[APU] Initial version of new XMA decoder

Uplifting glslang is problematic due to changes in SPIRV handling

.gitmodules

@@ -1,6 +1,6 @@
 [submodule "third_party/xbyak"]
 	path = third_party/xbyak
-	url = https://github.com/xenia-project/xbyak.git
+	url = https://github.com/herumi/xbyak.git
 [submodule "third_party/imgui"]
 	path = third_party/imgui
 	url = https://github.com/ocornut/imgui.git
@@ -30,7 +30,7 @@
 	url = https://github.com/openluopworld/aes_128.git
 [submodule "third_party/capstone"]
 	path = third_party/capstone
-	url = https://github.com/xenia-project/capstone.git
+	url = https://github.com/capstone-engine/capstone.git
 [submodule "third_party/cpptoml"]
 	path = third_party/cpptoml
 	url = https://github.com/skystrife/cpptoml.git

src/xenia/apu/xma_context.h

@@ -2,7 +2,7 @@
  ******************************************************************************
  * Xenia : Xbox 360 Emulator Research Project                                 *
  ******************************************************************************
- * Copyright 2021 Ben Vanik. All rights reserved.                             *
+ * Copyright 2023 Ben Vanik. All rights reserved.                             *
  * Released under the BSD license - see LICENSE in the root for more details. *
  ******************************************************************************
  */
@@ -14,8 +14,9 @@
 #include <atomic>
 #include <mutex>
 #include <queue>
-//#include <vector>
 
+#include "xenia/base/bit_stream.h"
+#include "xenia/base/ring_buffer.h"
 #include "xenia/memory.h"
 #include "xenia/xbox.h"
 
@@ -76,7 +77,7 @@ struct XMA_CONTEXT_DATA {
 
   // DWORD 2
   uint32_t input_buffer_read_offset : 26;  // XMAGetInputBufferReadOffset
-  uint32_t unk_dword_2 : 6;                // ErrorStatus/ErrorSet (?)
+  uint32_t error_status : 6;               // ErrorStatus/ErrorSet (?)
 
   // DWORD 3
   uint32_t loop_start : 26;  // XMASetLoopData LoopStartOffset
@@ -96,7 +97,7 @@ struct XMA_CONTEXT_DATA {
   // DWORD 7
   uint32_t output_buffer_ptr;  // physical address
   // DWORD 8
-  uint32_t work_buffer_ptr;  // PtrOverlapAdd(?)
+  uint32_t work_buffer_ptr;  // Stores currently processed subframe
 
   // DWORD 9
   // +0bit, XMAGetOutputBufferReadOffset AKA WriteBufferOffsetRead
@@ -115,10 +116,40 @@ struct XMA_CONTEXT_DATA {
   }
 
   void Store(void* ptr) {
+    // TODO: Compare current ptr state with this and see which dwords we should
+    // write
     xe::copy_and_swap(reinterpret_cast<uint32_t*>(ptr),
                       reinterpret_cast<const uint32_t*>(this),
                       sizeof(XMA_CONTEXT_DATA) / 4);
   }
+
+  bool IsInputBufferValid(uint8_t buffer_index) const {
+    return buffer_index == 0 ? input_buffer_0_valid : input_buffer_1_valid;
+  }
+
+  bool IsCurrentInputBufferValid() const {
+    return IsInputBufferValid(current_buffer);
+  }
+
+  bool IsAnyInputBufferValid() const {
+    return input_buffer_0_valid || input_buffer_1_valid;
+  }
+
+  const uint32_t GetInputBufferAddress(uint8_t buffer_index) const {
+    return buffer_index == 0 ? input_buffer_0_ptr : input_buffer_1_ptr;
+  }
+
+  const uint32_t GetCurrentInputBufferAddress() const {
+    return GetInputBufferAddress(current_buffer);
+  }
+
+  const uint32_t GetInputBufferPacketCount(uint8_t buffer_index) const {
+    return buffer_index == 0 ? input_buffer_0_packet_count
+                             : input_buffer_1_packet_count;
+  }
+  const uint32_t GetCurrentInputBufferPacketCount() const {
+    return GetInputBufferPacketCount(current_buffer);
+  }
 };
 static_assert_size(XMA_CONTEXT_DATA, 64);
 
@@ -130,11 +161,28 @@ struct Xma2ExtraData {
 static_assert_size(Xma2ExtraData, 34);
 #pragma pack(pop)
 
+struct kPacketInfo {
+  uint8_t frame_count_;
+  uint8_t current_frame_;
+  uint32_t current_frame_size_;
+
+  const bool isLastFrameInPacket() const {
+    return current_frame_ == frame_count_ - 1;
+  }
+};
+
+static constexpr int kIdToSampleRate[4] = {24000, 32000, 44100, 48000};
+
 class XmaContext {
  public:
   static const uint32_t kBytesPerPacket = 2048;
+  static const uint32_t kBytesPerPacketHeader = 4;
+  static const uint32_t kBytesPerPacketData =
+      kBytesPerPacket - kBytesPerPacketHeader;
+
   static const uint32_t kBitsPerPacket = kBytesPerPacket * 8;
-  static const uint32_t kBitsPerHeader = 32;
+  static const uint32_t kBitsPerPacketHeader = 32;
+  static const uint32_t kBitsPerFrameHeader = 15;
 
   static const uint32_t kBytesPerSample = 2;
   static const uint32_t kSamplesPerFrame = 512;
@@ -144,8 +192,11 @@ class XmaContext {
   static const uint32_t kBytesPerSubframeChannel =
       kSamplesPerSubframe * kBytesPerSample;
 
-  // static const uint32_t kOutputBytesPerBlock = 256;
-  // static const uint32_t kOutputMaxSizeBytes = 31 * kOutputBytesPerBlock;
+  static const uint32_t kOutputBytesPerBlock = 256;
+  static const uint32_t kOutputMaxSizeBytes = 31 * kOutputBytesPerBlock;
+
+  static const uint32_t kLastFrameMarker = 0x7FFF;
+  static const uint32_t kMaxFrameSizeinBits = 0x4000 - kBitsPerPacketHeader;
 
   explicit XmaContext();
   ~XmaContext();
@@ -171,30 +222,44 @@ class XmaContext {
 
  private:
   static void SwapInputBuffer(XMA_CONTEXT_DATA* data);
-  static bool TrySetupNextLoop(XMA_CONTEXT_DATA* data,
-                               bool ignore_input_buffer_offset);
-  static void NextPacket(XMA_CONTEXT_DATA* data);
+  // Convert sampling rate from ID to frequency.
   static int GetSampleRate(int id);
-  // Get the offset of the next frame. Does not traverse packets.
-  static size_t GetNextFrame(uint8_t* block, size_t size, size_t bit_offset);
   // Get the containing packet number of the frame pointed to by the offset.
-  static int GetFramePacketNumber(uint8_t* block, size_t size,
-                                  size_t bit_offset);
-  // Get the packet number and the index of the frame inside that packet
-  static std::tuple<int, int> GetFrameNumber(uint8_t* block, size_t size,
-                                             size_t bit_offset);
-  // Get the number of frames contained in the packet (including truncated) and
-  // if the last frame is split.
-  static std::tuple<int, bool> GetPacketFrameCount(uint8_t* packet);
+  static int16_t GetPacketNumber(size_t size, size_t bit_offset);
+
+  const kPacketInfo GetPacketInfo(uint8_t* packet, uint32_t frame_offset);
+
+  const uint32_t GetAmountOfBitsToRead(const uint32_t remaining_stream_bits,
+                                       const uint32_t frame_size);
+
+  const uint8_t* GetNextPacket(XMA_CONTEXT_DATA* data,
+                               uint32_t next_packet_index,
+                               uint32_t current_input_packet_count);
+
+  const uint32_t GetNextPacketReadOffset(uint8_t* buffer,
+                                         uint32_t next_packet_index,
+                                         uint32_t current_input_packet_count);
+
+  // Returns currently used buffer
+  uint8_t* GetCurrentInputBuffer(XMA_CONTEXT_DATA* data);
+
+  static uint32_t GetCurrentInputBufferSize(XMA_CONTEXT_DATA* data);
 
   // Convert sample format and swap bytes
   static void ConvertFrame(const uint8_t** samples, bool is_two_channel,
                            uint8_t* output_buffer);
 
-  bool ValidFrameOffset(uint8_t* block, size_t size_bytes,
-                        size_t frame_offset_bits);
   void Decode(XMA_CONTEXT_DATA* data);
-  int PrepareDecoder(uint8_t* packet, int sample_rate, bool is_two_channel);
+  void Consume(RingBuffer* output_rb, XMA_CONTEXT_DATA* data);
+
+  void UpdateLoopStatus(XMA_CONTEXT_DATA* data);
+  int PrepareDecoder(int sample_rate, bool is_two_channel);
+  void PreparePacket(const uint32_t frame_size, const uint32_t frame_padding);
+
+  RingBuffer PrepareOutputRingBuffer(XMA_CONTEXT_DATA* data);
+
+  bool DecodePacket(AVCodecContext* av_context, const AVPacket* av_packet,
+                    AVFrame* av_frame);
 
   // This method should be used ONLY when we're at the last packet of the stream
   // and we want to find offset in next buffer
@@ -204,39 +269,24 @@ class XmaContext {
 
   uint32_t id_ = 0;
   uint32_t guest_ptr_ = 0;
+
   xe_mutex lock_;
   volatile bool is_allocated_ = false;
   volatile bool is_enabled_ = false;
-  // bool is_dirty_ = true;
 
   // ffmpeg structures
   AVPacket* av_packet_ = nullptr;
   AVCodec* av_codec_ = nullptr;
   AVCodecContext* av_context_ = nullptr;
   AVFrame* av_frame_ = nullptr;
-  // uint32_t decoded_consumed_samples_ = 0; // TODO do this dynamically
-  // int decoded_idx_ = -1;
 
-  // bool partial_frame_saved_ = false;
-  // bool partial_frame_size_known_ = false;
-  // size_t partial_frame_total_size_bits_ = 0;
-  // size_t partial_frame_start_offset_bits_ = 0;
-  // size_t partial_frame_offset_bits_ = 0;  // blah internal don't use this
-  // std::vector<uint8_t> partial_frame_buffer_;
-  uint32_t packets_skip_ = 0;
-
-  bool is_stream_done_ = false;
-  // bool split_frame_pending_ = false;
-  uint32_t split_frame_len_ = 0;
-  uint32_t split_frame_len_partial_ = 0;
-  uint8_t split_frame_padding_start_ = 0;
+  std::array<uint8_t, kBytesPerPacketData * 2> input_buffer_;
   // first byte contains bit offset information
   std::array<uint8_t, 1 + 4096> xma_frame_;
-
-  // uint8_t* current_frame_ = nullptr;
-  // conversion buffer for 2 channel frame
   std::array<uint8_t, kBytesPerFrameChannel * 2> raw_frame_;
-  // std::vector<uint8_t> current_frame_ = std::vector<uint8_t>(0);
+
+  int32_t remaining_subframe_blocks_in_output_buffer_ = 0;
+  uint8_t current_frame_remaining_subframes_ = 0;
 };
 
 }  // namespace apu

src/xenia/apu/xma_decoder.cc

@@ -143,7 +143,7 @@ X_STATUS XmaDecoder::Setup(kernel::KernelState* kernel_state) {
   assert_not_null(work_event_);
   worker_thread_ = kernel::object_ref<kernel::XHostThread>(
       new kernel::XHostThread(kernel_state, 128 * 1024, 0, [this]() {
-        WorkerThreadMain();
+        //WorkerThreadMain();
         return 0;
       }, kernel_state->GetIdleProcess()));//this one doesnt need any process actually. never calls any guest code
   worker_thread_->set_name("XMA Decoder");
@@ -307,6 +307,7 @@ void XmaDecoder::WriteRegister(uint32_t addr, uint32_t value) {
         uint32_t context_id = base_context_id + i;
         auto& context = contexts_[context_id];
         context.Enable();
+        context.Work();
       }
     }
     // Signal the decoder thread to start processing.

src/xenia/apu/xma_helpers.h

@@ -2,7 +2,7 @@
  ******************************************************************************
  * Xenia : Xbox 360 Emulator Research Project                                 *
  ******************************************************************************
- * Copyright 2021 Ben Vanik. All rights reserved.                             *
+ * Copyright 2023 Ben Vanik. All rights reserved.                             *
  * Released under the BSD license - see LICENSE in the root for more details. *
  ******************************************************************************
  */
@@ -20,29 +20,29 @@ namespace xma {
 
 static const uint32_t kMaxFrameLength = 0x7FFF;
 
-// Get number of frames that /begin/ in this packet.
-uint32_t GetPacketFrameCount(uint8_t* packet) {
-  return (uint8_t)(packet[0] >> 2);
+// Get number of frames that /begin/ in this packet. This is valid only for XMA2
+// packets
+const uint8_t GetPacketFrameCount(const uint8_t* packet) {
+  return packet[0] >> 2;
 }
 
+const uint8_t GetPacketMetadata(const uint8_t* packet) {
+  return packet[2] & 0x7;
+}
+
+const bool IsPacketXma2Type(const uint8_t* packet) {
+  return GetPacketMetadata(packet) == 1;
+}
+
+const uint8_t GetPacketSkipCount(const uint8_t* packet) { return packet[3]; }
+
 // Get the first frame offset in bits
-uint32_t GetPacketFrameOffset(uint8_t* packet) {
+uint32_t GetPacketFrameOffset(const uint8_t* packet) {
   uint32_t val = (uint16_t)(((packet[0] & 0x3) << 13) | (packet[1] << 5) |
                             (packet[2] >> 3));
-  // if (val > kBitsPerPacket - kBitsPerHeader) {
-  //   // There is no data in this packet
-  //   return -1;
-  // } else {
   return val + 32;
-  // }
 }
 
-uint32_t GetPacketMetadata(uint8_t* packet) {
-  return (uint8_t)(packet[2] & 0x7);
-}
-
-uint32_t GetPacketSkipCount(uint8_t* packet) { return (uint8_t)(packet[3]); }
-
 }  // namespace xma
 }  // namespace apu
 }  // namespace xe

src/xenia/base/bit_stream.cc

@@ -20,6 +20,9 @@ namespace xe {
 BitStream::BitStream(uint8_t* buffer, size_t size_in_bits)
     : buffer_(buffer), size_bits_(size_in_bits) {}
 
+BitStream::BitStream(const uint8_t* buffer, size_t size_in_bits)
+    : buffer_(const_cast<uint8_t*>(buffer)), size_bits_(size_in_bits) {}
+
 BitStream::~BitStream() = default;
 
 void BitStream::SetOffset(size_t offset_bits) {

src/xenia/base/bit_stream.h

@@ -18,6 +18,7 @@ namespace xe {
 class BitStream {
  public:
   BitStream(uint8_t* buffer, size_t size_in_bits);
+  BitStream(const uint8_t* buffer, size_t size_in_bits);
   ~BitStream();
 
   const uint8_t* buffer() const { return buffer_; }

src/xenia/base/platform_win.h

@@ -56,7 +56,7 @@
     }                                                                        \
     template <typename TRet = void, typename... TArgs>                       \
     inline TRet invoke(TArgs... args) {                                      \
-      return reinterpret_cast<NTSYSAPI TRet(NTAPI*)(TArgs...)>(fn)(args...); \
+      return reinterpret_cast<TRet(NTAPI*)(TArgs...)>(fn)(args...);          \
     }                                                                        \
     inline operator bool() const { return fn != nullptr; }                   \
   } clsvar

src/xenia/gpu/dxbc_shader_translator_om.cc

@@ -2940,7 +2940,7 @@ void DxbcShaderTranslator::CompletePixelShader() {
           SystemConstants::Index::kAlphaTestReference,
           offsetof(SystemConstants, alpha_test_reference), dxbc::Src::kXXXX));
       // Epsilon for alpha checks
-      dxbc::Src fuzzy_epsilon = dxbc::Src::LF(0.01f);
+      dxbc::Src fuzzy_epsilon = dxbc::Src::LF(1e-3f);
       // Handle "not equal" specially (specifically as "not equal" so it's true
       // for NaN, not "less or greater" which is false for NaN).
       a_.OpIEq(alpha_test_op_dest, alpha_test_mask_src,

src/xenia/hid/sdl/sdl_input_driver.cc

@@ -40,7 +40,6 @@ SDLInputDriver::SDLInputDriver(xe::ui::Window* window, size_t window_z_order)
       sdl_events_unflushed_(0),
       sdl_pumpevents_queued_(false),
       controllers_(),
-      controllers_mutex_(),
       keystroke_states_() {}
 
 SDLInputDriver::~SDLInputDriver() {
@@ -151,8 +150,6 @@ X_RESULT SDLInputDriver::GetCapabilities(uint32_t user_index, uint32_t flags,
 
   QueueControllerUpdate();
 
-  std::unique_lock<std::mutex> guard(controllers_mutex_);
-
   auto controller = GetControllerState(user_index);
   if (!controller) {
     return X_ERROR_DEVICE_NOT_CONNECTED;
@@ -180,8 +177,6 @@ X_RESULT SDLInputDriver::GetState(uint32_t user_index,
     QueueControllerUpdate();
   }
 
-  std::unique_lock<std::mutex> guard(controllers_mutex_);
-
   auto controller = GetControllerState(user_index);
   if (!controller) {
     return X_ERROR_DEVICE_NOT_CONNECTED;
@@ -295,8 +290,6 @@ X_RESULT SDLInputDriver::GetKeystroke(uint32_t users, uint32_t flags,
     QueueControllerUpdate();
   }
 
-  std::unique_lock<std::mutex> guard(controllers_mutex_);
-
   for (uint32_t user_index = (user_any ? 0 : users);
        user_index < (user_any ? HID_SDL_USER_COUNT : users + 1); user_index++) {
     auto controller = GetControllerState(user_index);
@@ -420,8 +413,6 @@ void SDLInputDriver::HandleEvent(const SDL_Event& event) {
 }
 
 void SDLInputDriver::OnControllerDeviceAdded(const SDL_Event& event) {
-  std::unique_lock<std::mutex> guard(controllers_mutex_);
-
   // Open the controller.
   const auto controller = SDL_GameControllerOpen(event.cdevice.which);
   if (!controller) {
@@ -485,8 +476,6 @@ void SDLInputDriver::OnControllerDeviceAdded(const SDL_Event& event) {
 }
 
 void SDLInputDriver::OnControllerDeviceRemoved(const SDL_Event& event) {
-  std::unique_lock<std::mutex> guard(controllers_mutex_);
-
   // Find the disconnected gamecontroller and close it.
   auto idx = GetControllerIndexFromInstanceID(event.cdevice.which);
   if (idx) {
@@ -501,8 +490,6 @@ void SDLInputDriver::OnControllerDeviceRemoved(const SDL_Event& event) {
 }
 
 void SDLInputDriver::OnControllerDeviceAxisMotion(const SDL_Event& event) {
-  std::unique_lock<std::mutex> guard(controllers_mutex_);
-
   auto idx = GetControllerIndexFromInstanceID(event.caxis.which);
   assert(idx);
   auto& pad = controllers_.at(*idx).state.gamepad;
@@ -533,8 +520,6 @@ void SDLInputDriver::OnControllerDeviceAxisMotion(const SDL_Event& event) {
 }
 
 void SDLInputDriver::OnControllerDeviceButtonChanged(const SDL_Event& event) {
-  std::unique_lock<std::mutex> guard(controllers_mutex_);
-
   // Define a lookup table to map between SDL and XInput button codes.
   // These need to be in the order of the SDL_GameControllerButton enum.
   static constexpr std::array<

src/xenia/hid/sdl/sdl_input_driver.h

@@ -84,7 +84,6 @@ class SDLInputDriver final : public InputDriver {
   int sdl_events_unflushed_;
   std::atomic<bool> sdl_pumpevents_queued_;
   std::array<ControllerState, HID_SDL_USER_COUNT> controllers_;
-  std::mutex controllers_mutex_;
   std::array<KeystrokeState, HID_SDL_USER_COUNT> keystroke_states_;
 };
 

src/xenia/kernel/kernel_state.cc

@@ -38,6 +38,9 @@ DEFINE_uint32(max_signed_profiles, 4,
               "Limits how many profiles can be assigned. Possible values: 1-4",
               "Kernel");
 
+DEFINE_uint32(kernel_build_version, 1888, "Define current kernel version",
+              "Kernel");
+
 namespace xe {
 namespace kernel {
 
@@ -66,6 +69,7 @@ KernelState::KernelState(Emulator* emulator)
   user_profiles_.emplace(0, std::make_unique<xam::UserProfile>(0));
 
   InitializeKernelGuestGlobals();
+  kernel_version_ = KernelVersion(cvars::kernel_build_version);
 
   auto content_root = emulator_->content_root();
   if (!content_root.empty()) {

src/xenia/kernel/kernel_state.h

@@ -48,6 +48,8 @@ namespace kernel {
 
 constexpr fourcc_t kKernelSaveSignature = make_fourcc("KRNL");
 
+static constexpr const uint16_t kBaseKernelBuildVersion = 1888;
+
 // (?), used by KeGetCurrentProcessType
 constexpr uint32_t X_PROCTYPE_IDLE = 0;
 constexpr uint32_t X_PROCTYPE_TITLE = 1;
@@ -145,6 +147,27 @@ struct KernelGuestGlobals {
 struct DPCImpersonationScope {
   uint8_t previous_irql_;
 };
+
+struct KernelVersion {
+  union {
+    xe::be<uint64_t> value;
+
+    struct {
+      xe::be<uint16_t> major;
+      xe::be<uint16_t> minor;
+      xe::be<uint16_t> build;
+      xe::be<uint16_t> qfe;
+    };
+  };
+
+  KernelVersion(uint16_t build_ver = kBaseKernelBuildVersion) {
+    major = 2;
+    minor = 0;
+    build = std::max(kBaseKernelBuildVersion, build_ver);
+    qfe = 0;
+  }
+};
+
 class KernelState {
  public:
   explicit KernelState(Emulator* emulator);
@@ -200,6 +223,8 @@ class KernelState {
   // Access must be guarded by the global critical region.
   util::ObjectTable* object_table() { return &object_table_; }
 
+  const KernelVersion* GetKernelVersion() const { return &kernel_version_; }
+
   uint32_t GetSystemProcess() const {
     return kernel_guest_globals_ + offsetof(KernelGuestGlobals, system_process);
   }
@@ -332,6 +357,8 @@ class KernelState {
   std::map<uint8_t, std::unique_ptr<xam::UserProfile>> user_profiles_;
   std::unique_ptr<AchievementManager> achievement_manager_;
 
+  KernelVersion kernel_version_;
+
   xe::global_critical_region global_critical_region_;
 
   // Must be guarded by the global critical region.

src/xenia/kernel/xbdm/xbdm_misc.cc

@@ -13,8 +13,10 @@
 #include "xenia/kernel/xbdm/xbdm_private.h"
 #include "xenia/kernel/xthread.h"
 #include "xenia/xbox.h"
+
 // chrispy: no idea what a real valid value is for this
 static constexpr const char DmXboxName[] = "Xbox360Name";
+
 namespace xe {
 namespace kernel {
 namespace xbdm {
@@ -142,8 +144,8 @@ dword_result_t DmGetSystemInfo_entry(pointer_t<XBDM_SYSTEM_INFO> info) {
   info->base_kernel_version.minor = info->kernel_version.minor = 0;
   info->base_kernel_version.qfe = info->kernel_version.qfe = 0;
 
-  info->base_kernel_version.build = 1888;
-  info->kernel_version.build = 13139;
+  info->base_kernel_version.build = kBaseKernelBuildVersion;
+  info->kernel_version.build = kernel_state()->GetKernelVersion()->build;
 
   return XBDM_SUCCESSFUL;
 }

src/xenia/kernel/xboxkrnl/xboxkrnl_module.cc

@@ -205,15 +205,12 @@ XboxkrnlModule::XboxkrnlModule(Emulator* emulator, KernelState* kernel_state)
   // XboxKrnlVersion (8b)
   // Kernel version, looks like 2b.2b.2b.2b.
   // I've only seen games check >=, so we just fake something here.
-  uint32_t pXboxKrnlVersion = memory_->SystemHeapAlloc(8);
+  uint32_t pXboxKrnlVersion = memory_->SystemHeapAlloc(sizeof(KernelVersion));
   auto lpXboxKrnlVersion = memory_->TranslateVirtual(pXboxKrnlVersion);
   export_resolver_->SetVariableMapping(
       "xboxkrnl.exe", ordinals::XboxKrnlVersion, pXboxKrnlVersion);
-  xe::store_and_swap<uint16_t>(lpXboxKrnlVersion + 0, 2);
-  xe::store_and_swap<uint16_t>(lpXboxKrnlVersion + 2, 0xFFFF);
-  xe::store_and_swap<uint16_t>(lpXboxKrnlVersion + 4, 0xFFFF);
-  xe::store_and_swap<uint8_t>(lpXboxKrnlVersion + 6, 0x80);
-  xe::store_and_swap<uint8_t>(lpXboxKrnlVersion + 7, 0x00);
+  std::memcpy(lpXboxKrnlVersion, kernel_state_->GetKernelVersion(),
+              sizeof(KernelVersion));
 
   export_resolver_->SetVariableMapping("xboxkrnl.exe",
                                        ordinals::KeTimeStampBundle,

third_party/DirectXShaderCompiler

@@ -1 +1 @@
-Subproject commit 6b6f40200bea5ed99367513f53f6a28e52fd3d0e
+Subproject commit 69e54e29086b7035acffb304cec57a350225f8b0

third_party/SPIRV-Tools

@@ -1 +1 @@
-Subproject commit dd534e877e725c9bb6f751c427442456a05384e4
+Subproject commit 44d72a9b36702f093dd20815561a56778b2d181e

third_party/Vulkan-Headers

@@ -1 +1 @@
-Subproject commit b32da5329b50e3cb96229aaecba9ded032fe29cc
+Subproject commit 46dc0f6e514f5730784bb2cac2a7c731636839e8

third_party/VulkanMemoryAllocator

@@ -1 +1 @@
-Subproject commit 51c8b56011303e94840370089f816b19dbe7edf0
+Subproject commit 19b940e864bd3a5afb3c79e3c6788869d01a19eb

third_party/capstone

@@ -1 +1 @@
-Subproject commit f9c6a90489be7b3637ff1c7298e45efafe7cf1b9
+Subproject commit 097c04d9413c59a58b00d4d1c8d5dc0ac158ffaa

third_party/capstone.lua

@@ -39,6 +39,8 @@ project("capstone")
     "capstone/SStream.h",
     "capstone/utils.c",
     "capstone/utils.h",
+    "capstone/Mapping.c",
+    "capstone/Mapping.h",
 
     "capstone/arch/X86/*.c",
     "capstone/arch/X86/*.h",

third_party/cxxopts

@@ -1 +1 @@
-Subproject commit b2b8cf2f50a449720874f43445e23d75b77dcc43
+Subproject commit 3bf268481da8208d171d8908e6491459de3651d7

third_party/fmt

@@ -1 +1 @@
-Subproject commit 27e3c0fe9b5dc99f339637ec2ea7efae5b945ab8
+Subproject commit 7bdf0628b1276379886c7f6dda2cef2b3b374f0b

third_party/xbyak

@@ -1 +1 @@
-Subproject commit bc70e7e11ad37f207ece73330c5b72a55fef1ceb
+Subproject commit 63e6ee92cc906a3dbf6b333b2fd51ac1ca95f891

third_party/xxhash

@@ -1 +1 @@
-Subproject commit 4c881f796d6af27ef7d9c48f87817da0d3d75dc1
+Subproject commit bbb27a5efb85b92a0486cf361a8635715a53f6ba