Bump CI clang version to 15 + workaround for unsafe fiber optimizations (#982)

.github/workflows/build.yml

@@ -16,7 +16,7 @@ env:
 
 jobs:
   build-ubuntu:
-    runs-on: ubuntu-20.04
+    runs-on: ubuntu-22.04
     steps:
     - name: "Checkout repo"
       uses: actions/checkout@v3
@@ -53,7 +53,7 @@ jobs:
     - name: "Install system dependencies"
       run: |
         sudo apt update -qq
-        sudo apt install -y clang-12 cmake freeglut3-dev libgcrypt20-dev libglm-dev libgtk-3-dev libpulse-dev libsecret-1-dev libsystemd-dev libudev-dev nasm ninja-build
+        sudo apt install -y clang-15 cmake freeglut3-dev libgcrypt20-dev libglm-dev libgtk-3-dev libpulse-dev libsecret-1-dev libsystemd-dev libudev-dev nasm ninja-build
 
     - name: "Bootstrap vcpkg"
       run: |
@@ -75,7 +75,7 @@ jobs:
         
     - name: "cmake"
       run: |
-        cmake -S . -B build ${{ env.BUILD_FLAGS }} -DCMAKE_BUILD_TYPE=${{ env.BUILD_MODE }} -DPORTABLE=OFF -DCMAKE_C_COMPILER=/usr/bin/clang-12 -DCMAKE_CXX_COMPILER=/usr/bin/clang++-12 -G Ninja -DCMAKE_MAKE_PROGRAM=/usr/bin/ninja
+        cmake -S . -B build ${{ env.BUILD_FLAGS }} -DCMAKE_BUILD_TYPE=${{ env.BUILD_MODE }} -DPORTABLE=OFF -DCMAKE_C_COMPILER=/usr/bin/clang-15 -DCMAKE_CXX_COMPILER=/usr/bin/clang++-15 -G Ninja -DCMAKE_MAKE_PROGRAM=/usr/bin/ninja
         
     - name: "Build Cemu"
       run: |
@@ -93,7 +93,7 @@ jobs:
         path: ./bin/Cemu
 
   build-appimage:
-    runs-on: ubuntu-20.04
+    runs-on: ubuntu-22.04
     needs: build-ubuntu
     steps:
     - name: Checkout Upstream Repo
@@ -107,7 +107,7 @@ jobs:
     - name: "Install system dependencies"
       run: |
         sudo apt update -qq
-        sudo apt install -y clang-12 cmake freeglut3-dev libgcrypt20-dev libglm-dev libgtk-3-dev libpulse-dev libsecret-1-dev libsystemd-dev nasm ninja-build appstream
+        sudo apt install -y clang-15 cmake freeglut3-dev libgcrypt20-dev libglm-dev libgtk-3-dev libpulse-dev libsecret-1-dev libsystemd-dev nasm ninja-build appstream
 
     - name: "Build AppImage"
       run: |

.github/workflows/deploy_experimental_release.yml

@@ -10,7 +10,7 @@ jobs:
       experimentalversion: ${{ github.run_number }}
   deploy:
     name: Deploy experimental release
-    runs-on: ubuntu-20.04
+    runs-on: ubuntu-22.04
     needs: call-release-build
     steps:
       - uses: actions/checkout@v3
@@ -72,7 +72,7 @@ jobs:
           ls ./bin/
           cp -R ./bin ./${{ env.CEMU_FOLDER_NAME }}
           mv cemu-bin-linux-x64/Cemu ./${{ env.CEMU_FOLDER_NAME }}/Cemu
-          zip -9 -r upload/cemu-${{ env.CEMU_VERSION }}-ubuntu-20.04-x64.zip ${{ env.CEMU_FOLDER_NAME }}
+          zip -9 -r upload/cemu-${{ env.CEMU_VERSION }}-ubuntu-22.04-x64.zip ${{ env.CEMU_FOLDER_NAME }}
           rm -r ./${{ env.CEMU_FOLDER_NAME }}
           
       - name: Create release from macos-bin

.gitignore

@@ -17,7 +17,7 @@
 .idea/
 
 build/
-cmake-build-*-*/
+cmake-build-*/
 out/
 .cache/
 bin/Cemu_*

BUILD.md

@@ -19,17 +19,17 @@ Any other IDE should also work as long as it has CMake and MSVC support. CLion a
 
 ## Linux
 
-To compile Cemu, a recent enough compiler and STL with C++20 support is required! clang-12 or higher is what we recommend.
+To compile Cemu, a recent enough compiler and STL with C++20 support is required! clang-15 or higher is what we recommend.
 
 ### Installing dependencies
 
 #### For Ubuntu and derivatives:
-`sudo apt install -y cmake curl freeglut3-dev git libgcrypt20-dev libglm-dev libgtk-3-dev libpulse-dev libsecret-1-dev libsystemd-dev nasm ninja-build` 
+`sudo apt install -y cmake curl clang-15 freeglut3-dev git libgcrypt20-dev libglm-dev libgtk-3-dev libpulse-dev libsecret-1-dev libsystemd-dev nasm ninja-build`
 
-*Additionally, for Ubuntu 22.04 only:*
+You may also need to install `libusb-1.0-0-dev` as a workaround for an issue with the vcpkg hidapi package.
- - `sudo apt install -y clang-12`
+
- - At step 3 while building, use  
+At step 3 while building, use:
-   `cmake -S . -B build -DCMAKE_BUILD_TYPE=release -DCMAKE_C_COMPILER=/usr/bin/clang-12 -DCMAKE_CXX_COMPILER=/usr/bin/clang++-12 -G Ninja -DCMAKE_MAKE_PROGRAM=/usr/bin/ninja`
+   `cmake -S . -B build -DCMAKE_BUILD_TYPE=release -DCMAKE_C_COMPILER=/usr/bin/clang-15 -DCMAKE_CXX_COMPILER=/usr/bin/clang++-15 -G Ninja -DCMAKE_MAKE_PROGRAM=/usr/bin/ninja`
 
 #### For Arch and derivatives:
 `sudo pacman -S --needed base-devel clang cmake freeglut git glm gtk3 libgcrypt libpulse libsecret linux-headers llvm nasm ninja systemd unzip zip`

generate_vs_solution.bat

@@ -1,2 +0,0 @@
-"C:\PROGRAM FILES\MICROSOFT VISUAL STUDIO\2022\COMMUNITY\COMMON7\IDE\COMMONEXTENSIONS\MICROSOFT\CMAKE\CMake\bin\cmake.exe" -B build/
-pause

src/Cafe/HW/Espresso/Debugger/Debugger.cpp

@@ -210,7 +210,8 @@ void debugger_handleSingleStepException(uint64 dr6)
 	}
 	if (catchBP)
 	{
-		debugger_createCodeBreakpoint(ppcInterpreterCurrentInstance->instructionPointer + 4, DEBUGGER_BP_T_ONE_SHOT);
+		PPCInterpreter_t* hCPU = PPCInterpreter_getCurrentInstance();
+		debugger_createCodeBreakpoint(hCPU->instructionPointer + 4, DEBUGGER_BP_T_ONE_SHOT);
 	}
 }
 

src/Cafe/HW/Espresso/Debugger/GDBStub.cpp

@@ -959,8 +959,9 @@ void GDBServer::HandleAccessException(uint64 dr6)
 
 	if (!response.empty())
 	{
+		PPCInterpreter_t* hCPU = PPCInterpreter_getCurrentInstance();
 		cemuLog_logDebug(LogType::Force, "Received matching breakpoint exception: {}", response);
-		auto nextInstructions = findNextInstruction(ppcInterpreterCurrentInstance->instructionPointer, ppcInterpreterCurrentInstance->spr.LR, ppcInterpreterCurrentInstance->spr.CTR);
+		auto nextInstructions = findNextInstruction(hCPU->instructionPointer, hCPU->spr.LR, hCPU->spr.CTR);
 		for (MPTR nextInstr : nextInstructions)
 		{
 			auto bpIt = m_patchedInstructions.find(nextInstr);

src/Cafe/HW/Espresso/Interpreter/PPCInterpreterMain.cpp

@@ -6,7 +6,6 @@
 thread_local PPCInterpreter_t* ppcInterpreterCurrentInstance;
 
 // main thread instruction counter and timing
-volatile uint64 ppcMainThreadCycleCounter = 0;
 uint64 ppcMainThreadDECCycleValue = 0; // value that was set to dec register
 uint64 ppcMainThreadDECCycleStart = 0; // at which cycle the dec register was set, if == 0 -> dec is 0
 uint64 ppcCyclesSince2000 = 0;
@@ -29,11 +28,16 @@ PPCInterpreter_t* PPCInterpreter_createInstance(unsigned int Entrypoint)
 	return pData;
 }
 
-PPCInterpreter_t* PPCInterpreter_getCurrentInstance()
+TLS_WORKAROUND_NOINLINE PPCInterpreter_t* PPCInterpreter_getCurrentInstance()
 {
 	return ppcInterpreterCurrentInstance;
 }
 
+TLS_WORKAROUND_NOINLINE void PPCInterpreter_setCurrentInstance(PPCInterpreter_t* hCPU)
+{
+	ppcInterpreterCurrentInstance = hCPU;
+}
+
 uint64 PPCInterpreter_getMainCoreCycleCounter()
 {
 	return PPCTimer_getFromRDTSC();
@@ -78,24 +82,25 @@ uint32 PPCInterpreter_getCoreIndex(PPCInterpreter_t* hCPU)
 
 uint32 PPCInterpreter_getCurrentCoreIndex()
 {
-	return ppcInterpreterCurrentInstance->spr.UPIR;
+	return PPCInterpreter_getCurrentInstance()->spr.UPIR;
 };
 
 uint8* PPCInterpreterGetStackPointer()
 {
-	return memory_getPointerFromVirtualOffset(ppcInterpreterCurrentInstance->gpr[1]);
+	return memory_getPointerFromVirtualOffset(PPCInterpreter_getCurrentInstance()->gpr[1]);
 }
 
 uint8* PPCInterpreterGetAndModifyStackPointer(sint32 offset)
 {
-	uint8* result = memory_getPointerFromVirtualOffset(ppcInterpreterCurrentInstance->gpr[1] - offset);
+	PPCInterpreter_t* hCPU = PPCInterpreter_getCurrentInstance();
-	ppcInterpreterCurrentInstance->gpr[1] -= offset;
+	uint8* result = memory_getPointerFromVirtualOffset(hCPU->gpr[1] - offset);
+	hCPU->gpr[1] -= offset;
 	return result;
 }
 
 void PPCInterpreterModifyStackPointer(sint32 offset)
 {
-	ppcInterpreterCurrentInstance->gpr[1] -= offset;
+	PPCInterpreter_getCurrentInstance()->gpr[1] -= offset;
 }
 
 uint32 RPLLoader_MakePPCCallable(void(*ppcCallableExport)(PPCInterpreter_t* hCPU));

src/Cafe/HW/Espresso/PPCCallback.h

@@ -18,19 +18,20 @@ uint32 PPCCoreCallback(MPTR function, PPCCoreCallbackData_t& data, T currentArg,
 {
 	cemu_assert_debug(data.gprCount <= 8);
 	cemu_assert_debug(data.floatCount <= 8);
+	PPCInterpreter_t* hCPU = PPCInterpreter_getCurrentInstance();
 	if constexpr (std::is_pointer_v<T>)
 	{
-		ppcInterpreterCurrentInstance->gpr[3 + data.gprCount] = MEMPTR(currentArg).GetMPTR();
+		hCPU->gpr[3 + data.gprCount] = MEMPTR(currentArg).GetMPTR();
 		data.gprCount++;
 	}
 	else if constexpr (std::is_base_of_v<MEMPTRBase, std::remove_reference_t<T>>)
 	{
-		ppcInterpreterCurrentInstance->gpr[3 + data.gprCount] = currentArg.GetMPTR();
+		hCPU->gpr[3 + data.gprCount] = currentArg.GetMPTR();
 		data.gprCount++;
 	}
 	else if constexpr (std::is_reference_v<T>)
 	{
-		ppcInterpreterCurrentInstance->gpr[3 + data.gprCount] = MEMPTR(&currentArg).GetMPTR();
+		hCPU->gpr[3 + data.gprCount] = MEMPTR(&currentArg).GetMPTR();
 		data.gprCount++;
 	}
 	else if constexpr(std::is_enum_v<T>)
@@ -40,19 +41,19 @@ uint32 PPCCoreCallback(MPTR function, PPCCoreCallbackData_t& data, T currentArg,
 	}
 	else if constexpr (std::is_floating_point_v<T>)
 	{
-		ppcInterpreterCurrentInstance->fpr[1 + data.floatCount].fpr = (double)currentArg;
+		hCPU->fpr[1 + data.floatCount].fpr = (double)currentArg;
 		data.floatCount++;
 	}
 	else if constexpr (std::is_integral_v<T> && sizeof(T) == sizeof(uint64))
 	{
-		ppcInterpreterCurrentInstance->gpr[3 + data.gprCount] = (uint32)(currentArg >> 32); // high
+		hCPU->gpr[3 + data.gprCount] = (uint32)(currentArg >> 32); // high
-		ppcInterpreterCurrentInstance->gpr[3 + data.gprCount + 1] = (uint32)currentArg; // low
+		hCPU->gpr[3 + data.gprCount + 1] = (uint32)currentArg; // low
 
 		data.gprCount += 2;
 	}
 	else
 	{
-		ppcInterpreterCurrentInstance->gpr[3 + data.gprCount] = (uint32)currentArg;
+		hCPU->gpr[3 + data.gprCount] = (uint32)currentArg;
 		data.gprCount++;
 	}
 	

src/Cafe/HW/Espresso/PPCScheduler.cpp

@@ -11,21 +11,24 @@ uint32 ppcThreadQuantum = 45000; // execute 45000 instructions before thread res
 
 void PPCInterpreter_relinquishTimeslice()
 {
-	if( ppcInterpreterCurrentInstance->remainingCycles >= 0 )
+	PPCInterpreter_t* hCPU = PPCInterpreter_getCurrentInstance();
+	if( hCPU->remainingCycles >= 0 )
 	{
-		ppcInterpreterCurrentInstance->skippedCycles = ppcInterpreterCurrentInstance->remainingCycles + 1;
+		hCPU->skippedCycles = hCPU->remainingCycles + 1;
-		ppcInterpreterCurrentInstance->remainingCycles = -1;
+		hCPU->remainingCycles = -1;
 	}
 }
 
 void PPCCore_boostQuantum(sint32 numCycles)
 {
-	ppcInterpreterCurrentInstance->remainingCycles += numCycles;
+	PPCInterpreter_t* hCPU = PPCInterpreter_getCurrentInstance();
+	hCPU->remainingCycles += numCycles;
 }
 
 void PPCCore_deboostQuantum(sint32 numCycles)
 {
-	ppcInterpreterCurrentInstance->remainingCycles -= numCycles;
+	PPCInterpreter_t* hCPU = PPCInterpreter_getCurrentInstance();
+	hCPU->remainingCycles -= numCycles;
 }
 
 namespace coreinit
@@ -36,7 +39,7 @@ namespace coreinit
 void PPCCore_switchToScheduler()
 {
 	cemu_assert_debug(__OSHasSchedulerLock() == false); // scheduler lock must not be hold past thread time slice
-	cemu_assert_debug(ppcInterpreterCurrentInstance->coreInterruptMask != 0 || CafeSystem::GetForegroundTitleId() == 0x000500001019e600);
+	cemu_assert_debug(PPCInterpreter_getCurrentInstance()->coreInterruptMask != 0 || CafeSystem::GetForegroundTitleId() == 0x000500001019e600);
 	__OSLockScheduler();
 	coreinit::__OSThreadSwitchToNext();
 	__OSUnlockScheduler();
@@ -45,7 +48,7 @@ void PPCCore_switchToScheduler()
 void PPCCore_switchToSchedulerWithLock()
 {
 	cemu_assert_debug(__OSHasSchedulerLock() == true); // scheduler lock must be hold
-	cemu_assert_debug(ppcInterpreterCurrentInstance->coreInterruptMask != 0 || CafeSystem::GetForegroundTitleId() == 0x000500001019e600);
+	cemu_assert_debug(PPCInterpreter_getCurrentInstance()->coreInterruptMask != 0 || CafeSystem::GetForegroundTitleId() == 0x000500001019e600);
 	coreinit::__OSThreadSwitchToNext();
 }
 
@@ -58,7 +61,7 @@ void _PPCCore_callbackExit(PPCInterpreter_t* hCPU)
 PPCInterpreter_t* PPCCore_executeCallbackInternal(uint32 functionMPTR)
 {
 	cemu_assert_debug(functionMPTR != 0);
-	PPCInterpreter_t* hCPU = ppcInterpreterCurrentInstance;
+	PPCInterpreter_t* hCPU = PPCInterpreter_getCurrentInstance();
 	// remember LR and instruction pointer
 	uint32 lr = hCPU->spr.LR;
 	uint32 ip = hCPU->instructionPointer;

src/Cafe/HW/Espresso/PPCSchedulerLLE.cpp

@@ -220,7 +220,7 @@ void PPCCoreLLE_startSingleCoreScheduler(uint32 entrypoint)
 		for (uint32 coreIndex = 0; coreIndex < 3; coreIndex++)
 		{
 			PPCInterpreter_t* hCPU = cpuContext->cores+coreIndex;
-			ppcInterpreterCurrentInstance = hCPU;
+			PPCInterpreter_setCurrentInstance(hCPU);
 			if (coreIndex == 1)
 			{
 				// check SCR core 1 enable bit

src/Cafe/HW/Espresso/PPCState.h

@@ -149,6 +149,7 @@ static uint64 PPCInterpreter_getCallParamU64(PPCInterpreter_t* hCPU, uint32 inde
 
 PPCInterpreter_t* PPCInterpreter_createInstance(unsigned int Entrypoint);
 PPCInterpreter_t* PPCInterpreter_getCurrentInstance();
+void PPCInterpreter_setCurrentInstance(PPCInterpreter_t* hCPU);
 
 uint64 PPCInterpreter_getMainCoreCycleCounter();
 
@@ -192,7 +193,6 @@ uint32 PPCInterpreter_getCurrentCoreIndex();
 void PPCInterpreter_setDEC(PPCInterpreter_t* hCPU, uint32 newValue);
 
 // timing for main processor
-extern volatile uint64 ppcMainThreadCycleCounter;
 extern uint64 ppcCyclesSince2000; // on init this is set to the cycles that passed since 1.1.2000
 extern uint64 ppcCyclesSince2000TimerClock; // on init this is set to the cycles that passed since 1.1.2000 / 20
 extern uint64 ppcCyclesSince2000_UTC;
@@ -213,7 +213,6 @@ void PPCTimer_start();
 // core info and control
 extern uint32 ppcThreadQuantum;
 
-extern thread_local PPCInterpreter_t *ppcInterpreterCurrentInstance;
 uint8* PPCInterpreterGetAndModifyStackPointer(sint32 offset);
 uint8* PPCInterpreterGetStackPointer();
 void PPCInterpreterModifyStackPointer(sint32 offset);

src/Cafe/HW/Espresso/Recompiler/PPCRecompilerX64.cpp

@@ -100,7 +100,7 @@ void* ATTR_MS_ABI PPCRecompiler_virtualHLE(PPCInterpreter_t* hCPU, uint32 hleFun
 		hCPU->remainingCycles -= 500; // let subtract about 500 cycles for each HLE call
 		hCPU->gpr[3] = 0;
 		PPCInterpreter_nextInstruction(hCPU);
-		return ppcInterpreterCurrentInstance;
+		return hCPU;
 	}
 	else
 	{
@@ -109,7 +109,7 @@ void* ATTR_MS_ABI PPCRecompiler_virtualHLE(PPCInterpreter_t* hCPU, uint32 hleFun
 		hleCall(hCPU);
 	}
 	hCPU->rspTemp = prevRSPTemp;
-	return ppcInterpreterCurrentInstance;
+	return PPCInterpreter_getCurrentInstance();
 }
 
 void ATTR_MS_ABI PPCRecompiler_getTBL(PPCInterpreter_t* hCPU, uint32 gprIndex)

src/Cafe/IOSU/legacy/iosu_ioctl.cpp

@@ -23,7 +23,7 @@ sint32 iosuIoctl_pushAndWait(uint32 ioctlHandle, ioQueueEntry_t* ioQueueEntry)
 	}
 	__OSLockScheduler();
 	ioctlMutex.lock();
-	ioQueueEntry->ppcThread = coreinitThread_getCurrentThreadDepr(ppcInterpreterCurrentInstance);
+	ioQueueEntry->ppcThread = coreinitThread_getCurrentThreadDepr(PPCInterpreter_getCurrentInstance());
 	
 	_ioctlRingbuffer[ioctlHandle].Push(ioQueueEntry);
 	ioctlMutex.unlock();

src/Cafe/OS/common/OSUtil.h

@@ -65,7 +65,7 @@ public:
 		}
 		else if constexpr (std::is_floating_point_v<T>)
 		{
-			v = (T)ppcInterpreterCurrentInstance->fpr[1 + fprIndex].fpr;
+			v = (T)hCPU->fpr[1 + fprIndex].fpr;
 			fprIndex++;
 		}
 		else

src/Cafe/OS/libs/coreinit/coreinit.cpp

@@ -204,7 +204,7 @@ namespace coreinit
 {
 	sint32 OSGetCoreId()
 	{
-		return PPCInterpreter_getCoreIndex(ppcInterpreterCurrentInstance);
+		return PPCInterpreter_getCoreIndex(PPCInterpreter_getCurrentInstance());
 	}
 
 	uint32 OSGetCoreCount()
@@ -239,7 +239,7 @@ namespace coreinit
 
 	uint32 OSGetStackPointer()
 	{
-		return ppcInterpreterCurrentInstance->gpr[1];
+		return PPCInterpreter_getCurrentInstance()->gpr[1];
 	}
 
 	void coreinitExport_ENVGetEnvironmentVariable(PPCInterpreter_t* hCPU)

src/Cafe/OS/libs/coreinit/coreinit_FS.cpp

@@ -691,7 +691,7 @@ namespace coreinit
 			while (OSSendMessage(ioMsgQueue, &fsCmdBlockBody->asyncResult.msgUnion.osMsg, 0) == 0)
 			{
 				cemuLog_log(LogType::Force, "FS driver: Failed to add message to result queue. Retrying...");
-				if (ppcInterpreterCurrentInstance)
+				if (PPCInterpreter_getCurrentInstance())
 					PPCCore_switchToScheduler();
 				else
 					std::this_thread::sleep_for(std::chrono::milliseconds(10));

src/Cafe/OS/libs/coreinit/coreinit_Init.cpp

@@ -142,10 +142,11 @@ void CafeInit()
 		}
 	}
 	// setup UGQR
-	ppcInterpreterCurrentInstance->spr.UGQR[0 + 2] = 0x00040004;
+	PPCInterpreter_t* hCPU = PPCInterpreter_getCurrentInstance();
-	ppcInterpreterCurrentInstance->spr.UGQR[0 + 3] = 0x00050005;
+	hCPU->spr.UGQR[0 + 2] = 0x00040004;
-	ppcInterpreterCurrentInstance->spr.UGQR[0 + 4] = 0x00060006;
+	hCPU->spr.UGQR[0 + 3] = 0x00050005;
-	ppcInterpreterCurrentInstance->spr.UGQR[0 + 5] = 0x00070007;
+	hCPU->spr.UGQR[0 + 4] = 0x00060006;
+	hCPU->spr.UGQR[0 + 5] = 0x00070007;
 	coreinit::InitForegroundBucket();
 	coreinit::InitSysHeap();
 }

src/Cafe/OS/libs/coreinit/coreinit_Misc.cpp

@@ -235,7 +235,7 @@ namespace coreinit
 
 	sint32 __os_snprintf(char* outputStr, sint32 maxLength, const char* formatStr)
 	{
-		sint32 r = ppcSprintf(formatStr, outputStr, maxLength, ppcInterpreterCurrentInstance, 3);
+		sint32 r = ppcSprintf(formatStr, outputStr, maxLength, PPCInterpreter_getCurrentInstance(), 3);
 		return r;
 	}
 
@@ -303,7 +303,7 @@ namespace coreinit
 	void OSReport(const char* format)
 	{
 		char buffer[1024 * 2];
-		sint32 len = ppcSprintf(format, buffer, sizeof(buffer), ppcInterpreterCurrentInstance, 1);
+		sint32 len = ppcSprintf(format, buffer, sizeof(buffer), PPCInterpreter_getCurrentInstance(), 1);
 		WriteCafeConsole(CafeLogType::OSCONSOLE, buffer, len);
 	}
 
@@ -316,7 +316,7 @@ namespace coreinit
 	{
 		char buffer[1024 * 2];
 		int prefixLen = sprintf(buffer, "[COSWarn-%d] ", moduleId);
-		sint32 len = ppcSprintf(format, buffer + prefixLen, sizeof(buffer) - prefixLen, ppcInterpreterCurrentInstance, 2);
+		sint32 len = ppcSprintf(format, buffer + prefixLen, sizeof(buffer) - prefixLen, PPCInterpreter_getCurrentInstance(), 2);
 		WriteCafeConsole(CafeLogType::OSCONSOLE, buffer, len + prefixLen);
 	}
 
@@ -324,7 +324,7 @@ namespace coreinit
 	{
 		char buffer[1024 * 2];
 		int prefixLen = sprintf(buffer, "[OSLogPrintf-%d-%d-%d] ", ukn1, ukn2, ukn3);
-		sint32 len = ppcSprintf(format, buffer + prefixLen, sizeof(buffer) - prefixLen, ppcInterpreterCurrentInstance, 4);
+		sint32 len = ppcSprintf(format, buffer + prefixLen, sizeof(buffer) - prefixLen, PPCInterpreter_getCurrentInstance(), 4);
 		WriteCafeConsole(CafeLogType::OSCONSOLE, buffer, len + prefixLen);
 	}
 

src/Cafe/OS/libs/coreinit/coreinit_Thread.cpp

@@ -305,7 +305,7 @@ namespace coreinit
 		affinityMask = attr & 0x7;
 		// if no core is selected -> set current one
 		if (affinityMask == 0)
-			affinityMask |= (1 << PPCInterpreter_getCoreIndex(ppcInterpreterCurrentInstance));
+			affinityMask |= (1 << PPCInterpreter_getCoreIndex(PPCInterpreter_getCurrentInstance()));
 		// set attr
 		// todo: Support for other attr bits
 		thread->attr = (affinityMask & 0xFF) | (attr & OSThread_t::ATTR_BIT::ATTR_DETACHED);
@@ -325,7 +325,7 @@ namespace coreinit
 	{
 		__OSLockScheduler();
 
-		cemu_assert_debug(ppcInterpreterCurrentInstance == nullptr || OSGetCurrentThread() != thread); // called on self, what should this function do?
+		cemu_assert_debug(PPCInterpreter_getCurrentInstance() == nullptr || OSGetCurrentThread() != thread); // called on self, what should this function do?
 
 		if (thread->state != OSThread_t::THREAD_STATE::STATE_NONE && thread->state != OSThread_t::THREAD_STATE::STATE_MORIBUND)
 		{
@@ -607,7 +607,7 @@ namespace coreinit
 			// todo - only set this once?
 			thread->wakeUpTime = PPCInterpreter_getMainCoreCycleCounter();
 			// reschedule if thread has higher priority
-			if (ppcInterpreterCurrentInstance && __OSCoreShouldSwitchToThread(coreinit::OSGetCurrentThread(), thread))
+			if (PPCInterpreter_getCurrentInstance() && __OSCoreShouldSwitchToThread(coreinit::OSGetCurrentThread(), thread))
 				PPCCore_switchToSchedulerWithLock();
 		}
 		return previousSuspendCount;
@@ -930,17 +930,17 @@ namespace coreinit
 		thread->requestFlags = (OSThread_t::REQUEST_FLAG_BIT)(thread->requestFlags & OSThread_t::REQUEST_FLAG_CANCEL); // remove all flags except cancel flag
 
 		// update total cycles
-		uint64 remainingCycles = std::min((uint64)ppcInterpreterCurrentInstance->remainingCycles, (uint64)thread->quantumTicks);
+		uint64 remainingCycles = std::min((uint64)hCPU->remainingCycles, (uint64)thread->quantumTicks);
 		uint64 executedCycles = thread->quantumTicks - remainingCycles;
-		if (executedCycles < ppcInterpreterCurrentInstance->skippedCycles)
+		if (executedCycles < hCPU->skippedCycles)
 			executedCycles = 0;
 		else
-			executedCycles -= ppcInterpreterCurrentInstance->skippedCycles;
+			executedCycles -= hCPU->skippedCycles;
 		thread->totalCycles += executedCycles;
 		// store context and set current thread to null
 		__OSThreadStoreContext(hCPU, thread);
 		OSSetCurrentThread(OSGetCoreId(), nullptr);
-		ppcInterpreterCurrentInstance = nullptr;
+		PPCInterpreter_setCurrentInstance(nullptr);
 	}
 
 	void __OSLoadThread(OSThread_t* thread, PPCInterpreter_t* hCPU, uint32 coreIndex)
@@ -951,7 +951,7 @@ namespace coreinit
 		hCPU->reservedMemValue = 0;
 		hCPU->spr.UPIR = coreIndex;
 		hCPU->coreInterruptMask = 1;
-		ppcInterpreterCurrentInstance = hCPU;
+		PPCInterpreter_setCurrentInstance(hCPU);
 		OSSetCurrentThread(OSGetCoreId(), thread);
 		__OSThreadLoadContext(hCPU, thread);
 		thread->context.upir = coreIndex;
@@ -1076,7 +1076,7 @@ namespace coreinit
 
 		// store context of current thread
 		__OSStoreThread(OSGetCurrentThread(), &hostThread->ppcInstance);
-		cemu_assert_debug(ppcInterpreterCurrentInstance == nullptr);
+		cemu_assert_debug(PPCInterpreter_getCurrentInstance() == nullptr);
 
 		if (!sSchedulerActive.load(std::memory_order::relaxed))
 		{
@@ -1165,7 +1165,7 @@ namespace coreinit
 		
 		// create scheduler idle fiber and switch to it
 		g_idleLoopFiber[t_assignedCoreIndex] = new Fiber(__OSThreadCoreIdle, nullptr, nullptr);
-		cemu_assert_debug(ppcInterpreterCurrentInstance == nullptr);
+		cemu_assert_debug(PPCInterpreter_getCurrentInstance() == nullptr);
 		__OSLockScheduler();
 		Fiber::Switch(*g_idleLoopFiber[t_assignedCoreIndex]);
 		// returned from scheduler loop, exit thread

src/Cafe/OS/libs/coreinit/coreinit_ThreadQueue.cpp

@@ -139,7 +139,7 @@ namespace coreinit
 			thread->state = OSThread_t::THREAD_STATE::STATE_READY;
 			thread->currentWaitQueue = nullptr;
 			coreinit::__OSAddReadyThreadToRunQueue(thread);
-			if (reschedule && thread->suspendCounter == 0 && ppcInterpreterCurrentInstance && __OSCoreShouldSwitchToThread(coreinit::OSGetCurrentThread(), thread))
+			if (reschedule && thread->suspendCounter == 0 && PPCInterpreter_getCurrentInstance() && __OSCoreShouldSwitchToThread(coreinit::OSGetCurrentThread(), thread))
 				shouldReschedule = true;
 		}
 		if (shouldReschedule)
@@ -159,7 +159,7 @@ namespace coreinit
 			thread->state = OSThread_t::THREAD_STATE::STATE_READY;
 			thread->currentWaitQueue = nullptr;
 			coreinit::__OSAddReadyThreadToRunQueue(thread);
-			if (reschedule && thread->suspendCounter == 0 && ppcInterpreterCurrentInstance && __OSCoreShouldSwitchToThread(coreinit::OSGetCurrentThread(), thread))
+			if (reschedule && thread->suspendCounter == 0 && PPCInterpreter_getCurrentInstance() && __OSCoreShouldSwitchToThread(coreinit::OSGetCurrentThread(), thread))
 				shouldReschedule = true;
 		}
 		if (shouldReschedule)

src/Cafe/OS/libs/gx2/GX2.cpp

@@ -4,8 +4,8 @@
 #include "GX2.h"
 #include "Cafe/HW/Latte/Core/Latte.h"
 #include "Cafe/OS/libs/coreinit/coreinit_Time.h"
+#include "Cafe/OS/libs/coreinit/coreinit_Thread.h"
 #include "Cafe/CafeSystem.h"
-
 #include "Cafe/HW/Latte/Core/LattePM4.h"
 
 #include "GX2_Command.h"
@@ -68,7 +68,7 @@ void gx2Export_GX2SwapScanBuffers(PPCInterpreter_t* hCPU)
 
 	// Orochi Warriors seems to call GX2SwapScanBuffers on arbitrary threads/cores. The PM4 commands should go through to the GPU as long as there is no active display list and no other core is submitting commands simultaneously
 	// right now, we work around this by avoiding the infinite loop below (request counter incremented, but PM4 not sent)
-	uint32 coreIndex = PPCInterpreter_getCoreIndex(ppcInterpreterCurrentInstance);
+	uint32 coreIndex = coreinit::OSGetCoreId();
 	if (GX2::sGX2MainCoreIndex == coreIndex)
 		LatteGPUState.sharedArea->flipRequestCountBE = _swapEndianU32(_swapEndianU32(LatteGPUState.sharedArea->flipRequestCountBE) + 1);
 
@@ -332,7 +332,7 @@ uint64 Latte_GetTime()
 
 void _GX2SubmitToTCL()
 {
-	uint32 coreIndex = PPCInterpreter_getCoreIndex(ppcInterpreterCurrentInstance);
+	uint32 coreIndex = PPCInterpreter_getCoreIndex(PPCInterpreter_getCurrentInstance());
 	// do nothing if called from non-main GX2 core
 	if (GX2::sGX2MainCoreIndex != coreIndex)
 	{
@@ -373,7 +373,7 @@ uint32 _GX2GetUnflushedBytes(uint32 coreIndex)
  */
 void GX2ReserveCmdSpace(uint32 reservedFreeSpaceInU32)
 {
-	uint32 coreIndex = PPCInterpreter_getCoreIndex(ppcInterpreterCurrentInstance);
+	uint32 coreIndex = coreinit::OSGetCoreId();
 	// if we are in a display list then do nothing
 	if( gx2WriteGatherPipe.displayListStart[coreIndex] != MPTR_NULL )
 		return;

src/Cafe/OS/libs/gx2/GX2_Command.cpp

@@ -3,6 +3,7 @@
 #include "Cafe/OS/common/OSCommon.h"
 #include "Cafe/HW/Latte/Core/LattePM4.h"
 #include "Cafe/OS/libs/coreinit/coreinit.h"
+#include "Cafe/OS/libs/coreinit/coreinit_Thread.h"
 #include "Cafe/HW/Latte/ISA/RegDefines.h"
 #include "GX2.h"
 #include "GX2_Command.h"
@@ -15,7 +16,7 @@ GX2WriteGatherPipeState gx2WriteGatherPipe = { 0 };
 
 void gx2WriteGather_submitU32AsBE(uint32 v)
 {
-	uint32 coreIndex = PPCInterpreter_getCoreIndex(ppcInterpreterCurrentInstance);
+	uint32 coreIndex = PPCInterpreter_getCoreIndex(PPCInterpreter_getCurrentInstance());
 	if (gx2WriteGatherPipe.writeGatherPtrWrite[coreIndex] == NULL)
 		return;
 	*(uint32*)(*gx2WriteGatherPipe.writeGatherPtrWrite[coreIndex]) = _swapEndianU32(v);
@@ -24,7 +25,7 @@ void gx2WriteGather_submitU32AsBE(uint32 v)
 
 void gx2WriteGather_submitU32AsLE(uint32 v)
 {
-	uint32 coreIndex = PPCInterpreter_getCoreIndex(ppcInterpreterCurrentInstance);
+	uint32 coreIndex = PPCInterpreter_getCoreIndex(PPCInterpreter_getCurrentInstance());
 	if (gx2WriteGatherPipe.writeGatherPtrWrite[coreIndex] == NULL)
 		return;
 	*(uint32*)(*gx2WriteGatherPipe.writeGatherPtrWrite[coreIndex]) = v;
@@ -33,7 +34,7 @@ void gx2WriteGather_submitU32AsLE(uint32 v)
 
 void gx2WriteGather_submitU32AsLEArray(uint32* v, uint32 numValues)
 {
-	uint32 coreIndex = PPCInterpreter_getCoreIndex(ppcInterpreterCurrentInstance);
+	uint32 coreIndex = PPCInterpreter_getCoreIndex(PPCInterpreter_getCurrentInstance());
 	if (gx2WriteGatherPipe.writeGatherPtrWrite[coreIndex] == NULL)
 		return;
 	memcpy_dwords((*gx2WriteGatherPipe.writeGatherPtrWrite[coreIndex]), v, numValues);
@@ -134,7 +135,7 @@ namespace GX2
 
 	bool GX2GetCurrentDisplayList(betype<MPTR>* displayListAddr, uint32be* displayListSize)
 	{
-		uint32 coreIndex = PPCInterpreter_getCoreIndex(ppcInterpreterCurrentInstance);
+		uint32 coreIndex = coreinit::OSGetCoreId();
 		if (gx2WriteGatherPipe.displayListStart[coreIndex] == MPTR_NULL)
 			return false;
 
@@ -149,13 +150,13 @@ namespace GX2
 	bool GX2GetDisplayListWriteStatus()
 	{
 		// returns true if we are writing to a display list
-		uint32 coreIndex = PPCInterpreter_getCoreIndex(ppcInterpreterCurrentInstance);
+		uint32 coreIndex = coreinit::OSGetCoreId();
 		return gx2WriteGatherPipe.displayListStart[coreIndex] != MPTR_NULL;
 	}
 
 	bool GX2WriteGather_isDisplayListActive()
 	{
-		uint32 coreIndex = PPCInterpreter_getCoreIndex(ppcInterpreterCurrentInstance);
+		uint32 coreIndex = coreinit::OSGetCoreId();
 		if (gx2WriteGatherPipe.displayListStart[coreIndex] != MPTR_NULL)
 			return true;
 		return false;
@@ -171,7 +172,7 @@ namespace GX2
 
 	void GX2WriteGather_checkAndInsertWrapAroundMark()
 	{
-		uint32 coreIndex = PPCInterpreter_getCoreIndex(ppcInterpreterCurrentInstance);
+		uint32 coreIndex = coreinit::OSGetCoreId();
 		if (coreIndex != sGX2MainCoreIndex) // only if main gx2 core
 			return;
 		if (gx2WriteGatherPipe.displayListStart[coreIndex] != MPTR_NULL)
@@ -187,18 +188,18 @@ namespace GX2
 
 	void GX2BeginDisplayList(MEMPTR<void> displayListAddr, uint32 size)
 	{
-		GX2WriteGather_beginDisplayList(ppcInterpreterCurrentInstance, displayListAddr.GetMPTR(), size);
+		GX2WriteGather_beginDisplayList(PPCInterpreter_getCurrentInstance(), displayListAddr.GetMPTR(), size);
 	}
 
 	void GX2BeginDisplayListEx(MEMPTR<void> displayListAddr, uint32 size, bool profiling)
 	{
-		GX2WriteGather_beginDisplayList(ppcInterpreterCurrentInstance, displayListAddr.GetMPTR(), size);
+		GX2WriteGather_beginDisplayList(PPCInterpreter_getCurrentInstance(), displayListAddr.GetMPTR(), size);
 	}
 
 	uint32 GX2EndDisplayList(MEMPTR<void> displayListAddr)
 	{
 		cemu_assert_debug(displayListAddr != nullptr);
-		uint32 displayListSize = GX2WriteGather_endDisplayList(ppcInterpreterCurrentInstance, displayListAddr.GetMPTR());
+		uint32 displayListSize = GX2WriteGather_endDisplayList(PPCInterpreter_getCurrentInstance(), displayListAddr.GetMPTR());
 		return displayListSize;
 	}
 
@@ -220,7 +221,7 @@ namespace GX2
 		// its basically a way to manually submit a command buffer to the GPU
 		// as such it also affects the submission and retire timestamps
 
-		uint32 coreIndex = PPCInterpreter_getCoreIndex(ppcInterpreterCurrentInstance);
+		uint32 coreIndex = PPCInterpreter_getCoreIndex(PPCInterpreter_getCurrentInstance());
 		cemu_assert_debug(coreIndex == sGX2MainCoreIndex);
 		coreIndex = sGX2MainCoreIndex; // always submit to main queue which is owned by GX2 main core (TCLSubmitToRing does not need this workaround)
 

src/Cafe/OS/libs/gx2/GX2_Event.cpp

@@ -263,7 +263,7 @@ namespace GX2
 			gx2WriteGather_submitU32AsBE(0x00000000); // unused
 		}
 		// flush pipeline
-		if (_GX2GetUnflushedBytes(PPCInterpreter_getCoreIndex(ppcInterpreterCurrentInstance)) > 0)
+		if (_GX2GetUnflushedBytes(coreinit::OSGetCoreId()) > 0)
 			_GX2SubmitToTCL();
 
 		uint64 ts = GX2GetLastSubmittedTimeStamp();

src/Cafe/OS/libs/nlibcurl/nlibcurl.cpp

@@ -225,7 +225,7 @@ void CurlWorkerThread(CURL_t* curl, PPCConcurrentQueue<QueueMsg_t>* callerQueue,
 
 uint32 SendOrderToWorker(CURL_t* curl, QueueOrder order, uint32 arg1 = 0)
 {
-	OSThread_t* currentThread = coreinitThread_getCurrentThreadDepr(ppcInterpreterCurrentInstance);
+	OSThread_t* currentThread = coreinitThread_getCurrentThreadDepr(PPCInterpreter_getCurrentInstance());
 	curl->curlThread = currentThread;
 
 	// cemuLog_logDebug(LogType::Force, "CURRENTTHREAD: 0x{} -> {}",currentThread, order)
@@ -707,7 +707,7 @@ void export_curl_easy_init(PPCInterpreter_t* hCPU)
 		memset(result.GetPtr(), 0, sizeof(CURL_t));
 		*result = {};
 		result->curl = curl_easy_init();
-		result->curlThread = coreinitThread_getCurrentThreadDepr(ppcInterpreterCurrentInstance);
+		result->curlThread = coreinitThread_getCurrentThreadDepr(PPCInterpreter_getCurrentInstance());
 
 		result->info_contentType = nullptr;
 		result->info_redirectUrl = nullptr;

src/Cafe/OS/libs/nn_save/nn_save.cpp

@@ -451,14 +451,14 @@ namespace save
 		asyncParams.userCallback = PPCInterpreter_makeCallableExportDepr(AsyncCallback);
 
 		StackAllocator<AsyncCallbackParam_t> param;
-		param->thread = coreinitThread_getCurrentThreadMPTRDepr(ppcInterpreterCurrentInstance);
+		param->thread = coreinitThread_getCurrentThreadMPTRDepr(PPCInterpreter_getCurrentInstance());
 		param->returnStatus = (FSStatus)FS_RESULT::SUCCESS;
 		asyncParams.userContext = param.GetPointer();
 
 		SAVEStatus status = SAVEOpenFileOtherApplicationAsync(client, block, titleId, accountSlot, path, mode, hFile, errHandling, &asyncParams);
 		if (status == (FSStatus)FS_RESULT::SUCCESS)
 		{
-			coreinit_suspendThread(coreinitThread_getCurrentThreadDepr(ppcInterpreterCurrentInstance), 1000);
+			coreinit_suspendThread(coreinitThread_getCurrentThreadDepr(PPCInterpreter_getCurrentInstance()), 1000);
 			PPCCore_switchToScheduler();
 			return param->returnStatus;
 		}
@@ -685,14 +685,14 @@ namespace save
 		asyncParams.userCallback = _swapEndianU32(PPCInterpreter_makeCallableExportDepr(AsyncCallback));
 
 		StackAllocator<AsyncCallbackParam_t> param;
-		param->thread = coreinitThread_getCurrentThreadMPTRDepr(ppcInterpreterCurrentInstance);
+		param->thread = coreinitThread_getCurrentThreadMPTRDepr(PPCInterpreter_getCurrentInstance());
 		param->returnStatus = (FSStatus)FS_RESULT::SUCCESS;
 		asyncParams.userContext = param.GetMPTRBE();
 
 		SAVEStatus status = SAVEGetFreeSpaceSizeAsync(client, block, accountSlot, freeSize, errHandling, &asyncParams);
 		if (status == (FSStatus)FS_RESULT::SUCCESS)
 		{
-			coreinit_suspendThread(coreinitThread_getCurrentThreadDepr(ppcInterpreterCurrentInstance), 1000);
+			coreinit_suspendThread(coreinitThread_getCurrentThreadDepr(PPCInterpreter_getCurrentInstance()), 1000);
 			PPCCore_switchToScheduler();
 			return param->returnStatus;
 		}
@@ -754,14 +754,14 @@ namespace save
 		asyncParams.userCallback = _swapEndianU32(PPCInterpreter_makeCallableExportDepr(AsyncCallback));
 
 		StackAllocator<AsyncCallbackParam_t> param;
-		param->thread = coreinitThread_getCurrentThreadMPTRDepr(ppcInterpreterCurrentInstance);
+		param->thread = coreinitThread_getCurrentThreadMPTRDepr(PPCInterpreter_getCurrentInstance());
 		param->returnStatus = (FSStatus)FS_RESULT::SUCCESS;
 		asyncParams.userContext = param.GetMPTRBE();
 
 		SAVEStatus status = SAVERemoveAsync(client, block, accountSlot, path, errHandling, &asyncParams);
 		if (status == (FSStatus)FS_RESULT::SUCCESS)
 		{
-			coreinit_suspendThread(coreinitThread_getCurrentThreadDepr(ppcInterpreterCurrentInstance), 1000);
+			coreinit_suspendThread(coreinitThread_getCurrentThreadDepr(PPCInterpreter_getCurrentInstance()), 1000);
 			PPCCore_switchToScheduler();
 			return param->returnStatus;
 		}
@@ -867,14 +867,14 @@ namespace save
 		asyncParams.userCallback = _swapEndianU32(PPCInterpreter_makeCallableExportDepr(AsyncCallback));
 
 		StackAllocator<AsyncCallbackParam_t> param;
-		param->thread = coreinitThread_getCurrentThreadMPTRDepr(ppcInterpreterCurrentInstance);
+		param->thread = coreinitThread_getCurrentThreadMPTRDepr(PPCInterpreter_getCurrentInstance());
 		param->returnStatus = (FSStatus)FS_RESULT::SUCCESS;
 		asyncParams.userContext = param.GetMPTRBE();
 
 		SAVEStatus status = SAVEOpenDirAsync(client, block, accountSlot, path, hDir, errHandling, &asyncParams);
 		if (status == (FSStatus)FS_RESULT::SUCCESS)
 		{
-			coreinit_suspendThread(coreinitThread_getCurrentThreadDepr(ppcInterpreterCurrentInstance), 1000);
+			coreinit_suspendThread(coreinitThread_getCurrentThreadDepr(PPCInterpreter_getCurrentInstance()), 1000);
 			PPCCore_switchToScheduler();
 			return param->returnStatus;
 		}
@@ -940,14 +940,14 @@ namespace save
 		asyncParams.userCallback = _swapEndianU32(PPCInterpreter_makeCallableExportDepr(AsyncCallback));
 
 		StackAllocator<AsyncCallbackParam_t> param;
-		param->thread = coreinitThread_getCurrentThreadMPTRDepr(ppcInterpreterCurrentInstance);
+		param->thread = coreinitThread_getCurrentThreadMPTRDepr(PPCInterpreter_getCurrentInstance());
 		param->returnStatus = (FSStatus)FS_RESULT::SUCCESS;
 		asyncParams.userContext = param.GetMPTRBE();
 
 		SAVEStatus status = SAVEOpenDirOtherApplicationAsync(client, block, titleId, accountSlot, path, hDir, errHandling, &asyncParams);
 		if (status == (FSStatus)FS_RESULT::SUCCESS)
 		{
-			coreinit_suspendThread(coreinitThread_getCurrentThreadDepr(ppcInterpreterCurrentInstance), 1000);
+			coreinit_suspendThread(coreinitThread_getCurrentThreadDepr(PPCInterpreter_getCurrentInstance()), 1000);
 			PPCCore_switchToScheduler();
 			return param->returnStatus;
 		}
@@ -1076,14 +1076,14 @@ namespace save
 		asyncParams.userCallback = _swapEndianU32(PPCInterpreter_makeCallableExportDepr(AsyncCallback));
 
 		StackAllocator<AsyncCallbackParam_t> param;
-		param->thread = coreinitThread_getCurrentThreadMPTRDepr(ppcInterpreterCurrentInstance);
+		param->thread = coreinitThread_getCurrentThreadMPTRDepr(PPCInterpreter_getCurrentInstance());
 		param->returnStatus = (FSStatus)FS_RESULT::SUCCESS;
 		asyncParams.userContext = param.GetMPTRBE();
 
 		SAVEStatus status = SAVEMakeDirAsync(client, block, accountSlot, path, errHandling, &asyncParams);
 		if (status == (FSStatus)FS_RESULT::SUCCESS)
 		{
-			coreinit_suspendThread(coreinitThread_getCurrentThreadDepr(ppcInterpreterCurrentInstance), 1000);
+			coreinit_suspendThread(coreinitThread_getCurrentThreadDepr(PPCInterpreter_getCurrentInstance()), 1000);
 			PPCCore_switchToScheduler();
 			return param->returnStatus;
 		}
@@ -1127,14 +1127,14 @@ namespace save
 		asyncParams.userCallback = PPCInterpreter_makeCallableExportDepr(AsyncCallback);
 
 		StackAllocator<AsyncCallbackParam_t> param;
-		param->thread = coreinitThread_getCurrentThreadMPTRDepr(ppcInterpreterCurrentInstance);
+		param->thread = coreinitThread_getCurrentThreadMPTRDepr(PPCInterpreter_getCurrentInstance());
 		param->returnStatus = (FSStatus)FS_RESULT::SUCCESS;
 		asyncParams.userContext = param.GetPointer();
 
 		SAVEStatus status = SAVEOpenFileAsync(client, block, accountSlot, path, mode, hFile, errHandling, &asyncParams);
 		if (status == (FSStatus)FS_RESULT::SUCCESS)
 		{
-			coreinit_suspendThread(coreinitThread_getCurrentThreadDepr(ppcInterpreterCurrentInstance), 1000);
+			coreinit_suspendThread(coreinitThread_getCurrentThreadDepr(PPCInterpreter_getCurrentInstance()), 1000);
 			PPCCore_switchToScheduler();
 			return param->returnStatus;
 		}
@@ -1187,14 +1187,14 @@ namespace save
 		asyncParams.userCallback = _swapEndianU32(PPCInterpreter_makeCallableExportDepr(AsyncCallback));
 
 		StackAllocator<AsyncCallbackParam_t> param;
-		param->thread = coreinitThread_getCurrentThreadMPTRDepr(ppcInterpreterCurrentInstance);
+		param->thread = coreinitThread_getCurrentThreadMPTRDepr(PPCInterpreter_getCurrentInstance());
 		param->returnStatus = (FSStatus)FS_RESULT::SUCCESS;
 		asyncParams.userContext = param.GetMPTRBE();
 
 		SAVEStatus status = SAVEGetStatAsync(client, block, accountSlot, path, stat, errHandling, &asyncParams);
 		if (status == (FSStatus)FS_RESULT::SUCCESS)
 		{
-			coreinit_suspendThread(coreinitThread_getCurrentThreadDepr(ppcInterpreterCurrentInstance), 1000);
+			coreinit_suspendThread(coreinitThread_getCurrentThreadDepr(PPCInterpreter_getCurrentInstance()), 1000);
 			PPCCore_switchToScheduler();
 			return param->returnStatus;
 		}
@@ -1238,14 +1238,14 @@ namespace save
 		asyncParams.userCallback = _swapEndianU32(PPCInterpreter_makeCallableExportDepr(AsyncCallback));
 
 		StackAllocator<AsyncCallbackParam_t> param;
-		param->thread = coreinitThread_getCurrentThreadMPTRDepr(ppcInterpreterCurrentInstance);
+		param->thread = coreinitThread_getCurrentThreadMPTRDepr(PPCInterpreter_getCurrentInstance());
 		param->returnStatus = (FSStatus)FS_RESULT::SUCCESS;
 		asyncParams.userContext = param.GetMPTRBE();
 
 		SAVEStatus status = SAVEGetStatOtherApplicationAsync(client, block, titleId, accountSlot, path, stat, errHandling, &asyncParams);
 		if (status == (FSStatus)FS_RESULT::SUCCESS)
 		{
-			coreinit_suspendThread(coreinitThread_getCurrentThreadDepr(ppcInterpreterCurrentInstance), 1000);
+			coreinit_suspendThread(coreinitThread_getCurrentThreadDepr(PPCInterpreter_getCurrentInstance()), 1000);
 			PPCCore_switchToScheduler();
 			return param->returnStatus;
 		}
@@ -1427,14 +1427,14 @@ namespace save
 		asyncParams.userCallback = _swapEndianU32(PPCInterpreter_makeCallableExportDepr(AsyncCallback));
 
 		StackAllocator<AsyncCallbackParam_t> param;
-		param->thread = coreinitThread_getCurrentThreadMPTRDepr(ppcInterpreterCurrentInstance);
+		param->thread = coreinitThread_getCurrentThreadMPTRDepr(PPCInterpreter_getCurrentInstance());
 		param->returnStatus = (FSStatus)FS_RESULT::SUCCESS;
 		asyncParams.userContext = param.GetMPTRBE();
 
 		SAVEStatus status = SAVEChangeDirAsync(client, block, accountSlot, path, errHandling, &asyncParams);
 		if (status == (FSStatus)FS_RESULT::SUCCESS)
 		{
-			coreinit_suspendThread(coreinitThread_getCurrentThreadDepr(ppcInterpreterCurrentInstance), 1000);
+			coreinit_suspendThread(coreinitThread_getCurrentThreadDepr(PPCInterpreter_getCurrentInstance()), 1000);
 			PPCCore_switchToScheduler();
 			return param->returnStatus;
 		}
@@ -1496,14 +1496,14 @@ namespace save
 		asyncParams.userCallback = _swapEndianU32(PPCInterpreter_makeCallableExportDepr(AsyncCallback));
 
 		StackAllocator<AsyncCallbackParam_t> param;
-		param->thread = coreinitThread_getCurrentThreadMPTRDepr(ppcInterpreterCurrentInstance);
+		param->thread = coreinitThread_getCurrentThreadMPTRDepr(PPCInterpreter_getCurrentInstance());
 		param->returnStatus = (FSStatus)FS_RESULT::SUCCESS;
 		asyncParams.userContext = param.GetMPTRBE();
 
 		SAVEStatus status = SAVEFlushQuotaAsync(client, block, accountSlot, errHandling, &asyncParams);
 		if (status == (FSStatus)FS_RESULT::SUCCESS)
 		{
-			coreinit_suspendThread(coreinitThread_getCurrentThreadDepr(ppcInterpreterCurrentInstance), 1000);
+			coreinit_suspendThread(coreinitThread_getCurrentThreadDepr(PPCInterpreter_getCurrentInstance()), 1000);
 			PPCCore_switchToScheduler();
 			return param->returnStatus;
 		}

src/Cafe/OS/libs/snd_core/ax_aux.cpp

@@ -218,9 +218,10 @@ namespace snd_core
 				AXAUXCBCHANNELINFO* cbStruct = __AXAuxCB_auxCBStruct.GetPtr();
 				cbStruct->numChannels = tvChannelCount;
 				cbStruct->numSamples = sampleCount;
-				ppcInterpreterCurrentInstance->gpr[3] = __AXAuxCB_dataPtrs.GetMPTR();
+				PPCInterpreter_t* hCPU = PPCInterpreter_getCurrentInstance();
-				ppcInterpreterCurrentInstance->gpr[4] = __AXAuxTVCallbackUserParam[auxBusIndex];
+				hCPU->gpr[3] = __AXAuxCB_dataPtrs.GetMPTR();
-				ppcInterpreterCurrentInstance->gpr[5] = __AXAuxCB_auxCBStruct.GetMPTR();
+				hCPU->gpr[4] = __AXAuxTVCallbackUserParam[auxBusIndex];
+				hCPU->gpr[5] = __AXAuxCB_auxCBStruct.GetMPTR();
 				PPCCore_executeCallbackInternal(auxCBFuncMPTR);
 			}
 			else
@@ -255,9 +256,10 @@ namespace snd_core
 					AXAUXCBCHANNELINFO* cbStruct = __AXAuxCB_auxCBStruct.GetPtr();
 					cbStruct->numChannels = drcChannelCount;
 					cbStruct->numSamples = sampleCount;
-					ppcInterpreterCurrentInstance->gpr[3] = __AXAuxCB_dataPtrs.GetMPTR();
+					PPCInterpreter_t* hCPU = PPCInterpreter_getCurrentInstance();
-					ppcInterpreterCurrentInstance->gpr[4] = __AXAuxDRCCallbackUserParam[auxBusIndex + drcIndex * 3];
+					hCPU->gpr[3] = __AXAuxCB_dataPtrs.GetMPTR();
-					ppcInterpreterCurrentInstance->gpr[5] = __AXAuxCB_auxCBStruct.GetMPTR();
+					hCPU->gpr[4] = __AXAuxDRCCallbackUserParam[auxBusIndex + drcIndex * 3];
+					hCPU->gpr[5] = __AXAuxCB_auxCBStruct.GetMPTR();
 					PPCCore_executeCallbackInternal(auxCBFuncMPTR);
 				}
 				else

src/Cafe/OS/libs/snd_core/ax_voice.cpp

@@ -3,6 +3,7 @@
 #include "Cafe/HW/Espresso/PPCCallback.h"
 #include "Cafe/OS/libs/snd_core/ax.h"
 #include "Cafe/OS/libs/snd_core/ax_internal.h"
+#include "Cafe/OS/libs/coreinit/coreinit_Thread.h"
 #include "util/helpers/fspinlock.h"
 
 namespace snd_core
@@ -120,7 +121,7 @@ namespace snd_core
 		{
 			return -2;
 		}
-		MPTR currentThreadMPTR = coreinitThread_getCurrentThreadMPTRDepr(ppcInterpreterCurrentInstance);
+		MPTR currentThreadMPTR = memory_getVirtualOffsetFromPointer(coreinit::OSGetCurrentThread());
 		for (sint32 i = __AXUserProtectionArraySize - 1; i >= 0; i--)
 		{
 			if (__AXUserProtectionArray[i].threadMPTR == currentThreadMPTR)
@@ -151,7 +152,7 @@ namespace snd_core
 		PPCCore_deboostQuantum(10000);
 		if (AXIst_IsFrameBeingProcessed())
 			return -2;
-		MPTR currentThreadMPTR = coreinitThread_getCurrentThreadMPTRDepr(ppcInterpreterCurrentInstance);
+		MPTR currentThreadMPTR = memory_getVirtualOffsetFromPointer(coreinit::OSGetCurrentThread());
 		for (sint32 i = __AXUserProtectionArraySize - 1; i >= 0; i--)
 		{
 			if (__AXUserProtectionArray[i].threadMPTR == currentThreadMPTR)
@@ -206,7 +207,7 @@ namespace snd_core
 		if (AXIst_IsFrameBeingProcessed())
 			isProtected = __AXVoiceProtection[index].threadMPTR != MPTR_NULL;
 		else
-			isProtected = __AXVoiceProtection[index].threadMPTR != coreinitThread_getCurrentThreadMPTRDepr(ppcInterpreterCurrentInstance);
+			isProtected = __AXVoiceProtection[index].threadMPTR != memory_getVirtualOffsetFromPointer(coreinit::OSGetCurrentThread());
 		return isProtected;
 	}
 
@@ -219,7 +220,7 @@ namespace snd_core
 			return;
 		if (__AXVoiceProtection[index].threadMPTR == MPTR_NULL)
 		{
-			__AXVoiceProtection[index].threadMPTR = coreinitThread_getCurrentThreadMPTRDepr(ppcInterpreterCurrentInstance);
+			__AXVoiceProtection[index].threadMPTR = memory_getVirtualOffsetFromPointer(coreinit::OSGetCurrentThread());
 			// does not set count?
 		}
 	}
@@ -246,7 +247,7 @@ namespace snd_core
 		}
 		if (AXIst_IsFrameBeingProcessed())
 			return -2;
-		MPTR currentThreadMPTR = coreinitThread_getCurrentThreadMPTRDepr(ppcInterpreterCurrentInstance);
+		MPTR currentThreadMPTR = memory_getVirtualOffsetFromPointer(coreinit::OSGetCurrentThread());
 		if (__AXVoiceProtection[index].threadMPTR == MPTR_NULL)
 		{
 			__AXVoiceProtection[index].threadMPTR = currentThreadMPTR;
@@ -286,7 +287,7 @@ namespace snd_core
 		}
 		if (AXIst_IsFrameBeingProcessed())
 			return -2;
-		MPTR currentThreadMPTR = coreinitThread_getCurrentThreadMPTRDepr(ppcInterpreterCurrentInstance);
+		MPTR currentThreadMPTR = memory_getVirtualOffsetFromPointer(coreinit::OSGetCurrentThread());
 		if (__AXVoiceProtection[index].threadMPTR == currentThreadMPTR)
 		{
 			if (__AXVoiceProtection[index].count > 0)

src/Cafe/OS/libs/zlib125/zlib125.cpp

@@ -28,8 +28,7 @@ static_assert(sizeof(z_stream_ppc2) == 0x38);
 voidpf zcallocWrapper(voidpf opaque, uInt items, uInt size)
 {
 	z_stream_ppc2* zstream = (z_stream_ppc2*)opaque;
-	
+	PPCInterpreter_t* hCPU = PPCInterpreter_getCurrentInstance();
-	PPCInterpreter_t* hCPU = ppcInterpreterCurrentInstance;
 	hCPU->gpr[3] = zstream->opaque.GetMPTR();
 	hCPU->gpr[4] = items;
 	hCPU->gpr[5] = size;
@@ -41,7 +40,7 @@ voidpf zcallocWrapper(voidpf opaque, uInt items, uInt size)
 void zcfreeWrapper(voidpf opaque, voidpf baseIndex)
 {
 	z_stream_ppc2* zstream = (z_stream_ppc2*)opaque;
-	PPCInterpreter_t* hCPU = ppcInterpreterCurrentInstance;
+	PPCInterpreter_t* hCPU = PPCInterpreter_getCurrentInstance();
 	hCPU->gpr[3] = zstream->opaque.GetMPTR();
 	hCPU->gpr[4] = memory_getVirtualOffsetFromPointer(baseIndex);
 	PPCCore_executeCallbackInternal(zstream->zfree.GetMPTR());

src/Common/ExceptionHandler/ExceptionHandler.cpp

@@ -73,16 +73,17 @@ void ExceptionHandler_LogGeneralInfo()
     // info about active PPC instance:
     CrashLog_WriteLine("");
     CrashLog_WriteHeader("Active PPC instance");
-    if (ppcInterpreterCurrentInstance)
+	PPCInterpreter_t* hCPU = PPCInterpreter_getCurrentInstance();
+    if (hCPU)
     {
         OSThread_t* currentThread = coreinit::OSGetCurrentThread();
         uint32 threadPtr = memory_getVirtualOffsetFromPointer(coreinit::OSGetCurrentThread());
-        sprintf(dumpLine, "IP 0x%08x LR 0x%08x Thread 0x%08x", ppcInterpreterCurrentInstance->instructionPointer, ppcInterpreterCurrentInstance->spr.LR, threadPtr);
+        sprintf(dumpLine, "IP 0x%08x LR 0x%08x Thread 0x%08x", hCPU->instructionPointer, hCPU->spr.LR, threadPtr);
         CrashLog_WriteLine(dumpLine);
 
         // GPR info
         CrashLog_WriteLine("");
-        auto gprs = ppcInterpreterCurrentInstance->gpr;
+        auto gprs = hCPU->gpr;
         sprintf(dumpLine, "r0 =%08x r1 =%08x r2 =%08x r3 =%08x r4 =%08x r5 =%08x r6 =%08x r7 =%08x", gprs[0], gprs[1], gprs[2], gprs[3], gprs[4], gprs[5], gprs[6], gprs[7]);
         CrashLog_WriteLine(dumpLine);
         sprintf(dumpLine, "r8 =%08x r9 =%08x r10=%08x r11=%08x r12=%08x r13=%08x r14=%08x r15=%08x", gprs[8], gprs[9], gprs[10], gprs[11], gprs[12], gprs[13], gprs[14], gprs[15]);
@@ -93,7 +94,7 @@ void ExceptionHandler_LogGeneralInfo()
         CrashLog_WriteLine(dumpLine);
 
         // stack trace
-        MPTR currentStackVAddr = ppcInterpreterCurrentInstance->gpr[1];
+        MPTR currentStackVAddr = hCPU->gpr[1];
         CrashLog_WriteLine("");
         CrashLog_WriteHeader("PPC stack trace");
         DebugLogStackTrace(currentThread, currentStackVAddr);

src/Common/precompiled.h

@@ -236,12 +236,17 @@ inline uint64 _udiv128(uint64 highDividend, uint64 lowDividend, uint64 divisor,
 #if defined(_MSC_VER)
     #define UNREACHABLE __assume(false)
 	#define ASSUME(__cond) __assume(__cond)
-#elif defined(__GNUC__)
+	#define TLS_WORKAROUND_NOINLINE // no-op for MSVC as it has a flag for fiber-safe TLS optimizations
+#elif defined(__GNUC__) && !defined(__llvm__)
     #define UNREACHABLE __builtin_unreachable()
 	#define ASSUME(__cond) __attribute__((assume(__cond)))
+	#define TLS_WORKAROUND_NOINLINE __attribute__((noinline))
+#elif defined(__clang__)
+	#define UNREACHABLE __builtin_unreachable()
+	#define ASSUME(__cond) __builtin_assume(__cond)
+	#define TLS_WORKAROUND_NOINLINE __attribute__((noinline))
 #else
-    #define UNREACHABLE
+    #error Unknown compiler
-	#define ASSUME(__cond)
 #endif
 
 #if defined(_MSC_VER)