From 5eb24a51cfa57ad4bbdc6b9ccc77af4a4f2b289e Mon Sep 17 00:00:00 2001
From: NeoBrainX <NeoBrainX@gmail.com>
Date: Thu, 5 Jan 2012 17:49:46 +0100
Subject: FifoPlayer: Add a description label for the currently selected object
 command. Requires adding such descriptions for all BP/CP/XF registers. For
 now, I added descriptions for EFB copy related BP registers.

---
 Source/Core/VideoCommon/Src/BPMemory.cpp | 110 +++++++++++++++++++++++++++++++
 Source/Core/VideoCommon/Src/BPMemory.h   |   2 +
 2 files changed, 112 insertions(+)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/BPMemory.cpp b/Source/Core/VideoCommon/Src/BPMemory.cpp
index da09893f35..95e801a433 100644
--- a/Source/Core/VideoCommon/Src/BPMemory.cpp
+++ b/Source/Core/VideoCommon/Src/BPMemory.cpp
@@ -66,3 +66,113 @@ void BPReload()
 		}
 	}
 }
+
+
+void GetBPRegInfo(const u8* data, char* name, size_t name_size, char* desc, size_t desc_size)
+{
+	const char* yes_no[2] = { "Yes", "No" };
+
+	u32 cmddata = Common::swap32(*(u32*)data) & 0xFFFFFF;
+	switch (data[0])
+	{
+	 // Macro to set the register name and make sure it was written correctly via compile time assertion
+	#define SetRegName(reg) \
+		snprintf(name, name_size, #reg); \
+		(void)(reg);
+
+	case BPMEM_DISPLAYCOPYFILER: // 0x01
+		// TODO: This is actually the sample pattern used for copies from an antialiased EFB
+		SetRegName(BPMEM_DISPLAYCOPYFILER);
+		// TODO: Description
+		break;
+
+	case 0x02: // 0x02
+	case 0x03: // 0x03
+	case 0x04: // 0x04
+		// TODO: same as BPMEM_DISPLAYCOPYFILER
+		break;
+
+	case BPMEM_EFB_TL: // 0x49
+		{
+			SetRegName(BPMEM_EFB_TL);
+			X10Y10 left_top; left_top.hex = cmddata;
+			snprintf(desc, desc_size, "Left: %d\nTop: %d", left_top.x, left_top.y);
+		}
+		break;
+
+	case BPMEM_EFB_BR: // 0x4A
+		{
+			// TODO: Misleading name, should be BPMEM_EFB_WH instead
+			SetRegName(BPMEM_EFB_BR);
+			X10Y10 width_height; width_height.hex = cmddata;
+			snprintf(desc, desc_size, "Width: %d\nHeight: %d", width_height.x+1, width_height.y+1);
+		}
+		break;
+
+	case BPMEM_EFB_ADDR: // 0x4B
+		SetRegName(BPMEM_EFB_ADDR);
+		snprintf(desc, desc_size, "Target address (32 byte aligned): %06x", cmddata << 5);
+		break;
+
+	case BPMEM_COPYYSCALE: // 0x4E
+		SetRegName(BPMEM_COPYYSCALE);
+		snprintf(desc, desc_size, "Scaling factor (XFB copy only): 0x%X (%f or inverted %f)", cmddata, (float)cmddata/256.f, 256.f/(float)cmddata);
+		break;
+
+	case BPMEM_CLEAR_AR: // 0x4F
+		SetRegName(BPMEM_CLEAR_AR);
+		snprintf(desc, desc_size, "Alpha: %02X\nRed: %02x", (cmddata&0xFF00)>>8, cmddata&0xFF);
+		break;
+
+	case BPMEM_CLEAR_GB: // 0x50
+		SetRegName(BPMEM_CLEAR_GB);
+		snprintf(desc, desc_size, "Green: %02X\nBlue: %02x", (cmddata&0xFF00)>>8, cmddata&0xFF);
+		break;
+
+	case BPMEM_CLEAR_Z: // 0x51
+		SetRegName(BPMEM_CLEAR_Z);
+		snprintf(desc, desc_size, "Z value: %06X", cmddata);
+		break;
+
+	case BPMEM_TRIGGER_EFB_COPY: // 0x52
+		{
+			SetRegName(BPMEM_TRIGGER_EFB_COPY);
+			UPE_Copy copy; copy.Hex = cmddata;
+			snprintf(desc, desc_size, "Clamping: %s\n"
+								"Converting from RGB to YUV: %s\n"
+								"Target pixel format: 0x%X\n"
+								"Gamma correction: %s\n"
+								"Mipmap filter: %s\n"
+								"Vertical scaling: %s\n"
+								"Clear: %s\n"
+								"Frame to field: 0x%01X\n"
+								"Copy to XFB: %s\n"
+								"Intensity format: %s\n"
+								"Automatic color conversion: %s\n",
+								(copy.clamp0 && copy.clamp1) ? "Top and Bottom" : (copy.clamp0) ? "Top only" : (copy.clamp1) ? "Bottom only" : "None",
+								yes_no[copy.yuv],
+								copy.tp_realFormat(),
+								(copy.gamma==0)?"1.0":(copy.gamma==1)?"1.7":(copy.gamma==2)?"2.2":"Invalid value 0x3?",
+								yes_no[copy.half_scale],
+								yes_no[copy.scale_invert],
+								yes_no[copy.clear],
+								copy.frame_to_field,
+								yes_no[copy.copy_to_xfb],
+								yes_no[copy.intensity_fmt],
+								yes_no[copy.auto_conv]);
+		}
+		break;
+
+	case BPMEM_COPYFILTER0: // 0x53
+		SetRegName(BPMEM_COPYFILTER0);
+		// TODO: Description
+		break;
+
+	case BPMEM_COPYFILTER1: // 0x54
+		SetRegName(BPMEM_COPYFILTER1);
+		// TODO: Description
+		break;
+
+#undef SET_REG_NAME
+	}
+}
diff --git a/Source/Core/VideoCommon/Src/BPMemory.h b/Source/Core/VideoCommon/Src/BPMemory.h
index 1f6f453a77..51f2c0a5b5 100644
--- a/Source/Core/VideoCommon/Src/BPMemory.h
+++ b/Source/Core/VideoCommon/Src/BPMemory.h
@@ -978,4 +978,6 @@ extern BPMemory bpmem;
 
 void LoadBPReg(u32 value0);
 
+void GetBPRegInfo(const u8* data, char* name, size_t name_size, char* desc, size_t desc_size);
+
 #endif // _BPMEMORY_H
-- 
cgit v1.2.3


From ecb616f7fff1a3e01452367af9204f0a547d24cd Mon Sep 17 00:00:00 2001
From: NeoBrainX <NeoBrainX@googlemail.com>
Date: Sat, 7 Jan 2012 17:38:00 +0100
Subject: FifoPlayer: Make sure that we actually display "Yes" if some variable
 is true... FifoPlayer: More consistency of output (display "0x" in front of
 hex numbers, write hex numbers in capital letters)

---
 Source/Core/VideoCommon/Src/BPMemory.cpp | 24 ++++++++++++------------
 1 file changed, 12 insertions(+), 12 deletions(-)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/BPMemory.cpp b/Source/Core/VideoCommon/Src/BPMemory.cpp
index 95e801a433..3f21e1acb0 100644
--- a/Source/Core/VideoCommon/Src/BPMemory.cpp
+++ b/Source/Core/VideoCommon/Src/BPMemory.cpp
@@ -70,7 +70,7 @@ void BPReload()
 
 void GetBPRegInfo(const u8* data, char* name, size_t name_size, char* desc, size_t desc_size)
 {
-	const char* yes_no[2] = { "Yes", "No" };
+	const char* no_yes[2] = { "No", "Yes" };
 
 	u32 cmddata = Common::swap32(*(u32*)data) & 0xFFFFFF;
 	switch (data[0])
@@ -111,7 +111,7 @@ void GetBPRegInfo(const u8* data, char* name, size_t name_size, char* desc, size
 
 	case BPMEM_EFB_ADDR: // 0x4B
 		SetRegName(BPMEM_EFB_ADDR);
-		snprintf(desc, desc_size, "Target address (32 byte aligned): %06x", cmddata << 5);
+		snprintf(desc, desc_size, "Target address (32 byte aligned): 0x%06X", cmddata << 5);
 		break;
 
 	case BPMEM_COPYYSCALE: // 0x4E
@@ -121,17 +121,17 @@ void GetBPRegInfo(const u8* data, char* name, size_t name_size, char* desc, size
 
 	case BPMEM_CLEAR_AR: // 0x4F
 		SetRegName(BPMEM_CLEAR_AR);
-		snprintf(desc, desc_size, "Alpha: %02X\nRed: %02x", (cmddata&0xFF00)>>8, cmddata&0xFF);
+		snprintf(desc, desc_size, "Alpha: 0x%02X\nRed: 0x%02X", (cmddata&0xFF00)>>8, cmddata&0xFF);
 		break;
 
 	case BPMEM_CLEAR_GB: // 0x50
 		SetRegName(BPMEM_CLEAR_GB);
-		snprintf(desc, desc_size, "Green: %02X\nBlue: %02x", (cmddata&0xFF00)>>8, cmddata&0xFF);
+		snprintf(desc, desc_size, "Green: 0x%02X\nBlue: 0x%02X", (cmddata&0xFF00)>>8, cmddata&0xFF);
 		break;
 
 	case BPMEM_CLEAR_Z: // 0x51
 		SetRegName(BPMEM_CLEAR_Z);
-		snprintf(desc, desc_size, "Z value: %06X", cmddata);
+		snprintf(desc, desc_size, "Z value: 0x%06X", cmddata);
 		break;
 
 	case BPMEM_TRIGGER_EFB_COPY: // 0x52
@@ -150,16 +150,16 @@ void GetBPRegInfo(const u8* data, char* name, size_t name_size, char* desc, size
 								"Intensity format: %s\n"
 								"Automatic color conversion: %s\n",
 								(copy.clamp0 && copy.clamp1) ? "Top and Bottom" : (copy.clamp0) ? "Top only" : (copy.clamp1) ? "Bottom only" : "None",
-								yes_no[copy.yuv],
+								no_yes[copy.yuv],
 								copy.tp_realFormat(),
 								(copy.gamma==0)?"1.0":(copy.gamma==1)?"1.7":(copy.gamma==2)?"2.2":"Invalid value 0x3?",
-								yes_no[copy.half_scale],
-								yes_no[copy.scale_invert],
-								yes_no[copy.clear],
+								no_yes[copy.half_scale],
+								no_yes[copy.scale_invert],
+								no_yes[copy.clear],
 								copy.frame_to_field,
-								yes_no[copy.copy_to_xfb],
-								yes_no[copy.intensity_fmt],
-								yes_no[copy.auto_conv]);
+								no_yes[copy.copy_to_xfb],
+								no_yes[copy.intensity_fmt],
+								no_yes[copy.auto_conv]);
 		}
 		break;
 
-- 
cgit v1.2.3


From ad1a4d7ce318232acf2086bc59c44d66d83e6180 Mon Sep 17 00:00:00 2001
From: NeoBrainX <NeoBrainX@googlemail.com>
Date: Sat, 21 Jan 2012 13:40:14 +0100
Subject: FifoPlayer: Add a search function for the analyzer tab. Also some
 tiny fixes.

---
 Source/Core/VideoCommon/Src/BPMemory.cpp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/BPMemory.cpp b/Source/Core/VideoCommon/Src/BPMemory.cpp
index 3f21e1acb0..43a4facacf 100644
--- a/Source/Core/VideoCommon/Src/BPMemory.cpp
+++ b/Source/Core/VideoCommon/Src/BPMemory.cpp
@@ -148,7 +148,7 @@ void GetBPRegInfo(const u8* data, char* name, size_t name_size, char* desc, size
 								"Frame to field: 0x%01X\n"
 								"Copy to XFB: %s\n"
 								"Intensity format: %s\n"
-								"Automatic color conversion: %s\n",
+								"Automatic color conversion: %s",
 								(copy.clamp0 && copy.clamp1) ? "Top and Bottom" : (copy.clamp0) ? "Top only" : (copy.clamp1) ? "Bottom only" : "None",
 								no_yes[copy.yuv],
 								copy.tp_realFormat(),
-- 
cgit v1.2.3


From 9e398fd418027e4932a1539f7d382b70e8acc408 Mon Sep 17 00:00:00 2001
From: skidau <skidau@gmail.com>
Date: Fri, 2 Mar 2012 18:53:41 +1100
Subject: Added an external exception check when the CPU writes to the FIFO. 
 This allows the CPU time to service FIFO overflows.  Fixes random hangs
 caused by FIFO overflows and desyncs like in "The Last Story" and "Battalion
 Wars 2".  Thanks to marcosvitali for the research.

Fixes issue 5209.
Fixes issue 5150.
Fixes issue 5055.
Fixes issue 4889.
Fixes issue 4061.
Fixes issue 4010.
Fixes issue 3902.
---
 Source/Core/VideoCommon/Src/CommandProcessor.cpp | 22 ----------------------
 Source/Core/VideoCommon/Src/CommandProcessor.h   |  1 -
 Source/Core/VideoCommon/Src/Fifo.cpp             |  3 +--
 3 files changed, 1 insertion(+), 25 deletions(-)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/CommandProcessor.cpp b/Source/Core/VideoCommon/Src/CommandProcessor.cpp
index fc85b8e39d..dbe09aaa0e 100644
--- a/Source/Core/VideoCommon/Src/CommandProcessor.cpp
+++ b/Source/Core/VideoCommon/Src/CommandProcessor.cpp
@@ -60,7 +60,6 @@ volatile bool interruptSet= false;
 volatile bool interruptWaiting= false;
 volatile bool interruptTokenWaiting = false;
 volatile bool interruptFinishWaiting = false;
-volatile bool OnOverflow = false;
 
 bool IsOnThread()
 {
@@ -92,7 +91,6 @@ void DoState(PointerWrap &p)
 	p.Do(interruptWaiting);
 	p.Do(interruptTokenWaiting);
 	p.Do(interruptFinishWaiting);
-	p.Do(OnOverflow);
 }
 
 inline void WriteLow (volatile u32& _reg, u16 lowbits)  {Common::AtomicStore(_reg,(_reg & 0xFFFF0000) | lowbits);}
@@ -135,7 +133,6 @@ void Init()
 	bProcessFifoToLoWatermark = false;
 	bProcessFifoAllDistance = false;
 	isPossibleWaitingSetDrawDone = false;
-	OnOverflow = false;
 
     et_UpdateInterrupts = CoreTiming::RegisterEvent("UpdateInterrupts", UpdateInterrupts_Wrapper);
 }
@@ -449,26 +446,7 @@ void STACKALIGN GatherPipeBursted()
 	Common::AtomicAdd(fifo.CPReadWriteDistance, GATHER_PIPE_SIZE);
 
 	if (!IsOnThread())
-	{
 		RunGpu();
-	}
-	else
-	{
-		if(fifo.CPReadWriteDistance	== fifo.CPEnd - fifo.CPBase - 32)
-		{
-			if(!OnOverflow)
-				NOTICE_LOG(COMMANDPROCESSOR,"FIFO is almost in overflown, BreakPoint: %i", fifo.bFF_Breakpoint);
-			OnOverflow = true;
-			while (!CommandProcessor::interruptWaiting && fifo.bFF_GPReadEnable &&
-			fifo.CPReadWriteDistance > fifo.CPEnd - fifo.CPBase - 64)
-			Common::YieldCPU();														
-		}
-		else
-		{
-			OnOverflow = false;
-		}
-	}
-	
 
 	_assert_msg_(COMMANDPROCESSOR, fifo.CPReadWriteDistance	<= fifo.CPEnd - fifo.CPBase,
 	"FIFO is overflown by GatherPipe !\nCPU thread is too fast!");
diff --git a/Source/Core/VideoCommon/Src/CommandProcessor.h b/Source/Core/VideoCommon/Src/CommandProcessor.h
index db6772d66d..bb969fcb95 100644
--- a/Source/Core/VideoCommon/Src/CommandProcessor.h
+++ b/Source/Core/VideoCommon/Src/CommandProcessor.h
@@ -35,7 +35,6 @@ extern volatile bool interruptSet;
 extern volatile bool interruptWaiting;
 extern volatile bool interruptTokenWaiting;
 extern volatile bool interruptFinishWaiting;
-extern volatile bool OnOverflow;
 
 // internal hardware addresses
 enum
diff --git a/Source/Core/VideoCommon/Src/Fifo.cpp b/Source/Core/VideoCommon/Src/Fifo.cpp
index 842ff49e78..5cb3252816 100644
--- a/Source/Core/VideoCommon/Src/Fifo.cpp
+++ b/Source/Core/VideoCommon/Src/Fifo.cpp
@@ -137,8 +137,7 @@ void RunGpuLoop()
 				
 		CommandProcessor::SetCpStatus();
 		// check if we are able to run this buffer	
-		while (!CommandProcessor::interruptWaiting && fifo.bFF_GPReadEnable &&
-			fifo.CPReadWriteDistance && (!AtBreakpoint() || CommandProcessor::OnOverflow))
+		while (!CommandProcessor::interruptWaiting && fifo.bFF_GPReadEnable && fifo.CPReadWriteDistance && !AtBreakpoint())
 		{
 			if (!GpuRunningState) break;
 
-- 
cgit v1.2.3


From f5d4fe0bfe13937fa3736aa6b78974ceb6e190ad Mon Sep 17 00:00:00 2001
From: Henrik Rydgard <hrydgard@gmail.com>
Date: Sat, 3 Mar 2012 20:07:20 +0100
Subject: Fix some minor bugs pointed out by PVS Studio (thanks!)

---
 Source/Core/VideoCommon/Src/PixelShaderGen.cpp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/PixelShaderGen.cpp b/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
index fe5827de05..3ef923dc3e 100644
--- a/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
+++ b/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
@@ -158,7 +158,7 @@ void GetPixelShaderId(PIXELSHADERUID *uid, DSTALPHA_MODE dstAlphaMode, u32 compo
 	}
 
 	u32* ptr = &uid->values[2];
-	for (int i = 0; i < bpmem.genMode.numtevstages+1; ++i)
+	for (unsigned int i = 0; i < bpmem.genMode.numtevstages+1; ++i)
 	{
 		StageHash(i, ptr);
 		ptr += 4; // max: ptr = &uid->values[66]
-- 
cgit v1.2.3


From b0f75f17ae19484735675fea348f7e95740f597f Mon Sep 17 00:00:00 2001
From: marcosvitali <marcosvitali@gmail.com>
Date: Mon, 5 Mar 2012 02:40:10 -0300
Subject: This release still fixed the hangs produced by fifo overflow without
 sacrifice performance. For example you can test Tutorial moves at the
 beginning of The last history now is fluid 30/60. Shuffle2: I've delete the
 hacky line, I think is not necessary anymore. Additional some clean in
 CommandProcessor. Please test The Last Story and others games affected in the
 previous commits and give me a feedback.

---
 Source/Core/VideoCommon/Src/CommandProcessor.cpp | 39 +++++++++++++-----------
 Source/Core/VideoCommon/Src/CommandProcessor.h   |  1 +
 Source/Core/VideoCommon/Src/MainBase.cpp         |  5 +++
 Source/Core/VideoCommon/Src/PixelEngine.cpp      |  2 --
 4 files changed, 28 insertions(+), 19 deletions(-)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/CommandProcessor.cpp b/Source/Core/VideoCommon/Src/CommandProcessor.cpp
index dbe09aaa0e..38b342b76c 100644
--- a/Source/Core/VideoCommon/Src/CommandProcessor.cpp
+++ b/Source/Core/VideoCommon/Src/CommandProcessor.cpp
@@ -56,6 +56,7 @@ static bool bProcessFifoToLoWatermark = false;
 static bool bProcessFifoAllDistance = false;
 
 volatile bool isPossibleWaitingSetDrawDone = false;
+volatile bool isHiWatermarkActive = false;
 volatile bool interruptSet= false;
 volatile bool interruptWaiting= false;
 volatile bool interruptTokenWaiting = false;
@@ -85,7 +86,7 @@ void DoState(PointerWrap &p)
 
 	p.Do(bProcessFifoToLoWatermark);
 	p.Do(bProcessFifoAllDistance);
-
+	p.Do(isHiWatermarkActive);
 	p.Do(isPossibleWaitingSetDrawDone);
 	p.Do(interruptSet);
 	p.Do(interruptWaiting);
@@ -133,6 +134,7 @@ void Init()
 	bProcessFifoToLoWatermark = false;
 	bProcessFifoAllDistance = false;
 	isPossibleWaitingSetDrawDone = false;
+	isHiWatermarkActive = false;
 
     et_UpdateInterrupts = CoreTiming::RegisterEvent("UpdateInterrupts", UpdateInterrupts_Wrapper);
 }
@@ -141,7 +143,6 @@ void Read16(u16& _rReturnValue, const u32 _Address)
 {
 
 	INFO_LOG(COMMANDPROCESSOR, "(r): 0x%08x", _Address);
-	ProcessFifoEvents();
 	switch (_Address & 0xFFF)
 	{
 	case STATUS_REGISTER:		
@@ -409,7 +410,6 @@ void Write16(const u16 _Value, const u32 _Address)
 
 	if (!IsOnThread())
 		RunGpu();
-	ProcessFifoEvents();
 }
 
 void Read32(u32& _rReturnValue, const u32 _Address)
@@ -425,7 +425,6 @@ void Write32(const u32 _Data, const u32 _Address)
 
 void STACKALIGN GatherPipeBursted()
 {
-	ProcessFifoEvents();
 	// if we aren't linked, we don't care about gather pipe data
 	if (!m_CPCtrlReg.GPLinkEnable)
 	{
@@ -487,17 +486,18 @@ void AbortFrame()
 
 void SetOverflowStatusFromGatherPipe()
 {
-	if (!fifo.bFF_HiWatermarkInt) return;
+	
 
 	fifo.bFF_HiWatermark = (fifo.CPReadWriteDistance > fifo.CPHiWatermark);
-    fifo.bFF_LoWatermark = (fifo.CPReadWriteDistance < fifo.CPLoWatermark);
+	isHiWatermarkActive = fifo.bFF_HiWatermark && fifo.bFF_HiWatermarkInt && m_CPCtrlReg.GPReadEnable;
 	
-	bool interrupt = fifo.bFF_HiWatermark && fifo.bFF_HiWatermarkInt &&
-		m_CPCtrlReg.GPLinkEnable && m_CPCtrlReg.GPReadEnable;
-
-    if (interrupt != interruptSet && interrupt)
-            CommandProcessor::UpdateInterrupts(true);
-    
+    if (isHiWatermarkActive)
+	{
+		interruptSet = true;
+        INFO_LOG(COMMANDPROCESSOR,"Interrupt set");
+        ProcessorInterface::SetInterrupt(INT_CAUSE_CP, true);    
+	}
+			   
 }
 
 void SetCpStatus()
@@ -505,7 +505,7 @@ void SetCpStatus()
     // overflow & underflow check
 	fifo.bFF_HiWatermark = (fifo.CPReadWriteDistance > fifo.CPHiWatermark);
     fifo.bFF_LoWatermark = (fifo.CPReadWriteDistance < fifo.CPLoWatermark);
-
+	
     // breakpoint     
 	
 	if (fifo.bFF_BPEnable)
@@ -540,13 +540,18 @@ void SetCpStatus()
 	
 	bool interrupt = (bpInt || ovfInt || undfInt) && m_CPCtrlReg.GPReadEnable;
 
+	isHiWatermarkActive = ovfInt  && m_CPCtrlReg.GPReadEnable;
+
     if (interrupt != interruptSet && !interruptWaiting)
     {
         u64 userdata = interrupt?1:0;
         if (IsOnThread())
         {
-            interruptWaiting = true;
-            CommandProcessor::UpdateInterruptsFromVideoBackend(userdata);
+            if(!interrupt || bpInt || undfInt)
+			{
+				interruptWaiting = true;
+				CommandProcessor::UpdateInterruptsFromVideoBackend(userdata);
+			}
         }
         else
             CommandProcessor::UpdateInterrupts(userdata);
@@ -631,8 +636,8 @@ void SetCpControlRegister()
 		ProcessorInterface::Fifo_CPUEnd = fifo.CPEnd;
 	}
 	// If overflown happens process the fifo to LoWatemark
-	if (bProcessFifoToLoWatermark)
-		ProcessFifoToLoWatermark();
+	//if (bProcessFifoToLoWatermark)
+	//	ProcessFifoToLoWatermark();
 			
 	if(fifo.bFF_GPReadEnable && !m_CPCtrlReg.GPReadEnable)
 	{
diff --git a/Source/Core/VideoCommon/Src/CommandProcessor.h b/Source/Core/VideoCommon/Src/CommandProcessor.h
index bb969fcb95..2bf965a3ba 100644
--- a/Source/Core/VideoCommon/Src/CommandProcessor.h
+++ b/Source/Core/VideoCommon/Src/CommandProcessor.h
@@ -31,6 +31,7 @@ namespace CommandProcessor
 
 extern SCPFifoStruct fifo; //This one is shared between gfx thread and emulator thread.
 extern volatile bool isPossibleWaitingSetDrawDone; //This one is used for sync gfx thread and emulator thread.
+extern volatile bool isHiWatermarkActive;
 extern volatile bool interruptSet;
 extern volatile bool interruptWaiting;
 extern volatile bool interruptTokenWaiting;
diff --git a/Source/Core/VideoCommon/Src/MainBase.cpp b/Source/Core/VideoCommon/Src/MainBase.cpp
index af21ebbb94..0b1258a662 100644
--- a/Source/Core/VideoCommon/Src/MainBase.cpp
+++ b/Source/Core/VideoCommon/Src/MainBase.cpp
@@ -250,6 +250,11 @@ bool VideoBackendHardware::Video_IsPossibleWaitingSetDrawDone()
 	return CommandProcessor::isPossibleWaitingSetDrawDone;
 }
 
+bool VideoBackendHardware::Video_IsHiWatermarkActive()
+{
+	return CommandProcessor::isHiWatermarkActive;
+}
+
 void VideoBackendHardware::Video_AbortFrame()
 {
 	CommandProcessor::AbortFrame();
diff --git a/Source/Core/VideoCommon/Src/PixelEngine.cpp b/Source/Core/VideoCommon/Src/PixelEngine.cpp
index 456e0fb535..ec0683742f 100644
--- a/Source/Core/VideoCommon/Src/PixelEngine.cpp
+++ b/Source/Core/VideoCommon/Src/PixelEngine.cpp
@@ -180,7 +180,6 @@ void Init()
 void Read16(u16& _uReturnValue, const u32 _iAddress)
 {
 	DEBUG_LOG(PIXELENGINE, "(r16) 0x%08x", _iAddress);
-	CommandProcessor::ProcessFifoEvents();
 	switch (_iAddress & 0xFFF)
 	{
 		// CPU Direct Access EFB Raster State Config
@@ -334,7 +333,6 @@ void Write16(const u16 _iValue, const u32 _iAddress)
 		break;
 	}
 
-	CommandProcessor::ProcessFifoEvents();
 }
 
 void Write32(const u32 _iValue, const u32 _iAddress)
-- 
cgit v1.2.3


From 6fc8cdf12d41ea9f9be2d345face72ddf1fe045d Mon Sep 17 00:00:00 2001
From: LPFaint99 <lpfaint99@gmail.com>
Date: Wed, 7 Mar 2012 20:08:14 -0800
Subject: revert r90a2096a24f4 behavior changes in PixelEngine.cpp. Thanks to
 Autoran1 for finding the responsible commit. Fixes Scrubbing Serena Beach
 Countdown in both SC and DC.

Fixes issue 1498.
---
 Source/Core/VideoCommon/Src/PixelEngine.cpp | 4 ++++
 1 file changed, 4 insertions(+)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/PixelEngine.cpp b/Source/Core/VideoCommon/Src/PixelEngine.cpp
index 456e0fb535..8164bc41da 100644
--- a/Source/Core/VideoCommon/Src/PixelEngine.cpp
+++ b/Source/Core/VideoCommon/Src/PixelEngine.cpp
@@ -269,6 +269,10 @@ void Read16(u16& _uReturnValue, const u32 _iAddress)
 	case PE_PERF_5L:
 	case PE_PERF_5H:
 		INFO_LOG(PIXELENGINE, "(r16) perf counter @ %08x", _iAddress);
+		// git r90a2096a24f4 (svn r3663) added the PE_PERF cases, without setting
+		// _uReturnValue to anything, this reverts to the previous behaviour which allows
+		// The timer in SMS:Scrubbing Serena Beach to countdown correctly
+		_uReturnValue = 1;
 		break;
 
 	default:
-- 
cgit v1.2.3


From c2e6fdf09f33f7765526fac680d13af9ba744125 Mon Sep 17 00:00:00 2001
From: marcosvitali <marcosvitali@gmail.com>
Date: Thu, 8 Mar 2012 02:47:55 -0300
Subject: - I've fixed possibles random hangs in DC mode. - I've fixed hangs in
 DC mode in (Simpsons, Monkey Island, Pokemon XD, etc) - I've implemented
 accurate manage of Pixel Engine Interrupts, now the GPU loop is stopped when
 a PE Interrupt needs to be managed and resume when Pixel Engine finish, I
 think now, the Fifo in DC mode is more accurate than SC mode. :)

Time to close the big fifo Issue 3694 (snif), please if you have a possible fifo issue report this like a game issue.

I was working with Skid_AU together, especially thanks for him.

Test a lot all games, and compare the performance with the master maybe this accuracy has a cost (not a lot).
I think now the fifo is very stable, overflow fixed, random hang fixed, if you have a game with a hang with this rev and not in master please report this.
---
 Source/Core/VideoCommon/Src/CommandProcessor.cpp | 5 +++--
 Source/Core/VideoCommon/Src/Fifo.cpp             | 3 ++-
 Source/Core/VideoCommon/Src/PixelEngine.cpp      | 5 +++++
 Source/Core/VideoCommon/Src/PixelEngine.h        | 2 +-
 4 files changed, 11 insertions(+), 4 deletions(-)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/CommandProcessor.cpp b/Source/Core/VideoCommon/Src/CommandProcessor.cpp
index 38b342b76c..4fd7e88f5e 100644
--- a/Source/Core/VideoCommon/Src/CommandProcessor.cpp
+++ b/Source/Core/VideoCommon/Src/CommandProcessor.cpp
@@ -596,8 +596,8 @@ void SetCpStatusRegister()
 	// Here always there is one fifo attached to the GPU
 		
 	m_CPStatusReg.Breakpoint = fifo.bFF_Breakpoint;
-	m_CPStatusReg.ReadIdle = (fifo.CPReadPointer == fifo.CPWritePointer) || (fifo.CPReadPointer == fifo.CPBreakpoint);
-	m_CPStatusReg.CommandIdle = !fifo.CPReadWriteDistance;
+	m_CPStatusReg.ReadIdle = (fifo.CPReadPointer == fifo.CPWritePointer) || (fifo.CPReadPointer == fifo.CPBreakpoint) ;
+	m_CPStatusReg.CommandIdle = !fifo.CPReadWriteDistance || (IsOnThread() && PixelEngine::WaitingForPEInterrupt());
 	m_CPStatusReg.UnderflowLoWatermark = fifo.bFF_LoWatermark;
 	m_CPStatusReg.OverflowHiWatermark = fifo.bFF_HiWatermark;
 
@@ -620,6 +620,7 @@ void SetCpControlRegister()
 
 	if (!fifo.bFF_GPReadEnable && m_CPCtrlReg.GPReadEnable && !m_CPCtrlReg.BPEnable)
 	{
+		ProcessFifoEvents();
 		PixelEngine::ResetSetFinish();			
 	}
 
diff --git a/Source/Core/VideoCommon/Src/Fifo.cpp b/Source/Core/VideoCommon/Src/Fifo.cpp
index 5cb3252816..2e60055c97 100644
--- a/Source/Core/VideoCommon/Src/Fifo.cpp
+++ b/Source/Core/VideoCommon/Src/Fifo.cpp
@@ -22,6 +22,7 @@
 #include "Atomic.h"
 #include "OpcodeDecoding.h"
 #include "CommandProcessor.h"
+#include "PixelEngine.h"
 #include "ChunkFile.h"
 #include "Fifo.h"
 #include "HW/Memmap.h"
@@ -137,7 +138,7 @@ void RunGpuLoop()
 				
 		CommandProcessor::SetCpStatus();
 		// check if we are able to run this buffer	
-		while (!CommandProcessor::interruptWaiting && fifo.bFF_GPReadEnable && fifo.CPReadWriteDistance && !AtBreakpoint())
+		while (GpuRunningState && !CommandProcessor::interruptWaiting && fifo.bFF_GPReadEnable && fifo.CPReadWriteDistance && !AtBreakpoint() && !PixelEngine::WaitingForPEInterrupt())
 		{
 			if (!GpuRunningState) break;
 
diff --git a/Source/Core/VideoCommon/Src/PixelEngine.cpp b/Source/Core/VideoCommon/Src/PixelEngine.cpp
index ec0683742f..09cb9b8a5c 100644
--- a/Source/Core/VideoCommon/Src/PixelEngine.cpp
+++ b/Source/Core/VideoCommon/Src/PixelEngine.cpp
@@ -468,4 +468,9 @@ void ResetSetToken()
 	}
 	CommandProcessor::interruptTokenWaiting = false;
 }
+
+bool WaitingForPEInterrupt()
+{
+	return CommandProcessor::interruptFinishWaiting  || CommandProcessor::interruptTokenWaiting || interruptSetFinish || interruptSetToken;
+}
 } // end of namespace PixelEngine
diff --git a/Source/Core/VideoCommon/Src/PixelEngine.h b/Source/Core/VideoCommon/Src/PixelEngine.h
index dd0304fbe1..5e64300ef3 100644
--- a/Source/Core/VideoCommon/Src/PixelEngine.h
+++ b/Source/Core/VideoCommon/Src/PixelEngine.h
@@ -80,7 +80,7 @@ void SetToken(const u16 _token, const int _bSetTokenAcknowledge);
 void SetFinish(void);
 void ResetSetFinish(void);
 void ResetSetToken(void);
-bool AllowIdleSkipping();
+bool WaitingForPEInterrupt();
 
 // Bounding box functionality. Paper Mario (both) are a couple of the few games that use it.
 extern u16 bbox[4];
-- 
cgit v1.2.3


From 41652d6b1f391fdcd0f599463f689c0299cdc007 Mon Sep 17 00:00:00 2001
From: marcosvitali <marcosvitali@gmail.com>
Date: Fri, 9 Mar 2012 01:33:29 -0300
Subject: I've fixed Metroid Prime 3 and 2 desync. And other games with desync
 because of FIFO Reset. That happens because  FIFO_RW_DISTANCE_HI must be
 written first, for checking fifo.CPReadWriteDistance == 0, so some fifo
 resets was not managed in the right way. I didn't test Metroid 2 desync
 reported in Issue 4336 but I think is the same. About the flickering in MP2,
 I don't know for my is not related or yes, but you can test anyway.

Fixed Issue 3902

Well now the FIFO is 99.99% finished :)
---
 Source/Core/VideoCommon/Src/CommandProcessor.cpp | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/CommandProcessor.cpp b/Source/Core/VideoCommon/Src/CommandProcessor.cpp
index 4fd7e88f5e..a6e68f6298 100644
--- a/Source/Core/VideoCommon/Src/CommandProcessor.cpp
+++ b/Source/Core/VideoCommon/Src/CommandProcessor.cpp
@@ -388,10 +388,6 @@ void Write16(const u16 _Value, const u32 _Address)
 
 	case FIFO_RW_DISTANCE_HI:
 		WriteHigh((u32 &)fifo.CPReadWriteDistance, _Value);
-		DEBUG_LOG(COMMANDPROCESSOR,"try to write to FIFO_RW_DISTANCE_HI : %04x", _Value);
-		break;
-	case FIFO_RW_DISTANCE_LO:
-		WriteLow((u32 &)fifo.CPReadWriteDistance, _Value & 0xFFE0);
 		if (fifo.CPReadWriteDistance == 0)
 		{
 			GPFifo::ResetGatherPipe();
@@ -401,6 +397,10 @@ void Write16(const u16 _Value, const u32 _Address)
 			ResetVideoBuffer();		
 		}
 		IncrementCheckContextId();
+		DEBUG_LOG(COMMANDPROCESSOR,"try to write to FIFO_RW_DISTANCE_HI : %04x", _Value);
+		break;
+	case FIFO_RW_DISTANCE_LO:
+		WriteLow((u32 &)fifo.CPReadWriteDistance, _Value & 0xFFE0);
 		DEBUG_LOG(COMMANDPROCESSOR,"try to write to FIFO_RW_DISTANCE_LO : %04x", _Value);
 		break;
 
-- 
cgit v1.2.3


From 104603467be6cc15e63dfb5cadec8044edd1da46 Mon Sep 17 00:00:00 2001
From: marcosvitali <marcosvitali@gmail.com>
Date: Fri, 9 Mar 2012 18:58:23 -0300
Subject: This commit fix games hanging because of my prior Revision
 c2e6fdf09f33 The external exceptions in dolphin are checking frequently but
 is different to real HW, so sometime the game is in a loop checking GPU
 STATUS, the exceptions doesn't checked, and the game hang.\ For solve this I
 need a trick: still waiting for the exception handler be linked but if
 CommandProcecsor is reading the GPStatus, resume this.

This fixed "TimeSplitters: Future Perfect" broken in the Revision c2e6fdf09f33 and surely others games.
---
 Source/Core/VideoCommon/Src/CommandProcessor.cpp | 10 +++++++---
 Source/Core/VideoCommon/Src/PixelEngine.cpp      | 14 ++++++++------
 Source/Core/VideoCommon/Src/PixelEngine.h        |  1 +
 3 files changed, 16 insertions(+), 9 deletions(-)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/CommandProcessor.cpp b/Source/Core/VideoCommon/Src/CommandProcessor.cpp
index a6e68f6298..77c02d1692 100644
--- a/Source/Core/VideoCommon/Src/CommandProcessor.cpp
+++ b/Source/Core/VideoCommon/Src/CommandProcessor.cpp
@@ -574,7 +574,7 @@ void ProcessFifoAllDistance()
 	if (IsOnThread())
 	{
 		while (!CommandProcessor::interruptWaiting && fifo.bFF_GPReadEnable &&
-			fifo.CPReadWriteDistance && !AtBreakpoint())
+			fifo.CPReadWriteDistance && !AtBreakpoint() && !PixelEngine::WaitingForPEInterrupt())
 			Common::YieldCPU();
 	}
 	bProcessFifoAllDistance = false;
@@ -597,10 +597,12 @@ void SetCpStatusRegister()
 		
 	m_CPStatusReg.Breakpoint = fifo.bFF_Breakpoint;
 	m_CPStatusReg.ReadIdle = (fifo.CPReadPointer == fifo.CPWritePointer) || (fifo.CPReadPointer == fifo.CPBreakpoint) ;
-	m_CPStatusReg.CommandIdle = !fifo.CPReadWriteDistance || (IsOnThread() && PixelEngine::WaitingForPEInterrupt());
+	m_CPStatusReg.CommandIdle = !fifo.CPReadWriteDistance;
 	m_CPStatusReg.UnderflowLoWatermark = fifo.bFF_LoWatermark;
 	m_CPStatusReg.OverflowHiWatermark = fifo.bFF_HiWatermark;
-
+	
+	PixelEngine::ResumeWaitingForPEInterrupt();
+	
 	INFO_LOG(COMMANDPROCESSOR,"\t Read from STATUS_REGISTER : %04x", m_CPStatusReg.Hex);
 	DEBUG_LOG(COMMANDPROCESSOR, "(r) status: iBP %s | fReadIdle %s | fCmdIdle %s | iOvF %s | iUndF %s"
 		, m_CPStatusReg.Breakpoint ?			"ON" : "OFF"
@@ -609,6 +611,8 @@ void SetCpStatusRegister()
 		, m_CPStatusReg.OverflowHiWatermark ?	"ON" : "OFF"
 		, m_CPStatusReg.UnderflowLoWatermark ?	"ON" : "OFF"
 			);
+
+	
 }
 
 void SetCpControlRegister()
diff --git a/Source/Core/VideoCommon/Src/PixelEngine.cpp b/Source/Core/VideoCommon/Src/PixelEngine.cpp
index 09cb9b8a5c..750e8fbb40 100644
--- a/Source/Core/VideoCommon/Src/PixelEngine.cpp
+++ b/Source/Core/VideoCommon/Src/PixelEngine.cpp
@@ -356,22 +356,16 @@ void UpdateInterrupts()
 
 void UpdateTokenInterrupt(bool active)
 {
-	if(interruptSetToken != active)
-	{
 		ProcessorInterface::SetInterrupt(INT_CAUSE_PE_TOKEN, active);
 		interruptSetToken = active;
-	}
 }
 
 void UpdateFinishInterrupt(bool active)
 {
-	if(interruptSetFinish != active)
-	{
 		ProcessorInterface::SetInterrupt(INT_CAUSE_PE_FINISH, active);
 		interruptSetFinish = active;
 		if (active)
 			State::ProcessRequestedStates(0);
-	}
 }
 
 // TODO(mb2): Refactor SetTokenINT_OnMainThread(u64 userdata, int cyclesLate).
@@ -473,4 +467,12 @@ bool WaitingForPEInterrupt()
 {
 	return CommandProcessor::interruptFinishWaiting  || CommandProcessor::interruptTokenWaiting || interruptSetFinish || interruptSetToken;
 }
+
+void ResumeWaitingForPEInterrupt()
+{
+	interruptSetFinish = false;
+	interruptSetToken = false;
+	CommandProcessor::interruptFinishWaiting = false;
+	CommandProcessor::interruptTokenWaiting = false;
+}
 } // end of namespace PixelEngine
diff --git a/Source/Core/VideoCommon/Src/PixelEngine.h b/Source/Core/VideoCommon/Src/PixelEngine.h
index 5e64300ef3..64f959009f 100644
--- a/Source/Core/VideoCommon/Src/PixelEngine.h
+++ b/Source/Core/VideoCommon/Src/PixelEngine.h
@@ -81,6 +81,7 @@ void SetFinish(void);
 void ResetSetFinish(void);
 void ResetSetToken(void);
 bool WaitingForPEInterrupt();
+void ResumeWaitingForPEInterrupt();
 
 // Bounding box functionality. Paper Mario (both) are a couple of the few games that use it.
 extern u16 bbox[4];
-- 
cgit v1.2.3


From fedf6055ce2d1ac9862eac8c2c112160b8d7fbe8 Mon Sep 17 00:00:00 2001
From: marcosvitali <marcosvitali@gmail.com>
Date: Sun, 11 Mar 2012 12:40:39 -0300
Subject: I've fixed Super Monkey Ball in some cases when the game write the
 WriteReadDistance need to be safe like the SafeCPRead. This fix is not
 related with the previous commits, but the previous commits help me to see
 that because in the new scenery SMB was hanging. May be in the past also
 doesn't boot some times because of that.

Please Test FZero boot also. Thanks.
---
 Source/Core/VideoCommon/Src/CommandProcessor.cpp | 17 +++++++++++++++--
 1 file changed, 15 insertions(+), 2 deletions(-)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/CommandProcessor.cpp b/Source/Core/VideoCommon/Src/CommandProcessor.cpp
index 77c02d1692..dbf4907160 100644
--- a/Source/Core/VideoCommon/Src/CommandProcessor.cpp
+++ b/Source/Core/VideoCommon/Src/CommandProcessor.cpp
@@ -171,11 +171,23 @@ void Read16(u16& _rReturnValue, const u32 _Address)
 	case FIFO_LO_WATERMARK_HI:	_rReturnValue = ReadHigh(fifo.CPLoWatermark); return;
 
 	case FIFO_RW_DISTANCE_LO:
+		if (IsOnThread())
+			if(fifo.CPWritePointer >= fifo.SafeCPReadPointer)
+				_rReturnValue = ReadLow (fifo.CPWritePointer - fifo.SafeCPReadPointer);
+			else
+				_rReturnValue = ReadLow (fifo.CPEnd - fifo.CPWritePointer + fifo.SafeCPReadPointer);
+		else
 		_rReturnValue = ReadLow (fifo.CPReadWriteDistance);
 		DEBUG_LOG(COMMANDPROCESSOR, "read FIFO_RW_DISTANCE_LO : %04x", _rReturnValue);
 		return;
 	case FIFO_RW_DISTANCE_HI:
-		_rReturnValue = ReadHigh(fifo.CPReadWriteDistance);
+		if (IsOnThread())
+			if(fifo.CPWritePointer >= fifo.SafeCPReadPointer)
+				_rReturnValue = ReadHigh (fifo.CPWritePointer - fifo.SafeCPReadPointer);
+			else
+				_rReturnValue = ReadHigh (fifo.CPEnd - fifo.CPWritePointer + fifo.SafeCPReadPointer);
+		else
+ 		_rReturnValue = ReadHigh(fifo.CPReadWriteDistance);
 		DEBUG_LOG(COMMANDPROCESSOR, "read FIFO_RW_DISTANCE_HI : %04x", _rReturnValue);
 		return;
 	case FIFO_WRITE_POINTER_LO:
@@ -356,6 +368,7 @@ void Write16(const u16 _Value, const u32 _Address)
 		break;
 	case FIFO_READ_POINTER_HI:
 		WriteHigh((u32 &)fifo.CPReadPointer, _Value);
+		fifo.SafeCPReadPointer = fifo.CPReadPointer;
 		DEBUG_LOG(COMMANDPROCESSOR,"\t write to FIFO_READ_POINTER_HI : %04x", _Value);
 		break;
 
@@ -596,7 +609,7 @@ void SetCpStatusRegister()
 	// Here always there is one fifo attached to the GPU
 		
 	m_CPStatusReg.Breakpoint = fifo.bFF_Breakpoint;
-	m_CPStatusReg.ReadIdle = (fifo.CPReadPointer == fifo.CPWritePointer) || (fifo.CPReadPointer == fifo.CPBreakpoint) ;
+	m_CPStatusReg.ReadIdle = !fifo.CPReadWriteDistance || (fifo.CPReadPointer == fifo.CPWritePointer) || (fifo.CPReadPointer == fifo.CPBreakpoint) ;
 	m_CPStatusReg.CommandIdle = !fifo.CPReadWriteDistance;
 	m_CPStatusReg.UnderflowLoWatermark = fifo.bFF_LoWatermark;
 	m_CPStatusReg.OverflowHiWatermark = fifo.bFF_HiWatermark;
-- 
cgit v1.2.3


From dc79d68e72f04f60366484f60bff1e39d7e6babe Mon Sep 17 00:00:00 2001
From: skidau <skidau@gmail.com>
Date: Tue, 13 Mar 2012 22:35:11 +1100
Subject: Added the corresponding change from r352ab2ba4394 into JITIL.

Tidied some code.
---
 Source/Core/VideoCommon/Src/CommandProcessor.cpp | 34 ++++++++----------------
 Source/Core/VideoCommon/Src/CommandProcessor.h   |  1 -
 Source/Core/VideoCommon/Src/PixelEngine.cpp      |  3 ---
 3 files changed, 11 insertions(+), 27 deletions(-)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/CommandProcessor.cpp b/Source/Core/VideoCommon/Src/CommandProcessor.cpp
index dbf4907160..3744acd200 100644
--- a/Source/Core/VideoCommon/Src/CommandProcessor.cpp
+++ b/Source/Core/VideoCommon/Src/CommandProcessor.cpp
@@ -141,7 +141,6 @@ void Init()
 
 void Read16(u16& _rReturnValue, const u32 _Address)
 {
-
 	INFO_LOG(COMMANDPROCESSOR, "(r): 0x%08x", _Address);
 	switch (_Address & 0xFFF)
 	{
@@ -177,7 +176,7 @@ void Read16(u16& _rReturnValue, const u32 _Address)
 			else
 				_rReturnValue = ReadLow (fifo.CPEnd - fifo.CPWritePointer + fifo.SafeCPReadPointer);
 		else
-		_rReturnValue = ReadLow (fifo.CPReadWriteDistance);
+			_rReturnValue = ReadLow (fifo.CPReadWriteDistance);
 		DEBUG_LOG(COMMANDPROCESSOR, "read FIFO_RW_DISTANCE_LO : %04x", _rReturnValue);
 		return;
 	case FIFO_RW_DISTANCE_HI:
@@ -187,7 +186,7 @@ void Read16(u16& _rReturnValue, const u32 _Address)
 			else
 				_rReturnValue = ReadHigh (fifo.CPEnd - fifo.CPWritePointer + fifo.SafeCPReadPointer);
 		else
- 		_rReturnValue = ReadHigh(fifo.CPReadWriteDistance);
+			_rReturnValue = ReadHigh(fifo.CPReadWriteDistance);
 		DEBUG_LOG(COMMANDPROCESSOR, "read FIFO_RW_DISTANCE_HI : %04x", _rReturnValue);
 		return;
 	case FIFO_WRITE_POINTER_LO:
@@ -499,18 +498,15 @@ void AbortFrame()
 
 void SetOverflowStatusFromGatherPipe()
 {
-	
-
 	fifo.bFF_HiWatermark = (fifo.CPReadWriteDistance > fifo.CPHiWatermark);
 	isHiWatermarkActive = fifo.bFF_HiWatermark && fifo.bFF_HiWatermarkInt && m_CPCtrlReg.GPReadEnable;
-	
+
     if (isHiWatermarkActive)
 	{
 		interruptSet = true;
         INFO_LOG(COMMANDPROCESSOR,"Interrupt set");
-        ProcessorInterface::SetInterrupt(INT_CAUSE_CP, true);    
+        ProcessorInterface::SetInterrupt(INT_CAUSE_CP, true);
 	}
-			   
 }
 
 void SetCpStatus()
@@ -520,12 +516,10 @@ void SetCpStatus()
     fifo.bFF_LoWatermark = (fifo.CPReadWriteDistance < fifo.CPLoWatermark);
 	
     // breakpoint     
-	
 	if (fifo.bFF_BPEnable)
     {
 		if (fifo.CPBreakpoint == fifo.CPReadPointer)
-        {        
-			
+        {
             if (!fifo.bFF_Breakpoint)
 			{
 				INFO_LOG(COMMANDPROCESSOR, "Hit breakpoint at %i", fifo.CPReadPointer);
@@ -607,15 +601,16 @@ void Shutdown()
 void SetCpStatusRegister()
 {
 	// Here always there is one fifo attached to the GPU
-		
 	m_CPStatusReg.Breakpoint = fifo.bFF_Breakpoint;
 	m_CPStatusReg.ReadIdle = !fifo.CPReadWriteDistance || (fifo.CPReadPointer == fifo.CPWritePointer) || (fifo.CPReadPointer == fifo.CPBreakpoint) ;
 	m_CPStatusReg.CommandIdle = !fifo.CPReadWriteDistance;
 	m_CPStatusReg.UnderflowLoWatermark = fifo.bFF_LoWatermark;
 	m_CPStatusReg.OverflowHiWatermark = fifo.bFF_HiWatermark;
-	
-	PixelEngine::ResumeWaitingForPEInterrupt();
-	
+
+	// HACK to compensate for slow response to PE interrupts in Time Splitters: Future Perfect
+	if (IsOnThread())
+		PixelEngine::ResumeWaitingForPEInterrupt();
+
 	INFO_LOG(COMMANDPROCESSOR,"\t Read from STATUS_REGISTER : %04x", m_CPStatusReg.Hex);
 	DEBUG_LOG(COMMANDPROCESSOR, "(r) status: iBP %s | fReadIdle %s | fCmdIdle %s | iOvF %s | iUndF %s"
 		, m_CPStatusReg.Breakpoint ?			"ON" : "OFF"
@@ -624,13 +619,10 @@ void SetCpStatusRegister()
 		, m_CPStatusReg.OverflowHiWatermark ?	"ON" : "OFF"
 		, m_CPStatusReg.UnderflowLoWatermark ?	"ON" : "OFF"
 			);
-
-	
 }
 
 void SetCpControlRegister()
 {
-
 	// If the new fifo is being attached We make sure there wont be SetFinish event pending.
 	// This protection fix eternal darkness booting, because the second SetFinish event when it is booting
 	// seems invalid or has a bug and hang the game.
@@ -638,7 +630,7 @@ void SetCpControlRegister()
 	if (!fifo.bFF_GPReadEnable && m_CPCtrlReg.GPReadEnable && !m_CPCtrlReg.BPEnable)
 	{
 		ProcessFifoEvents();
-		PixelEngine::ResetSetFinish();			
+		PixelEngine::ResetSetFinish();
 	}
 
 	fifo.bFF_BPInt = m_CPCtrlReg.BPInt;
@@ -653,9 +645,6 @@ void SetCpControlRegister()
 		ProcessorInterface::Fifo_CPUBase = fifo.CPBase;
 		ProcessorInterface::Fifo_CPUEnd = fifo.CPEnd;
 	}
-	// If overflown happens process the fifo to LoWatemark
-	//if (bProcessFifoToLoWatermark)
-	//	ProcessFifoToLoWatermark();
 			
 	if(fifo.bFF_GPReadEnable && !m_CPCtrlReg.GPReadEnable)
 	{
@@ -667,7 +656,6 @@ void SetCpControlRegister()
 		fifo.bFF_GPReadEnable = m_CPCtrlReg.GPReadEnable;
 	}
 
-			
 	DEBUG_LOG(COMMANDPROCESSOR, "\t GPREAD %s | BP %s | Int %s | OvF %s | UndF %s | LINK %s"
 		, fifo.bFF_GPReadEnable ?				"ON" : "OFF"
 		, fifo.bFF_BPEnable ?					"ON" : "OFF"
diff --git a/Source/Core/VideoCommon/Src/CommandProcessor.h b/Source/Core/VideoCommon/Src/CommandProcessor.h
index 2bf965a3ba..440c900dc3 100644
--- a/Source/Core/VideoCommon/Src/CommandProcessor.h
+++ b/Source/Core/VideoCommon/Src/CommandProcessor.h
@@ -25,7 +25,6 @@ class PointerWrap;
 
 extern bool MT;
 
-
 namespace CommandProcessor
 {
 
diff --git a/Source/Core/VideoCommon/Src/PixelEngine.cpp b/Source/Core/VideoCommon/Src/PixelEngine.cpp
index 750e8fbb40..38c85cc10c 100644
--- a/Source/Core/VideoCommon/Src/PixelEngine.cpp
+++ b/Source/Core/VideoCommon/Src/PixelEngine.cpp
@@ -322,7 +322,6 @@ void Write16(const u16 _iValue, const u32 _iAddress)
 		break;
 
 	case PE_TOKEN_REG:
-		//LOG(PIXELENGINE,"WEIRD: program wrote token: %i",_iValue);
 		PanicAlert("(w16) WTF? PowerPC program wrote token: %i", _iValue);
 		//only the gx pipeline is supposed to be able to write here
 		//g_token = _iValue;
@@ -384,8 +383,6 @@ void SetToken_OnMainThread(u64 userdata, int cyclesLate)
 		CommandProcessor::interruptTokenWaiting = false;
 		IncrementCheckContextId();
 	//}
-	//else
-	//	LOGV(PIXELENGINE, 1, "VIDEO Backend wrote token: %i", CommandProcessor::fifo.PEToken);
 }
 
 void SetFinish_OnMainThread(u64 userdata, int cyclesLate)
-- 
cgit v1.2.3


From 20eca1bf7e2e02dd18229dc2b6b83935ce0d4047 Mon Sep 17 00:00:00 2001
From: marcosvitali <marcosvitali@gmail.com>
Date: Sun, 18 Mar 2012 22:54:58 -0300
Subject: Ive fixed definitely Pokemon XD in dual core mode. This game is doing
 something not allowed. It attach to CPU the same fifo attached to the GPU in
 multibuffer mode. I added a check to prevent overwrite the GPU FIFO with the
 CPU FIFO. If the game do that on breakpoint the solution can fail. Fixed
 ReadWriteDistance calc when CPRead > CPWrite. Added Token and Finish cause to
 GP Jit checking.

---
 Source/Core/VideoCommon/Src/CommandProcessor.cpp | 19 +++++++++++++++++--
 Source/Core/VideoCommon/Src/CommandProcessor.h   |  3 ++-
 Source/Core/VideoCommon/Src/PixelEngine.cpp      |  2 +-
 3 files changed, 20 insertions(+), 4 deletions(-)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/CommandProcessor.cpp b/Source/Core/VideoCommon/Src/CommandProcessor.cpp
index 3744acd200..a4dddf537f 100644
--- a/Source/Core/VideoCommon/Src/CommandProcessor.cpp
+++ b/Source/Core/VideoCommon/Src/CommandProcessor.cpp
@@ -61,6 +61,7 @@ volatile bool interruptSet= false;
 volatile bool interruptWaiting= false;
 volatile bool interruptTokenWaiting = false;
 volatile bool interruptFinishWaiting = false;
+volatile bool waitingForPEInterruptDisable = false;
 
 bool IsOnThread()
 {
@@ -174,7 +175,7 @@ void Read16(u16& _rReturnValue, const u32 _Address)
 			if(fifo.CPWritePointer >= fifo.SafeCPReadPointer)
 				_rReturnValue = ReadLow (fifo.CPWritePointer - fifo.SafeCPReadPointer);
 			else
-				_rReturnValue = ReadLow (fifo.CPEnd - fifo.CPWritePointer + fifo.SafeCPReadPointer);
+				_rReturnValue = ReadLow (fifo.CPEnd - fifo.SafeCPReadPointer + fifo.CPWritePointer - fifo.CPBase + 32);
 		else
 			_rReturnValue = ReadLow (fifo.CPReadWriteDistance);
 		DEBUG_LOG(COMMANDPROCESSOR, "read FIFO_RW_DISTANCE_LO : %04x", _rReturnValue);
@@ -184,7 +185,7 @@ void Read16(u16& _rReturnValue, const u32 _Address)
 			if(fifo.CPWritePointer >= fifo.SafeCPReadPointer)
 				_rReturnValue = ReadHigh (fifo.CPWritePointer - fifo.SafeCPReadPointer);
 			else
-				_rReturnValue = ReadHigh (fifo.CPEnd - fifo.CPWritePointer + fifo.SafeCPReadPointer);
+				_rReturnValue = ReadHigh (fifo.CPEnd - fifo.SafeCPReadPointer + fifo.CPWritePointer - fifo.CPBase + 32);
 		else
 			_rReturnValue = ReadHigh(fifo.CPReadWriteDistance);
 		DEBUG_LOG(COMMANDPROCESSOR, "read FIFO_RW_DISTANCE_HI : %04x", _rReturnValue);
@@ -437,11 +438,25 @@ void Write32(const u32 _Data, const u32 _Address)
 
 void STACKALIGN GatherPipeBursted()
 {
+	ProcessFifoEvents();
 	// if we aren't linked, we don't care about gather pipe data
 	if (!m_CPCtrlReg.GPLinkEnable)
 	{
 		if (!IsOnThread())
 			RunGpu();
+		else
+		{
+			// In multibuffer mode is not allowed write in the same fifo attached to the GPU.
+			// Fix Pokemon XD in DC mode.
+			if((ProcessorInterface::Fifo_CPUEnd == fifo.CPEnd) && (ProcessorInterface::Fifo_CPUBase == fifo.CPBase)
+				 && fifo.CPReadWriteDistance > 0)
+			{
+				waitingForPEInterruptDisable = true;
+				ProcessFifoAllDistance();
+				waitingForPEInterruptDisable = false;
+			}
+		
+		}
 		return;
 	}
 
diff --git a/Source/Core/VideoCommon/Src/CommandProcessor.h b/Source/Core/VideoCommon/Src/CommandProcessor.h
index 440c900dc3..5d31453537 100644
--- a/Source/Core/VideoCommon/Src/CommandProcessor.h
+++ b/Source/Core/VideoCommon/Src/CommandProcessor.h
@@ -35,7 +35,8 @@ extern volatile bool interruptSet;
 extern volatile bool interruptWaiting;
 extern volatile bool interruptTokenWaiting;
 extern volatile bool interruptFinishWaiting;
-
+extern volatile bool waitingForPEInterruptDisable;
+ 
 // internal hardware addresses
 enum
 {
diff --git a/Source/Core/VideoCommon/Src/PixelEngine.cpp b/Source/Core/VideoCommon/Src/PixelEngine.cpp
index 38c85cc10c..e6dbdb8c26 100644
--- a/Source/Core/VideoCommon/Src/PixelEngine.cpp
+++ b/Source/Core/VideoCommon/Src/PixelEngine.cpp
@@ -462,7 +462,7 @@ void ResetSetToken()
 
 bool WaitingForPEInterrupt()
 {
-	return CommandProcessor::interruptFinishWaiting  || CommandProcessor::interruptTokenWaiting || interruptSetFinish || interruptSetToken;
+	return !CommandProcessor::waitingForPEInterruptDisable && (CommandProcessor::interruptFinishWaiting  || CommandProcessor::interruptTokenWaiting || interruptSetFinish || interruptSetToken);
 }
 
 void ResumeWaitingForPEInterrupt()
-- 
cgit v1.2.3


From 6202714efdf66987e57846d3297a5972c8b75c38 Mon Sep 17 00:00:00 2001
From: NeoBrainX <NeoBrainX@gmail.com>
Date: Tue, 20 Mar 2012 22:36:21 +0100
Subject: Fix some regressions from r3b38295cbd08.

---
 Source/Core/VideoCommon/Src/TextureCacheBase.cpp | 18 +++++++++++++-----
 1 file changed, 13 insertions(+), 5 deletions(-)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/TextureCacheBase.cpp b/Source/Core/VideoCommon/Src/TextureCacheBase.cpp
index cc14499330..00ff9610be 100644
--- a/Source/Core/VideoCommon/Src/TextureCacheBase.cpp
+++ b/Source/Core/VideoCommon/Src/TextureCacheBase.cpp
@@ -172,8 +172,10 @@ void TextureCache::ClearRenderTargets()
 	TexCache::iterator
 		iter = textures.begin(),
 		tcend = textures.end();
+
 	for (; iter!=tcend; ++iter)
-		iter->second->type = TCET_NORMAL;
+		if (iter->second->type != TCET_EC_DYNAMIC)
+			iter->second->type = TCET_NORMAL;
 }
 
 TextureCache::TCacheEntryBase* TextureCache::Load(unsigned int stage,
@@ -238,6 +240,9 @@ TextureCache::TCacheEntryBase* TextureCache::Load(unsigned int stage,
 		// 2. a) For EFB copies, only the hash and the texture address need to match
 		if (entry->IsEfbCopy() && tex_hash == entry->hash && address == entry->addr)
 		{
+			if (entry->type != TCET_EC_VRAM)
+				entry->type = TCET_NORMAL;
+
 			// TODO: Print a warning if the format changes! In this case, we could reinterpret the internal texture object data to the new pixel format (similiar to what is already being done in Renderer::ReinterpretPixelFormat())
 			goto return_entry;
 		}
@@ -318,8 +323,8 @@ TextureCache::TCacheEntryBase* TextureCache::Load(unsigned int stage,
 	entry->SetGeneralParameters(address, texture_size, full_format, entry->num_mipmaps);
 	entry->SetDimensions(nativeW, nativeH, width, height);
 	entry->hash = tex_hash;
-	if (g_ActiveConfig.bCopyEFBToTexture) entry->type = TCET_NORMAL;
-	else if (entry->IsEfbCopy()) entry->type = TCET_EC_DYNAMIC;
+	if (entry->IsEfbCopy() && !g_ActiveConfig.bCopyEFBToTexture) entry->type = TCET_EC_DYNAMIC;
+	else entry->type = TCET_NORMAL;
 
 	// load texture
 	entry->Load(width, height, expandedWidth, 0, (texLevels == 0));
@@ -647,8 +652,11 @@ void TextureCache::CopyRenderTargetToTexture(u32 dstAddr, unsigned int dstFormat
 		if ((entry->type == TCET_EC_VRAM && entry->virtual_width == scaled_tex_w && entry->virtual_height == scaled_tex_h) 
 			|| (entry->type == TCET_EC_DYNAMIC && entry->native_width == tex_w && entry->native_height == tex_h))
 		{
-			scaled_tex_w = tex_w;
-			scaled_tex_h = tex_h;
+			if (entry->type == TCET_EC_DYNAMIC)
+			{
+				scaled_tex_w = tex_w;
+				scaled_tex_h = tex_h;
+			}
 		}
 		else
 		{
-- 
cgit v1.2.3


From 006923e871e496a7c6c0fe9ba76da4ee939ce14e Mon Sep 17 00:00:00 2001
From: skidau <skidau@gmail.com>
Date: Thu, 22 Mar 2012 20:36:41 +1100
Subject: Disabled zcomploc while update_enable has been set in zmode.  Fixes
 zcomploc in F-Zero GX.

---
 Source/Core/VideoCommon/Src/PixelShaderGen.cpp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/PixelShaderGen.cpp b/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
index e4944c842c..ec4eb2c70b 100644
--- a/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
+++ b/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
@@ -1147,7 +1147,7 @@ static bool WriteAlphaTest(char *&p, API_TYPE ApiType,DSTALPHA_MODE dstAlphaMode
 		WRITE(p, "ocol1 = 0;\n");
 	if (DepthTextureEnable)
 		WRITE(p, "depth = 1.f;\n");
-	if (!bpmem.zcontrol.zcomploc)
+	if (!(bpmem.zcontrol.zcomploc && bpmem.zmode.updateenable))
 	{
 		WRITE(p, "discard;\n");
 		if (ApiType != API_D3D11)
-- 
cgit v1.2.3


From 339ee98e6258024622ef509aafe25ff46ad19341 Mon Sep 17 00:00:00 2001
From: Pierre Bourdon <delroth@gmail.com>
Date: Thu, 22 Mar 2012 22:10:23 +0100
Subject: Add a comment before the zcomploc test explaining the
 implementation/hack

---
 Source/Core/VideoCommon/Src/PixelShaderGen.cpp | 10 ++++++++++
 1 file changed, 10 insertions(+)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/PixelShaderGen.cpp b/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
index 46ac264889..79afc45974 100644
--- a/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
+++ b/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
@@ -1198,6 +1198,16 @@ static bool WriteAlphaTest(char *&p, API_TYPE ApiType,DSTALPHA_MODE dstAlphaMode
 		WRITE(p, "ocol1 = 0;\n");
 	if (DepthTextureEnable)
 		WRITE(p, "depth = 1.f;\n");
+
+	// HAXX: zcomploc is a way to control whether depth test is done before
+	// or after texturing and alpha test. PC GPU have no way to support this
+	// feature properly as of 2012: depth buffer and depth test are not
+	// programmable and the depth test is always done after texturing.
+	//
+	// We implement "depth test before texturing" by discarding the fragment
+	// when the alpha test fail. This is not a correct implementation because
+	// even if the depth test fails the fragment could be alpha blended, but
+	// we don't have a choice.
 	if (!(bpmem.zcontrol.zcomploc && bpmem.zmode.updateenable))
 	{
 		WRITE(p, "discard;\n");
-- 
cgit v1.2.3


From c4fbb6e3771960b1bb1f28571612e4c60d0c9817 Mon Sep 17 00:00:00 2001
From: skidau <skidau@gmail.com>
Date: Fri, 23 Mar 2012 22:10:48 +1100
Subject: Checked PRELOAD_MODE to determine if TMEMODD or TMEMEVEN should be
 used.

Fixes issue 5212.
---
 Source/Core/VideoCommon/Src/BPStructs.cpp | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/BPStructs.cpp b/Source/Core/VideoCommon/Src/BPStructs.cpp
index 6cb102555b..01b09e351b 100644
--- a/Source/Core/VideoCommon/Src/BPStructs.cpp
+++ b/Source/Core/VideoCommon/Src/BPStructs.cpp
@@ -476,10 +476,15 @@ void BPWritten(const BPCmd& bp)
 		// if this is different from 0, manual TMEM management is used.
 		if (bp.newvalue != 0)
 		{
-			// NOTE(neobrain): Apparently tmemodd doesn't affect hardware behavior at all (libogc uses it just as a buffer and switches its contents with tmemeven whenever this is called)
 			BPS_TmemConfig& tmem_cfg = bpmem.tmem_config;
 			u8* ram_ptr = Memory::GetPointer(tmem_cfg.preload_addr << 5);
-			u32 tmem_addr = tmem_cfg.preload_tmem_even * TMEM_LINE_SIZE;
+			u32 tmem_addr = 0;
+
+			if (bp.newvalue >> 16)
+				tmem_addr = tmem_cfg.preload_tmem_odd * TMEM_LINE_SIZE;
+			else
+				tmem_addr = tmem_cfg.preload_tmem_even * TMEM_LINE_SIZE;
+
 			u32 size = tmem_cfg.preload_tile_info.count * 32;
 			memcpy(texMem + tmem_addr, ram_ptr, size);
 		}
-- 
cgit v1.2.3


From 41bcf657f89d912bb048247767c211f63eb13e94 Mon Sep 17 00:00:00 2001
From: skidau <skidau@gmail.com>
Date: Fri, 23 Mar 2012 23:20:19 +1100
Subject: Adjusted the DX9 offset of the Sampling address to more closely match
 what we are seeing in games.

Fixes issue 5305.
---
 Source/Core/VideoCommon/Src/PixelShaderGen.cpp | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/PixelShaderGen.cpp b/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
index 3ef923dc3e..da78c01741 100644
--- a/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
+++ b/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
@@ -1106,9 +1106,9 @@ void SampleTexture(char *&p, const char *destination, const char *texcoords, con
 		WRITE(p, "%s=Tex%d.Sample(samp%d, %s.xy * "I_TEXDIMS"[%d].xy).%s;\n", destination, texmap,texmap, texcoords, texmap, texswap);
 	else if (ApiType & API_D3D9)
 	{
-		// D3D9 uses different pixel to texel mapping, so we need to offset our sampling address by half a pixel (assuming native and virtual texture dimensions match each other, otherwise some math is involved).
+		// D3D9 uses different pixel to texel mapping, so we need to offset our sampling address.
 		// Read the MSDN article "Directly Mapping Texels to Pixels (Direct3D 9)" for further info.
-		WRITE(p, "%s=tex2D(samp%d, (%s.xy + 0.5f*"I_VTEXSCALE"[%d].%s) * "I_TEXDIMS"[%d].xy).%s;\n", destination, texmap, texcoords, texmap/2, (texmap&1)?"zw":"xy", texmap, texswap);
+		WRITE(p, "%s=tex2D(samp%d, (%s.xy + 0.0005f*"I_VTEXSCALE"[%d].%s) * "I_TEXDIMS"[%d].xy).%s;\n", destination, texmap, texcoords, texmap/2, (texmap&1)?"zw":"xy", texmap, texswap);
 	}
 	else
 		WRITE(p, "%s=tex2D(samp%d, %s.xy * "I_TEXDIMS"[%d].xy).%s;\n", destination, texmap, texcoords, texmap, texswap);
-- 
cgit v1.2.3


From fee98b426b5a0c4d78042a2db3b352c4fe79ef56 Mon Sep 17 00:00:00 2001
From: NeoBrainX <NeoBrainX@gmail.com>
Date: Sat, 24 Mar 2012 04:47:28 +0100
Subject: Revert r41bcf657f89d, r3a9fed0ba285, r9adc119e3c2f and re58692653afd
 (all of them are wrong).

---
 Source/Core/VideoCommon/Src/PixelShaderGen.cpp     | 14 ++------
 Source/Core/VideoCommon/Src/PixelShaderGen.h       | 19 +++++-----
 Source/Core/VideoCommon/Src/PixelShaderManager.cpp | 42 +++-------------------
 Source/Core/VideoCommon/Src/PixelShaderManager.h   |  6 ++--
 4 files changed, 18 insertions(+), 63 deletions(-)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/PixelShaderGen.cpp b/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
index da78c01741..662c9c59a8 100644
--- a/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
+++ b/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
@@ -560,10 +560,6 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType
 	WRITE(p, "uniform float4 "I_KCOLORS"[4] : register(c%d);\n", C_KCOLORS);
 	WRITE(p, "uniform float4 "I_ALPHA"[1] : register(c%d);\n", C_ALPHA);
 	WRITE(p, "uniform float4 "I_TEXDIMS"[8] : register(c%d);\n", C_TEXDIMS);
-	if (ApiType & API_D3D9)
-	{
-		WRITE(p, "uniform float4 "I_VTEXSCALE"[4] : register(c%d);\n", C_VTEXSCALE);
-	}
 	WRITE(p, "uniform float4 "I_ZBIAS"[2] : register(c%d);\n", C_ZBIAS);
 	WRITE(p, "uniform float4 "I_INDTEXSCALE"[2] : register(c%d);\n", C_INDTEXSCALE);
 	WRITE(p, "uniform float4 "I_INDTEXMTX"[6] : register(c%d);\n", C_INDTEXMTX);
@@ -1103,15 +1099,9 @@ static void WriteStage(char *&p, int n, API_TYPE ApiType)
 void SampleTexture(char *&p, const char *destination, const char *texcoords, const char *texswap, int texmap, API_TYPE ApiType)
 {
 	if (ApiType == API_D3D11)
-		WRITE(p, "%s=Tex%d.Sample(samp%d, %s.xy * "I_TEXDIMS"[%d].xy).%s;\n", destination, texmap,texmap, texcoords, texmap, texswap);
-	else if (ApiType & API_D3D9)
-	{
-		// D3D9 uses different pixel to texel mapping, so we need to offset our sampling address.
-		// Read the MSDN article "Directly Mapping Texels to Pixels (Direct3D 9)" for further info.
-		WRITE(p, "%s=tex2D(samp%d, (%s.xy + 0.0005f*"I_VTEXSCALE"[%d].%s) * "I_TEXDIMS"[%d].xy).%s;\n", destination, texmap, texcoords, texmap/2, (texmap&1)?"zw":"xy", texmap, texswap);
-	}
+		WRITE(p, "%s=Tex%d.Sample(samp%d,%s.xy * "I_TEXDIMS"[%d].xy).%s;\n", destination, texmap,texmap, texcoords, texmap, texswap);
 	else
-		WRITE(p, "%s=tex2D(samp%d, %s.xy * "I_TEXDIMS"[%d].xy).%s;\n", destination, texmap, texcoords, texmap, texswap);
+		WRITE(p, "%s=tex2D(samp%d,%s.xy * "I_TEXDIMS"[%d].xy).%s;\n", destination, texmap, texcoords, texmap, texswap);
 }
 
 static const char *tevAlphaFuncsTable[] =
diff --git a/Source/Core/VideoCommon/Src/PixelShaderGen.h b/Source/Core/VideoCommon/Src/PixelShaderGen.h
index 7374251c96..31242a916e 100644
--- a/Source/Core/VideoCommon/Src/PixelShaderGen.h
+++ b/Source/Core/VideoCommon/Src/PixelShaderGen.h
@@ -24,27 +24,26 @@
 #define I_KCOLORS     "k"
 #define I_ALPHA       "alphaRef"
 #define I_TEXDIMS     "texdim"
-#define I_VTEXSCALE   "vtexscale"
 #define I_ZBIAS       "czbias"
 #define I_INDTEXSCALE "cindscale"
 #define I_INDTEXMTX   "cindmtx"
 #define I_FOG         "cfog"
 #define I_PLIGHTS	  "cLights"
-#define I_PMATERIALS  "cmtrl"
+#define I_PMATERIALS   "cmtrl"
 
 #define C_COLORMATRIX	0						// 0
 #define C_COLORS		0						// 0
 #define C_KCOLORS		(C_COLORS + 4)			// 4
 #define C_ALPHA			(C_KCOLORS + 4)			// 8
 #define C_TEXDIMS		(C_ALPHA + 1)			// 9
-#define C_VTEXSCALE		(C_TEXDIMS + 8)			//17 - virtual texture scaling factor (e.g. custom textures, scaled EFB copies)
-#define C_ZBIAS			(C_VTEXSCALE + 4)		//21
-#define C_INDTEXSCALE	(C_ZBIAS + 2)			//23
-#define C_INDTEXMTX		(C_INDTEXSCALE + 2)		//25
-#define C_FOG			(C_INDTEXMTX + 6)		//31
-#define C_PLIGHTS		(C_FOG + 3)				//34
-#define C_PMATERIALS	(C_PLIGHTS + 40)		//74
-#define C_PENVCONST_END (C_PMATERIALS + 4)		//78
+#define C_ZBIAS			(C_TEXDIMS + 8)			//17
+#define C_INDTEXSCALE	(C_ZBIAS + 2)			//19
+#define C_INDTEXMTX		(C_INDTEXSCALE + 2)		//21
+#define C_FOG			(C_INDTEXMTX + 6)		//27
+
+#define C_PLIGHTS		(C_FOG + 3)
+#define C_PMATERIALS	(C_PLIGHTS + 40)
+#define C_PENVCONST_END (C_PMATERIALS + 4)
 #define PIXELSHADERUID_MAX_VALUES 70
 #define PIXELSHADERUID_MAX_VALUES_SAFE 120
 
diff --git a/Source/Core/VideoCommon/Src/PixelShaderManager.cpp b/Source/Core/VideoCommon/Src/PixelShaderManager.cpp
index 3e0f6d73c4..2521f80500 100644
--- a/Source/Core/VideoCommon/Src/PixelShaderManager.cpp
+++ b/Source/Core/VideoCommon/Src/PixelShaderManager.cpp
@@ -36,11 +36,9 @@ static bool s_bFogRangeAdjustChanged;
 static int nLightsChanged[2]; // min,max
 static float lastRGBAfull[2][4][4];
 static u8 s_nTexDimsChanged;
-static u8 s_nVirtualTexScalesChanged;
 static u8 s_nIndTexScaleChanged;
 static u32 lastAlpha;
 static u32 lastTexDims[8]; // width | height << 16 | wrap_s << 28 | wrap_t << 30
-static float lastVirtualTexScales[16]; // even fields: width ratio;   odd fields: height ratio
 static u32 lastZBias;
 static int nMaterialsChanged;
 
@@ -63,7 +61,6 @@ void PixelShaderManager::Init()
 {
 	lastAlpha = 0;
 	memset(lastTexDims, 0, sizeof(lastTexDims));
-	memset(lastVirtualTexScales, 0, sizeof(lastVirtualTexScales));
 	lastZBias = 0;
 	memset(lastRGBAfull, 0, sizeof(lastRGBAfull));
 	Dirty();
@@ -73,7 +70,6 @@ void PixelShaderManager::Dirty()
 {
 	s_nColorsChanged[0] = s_nColorsChanged[1] = 15;
 	s_nTexDimsChanged = 0xFF;
-	s_nVirtualTexScalesChanged = 0xFF;
 	s_nIndTexScaleChanged = 0xFF;
 	s_nIndTexMtxChanged = 15;
 	s_bAlphaChanged = s_bZBiasChanged = s_bZTextureTypeChanged = s_bDepthRangeChanged = true;
@@ -87,7 +83,7 @@ void PixelShaderManager::Shutdown()
 
 }
 
-void PixelShaderManager::SetConstants(API_TYPE api_type)
+void PixelShaderManager::SetConstants()
 {
     for (int i = 0; i < 2; ++i)
 	{
@@ -113,16 +109,6 @@ void PixelShaderManager::SetConstants(API_TYPE api_type)
         s_nTexDimsChanged = 0;
     }
 
-	if ((api_type & API_D3D9) && s_nVirtualTexScalesChanged)
-	{
-		for (int i = 0; i < 8; i += 2)
-		{
-			if (s_nVirtualTexScalesChanged & (3<<i))
-				SetPSVirtualTexScalePair(i/2);
-		}
-		s_nVirtualTexScalesChanged = 0;
-	}
-
     if (s_bAlphaChanged)
 	{
 		SetPSConstant4f(C_ALPHA, (lastAlpha&0xff)/255.0f, ((lastAlpha>>8)&0xff)/255.0f, 0, ((lastAlpha>>16)&0xff)/255.0f);
@@ -352,13 +338,6 @@ void PixelShaderManager::SetPSTextureDims(int texid)
 	SetPSConstant4fv(C_TEXDIMS + texid, fdims);
 }
 
-void PixelShaderManager::SetPSVirtualTexScalePair(int texpairid)
-{
-	PRIM_LOG("vtexscale%d: %f %f %f %f\n", texpairid, lastVirtualTexScales[texpairid*4], lastVirtualTexScales[texpairid*4+1],
-														lastVirtualTexScales[texpairid*4+2], lastVirtualTexScales[texpairid*4+3]);
-	SetPSConstant4fv(C_VTEXSCALE + texpairid, &lastVirtualTexScales[texpairid*4]);
-}
-
 // This one is high in profiles (0.5%). TODO: Move conversion out, only store the raw color value
 // and update it when the shader constant is set, only.
 void PixelShaderManager::SetColorChanged(int type, int num, bool high)
@@ -397,25 +376,14 @@ void PixelShaderManager::SetDestAlpha(const ConstantAlpha& alpha)
     }
 }
 
-void PixelShaderManager::SetTexDims(int texmapid, u32 width, u32 height, u32 virtual_width, u32 virtual_height, u32 wraps, u32 wrapt, API_TYPE api_type)
+void PixelShaderManager::SetTexDims(int texmapid, u32 width, u32 height, u32 wraps, u32 wrapt)
 {
     u32 wh = width | (height << 16) | (wraps << 28) | (wrapt << 30);
-
-	bool refresh = lastTexDims[texmapid] != wh;
-	if (api_type & API_D3D9)
+    if (lastTexDims[texmapid] != wh)
 	{
-		refresh |= (lastVirtualTexScales[texmapid*2] != (float)width / (float)virtual_width);
-		refresh |= (lastVirtualTexScales[texmapid*2+1] != (float)height / (float)virtual_height);
-	}
-
-	if (refresh)
-	{
-		lastTexDims[texmapid] = wh;
-		lastVirtualTexScales[texmapid*2] = (float)width / (float)virtual_width;
-		lastVirtualTexScales[texmapid*2+1] = (float)height / (float)virtual_height;
+        lastTexDims[texmapid] = wh;
 		s_nTexDimsChanged |= 1 << texmapid;
-		s_nVirtualTexScalesChanged |= 1 << texmapid;
-	}
+    }
 }
 
 void PixelShaderManager::SetZTextureBias(u32 bias)
diff --git a/Source/Core/VideoCommon/Src/PixelShaderManager.h b/Source/Core/VideoCommon/Src/PixelShaderManager.h
index 5cbe86cae4..2d1c01cad6 100644
--- a/Source/Core/VideoCommon/Src/PixelShaderManager.h
+++ b/Source/Core/VideoCommon/Src/PixelShaderManager.h
@@ -26,20 +26,18 @@
 class PixelShaderManager
 {
 	static void SetPSTextureDims(int texid);
-	static void SetPSVirtualTexScalePair(int texpairid);
-
 public:
 	static void Init();
 	static void Dirty();
 	static void Shutdown();
 
-	static void SetConstants(API_TYPE api_type); // sets pixel shader constants
+	static void SetConstants(); // sets pixel shader constants
 
 	// constant management, should be called after memory is committed
 	static void SetColorChanged(int type, int index, bool high);
 	static void SetAlpha(const AlphaFunc& alpha);
 	static void SetDestAlpha(const ConstantAlpha& alpha);
-	static void SetTexDims(int texmapid, u32 width, u32 height, u32 virtual_width, u32 virtual_height, u32 wraps, u32 wrapt, API_TYPE api_type);
+	static void SetTexDims(int texmapid, u32 width, u32 height, u32 wraps, u32 wrapt);
 	static void SetZTextureBias(u32 bias);
 	static void SetViewportChanged();
 	static void SetIndMatrixChanged(int matrixidx);
-- 
cgit v1.2.3


From 7a1744575d7569dd52739836685b6e6643ace61a Mon Sep 17 00:00:00 2001
From: NeoBrainX <NeoBrainX@gmail.com>
Date: Sat, 24 Mar 2012 16:30:00 +0100
Subject: D3D9: Fix texel to pixel mapping when sampling textures properly.

---
 Source/Core/VideoCommon/Src/RenderBase.cpp          | 2 ++
 Source/Core/VideoCommon/Src/VertexShaderGen.cpp     | 7 +++++++
 Source/Core/VideoCommon/Src/VertexShaderGen.h       | 2 +-
 Source/Core/VideoCommon/Src/VertexShaderManager.cpp | 6 +++++-
 4 files changed, 15 insertions(+), 2 deletions(-)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/RenderBase.cpp b/Source/Core/VideoCommon/Src/RenderBase.cpp
index 3b06bbb1c0..ef0fff57f7 100644
--- a/Source/Core/VideoCommon/Src/RenderBase.cpp
+++ b/Source/Core/VideoCommon/Src/RenderBase.cpp
@@ -43,6 +43,7 @@
 #include "XFMemory.h"
 #include "FifoPlayer/FifoRecorder.h"
 #include "AVIDump.h"
+#include "VertexShaderManager.h"
 
 #include <cmath>
 #include <string>
@@ -193,6 +194,7 @@ bool Renderer::CalculateTargetSize(int multiplier)
 	{
 		s_target_width  = newEFBWidth;
 		s_target_height = newEFBHeight;
+		VertexShaderManager::SetViewportChanged();
 		return true;
 	}
 	return false;
diff --git a/Source/Core/VideoCommon/Src/VertexShaderGen.cpp b/Source/Core/VideoCommon/Src/VertexShaderGen.cpp
index 5076f9f044..3df779a7ba 100644
--- a/Source/Core/VideoCommon/Src/VertexShaderGen.cpp
+++ b/Source/Core/VideoCommon/Src/VertexShaderGen.cpp
@@ -493,6 +493,13 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type)
 	    //seems to get rather complicated
 	}
 
+	if (api_type & API_D3D9)
+	{
+		// D3D9 is addressing pixel centers instead of pixel boundaries in clip space.
+		// Thus we need to offset the final position by half a pixel
+		WRITE(p, "o.pos = o.pos + float4("I_DEPTHPARAMS".z, "I_DEPTHPARAMS".w, 0.f, 0.f);\n");
+	}
+
 	WRITE(p, "return o;\n}\n");
 
 
diff --git a/Source/Core/VideoCommon/Src/VertexShaderGen.h b/Source/Core/VideoCommon/Src/VertexShaderGen.h
index 0c522286b1..cb253a9b6c 100644
--- a/Source/Core/VideoCommon/Src/VertexShaderGen.h
+++ b/Source/Core/VideoCommon/Src/VertexShaderGen.h
@@ -35,7 +35,7 @@
 #define I_TRANSFORMMATRICES     "ctrmtx"
 #define I_NORMALMATRICES        "cnmtx"
 #define I_POSTTRANSFORMMATRICES "cpostmtx"
-#define I_DEPTHPARAMS           "cDepth"
+#define I_DEPTHPARAMS           "cDepth" // farZ, zRange, scaled viewport width, scaled viewport height
 
 #define C_POSNORMALMATRIX        0
 #define C_PROJECTION            (C_POSNORMALMATRIX + 6)
diff --git a/Source/Core/VideoCommon/Src/VertexShaderManager.cpp b/Source/Core/VideoCommon/Src/VertexShaderManager.cpp
index d7d25c5c85..b6afbc61c6 100644
--- a/Source/Core/VideoCommon/Src/VertexShaderManager.cpp
+++ b/Source/Core/VideoCommon/Src/VertexShaderManager.cpp
@@ -306,7 +306,11 @@ void VertexShaderManager::SetConstants()
 	if (bViewportChanged)
 	{
 		bViewportChanged = false;
-		SetVSConstant4f(C_DEPTHPARAMS,xfregs.viewport.farZ / 16777216.0f,xfregs.viewport.zRange / 16777216.0f,0.0f,0.0f);
+		SetVSConstant4f(C_DEPTHPARAMS,
+						xfregs.viewport.farZ / 16777216.0f,
+						xfregs.viewport.zRange / 16777216.0f,
+						-1.f / (float)g_renderer->EFBToScaledX((int)ceil(2.0f * xfregs.viewport.wd)),
+						1.f / (float)g_renderer->EFBToScaledY((int)ceil(-2.0f * xfregs.viewport.ht)));
 		// This is so implementation-dependent that we can't have it here.
 		UpdateViewport(s_viewportCorrection);
 		bProjectionChanged = true;
-- 
cgit v1.2.3


From 11959a378169eec95f868b346ddc9ac72cb92689 Mon Sep 17 00:00:00 2001
From: Timo Wiren <timo.wiren@gmail.com>
Date: Sat, 24 Mar 2012 19:41:13 +0200
Subject: Matching new and delete.

---
 Source/Core/VideoCommon/Src/OpenCL/OCLTextureDecoder.cpp | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/OpenCL/OCLTextureDecoder.cpp b/Source/Core/VideoCommon/Src/OpenCL/OCLTextureDecoder.cpp
index 0f4809a6cc..8aabf2a066 100644
--- a/Source/Core/VideoCommon/Src/OpenCL/OCLTextureDecoder.cpp
+++ b/Source/Core/VideoCommon/Src/OpenCL/OCLTextureDecoder.cpp
@@ -143,8 +143,8 @@ void TexDecoder_OpenCL_Initialize()
 					}
 				}
 			}
-			delete header;
-			delete binary;
+			delete [] header;
+			delete [] binary;
 		}
 
 		// If an error occurred using the kernel binary, recompile the kernels
-- 
cgit v1.2.3


From 9ed83765e69f81ae1799e8e8ff8cfada2f53e9a2 Mon Sep 17 00:00:00 2001
From: rodolfoosvaldobogado <rodolfoosvaldobogado@gmail.com>
Date: Sun, 25 Mar 2012 00:01:47 -0300
Subject: long time no commits :) a little code cleaning to avoid duplicated
 execution of AlphaPreTest and a little correction to some comments from the
 previous commits. this change must behave exactly like last revision, if
 something is broken please let me know

---
 Source/Core/VideoCommon/Src/PixelShaderGen.cpp | 134 ++++++++++++-------------
 1 file changed, 63 insertions(+), 71 deletions(-)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/PixelShaderGen.cpp b/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
index 89032c67ea..76964f248c 100644
--- a/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
+++ b/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
@@ -299,7 +299,7 @@ void ValidatePixelShaderIDs(API_TYPE api, PIXELSHADERUIDSAFE old_id, const std::
 static void WriteStage(char *&p, int n, API_TYPE ApiType);
 static void SampleTexture(char *&p, const char *destination, const char *texcoords, const char *texswap, int texmap, API_TYPE ApiType);
 // static void WriteAlphaCompare(char *&p, int num, int comp);
-static bool WriteAlphaTest(char *&p, API_TYPE ApiType,DSTALPHA_MODE dstAlphaMode);
+static void WriteAlphaTest(char *&p, API_TYPE ApiType,DSTALPHA_MODE dstAlphaMode);
 static void WriteFog(char *&p);
 
 static const char *tevKSelTableC[] = // KCSEL
@@ -619,22 +619,29 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType
 
 	char* pmainstart = p;
 	int Pretest = AlphaPreTest();
-	if (dstAlphaMode == DSTALPHA_ALPHA_PASS && !DepthTextureEnable && Pretest >= 0)
+	if(Pretest >= 0 && !DepthTextureEnable)
 	{
 		if (!Pretest)
 		{
-			// alpha test will always fail, so restart the shader and just make it an empty function
+			// alpha test will always fail, so restart the shader and just make it an empty function		
 			WRITE(p, "ocol0 = 0;\n");
+			if(DepthTextureEnable)
+				WRITE(p, "depth = 1.f;\n");
+			if(dstAlphaMode == DSTALPHA_DUAL_SOURCE_BLEND)
+					WRITE(p, "ocol1 = 0;\n");
 			WRITE(p, "discard;\n");
 			if(ApiType != API_D3D11)
 				WRITE(p, "return;\n");
 		}
-		else
+		else if (dstAlphaMode == DSTALPHA_ALPHA_PASS)
 		{
 			WRITE(p, "  ocol0 = "I_ALPHA"[0].aaaa;\n");
 		}
-		WRITE(p, "}\n");
-		return text;
+		if(!Pretest || dstAlphaMode == DSTALPHA_ALPHA_PASS)
+		{
+			WRITE(p, "}\n");
+			return text;
+		}
 	}
 
 	WRITE(p, "  float4 c0 = "I_COLORS"[1], c1 = "I_COLORS"[2], c2 = "I_COLORS"[3], prev = float4(0.0f, 0.0f, 0.0f, 0.0f), textemp = float4(0.0f, 0.0f, 0.0f, 0.0f), rastemp = float4(0.0f, 0.0f, 0.0f, 0.0f), konsttemp = float4(0.0f, 0.0f, 0.0f, 0.0f);\n"
@@ -723,65 +730,53 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType
 	// emulation of unsigned 8 overflow when casting
 	WRITE(p, "prev = frac(4.0f + prev * (255.0f/256.0f)) * (256.0f/255.0f);\n");
 
-	// TODO: Why are we doing a second alpha pretest here?
-	if (!WriteAlphaTest(p, ApiType, dstAlphaMode))
+	if(Pretest ==  -1)
 	{
-		// alpha test will always fail, so restart the shader and just make it an empty function
-		p = pmainstart;
-		WRITE(p, "ocol0 = 0;\n");
-		if(DepthTextureEnable)
-			WRITE(p, "depth = 1.f;\n");
-		if(dstAlphaMode == DSTALPHA_DUAL_SOURCE_BLEND)
-				WRITE(p, "ocol1 = 0;\n");
-		WRITE(p, "discard;\n");
-		if(ApiType != API_D3D11)
-			WRITE(p, "return;\n");
+		WriteAlphaTest(p, ApiType, dstAlphaMode);
 	}
-	else
-	{
-		if((bpmem.fog.c_proj_fsel.fsel != 0) || DepthTextureEnable)
-		{
 
-			// the screen space depth value = far z + (clip z / clip w) * z range
-			WRITE(p, "float zCoord = "I_ZBIAS"[1].x + (clipPos.z / clipPos.w) * "I_ZBIAS"[1].y;\n");
-		}
+	if((bpmem.fog.c_proj_fsel.fsel != 0) || DepthTextureEnable)
+	{
+		// the screen space depth value = far z + (clip z / clip w) * z range
+		WRITE(p, "float zCoord = "I_ZBIAS"[1].x + (clipPos.z / clipPos.w) * "I_ZBIAS"[1].y;\n");
+	}
 
-		if (DepthTextureEnable)
+	if (DepthTextureEnable)
+	{
+		// use the texture input of the last texture stage (textemp), hopefully this has been read and is in correct format...
+		if (bpmem.ztex2.op != ZTEXTURE_DISABLE && !bpmem.zcontrol.zcomploc && bpmem.zmode.testenable && bpmem.zmode.updateenable)
 		{
-			// use the texture input of the last texture stage (textemp), hopefully this has been read and is in correct format...
-			if (bpmem.ztex2.op != ZTEXTURE_DISABLE && !bpmem.zcontrol.zcomploc && bpmem.zmode.testenable && bpmem.zmode.updateenable)
-			{
-				if (bpmem.ztex2.op == ZTEXTURE_ADD)
-					WRITE(p, "zCoord = dot("I_ZBIAS"[0].xyzw, textemp.xyzw) + "I_ZBIAS"[1].w + zCoord;\n");
-				else
-					WRITE(p, "zCoord = dot("I_ZBIAS"[0].xyzw, textemp.xyzw) + "I_ZBIAS"[1].w;\n");
-
-				// scale to make result from frac correct
-				WRITE(p, "zCoord = zCoord * (16777215.0f/16777216.0f);\n");
-				WRITE(p, "zCoord = frac(zCoord);\n");
-				WRITE(p, "zCoord = zCoord * (16777216.0f/16777215.0f);\n");
-			}
-			WRITE(p, "depth = zCoord;\n");
-		}
+			if (bpmem.ztex2.op == ZTEXTURE_ADD)
+				WRITE(p, "zCoord = dot("I_ZBIAS"[0].xyzw, textemp.xyzw) + "I_ZBIAS"[1].w + zCoord;\n");
+			else
+				WRITE(p, "zCoord = dot("I_ZBIAS"[0].xyzw, textemp.xyzw) + "I_ZBIAS"[1].w;\n");
 
-		if (dstAlphaMode == DSTALPHA_ALPHA_PASS)
-			WRITE(p, "  ocol0 = float4(prev.rgb, "I_ALPHA"[0].a);\n");
-		else
-		{
-			WriteFog(p);
-			WRITE(p, "  ocol0 = prev;\n");
+			// scale to make result from frac correct
+			WRITE(p, "zCoord = zCoord * (16777215.0f/16777216.0f);\n");
+			WRITE(p, "zCoord = frac(zCoord);\n");
+			WRITE(p, "zCoord = zCoord * (16777216.0f/16777215.0f);\n");
 		}
+		WRITE(p, "depth = zCoord;\n");
+	}
 
-		// On D3D11, use dual-source color blending to perform dst alpha in a
-		// single pass
-		if (dstAlphaMode == DSTALPHA_DUAL_SOURCE_BLEND)
-		{
-			// Colors will be blended against the alpha from ocol1...
-			WRITE(p, "  ocol1 = ocol0;\n");
-			// ...and the alpha from ocol0 will be written to the framebuffer.
-			WRITE(p, "  ocol0.a = "I_ALPHA"[0].a;\n");
-		}
+	if (dstAlphaMode == DSTALPHA_ALPHA_PASS)
+		WRITE(p, "  ocol0 = float4(prev.rgb, "I_ALPHA"[0].a);\n");
+	else
+	{
+		WriteFog(p);
+		WRITE(p, "  ocol0 = prev;\n");
 	}
+
+	// On D3D11, use dual-source color blending to perform dst alpha in a
+	// single pass
+	if (dstAlphaMode == DSTALPHA_DUAL_SOURCE_BLEND)
+	{
+		// Colors will be blended against the alpha from ocol1...
+		WRITE(p, "  ocol1 = ocol0;\n");
+		// ...and the alpha from ocol0 will be written to the framebuffer.
+		WRITE(p, "  ocol0.a = "I_ALPHA"[0].a;\n");
+	}
+	
 	WRITE(p, "}\n");
 	if (text[sizeof(text) - 1] != 0x7C)
 		PanicAlert("PixelShader generator - buffer too small, canary has been eaten!");
@@ -1157,19 +1152,13 @@ static int AlphaPreTest()
 }
 
 
-static bool WriteAlphaTest(char *&p, API_TYPE ApiType,DSTALPHA_MODE dstAlphaMode)
+static void WriteAlphaTest(char *&p, API_TYPE ApiType,DSTALPHA_MODE dstAlphaMode)
 {
 	static const char *alphaRef[2] =
 	{
 		I_ALPHA"[0].r",
 		I_ALPHA"[0].g"
-	};
-
-	int Pretest = AlphaPreTest();
-	if(Pretest >= 0)
-	{
-		return Pretest != 0;
-	}
+	};	
 
 	// using discard then return works the same in cg and dx9 but not in dx11
 	WRITE(p, "if(!( ");
@@ -1190,14 +1179,17 @@ static bool WriteAlphaTest(char *&p, API_TYPE ApiType,DSTALPHA_MODE dstAlphaMode
 		WRITE(p, "depth = 1.f;\n");
 
 	// HAXX: zcomploc is a way to control whether depth test is done before
-	// or after texturing and alpha test. PC GPU have no way to support this
-	// feature properly as of 2012: depth buffer and depth test are not
-	// programmable and the depth test is always done after texturing.
-	//
+	// or after texturing and alpha test. PC GPU does depth test before texturing ONLY if depth value is
+	// not updated during shader execution.
 	// We implement "depth test before texturing" by discarding the fragment
 	// when the alpha test fail. This is not a correct implementation because
-	// even if the depth test fails the fragment could be alpha blended, but
-	// we don't have a choice.
+	// even if the depth test fails the fragment could be alpha blended.
+	// this implemnetation is a trick to  keep speed.
+	// the correct, but slow, way to implement a correct zComploc is :
+	// 1 - if zcomplock is enebled make a first pass, with color channel write disabled updating only 
+	// depth channel.
+	// 2 - in the next pass disable depth chanel update, but proccess the color data normally
+	// this way is the only CORRECT way to emulate perfectly the zcomplock behaviour
 	if (!(bpmem.zcontrol.zcomploc && bpmem.zmode.updateenable))
 	{
 		WRITE(p, "discard;\n");
@@ -1206,7 +1198,7 @@ static bool WriteAlphaTest(char *&p, API_TYPE ApiType,DSTALPHA_MODE dstAlphaMode
 	}
 
 	WRITE(p, "}\n");
-	return true;
+	
 }
 
 static const char *tevFogFuncsTable[] =
-- 
cgit v1.2.3


From cb92805445856e372ccbd156b4c11740525d8e55 Mon Sep 17 00:00:00 2001
From: Timo Wiren <timo.wiren@gmail.com>
Date: Sun, 25 Mar 2012 13:01:26 +0300
Subject: Spaces to tabs. Removed a comment.

---
 Source/Core/VideoCommon/Src/OpenCL/OCLTextureDecoder.cpp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/OpenCL/OCLTextureDecoder.cpp b/Source/Core/VideoCommon/Src/OpenCL/OCLTextureDecoder.cpp
index 8aabf2a066..440e69efd6 100644
--- a/Source/Core/VideoCommon/Src/OpenCL/OCLTextureDecoder.cpp
+++ b/Source/Core/VideoCommon/Src/OpenCL/OCLTextureDecoder.cpp
@@ -114,7 +114,7 @@ void TexDecoder_OpenCL_Initialize()
 		else
 		{
 			binary_size = input.GetSize();
-			header = new char[HEADER_SIZE];	// TODO: memleak possible
+			header = new char[HEADER_SIZE];
 			binary = new char[binary_size];
 			input.ReadBytes(header, HEADER_SIZE);
 			input.ReadBytes(binary, binary_size);
-- 
cgit v1.2.3


From f30aebf8d7bac892d08ef5585c2e858cf6d3f50b Mon Sep 17 00:00:00 2001
From: skidau <skidau@gmail.com>
Date: Sun, 25 Mar 2012 21:35:57 +1100
Subject: Added a check for TMEM overflows while preloading textures.  Thanks
 to NeoBrainX for the tip.

---
 Source/Core/VideoCommon/Src/BPMemory.h    |  2 +-
 Source/Core/VideoCommon/Src/BPStructs.cpp | 18 ++++++++++--------
 2 files changed, 11 insertions(+), 9 deletions(-)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/BPMemory.h b/Source/Core/VideoCommon/Src/BPMemory.h
index 0bb2bf9013..de5c616f0f 100644
--- a/Source/Core/VideoCommon/Src/BPMemory.h
+++ b/Source/Core/VideoCommon/Src/BPMemory.h
@@ -62,7 +62,7 @@
 #define BPMEM_COPYFILTER1      0x54
 #define BPMEM_CLEARBBOX1       0x55 
 #define BPMEM_CLEARBBOX2       0x56
-#define BPMEM_UNKOWN_57        0x57
+#define BPMEM_UNKNOWN_57       0x57
 #define BPMEM_REVBITS          0x58
 #define BPMEM_SCISSOROFFSET    0x59
 #define BPMEM_PRELOAD_ADDR     0x60
diff --git a/Source/Core/VideoCommon/Src/BPStructs.cpp b/Source/Core/VideoCommon/Src/BPStructs.cpp
index 01b09e351b..d59523e117 100644
--- a/Source/Core/VideoCommon/Src/BPStructs.cpp
+++ b/Source/Core/VideoCommon/Src/BPStructs.cpp
@@ -463,8 +463,8 @@ void BPWritten(const BPCmd& bp)
 	case BPMEM_REVBITS: // Always set to 0x0F when GX_InitRevBits() is called.
 		break;
 
-	case BPMEM_UNKOWN_57: // Sunshine alternates this register between values 0x000 and 0xAAA
-		DEBUG_LOG(VIDEO, "Uknown BP Reg 0x57: %08x", bp.newvalue);
+	case BPMEM_UNKNOWN_57: // Sunshine alternates this register between values 0x000 and 0xAAA
+		DEBUG_LOG(VIDEO, "Unknown BP Reg 0x57: %08x", bp.newvalue);
 		break;
 
 	case BPMEM_PRELOAD_ADDR:
@@ -476,16 +476,18 @@ void BPWritten(const BPCmd& bp)
 		// if this is different from 0, manual TMEM management is used.
 		if (bp.newvalue != 0)
 		{
+			// NOTE(neobrain): Apparently tmemodd doesn't affect hardware behavior at all (libogc uses it just as a buffer and switches its contents with tmemeven whenever this is called)
 			BPS_TmemConfig& tmem_cfg = bpmem.tmem_config;
 			u8* ram_ptr = Memory::GetPointer(tmem_cfg.preload_addr << 5);
-			u32 tmem_addr = 0;
+			u32 tmem_addr = tmem_cfg.preload_tmem_even * TMEM_LINE_SIZE;
+			u32 size = tmem_cfg.preload_tile_info.count * 32;
 
-			if (bp.newvalue >> 16)
-				tmem_addr = tmem_cfg.preload_tmem_odd * TMEM_LINE_SIZE;
-			else
-				tmem_addr = tmem_cfg.preload_tmem_even * TMEM_LINE_SIZE;
+			// Check if the game has overflowed TMEM, and copy up to the limit.
+			// Paper Mario does this when entering the Great Boogly Tree (Chap 2)
+			// TODO: Does this wrap?
+			if ((tmem_addr + size) > TMEM_SIZE)
+				size = TMEM_SIZE - tmem_addr;
 
-			u32 size = tmem_cfg.preload_tile_info.count * 32;
 			memcpy(texMem + tmem_addr, ram_ptr, size);
 		}
 		break;
-- 
cgit v1.2.3


From 450dcc9d2cd874f6ab40d4a339afc29f55751319 Mon Sep 17 00:00:00 2001
From: rodolfoosvaldobogado <rodolfoosvaldobogado@gmail.com>
Date: Tue, 27 Mar 2012 15:44:58 -0300
Subject: As requested, this is my implementation of zcomploc using a
 multi-pass algorithm. My apologize to the others devs for committing in the
 main branch but is the only way to get this tested as soon as possible.
 please test for regressions, speed and for other issues fixed, as a example,
 the black color in water splash in super mario galaxy are fixed with this
 rev. please as soon as yo find a bug let me know.

---
 Source/Core/VideoCommon/Src/PixelShaderGen.cpp | 46 +++++++++++++++-----------
 Source/Core/VideoCommon/Src/PixelShaderGen.h   |  1 +
 Source/Core/VideoCommon/Src/RenderBase.h       | 13 ++++++--
 Source/Core/VideoCommon/Src/VideoConfig.cpp    |  4 +++
 Source/Core/VideoCommon/Src/VideoConfig.h      |  1 +
 5 files changed, 43 insertions(+), 22 deletions(-)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/PixelShaderGen.cpp b/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
index 76964f248c..7d8a3734da 100644
--- a/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
+++ b/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
@@ -109,7 +109,7 @@ void GetPixelShaderId(PIXELSHADERUID *uid, DSTALPHA_MODE dstAlphaMode, u32 compo
 	uid->values[0] |= bpmem.genMode.numtexgens << 4; // 4
 	uid->values[0] |= dstAlphaMode << 8; // 2
 
-	bool DepthTextureEnable = (bpmem.ztex2.op != ZTEXTURE_DISABLE && !bpmem.zcontrol.zcomploc && bpmem.zmode.testenable && bpmem.zmode.updateenable) || g_ActiveConfig.bEnablePerPixelDepth;
+	bool DepthTextureEnable = (bpmem.ztex2.op != ZTEXTURE_DISABLE && !bpmem.zcontrol.zcomploc && bpmem.zmode.testenable) || g_ActiveConfig.bEnablePerPixelDepth;
 
 	uid->values[0] |= DepthTextureEnable << 10; // 1
 
@@ -121,7 +121,7 @@ void GetPixelShaderId(PIXELSHADERUID *uid, DSTALPHA_MODE dstAlphaMode, u32 compo
 
 	uid->values[0] |= alphaPreTest << 16; // 2
 
-	if (alphaPreTest == 1 || (alphaPreTest && !DepthTextureEnable && dstAlphaMode == DSTALPHA_ALPHA_PASS))
+	if (alphaPreTest == 1 || (alphaPreTest == 2 && !DepthTextureEnable && dstAlphaMode == DSTALPHA_ALPHA_PASS) || dstAlphaMode == DSTALPHA_ZCOMPLOC)
 	{
 		// Courtesy of PreAlphaTest, we're done already ;)
 		// NOTE: The comment header of generated shaders depends on the value of bpmem.genmode.numindstages.. shouldnt really bother about that though.
@@ -177,6 +177,8 @@ void GetPixelShaderId(PIXELSHADERUID *uid, DSTALPHA_MODE dstAlphaMode, u32 compo
 			ptr[0] |= bpmem.zcontrol.zcomploc << 13; // 1
 			ptr[0] |= bpmem.zmode.testenable << 14; // 1
 			ptr[0] |= bpmem.zmode.updateenable << 15; // 1
+			ptr[0] |= g_ActiveConfig.bAcurateZcomploc << 16; // 1
+			
 		}
 	}
 
@@ -184,8 +186,8 @@ void GetPixelShaderId(PIXELSHADERUID *uid, DSTALPHA_MODE dstAlphaMode, u32 compo
 	{
 		if (bpmem.fog.c_proj_fsel.fsel != 0)
 		{
-			ptr[0] |= bpmem.fog.c_proj_fsel.proj << 16; // 1
-			ptr[0] |= bpmem.fogRange.Base.Enabled << 17; // 1
+			ptr[0] |= bpmem.fog.c_proj_fsel.proj << 17; // 1
+			ptr[0] |= bpmem.fogRange.Base.Enabled << 18; // 1
 		}
 	}
 
@@ -523,7 +525,7 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType
 				nIndirectStagesUsed |= 1 << bpmem.tevind[i].bt;
 		}
 	}
-	DepthTextureEnable = (bpmem.ztex2.op != ZTEXTURE_DISABLE && !bpmem.zcontrol.zcomploc && bpmem.zmode.testenable && bpmem.zmode.updateenable) || g_ActiveConfig.bEnablePerPixelDepth ;
+	DepthTextureEnable = (bpmem.ztex2.op != ZTEXTURE_DISABLE && !bpmem.zcontrol.zcomploc && bpmem.zmode.testenable) || g_ActiveConfig.bEnablePerPixelDepth ;
 	// Declare samplers
 
 	if(ApiType != API_D3D11)
@@ -619,9 +621,9 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType
 
 	char* pmainstart = p;
 	int Pretest = AlphaPreTest();
-	if(Pretest >= 0 && !DepthTextureEnable)
+	if(Pretest >= 0 || dstAlphaMode == DSTALPHA_ZCOMPLOC)
 	{
-		if (!Pretest)
+		if (!Pretest || dstAlphaMode == DSTALPHA_ZCOMPLOC)
 		{
 			// alpha test will always fail, so restart the shader and just make it an empty function		
 			WRITE(p, "ocol0 = 0;\n");
@@ -629,15 +631,18 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType
 				WRITE(p, "depth = 1.f;\n");
 			if(dstAlphaMode == DSTALPHA_DUAL_SOURCE_BLEND)
 					WRITE(p, "ocol1 = 0;\n");
-			WRITE(p, "discard;\n");
-			if(ApiType != API_D3D11)
-				WRITE(p, "return;\n");
+			if(dstAlphaMode != DSTALPHA_ZCOMPLOC)
+			{
+				WRITE(p, "discard;\n");
+				if(ApiType != API_D3D11)
+					WRITE(p, "return;\n");
+			}
 		}
-		else if (dstAlphaMode == DSTALPHA_ALPHA_PASS)
+		else if (dstAlphaMode == DSTALPHA_ALPHA_PASS && !DepthTextureEnable)
 		{
 			WRITE(p, "  ocol0 = "I_ALPHA"[0].aaaa;\n");
 		}
-		if(!Pretest || dstAlphaMode == DSTALPHA_ALPHA_PASS)
+		if(!Pretest || (dstAlphaMode == DSTALPHA_ALPHA_PASS && !DepthTextureEnable) || dstAlphaMode == DSTALPHA_ZCOMPLOC)
 		{
 			WRITE(p, "}\n");
 			return text;
@@ -744,7 +749,7 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType
 	if (DepthTextureEnable)
 	{
 		// use the texture input of the last texture stage (textemp), hopefully this has been read and is in correct format...
-		if (bpmem.ztex2.op != ZTEXTURE_DISABLE && !bpmem.zcontrol.zcomploc && bpmem.zmode.testenable && bpmem.zmode.updateenable)
+		if (bpmem.ztex2.op != ZTEXTURE_DISABLE && !bpmem.zcontrol.zcomploc && bpmem.zmode.testenable)
 		{
 			if (bpmem.ztex2.op == ZTEXTURE_ADD)
 				WRITE(p, "zCoord = dot("I_ZBIAS"[0].xyzw, textemp.xyzw) + "I_ZBIAS"[1].w + zCoord;\n");
@@ -1180,17 +1185,18 @@ static void WriteAlphaTest(char *&p, API_TYPE ApiType,DSTALPHA_MODE dstAlphaMode
 
 	// HAXX: zcomploc is a way to control whether depth test is done before
 	// or after texturing and alpha test. PC GPU does depth test before texturing ONLY if depth value is
-	// not updated during shader execution.
+	// not updated during shader execution. but GC hardware acts different:
+	// if zcomlock is enabled, depth value is updated even if alpha test fails
 	// We implement "depth test before texturing" by discarding the fragment
 	// when the alpha test fail. This is not a correct implementation because
-	// even if the depth test fails the fragment could be alpha blended.
-	// this implemnetation is a trick to  keep speed.
-	// the correct, but slow, way to implement a correct zComploc is :
+	// even if the depth test fails the fragment could be alpha blended.	
+	// the more correct, but slow, way is:
 	// 1 - if zcomplock is enebled make a first pass, with color channel write disabled updating only 
 	// depth channel.
-	// 2 - in the next pass disable depth chanel update, but proccess the color data normally
-	// this way is the only CORRECT way to emulate perfectly the zcomplock behaviour
-	if (!(bpmem.zcontrol.zcomploc && bpmem.zmode.updateenable))
+	// 2 - in the next pass disable depth chanel update, but proccess the color data normally an update only the pixels 
+	// updated by the depth pass
+	
+	if (!(bpmem.zcontrol.zcomploc && bpmem.zmode.updateenable) || g_ActiveConfig.bAcurateZcomploc)
 	{
 		WRITE(p, "discard;\n");
 		if (ApiType != API_D3D11)
diff --git a/Source/Core/VideoCommon/Src/PixelShaderGen.h b/Source/Core/VideoCommon/Src/PixelShaderGen.h
index 31242a916e..99b0c60d45 100644
--- a/Source/Core/VideoCommon/Src/PixelShaderGen.h
+++ b/Source/Core/VideoCommon/Src/PixelShaderGen.h
@@ -109,6 +109,7 @@ typedef _PIXELSHADERUID<true> PIXELSHADERUIDSAFE;
 enum DSTALPHA_MODE
 {
 	DSTALPHA_NONE, // Render normally, without destination alpha
+	DSTALPHA_ZCOMPLOC,
 	DSTALPHA_ALPHA_PASS, // Render normally first, then render again for alpha
 	DSTALPHA_DUAL_SOURCE_BLEND // Use dual-source blending
 };
diff --git a/Source/Core/VideoCommon/Src/RenderBase.h b/Source/Core/VideoCommon/Src/RenderBase.h
index e8d4c55a20..80d710efde 100644
--- a/Source/Core/VideoCommon/Src/RenderBase.h
+++ b/Source/Core/VideoCommon/Src/RenderBase.h
@@ -43,12 +43,21 @@ extern int OSDChoice, OSDTime;
 
 extern bool bLastFrameDumped;
 
+
+enum RenderStateMode
+	{
+		RSM_None,
+		RSM_Multipass,
+		RSM_Zcomploc,
+		RSM_UseDstAlpha
+	};
 // Renderer really isn't a very good name for this class - it's more like "Misc".
 // The long term goal is to get rid of this class and replace it with others that make
 // more sense.
 class Renderer
 {
 public:
+	
 	Renderer();
 	virtual ~Renderer();
 
@@ -63,8 +72,8 @@ public:
 	virtual void SetSamplerState(int stage,int texindex) = 0;
 	virtual void SetInterlacingMode() = 0;
 
-	virtual void ApplyState(bool bUseDstAlpha) = 0;
-	virtual void RestoreState() = 0;
+	virtual void ApplyState(RenderStateMode mode) = 0;
+	virtual void RestoreState(RenderStateMode mode) = 0;
 
 	// Ideal internal resolution - determined by display resolution (automatic scaling) and/or a multiple of the native EFB resolution
 	static int GetTargetWidth() { return s_target_width; }
diff --git a/Source/Core/VideoCommon/Src/VideoConfig.cpp b/Source/Core/VideoCommon/Src/VideoConfig.cpp
index 5bb6e4dd99..b04aa0d9cf 100644
--- a/Source/Core/VideoCommon/Src/VideoConfig.cpp
+++ b/Source/Core/VideoCommon/Src/VideoConfig.cpp
@@ -76,6 +76,7 @@ void VideoConfig::Load(const char *ini_file)
 	iniFile.Get("Settings", "AnaglyphFocalAngle", &iAnaglyphFocalAngle, 0);
 	iniFile.Get("Settings", "EnablePixelLighting", &bEnablePixelLighting, 0);
 	iniFile.Get("Settings", "EnablePerPixelDepth", &bEnablePerPixelDepth, 0);
+	iniFile.Get("Settings", "AcurateZcomploc", &bAcurateZcomploc, 0);
 	
 	iniFile.Get("Settings", "MSAA", &iMultisampleMode, 0);
 	iniFile.Get("Settings", "EFBScale", &iEFBScale, 2); // native
@@ -138,6 +139,7 @@ void VideoConfig::GameIniLoad(const char *ini_file)
 	iniFile.GetIfExists("Video_Settings", "AnaglyphFocalAngle", &iAnaglyphFocalAngle);
 	iniFile.GetIfExists("Video_Settings", "EnablePixelLighting", &bEnablePixelLighting);
 	iniFile.GetIfExists("Video_Settings", "EnablePerPixelDepth", &bEnablePerPixelDepth);
+	iniFile.GetIfExists("Video_Settings", "AcurateZcomploc", &bAcurateZcomploc);
 	iniFile.GetIfExists("Video_Settings", "MSAA", &iMultisampleMode);
 	iniFile.GetIfExists("Video_Settings", "EFBScale", &iEFBScale); // integral
 	iniFile.GetIfExists("Video_Settings", "DstAlphaPass", &bDstAlphaPass);
@@ -209,6 +211,7 @@ void VideoConfig::Save(const char *ini_file)
 	iniFile.Set("Settings", "AnaglyphFocalAngle", iAnaglyphFocalAngle);
 	iniFile.Set("Settings", "EnablePixelLighting", bEnablePixelLighting);
 	iniFile.Set("Settings", "EnablePerPixelDepth", bEnablePerPixelDepth);
+	iniFile.Set("Settings", "AcurateZcomploc", bAcurateZcomploc);
 	
 
 	iniFile.Set("Settings", "ShowEFBCopyRegions", bShowEFBCopyRegions);
@@ -278,6 +281,7 @@ void VideoConfig::GameIniSave(const char* default_ini, const char* game_ini)
 	SET_IF_DIFFERS("Video_Settings", "AnaglyphFocalAngle", iAnaglyphFocalAngle);
 	SET_IF_DIFFERS("Video_Settings", "EnablePixelLighting", bEnablePixelLighting);
 	SET_IF_DIFFERS("Video_Settings", "EnablePerPixelDepth", bEnablePerPixelDepth);
+	SET_IF_DIFFERS("Video_Settings", "AcurateZcomploc", bAcurateZcomploc);
 	SET_IF_DIFFERS("Video_Settings", "MSAA", iMultisampleMode);
 	SET_IF_DIFFERS("Video_Settings", "EFBScale", iEFBScale); // integral
 	SET_IF_DIFFERS("Video_Settings", "DstAlphaPass", bDstAlphaPass);
diff --git a/Source/Core/VideoCommon/Src/VideoConfig.h b/Source/Core/VideoCommon/Src/VideoConfig.h
index 7c846676bb..e330de54ad 100644
--- a/Source/Core/VideoCommon/Src/VideoConfig.h
+++ b/Source/Core/VideoCommon/Src/VideoConfig.h
@@ -137,6 +137,7 @@ struct VideoConfig
 	bool bUseBBox;
 	bool bEnablePixelLighting;
 	bool bEnablePerPixelDepth;
+	bool bAcurateZcomploc;
 
 	int iLog; // CONF_ bits
 	int iSaveTargetId;
-- 
cgit v1.2.3


From 402006a83a89bc0f790934096e915f0e8c23f808 Mon Sep 17 00:00:00 2001
From: rodolfoosvaldobogado <rodolfoosvaldobogado@gmail.com>
Date: Thu, 29 Mar 2012 18:26:58 -0300
Subject: more fixes to zcomplock and opengl implementation

---
 Source/Core/VideoCommon/Src/PixelShaderGen.cpp | 38 ++++++++++++--------------
 Source/Core/VideoCommon/Src/VideoConfig.cpp    |  8 +++---
 Source/Core/VideoCommon/Src/VideoConfig.h      |  2 +-
 3 files changed, 23 insertions(+), 25 deletions(-)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/PixelShaderGen.cpp b/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
index 7d8a3734da..5340b84f1a 100644
--- a/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
+++ b/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
@@ -177,7 +177,7 @@ void GetPixelShaderId(PIXELSHADERUID *uid, DSTALPHA_MODE dstAlphaMode, u32 compo
 			ptr[0] |= bpmem.zcontrol.zcomploc << 13; // 1
 			ptr[0] |= bpmem.zmode.testenable << 14; // 1
 			ptr[0] |= bpmem.zmode.updateenable << 15; // 1
-			ptr[0] |= g_ActiveConfig.bAcurateZcomploc << 16; // 1
+			ptr[0] |= g_ActiveConfig.bEnableFastZcomploc << 16; // 1
 			
 		}
 	}
@@ -212,7 +212,8 @@ void GetSafePixelShaderId(PIXELSHADERUIDSAFE *uid, DSTALPHA_MODE dstAlphaMode, u
 	*ptr++ = bpmem.zmode.hex; // 4
 	*ptr++ = g_ActiveConfig.bEnablePerPixelDepth; // 5
 	*ptr++ = g_ActiveConfig.bEnablePixelLighting && g_ActiveConfig.backend_info.bSupportsPixelLighting; // 6
-	*ptr++ = xfregs.numTexGen.hex; // 7
+	*ptr++ = g_ActiveConfig.bEnableFastZcomploc ; // 7
+	*ptr++ = xfregs.numTexGen.hex; // 8
 
 	if (g_ActiveConfig.bEnablePixelLighting && g_ActiveConfig.backend_info.bSupportsPixelLighting)
 	{
@@ -224,28 +225,28 @@ void GetSafePixelShaderId(PIXELSHADERUIDSAFE *uid, DSTALPHA_MODE dstAlphaMode, u
 	}
 
 	for (unsigned int i = 0; i < 8; ++i)
-		*ptr++ = xfregs.texMtxInfo[i].hex; // 8-15
+		*ptr++ = xfregs.texMtxInfo[i].hex; // 9-16
 
 	for (unsigned int i = 0; i < 16; ++i)
-		*ptr++ = bpmem.tevind[i].hex; // 16-31
+		*ptr++ = bpmem.tevind[i].hex; // 17-32
 
-	*ptr++ = bpmem.tevindref.hex; // 32
+	*ptr++ = bpmem.tevindref.hex; // 33
 
 	for (int i = 0; i < bpmem.genMode.numtevstages+1; ++i) // up to 16 times
 	{
-		*ptr++ = bpmem.combiners[i].colorC.hex; // 33+5*i
-		*ptr++ = bpmem.combiners[i].alphaC.hex; // 34+5*i
-		*ptr++ = bpmem.tevind[i].hex; // 35+5*i
-		*ptr++ = bpmem.tevksel[i/2].hex; // 36+5*i
-		*ptr++ = bpmem.tevorders[i/2].hex; // 37+5*i
+		*ptr++ = bpmem.combiners[i].colorC.hex; // 34+5*i
+		*ptr++ = bpmem.combiners[i].alphaC.hex; // 35+5*i
+		*ptr++ = bpmem.tevind[i].hex; // 36+5*i
+		*ptr++ = bpmem.tevksel[i/2].hex; // 37+5*i
+		*ptr++ = bpmem.tevorders[i/2].hex; // 38+5*i
 	}
 
-	ptr = &uid->values[113];
+	ptr = &uid->values[114];
 
-	*ptr++ = bpmem.alphaFunc.hex; // 113
+	*ptr++ = bpmem.alphaFunc.hex; // 114
 
-	*ptr++ = bpmem.fog.c_proj_fsel.hex; // 114
-	*ptr++ = bpmem.fogRange.Base.hex; // 115
+	*ptr++ = bpmem.fog.c_proj_fsel.hex; // 115
+	*ptr++ = bpmem.fogRange.Base.hex; // 116
 
 	_assert_((ptr - uid->values) == uid->GetNumValues());
 }
@@ -1191,12 +1192,9 @@ static void WriteAlphaTest(char *&p, API_TYPE ApiType,DSTALPHA_MODE dstAlphaMode
 	// when the alpha test fail. This is not a correct implementation because
 	// even if the depth test fails the fragment could be alpha blended.	
 	// the more correct, but slow, way is:
-	// 1 - if zcomplock is enebled make a first pass, with color channel write disabled updating only 
-	// depth channel.
-	// 2 - in the next pass disable depth chanel update, but proccess the color data normally an update only the pixels 
-	// updated by the depth pass
-	
-	if (!(bpmem.zcontrol.zcomploc && bpmem.zmode.updateenable) || g_ActiveConfig.bAcurateZcomploc)
+	// 1 - if zcomplock is enabled make a last pass, with color channel write disabled updating only 
+	// depth channel.	
+	if (!(g_ActiveConfig.bEnableFastZcomploc && bpmem.zcontrol.zcomploc && bpmem.zmode.updateenable))
 	{
 		WRITE(p, "discard;\n");
 		if (ApiType != API_D3D11)
diff --git a/Source/Core/VideoCommon/Src/VideoConfig.cpp b/Source/Core/VideoCommon/Src/VideoConfig.cpp
index b04aa0d9cf..5eb293251d 100644
--- a/Source/Core/VideoCommon/Src/VideoConfig.cpp
+++ b/Source/Core/VideoCommon/Src/VideoConfig.cpp
@@ -76,7 +76,7 @@ void VideoConfig::Load(const char *ini_file)
 	iniFile.Get("Settings", "AnaglyphFocalAngle", &iAnaglyphFocalAngle, 0);
 	iniFile.Get("Settings", "EnablePixelLighting", &bEnablePixelLighting, 0);
 	iniFile.Get("Settings", "EnablePerPixelDepth", &bEnablePerPixelDepth, 0);
-	iniFile.Get("Settings", "AcurateZcomploc", &bAcurateZcomploc, 0);
+	iniFile.Get("Settings", "EnableFastZcomploc", &bEnableFastZcomploc, 0);
 	
 	iniFile.Get("Settings", "MSAA", &iMultisampleMode, 0);
 	iniFile.Get("Settings", "EFBScale", &iEFBScale, 2); // native
@@ -139,7 +139,7 @@ void VideoConfig::GameIniLoad(const char *ini_file)
 	iniFile.GetIfExists("Video_Settings", "AnaglyphFocalAngle", &iAnaglyphFocalAngle);
 	iniFile.GetIfExists("Video_Settings", "EnablePixelLighting", &bEnablePixelLighting);
 	iniFile.GetIfExists("Video_Settings", "EnablePerPixelDepth", &bEnablePerPixelDepth);
-	iniFile.GetIfExists("Video_Settings", "AcurateZcomploc", &bAcurateZcomploc);
+	iniFile.GetIfExists("Video_Settings", "EnableFastZcomploc", &bEnableFastZcomploc);
 	iniFile.GetIfExists("Video_Settings", "MSAA", &iMultisampleMode);
 	iniFile.GetIfExists("Video_Settings", "EFBScale", &iEFBScale); // integral
 	iniFile.GetIfExists("Video_Settings", "DstAlphaPass", &bDstAlphaPass);
@@ -211,7 +211,7 @@ void VideoConfig::Save(const char *ini_file)
 	iniFile.Set("Settings", "AnaglyphFocalAngle", iAnaglyphFocalAngle);
 	iniFile.Set("Settings", "EnablePixelLighting", bEnablePixelLighting);
 	iniFile.Set("Settings", "EnablePerPixelDepth", bEnablePerPixelDepth);
-	iniFile.Set("Settings", "AcurateZcomploc", bAcurateZcomploc);
+	iniFile.Set("Settings", "EnableFastZcomploc", bEnableFastZcomploc);
 	
 
 	iniFile.Set("Settings", "ShowEFBCopyRegions", bShowEFBCopyRegions);
@@ -281,7 +281,7 @@ void VideoConfig::GameIniSave(const char* default_ini, const char* game_ini)
 	SET_IF_DIFFERS("Video_Settings", "AnaglyphFocalAngle", iAnaglyphFocalAngle);
 	SET_IF_DIFFERS("Video_Settings", "EnablePixelLighting", bEnablePixelLighting);
 	SET_IF_DIFFERS("Video_Settings", "EnablePerPixelDepth", bEnablePerPixelDepth);
-	SET_IF_DIFFERS("Video_Settings", "AcurateZcomploc", bAcurateZcomploc);
+	SET_IF_DIFFERS("Video_Settings", "EnableFastZcomploc", bEnableFastZcomploc);
 	SET_IF_DIFFERS("Video_Settings", "MSAA", iMultisampleMode);
 	SET_IF_DIFFERS("Video_Settings", "EFBScale", iEFBScale); // integral
 	SET_IF_DIFFERS("Video_Settings", "DstAlphaPass", bDstAlphaPass);
diff --git a/Source/Core/VideoCommon/Src/VideoConfig.h b/Source/Core/VideoCommon/Src/VideoConfig.h
index e330de54ad..d6dc3cefe0 100644
--- a/Source/Core/VideoCommon/Src/VideoConfig.h
+++ b/Source/Core/VideoCommon/Src/VideoConfig.h
@@ -137,7 +137,7 @@ struct VideoConfig
 	bool bUseBBox;
 	bool bEnablePixelLighting;
 	bool bEnablePerPixelDepth;
-	bool bAcurateZcomploc;
+	bool bEnableFastZcomploc;
 
 	int iLog; // CONF_ bits
 	int iSaveTargetId;
-- 
cgit v1.2.3


From 460610ea0e0d5083e261ed9d137048f2a5fc7e59 Mon Sep 17 00:00:00 2001
From: NeoBrainX <NeoBrainX@gmail.com>
Date: Fri, 30 Mar 2012 01:56:24 +0200
Subject: Revert Rodolfo's recent zcomploc commits until they actually work
 correctly.

This reverts commit 402006a83a89bc0f790934096e915f0e8c23f808.
This reverts commit 48d8d71391fc2cc40d85732680eb6421164bf87c.
This reverts commit 450dcc9d2cd874f6ab40d4a339afc29f55751319.
---
 Source/Core/VideoCommon/Src/PixelShaderGen.cpp | 74 ++++++++++++--------------
 Source/Core/VideoCommon/Src/PixelShaderGen.h   |  1 -
 Source/Core/VideoCommon/Src/RenderBase.h       | 13 +----
 Source/Core/VideoCommon/Src/VideoConfig.cpp    |  4 --
 Source/Core/VideoCommon/Src/VideoConfig.h      |  1 -
 5 files changed, 37 insertions(+), 56 deletions(-)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/PixelShaderGen.cpp b/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
index 5340b84f1a..76964f248c 100644
--- a/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
+++ b/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
@@ -109,7 +109,7 @@ void GetPixelShaderId(PIXELSHADERUID *uid, DSTALPHA_MODE dstAlphaMode, u32 compo
 	uid->values[0] |= bpmem.genMode.numtexgens << 4; // 4
 	uid->values[0] |= dstAlphaMode << 8; // 2
 
-	bool DepthTextureEnable = (bpmem.ztex2.op != ZTEXTURE_DISABLE && !bpmem.zcontrol.zcomploc && bpmem.zmode.testenable) || g_ActiveConfig.bEnablePerPixelDepth;
+	bool DepthTextureEnable = (bpmem.ztex2.op != ZTEXTURE_DISABLE && !bpmem.zcontrol.zcomploc && bpmem.zmode.testenable && bpmem.zmode.updateenable) || g_ActiveConfig.bEnablePerPixelDepth;
 
 	uid->values[0] |= DepthTextureEnable << 10; // 1
 
@@ -121,7 +121,7 @@ void GetPixelShaderId(PIXELSHADERUID *uid, DSTALPHA_MODE dstAlphaMode, u32 compo
 
 	uid->values[0] |= alphaPreTest << 16; // 2
 
-	if (alphaPreTest == 1 || (alphaPreTest == 2 && !DepthTextureEnable && dstAlphaMode == DSTALPHA_ALPHA_PASS) || dstAlphaMode == DSTALPHA_ZCOMPLOC)
+	if (alphaPreTest == 1 || (alphaPreTest && !DepthTextureEnable && dstAlphaMode == DSTALPHA_ALPHA_PASS))
 	{
 		// Courtesy of PreAlphaTest, we're done already ;)
 		// NOTE: The comment header of generated shaders depends on the value of bpmem.genmode.numindstages.. shouldnt really bother about that though.
@@ -177,8 +177,6 @@ void GetPixelShaderId(PIXELSHADERUID *uid, DSTALPHA_MODE dstAlphaMode, u32 compo
 			ptr[0] |= bpmem.zcontrol.zcomploc << 13; // 1
 			ptr[0] |= bpmem.zmode.testenable << 14; // 1
 			ptr[0] |= bpmem.zmode.updateenable << 15; // 1
-			ptr[0] |= g_ActiveConfig.bEnableFastZcomploc << 16; // 1
-			
 		}
 	}
 
@@ -186,8 +184,8 @@ void GetPixelShaderId(PIXELSHADERUID *uid, DSTALPHA_MODE dstAlphaMode, u32 compo
 	{
 		if (bpmem.fog.c_proj_fsel.fsel != 0)
 		{
-			ptr[0] |= bpmem.fog.c_proj_fsel.proj << 17; // 1
-			ptr[0] |= bpmem.fogRange.Base.Enabled << 18; // 1
+			ptr[0] |= bpmem.fog.c_proj_fsel.proj << 16; // 1
+			ptr[0] |= bpmem.fogRange.Base.Enabled << 17; // 1
 		}
 	}
 
@@ -212,8 +210,7 @@ void GetSafePixelShaderId(PIXELSHADERUIDSAFE *uid, DSTALPHA_MODE dstAlphaMode, u
 	*ptr++ = bpmem.zmode.hex; // 4
 	*ptr++ = g_ActiveConfig.bEnablePerPixelDepth; // 5
 	*ptr++ = g_ActiveConfig.bEnablePixelLighting && g_ActiveConfig.backend_info.bSupportsPixelLighting; // 6
-	*ptr++ = g_ActiveConfig.bEnableFastZcomploc ; // 7
-	*ptr++ = xfregs.numTexGen.hex; // 8
+	*ptr++ = xfregs.numTexGen.hex; // 7
 
 	if (g_ActiveConfig.bEnablePixelLighting && g_ActiveConfig.backend_info.bSupportsPixelLighting)
 	{
@@ -225,28 +222,28 @@ void GetSafePixelShaderId(PIXELSHADERUIDSAFE *uid, DSTALPHA_MODE dstAlphaMode, u
 	}
 
 	for (unsigned int i = 0; i < 8; ++i)
-		*ptr++ = xfregs.texMtxInfo[i].hex; // 9-16
+		*ptr++ = xfregs.texMtxInfo[i].hex; // 8-15
 
 	for (unsigned int i = 0; i < 16; ++i)
-		*ptr++ = bpmem.tevind[i].hex; // 17-32
+		*ptr++ = bpmem.tevind[i].hex; // 16-31
 
-	*ptr++ = bpmem.tevindref.hex; // 33
+	*ptr++ = bpmem.tevindref.hex; // 32
 
 	for (int i = 0; i < bpmem.genMode.numtevstages+1; ++i) // up to 16 times
 	{
-		*ptr++ = bpmem.combiners[i].colorC.hex; // 34+5*i
-		*ptr++ = bpmem.combiners[i].alphaC.hex; // 35+5*i
-		*ptr++ = bpmem.tevind[i].hex; // 36+5*i
-		*ptr++ = bpmem.tevksel[i/2].hex; // 37+5*i
-		*ptr++ = bpmem.tevorders[i/2].hex; // 38+5*i
+		*ptr++ = bpmem.combiners[i].colorC.hex; // 33+5*i
+		*ptr++ = bpmem.combiners[i].alphaC.hex; // 34+5*i
+		*ptr++ = bpmem.tevind[i].hex; // 35+5*i
+		*ptr++ = bpmem.tevksel[i/2].hex; // 36+5*i
+		*ptr++ = bpmem.tevorders[i/2].hex; // 37+5*i
 	}
 
-	ptr = &uid->values[114];
+	ptr = &uid->values[113];
 
-	*ptr++ = bpmem.alphaFunc.hex; // 114
+	*ptr++ = bpmem.alphaFunc.hex; // 113
 
-	*ptr++ = bpmem.fog.c_proj_fsel.hex; // 115
-	*ptr++ = bpmem.fogRange.Base.hex; // 116
+	*ptr++ = bpmem.fog.c_proj_fsel.hex; // 114
+	*ptr++ = bpmem.fogRange.Base.hex; // 115
 
 	_assert_((ptr - uid->values) == uid->GetNumValues());
 }
@@ -526,7 +523,7 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType
 				nIndirectStagesUsed |= 1 << bpmem.tevind[i].bt;
 		}
 	}
-	DepthTextureEnable = (bpmem.ztex2.op != ZTEXTURE_DISABLE && !bpmem.zcontrol.zcomploc && bpmem.zmode.testenable) || g_ActiveConfig.bEnablePerPixelDepth ;
+	DepthTextureEnable = (bpmem.ztex2.op != ZTEXTURE_DISABLE && !bpmem.zcontrol.zcomploc && bpmem.zmode.testenable && bpmem.zmode.updateenable) || g_ActiveConfig.bEnablePerPixelDepth ;
 	// Declare samplers
 
 	if(ApiType != API_D3D11)
@@ -622,9 +619,9 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType
 
 	char* pmainstart = p;
 	int Pretest = AlphaPreTest();
-	if(Pretest >= 0 || dstAlphaMode == DSTALPHA_ZCOMPLOC)
+	if(Pretest >= 0 && !DepthTextureEnable)
 	{
-		if (!Pretest || dstAlphaMode == DSTALPHA_ZCOMPLOC)
+		if (!Pretest)
 		{
 			// alpha test will always fail, so restart the shader and just make it an empty function		
 			WRITE(p, "ocol0 = 0;\n");
@@ -632,18 +629,15 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType
 				WRITE(p, "depth = 1.f;\n");
 			if(dstAlphaMode == DSTALPHA_DUAL_SOURCE_BLEND)
 					WRITE(p, "ocol1 = 0;\n");
-			if(dstAlphaMode != DSTALPHA_ZCOMPLOC)
-			{
-				WRITE(p, "discard;\n");
-				if(ApiType != API_D3D11)
-					WRITE(p, "return;\n");
-			}
+			WRITE(p, "discard;\n");
+			if(ApiType != API_D3D11)
+				WRITE(p, "return;\n");
 		}
-		else if (dstAlphaMode == DSTALPHA_ALPHA_PASS && !DepthTextureEnable)
+		else if (dstAlphaMode == DSTALPHA_ALPHA_PASS)
 		{
 			WRITE(p, "  ocol0 = "I_ALPHA"[0].aaaa;\n");
 		}
-		if(!Pretest || (dstAlphaMode == DSTALPHA_ALPHA_PASS && !DepthTextureEnable) || dstAlphaMode == DSTALPHA_ZCOMPLOC)
+		if(!Pretest || dstAlphaMode == DSTALPHA_ALPHA_PASS)
 		{
 			WRITE(p, "}\n");
 			return text;
@@ -750,7 +744,7 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType
 	if (DepthTextureEnable)
 	{
 		// use the texture input of the last texture stage (textemp), hopefully this has been read and is in correct format...
-		if (bpmem.ztex2.op != ZTEXTURE_DISABLE && !bpmem.zcontrol.zcomploc && bpmem.zmode.testenable)
+		if (bpmem.ztex2.op != ZTEXTURE_DISABLE && !bpmem.zcontrol.zcomploc && bpmem.zmode.testenable && bpmem.zmode.updateenable)
 		{
 			if (bpmem.ztex2.op == ZTEXTURE_ADD)
 				WRITE(p, "zCoord = dot("I_ZBIAS"[0].xyzw, textemp.xyzw) + "I_ZBIAS"[1].w + zCoord;\n");
@@ -1186,15 +1180,17 @@ static void WriteAlphaTest(char *&p, API_TYPE ApiType,DSTALPHA_MODE dstAlphaMode
 
 	// HAXX: zcomploc is a way to control whether depth test is done before
 	// or after texturing and alpha test. PC GPU does depth test before texturing ONLY if depth value is
-	// not updated during shader execution. but GC hardware acts different:
-	// if zcomlock is enabled, depth value is updated even if alpha test fails
+	// not updated during shader execution.
 	// We implement "depth test before texturing" by discarding the fragment
 	// when the alpha test fail. This is not a correct implementation because
-	// even if the depth test fails the fragment could be alpha blended.	
-	// the more correct, but slow, way is:
-	// 1 - if zcomplock is enabled make a last pass, with color channel write disabled updating only 
-	// depth channel.	
-	if (!(g_ActiveConfig.bEnableFastZcomploc && bpmem.zcontrol.zcomploc && bpmem.zmode.updateenable))
+	// even if the depth test fails the fragment could be alpha blended.
+	// this implemnetation is a trick to  keep speed.
+	// the correct, but slow, way to implement a correct zComploc is :
+	// 1 - if zcomplock is enebled make a first pass, with color channel write disabled updating only 
+	// depth channel.
+	// 2 - in the next pass disable depth chanel update, but proccess the color data normally
+	// this way is the only CORRECT way to emulate perfectly the zcomplock behaviour
+	if (!(bpmem.zcontrol.zcomploc && bpmem.zmode.updateenable))
 	{
 		WRITE(p, "discard;\n");
 		if (ApiType != API_D3D11)
diff --git a/Source/Core/VideoCommon/Src/PixelShaderGen.h b/Source/Core/VideoCommon/Src/PixelShaderGen.h
index 99b0c60d45..31242a916e 100644
--- a/Source/Core/VideoCommon/Src/PixelShaderGen.h
+++ b/Source/Core/VideoCommon/Src/PixelShaderGen.h
@@ -109,7 +109,6 @@ typedef _PIXELSHADERUID<true> PIXELSHADERUIDSAFE;
 enum DSTALPHA_MODE
 {
 	DSTALPHA_NONE, // Render normally, without destination alpha
-	DSTALPHA_ZCOMPLOC,
 	DSTALPHA_ALPHA_PASS, // Render normally first, then render again for alpha
 	DSTALPHA_DUAL_SOURCE_BLEND // Use dual-source blending
 };
diff --git a/Source/Core/VideoCommon/Src/RenderBase.h b/Source/Core/VideoCommon/Src/RenderBase.h
index 80d710efde..e8d4c55a20 100644
--- a/Source/Core/VideoCommon/Src/RenderBase.h
+++ b/Source/Core/VideoCommon/Src/RenderBase.h
@@ -43,21 +43,12 @@ extern int OSDChoice, OSDTime;
 
 extern bool bLastFrameDumped;
 
-
-enum RenderStateMode
-	{
-		RSM_None,
-		RSM_Multipass,
-		RSM_Zcomploc,
-		RSM_UseDstAlpha
-	};
 // Renderer really isn't a very good name for this class - it's more like "Misc".
 // The long term goal is to get rid of this class and replace it with others that make
 // more sense.
 class Renderer
 {
 public:
-	
 	Renderer();
 	virtual ~Renderer();
 
@@ -72,8 +63,8 @@ public:
 	virtual void SetSamplerState(int stage,int texindex) = 0;
 	virtual void SetInterlacingMode() = 0;
 
-	virtual void ApplyState(RenderStateMode mode) = 0;
-	virtual void RestoreState(RenderStateMode mode) = 0;
+	virtual void ApplyState(bool bUseDstAlpha) = 0;
+	virtual void RestoreState() = 0;
 
 	// Ideal internal resolution - determined by display resolution (automatic scaling) and/or a multiple of the native EFB resolution
 	static int GetTargetWidth() { return s_target_width; }
diff --git a/Source/Core/VideoCommon/Src/VideoConfig.cpp b/Source/Core/VideoCommon/Src/VideoConfig.cpp
index 5eb293251d..5bb6e4dd99 100644
--- a/Source/Core/VideoCommon/Src/VideoConfig.cpp
+++ b/Source/Core/VideoCommon/Src/VideoConfig.cpp
@@ -76,7 +76,6 @@ void VideoConfig::Load(const char *ini_file)
 	iniFile.Get("Settings", "AnaglyphFocalAngle", &iAnaglyphFocalAngle, 0);
 	iniFile.Get("Settings", "EnablePixelLighting", &bEnablePixelLighting, 0);
 	iniFile.Get("Settings", "EnablePerPixelDepth", &bEnablePerPixelDepth, 0);
-	iniFile.Get("Settings", "EnableFastZcomploc", &bEnableFastZcomploc, 0);
 	
 	iniFile.Get("Settings", "MSAA", &iMultisampleMode, 0);
 	iniFile.Get("Settings", "EFBScale", &iEFBScale, 2); // native
@@ -139,7 +138,6 @@ void VideoConfig::GameIniLoad(const char *ini_file)
 	iniFile.GetIfExists("Video_Settings", "AnaglyphFocalAngle", &iAnaglyphFocalAngle);
 	iniFile.GetIfExists("Video_Settings", "EnablePixelLighting", &bEnablePixelLighting);
 	iniFile.GetIfExists("Video_Settings", "EnablePerPixelDepth", &bEnablePerPixelDepth);
-	iniFile.GetIfExists("Video_Settings", "EnableFastZcomploc", &bEnableFastZcomploc);
 	iniFile.GetIfExists("Video_Settings", "MSAA", &iMultisampleMode);
 	iniFile.GetIfExists("Video_Settings", "EFBScale", &iEFBScale); // integral
 	iniFile.GetIfExists("Video_Settings", "DstAlphaPass", &bDstAlphaPass);
@@ -211,7 +209,6 @@ void VideoConfig::Save(const char *ini_file)
 	iniFile.Set("Settings", "AnaglyphFocalAngle", iAnaglyphFocalAngle);
 	iniFile.Set("Settings", "EnablePixelLighting", bEnablePixelLighting);
 	iniFile.Set("Settings", "EnablePerPixelDepth", bEnablePerPixelDepth);
-	iniFile.Set("Settings", "EnableFastZcomploc", bEnableFastZcomploc);
 	
 
 	iniFile.Set("Settings", "ShowEFBCopyRegions", bShowEFBCopyRegions);
@@ -281,7 +278,6 @@ void VideoConfig::GameIniSave(const char* default_ini, const char* game_ini)
 	SET_IF_DIFFERS("Video_Settings", "AnaglyphFocalAngle", iAnaglyphFocalAngle);
 	SET_IF_DIFFERS("Video_Settings", "EnablePixelLighting", bEnablePixelLighting);
 	SET_IF_DIFFERS("Video_Settings", "EnablePerPixelDepth", bEnablePerPixelDepth);
-	SET_IF_DIFFERS("Video_Settings", "EnableFastZcomploc", bEnableFastZcomploc);
 	SET_IF_DIFFERS("Video_Settings", "MSAA", iMultisampleMode);
 	SET_IF_DIFFERS("Video_Settings", "EFBScale", iEFBScale); // integral
 	SET_IF_DIFFERS("Video_Settings", "DstAlphaPass", bDstAlphaPass);
diff --git a/Source/Core/VideoCommon/Src/VideoConfig.h b/Source/Core/VideoCommon/Src/VideoConfig.h
index d6dc3cefe0..7c846676bb 100644
--- a/Source/Core/VideoCommon/Src/VideoConfig.h
+++ b/Source/Core/VideoCommon/Src/VideoConfig.h
@@ -137,7 +137,6 @@ struct VideoConfig
 	bool bUseBBox;
 	bool bEnablePixelLighting;
 	bool bEnablePerPixelDepth;
-	bool bEnableFastZcomploc;
 
 	int iLog; // CONF_ bits
 	int iSaveTargetId;
-- 
cgit v1.2.3


From 722480cb2e864ea7dd03427e509470ec09ce6f40 Mon Sep 17 00:00:00 2001
From: Jordan Woyak <jordan.woyak@gmail.com>
Date: Sat, 7 Apr 2012 14:45:12 -0500
Subject: Fix compilation with g++ 4.7 and some warnings. Fixes issue 5347.

---
 Source/Core/VideoCommon/Src/PixelShaderGen.cpp  | 64 +++++++++++------------
 Source/Core/VideoCommon/Src/VertexShaderGen.cpp | 68 ++++++++++++-------------
 2 files changed, 66 insertions(+), 66 deletions(-)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/PixelShaderGen.cpp b/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
index 76964f248c..7500997fef 100644
--- a/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
+++ b/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
@@ -556,22 +556,22 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType
 
 	WRITE(p, "\n");
 
-	WRITE(p, "uniform float4 "I_COLORS"[4] : register(c%d);\n", C_COLORS);
-	WRITE(p, "uniform float4 "I_KCOLORS"[4] : register(c%d);\n", C_KCOLORS);
-	WRITE(p, "uniform float4 "I_ALPHA"[1] : register(c%d);\n", C_ALPHA);
-	WRITE(p, "uniform float4 "I_TEXDIMS"[8] : register(c%d);\n", C_TEXDIMS);
-	WRITE(p, "uniform float4 "I_ZBIAS"[2] : register(c%d);\n", C_ZBIAS);
-	WRITE(p, "uniform float4 "I_INDTEXSCALE"[2] : register(c%d);\n", C_INDTEXSCALE);
-	WRITE(p, "uniform float4 "I_INDTEXMTX"[6] : register(c%d);\n", C_INDTEXMTX);
-	WRITE(p, "uniform float4 "I_FOG"[3] : register(c%d);\n", C_FOG);
+	WRITE(p, "uniform float4 " I_COLORS"[4] : register(c%d);\n", C_COLORS);
+	WRITE(p, "uniform float4 " I_KCOLORS"[4] : register(c%d);\n", C_KCOLORS);
+	WRITE(p, "uniform float4 " I_ALPHA"[1] : register(c%d);\n", C_ALPHA);
+	WRITE(p, "uniform float4 " I_TEXDIMS"[8] : register(c%d);\n", C_TEXDIMS);
+	WRITE(p, "uniform float4 " I_ZBIAS"[2] : register(c%d);\n", C_ZBIAS);
+	WRITE(p, "uniform float4 " I_INDTEXSCALE"[2] : register(c%d);\n", C_INDTEXSCALE);
+	WRITE(p, "uniform float4 " I_INDTEXMTX"[6] : register(c%d);\n", C_INDTEXMTX);
+	WRITE(p, "uniform float4 " I_FOG"[3] : register(c%d);\n", C_FOG);
 
 	if(g_ActiveConfig.bEnablePixelLighting && g_ActiveConfig.backend_info.bSupportsPixelLighting)
 	{
 		WRITE(p,"typedef struct { float4 col; float4 cosatt; float4 distatt; float4 pos; float4 dir; } Light;\n");
-		WRITE(p,"typedef struct { Light lights[8]; } s_"I_PLIGHTS";\n");
-		WRITE(p, "uniform s_"I_PLIGHTS" "I_PLIGHTS" : register(c%d);\n", C_PLIGHTS);
-		WRITE(p, "typedef struct { float4 C0, C1, C2, C3; } s_"I_PMATERIALS";\n");
-		WRITE(p, "uniform s_"I_PMATERIALS" "I_PMATERIALS" : register(c%d);\n", C_PMATERIALS);
+		WRITE(p,"typedef struct { Light lights[8]; } s_" I_PLIGHTS";\n");
+		WRITE(p, "uniform s_" I_PLIGHTS" " I_PLIGHTS" : register(c%d);\n", C_PLIGHTS);
+		WRITE(p, "typedef struct { float4 C0, C1, C2, C3; } s_" I_PMATERIALS";\n");
+		WRITE(p, "uniform s_" I_PMATERIALS" " I_PMATERIALS" : register(c%d);\n", C_PMATERIALS);
 	}
 
 	WRITE(p, "void main(\n");
@@ -635,7 +635,7 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType
 		}
 		else if (dstAlphaMode == DSTALPHA_ALPHA_PASS)
 		{
-			WRITE(p, "  ocol0 = "I_ALPHA"[0].aaaa;\n");
+			WRITE(p, "  ocol0 = " I_ALPHA"[0].aaaa;\n");
 		}
 		if(!Pretest || dstAlphaMode == DSTALPHA_ALPHA_PASS)
 		{
@@ -644,7 +644,7 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType
 		}
 	}
 
-	WRITE(p, "  float4 c0 = "I_COLORS"[1], c1 = "I_COLORS"[2], c2 = "I_COLORS"[3], prev = float4(0.0f, 0.0f, 0.0f, 0.0f), textemp = float4(0.0f, 0.0f, 0.0f, 0.0f), rastemp = float4(0.0f, 0.0f, 0.0f, 0.0f), konsttemp = float4(0.0f, 0.0f, 0.0f, 0.0f);\n"
+	WRITE(p, "  float4 c0 = " I_COLORS"[1], c1 = " I_COLORS"[2], c2 = " I_COLORS"[3], prev = float4(0.0f, 0.0f, 0.0f, 0.0f), textemp = float4(0.0f, 0.0f, 0.0f, 0.0f), rastemp = float4(0.0f, 0.0f, 0.0f, 0.0f), konsttemp = float4(0.0f, 0.0f, 0.0f, 0.0f);\n"
 			"  float3 comp16 = float3(1.0f, 255.0f, 0.0f), comp24 = float3(1.0f, 255.0f, 255.0f*255.0f);\n"
 			"  float4 alphabump=float4(0.0f,0.0f,0.0f,0.0f);\n"
 			"  float3 tevcoord=float3(0.0f, 0.0f, 0.0f);\n"
@@ -695,7 +695,7 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType
 				WRITE(p, "	uv%d.xy = uv%d.xy / uv%d.z;\n", i, i, i);
 			}
 
-			WRITE(p, "uv%d.xy = uv%d.xy * "I_TEXDIMS"[%d].zw;\n", i, i, i);
+			WRITE(p, "uv%d.xy = uv%d.xy * " I_TEXDIMS"[%d].zw;\n", i, i, i);
 		}
 	}
 
@@ -707,7 +707,7 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType
 			int texcoord = bpmem.tevindref.getTexCoord(i);
 
 			if (texcoord < numTexgen)
-				WRITE(p, "tempcoord = uv%d.xy * "I_INDTEXSCALE"[%d].%s;\n", texcoord, i/2, (i&1)?"zw":"xy");
+				WRITE(p, "tempcoord = uv%d.xy * " I_INDTEXSCALE"[%d].%s;\n", texcoord, i/2, (i&1)?"zw":"xy");
 			else
 				WRITE(p, "tempcoord = float2(0.0f, 0.0f);\n");
 
@@ -738,7 +738,7 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType
 	if((bpmem.fog.c_proj_fsel.fsel != 0) || DepthTextureEnable)
 	{
 		// the screen space depth value = far z + (clip z / clip w) * z range
-		WRITE(p, "float zCoord = "I_ZBIAS"[1].x + (clipPos.z / clipPos.w) * "I_ZBIAS"[1].y;\n");
+		WRITE(p, "float zCoord = " I_ZBIAS"[1].x + (clipPos.z / clipPos.w) * " I_ZBIAS"[1].y;\n");
 	}
 
 	if (DepthTextureEnable)
@@ -747,9 +747,9 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType
 		if (bpmem.ztex2.op != ZTEXTURE_DISABLE && !bpmem.zcontrol.zcomploc && bpmem.zmode.testenable && bpmem.zmode.updateenable)
 		{
 			if (bpmem.ztex2.op == ZTEXTURE_ADD)
-				WRITE(p, "zCoord = dot("I_ZBIAS"[0].xyzw, textemp.xyzw) + "I_ZBIAS"[1].w + zCoord;\n");
+				WRITE(p, "zCoord = dot(" I_ZBIAS"[0].xyzw, textemp.xyzw) + " I_ZBIAS"[1].w + zCoord;\n");
 			else
-				WRITE(p, "zCoord = dot("I_ZBIAS"[0].xyzw, textemp.xyzw) + "I_ZBIAS"[1].w;\n");
+				WRITE(p, "zCoord = dot(" I_ZBIAS"[0].xyzw, textemp.xyzw) + " I_ZBIAS"[1].w;\n");
 
 			// scale to make result from frac correct
 			WRITE(p, "zCoord = zCoord * (16777215.0f/16777216.0f);\n");
@@ -760,7 +760,7 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType
 	}
 
 	if (dstAlphaMode == DSTALPHA_ALPHA_PASS)
-		WRITE(p, "  ocol0 = float4(prev.rgb, "I_ALPHA"[0].a);\n");
+		WRITE(p, "  ocol0 = float4(prev.rgb, " I_ALPHA"[0].a);\n");
 	else
 	{
 		WriteFog(p);
@@ -774,7 +774,7 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType
 		// Colors will be blended against the alpha from ocol1...
 		WRITE(p, "  ocol1 = ocol0;\n");
 		// ...and the alpha from ocol0 will be written to the framebuffer.
-		WRITE(p, "  ocol0.a = "I_ALPHA"[0].a;\n");
+		WRITE(p, "  ocol0.a = " I_ALPHA"[0].a;\n");
 	}
 	
 	WRITE(p, "}\n");
@@ -867,20 +867,20 @@ static void WriteStage(char *&p, int n, API_TYPE ApiType)
 			if (bpmem.tevind[n].mid <= 3)
 			{
 				int mtxidx = 2*(bpmem.tevind[n].mid-1);
-				WRITE(p, "float2 indtevtrans%d = float2(dot("I_INDTEXMTX"[%d].xyz, indtevcrd%d), dot("I_INDTEXMTX"[%d].xyz, indtevcrd%d));\n",
+				WRITE(p, "float2 indtevtrans%d = float2(dot(" I_INDTEXMTX"[%d].xyz, indtevcrd%d), dot(" I_INDTEXMTX"[%d].xyz, indtevcrd%d));\n",
 					n, mtxidx, n, mtxidx+1, n);
 			}
 			else if (bpmem.tevind[n].mid <= 7 && bHasTexCoord)
 			{ // s matrix
 				_assert_(bpmem.tevind[n].mid >= 5);
 				int mtxidx = 2*(bpmem.tevind[n].mid-5);
-				WRITE(p, "float2 indtevtrans%d = "I_INDTEXMTX"[%d].ww * uv%d.xy * indtevcrd%d.xx;\n", n, mtxidx, texcoord, n);
+				WRITE(p, "float2 indtevtrans%d = " I_INDTEXMTX"[%d].ww * uv%d.xy * indtevcrd%d.xx;\n", n, mtxidx, texcoord, n);
 			}
 			else if (bpmem.tevind[n].mid <= 11 && bHasTexCoord)
 			{ // t matrix
 				_assert_(bpmem.tevind[n].mid >= 9);
 				int mtxidx = 2*(bpmem.tevind[n].mid-9);
-				WRITE(p, "float2 indtevtrans%d = "I_INDTEXMTX"[%d].ww * uv%d.xy * indtevcrd%d.yy;\n", n, mtxidx, texcoord, n);
+				WRITE(p, "float2 indtevtrans%d = " I_INDTEXMTX"[%d].ww * uv%d.xy * indtevcrd%d.yy;\n", n, mtxidx, texcoord, n);
 			}
 			else
 				WRITE(p, "float2 indtevtrans%d = 0;\n", n);
@@ -1094,9 +1094,9 @@ static void WriteStage(char *&p, int n, API_TYPE ApiType)
 void SampleTexture(char *&p, const char *destination, const char *texcoords, const char *texswap, int texmap, API_TYPE ApiType)
 {
 	if (ApiType == API_D3D11)
-		WRITE(p, "%s=Tex%d.Sample(samp%d,%s.xy * "I_TEXDIMS"[%d].xy).%s;\n", destination, texmap,texmap, texcoords, texmap, texswap);
+		WRITE(p, "%s=Tex%d.Sample(samp%d,%s.xy * " I_TEXDIMS"[%d].xy).%s;\n", destination, texmap,texmap, texcoords, texmap, texswap);
 	else
-		WRITE(p, "%s=tex2D(samp%d,%s.xy * "I_TEXDIMS"[%d].xy).%s;\n", destination, texmap, texcoords, texmap, texswap);
+		WRITE(p, "%s=tex2D(samp%d,%s.xy * " I_TEXDIMS"[%d].xy).%s;\n", destination, texmap, texcoords, texmap, texswap);
 }
 
 static const char *tevAlphaFuncsTable[] =
@@ -1221,13 +1221,13 @@ static void WriteFog(char *&p)
 	{
 		// perspective
 		// ze = A/(B - (Zs >> B_SHF)
-		WRITE (p, "  float ze = "I_FOG"[1].x / ("I_FOG"[1].y - (zCoord / "I_FOG"[1].w));\n");
+		WRITE (p, "  float ze = " I_FOG"[1].x / (" I_FOG"[1].y - (zCoord / " I_FOG"[1].w));\n");
 	}
 	else
 	{
 		// orthographic
 		// ze = a*Zs	(here, no B_SHF)
-		WRITE (p, "  float ze = "I_FOG"[1].x * zCoord;\n");
+		WRITE (p, "  float ze = " I_FOG"[1].x * zCoord;\n");
 	}
 
 	// x_adjust = sqrt((x-center)^2 + k^2)/k
@@ -1235,12 +1235,12 @@ static void WriteFog(char *&p)
 	//this is complitly teorical as the real hard seems to use a table intead of calculate the values.
 	if(bpmem.fogRange.Base.Enabled)
 	{
-		WRITE (p, "  float x_adjust = (2.0f * (clipPos.x / "I_FOG"[2].y)) - 1.0f - "I_FOG"[2].x;\n");
-		WRITE (p, "  x_adjust = sqrt(x_adjust * x_adjust + "I_FOG"[2].z * "I_FOG"[2].z) / "I_FOG"[2].z;\n");
+		WRITE (p, "  float x_adjust = (2.0f * (clipPos.x / " I_FOG"[2].y)) - 1.0f - " I_FOG"[2].x;\n");
+		WRITE (p, "  x_adjust = sqrt(x_adjust * x_adjust + " I_FOG"[2].z * " I_FOG"[2].z) / " I_FOG"[2].z;\n");
 		WRITE (p, "  ze *= x_adjust;\n");
 	}
 
-	WRITE (p, "  float fog = saturate(ze - "I_FOG"[1].z);\n");
+	WRITE (p, "  float fog = saturate(ze - " I_FOG"[1].z);\n");
 
 	if(bpmem.fog.c_proj_fsel.fsel > 3)
 	{
@@ -1252,7 +1252,7 @@ static void WriteFog(char *&p)
 			WARN_LOG(VIDEO, "Unknown Fog Type! %08x", bpmem.fog.c_proj_fsel.fsel);
 	}
 
-	WRITE(p, "  prev.rgb = lerp(prev.rgb,"I_FOG"[0].rgb,fog);\n");
+	WRITE(p, "  prev.rgb = lerp(prev.rgb," I_FOG"[0].rgb,fog);\n");
 
 
 }
diff --git a/Source/Core/VideoCommon/Src/VertexShaderGen.cpp b/Source/Core/VideoCommon/Src/VertexShaderGen.cpp
index 3df779a7ba..07ace04b97 100644
--- a/Source/Core/VideoCommon/Src/VertexShaderGen.cpp
+++ b/Source/Core/VideoCommon/Src/VertexShaderGen.cpp
@@ -178,31 +178,31 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type)
 
 	char *p = text;
 	WRITE(p, "//Vertex Shader: comp:%x, \n", components);
-	WRITE(p, "typedef struct { float4 T0, T1, T2; float4 N0, N1, N2; } s_"I_POSNORMALMATRIX";\n"
+	WRITE(p, "typedef struct { float4 T0, T1, T2; float4 N0, N1, N2; } s_" I_POSNORMALMATRIX";\n"
 		"typedef struct { float4 t; } FLT4;\n"
-		"typedef struct { FLT4 T[24]; } s_"I_TEXMATRICES";\n"
-		"typedef struct { FLT4 T[64]; } s_"I_TRANSFORMMATRICES";\n"
-		"typedef struct { FLT4 T[32]; } s_"I_NORMALMATRICES";\n"
-		"typedef struct { FLT4 T[64]; } s_"I_POSTTRANSFORMMATRICES";\n"
+		"typedef struct { FLT4 T[24]; } s_" I_TEXMATRICES";\n"
+		"typedef struct { FLT4 T[64]; } s_" I_TRANSFORMMATRICES";\n"
+		"typedef struct { FLT4 T[32]; } s_" I_NORMALMATRICES";\n"
+		"typedef struct { FLT4 T[64]; } s_" I_POSTTRANSFORMMATRICES";\n"
 		"typedef struct { float4 col; float4 cosatt; float4 distatt; float4 pos; float4 dir; } Light;\n"
-		"typedef struct { Light lights[8]; } s_"I_LIGHTS";\n"
-		"typedef struct { float4 C0, C1, C2, C3; } s_"I_MATERIALS";\n"
-		"typedef struct { float4 T0, T1, T2, T3; } s_"I_PROJECTION";\n"
+		"typedef struct { Light lights[8]; } s_" I_LIGHTS";\n"
+		"typedef struct { float4 C0, C1, C2, C3; } s_" I_MATERIALS";\n"
+		"typedef struct { float4 T0, T1, T2, T3; } s_" I_PROJECTION";\n"
 		);
 
 	p = GenerateVSOutputStruct(p, components, api_type);
 
 	// uniforms
 
-	WRITE(p, "uniform s_"I_TRANSFORMMATRICES" "I_TRANSFORMMATRICES" : register(c%d);\n", C_TRANSFORMMATRICES);
-	WRITE(p, "uniform s_"I_TEXMATRICES" "I_TEXMATRICES" : register(c%d);\n", C_TEXMATRICES); // also using tex matrices
-	WRITE(p, "uniform s_"I_NORMALMATRICES" "I_NORMALMATRICES" : register(c%d);\n", C_NORMALMATRICES);
-	WRITE(p, "uniform s_"I_POSNORMALMATRIX" "I_POSNORMALMATRIX" : register(c%d);\n", C_POSNORMALMATRIX);
-	WRITE(p, "uniform s_"I_POSTTRANSFORMMATRICES" "I_POSTTRANSFORMMATRICES" : register(c%d);\n", C_POSTTRANSFORMMATRICES);
-	WRITE(p, "uniform s_"I_LIGHTS" "I_LIGHTS" : register(c%d);\n", C_LIGHTS);
-	WRITE(p, "uniform s_"I_MATERIALS" "I_MATERIALS" : register(c%d);\n", C_MATERIALS);
-	WRITE(p, "uniform s_"I_PROJECTION" "I_PROJECTION" : register(c%d);\n", C_PROJECTION);
-	WRITE(p, "uniform float4 "I_DEPTHPARAMS" : register(c%d);\n", C_DEPTHPARAMS);
+	WRITE(p, "uniform s_" I_TRANSFORMMATRICES" " I_TRANSFORMMATRICES" : register(c%d);\n", C_TRANSFORMMATRICES);
+	WRITE(p, "uniform s_" I_TEXMATRICES" " I_TEXMATRICES" : register(c%d);\n", C_TEXMATRICES); // also using tex matrices
+	WRITE(p, "uniform s_" I_NORMALMATRICES" " I_NORMALMATRICES" : register(c%d);\n", C_NORMALMATRICES);
+	WRITE(p, "uniform s_" I_POSNORMALMATRIX" " I_POSNORMALMATRIX" : register(c%d);\n", C_POSNORMALMATRIX);
+	WRITE(p, "uniform s_" I_POSTTRANSFORMMATRICES" " I_POSTTRANSFORMMATRICES" : register(c%d);\n", C_POSTTRANSFORMMATRICES);
+	WRITE(p, "uniform s_" I_LIGHTS" " I_LIGHTS" : register(c%d);\n", C_LIGHTS);
+	WRITE(p, "uniform s_" I_MATERIALS" " I_MATERIALS" : register(c%d);\n", C_MATERIALS);
+	WRITE(p, "uniform s_" I_PROJECTION" " I_PROJECTION" : register(c%d);\n", C_PROJECTION);
+	WRITE(p, "uniform float4 " I_DEPTHPARAMS" : register(c%d);\n", C_DEPTHPARAMS);
 
 	WRITE(p, "VS_OUTPUT main(\n");	
 	
@@ -258,11 +258,11 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type)
 			WRITE(p, "int posmtx = fposmtx;\n");
 		}
 
-		WRITE(p, "float4 pos = float4(dot("I_TRANSFORMMATRICES".T[posmtx].t, rawpos), dot("I_TRANSFORMMATRICES".T[posmtx+1].t, rawpos), dot("I_TRANSFORMMATRICES".T[posmtx+2].t, rawpos), 1);\n");		
+		WRITE(p, "float4 pos = float4(dot(" I_TRANSFORMMATRICES".T[posmtx].t, rawpos), dot(" I_TRANSFORMMATRICES".T[posmtx+1].t, rawpos), dot(" I_TRANSFORMMATRICES".T[posmtx+2].t, rawpos), 1);\n");		
 
 		if (components & VB_HAS_NRMALL) {
 			WRITE(p, "int normidx = posmtx >= 32 ? (posmtx-32) : posmtx;\n");
-			WRITE(p, "float3 N0 = "I_NORMALMATRICES".T[normidx].t.xyz, N1 = "I_NORMALMATRICES".T[normidx+1].t.xyz, N2 = "I_NORMALMATRICES".T[normidx+2].t.xyz;\n");
+			WRITE(p, "float3 N0 = " I_NORMALMATRICES".T[normidx].t.xyz, N1 = " I_NORMALMATRICES".T[normidx+1].t.xyz, N2 = " I_NORMALMATRICES".T[normidx+2].t.xyz;\n");
 		}
 
 		if (components & VB_HAS_NRM0)
@@ -274,13 +274,13 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type)
 	}
 	else
 	{
-		WRITE(p, "float4 pos = float4(dot("I_POSNORMALMATRIX".T0, rawpos), dot("I_POSNORMALMATRIX".T1, rawpos), dot("I_POSNORMALMATRIX".T2, rawpos), 1.0f);\n");
+		WRITE(p, "float4 pos = float4(dot(" I_POSNORMALMATRIX".T0, rawpos), dot(" I_POSNORMALMATRIX".T1, rawpos), dot(" I_POSNORMALMATRIX".T2, rawpos), 1.0f);\n");
 		if (components & VB_HAS_NRM0)
-			WRITE(p, "float3 _norm0 = normalize(float3(dot("I_POSNORMALMATRIX".N0.xyz, rawnorm0), dot("I_POSNORMALMATRIX".N1.xyz, rawnorm0), dot("I_POSNORMALMATRIX".N2.xyz, rawnorm0)));\n");
+			WRITE(p, "float3 _norm0 = normalize(float3(dot(" I_POSNORMALMATRIX".N0.xyz, rawnorm0), dot(" I_POSNORMALMATRIX".N1.xyz, rawnorm0), dot(" I_POSNORMALMATRIX".N2.xyz, rawnorm0)));\n");
 		if (components & VB_HAS_NRM1)
-			WRITE(p, "float3 _norm1 = float3(dot("I_POSNORMALMATRIX".N0.xyz, rawnorm1), dot("I_POSNORMALMATRIX".N1.xyz, rawnorm1), dot("I_POSNORMALMATRIX".N2.xyz, rawnorm1));\n");
+			WRITE(p, "float3 _norm1 = float3(dot(" I_POSNORMALMATRIX".N0.xyz, rawnorm1), dot(" I_POSNORMALMATRIX".N1.xyz, rawnorm1), dot(" I_POSNORMALMATRIX".N2.xyz, rawnorm1));\n");
 		if (components & VB_HAS_NRM2)
-			WRITE(p, "float3 _norm2 = float3(dot("I_POSNORMALMATRIX".N0.xyz, rawnorm2), dot("I_POSNORMALMATRIX".N1.xyz, rawnorm2), dot("I_POSNORMALMATRIX".N2.xyz, rawnorm2));\n");
+			WRITE(p, "float3 _norm2 = float3(dot(" I_POSNORMALMATRIX".N0.xyz, rawnorm2), dot(" I_POSNORMALMATRIX".N1.xyz, rawnorm2), dot(" I_POSNORMALMATRIX".N2.xyz, rawnorm2));\n");
 	}
 
 	if (!(components & VB_HAS_NRM0))
@@ -288,7 +288,7 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type)
 
 	
 
-	WRITE(p, "o.pos = float4(dot("I_PROJECTION".T0, pos), dot("I_PROJECTION".T1, pos), dot("I_PROJECTION".T2, pos), dot("I_PROJECTION".T3, pos));\n");
+	WRITE(p, "o.pos = float4(dot(" I_PROJECTION".T0, pos), dot(" I_PROJECTION".T1, pos), dot(" I_PROJECTION".T2, pos), dot(" I_PROJECTION".T3, pos));\n");
 
 	WRITE(p, "float4 mat, lacc;\n"
 	"float3 ldir, h;\n"
@@ -367,7 +367,7 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type)
 
 				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, "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
@@ -389,16 +389,16 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type)
 			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);
+						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);
 					else {
-						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), 1);\n", i, i, i);
+						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), 1);\n", i, i, i);
 					}
 				}
 				else {
 					if (texinfo.projection == XF_TEXPROJ_STQ)
-						WRITE(p, "o.tex%d.xyz = float3(dot(coord, "I_TEXMATRICES".T[%d].t), dot(coord, "I_TEXMATRICES".T[%d].t), dot(coord, "I_TEXMATRICES".T[%d].t));\n", i, 3*i, 3*i+1, 3*i+2);
+						WRITE(p, "o.tex%d.xyz = float3(dot(coord, " I_TEXMATRICES".T[%d].t), dot(coord, " I_TEXMATRICES".T[%d].t), dot(coord, " I_TEXMATRICES".T[%d].t));\n", i, 3*i, 3*i+1, 3*i+2);
 					else
-						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);
+						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;
 		}
@@ -407,9 +407,9 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type)
 			const PostMtxInfo& postInfo = xfregs.postMtxInfo[i];
 
 			int postidx = postInfo.index;
-			WRITE(p, "float4 P0 = "I_POSTTRANSFORMMATRICES".T[%d].t;\n"
-				"float4 P1 = "I_POSTTRANSFORMMATRICES".T[%d].t;\n"
-				"float4 P2 = "I_POSTTRANSFORMMATRICES".T[%d].t;\n",
+			WRITE(p, "float4 P0 = " I_POSTTRANSFORMMATRICES".T[%d].t;\n"
+				"float4 P1 = " I_POSTTRANSFORMMATRICES".T[%d].t;\n"
+				"float4 P2 = " I_POSTTRANSFORMMATRICES".T[%d].t;\n",
 				postidx&0x3f, (postidx+1)&0x3f, (postidx+2)&0x3f);
 
 			if (texGenSpecialCase) {
@@ -467,7 +467,7 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type)
 	//if not early z culling will improve speed
 	if (is_d3d)
 	{
-		WRITE(p, "o.pos.z = "I_DEPTHPARAMS".x * o.pos.w + o.pos.z * "I_DEPTHPARAMS".y;\n");
+		WRITE(p, "o.pos.z = " I_DEPTHPARAMS".x * o.pos.w + o.pos.z * " I_DEPTHPARAMS".y;\n");
 	}
 	else
 	{
@@ -497,7 +497,7 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type)
 	{
 		// D3D9 is addressing pixel centers instead of pixel boundaries in clip space.
 		// Thus we need to offset the final position by half a pixel
-		WRITE(p, "o.pos = o.pos + float4("I_DEPTHPARAMS".z, "I_DEPTHPARAMS".w, 0.f, 0.f);\n");
+		WRITE(p, "o.pos = o.pos + float4(" I_DEPTHPARAMS".z, " I_DEPTHPARAMS".w, 0.f, 0.f);\n");
 	}
 
 	WRITE(p, "return o;\n}\n");
-- 
cgit v1.2.3


From beb4204bbbd42cadd057358e79966d4350c3a3e8 Mon Sep 17 00:00:00 2001
From: Pierre Bourdon <delroth@gmail.com>
Date: Wed, 2 May 2012 07:29:15 +0200
Subject: Include libavutil/mathematics.h explictly in AVIDump.cpp. Fixes issue
 #5266.

---
 Source/Core/VideoCommon/Src/AVIDump.cpp | 1 +
 1 file changed, 1 insertion(+)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/AVIDump.cpp b/Source/Core/VideoCommon/Src/AVIDump.cpp
index 670693690b..4c31798d9f 100644
--- a/Source/Core/VideoCommon/Src/AVIDump.cpp
+++ b/Source/Core/VideoCommon/Src/AVIDump.cpp
@@ -211,6 +211,7 @@ extern "C" {
 #include <libavcodec/avcodec.h>
 #include <libavformat/avformat.h>
 #include <libswscale/swscale.h>
+#include <libavutil/mathematics.h>
 }
 
 AVFormatContext *s_FormatContext = NULL;
-- 
cgit v1.2.3


From 1587cb3738ee7bcf156a4a1257a67730cc7ee9c3 Mon Sep 17 00:00:00 2001
From: skidau <skidau@gmail.com>
Date: Sat, 5 May 2012 11:21:05 +1000
Subject: Fixed texture encoding.  Fixes the interaction with objects in
 Another Code R. Thanks to wordmanwords for the patch.

Fixes issue 5405.
---
 Source/Core/VideoCommon/Src/TextureConversionShader.cpp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'Source/Core/VideoCommon/Src')

diff --git a/Source/Core/VideoCommon/Src/TextureConversionShader.cpp b/Source/Core/VideoCommon/Src/TextureConversionShader.cpp
index c921a1a9fd..2132a25258 100644
--- a/Source/Core/VideoCommon/Src/TextureConversionShader.cpp
+++ b/Source/Core/VideoCommon/Src/TextureConversionShader.cpp
@@ -259,7 +259,7 @@ void WriteIncrementSampleX(char*& p,API_TYPE ApiType)
 
 void WriteToBitDepth(char*& p, u8 depth, const char* src, const char* dest)
 {
-	float result = pow(2.0f, depth) - 1.0f;
+	float result = 255 / pow(2.0f, (8 - depth));
 	WRITE(p, "  %s = floor(%s * %ff);\n", dest, src, result);
 }
 
-- 
cgit v1.2.3