diff --git a/Core/HLE/sceMpeg.cpp b/Core/HLE/sceMpeg.cpp index 2f8cb86416..cecc9e4d15 100644 --- a/Core/HLE/sceMpeg.cpp +++ b/Core/HLE/sceMpeg.cpp @@ -2091,7 +2091,7 @@ static u32 sceMpegAvcInitYCbCr(u32 mpeg, int mode, int width, int height, u32 yc return -1; } - ERROR_LOG(ME, "UNIMPL sceMpegAvcInitYCbCr(%08x, %i, %i, %i, %08x)", mpeg, mode, width, height, ycbcr_addr); + WARN_LOG_ONCE(sceMpegAvcInitYCbCr, ME, "UNIMPL sceMpegAvcInitYCbCr(%08x, %i, %i, %i, %08x)", mpeg, mode, width, height, ycbcr_addr); return 0; } diff --git a/GPU/Common/FragmentShaderGenerator.cpp b/GPU/Common/FragmentShaderGenerator.cpp index d2b5315d82..efa17e6500 100644 --- a/GPU/Common/FragmentShaderGenerator.cpp +++ b/GPU/Common/FragmentShaderGenerator.cpp @@ -96,6 +96,7 @@ bool GenerateFragmentShader(const FShaderID &id, char *buffer, const ShaderLangu p.ApplySamplerMetadata(arrayTexture ? samplersStereo : samplersMono); + bool lmode = id.Bit(FS_BIT_LMODE); bool doTexture = id.Bit(FS_BIT_DO_TEXTURE); bool enableFog = id.Bit(FS_BIT_ENABLE_FOG); bool enableAlphaTest = id.Bit(FS_BIT_ALPHA_TEST); @@ -173,6 +174,10 @@ bool GenerateFragmentShader(const FShaderID &id, char *buffer, const ShaderLangu bool needFragCoord = readFramebufferTex || gstate_c.Use(GPU_ROUND_FRAGMENT_DEPTH_TO_16BIT); bool writeDepth = gstate_c.Use(GPU_ROUND_FRAGMENT_DEPTH_TO_16BIT); + // TODO: We could have a separate mechanism to support more ops using the shader blending mechanism, +// on hardware that can do proper bit math in fragment shaders. + SimulateLogicOpType simulateLogicOpType = (SimulateLogicOpType)id.Bits(FS_BIT_SIMULATE_LOGIC_OP_TYPE, 2); + if (shaderDepalMode != ShaderDepalMode::OFF && !doTexture) { *errorString = "depal requires a texture"; return false; @@ -205,7 +210,9 @@ bool GenerateFragmentShader(const FShaderID &id, char *buffer, const ShaderLangu // Note: the precision qualifiers must match the vertex shader! WRITE(p, "layout (location = 1) %s in lowp vec4 v_color0;\n", shading); - WRITE(p, "layout (location = 2) %s in lowp vec3 v_color1;\n", shading); + if (lmode) { + WRITE(p, "layout (location = 2) %s in lowp vec3 v_color1;\n", shading); + } WRITE(p, "layout (location = 3) in highp float v_fogdepth;\n"); if (doTexture) { WRITE(p, "layout (location = 0) in highp vec3 v_texcoord;\n"); @@ -311,7 +318,9 @@ bool GenerateFragmentShader(const FShaderID &id, char *buffer, const ShaderLangu } const char *colorInterpolation = doFlatShading && compat.shaderLanguage == HLSL_D3D11 ? "nointerpolation " : ""; WRITE(p, " %svec4 v_color0: COLOR0;\n", colorInterpolation); - WRITE(p, " vec3 v_color1: COLOR1;\n"); + if (lmode) { + WRITE(p, " vec3 v_color1: COLOR1;\n"); + } WRITE(p, " float v_fogdepth: TEXCOORD1;\n"); if (needFragCoord) { if (compat.shaderLanguage == HLSL_D3D11) { @@ -382,6 +391,7 @@ bool GenerateFragmentShader(const FShaderID &id, char *buffer, const ShaderLangu WRITE(p, "uniform vec2 u_texclampoff;\n"); } + // TODO: Can get rid of some of this in the != 0 cases. if (enableAlphaTest || enableColorTest) { if (enableFragmentTestCache) { WRITE(p, "uniform sampler2D testtex;\n"); @@ -425,7 +435,9 @@ bool GenerateFragmentShader(const FShaderID &id, char *buffer, const ShaderLangu } WRITE(p, "%s %s lowp vec4 v_color0;\n", shading, compat.varying_fs); - WRITE(p, "%s %s lowp vec3 v_color1;\n", shading, compat.varying_fs); + if (lmode) { + WRITE(p, "%s %s lowp vec3 v_color1;\n", shading, compat.varying_fs); + } if (enableFog) { *uniformMask |= DIRTY_FOGCOLOR; WRITE(p, "uniform vec3 u_fogcolor;\n"); @@ -535,7 +547,9 @@ bool GenerateFragmentShader(const FShaderID &id, char *buffer, const ShaderLangu if (compat.shaderLanguage == HLSL_D3D11 || compat.shaderLanguage == HLSL_D3D9) { WRITE(p, " vec4 v_color0 = In.v_color0;\n"); - WRITE(p, " vec3 v_color1 = In.v_color1;\n"); + if (lmode) { + WRITE(p, " vec3 v_color1 = In.v_color1;\n"); + } if (enableFog) { WRITE(p, " float v_fogdepth = In.v_fogdepth;\n"); } @@ -573,8 +587,12 @@ bool GenerateFragmentShader(const FShaderID &id, char *buffer, const ShaderLangu // Clear mode does not allow any fancy shading. WRITE(p, " vec4 v = v_color0;\n"); } else { + const char *secondary = ""; // Secondary color for specular on top of texture - WRITE(p, " vec4 s = vec4(v_color1, 0.0);\n"); + if (lmode) { + WRITE(p, " vec4 s = vec4(v_color1, 0.0);\n"); + secondary = " + s"; + } if (doTexture) { char texcoord[64] = "v_texcoord"; @@ -831,37 +849,43 @@ bool GenerateFragmentShader(const FShaderID &id, char *buffer, const ShaderLangu switch (texFunc) { case GE_TEXFUNC_MODULATE: - WRITE(p, " vec4 v = p * t + s;\n"); + WRITE(p, " vec4 v = p * t%s;\n", secondary); break; case GE_TEXFUNC_DECAL: - WRITE(p, " vec4 v = vec4(mix(p.rgb, t.rgb, t.a), p.a) + s;\n"); + WRITE(p, " vec4 v = vec4(mix(p.rgb, t.rgb, t.a), p.a)%s;\n", secondary); break; case GE_TEXFUNC_BLEND: - WRITE(p, " vec4 v = vec4(mix(p.rgb, u_texenv.rgb, t.rgb), p.a * t.a) + s;\n"); + WRITE(p, " vec4 v = vec4(mix(p.rgb, u_texenv.rgb, t.rgb), p.a * t.a)%s;\n", secondary); break; case GE_TEXFUNC_REPLACE: WRITE(p, " vec4 r = t;\n"); WRITE(p, " r.a = mix(r.a, p.a, u_texNoAlpha);\n"); - WRITE(p, " vec4 v = r + s;\n"); + WRITE(p, " vec4 v = r%s;\n", secondary); break; case GE_TEXFUNC_ADD: case GE_TEXFUNC_UNKNOWN1: case GE_TEXFUNC_UNKNOWN2: case GE_TEXFUNC_UNKNOWN3: - WRITE(p, " vec4 v = vec4(p.rgb + t.rgb, p.a * t.a) + s;\n"); + WRITE(p, " vec4 v = vec4(p.rgb + t.rgb, p.a * t.a)%s;\n", secondary); break; default: // Doesn't happen - WRITE(p, " vec4 v = p + s;\n"); break; + WRITE(p, " vec4 v = p%s;\n", secondary); break; break; } // This happens before fog is applied. *uniformMask |= DIRTY_TEX_ALPHA_MUL; - WRITE(p, " v.rgb = clamp(v.rgb * u_texMul, 0.0, 1.0);\n"); + + // We only need a clamp if the color will be further processed. Otherwise the hardware color conversion will clamp for us. + if (enableFog || enableColorTest || replaceBlend != REPLACE_BLEND_NO || simulateLogicOpType != LOGICOPTYPE_NORMAL || colorWriteMask || blueToAlpha) { + WRITE(p, " v.rgb = clamp(v.rgb * u_texMul, 0.0, 1.0);\n"); + } else { + WRITE(p, " v.rgb *= u_texMul;\n"); + } } else { // No texture mapping - WRITE(p, " vec4 v = v_color0 + s;\n"); + WRITE(p, " vec4 v = v_color0%s;\n", secondary); } if (enableFog) { @@ -1134,9 +1158,6 @@ bool GenerateFragmentShader(const FShaderID &id, char *buffer, const ShaderLangu return false; } - // TODO: We could have a separate mechanism to support more ops using the shader blending mechanism, - // on hardware that can do proper bit math in fragment shaders. - SimulateLogicOpType simulateLogicOpType = (SimulateLogicOpType)id.Bits(FS_BIT_SIMULATE_LOGIC_OP_TYPE, 2); switch (simulateLogicOpType) { case LOGICOPTYPE_ONE: WRITE(p, " %s.rgb = splat3(1.0);\n", compat.fragColor0); diff --git a/GPU/Common/GeometryShaderGenerator.cpp b/GPU/Common/GeometryShaderGenerator.cpp index 3f564dc012..d09775cf76 100644 --- a/GPU/Common/GeometryShaderGenerator.cpp +++ b/GPU/Common/GeometryShaderGenerator.cpp @@ -74,8 +74,10 @@ bool GenerateGeometryShader(const GShaderID &id, char *buffer, const ShaderLangu } varyings.push_back(VaryingDef{ "vec4", "v_color0", Draw::SEM_COLOR0, 1, "lowp" }); outVaryings.push_back(VaryingDef{ "vec4", "v_color0Out", Draw::SEM_COLOR0, 1, "lowp" }); - varyings.push_back(VaryingDef{ "vec3", "v_color1", Draw::SEM_COLOR1, 2, "lowp" }); - outVaryings.push_back(VaryingDef{ "vec3", "v_color1Out", Draw::SEM_COLOR1, 2, "lowp" }); + if (id.Bit(GS_BIT_LMODE)) { + varyings.push_back(VaryingDef{ "vec3", "v_color1", Draw::SEM_COLOR1, 2, "lowp" }); + outVaryings.push_back(VaryingDef{ "vec3", "v_color1Out", Draw::SEM_COLOR1, 2, "lowp" }); + } varyings.push_back(VaryingDef{ "float", "v_fogdepth", Draw::SEM_TEXCOORD1, 3, "highp" }); outVaryings.push_back(VaryingDef{ "float", "v_fogdepthOut", Draw::SEM_TEXCOORD1, 3, "highp" }); diff --git a/GPU/Common/ShaderId.cpp b/GPU/Common/ShaderId.cpp index 75c9d72c88..b435cd2d74 100644 --- a/GPU/Common/ShaderId.cpp +++ b/GPU/Common/ShaderId.cpp @@ -90,6 +90,7 @@ void ComputeVertexShaderID(VShaderID *id_out, VertexDecoder *vertexDecoder, bool !isModeThrough && gstate_c.submitType == SubmitType::DRAW; // neither hw nor sw spline/bezier. See #11692 VShaderID id; + id.SetBit(VS_BIT_LMODE, lmode); id.SetBit(VS_BIT_IS_THROUGH, isModeThrough); id.SetBit(VS_BIT_HAS_COLOR, vtypeHasColor); id.SetBit(VS_BIT_VERTEX_RANGE_CULLING, vertexRangeCulling); @@ -130,7 +131,6 @@ void ComputeVertexShaderID(VShaderID *id_out, VertexDecoder *vertexDecoder, bool // doShadeMapping is stored as UVGenMode, and light type doesn't matter for shade mapping. id.SetBit(VS_BIT_LIGHTING_ENABLE); if (gstate_c.Use(GPU_USE_LIGHT_UBERSHADER)) { - lmode = false; // handled dynamically. id.SetBit(VS_BIT_LIGHT_UBERSHADER); } else { id.SetBits(VS_BIT_MATERIAL_UPDATE, 3, gstate.getMaterialUpdate()); @@ -162,7 +162,6 @@ void ComputeVertexShaderID(VShaderID *id_out, VertexDecoder *vertexDecoder, bool } } - id.SetBit(VS_BIT_LMODE, lmode); id.SetBit(VS_BIT_FLATSHADE, doFlatShading); // These two bits cannot be combined, otherwise havoc occurs. We get reports that indicate this happened somehow... "ERROR: 0:14: 'u_proj' : undeclared identifier" @@ -190,6 +189,7 @@ std::string FragmentShaderDesc(const FShaderID &id) { if (id.Bit(FS_BIT_DO_TEXTURE)) desc << (id.Bit(FS_BIT_3D_TEXTURE) ? "Tex3D " : "Tex "); if (id.Bit(FS_BIT_DO_TEXTURE_PROJ)) desc << "TexProj "; if (id.Bit(FS_BIT_ENABLE_FOG)) desc << "Fog "; + if (id.Bit(FS_BIT_LMODE)) desc << "LM "; if (id.Bit(FS_BIT_FLATSHADE)) desc << "Flat "; if (id.Bit(FS_BIT_BGRA_TEXTURE)) desc << "BGRA "; switch ((ShaderDepalMode)id.Bits(FS_BIT_SHADER_DEPAL_MODE, 2)) { @@ -281,6 +281,7 @@ void ComputeFragmentShaderID(FShaderID *id_out, const ComputedPipelineState &pip id.SetBit(FS_BIT_CLEARMODE); } else { bool isModeThrough = gstate.isModeThrough(); + bool lmode = gstate.isUsingSecondaryColor() && gstate.isLightingEnabled() && !isModeThrough; bool enableFog = gstate.isFogEnabled() && !isModeThrough; bool enableAlphaTest = gstate.isAlphaTestEnabled() && !IsAlphaTestTriviallyTrue(); bool enableColorTest = gstate.isColorTestEnabled() && !IsColorTestTriviallyTrue(); @@ -310,6 +311,8 @@ void ComputeFragmentShaderID(FShaderID *id_out, const ComputedPipelineState &pip id.SetBit(FS_BIT_3D_TEXTURE, gstate_c.curTextureIs3D); } + id.SetBit(FS_BIT_LMODE, lmode); + if (enableAlphaTest) { // 5 bits total. id.SetBit(FS_BIT_ALPHA_TEST); @@ -392,6 +395,7 @@ std::string GeometryShaderDesc(const GShaderID &id) { desc << StringFromFormat("%08x:%08x ", id.d[1], id.d[0]); if (id.Bit(GS_BIT_ENABLED)) desc << "ENABLED "; if (id.Bit(GS_BIT_DO_TEXTURE)) desc << "TEX "; + if (id.Bit(GS_BIT_LMODE)) desc << "LM "; return desc.str(); } @@ -423,6 +427,9 @@ void ComputeGeometryShaderID(GShaderID *id_out, const Draw::Bugs &bugs, int prim if (gstate.isModeClear()) { // No attribute bits. } else { + bool lmode = gstate.isUsingSecondaryColor() && gstate.isLightingEnabled() && !isModeThrough; + + id.SetBit(GS_BIT_LMODE, lmode); if (gstate.isTextureMapEnabled()) { id.SetBit(GS_BIT_DO_TEXTURE); } diff --git a/GPU/Common/ShaderId.h b/GPU/Common/ShaderId.h index 45bd12dc6e..1dc2781b5f 100644 --- a/GPU/Common/ShaderId.h +++ b/GPU/Common/ShaderId.h @@ -73,7 +73,8 @@ enum FShaderBit : uint8_t { FS_BIT_SHADER_TEX_CLAMP = 7, FS_BIT_CLAMP_S = 8, FS_BIT_CLAMP_T = 9, - // 2 bits free + // 1 bit free + FS_BIT_LMODE = 11, FS_BIT_ALPHA_TEST = 12, FS_BIT_ALPHA_TEST_FUNC = 13, // 3 bits FS_BIT_ALPHA_AGAINST_ZERO = 16, @@ -110,7 +111,8 @@ static inline FShaderBit operator +(FShaderBit bit, int i) { enum GShaderBit : uint8_t { GS_BIT_ENABLED = 0, // If not set, we don't use a geo shader. GS_BIT_DO_TEXTURE = 1, // presence of texcoords - GS_BIT_CURVE = 2, // curve, which means don't do range culling. + GS_BIT_LMODE = 2, + GS_BIT_CURVE = 3, // curve, which means don't do range culling. }; static inline GShaderBit operator +(GShaderBit bit, int i) { diff --git a/GPU/Common/ShaderUniforms.cpp b/GPU/Common/ShaderUniforms.cpp index ee952d1f97..c020eb1088 100644 --- a/GPU/Common/ShaderUniforms.cpp +++ b/GPU/Common/ShaderUniforms.cpp @@ -298,8 +298,6 @@ uint32_t PackLightControlBits() { // Material update is 3 bits. lightControl |= gstate.getMaterialUpdate() << 20; - // LMODE is 1 bit. - lightControl |= gstate.isUsingSecondaryColor() << 23; return lightControl; } diff --git a/GPU/Common/VertexShaderGenerator.cpp b/GPU/Common/VertexShaderGenerator.cpp index 1027a15632..9d724e87e3 100644 --- a/GPU/Common/VertexShaderGenerator.cpp +++ b/GPU/Common/VertexShaderGenerator.cpp @@ -285,7 +285,9 @@ bool GenerateVertexShader(const VShaderID &id, char *buffer, const ShaderLanguag } WRITE(p, "layout (location = 1) %sout lowp vec4 v_color0;\n", shading); - WRITE(p, "layout (location = 2) %sout lowp vec3 v_color1;\n", shading); + if (lmode) { + WRITE(p, "layout (location = 2) %sout lowp vec3 v_color1;\n", shading); + } WRITE(p, "layout (location = 0) out highp vec3 v_texcoord;\n"); @@ -416,7 +418,9 @@ bool GenerateVertexShader(const VShaderID &id, char *buffer, const ShaderLanguag WRITE(p, " vec3 v_texcoord : TEXCOORD0;\n"); const char *colorInterpolation = doFlatShading && compat.shaderLanguage == HLSL_D3D11 ? "nointerpolation " : ""; WRITE(p, " %svec4 v_color0 : COLOR0;\n", colorInterpolation); - WRITE(p, " vec3 v_color1 : COLOR1;\n"); + if (lmode) { + WRITE(p, " vec3 v_color1 : COLOR1;\n"); + } WRITE(p, " float v_fogdepth : TEXCOORD1;\n"); if (compat.shaderLanguage == HLSL_D3D9) { @@ -587,7 +591,9 @@ bool GenerateVertexShader(const VShaderID &id, char *buffer, const ShaderLanguag } WRITE(p, "%s%s lowp vec4 v_color0;\n", shading, compat.varying_vs); - WRITE(p, "%s%s lowp vec3 v_color1;\n", shading, compat.varying_vs); + if (lmode) { + WRITE(p, "%s%s lowp vec3 v_color1;\n", shading, compat.varying_vs); + } WRITE(p, "%s %s vec3 v_texcoord;\n", compat.varying_vs, highpTexcoord ? "highp" : "mediump"); @@ -819,13 +825,14 @@ bool GenerateVertexShader(const VShaderID &id, char *buffer, const ShaderLanguag } if (hasColor) { WRITE(p, " %sv_color0 = color0;\n", compat.vsOutPrefix); - if (lmode) + if (lmode) { WRITE(p, " %sv_color1 = color1;\n", compat.vsOutPrefix); - else - WRITE(p, " %sv_color1 = splat3(0.0);\n", compat.vsOutPrefix); + } } else { WRITE(p, " %sv_color0 = u_matambientalpha;\n", compat.vsOutPrefix); - WRITE(p, " %sv_color1 = splat3(0.0);\n", compat.vsOutPrefix); + if (lmode) { + WRITE(p, " %sv_color1 = splat3(0.0);\n", compat.vsOutPrefix); + } } WRITE(p, " %sv_fogdepth = fog;\n", compat.vsOutPrefix); if (isModeThrough) { @@ -970,6 +977,7 @@ bool GenerateVertexShader(const VShaderID &id, char *buffer, const ShaderLanguag bool distanceNeeded = false; bool anySpots = false; if (enableLighting) { + p.C(" lowp vec4 lightSum0 = u_ambient * ambientColor + vec4(u_matemissive, 0.0);\n"); for (int i = 0; i < 4; i++) { @@ -1153,29 +1161,20 @@ bool GenerateVertexShader(const VShaderID &id, char *buffer, const ShaderLanguag } if (enableLighting) { - WRITE(p, " lightSum0 = clamp(lightSum0, 0.0, 1.0);\n"); - if (specularIsZero) { - WRITE(p, " %sv_color0 = lightSum0;\n", compat.vsOutPrefix); - WRITE(p, " %sv_color1 = splat3(0.0);\n", compat.vsOutPrefix); - } else { - if (lightUberShader) { - p.C(" bool lmode = (u_lightControl & (0x1u << 0x17u)) != 0x0u;\n"); - p.C(" if (lmode) {\n"); - p.F(" %sv_color0 = lightSum0;\n", compat.vsOutPrefix); - p.F(" %sv_color1 = clamp(lightSum1, 0.0, 1.0);\n", compat.vsOutPrefix); - p.C(" } else {\n"); - p.F(" %sv_color0 = clamp(lightSum0 + vec4(lightSum1, 0.0), 0.0, 1.0);\n", compat.vsOutPrefix); - p.F(" %sv_color1 = splat3(0.0);\n", compat.vsOutPrefix); - p.C(" }"); + // Sum up ambient, emissive here. + if (lmode) { + WRITE(p, " %sv_color0 = clamp(lightSum0, 0.0, 1.0);\n", compat.vsOutPrefix); + // v_color1 only exists when lmode = 1. + if (specularIsZero) { + WRITE(p, " %sv_color1 = splat3(0.0);\n", compat.vsOutPrefix); } else { - if (lmode) { - WRITE(p, " %sv_color0 = lightSum0;\n", compat.vsOutPrefix); - // v_color1 only exists when lmode = 1. - WRITE(p, " %sv_color1 = clamp(lightSum1, 0.0, 1.0);\n", compat.vsOutPrefix); - } else { - WRITE(p, " %sv_color0 = clamp(lightSum0 + vec4(lightSum1, 0.0), 0.0, 1.0);\n", compat.vsOutPrefix); - WRITE(p, " %sv_color1 = splat3(0.0);\n", compat.vsOutPrefix); - } + WRITE(p, " %sv_color1 = clamp(lightSum1, 0.0, 1.0);\n", compat.vsOutPrefix); + } + } else { + if (specularIsZero) { + WRITE(p, " %sv_color0 = clamp(lightSum0, 0.0, 1.0);\n", compat.vsOutPrefix); + } else { + WRITE(p, " %sv_color0 = clamp(clamp(lightSum0, 0.0, 1.0) + vec4(lightSum1, 0.0), 0.0, 1.0);\n", compat.vsOutPrefix); } } } else { @@ -1191,7 +1190,9 @@ bool GenerateVertexShader(const VShaderID &id, char *buffer, const ShaderLanguag WRITE(p, " %sv_color0.r += 0.000001;\n", compat.vsOutPrefix); } } - WRITE(p, " %sv_color1 = splat3(0.0);\n", compat.vsOutPrefix); + if (lmode) { + WRITE(p, " %sv_color1 = splat3(0.0);\n", compat.vsOutPrefix); + } } bool scaleUV = !isModeThrough && (uvGenMode == GE_TEXMAP_TEXTURE_COORDS || uvGenMode == GE_TEXMAP_UNKNOWN); diff --git a/GPU/GLES/ShaderManagerGLES.cpp b/GPU/GLES/ShaderManagerGLES.cpp index ee4b99ff8a..248347a07b 100644 --- a/GPU/GLES/ShaderManagerGLES.cpp +++ b/GPU/GLES/ShaderManagerGLES.cpp @@ -869,6 +869,7 @@ LinkedShader *ShaderManagerGLES::ApplyFragmentShader(VShaderID VSID, Shader *vs, shaderSwitchDirtyUniforms_ = 0; if (ls == nullptr) { + _dbg_assert_(FSID.Bit(FS_BIT_LMODE) == VSID.Bit(VS_BIT_LMODE)); _dbg_assert_(FSID.Bit(FS_BIT_FLATSHADE) == VSID.Bit(VS_BIT_FLATSHADE)); if (vs == nullptr || fs == nullptr) { @@ -965,7 +966,7 @@ enum class CacheDetectFlags { }; #define CACHE_HEADER_MAGIC 0x83277592 -#define CACHE_VERSION 29 +#define CACHE_VERSION 30 struct CacheHeader { uint32_t magic; diff --git a/GPU/GPUCommonHW.cpp b/GPU/GPUCommonHW.cpp index 02f4065d29..85bbe1e530 100644 --- a/GPU/GPUCommonHW.cpp +++ b/GPU/GPUCommonHW.cpp @@ -100,8 +100,8 @@ const CommonCommandTableEntry commonCommandTable[] = { { GE_CMD_LIGHTTYPE3, FLAG_FLUSHBEFOREONCHANGE, DIRTY_VERTEXSHADER_STATE | DIRTY_LIGHT3 }, { GE_CMD_MATERIALUPDATE, FLAG_FLUSHBEFOREONCHANGE, DIRTY_VERTEXSHADER_STATE }, - // These change vertex shaders (in non uber shader mode) so need flushing. - { GE_CMD_LIGHTMODE, FLAG_FLUSHBEFOREONCHANGE, DIRTY_VERTEXSHADER_STATE }, + // These change all shaders so need flushing. + { GE_CMD_LIGHTMODE, FLAG_FLUSHBEFOREONCHANGE, DIRTY_VERTEXSHADER_STATE | DIRTY_FRAGMENTSHADER_STATE | DIRTY_GEOMETRYSHADER_STATE }, { GE_CMD_TEXFILTER, FLAG_FLUSHBEFOREONCHANGE, DIRTY_TEXTURE_PARAMS }, { GE_CMD_TEXWRAP, FLAG_FLUSHBEFOREONCHANGE, DIRTY_TEXTURE_PARAMS | DIRTY_FRAGMENTSHADER_STATE }, @@ -460,11 +460,9 @@ void GPUCommonHW::UpdateCmdInfo() { if (gstate_c.Use(GPU_USE_LIGHT_UBERSHADER)) { cmdInfo_[GE_CMD_MATERIALUPDATE].RemoveDirty(DIRTY_VERTEXSHADER_STATE); cmdInfo_[GE_CMD_MATERIALUPDATE].AddDirty(DIRTY_LIGHT_CONTROL); - cmdInfo_[GE_CMD_LIGHTMODE].AddDirty(DIRTY_LIGHT_CONTROL); } else { cmdInfo_[GE_CMD_MATERIALUPDATE].RemoveDirty(DIRTY_LIGHT_CONTROL); cmdInfo_[GE_CMD_MATERIALUPDATE].AddDirty(DIRTY_VERTEXSHADER_STATE); - cmdInfo_[GE_CMD_LIGHTMODE].RemoveDirty(DIRTY_LIGHT_CONTROL); } } diff --git a/GPU/Vulkan/ShaderManagerVulkan.cpp b/GPU/Vulkan/ShaderManagerVulkan.cpp index 23759f4962..dd8442a0b3 100644 --- a/GPU/Vulkan/ShaderManagerVulkan.cpp +++ b/GPU/Vulkan/ShaderManagerVulkan.cpp @@ -326,6 +326,10 @@ void ShaderManagerVulkan::GetShaders(int prim, VertexDecoder *decoder, VulkanVer } _dbg_assert_(FSID.Bit(FS_BIT_FLATSHADE) == VSID.Bit(VS_BIT_FLATSHADE)); + _dbg_assert_(FSID.Bit(FS_BIT_LMODE) == VSID.Bit(VS_BIT_LMODE)); + if (GSID.Bit(GS_BIT_ENABLED)) { + _dbg_assert_(GSID.Bit(GS_BIT_LMODE) == VSID.Bit(VS_BIT_LMODE)); + } // Just update uniforms if this is the same shader as last time. if (lastVShader_ != nullptr && lastFShader_ != nullptr && VSID == lastVSID_ && FSID == lastFSID_ && GSID == lastGSID_) { @@ -516,7 +520,7 @@ enum class VulkanCacheDetectFlags { }; #define CACHE_HEADER_MAGIC 0xff51f420 -#define CACHE_VERSION 43 +#define CACHE_VERSION 44 struct VulkanCacheHeader { uint32_t magic;