this commit is divided in 4 parts:

1 - Optimize pixelshadergen to avoid redundant register overflow math, with this if a game don't need this will be not applied.
this must bring some fill rate back and improve speed a little in fill rate limited systems.
2- some corrections to vertexshadergen to avoid uninitialized texture coordinates, dono if is the correct way to fix it but t least it will make house of the dead overkill playable in dx11.
the bad thing: still missing geometri in dx9, in dx11 it works exactly as on opengl.
3 - some optimization made to improve fps a little wih the latests changes made to fifo.
* back to the original code in beginfield as now it will work right.
* check for efb access more often as a lot of time is lost waiting for efb access
4 - apply a little fix for missing textures in nvidia opengl tanks to  Wagnard28 for finding that nvidia does not like invalid shader id :)
please test for any regression

git-svn-id: https://dolphin-emu.googlecode.com/svn/trunk@5812 8ced0084-cf51-0410-be5f-012b33b47a6e

Source/Core/Common/Src/LinearDiskCache.cpp

@@ -22,7 +22,7 @@ static const char ID[4] = {'D', 'C', 'A', 'C'};
 // Update this to the current SVN revision every time you change shader generation code.
 // We don't automatically get this from SVN_REV because that would mean regenerating the
 // shader cache for every revision, graphics-related or not, which is simply annoying.
-const int version = 5691;
+const int version = 5812;
 
 LinearDiskCache::LinearDiskCache() 
 	: file_(NULL), num_entries_(0) {

Source/Core/Core/Src/HW/VideoInterface.cpp

@@ -729,7 +729,7 @@ void UpdateParameters()
 		TargetRefreshRate = PAL_FIELD_RATE;
 		// AyuanX: Some games are pretty sensitive to this value
 		// So we have to make it run a little faster to prevent potential time out
-		TicksPerFrame = SystemTimers::GetTicksPerSecond() / (PAL_FIELD_RATE + 1);
+		TicksPerFrame = SystemTimers::GetTicksPerSecond() / (PAL_FIELD_RATE  + 1);
 		s_lineCount = m_DisplayControlRegister.NIN ? PAL_LINE_COUNT : (PAL_LINE_COUNT+1)/2;
 		//s_upperFieldBegin = PAL_UPPER_BEGIN;
 		//s_lowerFieldBegin = PAL_LOWER_BEGIN;
@@ -800,19 +800,16 @@ void Update()
 	{
 		// Progressive
 		NewVBeamPos = s_lineCount + 1;
-		BeginField(FIELD_PROGRESSIVE);
 	}
 	else if (m_VBeamPos == s_lineCount)
 	{
 		// Interlace Upper
 		NewVBeamPos = s_lineCount * 2;
-		BeginField(FIELD_UPPER);
 	}
 	else
 	{
 		// Interlace Lower
 		NewVBeamPos = s_lineCount;
-		BeginField(FIELD_LOWER);
 	}
 
 	for (int i = 0; i < 4; i++)
@@ -821,8 +818,25 @@ void Update()
 			m_InterruptRegister[i].IR_INT = 1;
 	}
 	UpdateInterrupts();
+	
+	if (m_DisplayControlRegister.NIN)
+	{
+		// Progressive	
+		BeginField(FIELD_PROGRESSIVE);
+	}
+	else if (m_VBeamPos == s_lineCount)
+	{
+		// Interlace Upper
+		BeginField(FIELD_UPPER);
+	}
+	else
+	{
+		// Interlace Lower
+		BeginField(FIELD_LOWER);
+	}
 
 	m_VBeamPos = (NewVBeamPos > s_lineCount) ? 1 : NewVBeamPos;
+	
 
 	Core::VideoThrottle();
 }

Source/Core/VideoCommon/Src/PixelShaderGen.cpp

@@ -345,6 +345,15 @@ static char swapModeTable[4][5];
 
 static char text[16384];
 
+struct RegisterState
+{
+	bool ColorNeedOverflowControl;
+	bool AlphaNeedOverflowControl;
+	bool AuxStored;
+};
+
+static RegisterState RegisterStates[4];
+
 static void BuildSwapModeTable()
 {
     for (int i = 0; i < 4; i++)
@@ -383,21 +392,10 @@ const char *GeneratePixelShaderCode(u32 texture_mask, bool dstAlphaEnable,  API_
     }
 
     // Declare samplers
-    if (texture_mask)
+    if (texture_mask && ApiType == API_OPENGL)
 	{
-		if (ApiType == API_D3D11)
+		WRITE(p, "uniform samplerRECT ");
-		{
+		bool bfirst = true;
-			WRITE(p, "sampler ");
-		}
-		else if (ApiType == API_D3D9)
-		{
-			WRITE(p, "uniform sampler ");
-		}
-		else
-		{
-			WRITE(p, "uniform samplerRECT ");
-		}
-        bool bfirst = true;
 		for (int i = 0; i < 8; ++i)
 		{
             if (texture_mask & (1<<i))
@@ -406,21 +404,7 @@ const char *GeneratePixelShaderCode(u32 texture_mask, bool dstAlphaEnable,  API_
                 bfirst = false;
             }
 		}
-		WRITE(p, ";\n");
+		WRITE(p, ";\n");		
-		if(ApiType == API_D3D11)
-		{
-			bfirst = true;
-			WRITE(p, "Texture2D ");
-			for (int i = 0; i < 8; ++i)
-			{
-				if (texture_mask & (1<<i))
-				{
-					WRITE(p, "%s Tex%d : register(t%d)", bfirst?"":",", i, i);
-					bfirst = false;
-				}
-			}
-			WRITE(p, ";\n");
-		}        
 	}
 
     if (texture_mask != 0xff) 
@@ -549,19 +533,39 @@ const char *GeneratePixelShaderCode(u32 texture_mask, bool dstAlphaEnable,  API_
         }
     }
 
+	RegisterStates[0].AlphaNeedOverflowControl = false;
+	RegisterStates[0].ColorNeedOverflowControl = false;
+	RegisterStates[0].AuxStored = false;
+	for(int i = 0 ; i < 4 ; i++)
+	{
+		RegisterStates[i].AlphaNeedOverflowControl = true;
+		RegisterStates[i].ColorNeedOverflowControl = true;
+		RegisterStates[i].AuxStored = false;
+	}
 
 	for (int i = 0; i < numStages; i++)
 		WriteStage(p, i, texture_mask,ApiType); //build the equation for this stage
+
 	if(numStages)
 	{
 		// The results of the last texenv stage are put onto the screen,
 		// regardless of the used destination register
-		WRITE(p, "prev.rgb = %s;\n",tevCOutputTable[bpmem.combiners[numStages-1].colorC.dest]);
+		if(bpmem.combiners[numStages - 1].colorC.dest != 0)
-		WRITE(p, "prev.a = %s;\n",tevAOutputTable[bpmem.combiners[numStages-1].alphaC.dest]);
+		{
+			bool retrieveFromAuxRegister = !RegisterStates[bpmem.combiners[numStages - 1].colorC.dest].ColorNeedOverflowControl && RegisterStates[bpmem.combiners[numStages - 1].colorC.dest].AuxStored;
+			WRITE(p, "prev.rgb = %s%s;\n", retrieveFromAuxRegister ? "c" : "" ,tevCOutputTable[bpmem.combiners[numStages - 1].colorC.dest]);
+			RegisterStates[0].ColorNeedOverflowControl = RegisterStates[bpmem.combiners[numStages - 1].colorC.dest].ColorNeedOverflowControl;
+		}
+		if(bpmem.combiners[numStages - 1].alphaC.dest != 0)
+		{
+			bool retrieveFromAuxRegister = !RegisterStates[bpmem.combiners[numStages - 1].alphaC.dest].AlphaNeedOverflowControl && RegisterStates[bpmem.combiners[numStages - 1].alphaC.dest].AuxStored;
+			WRITE(p, "prev.a = %s%s;\n", retrieveFromAuxRegister ? "c" : "" ,tevAOutputTable[bpmem.combiners[numStages - 1].alphaC.dest]);		
+			RegisterStates[0].AlphaNeedOverflowControl = RegisterStates[bpmem.combiners[numStages - 1].alphaC.dest].AlphaNeedOverflowControl;		
+		}		
 	}
-
+	// emulation of unisgned 8 overflow when casting if needed
-	// emulation of unisgned 8 overflow when casting
+	if(RegisterStates[0].AlphaNeedOverflowControl || RegisterStates[0].ColorNeedOverflowControl)
-	WRITE(p, "prev = frac(4.0f + prev * (255.0f/256.0f)) * (256.0f/255.0f);\n");
+		WRITE(p, "prev = frac(4.0f + prev * (255.0f/256.0f)) * (256.0f/255.0f);\n");	
 		
     if (!WriteAlphaTest(p, ApiType))
 	{
@@ -782,7 +786,19 @@ static void WriteStage(char *&p, int n, u32 texture_mask, API_TYPE ApiType)
 	|| ac.a == TEVALPHAARG_APREV 
 	|| ac.b == TEVALPHAARG_APREV 
 	|| ac.c == TEVALPHAARG_APREV)
-		WRITE(p, "cprev = frac(4.0f + prev * (255.0f/256.0f)) * (256.0f/255.0f);\n");  
+	{
+		if(RegisterStates[0].AlphaNeedOverflowControl || RegisterStates[0].ColorNeedOverflowControl)
+		{
+			WRITE(p, "cprev = frac(4.0f + prev * (255.0f/256.0f)) * (256.0f/255.0f);\n");  
+			RegisterStates[0].AlphaNeedOverflowControl = false;
+			RegisterStates[0].ColorNeedOverflowControl = false;			
+		}
+		else
+		{
+			WRITE(p, "cprev = prev;\n");  
+		}
+		RegisterStates[0].AuxStored = true;
+	}
 
 	if(cc.a == TEVCOLORARG_C0 
 	|| cc.a == TEVCOLORARG_A0 
@@ -793,7 +809,20 @@ static void WriteStage(char *&p, int n, u32 texture_mask, API_TYPE ApiType)
 	|| ac.a == TEVALPHAARG_A0 
 	|| ac.b == TEVALPHAARG_A0 
 	|| ac.c == TEVALPHAARG_A0)
-		WRITE(p, "cc0 = frac(4.0f + c0 * (255.0f/256.0f)) * (256.0f/255.0f);\n");
+	{
+		if(RegisterStates[1].AlphaNeedOverflowControl || RegisterStates[1].ColorNeedOverflowControl)
+		{
+			WRITE(p, "cc0 = frac(4.0f + c0 * (255.0f/256.0f)) * (256.0f/255.0f);\n");
+			RegisterStates[1].AlphaNeedOverflowControl = false;
+			RegisterStates[1].ColorNeedOverflowControl = false;
+		}
+		else
+		{
+			WRITE(p, "cc0 = c0;\n");
+		}
+		RegisterStates[1].AuxStored = true;
+	}
+		
 
 	if(cc.a == TEVCOLORARG_C1 
 	|| cc.a == TEVCOLORARG_A1 
@@ -804,7 +833,19 @@ static void WriteStage(char *&p, int n, u32 texture_mask, API_TYPE ApiType)
 	|| ac.a == TEVALPHAARG_A1 
 	|| ac.b == TEVALPHAARG_A1 
 	|| ac.c == TEVALPHAARG_A1)
-		WRITE(p, "cc1 = frac(4.0f + c1 * (255.0f/256.0f)) * (256.0f/255.0f);\n");  
+	{
+		if(RegisterStates[2].AlphaNeedOverflowControl || RegisterStates[2].ColorNeedOverflowControl)
+		{
+			WRITE(p, "cc1 = frac(4.0f + c1 * (255.0f/256.0f)) * (256.0f/255.0f);\n");
+			RegisterStates[2].AlphaNeedOverflowControl = false;
+			RegisterStates[2].ColorNeedOverflowControl = false;
+		}
+		else
+		{
+			WRITE(p, "cc1 = c1;\n");
+		}
+		RegisterStates[2].AuxStored = true;
+	}
 	
 	if(cc.a == TEVCOLORARG_C2 
 	|| cc.a == TEVCOLORARG_A2 
@@ -815,8 +856,22 @@ static void WriteStage(char *&p, int n, u32 texture_mask, API_TYPE ApiType)
 	|| ac.a == TEVALPHAARG_A2 
 	|| ac.b == TEVALPHAARG_A2 
 	|| ac.c == TEVALPHAARG_A2)
-		WRITE(p, "cc2 = frac(4.0f + c2 * (255.0f/256.0f)) * (256.0f/255.0f);\n");  	
+	{
+		if(RegisterStates[3].AlphaNeedOverflowControl || RegisterStates[3].ColorNeedOverflowControl)
+		{
+			WRITE(p, "cc2 = frac(4.0f + c2 * (255.0f/256.0f)) * (256.0f/255.0f);\n");
+			RegisterStates[3].AlphaNeedOverflowControl = false;
+			RegisterStates[3].ColorNeedOverflowControl = false;
+		}
+		else
+		{
+			WRITE(p, "cc2 = c2;\n");
+		}
+		RegisterStates[3].AuxStored = true;
+	}
 
+	RegisterStates[cc.dest].ColorNeedOverflowControl = (cc.clamp == 0);
+	RegisterStates[cc.dest].AuxStored = false;
     if (cc.clamp)
 		WRITE(p, "%s=saturate(", tevCOutputTable[cc.dest]);
 	else
@@ -863,6 +918,8 @@ static void WriteStage(char *&p, int n, u32 texture_mask, API_TYPE ApiType)
 		WRITE(p,")");
 	WRITE(p,";\n");	
     
+	RegisterStates[ac.dest].AlphaNeedOverflowControl = (ac.clamp == 0);
+	RegisterStates[ac.dest].AuxStored = false;
     // combine the alpha channel
     if (ac.clamp)
 	    WRITE(p, "%s=saturate(", tevAOutputTable[ac.dest]);

Source/Core/VideoCommon/Src/VertexShaderGen.cpp

@@ -332,7 +332,7 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type)
 
     // zero left over channels
     for (int i = xfregs.nNumChans; i < 2; ++i)
-		WRITE(p, "o.colors[%d] = float4(0.0f,0.0f,0.0f,0.0f);\n", i);
+		WRITE(p, "o.colors[%d] = float4(0.0f,0.0f,0.0f,1.0f);\n", i);
 
 	// special case if only pos and tex coord 0 and tex coord input is AB11	
 	// donko - this has caused problems in some games. removed for now.
@@ -349,6 +349,7 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type)
         TexMtxInfo& texinfo = xfregs.texcoords[i].texmtxinfo;
 
         WRITE(p, "{\n");
+		WRITE(p, "coord = float4(0.0f,0.0f,1.0f,1.0f);\n"); 
         switch (texinfo.sourcerow) {
         case XF_SRCGEOM_INROW:
             _assert_( texinfo.inputform == XF_TEXINPUT_ABC1 );
@@ -384,7 +385,26 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type)
 
         // firs transformation
         switch (texinfo.texgentype) {
-            case XF_TEXGEN_REGULAR:
+            case XF_TEXGEN_EMBOSS_MAP: // calculate tex coords into bump map
+
+                if (components & (VB_HAS_NRM1|VB_HAS_NRM2)) {
+                    // transform the light dir into tangent space
+                    WRITE(p, "ldir = normalize("I_LIGHTS".lights[%d].pos.xyz - pos.xyz);\n", texinfo.embosslightshift); 
+                    WRITE(p, "o.tex%d.xyz = coord.xyz + float3(dot(ldir, _norm1), dot(ldir, _norm2), 0.0f);\n", i, texinfo.embosssourceshift);
+                }
+                else _assert_(0); // should have normals
+
+                break;
+            case XF_TEXGEN_COLOR_STRGBC0:
+                _assert_(texinfo.sourcerow == XF_SRCCOLORS_INROW);
+                WRITE(p, "o.tex%d.xyz = float3(o.colors[0].x, o.colors[0].y, 1);\n", i);
+                break;
+            case XF_TEXGEN_COLOR_STRGBC1:
+                _assert_(texinfo.sourcerow == XF_SRCCOLORS_INROW);
+                WRITE(p, "o.tex%d.xyz = float3(o.colors[1].x, o.colors[1].y, 1);\n", i);
+                break;
+			case XF_TEXGEN_REGULAR:
+			default:
                 if (components & (VB_HAS_TEXMTXIDX0<<i)) {
                     if (texinfo.projection == XF_TEXPROJ_STQ )
                         WRITE(p, "o.tex%d.xyz = float3(dot(coord, "I_TRANSFORMMATRICES".T[tex%d.z].t), dot(coord, "I_TRANSFORMMATRICES".T[tex%d.z+1].t), dot(coord, "I_TRANSFORMMATRICES".T[tex%d.z+2].t));\n", i, i, i, i);
@@ -399,24 +419,6 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type)
                         WRITE(p, "o.tex%d.xyz = float3(dot(coord, "I_TEXMATRICES".T[%d].t), dot(coord, "I_TEXMATRICES".T[%d].t), 1);\n", i, 3*i, 3*i+1);
                 }
                 break;
-            case XF_TEXGEN_EMBOSS_MAP: // calculate tex coords into bump map
-
-                if (components & (VB_HAS_NRM1|VB_HAS_NRM2)) {
-                    // transform the light dir into tangent space
-                    WRITE(p, "ldir = normalize("I_LIGHTS".lights[%d].pos.xyz - pos.xyz);\n", texinfo.embosslightshift); 
-                    WRITE(p, "o.tex%d.xyz = o.tex%d.xyz + float3(dot(ldir, _norm1), dot(ldir, _norm2), 0.0f);\n", i, texinfo.embosssourceshift);
-                }
-                else _assert_(0); // should have normals
-
-                break;
-            case XF_TEXGEN_COLOR_STRGBC0:
-                _assert_(texinfo.sourcerow == XF_SRCCOLORS_INROW);
-                WRITE(p, "o.tex%d.xyz = float3(o.colors[0].x, o.colors[0].y, 1);\n", i);
-                break;
-            case XF_TEXGEN_COLOR_STRGBC1:
-                _assert_(texinfo.sourcerow == XF_SRCCOLORS_INROW);
-                WRITE(p, "o.tex%d.xyz = float3(o.colors[1].x, o.colors[1].y, 1);\n", i);
-                break;
         }
 
         if (xfregs.bEnableDualTexTransform && texinfo.texgentype == XF_TEXGEN_REGULAR) { // only works for regular tex gen types?

Source/Plugins/Plugin_VideoDX11/Src/VertexManager.cpp

@@ -21,7 +21,7 @@
 #include "D3DBase.h"
 #include "D3DShader.h"
 #include "D3DUtil.h"
-
+#include "Fifo.h"
 #include "Statistics.h"
 #include "Profiler.h"
 #include "FBManager.h"
@@ -304,6 +304,7 @@ void Flush()
 	if (LocalVBuffer == s_pCurBufferPointer) return;
 	if (Flushed) return;
 	Flushed=true;
+	VideoFifo_CheckEFBAccess();
 
 	DVSTARTPROFILE();
 

Source/Plugins/Plugin_VideoDX11/Src/main.cpp

@@ -71,7 +71,6 @@ static bool s_PluginInitialized = false;
 volatile u32 s_swapRequested = FALSE;
 static u32 s_efbAccessRequested = FALSE;
 static volatile u32 s_FifoShuttingDown = FALSE;
-static bool ForceSwap = true;
 
 static volatile struct
 {
@@ -234,7 +233,6 @@ void Video_Prepare()
 	s_efbAccessRequested = FALSE;
 	s_FifoShuttingDown = FALSE;
 	s_swapRequested = FALSE;
-	ForceSwap = true;
 	Renderer::Init();
 	TextureCache::Init();
 	BPInit();
@@ -307,16 +305,16 @@ void Video_SetRendering(bool bEnabled)
 	Fifo_SetRendering(bEnabled);
 }
 
-// run from the graphics thread
+// Run from the graphics thread (from Fifo.cpp)
 void VideoFifo_CheckSwapRequest()
 {
-	if (Common::AtomicLoadAcquire(s_swapRequested))
+	if(g_ActiveConfig.bUseXFB)
 	{
-		if (ForceSwap || g_ActiveConfig.bUseXFB)
+		if (Common::AtomicLoadAcquire(s_swapRequested))
 		{
 			Renderer::Swap(s_beginFieldArgs.xfbAddr, s_beginFieldArgs.field, s_beginFieldArgs.fbWidth, s_beginFieldArgs.fbHeight);
+			Common::AtomicStoreRelease(s_swapRequested, FALSE);
 		}
-		Common::AtomicStoreRelease(s_swapRequested, FALSE);
 	}
 }
 
@@ -325,33 +323,29 @@ inline bool addrRangesOverlap(u32 aLower, u32 aUpper, u32 bLower, u32 bUpper)
 	return !((aLower >= bUpper) || (bLower >= aUpper));
 }
 
-// Run from the graphics thread
+// Run from the graphics thread (from Fifo.cpp)
 void VideoFifo_CheckSwapRequestAt(u32 xfbAddr, u32 fbWidth, u32 fbHeight)
 {
-	if (Common::AtomicLoadAcquire(s_swapRequested) && g_ActiveConfig.bUseXFB)
+	if (g_ActiveConfig.bUseXFB)
 	{
-		u32 aLower = xfbAddr;
+		if(Common::AtomicLoadAcquire(s_swapRequested))
-		u32 aUpper = xfbAddr + 2 * fbWidth * fbHeight;
+		{
-		u32 bLower = s_beginFieldArgs.xfbAddr;
+			u32 aLower = xfbAddr;
-		u32 bUpper = s_beginFieldArgs.xfbAddr + 2 * s_beginFieldArgs.fbWidth * s_beginFieldArgs.fbHeight;
+			u32 aUpper = xfbAddr + 2 * fbWidth * fbHeight;
+			u32 bLower = s_beginFieldArgs.xfbAddr;
+			u32 bUpper = s_beginFieldArgs.xfbAddr + 2 * s_beginFieldArgs.fbWidth * s_beginFieldArgs.fbHeight;
 
-		if (addrRangesOverlap(aLower, aUpper, bLower, bUpper))
+			if (addrRangesOverlap(aLower, aUpper, bLower, bUpper))
-			VideoFifo_CheckSwapRequest();
+				VideoFifo_CheckSwapRequest();
-	}
+		}
-	ForceSwap = false;
+	}	
 }
 
 // Run from the CPU thread (from VideoInterface.cpp)
 void Video_BeginField(u32 xfbAddr, FieldType field, u32 fbWidth, u32 fbHeight)
 {
 	if (s_PluginInitialized && g_ActiveConfig.bUseXFB)
-	{
+	{		
-		s_beginFieldArgs.xfbAddr = xfbAddr;
-		s_beginFieldArgs.field = field;
-		s_beginFieldArgs.fbWidth = fbWidth;
-		s_beginFieldArgs.fbHeight = fbHeight;
-
-		Common::AtomicStoreRelease(s_swapRequested, TRUE);
 		if (g_VideoInitialize.bOnThread)
 		{
 			while (Common::AtomicLoadAcquire(s_swapRequested) && !s_FifoShuttingDown)
@@ -359,7 +353,13 @@ void Video_BeginField(u32 xfbAddr, FieldType field, u32 fbWidth, u32 fbHeight)
 				Common::YieldCPU();
 		}
 		else
-			VideoFifo_CheckSwapRequest();
+			VideoFifo_CheckSwapRequest();				
+		s_beginFieldArgs.xfbAddr = xfbAddr;
+		s_beginFieldArgs.field = field;
+		s_beginFieldArgs.fbWidth = fbWidth;
+		s_beginFieldArgs.fbHeight = fbHeight;
+
+		Common::AtomicStoreRelease(s_swapRequested, TRUE);
 	}
 }
 

Source/Plugins/Plugin_VideoDX9/Src/VertexManager.cpp

@@ -19,7 +19,7 @@
 #include "FileUtil.h"
 
 #include "D3DBase.h"
-
+#include "Fifo.h"
 #include "Statistics.h"
 #include "Profiler.h"
 #include "VertexManager.h"
@@ -226,7 +226,7 @@ void Flush()
 	if (LocalVBuffer == s_pCurBufferPointer) return;
 	if(Flushed) return;
 	Flushed=true;
-
+	VideoFifo_CheckEFBAccess();
 	DVSTARTPROFILE();
 	
 	u32 usedtextures = 0;

Source/Plugins/Plugin_VideoDX9/Src/main.cpp

@@ -65,7 +65,6 @@ static bool s_PluginInitialized = false;
 volatile u32 s_swapRequested = FALSE;
 static u32 s_efbAccessRequested = FALSE;
 static volatile u32 s_FifoShuttingDown = FALSE;
-static bool ForceSwap = true;
 
 static volatile struct
 {
@@ -275,7 +274,6 @@ void Video_Prepare()
 	s_efbAccessRequested = FALSE;
 	s_FifoShuttingDown = FALSE;
 	s_swapRequested = FALSE;
-	ForceSwap = true;
 	Renderer::Init();
 	TextureCache::Init();
 	BPInit();
@@ -342,16 +340,16 @@ void Video_SetRendering(bool bEnabled) {
 	Fifo_SetRendering(bEnabled);
 }
 
-// Run from the graphics thread
+// Run from the graphics thread (from Fifo.cpp)
 void VideoFifo_CheckSwapRequest()
 {
-	if (Common::AtomicLoadAcquire(s_swapRequested))
+	if(g_ActiveConfig.bUseXFB)
 	{
-		if (ForceSwap || g_ActiveConfig.bUseXFB)
+		if (Common::AtomicLoadAcquire(s_swapRequested))
 		{
 			Renderer::Swap(s_beginFieldArgs.xfbAddr, s_beginFieldArgs.field, s_beginFieldArgs.fbWidth, s_beginFieldArgs.fbHeight);
+			Common::AtomicStoreRelease(s_swapRequested, FALSE);
 		}
-		Common::AtomicStoreRelease(s_swapRequested, FALSE);
 	}
 }
 
@@ -360,34 +358,29 @@ inline bool addrRangesOverlap(u32 aLower, u32 aUpper, u32 bLower, u32 bUpper)
 	return !((aLower >= bUpper) || (bLower >= aUpper));
 }
 
-// Run from the graphics thread
+// Run from the graphics thread (from Fifo.cpp)
 void VideoFifo_CheckSwapRequestAt(u32 xfbAddr, u32 fbWidth, u32 fbHeight)
 {
-	if (Common::AtomicLoadAcquire(s_swapRequested) && g_ActiveConfig.bUseXFB)
+	if (g_ActiveConfig.bUseXFB)
 	{
-		u32 aLower = xfbAddr;
+		if(Common::AtomicLoadAcquire(s_swapRequested))
-		u32 aUpper = xfbAddr + 2 * fbWidth * fbHeight;
+		{
-		u32 bLower = s_beginFieldArgs.xfbAddr;
+			u32 aLower = xfbAddr;
-		u32 bUpper = s_beginFieldArgs.xfbAddr + 2 * s_beginFieldArgs.fbWidth * s_beginFieldArgs.fbHeight;
+			u32 aUpper = xfbAddr + 2 * fbWidth * fbHeight;
+			u32 bLower = s_beginFieldArgs.xfbAddr;
+			u32 bUpper = s_beginFieldArgs.xfbAddr + 2 * s_beginFieldArgs.fbWidth * s_beginFieldArgs.fbHeight;
 
-		if (addrRangesOverlap(aLower, aUpper, bLower, bUpper))
+			if (addrRangesOverlap(aLower, aUpper, bLower, bUpper))
-			VideoFifo_CheckSwapRequest();
+				VideoFifo_CheckSwapRequest();
-	}
+		}
-
+	}	
-	ForceSwap = false;
 }
 
 // Run from the CPU thread (from VideoInterface.cpp)
 void Video_BeginField(u32 xfbAddr, FieldType field, u32 fbWidth, u32 fbHeight)
 {
 	if (s_PluginInitialized && g_ActiveConfig.bUseXFB)
-	{
+	{		
-		s_beginFieldArgs.xfbAddr = xfbAddr;
-		s_beginFieldArgs.field = field;
-		s_beginFieldArgs.fbWidth = fbWidth;
-		s_beginFieldArgs.fbHeight = fbHeight;
-
-		Common::AtomicStoreRelease(s_swapRequested, TRUE);
 		if (g_VideoInitialize.bOnThread)
 		{
 			while (Common::AtomicLoadAcquire(s_swapRequested) && !s_FifoShuttingDown)
@@ -395,7 +388,13 @@ void Video_BeginField(u32 xfbAddr, FieldType field, u32 fbWidth, u32 fbHeight)
 				Common::YieldCPU();
 		}
 		else
-			VideoFifo_CheckSwapRequest();
+			VideoFifo_CheckSwapRequest();				
+		s_beginFieldArgs.xfbAddr = xfbAddr;
+		s_beginFieldArgs.field = field;
+		s_beginFieldArgs.fbWidth = fbWidth;
+		s_beginFieldArgs.fbHeight = fbHeight;
+
+		Common::AtomicStoreRelease(s_swapRequested, TRUE);
 	}
 }
 

Source/Plugins/Plugin_VideoOGL/Src/Render.cpp

@@ -557,8 +557,6 @@ void Renderer::ResetAPIState()
 {
 	// Gets us to a reasonably sane state where it's possible to do things like
 	// image copies with textured quads, etc.
-	glDisable(GL_VERTEX_PROGRAM_ARB);// needed by nvidia cards to avoid texture problems
-	glDisable(GL_FRAGMENT_PROGRAM_ARB);// needed by nvidia cards to avoid texture problems
 	VertexShaderCache::DisableShader();
 	PixelShaderCache::DisableShader();
 	glDisable(GL_SCISSOR_TEST);
@@ -586,10 +584,8 @@ void Renderer::RestoreAPIState()
 	SetColorMask();
 	SetBlendMode(true);
 
-	glEnable(GL_VERTEX_PROGRAM_ARB);// needed by nvidia cards o avoid texture problems
+	VertexShaderCache::SetCurrentShader(1);
-	glEnable(GL_FRAGMENT_PROGRAM_ARB);// needed by nvidia cards o avoid texture problems	
+	PixelShaderCache::SetCurrentShader(1);
-	VertexShaderCache::SetCurrentShader(0);
-	PixelShaderCache::SetCurrentShader(0);
 }
 
 void Renderer::SetColorMask()

Source/Plugins/Plugin_VideoOGL/Src/VertexManager.cpp

@@ -202,6 +202,7 @@ void Flush()
 	if (LocalVBuffer == s_pCurBufferPointer) return;
 	if(Flushed) return;
 	Flushed=true;
+	VideoFifo_CheckEFBAccess();
 #if defined(_DEBUG) || defined(DEBUGFAST) 
 	PRIM_LOG("frame%d:\n texgen=%d, numchan=%d, dualtex=%d, ztex=%d, cole=%d, alpe=%d, ze=%d", g_ActiveConfig.iSaveTargetId, xfregs.numTexGens,
 		xfregs.nNumChans, (int)xfregs.bEnableDualTexTransform, bpmem.ztex2.op,

Source/Plugins/Plugin_VideoOGL/Src/main.cpp

@@ -106,7 +106,6 @@ static bool s_PluginInitialized = false;
 volatile u32 s_swapRequested = FALSE;
 static u32 s_efbAccessRequested = FALSE;
 static volatile u32 s_FifoShuttingDown = FALSE;
-static bool ForceSwap = true;
 
 bool IsD3D()
 {
@@ -301,8 +300,7 @@ void Video_Prepare(void)
 
 void Shutdown(void)
 {
-	s_PluginInitialized = false;
+	s_PluginInitialized = false;	
-	ForceSwap = true;
 
 	s_efbAccessRequested = FALSE;
 	s_swapRequested = FALSE;
@@ -368,14 +366,13 @@ static volatile struct
 // Run from the graphics thread (from Fifo.cpp)
 void VideoFifo_CheckSwapRequest()
 {
-	if (Common::AtomicLoadAcquire(s_swapRequested))
+	if(g_ActiveConfig.bUseXFB)
 	{
-		if (ForceSwap || g_ActiveConfig.bUseXFB)
+		if (Common::AtomicLoadAcquire(s_swapRequested))
 		{
 			Renderer::Swap(s_beginFieldArgs.xfbAddr, s_beginFieldArgs.field, s_beginFieldArgs.fbWidth, s_beginFieldArgs.fbHeight);
+			Common::AtomicStoreRelease(s_swapRequested, FALSE);
 		}
-
-		Common::AtomicStoreRelease(s_swapRequested, FALSE);
 	}
 }
 
@@ -387,31 +384,26 @@ inline bool addrRangesOverlap(u32 aLower, u32 aUpper, u32 bLower, u32 bUpper)
 // Run from the graphics thread (from Fifo.cpp)
 void VideoFifo_CheckSwapRequestAt(u32 xfbAddr, u32 fbWidth, u32 fbHeight)
 {
-	if (Common::AtomicLoadAcquire(s_swapRequested) && g_ActiveConfig.bUseXFB)
+	if (g_ActiveConfig.bUseXFB)
 	{
-		u32 aLower = xfbAddr;
+		if(Common::AtomicLoadAcquire(s_swapRequested))
-		u32 aUpper = xfbAddr + 2 * fbWidth * fbHeight;
+		{
-		u32 bLower = s_beginFieldArgs.xfbAddr;
+			u32 aLower = xfbAddr;
-		u32 bUpper = s_beginFieldArgs.xfbAddr + 2 * s_beginFieldArgs.fbWidth * s_beginFieldArgs.fbHeight;
+			u32 aUpper = xfbAddr + 2 * fbWidth * fbHeight;
+			u32 bLower = s_beginFieldArgs.xfbAddr;
+			u32 bUpper = s_beginFieldArgs.xfbAddr + 2 * s_beginFieldArgs.fbWidth * s_beginFieldArgs.fbHeight;
 
-		if (addrRangesOverlap(aLower, aUpper, bLower, bUpper))
+			if (addrRangesOverlap(aLower, aUpper, bLower, bUpper))
-			VideoFifo_CheckSwapRequest();
+				VideoFifo_CheckSwapRequest();
-	}
+		}
-
+	}	
-	ForceSwap = false;
 }
 
 // Run from the CPU thread (from VideoInterface.cpp)
 void Video_BeginField(u32 xfbAddr, FieldType field, u32 fbWidth, u32 fbHeight)
 {
 	if (s_PluginInitialized && g_ActiveConfig.bUseXFB)
-	{
+	{		
-		s_beginFieldArgs.xfbAddr = xfbAddr;
-		s_beginFieldArgs.field = field;
-		s_beginFieldArgs.fbWidth = fbWidth;
-		s_beginFieldArgs.fbHeight = fbHeight;
-
-		Common::AtomicStoreRelease(s_swapRequested, TRUE);
 		if (g_VideoInitialize.bOnThread)
 		{
 			while (Common::AtomicLoadAcquire(s_swapRequested) && !s_FifoShuttingDown)
@@ -420,6 +412,12 @@ void Video_BeginField(u32 xfbAddr, FieldType field, u32 fbWidth, u32 fbHeight)
 		}
 		else
 			VideoFifo_CheckSwapRequest();				
+		s_beginFieldArgs.xfbAddr = xfbAddr;
+		s_beginFieldArgs.field = field;
+		s_beginFieldArgs.fbWidth = fbWidth;
+		s_beginFieldArgs.fbHeight = fbHeight;
+
+		Common::AtomicStoreRelease(s_swapRequested, TRUE);
 	}
 }