Merge branch 'FIFO-BP'

# By skidau (30) and Pierre Bourdon (1)
* FIFO-BP: (31 commits)
  Set g_bSignalTokenInterrupt on the main thread.  Fixes the random hang in Harry Potter: Prisoner of Azkaban.
  Used a scheduled event to generate the ARAM DMA interrupt if the DMA is greater than a certain size.  Fixes NFS:HP2 GC.
  Bumped up the disc transfer speed enough to prevent audio stuttering in Gauntlet: Dark Legacy.
  Enabled Synchronise GPU on "SPEED CHALLENGE - Jacques Villeneuve's Racing Vision".  Required to go in-game.
  Added direct GameCube controller commands to the Serial Interface emulation.  Fixes the controls in MaxPlay Classic Games Volume 1 and the Action Replay disc.
  Increased the FIFO buffer size to 2MB from 1MB.  Fixes Killer 7's Angel boss.
  Used an immediate GenerateDSPInterrupt when transferring data from ARAM to MRAM and a scheduled DSP interrupt when transferring data from MRAM to ARAM.
  Fixes the audio cutting in and out in the Resident Evil GC games using DSP HLE. Triggered the ARAM interrupt by the scheduler instead of directly in function.
  Implemented proper timing for the sample counter in the AudioInterface, removing the previous hack. Cleaned up some of the audio streaming code.
  Skipped the EE check if there is a CP interrupt pending.
  Disabled "Speed up disc transfer" from the ZTP GC game ini.
  Removed the disc seek times for GC games and removed the disc speed option on Wii games. Checked for external exceptions only in mtmsr.
  Delayed the interrupts in the EXI Channel.
  Merge aram-dma-fixes (r76a13604ef49b522281af75675f044d59a74e871)
  Added a patch that bypasses the FIFO reset code in Wallace and Gromit: Project Zoo, allowing it to go in-game.
  Made vertex loading take constant time.
  Increased the cycle time of the vertex command.  Fixes "Speed Challenge: Jacques Villeneuve's Racing Vision".
  Moved the setting of the Finish interrupt signal back to the main thread as it was causing Wii games like Resident Evil 4 (Wii) to hang.
  Profile stores, fp stores and ps stores only to the fifo write addresses list.  This should make the JIT a little faster as it will not be checking for external exceptions unnecessarily.
  ...

Conflicts:
	Source/Core/VideoCommon/Src/PixelEngine.cpp

Source/Core/Core/Src/BootManager.cpp

@@ -55,7 +55,7 @@ namespace BootManager
 struct ConfigCache
 {
 	bool valid, bCPUThread, bSkipIdle, bEnableFPRF, bMMU, bDCBZOFF,
-		bVBeam, bFastDiscSpeed, bMergeBlocks, bDSPHLE, bHLE_BS2;
+		bVBeam, bSyncGPU, bFastDiscSpeed, bMergeBlocks, bDSPHLE, bHLE_BS2;
 	int iTLBHack, iCPUCore;
 	std::string strBackend;
 };
@@ -95,6 +95,7 @@ bool BootCore(const std::string& _rFilename)
 		config_cache.bDCBZOFF = StartUp.bDCBZOFF;
 		config_cache.iTLBHack = StartUp.iTLBHack;
 		config_cache.bVBeam = StartUp.bVBeam;
+		config_cache.bSyncGPU = StartUp.bSyncGPU;
 		config_cache.bFastDiscSpeed = StartUp.bFastDiscSpeed;
 		config_cache.bMergeBlocks = StartUp.bMergeBlocks;
 		config_cache.bDSPHLE = StartUp.bDSPHLE;
@@ -109,6 +110,7 @@ bool BootCore(const std::string& _rFilename)
 		game_ini.Get("Core", "TLBHack",				&StartUp.iTLBHack, StartUp.iTLBHack);
 		game_ini.Get("Core", "DCBZ",				&StartUp.bDCBZOFF, StartUp.bDCBZOFF);
 		game_ini.Get("Core", "VBeam",				&StartUp.bVBeam, StartUp.bVBeam);
+		game_ini.Get("Core", "SyncGPU",				&StartUp.bSyncGPU, StartUp.bSyncGPU);
 		game_ini.Get("Core", "FastDiscSpeed",		&StartUp.bFastDiscSpeed, StartUp.bFastDiscSpeed);
 		game_ini.Get("Core", "BlockMerging",		&StartUp.bMergeBlocks, StartUp.bMergeBlocks);
 		game_ini.Get("Core", "DSPHLE",				&StartUp.bDSPHLE, StartUp.bDSPHLE);
@@ -168,6 +170,7 @@ void Stop()
 		StartUp.bDCBZOFF = config_cache.bDCBZOFF;
 		StartUp.iTLBHack = config_cache.iTLBHack;
 		StartUp.bVBeam = config_cache.bVBeam;
+		StartUp.bSyncGPU = config_cache.bSyncGPU;
 		StartUp.bFastDiscSpeed = config_cache.bFastDiscSpeed;
 		StartUp.bMergeBlocks = config_cache.bMergeBlocks;
 		StartUp.bDSPHLE = config_cache.bDSPHLE;

Source/Core/Core/Src/ConfigManager.cpp

@@ -404,6 +404,7 @@ void SConfig::LoadSettings()
 		ini.Get("Core", "MMU",				&m_LocalCoreStartupParameter.bMMU,				false);
 		ini.Get("Core", "TLBHack",			&m_LocalCoreStartupParameter.iTLBHack,			0);
 		ini.Get("Core", "VBeam",			&m_LocalCoreStartupParameter.bVBeam,			false);
+		ini.Get("Core", "SyncGPU",			&m_LocalCoreStartupParameter.bSyncGPU,			false);
 		ini.Get("Core", "FastDiscSpeed",	&m_LocalCoreStartupParameter.bFastDiscSpeed,	false);
 		ini.Get("Core", "DCBZ",				&m_LocalCoreStartupParameter.bDCBZOFF,			false);
 		ini.Get("Core", "FrameLimit",		&m_Framelimit,									1); // auto frame limit by default

Source/Core/Core/Src/CoreParameter.cpp

@@ -50,7 +50,7 @@ SCoreStartupParameter::SCoreStartupParameter()
   bDPL2Decoder(false), iLatency(14),
   bRunCompareServer(false), bRunCompareClient(false),
   bMMU(false), bDCBZOFF(false), iTLBHack(0), bVBeam(false),
-  bFastDiscSpeed(false),
+  bSyncGPU(false), bFastDiscSpeed(false),
   SelectedLanguage(0), bWii(false),
   bConfirmStop(false), bHideCursor(false),
   bAutoHideCursor(false), bUsePanicHandlers(true), bOnScreenDisplayMessages(true),
@@ -78,6 +78,7 @@ void SCoreStartupParameter::LoadDefaults()
 	bDCBZOFF = false;
 	iTLBHack = 0;
 	bVBeam = false;
+	bSyncGPU = false;
 	bFastDiscSpeed = false;
 	bMergeBlocks = false;
 	SelectedLanguage = 0;

Source/Core/Core/Src/CoreParameter.h

@@ -116,6 +116,7 @@ struct SCoreStartupParameter
 	bool bDCBZOFF;
 	int iTLBHack;
 	bool bVBeam;
+	bool bSyncGPU;
 	bool bFastDiscSpeed;
 
 	int SelectedLanguage;

Source/Core/Core/Src/HW/AudioInterface.cpp

@@ -145,6 +145,8 @@ static void GenerateAudioInterrupt();
 static void UpdateInterrupts();
 static void IncreaseSampleCount(const u32 _uAmount);
 void ReadStreamBlock(s16* _pPCM);
+u64 GetAIPeriod();
+int et_AI;
 
 void Init()
 {
@@ -159,6 +161,8 @@ void Init()
 
 	g_AISSampleRate = 48000;
 	g_AIDSampleRate = 32000;
+
+	et_AI = CoreTiming::RegisterEvent("AICallback", Update);
 }
 
 void Shutdown()
@@ -178,11 +182,8 @@ void Read32(u32& _rReturnValue, const u32 _Address)
 		break;
 
 	case AI_SAMPLE_COUNTER:
+		Update(0, 0);
         _rReturnValue = m_SampleCounter;
-		// HACK - AI SRC init will do while (oldval == sample_counter) {}
-		// in order to pass this, we need to increment the counter whenever read
-		if (m_Control.PSTAT)
-			m_SampleCounter++;
 		break;
 
 	case AI_INTERRUPT_TIMING:
@@ -236,6 +237,9 @@ void Write32(const u32 _Value, const u32 _Address)
 
 				// Tell Drive Interface to start/stop streaming
 				DVDInterface::g_bStream = tmpAICtrl.PSTAT;
+
+				CoreTiming::RemoveEvent(et_AI);
+				CoreTiming::ScheduleEvent(((int)GetAIPeriod() / 2), et_AI);
             }
 
             // AI Interrupt
@@ -271,6 +275,8 @@ void Write32(const u32 _Value, const u32 _Address)
 
 	case AI_INTERRUPT_TIMING:		
 		m_InterruptTiming = _Value;
+		CoreTiming::RemoveEvent(et_AI);
+		CoreTiming::ScheduleEvent(((int)GetAIPeriod() / 2), et_AI);
 		DEBUG_LOG(AUDIO_INTERFACE, "Set interrupt: %08x samples", m_InterruptTiming);
 		break;
 
@@ -448,7 +454,7 @@ unsigned int GetAIDSampleRate()
 	return g_AIDSampleRate;
 }
 
-void Update()
+void Update(u64 userdata, int cyclesLate)
 {
     if (m_Control.PSTAT)
     {
@@ -458,8 +464,17 @@ void Update()
             const u32 Samples = static_cast<u32>(Diff / g_CPUCyclesPerSample);
             g_LastCPUTime += Samples * g_CPUCyclesPerSample;
 			IncreaseSampleCount(Samples);
-        } 
+        }
+		CoreTiming::ScheduleEvent(((int)GetAIPeriod() / 2) - cyclesLate, et_AI);
     }
 }
 
+u64 GetAIPeriod()
+{
+	u64 period = g_CPUCyclesPerSample * m_InterruptTiming;
+	if (period == 0)
+		period = 32000 * g_CPUCyclesPerSample;
+	return period;
+}
+
 } // end of namespace AudioInterface

Source/Core/Core/Src/HW/AudioInterface.h

@@ -31,7 +31,7 @@ void Init();
 void Shutdown();	
 void DoState(PointerWrap &p);
 
-void Update();
+void Update(u64 userdata, int cyclesLate);
 
 // Called by DSP emulator
 void Callback_GetSampleRate(unsigned int &_AISampleRate, unsigned int &_DACSampleRate);

Source/Core/Core/Src/HW/DSP.cpp

@@ -436,10 +436,6 @@ void Write16(const u16 _Value, const u32 _Address)
 			if (tmpControl.ARAM) g_dspState.DSPControl.ARAM = 0;
 			if (tmpControl.DSP)  g_dspState.DSPControl.DSP  = 0;
 
-			// Tracking DMAState fixes Knockout Kings 2003 in DSP HLE mode
-			if (GetDSPEmulator()->IsLLE())
-				g_dspState.DSPControl.DMAState = 0;	// keep g_ARAM DMA State zero
-
 			// unknown
 			g_dspState.DSPControl.unk3	= tmpControl.unk3;
 			g_dspState.DSPControl.pad   = tmpControl.pad;
@@ -691,7 +687,7 @@ void UpdateAudioDMA()
 	{
 		// Send silence. Yeah, it's a bit of a waste to sample rate convert
 		// silence.  or hm. Maybe we shouldn't do this :)
-		//dsp_emulator->DSP_SendAIBuffer(0, AudioInterface::GetAIDSampleRate());
+		dsp_emulator->DSP_SendAIBuffer(0, AudioInterface::GetAIDSampleRate());
 	}
 }
 
@@ -701,7 +697,10 @@ void Do_ARAM_DMA()
 	if (!GetDSPEmulator()->IsLLE())
 		g_dspState.DSPControl.DMAState = 1;
 
-	GenerateDSPInterrupt(INT_ARAM, true);
+	if (g_arDMA.Cnt.dir || g_arDMA.Cnt.count > 10240)
+		CoreTiming::ScheduleEvent_Threadsafe(0, et_GenerateDSPInterrupt, INT_ARAM | (1<<16));
+	else
+		GenerateDSPInterrupt(INT_ARAM);
 
 	// Real hardware DMAs in 32byte chunks, but we can get by with 8byte chunks
 	if (g_arDMA.Cnt.dir)

Source/Core/Core/Src/HW/DVDInterface.cpp

@@ -32,11 +32,7 @@
 #include "../Movie.h"
 
 // Disc transfer rate measured in bytes per second
-static const u32 DISC_TRANSFER_RATE_GC = 3125 * 1024;
-static const u32 DISC_TRANSFER_RATE_WII = 7926 * 1024;
-
-// Disc access time measured in milliseconds
-static const u32 DISC_ACCESS_TIME_MS = 128;
+static const u32 DISC_TRANSFER_RATE_GC = 4 * 1024 * 1024;
 
 namespace DVDInterface
 {
@@ -501,8 +497,7 @@ void Write32(const u32 _iValue, const u32 _iAddress)
 				if (!SConfig::GetInstance().m_LocalCoreStartupParameter.bFastDiscSpeed)
 				{
 					u64 ticksUntilTC = m_DILENGTH.Length * 
-						(SystemTimers::GetTicksPerSecond() / (SConfig::GetInstance().m_LocalCoreStartupParameter.bWii?DISC_TRANSFER_RATE_WII:DISC_TRANSFER_RATE_GC)) + 
-						(SystemTimers::GetTicksPerSecond() * DISC_ACCESS_TIME_MS / 1000);
+						(SystemTimers::GetTicksPerSecond() / (SConfig::GetInstance().m_LocalCoreStartupParameter.bWii ? 1 : DISC_TRANSFER_RATE_GC));
 					CoreTiming::ScheduleEvent((int)ticksUntilTC, tc);
 				}
 				else

Source/Core/Core/Src/HW/EXI_Channel.cpp

@@ -25,9 +25,9 @@
 #define EXI_WRITE		1
 #define EXI_READWRITE	2
 
-
 #include "ProcessorInterface.h"
 #include "../PowerPC/PowerPC.h"
+#include "CoreTiming.h"
 
 CEXIChannel::CEXIChannel(u32 ChannelId) :
 	m_DMAMemoryAddress(0),
@@ -45,6 +45,8 @@ CEXIChannel::CEXIChannel(u32 ChannelId) :
 
 	for (int i = 0; i < NUM_DEVICES; i++)
 		m_pDevices[i] = EXIDevice_Create(EXIDEVICE_NONE, m_ChannelId);
+
+	updateInterrupts = CoreTiming::RegisterEvent("EXIInterrupt", UpdateInterrupts);
 }
 
 CEXIChannel::~CEXIChannel()
@@ -88,12 +90,12 @@ void CEXIChannel::AddDevice(IEXIDevice* pDevice, const int device_num, bool noti
 		if (m_ChannelId != 2)
 		{
 			m_Status.EXTINT = 1;
-			UpdateInterrupts();
+			CoreTiming::ScheduleEvent_Threadsafe_Immediate(updateInterrupts, 0);
 		}
 	}
 }
 
-void CEXIChannel::UpdateInterrupts()
+void CEXIChannel::UpdateInterrupts(u64 userdata, int cyclesLate)
 {
 	ExpansionInterface::UpdateInterrupts();
 }
@@ -149,7 +151,9 @@ void CEXIChannel::Read32(u32& _uReturnValue, const u32 _iRegister)
 			if (m_ChannelId == 2)
 				m_Status.EXT = 0;
 			else
+			{
 				m_Status.EXT = GetDevice(1)->IsPresent() ? 1 : 0;
+			}
 
 			_uReturnValue = m_Status.Hex;
 			break;
@@ -213,7 +217,7 @@ void CEXIChannel::Write32(const u32 _iValue, const u32 _iRegister)
 			if (pDevice != NULL)
 				pDevice->SetCS(m_Status.CHIP_SELECT);
 
-			UpdateInterrupts();
+			CoreTiming::ScheduleEvent_Threadsafe_Immediate(updateInterrupts, 0);
 		}
 		break;
 
@@ -264,7 +268,7 @@ void CEXIChannel::Write32(const u32 _iValue, const u32 _iRegister)
 			if(!m_Control.TSTART) // completed !
 			{
 				m_Status.TCINT = 1;
-				UpdateInterrupts();
+				CoreTiming::ScheduleEvent_Threadsafe_Immediate(updateInterrupts, 0);
 			}
 		}
 		break;

Source/Core/Core/Src/HW/EXI_Channel.h

@@ -110,6 +110,9 @@ private:
 	// Since channels operate a bit differently from each other
 	u32 m_ChannelId;
 
+	int updateInterrupts;
+
+	static void UpdateInterrupts(u64 userdata, int cyclesLate);
 public:
 	// get device
 	IEXIDevice* GetDevice(const u8 _CHIP_SELECT);
@@ -129,7 +132,6 @@ public:
 
 	void Update();
 	bool IsCausingInterrupt();
-	void UpdateInterrupts();
 	void DoState(PointerWrap &p);
 	void PauseAndLock(bool doLock, bool unpauseOnUnlock);
 

Source/Core/Core/Src/HW/EXI_DeviceMemoryCard.cpp

@@ -47,6 +47,14 @@ void CEXIMemoryCard::FlushCallback(u64 userdata, int cyclesLate)
 		pThis->Flush();
 }
 
+void CEXIMemoryCard::CmdDoneCallback(u64 userdata, int cyclesLate)
+{
+	int card_index = (int)userdata;
+	CEXIMemoryCard* pThis = (CEXIMemoryCard*)ExpansionInterface::FindDevice(EXIDEVICE_MEMORYCARD, card_index);
+	if (pThis)
+		pThis->CmdDone();
+}
+
 CEXIMemoryCard::CEXIMemoryCard(const int index)
 	: card_index(index)
 	, m_bDirty(false)
@@ -56,7 +64,8 @@ CEXIMemoryCard::CEXIMemoryCard(const int index)
 		m_strFilename = "Movie.raw";
 
 	// we're potentially leaking events here, since there's no UnregisterEvent until emu shutdown, but I guess it's inconsequential
-	et_this_card = CoreTiming::RegisterEvent((card_index == 0) ? "memcardA" : "memcardB", FlushCallback);
+	et_this_card = CoreTiming::RegisterEvent((card_index == 0) ? "memcardFlushA" : "memcardFlushB", FlushCallback);
+	et_cmd_done = CoreTiming::RegisterEvent((card_index == 0) ? "memcardDoneA" : "memcardDoneB", CmdDoneCallback);
  
 	interruptSwitch = 0;
 	m_bInterruptSet = 0;
@@ -178,6 +187,21 @@ bool CEXIMemoryCard::IsPresent()
 	return true;
 }
 
+void CEXIMemoryCard::CmdDone()
+{
+	status |= MC_STATUS_READY;
+	status &= ~MC_STATUS_BUSY;
+
+	m_bInterruptSet = 1;
+	m_bDirty = true;
+}
+
+void CEXIMemoryCard::CmdDoneLater(u64 cycles)
+{
+	CoreTiming::RemoveEvent(et_cmd_done);
+	CoreTiming::ScheduleEvent(cycles, et_cmd_done, (u64)card_index);
+}
+
 void CEXIMemoryCard::SetCS(int cs)
 {
 	// So that memory card won't be invalidated during flushing
@@ -201,11 +225,7 @@ void CEXIMemoryCard::SetCS(int cs)
 
 				//???
 
-				status |= MC_STATUS_READY;
-				status &= ~MC_STATUS_BUSY;
-
-				m_bInterruptSet = 1;
-				m_bDirty = true;
+				CmdDoneLater(5000);
 			}
 			break;
 
@@ -232,11 +252,7 @@ void CEXIMemoryCard::SetCS(int cs)
 					address = (address & ~0x1FF) | ((address+1) & 0x1FF);
 				}
 
-				status |= MC_STATUS_READY;
-				status &= ~MC_STATUS_BUSY;
-
-				m_bInterruptSet = 1;
-				m_bDirty = true;
+				CmdDoneLater(5000);
 			}
 			
 			// Page written to memory card, not just to buffer - let's schedule a flush 0.5b cycles into the future (1 sec)

Source/Core/Core/Src/HW/EXI_DeviceMemoryCard.h

@@ -47,9 +47,18 @@ private:
 	// through the userdata parameter, so that it can then call Flush on the right card.
 	static void FlushCallback(u64 userdata, int cyclesLate);
 
+	// Scheduled when a command that required delayed end signaling is done.
+	static void CmdDoneCallback(u64 userdata, int cyclesLate);
+
 	// Flushes the memory card contents to disk.
 	void Flush(bool exiting = false);
 
+	// Signals that the command that was previously executed is now done.
+	void CmdDone();
+
+	// Variant of CmdDone which schedules an event later in the future to complete the command.
+	void CmdDoneLater(u64 cycles);
+
 	enum 
 	{
 		cmdNintendoID			= 0x00,
@@ -71,7 +80,7 @@ private:
 
 	std::string m_strFilename;
 	int card_index;
-	int et_this_card;
+	int et_this_card, et_cmd_done;
 	//! memory card state
 
 	// STATE_TO_SAVE

Source/Core/Core/Src/HW/GPFifo.cpp

@@ -98,12 +98,17 @@ void STACKALIGN CheckGatherPipe()
 		memmove(m_gatherPipe, m_gatherPipe + cnt, m_gatherPipeCount);
 		
 		// Profile where the FIFO writes are occurring.
-		if (jit && (jit->js.fifoWriteAddresses.find(PC)) == (jit->js.fifoWriteAddresses.end()))
+		if (jit && PC != 0 && (jit->js.fifoWriteAddresses.find(PC)) == (jit->js.fifoWriteAddresses.end()))
 		{
-			jit->js.fifoWriteAddresses.insert(PC);
+			// Log only stores, fp stores and ps stores, filtering out other instructions arrived via optimizeGatherPipe
+			int type = GetOpInfo(Memory::ReadUnchecked_U32(PC))->type;
+			if (type == OPTYPE_STORE || type == OPTYPE_STOREFP || (type == OPTYPE_PS && GetOpInfo(Memory::ReadUnchecked_U32(PC))->opname=="psq_st"))
+			{
+				jit->js.fifoWriteAddresses.insert(PC);
 
-			// Invalidate the JIT block so that it gets recompiled with the external exception check included.
-			jit->GetBlockCache()->InvalidateICache(PC, 4);
+				// Invalidate the JIT block so that it gets recompiled with the external exception check included.
+				jit->GetBlockCache()->InvalidateICache(PC, 4);
+			}
 		}
 	}
 }

Source/Core/Core/Src/HW/ProcessorInterface.cpp

@@ -99,7 +99,7 @@ void Init()
 	m_ResetCode = 0x80000000; // Cold reset
 	m_InterruptCause = INT_CAUSE_RST_BUTTON | INT_CAUSE_VI;
 
-	toggleResetButton = CoreTiming::RegisterEvent("ToggleResetButton", &ToggleResetButtonCallback);
+	toggleResetButton = CoreTiming::RegisterEvent("ToggleResetButton", ToggleResetButtonCallback);
 }
 
 void Read16(u16& _uReturnValue, const u32 _iAddress)

Source/Core/Core/Src/HW/SI_DeviceGCController.cpp

@@ -69,6 +69,19 @@ int CSIDevice_GCController::RunBuffer(u8* _pBuffer, int _iLength)
 		*(u32*)&_pBuffer[0] = SI_GC_CONTROLLER;
 		break;
 
+	case CMD_DIRECT:
+		{
+			INFO_LOG(SERIALINTERFACE, "PAD - Direct (Length: %d)", _iLength);
+			u32 high, low;
+			GetData(high, low);
+			for (int i = 0; i < (_iLength - 1) / 2; i++)
+			{
+				_pBuffer[0 + i] = (high >> (i * 8)) & 0xff;
+				_pBuffer[4 + i] = (low >> (i * 8)) & 0xff;
+			}
+		}
+		break;
+
 	case CMD_ORIGIN:
 		{
 			INFO_LOG(SERIALINTERFACE, "PAD - Get Origin");
@@ -96,7 +109,7 @@ int CSIDevice_GCController::RunBuffer(u8* _pBuffer, int _iLength)
 	default:
 		{
 			ERROR_LOG(SERIALINTERFACE, "unknown SI command     (0x%x)", command);
-			PanicAlert("SI: Unknown command");
+			PanicAlert("SI: Unknown command (0x%x)", command);
 		}			
 		break;
 	}

Source/Core/Core/Src/HW/SI_DeviceGCController.h

@@ -31,6 +31,7 @@ private:
 	enum EBufferCommands
 	{
 		CMD_RESET		= 0x00,
+		CMD_DIRECT		= 0x40,
 		CMD_ORIGIN		= 0x41,
 		CMD_RECALIBRATE	= 0x42,
 	};

Source/Core/Core/Src/HW/SystemTimers.cpp

@@ -74,6 +74,7 @@ IPC_HLE_PERIOD: For the Wiimote this is the call schedule:
 #include "Thread.h"
 #include "Timer.h"
 #include "VideoBackendBase.h"
+#include "CommandProcessor.h"
 
 
 namespace SystemTimers
@@ -109,7 +110,7 @@ enum
 int et_Dec;
 int et_VI;
 int et_SI;
-int et_AI;
+int et_CP;
 int et_AudioDMA;
 int et_DSP;
 int et_IPC_HLE;
@@ -118,15 +119,15 @@ int et_PatchEngine;	// PatchEngine updates every 1/60th of a second by default
 // These are badly educated guesses
 // Feel free to experiment. Set these in Init below.
 int
-	// This one should simply be determined by the increasing counter in AI.
-	AI_PERIOD,
-
 	// These shouldn't be period controlled either, most likely.
 	DSP_PERIOD,
 
 	// This is a fixed value, don't change it 
 	AUDIO_DMA_PERIOD,
 
+	// Regulates the speed of the Command Processor
+	CP_PERIOD,
+
 	// This is completely arbitrary. If we find that we need lower latency, we can just
 	// increase this number.
 	IPC_HLE_PERIOD;
@@ -143,12 +144,6 @@ u32 ConvertMillisecondsToTicks(u32 _Milliseconds)
 	return GetTicksPerSecond() / 1000 * _Milliseconds;
 }
 
-void AICallback(u64 userdata, int cyclesLate)
-{
-	AudioInterface::Update();
-	CoreTiming::ScheduleEvent(AI_PERIOD - cyclesLate, et_AI);
-}
-
 // DSP/CPU timeslicing.
 void DSPCallback(u64 userdata, int cyclesLate)
 {
@@ -187,6 +182,12 @@ void SICallback(u64 userdata, int cyclesLate)
 	CoreTiming::ScheduleEvent(SerialInterface::GetTicksToNextSIPoll() - cyclesLate, et_SI);
 }
 
+void CPCallback(u64 userdata, int cyclesLate)
+{
+	CommandProcessor::Update();
+	CoreTiming::ScheduleEvent(CP_PERIOD - cyclesLate, et_CP);
+}
+
 void DecrementerCallback(u64 userdata, int cyclesLate)
 {
 	PowerPC::ppcState.spr[SPR_DEC] = 0xFFFFFFFF;
@@ -266,12 +267,12 @@ void Init()
 	if (DSP::GetDSPEmulator()->IsLLE())
 		DSP_PERIOD = 12000; // TO BE TWEAKED
 
-	// This is the biggest question mark.
-	AI_PERIOD = GetTicksPerSecond() / 80;
-
 	// System internal sample rate is fixed at 32KHz * 4 (16bit Stereo) / 32 bytes DMA
 	AUDIO_DMA_PERIOD = CPU_CORE_CLOCK / (AudioInterface::GetAIDSampleRate() * 4 / 32);
 
+	// Emulated gekko <-> flipper bus speed ratio (cpu clock / flipper clock)
+	CP_PERIOD = GetTicksPerSecond() / 10000;
+
 	Common::Timer::IncreaseResolution();
 	// store and convert localtime at boot to timebase ticks
 	CoreTiming::SetFakeTBStartValue((u64)(CPU_CORE_CLOCK / TIMER_RATIO) * (u64)CEXIIPL::GetGCTime());
@@ -281,19 +282,21 @@ void Init()
 	CoreTiming::SetFakeDecStartTicks(CoreTiming::GetTicks());
 
 	et_Dec = CoreTiming::RegisterEvent("DecCallback", DecrementerCallback);
-	et_AI = CoreTiming::RegisterEvent("AICallback", AICallback);
 	et_VI = CoreTiming::RegisterEvent("VICallback", VICallback);
 	et_SI = CoreTiming::RegisterEvent("SICallback", SICallback);
+	if (SConfig::GetInstance().m_LocalCoreStartupParameter.bSyncGPU)
+		et_CP = CoreTiming::RegisterEvent("CPCallback", CPCallback);
 	et_DSP = CoreTiming::RegisterEvent("DSPCallback", DSPCallback);
 	et_AudioDMA = CoreTiming::RegisterEvent("AudioDMACallback", AudioDMACallback);
 	et_IPC_HLE = CoreTiming::RegisterEvent("IPC_HLE_UpdateCallback", IPC_HLE_UpdateCallback);
 	et_PatchEngine = CoreTiming::RegisterEvent("PatchEngine", PatchEngineCallback);
 
-	CoreTiming::ScheduleEvent(AI_PERIOD, et_AI);
 	CoreTiming::ScheduleEvent(VideoInterface::GetTicksPerLine(), et_VI);
 	CoreTiming::ScheduleEvent(DSP_PERIOD, et_DSP);
 	CoreTiming::ScheduleEvent(VideoInterface::GetTicksPerFrame(), et_SI);
 	CoreTiming::ScheduleEvent(AUDIO_DMA_PERIOD, et_AudioDMA);
+	if (SConfig::GetInstance().m_LocalCoreStartupParameter.bSyncGPU)
+		CoreTiming::ScheduleEvent(CP_PERIOD, et_CP);
 
 	CoreTiming::ScheduleEvent(VideoInterface::GetTicksPerFrame(), et_PatchEngine);
 

Source/Core/Core/Src/PowerPC/Interpreter/Interpreter_SystemRegisters.cpp

@@ -193,6 +193,7 @@ void Interpreter::mtmsr(UGeckoInstruction _inst)
 {
 	// Privileged?
 	MSR = m_GPR[_inst.RS];
+	PowerPC::CheckExceptions();
 	m_EndBlock = true;
 }
 

Source/Core/Core/Src/PowerPC/Jit64/Jit.cpp

@@ -291,7 +291,9 @@ static void ImHere()
 void Jit64::Cleanup()
 {
 	if (jo.optimizeGatherPipe && js.fifoBytesThisBlock > 0)
+	{
 		ABI_CallFunction((void *)&GPFifo::CheckGatherPipe);
+	}
 
 	// SPEED HACK: MMCR0/MMCR1 should be checked at run-time, not at compile time.
 	if (MMCR0.Hex || MMCR1.Hex)
@@ -301,7 +303,8 @@ void Jit64::Cleanup()
 void Jit64::WriteExit(u32 destination, int exit_num)
 {
 	Cleanup();
-	SUB(32, M(&CoreTiming::downcount), js.downcountAmount > 127 ? Imm32(js.downcountAmount) : Imm8(js.downcountAmount)); 
+
+	SUB(32, M(&CoreTiming::downcount), js.downcountAmount > 127 ? Imm32(js.downcountAmount) : Imm8(js.downcountAmount));
 
 	//If nobody has taken care of this yet (this can be removed when all branches are done)
 	JitBlock *b = js.curBlock;
@@ -561,6 +564,7 @@ const u8* Jit64::DoJit(u32 em_address, PPCAnalyst::CodeBuffer *code_buf, JitBloc
 		if (jo.optimizeGatherPipe && js.fifoBytesThisBlock >= 32)
 		{
 			js.fifoBytesThisBlock -= 32;
+			MOV(32, M(&PC), Imm32(jit->js.compilerPC)); // Helps external systems know which instruction triggered the write
 			ABI_CallFunction(thunks.ProtectFunction((void *)&GPFifo::CheckGatherPipe, 0));
 		}
 

Source/Core/Core/Src/PowerPC/Jit64/Jit_LoadStore.cpp

@@ -302,6 +302,7 @@ void Jit64::stX(UGeckoInstruction inst)
 			addr += offset;
 			if ((addr & 0xFFFFF000) == 0xCC008000 && jo.optimizeGatherPipe)
 			{
+				MOV(32, M(&PC), Imm32(jit->js.compilerPC)); // Helps external systems know which instruction triggered the write
 				gpr.FlushLockX(ABI_PARAM1);
 				MOV(32, R(ABI_PARAM1), gpr.R(s));
 				if (update)
@@ -329,6 +330,7 @@ void Jit64::stX(UGeckoInstruction inst)
 			}
 			else
 			{
+				MOV(32, M(&PC), Imm32(jit->js.compilerPC)); // Helps external systems know which instruction triggered the write
 				switch (accessSize)
 				{
 				case 32: ABI_CallFunctionAC(thunks.ProtectFunction(true ? ((void *)&Memory::Write_U32) : ((void *)&Memory::Write_U32_Swap), 2), gpr.R(s), addr); break;

Source/Core/Core/Src/PowerPC/Jit64/Jit_LoadStorePaired.cpp

@@ -42,6 +42,7 @@ const u8 GC_ALIGNED16(pbswapShuffle2x4[16]) = {3, 2, 1, 0, 7, 6, 5, 4, 8, 9, 10,
 #if 0
 static void WriteDual32(u64 value, u32 address)
 {
+	MOV(32, M(&PC), Imm32(jit->js.compilerPC)); // Helps external systems know which instruction triggered the write
 	Memory::Write_U32((u32)(value >> 32), address);
 	Memory::Write_U32((u32)value, address + 4);
 }

Source/Core/Core/Src/PowerPC/Jit64/Jit_SystemRegisters.cpp

@@ -28,6 +28,7 @@
 
 #include "Jit.h"
 #include "JitRegCache.h"
+#include "HW/ProcessorInterface.h"
 
 void Jit64::mtspr(UGeckoInstruction inst)
 {
@@ -122,7 +123,29 @@ void Jit64::mtmsr(UGeckoInstruction inst)
 	gpr.UnlockAll();
 	gpr.Flush(FLUSH_ALL);
 	fpr.Flush(FLUSH_ALL);
+
+	// If some exceptions are pending and EE are now enabled, force checking
+	// external exceptions when going out of mtmsr in order to execute delayed
+	// interrupts as soon as possible.
+	TEST(32, M(&MSR), Imm32(0x8000));
+	FixupBranch eeDisabled = J_CC(CC_Z);
+
+	TEST(32, M((void*)&PowerPC::ppcState.Exceptions), Imm32(EXCEPTION_EXTERNAL_INT | EXCEPTION_PERFORMANCE_MONITOR | EXCEPTION_DECREMENTER));
+	FixupBranch noExceptionsPending = J_CC(CC_Z);
+
+	// Check if a CP interrupt is waiting and keep the GPU emulation in sync (issue 4336)
+	TEST(32, M((void *)&ProcessorInterface::m_InterruptCause), Imm32(ProcessorInterface::INT_CAUSE_CP));
+	FixupBranch cpInt = J_CC(CC_NZ);
+
+	MOV(32, M(&PC), Imm32(js.compilerPC + 4));
+	WriteExternalExceptionExit();
+
+	SetJumpTarget(cpInt);
+	SetJumpTarget(noExceptionsPending);
+	SetJumpTarget(eeDisabled);
+
 	WriteExit(js.compilerPC + 4, 0);
+
 	js.firstFPInstructionFound = false;
 }
 

Source/Core/Core/Src/PowerPC/Jit64IL/IR.h

@@ -276,8 +276,8 @@ public:
 	InstLoc EmitLoadMSR() {
 		return FoldZeroOp(LoadMSR, 0);
 	}
-	InstLoc EmitStoreMSR(InstLoc val) {
-		return FoldUOp(StoreMSR, val);
+	InstLoc EmitStoreMSR(InstLoc val, InstLoc pc) {
+		return FoldBiOp(StoreMSR, val, pc);
 	}
 	InstLoc EmitStoreFPRF(InstLoc value) {
 		return FoldUOp(StoreFPRF, value);

Source/Core/Core/Src/PowerPC/Jit64IL/IR_X86.cpp

@@ -994,8 +994,26 @@ static void DoWriteCode(IRBuilder* ibuild, JitIL* Jit, bool UseProfile, bool Mak
 			break;
 		}
 		case StoreMSR: {
+			unsigned InstLoc = ibuild->GetImmValue(getOp2(I));
 			regStoreInstToConstLoc(RI, 32, getOp1(I), &MSR);
 			regNormalRegClear(RI, I);
+
+			// If some exceptions are pending and EE are now enabled, force checking
+			// external exceptions when going out of mtmsr in order to execute delayed
+			// interrupts as soon as possible.
+			Jit->MOV(32, R(EAX), M(&MSR));
+			Jit->TEST(32, R(EAX), Imm32(0x8000));
+			FixupBranch eeDisabled = Jit->J_CC(CC_Z);
+
+			Jit->MOV(32, R(EAX), M((void*)&PowerPC::ppcState.Exceptions));
+			Jit->TEST(32, R(EAX), R(EAX));
+			FixupBranch noExceptionsPending = Jit->J_CC(CC_Z);
+
+			Jit->MOV(32, M(&PC), Imm32(InstLoc + 4));
+			Jit->WriteExceptionExit(); // TODO: Implement WriteExternalExceptionExit for JitIL
+
+			Jit->SetJumpTarget(eeDisabled);
+			Jit->SetJumpTarget(noExceptionsPending);
 			break;
 		}
 		case StoreGQR: {

Source/Core/Core/Src/PowerPC/Jit64IL/JitIL.cpp

@@ -389,7 +389,9 @@ static void ImHere()
 void JitIL::Cleanup()
 {
 	if (jo.optimizeGatherPipe && js.fifoBytesThisBlock > 0)
+	{
 		ABI_CallFunction((void *)&GPFifo::CheckGatherPipe);
+	}
 
 	// SPEED HACK: MMCR0/MMCR1 should be checked at run-time, not at compile time.
 	if (MMCR0.Hex || MMCR1.Hex)

Source/Core/Core/Src/PowerPC/Jit64IL/JitIL_SystemRegisters.cpp

@@ -106,7 +106,7 @@ void JitIL::mfspr(UGeckoInstruction inst)
 // --------------
 void JitIL::mtmsr(UGeckoInstruction inst)
 {
-	ibuild.EmitStoreMSR(ibuild.EmitLoadGReg(inst.RS));
+	ibuild.EmitStoreMSR(ibuild.EmitLoadGReg(inst.RS), ibuild.EmitIntConst(js.compilerPC));
 	ibuild.EmitBranchUncond(ibuild.EmitIntConst(js.compilerPC + 4));
 }
 // ==============

Source/Core/Core/Src/PowerPC/JitCommon/JitAsmCommon.cpp

@@ -32,6 +32,7 @@
 #include "../../HW/GPFifo.h"
 #include "../../Core.h"
 #include "JitAsmCommon.h"
+#include "JitBase.h"
 
 using namespace Gen;