From a438791e7c250e61409c1fe658b75a6dbd340633 Mon Sep 17 00:00:00 2001
From: "Unknown W. Brackets" <checkins@unknownbrackets.org>
Date: Mon, 18 Feb 2013 23:21:18 -0800
Subject: [PATCH] Initial (very inefficient) vmmov for x86 jit.

This makes #464 work (at least LittleBigPlanet), but only in x86 jit.
---
 Core/MIPS/ARM/ArmCompVFPU.cpp |  4 ++++
 Core/MIPS/ARM/ArmJit.h        |  1 +
 Core/MIPS/MIPSTables.cpp      |  2 +-
 Core/MIPS/x86/CompVFPU.cpp    | 42 +++++++++++++++++++++++++++++++++++
 Core/MIPS/x86/Jit.h           |  1 +
 5 files changed, 49 insertions(+), 1 deletion(-)

diff --git a/Core/MIPS/ARM/ArmCompVFPU.cpp b/Core/MIPS/ARM/ArmCompVFPU.cpp
index 79a30cf52c..ca909fe193 100644
--- a/Core/MIPS/ARM/ArmCompVFPU.cpp
+++ b/Core/MIPS/ARM/ArmCompVFPU.cpp
@@ -63,4 +63,8 @@ namespace MIPSComp
 		DISABLE;
 	}
 
+	void Comp_Vmmov(u32 op) {
+		DISABLE;
+	}
+
 }
diff --git a/Core/MIPS/ARM/ArmJit.h b/Core/MIPS/ARM/ArmJit.h
index 80a6a0d3c4..194c557139 100644
--- a/Core/MIPS/ARM/ArmJit.h
+++ b/Core/MIPS/ARM/ArmJit.h
@@ -116,6 +116,7 @@ public:
 	void Comp_VV2Op(u32 op);
 	void Comp_Mftv(u32 op);
 	void Comp_Vmtvc(u32 op);
+	void Comp_Vmmov(u32 op);
 
 	ArmJitBlockCache *GetBlockCache() { return &blocks; }
 
diff --git a/Core/MIPS/MIPSTables.cpp b/Core/MIPS/MIPSTables.cpp
index 1dde164e39..ed6be36317 100644
--- a/Core/MIPS/MIPSTables.cpp
+++ b/Core/MIPS/MIPSTables.cpp
@@ -669,7 +669,7 @@ const MIPSInstruction tableVFPU6[32] =  //111100 xxx
 
 const MIPSInstruction tableVFPUMatrixSet1[16] = //111100 11100 0xxxx   (rm x is 16)
 {
-	INSTR("vmmov",&Jit::Comp_Generic, Dis_MatrixSet2, Int_Vmmov, IS_VFPU|OUT_EAT_PREFIX),
+	INSTR("vmmov",&Jit::Comp_Vmmov, Dis_MatrixSet2, Int_Vmmov, IS_VFPU|OUT_EAT_PREFIX),
 	{-2},
 	{-2},
 	INSTR("vmidt",&Jit::Comp_Generic, Dis_MatrixSet1, Int_VMatrixInit, IS_VFPU|OUT_EAT_PREFIX),
diff --git a/Core/MIPS/x86/CompVFPU.cpp b/Core/MIPS/x86/CompVFPU.cpp
index 6928a57f01..928684fae9 100644
--- a/Core/MIPS/x86/CompVFPU.cpp
+++ b/Core/MIPS/x86/CompVFPU.cpp
@@ -693,4 +693,46 @@ void Jit::Comp_Vmtvc(u32 op) {
 	}
 }
 
+void Jit::Comp_Vmmov(u32 op) {
+	CONDITIONAL_DISABLE;
+
+	// TODO: This probably ignores prefixes?
+	if (js.MayHavePrefix())
+		DISABLE;
+
+	MatrixSize sz = GetMtxSize(op);
+	int n = GetMatrixSide(sz);
+
+	u8 sregs[16], dregs[16];
+	GetMatrixRegs(sregs, sz, _VS);
+	GetMatrixRegs(dregs, sz, _VD);
+
+	// TODO: gas doesn't allow overlap, what does the PSP do?
+	// Potentially detect overlap or the safe direction to move in, or just DISABLE?
+	// This is very not optimal, blows the regcache everytime.
+	u8 tempregs[16];
+	for (int a = 0; a < n; a++)
+	{
+		for (int b = 0; b < n; b++)
+		{
+			u8 temp = (u8) fpr.GetTempV();
+			fpr.MapRegV(temp, MAP_NOINIT | MAP_DIRTY);
+			MOVSS(fpr.VX(temp), fpr.V(sregs[a * 4 + b]));
+			fpr.StoreFromRegisterV(temp);
+			tempregs[a * 4 + b] = temp;
+		}
+	}
+	for (int a = 0; a < n; a++)
+	{
+		for (int b = 0; b < n; b++)
+		{
+			u8 temp = tempregs[a * 4 + b];
+			fpr.MapRegV(temp, 0);
+			MOVSS(fpr.V(dregs[a * 4 + b]), fpr.VX(temp));
+		}
+	}
+
+	fpr.ReleaseSpillLocks();
+}
+
 }
\ No newline at end of file
diff --git a/Core/MIPS/x86/Jit.h b/Core/MIPS/x86/Jit.h
index dc5082ff81..54c443b53a 100644
--- a/Core/MIPS/x86/Jit.h
+++ b/Core/MIPS/x86/Jit.h
@@ -191,6 +191,7 @@ public:
 	void Comp_VV2Op(u32 op);
 	void Comp_Mftv(u32 op);
 	void Comp_Vmtvc(u32 op);
+	void Comp_Vmmov(u32 op);
 
 	void Comp_DoNothing(u32 op);