fix: texture updates and buffer bindings

src/Cafe/HW/Latte/Core/LatteTexture.cpp

@@ -170,7 +170,7 @@ void LatteTexture_UnregisterTextureMemoryOccupancy(LatteTexture* texture)
 }
 
 // calculate the actually accessed data range
-// the resulting range is an estimate and may be smaller than the actual slice size (but not larger) 
+// the resulting range is an estimate and may be smaller than the actual slice size (but not larger)
 void LatteTexture_EstimateMipSliceAccessedDataRange(LatteTexture* texture, sint32 sliceIndex, sint32 mipIndex, LatteTextureSliceMipInfo* sliceMipInfo)
 {
 	uint32 estAddrStart;
@@ -222,7 +222,7 @@ void LatteTexture_InitSliceAndMipInfo(LatteTexture* texture)
 			LatteAddrLib::AddrSurfaceInfo_OUT surfaceInfo;
 			LatteAddrLib::GX2CalculateSurfaceInfo(texture->format, texture->width, texture->height, texture->depth, texture->dim, Latte::MakeGX2TileMode(texture->tileMode), 0, mipIndex, &surfaceInfo);
 			sliceMipInfo->tileMode = surfaceInfo.hwTileMode;
-			
+
 			if (mipIndex == 0)
 				sliceMipInfo->pitch = texture->pitch; // for the base level, use the pitch value configured in hardware
 			else
@@ -352,6 +352,7 @@ void LatteTexture_CopySlice(LatteTexture* srcTexture, sint32 srcSlice, sint32 sr
 	if (srcTexture->isDepth != dstTexture->isDepth)
 	{
 		g_renderer->surfaceCopy_copySurfaceWithFormatConversion(srcTexture, srcMip, srcSlice, dstTexture, dstMip, dstSlice, width, height);
+		throw std::runtime_error("1");
 		return;
 	}
 	// rescale copy size
@@ -384,6 +385,7 @@ void LatteTexture_CopySlice(LatteTexture* srcTexture, sint32 srcSlice, sint32 sr
 			cemuLog_log(LogType::Force, "Source:      {:08x} origResolution {:4}/{:4} effectiveResolution {:4}/{:4} fmt {:04x} mipIndex {} ratioW/H: {:.4}/{:.4}", srcTexture->physAddress, srcTexture->width, srcTexture->height, effectiveWidth_src, effectiveHeight_src, (uint32)srcTexture->format, srcMip, ratioWidth_src, ratioHeight_src);
 			cemuLog_log(LogType::Force, "Destination: {:08x} origResolution {:4}/{:4} effectiveResolution {:4}/{:4} fmt {:04x} mipIndex {} ratioW/H: {:.4}/{:.4}", dstTexture->physAddress, dstTexture->width, dstTexture->height, effectiveWidth_dst, effectiveHeight_dst, (uint32)dstTexture->format, dstMip, ratioWidth_dst, ratioHeight_dst);
 		}
+		throw std::runtime_error("2");
 		//cemuLog_logDebug(LogType::Force, "If these textures are not meant to share data you can ignore this");
 		return;
 	}
@@ -877,7 +879,7 @@ VIEWCOMPATIBILITY LatteTexture_CanTextureBeRepresentedAsView(LatteTexture* baseT
 			// check pitch
 			if(sliceMipInfo->pitch != pitch)
 				continue;
-			// check all slices			
+			// check all slices
 			if(LatteAddrLib::TM_IsThickAndMacroTiled(baseTexture->tileMode))
 				continue; // todo - check only every 4th slice?
 			for (sint32 s=0; s<baseTexture->GetMipDepth(m); s++)
@@ -978,7 +980,7 @@ LatteTextureView* LatteTexture_CreateMapping(MPTR physAddr, MPTR physMipAddr, si
 	}
 	// note: When creating an existing texture, we only allow mip and slice expansion at the end
 	cemu_assert_debug(depth);
-	
+
 	cemu_assert_debug(!(depth > 1 && dimBase == Latte::E_DIM::DIM_2D));
 	cemu_assert_debug(!(numSlice > 1 && dimView == Latte::E_DIM::DIM_2D));
 	// todo, depth and numSlice are redundant

src/Cafe/HW/Latte/LegacyShaderDecompiler/LatteDecompilerEmitMSLHeader.hpp

@@ -261,7 +261,7 @@ namespace LatteDecompiler
             // generate pixel outputs for pixel shader
             for (uint32 i = 0; i < LATTE_NUM_COLOR_TARGET; i++)
             {
-               	if ((decompilerContext->shader->pixelColorOutputMask&(1 << i)) != 0)
+               	if ((decompilerContext->shader->pixelColorOutputMask & (1 << i)) != 0)
                	{
               		src->addFmt("float4 passPixelColor{} [[color({})]];" _CRLF, i, i);
                	}

src/Cafe/HW/Latte/Renderer/Metal/LatteToMtl.cpp

@@ -260,7 +260,8 @@ MTL::PrimitiveType GetMtlPrimitiveType(LattePrimitiveMode mode)
         case LattePrimitiveMode::TRIANGLE_STRIP:
             return MTL::PrimitiveTypeTriangleStrip;
         default:
-            printf("unimplemented primitive type %u\n", (uint32)mode);
+            // TODO: uncomment
+            //printf("unimplemented primitive type %u\n", (uint32)mode);
             return MTL::PrimitiveTypeTriangle;
     }
 }

src/Cafe/HW/Latte/Renderer/Metal/MetalPipelineCache.cpp

@@ -58,7 +58,6 @@ MTL::RenderPipelineState* MetalPipelineCache::GetPipelineState(const LatteFetchS
 
 		uint32 bufferIndex = bufferGroup.attributeBufferIndex;
 		uint32 bufferBaseRegisterIndex = mmSQ_VTX_ATTRIBUTE_BLOCK_START + bufferIndex * 7;
-		// TODO: is LatteGPUState.contextNew correct?
 		uint32 bufferStride = (LatteGPUState.contextNew.GetRawView()[bufferBaseRegisterIndex + 2] >> 11) & 0xFFFF;
 
 		auto layout = vertexDescriptor->layouts()->object(GET_MTL_VERTEX_BUFFER_INDEX(bufferIndex));

src/Cafe/HW/Latte/Renderer/Metal/MetalRenderer.cpp

@@ -561,7 +561,6 @@ void MetalRenderer::draw_execute(uint32 baseVertex, uint32 baseInstance, uint32
 	auto renderCommandEncoder = GetRenderCommandEncoder(renderPassDescriptor, colorRenderTargets, depthRenderTarget);
 
 	// Shaders
-	LatteSHRC_UpdateActiveShaders();
 	LatteDecompilerShader* vertexShader = LatteSHRC_GetActiveVertexShader();
 	LatteDecompilerShader* pixelShader = LatteSHRC_GetActivePixelShader();
 	if (!vertexShader || !static_cast<RendererShaderMtl*>(vertexShader->shader)->GetFunction())
@@ -627,7 +626,16 @@ void MetalRenderer::draw_execute(uint32 baseVertex, uint32 baseInstance, uint32
 
 void MetalRenderer::draw_endSequence()
 {
-    // TODO: do something?
+    LatteDecompilerShader* pixelShader = LatteSHRC_GetActivePixelShader();
+	// post-drawcall logic
+	if (pixelShader)
+		LatteRenderTarget_trackUpdates();
+	bool hasReadback = LatteTextureReadback_Update();
+	//m_recordedDrawcalls++;
+	//if (m_recordedDrawcalls >= m_submitThreshold || hasReadback)
+	//{
+	//	SubmitCommandBuffer();
+	//}
 }
 
 void* MetalRenderer::indexData_reserveIndexMemory(uint32 size, uint32& offset, uint32& bufferIndex)
@@ -1095,33 +1103,39 @@ void MetalRenderer::BindStageResources(MTL::RenderCommandEncoder* renderCommandE
 	{
 		if (shader->resourceMapping.uniformBuffersBindingPoint[i] >= 0)
 		{
-			uint32 binding = shader->resourceMapping.uniformBuffersBindingPoint[i];
+    		uint32 binding = shader->resourceMapping.uniformBuffersBindingPoint[i];
-			if (binding >= MAX_MTL_BUFFERS)
+    		if (binding >= MAX_MTL_BUFFERS)
-			{
+    		{
-			    debug_printf("too big buffer index (%u), skipping binding\n", binding);
+    		    debug_printf("too big buffer index (%u), skipping binding\n", binding);
-				continue;
+    			continue;
-			}
+    		}
-			size_t offset = m_state.uniformBufferOffsets[(uint32)shader->shaderType][binding];
+    		size_t offset = m_state.uniformBufferOffsets[(uint32)shader->shaderType][i];
-			if (offset != INVALID_OFFSET)
+    		if (offset != INVALID_OFFSET)
-			{
+    		{
-    			switch (shader->shaderType)
+     			switch (shader->shaderType)
-    			{
+     			{
-    			case LatteConst::ShaderType::Vertex:
+     			case LatteConst::ShaderType::Vertex:
-    			{
+     			{
     				renderCommandEncoder->setVertexBuffer(m_memoryManager->GetBufferCache(), offset, binding);
     				break;
-    			}
+     			}
-    			case LatteConst::ShaderType::Pixel:
+     			case LatteConst::ShaderType::Pixel:
-    			{
+     			{
-    			    renderCommandEncoder->setFragmentBuffer(m_memoryManager->GetBufferCache(), offset, binding);
+     			    renderCommandEncoder->setFragmentBuffer(m_memoryManager->GetBufferCache(), offset, binding);
     				break;
-    			}
+     			}
-    			default:
+     			default:
     				UNREACHABLE;
-    			}
+     			}
-			}
+    		}
 		}
 	}
+
+	// Storage buffer
+	if (shader->resourceMapping.tfStorageBindingPoint >= 0)
+	{
+		debug_printf("storage buffer not implemented, index: %i\n", shader->resourceMapping.tfStorageBindingPoint);
+	}
 }
 
 void MetalRenderer::RebindRenderState(MTL::RenderCommandEncoder* renderCommandEncoder)