD24(value, dst_pixel); - break; - - case Regs::DepthFormat::D24S8: - Color::EncodeD24X8(value, dst_pixel); - break; - - default: - LOG_CRITICAL(HW_GPU, "Unimplemented depth format %u", framebuffer.depth_format); - UNIMPLEMENTED(); - break; - } -} - -static void SetStencil(int x, int y, u8 value) { - const auto& framebuffer = g_state.regs.framebuffer; - const PAddr addr = framebuffer.GetDepthBufferPhysicalAddress(); - u8* depth_buffer = Memory::GetPhysicalPointer(addr); - - y = framebuffer.height - y; - - const u32 coarse_y = y & ~7; - u32 bytes_per_pixel = Pica::Regs::BytesPerDepthPixel(framebuffer.depth_format); - u32 stride = framebuffer.width * bytes_per_pixel; - - u32 dst_offset = VideoCore::GetMortonOffset(x, y, bytes_per_pixel) + coarse_y * stride; - u8* dst_pixel = depth_buffer + dst_offset; - - switch (framebuffer.depth_format) { - case Pica::Regs::DepthFormat::D16: - case Pica::Regs::DepthFormat::D24: - // Nothing to do - break; - - case Pica::Regs::DepthFormat::D24S8: - Color::EncodeX24S8(value, dst_pixel); - break; - - default: - LOG_CRITICAL(HW_GPU, "Unimplemented depth format %u", framebuffer.depth_format); - UNIMPLEMENTED(); - break; - } -} - -static u8 PerformStencilAction(Regs::StencilAction action, u8 old_stencil, u8 ref) { - switch (action) { - case Regs::StencilAction::Keep: - return old_stencil; - - case Regs::StencilAction::Zero: - return 0; - - case Regs::StencilAction::Replace: - return ref; - - case Regs::StencilAction::Increment: - // Saturated increment - return std::min<u8>(old_stencil, 254) + 1; - - case Regs::StencilAction::Decrement: - // Saturated decrement - return std::max<u8>(old_stencil, 1) - 1; - - case Regs::StencilAction::Invert: - return ~old_stencil; - - case Regs::StencilAction::IncrementWrap: - return old_stencil + 1; - - case Regs::StencilAction::DecrementWrap: - return old_stencil - 1; - - default: - LOG_CRITICAL(HW_GPU, "Unknown stencil action %x", (int)action); - UNIMPLEMENTED(); - return 0; - } -} - // NOTE: Assuming that rasterizer coordinates are 12.4 fixed-point values struct Fix12P4 { Fix12P4() {} @@ -307,8 +75,8 @@ MICROPROFILE_DEFINE(GPU_Rasterization, "GPU", "Rasterization", MP_RGB(50, 50, 24 * Helper function for ProcessTriangle with the "reversed" flag to allow for implementing * culling via recursion. */ -static void ProcessTriangleInternal(const Shader::OutputVertex& v0, const Shader::OutputVertex& v1, - const Shader::OutputVertex& v2, bool reversed = false) { +static void ProcessTriangleInternal(const Vertex& v0, const Vertex& v1, const Vertex& v2, + bool reversed = false) { const auto& regs = g_state.regs; MICROPROFILE_SCOPE(GPU_Rasterization); @@ -326,14 +94,14 @@ static void ProcessTriangleInternal(const Shader::OutputVertex& v0, const Shader ScreenToRasterizerCoordinates(v1.screenpos), ScreenToRasterizerCoordinates(v2.screenpos)}; - if (regs.cull_mode == Regs::CullMode::KeepAll) { + if (regs.rasterizer.cull_mode == RasterizerRegs::CullMode::KeepAll) { // Make sure we always end up with a triangle wound counter-clockwise if (!reversed && SignedArea(vtxpos[0].xy(), vtxpos[1].xy(), vtxpos[2].xy()) <= 0) { ProcessTriangleInternal(v0, v2, v1, true); return; } } else { - if (!reversed && regs.cull_mode == Regs::CullMode::KeepClockWise) { + if (!reversed && regs.rasterizer.cull_mode == RasterizerRegs::CullMode::KeepClockWise) { // Reverse vertex order and use the CCW code path. ProcessTriangleInternal(v0, v2, v1, true); return; @@ -350,13 +118,13 @@ static void ProcessTriangleInternal(const Shader::OutputVertex& v0, const Shader u16 max_y = std::max({vtxpos[0].y, vtxpos[1].y, vtxpos[2].y}); // Convert the scissor box coordinates to 12.4 fixed point - u16 scissor_x1 = (u16)(regs.scissor_test.x1 << 4); - u16 scissor_y1 = (u16)(regs.scissor_test.y1 << 4); + u16 scissor_x1 = (u16)(regs.rasterizer.scissor_test.x1 << 4); + u16 scissor_y1 = (u16)(regs.rasterizer.scissor_test.y1 << 4); // x2,y2 have +1 added to cover the entire sub-pixel area - u16 scissor_x2 = (u16)((regs.scissor_test.x2 + 1) << 4); - u16 scissor_y2 = (u16)((regs.scissor_test.y2 + 1) << 4); + u16 scissor_x2 = (u16)((regs.rasterizer.scissor_test.x2 + 1) << 4); + u16 scissor_y2 = (u16)((regs.rasterizer.scissor_test.y2 + 1) << 4); - if (regs.scissor_test.mode == Regs::ScissorMode::Include) { + if (regs.rasterizer.scissor_test.mode == RasterizerRegs::ScissorMode::Include) { // Calculate the new bounds min_x = std::max(min_x, scissor_x1); min_y = std::max(min_y, scissor_y1); @@ -396,12 +164,13 @@ static void ProcessTriangleInternal(const Shader::OutputVertex& v0, const Shader auto w_inverse = Math::MakeVec(v0.pos.w, v1.pos.w, v2.pos.w); - auto textures = regs.GetTextures(); - auto tev_stages = regs.GetTevStages(); + auto textures = regs.texturing.GetTextures(); + auto tev_stages = regs.texturing.GetTevStages(); - bool stencil_action_enable = g_state.regs.output_merger.stencil_test.enable && - g_state.regs.framebuffer.depth_format == Regs::DepthFormat::D24S8; - const auto stencil_test = g_state.regs.output_merger.stencil_test; + bool stencil_action_enable = + g_state.regs.framebuffer.output_merger.stencil_test.enable && + g_state.regs.framebuffer.framebuffer.depth_format == FramebufferRegs::DepthFormat::D24S8; + const auto stencil_test = g_state.regs.framebuffer.output_merger.stencil_test; // Enter rasterization loop, starting at the center of the topleft bounding box corner. // TODO: Not sure if looping through x first might be faster @@ -410,7 +179,7 @@ static void ProcessTriangleInternal(const Shader::OutputVertex& v0, const Shader // Do not process the pixel if it's inside the scissor box and the scissor mode is set // to Exclude - if (regs.scissor_test.mode == Regs::ScissorMode::Exclude) { + if (regs.rasterizer.scissor_test.mode == RasterizerRegs::ScissorMode::Exclude) { if (x >= scissor_x1 && x < scissor_x2 && y >= scissor_y1 && y < scissor_y2) continue; } @@ -440,12 +209,14 @@ static void ProcessTriangleInternal(const Shader::OutputVertex& v0, const Shader // Not fully accurate. About 3 bits in precision are missing. // Z-Buffer (z / w * scale + offset) - float depth_scale = float24::FromRaw(regs.viewport_depth_range).ToFloat32(); - float depth_offset = float24::FromRaw(regs.viewport_depth_near_plane).ToFloat32(); + float depth_scale = float24::FromRaw(regs.rasterizer.viewport_depth_range).ToFloat32(); + float depth_offset = + float24::FromRaw(regs.rasterizer.viewport_depth_near_plane).ToFloat32(); float depth = interpolated_z_over_w * depth_scale + depth_offset; // Potentially switch to W-Buffer - if (regs.depthmap_enable == Pica::Regs::DepthBuffering::WBuffering) { + if (regs.rasterizer.depthmap_enable == + Pica::RasterizerRegs::DepthBuffering::WBuffering) { // W-Buffer (z * scale + w * offset = (z / w * scale + offset) * w) depth *= interpolated_w_inverse.ToFloat32() * wsum; } @@ -512,9 +283,9 @@ static void ProcessTriangleInternal(const Shader::OutputVertex& v0, const Shader // TODO: Refactor so cubemaps and shadowmaps can be handled if (i == 0) { switch (texture.config.type) { - case Regs::TextureConfig::Texture2D: + case TexturingRegs::TextureConfig::Texture2D: break; - case Regs::TextureConfig::Projection2D: { + case TexturingRegs::TextureConfig::Projection2D: { auto tc0_w = GetInterpolatedAttribute(v0.tc0_w, v1.tc0_w, v2.tc0_w); u /= tc0_w; v /= tc0_w; @@ -533,37 +304,9 @@ static void ProcessTriangleInternal(const Shader::OutputVertex& v0, const Shader int t = (int)(v * float24::FromFloat32(static_cast<float>(texture.config.height))) .ToFloat32(); - static auto GetWrappedTexCoord = [](Regs::TextureConfig::WrapMode mode, int val, - unsigned size) { - switch (mode) { - case Regs::TextureConfig::ClampToEdge: - val = std::max(val, 0); - val = std::min(val, (int)size - 1); - return val; - - case Regs::TextureConfig::ClampToBorder: - return val; - - case Regs::TextureConfig::Repeat: - return (int)((unsigned)val % size); - - case Regs::TextureConfig::MirroredRepeat: { - unsigned int coord = ((unsigned)val % (2 * size)); - if (coord >= size) - coord = 2 * size - 1 - coord; - return (int)coord; - } - - default: - LOG_ERROR(HW_GPU, "Unknown texture coordinate wrapping mode %x", (int)mode); - UNIMPLEMENTED(); - return 0; - } - }; - - if ((texture.config.wrap_s == Regs::TextureConfig::ClampToBorder && + if ((texture.config.wrap_s == TexturingRegs::TextureConfig::ClampToBorder && (s < 0 || static_cast<u32>(s) >= texture.config.width)) || - (texture.config.wrap_t == Regs::TextureConfig::ClampToBorder && + (texture.config.wrap_t == TexturingRegs::TextureConfig::ClampToBorder && (t < 0 || static_cast<u32>(t) >= texture.config.height))) { auto border_color = texture.config.border_color; texture_color[i] = {border_color.r, border_color.g, border_color.b, @@ -579,10 +322,10 @@ static void ProcessTriangleInternal(const Shader::OutputVertex& v0, const Shader u8* texture_data = Memory::GetPhysicalPointer(texture.config.GetPhysicalAddress()); auto info = - DebugUtils::TextureInfo::FromPicaRegister(texture.config, texture.format); + Texture::TextureInfo::FromPicaRegister(texture.config, texture.format); // TODO: Apply the min and mag filters to the texture - texture_color[i] = DebugUtils::LookupTexture(texture_data, s, t, info); + texture_color[i] = Texture::LookupTexture(texture_data, s, t, info); #if PICA_DUMP_TEXTURES DebugUtils::DumpTexture(texture.config, texture_data); #endif @@ -599,17 +342,16 @@ static void ProcessTriangleInternal(const Shader::OutputVertex& v0, const Shader Math::Vec4<u8> combiner_output; Math::Vec4<u8> combiner_buffer = {0, 0, 0, 0}; Math::Vec4<u8> next_combiner_buffer = { - regs.tev_combiner_buffer_color.r, regs.tev_combiner_buffer_color.g, - regs.tev_combiner_buffer_color.b, regs.tev_combiner_buffer_color.a, + regs.texturing.tev_combiner_buffer_color.r, + regs.texturing.tev_combiner_buffer_color.g, + regs.texturing.tev_combiner_buffer_color.b, + regs.texturing.tev_combiner_buffer_color.a, }; for (unsigned tev_stage_index = 0; tev_stage_index < tev_stages.size(); ++tev_stage_index) { const auto& tev_stage = tev_stages[tev_stage_index]; - using Source = Regs::TevStageConfig::Source; - using ColorModifier = Regs::TevStageConfig::ColorModifier; - using AlphaModifier = Regs::TevStageConfig::AlphaModifier; - using Operation = Regs::TevStageConfig::Operation; + using Source = TexturingRegs::TevStageConfig::Source; auto GetSource = [&](Source source) -> Math::Vec4<u8> { switch (source) { @@ -649,187 +391,6 @@ static void ProcessTriangleInternal(const Shader::OutputVertex& v0, const Shader } }; - static auto GetColorModifier = [](ColorModifier factor, - const Math::Vec4<u8>& values) -> Math::Vec3<u8> { - switch (factor) { - case ColorModifier::SourceColor: - return values.rgb(); - - case ColorModifier::OneMinusSourceColor: - return (Math::Vec3<u8>(255, 255, 255) - values.rgb()).Cast<u8>(); - - case ColorModifier::SourceAlpha: - return values.aaa(); - - case ColorModifier::OneMinusSourceAlpha: - return (Math::Vec3<u8>(255, 255, 255) - values.aaa()).Cast<u8>(); - - case ColorModifier::SourceRed: - return values.rrr(); - - case ColorModifier::OneMinusSourceRed: - return (Math::Vec3<u8>(255, 255, 255) - values.rrr()).Cast<u8>(); - - case ColorModifier::SourceGreen: - return values.ggg(); - - case ColorModifier::OneMinusSourceGreen: - return (Math::Vec3<u8>(255, 255, 255) - values.ggg()).Cast<u8>(); - - case ColorModifier::SourceBlue: - return values.bbb(); - - case ColorModifier::OneMinusSourceBlue: - return (Math::Vec3<u8>(255, 255, 255) - values.bbb()).Cast<u8>(); - } - }; - - static auto GetAlphaModifier = [](AlphaModifier factor, - const Math::Vec4<u8>& values) -> u8 { - switch (factor) { - case AlphaModifier::SourceAlpha: - return values.a(); - - case AlphaModifier::OneMinusSourceAlpha: - return 255 - values.a(); - - case AlphaModifier::SourceRed: - return values.r(); - - case AlphaModifier::OneMinusSourceRed: - return 255 - values.r(); - - case AlphaModifier::SourceGreen: - return values.g(); - - case AlphaModifier::OneMinusSourceGreen: - return 255 - values.g(); - - case AlphaModifier::SourceBlue: - return values.b(); - - case AlphaModifier::OneMinusSourceBlue: - return 255 - values.b(); - } - }; - - static auto ColorCombine = [](Operation op, - const Math::Vec3<u8> input[3]) -> Math::Vec3<u8> { - switch (op) { - case Operation::Replace: - return input[0]; - - case Operation::Modulate: - return ((input[0] * input[1]) / 255).Cast<u8>(); - - case Operation::Add: { - auto result = input[0] + input[1]; - result.r() = std::min(255, result.r()); - result.g() = std::min(255, result.g()); - result.b() = std::min(255, result.b()); - return result.Cast<u8>(); - } - - case Operation::AddSigned: { - // TODO(bunnei): Verify that the color conversion from (float) 0.5f to - // (byte) 128 is correct - auto result = input[0].Cast<int>() + input[1].Cast<int>() - - Math::MakeVec<int>(128, 128, 128); - result.r() = MathUtil::Clamp<int>(result.r(), 0, 255); - result.g() = MathUtil::Clamp<int>(result.g(), 0, 255); - result.b() = MathUtil::Clamp<int>(result.b(), 0, 255); - return result.Cast<u8>(); - } - - case Operation::Lerp: - return ((input[0] * input[2] + - input[1] * - (Math::MakeVec<u8>(255, 255, 255) - input[2]).Cast<u8>()) / - 255) - .Cast<u8>(); - - case Operation::Subtract: { - auto result = input[0].Cast<int>() - input[1].Cast<int>(); - result.r() = std::max(0, result.r()); - result.g() = std::max(0, result.g()); - result.b() = std::max(0, result.b()); - return result.Cast<u8>(); - } - - case Operation::MultiplyThenAdd: { - auto result = (input[0] * input[1] + 255 * input[2].Cast<int>()) / 255; - result.r() = std::min(255, result.r()); - result.g() = std::min(255, result.g()); - result.b() = std::min(255, result.b()); - return result.Cast<u8>(); - } - - case Operation::AddThenMultiply: { - auto result = input[0] + input[1]; - result.r() = std::min(255, result.r()); - result.g() = std::min(255, result.g()); - result.b() = std::min(255, result.b()); - result = (result * input[2].Cast<int>()) / 255; - return result.Cast<u8>(); - } - case Operation::Dot3_RGB: { - // Not fully accurate. - // Worst case scenario seems to yield a +/-3 error - // Some HW results indicate that the per-component computation can't have a - // higher precision than 1/256, - // while dot3_rgb( (0x80,g0,b0),(0x7F,g1,b1) ) and dot3_rgb( - // (0x80,g0,b0),(0x80,g1,b1) ) give different results - int result = - ((input[0].r() * 2 - 255) * (input[1].r() * 2 - 255) + 128) / 256 + - ((input[0].g() * 2 - 255) * (input[1].g() * 2 - 255) + 128) / 256 + - ((input[0].b() * 2 - 255) * (input[1].b() * 2 - 255) + 128) / 256; - result = std::max(0, std::min(255, result)); - return {(u8)result, (u8)result, (u8)result}; - } - default: - LOG_ERROR(HW_GPU, "Unknown color combiner operation %d", (int)op); - UNIMPLEMENTED(); - return {0, 0, 0}; - } - }; - - static auto AlphaCombine = [](Operation op, const std::array<u8, 3>& input) -> u8 { - switch (op) { - case Operation::Replace: - return input[0]; - - case Operation::Modulate: - return input[0] * input[1] / 255; - - case Operation::Add: - return std::min(255, input[0] + input[1]); - - case Operation::AddSigned: { - // TODO(bunnei): Verify that the color conversion from (float) 0.5f to - // (byte) 128 is correct - auto result = static_cast<int>(input[0]) + static_cast<int>(input[1]) - 128; - return static_cast<u8>(MathUtil::Clamp<int>(result, 0, 255)); - } - - case Operation::Lerp: - return (input[0] * input[2] + input[1] * (255 - input[2])) / 255; - - case Operation::Subtract: - return std::max(0, (int)input[0] - (int)input[1]); - - case Operation::MultiplyThenAdd: - return std::min(255, (input[0] * input[1] + 255 * input[2]) / 255); - - case Operation::AddThenMultiply: - return (std::min(255, (input[0] + input[1])) * input[2]) / 255; - - default: - LOG_ERROR(HW_GPU, "Unknown alpha combiner operation %d", (int)op); - UNIMPLEMENTED(); - return 0; - } - }; - // color combiner // NOTE: Not sure if the alpha combiner might use the color output of the previous // stage as input. Hence, we currently don't directly write the result to @@ -861,54 +422,54 @@ static void ProcessTriangleInternal(const Shader::OutputVertex& v0, const Shader combiner_buffer = next_combiner_buffer; - if (regs.tev_combiner_buffer_input.TevStageUpdatesCombinerBufferColor( + if (regs.texturing.tev_combiner_buffer_input.TevStageUpdatesCombinerBufferColor( tev_stage_index)) { next_combiner_buffer.r() = combiner_output.r(); next_combiner_buffer.g() = combiner_output.g(); next_combiner_buffer.b() = combiner_output.b(); } - if (regs.tev_combiner_buffer_input.TevStageUpdatesCombinerBufferAlpha( + if (regs.texturing.tev_combiner_buffer_input.TevStageUpdatesCombinerBufferAlpha( tev_stage_index)) { next_combiner_buffer.a() = combiner_output.a(); } } - const auto& output_merger = regs.output_merger; + const auto& output_merger = regs.framebuffer.output_merger; // TODO: Does alpha testing happen before or after stencil? if (output_merger.alpha_test.enable) { bool pass = false; switch (output_merger.alpha_test.func) { - case Regs::CompareFunc::Never: + case FramebufferRegs::CompareFunc::Never: pass = false; break; - case Regs::CompareFunc::Always: + case FramebufferRegs::CompareFunc::Always: pass = true; break; - case Regs::CompareFunc::Equal: + case FramebufferRegs::CompareFunc::Equal: pass = combiner_output.a() == output_merger.alpha_test.ref; break; - case Regs::CompareFunc::NotEqual: + case FramebufferRegs::CompareFunc::NotEqual: pass = combiner_output.a() != output_merger.alpha_test.ref; break; - case Regs::CompareFunc::LessThan: + case FramebufferRegs::CompareFunc::LessThan: pass = combiner_output.a() < output_merger.alpha_test.ref; break; - case Regs::CompareFunc::LessThanOrEqual: + case FramebufferRegs::CompareFunc::LessThanOrEqual: pass = combiner_output.a() <= output_merger.alpha_test.ref; break; - case Regs::CompareFunc::GreaterThan: + case FramebufferRegs::CompareFunc::GreaterThan: pass = combiner_output.a() > output_merger.alpha_test.ref; break; - case Regs::CompareFunc::GreaterThanOrEqual: + case FramebufferRegs::CompareFunc::GreaterThanOrEqual: pass = combiner_output.a() >= output_merger.alpha_test.ref; break; } @@ -921,16 +482,16 @@ static void ProcessTriangleInternal(const Shader::OutputVertex& v0, const Shader // Not fully accurate. We'd have to know what data type is used to // store the depth etc. Using float for now until we know more // about Pica datatypes - if (regs.fog_mode == Regs::FogMode::Fog) { + if (regs.texturing.fog_mode == TexturingRegs::FogMode::Fog) { const Math::Vec3<u8> fog_color = { - static_cast<u8>(regs.fog_color.r.Value()), - static_cast<u8>(regs.fog_color.g.Value()), - static_cast<u8>(regs.fog_color.b.Value()), + static_cast<u8>(regs.texturing.fog_color.r.Value()), + static_cast<u8>(regs.texturing.fog_color.g.Value()), + static_cast<u8>(regs.texturing.fog_color.b.Value()), }; // Get index into fog LUT float fog_index; - if (g_state.regs.fog_flip) { + if (g_state.regs.texturing.fog_flip) { fog_index = (1.0f - depth) * 128.0f; } else { fog_index = depth * 128.0f; @@ -954,10 +515,10 @@ static void ProcessTriangleInternal(const Shader::OutputVertex& v0, const Shader u8 old_stencil = 0; auto UpdateStencil = [stencil_test, x, y, - &old_stencil](Pica::Regs::StencilAction action) { + &old_stencil](Pica::FramebufferRegs::StencilAction action) { u8 new_stencil = PerformStencilAction(action, old_stencil, stencil_test.reference_value); - if (g_state.regs.framebuffer.allow_depth_stencil_write != 0) + if (g_state.regs.framebuffer.framebuffer.allow_depth_stencil_write != 0) SetStencil(x >> 4, y >> 4, (new_stencil & stencil_test.write_mask) | (old_stencil & ~stencil_test.write_mask)); }; @@ -969,35 +530,35 @@ static void ProcessTriangleInternal(const Shader::OutputVertex& v0, const Shader bool pass = false; switch (stencil_test.func) { - case Regs::CompareFunc::Never: + case FramebufferRegs::CompareFunc::Never: pass = false; break; - case Regs::CompareFunc::Always: + case FramebufferRegs::CompareFunc::Always: pass = true; break; - case Regs::CompareFunc::Equal: + case FramebufferRegs::CompareFunc::Equal: pass = (ref == dest); break; - case Regs::CompareFunc::NotEqual: + case FramebufferRegs::CompareFunc::NotEqual: pass = (ref != dest); break; - case Regs::CompareFunc::LessThan: + case FramebufferRegs::CompareFunc::LessThan: pass = (ref < dest); break; - case Regs::CompareFunc::LessThanOrEqual: + case FramebufferRegs::CompareFunc::LessThanOrEqual: pass = (ref <= dest); break; - case Regs::CompareFunc::GreaterThan: + case FramebufferRegs::CompareFunc::GreaterThan: pass = (ref > dest); break; - case Regs::CompareFunc::GreaterThanOrEqual: + case FramebufferRegs::CompareFunc::GreaterThanOrEqual: pass = (ref >= dest); break; } @@ -1009,7 +570,8 @@ static void ProcessTriangleInternal(const Shader::OutputVertex& v0, const Shader } // Convert float to integer - unsigned num_bits = Regs::DepthBitsPerPixel(regs.framebuffer.depth_format); + unsigned num_bits = + FramebufferRegs::DepthBitsPerPixel(regs.framebuffer.framebuffer.depth_format); u32 z = (u32)(depth * ((1 << num_bits) - 1)); if (output_merger.depth_test_enable) { @@ -1018,35 +580,35 @@ static void ProcessTriangleInternal(const Shader::OutputVertex& v0, const Shader bool pass = false; switch (output_merger.depth_test_func) { - case Regs::CompareFunc::Never: + case FramebufferRegs::CompareFunc::Never: pass = false; break; - case Regs::CompareFunc::Always: + case FramebufferRegs::CompareFunc::Always: pass = true; break; - case Regs::CompareFunc::Equal: + case FramebufferRegs::CompareFunc::Equal: pass = z == ref_z; break; - case Regs::CompareFunc::NotEqual: + case FramebufferRegs::CompareFunc::NotEqual: pass = z != ref_z; break; - case Regs::CompareFunc::LessThan: + case FramebufferRegs::CompareFunc::LessThan: pass = z < ref_z; break; - case Regs::CompareFunc::LessThanOrEqual: + case FramebufferRegs::CompareFunc::LessThanOrEqual: pass = z <= ref_z; break; - case Regs::CompareFunc::GreaterThan: + case FramebufferRegs::CompareFunc::GreaterThan: pass = z > ref_z; break; - case Regs::CompareFunc::GreaterThanOrEqual: + case FramebufferRegs::CompareFunc::GreaterThanOrEqual: pass = z >= ref_z; break; } @@ -1058,8 +620,11 @@ static void ProcessTriangleInternal(const Shader::OutputVertex& v0, const Shader } } - if (regs.framebuffer.allow_depth_stencil_write != 0 && output_merger.depth_write_enable) + if (regs.framebuffer.framebuffer.allow_depth_stencil_write != 0 && + output_merger.depth_write_enable) { + SetDepth(x >> 4, y >> 4, z); + } // The stencil depth_pass action is executed even if depth testing is disabled if (stencil_action_enable) @@ -1071,7 +636,8 @@ static void ProcessTriangleInternal(const Shader::OutputVertex& v0, const Shader if (output_merger.alphablend_enable) { auto params = output_merger.alpha_blending; - auto LookupFactor = [&](unsigned channel, Regs::BlendFactor factor) -> u8 { + auto LookupFactor = [&](unsigned channel, + FramebufferRegs::BlendFactor factor) -> u8 { DEBUG_ASSERT(channel < 4); const Math::Vec4<u8> blend_const = { @@ -1082,49 +648,49 @@ static void ProcessTriangleInternal(const Shader::OutputVertex& v0, const Shader }; switch (factor) { - case Regs::BlendFactor::Zero: + case FramebufferRegs::BlendFactor::Zero: return 0; - case Regs::BlendFactor::One: + case FramebufferRegs::BlendFactor::One: return 255; - case Regs::BlendFactor::SourceColor: + case FramebufferRegs::BlendFactor::SourceColor: return combiner_output[channel]; - case Regs::BlendFactor::OneMinusSourceColor: + case FramebufferRegs::BlendFactor::OneMinusSourceColor: return 255 - combiner_output[channel]; - case Regs::BlendFactor::DestColor: + case FramebufferRegs::BlendFactor::DestColor: return dest[channel]; - case Regs::BlendFactor::OneMinusDestColor: + case FramebufferRegs::BlendFactor::OneMinusDestColor: return 255 - dest[channel]; - case Regs::BlendFactor::SourceAlpha: + case FramebufferRegs::BlendFactor::SourceAlpha: return combiner_output.a(); - case Regs::BlendFactor::OneMinusSourceAlpha: + case FramebufferRegs::BlendFactor::OneMinusSourceAlpha: return 255 - combiner_output.a(); - case Regs::BlendFactor::DestAlpha: + case FramebufferRegs::BlendFactor::DestAlpha: return dest.a(); - case Regs::BlendFactor::OneMinusDestAlpha: + case FramebufferRegs::BlendFactor::OneMinusDestAlpha: return 255 - dest.a(); - case Regs::BlendFactor::ConstantColor: + case FramebufferRegs::BlendFactor::ConstantColor: return blend_const[channel]; - case Regs::BlendFactor::OneMinusConstantColor: + case FramebufferRegs::BlendFactor::OneMinusConstantColor: return 255 - blend_const[channel]; - case Regs::BlendFactor::ConstantAlpha: + case FramebufferRegs::BlendFactor::ConstantAlpha: return blend_const.a(); - case Regs::BlendFactor::OneMinusConstantAlpha: + case FramebufferRegs::BlendFactor::OneMinusConstantAlpha: return 255 - blend_const.a(); - case Regs::BlendFactor::SourceAlphaSaturate: + case FramebufferRegs::BlendFactor::SourceAlphaSaturate: // Returns 1.0 for the alpha channel if (channel == 3) return 255; @@ -1139,55 +705,6 @@ static void ProcessTriangleInternal(const Shader::OutputVertex& v0, const Shader return combiner_output[channel]; }; - static auto EvaluateBlendEquation = []( - const Math::Vec4<u8>& src, const Math::Vec4<u8>& srcfactor, - const Math::Vec4<u8>& dest, const Math::Vec4<u8>& destfactor, - Regs::BlendEquation equation) { - Math::Vec4<int> result; - - auto src_result = (src * srcfactor).Cast<int>(); - auto dst_result = (dest * destfactor).Cast<int>(); - - switch (equation) { - case Regs::BlendEquation::Add: - result = (src_result + dst_result) / 255; - break; - - case Regs::BlendEquation::Subtract: - result = (src_result - dst_result) / 255; - break; - - case Regs::BlendEquation::ReverseSubtract: - result = (dst_result - src_result) / 255; - break; - - // TODO: How do these two actually work? - // OpenGL doesn't include the blend factors in the min/max computations, - // but is this what the 3DS actually does? - case Regs::BlendEquation::Min: - result.r() = std::min(src.r(), dest.r()); - result.g() = std::min(src.g(), dest.g()); - result.b() = std::min(src.b(), dest.b()); - result.a() = std::min(src.a(), dest.a()); - break; - - case Regs::BlendEquation::Max: - result.r() = std::max(src.r(), dest.r()); - result.g() = std::max(src.g(), dest.g()); - result.b() = std::max(src.b(), dest.b()); - result.a() = std::max(src.a(), dest.a()); - break; - - default: - LOG_CRITICAL(HW_GPU, "Unknown RGB blend equation %x", equation); - UNIMPLEMENTED(); - } - - return Math::Vec4<u8>( - MathUtil::Clamp(result.r(), 0, 255), MathUtil::Clamp(result.g(), 0, 255), - MathUtil::Clamp(result.b(), 0, 255), MathUtil::Clamp(result.a(), 0, 255)); - }; - auto srcfactor = Math::MakeVec(LookupFactor(0, params.factor_source_rgb), LookupFactor(1, params.factor_source_rgb), LookupFactor(2, params.factor_source_rgb), @@ -1204,58 +721,6 @@ static void ProcessTriangleInternal(const Shader::OutputVertex& v0, const Shader dstfactor, params.blend_equation_a) .a(); } else { - static auto LogicOp = [](u8 src, u8 dest, Regs::LogicOp op) -> u8 { - switch (op) { - case Regs::LogicOp::Clear: - return 0; - - case Regs::LogicOp::And: - return src & dest; - - case Regs::LogicOp::AndReverse: - return src & ~dest; - - case Regs::LogicOp::Copy: - return src; - - case Regs::LogicOp::Set: - return 255; - - case Regs::LogicOp::CopyInverted: - return ~src; - - case Regs::LogicOp::NoOp: - return dest; - - case Regs::LogicOp::Invert: - return ~dest; - - case Regs::LogicOp::Nand: - return ~(src & dest); - - case Regs::LogicOp::Or: - return src | dest; - - case Regs::LogicOp::Nor: - return ~(src | dest); - - case Regs::LogicOp::Xor: - return src ^ dest; - - case Regs::LogicOp::Equiv: - return ~(src ^ dest); - - case Regs::LogicOp::AndInverted: - return ~src & dest; - - case Regs::LogicOp::OrReverse: - return src | ~dest; - - case Regs::LogicOp::OrInverted: - return ~src | dest; - } - }; - blend_output = Math::MakeVec(LogicOp(combiner_output.r(), dest.r(), output_merger.logic_op), LogicOp(combiner_output.g(), dest.g(), output_merger.logic_op), @@ -1270,14 +735,13 @@ static void ProcessTriangleInternal(const Shader::OutputVertex& v0, const Shader output_merger.alpha_enable ? blend_output.a() : dest.a(), }; - if (regs.framebuffer.allow_color_write != 0) + if (regs.framebuffer.framebuffer.allow_color_write != 0) DrawPixel(x >> 4, y >> 4, result); } } } -void ProcessTriangle(const Shader::OutputVertex& v0, const Shader::OutputVertex& v1, - const Shader::OutputVertex& v2) { +void ProcessTriangle(const Vertex& v0, const Vertex& v1, const Vertex& v2) { ProcessTriangleInternal(v0, v1, v2); } diff --git a/src/video_core/swrasterizer/rasterizer.h b/src/video_core/swrasterizer/rasterizer.h new file mode 100644 index 000000000..3a72ac343 --- /dev/null +++ b/src/video_core/swrasterizer/rasterizer.h @@ -0,0 +1,48 @@ +// Copyright 2014 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include "video_core/shader/shader.h" + +namespace Pica { + +namespace Rasterizer { + +struct Vertex : Shader::OutputVertex { + Vertex(const OutputVertex& v) : OutputVertex(v) {} + + // Attributes used to store intermediate results + // position after perspective divide + Math::Vec3<float24> screenpos; + + // Linear interpolation + // factor: 0=this, 1=vtx + void Lerp(float24 factor, const Vertex& vtx) { + pos = pos * factor + vtx.pos * (float24::FromFloat32(1) - factor); + + // TODO: Should perform perspective correct interpolation here... + tc0 = tc0 * factor + vtx.tc0 * (float24::FromFloat32(1) - factor); + tc1 = tc1 * factor + vtx.tc1 * (float24::FromFloat32(1) - factor); + tc2 = tc2 * factor + vtx.tc2 * (float24::FromFloat32(1) - factor); + + screenpos = screenpos * factor + vtx.screenpos * (float24::FromFloat32(1) - factor); + + color = color * factor + vtx.color * (float24::FromFloat32(1) - factor); + } + + // Linear interpolation + // factor: 0=v0, 1=v1 + static Vertex Lerp(float24 factor, const Vertex& v0, const Vertex& v1) { + Vertex ret = v0; + ret.Lerp(factor, v1); + return ret; + } +}; + +void ProcessTriangle(const Vertex& v0, const Vertex& v1, const Vertex& v2); + +} // namespace Rasterizer + +} // namespace Pica diff --git a/src/video_core/swrasterizer.cpp b/src/video_core/swrasterizer/swrasterizer.cpp index 9cd21f72b..402b705dd 100644 --- a/src/video_core/swrasterizer.cpp +++ b/src/video_core/swrasterizer/swrasterizer.cpp @@ -2,8 +2,8 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. -#include "video_core/clipper.h" -#include "video_core/swrasterizer.h" +#include "video_core/swrasterizer/clipper.h" +#include "video_core/swrasterizer/swrasterizer.h" namespace VideoCore { diff --git a/src/video_core/swrasterizer.h b/src/video_core/swrasterizer/swrasterizer.h index 6d42d7409..6d42d7409 100644 --- a/src/video_core/swrasterizer.h +++ b/src/video_core/swrasterizer/swrasterizer.h diff --git a/src/video_core/swrasterizer/texturing.cpp b/src/video_core/swrasterizer/texturing.cpp new file mode 100644 index 000000000..eb18e4ba4 --- /dev/null +++ b/src/video_core/swrasterizer/texturing.cpp @@ -0,0 +1,228 @@ +// Copyright 2017 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <algorithm> + +#include "common/assert.h" +#include "common/common_types.h" +#include "common/math_util.h" +#include "common/vector_math.h" +#include "video_core/regs_texturing.h" +#include "video_core/swrasterizer/texturing.h" + +namespace Pica { +namespace Rasterizer { + +using TevStageConfig = TexturingRegs::TevStageConfig; + +int GetWrappedTexCoord(TexturingRegs::TextureConfig::WrapMode mode, int val, unsigned size) { + switch (mode) { + case TexturingRegs::TextureConfig::ClampToEdge: + val = std::max(val, 0); + val = std::min(val, (int)size - 1); + return val; + + case TexturingRegs::TextureConfig::ClampToBorder: + return val; + + case TexturingRegs::TextureConfig::Repeat: + return (int)((unsigned)val % size); + + case TexturingRegs::TextureConfig::MirroredRepeat: { + unsigned int coord = ((unsigned)val % (2 * size)); + if (coord >= size) + coord = 2 * size - 1 - coord; + return (int)coord; + } + + default: + LOG_ERROR(HW_GPU, "Unknown texture coordinate wrapping mode %x", (int)mode); + UNIMPLEMENTED(); + return 0; + } +}; + +Math::Vec3<u8> GetColorModifier(TevStageConfig::ColorModifier factor, + const Math::Vec4<u8>& values) { + using ColorModifier = TevStageConfig::ColorModifier; + + switch (factor) { + case ColorModifier::SourceColor: + return values.rgb(); + + case ColorModifier::OneMinusSourceColor: + return (Math::Vec3<u8>(255, 255, 255) - values.rgb()).Cast<u8>(); + + case ColorModifier::SourceAlpha: + return values.aaa(); + + case ColorModifier::OneMinusSourceAlpha: + return (Math::Vec3<u8>(255, 255, 255) - values.aaa()).Cast<u8>(); + + case ColorModifier::SourceRed: + return values.rrr(); + + case ColorModifier::OneMinusSourceRed: + return (Math::Vec3<u8>(255, 255, 255) - values.rrr()).Cast<u8>(); + + case ColorModifier::SourceGreen: + return values.ggg(); + + case ColorModifier::OneMinusSourceGreen: + return (Math::Vec3<u8>(255, 255, 255) - values.ggg()).Cast<u8>(); + + case ColorModifier::SourceBlue: + return values.bbb(); + + case ColorModifier::OneMinusSourceBlue: + return (Math::Vec3<u8>(255, 255, 255) - values.bbb()).Cast<u8>(); + } +}; + +u8 GetAlphaModifier(TevStageConfig::AlphaModifier factor, const Math::Vec4<u8>& values) { + using AlphaModifier = TevStageConfig::AlphaModifier; + + switch (factor) { + case AlphaModifier::SourceAlpha: + return values.a(); + + case AlphaModifier::OneMinusSourceAlpha: + return 255 - values.a(); + + case AlphaModifier::SourceRed: + return values.r(); + + case AlphaModifier::OneMinusSourceRed: + return 255 - values.r(); + + case AlphaModifier::SourceGreen: + return values.g(); + + case AlphaModifier::OneMinusSourceGreen: + return 255 - values.g(); + + case AlphaModifier::SourceBlue: + return values.b(); + + case AlphaModifier::OneMinusSourceBlue: + return 255 - values.b(); + } +}; + +Math::Vec3<u8> ColorCombine(TevStageConfig::Operation op, const Math::Vec3<u8> input[3]) { + using Operation = TevStageConfig::Operation; + + switch (op) { + case Operation::Replace: + return input[0]; + + case Operation::Modulate: + return ((input[0] * input[1]) / 255).Cast<u8>(); + + case Operation::Add: { + auto result = input[0] + input[1]; + result.r() = std::min(255, result.r()); + result.g() = std::min(255, result.g()); + result.b() = std::min(255, result.b()); + return result.Cast<u8>(); + } + + case Operation::AddSigned: { + // TODO(bunnei): Verify that the color conversion from (float) 0.5f to + // (byte) 128 is correct + auto result = + input[0].Cast<int>() + input[1].Cast<int>() - Math::MakeVec<int>(128, 128, 128); + result.r() = MathUtil::Clamp<int>(result.r(), 0, 255); + result.g() = MathUtil::Clamp<int>(result.g(), 0, 255); + result.b() = MathUtil::Clamp<int>(result.b(), 0, 255); + return result.Cast<u8>(); + } + + case Operation::Lerp: + return ((input[0] * input[2] + + input[1] * (Math::MakeVec<u8>(255, 255, 255) - input[2]).Cast<u8>()) / + 255) + .Cast<u8>(); + + case Operation::Subtract: { + auto result = input[0].Cast<int>() - input[1].Cast<int>(); + result.r() = std::max(0, result.r()); + result.g() = std::max(0, result.g()); + result.b() = std::max(0, result.b()); + return result.Cast<u8>(); + } + + case Operation::MultiplyThenAdd: { + auto result = (input[0] * input[1] + 255 * input[2].Cast<int>()) / 255; + result.r() = std::min(255, result.r()); + result.g() = std::min(255, result.g()); + result.b() = std::min(255, result.b()); + return result.Cast<u8>(); + } + + case Operation::AddThenMultiply: { + auto result = input[0] + input[1]; + result.r() = std::min(255, result.r()); + result.g() = std::min(255, result.g()); + result.b() = std::min(255, result.b()); + result = (result * input[2].Cast<int>()) / 255; + return result.Cast<u8>(); + } + case Operation::Dot3_RGB: { + // Not fully accurate. Worst case scenario seems to yield a +/-3 error. Some HW results + // indicate that the per-component computation can't have a higher precision than 1/256, + // while dot3_rgb((0x80,g0,b0), (0x7F,g1,b1)) and dot3_rgb((0x80,g0,b0), (0x80,g1,b1)) give + // different results. + int result = ((input[0].r() * 2 - 255) * (input[1].r() * 2 - 255) + 128) / 256 + + ((input[0].g() * 2 - 255) * (input[1].g() * 2 - 255) + 128) / 256 + + ((input[0].b() * 2 - 255) * (input[1].b() * 2 - 255) + 128) / 256; + result = std::max(0, std::min(255, result)); + return {(u8)result, (u8)result, (u8)result}; + } + default: + LOG_ERROR(HW_GPU, "Unknown color combiner operation %d", (int)op); + UNIMPLEMENTED(); + return {0, 0, 0}; + } +}; + +u8 AlphaCombine(TevStageConfig::Operation op, const std::array<u8, 3>& input) { + switch (op) { + using Operation = TevStageConfig::Operation; + case Operation::Replace: + return input[0]; + + case Operation::Modulate: + return input[0] * input[1] / 255; + + case Operation::Add: + return std::min(255, input[0] + input[1]); + + case Operation::AddSigned: { + // TODO(bunnei): Verify that the color conversion from (float) 0.5f to (byte) 128 is correct + auto result = static_cast<int>(input[0]) + static_cast<int>(input[1]) - 128; + return static_cast<u8>(MathUtil::Clamp<int>(result, 0, 255)); + } + + case Operation::Lerp: + return (input[0] * input[2] + input[1] * (255 - input[2])) / 255; + + case Operation::Subtract: + return std::max(0, (int)input[0] - (int)input[1]); + + case Operation::MultiplyThenAdd: + return std::min(255, (input[0] * input[1] + 255 * input[2]) / 255); + + case Operation::AddThenMultiply: + return (std::min(255, (input[0] + input[1])) * input[2]) / 255; + + default: + LOG_ERROR(HW_GPU, "Unknown alpha combiner operation %d", (int)op); + UNIMPLEMENTED(); + return 0; + } +}; + +} // namespace Rasterizer +} // namespace Pica diff --git a/src/video_core/swrasterizer/texturing.h b/src/video_core/swrasterizer/texturing.h new file mode 100644 index 000000000..24f74a5a3 --- /dev/null +++ b/src/video_core/swrasterizer/texturing.h @@ -0,0 +1,28 @@ +// Copyright 2017 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 "common/vector_math.h" +#include "video_core/regs_texturing.h" + +namespace Pica { +namespace Rasterizer { + +int GetWrappedTexCoord(TexturingRegs::TextureConfig::WrapMode mode, int val, unsigned size); + +Math::Vec3<u8> GetColorModifier(TexturingRegs::TevStageConfig::ColorModifier factor, + const Math::Vec4<u8>& values); + +u8 GetAlphaModifier(TexturingRegs::TevStageConfig::AlphaModifier factor, + const Math::Vec4<u8>& values); + +Math::Vec3<u8> ColorCombine(TexturingRegs::TevStageConfig::Operation op, + const Math::Vec3<u8> input[3]); + +u8 AlphaCombine(TexturingRegs::TevStageConfig::Operation op, const std::array<u8, 3>& input); + +} // namespace Rasterizer +} // namespace Pica diff --git a/src/video_core/texture/etc1.cpp b/src/video_core/texture/etc1.cpp new file mode 100644 index 000000000..43f7f56db --- /dev/null +++ b/src/video_core/texture/etc1.cpp @@ -0,0 +1,122 @@ +// Copyright 2017 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <array> +#include "common/bit_field.h" +#include "common/color.h" +#include "common/common_types.h" +#include "common/math_util.h" +#include "common/vector_math.h" +#include "video_core/texture/etc1.h" + +namespace Pica { +namespace Texture { + +namespace { + +constexpr std::array<std::array<u8, 2>, 8> etc1_modifier_table = {{ + {2, 8}, {5, 17}, {9, 29}, {13, 42}, {18, 60}, {24, 80}, {33, 106}, {47, 183}, +}}; + +union ETC1Tile { + u64 raw; + + // Each of these two is a collection of 16 bits (one per lookup value) + BitField<0, 16, u64> table_subindexes; + BitField<16, 16, u64> negation_flags; + + unsigned GetTableSubIndex(unsigned index) const { + return (table_subindexes >> index) & 1; + } + + bool GetNegationFlag(unsigned index) const { + return ((negation_flags >> index) & 1) == 1; + } + + BitField<32, 1, u64> flip; + BitField<33, 1, u64> differential_mode; + + BitField<34, 3, u64> table_index_2; + BitField<37, 3, u64> table_index_1; + + union { + // delta value + base value + BitField<40, 3, s64> db; + BitField<43, 5, u64> b; + + BitField<48, 3, s64> dg; + BitField<51, 5, u64> g; + + BitField<56, 3, s64> dr; + BitField<59, 5, u64> r; + } differential; + + union { + BitField<40, 4, u64> b2; + BitField<44, 4, u64> b1; + + BitField<48, 4, u64> g2; + BitField<52, 4, u64> g1; + + BitField<56, 4, u64> r2; + BitField<60, 4, u64> r1; + } separate; + + const Math::Vec3<u8> GetRGB(unsigned int x, unsigned int y) const { + int texel = 4 * x + y; + + if (flip) + std::swap(x, y); + + // Lookup base value + Math::Vec3<int> ret; + if (differential_mode) { + ret.r() = static_cast<int>(differential.r); + ret.g() = static_cast<int>(differential.g); + ret.b() = static_cast<int>(differential.b); + if (x >= 2) { + ret.r() += static_cast<int>(differential.dr); + ret.g() += static_cast<int>(differential.dg); + ret.b() += static_cast<int>(differential.db); + } + ret.r() = Color::Convert5To8(ret.r()); + ret.g() = Color::Convert5To8(ret.g()); + ret.b() = Color::Convert5To8(ret.b()); + } else { + if (x < 2) { + ret.r() = Color::Convert4To8(static_cast<u8>(separate.r1)); + ret.g() = Color::Convert4To8(static_cast<u8>(separate.g1)); + ret.b() = Color::Convert4To8(static_cast<u8>(separate.b1)); + } else { + ret.r() = Color::Convert4To8(static_cast<u8>(separate.r2)); + ret.g() = Color::Convert4To8(static_cast<u8>(separate.g2)); + ret.b() = Color::Convert4To8(static_cast<u8>(separate.b2)); + } + } + + // Add modifier + unsigned table_index = + static_cast<int>((x < 2) ? table_index_1.Value() : table_index_2.Value()); + + int modifier = etc1_modifier_table[table_index][GetTableSubIndex(texel)]; + if (GetNegationFlag(texel)) + modifier *= -1; + + ret.r() = MathUtil::Clamp(ret.r() + modifier, 0, 255); + ret.g() = MathUtil::Clamp(ret.g() + modifier, 0, 255); + ret.b() = MathUtil::Clamp(ret.b() + modifier, 0, 255); + + return ret.Cast<u8>(); + } +}; + +} // anonymous namespace + +Math::Vec3<u8> SampleETC1Subtile(u64 value, unsigned int x, unsigned int y) { + ETC1Tile tile{value}; + return tile.GetRGB(x, y); +} + +} // namespace Texture +} // namespace Pica diff --git a/src/video_core/texture/etc1.h b/src/video_core/texture/etc1.h new file mode 100644 index 000000000..e188b19df --- /dev/null +++ b/src/video_core/texture/etc1.h @@ -0,0 +1,16 @@ +// Copyright 2017 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 "common/vector_math.h" + +namespace Pica { +namespace Texture { + +Math::Vec3<u8> SampleETC1Subtile(u64 value, unsigned int x, unsigned int y); + +} // namespace Texture +} // namespace Pica diff --git a/src/video_core/texture/texture_decode.cpp b/src/video_core/texture/texture_decode.cpp new file mode 100644 index 000000000..0818d652c --- /dev/null +++ b/src/video_core/texture/texture_decode.cpp @@ -0,0 +1,227 @@ +// Copyright 2017 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include "common/assert.h" +#include "common/color.h" +#include "common/logging/log.h" +#include "common/math_util.h" +#include "common/swap.h" +#include "common/vector_math.h" +#include "video_core/regs_texturing.h" +#include "video_core/texture/etc1.h" +#include "video_core/texture/texture_decode.h" +#include "video_core/utils.h" + +using TextureFormat = Pica::TexturingRegs::TextureFormat; + +namespace Pica { +namespace Texture { + +constexpr size_t TILE_SIZE = 8 * 8; +constexpr size_t ETC1_SUBTILES = 2 * 2; + +size_t CalculateTileSize(TextureFormat format) { + switch (format) { + case TextureFormat::RGBA8: + return 4 * TILE_SIZE; + + case TextureFormat::RGB8: + return 3 * TILE_SIZE; + + case TextureFormat::RGB5A1: + case TextureFormat::RGB565: + case TextureFormat::RGBA4: + case TextureFormat::IA8: + case TextureFormat::RG8: + return 2 * TILE_SIZE; + + case TextureFormat::I8: + case TextureFormat::A8: + case TextureFormat::IA4: + return 1 * TILE_SIZE; + + case TextureFormat::I4: + case TextureFormat::A4: + return TILE_SIZE / 2; + + case TextureFormat::ETC1: + return ETC1_SUBTILES * 8; + + case TextureFormat::ETC1A4: + return ETC1_SUBTILES * 16; + + default: // placeholder for yet unknown formats + UNIMPLEMENTED(); + return 0; + } +} + +Math::Vec4<u8> LookupTexture(const u8* source, unsigned int x, unsigned int y, + const TextureInfo& info, bool disable_alpha) { + // Coordinate in tiles + const unsigned int coarse_x = x / 8; + const unsigned int coarse_y = y / 8; + + // Coordinate inside the tile + const unsigned int fine_x = x % 8; + const unsigned int fine_y = y % 8; + + const u8* line = source + coarse_y * info.stride; + const u8* tile = line + coarse_x * CalculateTileSize(info.format); + return LookupTexelInTile(tile, fine_x, fine_y, info, disable_alpha); +} + +Math::Vec4<u8> LookupTexelInTile(const u8* source, unsigned int x, unsigned int y, + const TextureInfo& info, bool disable_alpha) { + DEBUG_ASSERT(x < 8); + DEBUG_ASSERT(y < 8); + + using VideoCore::MortonInterleave; + + switch (info.format) { + case TextureFormat::RGBA8: { + auto res = Color::DecodeRGBA8(source + MortonInterleave(x, y) * 4); + return {res.r(), res.g(), res.b(), static_cast<u8>(disable_alpha ? 255 : res.a())}; + } + + case TextureFormat::RGB8: { + auto res = Color::DecodeRGB8(source + MortonInterleave(x, y) * 3); + return {res.r(), res.g(), res.b(), 255}; + } + + case TextureFormat::RGB5A1: { + auto res = Color::DecodeRGB5A1(source + MortonInterleave(x, y) * 2); + return {res.r(), res.g(), res.b(), static_cast<u8>(disable_alpha ? 255 : res.a())}; + } + + case TextureFormat::RGB565: { + auto res = Color::DecodeRGB565(source + MortonInterleave(x, y) * 2); + return {res.r(), res.g(), res.b(), 255}; + } + + case TextureFormat::RGBA4: { + auto res = Color::DecodeRGBA4(source + MortonInterleave(x, y) * 2); + return {res.r(), res.g(), res.b(), static_cast<u8>(disable_alpha ? 255 : res.a())}; + } + + case TextureFormat::IA8: { + const u8* source_ptr = source + MortonInterleave(x, y) * 2; + + if (disable_alpha) { + // Show intensity as red, alpha as green + return {source_ptr[1], source_ptr[0], 0, 255}; + } else { + return {source_ptr[1], source_ptr[1], source_ptr[1], source_ptr[0]}; + } + } + + case TextureFormat::RG8: { + auto res = Color::DecodeRG8(source + MortonInterleave(x, y) * 2); + return {res.r(), res.g(), 0, 255}; + } + + case TextureFormat::I8: { + const u8* source_ptr = source + MortonInterleave(x, y); + return {*source_ptr, *source_ptr, *source_ptr, 255}; + } + + case TextureFormat::A8: { + const u8* source_ptr = source + MortonInterleave(x, y); + + if (disable_alpha) { + return {*source_ptr, *source_ptr, *source_ptr, 255}; + } else { + return {0, 0, 0, *source_ptr}; + } + } + + case TextureFormat::IA4: { + const u8* source_ptr = source + MortonInterleave(x, y); + + u8 i = Color::Convert4To8(((*source_ptr) & 0xF0) >> 4); + u8 a = Color::Convert4To8((*source_ptr) & 0xF); + + if (disable_alpha) { + // Show intensity as red, alpha as green + return {i, a, 0, 255}; + } else { + return {i, i, i, a}; + } + } + + case TextureFormat::I4: { + u32 morton_offset = MortonInterleave(x, y); + const u8* source_ptr = source + morton_offset / 2; + + u8 i = (morton_offset % 2) ? ((*source_ptr & 0xF0) >> 4) : (*source_ptr & 0xF); + i = Color::Convert4To8(i); + + return {i, i, i, 255}; + } + + case TextureFormat::A4: { + u32 morton_offset = MortonInterleave(x, y); + const u8* source_ptr = source + morton_offset / 2; + + u8 a = (morton_offset % 2) ? ((*source_ptr & 0xF0) >> 4) : (*source_ptr & 0xF); + a = Color::Convert4To8(a); + + if (disable_alpha) { + return {a, a, a, 255}; + } else { + return {0, 0, 0, a}; + } + } + + case TextureFormat::ETC1: + case TextureFormat::ETC1A4: { + bool has_alpha = (info.format == TextureFormat::ETC1A4); + size_t subtile_size = has_alpha ? 16 : 8; + + // ETC1 further subdivides each 8x8 tile into four 4x4 subtiles + constexpr unsigned int subtile_width = 4; + constexpr unsigned int subtile_height = 4; + + unsigned int subtile_index = (x / subtile_width) + 2 * (y / subtile_height); + x %= subtile_width; + y %= subtile_height; + + const u8* subtile_ptr = source + subtile_index * subtile_size; + + u8 alpha = 255; + if (has_alpha) { + u64_le packed_alpha; + memcpy(&packed_alpha, subtile_ptr, sizeof(u64)); + subtile_ptr += sizeof(u64); + + alpha = Color::Convert4To8((packed_alpha >> (4 * (x * subtile_width + y))) & 0xF); + } + + u64_le subtile_data; + memcpy(&subtile_data, subtile_ptr, sizeof(u64)); + + return Math::MakeVec(SampleETC1Subtile(subtile_data, x, y), + disable_alpha ? (u8)255 : alpha); + } + + default: + LOG_ERROR(HW_GPU, "Unknown texture format: %x", (u32)info.format); + DEBUG_ASSERT(false); + return {}; + } +} + +TextureInfo TextureInfo::FromPicaRegister(const TexturingRegs::TextureConfig& config, + const TexturingRegs::TextureFormat& format) { + TextureInfo info; + info.physical_address = config.GetPhysicalAddress(); + info.width = config.width; + info.height = config.height; + info.format = format; + info.SetDefaultStride(); + return info; +} + +} // namespace Texture +} // namespace Pica diff --git a/src/video_core/texture/texture_decode.h b/src/video_core/texture/texture_decode.h new file mode 100644 index 000000000..8507cfeb8 --- /dev/null +++ b/src/video_core/texture/texture_decode.h @@ -0,0 +1,60 @@ +// Copyright 2017 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 "common/vector_math.h" +#include "video_core/regs_texturing.h" + +namespace Pica { +namespace Texture { + +/// Returns the byte size of a 8*8 tile of the specified texture format. +size_t CalculateTileSize(TexturingRegs::TextureFormat format); + +struct TextureInfo { + PAddr physical_address; + unsigned int width; + unsigned int height; + ptrdiff_t stride; + TexturingRegs::TextureFormat format; + + static TextureInfo FromPicaRegister(const TexturingRegs::TextureConfig& config, + const TexturingRegs::TextureFormat& format); + + /// Calculates stride from format and width, assuming that the entire texture is contiguous. + void SetDefaultStride() { + stride = CalculateTileSize(format) * (width / 8); + } +}; + +/** + * Lookup texel located at the given coordinates and return an RGBA vector of its color. + * @param source Source pointer to read data from + * @param x,y Texture coordinates to read from + * @param info TextureInfo object describing the texture setup + * @param disable_alpha This is used for debug widgets which use this method to display textures + * without providing a good way to visualize alpha by themselves. If true, this will return 255 for + * the alpha component, and either drop the information entirely or store it in an "unused" color + * channel. + * @todo Eventually we should get rid of the disable_alpha parameter. + */ +Math::Vec4<u8> LookupTexture(const u8* source, unsigned int x, unsigned int y, + const TextureInfo& info, bool disable_alpha = false); + +/** + * Looks up a texel from a single 8x8 texture tile. + * + * @param source Pointer to the beginning of the tile. + * @param x, y In-tile coordinates to read from. Must be < 8. + * @param info TextureInfo describing the texture format. + * @param disable_alpha Used for debugging. Sets the result alpha to 255 and either discards the + * real alpha or inserts it in an otherwise unused channel. + */ +Math::Vec4<u8> LookupTexelInTile(const u8* source, unsigned int x, unsigned int y, + const TextureInfo& info, bool disable_alpha); + +} // namespace Texture +} // namespace Pica diff --git a/src/video_core/vertex_loader.cpp b/src/video_core/vertex_loader.cpp index 2b8ef7018..37c5224a9 100644 --- a/src/video_core/vertex_loader.cpp +++ b/src/video_core/vertex_loader.cpp @@ -8,15 +8,15 @@ #include "common/vector_math.h" #include "core/memory.h" #include "video_core/debug_utils/debug_utils.h" -#include "video_core/pica.h" #include "video_core/pica_state.h" #include "video_core/pica_types.h" +#include "video_core/regs_pipeline.h" #include "video_core/shader/shader.h" #include "video_core/vertex_loader.h" namespace Pica { -void VertexLoader::Setup(const Pica::Regs& regs) { +void VertexLoader::Setup(const PipelineRegs& regs) { ASSERT_MSG(!is_setup, "VertexLoader is not intended to be setup more than once."); const auto& attribute_config = regs.vertex_attributes; @@ -70,7 +70,8 @@ void VertexLoader::Setup(const Pica::Regs& regs) { is_setup = true; } -void VertexLoader::LoadVertex(u32 base_address, int index, int vertex, Shader::InputVertex& input, +void VertexLoader::LoadVertex(u32 base_address, int index, int vertex, + Shader::AttributeBuffer& input, DebugUtils::MemoryAccessTracker& memory_accesses) { ASSERT_MSG(is_setup, "A VertexLoader needs to be setup before loading vertices."); @@ -84,15 +85,16 @@ void VertexLoader::LoadVertex(u32 base_address, int index, int vertex, Shader::I memory_accesses.AddAccess( source_addr, vertex_attribute_elements[i] * - ((vertex_attribute_formats[i] == Regs::VertexAttributeFormat::FLOAT) + ((vertex_attribute_formats[i] == PipelineRegs::VertexAttributeFormat::FLOAT) ? 4 - : (vertex_attribute_formats[i] == Regs::VertexAttributeFormat::SHORT) + : (vertex_attribute_formats[i] == + PipelineRegs::VertexAttributeFormat::SHORT) ? 2 : 1)); } switch (vertex_attribute_formats[i]) { - case Regs::VertexAttributeFormat::BYTE: { + case PipelineRegs::VertexAttributeFormat::BYTE: { const s8* srcdata = reinterpret_cast<const s8*>(Memory::GetPhysicalPointer(source_addr)); for (unsigned int comp = 0; comp < vertex_attribute_elements[i]; ++comp) { @@ -100,7 +102,7 @@ void VertexLoader::LoadVertex(u32 base_address, int index, int vertex, Shader::I } break; } - case Regs::VertexAttributeFormat::UBYTE: { + case PipelineRegs::VertexAttributeFormat::UBYTE: { const u8* srcdata = reinterpret_cast<const u8*>(Memory::GetPhysicalPointer(source_addr)); for (unsigned int comp = 0; comp < vertex_attribute_elements[i]; ++comp) { @@ -108,7 +110,7 @@ void VertexLoader::LoadVertex(u32 base_address, int index, int vertex, Shader::I } break; } - case Regs::VertexAttributeFormat::SHORT: { + case PipelineRegs::VertexAttributeFormat::SHORT: { const s16* srcdata = reinterpret_cast<const s16*>(Memory::GetPhysicalPointer(source_addr)); for (unsigned int comp = 0; comp < vertex_attribute_elements[i]; ++comp) { @@ -116,7 +118,7 @@ void VertexLoader::LoadVertex(u32 base_address, int index, int vertex, Shader::I } break; } - case Regs::VertexAttributeFormat::FLOAT: { + case PipelineRegs::VertexAttributeFormat::FLOAT: { const float* srcdata = reinterpret_cast<const float*>(Memory::GetPhysicalPointer(source_addr)); for (unsigned int comp = 0; comp < vertex_attribute_elements[i]; ++comp) { @@ -142,7 +144,7 @@ void VertexLoader::LoadVertex(u32 base_address, int index, int vertex, Shader::I input.attr[i][2].ToFloat32(), input.attr[i][3].ToFloat32()); } else if (vertex_attribute_is_default[i]) { // Load the default attribute if we're configured to do so - input.attr[i] = g_state.vs_default_attributes[i]; + input.attr[i] = g_state.input_default_attributes.attr[i]; LOG_TRACE(HW_GPU, "Loaded default attribute %x for vertex %x (index %x): (%f, %f, %f, %f)", i, vertex, index, input.attr[i][0].ToFloat32(), input.attr[i][1].ToFloat32(), diff --git a/src/video_core/vertex_loader.h b/src/video_core/vertex_loader.h index 9f2098bb2..02db10aee 100644 --- a/src/video_core/vertex_loader.h +++ b/src/video_core/vertex_loader.h @@ -2,7 +2,7 @@ #include <array> #include "common/common_types.h" -#include "video_core/pica.h" +#include "video_core/regs_pipeline.h" namespace Pica { @@ -11,18 +11,18 @@ class MemoryAccessTracker; } namespace Shader { -struct InputVertex; +struct AttributeBuffer; } class VertexLoader { public: VertexLoader() = default; - explicit VertexLoader(const Pica::Regs& regs) { + explicit VertexLoader(const PipelineRegs& regs) { Setup(regs); } - void Setup(const Pica::Regs& regs); - void LoadVertex(u32 base_address, int index, int vertex, Shader::InputVertex& input, + void Setup(const PipelineRegs& regs); + void LoadVertex(u32 base_address, int index, int vertex, Shader::AttributeBuffer& input, DebugUtils::MemoryAccessTracker& memory_accesses); int GetNumTotalAttributes() const { @@ -32,7 +32,7 @@ public: private: std::array<u32, 16> vertex_attribute_sources; std::array<u32, 16> vertex_attribute_strides{}; - std::array<Regs::VertexAttributeFormat, 16> vertex_attribute_formats; + std::array<PipelineRegs::VertexAttributeFormat, 16> vertex_attribute_formats; std::array<u32, 16> vertex_attribute_elements{}; std::array<bool, 16> vertex_attribute_is_default; int num_total_attributes = 0; |