From 3c0ca97b5504d817fd09984c452e8e512ae4d1d6 Mon Sep 17 00:00:00 2001
From: wheremyfoodat <gponiris2004@gmail.com>
Date: Sat, 10 Jun 2023 17:54:07 +0300
Subject: [PATCH] [Shader JIT] Add indexed register addressing
---
.../PICA/dynapica/shader_rec_emitter_x64.cpp | 58 +++++++++++++++++++
1 file changed, 58 insertions(+)
diff --git a/src/core/PICA/dynapica/shader_rec_emitter_x64.cpp b/src/core/PICA/dynapica/shader_rec_emitter_x64.cpp
index 6d51bc87..d84de99d 100644
--- a/src/core/PICA/dynapica/shader_rec_emitter_x64.cpp
+++ b/src/core/PICA/dynapica/shader_rec_emitter_x64.cpp
@@ -171,6 +171,23 @@ void ShaderEmitter::loadRegister(Xmm dest, const PICAShader& shader, u32 src, u3
movaps(dest, xword[statePointer + offset]);
else // Swizzle is not trivial so we need to emit a shuffle instruction
pshufd(dest, xword[statePointer + offset], convertedSwizzle);
+
+ // Negate the register if necessary
+ if (negate) {
+ pxor(dest, xword[rip + negateVector]);
+ }
+ return; // Return. Rest of the function handles indexing which is not used if index == 0
+ }
+
+ case 1: {
+ const uintptr_t addrXOffset = uintptr_t(&shader.addrRegister[0]) - uintptr_t(&shader);
+ movsxd(rax, dword[statePointer + addrXOffset]); // rax = address register x
+ break;
+ }
+
+ case 2: {
+ const uintptr_t addrYOffset = uintptr_t(&shader.addrRegister[1]) - uintptr_t(&shader);
+ movsxd(rax, dword[statePointer + addrYOffset]); // rax = address register y
break;
}
@@ -178,6 +195,47 @@ void ShaderEmitter::loadRegister(Xmm dest, const PICAShader& shader, u32 src, u3
Helpers::panic("[ShaderJIT]: Unimplemented source index type %d", index);
}
+ // Here we handle what happens when using indexed addressing & we can't predict what register will be read at compile time
+ // The index of the access is assumed to be in rax
+ // Add source register (src) and index (rax) to form the final register
+ add(rax, src);
+
+ Label maybeTemp, maybeUniform, unknownReg, end;
+ const uintptr_t inputOffset = uintptr_t(&shader.inputs) - uintptr_t(&shader);
+ const uintptr_t tempOffset = uintptr_t(&shader.tempRegisters) - uintptr_t(&shader);
+ const uintptr_t uniformOffset = uintptr_t(&shader.floatUniforms) - uintptr_t(&shader);
+
+ // If reg < 0x10, return inputRegisters[reg]
+ cmp(rax, 0x10);
+ jae(maybeTemp);
+ mov(rcx, rax);
+ shl(rcx, 4); // rcx = rax * sizeof(vec4 of floats) = rax * 16
+ movaps(dest, xword[statePointer + rcx + inputOffset]);
+ jmp(end);
+
+ // If (reg < 0x1F) return tempRegisters[reg - 0x10]
+ L(maybeTemp);
+ cmp(rax, 0x20);
+ jae(maybeUniform);
+ lea(rcx, qword[rax - 0x10]);
+ shl(rcx, 4);
+ movaps(dest, xword[statePointer + rcx + tempOffset]);
+ jmp(end);
+
+ // If (reg < 0x80) return floatUniforms[reg - 0x20]
+ L(maybeUniform);
+ cmp(rax, 0x80);
+ jae(unknownReg);
+ lea(rcx, qword[rax - 0x20]);
+ shl(rcx, 4);
+ movaps(dest, xword[statePointer + rcx + uniformOffset]);
+ jmp(end);
+
+ L(unknownReg);
+ pxor(dest, dest); // Set dest to 0 if we're reading from a garbage register
+
+ L(end);
+ // Negate the register if necessary
if (negate) {
pxor(dest, xword[rip + negateVector]);
}