Merge pull request #708 from wheremyfoodat/icache

GPU: Add sw texture copies

include/renderer.hpp

@@ -53,6 +53,7 @@ class Renderer {
 
 	EmulatorConfig* emulatorConfig = nullptr;
 
+	void doSoftwareTextureCopy(u32 inputAddr, u32 outputAddr, u32 copySize, u32 inputWidth, u32 inputGap, u32 outputWidth, u32 outputGap);
   public:
 	Renderer(GPU& gpu, const std::array<u32, regNum>& internalRegs, const std::array<u32, extRegNum>& externalRegs);
 	virtual ~Renderer();

src/core/renderer_gl/renderer_gl.cpp

@@ -792,6 +792,8 @@ void RendererGL::textureCopy(u32 inputAddr, u32 outputAddr, u32 totalBytes, u32
 			shutUpCounter++;
 			printf("RendererGL::TextureCopy failed to locate src framebuffer!\n");
 		}
+
+		doSoftwareTextureCopy(inputAddr, outputAddr, copySize, inputWidth, inputGap, outputWidth, outputGap);
 		return;
 	}
 

src/core/renderer_mtl/renderer_mtl.cpp

@@ -426,7 +426,7 @@ void RendererMTL::textureCopy(u32 inputAddr, u32 outputAddr, u32 totalBytes, u32
 	// Find the source surface.
 	auto srcFramebuffer = getColorRenderTarget(inputAddr, PICA::ColorFmt::RGBA8, copyStride, copyHeight, false);
 	if (!srcFramebuffer) {
-		Helpers::warn("RendererMTL::TextureCopy failed to locate src framebuffer!\n");
+		doSoftwareTextureCopy(inputAddr, outputAddr, copySize, inputWidth, inputGap, outputWidth, outputGap);
 		return;
 	}
 	nextRenderPassName = "Clear before texture copy";

src/renderer.cpp

@@ -3,6 +3,8 @@
 #include <algorithm>
 #include <unordered_map>
 
+#include "PICA/gpu.hpp"
+
 Renderer::Renderer(GPU& gpu, const std::array<u32, regNum>& internalRegs, const std::array<u32, extRegNum>& externalRegs)
 	: gpu(gpu), regs(internalRegs), externalRegs(externalRegs) {}
 Renderer::~Renderer() {}
@@ -39,3 +41,39 @@ const char* Renderer::typeToString(RendererType rendererType) {
 		default: return "Invalid";
 	}
 }
+
+void Renderer::doSoftwareTextureCopy(u32 inputAddr, u32 outputAddr, u32 copySize, u32 inputWidth, u32 inputGap, u32 outputWidth, u32 outputGap) {
+	u8* inputPointer = gpu.getPointerPhys<u8>(inputAddr);
+	u8* outputPointer = gpu.getPointerPhys<u8>(outputAddr);
+
+	if (inputPointer == nullptr || outputPointer == nullptr) {
+		return;
+	}
+
+	u32 inputBytesLeft = inputWidth;
+	u32 outputBytesLeft = outputWidth;
+	u32 copyBytesLeft = copySize;
+
+	while (copyBytesLeft > 0) {
+		const u32 bytes = std::min<u32>({inputBytesLeft, outputBytesLeft, copyBytesLeft});
+		std::memcpy(outputPointer, inputPointer, bytes);
+
+		inputPointer += bytes;
+		outputPointer += bytes;
+
+		inputBytesLeft -= bytes;
+		outputBytesLeft -= bytes;
+		copyBytesLeft -= bytes;
+
+		// Apply input and output gap when an input or output line ends
+		if (inputBytesLeft == 0) {
+			inputBytesLeft = inputWidth;
+			inputPointer += inputGap;
+		}
+
+		if (outputBytesLeft == 0) {
+			outputBytesLeft = outputWidth;
+			outputPointer += outputGap;
+		}
+	}
+}