figure out this mov addressing thing fast please

CMakeLists.txt

@@ -13,6 +13,7 @@ if(APPLE)
   enable_language(OBJC)
 endif()
 
+set(VULKAN_VALIDATION FALSE)
 set(SANITIZERS FALSE)
 set(CMAKE_CXX_STANDARD 23)
 set(CMAKE_CXX_STANDARD_REQUIRED ON)
@@ -102,7 +103,8 @@ if (HAS_SIMD)
   add_compile_options(${SIMD_FLAG})
 endif ()
 
-if (${CMAKE_BUILD_TYPE} MATCHES Debug)
+if (${CMAKE_BUILD_TYPE} MATCHES Debug AND VULKAN_VALIDATION)
+  message("VALIDATION LAYERS: ON")
   add_compile_definitions(VULKAN_DEBUG)
 endif ()
 
@@ -152,6 +154,7 @@ target_link_libraries(kaizen PUBLIC imgui SDL3::SDL3 SDL3::SDL3-static cflags::c
 target_compile_definitions(kaizen PUBLIC SDL_MAIN_HANDLED)
 
 if (SANITIZERS)
+  message("UBSAN AND ASAN: ON")
   target_compile_options(kaizen PUBLIC -fsanitize=undefined -fsanitize=address)
   target_link_options(kaizen PUBLIC -fsanitize=undefined -fsanitize=address)
 endif ()

external/parallel-rdp/ParallelRDPWrapper.cpp

@@ -6,14 +6,13 @@
 #include <resources/frag.spv.h>
 #include <KaizenGui.hpp>
 #include <imgui_impl_vulkan.h>
-#include <mutex>
 
 using namespace Vulkan;
 using namespace RDP;
 
 bool ParallelRDP::IsFramerateUnlocked() const { return wsi->get_present_mode() != PresentMode::SyncToVBlank; }
 
-void ParallelRDP::SetFramerateUnlocked(bool unlocked) const {
+void ParallelRDP::SetFramerateUnlocked(const bool unlocked) const {
   if (unlocked) {
     wsi->set_present_mode(PresentMode::UnlockedForceTearing);
   } else {
@@ -24,11 +23,11 @@ void ParallelRDP::SetFramerateUnlocked(bool unlocked) const {
 Program *fullscreen_quad_program;
 
 // Copied and modified from WSI::init_context_from_platform
-Util::IntrusivePtr<Context> InitVulkanContext(WSIPlatform *platform, unsigned num_thread_indices,
+Util::IntrusivePtr<Context> InitVulkanContext(WSIPlatform *platform, const unsigned num_thread_indices,
                                               const Context::SystemHandles &system_handles) {
   VK_ASSERT(platform);
-  auto instance_ext = platform->get_instance_extensions();
+  const auto instance_ext = platform->get_instance_extensions();
-  auto device_ext = platform->get_device_extensions();
+  const auto device_ext = platform->get_device_extensions();
   auto new_context = Util::make_handle<Context>();
 
   new_context->set_application_info(platform->get_application_info());
@@ -39,9 +38,9 @@ Util::IntrusivePtr<Context> InitVulkanContext(WSIPlatform *platform, unsigned nu
     panic("Failed to create Vulkan instance.\n");
   }
 
-  VkSurfaceKHR tmp_surface = platform->create_surface(new_context->get_instance(), VK_NULL_HANDLE);
+  const auto tmp_surface = platform->create_surface(new_context->get_instance(), VK_NULL_HANDLE);
 
-  bool ret = new_context->init_device(VK_NULL_HANDLE, tmp_surface, device_ext.data(), device_ext.size(),
+  const bool ret = new_context->init_device(VK_NULL_HANDLE, tmp_surface, device_ext.data(), device_ext.size(),
                                       CONTEXT_CREATION_ENABLE_ADVANCED_WSI_BIT);
 
   if (tmp_surface) {
@@ -61,9 +60,9 @@ void ParallelRDP::LoadWSIPlatform(const std::shared_ptr<WSIPlatform> &wsi_platfo
   wsi->set_backbuffer_srgb(false);
   wsi->set_platform(wsi_platform.get());
   wsi->set_present_mode(PresentMode::SyncToVBlank);
-  Context::SystemHandles handles;
 
-  if (!wsi->init_from_existing_context(InitVulkanContext(wsi_platform.get(), 1, handles))) {
+  if (constexpr Context::SystemHandles handles;
+      !wsi->init_from_existing_context(InitVulkanContext(wsi_platform.get(), 1, handles))) {
     panic("Failed to initialize WSI: init_from_existing_context() failed");
   }
 
@@ -97,8 +96,8 @@ void ParallelRDP::Init(const std::shared_ptr<WSIPlatform> &wsiPlatform,
   fragLayout.sets[0].fp_mask = 1;
   fragLayout.sets[0].array_size[0] = 1;
 
-  auto sizeVert = sizeof(vertex_shader);
+  constexpr auto sizeVert = sizeof(vertex_shader);
-  auto sizeFrag = sizeof(fragment_shader);
+  constexpr auto sizeFrag = sizeof(fragment_shader);
 
   fullscreen_quad_program = wsi->get_device().request_program(
     reinterpret_cast<const u32 *>(vertex_shader), sizeVert, reinterpret_cast<const u32 *>(fragment_shader),
@@ -108,8 +107,8 @@ void ParallelRDP::Init(const std::shared_ptr<WSIPlatform> &wsiPlatform,
   uintptr_t offset = 0;
 
   if (wsi->get_device().get_device_features().supports_external_memory_host) {
-    size_t align = wsi->get_device().get_device_features().host_memory_properties.minImportedHostPointerAlignment;
+    const size_t align = wsi->get_device().get_device_features().host_memory_properties.minImportedHostPointerAlignment;
-    offset = aligned_rdram & (align - 1);
+    offset = aligned_rdram & align - 1;
     aligned_rdram -= offset;
   }
 
@@ -128,7 +127,7 @@ void ParallelRDP::DrawFullscreenTexturedQuad(Util::IntrusivePtr<Image> image,
   cmd->set_texture(0, 0, image->get_view(), StockSampler::LinearClamp);
   cmd->set_program(fullscreen_quad_program);
   cmd->set_quad_state();
-  auto data = static_cast<float *>(cmd->allocate_vertex_data(0, 6 * sizeof(float), 2 * sizeof(float)));
+  const auto data = static_cast<float *>(cmd->allocate_vertex_data(0, 6 * sizeof(float), 2 * sizeof(float)));
   data[0] = -1.0f;
   data[1] = -3.0f;
   data[2] = -1.0f;
@@ -136,15 +135,15 @@ void ParallelRDP::DrawFullscreenTexturedQuad(Util::IntrusivePtr<Image> image,
   data[4] = +3.0f;
   data[5] = +1.0f;
 
-  auto windowSize = windowInfo->get_window_size();
+  const auto [x, y] = windowInfo->get_window_size();
 
-  float zoom = std::min(windowSize.x / wsi->get_platform().get_surface_width(),
+  const float zoom = std::min(x / static_cast<float>(wsi->get_platform().get_surface_width()),
-                        windowSize.y / wsi->get_platform().get_surface_height());
+                        y / static_cast<float>(wsi->get_platform().get_surface_height()));
 
-  float width = (wsi->get_platform().get_surface_width() / windowSize.x) * zoom;
+  const float width = static_cast<float>(wsi->get_platform().get_surface_width()) / x * zoom;
-  float height = (wsi->get_platform().get_surface_height() / windowSize.y) * zoom;
+  const float height = static_cast<float>(wsi->get_platform().get_surface_height()) / y * zoom;
 
-  float uniform_data[] = {// Size
+  const float uniform_data[] = {// Size
                           width, height,
                           // Offset
                           (1.0f - width) * 0.5f, (1.0f - height) * 0.5f};
@@ -190,43 +189,43 @@ void ParallelRDP::UpdateScreen(Util::IntrusivePtr<Image> image) const {
   wsi->end_frame();
 }
 
-void ParallelRDP::UpdateScreen(bool playing) const {  
+template <>
-  if(playing) {
+void ParallelRDP::UpdateScreen<true>() const {
-    n64::Core& core = n64::Core::GetInstance();
+  const n64::VI& vi = n64::Core::GetMem().mmio.vi;
-    n64::VI& vi = core.GetMem().mmio.vi;
+  command_processor->set_vi_register(VIRegister::Control, vi.status.raw);
-    command_processor->set_vi_register(VIRegister::Control, vi.status.raw);
+  command_processor->set_vi_register(VIRegister::Origin, vi.origin);
-    command_processor->set_vi_register(VIRegister::Origin, vi.origin);
+  command_processor->set_vi_register(VIRegister::Width, vi.width);
-    command_processor->set_vi_register(VIRegister::Width, vi.width);
+  command_processor->set_vi_register(VIRegister::Intr, vi.intr);
-    command_processor->set_vi_register(VIRegister::Intr, vi.intr);
+  command_processor->set_vi_register(VIRegister::VCurrentLine, vi.current);
-    command_processor->set_vi_register(VIRegister::VCurrentLine, vi.current);
+  command_processor->set_vi_register(VIRegister::Timing, vi.burst.raw);
-    command_processor->set_vi_register(VIRegister::Timing, vi.burst.raw);
+  command_processor->set_vi_register(VIRegister::VSync, vi.vsync);
-    command_processor->set_vi_register(VIRegister::VSync, vi.vsync);
+  command_processor->set_vi_register(VIRegister::HSync, vi.hsync);
-    command_processor->set_vi_register(VIRegister::HSync, vi.hsync);
+  command_processor->set_vi_register(VIRegister::Leap, vi.hsyncLeap.raw);
-    command_processor->set_vi_register(VIRegister::Leap, vi.hsyncLeap.raw);
+  command_processor->set_vi_register(VIRegister::HStart, vi.hstart.raw);
-    command_processor->set_vi_register(VIRegister::HStart, vi.hstart.raw);
+  command_processor->set_vi_register(VIRegister::VStart, vi.vstart.raw);
-    command_processor->set_vi_register(VIRegister::VStart, vi.vstart.raw);
+  command_processor->set_vi_register(VIRegister::VBurst, vi.vburst);
-    command_processor->set_vi_register(VIRegister::VBurst, vi.vburst);
+  command_processor->set_vi_register(VIRegister::XScale, vi.xscale.raw);
-    command_processor->set_vi_register(VIRegister::XScale, vi.xscale.raw);
+  command_processor->set_vi_register(VIRegister::YScale, vi.yscale.raw);
-    command_processor->set_vi_register(VIRegister::YScale, vi.yscale.raw);
+  ScanoutOptions opts;
-    ScanoutOptions opts;
+  opts.persist_frame_on_invalid_input = true;
-    opts.persist_frame_on_invalid_input = true;
+  opts.vi.aa = true;
-    opts.vi.aa = true;
+  opts.vi.scale = true;
-    opts.vi.scale = true;
+  opts.vi.dither_filter = true;
-    opts.vi.dither_filter = true;
+  opts.vi.divot_filter = true;
-    opts.vi.divot_filter = true;
+  opts.vi.gamma_dither = true;
-    opts.vi.gamma_dither = true;
+  opts.downscale_steps = true;
-    opts.downscale_steps = true;
+  opts.crop_overscan_pixels = true;
-    opts.crop_overscan_pixels = true;
+  const Util::IntrusivePtr<Image> image = command_processor->scanout(opts);
-    Util::IntrusivePtr<Image> image = command_processor->scanout(opts);
+  UpdateScreen(image);
-    UpdateScreen(image);
+  command_processor->begin_frame_context();
-    command_processor->begin_frame_context();
+}
-    return;
+
-  }
+template <>
-  
+void ParallelRDP::UpdateScreen<false>() const {
   UpdateScreen(static_cast<Util::IntrusivePtr<Image>>(nullptr));
 }
 
-void ParallelRDP::EnqueueCommand(int command_length, const u32 *buffer) const {
+void ParallelRDP::EnqueueCommand(const int command_length, const u32 *buffer) const {
   command_processor->enqueue_command(command_length, buffer);
 }
 

external/parallel-rdp/ParallelRDPWrapper.hpp

@@ -17,11 +17,12 @@ public:
 
   void Init(const std::shared_ptr<Vulkan::WSIPlatform> &,
             const std::shared_ptr<WindowInfo> &, const u8 *);
-  
+
-  void UpdateScreen(bool = true) const;
+  template <bool>
+  void UpdateScreen() const;
   void EnqueueCommand(int, const u32 *) const;
   void OnFullSync() const;
-  bool IsFramerateUnlocked() const;
+  [[nodiscard]] bool IsFramerateUnlocked() const;
   void SetFramerateUnlocked(bool) const;
 
   std::shared_ptr<Vulkan::WSI> wsi;

src/backend/Core.cpp

@@ -5,14 +5,14 @@
 
 namespace n64 {
 Core::Core() {
-  auto cpuType = Options::GetInstance().GetValue<std::string>("cpu", "type");
+  const auto selectedCpu = Options::GetInstance().GetValue<std::string>("cpu", "type");
-  if (cpuType == "interpreter") {
+  if (selectedCpu == "interpreter") {
     cpuType = Interpreted;
-    cpu = std::make_unique<Interpreter>(parallel, *mem, regs);
+    cpu = std::make_unique<Interpreter>(*mem, regs);
-  } else if(cpuType == "jit") {
+  } else if(selectedCpu == "jit") {
     #ifndef __aarch64__
     cpuType = DynamicRecompiler;
-    cpu = std::make_unique<JIT>(parallel, *mem, regs);
+    cpu = std::make_unique<JIT>(*mem, regs);
     #else
     panic("JIT currently unsupported on aarch64");
     #endif
@@ -62,11 +62,11 @@ void Core::LoadROM(const std::string &rom_) {
   romLoaded = true;
 }
 
-int Core::StepCPU() {
+u32 Core::StepCPU() {
   return cpu->Step() + regs.PopStalledCycles();
 }
 
-void Core::StepRSP(int cpuCycles) {
+void Core::StepRSP(const u32 cpuCycles) {
   MMIO &mmio = mem->mmio;
 
   if (mmio.rsp.spStatus.halt) {
@@ -75,10 +75,10 @@ void Core::StepRSP(int cpuCycles) {
     return;
   }
 
-  static constexpr int cpuRatio = 3, rspRatio = 2;
+  static constexpr u32 cpuRatio = 3, rspRatio = 2;
   
   regs.steps += cpuCycles;
-  int sets = regs.steps / cpuRatio;
+  const auto sets = regs.steps / cpuRatio;
   mmio.rsp.steps += sets * rspRatio;
   regs.steps -= sets * cpuRatio;
 
@@ -88,7 +88,7 @@ void Core::StepRSP(int cpuCycles) {
   }
 }
 
-void Core::Run(float volumeL, float volumeR) {
+void Core::Run(const float volumeL, const float volumeR) {
   MMIO &mmio = mem->mmio;
 
   bool broken = false;
@@ -102,7 +102,7 @@ void Core::Run(float volumeL, float volumeR) {
       }
 
       while(cycles < mem->mmio.vi.cyclesPerHalfline) {
-        u32 taken = StepCPU();
+        const u32 taken = StepCPU();
         cycles += taken;
 
         if((broken = breakpoints.contains(regs.nextPC)))

src/backend/Core.hpp

@@ -29,8 +29,8 @@ struct Core {
     return *GetInstance().mem;
   }
 
-  int StepCPU();
+  u32 StepCPU();
-  void StepRSP(int cpuCycles);
+  void StepRSP(u32 cpuCycles);
   void Stop();
   void Reset();
   void LoadROM(const std::string &);

src/backend/core/BaseCPU.hpp

@@ -6,7 +6,7 @@
 namespace n64 {
 struct BaseCPU {
   virtual ~BaseCPU() = default;
-  virtual int Step() = 0;
+  virtual u32 Step() = 0;
   virtual void Reset() = 0;
 };
 } // namespace n64

src/backend/core/Interpreter.cpp

@@ -1,7 +1,7 @@
 #include <Core.hpp>
 
 namespace n64 {
-Interpreter::Interpreter(ParallelRDP& parallel, Mem& mem, Registers& regs) : mem(mem), regs(regs) {}
+Interpreter::Interpreter(Mem& mem, Registers& regs) : regs(regs), mem(mem) {}
 
 bool Interpreter::ShouldServiceInterrupt() const {
   const bool interrupts_pending = (regs.cop0.status.im & regs.cop0.cause.interruptPending) != 0;
@@ -12,7 +12,7 @@ bool Interpreter::ShouldServiceInterrupt() const {
   return interrupts_pending && interrupts_enabled && !currently_handling_exception && !currently_handling_error;
 }
 
-void Interpreter::CheckCompareInterrupt() {
+void Interpreter::CheckCompareInterrupt() const {
   regs.cop0.count++;
   regs.cop0.count &= 0x1FFFFFFFF;
   if (regs.cop0.count == static_cast<u64>(regs.cop0.compare) << 1) {
@@ -21,7 +21,7 @@ void Interpreter::CheckCompareInterrupt() {
   }
 }
 
-int Interpreter::Step() {
+u32 Interpreter::Step() {
   CheckCompareInterrupt();
 
   regs.prevDelaySlot = regs.delaySlot;

src/backend/core/Interpreter.hpp

@@ -6,10 +6,10 @@
 namespace n64 {
 struct Core;
 
-struct Interpreter : BaseCPU {
+struct Interpreter final : BaseCPU {
-  explicit Interpreter(ParallelRDP&, Mem&, Registers&);
+  explicit Interpreter(Mem&, Registers&);
   ~Interpreter() override = default;
-  int Step() override;
+  u32 Step() override;
 
   void Reset() override {
     cop2Latch = {};
@@ -22,8 +22,8 @@ private:
   friend struct Cop1;
 #define check_address_error(mask, vaddr)                                                                               \
   (((!regs.cop0.is64BitAddressing) && (s32)(vaddr) != (vaddr)) || (((vaddr) & (mask)) != 0))
-  bool ShouldServiceInterrupt() const;
+  [[nodiscard]] bool ShouldServiceInterrupt() const;
-  void CheckCompareInterrupt();
+  void CheckCompareInterrupt() const;
 
   void cop2Decode(Instruction);
   void special(Instruction);

src/backend/core/JIT.cpp

@@ -3,8 +3,9 @@
 
 namespace n64 {
 #ifndef __aarch64__
-JIT::JIT(ParallelRDP& parallel, Mem& mem, Registers& regs) : mem(mem), regs(regs) {
+JIT::JIT(Mem& mem, Registers& regs) : regs(regs), mem(mem) {
   regs.SetJIT(this);
+  mem.SetJIT(this);
   blockCache.resize(kUpperSize);
   if (cs_open(CS_ARCH_MIPS, static_cast<cs_mode>(CS_MODE_MIPS64 | CS_MODE_BIG_ENDIAN), &disassemblerMips) !=
       CS_ERR_OK) {
@@ -25,7 +26,7 @@ bool JIT::ShouldServiceInterrupt() const {
   return interrupts_pending && interrupts_enabled && !currently_handling_exception && !currently_handling_error;
 }
 
-void JIT::CheckCompareInterrupt() {
+void JIT::CheckCompareInterrupt() const {
   regs.cop0.count++;
   regs.cop0.count &= 0x1FFFFFFFF;
   if (regs.cop0.count == static_cast<u64>(regs.cop0.compare) << 1) {
@@ -39,20 +40,20 @@ void JIT::InvalidateBlock(const u32 paddr) {
     blockCache[index] = {};
 }
 
-u32 JIT::FetchInstruction() {
+std::optional<u32> JIT::FetchInstruction(s64 vaddr) {
   u32 paddr = 0;
 
-  if (check_address_error(0b11, u64(blockPC))) [[unlikely]] {
+  if (check_address_error(0b11, vaddr)) [[unlikely]] {
     /*regs.cop0.HandleTLBException(blockPC);
     regs.cop0.FireException(ExceptionCode::AddressErrorLoad, 0, blockPC);
     return 1;*/
     Util::Error::GetInstance().Throw(
       {Util::Error::Severity::NON_FATAL}, {Util::Error::Type::UNHANDLED_EXCEPTION}, blockPC, {},
       "[JIT]: Unhandled exception ADL due to unaligned PC virtual value!");
-    return 0;
+    return std::nullopt;
   }
 
-  if (!regs.cop0.MapVAddr(Cop0::LOAD, blockPC, paddr)) {
+  if (!regs.cop0.MapVAddr(Cop0::LOAD, vaddr, paddr)) {
     /*regs.cop0.HandleTLBException(blockPC);
     regs.cop0.FireException(Cop0::GetTLBExceptionCode(regs.cop0.tlbError, Cop0::LOAD), 0, blockPC);
     return 1;*/
@@ -60,48 +61,54 @@ u32 JIT::FetchInstruction() {
       {Util::Error::Severity::NON_FATAL}, {Util::Error::Type::UNHANDLED_EXCEPTION}, blockPC, {},
       "[JIT]: Unhandled exception TLB exception {} when retrieving PC physical address!",
       static_cast<int>(Cop0::GetTLBExceptionCode(regs.cop0.tlbError, Cop0::LOAD)));
-    return 0;
+    return std::nullopt;
   }
 
-  return Core::GetMem().Read<u32>(paddr);
+  const u32 instr = Core::GetMem().Read<u32>(paddr);
+
+  info("{}", Disassembler::GetInstance().DisassembleSimple(paddr, instr).full);
+
+  return instr;
 }
 
-void JIT::SetPC32(s32 val) {
+void JIT::SetPC32(const s32 val) {
-  code.mov(code.SCR1, REG(qword, pc));
+  code.mov(code.SCR1, code.ptr[PC_OFFSET()]);
-  code.mov(REG(qword, oldPC), code.SCR1);
+  code.mov(code.ptr[OLD_PC_OFFSET()], code.SCR1);
-  code.mov(code.SCR1, s64(val));
+  code.mov(code.SCR1.cvt32(), val);
-  code.mov(REG(qword, pc), code.SCR1);
+  code.movsxd(code.SCR1.cvt64(), code.SCR1.cvt32());
-  code.mov(code.SCR1, s64(val) + 4);
+  code.mov(code.ptr[PC_OFFSET()], code.SCR1);
-  code.mov(REG(qword, nextPC), code.SCR1);
+  code.mov(code.SCR1.cvt32(), val + 4);
+  code.movsxd(code.SCR1.cvt64(), code.SCR1.cvt32());
+  code.mov(code.ptr[NEXT_PC_OFFSET()], code.SCR1);
 }
 
-void JIT::SetPC64(s64 val) {
+void JIT::SetPC64(const s64 val) {
-  code.mov(code.SCR1, REG(qword, pc));
+  code.mov(code.SCR1, code.ptr[PC_OFFSET()]);
-  code.mov(REG(qword, oldPC), code.SCR1);
+  code.mov(code.ptr[OLD_PC_OFFSET()], code.SCR1);
   code.mov(code.SCR1, val);
-  code.mov(REG(qword, pc), code.SCR1);
+  code.mov(code.ptr[PC_OFFSET()], code.SCR1);
   code.mov(code.SCR1, val + 4);
-  code.mov(REG(qword, nextPC), code.SCR1);
+  code.mov(code.ptr[NEXT_PC_OFFSET()], code.SCR1);
 }
 
 void JIT::SetPC32(const Xbyak::Reg32& val) {
-  code.mov(code.SCR1, REG(qword, pc));
+  code.mov(code.SCR1, code.ptr[PC_OFFSET()]);
-  code.mov(REG(qword, oldPC), code.SCR1);
+  code.mov(code.ptr[OLD_PC_OFFSET()], code.SCR1);
   code.movsxd(val.cvt64(), val);
-  code.mov(REG(qword, pc), val);
+  code.mov(code.ptr[PC_OFFSET()], val);
   code.add(val, 4);
-  code.mov(REG(qword, nextPC), val);
+  code.mov(code.ptr[NEXT_PC_OFFSET()], val);
 }
 
 void JIT::SetPC64(const Xbyak::Reg64& val) {
-  code.mov(code.SCR1, REG(qword, pc));
+  code.mov(code.SCR1, code.ptr[PC_OFFSET()]);
-  code.mov(REG(qword, oldPC), code.SCR1);
+  code.mov(code.ptr[OLD_PC_OFFSET()], code.SCR1);
-  code.mov(REG(qword, pc), val);
+  code.mov(code.ptr[PC_OFFSET()], val);
   code.add(val, 4);
-  code.mov(REG(qword, nextPC), val);
+  code.mov(code.ptr[NEXT_PC_OFFSET()], val);
 }
 
-int JIT::Step() {
+u32 JIT::Step() {
   blockOldPC = regs.oldPC;
   blockPC = regs.pc;
   blockNextPC = regs.nextPC;
@@ -111,15 +118,15 @@ int JIT::Step() {
     /*regs.cop0.HandleTLBException(blockPC);
     regs.cop0.FireException(Cop0::GetTLBExceptionCode(regs.cop0.tlbError, Cop0::LOAD), 0, blockPC);
     return 1;*/
-    Util::Error::GetInstance().Throw(
+    Util::Error::GetInstance().Throw({Util::Error::Severity::NON_FATAL}, {Util::Error::Type::UNHANDLED_EXCEPTION},
-      {Util::Error::Severity::NON_FATAL}, {Util::Error::Type::UNHANDLED_EXCEPTION}, blockPC, {},
+                                     blockPC, {},
-      "[JIT]: Unhandled exception TLB exception {} when retrieving PC physical address!",
+                                     "[JIT]: Unhandled exception TLB exception {} when retrieving PC physical address!",
-      static_cast<int>(Cop0::GetTLBExceptionCode(regs.cop0.tlbError, Cop0::LOAD)));
+                                     static_cast<int>(Cop0::GetTLBExceptionCode(regs.cop0.tlbError, Cop0::LOAD)));
     return 0;
   }
 
-  u32 upperIndex = paddr >> kUpperShift;
+  const u32 upperIndex = paddr >> kUpperShift;
-  u32 lowerIndex = paddr & kLowerMask;
+  const u32 lowerIndex = paddr & kLowerMask;
 
   if (!blockCache[upperIndex].empty()) {
     if (blockCache[upperIndex][lowerIndex]) {
@@ -130,7 +137,7 @@ int JIT::Step() {
     blockCache[upperIndex].resize(kLowerSize);
   }
 
-  trace("[JIT]: Compiling block @ 0x{:016X}:", blockPC);
+  info("[JIT]: Compiling block @ 0x{:016X}:", static_cast<u64>(blockPC));
   const auto blockInfo = code.getCurr();
   const auto block = code.getCurr<BlockFn>();
   blockCache[upperIndex][lowerIndex] = block;
@@ -138,7 +145,6 @@ int JIT::Step() {
   code.setProtectModeRW();
 
   u32 instructionsInBlock = 0;
-  u32 instruction = 0;
 
   bool instrEndsBlock = false;
 
@@ -146,78 +152,70 @@ int JIT::Step() {
   code.push(code.rbp);
   code.mov(code.rbp, reinterpret_cast<uintptr_t>(this)); // Load context pointer
 
-  //cs_insn *insn;
+  cs_insn *insn;
-  trace("\tMIPS code (guest PC = 0x{:016X}):", blockPC);
+  info("\tMIPS code (guest PC = 0x{:016X}):", static_cast<u64>(blockPC));
-  while (!instrEndsBlock) {
+  
-    // CheckCompareInterrupt();
+  emitMemberFunctionCall(&JIT::AdvanceDelaySlot, this);
+
+  while (true) {
     paddr = 0;
 
-    if (check_address_error(0b11, u64(blockPC))) [[unlikely]] {
+    auto instruction = FetchInstruction(blockPC);
-      /*regs.cop0.HandleTLBException(blockPC);
-      regs.cop0.FireException(ExceptionCode::AddressErrorLoad, 0, blockPC);
-      return 1;*/
-      Util::Error::GetInstance().Throw(
-        {Util::Error::Severity::NON_FATAL}, {Util::Error::Type::UNHANDLED_EXCEPTION}, blockPC, {},
-        "[JIT]: Unhandled exception ADL due to unaligned PC virtual value!");
-      return 0;
-    }
 
-    if (!regs.cop0.MapVAddr(Cop0::LOAD, blockPC, paddr)) {
+    if(!instruction)
-      /*regs.cop0.HandleTLBException(blockPC);
-      regs.cop0.FireException(Cop0::GetTLBExceptionCode(regs.cop0.tlbError, Cop0::LOAD), 0, blockPC);
-      return 1;*/
-      Util::Error::GetInstance().Throw(
-        {Util::Error::Severity::NON_FATAL}, {Util::Error::Type::UNHANDLED_EXCEPTION}, blockPC, {},
-        "[JIT]: Unhandled exception TLB exception {} when retrieving PC physical address!",
-        static_cast<int>(Cop0::GetTLBExceptionCode(regs.cop0.tlbError, Cop0::LOAD)));
       return 0;
-    }
 
-    instruction = Core::GetMem().Read<u32>(paddr);
     instructionsInBlock++;
 
     blockOldPC = blockPC;
     blockPC = blockNextPC;
     blockNextPC += 4;
 
-    if((instrEndsBlock = InstrEndsBlock(instruction)))
+    if(InstrEndsBlock(instruction.value())) {
-      continue;
+      const auto delay_instruction = FetchInstruction(blockPC); // get instruction in delay slot
+      if(!delay_instruction)
+        return 0;
 
-    trace("{}", Disassembler::GetInstance().DisassembleSimple(paddr, instruction).full);
+      if(InstrEndsBlock(delay_instruction.value())) {
+        Util::Error::GetInstance().Throw(
+          {Util::Error::Severity::NON_FATAL}, {Util::Error::Type::JIT_BRANCH_INSIDE_DELAY_SLOT},
+          blockPC, {}, "[JIT]: Unhandled case of branch from delay slot!");
+        return 0;
+      }
 
-    /*if(ShouldServiceInterrupt()) {
+      instructionsInBlock++;
-      regs.cop0.FireException(ExceptionCode::Interrupt, 0, blockPC);
-      return 1;
-    }*/
 
-    Emit(instruction);
+      blockOldPC = blockPC;
-  }
+      blockPC = blockNextPC;
+      blockNextPC += 4;
 
-  instructionsInBlock++;
+      Emit(delay_instruction.value());
-  const u32 delay_instruction = FetchInstruction();
+      Emit(instruction.value());
-  instrEndsBlock = InstrEndsBlock(delay_instruction);
-  if(instrEndsBlock) {
-    Util::Error::GetInstance().Throw(
-      {Util::Error::Severity::NON_FATAL}, {Util::Error::Type::JIT_BRANCH_INSIDE_DELAY_SLOT},
-      blockPC, {}, "[JIT]: Unhandled case of branch from delay slot!");
-    return 0;
-  }
 
-  blockOldPC = blockPC;
+      if(!branch_taken) {
-  blockPC = blockNextPC;
+        Xbyak::Label runtime_branch_taken;
-  blockNextPC += 4;
+        code.mov(code.SCR1, code.byte[code.rbp + BRANCH_TAKEN_OFFSET()]);
+        code.cmp(code.SCR1, 0);
+        code.jne(runtime_branch_taken);
+        code.mov(code.SCR1, blockOldPC);
+        code.mov(code.ptr[OLD_PC_OFFSET()], code.SCR1);
+        code.mov(code.SCR1, blockPC);
+        code.mov(code.ptr[PC_OFFSET()], code.SCR1);
+        code.mov(code.SCR1, blockNextPC);
+        code.mov(code.ptr[NEXT_PC_OFFSET()], code.SCR1);
+        code.L(runtime_branch_taken);
+      }
 
-  Emit(delay_instruction);
+      if(branch_taken) branch_taken = false;
+      
+      emitMemberFunctionCall(&JIT::AdvanceDelaySlot, this);
 
-  Emit(instruction);
+      break;
-  
+    }
-  if(!regs.delaySlot) {
+
-    code.mov(code.SCR1, blockOldPC);
+    Emit(instruction.value());
-    code.mov(REG(qword, oldPC), code.SCR1);
+
-    code.mov(code.SCR1, blockPC);
+    emitMemberFunctionCall(&JIT::AdvanceDelaySlot, this);
-    code.mov(REG(qword, pc), code.SCR1);
-    code.mov(code.SCR1, blockNextPC);
-    code.mov(REG(qword, nextPC), code.SCR1);
   }
 
   code.mov(code.rax, instructionsInBlock);
@@ -225,22 +223,24 @@ int JIT::Step() {
   code.add(code.rsp, 8);
   code.ret();
   code.setProtectModeRE();
-  //static auto blockInfoSize = 0;
+  static size_t blockInfoSize = 0;
-  //blockInfoSize = code.getSize() - blockInfoSize;
+  blockInfoSize = code.getSize() - blockInfoSize;
 
-  //trace("\tX86 code (block address = 0x{:016X}):", (uintptr_t)block);
+  info("\tX86 code (block address = 0x{:016X}):", reinterpret_cast<uintptr_t>(block));
-  //auto count = cs_disasm(disassemblerX86, blockInfo, blockInfoSize, (uintptr_t)block, 0, &insn);
+  const auto count = cs_disasm(disassemblerX86, blockInfo, blockInfoSize, reinterpret_cast<uintptr_t>(block), 0, &insn);
-  //if (count > 0) {
+  if (count > 0) {
-  //  for (size_t j = 0; j < count; j++) {
+    for (size_t j = 0; j < count; j++) {
-  //    trace("\t\t0x{:016X}:\t{}\t\t{}\n", insn[j].address, insn[j].mnemonic, insn[j].op_str);
+      info("\t\t0x{:016X}:\t{}\t\t{}", insn[j].address, insn[j].mnemonic, insn[j].op_str);
-  //  }
+    }
-  //
+  
-  //  cs_free(insn, count);
+    cs_free(insn, count);
-  //}
+  }
-  // const auto dump = code.getCode();
-  // Util::WriteFileBinary(dump, code.getSize(), "jit.dump");
   //  panic("");
   return block();
 }
 #endif
+
+void JIT::DumpBlockCacheToDisk() const {
+  Util::WriteFileBinary(code.getCode<u8*>(), code.getSize(), "jit.dump");
+}
 } // namespace n64

src/backend/core/JIT.hpp

@@ -16,15 +16,21 @@ static constexpr u32 kUpperSize = kAddressSpaceSize >> kUpperShift; // 0x800000
 static constexpr u32 kLowerSize = 0x100; // 0x80
 static constexpr u32 kCodeCacheSize = 32_mb;
 static constexpr u32 kCodeCacheAllocSize = kCodeCacheSize + 4_kb;
-#define REG(acc, x) code.acc[code.rbp + (reinterpret_cast<uintptr_t>(&regs.x) - (uintptr_t)this)]
+#define OLD_PC_OFFSET() (reinterpret_cast<uintptr_t>(&regs.oldPC))
+#define PC_OFFSET() (reinterpret_cast<uintptr_t>(&regs.pc))
+#define NEXT_PC_OFFSET() (reinterpret_cast<uintptr_t>(&regs.nextPC))
+#define GPR_OFFSET(x) (reinterpret_cast<uintptr_t>(&regs.gpr[(x)]))
+#define BRANCH_TAKEN_OFFSET() (reinterpret_cast<uintptr_t>(&branch_taken) - reinterpret_cast<uintptr_t>(this))
+#define HI_OFFSET() (reinterpret_cast<uintptr_t>(&regs.hi))
+#define LO_OFFSET() (reinterpret_cast<uintptr_t>(&regs.lo))
 
 #ifdef __aarch64__
 struct JIT : BaseCPU {};
 #else
-struct JIT : BaseCPU {
+struct JIT final : BaseCPU {
-  explicit JIT(ParallelRDP&, Mem&, Registers&);
+  explicit JIT(Mem&, Registers&);
   ~JIT() override = default;
-  int Step() override;
+  u32 Step() override;
 
   void Reset() override {
     code.reset();
@@ -32,30 +38,38 @@ struct JIT : BaseCPU {
     blockCache.resize(kUpperSize);
   }
 
+  void DumpBlockCacheToDisk() const;
+
+  void AdvanceDelaySlot() {
+    regs.prevDelaySlot = regs.delaySlot;
+    regs.delaySlot = false;
+  }
+
   void InvalidateBlock(u32);
 private:
   Registers& regs;
   Mem& mem;
   Xbyak::CodeGenerator code{kCodeCacheAllocSize};
   u64 cop2Latch{};
-  u64 blockOldPC = 0, blockPC = 0, blockNextPC = 0;
+  s64 blockOldPC = 0, blockPC = 0, blockNextPC = 0;
   friend struct Cop1;
   friend struct Registers;
   using BlockFn = int (*)();
   std::vector<std::vector<BlockFn>> blockCache;
   Xbyak::Label branch_likely_not_taken;
-  csh disassemblerMips, disassemblerX86;
+  bool branch_taken;
+  csh disassemblerMips{}, disassemblerX86{};
 
   template <typename T>
-  Xbyak::Address GPR(const size_t index) const {
+  Xbyak::Address GPR(const size_t index) {
     if constexpr (sizeof(T) == 1) {
-      return code.byte[code.rbp + (reinterpret_cast<uintptr_t>(&regs.gpr[index]) - reinterpret_cast<uintptr_t>(this))];
+      return code.byte[GPR_OFFSET(index)];
     } else if constexpr (sizeof(T) == 2) {
-      return code.word[code.rbp + (reinterpret_cast<uintptr_t>(&regs.gpr[index]) - reinterpret_cast<uintptr_t>(this))];
+      return code.word[GPR_OFFSET(index)];
     } else if constexpr (sizeof(T) == 4) {
-      return code.dword[code.rbp + (reinterpret_cast<uintptr_t>(&regs.gpr[index]) - reinterpret_cast<uintptr_t>(this))];
+      return code.dword[GPR_OFFSET(index)];
     } else if constexpr (sizeof(T) == 8) {
-      return code.qword[code.rbp + (reinterpret_cast<uintptr_t>(&regs.gpr[index]) - reinterpret_cast<uintptr_t>(this))];
+      return code.ptr[GPR_OFFSET(index)];
     }
 
     Util::Error::GetInstance().Throw(
@@ -106,89 +120,89 @@ private:
   (((!regs.cop0.is64BitAddressing) && (s32)(vaddr) != (vaddr)) || (((vaddr) & (mask)) != 0))
 
   [[nodiscard]] bool ShouldServiceInterrupt() const;
-  void CheckCompareInterrupt();
+  void CheckCompareInterrupt() const;
-  u32 FetchInstruction();
+  std::optional<u32> FetchInstruction(s64);
 
-  void Emit(const Instruction);
+  void Emit(Instruction);
-  void special(const Instruction);
+  void special(Instruction);
-  void regimm(const Instruction);
+  void regimm(Instruction);
-  void add(const Instruction);
+  void add(Instruction);
-  void addu(const Instruction);
+  void addu(Instruction);
-  void addi(const Instruction);
+  void addi(Instruction);
-  void addiu(const Instruction);
+  void addiu(Instruction);
-  void andi(const Instruction);
+  void andi(Instruction);
-  void and_(const Instruction);
+  void and_(Instruction);
   void branch_constant(bool cond, s64 offset);
   void branch_likely_constant(bool cond, s64 offset);
   void branch_abs_constant(bool cond, s64 address);
-  void bltz(const Instruction);
+  void bltz(Instruction);
-  void bgez(const Instruction);
+  void bgez(Instruction);
-  void bltzl(const Instruction);
+  void bltzl(Instruction);
-  void bgezl(const Instruction);
+  void bgezl(Instruction);
-  void bltzal(const Instruction);
+  void bltzal(Instruction);
-  void bgezal(const Instruction);
+  void bgezal(Instruction);
-  void bltzall(const Instruction);
+  void bltzall(Instruction);
-  void bgezall(const Instruction);
+  void bgezall(Instruction);
-  void beq(const Instruction);
+  void beq(Instruction);
-  void beql(const Instruction);
+  void beql(Instruction);
-  void bne(const Instruction);
+  void bne(Instruction);
-  void bnel(const Instruction);
+  void bnel(Instruction);
-  void blez(const Instruction);
+  void blez(Instruction);
-  void blezl(const Instruction);
+  void blezl(Instruction);
-  void bgtz(const Instruction);
+  void bgtz(Instruction);
-  void bgtzl(const Instruction);
+  void bgtzl(Instruction);
-  void bfc1(const Instruction);
+  void bfc1(Instruction);
-  void blfc1(const Instruction);
+  void blfc1(Instruction);
-  void bfc0(const Instruction);
+  void bfc0(Instruction);
-  void blfc0(const Instruction);
+  void blfc0(Instruction);
-  void dadd(const Instruction);
+  void dadd(Instruction);
-  void daddu(const Instruction);
+  void daddu(Instruction);
-  void daddi(const Instruction);
+  void daddi(Instruction);
-  void daddiu(const Instruction);
+  void daddiu(Instruction);
-  void ddiv(const Instruction);
+  void ddiv(Instruction);
-  void ddivu(const Instruction);
+  void ddivu(Instruction);
-  void div(const Instruction);
+  void div(Instruction);
-  void divu(const Instruction);
+  void divu(Instruction);
-  void dmult(const Instruction);
+  void dmult(Instruction);
-  void dmultu(const Instruction);
+  void dmultu(Instruction);
-  void dsll(const Instruction);
+  void dsll(Instruction);
-  void dsllv(const Instruction);
+  void dsllv(Instruction);
-  void dsll32(const Instruction);
+  void dsll32(Instruction);
-  void dsra(const Instruction);
+  void dsra(Instruction);
-  void dsrav(const Instruction);
+  void dsrav(Instruction);
-  void dsra32(const Instruction);
+  void dsra32(Instruction);
-  void dsrl(const Instruction);
+  void dsrl(Instruction);
-  void dsrlv(const Instruction);
+  void dsrlv(Instruction);
-  void dsrl32(const Instruction);
+  void dsrl32(Instruction);
-  void dsub(const Instruction);
+  void dsub(Instruction);
-  void dsubu(const Instruction);
+  void dsubu(Instruction);
-  void j(const Instruction);
+  void j(Instruction);
-  void jr(const Instruction);
+  void jr(Instruction);
-  void jal(const Instruction);
+  void jal(Instruction);
-  void jalr(const Instruction);
+  void jalr(Instruction);
-  void lui(const Instruction);
+  void lui(Instruction);
-  void lbu(const Instruction);
+  void lbu(Instruction);
-  void lb(const Instruction);
+  void lb(Instruction);
-  void ld(const Instruction);
+  void ld(Instruction);
-  void ldc1(const Instruction);
+  void ldc1(Instruction);
-  void ldl(const Instruction);
+  void ldl(Instruction);
-  void ldr(const Instruction);
+  void ldr(Instruction);
-  void lh(const Instruction);
+  void lh(Instruction);
-  void lhu(const Instruction);
+  void lhu(Instruction);
-  void ll(const Instruction);
+  void ll(Instruction);
-  void lld(const Instruction);
+  void lld(Instruction);
-  void lw(const Instruction);
+  void lw(Instruction);
-  void lwc1(const Instruction);
+  void lwc1(Instruction);
-  void lwl(const Instruction);
+  void lwl(Instruction);
-  void lwu(const Instruction);
+  void lwu(Instruction);
-  void lwr(const Instruction);
+  void lwr(Instruction);
-  void mfhi(const Instruction);
+  void mfhi(Instruction);
-  void mflo(const Instruction);
+  void mflo(Instruction);
-  void mult(const Instruction);
+  void mult(Instruction);
-  void multu(const Instruction);
+  void multu(Instruction);
-  void mthi(const Instruction);
+  void mthi(Instruction);
-  void mtlo(const Instruction);
+  void mtlo(Instruction);
-  void nor(const Instruction);
+  void nor(Instruction);
   void sb(const Instruction) {
     Util::Error::GetInstance().Throw(
       {Util::Error::Severity::NON_FATAL}, {Util::Error::Type::UNHANDLED_INSTRUCTION},
@@ -229,7 +243,7 @@ private:
       {Util::Error::Severity::NON_FATAL}, {Util::Error::Type::UNHANDLED_INSTRUCTION},
       blockPC, {}, "[JIT]: Unhandled 'sh'!");
   }
-  void sw(const Instruction);
+  void sw(Instruction);
   void swl(const Instruction) {
     Util::Error::GetInstance().Throw(
       {Util::Error::Severity::NON_FATAL}, {Util::Error::Type::UNHANDLED_INSTRUCTION},
@@ -240,32 +254,32 @@ private:
       {Util::Error::Severity::NON_FATAL}, {Util::Error::Type::UNHANDLED_INSTRUCTION},
       blockPC, {}, "[JIT]: Unhandled 'swr'!");
   }
-  void slti(const Instruction);
+  void slti(Instruction);
-  void sltiu(const Instruction);
+  void sltiu(Instruction);
-  void slt(const Instruction);
+  void slt(Instruction);
-  void sltu(const Instruction);
+  void sltu(Instruction);
-  void sll(const Instruction);
+  void sll(Instruction);
-  void sllv(const Instruction);
+  void sllv(Instruction);
-  void sub(const Instruction);
+  void sub(Instruction);
-  void subu(const Instruction);
+  void subu(Instruction);
   void swc1(const Instruction) {
     Util::Error::GetInstance().Throw(
       {Util::Error::Severity::NON_FATAL}, {Util::Error::Type::JIT_BRANCH_INSIDE_DELAY_SLOT},
       blockPC, {}, "[JIT]: Unhandled case of branch from delay slot!");
   }
-  void sra(const Instruction);
+  void sra(Instruction);
-  void srav(const Instruction);
+  void srav(Instruction);
-  void srl(const Instruction);
+  void srl(Instruction);
-  void srlv(const Instruction);
+  void srlv(Instruction);
   void trap(bool) {
     Util::Error::GetInstance().Throw(
       {Util::Error::Severity::NON_FATAL}, {Util::Error::Type::JIT_BRANCH_INSIDE_DELAY_SLOT},
       blockPC, {}, "[JIT]: Unhandled case of branch from delay slot!");
   }
-  void or_(const Instruction);
+  void or_(Instruction);
-  void ori(const Instruction);
+  void ori(Instruction);
-  void xor_(const Instruction);
+  void xor_(Instruction);
-  void xori(const Instruction);
+  void xori(Instruction);
 };
 #endif
 } // namespace n64

src/backend/core/Mem.cpp

@@ -6,7 +6,7 @@
 #include <Options.hpp>
 
 namespace n64 {
-Mem::Mem(JIT *jit) : flash(saveData), jit(jit) {
+Mem::Mem() : flash(saveData) {
   rom.cart.resize(CART_SIZE);
   std::ranges::fill(rom.cart, 0);
 }

src/backend/core/Mem.hpp

@@ -78,10 +78,11 @@ struct JIT;
 
 struct Mem {
   ~Mem() = default;
-  Mem(JIT * = nullptr);
+  Mem();
   void Reset();
   void LoadSRAM(SaveType, fs::path);
   void LoadROM(bool, const std::string &);
+  void SetJIT(JIT* jit) { this->jit = jit; }
   [[nodiscard]] auto GetRDRAMPtr() -> u8 * { return mmio.rdp.rdram.data(); }
 
   [[nodiscard]] auto GetRDRAM() -> std::vector<u8> & { return mmio.rdp.rdram; }

src/backend/core/RSP.hpp

@@ -135,7 +135,7 @@ struct RSP {
   void SetVTE(const VPR &vt, u8 e);
   auto Read(u32 addr) -> u32;
   void Write(u32 addr, u32 val);
-  void Exec(const Instruction instr);
+  void Exec(Instruction instr);
   SPStatus spStatus{};
   u16 oldPC{}, pc{}, nextPC{};
   SPDMASPAddr spDMASPAddr{};
@@ -151,7 +151,7 @@ struct RSP {
   VPR vce{};
   s16 divIn{}, divOut{};
   bool divInLoaded = false;
-  int steps = 0;
+  u32 steps = 0;
 
   struct {
     VPR h{}, m{}, l{};
@@ -257,12 +257,12 @@ struct RSP {
 
   FORCE_INLINE void ReleaseSemaphore() { semaphore = false; }
 
-  void special(const Instruction instr);
+  void special(Instruction instr);
-  void regimm(const Instruction instr);
+  void regimm(Instruction instr);
-  void lwc2(const Instruction instr);
+  void lwc2(Instruction instr);
-  void swc2(const Instruction instr);
+  void swc2(Instruction instr);
-  void cop2(const Instruction instr);
+  void cop2(Instruction instr);
-  void cop0(const Instruction instr);
+  void cop0(Instruction instr);
   
   void add(Instruction instr);
   void addi(Instruction instr);

src/backend/core/jit/decode.cpp

@@ -270,7 +270,54 @@ void JIT::Emit(const Instruction instr) {
     lui(instr);
     break;
   case Instruction::COP0:
-    panic("[JIT]: Unimplemented Cop0 decode");
+    switch (instr.cop_rs()) {
+      case 0x00:
+        code.mov(code.ARG2, instr);
+        emitMemberFunctionCall(&Cop0::mfc0, &regs.cop0);
+        break;
+      case 0x01:
+        code.mov(code.ARG2, instr);
+        emitMemberFunctionCall(&Cop0::dmfc0, &regs.cop0);
+        break;
+      case 0x04:
+        code.mov(code.ARG2, instr);
+        emitMemberFunctionCall(&Cop0::mtc0, &regs.cop0);
+        break;
+      case 0x05:
+        code.mov(code.ARG2, instr);
+        emitMemberFunctionCall(&Cop0::dmtc0, &regs.cop0);
+        break;
+      case 0x10 ... 0x1F:
+        switch (instr.cop_funct()) {
+          case 0x01:
+            emitMemberFunctionCall(&Cop0::tlbr, &regs.cop0);
+            break;
+          case 0x02:
+            code.mov(code.ARG2, COP0_REG_INDEX);
+            emitMemberFunctionCall(&Cop0::GetReg32, &regs.cop0);
+            code.mov(code.ARG2, code.rax);
+            code.and_(code.ARG2, 0x3F);
+            emitMemberFunctionCall(&Cop0::tlbw, &regs.cop0);
+            break;
+          case 0x06:
+            emitMemberFunctionCall(&Cop0::GetRandom, &regs.cop0);
+            code.mov(code.ARG2, code.rax);
+            emitMemberFunctionCall(&Cop0::tlbw, &regs.cop0);
+            break;
+          case 0x08:
+            emitMemberFunctionCall(&Cop0::tlbp, &regs.cop0);
+            break;
+          case 0x18:
+            emitMemberFunctionCall(&Cop0::eret, &regs.cop0);
+            break;
+          default:
+            panic("Unimplemented COP0 function {} ({:08X}) ({:016X})", instr.cop_funct(), u32(instr),
+                        regs.oldPC);
+        }
+        break;
+      default:
+        panic("Unimplemented COP0 instruction {}", instr.cop_rs());
+    }
     break;
   case Instruction::COP1:
     {
@@ -411,6 +458,7 @@ void JIT::Emit(const Instruction instr) {
     sd(instr);
     break;
   default:
+    DumpBlockCacheToDisk();
     panic("Unimplemented instruction {:02X} ({:08X}) (pc: {:016X})", instr.opcode(), u32(instr), static_cast<u64>(regs.oldPC));
   }
 }

src/backend/core/jit/instructions.cpp

@@ -159,7 +159,7 @@ void JIT::bfc0(const Instruction instr) {
   const s16 imm = instr;
   const s64 offset = u64((s64)imm) << 2;
   const s64 address = blockPC + offset;
-  // code.mov(code.al, REG(byte, cop1.fcr31.compare));
+  // code.mov(code.al, code.byte[code.rbp + REG_OFFSET(cop1.fcr31.compare));
   // code.test(code.al, code.al);
   // branch(address, z);
 }
@@ -168,7 +168,7 @@ void JIT::blfc0(const Instruction instr) {
   const s16 imm = instr;
   const s64 offset = u64((s64)imm) << 2;
   const s64 address = blockPC + offset;
-  // code.mov(code.al, REG(byte, cop1.fcr31.compare));
+  // code.mov(code.al, code.byte[code.rbp + REG_OFFSET(cop1.fcr31.compare));
   // code.test(code.al, code.al);
   // branch_likely(address, z);
 }
@@ -177,7 +177,7 @@ void JIT::bfc1(const Instruction instr) {
   const s16 imm = instr;
   const s64 offset = u64((s64)imm) << 2;
   const s64 address = blockPC + offset;
-  // code.mov(code.al, REG(byte, cop1.fcr31.compare));
+  // code.mov(code.al, code.byte[code.rbp + REG_OFFSET(cop1.fcr31.compare));
   // code.test(code.al, code.al);
   // branch(address, nz);
 }
@@ -186,87 +186,95 @@ void JIT::blfc1(const Instruction instr) {
   const s16 imm = instr;
   const s64 offset = u64((s64)imm) << 2;
   const s64 address = blockPC + offset;
-  // code.mov(code.al, REG(byte, cop1.fcr31.compare));
+  // code.mov(code.al, code.byte[code.rbp + REG_OFFSET(cop1.fcr31.compare));
   // code.test(code.al, code.al);
   // branch_likely(address, nz);
 }
 
 void JIT::BranchNotTaken() {}
-void JIT::BranchTaken(s64 offs) {
+
-  code.mov(code.SCR1, REG(qword, pc));
+void JIT::BranchTaken(const s64 offs) {
-  code.add(code.SCR1, offs);
+  code.mov(code.SCR2, code.qword[PC_OFFSET()]);
-  SetPC64(offs);
+  code.add(code.SCR2, offs);
+  SetPC64(code.SCR2);
 }
 
 void JIT::BranchTaken(const Xbyak::Reg64 &offs) {
-  code.mov(code.SCR1, REG(qword, pc));
+  code.mov(code.SCR2, code.qword[PC_OFFSET()]);
-  code.add(code.SCR1, offs);
+  code.add(code.SCR2, offs);
-  SetPC64(offs);
+  SetPC64(code.SCR2);
 }
 
-void JIT::BranchAbsTaken(s64 addr) {
+void JIT::BranchAbsTaken(const s64 addr) {
-  code.add(code.SCR1, addr);
+  SetPC64(addr);
-  code.mov(REG(qword, nextPC), code.SCR1);
 }
 
 void JIT::BranchAbsTaken(const Xbyak::Reg64 &addr) {
-  code.mov(REG(qword, nextPC), addr);
+  SetPC64(addr);
 }
 
-void JIT::branch_constant(const bool cond, s64 offset) {
+void JIT::branch_constant(const bool cond, const s64 offset) {
   if(cond) {
     regs.delaySlot = true;
     BranchTaken(offset);
+    branch_taken = true;
   }
 }
 
-void JIT::branch_likely_constant(bool cond, s64 offset) {
+void JIT::branch_abs_constant(const bool cond, const s64 address) {
+  if(cond) {
+    regs.delaySlot = true;
+    BranchAbsTaken(address);
+    branch_taken = true;
+  }
+}
+
+void JIT::branch_likely_constant(const bool cond, const s64 offset) {
   if(cond) {
     regs.delaySlot = true;
     BranchTaken(offset);
+    branch_taken = true;
   } else {
     SetPC64(blockNextPC);
   }
 }
 
-void JIT::branch_abs_constant(bool cond, s64 address) {
-  if(cond) {
-    regs.delaySlot = true;
-    BranchAbsTaken(address);
-  }
-}
-
 #define branch(offs, cond) do { \
   Xbyak::Label taken, not_taken; \
-  code.j##cond(taken); \
+  code.j## cond(taken); \
   code.jmp(not_taken); \
   code.L(taken); \
   BranchTaken(offs); \
+  code.mov(code.byte[code.rbp + BRANCH_TAKEN_OFFSET()], 1); \
   code.L(not_taken); \
 } while(0)
 
 #define branch_abs(addr, cond) do { \
   Xbyak::Label taken, not_taken; \
-  code.j##cond(taken); \
+  code.j## cond(taken); \
   code.jmp(not_taken); \
   code.L(taken); \
   BranchAbsTaken(addr); \
+  code.mov(code.byte[code.rbp + BRANCH_TAKEN_OFFSET()], 1); \
   code.L(not_taken); \
 } while(0)
 
 #define branch_likely(offs, cond) do { \
-  Xbyak::Label taken, not_taken; \
+  Xbyak::Label taken, not_taken, end; \
-  code.j##cond(taken); \
+  code.j## cond(taken); \
   code.jmp(not_taken); \
   code.L(taken); \
   BranchTaken(offs); \
+  code.mov(code.byte[code.rbp + BRANCH_TAKEN_OFFSET()], 1); \
+  code.jmp(end); \
   code.L(not_taken); \
   SetPC64(blockNextPC); \
+  code.L(end); \
 } while(0)
 
 void JIT::bltz(const Instruction instr) {
   const s16 imm = instr;
-  const s64 offset = u64((s64)imm) << 2;
+  const s64 offset = static_cast<s64>(imm) << 2;
   if (regs.IsRegConstant(instr.rs())) {
     branch_constant(regs.Read<s64>(instr.rs()) < 0, offset);
     return;
@@ -435,7 +443,7 @@ void JIT::beql(const Instruction instr) {
 
 void JIT::bne(const Instruction instr) {
   const s16 imm = instr;
-  const s64 offset = u64((s64)imm) << 2;
+  const s64 offset = static_cast<s64>(imm) << 2;
   if (regs.IsRegConstant(instr.rs()) && regs.IsRegConstant(instr.rt())) {
     branch_constant(regs.Read<s64>(instr.rs()) != regs.Read<s64>(instr.rt()), offset);
     return;
@@ -857,13 +865,13 @@ void JIT::dsubu(const Instruction instr) {
 
 void JIT::j(const Instruction instr) {
   const s32 target = (instr & 0x3ffffff) << 2;
-  const s64 address = (blockOldPC & ~0xfffffff) | target;
+  const s64 address = blockOldPC & ~0xfffffff | target;
   branch_abs_constant(true, address);
 }
 
 void JIT::jr(const Instruction instr) {
   if (regs.IsRegConstant(instr.rs())) {
-    const u64 address = regs.Read<s64>(instr.rs());
+    const auto address = regs.Read<s64>(instr.rs());
     branch_abs_constant(true, address);
     return;
   }
@@ -891,9 +899,7 @@ void JIT::lbu(const Instruction instr) {
       // regs.cop0.FireException(Cop0::GetTLBExceptionCode(regs.cop0.tlbError, Cop0::LOAD), 0, regs.oldPC);
       panic("[JIT]: Unhandled TLBL exception in LBU!");
     } else {
-      code.mov(code.ARG2,
+      code.mov(code.ARG2, paddr);
-               code.ptr[code.rbp + (reinterpret_cast<uintptr_t>(&regs) - reinterpret_cast<uintptr_t>(this))]);
-      code.mov(code.ARG3, paddr);
       emitMemberFunctionCall(&Mem::Read<u8>, &mem);
       regs.Write<u8>(instr.rt(), code.rax);
     }
@@ -908,11 +914,9 @@ void JIT::lb(const Instruction instr) {
     if (u32 paddr = 0; !regs.cop0.MapVAddr(Cop0::LOAD, address, paddr)) {
       // regs.cop0.HandleTLBException(address);
       // regs.cop0.FireException(Cop0::GetTLBExceptionCode(regs.cop0.tlbError, Cop0::LOAD), 0, regs.oldPC);
-      panic("[JIT]: Unhandled TLBL exception in LB (pc: 0x{:016X})!", blockPC);
+      panic("[JIT]: Unhandled TLBL exception in LB (pc: 0x{:016X})!", static_cast<u64>(blockPC));
     } else {
-      code.mov(code.ARG2,
+      code.mov(code.ARG2, paddr);
-               code.ptr[code.rbp + (reinterpret_cast<uintptr_t>(&regs) - reinterpret_cast<uintptr_t>(this))]);
-      code.mov(code.ARG3, paddr);
       emitMemberFunctionCall(&Mem::Read<u8>, &mem);
       regs.Write<s8>(instr.rt(), code.rax);
     }
@@ -937,9 +941,7 @@ void JIT::ld(const Instruction instr) {
       // regs.cop0.FireException(Cop0::GetTLBExceptionCode(regs.cop0.tlbError, Cop0::LOAD), 0, regs.oldPC);
       panic("[JIT]: Unhandled TLBL exception in LD!");
     } else {
-      code.mov(code.ARG2,
+      code.mov(code.ARG2, paddr);
-               code.ptr[code.rbp + (reinterpret_cast<uintptr_t>(&regs) - reinterpret_cast<uintptr_t>(this))]);
-      code.mov(code.ARG3, paddr);
       emitMemberFunctionCall(&Mem::Read<u64>, &mem);
       regs.Write<u64>(instr.rt(), code.rax);
     }
@@ -1026,8 +1028,7 @@ void JIT::lh(const Instruction instr) {
       return;
     }
 
-    code.mov(code.ARG2, code.ptr[code.rbp + (reinterpret_cast<uintptr_t>(&regs) - reinterpret_cast<uintptr_t>(this))]);
+    code.mov(code.ARG2, paddr);
-    code.mov(code.ARG3, paddr);
     emitMemberFunctionCall(&Mem::Read<u16>, &mem);
     regs.Write<s16>(instr.rt(), code.rax);
     return;
@@ -1052,8 +1053,7 @@ void JIT::lhu(const Instruction instr) {
       return;
     }
 
-    code.mov(code.ARG2, code.ptr[code.rbp + (reinterpret_cast<uintptr_t>(&regs) - reinterpret_cast<uintptr_t>(this))]);
+    code.mov(code.ARG2, paddr);
-    code.mov(code.ARG3, paddr);
     emitMemberFunctionCall(&Mem::Read<u16>, &mem);
     regs.Write<u16>(instr.rt(), code.rax);
   }
@@ -1064,8 +1064,7 @@ void JIT::lhu(const Instruction instr) {
   code.mov(code.ARG4, reinterpret_cast<uintptr_t>(&paddr));
   emitMemberFunctionCall(&Cop0::MapVAddr, &regs.cop0);
 
-  code.mov(code.ARG2, code.ptr[code.rbp + (reinterpret_cast<uintptr_t>(&regs) - reinterpret_cast<uintptr_t>(this))]);
+  code.mov(code.ARG2, code.ptr[code.ARG4]);
-  code.mov(code.ARG3, code.qword[code.ARG4]);
   emitMemberFunctionCall(&Mem::Read<u16>, &mem);
   regs.Write<u16>(instr.rt(), code.rax);
 }
@@ -1094,8 +1093,7 @@ void JIT::lw(const Instruction instr) {
       return;
     }
 
-    code.mov(code.ARG2, code.ptr[code.rbp + (reinterpret_cast<uintptr_t>(&regs) - reinterpret_cast<uintptr_t>(this))]);
+    code.mov(code.ARG2, paddr);
-    code.mov(code.ARG3, paddr);
     emitMemberFunctionCall(&Mem::Read<u32>, &mem);
     regs.Write<s32>(instr.rt(), code.rax);
     return;
@@ -1109,8 +1107,7 @@ void JIT::lw(const Instruction instr) {
   code.mov(code.ARG4, reinterpret_cast<uintptr_t>(&paddr));
   emitMemberFunctionCall(&Cop0::MapVAddr, &regs.cop0);
 
-  code.mov(code.ARG2, code.ptr[code.rbp + (reinterpret_cast<uintptr_t>(&regs) - reinterpret_cast<uintptr_t>(this))]);
+  code.mov(code.ARG2, code.ptr[code.ARG4]);
-  code.mov(code.ARG3, code.qword[code.ARG4]);
   emitMemberFunctionCall(&Mem::Read<u32>, &mem);
   regs.Write<s32>(instr.rt(), code.rax);
 }
@@ -1127,7 +1124,7 @@ void JIT::mfhi(const Instruction instr) {
   if (regs.GetHIConstant()) {
     regs.Write(instr.rd(), regs.hi);
   } else {
-    code.mov(code.SCR1, REG(qword, hi));
+    code.mov(code.SCR1, code.ptr[HI_OFFSET()]);
     regs.Write<s64>(instr.rd(), code.SCR1);
   }
 }
@@ -1136,7 +1133,7 @@ void JIT::mflo(const Instruction instr) {
   if (regs.GetLOConstant()) {
     regs.Write(instr.rd(), regs.lo);
   } else {
-    code.mov(code.SCR1, REG(qword, lo));
+    code.mov(code.SCR1, code.ptr[LO_OFFSET()]);
     regs.Write<s64>(instr.rd(), code.SCR1);
   }
 }
@@ -1175,7 +1172,7 @@ void JIT::mthi(const Instruction instr) {
     regs.SetHIConstant();
   } else {
     regs.Read<s64>(instr.rs(), code.SCR1);
-    code.mov(REG(qword, hi), code.SCR1);
+    code.mov(code.ptr[HI_OFFSET()], code.SCR1);
     regs.UnsetHIConstant();
   }
 }
@@ -1186,7 +1183,7 @@ void JIT::mtlo(const Instruction instr) {
     regs.SetLOConstant();
   } else {
     regs.Read<s64>(instr.rs(), code.SCR1);
-    code.mov(REG(qword, lo), code.SCR1);
+    code.mov(code.ptr[LO_OFFSET()], code.SCR1);
     regs.UnsetLOConstant();
   }
 }
@@ -1357,10 +1354,8 @@ void JIT::sw(const Instruction instr) {
       // regs.cop0.FireException(Cop0::GetTLBExceptionCode(regs.cop0.tlbError, Cop0::STORE), 0, regs.oldPC);
       panic("[JIT]: Unhandled TLBS exception in SW!");
     } else {
-      code.lea(code.ARG2,
+      code.mov(code.ARG2, physical);
-               code.ptr[code.rbp + (reinterpret_cast<uintptr_t>(&regs) - reinterpret_cast<uintptr_t>(this))]);
+      regs.Read<s64>(instr.rt(), code.ARG3);
-      code.mov(code.ARG3, physical);
-      regs.Read<s64>(instr.rt(), code.ARG4);
       emitMemberFunctionCall(&Mem::WriteJIT<u32>, &mem);
     }
 
@@ -1382,10 +1377,8 @@ void JIT::sw(const Instruction instr) {
       // regs.cop0.FireException(Cop0::GetTLBExceptionCode(regs.cop0.tlbError, Cop0::STORE), 0, regs.oldPC);
       panic("[JIT]: Unhandled TLBS exception in SW!");
     } else {
-      code.mov(code.ARG2,
+      code.mov(code.ARG2, physical);
-               code.ptr[code.rbp + (reinterpret_cast<uintptr_t>(&regs) - reinterpret_cast<uintptr_t>(this))]);
+      regs.Read<s64>(instr.rt(), code.ARG3);
-      code.mov(code.ARG3, physical);
-      regs.Read<s64>(instr.rt(), code.ARG4);
       emitMemberFunctionCall(&Mem::WriteJIT<u32>, &mem);
     }
 
@@ -1402,9 +1395,8 @@ void JIT::sw(const Instruction instr) {
     code.mov(code.ARG4, reinterpret_cast<uintptr_t>(&physical));
     emitMemberFunctionCall(&Cop0::MapVAddr, &regs.cop0);
 
-    code.mov(code.ARG2, code.ptr[code.rbp + (reinterpret_cast<uintptr_t>(&regs) - reinterpret_cast<uintptr_t>(this))]);
+    code.mov(code.ARG2, code.ptr[code.ARG4]);
-    code.mov(code.ARG3, code.qword[code.ARG4]);
+    regs.Read<s64>(instr.rt(), code.ARG3);
-    regs.Read<s64>(instr.rt(), code.ARG4);
     emitMemberFunctionCall(&Mem::WriteJIT<u32>, &mem);
 
     return;
@@ -1419,9 +1411,8 @@ void JIT::sw(const Instruction instr) {
   code.mov(code.ARG4, reinterpret_cast<uintptr_t>(&physical));
   emitMemberFunctionCall(&Cop0::MapVAddr, &regs.cop0);
 
-  code.mov(code.ARG2, code.ptr[code.rbp + (reinterpret_cast<uintptr_t>(&regs) - reinterpret_cast<uintptr_t>(this))]);
+  code.mov(code.ARG2, code.ptr[code.ARG4]);
-  code.mov(code.ARG3, code.qword[code.ARG4]);
+  regs.Read<s64>(instr.rt(), code.ARG3);
-  regs.Read<s64>(instr.rt(), code.ARG4);
   emitMemberFunctionCall(&Mem::WriteJIT<u32>, &mem);
 }
 

src/backend/core/registers/Cop0.hpp

@@ -264,6 +264,8 @@ struct Cop0 {
   void decode(const Instruction);
 
 private:
+  friend struct JIT;
+
   [[nodiscard]] FORCE_INLINE u32 GetWired() const { return wired & 0x3F; }
   [[nodiscard]] FORCE_INLINE u32 GetCount() const { return u32(u64(count >> 1)); }
 

src/backend/core/registers/Registers.hpp

@@ -18,8 +18,8 @@ struct Registers {
     return regIsConstant & (1 << index);
   }
 
-  [[nodiscard]] bool IsRegConstant(const u32 first, const u32 second) const {
+  [[nodiscard]] bool IsRegConstant(const u32 index1, const u32 index2) const {
-    return IsRegConstant(first) && IsRegConstant(second);
+    return IsRegConstant(index1) && IsRegConstant(index2);
   }
 
   bool GetLOConstant() {

src/frontend/ImGuiImpl/ProgressIndicators.hpp

@@ -0,0 +1,43 @@
+#pragma once
+#include <imgui.h>
+#include <imgui_internal.h>
+
+namespace ImGui {
+inline bool Spinner(const char* label, const float radius, const int thickness, const ImU32& color) {
+  ImGuiWindow* window = GetCurrentWindow();
+  if (window->SkipItems)
+    return false;
+
+  const ImGuiContext & g = *GImGui;
+  const ImGuiStyle& style = g.Style;
+  const ImGuiID id = window->GetID(label);
+
+  const ImVec2 pos = window->DC.CursorPos;
+  const ImVec2 size(radius*2, (radius + style.FramePadding.y)*2);
+
+  const ImRect bb(pos, ImVec2(pos.x + size.x, pos.y + size.y));
+  ItemSize(bb, style.FramePadding.y);
+  if (!ItemAdd(bb, id))
+    return false;
+
+  // Render
+  window->DrawList->PathClear();
+
+  constexpr int num_segments = 30;
+  const int start = abs(ImSin(g.Time*1.8f)*(num_segments-5));
+
+  const float a_min = IM_PI * 2.0f * static_cast<float>(start) / static_cast<float>(num_segments);
+  constexpr float a_max = IM_PI*2.0f * (static_cast<float>(num_segments) -3) / static_cast<float>(num_segments);
+
+  const auto centre = ImVec2(pos.x+radius, pos.y+radius+style.FramePadding.y);
+
+  for (int i = 0; i < num_segments; i++) {
+    const float a = a_min + static_cast<float>(i) / static_cast<float>(num_segments) * (a_max - a_min);
+    window->DrawList->PathLineTo(ImVec2(centre.x + ImCos(a+g.Time*8) * radius,
+                                        centre.y + ImSin(a+g.Time*8) * radius));
+  }
+
+  window->DrawList->PathStroke(color, false, thickness);
+  return true;
+}
+}

src/frontend/ImGuiImpl/StatusBar.hpp

@@ -1,7 +1,6 @@
 #pragma once
 #include <imgui.h>
 #include <imgui_internal.h>
-#include <functional>
 
 namespace ImGui {
 inline bool BeginMainStatusBar()

src/frontend/KaizenGui.cpp

@@ -1,6 +1,7 @@
 #include <KaizenGui.hpp>
 #include <backend/Core.hpp>
 #include <ImGuiImpl/GUI.hpp>
+#include <ImGuiImpl/ProgressIndicators.hpp>
 #include <ImGuiImpl/StatusBar.hpp>
 #include <resources/gamecontrollerdb.h>
 
@@ -16,7 +17,7 @@ KaizenGui::~KaizenGui() {
   SDL_Quit();
 }
 
-void KaizenGui::QueryDevices(SDL_Event event) {
+void KaizenGui::QueryDevices(const SDL_Event &event) {
   switch (event.type) {
   case SDL_EVENT_GAMEPAD_ADDED:
     if (!gamepad) {
@@ -36,8 +37,8 @@ void KaizenGui::QueryDevices(SDL_Event event) {
   }
 }
 
-void KaizenGui::HandleInput(SDL_Event event) {
+void KaizenGui::HandleInput(const SDL_Event &event) {
-  n64::Core& core = n64::Core::GetInstance();
+  const n64::Core& core = n64::Core::GetInstance();
   n64::PIF &pif = n64::Core::GetMem().mmio.si.pif;
   switch(event.type) {
     case SDL_EVENT_GAMEPAD_AXIS_MOTION:
@@ -52,7 +53,7 @@ void KaizenGui::HandleInput(SDL_Event event) {
 
         float xclamped = SDL_GetGamepadAxis(gamepad, SDL_GAMEPAD_AXIS_LEFTX);
         if (xclamped < 0) {
-          xclamped /= float(std::abs(SDL_JOYSTICK_AXIS_MAX));
+          xclamped /= static_cast<float>(std::abs(SDL_JOYSTICK_AXIS_MAX));
         } else {
           xclamped /= SDL_JOYSTICK_AXIS_MAX;
         }
@@ -61,15 +62,15 @@ void KaizenGui::HandleInput(SDL_Event event) {
 
         float yclamped = SDL_GetGamepadAxis(gamepad, SDL_GAMEPAD_AXIS_LEFTY);
         if (yclamped < 0) {
-          yclamped /= float(std::abs(SDL_JOYSTICK_AXIS_MIN));
+          yclamped /= static_cast<float>(std::abs(SDL_JOYSTICK_AXIS_MIN));
         } else {
           yclamped /= SDL_JOYSTICK_AXIS_MAX;
         }
 
         yclamped *= 86;
 
-        pif.UpdateAxis(0, n64::Controller::Axis::Y, -yclamped);
+        pif.UpdateAxis(0, n64::Controller::Axis::Y, static_cast<s8>(-yclamped));
-        pif.UpdateAxis(0, n64::Controller::Axis::X, xclamped);
+        pif.UpdateAxis(0, n64::Controller::Axis::X, static_cast<s8>( xclamped));
       }
       break;
     case SDL_EVENT_GAMEPAD_BUTTON_DOWN:
@@ -90,7 +91,7 @@ void KaizenGui::HandleInput(SDL_Event event) {
     case SDL_EVENT_KEY_DOWN:
     case SDL_EVENT_KEY_UP:
       {
-        auto keys = SDL_GetKeyboardState(nullptr);
+        const auto keys = SDL_GetKeyboardState(nullptr);
         if((keys[SDL_SCANCODE_LCTRL] || keys[SDL_SCANCODE_RCTRL]) && keys[SDL_SCANCODE_O]) {
           fileDialogOpen = true;
         }
@@ -130,15 +131,25 @@ void KaizenGui::HandleInput(SDL_Event event) {
         pif.UpdateButton(0, n64::Controller::Key::DRight, keys[SDL_SCANCODE_L]);
         pif.UpdateButton(0, n64::Controller::Key::LT, keys[SDL_SCANCODE_A]);
         pif.UpdateButton(0, n64::Controller::Key::RT, keys[SDL_SCANCODE_S]);
-        pif.UpdateAxis(0, n64::Controller::Axis::Y, keys[SDL_SCANCODE_UP] ? 86 : keys[SDL_SCANCODE_DOWN] ? -86 : 0);
+
-        pif.UpdateAxis(0, n64::Controller::Axis::X, keys[SDL_SCANCODE_LEFT] ? -86 : keys[SDL_SCANCODE_RIGHT] ? 86 : 0);
+        if (keys[SDL_SCANCODE_UP]) pif.UpdateAxis(0, n64::Controller::Axis::Y, 86);
+        else pif.UpdateAxis(0, n64::Controller::Axis::Y, 0);
+
+        if (keys[SDL_SCANCODE_DOWN]) pif.UpdateAxis(0, n64::Controller::Axis::Y, -86);
+        else pif.UpdateAxis(0, n64::Controller::Axis::Y, 0);
+
+        if (keys[SDL_SCANCODE_LEFT]) pif.UpdateAxis(0, n64::Controller::Axis::X, -86);
+        else pif.UpdateAxis(0, n64::Controller::Axis::X, 0);
+
+        if (keys[SDL_SCANCODE_RIGHT]) pif.UpdateAxis(0, n64::Controller::Axis::X, 86);
+        else pif.UpdateAxis(0, n64::Controller::Axis::X, 0);
       }
       break;
     default: break;
   }
 }
 
-std::tuple<std::optional<u64>, std::optional<Util::Error::MemoryAccess>> RenderErrorMessageDetails() {
+std::pair<std::optional<s64>, std::optional<Util::Error::MemoryAccess>> RenderErrorMessageDetails() {
   auto lastPC = Util::Error::GetLastPC();
   if(lastPC.has_value()) {
     ImGui::Text("%s", std::format("Occurred @ PC = {:016X}", Util::Error::GetLastPC().value()).c_str());
@@ -146,10 +157,10 @@ std::tuple<std::optional<u64>, std::optional<Util::Error::MemoryAccess>> RenderE
 
   auto memoryAccess = Util::Error::GetMemoryAccess();
   if(memoryAccess.has_value()) {
-    auto memoryAccessV = memoryAccess.value();
+    const auto [is_write, size, address, written_val] = memoryAccess.value();
-    ImGui::Text("%s", std::format("{} {}-bit value @ {:08X}{}", memoryAccessV.is_write ? "Writing" : "Reading",
+    ImGui::Text("%s", std::format("{} {}-bit value @ {:08X}{}", is_write ? "Writing" : "Reading",
-                                  (u8)memoryAccessV.size, memoryAccessV.address,
+                                  static_cast<u8>(size), address,
-                                  memoryAccessV.is_write ? std::format(" (value = 0x{:X})", memoryAccessV.written_val) : "")
+                                  is_write ? std::format(" (value = 0x{:X})", written_val) : "")
                                   .c_str());
   }
 
@@ -227,7 +238,7 @@ void KaizenGui::RenderUI() {
   settingsWindow.render();
   debugger.render();
 
-  ImVec2 center = ImGui::GetMainViewport()->GetCenter();
+  const ImVec2 center = ImGui::GetMainViewport()->GetCenter();
   ImGui::SetNextWindowPos(center, ImGuiCond_Appearing, ImVec2(0.5f, 0.5f));
 
   if (ImGui::BeginPopupModal("About Kaizen", &aboutOpen, ImGuiWindowFlags_AlwaysAutoResize)) {
@@ -258,9 +269,9 @@ void KaizenGui::RenderUI() {
         RenderErrorMessageDetails();
 
         if(n64::Core::GetInstance().romLoaded && !n64::Core::GetInstance().pause) {
-          bool ignore = ImGui::Button("Try continuing"); ImGui::SameLine();
+          const bool ignore = ImGui::Button("Try continuing"); ImGui::SameLine();
-          bool stop = ImGui::Button("Stop emulation"); ImGui::SameLine();
+          const bool stop = ImGui::Button("Stop emulation"); ImGui::SameLine();
-          bool chooseAnother = ImGui::Button("Choose another ROM");
+          const bool chooseAnother = ImGui::Button("Choose another ROM");
           if(ignore || stop || chooseAnother) {
             Util::Error::SetHandled();
             ImGui::CloseCurrentPopup();
@@ -306,10 +317,10 @@ void KaizenGui::RenderUI() {
         ImGui::Text(R"(Error message: "%s")", Util::Error::GetError().c_str());
         auto [lastPC, memoryAccess] = RenderErrorMessageDetails();
 
-        bool ignore = ImGui::Button("Try continuing"); ImGui::SameLine();
+        const bool ignore = ImGui::Button("Try continuing"); ImGui::SameLine();
-        bool stop = ImGui::Button("Stop emulation"); ImGui::SameLine();
+        const bool stop = ImGui::Button("Stop emulation"); ImGui::SameLine();
-        bool chooseAnother = ImGui::Button("Choose another ROM");
+        const bool chooseAnother = ImGui::Button("Choose another ROM");
-        bool openInDebugger = lastPC.has_value() ? ImGui::Button("Add breakpoint at this PC and open the debugger") : false;
+        const bool openInDebugger = lastPC.has_value() ? ImGui::Button("Add breakpoint at this PC and open the debugger") : false;
         if(ignore || stop || chooseAnother || openInDebugger) {
           Util::Error::SetHandled();
           ImGui::CloseCurrentPopup();
@@ -346,6 +357,26 @@ void KaizenGui::RenderUI() {
     ImGui::EndMainStatusBar();
   }
 
+  if (shouldDisplaySpinner) {
+    ImGui::SetNextWindowPos({static_cast<float>(width) * 0.5f, static_cast<float>(height) * 0.5f}, 0, ImVec2(0.5f, 0.5f));
+    ImGui::PushStyleColor(ImGuiCol_WindowBg, IM_COL32_BLACK_TRANS);
+    ImGui::PushStyleVar(ImGuiStyleVar_WindowBorderSize, 0.0f);
+
+    ImGui::Begin("##spinnerContainer", nullptr, ImGuiWindowFlags_AlwaysAutoResize | ImGuiWindowFlags_NoDecoration);
+
+    ImGui::Spinner("##spinner", 10.f, 4.f, ImGui::GetColorU32(ImGui::GetStyle().Colors[ImGuiCol_TitleBgActive]));
+    ImGui::SameLine();
+
+    ImGui::PushFont(nullptr, ImGui::GetStyle().FontSizeBase * 2.f);
+    ImGui::Text("Loading \"%s\"...", fs::path(fileToLoad).filename().string().c_str());
+    ImGui::PopFont();
+
+    ImGui::End();
+
+    ImGui::PopStyleVar();
+    ImGui::PopStyleColor();
+  }
+
   ImGui::Render();
   if (ImGui::GetIO().ConfigFlags & ImGuiConfigFlags_ViewportsEnable) {
     ImGui::UpdatePlatformWindows();
@@ -354,20 +385,25 @@ void KaizenGui::RenderUI() {
 
   if(fileDialogOpen) {
     fileDialogOpen = false;
-    const SDL_DialogFileFilter filters[] = {{"All files", "*"}, {"Nintendo 64 executable", "n64;z64;v64"}, {"Nintendo 64 executable archive", "rar;tar;zip;7z"}};
+    constexpr SDL_DialogFileFilter filters[] = {{"All files", "*"}, {"Nintendo 64 executable", "n64;z64;v64"}, {"Nintendo 64 executable archive", "rar;tar;zip;7z"}};
-    SDL_ShowOpenFileDialog([](void *userdata, const char * const *filelist, int filter) {
+    SDL_ShowOpenFileDialog([](void *userdata, const char * const *filelist, int) {
-      auto kaizen = (KaizenGui*)userdata;
+      auto kaizen = static_cast<KaizenGui*>(userdata);
       
       if (!filelist) {
         panic("An error occured: {}", SDL_GetError());
-        return;
+      }
-      } else if (!*filelist) {
+
+      if (!*filelist) {
         warn("The user did not select any file.");
         warn("Most likely, the dialog was canceled.");
         return;
       }
 
-      kaizen->LoadROM(*filelist);
+      kaizen->fileToLoad = *filelist;
+      kaizen->shouldDisplaySpinner = true;
+
+      std::thread fileWorker(&KaizenGui::FileWorker, kaizen);
+      fileWorker.detach();
     }, this, window.getHandle(), filters, 3, nullptr, false);
   }
 
@@ -375,17 +411,17 @@ void KaizenGui::RenderUI() {
     return;
 
   if(core.romLoaded) {
-    core.parallel.UpdateScreen();
+    core.parallel.UpdateScreen<true>();
     return;
   }
 
-  core.parallel.UpdateScreen(false);
+  core.parallel.UpdateScreen<false>();
 }
 
 void KaizenGui::LoadROM(const std::string &path) noexcept {
   n64::Core& core = n64::Core::GetInstance();
   core.LoadROM(path);
-  const auto gameNameDB = core.GetMem().rom.gameNameDB;
+  const auto gameNameDB = n64::Core::GetMem().rom.gameNameDB;
   SDL_SetWindowTitle(window.getHandle(), ("Kaizen - " + gameNameDB).c_str());
 }
 
@@ -405,11 +441,14 @@ void KaizenGui::run() {
         case SDL_EVENT_WINDOW_RESTORED:
           minimized = false;
           break;
+        default:
       }
       QueryDevices(e);
       HandleInput(e);
     }
 
+    SDL_GetWindowSize(window.getHandle(), &width, &height);
+
     emuThread.run();
     RenderUI();
   }

src/frontend/KaizenGui.hpp

@@ -27,10 +27,23 @@ public:
   static void LoadTAS(const std::string &path) noexcept;
   void LoadROM(const std::string &path) noexcept;
 private:
+  int width{}, height{};
   bool aboutOpen = false;
   bool fileDialogOpen = false;
   bool quit = false;
+  bool shouldDisplaySpinner = false;
+  std::string fileToLoad = "";
   void RenderUI();
-  void HandleInput(SDL_Event event);
+  void HandleInput(const SDL_Event &event);
-  void QueryDevices(SDL_Event event);
+  void QueryDevices(const SDL_Event &event);
+
+  [[noreturn]] void FileWorker() {
+    while (true) {
+      if (!fileToLoad.empty()) {
+        LoadROM(fileToLoad);
+        shouldDisplaySpinner = false;
+        fileToLoad = "";
+      }
+    }
+  }
 };

src/frontend/Settings/GeneralSettings.cpp

@@ -23,7 +23,7 @@ void GeneralSettings::render() {
         return;
       }
 
-      general->savesPath = *filelist;
+      general->savesPath = fs::absolute(*filelist).string();
       Options::GetInstance().SetValue<std::string>("general", "savePath", general->savesPath);
       general->modified = true;
     }, this, window.getHandle(), nullptr, false);

src/frontend/SettingsWindow.cpp

@@ -7,7 +7,7 @@ bool SettingsWindow::render() {
   const ImVec2 center = ImGui::GetMainViewport()->GetCenter();
   ImGui::SetNextWindowPos(center, ImGuiCond_Appearing, ImVec2(0.5f, 0.5f));
 
-  if(!ImGui::BeginPopupModal("Settings", &isOpen))
+  if(!ImGui::BeginPopupModal("Settings", &isOpen, ImGuiWindowFlags_AlwaysAutoResize))
     return false;
   
   if(!ImGui::BeginTabBar("SettingsTabBar"))

src/frontend/main.cpp

@@ -1,7 +1,7 @@
 #include <KaizenGui.hpp>
 #include <cflags.hpp>
 
-int main(int argc, char **argv) {
+int main(const int argc, char **argv) {
   KaizenGui kaizenGui; 
   cflags::cflags flags;
   flags.add_string_callback('\0', "rom", [&kaizenGui](const std::string& v) { kaizenGui.LoadROM(v); }, "Rom to launch from command-line");

src/utils/ErrorData.hpp

@@ -14,7 +14,7 @@ struct Error {
             UNRECOVERABLE,
         } as_enum;
 
-        const char* as_c_str() {
+        [[nodiscard]] const char* as_c_str() const {
             switch(as_enum) {
                 case NONE: return "";
                 case WARN: return "Warning";
@@ -70,7 +70,7 @@ struct Error {
         } as_enum;
 
         
-        const char* as_c_str() {
+        [[nodiscard]] const char* as_c_str() const {
             switch(as_enum) {
                 case SCHEDULER_EOL: return "SCHEDULER_EOL";
                 case SCHEDULER_UNKNOWN: return "SCHEDULER_UNKNOWN";
@@ -108,11 +108,9 @@ struct Error {
     };
 
     template <class... Args>
-    void Throw (Severity severity,
+    void Throw (const Severity severity, const Type type, const std::optional<u64> lastPC,
-                Type type,
+               const std::optional<MemoryAccess> memoryAccess,
-                std::optional<u64> lastPC,
+               const std::format_string<Args...> fmt, Args... args) {
-                std::optional<MemoryAccess> memoryAccess,
-                const std::format_string<Args...> fmt, Args... args) {
         this->severity = severity;
         this->lastPC = lastPC;
         this->memoryAccess = memoryAccess;
@@ -128,7 +126,7 @@ struct Error {
     static std::string& GetError() { return GetInstance().err; }
     static Severity& GetSeverity() { return GetInstance().severity; }
     static Type& GetType() { return GetInstance().type; }
-    static std::optional<u64>& GetLastPC() { return GetInstance().lastPC; }
+    static std::optional<s64>& GetLastPC() { return GetInstance().lastPC; }
     static std::optional<MemoryAccess>& GetMemoryAccess() { return GetInstance().memoryAccess; }
     static bool IsHandled() {
         return GetSeverity().as_enum == Severity::NONE;
@@ -145,7 +143,7 @@ private:
     std::string err;
     Severity severity = {};
     Type type = {};
-    std::optional<u64> lastPC = {};
+    std::optional<s64> lastPC = {};
     std::optional<MemoryAccess> memoryAccess = {};
 };
 }