Panda3DS/src/core/PICA/regs.cpp

#include "PICA/regs.hpp"

#include "PICA/gpu.hpp"

using namespace Floats;
using namespace Helpers;

u32 GPU::readReg(u32 address) {
	if (address >= 0x1EF01000 && address < 0x1EF01C00) {  // Internal registers
		const u32 index = (address - 0x1EF01000) / sizeof(u32);
		return readInternalReg(index);
	} else {
		log("Ignoring read to external GPU register %08X.\n", address);
		return 0;
	}
}

void GPU::writeReg(u32 address, u32 value) {
	if (address >= 0x1EF01000 && address < 0x1EF01C00) {  // Internal registers
		const u32 index = (address - 0x1EF01000) / sizeof(u32);
		writeInternalReg(index, value, 0xffffffff);
	} else if (address >= 0x1EF00004 && address < 0x1EF01000) {
		const u32 index = (address - 0x1EF00004) / sizeof(u32);
		writeExternalReg(index, value);
	} else {
		log("Ignoring write to unknown GPU register %08X. Value: %08X\n", address, value);
	}
}

u32 GPU::readExternalReg(u32 index) {
	using namespace PICA::ExternalRegs;

	if (index > 0x1000) [[unlikely]] {
		Helpers::panic("Tried to read invalid external GPU register. Index: %X\n", index);
		return -1;
	}

	return externalRegs[index];
}

void GPU::writeExternalReg(u32 index, u32 value) {
	using namespace PICA::ExternalRegs;

	if (index > 0x1000) [[unlikely]] {
		Helpers::panic("Tried to write to invalid external GPU register. Index: %X, value: %08X\n", index, value);
		return;
	}

	externalRegs[index] = value;
}

u32 GPU::readInternalReg(u32 index) {
	using namespace PICA::InternalRegs;

	if (index > regNum) [[unlikely]] {
		Helpers::panic("Tried to read invalid GPU register. Index: %X\n", index);
		return 0;
	}

	else if (index >= LightingLUTData0 && index <= LightingLUTData7) [[unlikely]] {
		const uint32_t index = regs[LightingLUTIndex];  // Get full LUT index register
		const uint32_t lutID = getBits<8, 5>(index);    // Get which LUT we're actually writing to
		uint32_t lutIndex = getBits<0, 8>(index);       // And get the index inside the LUT we're writing to
		uint32_t value = 0xffffffff;                    // Return value

		if (lutID < PICA::Lights::LUT_Count) {
			value = lightingLUT[lutID * 256 + lutIndex];
		}

		// Increment the bottom 8 bits of the lighting LUT index register
		lutIndex += 1;
		regs[LightingLUTIndex] = (index & ~0xff) | (lutIndex & 0xff);
		return value;
	}

	return regs[index];
}

void GPU::writeInternalReg(u32 index, u32 value, u32 mask) {
	using namespace PICA::InternalRegs;

	if (index > regNum) [[unlikely]] {
		Helpers::panic("Tried to write to invalid GPU register. Index: %X, value: %08X\n", index, value);
		return;
	}

	u32 currentValue = regs[index];
	u32 newValue = (currentValue & ~mask) | (value & mask);  // Only overwrite the bits specified by "mask"
	regs[index] = newValue;

	// TODO: Figure out if things like the shader index use the unmasked value or the masked one
	// We currently use the unmasked value like Citra does
	switch (index) {
		case SignalDrawArrays:
			if (value != 0) drawArrays(false);
			break;

		case SignalDrawElements:
			if (value != 0) drawArrays(true);
			break;

		case AttribFormatHigh:
			totalAttribCount = (value >> 28) + 1;      // Total number of vertex attributes
			fixedAttribMask = getBits<16, 12>(value);  // Determines which vertex attributes are fixed for all vertices
			break;

		case ColourBufferLoc: {
			u32 loc = (value & 0x0fffffff) << 3;
			renderer->setColourBufferLoc(loc);
			break;
		};

		case ColourBufferFormat: {
			u32 format = getBits<16, 3>(value);
			renderer->setColourFormat(static_cast<PICA::ColorFmt>(format));
			break;
		}

		case DepthBufferLoc: {
			u32 loc = (value & 0x0fffffff) << 3;
			renderer->setDepthBufferLoc(loc);
			break;
		}

		case DepthBufferFormat: {
			u32 format = value & 0x3;
			renderer->setDepthFormat(static_cast<PICA::DepthFmt>(format));
			break;
		}

		case FramebufferSize: {
			const u32 width = value & 0x7ff;
			const u32 height = getBits<12, 10>(value) + 1;
			renderer->setFBSize(width, height);
			break;
		}

		case FogLUTData0:
		case FogLUTData1:
		case FogLUTData2:
		case FogLUTData3:
		case FogLUTData4:
		case FogLUTData5:
		case FogLUTData6:
		case FogLUTData7: {
			const uint32_t index = regs[FogLUTIndex] & 0x7F;
			fogLUT[index] = value;
			fogLUTDirty = true;
			regs[FogLUTIndex] = (index + 1) & 0x7F;
			break;
		}

		case LightingLUTData0:
		case LightingLUTData1:
		case LightingLUTData2:
		case LightingLUTData3:
		case LightingLUTData4:
		case LightingLUTData5:
		case LightingLUTData6:
		case LightingLUTData7: {
			const uint32_t index = regs[LightingLUTIndex];  // Get full LUT index register
			const uint32_t lutID = getBits<8, 5>(index);    // Get which LUT we're actually writing to
			uint32_t lutIndex = getBits<0, 8>(index);       // And get the index inside the LUT we're writing to

			if (lutID < PICA::Lights::LUT_Count) {
				lightingLUT[lutID * 256 + lutIndex] = newValue;
				lightingLUTDirty = true;
			}

			// Increment the bottom 8 bits of the lighting LUT index register
			lutIndex += 1;
			regs[LightingLUTIndex] = (index & ~0xff) | (lutIndex & 0xff);

			break;
		}

		case VertexFloatUniformIndex: {
			shaderUnit.vs.setFloatUniformIndex(value);
			break;
		}

		case VertexFloatUniformData0:
		case VertexFloatUniformData1:
		case VertexFloatUniformData2:
		case VertexFloatUniformData3:
		case VertexFloatUniformData4:
		case VertexFloatUniformData5:
		case VertexFloatUniformData6:
		case VertexFloatUniformData7: {
			shaderUnit.vs.uploadFloatUniform(value);
			break;
		}

		case FixedAttribIndex:
			fixedAttribCount = 0;
			fixedAttribIndex = value & 0xf;

			if (fixedAttribIndex == 0xf) {
				log("[PICA] Immediate mode vertex submission enabled");
				immediateModeAttrIndex = 0;
				immediateModeVertIndex = 0;
			}
			break;

		// Restart immediate mode primitive drawing
		case PrimitiveRestart:
			if (value & 1) {
				immediateModeAttrIndex = 0;
				immediateModeVertIndex = 0;
			}
			break;

		case FixedAttribData0:
		case FixedAttribData1:
		case FixedAttribData2:
			fixedAttrBuff[fixedAttribCount++] = value;

			if (fixedAttribCount == 3) {
				fixedAttribCount = 0;

				vec4f attr;
				// These are stored in the reverse order anyone would expect them to be in
				attr[0] = f24::fromRaw(fixedAttrBuff[2] & 0xffffff);
				attr[1] = f24::fromRaw(((fixedAttrBuff[1] & 0xffff) << 8) | (fixedAttrBuff[2] >> 24));
				attr[2] = f24::fromRaw(((fixedAttrBuff[0] & 0xff) << 16) | (fixedAttrBuff[1] >> 16));
				attr[3] = f24::fromRaw(fixedAttrBuff[0] >> 8);

				// If the fixed attribute index is < 12, we're just writing to one of the fixed attributes
				if (fixedAttribIndex < 12) [[likely]] {
					shaderUnit.vs.fixedAttributes[fixedAttribIndex++] = attr;
				} else if (fixedAttribIndex == 15) {  // Otherwise if it's 15, we're submitting an immediate mode vertex
					const uint totalAttrCount = (regs[PICA::InternalRegs::VertexShaderAttrNum] & 0xf) + 1;
					if (totalAttrCount <= immediateModeAttrIndex) {
						printf("Broken state in the immediate mode vertex submission pipeline. Failing silently\n");
						immediateModeAttrIndex = 0;
						immediateModeVertIndex = 0;
					}

					immediateModeAttributes[immediateModeAttrIndex++] = attr;
					if (immediateModeAttrIndex == totalAttrCount) {
						PICA::Vertex v = getImmediateModeVertex();
						immediateModeAttrIndex = 0;
						immediateModeVertices[immediateModeVertIndex++] = v;

						// Get primitive type
						const u32 primConfig = regs[PICA::InternalRegs::PrimitiveConfig];
						const u32 primType = getBits<8, 2>(primConfig);

						// If we've reached 3 verts, issue a draw call
						// Handle rendering depending on the primitive type
						if (immediateModeVertIndex == 3) {
							renderer->prepareForDraw(shaderUnit, nullptr);
							renderer->drawVertices(PICA::PrimType::TriangleList, immediateModeVertices);

							switch (primType) {
								// Triangle or geometry primitive. Draw a triangle and discard all vertices
								case 0:
								case 3: {
									immediateModeVertIndex = 0;
									break;
								}

								// Triangle strip. Draw triangle, discard first vertex and keep the last 2
								case 1:
									immediateModeVertIndex = 2;

									immediateModeVertices[0] = immediateModeVertices[1];
									immediateModeVertices[1] = immediateModeVertices[2];
									break;

								// Triangle fan. Draw triangle, keep first vertex and last vertex, discard second vertex
								case 2:
									immediateModeVertIndex = 2;
									immediateModeVertices[1] = immediateModeVertices[2];
									break;
							}
						}
					}
				} else {  // Writing to fixed attributes 13 and 14 probably does nothing, but we'll see
					log("Wrote to invalid fixed vertex attribute %d\n", fixedAttribIndex);
				}
			}

			break;

		case VertexShaderOpDescriptorIndex: {
			shaderUnit.vs.setOpDescriptorIndex(newValue);
			break;
		}

		case VertexShaderOpDescriptorData0:
		case VertexShaderOpDescriptorData1:
		case VertexShaderOpDescriptorData2:
		case VertexShaderOpDescriptorData3:
		case VertexShaderOpDescriptorData4:
		case VertexShaderOpDescriptorData5:
		case VertexShaderOpDescriptorData6:
		case VertexShaderOpDescriptorData7: {
			shaderUnit.vs.uploadDescriptor(value);
			break;
		}

		case VertexBoolUniform: {
			shaderUnit.vs.uploadBoolUniform(newValue & 0xffff);
			break;
		}

		case VertexIntUniform0:
		case VertexIntUniform1:
		case VertexIntUniform2:
		case VertexIntUniform3: {
			shaderUnit.vs.uploadIntUniform(index - VertexIntUniform0, newValue);
			break;
		}

		case VertexShaderData0:
		case VertexShaderData1:
		case VertexShaderData2:
		case VertexShaderData3:
		case VertexShaderData4:
		case VertexShaderData5:
		case VertexShaderData6:
		case VertexShaderData7: {
			shaderUnit.vs.uploadWord(value);
			break;
		}

		case VertexShaderEntrypoint: {
			shaderUnit.vs.entrypoint = newValue & 0xffff;
			break;
		}

		/* TODO: Find out if this actually does anything
		case VertexShaderTransferEnd:
			if (value != 0) shaderUnit.vs.finalize();
			break;
		*/

		case VertexShaderTransferIndex: shaderUnit.vs.setBufferIndex(newValue); break;

		// Command lists can write to the command processor registers and change the command list stream
		// Several games are known to do this, including New Super Mario Bros 2 and Super Mario 3D Land
		case CmdBufTrigger0:
		case CmdBufTrigger1: {
			if (newValue != 0) {                           // A non-zero value triggers command list processing
				int bufferIndex = index - CmdBufTrigger0;  // Index of the command buffer to execute (0 or 1)
				u32 addr = (regs[CmdBufAddr0 + bufferIndex] & 0xfffffff) << 3;
				u32 size = (regs[CmdBufSize0 + bufferIndex] & 0xfffff) << 3;

				// Set command buffer state to execute the new buffer
				cmdBuffStart = getPointerPhys<u32>(addr);
				cmdBuffCurr = cmdBuffStart;
				cmdBuffEnd = cmdBuffStart + (size / sizeof(u32));
			}
			break;
		}

		default:
			// Vertex attribute registers
			if (index >= AttribInfoStart && index <= AttribInfoEnd) {
				uint attributeIndex = (index - AttribInfoStart) / 3;  // Which attribute are we writing to
				uint reg = (index - AttribInfoStart) % 3;             // Which of this attribute's registers are we writing to?
				auto& attr = attributeInfo[attributeIndex];

				switch (reg) {
					case 0: attr.offset = value & 0xfffffff; break;  // Attribute offset
					case 1: attr.config1 = value; break;
					case 2:
						attr.config2 = value;
						attr.size = getBits<16, 8>(value);
						attr.componentCount = value >> 28;
						break;
				}
			} else {
				log("GPU: Wrote to unimplemented internal reg: %X, value: %08X\n", index, newValue);
			}
			break;
	}
}

void GPU::startCommandList(u32 addr, u32 size) {
	cmdBuffStart = static_cast<u32*>(mem.getReadPointer(addr));
	if (!cmdBuffStart) Helpers::panic("Couldn't get buffer for command list");
	// TODO: This is very memory unsafe. We get a pointer to FCRAM and just keep writing without checking if we're gonna go OoB

	cmdBuffCurr = cmdBuffStart;
	cmdBuffEnd = cmdBuffStart + (size / sizeof(u32));

	// LUT for converting the parameter mask to an actual 32-bit mask
	// The parameter mask is 4 bits long, each bit corresponding to one byte of the mask
	// If the bit is 0 then the corresponding mask byte is 0, otherwise the mask byte is 0xff
	// So for example if the parameter mask is 0b1001, the full mask is 0xff'00'00'ff
	static constexpr std::array<u32, 16> maskLUT = {
		0x00000000, 0x000000ff, 0x0000ff00, 0x0000ffff, 0x00ff0000, 0x00ff00ff, 0x00ffff00, 0x00ffffff,
		0xff000000, 0xff0000ff, 0xff00ff00, 0xff00ffff, 0xffff0000, 0xffff00ff, 0xffffff00, 0xffffffff,
	};

	while (cmdBuffCurr < cmdBuffEnd) {
		// If the buffer is not aligned to an 8 byte boundary, force align it by moving the pointer up a word
		// The curr pointer starts out doubleword-aligned and is increased by 4 bytes each time
		// So to check if it is aligned, we get the number of words it's been incremented by
		// If that number is an odd value then the buffer is not aligned, otherwise it is
		if ((cmdBuffCurr - cmdBuffStart) % 2 != 0) {
			cmdBuffCurr++;
		}

		// The first word of a command is the command parameter and the second one is the header
		u32 param1 = *cmdBuffCurr++;
		u32 header = *cmdBuffCurr++;

		u32 id = header & 0xffff;
		u32 paramMaskIndex = getBits<16, 4>(header);
		u32 paramCount = getBits<20, 8>(header);  // Number of additional parameters
		// Bit 31 tells us whether this command is going to write to multiple sequential registers (if the bit is 1)
		// Or if all written values will go to the same register (If the bit is 0). It's essentially the value that
		// gets added to the "id" field after each register write
		bool consecutiveWritingMode = (header >> 31) != 0;

		u32 mask = maskLUT[paramMaskIndex];  // Actual parameter mask
		// Increment the ID by 1 after each write if we're in consecutive mode, or 0 otherwise
		u32 idIncrement = (consecutiveWritingMode) ? 1 : 0;

		writeInternalReg(id, param1, mask);
		for (u32 i = 0; i < paramCount; i++) {
			id += idIncrement;
			u32 param = *cmdBuffCurr++;
			writeInternalReg(id, param, mask);
		}
	}
}