mixed commit:

in D3D and Opengl:
fixed one nasty bug in texture loading where if a dynamic texture keeps his format but the tlut format is changed, the try or reloading the texture in the same texture could cause a hang if the size of the resulting texture is different than the original (size in bytes)
Applied a ugly temporal hack to the texture conversor to solve efb to ram misalignments and effect distortions.
in D3D:
Pseudo implementation of logic ops using basic blending: the first 8 operations are "good approximations", the remaining 8 are bullshit :) if someone have a better approximation to emulate this logic please let me know.
please test if i don't break anything in the process and test Mario kart wee you will get a nice surprise.:)
 

git-svn-id: https://dolphin-emu.googlecode.com/svn/trunk@4656 8ced0084-cf51-0410-be5f-012b33b47a6e

4 files changed, 324 insertions, 28 deletions

diff --git a/Source/Core/VideoCommon/Src/PixelShaderGen.cpp b/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
index 439400e347..1249011538 100644
--- a/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
+++ b/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
@@ -545,7 +545,7 @@ const char *GeneratePixelShader(u32 texture_mask, bool dstAlphaEnable, u32 HLSL)
         // alpha test will always fail, so restart the shader and just make it an empty function
         p = pmainstart;
         WRITE(p, "ocol0 = 0;\n");
-		WRITE(p, HLSL ? "clip(-1);" : "discard;\n");
+		WRITE(p, "discard;return;\n");
     }
     else
 	{
@@ -582,9 +582,9 @@ static const char *TEVCMPColorOPTable[16] =
 	"float3(0.0f,0.0f,0.0f)",//7
 	"   %s + ((%s.r > %s.r + (1.0f/510.0f)) ? %s : float3(0.0f,0.0f,0.0f))",//#define TEVCMP_R8_GT 8
 	"   %s + ((abs(%s.r - %s.r) < (1.0f/255.0f)) ? %s : float3(0.0f,0.0f,0.0f))",//#define TEVCMP_R8_EQ 9
-	"   %s + (( dot(%s.rgb, comp16) >= (dot(%s.rgb, comp16) + (1.0f/510.0f))) ? %s : float3(0.0f,0.0f,0.0f))",//#define TEVCMP_GR16_GT 10
+	"   %s + (( dot(%s.rgb, comp16) > (dot(%s.rgb, comp16) + (1.0f/510.0f))) ? %s : float3(0.0f,0.0f,0.0f))",//#define TEVCMP_GR16_GT 10
 	"   %s + (abs(dot(%s.rgb, comp16) - dot(%s.rgb, comp16)) < (1.0f/255.0f) ? %s : float3(0.0f,0.0f,0.0f))",//#define TEVCMP_GR16_EQ 11
-	"   %s + (( dot(%s.rgb, comp24) >= (dot(%s.rgb, comp24) + (1.0f/510.0f))) ? %s : float3(0.0f,0.0f,0.0f))",//#define TEVCMP_BGR24_GT 12
+	"   %s + (( dot(%s.rgb, comp24) > (dot(%s.rgb, comp24) + (1.0f/510.0f))) ? %s : float3(0.0f,0.0f,0.0f))",//#define TEVCMP_BGR24_GT 12
 	"   %s + (abs(dot(%s.rgb, comp24) - dot(%s.rgb, comp24)) < (1.0f/255.0f) ? %s : float3(0.0f,0.0f,0.0f))",//#define TEVCMP_BGR24_EQ 13
 	"   %s + (max(sign(%s.rgb - %s.rgb - (1.0f/510.0f)),float3(0.0f,0.0f,0.0f)) * %s)",//#define TEVCMP_RGB8_GT  14
 	"   %s + ((float3(1.0f,1.0f,1.0f) - max(sign(abs(%s.rgb - %s.rgb) - (1.0f/255.0f)),float3(0.0f,0.0f,0.0f))) * %s)"//#define TEVCMP_RGB8_EQ  15
@@ -601,13 +601,13 @@ static const char *TEVCMPAlphaOPTable[16] =
 	"0.0f",//5
 	"0.0f",//6
 	"0.0f",//7
-	"   %s + ((%s.r >= (%s.r + (1.0f/510.0f))) ? %s : 0.0f)",//#define TEVCMP_R8_GT 8
+	"   %s + ((%s.r > (%s.r + (1.0f/510.0f))) ? %s : 0.0f)",//#define TEVCMP_R8_GT 8
 	"   %s + (abs(%s.r - %s.r) < (1.0f/255.0f) ? %s : 0.0f)",//#define TEVCMP_R8_EQ 9
-	"   %s + ((dot(%s.rgb, comp16) >= (dot(%s.rgb, comp16) + (1.0f/510.0f))) ? %s : 0.0f)",//#define TEVCMP_GR16_GT 10
+	"   %s + ((dot(%s.rgb, comp16) > (dot(%s.rgb, comp16) + (1.0f/510.0f))) ? %s : 0.0f)",//#define TEVCMP_GR16_GT 10
 	"   %s + (abs(dot(%s.rgb, comp16) - dot(%s.rgb, comp16)) < (1.0f/255.0f) ? %s : 0.0f)",//#define TEVCMP_GR16_EQ 11
-	"   %s + ((dot(%s.rgb, comp24) >= (dot(%s.rgb, comp24) + (1.0f/510.0f))) ? %s : 0.0f)",//#define TEVCMP_BGR24_GT 12
+	"   %s + ((dot(%s.rgb, comp24) > (dot(%s.rgb, comp24) + (1.0f/510.0f))) ? %s : 0.0f)",//#define TEVCMP_BGR24_GT 12
 	"   %s + (abs(dot(%s.rgb, comp24) - dot(%s.rgb, comp24)) < (1.0f/255.0f) ? %s : 0.0f)",//#define TEVCMP_BGR24_EQ 13	
-	"   %s + ((%s.a >= (%s.a + (1.0f/510.0f))) ? %s : 0.0f)",//#define TEVCMP_A8_GT 14
+	"   %s + ((%s.a > (%s.a + (1.0f/510.0f))) ? %s : 0.0f)",//#define TEVCMP_A8_GT 14
 	"   %s + (abs(%s.a - %s.a) < (1.0f/255.0f) ? %s : 0.0f)"//#define TEVCMP_A8_EQ 15
 
 };
@@ -862,6 +862,8 @@ static const char *tevAlphaFuncsTable[] =
 	"(true)"									//ALPHACMP_ALWAYS 7
 };
 
+
+
 static const char *tevAlphaFunclogicTable[] =
 {
 	" && ", // and
@@ -903,10 +905,7 @@ static bool WriteAlphaTest(char *&p, u32 HLSL)
 
 
 	// Seems we need discard for Cg and clip for d3d. sigh.
-	if (HLSL) 
-		WRITE(p, "clip( (");
-	else 
-		WRITE(p, "discard(!( ");
+	WRITE(p, "if(!( ");
 
 	int compindex = bpmem.alphaFunc.comp0 % 8;
 	WRITE(p, tevAlphaFuncsTable[compindex],alphaRef[0]);//lookup the first component from the alpha function table
@@ -916,12 +915,8 @@ static bool WriteAlphaTest(char *&p, u32 HLSL)
     compindex = bpmem.alphaFunc.comp1 % 8;
 	WRITE(p, tevAlphaFuncsTable[compindex],alphaRef[1]);//lookup the second component from the alpha function table    
 
-	if (HLSL) {
-		// clip works differently than discard - discard takes a bool, clip takes a value that kills the pixel on negative
-		WRITE(p, ") ? 1 : -1);\n");
-	} else {
-		WRITE(p, "));\n");
-	}
+	WRITE(p, ")){ocol0 = 0;discard;return;}\n");
+	
     return true;
 }
 
@@ -967,6 +962,6 @@ static void WriteFog(char *&p)
 			WARN_LOG(VIDEO, "Unknown Fog Type! %08x", bpmem.fog.c_proj_fsel.fsel);
 	}		
 
-    WRITE(p, "  prev.rgb = (1.0f - fog) * prev.rgb + (fog * "I_FOG"[0].rgb);\n");
+    WRITE(p, "  prev.rgb = lerp(prev.rgb,"I_FOG"[0].rgb,fog);\n");
     
 }
diff --git a/Source/Core/VideoCommon/Src/TextureConversionShader.cpp b/Source/Core/VideoCommon/Src/TextureConversionShader.cpp
index 4db14bdeaa..eb6b83e234 100644
--- a/Source/Core/VideoCommon/Src/TextureConversionShader.cpp
+++ b/Source/Core/VideoCommon/Src/TextureConversionShader.cpp
@@ -107,12 +107,16 @@ void WriteSwizzler(char*& p, u32 format,bool HLSL)
 	if(!HLSL)
 		WRITE(p,"  sampleUv.y = textureDims.y - sampleUv.y;\n");
 	
-	WRITE(p, "  sampleUv.x = sampleUv.x + textureDims.z;\n"
-	         "  sampleUv.y = sampleUv.y + textureDims.w;\n");
+	WRITE(p, "  sampleUv = sampleUv + textureDims.zw;\n");
+
 	if(HLSL)
 	{
-		WRITE(p, "  sampleUv.x = sampleUv.x / blkDims.z;\n"
-				"  sampleUv.y = sampleUv.y / blkDims.w;\n");
+		WRITE(p, "  sampleUv = sampleUv + float2(1.0f,1.0f);\n"
+				 "  sampleUv = sampleUv / blkDims.zw;\n");
+	}
+	else
+	{
+		WRITE(p, "  sampleUv = sampleUv + float2(1.0f,-1.0f);\n");
 	}	
 }
 
@@ -158,15 +162,18 @@ void Write32BitSwizzler(char*& p, u32 format, bool HLSL)
 	WRITE(p, "  sampleUv = sampleUv * blkDims.xy;\n");
 	
 	if(!HLSL)
-		WRITE(p, "  sampleUv.y = textureDims.y - sampleUv.y;\n");
+		WRITE(p,"  sampleUv.y = textureDims.y - sampleUv.y;\n");
 	
-	WRITE(p, "  sampleUv.x = sampleUv.x + textureDims.z;\n");
-	WRITE(p, "  sampleUv.y = sampleUv.y + textureDims.w;\n");
+	WRITE(p, "  sampleUv = sampleUv + textureDims.zw;\n");
 
 	if(HLSL)
 	{
-		WRITE(p, "  sampleUv.x = sampleUv.x / blkDims.z;\n"
-				"  sampleUv.y = sampleUv.y / blkDims.w;\n");
+		WRITE(p, "  sampleUv = sampleUv + float2(1.0f,1.0f);\n"
+				 "  sampleUv = sampleUv / blkDims.zw;\n");
+	}
+	else
+	{
+		WRITE(p, "  sampleUv = sampleUv + float2(1.0f,-1.0f);\n");
 	}
 }
 
diff --git a/Source/Core/VideoCommon/Src/TextureDecoder.cpp b/Source/Core/VideoCommon/Src/TextureDecoder.cpp
index bbed1cc3cf..01d1c50d4c 100644
--- a/Source/Core/VideoCommon/Src/TextureDecoder.cpp
+++ b/Source/Core/VideoCommon/Src/TextureDecoder.cpp
@@ -418,6 +418,219 @@ static PC_TexFormat GetPCFormatFromTLUTFormat(int tlutfmt)
 	return PC_TEX_FMT_NONE; // Error
 }
 
+PC_TexFormat GetPC_TexFormat(int texformat, int tlutfmt)
+{
+    switch (texformat)
+    {
+    case GX_TF_C4:
+		return GetPCFormatFromTLUTFormat(tlutfmt);
+    case GX_TF_I4:
+		return PC_TEX_FMT_I4_AS_I8;
+	case GX_TF_I8:  // speed critical
+		return PC_TEX_FMT_I8;
+    case GX_TF_C8:
+		return GetPCFormatFromTLUTFormat(tlutfmt);
+    case GX_TF_IA4:
+        return PC_TEX_FMT_IA4_AS_IA8;
+    case GX_TF_IA8:
+        return PC_TEX_FMT_IA8;
+    case GX_TF_C14X2: 
+		return GetPCFormatFromTLUTFormat(tlutfmt);
+    case GX_TF_RGB565:
+		return PC_TEX_FMT_RGB565;
+    case GX_TF_RGB5A3:
+        return PC_TEX_FMT_BGRA32;
+    case GX_TF_RGBA8:  // speed critical
+        return PC_TEX_FMT_BGRA32;
+    case GX_TF_CMPR:  // speed critical
+        // The metroid games use this format almost exclusively.
+		{
+			return PC_TEX_FMT_BGRA32;
+		}
+    }
+
+	// The "copy" texture formats, too?
+    return PC_TEX_FMT_NONE;
+}
+
+
+PC_TexFormat TexDecoder_DirectDecode_real(u8 *dst, const u8 *src, int width, int height,int Pitch, int texformat, int tlutaddr, int tlutfmt)
+{
+   switch (texformat)
+    {
+    case GX_TF_C4:
+		if (tlutfmt == 2)
+        {
+			// Special decoding is required for TLUT format 5A3
+            for (int y = 0; y < height; y += 8)
+                for (int x = 0; x < width; x += 8)
+                    for (int iy = 0; iy < 8; iy++, src += 4)
+                        decodebytesC4_5A3_To_BGRA32((u32*)dst + (y + iy) * width + x, src, tlutaddr);
+        }
+		else
+		{
+            for (int y = 0; y < height; y += 8)
+                for (int x = 0; x < width; x += 8)
+                    for (int iy = 0; iy < 8; iy++, src += 4)
+                        decodebytesC4_To_Raw16((u16*)dst + (y + iy) * width + x, src, tlutaddr);
+		}
+        return GetPCFormatFromTLUTFormat(tlutfmt);
+    case GX_TF_I4:
+		{
+			for (int y = 0; y < height; y += 8)
+				for (int x = 0; x < width; x += 8)
+					for (int iy = 0; iy < 8; iy++, src += 4)
+						for (int ix = 0; ix < 4; ix++)
+						{
+							int val = src[ix];
+							dst[(y + iy) * width + x + ix * 2] = Convert4To8(val >> 4);
+							dst[(y + iy) * width + x + ix * 2 + 1] = Convert4To8(val & 0xF);
+						}
+        }
+       return PC_TEX_FMT_I4_AS_I8;
+	case GX_TF_I8:  // speed critical
+		{
+			for (int y = 0; y < height; y += 4)
+				for (int x = 0; x < width; x += 8)
+					for (int iy = 0; iy < 4; iy++, src += 8)
+						memcpy(dst + (y + iy)*width+x, src, 8);
+		}
+		return PC_TEX_FMT_I8;
+    case GX_TF_C8:
+		if (tlutfmt == 2)
+        {
+			// Special decoding is required for TLUT format 5A3
+            for (int y = 0; y < height; y += 4)
+                for (int x = 0; x < width; x += 8)
+                    for (int iy = 0; iy < 4; iy++, src += 8)
+                        decodebytesC8_5A3_To_BGRA32((u32*)dst + (y + iy) * width + x, src, tlutaddr);
+        }
+		else
+		{
+            for (int y = 0; y < height; y += 4)
+                for (int x = 0; x < width; x += 8)
+                    for (int iy = 0; iy < 4; iy++, src += 8)
+                        decodebytesC8_To_Raw16((u16*)dst + (y + iy) * width + x, src, tlutaddr);
+		}
+        return GetPCFormatFromTLUTFormat(tlutfmt);
+    case GX_TF_IA4:
+        {
+            for (int y = 0; y < height; y += 4)
+                for (int x = 0; x < width; x += 8)
+					for (int iy = 0; iy < 4; iy++, src += 8)
+                        decodebytesIA4((u16*)dst + (y + iy) * width + x, src);
+        }
+		return PC_TEX_FMT_IA4_AS_IA8;
+    case GX_TF_IA8:
+        {
+			for (int y = 0; y < height; y += 4)
+				for (int x = 0; x < width; x += 4)
+					for (int iy = 0; iy < 4; iy++, src += 8)
+					{
+						u16 *ptr = (u16 *)dst + (y + iy) * width + x;
+						u16 *s = (u16 *)src;
+						for(int j = 0; j < 4; j++)
+							*ptr++ = Common::swap16(*s++);
+					}
+
+        }
+		return PC_TEX_FMT_IA8;
+    case GX_TF_C14X2: 
+		if (tlutfmt == 2)
+        {
+			// Special decoding is required for TLUT format 5A3
+            for (int y = 0; y < height; y += 4)
+                for (int x = 0; x < width; x += 4)
+                    for (int iy = 0; iy < 4; iy++, src += 8)
+                        decodebytesC14X2_5A3_To_BGRA32((u32*)dst + (y + iy) * width + x, (u16*)src, tlutaddr);
+        }
+		else
+		{
+            for (int y = 0; y < height; y += 4)
+                for (int x = 0; x < width; x += 4)
+                    for (int iy = 0; iy < 4; iy++, src += 8)
+                        decodebytesC14X2_To_Raw16((u16*)dst + (y + iy) * width + x, (u16*)src, tlutaddr);
+		}
+		return GetPCFormatFromTLUTFormat(tlutfmt);
+    case GX_TF_RGB565:
+		{
+			for (int y = 0; y < height; y += 4)
+				for (int x = 0; x < width; x += 4)
+					for (int iy = 0; iy < 4; iy++, src += 8)
+					{
+						u16 *ptr = (u16 *)dst + (y + iy) * width + x;
+						u16 *s = (u16 *)src;
+						for(int j = 0; j < 4; j++)
+							*ptr++ = Common::swap16(*s++);
+					}
+		}
+		return PC_TEX_FMT_RGB565;
+    case GX_TF_RGB5A3:
+        {
+            for (int y = 0; y < height; y += 4)
+                for (int x = 0; x < width; x += 4)
+                    for (int iy = 0; iy < 4; iy++, src += 8)
+                        //decodebytesRGB5A3((u32*)dst+(y+iy)*width+x, (u16*)src, 4);
+                        decodebytesRGB5A3((u32*)dst+(y+iy)*width+x, (u16*)src);
+        }
+        return PC_TEX_FMT_BGRA32;
+    case GX_TF_RGBA8:  // speed critical
+        {
+			for (int y = 0; y < height; y += 4)
+                for (int x = 0; x < width; x += 4)
+                {
+					for (int iy = 0; iy < 4; iy++)
+                        decodebytesARGB8_4((u32*)dst + (y+iy)*width + x, (u16*)src + 4 * iy, (u16*)src + 4 * iy + 16);
+					src += 64;
+                }
+        }
+        return PC_TEX_FMT_BGRA32;
+    case GX_TF_CMPR:  // speed critical
+        // The metroid games use this format almost exclusively.
+		{
+#if 0   // TODO - currently does not handle transparency correctly and causes problems when texture dimensions are not multiples of 8
+            // 11111111 22222222 55555555 66666666
+            // 33333333 44444444 77777777 88888888
+			for (int y = 0; y < height; y += 8)
+			{
+                for (int x = 0; x < width; x += 8)
+                {
+					copyDXTBlock(dst+(y/2)*width+x*2, src);
+					src += 8;
+					copyDXTBlock(dst+(y/2)*width+x*2+8, src);
+					src += 8;
+					copyDXTBlock(dst+(y/2+2)*width+x*2, src);
+					src += 8;
+					copyDXTBlock(dst+(y/2+2)*width+x*2+8, src);
+					src += 8;
+				}
+			}
+			return PC_TEX_FMT_DXT1;
+#else
+			for (int y = 0; y < height; y += 8)
+			{
+                for (int x = 0; x < width; x += 8)
+                {
+                    decodeDXTBlock((u32*)dst + y * width + x, (DXTBlock*)src, width);
+                                        src += sizeof(DXTBlock);
+                    decodeDXTBlock((u32*)dst + y * width + x + 4, (DXTBlock*)src, width);
+                                        src += sizeof(DXTBlock);
+                    decodeDXTBlock((u32*)dst + (y + 4) * width + x, (DXTBlock*)src, width);
+                                        src += sizeof(DXTBlock);
+                    decodeDXTBlock((u32*)dst + (y + 4) * width + x + 4, (DXTBlock*)src, width);
+                                        src += sizeof(DXTBlock);
+                }
+			}
+#endif
+			return PC_TEX_FMT_BGRA32;
+		}
+    }
+
+	// The "copy" texture formats, too?
+    return PC_TEX_FMT_NONE;
+}
+
+
 
 //switch endianness, unswizzle
 //TODO: to save memory, don't blindly convert everything to argb8888
@@ -599,12 +812,91 @@ PC_TexFormat TexDecoder_Decode_real(u8 *dst, const u8 *src, int width, int heigh
     return PC_TEX_FMT_NONE;
 }
 
+
 void TexDecoder_SetTexFmtOverlayOptions(bool enable, bool center)
 {
 	TexFmt_Overlay_Enable = enable;
 	TexFmt_Overlay_Center = center;
 }
 
+void TexDecoder_DirectDecode(u8 *dst, const u8 *src, int width, int height,int Pitch, int texformat, int tlutaddr, int tlutfmt)
+{
+	PC_TexFormat  retval = TexDecoder_DirectDecode_real(dst,src,width,height,Pitch,texformat,tlutaddr,tlutfmt);
+
+	if ((!TexFmt_Overlay_Enable)|| (retval == PC_TEX_FMT_NONE))
+		return;
+
+	int w = min(width, 40);
+	int h = min(height, 10);
+
+	int xoff = (width - w) >> 1;
+	int yoff = (height - h) >> 1;
+
+	if (!TexFmt_Overlay_Center)
+	{
+		xoff=0;
+		yoff=0;
+	}
+
+	const char* fmt = texfmt[texformat&15];
+	while (*fmt)
+	{
+		int xcnt = 0;
+		int nchar = sfont_map[(int)*fmt];
+		
+		const unsigned char *ptr = sfont_raw[nchar]; // each char is up to 9x10
+
+		for (int x = 0; x < 9;x++)
+		{
+			if (ptr[x] == 0x78)
+				break;
+			xcnt++;
+		}
+
+		for (int y=0; y < 10; y++)
+		{
+			for (int x=0; x < xcnt; x++)
+			{
+				switch(retval)
+				{
+				case PC_TEX_FMT_I8:
+					{
+						// TODO: Is this an acceptable way to draw in I8?
+						u8  *dtp = (u8*)dst;
+						dtp[(y + yoff) * width + x + xoff] = ptr[x] ? 0xFF : 0x88;
+						break;
+					}
+				case PC_TEX_FMT_IA8:
+				case PC_TEX_FMT_IA4_AS_IA8:
+					{
+						u16  *dtp = (u16*)dst;
+						dtp[(y + yoff) * width + x + xoff] = ptr[x] ? 0xFFFF : 0xFF00;
+						break;
+					}
+				case PC_TEX_FMT_RGB565:
+					{
+						u16  *dtp = (u16*)dst;
+						dtp[(y + yoff)*width + x + xoff] = ptr[x] ? 0xFFFF : 0x0000;
+						break;
+					}
+				default:				
+				case PC_TEX_FMT_BGRA32:
+					{
+						int  *dtp = (int*)dst;
+						dtp[(y + yoff) * width + x + xoff] = ptr[x] ? 0xFFFFFFFF : 0xFF000000;
+						break;
+					}
+				}
+			}
+			ptr += 9;
+		}
+		xoff += xcnt;
+		fmt++;
+	}
+}
+
+
+
 PC_TexFormat TexDecoder_Decode(u8 *dst, const u8 *src, int width, int height, int texformat, int tlutaddr, int tlutfmt)
 {
 #if defined(HAVE_OPENCL) && HAVE_OPENCL
diff --git a/Source/Core/VideoCommon/Src/TextureDecoder.h b/Source/Core/VideoCommon/Src/TextureDecoder.h
index 33ba57ee2f..068be12118 100644
--- a/Source/Core/VideoCommon/Src/TextureDecoder.h
+++ b/Source/Core/VideoCommon/Src/TextureDecoder.h
@@ -84,9 +84,11 @@ enum PC_TexFormat
 };
 
 PC_TexFormat TexDecoder_Decode(u8 *dst, const u8 *src, int width, int height, int texformat, int tlutaddr, int tlutfmt);
-
+void TexDecoder_DirectDecode(u8 *dst, const u8 *src, int width, int height,int Pitch, int texformat, int tlutaddr, int tlutfmt);
+PC_TexFormat GetPC_TexFormat(int texformat, int tlutfmt);
 void TexDecoder_DecodeTexel(u8 *dst, const u8 *src, int s, int t, int imageWidth, int texformat, int tlutaddr, int tlutfmt);
 
+
 u32 TexDecoder_GetSafeTextureHash(const u8 *src, int width, int height, int texformat, u32 seed=0);
 u32 TexDecoder_GetTlutHash(const u8* src, int len);