aligned_stack: no real progress... just reached a point where I need to move dev over to my windows so I can have real debugging/disasm/memory dumping ability. (this is the most horrible job I've ever undertaken, btw)

git-svn-id: http://pcsx2.googlecode.com/svn/branches/aligned_stack@2034 96395faa-99c1-11dd-bbfe-3dabce05a288

common/include/x86emitter/instructions.h

@@ -1,6 +1,6 @@
 /*  PCSX2 - PS2 Emulator for PCs
  *  Copyright (C) 2002-2009  PCSX2 Dev Team
- * 
+ *
  *  PCSX2 is free software: you can redistribute it and/or modify it under the terms
  *  of the GNU Lesser General Public License as published by the Free Software Found-
  *  ation, either version 3 of the License, or (at your option) any later version.
@@ -188,6 +188,7 @@ namespace x86Emitter
 	// ----- Miscellaneous Instructions  -----
 	// Various Instructions with no parameter and no special encoding logic.
 
+	extern void xLEAVE();
 	extern void xRET();
 	extern void xCBW();
 	extern void xCWD();
@@ -244,7 +245,7 @@ namespace x86Emitter
 
 	// ------------------------------------------------------------------------
 	// Forward Jump Helpers (act as labels!)
-	
+
 #define DEFINE_FORWARD_JUMP( label, cond ) \
 	template< typename OperandType > \
 	class xForward##label : public xForwardJump<OperandType> \
@@ -328,7 +329,7 @@ namespace x86Emitter
 
 	DEFINE_FORWARD_JUMP( JPE,	Jcc_ParityEven );
 	DEFINE_FORWARD_JUMP( JPO,	Jcc_ParityOdd );
-	
+
 	typedef xForwardJPE<s8>		xForwardJPE8;
 	typedef xForwardJPE<s32>	xForwardJPE32;
 	typedef xForwardJPO<s8>		xForwardJPO8;
@@ -362,7 +363,7 @@ namespace x86Emitter
 
 	extern void xMOVQZX( const xRegisterSSE& to, const ModSibBase& src );
 	extern void xMOVQZX( const xRegisterSSE& to, const xRegisterSSE& from );
-	
+
 	extern void xMOVSS( const xRegisterSSE& to, const xRegisterSSE& from );
 	extern void xMOVSS( const ModSibBase& to, const xRegisterSSE& from );
 	extern void xMOVSD( const xRegisterSSE& to, const xRegisterSSE& from );
@@ -421,12 +422,12 @@ namespace x86Emitter
 	extern void xEXTRACTPS( const ModSibStrict<u32>& dest, const xRegisterSSE& from, u8 imm8 );
 
 	// ------------------------------------------------------------------------
-	
+
 	extern const Internal::SimdImpl_DestRegEither<0x66,0xdb> xPAND;
 	extern const Internal::SimdImpl_DestRegEither<0x66,0xdf> xPANDN;
 	extern const Internal::SimdImpl_DestRegEither<0x66,0xeb> xPOR;
 	extern const Internal::SimdImpl_DestRegEither<0x66,0xef> xPXOR;
-	
+
 	extern const Internal::SimdImpl_AndNot			xANDN;
 
 	extern const Internal::SimdImpl_COMI<true>		xCOMI;
@@ -434,7 +435,7 @@ namespace x86Emitter
 	extern const Internal::SimdImpl_rSqrt<0x53>		xRCP;
 	extern const Internal::SimdImpl_rSqrt<0x52>		xRSQRT;
 	extern const Internal::SimdImpl_Sqrt<0x51>		xSQRT;
-	
+
 	extern const Internal::SimdImpl_MinMax<0x5f>	xMAX;
 	extern const Internal::SimdImpl_MinMax<0x5d>	xMIN;
 	extern const Internal::SimdImpl_Shuffle<0xc6>	xSHUF;
@@ -442,7 +443,7 @@ namespace x86Emitter
 	// ------------------------------------------------------------------------
 
 	extern const Internal::SimdImpl_DestRegSSE<0x66,0x1738> xPTEST;
-	
+
 	extern const Internal::SimdImpl_Compare<SSE2_Equal>			xCMPEQ;
 	extern const Internal::SimdImpl_Compare<SSE2_Less>			xCMPLT;
 	extern const Internal::SimdImpl_Compare<SSE2_LessOrEqual>	xCMPLE;
@@ -484,9 +485,9 @@ namespace x86Emitter
 
 	extern const Internal::SimdImpl_DestRegStrict<0xf2,0x2c,xRegister32, xRegisterSSE,u64>	xCVTTSD2SI;
 	extern const Internal::SimdImpl_DestRegStrict<0xf3,0x2c,xRegister32, xRegisterSSE,u32>	xCVTTSS2SI;
-	
+
 	// ------------------------------------------------------------------------
-	
+
 	extern const Internal::SimdImpl_Shift<0xf0, 6>		xPSLL;
 	extern const Internal::SimdImpl_Shift<0xd0, 2>		xPSRL;
 	extern const Internal::SimdImpl_ShiftWithoutQ<0xe0, 4> xPSRA;
@@ -502,7 +503,7 @@ namespace x86Emitter
 	extern const Internal::SimdImpl_PUnpack				xPUNPCK;
 	extern const Internal::SimdImpl_Unpack				xUNPCK;
 	extern const Internal::SimdImpl_Pack				xPACK;
-	
+
 	extern const Internal::SimdImpl_PAbsolute			xPABS;
 	extern const Internal::SimdImpl_PSign				xPSIGN;
 	extern const Internal::SimdImpl_PInsert				xPINSR;
@@ -513,7 +514,7 @@ namespace x86Emitter
 	extern const Internal::SimdImpl_Blend				xBLEND;
 	extern const Internal::SimdImpl_DotProduct			xDP;
 	extern const Internal::SimdImpl_Round				xROUND;
-	
+
 	extern const Internal::SimdImpl_PMove<true>			xPMOVSX;
 	extern const Internal::SimdImpl_PMove<false>		xPMOVZX;
 

common/src/x86emitter/x86emitter.cpp

@@ -729,6 +729,7 @@ __forceinline void xPOPFD()						{ xWrite8( 0x9D ); }
 //////////////////////////////////////////////////////////////////////////////////////////
 //
 
+__forceinline void xLEAVE()	{ xWrite8( 0xC9 ); }
 __forceinline void xRET()	{ xWrite8( 0xC3 ); }
 __forceinline void xCBW()	{ xWrite16( 0x9866 );  }
 __forceinline void xCWD()	{ xWrite8( 0x98 ); }

pcsx2/x86/ix86-32/iR5900-32.cpp

@@ -328,6 +328,7 @@ static DynGenFunc* DispatcherReg		= NULL;
 static DynGenFunc* JITCompile			= NULL;
 static DynGenFunc* JITCompileInBlock	= NULL;
 static DynGenFunc* EnterRecompiledCode	= NULL;
+static DynGenFunc* ExitRecompiledCode	= NULL;
 
 // parameters:
 //   espORebp - 0 for ESP, or 1 for EBP.
@@ -427,41 +428,45 @@ static DynGenFunc* _DynGen_EnterRecompiledCode()
 	// use for supplying parameters to cdecl functions.
 
 	xPUSH( ebp );
-	xPUSH( edi );
-	xPUSH( esi );
-	xPUSH( ebx );
-	
 	xMOV( ebp, esp );
+
 	xAND( esp, -0x10 );
-	xSUB( esp, 0x10 );
 
-	xMOV( &s_store_ebp, ebp );
+	// First 0x10 is for esi, edi, etc. Second 0x10 is for the return address and ebp.  The
+	// third second 0x10 is for C-style CDECL calls we might make from the recompiler
+	// (parameters for those calls can be stored there!)
+
+	xSUB( esp, 0x30 );
+
+	xMOV( ptr[ebp-12], edi );
+	xMOV( ptr[ebp-8], esi );
+	xMOV( ptr[ebp-4], ebx );
+
+	// Simulate a CALL function by pushing the call address and EBP onto the stack.
+	xMOV( ptr32[esp+0x10+12], 0xffeeff );
+	uptr& imm = *(uptr*)(xGetPtr()-4);
+	xMOV( ptr32[esp+0x10+8], ebp );
+
 	xMOV( &s_store_esp, esp );
-	xSUB( ptr32[&s_store_esp], 4 );		// account for the address pushed when we xCALL
+	xMOV( &s_store_ebp, ebp );
 
-	//xPUSH( edi );
+	xJMP( ptr32[&DispatcherReg] );
-	//xPUSH( esi );
+	imm = (uptr)xGetPtr();
-	//xPUSH( ebx );
+	ExitRecompiledCode = (DynGenFunc*)xGetPtr();
 
-	xCALL( ptr32[&DispatcherReg] );
+	xLEAVE();
+	//xMOV( esp, ebp );
+	//xPOP( ebp );
 
-	//xPOP( ebx );
+	//_DynGen_StackFrameCheck();
-	//xPOP( esi );
+
-	//xPOP( edi );
+	xMOV( edi, ptr[ebp-12] );
+	xMOV( esi, ptr[ebp-8] );
+	xMOV( ebx, ptr[ebp-4] );
 
 	//xMOV( esp, ebp );
 	//xPOP( ebp );
-	//xRET();
+	xLEAVE();
-
-	xADD( ptr32[&s_store_esp], 4 );		// account for the address pushed when we xCALL
-	_DynGen_StackFrameCheck();
-
-	xMOV( esp, ebp );
-	
-	xPOP( ebx );
-	xPOP( esi );
-	xPOP( edi );
-	xPOP( ebp );
 	xRET();
 
 	return (DynGenFunc*)retval;
@@ -808,18 +813,16 @@ void recClear(u32 addr, u32 size)
 
 	upperextent = min(upperextent, ceiling);
 
-#ifdef PCSX2_DEVBUILD
 	for (int i = 0; pexblock = recBlocks[i]; i++) {
 		if (s_pCurBlock == PC_GETBLOCK(pexblock->startpc))
 			continue;
 		u32 blockend = pexblock->startpc + pexblock->size * 4;
 		if (pexblock->startpc >= addr && pexblock->startpc < addr + size * 4
 		 || pexblock->startpc < addr && blockend > addr) {
-			Console.Error( "Impossible block clearing failure" );
+			DevCon.Error( "Impossible block clearing failure" );
-			pxFail( "Impossible block clearing failure" );
+			pxFailDev( "Impossible block clearing failure" );
 		}
 	}
-#endif
 
 	if (upperextent > lowerextent)
 		ClearRecLUT(PC_GETBLOCK(lowerextent), (upperextent - lowerextent) / 4);
@@ -1284,6 +1287,8 @@ void __fastcall dyna_page_reset(u32 start,u32 sz)
 
 	// Note: this function is accessed via a JMP, and thus the RET here will exit
 	// recompiled code and take us back to recExecute.
+
+	__asm__ __volatile__( "leave\n jmp %[exitRec]\n" : : [exitRec] "m" (ExitRecompiledCode) : );
 }
 
 void recRecompile( const u32 startpc )