GPU: Partially implemented the Fermi2D surface copy operation.

The hardware allows for some rather complicated operations to be performed on the data during the copy, this is not implemented.
Only same-format same-size raw copies are implemented for now.

src/core/CMakeLists.txt

@@ -42,8 +42,6 @@ add_library(core STATIC
     hle/kernel/client_port.h
     hle/kernel/client_session.cpp
     hle/kernel/client_session.h
-    hle/kernel/condition_variable.cpp
-    hle/kernel/condition_variable.h
     hle/kernel/errors.h
     hle/kernel/event.cpp
     hle/kernel/event.h

src/core/core.cpp

@@ -12,10 +12,13 @@
 #include "core/core.h"
 #include "core/core_timing.h"
 #include "core/gdbstub/gdbstub.h"
+#include "core/hle/kernel/client_port.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/thread.h"
 #include "core/hle/service/service.h"
+#include "core/hle/service/sm/controller.h"
+#include "core/hle/service/sm/sm.h"
 #include "core/hw/hw.h"
 #include "core/loader/loader.h"
 #include "core/memory_setup.h"
@@ -26,6 +29,8 @@ namespace Core {
 
 /*static*/ System System::s_instance;
 
+System::~System() = default;
+
 System::ResultStatus System::RunLoop(bool tight_loop) {
     status = ResultStatus::Success;
     if (!cpu_core) {
@@ -167,10 +172,12 @@ System::ResultStatus System::Init(EmuWindow* emu_window, u32 system_mode) {
 
     telemetry_session = std::make_unique<Core::TelemetrySession>();
 
+    service_manager = std::make_shared<Service::SM::ServiceManager>();
+
     HW::Init();
     Kernel::Init(system_mode);
     scheduler = std::make_unique<Kernel::Scheduler>(cpu_core.get());
-    Service::Init();
+    Service::Init(service_manager);
     GDBStub::Init();
 
     if (!VideoCore::Init(emu_window)) {
@@ -200,17 +207,26 @@ void System::Shutdown() {
     VideoCore::Shutdown();
     GDBStub::Shutdown();
     Service::Shutdown();
-    scheduler = nullptr;
+    scheduler.reset();
     Kernel::Shutdown();
     HW::Shutdown();
-    telemetry_session = nullptr;
-    gpu_core = nullptr;
-    cpu_core = nullptr;
+    service_manager.reset();
+    telemetry_session.reset();
+    gpu_core.reset();
+    cpu_core.reset();
     CoreTiming::Shutdown();
 
-    app_loader = nullptr;
+    app_loader.reset();
 
     LOG_DEBUG(Core, "Shutdown OK");
 }
 
+Service::SM::ServiceManager& System::ServiceManager() {
+    return *service_manager;
+}
+
+const Service::SM::ServiceManager& System::ServiceManager() const {
+    return *service_manager;
+}
+
 } // namespace Core

src/core/core.h

@@ -19,10 +19,16 @@
 class EmuWindow;
 class ARM_Interface;
 
+namespace Service::SM {
+class ServiceManager;
+}
+
 namespace Core {
 
 class System {
 public:
+    ~System();
+
     /**
      * Gets the instance of the System singleton class.
      * @returns Reference to the instance of the System singleton class.
@@ -137,6 +143,9 @@ public:
         return *app_loader;
     }
 
+    Service::SM::ServiceManager& ServiceManager();
+    const Service::SM::ServiceManager& ServiceManager() const;
+
     void SetGPUDebugContext(std::shared_ptr<Tegra::DebugContext> context) {
         debug_context = std::move(context);
     }
@@ -171,6 +180,9 @@ private:
     /// When true, signals that a reschedule should happen
     bool reschedule_pending{};
 
+    /// Service manager
+    std::shared_ptr<Service::SM::ServiceManager> service_manager;
+
     /// Telemetry session for this emulation session
     std::unique_ptr<Core::TelemetrySession> telemetry_session;
 

src/core/file_sys/disk_filesystem.cpp

@@ -67,10 +67,16 @@ ResultCode Disk_FileSystem::DeleteFile(const std::string& path) const {
     return RESULT_SUCCESS;
 }
 
-ResultCode Disk_FileSystem::RenameFile(const Path& src_path, const Path& dest_path) const {
-    LOG_WARNING(Service_FS, "(STUBBED) called");
+ResultCode Disk_FileSystem::RenameFile(const std::string& src_path,
+                                       const std::string& dest_path) const {
+    const std::string full_src_path = base_directory + src_path;
+    const std::string full_dest_path = base_directory + dest_path;
+
+    if (!FileUtil::Exists(full_src_path)) {
+        return ERROR_PATH_NOT_FOUND;
+    }
     // TODO(wwylele): Use correct error code
-    return ResultCode(-1);
+    return FileUtil::Rename(full_src_path, full_dest_path) ? RESULT_SUCCESS : ResultCode(-1);
 }
 
 ResultCode Disk_FileSystem::DeleteDirectory(const Path& path) const {

src/core/file_sys/disk_filesystem.h

@@ -26,7 +26,7 @@ public:
     ResultVal<std::unique_ptr<StorageBackend>> OpenFile(const std::string& path,
                                                         Mode mode) const override;
     ResultCode DeleteFile(const std::string& path) const override;
-    ResultCode RenameFile(const Path& src_path, const Path& dest_path) const override;
+    ResultCode RenameFile(const std::string& src_path, const std::string& dest_path) const override;
     ResultCode DeleteDirectory(const Path& path) const override;
     ResultCode DeleteDirectoryRecursively(const Path& path) const override;
     ResultCode CreateFile(const std::string& path, u64 size) const override;

src/core/file_sys/filesystem.h

@@ -126,7 +126,8 @@ public:
      * @param dest_path Destination path relative to the archive
      * @return Result of the operation
      */
-    virtual ResultCode RenameFile(const Path& src_path, const Path& dest_path) const = 0;
+    virtual ResultCode RenameFile(const std::string& src_path,
+                                  const std::string& dest_path) const = 0;
 
     /**
      * Rename a Directory specified by its path

src/core/file_sys/romfs_filesystem.cpp

@@ -27,7 +27,8 @@ ResultCode RomFS_FileSystem::DeleteFile(const std::string& path) const {
     return ResultCode(-1);
 }
 
-ResultCode RomFS_FileSystem::RenameFile(const Path& src_path, const Path& dest_path) const {
+ResultCode RomFS_FileSystem::RenameFile(const std::string& src_path,
+                                        const std::string& dest_path) const {
     LOG_CRITICAL(Service_FS, "Attempted to rename a file within an ROMFS archive (%s).",
                  GetName().c_str());
     // TODO(wwylele): Use correct error code

src/core/file_sys/romfs_filesystem.h

@@ -32,7 +32,7 @@ public:
     ResultVal<std::unique_ptr<StorageBackend>> OpenFile(const std::string& path,
                                                         Mode mode) const override;
     ResultCode DeleteFile(const std::string& path) const override;
-    ResultCode RenameFile(const Path& src_path, const Path& dest_path) const override;
+    ResultCode RenameFile(const std::string& src_path, const std::string& dest_path) const override;
     ResultCode DeleteDirectory(const Path& path) const override;
     ResultCode DeleteDirectoryRecursively(const Path& path) const override;
     ResultCode CreateFile(const std::string& path, u64 size) const override;

src/core/hle/kernel/condition_variable.cpp

@@ -1,64 +0,0 @@
-// Copyright 2018 yuzu emulator team
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include "common/assert.h"
-#include "core/hle/kernel/condition_variable.h"
-#include "core/hle/kernel/errors.h"
-#include "core/hle/kernel/kernel.h"
-#include "core/hle/kernel/object_address_table.h"
-#include "core/hle/kernel/thread.h"
-
-namespace Kernel {
-
-ConditionVariable::ConditionVariable() {}
-ConditionVariable::~ConditionVariable() {}
-
-ResultVal<SharedPtr<ConditionVariable>> ConditionVariable::Create(VAddr guest_addr,
-                                                                  std::string name) {
-    SharedPtr<ConditionVariable> condition_variable(new ConditionVariable);
-
-    condition_variable->name = std::move(name);
-    condition_variable->guest_addr = guest_addr;
-    condition_variable->mutex_addr = 0;
-
-    // Condition variables are referenced by guest address, so track this in the kernel
-    g_object_address_table.Insert(guest_addr, condition_variable);
-
-    return MakeResult<SharedPtr<ConditionVariable>>(std::move(condition_variable));
-}
-
-bool ConditionVariable::ShouldWait(Thread* thread) const {
-    return GetAvailableCount() <= 0;
-}
-
-void ConditionVariable::Acquire(Thread* thread) {
-    if (GetAvailableCount() <= 0)
-        return;
-
-    SetAvailableCount(GetAvailableCount() - 1);
-}
-
-ResultCode ConditionVariable::Release(s32 target) {
-    if (target == -1) {
-        // When -1, wake up all waiting threads
-        SetAvailableCount(static_cast<s32>(GetWaitingThreads().size()));
-        WakeupAllWaitingThreads();
-    } else {
-        // Otherwise, wake up just a single thread
-        SetAvailableCount(target);
-        WakeupWaitingThread(GetHighestPriorityReadyThread());
-    }
-
-    return RESULT_SUCCESS;
-}
-
-s32 ConditionVariable::GetAvailableCount() const {
-    return Memory::Read32(guest_addr);
-}
-
-void ConditionVariable::SetAvailableCount(s32 value) const {
-    Memory::Write32(guest_addr, value);
-}
-
-} // namespace Kernel

src/core/hle/kernel/condition_variable.h

@@ -1,63 +0,0 @@
-// Copyright 2018 yuzu emulator team
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <string>
-#include <queue>
-#include "common/common_types.h"
-#include "core/hle/kernel/kernel.h"
-#include "core/hle/kernel/wait_object.h"
-#include "core/hle/result.h"
-
-namespace Kernel {
-
-class ConditionVariable final : public WaitObject {
-public:
-    /**
-     * Creates a condition variable.
-     * @param guest_addr Address of the object tracking the condition variable in guest memory. If
-     * specified, this condition variable will update the guest object when its state changes.
-     * @param name Optional name of condition variable.
-     * @return The created condition variable.
-     */
-    static ResultVal<SharedPtr<ConditionVariable>> Create(VAddr guest_addr,
-                                                          std::string name = "Unknown");
-
-    std::string GetTypeName() const override {
-        return "ConditionVariable";
-    }
-    std::string GetName() const override {
-        return name;
-    }
-
-    static const HandleType HANDLE_TYPE = HandleType::ConditionVariable;
-    HandleType GetHandleType() const override {
-        return HANDLE_TYPE;
-    }
-
-    s32 GetAvailableCount() const;
-    void SetAvailableCount(s32 value) const;
-
-    std::string name; ///< Name of condition variable (optional)
-    VAddr guest_addr; ///< Address of the guest condition variable value
-    VAddr mutex_addr; ///< (optional) Address of guest mutex value associated with this condition
-                      ///< variable, used for implementing events
-
-    bool ShouldWait(Thread* thread) const override;
-    void Acquire(Thread* thread) override;
-
-    /**
-     * Releases a slot from a condition variable.
-     * @param target The number of threads to wakeup, -1 is all.
-     * @return ResultCode indicating if the operation succeeded.
-     */
-    ResultCode Release(s32 target);
-
-private:
-    ConditionVariable();
-    ~ConditionVariable() override;
-};
-
-} // namespace Kernel

src/core/hle/kernel/errors.h

@@ -20,6 +20,7 @@ enum {
     MaxConnectionsReached = 52,
 
     // Confirmed Switch OS error codes
+    MisalignedAddress = 102,
     InvalidHandle = 114,
     Timeout = 117,
     SynchronizationCanceled = 118,

src/core/hle/kernel/kernel.h

@@ -18,12 +18,10 @@ using Handle = u32;
 enum class HandleType : u32 {
     Unknown,
     Event,
-    Mutex,
     SharedMemory,
     Thread,
     Process,
     AddressArbiter,
-    ConditionVariable,
     Timer,
     ResourceLimit,
     CodeSet,
@@ -63,9 +61,7 @@ public:
     bool IsWaitable() const {
         switch (GetHandleType()) {
         case HandleType::Event:
-        case HandleType::Mutex:
         case HandleType::Thread:
-        case HandleType::ConditionVariable:
         case HandleType::Timer:
         case HandleType::ServerPort:
         case HandleType::ServerSession:

src/core/hle/kernel/mutex.cpp

@@ -7,6 +7,7 @@
 #include <boost/range/algorithm_ext/erase.hpp>
 #include "common/assert.h"
 #include "core/core.h"
+#include "core/hle/kernel/errors.h"
 #include "core/hle/kernel/handle_table.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/mutex.h"
@@ -15,124 +16,120 @@
 
 namespace Kernel {
 
-void ReleaseThreadMutexes(Thread* thread) {
-    for (auto& mtx : thread->held_mutexes) {
-        mtx->SetHasWaiters(false);
-        mtx->SetHoldingThread(nullptr);
-        mtx->WakeupAllWaitingThreads();
-    }
-    thread->held_mutexes.clear();
-}
+/// Returns the number of threads that are waiting for a mutex, and the highest priority one among
+/// those.
+static std::pair<SharedPtr<Thread>, u32> GetHighestPriorityMutexWaitingThread(
+    SharedPtr<Thread> current_thread, VAddr mutex_addr) {
 
-Mutex::Mutex() {}
-Mutex::~Mutex() {}
+    SharedPtr<Thread> highest_priority_thread;
+    u32 num_waiters = 0;
 
-SharedPtr<Mutex> Mutex::Create(SharedPtr<Kernel::Thread> holding_thread, VAddr guest_addr,
-                               std::string name) {
-    SharedPtr<Mutex> mutex(new Mutex);
+    for (auto& thread : current_thread->wait_mutex_threads) {
+        if (thread->mutex_wait_address != mutex_addr)
+            continue;
 
-    mutex->guest_addr = guest_addr;
-    mutex->name = std::move(name);
+        ASSERT(thread->status == THREADSTATUS_WAIT_MUTEX);
 
-    // If mutex was initialized with a holding thread, acquire it by the holding thread
-    if (holding_thread) {
-        mutex->Acquire(holding_thread.get());
+        ++num_waiters;
+        if (highest_priority_thread == nullptr ||
+            thread->GetPriority() < highest_priority_thread->GetPriority()) {
+            highest_priority_thread = thread;
+        }
     }
 
-    // Mutexes are referenced by guest address, so track this in the kernel
-    g_object_address_table.Insert(guest_addr, mutex);
-
-    return mutex;
+    return {highest_priority_thread, num_waiters};
 }
 
-bool Mutex::ShouldWait(Thread* thread) const {
-    auto holding_thread = GetHoldingThread();
-    return holding_thread != nullptr && thread != holding_thread;
+/// Update the mutex owner field of all threads waiting on the mutex to point to the new owner.
+static void TransferMutexOwnership(VAddr mutex_addr, SharedPtr<Thread> current_thread,
+                                   SharedPtr<Thread> new_owner) {
+    auto threads = current_thread->wait_mutex_threads;
+    for (auto& thread : threads) {
+        if (thread->mutex_wait_address != mutex_addr)
+            continue;
+
+        ASSERT(thread->lock_owner == current_thread);
+        current_thread->RemoveMutexWaiter(thread);
+        if (new_owner != thread)
+            new_owner->AddMutexWaiter(thread);
+    }
 }
 
-void Mutex::Acquire(Thread* thread) {
-    ASSERT_MSG(!ShouldWait(thread), "object unavailable!");
+ResultCode Mutex::TryAcquire(VAddr address, Handle holding_thread_handle,
+                             Handle requesting_thread_handle) {
+    // The mutex address must be 4-byte aligned
+    if ((address % sizeof(u32)) != 0) {
+        return ResultCode(ErrorModule::Kernel, ErrCodes::MisalignedAddress);
+    }
 
-    priority = thread->current_priority;
-    thread->held_mutexes.insert(this);
-    SetHoldingThread(thread);
-    thread->UpdatePriority();
-    Core::System::GetInstance().PrepareReschedule();
-}
+    SharedPtr<Thread> holding_thread = g_handle_table.Get<Thread>(holding_thread_handle);
+    SharedPtr<Thread> requesting_thread = g_handle_table.Get<Thread>(requesting_thread_handle);
 
-ResultCode Mutex::Release(Thread* thread) {
-    auto holding_thread = GetHoldingThread();
-    ASSERT(holding_thread);
+    // TODO(Subv): It is currently unknown if it is possible to lock a mutex in behalf of another
+    // thread.
+    ASSERT(requesting_thread == GetCurrentThread());
 
-    // We can only release the mutex if it's held by the calling thread.
-    ASSERT(thread == holding_thread);
+    u32 addr_value = Memory::Read32(address);
+
+    // If the mutex isn't being held, just return success.
+    if (addr_value != (holding_thread_handle | Mutex::MutexHasWaitersFlag)) {
+        return RESULT_SUCCESS;
+    }
+
+    if (holding_thread == nullptr)
+        return ERR_INVALID_HANDLE;
+
+    // Wait until the mutex is released
+    GetCurrentThread()->mutex_wait_address = address;
+    GetCurrentThread()->wait_handle = requesting_thread_handle;
+
+    GetCurrentThread()->status = THREADSTATUS_WAIT_MUTEX;
+    GetCurrentThread()->wakeup_callback = nullptr;
+
+    // Update the lock holder thread's priority to prevent priority inversion.
+    holding_thread->AddMutexWaiter(GetCurrentThread());
 
-    holding_thread->held_mutexes.erase(this);
-    holding_thread->UpdatePriority();
-    SetHoldingThread(nullptr);
-    SetHasWaiters(!GetWaitingThreads().empty());
-    WakeupAllWaitingThreads();
     Core::System::GetInstance().PrepareReschedule();
 
     return RESULT_SUCCESS;
 }
 
-void Mutex::AddWaitingThread(SharedPtr<Thread> thread) {
-    WaitObject::AddWaitingThread(thread);
-    thread->pending_mutexes.insert(this);
-    SetHasWaiters(true);
-    UpdatePriority();
-}
-
-void Mutex::RemoveWaitingThread(Thread* thread) {
-    WaitObject::RemoveWaitingThread(thread);
-    thread->pending_mutexes.erase(this);
-    if (!GetHasWaiters())
-        SetHasWaiters(!GetWaitingThreads().empty());
-    UpdatePriority();
-}
-
-void Mutex::UpdatePriority() {
-    if (!GetHoldingThread())
-        return;
-
-    u32 best_priority = THREADPRIO_LOWEST;
-    for (auto& waiter : GetWaitingThreads()) {
-        if (waiter->current_priority < best_priority)
-            best_priority = waiter->current_priority;
+ResultCode Mutex::Release(VAddr address) {
+    // The mutex address must be 4-byte aligned
+    if ((address % sizeof(u32)) != 0) {
+        return ResultCode(ErrorModule::Kernel, ErrCodes::MisalignedAddress);
     }
 
-    if (best_priority != priority) {
-        priority = best_priority;
-        GetHoldingThread()->UpdatePriority();
+    auto [thread, num_waiters] = GetHighestPriorityMutexWaitingThread(GetCurrentThread(), address);
+
+    // There are no more threads waiting for the mutex, release it completely.
+    if (thread == nullptr) {
+        ASSERT(GetCurrentThread()->wait_mutex_threads.empty());
+        Memory::Write32(address, 0);
+        return RESULT_SUCCESS;
     }
-}
 
-Handle Mutex::GetOwnerHandle() const {
-    GuestState guest_state{Memory::Read32(guest_addr)};
-    return guest_state.holding_thread_handle;
-}
+    // Transfer the ownership of the mutex from the previous owner to the new one.
+    TransferMutexOwnership(address, GetCurrentThread(), thread);
 
-SharedPtr<Thread> Mutex::GetHoldingThread() const {
-    GuestState guest_state{Memory::Read32(guest_addr)};
-    return g_handle_table.Get<Thread>(guest_state.holding_thread_handle);
-}
+    u32 mutex_value = thread->wait_handle;
 
-void Mutex::SetHoldingThread(SharedPtr<Thread> thread) {
-    GuestState guest_state{Memory::Read32(guest_addr)};
-    guest_state.holding_thread_handle.Assign(thread ? thread->guest_handle : 0);
-    Memory::Write32(guest_addr, guest_state.raw);
-}
+    if (num_waiters >= 2) {
+        // Notify the guest that there are still some threads waiting for the mutex
+        mutex_value |= Mutex::MutexHasWaitersFlag;
+    }
 
-bool Mutex::GetHasWaiters() const {
-    GuestState guest_state{Memory::Read32(guest_addr)};
-    return guest_state.has_waiters != 0;
-}
+    // Grant the mutex to the next waiting thread and resume it.
+    Memory::Write32(address, mutex_value);
 
-void Mutex::SetHasWaiters(bool has_waiters) {
-    GuestState guest_state{Memory::Read32(guest_addr)};
-    guest_state.has_waiters.Assign(has_waiters ? 1 : 0);
-    Memory::Write32(guest_addr, guest_state.raw);
-}
+    ASSERT(thread->status == THREADSTATUS_WAIT_MUTEX);
+    thread->ResumeFromWait();
 
+    thread->lock_owner = nullptr;
+    thread->condvar_wait_address = 0;
+    thread->mutex_wait_address = 0;
+    thread->wait_handle = 0;
+
+    return RESULT_SUCCESS;
+}
 } // namespace Kernel

src/core/hle/kernel/mutex.h

@@ -15,87 +15,23 @@ namespace Kernel {
 
 class Thread;
 
-class Mutex final : public WaitObject {
+class Mutex final {
 public:
-    /**
-     * Creates a mutex.
-     * @param holding_thread Specifies a thread already holding the mutex. If not nullptr, this
-     * thread will acquire the mutex.
-     * @param guest_addr Address of the object tracking the mutex in guest memory. If specified,
-     * this mutex will update the guest object when its state changes.
-     * @param name Optional name of mutex
-     * @return Pointer to new Mutex object
-     */
-    static SharedPtr<Mutex> Create(SharedPtr<Kernel::Thread> holding_thread, VAddr guest_addr = 0,
-                                   std::string name = "Unknown");
+    /// Flag that indicates that a mutex still has threads waiting for it.
+    static constexpr u32 MutexHasWaitersFlag = 0x40000000;
+    /// Mask of the bits in a mutex address value that contain the mutex owner.
+    static constexpr u32 MutexOwnerMask = 0xBFFFFFFF;
 
-    std::string GetTypeName() const override {
-        return "Mutex";
-    }
-    std::string GetName() const override {
-        return name;
-    }
+    /// Attempts to acquire a mutex at the specified address.
+    static ResultCode TryAcquire(VAddr address, Handle holding_thread_handle,
+                                 Handle requesting_thread_handle);
 
-    static const HandleType HANDLE_TYPE = HandleType::Mutex;
-    HandleType GetHandleType() const override {
-        return HANDLE_TYPE;
-    }
-
-    u32 priority;     ///< The priority of the mutex, used for priority inheritance.
-    std::string name; ///< Name of mutex (optional)
-    VAddr guest_addr; ///< Address of the guest mutex value
-
-    /**
-     * Elevate the mutex priority to the best priority
-     * among the priorities of all its waiting threads.
-     */
-    void UpdatePriority();
-
-    bool ShouldWait(Thread* thread) const override;
-    void Acquire(Thread* thread) override;
-
-    void AddWaitingThread(SharedPtr<Thread> thread) override;
-    void RemoveWaitingThread(Thread* thread) override;
-
-    /**
-     * Attempts to release the mutex from the specified thread.
-     * @param thread Thread that wants to release the mutex.
-     * @returns The result code of the operation.
-     */
-    ResultCode Release(Thread* thread);
-
-    /// Gets the handle to the holding process stored in the guest state.
-    Handle GetOwnerHandle() const;
-
-    /// Gets the Thread pointed to by the owner handle
-    SharedPtr<Thread> GetHoldingThread() const;
-    /// Sets the holding process handle in the guest state.
-    void SetHoldingThread(SharedPtr<Thread> thread);
-
-    /// Returns the has_waiters bit in the guest state.
-    bool GetHasWaiters() const;
-    /// Sets the has_waiters bit in the guest state.
-    void SetHasWaiters(bool has_waiters);
+    /// Releases the mutex at the specified address.
+    static ResultCode Release(VAddr address);
 
 private:
-    Mutex();
-    ~Mutex() override;
-
-    /// Object in guest memory used to track the mutex state
-    union GuestState {
-        u32_le raw;
-        /// Handle of the thread that currently holds the mutex, 0 if available
-        BitField<0, 30, u32_le> holding_thread_handle;
-        /// 1 when there are threads waiting for this mutex, otherwise 0
-        BitField<30, 1, u32_le> has_waiters;
-    };
-    static_assert(sizeof(GuestState) == 4, "GuestState size is incorrect");
+    Mutex() = default;
+    ~Mutex() = default;
 };
 
-/**
- * Releases all the mutexes held by the specified thread
- * @param thread Thread that is holding the mutexes
- */
-void ReleaseThreadMutexes(Thread* thread);
-
 } // namespace Kernel

src/core/hle/kernel/resource_limit.cpp

@@ -34,57 +34,57 @@ SharedPtr<ResourceLimit> ResourceLimit::GetForCategory(ResourceLimitCategory cat
     }
 }
 
-s32 ResourceLimit::GetCurrentResourceValue(u32 resource) const {
+s32 ResourceLimit::GetCurrentResourceValue(ResourceType resource) const {
     switch (resource) {
-    case COMMIT:
+    case ResourceType::Commit:
         return current_commit;
-    case THREAD:
+    case ResourceType::Thread:
         return current_threads;
-    case EVENT:
+    case ResourceType::Event:
         return current_events;
-    case MUTEX:
+    case ResourceType::Mutex:
         return current_mutexes;
-    case SEMAPHORE:
+    case ResourceType::Semaphore:
         return current_semaphores;
-    case TIMER:
+    case ResourceType::Timer:
         return current_timers;
-    case SHARED_MEMORY:
+    case ResourceType::SharedMemory:
         return current_shared_mems;
-    case ADDRESS_ARBITER:
+    case ResourceType::AddressArbiter:
         return current_address_arbiters;
-    case CPU_TIME:
+    case ResourceType::CPUTime:
         return current_cpu_time;
     default:
-        LOG_ERROR(Kernel, "Unknown resource type=%08X", resource);
+        LOG_ERROR(Kernel, "Unknown resource type=%08X", static_cast<u32>(resource));
         UNIMPLEMENTED();
         return 0;
     }
 }
 
-u32 ResourceLimit::GetMaxResourceValue(u32 resource) const {
+u32 ResourceLimit::GetMaxResourceValue(ResourceType resource) const {
     switch (resource) {
-    case PRIORITY:
+    case ResourceType::Priority:
         return max_priority;
-    case COMMIT:
+    case ResourceType::Commit:
         return max_commit;
-    case THREAD:
+    case ResourceType::Thread:
         return max_threads;
-    case EVENT:
+    case ResourceType::Event:
         return max_events;
-    case MUTEX:
+    case ResourceType::Mutex:
         return max_mutexes;
-    case SEMAPHORE:
+    case ResourceType::Semaphore:
         return max_semaphores;
-    case TIMER:
+    case ResourceType::Timer:
         return max_timers;
-    case SHARED_MEMORY:
+    case ResourceType::SharedMemory:
         return max_shared_mems;
-    case ADDRESS_ARBITER:
+    case ResourceType::AddressArbiter:
         return max_address_arbiters;
-    case CPU_TIME:
+    case ResourceType::CPUTime:
         return max_cpu_time;
     default:
-        LOG_ERROR(Kernel, "Unknown resource type=%08X", resource);
+        LOG_ERROR(Kernel, "Unknown resource type=%08X", static_cast<u32>(resource));
         UNIMPLEMENTED();
         return 0;
     }

src/core/hle/kernel/resource_limit.h

@@ -16,17 +16,17 @@ enum class ResourceLimitCategory : u8 {
     OTHER = 3
 };
 
-enum ResourceTypes {
-    PRIORITY = 0,
-    COMMIT = 1,
-    THREAD = 2,
-    EVENT = 3,
-    MUTEX = 4,
-    SEMAPHORE = 5,
-    TIMER = 6,
-    SHARED_MEMORY = 7,
-    ADDRESS_ARBITER = 8,
-    CPU_TIME = 9,
+enum class ResourceType {
+    Priority = 0,
+    Commit = 1,
+    Thread = 2,
+    Event = 3,
+    Mutex = 4,
+    Semaphore = 5,
+    Timer = 6,
+    SharedMemory = 7,
+    AddressArbiter = 8,
+    CPUTime = 9,
 };
 
 class ResourceLimit final : public Object {
@@ -60,14 +60,14 @@ public:
      * @param resource Requested resource type
      * @returns The current value of the resource type
      */
-    s32 GetCurrentResourceValue(u32 resource) const;
+    s32 GetCurrentResourceValue(ResourceType resource) const;
 
     /**
      * Gets the max value for the specified resource.
      * @param resource Requested resource type
      * @returns The max value of the resource type
      */
-    u32 GetMaxResourceValue(u32 resource) const;
+    u32 GetMaxResourceValue(ResourceType resource) const;
 
     /// Name of resource limit object.
     std::string name;

src/core/hle/kernel/svc.cpp

@@ -13,7 +13,6 @@
 #include "core/core_timing.h"
 #include "core/hle/kernel/client_port.h"
 #include "core/hle/kernel/client_session.h"
-#include "core/hle/kernel/condition_variable.h"
 #include "core/hle/kernel/event.h"
 #include "core/hle/kernel/handle_table.h"
 #include "core/hle/kernel/mutex.h"
@@ -262,32 +261,14 @@ static ResultCode ArbitrateLock(Handle holding_thread_handle, VAddr mutex_addr,
               "requesting_current_thread_handle=0x%08X",
               holding_thread_handle, mutex_addr, requesting_thread_handle);
 
-    SharedPtr<Thread> holding_thread = g_handle_table.Get<Thread>(holding_thread_handle);
-    SharedPtr<Thread> requesting_thread = g_handle_table.Get<Thread>(requesting_thread_handle);
-
-    ASSERT(requesting_thread);
-    ASSERT(requesting_thread == GetCurrentThread());
-
-    SharedPtr<Mutex> mutex = g_object_address_table.Get<Mutex>(mutex_addr);
-    if (!mutex) {
-        // Create a new mutex for the specified address if one does not already exist
-        mutex = Mutex::Create(holding_thread, mutex_addr);
-        mutex->name = Common::StringFromFormat("mutex-%llx", mutex_addr);
-    }
-
-    ASSERT(holding_thread == mutex->GetHoldingThread());
-
-    return WaitSynchronization1(mutex, requesting_thread.get());
+    return Mutex::TryAcquire(mutex_addr, holding_thread_handle, requesting_thread_handle);
 }
 
 /// Unlock a mutex
 static ResultCode ArbitrateUnlock(VAddr mutex_addr) {
     LOG_TRACE(Kernel_SVC, "called mutex_addr=0x%llx", mutex_addr);
 
-    SharedPtr<Mutex> mutex = g_object_address_table.Get<Mutex>(mutex_addr);
-    ASSERT(mutex);
-
-    return mutex->Release(GetCurrentThread());
+    return Mutex::Release(mutex_addr);
 }
 
 /// Break program execution
@@ -407,16 +388,11 @@ static ResultCode SetThreadPriority(Handle handle, u32 priority) {
     // Note: The kernel uses the current process's resource limit instead of
     // the one from the thread owner's resource limit.
     SharedPtr<ResourceLimit>& resource_limit = Core::CurrentProcess()->resource_limit;
-    if (resource_limit->GetMaxResourceValue(ResourceTypes::PRIORITY) > priority) {
+    if (resource_limit->GetMaxResourceValue(ResourceType::Priority) > priority) {
         return ERR_NOT_AUTHORIZED;
     }
 
     thread->SetPriority(priority);
-    thread->UpdatePriority();
-
-    // Update the mutexes that this thread is waiting for
-    for (auto& mutex : thread->pending_mutexes)
-        mutex->UpdatePriority();
 
     Core::System::GetInstance().PrepareReschedule();
     return RESULT_SUCCESS;
@@ -541,7 +517,7 @@ static ResultCode CreateThread(Handle* out_handle, VAddr entry_point, u64 arg, V
     }
 
     SharedPtr<ResourceLimit>& resource_limit = Core::CurrentProcess()->resource_limit;
-    if (resource_limit->GetMaxResourceValue(ResourceTypes::PRIORITY) > priority) {
+    if (resource_limit->GetMaxResourceValue(ResourceType::Priority) > priority) {
         return ERR_NOT_AUTHORIZED;
     }
 
@@ -634,77 +610,20 @@ static ResultCode WaitProcessWideKeyAtomic(VAddr mutex_addr, VAddr condition_var
     SharedPtr<Thread> thread = g_handle_table.Get<Thread>(thread_handle);
     ASSERT(thread);
 
-    SharedPtr<Mutex> mutex = g_object_address_table.Get<Mutex>(mutex_addr);
-    if (!mutex) {
-        // Create a new mutex for the specified address if one does not already exist
-        mutex = Mutex::Create(thread, mutex_addr);
-        mutex->name = Common::StringFromFormat("mutex-%llx", mutex_addr);
-    }
+    CASCADE_CODE(Mutex::Release(mutex_addr));
 
-    SharedPtr<ConditionVariable> condition_variable =
-        g_object_address_table.Get<ConditionVariable>(condition_variable_addr);
-    if (!condition_variable) {
-        // Create a new condition_variable for the specified address if one does not already exist
-        condition_variable = ConditionVariable::Create(condition_variable_addr).Unwrap();
-        condition_variable->name =
-            Common::StringFromFormat("condition-variable-%llx", condition_variable_addr);
-    }
+    SharedPtr<Thread> current_thread = GetCurrentThread();
+    current_thread->condvar_wait_address = condition_variable_addr;
+    current_thread->mutex_wait_address = mutex_addr;
+    current_thread->wait_handle = thread_handle;
+    current_thread->status = THREADSTATUS_WAIT_MUTEX;
+    current_thread->wakeup_callback = nullptr;
 
-    if (condition_variable->mutex_addr) {
-        // Previously created the ConditionVariable using WaitProcessWideKeyAtomic, verify
-        // everything is correct
-        ASSERT(condition_variable->mutex_addr == mutex_addr);
-    } else {
-        // Previously created the ConditionVariable using SignalProcessWideKey, set the mutex
-        // associated with it
-        condition_variable->mutex_addr = mutex_addr;
-    }
+    current_thread->WakeAfterDelay(nano_seconds);
 
-    if (mutex->GetOwnerHandle()) {
-        // Release the mutex if the current thread is holding it
-        mutex->Release(thread.get());
-    }
-
-    auto wakeup_callback = [mutex, nano_seconds](ThreadWakeupReason reason,
-                                                 SharedPtr<Thread> thread,
-                                                 SharedPtr<WaitObject> object, size_t index) {
-        ASSERT(thread->status == THREADSTATUS_WAIT_SYNCH_ANY);
-
-        if (reason == ThreadWakeupReason::Timeout) {
-            thread->SetWaitSynchronizationResult(RESULT_TIMEOUT);
-            return true;
-        }
-
-        ASSERT(reason == ThreadWakeupReason::Signal);
-
-        // Now try to acquire the mutex and don't resume if it's not available.
-        if (!mutex->ShouldWait(thread.get())) {
-            mutex->Acquire(thread.get());
-            thread->SetWaitSynchronizationResult(RESULT_SUCCESS);
-            return true;
-        }
-
-        if (nano_seconds == 0) {
-            thread->SetWaitSynchronizationResult(RESULT_TIMEOUT);
-            return true;
-        }
-
-        thread->wait_objects = {mutex};
-        mutex->AddWaitingThread(thread);
-        thread->status = THREADSTATUS_WAIT_SYNCH_ANY;
-
-        // Create an event to wake the thread up after the
-        // specified nanosecond delay has passed
-        thread->WakeAfterDelay(nano_seconds);
-        thread->wakeup_callback = DefaultThreadWakeupCallback;
-
-        Core::System::GetInstance().PrepareReschedule();
-
-        return false;
-    };
-    CASCADE_CODE(
-        WaitSynchronization1(condition_variable, thread.get(), nano_seconds, wakeup_callback));
+    // Note: Deliberately don't attempt to inherit the lock owner's priority.
 
+    Core::System::GetInstance().PrepareReschedule();
     return RESULT_SUCCESS;
 }
 
@@ -713,24 +632,53 @@ static ResultCode SignalProcessWideKey(VAddr condition_variable_addr, s32 target
     LOG_TRACE(Kernel_SVC, "called, condition_variable_addr=0x%llx, target=0x%08x",
               condition_variable_addr, target);
 
-    // Wakeup all or one thread - Any other value is unimplemented
-    ASSERT(target == -1 || target == 1);
+    u32 processed = 0;
+    auto& thread_list = Core::System::GetInstance().Scheduler().GetThreadList();
 
-    SharedPtr<ConditionVariable> condition_variable =
-        g_object_address_table.Get<ConditionVariable>(condition_variable_addr);
-    if (!condition_variable) {
-        // Create a new condition_variable for the specified address if one does not already exist
-        condition_variable = ConditionVariable::Create(condition_variable_addr).Unwrap();
-        condition_variable->name =
-            Common::StringFromFormat("condition-variable-%llx", condition_variable_addr);
-    }
+    for (auto& thread : thread_list) {
+        if (thread->condvar_wait_address != condition_variable_addr)
+            continue;
 
-    CASCADE_CODE(condition_variable->Release(target));
+        // Only process up to 'target' threads, unless 'target' is -1, in which case process
+        // them all.
+        if (target != -1 && processed >= target)
+            break;
 
-    if (condition_variable->mutex_addr) {
-        // If a mutex was created for this condition_variable, wait the current thread on it
-        SharedPtr<Mutex> mutex = g_object_address_table.Get<Mutex>(condition_variable->mutex_addr);
-        return WaitSynchronization1(mutex, GetCurrentThread());
+        // If the mutex is not yet acquired, acquire it.
+        u32 mutex_val = Memory::Read32(thread->mutex_wait_address);
+
+        if (mutex_val == 0) {
+            // We were able to acquire the mutex, resume this thread.
+            Memory::Write32(thread->mutex_wait_address, thread->wait_handle);
+            ASSERT(thread->status == THREADSTATUS_WAIT_MUTEX);
+            thread->ResumeFromWait();
+
+            auto lock_owner = thread->lock_owner;
+            if (lock_owner)
+                lock_owner->RemoveMutexWaiter(thread);
+
+            thread->lock_owner = nullptr;
+            thread->mutex_wait_address = 0;
+            thread->condvar_wait_address = 0;
+            thread->wait_handle = 0;
+        } else {
+            // Couldn't acquire the mutex, block the thread.
+            Handle owner_handle = static_cast<Handle>(mutex_val & Mutex::MutexOwnerMask);
+            auto owner = g_handle_table.Get<Thread>(owner_handle);
+            ASSERT(owner);
+            ASSERT(thread->status != THREADSTATUS_RUNNING);
+            thread->status = THREADSTATUS_WAIT_MUTEX;
+            thread->wakeup_callback = nullptr;
+
+            // Signal that the mutex now has a waiting thread.
+            Memory::Write32(thread->mutex_wait_address, mutex_val | Mutex::MutexHasWaitersFlag);
+
+            owner->AddMutexWaiter(thread);
+
+            Core::System::GetInstance().PrepareReschedule();
+        }
+
+        ++processed;
     }
 
     return RESULT_SUCCESS;

src/core/hle/kernel/thread.cpp

@@ -77,9 +77,6 @@ void Thread::Stop() {
     }
     wait_objects.clear();
 
-    // Release all the mutexes that this thread holds
-    ReleaseThreadMutexes(this);
-
     // Mark the TLS slot in the thread's page as free.
     u64 tls_page = (tls_address - Memory::TLS_AREA_VADDR) / Memory::PAGE_SIZE;
     u64 tls_slot =
@@ -126,6 +123,19 @@ static void ThreadWakeupCallback(u64 thread_handle, int cycles_late) {
             resume = thread->wakeup_callback(ThreadWakeupReason::Timeout, thread, nullptr, 0);
     }
 
+    if (thread->mutex_wait_address != 0 || thread->condvar_wait_address != 0 ||
+        thread->wait_handle) {
+        ASSERT(thread->status == THREADSTATUS_WAIT_MUTEX);
+        thread->mutex_wait_address = 0;
+        thread->condvar_wait_address = 0;
+        thread->wait_handle = 0;
+
+        auto lock_owner = thread->lock_owner;
+        // Threads waking up by timeout from WaitProcessWideKey do not perform priority inheritance
+        // and don't have a lock owner.
+        ASSERT(lock_owner == nullptr);
+    }
+
     if (resume)
         thread->ResumeFromWait();
 }
@@ -151,6 +161,7 @@ void Thread::ResumeFromWait() {
     case THREADSTATUS_WAIT_HLE_EVENT:
     case THREADSTATUS_WAIT_SLEEP:
     case THREADSTATUS_WAIT_IPC:
+    case THREADSTATUS_WAIT_MUTEX:
         break;
 
     case THREADSTATUS_READY:
@@ -256,7 +267,9 @@ ResultVal<SharedPtr<Thread>> Thread::Create(std::string name, VAddr entry_point,
     thread->last_running_ticks = CoreTiming::GetTicks();
     thread->processor_id = processor_id;
     thread->wait_objects.clear();
-    thread->wait_address = 0;
+    thread->mutex_wait_address = 0;
+    thread->condvar_wait_address = 0;
+    thread->wait_handle = 0;
     thread->name = std::move(name);
     thread->callback_handle = wakeup_callback_handle_table.Create(thread).Unwrap();
     thread->owner_process = owner_process;
@@ -317,17 +330,8 @@ ResultVal<SharedPtr<Thread>> Thread::Create(std::string name, VAddr entry_point,
 void Thread::SetPriority(u32 priority) {
     ASSERT_MSG(priority <= THREADPRIO_LOWEST && priority >= THREADPRIO_HIGHEST,
                "Invalid priority value.");
-    Core::System::GetInstance().Scheduler().SetThreadPriority(this, priority);
-    nominal_priority = current_priority = priority;
-}
-
-void Thread::UpdatePriority() {
-    u32 best_priority = nominal_priority;
-    for (auto& mutex : held_mutexes) {
-        if (mutex->priority < best_priority)
-            best_priority = mutex->priority;
-    }
-    BoostPriority(best_priority);
+    nominal_priority = priority;
+    UpdatePriority();
 }
 
 void Thread::BoostPriority(u32 priority) {
@@ -377,6 +381,38 @@ VAddr Thread::GetCommandBufferAddress() const {
     return GetTLSAddress() + CommandHeaderOffset;
 }
 
+void Thread::AddMutexWaiter(SharedPtr<Thread> thread) {
+    thread->lock_owner = this;
+    wait_mutex_threads.emplace_back(std::move(thread));
+    UpdatePriority();
+}
+
+void Thread::RemoveMutexWaiter(SharedPtr<Thread> thread) {
+    boost::remove_erase(wait_mutex_threads, thread);
+    thread->lock_owner = nullptr;
+    UpdatePriority();
+}
+
+void Thread::UpdatePriority() {
+    // Find the highest priority among all the threads that are waiting for this thread's lock
+    u32 new_priority = nominal_priority;
+    for (const auto& thread : wait_mutex_threads) {
+        if (thread->nominal_priority < new_priority)
+            new_priority = thread->nominal_priority;
+    }
+
+    if (new_priority == current_priority)
+        return;
+
+    Core::System::GetInstance().Scheduler().SetThreadPriority(this, new_priority);
+
+    current_priority = new_priority;
+
+    // Recursively update the priority of the thread that depends on the priority of this one.
+    if (lock_owner)
+        lock_owner->UpdatePriority();
+}
+
 ////////////////////////////////////////////////////////////////////////////////////////////////////
 
 /**

src/core/hle/kernel/thread.h

@@ -18,7 +18,7 @@
 enum ThreadPriority : u32 {
     THREADPRIO_HIGHEST = 0,       ///< Highest thread priority
     THREADPRIO_USERLAND_MAX = 24, ///< Highest thread priority for userland apps
-    THREADPRIO_DEFAULT = 48,      ///< Default thread priority for userland apps
+    THREADPRIO_DEFAULT = 44,      ///< Default thread priority for userland apps
     THREADPRIO_LOWEST = 63,       ///< Lowest thread priority
 };
 
@@ -43,6 +43,7 @@ enum ThreadStatus {
     THREADSTATUS_WAIT_IPC,       ///< Waiting for the reply from an IPC request
     THREADSTATUS_WAIT_SYNCH_ANY, ///< Waiting due to WaitSynch1 or WaitSynchN with wait_all = false
     THREADSTATUS_WAIT_SYNCH_ALL, ///< Waiting due to WaitSynchronizationN with wait_all = true
+    THREADSTATUS_WAIT_MUTEX,     ///< Waiting due to an ArbitrateLock/WaitProcessWideKey svc
     THREADSTATUS_DORMANT,        ///< Created but not yet made ready
     THREADSTATUS_DEAD            ///< Run to completion, or forcefully terminated
 };
@@ -54,7 +55,6 @@ enum class ThreadWakeupReason {
 
 namespace Kernel {
 
-class Mutex;
 class Process;
 
 class Thread final : public WaitObject {
@@ -103,18 +103,21 @@ public:
      */
     void SetPriority(u32 priority);
 
-    /**
-     * Boost's a thread's priority to the best priority among the thread's held mutexes.
-     * This prevents priority inversion via priority inheritance.
-     */
-    void UpdatePriority();
-
     /**
      * Temporarily boosts the thread's priority until the next time it is scheduled
      * @param priority The new priority
      */
     void BoostPriority(u32 priority);
 
+    /// Adds a thread to the list of threads that are waiting for a lock held by this thread.
+    void AddMutexWaiter(SharedPtr<Thread> thread);
+
+    /// Removes a thread from the list of threads that are waiting for a lock held by this thread.
+    void RemoveMutexWaiter(SharedPtr<Thread> thread);
+
+    /// Recalculates the current priority taking into account priority inheritance.
+    void UpdatePriority();
+
     /**
      * Gets the thread's thread ID
      * @return The thread's ID
@@ -205,19 +208,22 @@ public:
 
     VAddr tls_address; ///< Virtual address of the Thread Local Storage of the thread
 
-    /// Mutexes currently held by this thread, which will be released when it exits.
-    boost::container::flat_set<SharedPtr<Mutex>> held_mutexes;
-
-    /// Mutexes that this thread is currently waiting for.
-    boost::container::flat_set<SharedPtr<Mutex>> pending_mutexes;
-
     SharedPtr<Process> owner_process; ///< Process that owns this thread
 
     /// Objects that the thread is waiting on, in the same order as they were
     // passed to WaitSynchronization1/N.
     std::vector<SharedPtr<WaitObject>> wait_objects;
 
-    VAddr wait_address; ///< If waiting on an AddressArbiter, this is the arbitration address
+    /// List of threads that are waiting for a mutex that is held by this thread.
+    std::vector<SharedPtr<Thread>> wait_mutex_threads;
+
+    /// Thread that owns the lock that this thread is waiting for.
+    SharedPtr<Thread> lock_owner;
+
+    // If waiting on a ConditionVariable, this is the ConditionVariable  address
+    VAddr condvar_wait_address;
+    VAddr mutex_wait_address; ///< If waiting on a Mutex, this is the mutex address
+    Handle wait_handle;       ///< The handle used to wait for the mutex.
 
     std::string name;
 

src/core/hle/service/acc/acc.cpp

@@ -47,7 +47,7 @@ public:
 
 private:
     void GetBase(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_ACC, "(STUBBED) called");
+        NGLOG_WARNING(Service_ACC, "(STUBBED) called");
         ProfileBase profile_base{};
         IPC::ResponseBuilder rb{ctx, 16};
         rb.Push(RESULT_SUCCESS);
@@ -72,14 +72,14 @@ public:
 
 private:
     void CheckAvailability(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_ACC, "(STUBBED) called");
+        NGLOG_WARNING(Service_ACC, "(STUBBED) called");
         IPC::ResponseBuilder rb{ctx, 3};
         rb.Push(RESULT_SUCCESS);
         rb.Push(true); // TODO: Check when this is supposed to return true and when not
     }
 
     void GetAccountId(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_ACC, "(STUBBED) called");
+        NGLOG_WARNING(Service_ACC, "(STUBBED) called");
         IPC::ResponseBuilder rb{ctx, 4};
         rb.Push(RESULT_SUCCESS);
         rb.Push<u64>(0x12345678ABCDEF);
@@ -87,14 +87,14 @@ private:
 };
 
 void Module::Interface::GetUserExistence(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_ACC, "(STUBBED) called");
+    NGLOG_WARNING(Service_ACC, "(STUBBED) called");
     IPC::ResponseBuilder rb{ctx, 3};
     rb.Push(RESULT_SUCCESS);
     rb.Push(true); // TODO: Check when this is supposed to return true and when not
 }
 
 void Module::Interface::ListAllUsers(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_ACC, "(STUBBED) called");
+    NGLOG_WARNING(Service_ACC, "(STUBBED) called");
     constexpr std::array<u128, 10> user_ids{DEFAULT_USER_ID};
     ctx.WriteBuffer(user_ids.data(), user_ids.size());
     IPC::ResponseBuilder rb{ctx, 2};
@@ -102,7 +102,7 @@ void Module::Interface::ListAllUsers(Kernel::HLERequestContext& ctx) {
 }
 
 void Module::Interface::ListOpenUsers(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_ACC, "(STUBBED) called");
+    NGLOG_WARNING(Service_ACC, "(STUBBED) called");
     constexpr std::array<u128, 10> user_ids{DEFAULT_USER_ID};
     ctx.WriteBuffer(user_ids.data(), user_ids.size());
     IPC::ResponseBuilder rb{ctx, 2};
@@ -113,11 +113,11 @@ void Module::Interface::GetProfile(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
     rb.PushIpcInterface<IProfile>();
-    LOG_DEBUG(Service_ACC, "called");
+    NGLOG_DEBUG(Service_ACC, "called");
 }
 
 void Module::Interface::InitializeApplicationInfo(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_ACC, "(STUBBED) called");
+    NGLOG_WARNING(Service_ACC, "(STUBBED) called");
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
 }
@@ -126,11 +126,11 @@ void Module::Interface::GetBaasAccountManagerForApplication(Kernel::HLERequestCo
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
     rb.PushIpcInterface<IManagerForApplication>();
-    LOG_DEBUG(Service_ACC, "called");
+    NGLOG_DEBUG(Service_ACC, "called");
 }
 
 void Module::Interface::GetLastOpenedUser(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_ACC, "(STUBBED) called");
+    NGLOG_WARNING(Service_ACC, "(STUBBED) called");
     IPC::ResponseBuilder rb{ctx, 6};
     rb.Push(RESULT_SUCCESS);
     rb.PushRaw(DEFAULT_USER_ID);

src/core/hle/service/am/am.cpp

@@ -28,14 +28,14 @@ IWindowController::IWindowController() : ServiceFramework("IWindowController") {
 }
 
 void IWindowController::GetAppletResourceUserId(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    NGLOG_WARNING(Service_AM, "(STUBBED) called");
     IPC::ResponseBuilder rb{ctx, 4};
     rb.Push(RESULT_SUCCESS);
     rb.Push<u64>(0);
 }
 
 void IWindowController::AcquireForegroundRights(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    NGLOG_WARNING(Service_AM, "(STUBBED) called");
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
 }
@@ -54,20 +54,20 @@ IAudioController::IAudioController() : ServiceFramework("IAudioController") {
 }
 
 void IAudioController::SetExpectedMasterVolume(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    NGLOG_WARNING(Service_AM, "(STUBBED) called");
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
 }
 
 void IAudioController::GetMainAppletExpectedMasterVolume(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    NGLOG_WARNING(Service_AM, "(STUBBED) called");
     IPC::ResponseBuilder rb{ctx, 3};
     rb.Push(RESULT_SUCCESS);
     rb.Push(volume);
 }
 
 void IAudioController::GetLibraryAppletExpectedMasterVolume(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    NGLOG_WARNING(Service_AM, "(STUBBED) called");
     IPC::ResponseBuilder rb{ctx, 3};
     rb.Push(RESULT_SUCCESS);
     rb.Push(volume);
@@ -139,14 +139,14 @@ void ISelfController::SetFocusHandlingMode(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
 
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    NGLOG_WARNING(Service_AM, "(STUBBED) called");
 }
 
 void ISelfController::SetRestartMessageEnabled(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
 
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    NGLOG_WARNING(Service_AM, "(STUBBED) called");
 }
 
 void ISelfController::SetPerformanceModeChangedNotification(Kernel::HLERequestContext& ctx) {
@@ -157,14 +157,14 @@ void ISelfController::SetPerformanceModeChangedNotification(Kernel::HLERequestCo
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
 
-    LOG_WARNING(Service_AM, "(STUBBED) called flag=%u", static_cast<u32>(flag));
+    NGLOG_WARNING(Service_AM, "(STUBBED) called flag={}", flag);
 }
 
 void ISelfController::SetScreenShotPermission(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
 
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    NGLOG_WARNING(Service_AM, "(STUBBED) called");
 }
 
 void ISelfController::SetOperationModeChangedNotification(Kernel::HLERequestContext& ctx) {
@@ -175,7 +175,7 @@ void ISelfController::SetOperationModeChangedNotification(Kernel::HLERequestCont
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
 
-    LOG_WARNING(Service_AM, "(STUBBED) called flag=%u", static_cast<u32>(flag));
+    NGLOG_WARNING(Service_AM, "(STUBBED) called flag={}", flag);
 }
 
 void ISelfController::SetOutOfFocusSuspendingEnabled(Kernel::HLERequestContext& ctx) {
@@ -188,21 +188,21 @@ void ISelfController::SetOutOfFocusSuspendingEnabled(Kernel::HLERequestContext&
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
 
-    LOG_WARNING(Service_AM, "(STUBBED) called enabled=%u", static_cast<u32>(enabled));
+    NGLOG_WARNING(Service_AM, "(STUBBED) called enabled={}", enabled);
 }
 
 void ISelfController::LockExit(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
 
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    NGLOG_WARNING(Service_AM, "(STUBBED) called");
 }
 
 void ISelfController::UnlockExit(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
 
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    NGLOG_WARNING(Service_AM, "(STUBBED) called");
 }
 
 void ISelfController::GetLibraryAppletLaunchableEvent(Kernel::HLERequestContext& ctx) {
@@ -212,7 +212,7 @@ void ISelfController::GetLibraryAppletLaunchableEvent(Kernel::HLERequestContext&
     rb.Push(RESULT_SUCCESS);
     rb.PushCopyObjects(launchable_event);
 
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    NGLOG_WARNING(Service_AM, "(STUBBED) called");
 }
 
 void ISelfController::CreateManagedDisplayLayer(Kernel::HLERequestContext& ctx) {
@@ -225,7 +225,7 @@ void ISelfController::CreateManagedDisplayLayer(Kernel::HLERequestContext& ctx)
     rb.Push(RESULT_SUCCESS);
     rb.Push(layer_id);
 
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    NGLOG_WARNING(Service_AM, "(STUBBED) called");
 }
 
 ICommonStateGetter::ICommonStateGetter() : ServiceFramework("ICommonStateGetter") {
@@ -269,7 +269,7 @@ void ICommonStateGetter::GetEventHandle(Kernel::HLERequestContext& ctx) {
     rb.Push(RESULT_SUCCESS);
     rb.PushCopyObjects(event);
 
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    NGLOG_WARNING(Service_AM, "(STUBBED) called");
 }
 
 void ICommonStateGetter::ReceiveMessage(Kernel::HLERequestContext& ctx) {
@@ -277,7 +277,7 @@ void ICommonStateGetter::ReceiveMessage(Kernel::HLERequestContext& ctx) {
     rb.Push(RESULT_SUCCESS);
     rb.Push<u32>(15);
 
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    NGLOG_WARNING(Service_AM, "(STUBBED) called");
 }
 
 void ICommonStateGetter::GetCurrentFocusState(Kernel::HLERequestContext& ctx) {
@@ -285,7 +285,7 @@ void ICommonStateGetter::GetCurrentFocusState(Kernel::HLERequestContext& ctx) {
     rb.Push(RESULT_SUCCESS);
     rb.Push(static_cast<u8>(FocusState::InFocus));
 
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    NGLOG_WARNING(Service_AM, "(STUBBED) called");
 }
 
 void ICommonStateGetter::GetOperationMode(Kernel::HLERequestContext& ctx) {
@@ -294,7 +294,7 @@ void ICommonStateGetter::GetOperationMode(Kernel::HLERequestContext& ctx) {
     rb.Push(RESULT_SUCCESS);
     rb.Push(static_cast<u8>(use_docked_mode ? OperationMode::Docked : OperationMode::Handheld));
 
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    NGLOG_WARNING(Service_AM, "(STUBBED) called");
 }
 
 void ICommonStateGetter::GetPerformanceMode(Kernel::HLERequestContext& ctx) {
@@ -304,7 +304,7 @@ void ICommonStateGetter::GetPerformanceMode(Kernel::HLERequestContext& ctx) {
     rb.Push(static_cast<u32>(use_docked_mode ? APM::PerformanceMode::Docked
                                              : APM::PerformanceMode::Handheld));
 
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    NGLOG_WARNING(Service_AM, "(STUBBED) called");
 }
 
 class ILibraryAppletAccessor final : public ServiceFramework<ILibraryAppletAccessor> {
@@ -344,7 +344,7 @@ private:
         rb.Push(RESULT_SUCCESS);
         rb.PushCopyObjects(state_changed_event);
 
-        LOG_WARNING(Service_AM, "(STUBBED) called");
+        NGLOG_WARNING(Service_AM, "(STUBBED) called");
     }
 
     Kernel::SharedPtr<Kernel::Event> state_changed_event;
@@ -368,7 +368,7 @@ void ILibraryAppletCreator::CreateLibraryApplet(Kernel::HLERequestContext& ctx)
     rb.Push(RESULT_SUCCESS);
     rb.PushIpcInterface<AM::ILibraryAppletAccessor>();
 
-    LOG_DEBUG(Service_AM, "called");
+    NGLOG_DEBUG(Service_AM, "called");
 }
 
 class IStorageAccessor final : public ServiceFramework<IStorageAccessor> {
@@ -392,7 +392,7 @@ private:
         rb.Push(RESULT_SUCCESS);
         rb.Push(static_cast<u64>(buffer.size()));
 
-        LOG_DEBUG(Service_AM, "called");
+        NGLOG_DEBUG(Service_AM, "called");
     }
 
     void Read(Kernel::HLERequestContext& ctx) {
@@ -410,7 +410,7 @@ private:
 
         rb.Push(RESULT_SUCCESS);
 
-        LOG_DEBUG(Service_AM, "called");
+        NGLOG_DEBUG(Service_AM, "called");
     }
 };
 
@@ -434,7 +434,7 @@ private:
         rb.Push(RESULT_SUCCESS);
         rb.PushIpcInterface<AM::IStorageAccessor>(buffer);
 
-        LOG_DEBUG(Service_AM, "called");
+        NGLOG_DEBUG(Service_AM, "called");
     }
 };
 
@@ -498,14 +498,14 @@ void IApplicationFunctions::PopLaunchParameter(Kernel::HLERequestContext& ctx) {
     rb.Push(RESULT_SUCCESS);
     rb.PushIpcInterface<AM::IStorage>(buffer);
 
-    LOG_DEBUG(Service_AM, "called");
+    NGLOG_DEBUG(Service_AM, "called");
 }
 
 void IApplicationFunctions::EnsureSaveData(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
     u128 uid = rp.PopRaw<u128>();
 
-    LOG_WARNING(Service, "(STUBBED) called uid = %016" PRIX64 "%016" PRIX64, uid[1], uid[0]);
+    NGLOG_WARNING(Service, "(STUBBED) called uid = {:016X}{:016X}", uid[1], uid[0]);
 
     IPC::ResponseBuilder rb{ctx, 4};
 
@@ -533,27 +533,27 @@ void IApplicationFunctions::SetTerminateResult(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
 
-    LOG_WARNING(Service_AM, "(STUBBED) called, result=0x%08X", result);
+    NGLOG_WARNING(Service_AM, "(STUBBED) called, result={:#010}", result);
 }
 
 void IApplicationFunctions::GetDesiredLanguage(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 4};
     rb.Push(RESULT_SUCCESS);
     rb.Push<u64>(SystemLanguage::English);
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    NGLOG_WARNING(Service_AM, "(STUBBED) called");
 }
 
 void IApplicationFunctions::InitializeGamePlayRecording(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    NGLOG_WARNING(Service_AM, "(STUBBED) called");
 }
 
 void IApplicationFunctions::SetGamePlayRecordingState(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
 
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    NGLOG_WARNING(Service_AM, "(STUBBED) called");
 }
 
 void IApplicationFunctions::NotifyRunning(Kernel::HLERequestContext& ctx) {
@@ -561,7 +561,7 @@ void IApplicationFunctions::NotifyRunning(Kernel::HLERequestContext& ctx) {
     rb.Push(RESULT_SUCCESS);
     rb.Push<u8>(0); // Unknown, seems to be ignored by official processes
 
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    NGLOG_WARNING(Service_AM, "(STUBBED) called");
 }
 
 void InstallInterfaces(SM::ServiceManager& service_manager,

src/core/hle/service/am/applet_ae.cpp

@@ -33,56 +33,56 @@ private:
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
         rb.PushIpcInterface<ICommonStateGetter>();
-        LOG_DEBUG(Service_AM, "called");
+        NGLOG_DEBUG(Service_AM, "called");
     }
 
     void GetSelfController(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
         rb.PushIpcInterface<ISelfController>(nvflinger);
-        LOG_DEBUG(Service_AM, "called");
+        NGLOG_DEBUG(Service_AM, "called");
     }
 
     void GetWindowController(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
         rb.PushIpcInterface<IWindowController>();
-        LOG_DEBUG(Service_AM, "called");
+        NGLOG_DEBUG(Service_AM, "called");
     }
 
     void GetAudioController(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
         rb.PushIpcInterface<IAudioController>();
-        LOG_DEBUG(Service_AM, "called");
+        NGLOG_DEBUG(Service_AM, "called");
     }
 
     void GetDisplayController(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
         rb.PushIpcInterface<IDisplayController>();
-        LOG_DEBUG(Service_AM, "called");
+        NGLOG_DEBUG(Service_AM, "called");
     }
 
     void GetDebugFunctions(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
         rb.PushIpcInterface<IDebugFunctions>();
-        LOG_DEBUG(Service_AM, "called");
+        NGLOG_DEBUG(Service_AM, "called");
     }
 
     void GetLibraryAppletCreator(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
         rb.PushIpcInterface<ILibraryAppletCreator>();
-        LOG_DEBUG(Service_AM, "called");
+        NGLOG_DEBUG(Service_AM, "called");
     }
 
     void GetApplicationFunctions(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
         rb.PushIpcInterface<IApplicationFunctions>();
-        LOG_DEBUG(Service_AM, "called");
+        NGLOG_DEBUG(Service_AM, "called");
     }
 
     std::shared_ptr<NVFlinger::NVFlinger> nvflinger;
@@ -92,7 +92,7 @@ void AppletAE::OpenLibraryAppletProxyOld(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
     rb.PushIpcInterface<ILibraryAppletProxy>(nvflinger);
-    LOG_DEBUG(Service_AM, "called");
+    NGLOG_DEBUG(Service_AM, "called");
 }
 
 AppletAE::AppletAE(std::shared_ptr<NVFlinger::NVFlinger> nvflinger)

src/core/hle/service/am/applet_oe.cpp

@@ -33,56 +33,56 @@ private:
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
         rb.PushIpcInterface<IAudioController>();
-        LOG_DEBUG(Service_AM, "called");
+        NGLOG_DEBUG(Service_AM, "called");
     }
 
     void GetDisplayController(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
         rb.PushIpcInterface<IDisplayController>();
-        LOG_DEBUG(Service_AM, "called");
+        NGLOG_DEBUG(Service_AM, "called");
     }
 
     void GetDebugFunctions(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
         rb.PushIpcInterface<IDebugFunctions>();
-        LOG_DEBUG(Service_AM, "called");
+        NGLOG_DEBUG(Service_AM, "called");
     }
 
     void GetWindowController(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
         rb.PushIpcInterface<IWindowController>();
-        LOG_DEBUG(Service_AM, "called");
+        NGLOG_DEBUG(Service_AM, "called");
     }
 
     void GetSelfController(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
         rb.PushIpcInterface<ISelfController>(nvflinger);
-        LOG_DEBUG(Service_AM, "called");
+        NGLOG_DEBUG(Service_AM, "called");
     }
 
     void GetCommonStateGetter(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
         rb.PushIpcInterface<ICommonStateGetter>();
-        LOG_DEBUG(Service_AM, "called");
+        NGLOG_DEBUG(Service_AM, "called");
     }
 
     void GetLibraryAppletCreator(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
         rb.PushIpcInterface<ILibraryAppletCreator>();
-        LOG_DEBUG(Service_AM, "called");
+        NGLOG_DEBUG(Service_AM, "called");
     }
 
     void GetApplicationFunctions(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
         rb.PushIpcInterface<IApplicationFunctions>();
-        LOG_DEBUG(Service_AM, "called");
+        NGLOG_DEBUG(Service_AM, "called");
     }
 
     std::shared_ptr<NVFlinger::NVFlinger> nvflinger;
@@ -92,7 +92,7 @@ void AppletOE::OpenApplicationProxy(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
     rb.PushIpcInterface<IApplicationProxy>(nvflinger);
-    LOG_DEBUG(Service_AM, "called");
+    NGLOG_DEBUG(Service_AM, "called");
 }
 
 AppletOE::AppletOE(std::shared_ptr<NVFlinger::NVFlinger> nvflinger)

src/core/hle/service/aoc/aoc_u.cpp

@@ -27,14 +27,14 @@ void AOC_U::CountAddOnContent(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 4};
     rb.Push(RESULT_SUCCESS);
     rb.Push<u64>(0);
-    LOG_WARNING(Service_AOC, "(STUBBED) called");
+    NGLOG_WARNING(Service_AOC, "(STUBBED) called");
 }
 
 void AOC_U::ListAddOnContent(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 4};
     rb.Push(RESULT_SUCCESS);
     rb.Push<u64>(0);
-    LOG_WARNING(Service_AOC, "(STUBBED) called");
+    NGLOG_WARNING(Service_AOC, "(STUBBED) called");
 }
 
 void InstallInterfaces(SM::ServiceManager& service_manager) {

src/core/hle/service/apm/interface.cpp

@@ -29,8 +29,8 @@ private:
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
 
-        LOG_WARNING(Service_APM, "(STUBBED) called mode=%u config=%u", static_cast<u32>(mode),
-                    config);
+        NGLOG_WARNING(Service_APM, "(STUBBED) called mode={} config={}", static_cast<u32>(mode),
+                      config);
     }
 
     void GetPerformanceConfiguration(Kernel::HLERequestContext& ctx) {
@@ -42,7 +42,7 @@ private:
         rb.Push(RESULT_SUCCESS);
         rb.Push<u32>(0); // Performance configuration
 
-        LOG_WARNING(Service_APM, "(STUBBED) called mode=%u", static_cast<u32>(mode));
+        NGLOG_WARNING(Service_APM, "(STUBBED) called mode={}", static_cast<u32>(mode));
     }
 };
 

src/core/hle/service/audio/audout_u.cpp

@@ -60,14 +60,14 @@ public:
 
 private:
     void GetAudioOutState(Kernel::HLERequestContext& ctx) {
-        LOG_DEBUG(Service_Audio, "called");
+        NGLOG_DEBUG(Service_Audio, "called");
         IPC::ResponseBuilder rb{ctx, 3};
         rb.Push(RESULT_SUCCESS);
         rb.Push(static_cast<u32>(audio_out_state));
     }
 
     void StartAudioOut(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_Audio, "(STUBBED) called");
+        NGLOG_WARNING(Service_Audio, "(STUBBED) called");
 
         // Start audio
         audio_out_state = AudioState::Started;
@@ -77,7 +77,7 @@ private:
     }
 
     void StopAudioOut(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_Audio, "(STUBBED) called");
+        NGLOG_WARNING(Service_Audio, "(STUBBED) called");
 
         // Stop audio
         audio_out_state = AudioState::Stopped;
@@ -89,7 +89,7 @@ private:
     }
 
     void RegisterBufferEvent(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_Audio, "(STUBBED) called");
+        NGLOG_WARNING(Service_Audio, "(STUBBED) called");
 
         IPC::ResponseBuilder rb{ctx, 2, 1};
         rb.Push(RESULT_SUCCESS);
@@ -97,7 +97,7 @@ private:
     }
 
     void AppendAudioOutBuffer(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_Audio, "(STUBBED) called");
+        NGLOG_WARNING(Service_Audio, "(STUBBED) called");
         IPC::RequestParser rp{ctx};
 
         const u64 key{rp.Pop<u64>()};
@@ -108,7 +108,7 @@ private:
     }
 
     void GetReleasedAudioOutBuffer(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_Audio, "(STUBBED) called");
+        NGLOG_WARNING(Service_Audio, "(STUBBED) called");
 
         // TODO(st4rk): This is how libtransistor currently implements the
         // GetReleasedAudioOutBuffer, it should return the key (a VAddr) to the app and this address
@@ -164,7 +164,7 @@ private:
 };
 
 void AudOutU::ListAudioOuts(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_Audio, "(STUBBED) called");
+    NGLOG_WARNING(Service_Audio, "(STUBBED) called");
     IPC::RequestParser rp{ctx};
 
     const std::string audio_interface = "AudioInterface";
@@ -180,7 +180,7 @@ void AudOutU::ListAudioOuts(Kernel::HLERequestContext& ctx) {
 }
 
 void AudOutU::OpenAudioOut(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_Audio, "(STUBBED) called");
+    NGLOG_WARNING(Service_Audio, "(STUBBED) called");
 
     if (!audio_out_interface) {
         audio_out_interface = std::make_shared<IAudioOut>();

src/core/hle/service/audio/audren_u.cpp

@@ -56,7 +56,7 @@ private:
     }
 
     void RequestUpdateAudioRenderer(Kernel::HLERequestContext& ctx) {
-        LOG_DEBUG(Service_Audio, "%s", ctx.Description().c_str());
+        NGLOG_DEBUG(Service_Audio, "{}", ctx.Description());
         AudioRendererResponseData response_data{};
 
         response_data.section_0_size =
@@ -79,7 +79,7 @@ private:
 
         rb.Push(RESULT_SUCCESS);
 
-        LOG_WARNING(Service_Audio, "(STUBBED) called");
+        NGLOG_WARNING(Service_Audio, "(STUBBED) called");
     }
 
     void StartAudioRenderer(Kernel::HLERequestContext& ctx) {
@@ -87,7 +87,7 @@ private:
 
         rb.Push(RESULT_SUCCESS);
 
-        LOG_WARNING(Service_Audio, "(STUBBED) called");
+        NGLOG_WARNING(Service_Audio, "(STUBBED) called");
     }
 
     void StopAudioRenderer(Kernel::HLERequestContext& ctx) {
@@ -95,7 +95,7 @@ private:
 
         rb.Push(RESULT_SUCCESS);
 
-        LOG_WARNING(Service_Audio, "(STUBBED) called");
+        NGLOG_WARNING(Service_Audio, "(STUBBED) called");
     }
 
     void QuerySystemEvent(Kernel::HLERequestContext& ctx) {
@@ -105,7 +105,7 @@ private:
         rb.Push(RESULT_SUCCESS);
         rb.PushCopyObjects(system_event);
 
-        LOG_WARNING(Service_Audio, "(STUBBED) called");
+        NGLOG_WARNING(Service_Audio, "(STUBBED) called");
     }
 
     struct AudioRendererStateEntry {
@@ -176,7 +176,7 @@ public:
 
 private:
     void ListAudioDeviceName(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_Audio, "(STUBBED) called");
+        NGLOG_WARNING(Service_Audio, "(STUBBED) called");
         IPC::RequestParser rp{ctx};
 
         const std::string audio_interface = "AudioInterface";
@@ -188,7 +188,7 @@ private:
     }
 
     void SetAudioDeviceOutputVolume(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_Audio, "(STUBBED) called");
+        NGLOG_WARNING(Service_Audio, "(STUBBED) called");
 
         IPC::RequestParser rp{ctx};
         f32 volume = static_cast<f32>(rp.Pop<u32>());
@@ -201,7 +201,7 @@ private:
     }
 
     void GetActiveAudioDeviceName(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_Audio, "(STUBBED) called");
+        NGLOG_WARNING(Service_Audio, "(STUBBED) called");
         IPC::RequestParser rp{ctx};
 
         const std::string audio_interface = "AudioDevice";
@@ -213,7 +213,7 @@ private:
     }
 
     void QueryAudioDeviceSystemEvent(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_Audio, "(STUBBED) called");
+        NGLOG_WARNING(Service_Audio, "(STUBBED) called");
 
         buffer_event->Signal();
 
@@ -223,7 +223,7 @@ private:
     }
 
     void GetActiveChannelCount(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_Audio, "(STUBBED) called");
+        NGLOG_WARNING(Service_Audio, "(STUBBED) called");
         IPC::ResponseBuilder rb{ctx, 3};
         rb.Push(RESULT_SUCCESS);
         rb.Push<u32>(1);
@@ -250,7 +250,7 @@ void AudRenU::OpenAudioRenderer(Kernel::HLERequestContext& ctx) {
     rb.Push(RESULT_SUCCESS);
     rb.PushIpcInterface<Audio::IAudioRenderer>();
 
-    LOG_DEBUG(Service_Audio, "called");
+    NGLOG_DEBUG(Service_Audio, "called");
 }
 
 void AudRenU::GetAudioRendererWorkBufferSize(Kernel::HLERequestContext& ctx) {
@@ -259,7 +259,7 @@ void AudRenU::GetAudioRendererWorkBufferSize(Kernel::HLERequestContext& ctx) {
     rb.Push(RESULT_SUCCESS);
     rb.Push<u64>(0x400);
 
-    LOG_WARNING(Service_Audio, "(STUBBED) called");
+    NGLOG_WARNING(Service_Audio, "(STUBBED) called");
 }
 
 void AudRenU::GetAudioDevice(Kernel::HLERequestContext& ctx) {
@@ -268,7 +268,7 @@ void AudRenU::GetAudioDevice(Kernel::HLERequestContext& ctx) {
     rb.Push(RESULT_SUCCESS);
     rb.PushIpcInterface<Audio::IAudioDevice>();
 
-    LOG_DEBUG(Service_Audio, "called");
+    NGLOG_DEBUG(Service_Audio, "called");
 }
 
 } // namespace Service::Audio

src/core/hle/service/fatal/fatal.cpp

@@ -16,13 +16,13 @@ Module::Interface::Interface(std::shared_ptr<Module> module, const char* name)
 void Module::Interface::FatalSimple(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp(ctx);
     u32 error_code = rp.Pop<u32>();
-    LOG_WARNING(Service_Fatal, "(STUBBED) called, error_code=0x%X", error_code);
+    NGLOG_WARNING(Service_Fatal, "(STUBBED) called, error_code={:#X}", error_code);
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
 }
 
 void Module::Interface::TransitionToFatalError(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_Fatal, "(STUBBED) called");
+    NGLOG_WARNING(Service_Fatal, "(STUBBED) called");
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
 }

src/core/hle/service/filesystem/filesystem.cpp

@@ -25,14 +25,14 @@ ResultCode RegisterFileSystem(std::unique_ptr<FileSys::FileSystemFactory>&& fact
     ASSERT_MSG(inserted, "Tried to register more than one system with same id code");
 
     auto& filesystem = result.first->second;
-    LOG_DEBUG(Service_FS, "Registered file system %s with id code 0x%08X",
-              filesystem->GetName().c_str(), static_cast<u32>(type));
+    NGLOG_DEBUG(Service_FS, "Registered file system {} with id code {:#010X}",
+                filesystem->GetName(), static_cast<u32>(type));
     return RESULT_SUCCESS;
 }
 
 ResultVal<std::unique_ptr<FileSys::FileSystemBackend>> OpenFileSystem(Type type,
                                                                       FileSys::Path& path) {
-    LOG_TRACE(Service_FS, "Opening FileSystem with type=%d", type);
+    NGLOG_TRACE(Service_FS, "Opening FileSystem with type={}", static_cast<u32>(type));
 
     auto itr = filesystem_map.find(type);
     if (itr == filesystem_map.end()) {
@@ -44,7 +44,7 @@ ResultVal<std::unique_ptr<FileSys::FileSystemBackend>> OpenFileSystem(Type type,
 }
 
 ResultCode FormatFileSystem(Type type) {
-    LOG_TRACE(Service_FS, "Formatting FileSystem with type=%d", type);
+    NGLOG_TRACE(Service_FS, "Formatting FileSystem with type={}", static_cast<u32>(type));
 
     auto itr = filesystem_map.find(type);
     if (itr == filesystem_map.end()) {

src/core/hle/service/filesystem/fsp_srv.cpp

@@ -35,7 +35,7 @@ private:
         const s64 offset = rp.Pop<s64>();
         const s64 length = rp.Pop<s64>();
 
-        LOG_DEBUG(Service_FS, "called, offset=0x%ld, length=0x%ld", offset, length);
+        NGLOG_DEBUG(Service_FS, "called, offset={:#X}, length={}", offset, length);
 
         // Error checking
         if (length < 0) {
@@ -87,7 +87,7 @@ private:
         const s64 offset = rp.Pop<s64>();
         const s64 length = rp.Pop<s64>();
 
-        LOG_DEBUG(Service_FS, "called, offset=0x%ld, length=0x%ld", offset, length);
+        NGLOG_DEBUG(Service_FS, "called, offset={:#X}, length={}", offset, length);
 
         // Error checking
         if (length < 0) {
@@ -124,7 +124,7 @@ private:
         const s64 offset = rp.Pop<s64>();
         const s64 length = rp.Pop<s64>();
 
-        LOG_DEBUG(Service_FS, "called, offset=0x%ld, length=0x%ld", offset, length);
+        NGLOG_DEBUG(Service_FS, "called, offset={:#X}, length={}", offset, length);
 
         // Error checking
         if (length < 0) {
@@ -152,7 +152,7 @@ private:
     }
 
     void Flush(Kernel::HLERequestContext& ctx) {
-        LOG_DEBUG(Service_FS, "called");
+        NGLOG_DEBUG(Service_FS, "called");
         backend->Flush();
 
         IPC::ResponseBuilder rb{ctx, 2};
@@ -163,7 +163,7 @@ private:
         IPC::RequestParser rp{ctx};
         const u64 size = rp.Pop<u64>();
         backend->SetSize(size);
-        LOG_DEBUG(Service_FS, "called, size=%" PRIu64, size);
+        NGLOG_DEBUG(Service_FS, "called, size={}", size);
 
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
@@ -171,7 +171,7 @@ private:
 
     void GetSize(Kernel::HLERequestContext& ctx) {
         const u64 size = backend->GetSize();
-        LOG_DEBUG(Service_FS, "called, size=%" PRIu64, size);
+        NGLOG_DEBUG(Service_FS, "called, size={}", size);
 
         IPC::ResponseBuilder rb{ctx, 4};
         rb.Push(RESULT_SUCCESS);
@@ -197,7 +197,7 @@ private:
         IPC::RequestParser rp{ctx};
         const u64 unk = rp.Pop<u64>();
 
-        LOG_DEBUG(Service_FS, "called, unk=0x%llx", unk);
+        NGLOG_DEBUG(Service_FS, "called, unk={:#X}", unk);
 
         // Calculate how many entries we can fit in the output buffer
         u64 count_entries = ctx.GetWriteBufferSize() / sizeof(FileSys::Entry);
@@ -219,7 +219,7 @@ private:
     }
 
     void GetEntryCount(Kernel::HLERequestContext& ctx) {
-        LOG_DEBUG(Service_FS, "called");
+        NGLOG_DEBUG(Service_FS, "called");
 
         u64 count = backend->GetEntryCount();
 
@@ -239,7 +239,7 @@ public:
             {2, &IFileSystem::CreateDirectory, "CreateDirectory"},
             {3, nullptr, "DeleteDirectory"},
             {4, nullptr, "DeleteDirectoryRecursively"},
-            {5, nullptr, "RenameFile"},
+            {5, &IFileSystem::RenameFile, "RenameFile"},
             {6, nullptr, "RenameDirectory"},
             {7, &IFileSystem::GetEntryType, "GetEntryType"},
             {8, &IFileSystem::OpenFile, "OpenFile"},
@@ -265,8 +265,7 @@ public:
         u64 mode = rp.Pop<u64>();
         u32 size = rp.Pop<u32>();
 
-        LOG_DEBUG(Service_FS, "called file %s mode 0x%" PRIX64 " size 0x%08X", name.c_str(), mode,
-                  size);
+        NGLOG_DEBUG(Service_FS, "called file {} mode {:#X} size {:#010X}", name, mode, size);
 
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(backend->CreateFile(name, size));
@@ -280,7 +279,7 @@ public:
 
         std::string name(file_buffer.begin(), end);
 
-        LOG_DEBUG(Service_FS, "called file %s", name.c_str());
+        NGLOG_DEBUG(Service_FS, "called file {}", name);
 
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(backend->DeleteFile(name));
@@ -294,12 +293,32 @@ public:
 
         std::string name(file_buffer.begin(), end);
 
-        LOG_DEBUG(Service_FS, "called directory %s", name.c_str());
+        NGLOG_DEBUG(Service_FS, "called directory {}", name);
 
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(backend->CreateDirectory(name));
     }
 
+    void RenameFile(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+
+        std::vector<u8> buffer;
+        buffer.resize(ctx.BufferDescriptorX()[0].Size());
+        Memory::ReadBlock(ctx.BufferDescriptorX()[0].Address(), buffer.data(), buffer.size());
+        auto end = std::find(buffer.begin(), buffer.end(), '\0');
+        std::string src_name(buffer.begin(), end);
+
+        buffer.resize(ctx.BufferDescriptorX()[1].Size());
+        Memory::ReadBlock(ctx.BufferDescriptorX()[1].Address(), buffer.data(), buffer.size());
+        end = std::find(buffer.begin(), buffer.end(), '\0');
+        std::string dst_name(buffer.begin(), end);
+
+        NGLOG_DEBUG(Service_FS, "called file '{}' to file '{}'", src_name, dst_name);
+
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(backend->RenameFile(src_name, dst_name));
+    }
+
     void OpenFile(Kernel::HLERequestContext& ctx) {
         IPC::RequestParser rp{ctx};
 
@@ -310,7 +329,7 @@ public:
 
         auto mode = static_cast<FileSys::Mode>(rp.Pop<u32>());
 
-        LOG_DEBUG(Service_FS, "called file %s mode %u", name.c_str(), static_cast<u32>(mode));
+        NGLOG_DEBUG(Service_FS, "called file {} mode {}", name, static_cast<u32>(mode));
 
         auto result = backend->OpenFile(name, mode);
         if (result.Failed()) {
@@ -337,7 +356,7 @@ public:
         // TODO(Subv): Implement this filter.
         u32 filter_flags = rp.Pop<u32>();
 
-        LOG_DEBUG(Service_FS, "called directory %s filter %u", name.c_str(), filter_flags);
+        NGLOG_DEBUG(Service_FS, "called directory {} filter {}", name, filter_flags);
 
         auto result = backend->OpenDirectory(name);
         if (result.Failed()) {
@@ -361,7 +380,7 @@ public:
 
         std::string name(file_buffer.begin(), end);
 
-        LOG_DEBUG(Service_FS, "called file %s", name.c_str());
+        NGLOG_DEBUG(Service_FS, "called file {}", name);
 
         auto result = backend->GetEntryType(name);
         if (result.Failed()) {
@@ -376,7 +395,7 @@ public:
     }
 
     void Commit(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_FS, "(STUBBED) called");
+        NGLOG_WARNING(Service_FS, "(STUBBED) called");
 
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
@@ -492,14 +511,14 @@ void FSP_SRV::TryLoadRomFS() {
 }
 
 void FSP_SRV::Initialize(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_FS, "(STUBBED) called");
+    NGLOG_WARNING(Service_FS, "(STUBBED) called");
 
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
 }
 
 void FSP_SRV::MountSdCard(Kernel::HLERequestContext& ctx) {
-    LOG_DEBUG(Service_FS, "called");
+    NGLOG_DEBUG(Service_FS, "called");
 
     FileSys::Path unused;
     auto filesystem = OpenFileSystem(Type::SDMC, unused).Unwrap();
@@ -516,14 +535,14 @@ void FSP_SRV::CreateSaveData(Kernel::HLERequestContext& ctx) {
     auto save_create_struct = rp.PopRaw<std::array<u8, 0x40>>();
     u128 uid = rp.PopRaw<u128>();
 
-    LOG_WARNING(Service_FS, "(STUBBED) called uid = %016" PRIX64 "%016" PRIX64, uid[1], uid[0]);
+    NGLOG_WARNING(Service_FS, "(STUBBED) called uid = {:016X}{:016X}", uid[1], uid[0]);
 
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
 }
 
 void FSP_SRV::MountSaveData(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_FS, "(STUBBED) called");
+    NGLOG_WARNING(Service_FS, "(STUBBED) called");
 
     FileSys::Path unused;
     auto filesystem = OpenFileSystem(Type::SaveData, unused).Unwrap();
@@ -534,7 +553,7 @@ void FSP_SRV::MountSaveData(Kernel::HLERequestContext& ctx) {
 }
 
 void FSP_SRV::GetGlobalAccessLogMode(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_FS, "(STUBBED) called");
+    NGLOG_WARNING(Service_FS, "(STUBBED) called");
 
     IPC::ResponseBuilder rb{ctx, 3};
     rb.Push(RESULT_SUCCESS);
@@ -542,12 +561,12 @@ void FSP_SRV::GetGlobalAccessLogMode(Kernel::HLERequestContext& ctx) {
 }
 
 void FSP_SRV::OpenDataStorageByCurrentProcess(Kernel::HLERequestContext& ctx) {
-    LOG_DEBUG(Service_FS, "called");
+    NGLOG_DEBUG(Service_FS, "called");
 
     TryLoadRomFS();
     if (!romfs) {
         // TODO (bunnei): Find the right error code to use here
-        LOG_CRITICAL(Service_FS, "no file system interface available!");
+        NGLOG_CRITICAL(Service_FS, "no file system interface available!");
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(ResultCode(-1));
         return;
@@ -556,7 +575,7 @@ void FSP_SRV::OpenDataStorageByCurrentProcess(Kernel::HLERequestContext& ctx) {
     // Attempt to open a StorageBackend interface to the RomFS
     auto storage = romfs->OpenFile({}, {});
     if (storage.Failed()) {
-        LOG_CRITICAL(Service_FS, "no storage interface available!");
+        NGLOG_CRITICAL(Service_FS, "no storage interface available!");
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(storage.Code());
         return;
@@ -568,7 +587,7 @@ void FSP_SRV::OpenDataStorageByCurrentProcess(Kernel::HLERequestContext& ctx) {
 }
 
 void FSP_SRV::OpenRomStorage(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_FS, "(STUBBED) called, using OpenDataStorageByCurrentProcess");
+    NGLOG_WARNING(Service_FS, "(STUBBED) called, using OpenDataStorageByCurrentProcess");
     OpenDataStorageByCurrentProcess(ctx);
 }
 

src/core/hle/service/friend/friend.cpp

@@ -13,7 +13,7 @@ namespace Service::Friend {
 void Module::Interface::CreateFriendService(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
-    LOG_WARNING(Service_Friend, "(STUBBED) called");
+    NGLOG_WARNING(Service_Friend, "(STUBBED) called");
 }
 
 Module::Interface::Interface(std::shared_ptr<Module> module, const char* name)

src/core/hle/service/hid/hid.cpp

@@ -53,7 +53,7 @@ private:
         IPC::ResponseBuilder rb{ctx, 2, 1};
         rb.Push(RESULT_SUCCESS);
         rb.PushCopyObjects(shared_mem);
-        LOG_DEBUG(Service_HID, "called");
+        NGLOG_DEBUG(Service_HID, "called");
     }
 
     void LoadInputDevices() {
@@ -184,7 +184,7 @@ private:
     void ActivateVibrationDevice(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        NGLOG_WARNING(Service_HID, "(STUBBED) called");
     }
 };
 
@@ -286,144 +286,144 @@ private:
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
         rb.PushIpcInterface<IAppletResource>(applet_resource);
-        LOG_DEBUG(Service_HID, "called");
+        NGLOG_DEBUG(Service_HID, "called");
     }
 
     void ActivateDebugPad(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        NGLOG_WARNING(Service_HID, "(STUBBED) called");
     }
 
     void ActivateTouchScreen(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        NGLOG_WARNING(Service_HID, "(STUBBED) called");
     }
 
     void ActivateMouse(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        NGLOG_WARNING(Service_HID, "(STUBBED) called");
     }
 
     void ActivateKeyboard(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        NGLOG_WARNING(Service_HID, "(STUBBED) called");
     }
 
     void StartSixAxisSensor(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        NGLOG_WARNING(Service_HID, "(STUBBED) called");
     }
 
     void SetGyroscopeZeroDriftMode(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        NGLOG_WARNING(Service_HID, "(STUBBED) called");
     }
 
     void SetSupportedNpadStyleSet(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        NGLOG_WARNING(Service_HID, "(STUBBED) called");
     }
 
     void GetSupportedNpadStyleSet(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 3};
         rb.Push(RESULT_SUCCESS);
         rb.Push<u32>(0);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        NGLOG_WARNING(Service_HID, "(STUBBED) called");
     }
 
     void SetSupportedNpadIdType(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        NGLOG_WARNING(Service_HID, "(STUBBED) called");
     }
 
     void ActivateNpad(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        NGLOG_WARNING(Service_HID, "(STUBBED) called");
     }
 
     void AcquireNpadStyleSetUpdateEventHandle(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 1};
         rb.Push(RESULT_SUCCESS);
         rb.PushCopyObjects(event);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        NGLOG_WARNING(Service_HID, "(STUBBED) called");
     }
 
     void GetPlayerLedPattern(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        NGLOG_WARNING(Service_HID, "(STUBBED) called");
     }
 
     void SetNpadJoyHoldType(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        NGLOG_WARNING(Service_HID, "(STUBBED) called");
     }
 
     void GetNpadJoyHoldType(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 3};
         rb.Push(RESULT_SUCCESS);
         rb.Push(joy_hold_type);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        NGLOG_WARNING(Service_HID, "(STUBBED) called");
     }
 
     void SetNpadJoyAssignmentModeSingleByDefault(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        NGLOG_WARNING(Service_HID, "(STUBBED) called");
     }
 
     void SendVibrationValue(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        NGLOG_WARNING(Service_HID, "(STUBBED) called");
     }
 
     void GetActualVibrationValue(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        NGLOG_WARNING(Service_HID, "(STUBBED) called");
     }
 
     void SetNpadJoyAssignmentModeDual(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        NGLOG_WARNING(Service_HID, "(STUBBED) called");
     }
 
     void SetNpadHandheldActivationMode(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        NGLOG_WARNING(Service_HID, "(STUBBED) called");
     }
 
     void GetVibrationDeviceInfo(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 4};
         rb.Push(RESULT_SUCCESS);
         rb.Push<u64>(0);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        NGLOG_WARNING(Service_HID, "(STUBBED) called");
     }
 
     void CreateActiveVibrationDeviceList(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
         rb.PushIpcInterface<IActiveVibrationDeviceList>();
-        LOG_DEBUG(Service_HID, "called");
+        NGLOG_DEBUG(Service_HID, "called");
     }
 
     void SendVibrationValues(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        NGLOG_WARNING(Service_HID, "(STUBBED) called");
     }
 };
 

src/core/hle/service/lm/lm.cpp

@@ -141,19 +141,19 @@ private:
         if (header.IsTailLog()) {
             switch (header.severity) {
             case MessageHeader::Severity::Trace:
-                LOG_TRACE(Debug_Emulated, "%s", log_stream.str().c_str());
+                NGLOG_TRACE(Debug_Emulated, "{}", log_stream.str());
                 break;
             case MessageHeader::Severity::Info:
-                LOG_INFO(Debug_Emulated, "%s", log_stream.str().c_str());
+                NGLOG_INFO(Debug_Emulated, "{}", log_stream.str());
                 break;
             case MessageHeader::Severity::Warning:
-                LOG_WARNING(Debug_Emulated, "%s", log_stream.str().c_str());
+                NGLOG_WARNING(Debug_Emulated, "{}", log_stream.str());
                 break;
             case MessageHeader::Severity::Error:
-                LOG_ERROR(Debug_Emulated, "%s", log_stream.str().c_str());
+                NGLOG_ERROR(Debug_Emulated, "{}", log_stream.str());
                 break;
             case MessageHeader::Severity::Critical:
-                LOG_CRITICAL(Debug_Emulated, "%s", log_stream.str().c_str());
+                NGLOG_CRITICAL(Debug_Emulated, "{}", log_stream.str());
                 break;
             }
         }
@@ -178,7 +178,7 @@ void LM::Initialize(Kernel::HLERequestContext& ctx) {
     rb.Push(RESULT_SUCCESS);
     rb.PushIpcInterface<Logger>();
 
-    LOG_DEBUG(Service_LM, "called");
+    NGLOG_DEBUG(Service_LM, "called");
 }
 
 LM::LM() : ServiceFramework("lm") {

src/core/hle/service/nfp/nfp.cpp

@@ -13,7 +13,7 @@ Module::Interface::Interface(std::shared_ptr<Module> module, const char* name)
     : ServiceFramework(name), module(std::move(module)) {}
 
 void Module::Interface::Unknown(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_NFP, "(STUBBED) called");
+    NGLOG_WARNING(Service_NFP, "(STUBBED) called");
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
 }

src/core/hle/service/nifm/nifm.cpp

@@ -62,24 +62,24 @@ public:
 
 private:
     void GetRequestState(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_NIFM, "(STUBBED) called");
+        NGLOG_WARNING(Service_NIFM, "(STUBBED) called");
         IPC::ResponseBuilder rb{ctx, 3};
         rb.Push(RESULT_SUCCESS);
         rb.Push<u32>(0);
     }
     void GetResult(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_NIFM, "(STUBBED) called");
+        NGLOG_WARNING(Service_NIFM, "(STUBBED) called");
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
     }
     void GetSystemEventReadableHandles(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_NIFM, "(STUBBED) called");
+        NGLOG_WARNING(Service_NIFM, "(STUBBED) called");
         IPC::ResponseBuilder rb{ctx, 2, 2};
         rb.Push(RESULT_SUCCESS);
         rb.PushCopyObjects(event1, event2);
     }
     void Cancel(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_NIFM, "(STUBBED) called");
+        NGLOG_WARNING(Service_NIFM, "(STUBBED) called");
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
     }
@@ -105,7 +105,7 @@ public:
 
 private:
     void GetClientId(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_NIFM, "(STUBBED) called");
+        NGLOG_WARNING(Service_NIFM, "(STUBBED) called");
         IPC::ResponseBuilder rb{ctx, 4};
         rb.Push(RESULT_SUCCESS);
         rb.Push<u64>(0);
@@ -116,7 +116,7 @@ private:
         rb.Push(RESULT_SUCCESS);
         rb.PushIpcInterface<IScanRequest>();
 
-        LOG_DEBUG(Service_NIFM, "called");
+        NGLOG_DEBUG(Service_NIFM, "called");
     }
     void CreateRequest(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
@@ -124,10 +124,10 @@ private:
         rb.Push(RESULT_SUCCESS);
         rb.PushIpcInterface<IRequest>();
 
-        LOG_DEBUG(Service_NIFM, "called");
+        NGLOG_DEBUG(Service_NIFM, "called");
     }
     void RemoveNetworkProfile(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_NIFM, "(STUBBED) called");
+        NGLOG_WARNING(Service_NIFM, "(STUBBED) called");
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
     }
@@ -137,7 +137,7 @@ private:
         rb.Push(RESULT_SUCCESS);
         rb.PushIpcInterface<INetworkProfile>();
 
-        LOG_DEBUG(Service_NIFM, "called");
+        NGLOG_DEBUG(Service_NIFM, "called");
     }
 };
 
@@ -187,14 +187,14 @@ void Module::Interface::CreateGeneralServiceOld(Kernel::HLERequestContext& ctx)
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
     rb.PushIpcInterface<IGeneralService>();
-    LOG_DEBUG(Service_NIFM, "called");
+    NGLOG_DEBUG(Service_NIFM, "called");
 }
 
 void Module::Interface::CreateGeneralService(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
     rb.PushIpcInterface<IGeneralService>();
-    LOG_DEBUG(Service_NIFM, "called");
+    NGLOG_DEBUG(Service_NIFM, "called");
 }
 
 Module::Interface::Interface(std::shared_ptr<Module> module, const char* name)

src/core/hle/service/ns/pl_u.cpp

@@ -52,7 +52,7 @@ PL_U::PL_U() : ServiceFramework("pl:u") {
         ASSERT(file.GetSize() == SHARED_FONT_MEM_SIZE);
         file.ReadBytes(shared_font->data(), shared_font->size());
     } else {
-        LOG_WARNING(Service_NS, "Unable to load shared font: %s", filepath.c_str());
+        NGLOG_WARNING(Service_NS, "Unable to load shared font: {}", filepath);
     }
 }
 
@@ -60,7 +60,7 @@ void PL_U::RequestLoad(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
     const u32 shared_font_type{rp.Pop<u32>()};
 
-    LOG_DEBUG(Service_NS, "called, shared_font_type=%d", shared_font_type);
+    NGLOG_DEBUG(Service_NS, "called, shared_font_type={}", shared_font_type);
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
 }
@@ -69,7 +69,7 @@ void PL_U::GetLoadState(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
     const u32 font_id{rp.Pop<u32>()};
 
-    LOG_DEBUG(Service_NS, "called, font_id=%d", font_id);
+    NGLOG_DEBUG(Service_NS, "called, font_id={}", font_id);
     IPC::ResponseBuilder rb{ctx, 3};
     rb.Push(RESULT_SUCCESS);
     rb.Push<u32>(static_cast<u32>(LoadState::Done));
@@ -79,7 +79,7 @@ void PL_U::GetSize(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
     const u32 font_id{rp.Pop<u32>()};
 
-    LOG_DEBUG(Service_NS, "called, font_id=%d", font_id);
+    NGLOG_DEBUG(Service_NS, "called, font_id={}", font_id);
     IPC::ResponseBuilder rb{ctx, 3};
     rb.Push(RESULT_SUCCESS);
     rb.Push<u32>(SHARED_FONT_REGIONS[font_id].size);
@@ -89,7 +89,7 @@ void PL_U::GetSharedMemoryAddressOffset(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
     const u32 font_id{rp.Pop<u32>()};
 
-    LOG_DEBUG(Service_NS, "called, font_id=%d", font_id);
+    NGLOG_DEBUG(Service_NS, "called, font_id={}", font_id);
     IPC::ResponseBuilder rb{ctx, 3};
     rb.Push(RESULT_SUCCESS);
     rb.Push<u32>(SHARED_FONT_REGIONS[font_id].offset);
@@ -110,7 +110,7 @@ void PL_U::GetSharedMemoryNativeHandle(Kernel::HLERequestContext& ctx) {
         Kernel::MemoryPermission::Read, SHARED_FONT_MEM_VADDR, Kernel::MemoryRegion::BASE,
         "PL_U:shared_font_mem");
 
-    LOG_DEBUG(Service_NS, "called");
+    NGLOG_DEBUG(Service_NS, "called");
     IPC::ResponseBuilder rb{ctx, 2, 1};
     rb.Push(RESULT_SUCCESS);
     rb.PushCopyObjects(shared_font_mem);

src/core/hle/service/nvdrv/devices/nvdisp_disp0.cpp

@@ -13,16 +13,16 @@
 namespace Service::Nvidia::Devices {
 
 u32 nvdisp_disp0::ioctl(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) {
-    UNIMPLEMENTED();
+    UNIMPLEMENTED_MSG("Unimplemented ioctl");
     return 0;
 }
 
 void nvdisp_disp0::flip(u32 buffer_handle, u32 offset, u32 format, u32 width, u32 height,
                         u32 stride, NVFlinger::BufferQueue::BufferTransformFlags transform) {
     VAddr addr = nvmap_dev->GetObjectAddress(buffer_handle);
-    LOG_WARNING(Service,
-                "Drawing from address %lx offset %08X Width %u Height %u Stride %u Format %u", addr,
-                offset, width, height, stride, format);
+    NGLOG_WARNING(Service,
+                  "Drawing from address {:X} offset {:08X} Width {} Height {} Stride {} Format {}",
+                  addr, offset, width, height, stride, format);
 
     using PixelFormat = Tegra::FramebufferConfig::PixelFormat;
     const Tegra::FramebufferConfig framebuffer{

src/core/hle/service/nvdrv/devices/nvhost_as_gpu.cpp

@@ -12,8 +12,8 @@
 namespace Service::Nvidia::Devices {
 
 u32 nvhost_as_gpu::ioctl(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) {
-    LOG_DEBUG(Service_NVDRV, "called, command=0x%08x, input_size=0x%zx, output_size=0x%zx",
-              command.raw, input.size(), output.size());
+    NGLOG_DEBUG(Service_NVDRV, "called, command={:#010X}, input_size={:#X}, output_size={:#X}",
+                command.raw, input.size(), output.size());
 
     switch (static_cast<IoctlCommand>(command.raw)) {
     case IoctlCommand::IocInitalizeExCommand:
@@ -27,13 +27,18 @@ u32 nvhost_as_gpu::ioctl(Ioctl command, const std::vector<u8>& input, std::vecto
     case IoctlCommand::IocGetVaRegionsCommand:
         return GetVARegions(input, output);
     }
+
+    if (static_cast<IoctlCommand>(command.cmd.Value()) == IoctlCommand::IocRemapCommand)
+        return Remap(input, output);
+
+    UNIMPLEMENTED_MSG("Unimplemented ioctl command");
     return 0;
 }
 
 u32 nvhost_as_gpu::InitalizeEx(const std::vector<u8>& input, std::vector<u8>& output) {
     IoctlInitalizeEx params{};
     std::memcpy(&params, input.data(), input.size());
-    LOG_WARNING(Service_NVDRV, "(STUBBED) called, big_page_size=0x%x", params.big_page_size);
+    NGLOG_WARNING(Service_NVDRV, "(STUBBED) called, big_page_size={:#X}", params.big_page_size);
     std::memcpy(output.data(), &params, output.size());
     return 0;
 }
@@ -41,8 +46,8 @@ u32 nvhost_as_gpu::InitalizeEx(const std::vector<u8>& input, std::vector<u8>& ou
 u32 nvhost_as_gpu::AllocateSpace(const std::vector<u8>& input, std::vector<u8>& output) {
     IoctlAllocSpace params{};
     std::memcpy(&params, input.data(), input.size());
-    LOG_DEBUG(Service_NVDRV, "called, pages=%x, page_size=%x, flags=%x", params.pages,
-              params.page_size, params.flags);
+    NGLOG_DEBUG(Service_NVDRV, "called, pages={:X}, page_size={:X}, flags={:X}", params.pages,
+                params.page_size, params.flags);
 
     auto& gpu = Core::System::GetInstance().GPU();
     const u64 size{static_cast<u64>(params.pages) * static_cast<u64>(params.page_size)};
@@ -56,15 +61,45 @@ u32 nvhost_as_gpu::AllocateSpace(const std::vector<u8>& input, std::vector<u8>&
     return 0;
 }
 
+u32 nvhost_as_gpu::Remap(const std::vector<u8>& input, std::vector<u8>& output) {
+    size_t num_entries = input.size() / sizeof(IoctlRemapEntry);
+
+    NGLOG_WARNING(Service_NVDRV, "(STUBBED) called, num_entries=0x{:X}", num_entries);
+
+    std::vector<IoctlRemapEntry> entries(num_entries);
+    std::memcpy(entries.data(), input.data(), input.size());
+
+    auto& gpu = Core::System::GetInstance().GPU();
+
+    for (const auto& entry : entries) {
+        NGLOG_WARNING(Service_NVDRV, "remap entry, offset=0x{:X} handle=0x{:X} pages=0x{:X}",
+                      entry.offset, entry.nvmap_handle, entry.pages);
+        Tegra::GPUVAddr offset = static_cast<Tegra::GPUVAddr>(entry.offset) << 0x10;
+
+        auto object = nvmap_dev->GetObject(entry.nvmap_handle);
+        ASSERT(object);
+
+        ASSERT(object->status == nvmap::Object::Status::Allocated);
+
+        u64 size = static_cast<u64>(entry.pages) << 0x10;
+        ASSERT(size <= object->size);
+
+        Tegra::GPUVAddr returned = gpu.memory_manager->MapBufferEx(object->addr, offset, size);
+        ASSERT(returned == offset);
+    }
+    std::memcpy(output.data(), entries.data(), output.size());
+    return 0;
+}
+
 u32 nvhost_as_gpu::MapBufferEx(const std::vector<u8>& input, std::vector<u8>& output) {
     IoctlMapBufferEx params{};
     std::memcpy(&params, input.data(), input.size());
 
-    LOG_DEBUG(Service_NVDRV,
-              "called, flags=%x, nvmap_handle=%x, buffer_offset=%" PRIu64 ", mapping_size=%" PRIu64
-              ", offset=%" PRIu64,
-              params.flags, params.nvmap_handle, params.buffer_offset, params.mapping_size,
-              params.offset);
+    NGLOG_DEBUG(Service_NVDRV,
+                "called, flags={:X}, nvmap_handle={:X}, buffer_offset={}, mapping_size={}"
+                ", offset={}",
+                params.flags, params.nvmap_handle, params.buffer_offset, params.mapping_size,
+                params.offset);
 
     if (!params.nvmap_handle) {
         return 0;
@@ -73,6 +108,16 @@ u32 nvhost_as_gpu::MapBufferEx(const std::vector<u8>& input, std::vector<u8>& ou
     auto object = nvmap_dev->GetObject(params.nvmap_handle);
     ASSERT(object);
 
+    // We can only map objects that have already been assigned a CPU address.
+    ASSERT(object->status == nvmap::Object::Status::Allocated);
+
+    ASSERT(params.buffer_offset == 0);
+
+    // The real nvservices doesn't make a distinction between handles and ids, and
+    // object can only have one handle and it will be the same as its id. Assert that this is the
+    // case to prevent unexpected behavior.
+    ASSERT(object->id == params.nvmap_handle);
+
     auto& gpu = Core::System::GetInstance().GPU();
 
     if (params.flags & 1) {
@@ -88,7 +133,7 @@ u32 nvhost_as_gpu::MapBufferEx(const std::vector<u8>& input, std::vector<u8>& ou
 u32 nvhost_as_gpu::BindChannel(const std::vector<u8>& input, std::vector<u8>& output) {
     IoctlBindChannel params{};
     std::memcpy(&params, input.data(), input.size());
-    LOG_DEBUG(Service_NVDRV, "called, fd=%x", params.fd);
+    NGLOG_DEBUG(Service_NVDRV, "called, fd={:X}", params.fd);
     channel = params.fd;
     std::memcpy(output.data(), &params, output.size());
     return 0;
@@ -97,8 +142,8 @@ u32 nvhost_as_gpu::BindChannel(const std::vector<u8>& input, std::vector<u8>& ou
 u32 nvhost_as_gpu::GetVARegions(const std::vector<u8>& input, std::vector<u8>& output) {
     IoctlGetVaRegions params{};
     std::memcpy(&params, input.data(), input.size());
-    LOG_WARNING(Service_NVDRV, "(STUBBED) called, buf_addr=%" PRIu64 ", buf_size=%x",
-                params.buf_addr, params.buf_size);
+    NGLOG_WARNING(Service_NVDRV, "(STUBBED) called, buf_addr={:X}, buf_size={:X}", params.buf_addr,
+                  params.buf_size);
 
     params.buf_size = 0x30;
     params.regions[0].offset = 0x04000000;

src/core/hle/service/nvdrv/devices/nvhost_as_gpu.h

@@ -26,6 +26,7 @@ private:
     enum class IoctlCommand : u32_le {
         IocInitalizeExCommand = 0x40284109,
         IocAllocateSpaceCommand = 0xC0184102,
+        IocRemapCommand = 0x00000014,
         IocMapBufferExCommand = 0xC0284106,
         IocBindChannelCommand = 0x40044101,
         IocGetVaRegionsCommand = 0xC0404108,
@@ -54,6 +55,16 @@ private:
     };
     static_assert(sizeof(IoctlAllocSpace) == 24, "IoctlInitalizeEx is incorrect size");
 
+    struct IoctlRemapEntry {
+        u16_le flags;
+        u16_le kind;
+        u32_le nvmap_handle;
+        INSERT_PADDING_WORDS(1);
+        u32_le offset;
+        u32_le pages;
+    };
+    static_assert(sizeof(IoctlRemapEntry) == 20, "IoctlRemapEntry is incorrect size");
+
     struct IoctlMapBufferEx {
         u32_le flags; // bit0: fixed_offset, bit2: cacheable
         u32_le kind;  // -1 is default
@@ -91,6 +102,7 @@ private:
 
     u32 InitalizeEx(const std::vector<u8>& input, std::vector<u8>& output);
     u32 AllocateSpace(const std::vector<u8>& input, std::vector<u8>& output);
+    u32 Remap(const std::vector<u8>& input, std::vector<u8>& output);
     u32 MapBufferEx(const std::vector<u8>& input, std::vector<u8>& output);
     u32 BindChannel(const std::vector<u8>& input, std::vector<u8>& output);
     u32 GetVARegions(const std::vector<u8>& input, std::vector<u8>& output);

src/core/hle/service/nvdrv/devices/nvhost_ctrl.cpp

@@ -9,8 +9,8 @@
 namespace Service::Nvidia::Devices {
 
 u32 nvhost_ctrl::ioctl(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) {
-    LOG_DEBUG(Service_NVDRV, "called, command=0x%08x, input_size=0x%zx, output_size=0x%zx",
-              command.raw, input.size(), output.size());
+    NGLOG_DEBUG(Service_NVDRV, "called, command={:#010X}, input_size={:#X}, output_size={:#X}",
+                command.raw, input.size(), output.size());
 
     switch (static_cast<IoctlCommand>(command.raw)) {
     case IoctlCommand::IocGetConfigCommand:
@@ -18,15 +18,15 @@ u32 nvhost_ctrl::ioctl(Ioctl command, const std::vector<u8>& input, std::vector<
     case IoctlCommand::IocCtrlEventWaitCommand:
         return IocCtrlEventWait(input, output);
     }
-    UNIMPLEMENTED();
+    UNIMPLEMENTED_MSG("Unimplemented ioctl");
     return 0;
 }
 
 u32 nvhost_ctrl::NvOsGetConfigU32(const std::vector<u8>& input, std::vector<u8>& output) {
     IocGetConfigParams params{};
     std::memcpy(&params, input.data(), sizeof(params));
-    LOG_DEBUG(Service_NVDRV, "called, setting=%s!%s", params.domain_str.data(),
-              params.param_str.data());
+    NGLOG_DEBUG(Service_NVDRV, "called, setting={}!{}", params.domain_str.data(),
+                params.param_str.data());
 
     if (!strcmp(params.domain_str.data(), "nv")) {
         if (!strcmp(params.param_str.data(), "NV_MEMORY_PROFILER")) {
@@ -48,8 +48,8 @@ u32 nvhost_ctrl::NvOsGetConfigU32(const std::vector<u8>& input, std::vector<u8>&
 u32 nvhost_ctrl::IocCtrlEventWait(const std::vector<u8>& input, std::vector<u8>& output) {
     IocCtrlEventWaitParams params{};
     std::memcpy(&params, input.data(), sizeof(params));
-    LOG_WARNING(Service_NVDRV, "(STUBBED) called, syncpt_id=%u threshold=%u timeout=%d",
-                params.syncpt_id, params.threshold, params.timeout);
+    NGLOG_WARNING(Service_NVDRV, "(STUBBED) called, syncpt_id={} threshold={} timeout={}",
+                  params.syncpt_id, params.threshold, params.timeout);
 
     // TODO(Subv): Implement actual syncpt waiting.
     params.value = 0;

src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.cpp

@@ -10,8 +10,8 @@
 namespace Service::Nvidia::Devices {
 
 u32 nvhost_ctrl_gpu::ioctl(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) {
-    LOG_DEBUG(Service_NVDRV, "called, command=0x%08x, input_size=0x%zx, output_size=0x%zx",
-              command.raw, input.size(), output.size());
+    NGLOG_DEBUG(Service_NVDRV, "called, command={:#010X}, input_size={:#X}, output_size={:#X}",
+                command.raw, input.size(), output.size());
 
     switch (static_cast<IoctlCommand>(command.raw)) {
     case IoctlCommand::IocGetCharacteristicsCommand:
@@ -25,12 +25,12 @@ u32 nvhost_ctrl_gpu::ioctl(Ioctl command, const std::vector<u8>& input, std::vec
     case IoctlCommand::IocZcullGetInfo:
         return ZCullGetInfo(input, output);
     }
-    UNIMPLEMENTED();
+    UNIMPLEMENTED_MSG("Unimplemented ioctl");
     return 0;
 }
 
 u32 nvhost_ctrl_gpu::GetCharacteristics(const std::vector<u8>& input, std::vector<u8>& output) {
-    LOG_DEBUG(Service_NVDRV, "called");
+    NGLOG_DEBUG(Service_NVDRV, "called");
     IoctlCharacteristics params{};
     std::memcpy(&params, input.data(), input.size());
     params.gc.arch = 0x120;
@@ -77,14 +77,14 @@ u32 nvhost_ctrl_gpu::GetCharacteristics(const std::vector<u8>& input, std::vecto
 u32 nvhost_ctrl_gpu::GetTPCMasks(const std::vector<u8>& input, std::vector<u8>& output) {
     IoctlGpuGetTpcMasksArgs params{};
     std::memcpy(&params, input.data(), input.size());
-    LOG_WARNING(Service_NVDRV, "(STUBBED) called, mask=0x%x, mask_buf_addr=0x%" PRIx64,
-                params.mask_buf_size, params.mask_buf_addr);
+    NGLOG_WARNING(Service_NVDRV, "(STUBBED) called, mask={:#X}, mask_buf_addr={:#X}",
+                  params.mask_buf_size, params.mask_buf_addr);
     std::memcpy(output.data(), &params, sizeof(params));
     return 0;
 }
 
 u32 nvhost_ctrl_gpu::GetActiveSlotMask(const std::vector<u8>& input, std::vector<u8>& output) {
-    LOG_DEBUG(Service_NVDRV, "called");
+    NGLOG_DEBUG(Service_NVDRV, "called");
     IoctlActiveSlotMask params{};
     std::memcpy(&params, input.data(), input.size());
     params.slot = 0x07;
@@ -94,7 +94,7 @@ u32 nvhost_ctrl_gpu::GetActiveSlotMask(const std::vector<u8>& input, std::vector
 }
 
 u32 nvhost_ctrl_gpu::ZCullGetCtxSize(const std::vector<u8>& input, std::vector<u8>& output) {
-    LOG_DEBUG(Service_NVDRV, "called");
+    NGLOG_DEBUG(Service_NVDRV, "called");
     IoctlZcullGetCtxSize params{};
     std::memcpy(&params, input.data(), input.size());
     params.size = 0x1;
@@ -103,7 +103,7 @@ u32 nvhost_ctrl_gpu::ZCullGetCtxSize(const std::vector<u8>& input, std::vector<u
 }
 
 u32 nvhost_ctrl_gpu::ZCullGetInfo(const std::vector<u8>& input, std::vector<u8>& output) {
-    LOG_DEBUG(Service_NVDRV, "called");
+    NGLOG_DEBUG(Service_NVDRV, "called");
     IoctlNvgpuGpuZcullGetInfoArgs params{};
     std::memcpy(&params, input.data(), input.size());
     params.width_align_pixels = 0x20;

src/core/hle/service/nvdrv/devices/nvhost_gpu.cpp

@@ -12,8 +12,8 @@
 namespace Service::Nvidia::Devices {
 
 u32 nvhost_gpu::ioctl(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) {
-    LOG_DEBUG(Service_NVDRV, "called, command=0x%08x, input_size=0x%zx, output_size=0x%zx",
-              command.raw, input.size(), output.size());
+    NGLOG_DEBUG(Service_NVDRV, "called, command={:#010X}, input_size={:#X}, output_size={:#X}",
+                command.raw, input.size(), output.size());
 
     switch (static_cast<IoctlCommand>(command.raw)) {
     case IoctlCommand::IocSetNVMAPfdCommand:
@@ -40,21 +40,21 @@ u32 nvhost_gpu::ioctl(Ioctl command, const std::vector<u8>& input, std::vector<u
         }
     }
 
-    UNIMPLEMENTED();
+    UNIMPLEMENTED_MSG("Unimplemented ioctl");
     return 0;
 };
 
 u32 nvhost_gpu::SetNVMAPfd(const std::vector<u8>& input, std::vector<u8>& output) {
     IoctlSetNvmapFD params{};
     std::memcpy(&params, input.data(), input.size());
-    LOG_DEBUG(Service_NVDRV, "called, fd=%x", params.nvmap_fd);
+    NGLOG_DEBUG(Service_NVDRV, "called, fd={}", params.nvmap_fd);
     nvmap_fd = params.nvmap_fd;
     std::memcpy(output.data(), &params, output.size());
     return 0;
 }
 
 u32 nvhost_gpu::SetClientData(const std::vector<u8>& input, std::vector<u8>& output) {
-    LOG_DEBUG(Service_NVDRV, "called");
+    NGLOG_DEBUG(Service_NVDRV, "called");
     IoctlClientData params{};
     std::memcpy(&params, input.data(), input.size());
     user_data = params.data;
@@ -63,7 +63,7 @@ u32 nvhost_gpu::SetClientData(const std::vector<u8>& input, std::vector<u8>& out
 }
 
 u32 nvhost_gpu::GetClientData(const std::vector<u8>& input, std::vector<u8>& output) {
-    LOG_DEBUG(Service_NVDRV, "called");
+    NGLOG_DEBUG(Service_NVDRV, "called");
     IoctlClientData params{};
     std::memcpy(&params, input.data(), input.size());
     params.data = user_data;
@@ -73,8 +73,8 @@ u32 nvhost_gpu::GetClientData(const std::vector<u8>& input, std::vector<u8>& out
 
 u32 nvhost_gpu::ZCullBind(const std::vector<u8>& input, std::vector<u8>& output) {
     std::memcpy(&zcull_params, input.data(), input.size());
-    LOG_DEBUG(Service_NVDRV, "called, gpu_va=%" PRIx64 ", mode=%x", zcull_params.gpu_va,
-              zcull_params.mode);
+    NGLOG_DEBUG(Service_NVDRV, "called, gpu_va={:X}, mode={:X}", zcull_params.gpu_va,
+                zcull_params.mode);
     std::memcpy(output.data(), &zcull_params, output.size());
     return 0;
 }
@@ -82,15 +82,15 @@ u32 nvhost_gpu::ZCullBind(const std::vector<u8>& input, std::vector<u8>& output)
 u32 nvhost_gpu::SetErrorNotifier(const std::vector<u8>& input, std::vector<u8>& output) {
     IoctlSetErrorNotifier params{};
     std::memcpy(&params, input.data(), input.size());
-    LOG_WARNING(Service_NVDRV, "(STUBBED) called, offset=%" PRIx64 ", size=%" PRIx64 ", mem=%x",
-                params.offset, params.size, params.mem);
+    NGLOG_WARNING(Service_NVDRV, "(STUBBED) called, offset={:X}, size={:X}, mem={:X}",
+                  params.offset, params.size, params.mem);
     std::memcpy(output.data(), &params, output.size());
     return 0;
 }
 
 u32 nvhost_gpu::SetChannelPriority(const std::vector<u8>& input, std::vector<u8>& output) {
     std::memcpy(&channel_priority, input.data(), input.size());
-    LOG_DEBUG(Service_NVDRV, "(STUBBED) called, priority=%x", channel_priority);
+    NGLOG_DEBUG(Service_NVDRV, "(STUBBED) called, priority={:X}", channel_priority);
     std::memcpy(output.data(), &channel_priority, output.size());
     return 0;
 }
@@ -98,10 +98,11 @@ u32 nvhost_gpu::SetChannelPriority(const std::vector<u8>& input, std::vector<u8>
 u32 nvhost_gpu::AllocGPFIFOEx2(const std::vector<u8>& input, std::vector<u8>& output) {
     IoctlAllocGpfifoEx2 params{};
     std::memcpy(&params, input.data(), input.size());
-    LOG_WARNING(Service_NVDRV,
-                "(STUBBED) called, num_entries=%x, flags=%x, unk0=%x, unk1=%x, unk2=%x, unk3=%x",
-                params.num_entries, params.flags, params.unk0, params.unk1, params.unk2,
-                params.unk3);
+    NGLOG_WARNING(Service_NVDRV,
+                  "(STUBBED) called, num_entries={:X}, flags={:X}, unk0={:X}, "
+                  "unk1={:X}, unk2={:X}, unk3={:X}",
+                  params.num_entries, params.flags, params.unk0, params.unk1, params.unk2,
+                  params.unk3);
     params.fence_out.id = 0;
     params.fence_out.value = 0;
     std::memcpy(output.data(), &params, output.size());
@@ -111,8 +112,8 @@ u32 nvhost_gpu::AllocGPFIFOEx2(const std::vector<u8>& input, std::vector<u8>& ou
 u32 nvhost_gpu::AllocateObjectContext(const std::vector<u8>& input, std::vector<u8>& output) {
     IoctlAllocObjCtx params{};
     std::memcpy(&params, input.data(), input.size());
-    LOG_WARNING(Service_NVDRV, "(STUBBED) called, class_num=%x, flags=%x", params.class_num,
-                params.flags);
+    NGLOG_WARNING(Service_NVDRV, "(STUBBED) called, class_num={:X}, flags={:X}", params.class_num,
+                  params.flags);
     params.obj_id = 0x0;
     std::memcpy(output.data(), &params, output.size());
     return 0;
@@ -123,8 +124,8 @@ u32 nvhost_gpu::SubmitGPFIFO(const std::vector<u8>& input, std::vector<u8>& outp
         UNIMPLEMENTED();
     IoctlSubmitGpfifo params{};
     std::memcpy(&params, input.data(), sizeof(IoctlSubmitGpfifo));
-    LOG_WARNING(Service_NVDRV, "(STUBBED) called, gpfifo=%" PRIx64 ", num_entries=%x, flags=%x",
-                params.gpfifo, params.num_entries, params.flags);
+    NGLOG_WARNING(Service_NVDRV, "(STUBBED) called, gpfifo={:X}, num_entries={:X}, flags={:X}",
+                  params.gpfifo, params.num_entries, params.flags);
 
     auto entries = std::vector<IoctlGpfifoEntry>();
     entries.resize(params.num_entries);

src/core/hle/service/nvdrv/devices/nvmap.cpp

@@ -32,7 +32,7 @@ u32 nvmap::ioctl(Ioctl command, const std::vector<u8>& input, std::vector<u8>& o
         return IocParam(input, output);
     }
 
-    UNIMPLEMENTED();
+    UNIMPLEMENTED_MSG("Unimplemented ioctl");
     return 0;
 }
 
@@ -49,7 +49,7 @@ u32 nvmap::IocCreate(const std::vector<u8>& input, std::vector<u8>& output) {
     u32 handle = next_handle++;
     handles[handle] = std::move(object);
 
-    LOG_DEBUG(Service_NVDRV, "size=0x%08X", params.size);
+    NGLOG_DEBUG(Service_NVDRV, "size={:#010X}", params.size);
 
     params.handle = handle;
 
@@ -70,7 +70,7 @@ u32 nvmap::IocAlloc(const std::vector<u8>& input, std::vector<u8>& output) {
     object->addr = params.addr;
     object->status = Object::Status::Allocated;
 
-    LOG_DEBUG(Service_NVDRV, "called, addr=0x%" PRIx64, params.addr);
+    NGLOG_DEBUG(Service_NVDRV, "called, addr={:X}", params.addr);
 
     std::memcpy(output.data(), &params, sizeof(params));
     return 0;
@@ -80,7 +80,7 @@ u32 nvmap::IocGetId(const std::vector<u8>& input, std::vector<u8>& output) {
     IocGetIdParams params;
     std::memcpy(&params, input.data(), sizeof(params));
 
-    LOG_WARNING(Service_NVDRV, "called");
+    NGLOG_WARNING(Service_NVDRV, "called");
 
     auto object = GetObject(params.handle);
     ASSERT(object);
@@ -95,7 +95,7 @@ u32 nvmap::IocFromId(const std::vector<u8>& input, std::vector<u8>& output) {
     IocFromIdParams params;
     std::memcpy(&params, input.data(), sizeof(params));
 
-    LOG_WARNING(Service_NVDRV, "(STUBBED) called");
+    NGLOG_WARNING(Service_NVDRV, "(STUBBED) called");
 
     auto itr = std::find_if(handles.begin(), handles.end(),
                             [&](const auto& entry) { return entry.second->id == params.id; });
@@ -114,7 +114,7 @@ u32 nvmap::IocParam(const std::vector<u8>& input, std::vector<u8>& output) {
     IocParamParams params;
     std::memcpy(&params, input.data(), sizeof(params));
 
-    LOG_WARNING(Service_NVDRV, "(STUBBED) called type=%u", params.type);
+    NGLOG_WARNING(Service_NVDRV, "(STUBBED) called type={}", params.type);
 
     auto object = GetObject(params.handle);
     ASSERT(object);

src/core/hle/service/nvdrv/interface.cpp

@@ -12,7 +12,7 @@
 namespace Service::Nvidia {
 
 void NVDRV::Open(Kernel::HLERequestContext& ctx) {
-    LOG_DEBUG(Service_NVDRV, "called");
+    NGLOG_DEBUG(Service_NVDRV, "called");
 
     const auto& buffer = ctx.ReadBuffer();
     std::string device_name(buffer.begin(), buffer.end());
@@ -25,7 +25,7 @@ void NVDRV::Open(Kernel::HLERequestContext& ctx) {
 }
 
 void NVDRV::Ioctl(Kernel::HLERequestContext& ctx) {
-    LOG_DEBUG(Service_NVDRV, "called");
+    NGLOG_DEBUG(Service_NVDRV, "called");
 
     IPC::RequestParser rp{ctx};
     u32 fd = rp.Pop<u32>();
@@ -41,7 +41,7 @@ void NVDRV::Ioctl(Kernel::HLERequestContext& ctx) {
 }
 
 void NVDRV::Close(Kernel::HLERequestContext& ctx) {
-    LOG_DEBUG(Service_NVDRV, "called");
+    NGLOG_DEBUG(Service_NVDRV, "called");
 
     IPC::RequestParser rp{ctx};
     u32 fd = rp.Pop<u32>();
@@ -53,7 +53,7 @@ void NVDRV::Close(Kernel::HLERequestContext& ctx) {
 }
 
 void NVDRV::Initialize(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_NVDRV, "(STUBBED) called");
+    NGLOG_WARNING(Service_NVDRV, "(STUBBED) called");
     IPC::ResponseBuilder rb{ctx, 3};
     rb.Push(RESULT_SUCCESS);
     rb.Push<u32>(0);
@@ -63,7 +63,7 @@ void NVDRV::QueryEvent(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
     u32 fd = rp.Pop<u32>();
     u32 event_id = rp.Pop<u32>();
-    LOG_WARNING(Service_NVDRV, "(STUBBED) called, fd=%x, event_id=%x", fd, event_id);
+    NGLOG_WARNING(Service_NVDRV, "(STUBBED) called, fd={:X}, event_id={:X}", fd, event_id);
 
     IPC::ResponseBuilder rb{ctx, 3, 1};
     rb.Push(RESULT_SUCCESS);
@@ -75,14 +75,14 @@ void NVDRV::SetClientPID(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
     pid = rp.Pop<u64>();
 
-    LOG_WARNING(Service_NVDRV, "(STUBBED) called, pid=0x%" PRIx64, pid);
+    NGLOG_WARNING(Service_NVDRV, "(STUBBED) called, pid={:#X}", pid);
     IPC::ResponseBuilder rb{ctx, 3};
     rb.Push(RESULT_SUCCESS);
     rb.Push<u32>(0);
 }
 
 void NVDRV::FinishInitialize(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_NVDRV, "(STUBBED) called");
+    NGLOG_WARNING(Service_NVDRV, "(STUBBED) called");
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
 }

src/core/hle/service/nvflinger/buffer_queue.cpp

@@ -9,7 +9,8 @@
 #include "core/core_timing.h"
 #include "core/hle/service/nvflinger/buffer_queue.h"
 
-namespace Service::NVFlinger {
+namespace Service {
+namespace NVFlinger {
 
 BufferQueue::BufferQueue(u32 id, u64 layer_id) : id(id), layer_id(layer_id) {
     native_handle = Kernel::Event::Create(Kernel::ResetType::OneShot, "BufferQueue NativeHandle");
@@ -22,7 +23,7 @@ void BufferQueue::SetPreallocatedBuffer(u32 slot, IGBPBuffer& igbp_buffer) {
     buffer.igbp_buffer = igbp_buffer;
     buffer.status = Buffer::Status::Free;
 
-    LOG_WARNING(Service, "Adding graphics buffer %u", slot);
+    NGLOG_WARNING(Service, "Adding graphics buffer {}", slot);
 
     queue.emplace_back(buffer);
 
@@ -93,7 +94,7 @@ void BufferQueue::ReleaseBuffer(u32 slot) {
 }
 
 u32 BufferQueue::Query(QueryType type) {
-    LOG_WARNING(Service, "(STUBBED) called type=%u", static_cast<u32>(type));
+    NGLOG_WARNING(Service, "(STUBBED) called type={}", static_cast<u32>(type));
     switch (type) {
     case QueryType::NativeWindowFormat:
         // TODO(Subv): Use an enum for this
@@ -110,4 +111,5 @@ void BufferQueue::SetBufferWaitEvent(Kernel::SharedPtr<Kernel::Event>&& wait_eve
     buffer_wait_event = std::move(wait_event);
 }
 
-} // namespace Service::NVFlinger
+} // namespace NVFlinger
+} // namespace Service

src/core/hle/service/nvflinger/buffer_queue.h

@@ -13,7 +13,8 @@ namespace CoreTiming {
 struct EventType;
 }
 
-namespace Service::NVFlinger {
+namespace Service {
+namespace NVFlinger {
 
 struct IGBPBuffer {
     u32_le magic;
@@ -97,4 +98,5 @@ private:
     Kernel::SharedPtr<Kernel::Event> buffer_wait_event;
 };
 
-} // namespace Service::NVFlinger
+} // namespace NVFlinger
+} // namespace Service

src/core/hle/service/nvflinger/nvflinger.cpp

@@ -48,7 +48,7 @@ NVFlinger::~NVFlinger() {
 }
 
 u64 NVFlinger::OpenDisplay(const std::string& name) {
-    LOG_WARNING(Service, "Opening display %s", name.c_str());
+    NGLOG_WARNING(Service, "Opening display {}", name);
 
     // TODO(Subv): Currently we only support the Default display.
     ASSERT(name == "Default");

src/core/hle/service/pctl/pctl_a.cpp

@@ -113,7 +113,7 @@ void PCTL_A::CreateService(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
     rb.PushIpcInterface<IParentalControlService>();
-    LOG_DEBUG(Service_PCTL, "called");
+    NGLOG_DEBUG(Service_PCTL, "called");
 }
 
 PCTL_A::PCTL_A() : ServiceFramework("pctl:a") {

src/core/hle/service/service.cpp

@@ -120,7 +120,7 @@ void ServiceFrameworkBase::ReportUnimplementedFunction(Kernel::HLERequestContext
     }
     buf.push_back('}');
 
-    LOG_ERROR(Service, "unknown / unimplemented %s", fmt::to_string(buf).c_str());
+    NGLOG_ERROR(Service, "unknown / unimplemented {}", fmt::to_string(buf));
     UNIMPLEMENTED();
 }
 
@@ -131,8 +131,8 @@ void ServiceFrameworkBase::InvokeRequest(Kernel::HLERequestContext& ctx) {
         return ReportUnimplementedFunction(ctx, info);
     }
 
-    LOG_TRACE(
-        Service, "%s",
+    NGLOG_TRACE(
+        Service, "{}",
         MakeFunctionString(info->name, GetServiceName().c_str(), ctx.CommandBuffer()).c_str());
     handler_invoker(this, info->handler_callback, ctx);
 }
@@ -145,7 +145,7 @@ ResultCode ServiceFrameworkBase::HandleSyncRequest(Kernel::HLERequestContext& co
         return ResultCode(ErrorModule::HIPC, ErrorDescription::RemoteProcessDead);
     }
     case IPC::CommandType::Control: {
-        SM::g_service_manager->InvokeControlRequest(context);
+        Core::System::GetInstance().ServiceManager().InvokeControlRequest(context);
         break;
     }
     case IPC::CommandType::Request: {
@@ -170,43 +170,41 @@ void AddNamedPort(std::string name, SharedPtr<ClientPort> port) {
 }
 
 /// Initialize ServiceManager
-void Init() {
+void Init(std::shared_ptr<SM::ServiceManager>& sm) {
     // NVFlinger needs to be accessed by several services like Vi and AppletOE so we instantiate it
     // here and pass it into the respective InstallInterfaces functions.
     auto nv_flinger = std::make_shared<NVFlinger::NVFlinger>();
 
-    SM::g_service_manager = std::make_shared<SM::ServiceManager>();
-    SM::ServiceManager::InstallInterfaces(SM::g_service_manager);
+    SM::ServiceManager::InstallInterfaces(sm);
 
-    Account::InstallInterfaces(*SM::g_service_manager);
-    AM::InstallInterfaces(*SM::g_service_manager, nv_flinger);
-    AOC::InstallInterfaces(*SM::g_service_manager);
-    APM::InstallInterfaces(*SM::g_service_manager);
-    Audio::InstallInterfaces(*SM::g_service_manager);
-    Fatal::InstallInterfaces(*SM::g_service_manager);
-    FileSystem::InstallInterfaces(*SM::g_service_manager);
-    Friend::InstallInterfaces(*SM::g_service_manager);
-    HID::InstallInterfaces(*SM::g_service_manager);
-    LM::InstallInterfaces(*SM::g_service_manager);
-    NFP::InstallInterfaces(*SM::g_service_manager);
-    NIFM::InstallInterfaces(*SM::g_service_manager);
-    NS::InstallInterfaces(*SM::g_service_manager);
-    Nvidia::InstallInterfaces(*SM::g_service_manager);
-    PCTL::InstallInterfaces(*SM::g_service_manager);
-    Sockets::InstallInterfaces(*SM::g_service_manager);
-    SPL::InstallInterfaces(*SM::g_service_manager);
-    SSL::InstallInterfaces(*SM::g_service_manager);
-    Time::InstallInterfaces(*SM::g_service_manager);
-    VI::InstallInterfaces(*SM::g_service_manager, nv_flinger);
-    Set::InstallInterfaces(*SM::g_service_manager);
+    Account::InstallInterfaces(*sm);
+    AM::InstallInterfaces(*sm, nv_flinger);
+    AOC::InstallInterfaces(*sm);
+    APM::InstallInterfaces(*sm);
+    Audio::InstallInterfaces(*sm);
+    Fatal::InstallInterfaces(*sm);
+    FileSystem::InstallInterfaces(*sm);
+    Friend::InstallInterfaces(*sm);
+    HID::InstallInterfaces(*sm);
+    LM::InstallInterfaces(*sm);
+    NFP::InstallInterfaces(*sm);
+    NIFM::InstallInterfaces(*sm);
+    NS::InstallInterfaces(*sm);
+    Nvidia::InstallInterfaces(*sm);
+    PCTL::InstallInterfaces(*sm);
+    Sockets::InstallInterfaces(*sm);
+    SPL::InstallInterfaces(*sm);
+    SSL::InstallInterfaces(*sm);
+    Time::InstallInterfaces(*sm);
+    VI::InstallInterfaces(*sm, nv_flinger);
+    Set::InstallInterfaces(*sm);
 
-    LOG_DEBUG(Service, "initialized OK");
+    NGLOG_DEBUG(Service, "initialized OK");
 }
 
 /// Shutdown ServiceManager
 void Shutdown() {
-    SM::g_service_manager = nullptr;
     g_kernel_named_ports.clear();
-    LOG_DEBUG(Service, "shutdown OK");
+    NGLOG_DEBUG(Service, "shutdown OK");
 }
 } // namespace Service

src/core/hle/service/service.h

@@ -178,7 +178,7 @@ private:
 };
 
 /// Initialize ServiceManager
-void Init();
+void Init(std::shared_ptr<SM::ServiceManager>& sm);
 
 /// Shutdown ServiceManager
 void Shutdown();

src/core/hle/service/set/set.cpp

@@ -22,7 +22,7 @@ void SET::GetAvailableLanguageCodes(Kernel::HLERequestContext& ctx) {
 
     rb.Push(RESULT_SUCCESS);
 
-    LOG_WARNING(Service_SET, "(STUBBED) called");
+    NGLOG_WARNING(Service_SET, "(STUBBED) called");
 }
 
 SET::SET() : ServiceFramework("set") {

src/core/hle/service/set/set_sys.cpp

@@ -16,7 +16,7 @@ void SET_SYS::GetColorSetId(Kernel::HLERequestContext& ctx) {
     rb.Push(RESULT_SUCCESS);
     rb.Push<u32>(0);
 
-    LOG_WARNING(Service_SET, "(STUBBED) called");
+    NGLOG_WARNING(Service_SET, "(STUBBED) called");
 }
 
 SET_SYS::SET_SYS() : ServiceFramework("set:sys") {

src/core/hle/service/sm/controller.cpp

@@ -17,7 +17,7 @@ void Controller::ConvertSessionToDomain(Kernel::HLERequestContext& ctx) {
     rb.Push(RESULT_SUCCESS);
     rb.Push<u32>(1); // Converted sessions start with 1 request handler
 
-    LOG_DEBUG(Service, "called, server_session=%d", ctx.Session()->GetObjectId());
+    NGLOG_DEBUG(Service, "called, server_session={}", ctx.Session()->GetObjectId());
 }
 
 void Controller::DuplicateSession(Kernel::HLERequestContext& ctx) {
@@ -29,11 +29,11 @@ void Controller::DuplicateSession(Kernel::HLERequestContext& ctx) {
     Kernel::SharedPtr<Kernel::ClientSession> session{ctx.Session()->parent->client};
     rb.PushMoveObjects(session);
 
-    LOG_DEBUG(Service, "called, session=%u", session->GetObjectId());
+    NGLOG_DEBUG(Service, "called, session={}", session->GetObjectId());
 }
 
 void Controller::DuplicateSessionEx(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service, "(STUBBED) called, using DuplicateSession");
+    NGLOG_WARNING(Service, "(STUBBED) called, using DuplicateSession");
 
     DuplicateSession(ctx);
 }
@@ -43,7 +43,7 @@ void Controller::QueryPointerBufferSize(Kernel::HLERequestContext& ctx) {
     rb.Push(RESULT_SUCCESS);
     rb.Push<u32>(0x500);
 
-    LOG_WARNING(Service, "(STUBBED) called");
+    NGLOG_WARNING(Service, "(STUBBED) called");
 }
 
 Controller::Controller() : ServiceFramework("IpcController") {

src/core/hle/service/sm/sm.cpp

@@ -14,6 +14,8 @@
 
 namespace Service::SM {
 
+ServiceManager::~ServiceManager() = default;
+
 void ServiceManager::InvokeControlRequest(Kernel::HLERequestContext& context) {
     controller_interface->InvokeRequest(context);
 }
@@ -72,7 +74,7 @@ ResultVal<Kernel::SharedPtr<Kernel::ClientSession>> ServiceManager::ConnectToSer
     return client_port->Connect();
 }
 
-std::shared_ptr<ServiceManager> g_service_manager;
+SM::~SM() = default;
 
 /**
  * SM::Initialize service function
@@ -84,7 +86,7 @@ std::shared_ptr<ServiceManager> g_service_manager;
 void SM::Initialize(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
-    LOG_DEBUG(Service_SM, "called");
+    NGLOG_DEBUG(Service_SM, "called");
 }
 
 void SM::GetService(Kernel::HLERequestContext& ctx) {
@@ -100,8 +102,8 @@ void SM::GetService(Kernel::HLERequestContext& ctx) {
     if (client_port.Failed()) {
         IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
         rb.Push(client_port.Code());
-        LOG_ERROR(Service_SM, "called service=%s -> error 0x%08X", name.c_str(),
-                  client_port.Code().raw);
+        NGLOG_ERROR(Service_SM, "called service={} -> error {:#010X}", name,
+                    client_port.Code().raw);
         if (name.length() == 0)
             return; // LibNX Fix
         UNIMPLEMENTED();
@@ -111,8 +113,7 @@ void SM::GetService(Kernel::HLERequestContext& ctx) {
     auto session = client_port.Unwrap()->Connect();
     ASSERT(session.Succeeded());
     if (session.Succeeded()) {
-        LOG_DEBUG(Service_SM, "called service=%s -> session=%u", name.c_str(),
-                  (*session)->GetObjectId());
+        NGLOG_DEBUG(Service_SM, "called service={} -> session={}", name, (*session)->GetObjectId());
         IPC::ResponseBuilder rb =
             rp.MakeBuilder(2, 0, 1, IPC::ResponseBuilder::Flags::AlwaysMoveHandles);
         rb.Push(session.Code());

src/core/hle/service/sm/sm.h

@@ -23,7 +23,7 @@ namespace Service::SM {
 class SM final : public ServiceFramework<SM> {
 public:
     SM(std::shared_ptr<ServiceManager> service_manager);
-    ~SM() = default;
+    ~SM() override;
 
 private:
     void Initialize(Kernel::HLERequestContext& ctx);
@@ -44,6 +44,8 @@ class ServiceManager {
 public:
     static void InstallInterfaces(std::shared_ptr<ServiceManager> self);
 
+    ~ServiceManager();
+
     ResultVal<Kernel::SharedPtr<Kernel::ServerPort>> RegisterService(std::string name,
                                                                      unsigned int max_sessions);
     ResultVal<Kernel::SharedPtr<Kernel::ClientPort>> GetServicePort(const std::string& name);
@@ -59,6 +61,4 @@ private:
     std::unordered_map<std::string, Kernel::SharedPtr<Kernel::ClientPort>> registered_services;
 };
 
-extern std::shared_ptr<ServiceManager> g_service_manager;
-
 } // namespace Service::SM

src/core/hle/service/sockets/bsd.cpp

@@ -8,7 +8,7 @@
 namespace Service::Sockets {
 
 void BSD::RegisterClient(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service, "(STUBBED) called");
+    NGLOG_WARNING(Service, "(STUBBED) called");
 
     IPC::ResponseBuilder rb{ctx, 3};
 
@@ -17,7 +17,7 @@ void BSD::RegisterClient(Kernel::HLERequestContext& ctx) {
 }
 
 void BSD::StartMonitoring(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service, "(STUBBED) called");
+    NGLOG_WARNING(Service, "(STUBBED) called");
 
     IPC::ResponseBuilder rb{ctx, 3};
 
@@ -32,7 +32,8 @@ void BSD::Socket(Kernel::HLERequestContext& ctx) {
     u32 type = rp.Pop<u32>();
     u32 protocol = rp.Pop<u32>();
 
-    LOG_WARNING(Service, "(STUBBED) called domain=%u type=%u protocol=%u", domain, type, protocol);
+    NGLOG_WARNING(Service, "(STUBBED) called domain={} type={} protocol={}", domain, type,
+                  protocol);
 
     u32 fd = next_fd++;
 
@@ -44,7 +45,7 @@ void BSD::Socket(Kernel::HLERequestContext& ctx) {
 }
 
 void BSD::Connect(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service, "(STUBBED) called");
+    NGLOG_WARNING(Service, "(STUBBED) called");
 
     IPC::ResponseBuilder rb{ctx, 4};
 
@@ -54,7 +55,7 @@ void BSD::Connect(Kernel::HLERequestContext& ctx) {
 }
 
 void BSD::SendTo(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service, "(STUBBED) called");
+    NGLOG_WARNING(Service, "(STUBBED) called");
 
     IPC::ResponseBuilder rb{ctx, 4};
 
@@ -64,7 +65,7 @@ void BSD::SendTo(Kernel::HLERequestContext& ctx) {
 }
 
 void BSD::Close(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service, "(STUBBED) called");
+    NGLOG_WARNING(Service, "(STUBBED) called");
 
     IPC::ResponseBuilder rb{ctx, 4};
 

src/core/hle/service/sockets/sfdnsres.cpp

@@ -10,7 +10,7 @@ namespace Service::Sockets {
 void SFDNSRES::GetAddrInfo(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
 
-    LOG_WARNING(Service, "(STUBBED) called");
+    NGLOG_WARNING(Service, "(STUBBED) called");
 
     IPC::ResponseBuilder rb{ctx, 2};
 

src/core/hle/service/spl/module.cpp

@@ -28,7 +28,7 @@ void Module::Interface::GetRandomBytes(Kernel::HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
-    LOG_DEBUG(Service_SPL, "called");
+    NGLOG_DEBUG(Service_SPL, "called");
 }
 
 void InstallInterfaces(SM::ServiceManager& service_manager) {

src/core/hle/service/ssl/ssl.cpp

@@ -65,7 +65,7 @@ public:
 
 private:
     void SetOption(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_SSL, "(STUBBED) called");
+        NGLOG_WARNING(Service_SSL, "(STUBBED) called");
         IPC::RequestParser rp{ctx};
 
         IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
@@ -73,7 +73,7 @@ private:
     }
 
     void CreateConnection(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_SSL, "(STUBBED) called");
+        NGLOG_WARNING(Service_SSL, "(STUBBED) called");
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
@@ -82,7 +82,7 @@ private:
 };
 
 void SSL::CreateContext(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_SSL, "(STUBBED) called");
+    NGLOG_WARNING(Service_SSL, "(STUBBED) called");
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);

src/core/hle/service/time/time.cpp

@@ -32,14 +32,14 @@ private:
         const s64 time_since_epoch{std::chrono::duration_cast<std::chrono::seconds>(
                                        std::chrono::system_clock::now().time_since_epoch())
                                        .count()};
-        LOG_DEBUG(Service_Time, "called");
+        NGLOG_DEBUG(Service_Time, "called");
         IPC::ResponseBuilder rb{ctx, 4};
         rb.Push(RESULT_SUCCESS);
         rb.Push<u64>(time_since_epoch);
     }
 
     void GetSystemClockContext(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_Time, "(STUBBED) called");
+        NGLOG_WARNING(Service_Time, "(STUBBED) called");
         SystemClockContext system_clock_ontext{};
         IPC::ResponseBuilder rb{ctx, (sizeof(SystemClockContext) / 4) + 2};
         rb.Push(RESULT_SUCCESS);
@@ -58,7 +58,7 @@ public:
 
 private:
     void GetCurrentTimePoint(Kernel::HLERequestContext& ctx) {
-        LOG_DEBUG(Service_Time, "called");
+        NGLOG_DEBUG(Service_Time, "called");
         SteadyClockTimePoint steady_clock_time_point{cyclesToMs(CoreTiming::GetTicks()) / 1000};
         IPC::ResponseBuilder rb{ctx, (sizeof(SteadyClockTimePoint) / 4) + 2};
         rb.Push(RESULT_SUCCESS);
@@ -86,7 +86,7 @@ public:
 
 private:
     void GetDeviceLocationName(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_Time, "(STUBBED) called");
+        NGLOG_WARNING(Service_Time, "(STUBBED) called");
         LocationName location_name{};
         IPC::ResponseBuilder rb{ctx, (sizeof(LocationName) / 4) + 2};
         rb.Push(RESULT_SUCCESS);
@@ -94,14 +94,14 @@ private:
     }
 
     void GetTotalLocationNameCount(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_Time, "(STUBBED) called");
+        NGLOG_WARNING(Service_Time, "(STUBBED) called");
         IPC::ResponseBuilder rb{ctx, 3};
         rb.Push(RESULT_SUCCESS);
         rb.Push<u32>(0);
     }
 
     void LoadTimeZoneRule(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_Time, "(STUBBED) called");
+        NGLOG_WARNING(Service_Time, "(STUBBED) called");
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
     }
@@ -110,7 +110,7 @@ private:
         IPC::RequestParser rp{ctx};
         u64 posix_time = rp.Pop<u64>();
 
-        LOG_WARNING(Service_Time, "(STUBBED) called, posix_time=0x%016lX", posix_time);
+        NGLOG_WARNING(Service_Time, "(STUBBED) called, posix_time={:#018X}", posix_time);
 
         CalendarTime calendar_time{2018, 1, 1, 0, 0, 0};
         CalendarAdditionalInfo additional_info{};
@@ -125,35 +125,35 @@ void Module::Interface::GetStandardUserSystemClock(Kernel::HLERequestContext& ct
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
     rb.PushIpcInterface<ISystemClock>();
-    LOG_DEBUG(Service_Time, "called");
+    NGLOG_DEBUG(Service_Time, "called");
 }
 
 void Module::Interface::GetStandardNetworkSystemClock(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
     rb.PushIpcInterface<ISystemClock>();
-    LOG_DEBUG(Service_Time, "called");
+    NGLOG_DEBUG(Service_Time, "called");
 }
 
 void Module::Interface::GetStandardSteadyClock(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
     rb.PushIpcInterface<ISteadyClock>();
-    LOG_DEBUG(Service_Time, "called");
+    NGLOG_DEBUG(Service_Time, "called");
 }
 
 void Module::Interface::GetTimeZoneService(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
     rb.PushIpcInterface<ITimeZoneService>();
-    LOG_DEBUG(Service_Time, "called");
+    NGLOG_DEBUG(Service_Time, "called");
 }
 
 void Module::Interface::GetStandardLocalSystemClock(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
     rb.PushIpcInterface<ISystemClock>();
-    LOG_DEBUG(Service_Time, "called");
+    NGLOG_DEBUG(Service_Time, "called");
 }
 
 Module::Interface::Interface(std::shared_ptr<Module> time, const char* name)

src/core/hle/service/vi/vi.cpp

@@ -470,7 +470,7 @@ private:
         u32 flags = rp.Pop<u32>();
         auto buffer_queue = nv_flinger->GetBufferQueue(id);
 
-        LOG_DEBUG(Service_VI, "called, transaction=%x", static_cast<u32>(transaction));
+        NGLOG_DEBUG(Service_VI, "called, transaction={:X}", static_cast<u32>(transaction));
 
         if (transaction == TransactionId::Connect) {
             IGBPConnectRequestParcel request{ctx.ReadBuffer()};
@@ -532,7 +532,7 @@ private:
             IGBPQueryResponseParcel response{value};
             ctx.WriteBuffer(response.Serialize());
         } else if (transaction == TransactionId::CancelBuffer) {
-            LOG_WARNING(Service_VI, "(STUBBED) called, transaction=CancelBuffer");
+            NGLOG_WARNING(Service_VI, "(STUBBED) called, transaction=CancelBuffer");
         } else {
             ASSERT_MSG(false, "Unimplemented");
         }
@@ -547,7 +547,8 @@ private:
         s32 addval = rp.PopRaw<s32>();
         u32 type = rp.Pop<u32>();
 
-        LOG_WARNING(Service_VI, "(STUBBED) called id=%u, addval=%08X, type=%08X", id, addval, type);
+        NGLOG_WARNING(Service_VI, "(STUBBED) called id={}, addval={:08X}, type={:08X}", id, addval,
+                      type);
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
     }
@@ -561,7 +562,7 @@ private:
 
         // TODO(Subv): Find out what this actually is.
 
-        LOG_WARNING(Service_VI, "(STUBBED) called id=%u, unknown=%08X", id, unknown);
+        NGLOG_WARNING(Service_VI, "(STUBBED) called id={}, unknown={:08X}", id, unknown);
         IPC::ResponseBuilder rb{ctx, 2, 1};
         rb.Push(RESULT_SUCCESS);
         rb.PushCopyObjects(buffer_queue->GetNativeHandle());
@@ -624,7 +625,7 @@ public:
 
 private:
     void SetLayerZ(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_VI, "(STUBBED) called");
+        NGLOG_WARNING(Service_VI, "(STUBBED) called");
         IPC::RequestParser rp{ctx};
         u64 layer_id = rp.Pop<u64>();
         u64 z_value = rp.Pop<u64>();
@@ -639,8 +640,8 @@ private:
         bool visibility = rp.Pop<bool>();
         IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_VI, "(STUBBED) called, layer_id=0x%x, visibility=%u", layer_id,
-                    visibility);
+        NGLOG_WARNING(Service_VI, "(STUBBED) called, layer_id={:#010X}, visibility={}", layer_id,
+                      visibility);
     }
 };
 
@@ -722,7 +723,7 @@ public:
 
 private:
     void CloseDisplay(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_VI, "(STUBBED) called");
+        NGLOG_WARNING(Service_VI, "(STUBBED) called");
         IPC::RequestParser rp{ctx};
         u64 display = rp.Pop<u64>();
 
@@ -731,7 +732,7 @@ private:
     }
 
     void CreateManagedLayer(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_VI, "(STUBBED) called");
+        NGLOG_WARNING(Service_VI, "(STUBBED) called");
         IPC::RequestParser rp{ctx};
         u32 unknown = rp.Pop<u32>();
         rp.Skip(1, false);
@@ -746,7 +747,7 @@ private:
     }
 
     void AddToLayerStack(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_VI, "(STUBBED) called");
+        NGLOG_WARNING(Service_VI, "(STUBBED) called");
         IPC::RequestParser rp{ctx};
         u32 stack = rp.Pop<u32>();
         u64 layer_id = rp.Pop<u64>();
@@ -761,8 +762,8 @@ private:
         bool visibility = rp.Pop<bool>();
         IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_VI, "(STUBBED) called, layer_id=0x%x, visibility=%u", layer_id,
-                    visibility);
+        NGLOG_WARNING(Service_VI, "(STUBBED) called, layer_id={:#X}, visibility={}", layer_id,
+                      visibility);
     }
 
     std::shared_ptr<NVFlinger::NVFlinger> nv_flinger;
@@ -775,7 +776,7 @@ public:
 
 private:
     void GetRelayService(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_VI, "(STUBBED) called");
+        NGLOG_WARNING(Service_VI, "(STUBBED) called");
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
@@ -783,7 +784,7 @@ private:
     }
 
     void GetSystemDisplayService(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_VI, "(STUBBED) called");
+        NGLOG_WARNING(Service_VI, "(STUBBED) called");
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
@@ -791,7 +792,7 @@ private:
     }
 
     void GetManagerDisplayService(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_VI, "(STUBBED) called");
+        NGLOG_WARNING(Service_VI, "(STUBBED) called");
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
@@ -799,7 +800,7 @@ private:
     }
 
     void GetIndirectDisplayTransactionService(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_VI, "(STUBBED) called");
+        NGLOG_WARNING(Service_VI, "(STUBBED) called");
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
@@ -807,7 +808,7 @@ private:
     }
 
     void OpenDisplay(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_VI, "(STUBBED) called");
+        NGLOG_WARNING(Service_VI, "(STUBBED) called");
         IPC::RequestParser rp{ctx};
         auto name_buf = rp.PopRaw<std::array<u8, 0x40>>();
         auto end = std::find(name_buf.begin(), name_buf.end(), '\0');
@@ -822,7 +823,7 @@ private:
     }
 
     void CloseDisplay(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_VI, "(STUBBED) called");
+        NGLOG_WARNING(Service_VI, "(STUBBED) called");
         IPC::RequestParser rp{ctx};
         u64 display_id = rp.Pop<u64>();
 
@@ -831,7 +832,7 @@ private:
     }
 
     void GetDisplayResolution(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_VI, "(STUBBED) called");
+        NGLOG_WARNING(Service_VI, "(STUBBED) called");
         IPC::RequestParser rp{ctx};
         u64 display_id = rp.Pop<u64>();
 
@@ -848,7 +849,7 @@ private:
     }
 
     void SetLayerScalingMode(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_VI, "(STUBBED) called");
+        NGLOG_WARNING(Service_VI, "(STUBBED) called");
         IPC::RequestParser rp{ctx};
         u32 scaling_mode = rp.Pop<u32>();
         u64 unknown = rp.Pop<u64>();
@@ -864,11 +865,11 @@ private:
         IPC::ResponseBuilder rb = rp.MakeBuilder(4, 0, 0);
         rb.Push(RESULT_SUCCESS);
         rb.Push<u64>(1);
-        LOG_WARNING(Service_VI, "(STUBBED) called");
+        NGLOG_WARNING(Service_VI, "(STUBBED) called");
     }
 
     void OpenLayer(Kernel::HLERequestContext& ctx) {
-        LOG_DEBUG(Service_VI, "called");
+        NGLOG_DEBUG(Service_VI, "called");
         IPC::RequestParser rp{ctx};
         auto name_buf = rp.PopRaw<std::array<u8, 0x40>>();
         auto end = std::find(name_buf.begin(), name_buf.end(), '\0');
@@ -888,7 +889,7 @@ private:
     }
 
     void CreateStrayLayer(Kernel::HLERequestContext& ctx) {
-        LOG_DEBUG(Service_VI, "called");
+        NGLOG_DEBUG(Service_VI, "called");
 
         IPC::RequestParser rp{ctx};
         u32 flags = rp.Pop<u32>();
@@ -908,7 +909,7 @@ private:
     }
 
     void DestroyStrayLayer(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_VI, "(STUBBED) called");
+        NGLOG_WARNING(Service_VI, "(STUBBED) called");
 
         IPC::RequestParser rp{ctx};
         u64 layer_id = rp.Pop<u64>();
@@ -918,7 +919,7 @@ private:
     }
 
     void GetDisplayVsyncEvent(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_VI, "(STUBBED) called");
+        NGLOG_WARNING(Service_VI, "(STUBBED) called");
         IPC::RequestParser rp{ctx};
         u64 display_id = rp.Pop<u64>();
 
@@ -967,7 +968,7 @@ Module::Interface::Interface(std::shared_ptr<Module> module, const char* name,
     : ServiceFramework(name), module(std::move(module)), nv_flinger(std::move(nv_flinger)) {}
 
 void Module::Interface::GetDisplayService(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_VI, "(STUBBED) called");
+    NGLOG_WARNING(Service_VI, "(STUBBED) called");
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);

src/core/loader/deconstructed_rom_directory.cpp

@@ -132,7 +132,7 @@ ResultStatus AppLoader_DeconstructedRomDirectory::Load(
         const VAddr load_addr = next_load_addr;
         next_load_addr = AppLoader_NSO::LoadModule(path, load_addr);
         if (next_load_addr) {
-            LOG_DEBUG(Loader, "loaded module %s @ 0x%" PRIx64, module, load_addr);
+            NGLOG_DEBUG(Loader, "loaded module {} @ {:#X}", module, load_addr);
         } else {
             next_load_addr = load_addr;
         }
@@ -163,7 +163,7 @@ ResultStatus AppLoader_DeconstructedRomDirectory::ReadRomFS(
     std::shared_ptr<FileUtil::IOFile>& romfs_file, u64& offset, u64& size) {
 
     if (filepath_romfs.empty()) {
-        LOG_DEBUG(Loader, "No RomFS available");
+        NGLOG_DEBUG(Loader, "No RomFS available");
         return ResultStatus::ErrorNotUsed;
     }
 
@@ -176,8 +176,8 @@ ResultStatus AppLoader_DeconstructedRomDirectory::ReadRomFS(
     offset = 0;
     size = romfs_file->GetSize();
 
-    LOG_DEBUG(Loader, "RomFS offset:           0x%016" PRIX64, offset);
-    LOG_DEBUG(Loader, "RomFS size:             0x%016" PRIX64, size);
+    NGLOG_DEBUG(Loader, "RomFS offset:           {:#018X}", offset);
+    NGLOG_DEBUG(Loader, "RomFS size:             {:#018X}", size);
 
     // Reset read pointer
     file.Seek(0, SEEK_SET);

src/core/loader/elf.cpp

@@ -273,18 +273,18 @@ const char* ElfReader::GetSectionName(int section) const {
 }
 
 SharedPtr<CodeSet> ElfReader::LoadInto(u32 vaddr) {
-    LOG_DEBUG(Loader, "String section: %i", header->e_shstrndx);
+    NGLOG_DEBUG(Loader, "String section: {}", header->e_shstrndx);
 
     // Should we relocate?
     relocate = (header->e_type != ET_EXEC);
 
     if (relocate) {
-        LOG_DEBUG(Loader, "Relocatable module");
+        NGLOG_DEBUG(Loader, "Relocatable module");
         entryPoint += vaddr;
     } else {
-        LOG_DEBUG(Loader, "Prerelocated executable");
+        NGLOG_DEBUG(Loader, "Prerelocated executable");
     }
-    LOG_DEBUG(Loader, "%i segments:", header->e_phnum);
+    NGLOG_DEBUG(Loader, "{} segments:", header->e_phnum);
 
     // First pass : Get the bits into RAM
     u32 base_addr = relocate ? vaddr : 0;
@@ -304,8 +304,8 @@ SharedPtr<CodeSet> ElfReader::LoadInto(u32 vaddr) {
 
     for (unsigned int i = 0; i < header->e_phnum; ++i) {
         Elf32_Phdr* p = &segments[i];
-        LOG_DEBUG(Loader, "Type: %i Vaddr: %08X Filesz: %8X Memsz: %8X ", p->p_type, p->p_vaddr,
-                  p->p_filesz, p->p_memsz);
+        NGLOG_DEBUG(Loader, "Type: {} Vaddr: {:08X} Filesz: {:08X} Memsz: {:08X} ", p->p_type,
+                    p->p_vaddr, p->p_filesz, p->p_memsz);
 
         if (p->p_type == PT_LOAD) {
             CodeSet::Segment* codeset_segment;
@@ -317,16 +317,16 @@ SharedPtr<CodeSet> ElfReader::LoadInto(u32 vaddr) {
             } else if (permission_flags == (PF_R | PF_W)) {
                 codeset_segment = &codeset->data;
             } else {
-                LOG_ERROR(Loader, "Unexpected ELF PT_LOAD segment id %u with flags %X", i,
-                          p->p_flags);
+                NGLOG_ERROR(Loader, "Unexpected ELF PT_LOAD segment id {} with flags {:X}", i,
+                            p->p_flags);
                 continue;
             }
 
             if (codeset_segment->size != 0) {
-                LOG_ERROR(Loader,
-                          "ELF has more than one segment of the same type. Skipping extra "
-                          "segment (id %i)",
-                          i);
+                NGLOG_ERROR(Loader,
+                            "ELF has more than one segment of the same type. Skipping extra "
+                            "segment (id {})",
+                            i);
                 continue;
             }
 
@@ -345,7 +345,7 @@ SharedPtr<CodeSet> ElfReader::LoadInto(u32 vaddr) {
     codeset->entrypoint = base_addr + header->e_entry;
     codeset->memory = std::make_shared<std::vector<u8>>(std::move(program_image));
 
-    LOG_DEBUG(Loader, "Done loading.");
+    NGLOG_DEBUG(Loader, "Done loading.");
 
     return codeset;
 }

src/core/loader/linker.cpp

@@ -84,7 +84,7 @@ void Linker::WriteRelocations(std::vector<u8>& program_image, const std::vector<
             }
             break;
         default:
-            LOG_CRITICAL(Loader, "Unknown relocation type: %d", static_cast<int>(rela.type));
+            NGLOG_CRITICAL(Loader, "Unknown relocation type: {}", static_cast<int>(rela.type));
             break;
         }
     }
@@ -141,7 +141,7 @@ void Linker::ResolveImports() {
         if (search != exports.end()) {
             Memory::Write64(import.second.ea, search->second + import.second.addend);
         } else {
-            LOG_ERROR(Loader, "Unresolved import: %s", import.first.c_str());
+            NGLOG_ERROR(Loader, "Unresolved import: {}", import.first);
         }
     }
 }

src/core/loader/loader.cpp

@@ -41,7 +41,7 @@ FileType IdentifyFile(FileUtil::IOFile& file, const std::string& filepath) {
 FileType IdentifyFile(const std::string& file_name) {
     FileUtil::IOFile file(file_name, "rb");
     if (!file.IsOpen()) {
-        LOG_ERROR(Loader, "Failed to load file %s", file_name.c_str());
+        NGLOG_ERROR(Loader, "Failed to load file {}", file_name);
         return FileType::Unknown;
     }
 
@@ -116,7 +116,7 @@ static std::unique_ptr<AppLoader> GetFileLoader(FileUtil::IOFile&& file, FileTyp
 std::unique_ptr<AppLoader> GetLoader(const std::string& filename) {
     FileUtil::IOFile file(filename, "rb");
     if (!file.IsOpen()) {
-        LOG_ERROR(Loader, "Failed to load file %s", filename.c_str());
+        NGLOG_ERROR(Loader, "Failed to load file {}", filename);
         return nullptr;
     }
 
@@ -127,12 +127,12 @@ std::unique_ptr<AppLoader> GetLoader(const std::string& filename) {
     FileType filename_type = GuessFromExtension(filename_extension);
 
     if (type != filename_type) {
-        LOG_WARNING(Loader, "File %s has a different type than its extension.", filename.c_str());
+        NGLOG_WARNING(Loader, "File {} has a different type than its extension.", filename);
         if (FileType::Unknown == type)
             type = filename_type;
     }
 
-    LOG_DEBUG(Loader, "Loading file %s as %s...", filename.c_str(), GetFileTypeString(type));
+    NGLOG_DEBUG(Loader, "Loading file {} as {}...", filename, GetFileTypeString(type));
 
     return GetFileLoader(std::move(file), type, filename_filename, filename);
 }

src/core/loader/nro.cpp

@@ -137,7 +137,7 @@ ResultStatus AppLoader_NRO::Load(Kernel::SharedPtr<Kernel::Process>& process) {
     process->address_mappings = default_address_mappings;
     process->resource_limit =
         Kernel::ResourceLimit::GetForCategory(Kernel::ResourceLimitCategory::APPLICATION);
-    process->Run(base_addr, 48, Memory::DEFAULT_STACK_SIZE);
+    process->Run(base_addr, THREADPRIO_DEFAULT, Memory::DEFAULT_STACK_SIZE);
 
     is_loaded = true;
     return ResultStatus::Success;

src/core/loader/nso.cpp

@@ -73,7 +73,7 @@ static std::vector<u8> ReadSegment(FileUtil::IOFile& file, const NsoSegmentHeade
 
     file.Seek(header.offset, SEEK_SET);
     if (compressed_size != file.ReadBytes(compressed_data.data(), compressed_size)) {
-        LOG_CRITICAL(Loader, "Failed to read %d NSO LZ4 compressed bytes", compressed_size);
+        NGLOG_CRITICAL(Loader, "Failed to read {} NSO LZ4 compressed bytes", compressed_size);
         return {};
     }
 
@@ -158,14 +158,13 @@ ResultStatus AppLoader_NSO::Load(Kernel::SharedPtr<Kernel::Process>& process) {
 
     // Load module
     LoadModule(filepath, Memory::PROCESS_IMAGE_VADDR);
-    LOG_DEBUG(Loader, "loaded module %s @ 0x%" PRIx64, filepath.c_str(),
-              Memory::PROCESS_IMAGE_VADDR);
+    NGLOG_DEBUG(Loader, "loaded module {} @ {:#X}", filepath, Memory::PROCESS_IMAGE_VADDR);
 
     process->svc_access_mask.set();
     process->address_mappings = default_address_mappings;
     process->resource_limit =
         Kernel::ResourceLimit::GetForCategory(Kernel::ResourceLimitCategory::APPLICATION);
-    process->Run(Memory::PROCESS_IMAGE_VADDR, 48, Memory::DEFAULT_STACK_SIZE);
+    process->Run(Memory::PROCESS_IMAGE_VADDR, THREADPRIO_DEFAULT, Memory::DEFAULT_STACK_SIZE);
 
     is_loaded = true;
     return ResultStatus::Success;

src/core/memory.cpp

@@ -325,15 +325,29 @@ u8* GetPhysicalPointer(PAddr address) {
     return target_pointer;
 }
 
-void RasterizerMarkRegionCached(VAddr start, u64 size, bool cached) {
-    if (start == 0) {
+void RasterizerMarkRegionCached(Tegra::GPUVAddr gpu_addr, u64 size, bool cached) {
+    if (gpu_addr == 0) {
         return;
     }
 
-    u64 num_pages = ((start + size - 1) >> PAGE_BITS) - (start >> PAGE_BITS) + 1;
-    VAddr vaddr = start;
+    // Iterate over a contiguous CPU address space, which corresponds to the specified GPU address
+    // space, marking the region as un/cached. The region is marked un/cached at a granularity of
+    // CPU pages, hence why we iterate on a CPU page basis (note: GPU page size is different). This
+    // assumes the specified GPU address region is contiguous as well.
+
+    u64 num_pages = ((gpu_addr + size - 1) >> PAGE_BITS) - (gpu_addr >> PAGE_BITS) + 1;
+    for (unsigned i = 0; i < num_pages; ++i, gpu_addr += PAGE_SIZE) {
+        boost::optional<VAddr> maybe_vaddr =
+            Core::System::GetInstance().GPU().memory_manager->GpuToCpuAddress(gpu_addr);
+        // The GPU <-> CPU virtual memory mapping is not 1:1
+        if (!maybe_vaddr) {
+            LOG_ERROR(HW_Memory,
+                      "Trying to flush a cached region to an invalid physical address %08X",
+                      gpu_addr);
+            continue;
+        }
+        VAddr vaddr = *maybe_vaddr;
 
-    for (unsigned i = 0; i < num_pages; ++i, vaddr += PAGE_SIZE) {
         PageType& page_type = current_page_table->attributes[vaddr >> PAGE_BITS];
 
         if (cached) {
@@ -347,6 +361,10 @@ void RasterizerMarkRegionCached(VAddr start, u64 size, bool cached) {
                 page_type = PageType::RasterizerCachedMemory;
                 current_page_table->pointers[vaddr >> PAGE_BITS] = nullptr;
                 break;
+            case PageType::RasterizerCachedMemory:
+                // There can be more than one GPU region mapped per CPU region, so it's common that
+                // this area is already marked as cached.
+                break;
             default:
                 UNREACHABLE();
             }
@@ -357,6 +375,10 @@ void RasterizerMarkRegionCached(VAddr start, u64 size, bool cached) {
                 // It is not necessary for a process to have this region mapped into its address
                 // space, for example, a system module need not have a VRAM mapping.
                 break;
+            case PageType::Memory:
+                // There can be more than one GPU region mapped per CPU region, so it's common that
+                // this area is already unmarked as cached.
+                break;
             case PageType::RasterizerCachedMemory: {
                 u8* pointer = GetPointerFromVMA(vaddr & ~PAGE_MASK);
                 if (pointer == nullptr) {
@@ -394,19 +416,29 @@ void RasterizerFlushVirtualRegion(VAddr start, u64 size, FlushMode mode) {
 
         VAddr overlap_start = std::max(start, region_start);
         VAddr overlap_end = std::min(end, region_end);
+
+        std::vector<Tegra::GPUVAddr> gpu_addresses =
+            Core::System::GetInstance().GPU().memory_manager->CpuToGpuAddress(overlap_start);
+
+        if (gpu_addresses.empty()) {
+            return;
+        }
+
         u64 overlap_size = overlap_end - overlap_start;
 
-        auto* rasterizer = VideoCore::g_renderer->Rasterizer();
-        switch (mode) {
-        case FlushMode::Flush:
-            rasterizer->FlushRegion(overlap_start, overlap_size);
-            break;
-        case FlushMode::Invalidate:
-            rasterizer->InvalidateRegion(overlap_start, overlap_size);
-            break;
-        case FlushMode::FlushAndInvalidate:
-            rasterizer->FlushAndInvalidateRegion(overlap_start, overlap_size);
-            break;
+        for (const auto& gpu_address : gpu_addresses) {
+            auto* rasterizer = VideoCore::g_renderer->Rasterizer();
+            switch (mode) {
+            case FlushMode::Flush:
+                rasterizer->FlushRegion(gpu_address, overlap_size);
+                break;
+            case FlushMode::Invalidate:
+                rasterizer->InvalidateRegion(gpu_address, overlap_size);
+                break;
+            case FlushMode::FlushAndInvalidate:
+                rasterizer->FlushAndInvalidateRegion(gpu_address, overlap_size);
+                break;
+            }
         }
     };
 
@@ -625,6 +657,10 @@ void CopyBlock(const Kernel::Process& process, VAddr dest_addr, VAddr src_addr,
     }
 }
 
+void CopyBlock(VAddr dest_addr, VAddr src_addr, size_t size) {
+    CopyBlock(*Core::CurrentProcess(), dest_addr, src_addr, size);
+}
+
 boost::optional<PAddr> TryVirtualToPhysicalAddress(const VAddr addr) {
     if (addr == 0) {
         return 0;

src/core/memory.h

@@ -14,6 +14,7 @@
 #include <boost/optional.hpp>
 #include "common/common_types.h"
 #include "core/memory_hook.h"
+#include "video_core/memory_manager.h"
 
 namespace Kernel {
 class Process;
@@ -258,7 +259,7 @@ enum class FlushMode {
 /**
  * Mark each page touching the region as cached.
  */
-void RasterizerMarkRegionCached(VAddr start, u64 size, bool cached);
+void RasterizerMarkRegionCached(Tegra::GPUVAddr start, u64 size, bool cached);
 
 /**
  * Flushes and invalidates any externally cached rasterizer resources touching the given virtual

src/video_core/command_processor.cpp

@@ -24,41 +24,18 @@ namespace Tegra {
 
 enum class BufferMethods {
     BindObject = 0,
-    SetGraphMacroCode = 0x45,
-    SetGraphMacroCodeArg = 0x46,
-    SetGraphMacroEntry = 0x47,
-    CountBufferMethods = 0x100,
+    CountBufferMethods = 0x40,
 };
 
 void GPU::WriteReg(u32 method, u32 subchannel, u32 value, u32 remaining_params) {
-    LOG_WARNING(HW_GPU, "Processing method %08X on subchannel %u value %08X remaining params %u",
-                method, subchannel, value, remaining_params);
-
-    if (method == static_cast<u32>(BufferMethods::SetGraphMacroEntry)) {
-        // Prepare to upload a new macro, reset the upload counter.
-        LOG_DEBUG(HW_GPU, "Uploading GPU macro %08X", value);
-        current_macro_entry = value;
-        current_macro_code.clear();
-        return;
-    }
-
-    if (method == static_cast<u32>(BufferMethods::SetGraphMacroCodeArg)) {
-        // Append a new code word to the current macro.
-        current_macro_code.push_back(value);
-
-        // There are no more params remaining, submit the code to the 3D engine.
-        if (remaining_params == 0) {
-            maxwell_3d->SubmitMacroCode(current_macro_entry, std::move(current_macro_code));
-            current_macro_entry = InvalidGraphMacroEntry;
-            current_macro_code.clear();
-        }
-
-        return;
-    }
+    NGLOG_WARNING(HW_GPU,
+                  "Processing method {:08X} on subchannel {} value "
+                  "{:08X} remaining params {}",
+                  method, subchannel, value, remaining_params);
 
     if (method == static_cast<u32>(BufferMethods::BindObject)) {
         // Bind the current subchannel to the desired engine id.
-        LOG_DEBUG(HW_GPU, "Binding subchannel %u to engine %u", subchannel, value);
+        NGLOG_DEBUG(HW_GPU, "Binding subchannel {} to engine {}", subchannel, value);
         ASSERT(bound_engines.find(subchannel) == bound_engines.end());
         bound_engines[subchannel] = static_cast<EngineID>(value);
         return;
@@ -66,7 +43,7 @@ void GPU::WriteReg(u32 method, u32 subchannel, u32 value, u32 remaining_params)
 
     if (method < static_cast<u32>(BufferMethods::CountBufferMethods)) {
         // TODO(Subv): Research and implement these methods.
-        LOG_ERROR(HW_GPU, "Special buffer methods other than Bind are not implemented");
+        NGLOG_ERROR(HW_GPU, "Special buffer methods other than Bind are not implemented");
         return;
     }
 
@@ -90,11 +67,9 @@ void GPU::WriteReg(u32 method, u32 subchannel, u32 value, u32 remaining_params)
 }
 
 void GPU::ProcessCommandList(GPUVAddr address, u32 size) {
-    // TODO(Subv): PhysicalToVirtualAddress is a misnomer, it converts a GPU VAddr into an
-    // application VAddr.
-    const VAddr head_address = memory_manager->PhysicalToVirtualAddress(address);
-    VAddr current_addr = head_address;
-    while (current_addr < head_address + size * sizeof(CommandHeader)) {
+    const boost::optional<VAddr> head_address = memory_manager->GpuToCpuAddress(address);
+    VAddr current_addr = *head_address;
+    while (current_addr < *head_address + size * sizeof(CommandHeader)) {
         const CommandHeader header = {Memory::Read32(current_addr)};
         current_addr += sizeof(u32);
 

src/video_core/engines/fermi_2d.cpp

@@ -2,12 +2,71 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include "core/memory.h"
 #include "video_core/engines/fermi_2d.h"
+#include "video_core/textures/decoders.h"
 
 namespace Tegra {
 namespace Engines {
 
-void Fermi2D::WriteReg(u32 method, u32 value) {}
+Fermi2D::Fermi2D(MemoryManager& memory_manager) : memory_manager(memory_manager) {}
+
+void Fermi2D::WriteReg(u32 method, u32 value) {
+    ASSERT_MSG(method < Regs::NUM_REGS,
+               "Invalid Fermi2D register, increase the size of the Regs structure");
+
+    regs.reg_array[method] = value;
+
+    switch (method) {
+    case FERMI2D_REG_INDEX(trigger): {
+        HandleSurfaceCopy();
+        break;
+    }
+    }
+}
+
+void Fermi2D::HandleSurfaceCopy() {
+    NGLOG_WARNING(HW_GPU, "Requested a surface copy with operation {}",
+                  static_cast<u32>(regs.operation));
+
+    const GPUVAddr source = regs.src.Address();
+    const GPUVAddr dest = regs.dst.Address();
+
+    // TODO(Subv): Only same-format and same-size copies are allowed for now.
+    ASSERT(regs.src.format == regs.dst.format);
+    ASSERT(regs.src.width * regs.src.height == regs.dst.width * regs.dst.height);
+
+    // TODO(Subv): Only raw copies are implemented.
+    ASSERT(regs.operation == Regs::Operation::SrcCopy);
+
+    const VAddr source_cpu = *memory_manager.GpuToCpuAddress(source);
+    const VAddr dest_cpu = *memory_manager.GpuToCpuAddress(dest);
+
+    u32 src_bytes_per_pixel = RenderTargetBytesPerPixel(regs.src.format);
+    u32 dst_bytes_per_pixel = RenderTargetBytesPerPixel(regs.dst.format);
+
+    if (regs.src.linear == regs.dst.linear) {
+        // If the input layout and the output layout are the same, just perform a raw copy.
+        Memory::CopyBlock(dest_cpu, source_cpu,
+                          src_bytes_per_pixel * regs.dst.width * regs.dst.height);
+        return;
+    }
+
+    u8* src_buffer = Memory::GetPointer(source_cpu);
+    u8* dst_buffer = Memory::GetPointer(dest_cpu);
+
+    if (!regs.src.linear && regs.dst.linear) {
+        // If the input is tiled and the output is linear, deswizzle the input and copy it over.
+        Texture::CopySwizzledData(regs.src.width, regs.src.height, src_bytes_per_pixel,
+                                  dst_bytes_per_pixel, src_buffer, dst_buffer, true,
+                                  regs.src.block_height);
+    } else {
+        // If the input is linear and the output is tiled, swizzle the input and copy it over.
+        Texture::CopySwizzledData(regs.src.width, regs.src.height, src_bytes_per_pixel,
+                                  dst_bytes_per_pixel, dst_buffer, src_buffer, false,
+                                  regs.dst.block_height);
+    }
+}
 
 } // namespace Engines
 } // namespace Tegra

src/video_core/engines/fermi_2d.h

@@ -4,19 +4,106 @@
 
 #pragma once
 
+#include <array>
+#include "common/assert.h"
+#include "common/bit_field.h"
+#include "common/common_funcs.h"
 #include "common/common_types.h"
+#include "video_core/gpu.h"
+#include "video_core/memory_manager.h"
 
 namespace Tegra {
 namespace Engines {
 
+#define FERMI2D_REG_INDEX(field_name)                                                              \
+    (offsetof(Tegra::Engines::Fermi2D::Regs, field_name) / sizeof(u32))
+
 class Fermi2D final {
 public:
-    Fermi2D() = default;
+    explicit Fermi2D(MemoryManager& memory_manager);
     ~Fermi2D() = default;
 
     /// Write the value to the register identified by method.
     void WriteReg(u32 method, u32 value);
+
+    struct Regs {
+        static constexpr size_t NUM_REGS = 0x258;
+
+        struct Surface {
+            RenderTargetFormat format;
+            BitField<0, 1, u32> linear;
+            union {
+                BitField<0, 4, u32> block_depth;
+                BitField<4, 4, u32> block_height;
+                BitField<8, 4, u32> block_width;
+            };
+            u32 depth;
+            u32 layer;
+            u32 pitch;
+            u32 width;
+            u32 height;
+            u32 address_high;
+            u32 address_low;
+
+            GPUVAddr Address() const {
+                return static_cast<GPUVAddr>((static_cast<GPUVAddr>(address_high) << 32) |
+                                             address_low);
+            }
+        };
+        static_assert(sizeof(Surface) == 0x28, "Surface has incorrect size");
+
+        enum class Operation : u32 {
+            SrcCopyAnd = 0,
+            ROPAnd = 1,
+            Blend = 2,
+            SrcCopy = 3,
+            ROP = 4,
+            SrcCopyPremult = 5,
+            BlendPremult = 6,
+        };
+
+        union {
+            struct {
+                INSERT_PADDING_WORDS(0x80);
+
+                Surface dst;
+
+                INSERT_PADDING_WORDS(2);
+
+                Surface src;
+
+                INSERT_PADDING_WORDS(0x15);
+
+                Operation operation;
+
+                INSERT_PADDING_WORDS(0x9);
+
+                // TODO(Subv): This is only a guess.
+                u32 trigger;
+
+                INSERT_PADDING_WORDS(0x1A3);
+            };
+            std::array<u32, NUM_REGS> reg_array;
+        };
+    } regs{};
+
+    MemoryManager& memory_manager;
+
+private:
+    /// Performs the copy from the source surface to the destination surface as configured in the
+    /// registers.
+    void HandleSurfaceCopy();
 };
 
+#define ASSERT_REG_POSITION(field_name, position)                                                  \
+    static_assert(offsetof(Fermi2D::Regs, field_name) == position * 4,                             \
+                  "Field " #field_name " has invalid position")
+
+ASSERT_REG_POSITION(dst, 0x80);
+ASSERT_REG_POSITION(src, 0x8C);
+ASSERT_REG_POSITION(operation, 0xAB);
+ASSERT_REG_POSITION(trigger, 0xB5);
+#undef ASSERT_REG_POSITION
+
 } // namespace Engines
 } // namespace Tegra

src/video_core/engines/maxwell_3d.cpp

@@ -22,10 +22,6 @@ constexpr u32 MacroRegistersStart = 0xE00;
 Maxwell3D::Maxwell3D(MemoryManager& memory_manager)
     : memory_manager(memory_manager), macro_interpreter(*this) {}
 
-void Maxwell3D::SubmitMacroCode(u32 entry, std::vector<u32> code) {
-    uploaded_macros[entry * 2 + MacroRegistersStart] = std::move(code);
-}
-
 void Maxwell3D::CallMacroMethod(u32 method, std::vector<u32> parameters) {
     auto macro_code = uploaded_macros.find(method);
     // The requested macro must have been uploaded already.
@@ -37,9 +33,6 @@ void Maxwell3D::CallMacroMethod(u32 method, std::vector<u32> parameters) {
 }
 
 void Maxwell3D::WriteReg(u32 method, u32 value, u32 remaining_params) {
-    ASSERT_MSG(method < Regs::NUM_REGS,
-               "Invalid Maxwell3D register, increase the size of the Regs structure");
-
     auto debug_context = Core::System::GetInstance().GetGPUDebugContext();
 
     // It is an error to write to a register other than the current macro's ARG register before it
@@ -68,6 +61,9 @@ void Maxwell3D::WriteReg(u32 method, u32 value, u32 remaining_params) {
         return;
     }
 
+    ASSERT_MSG(method < Regs::NUM_REGS,
+               "Invalid Maxwell3D register, increase the size of the Regs structure");
+
     if (debug_context) {
         debug_context->OnEvent(Tegra::DebugContext::Event::MaxwellCommandLoaded, nullptr);
     }
@@ -75,6 +71,10 @@ void Maxwell3D::WriteReg(u32 method, u32 value, u32 remaining_params) {
     regs.reg_array[method] = value;
 
     switch (method) {
+    case MAXWELL3D_REG_INDEX(macros.data): {
+        ProcessMacroUpload(value);
+        break;
+    }
     case MAXWELL3D_REG_INDEX(code_address.code_address_high):
     case MAXWELL3D_REG_INDEX(code_address.code_address_low): {
         // Note: For some reason games (like Puyo Puyo Tetris) seem to write 0 to the CODE_ADDRESS
@@ -141,17 +141,48 @@ void Maxwell3D::WriteReg(u32 method, u32 value, u32 remaining_params) {
     }
 }
 
+void Maxwell3D::ProcessMacroUpload(u32 data) {
+    // Store the uploaded macro code to interpret them when they're called.
+    auto& macro = uploaded_macros[regs.macros.entry * 2 + MacroRegistersStart];
+    macro.push_back(data);
+}
+
 void Maxwell3D::ProcessQueryGet() {
     GPUVAddr sequence_address = regs.query.QueryAddress();
     // Since the sequence address is given as a GPU VAddr, we have to convert it to an application
     // VAddr before writing.
-    VAddr address = memory_manager.PhysicalToVirtualAddress(sequence_address);
+    boost::optional<VAddr> address = memory_manager.GpuToCpuAddress(sequence_address);
+
+    // TODO(Subv): Support the other query units.
+    ASSERT_MSG(regs.query.query_get.unit == Regs::QueryUnit::Crop,
+               "Units other than CROP are unimplemented");
+    ASSERT_MSG(regs.query.query_get.short_query,
+               "Writing the entire query result structure is unimplemented");
+
+    u32 value = Memory::Read32(*address);
+    u32 result = 0;
+
+    // TODO(Subv): Support the other query variables
+    switch (regs.query.query_get.select) {
+    case Regs::QuerySelect::Zero:
+        result = 0;
+        break;
+    default:
+        UNIMPLEMENTED_MSG("Unimplemented query select type %u",
+                          static_cast<u32>(regs.query.query_get.select.Value()));
+    }
+
+    // TODO(Subv): Research and implement how query sync conditions work.
 
     switch (regs.query.query_get.mode) {
-    case Regs::QueryMode::Write: {
+    case Regs::QueryMode::Write:
+    case Regs::QueryMode::Write2: {
         // Write the current query sequence to the sequence address.
         u32 sequence = regs.query.query_sequence;
-        Memory::Write32(address, sequence);
+        Memory::Write32(*address, sequence);
+
+        // TODO(Subv): Write the proper query response structure to the address when not using short
+        // mode.
         break;
     }
     default:
@@ -161,8 +192,8 @@ void Maxwell3D::ProcessQueryGet() {
 }
 
 void Maxwell3D::DrawArrays() {
-    LOG_DEBUG(HW_GPU, "called, topology=%d, count=%d", regs.draw.topology.Value(),
-              regs.vertex_buffer.count);
+    NGLOG_DEBUG(HW_GPU, "called, topology={}, count={}",
+                static_cast<u32>(regs.draw.topology.Value()), regs.vertex_buffer.count);
     ASSERT_MSG(!(regs.index_array.count && regs.vertex_buffer.count), "Both indexed and direct?");
 
     auto debug_context = Core::System::GetInstance().GetGPUDebugContext();
@@ -200,10 +231,10 @@ void Maxwell3D::ProcessCBData(u32 value) {
     // Don't allow writing past the end of the buffer.
     ASSERT(regs.const_buffer.cb_pos + sizeof(u32) <= regs.const_buffer.cb_size);
 
-    VAddr address =
-        memory_manager.PhysicalToVirtualAddress(buffer_address + regs.const_buffer.cb_pos);
+    boost::optional<VAddr> address =
+        memory_manager.GpuToCpuAddress(buffer_address + regs.const_buffer.cb_pos);
 
-    Memory::Write32(address, value);
+    Memory::Write32(*address, value);
 
     // Increment the current buffer position.
     regs.const_buffer.cb_pos = regs.const_buffer.cb_pos + 4;
@@ -213,10 +244,10 @@ Texture::TICEntry Maxwell3D::GetTICEntry(u32 tic_index) const {
     GPUVAddr tic_base_address = regs.tic.TICAddress();
 
     GPUVAddr tic_address_gpu = tic_base_address + tic_index * sizeof(Texture::TICEntry);
-    VAddr tic_address_cpu = memory_manager.PhysicalToVirtualAddress(tic_address_gpu);
+    boost::optional<VAddr> tic_address_cpu = memory_manager.GpuToCpuAddress(tic_address_gpu);
 
     Texture::TICEntry tic_entry;
-    Memory::ReadBlock(tic_address_cpu, &tic_entry, sizeof(Texture::TICEntry));
+    Memory::ReadBlock(*tic_address_cpu, &tic_entry, sizeof(Texture::TICEntry));
 
     ASSERT_MSG(tic_entry.header_version == Texture::TICHeaderVersion::BlockLinear ||
                    tic_entry.header_version == Texture::TICHeaderVersion::Pitch,
@@ -243,10 +274,10 @@ Texture::TSCEntry Maxwell3D::GetTSCEntry(u32 tsc_index) const {
     GPUVAddr tsc_base_address = regs.tsc.TSCAddress();
 
     GPUVAddr tsc_address_gpu = tsc_base_address + tsc_index * sizeof(Texture::TSCEntry);
-    VAddr tsc_address_cpu = memory_manager.PhysicalToVirtualAddress(tsc_address_gpu);
+    boost::optional<VAddr> tsc_address_cpu = memory_manager.GpuToCpuAddress(tsc_address_gpu);
 
     Texture::TSCEntry tsc_entry;
-    Memory::ReadBlock(tsc_address_cpu, &tsc_entry, sizeof(Texture::TSCEntry));
+    Memory::ReadBlock(*tsc_address_cpu, &tsc_entry, sizeof(Texture::TSCEntry));
     return tsc_entry;
 }
 
@@ -268,7 +299,7 @@ std::vector<Texture::FullTextureInfo> Maxwell3D::GetStageTextures(Regs::ShaderSt
          current_texture < tex_info_buffer_end; current_texture += sizeof(Texture::TextureHandle)) {
 
         Texture::TextureHandle tex_handle{
-            Memory::Read32(memory_manager.PhysicalToVirtualAddress(current_texture))};
+            Memory::Read32(*memory_manager.GpuToCpuAddress(current_texture))};
 
         Texture::FullTextureInfo tex_info{};
         // TODO(Subv): Use the shader to determine which textures are actually accessed.

src/video_core/engines/maxwell_3d.h

@@ -31,7 +31,7 @@ public:
     /// Register structure of the Maxwell3D engine.
     /// TODO(Subv): This structure will need to be made bigger as more registers are discovered.
     struct Regs {
-        static constexpr size_t NUM_REGS = 0xE36;
+        static constexpr size_t NUM_REGS = 0xE00;
 
         static constexpr size_t NumRenderTargets = 8;
         static constexpr size_t NumViewports = 16;
@@ -46,6 +46,29 @@ public:
         enum class QueryMode : u32 {
             Write = 0,
             Sync = 1,
+            // TODO(Subv): It is currently unknown what the difference between method 2 and method 0
+            // is.
+            Write2 = 2,
+        };
+
+        enum class QueryUnit : u32 {
+            VFetch = 1,
+            VP = 2,
+            Rast = 4,
+            StrmOut = 5,
+            GP = 6,
+            ZCull = 7,
+            Prop = 10,
+            Crop = 15,
+        };
+
+        enum class QuerySelect : u32 {
+            Zero = 0,
+        };
+
+        enum class QuerySyncCondition : u32 {
+            NotEqual = 0,
+            GreaterThan = 1,
         };
 
         enum class ShaderProgram : u32 {
@@ -299,7 +322,15 @@ public:
 
         union {
             struct {
-                INSERT_PADDING_WORDS(0x200);
+                INSERT_PADDING_WORDS(0x45);
+
+                struct {
+                    INSERT_PADDING_WORDS(1);
+                    u32 data;
+                    u32 entry;
+                } macros;
+
+                INSERT_PADDING_WORDS(0x1B8);
 
                 struct {
                     u32 address_high;
@@ -476,7 +507,10 @@ public:
                         u32 raw;
                         BitField<0, 2, QueryMode> mode;
                         BitField<4, 1, u32> fence;
-                        BitField<12, 4, u32> unit;
+                        BitField<12, 4, QueryUnit> unit;
+                        BitField<16, 1, QuerySyncCondition> sync_cond;
+                        BitField<23, 5, QuerySelect> select;
+                        BitField<28, 1, u32> short_query;
                     } query_get;
 
                     GPUVAddr QueryAddress() const {
@@ -500,6 +534,11 @@ public:
                         return static_cast<GPUVAddr>((static_cast<GPUVAddr>(start_high) << 32) |
                                                      start_low);
                     }
+
+                    bool IsEnabled() const {
+                        return enable != 0 && StartAddress() != 0;
+                    }
+
                 } vertex_array[NumVertexArrays];
 
                 Blend blend;
@@ -574,7 +613,7 @@ public:
                     u32 size[MaxShaderStage];
                 } tex_info_buffers;
 
-                INSERT_PADDING_WORDS(0x102);
+                INSERT_PADDING_WORDS(0xCC);
             };
             std::array<u32, NUM_REGS> reg_array;
         };
@@ -606,9 +645,6 @@ public:
     /// Write the value to the register identified by method.
     void WriteReg(u32 method, u32 value, u32 remaining_params);
 
-    /// Uploads the code for a GPU macro program associated with the specified entry.
-    void SubmitMacroCode(u32 entry, std::vector<u32> code);
-
     /// Returns a list of enabled textures for the specified shader stage.
     std::vector<Texture::FullTextureInfo> GetStageTextures(Regs::ShaderStage stage) const;
 
@@ -639,6 +675,9 @@ private:
      */
     void CallMacroMethod(u32 method, std::vector<u32> parameters);
 
+    /// Handles writes to the macro uploading registers.
+    void ProcessMacroUpload(u32 data);
+
     /// Handles a write to the QUERY_GET register.
     void ProcessQueryGet();
 
@@ -656,6 +695,7 @@ private:
     static_assert(offsetof(Maxwell3D::Regs, field_name) == position * 4,                           \
                   "Field " #field_name " has invalid position")
 
+ASSERT_REG_POSITION(macros, 0x45);
 ASSERT_REG_POSITION(rt, 0x200);
 ASSERT_REG_POSITION(viewport_transform[0], 0x280);
 ASSERT_REG_POSITION(viewport, 0x300);

src/video_core/engines/shader_bytecode.h

@@ -4,15 +4,24 @@
 
 #pragma once
 
+#include <bitset>
 #include <cstring>
 #include <map>
 #include <string>
+#include <vector>
+
+#include <boost/optional.hpp>
+
 #include "common/bit_field.h"
+#include "common/common_types.h"
 
 namespace Tegra {
 namespace Shader {
 
 struct Register {
+    // Register 255 is special cased to always be 0
+    static constexpr size_t ZeroIndex = 255;
+
     constexpr Register() = default;
 
     constexpr Register(u64 value) : value(value) {}
@@ -86,181 +95,12 @@ union Uniform {
     BitField<34, 5, u64> index;
 };
 
-union OpCode {
-    enum class Id : u64 {
-        TEXS = 0x6C,
-        IPA = 0xE0,
-        FMUL32_IMM = 0x1E,
-        FFMA_IMM = 0x65,
-        FFMA_CR = 0x93,
-        FFMA_RC = 0xA3,
-        FFMA_RR = 0xB3,
-
-        FADD_C = 0x98B,
-        FMUL_C = 0x98D,
-        MUFU = 0xA10,
-        FADD_R = 0xB8B,
-        FMUL_R = 0xB8D,
-        LD_A = 0x1DFB,
-        ST_A = 0x1DFE,
-
-        FSETP_R = 0x5BB,
-        FSETP_C = 0x4BB,
-        EXIT = 0xE30,
-        KIL = 0xE33,
-
-        FMUL_IMM = 0x70D,
-        FMUL_IMM_x = 0x72D,
-        FADD_IMM = 0x70B,
-        FADD_IMM_x = 0x72B,
-    };
-
-    enum class Type {
-        Trivial,
-        Arithmetic,
-        Ffma,
-        Flow,
-        Memory,
-        Unknown,
-    };
-
-    struct Info {
-        Type type;
-        std::string name;
-    };
-
-    OpCode() = default;
-
-    constexpr OpCode(Id value) : value(static_cast<u64>(value)) {}
-
-    constexpr OpCode(u64 value) : value{value} {}
-
-    constexpr Id EffectiveOpCode() const {
-        switch (op1) {
-        case Id::TEXS:
-            return op1;
-        }
-
-        switch (op2) {
-        case Id::IPA:
-        case Id::FMUL32_IMM:
-            return op2;
-        }
-
-        switch (op3) {
-        case Id::FFMA_IMM:
-        case Id::FFMA_CR:
-        case Id::FFMA_RC:
-        case Id::FFMA_RR:
-            return op3;
-        }
-
-        switch (op4) {
-        case Id::EXIT:
-        case Id::FSETP_R:
-        case Id::FSETP_C:
-        case Id::KIL:
-            return op4;
-        }
-
-        switch (op5) {
-        case Id::MUFU:
-        case Id::LD_A:
-        case Id::ST_A:
-        case Id::FADD_R:
-        case Id::FADD_C:
-        case Id::FMUL_R:
-        case Id::FMUL_C:
-            return op5;
-
-        case Id::FMUL_IMM:
-        case Id::FMUL_IMM_x:
-            return Id::FMUL_IMM;
-
-        case Id::FADD_IMM:
-        case Id::FADD_IMM_x:
-            return Id::FADD_IMM;
-        }
-
-        return static_cast<Id>(value);
-    }
-
-    static const Info& GetInfo(const OpCode& opcode) {
-        static const std::map<Id, Info> info_table{BuildInfoTable()};
-        const auto& search{info_table.find(opcode.EffectiveOpCode())};
-        if (search != info_table.end()) {
-            return search->second;
-        }
-
-        static const Info unknown{Type::Unknown, "UNK"};
-        return unknown;
-    }
-
-    constexpr operator Id() const {
-        return static_cast<Id>(value);
-    }
-
-    constexpr OpCode operator<<(size_t bits) const {
-        return value << bits;
-    }
-
-    constexpr OpCode operator>>(size_t bits) const {
-        return value >> bits;
-    }
-
-    template <typename T>
-    constexpr u64 operator-(const T& oth) const {
-        return value - oth;
-    }
-
-    constexpr u64 operator&(const OpCode& oth) const {
-        return value & oth.value;
-    }
-
-    constexpr u64 operator~() const {
-        return ~value;
-    }
-
-    static std::map<Id, Info> BuildInfoTable() {
-        std::map<Id, Info> info_table;
-        info_table[Id::TEXS] = {Type::Memory, "texs"};
-        info_table[Id::LD_A] = {Type::Memory, "ld_a"};
-        info_table[Id::ST_A] = {Type::Memory, "st_a"};
-        info_table[Id::MUFU] = {Type::Arithmetic, "mufu"};
-        info_table[Id::FFMA_IMM] = {Type::Ffma, "ffma_imm"};
-        info_table[Id::FFMA_CR] = {Type::Ffma, "ffma_cr"};
-        info_table[Id::FFMA_RC] = {Type::Ffma, "ffma_rc"};
-        info_table[Id::FFMA_RR] = {Type::Ffma, "ffma_rr"};
-        info_table[Id::FADD_R] = {Type::Arithmetic, "fadd_r"};
-        info_table[Id::FADD_C] = {Type::Arithmetic, "fadd_c"};
-        info_table[Id::FADD_IMM] = {Type::Arithmetic, "fadd_imm"};
-        info_table[Id::FMUL_R] = {Type::Arithmetic, "fmul_r"};
-        info_table[Id::FMUL_C] = {Type::Arithmetic, "fmul_c"};
-        info_table[Id::FMUL_IMM] = {Type::Arithmetic, "fmul_imm"};
-        info_table[Id::FMUL32_IMM] = {Type::Arithmetic, "fmul32_imm"};
-        info_table[Id::FSETP_C] = {Type::Arithmetic, "fsetp_c"};
-        info_table[Id::FSETP_R] = {Type::Arithmetic, "fsetp_r"};
-        info_table[Id::EXIT] = {Type::Trivial, "exit"};
-        info_table[Id::IPA] = {Type::Trivial, "ipa"};
-        info_table[Id::KIL] = {Type::Flow, "kil"};
-        return info_table;
-    }
-
-    BitField<57, 7, Id> op1;
-    BitField<56, 8, Id> op2;
-    BitField<55, 9, Id> op3;
-    BitField<52, 12, Id> op4;
-    BitField<51, 13, Id> op5;
-    u64 value{};
-};
-static_assert(sizeof(OpCode) == 0x8, "Incorrect structure size");
-
 } // namespace Shader
 } // namespace Tegra
 
 namespace std {
 
-// TODO(bunne): The below is forbidden by the C++ standard, but works fine. See #330.
+// TODO(bunnei): The below is forbidden by the C++ standard, but works fine. See #330.
 template <>
 struct make_unsigned<Tegra::Shader::Attribute> {
     using type = Tegra::Shader::Attribute;
@@ -271,11 +111,6 @@ struct make_unsigned<Tegra::Shader::Register> {
     using type = Tegra::Shader::Register;
 };
 
-template <>
-struct make_unsigned<Tegra::Shader::OpCode> {
-    using type = Tegra::Shader::OpCode;
-};
-
 } // namespace std
 
 namespace Tegra {
@@ -283,7 +118,23 @@ namespace Shader {
 
 enum class Pred : u64 {
     UnusedIndex = 0x7,
-    NeverExecute = 0xf,
+    NeverExecute = 0xF,
+};
+
+enum class PredCondition : u64 {
+    LessThan = 1,
+    Equal = 2,
+    LessEqual = 3,
+    GreaterThan = 4,
+    NotEqual = 5,
+    GreaterEqual = 6,
+    // TODO(Subv): Other condition types
+};
+
+enum class PredOperation : u64 {
+    And = 0,
+    Or = 1,
+    Xor = 2,
 };
 
 enum class SubOp : u64 {
@@ -298,18 +149,24 @@ enum class SubOp : u64 {
 
 union Instruction {
     Instruction& operator=(const Instruction& instr) {
-        hex = instr.hex;
+        value = instr.value;
         return *this;
     }
 
-    OpCode opcode;
+    constexpr Instruction(u64 value) : value{value} {}
+
     BitField<0, 8, Register> gpr0;
     BitField<8, 8, Register> gpr8;
-    BitField<16, 4, Pred> pred;
+    union {
+        BitField<16, 4, Pred> full_pred;
+        BitField<16, 3, u64> pred_index;
+    } pred;
+    BitField<19, 1, u64> negate_pred;
     BitField<20, 8, Register> gpr20;
     BitField<20, 7, SubOp> sub_op;
     BitField<28, 8, Register> gpr28;
     BitField<39, 8, Register> gpr39;
+    BitField<48, 16, u64> opcode;
 
     union {
         BitField<20, 19, u64> imm20_19;
@@ -343,6 +200,20 @@ union Instruction {
         BitField<49, 1, u64> negate_c;
     } ffma;
 
+    union {
+        BitField<0, 3, u64> pred0;
+        BitField<3, 3, u64> pred3;
+        BitField<7, 1, u64> abs_a;
+        BitField<39, 3, u64> pred39;
+        BitField<42, 1, u64> neg_pred;
+        BitField<43, 1, u64> neg_a;
+        BitField<44, 1, u64> abs_b;
+        BitField<45, 2, PredOperation> op;
+        BitField<47, 1, u64> ftz;
+        BitField<48, 4, PredCondition> cond;
+        BitField<56, 1, u64> neg_b;
+    } fsetp;
+
     BitField<61, 1, u64> is_b_imm;
     BitField<60, 1, u64> is_b_gpr;
     BitField<59, 1, u64> is_c_gpr;
@@ -351,11 +222,218 @@ union Instruction {
     Uniform uniform;
     Sampler sampler;
 
-    u64 hex;
+    u64 value;
 };
 static_assert(sizeof(Instruction) == 0x8, "Incorrect structure size");
 static_assert(std::is_standard_layout<Instruction>::value,
               "Structure does not have standard layout");
 
+class OpCode {
+public:
+    enum class Id {
+        KIL,
+        LD_A,
+        ST_A,
+        TEXQ, // Texture Query
+        TEXS, // Texture Fetch with scalar/non-vec4 source/destinations
+        TLDS, // Texture Load with scalar/non-vec4 source/destinations
+        EXIT,
+        IPA,
+        FFMA_IMM, // Fused Multiply and Add
+        FFMA_CR,
+        FFMA_RC,
+        FFMA_RR,
+        FADD_C,
+        FADD_R,
+        FADD_IMM,
+        FMUL_C,
+        FMUL_R,
+        FMUL_IMM,
+        FMUL32_IMM,
+        MUFU, // Multi-Function Operator
+        RRO,  // Range Reduction Operator
+        F2F_C,
+        F2F_R,
+        F2F_IMM,
+        F2I_C,
+        F2I_R,
+        F2I_IMM,
+        I2F_C,
+        I2F_R,
+        I2F_IMM,
+        LOP32I,
+        MOV_C,
+        MOV_R,
+        MOV_IMM,
+        MOV32I,
+        SHR_C,
+        SHR_R,
+        SHR_IMM,
+        FSETP_C, // Set Predicate
+        FSETP_R,
+        FSETP_IMM,
+        ISETP_C,
+        ISETP_IMM,
+        ISETP_R,
+    };
+
+    enum class Type {
+        Trivial,
+        Arithmetic,
+        Ffma,
+        Flow,
+        Memory,
+        FloatPredicate,
+        IntegerPredicate,
+        Unknown,
+    };
+
+    class Matcher {
+    public:
+        Matcher(const char* const name, u16 mask, u16 expected, OpCode::Id id, OpCode::Type type)
+            : name{name}, mask{mask}, expected{expected}, id{id}, type{type} {}
+
+        const char* GetName() const {
+            return name;
+        }
+
+        u16 GetMask() const {
+            return mask;
+        }
+
+        Id GetId() const {
+            return id;
+        }
+
+        Type GetType() const {
+            return type;
+        }
+
+        /**
+         * Tests to see if the given instruction is the instruction this matcher represents.
+         * @param instruction The instruction to test
+         * @returns true if the given instruction matches.
+         */
+        bool Matches(u16 instruction) const {
+            return (instruction & mask) == expected;
+        }
+
+    private:
+        const char* name;
+        u16 mask;
+        u16 expected;
+        Id id;
+        Type type;
+    };
+
+    static boost::optional<const Matcher&> Decode(Instruction instr) {
+        static const auto table{GetDecodeTable()};
+
+        const auto matches_instruction = [instr](const auto& matcher) {
+            return matcher.Matches(static_cast<u16>(instr.opcode));
+        };
+
+        auto iter = std::find_if(table.begin(), table.end(), matches_instruction);
+        return iter != table.end() ? boost::optional<const Matcher&>(*iter) : boost::none;
+    }
+
+private:
+    struct Detail {
+    private:
+        static constexpr size_t opcode_bitsize = 16;
+
+        /**
+         * Generates the mask and the expected value after masking from a given bitstring.
+         * A '0' in a bitstring indicates that a zero must be present at that bit position.
+         * A '1' in a bitstring indicates that a one must be present at that bit position.
+         */
+        static auto GetMaskAndExpect(const char* const bitstring) {
+            u16 mask = 0, expect = 0;
+            for (size_t i = 0; i < opcode_bitsize; i++) {
+                const size_t bit_position = opcode_bitsize - i - 1;
+                switch (bitstring[i]) {
+                case '0':
+                    mask |= 1 << bit_position;
+                    break;
+                case '1':
+                    expect |= 1 << bit_position;
+                    mask |= 1 << bit_position;
+                    break;
+                default:
+                    // Ignore
+                    break;
+                }
+            }
+            return std::make_tuple(mask, expect);
+        }
+
+    public:
+        /// Creates a matcher that can match and parse instructions based on bitstring.
+        static auto GetMatcher(const char* const bitstring, OpCode::Id op, OpCode::Type type,
+                               const char* const name) {
+            const auto mask_expect = GetMaskAndExpect(bitstring);
+            return Matcher(name, std::get<0>(mask_expect), std::get<1>(mask_expect), op, type);
+        }
+    };
+
+    static std::vector<Matcher> GetDecodeTable() {
+        std::vector<Matcher> table = {
+#define INST(bitstring, op, type, name) Detail::GetMatcher(bitstring, op, type, name)
+            INST("111000110011----", Id::KIL, Type::Flow, "KIL"),
+            INST("1110111111011---", Id::LD_A, Type::Memory, "LD_A"),
+            INST("1110111111110---", Id::ST_A, Type::Memory, "ST_A"),
+            INST("1101111101001---", Id::TEXQ, Type::Memory, "TEXQ"),
+            INST("1101100---------", Id::TEXS, Type::Memory, "TEXS"),
+            INST("1101101---------", Id::TLDS, Type::Memory, "TLDS"),
+            INST("111000110000----", Id::EXIT, Type::Trivial, "EXIT"),
+            INST("11100000--------", Id::IPA, Type::Trivial, "IPA"),
+            INST("001100101-------", Id::FFMA_IMM, Type::Ffma, "FFMA_IMM"),
+            INST("010010011-------", Id::FFMA_CR, Type::Ffma, "FFMA_CR"),
+            INST("010100011-------", Id::FFMA_RC, Type::Ffma, "FFMA_RC"),
+            INST("010110011-------", Id::FFMA_RR, Type::Ffma, "FFMA_RR"),
+            INST("0100110001011---", Id::FADD_C, Type::Arithmetic, "FADD_C"),
+            INST("0101110001011---", Id::FADD_R, Type::Arithmetic, "FADD_R"),
+            INST("0011100-01011---", Id::FADD_IMM, Type::Arithmetic, "FADD_IMM"),
+            INST("0100110001101---", Id::FMUL_C, Type::Arithmetic, "FMUL_C"),
+            INST("0101110001101---", Id::FMUL_R, Type::Arithmetic, "FMUL_R"),
+            INST("0011100-01101---", Id::FMUL_IMM, Type::Arithmetic, "FMUL_IMM"),
+            INST("00011110--------", Id::FMUL32_IMM, Type::Arithmetic, "FMUL32_IMM"),
+            INST("0101000010000---", Id::MUFU, Type::Arithmetic, "MUFU"),
+            INST("0101110010010---", Id::RRO, Type::Arithmetic, "RRO"),
+            INST("0100110010101---", Id::F2F_C, Type::Arithmetic, "F2F_C"),
+            INST("0101110010101---", Id::F2F_R, Type::Arithmetic, "F2F_R"),
+            INST("0011100-10101---", Id::F2F_IMM, Type::Arithmetic, "F2F_IMM"),
+            INST("0100110010110---", Id::F2I_C, Type::Arithmetic, "F2I_C"),
+            INST("0101110010110---", Id::F2I_R, Type::Arithmetic, "F2I_R"),
+            INST("0011100-10110---", Id::F2I_IMM, Type::Arithmetic, "F2I_IMM"),
+            INST("0100110010111---", Id::I2F_C, Type::Arithmetic, "I2F_C"),
+            INST("0101110010111---", Id::I2F_R, Type::Arithmetic, "I2F_R"),
+            INST("0011100-10111---", Id::I2F_IMM, Type::Arithmetic, "I2F_IMM"),
+            INST("000001----------", Id::LOP32I, Type::Arithmetic, "LOP32I"),
+            INST("0100110010011---", Id::MOV_C, Type::Arithmetic, "MOV_C"),
+            INST("0101110010011---", Id::MOV_R, Type::Arithmetic, "MOV_R"),
+            INST("0011100-10011---", Id::MOV_IMM, Type::Arithmetic, "MOV_IMM"),
+            INST("000000010000----", Id::MOV32I, Type::Arithmetic, "MOV32I"),
+            INST("0100110000101---", Id::SHR_C, Type::Arithmetic, "SHR_C"),
+            INST("0101110000101---", Id::SHR_R, Type::Arithmetic, "SHR_R"),
+            INST("0011100-00101---", Id::SHR_IMM, Type::Arithmetic, "SHR_IMM"),
+            INST("010010111011----", Id::FSETP_C, Type::FloatPredicate, "FSETP_C"),
+            INST("010110111011----", Id::FSETP_R, Type::FloatPredicate, "FSETP_R"),
+            INST("0011011-1011----", Id::FSETP_IMM, Type::FloatPredicate, "FSETP_IMM"),
+            INST("010010110110----", Id::ISETP_C, Type::IntegerPredicate, "ISETP_C"),
+            INST("010110110110----", Id::ISETP_R, Type::IntegerPredicate, "ISETP_R"),
+            INST("0011011-0110----", Id::ISETP_IMM, Type::IntegerPredicate, "ISETP_IMM"),
+        };
+#undef INST
+        std::stable_sort(table.begin(), table.end(), [](const auto& a, const auto& b) {
+            // If a matcher has more bits in its mask it is more specific, so it
+            // should come first.
+            return std::bitset<16>(a.GetMask()).count() > std::bitset<16>(b.GetMask()).count();
+        });
+
+        return table;
+    }
+};
+
 } // namespace Shader
 } // namespace Tegra

src/video_core/gpu.cpp

@@ -12,7 +12,7 @@ namespace Tegra {
 GPU::GPU() {
     memory_manager = std::make_unique<MemoryManager>();
     maxwell_3d = std::make_unique<Engines::Maxwell3D>(*memory_manager);
-    fermi_2d = std::make_unique<Engines::Fermi2D>();
+    fermi_2d = std::make_unique<Engines::Fermi2D>(*memory_manager);
     maxwell_compute = std::make_unique<Engines::MaxwellCompute>();
 }
 
@@ -22,4 +22,16 @@ const Tegra::Engines::Maxwell3D& GPU::Get3DEngine() const {
     return *maxwell_3d;
 }
 
+u32 RenderTargetBytesPerPixel(RenderTargetFormat format) {
+    ASSERT(format != RenderTargetFormat::NONE);
+
+    switch (format) {
+    case RenderTargetFormat::RGBA8_UNORM:
+    case RenderTargetFormat::RGB10_A2_UNORM:
+        return 4;
+    default:
+        UNIMPLEMENTED_MSG("Unimplemented render target format %u", static_cast<u32>(format));
+    }
+}
+
 } // namespace Tegra

src/video_core/gpu.h

@@ -21,6 +21,9 @@ enum class RenderTargetFormat : u32 {
     RGBA8_SRGB = 0xD6,
 };
 
+/// Returns the number of bytes per pixel of each rendertarget format.
+u32 RenderTargetBytesPerPixel(RenderTargetFormat format);
+
 class DebugContext;
 
 /**
@@ -86,8 +89,6 @@ public:
     }
 
 private:
-    static constexpr u32 InvalidGraphMacroEntry = 0xFFFFFFFF;
-
     /// Writes a single register in the engine bound to the specified subchannel
     void WriteReg(u32 method, u32 subchannel, u32 value, u32 remaining_params);
 
@@ -100,11 +101,6 @@ private:
     std::unique_ptr<Engines::Fermi2D> fermi_2d;
     /// Compute engine
     std::unique_ptr<Engines::MaxwellCompute> maxwell_compute;
-
-    /// Entry of the macro that is currently being uploaded
-    u32 current_macro_entry = InvalidGraphMacroEntry;
-    /// Code being uploaded for the current macro
-    std::vector<u32> current_macro_code;
 };
 
 } // namespace Tegra

src/video_core/memory_manager.cpp

@@ -2,109 +2,118 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include "common/alignment.h"
 #include "common/assert.h"
 #include "video_core/memory_manager.h"
 
 namespace Tegra {
 
-PAddr MemoryManager::AllocateSpace(u64 size, u64 align) {
-    boost::optional<PAddr> paddr = FindFreeBlock(size, align);
-    ASSERT(paddr);
+GPUVAddr MemoryManager::AllocateSpace(u64 size, u64 align) {
+    boost::optional<GPUVAddr> gpu_addr = FindFreeBlock(size, align);
+    ASSERT(gpu_addr);
 
-    for (u64 offset = 0; offset < size; offset += Memory::PAGE_SIZE) {
-        PageSlot(*paddr + offset) = static_cast<u64>(PageStatus::Allocated);
+    for (u64 offset = 0; offset < size; offset += PAGE_SIZE) {
+        ASSERT(PageSlot(*gpu_addr + offset) == static_cast<u64>(PageStatus::Unmapped));
+        PageSlot(*gpu_addr + offset) = static_cast<u64>(PageStatus::Allocated);
     }
 
-    return *paddr;
+    return *gpu_addr;
 }
 
-PAddr MemoryManager::AllocateSpace(PAddr paddr, u64 size, u64 align) {
-    for (u64 offset = 0; offset < size; offset += Memory::PAGE_SIZE) {
-        if (IsPageMapped(paddr + offset)) {
-            return AllocateSpace(size, align);
-        }
+GPUVAddr MemoryManager::AllocateSpace(GPUVAddr gpu_addr, u64 size, u64 align) {
+    for (u64 offset = 0; offset < size; offset += PAGE_SIZE) {
+        ASSERT(PageSlot(gpu_addr + offset) == static_cast<u64>(PageStatus::Unmapped));
+        PageSlot(gpu_addr + offset) = static_cast<u64>(PageStatus::Allocated);
     }
 
-    for (u64 offset = 0; offset < size; offset += Memory::PAGE_SIZE) {
-        PageSlot(paddr + offset) = static_cast<u64>(PageStatus::Allocated);
-    }
-
-    return paddr;
+    return gpu_addr;
 }
 
-PAddr MemoryManager::MapBufferEx(VAddr vaddr, u64 size) {
-    vaddr &= ~Memory::PAGE_MASK;
+GPUVAddr MemoryManager::MapBufferEx(VAddr cpu_addr, u64 size) {
+    boost::optional<GPUVAddr> gpu_addr = FindFreeBlock(size, PAGE_SIZE);
+    ASSERT(gpu_addr);
 
-    boost::optional<PAddr> paddr = FindFreeBlock(size);
-    ASSERT(paddr);
-
-    for (u64 offset = 0; offset < size; offset += Memory::PAGE_SIZE) {
-        PageSlot(*paddr + offset) = vaddr + offset;
+    for (u64 offset = 0; offset < size; offset += PAGE_SIZE) {
+        ASSERT(PageSlot(*gpu_addr + offset) == static_cast<u64>(PageStatus::Unmapped));
+        PageSlot(*gpu_addr + offset) = cpu_addr + offset;
     }
 
-    return *paddr;
+    MappedRegion region{cpu_addr, *gpu_addr, size};
+    mapped_regions.push_back(region);
+
+    return *gpu_addr;
 }
 
-PAddr MemoryManager::MapBufferEx(VAddr vaddr, PAddr paddr, u64 size) {
-    vaddr &= ~Memory::PAGE_MASK;
-    paddr &= ~Memory::PAGE_MASK;
+GPUVAddr MemoryManager::MapBufferEx(VAddr cpu_addr, GPUVAddr gpu_addr, u64 size) {
+    ASSERT((gpu_addr & PAGE_MASK) == 0);
 
-    for (u64 offset = 0; offset < size; offset += Memory::PAGE_SIZE) {
-        if (PageSlot(paddr + offset) != static_cast<u64>(PageStatus::Allocated)) {
-            return MapBufferEx(vaddr, size);
-        }
+    for (u64 offset = 0; offset < size; offset += PAGE_SIZE) {
+        ASSERT(PageSlot(gpu_addr + offset) == static_cast<u64>(PageStatus::Allocated));
+        PageSlot(gpu_addr + offset) = cpu_addr + offset;
     }
 
-    for (u64 offset = 0; offset < size; offset += Memory::PAGE_SIZE) {
-        PageSlot(paddr + offset) = vaddr + offset;
-    }
+    MappedRegion region{cpu_addr, gpu_addr, size};
+    mapped_regions.push_back(region);
 
-    return paddr;
+    return gpu_addr;
 }
 
-boost::optional<PAddr> MemoryManager::FindFreeBlock(u64 size, u64 align) {
-    PAddr paddr{};
-    u64 free_space{};
-    align = (align + Memory::PAGE_MASK) & ~Memory::PAGE_MASK;
+boost::optional<GPUVAddr> MemoryManager::FindFreeBlock(u64 size, u64 align) {
+    GPUVAddr gpu_addr = 0;
+    u64 free_space = 0;
+    align = (align + PAGE_MASK) & ~PAGE_MASK;
 
-    while (paddr + free_space < MAX_ADDRESS) {
-        if (!IsPageMapped(paddr + free_space)) {
-            free_space += Memory::PAGE_SIZE;
+    while (gpu_addr + free_space < MAX_ADDRESS) {
+        if (!IsPageMapped(gpu_addr + free_space)) {
+            free_space += PAGE_SIZE;
             if (free_space >= size) {
-                return paddr;
+                return gpu_addr;
             }
         } else {
-            paddr += free_space + Memory::PAGE_SIZE;
+            gpu_addr += free_space + PAGE_SIZE;
             free_space = 0;
-            const u64 remainder{paddr % align};
-            if (!remainder) {
-                paddr = (paddr - remainder) + align;
-            }
+            gpu_addr = Common::AlignUp(gpu_addr, align);
         }
     }
 
     return {};
 }
 
-VAddr MemoryManager::PhysicalToVirtualAddress(PAddr paddr) {
-    VAddr base_addr = PageSlot(paddr);
+boost::optional<VAddr> MemoryManager::GpuToCpuAddress(GPUVAddr gpu_addr) {
+    VAddr base_addr = PageSlot(gpu_addr);
     ASSERT(base_addr != static_cast<u64>(PageStatus::Unmapped));
-    return base_addr + (paddr & Memory::PAGE_MASK);
+
+    if (base_addr == static_cast<u64>(PageStatus::Allocated)) {
+        return {};
+    }
+
+    return base_addr + (gpu_addr & PAGE_MASK);
 }
 
-bool MemoryManager::IsPageMapped(PAddr paddr) {
-    return PageSlot(paddr) != static_cast<u64>(PageStatus::Unmapped);
+std::vector<GPUVAddr> MemoryManager::CpuToGpuAddress(VAddr cpu_addr) const {
+    std::vector<GPUVAddr> results;
+    for (const auto& region : mapped_regions) {
+        if (cpu_addr >= region.cpu_addr && cpu_addr < (region.cpu_addr + region.size)) {
+            u64 offset = cpu_addr - region.cpu_addr;
+            results.push_back(region.gpu_addr + offset);
+        }
+    }
+    return results;
 }
 
-VAddr& MemoryManager::PageSlot(PAddr paddr) {
-    auto& block = page_table[(paddr >> (Memory::PAGE_BITS + PAGE_TABLE_BITS)) & PAGE_TABLE_MASK];
+bool MemoryManager::IsPageMapped(GPUVAddr gpu_addr) {
+    return PageSlot(gpu_addr) != static_cast<u64>(PageStatus::Unmapped);
+}
+
+VAddr& MemoryManager::PageSlot(GPUVAddr gpu_addr) {
+    auto& block = page_table[(gpu_addr >> (PAGE_BITS + PAGE_TABLE_BITS)) & PAGE_TABLE_MASK];
     if (!block) {
         block = std::make_unique<PageBlock>();
         for (unsigned index = 0; index < PAGE_BLOCK_SIZE; index++) {
             (*block)[index] = static_cast<u64>(PageStatus::Unmapped);
         }
     }
-    return (*block)[(paddr >> Memory::PAGE_BITS) & PAGE_BLOCK_MASK];
+    return (*block)[(gpu_addr >> PAGE_BITS) & PAGE_BLOCK_MASK];
 }
 
 } // namespace Tegra

src/video_core/memory_manager.h

@@ -6,8 +6,11 @@
 
 #include <array>
 #include <memory>
+#include <vector>
+
+#include <boost/optional.hpp>
+
 #include "common/common_types.h"
-#include "core/memory.h"
 
 namespace Tegra {
 
@@ -18,16 +21,21 @@ class MemoryManager final {
 public:
     MemoryManager() = default;
 
-    PAddr AllocateSpace(u64 size, u64 align);
-    PAddr AllocateSpace(PAddr paddr, u64 size, u64 align);
-    PAddr MapBufferEx(VAddr vaddr, u64 size);
-    PAddr MapBufferEx(VAddr vaddr, PAddr paddr, u64 size);
-    VAddr PhysicalToVirtualAddress(PAddr paddr);
+    GPUVAddr AllocateSpace(u64 size, u64 align);
+    GPUVAddr AllocateSpace(GPUVAddr gpu_addr, u64 size, u64 align);
+    GPUVAddr MapBufferEx(VAddr cpu_addr, u64 size);
+    GPUVAddr MapBufferEx(VAddr cpu_addr, GPUVAddr gpu_addr, u64 size);
+    boost::optional<VAddr> GpuToCpuAddress(GPUVAddr gpu_addr);
+    std::vector<GPUVAddr> CpuToGpuAddress(VAddr cpu_addr) const;
+
+    static constexpr u64 PAGE_BITS = 16;
+    static constexpr u64 PAGE_SIZE = 1 << PAGE_BITS;
+    static constexpr u64 PAGE_MASK = PAGE_SIZE - 1;
 
 private:
-    boost::optional<PAddr> FindFreeBlock(u64 size, u64 align = 1);
-    bool IsPageMapped(PAddr paddr);
-    VAddr& PageSlot(PAddr paddr);
+    boost::optional<GPUVAddr> FindFreeBlock(u64 size, u64 align = 1);
+    bool IsPageMapped(GPUVAddr gpu_addr);
+    VAddr& PageSlot(GPUVAddr gpu_addr);
 
     enum class PageStatus : u64 {
         Unmapped = 0xFFFFFFFFFFFFFFFFULL,
@@ -35,7 +43,7 @@ private:
     };
 
     static constexpr u64 MAX_ADDRESS{0x10000000000ULL};
-    static constexpr u64 PAGE_TABLE_BITS{14};
+    static constexpr u64 PAGE_TABLE_BITS{10};
     static constexpr u64 PAGE_TABLE_SIZE{1 << PAGE_TABLE_BITS};
     static constexpr u64 PAGE_TABLE_MASK{PAGE_TABLE_SIZE - 1};
     static constexpr u64 PAGE_BLOCK_BITS{14};
@@ -44,6 +52,14 @@ private:
 
     using PageBlock = std::array<VAddr, PAGE_BLOCK_SIZE>;
     std::array<std::unique_ptr<PageBlock>, PAGE_TABLE_SIZE> page_table{};
+
+    struct MappedRegion {
+        VAddr cpu_addr;
+        GPUVAddr gpu_addr;
+        u64 size;
+    };
+
+    std::vector<MappedRegion> mapped_regions;
 };
 
 } // namespace Tegra

src/video_core/rasterizer_interface.h

@@ -6,6 +6,7 @@
 
 #include "common/common_types.h"
 #include "video_core/gpu.h"
+#include "video_core/memory_manager.h"
 
 struct ScreenInfo;
 
@@ -25,14 +26,14 @@ public:
     virtual void FlushAll() = 0;
 
     /// Notify rasterizer that any caches of the specified region should be flushed to Switch memory
-    virtual void FlushRegion(VAddr addr, u64 size) = 0;
+    virtual void FlushRegion(Tegra::GPUVAddr addr, u64 size) = 0;
 
     /// Notify rasterizer that any caches of the specified region should be invalidated
-    virtual void InvalidateRegion(VAddr addr, u64 size) = 0;
+    virtual void InvalidateRegion(Tegra::GPUVAddr addr, u64 size) = 0;
 
     /// Notify rasterizer that any caches of the specified region should be flushed to Switch memory
     /// and invalidated
-    virtual void FlushAndInvalidateRegion(VAddr addr, u64 size) = 0;
+    virtual void FlushAndInvalidateRegion(Tegra::GPUVAddr addr, u64 size) = 0;
 
     /// Attempt to use a faster method to perform a display transfer with is_texture_copy = 0
     virtual bool AccelerateDisplayTransfer(const void* config) {

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -14,7 +14,6 @@
 #include "common/math_util.h"
 #include "common/microprofile.h"
 #include "common/scope_exit.h"
-#include "common/vector_math.h"
 #include "core/core.h"
 #include "core/hle/kernel/process.h"
 #include "core/settings.h"
@@ -117,7 +116,7 @@ RasterizerOpenGL::RasterizerOpenGL() {
 
     glEnable(GL_BLEND);
 
-    LOG_CRITICAL(Render_OpenGL, "Sync fixed function OpenGL state here!");
+    NGLOG_CRITICAL(Render_OpenGL, "Sync fixed function OpenGL state here!");
 }
 
 RasterizerOpenGL::~RasterizerOpenGL() {
@@ -128,7 +127,8 @@ RasterizerOpenGL::~RasterizerOpenGL() {
     }
 }
 
-void RasterizerOpenGL::SetupVertexArray(u8* array_ptr, GLintptr buffer_offset) {
+std::pair<u8*, GLintptr> RasterizerOpenGL::SetupVertexArrays(u8* array_ptr,
+                                                             GLintptr buffer_offset) {
     MICROPROFILE_SCOPE(OpenGL_VAO);
     const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
     const auto& memory_manager = Core::System().GetInstance().GPU().memory_manager;
@@ -137,43 +137,58 @@ void RasterizerOpenGL::SetupVertexArray(u8* array_ptr, GLintptr buffer_offset) {
     state.draw.vertex_buffer = stream_buffer->GetHandle();
     state.Apply();
 
-    // TODO(bunnei): Add support for 1+ vertex arrays
-    const auto& vertex_array{regs.vertex_array[0]};
-    const auto& vertex_array_limit{regs.vertex_array_limit[0]};
-    ASSERT_MSG(vertex_array.enable, "vertex array 0 is disabled?");
-    ASSERT_MSG(!vertex_array.divisor, "vertex array 0 divisor is unimplemented!");
-    for (unsigned index = 1; index < Maxwell::NumVertexArrays; ++index) {
-        ASSERT_MSG(!regs.vertex_array[index].enable, "vertex array %d is unimplemented!", index);
+    // Upload all guest vertex arrays sequentially to our buffer
+    for (u32 index = 0; index < Maxwell::NumVertexArrays; ++index) {
+        const auto& vertex_array = regs.vertex_array[index];
+        if (!vertex_array.IsEnabled())
+            continue;
+
+        const Tegra::GPUVAddr start = vertex_array.StartAddress();
+        const Tegra::GPUVAddr end = regs.vertex_array_limit[index].LimitAddress();
+
+        ASSERT(end > start);
+        u64 size = end - start + 1;
+
+        // Copy vertex array data
+        res_cache.FlushRegion(start, size, nullptr);
+        Memory::ReadBlock(*memory_manager->GpuToCpuAddress(start), array_ptr, size);
+
+        // Bind the vertex array to the buffer at the current offset.
+        glBindVertexBuffer(index, stream_buffer->GetHandle(), buffer_offset, vertex_array.stride);
+
+        ASSERT_MSG(vertex_array.divisor == 0, "Vertex buffer divisor unimplemented");
+
+        array_ptr += size;
+        buffer_offset += size;
     }
 
     // Use the vertex array as-is, assumes that the data is formatted correctly for OpenGL.
     // Enables the first 16 vertex attributes always, as we don't know which ones are actually used
-    // until shader time. Note, Tegra technically supports 32, but we're cappinig this to 16 for now
+    // until shader time. Note, Tegra technically supports 32, but we're capping this to 16 for now
     // to avoid OpenGL errors.
+    // TODO(Subv): Analyze the shader to identify which attributes are actually used and don't
+    // assume every shader uses them all.
     for (unsigned index = 0; index < 16; ++index) {
         auto& attrib = regs.vertex_attrib_format[index];
         NGLOG_DEBUG(HW_GPU, "vertex attrib {}, count={}, size={}, type={}, offset={}, normalize={}",
                     index, attrib.ComponentCount(), attrib.SizeString(), attrib.TypeString(),
                     attrib.offset.Value(), attrib.IsNormalized());
 
-        glVertexAttribPointer(index, attrib.ComponentCount(), MaxwellToGL::VertexType(attrib),
-                              attrib.IsNormalized() ? GL_TRUE : GL_FALSE, vertex_array.stride,
-                              reinterpret_cast<GLvoid*>(buffer_offset + attrib.offset));
+        auto& buffer = regs.vertex_array[attrib.buffer];
+        ASSERT(buffer.IsEnabled());
+
         glEnableVertexAttribArray(index);
+        glVertexAttribFormat(index, attrib.ComponentCount(), MaxwellToGL::VertexType(attrib),
+                             attrib.IsNormalized() ? GL_TRUE : GL_FALSE, attrib.offset);
+        glVertexAttribBinding(index, attrib.buffer);
+
         hw_vao_enabled_attributes[index] = true;
     }
 
-    // Copy vertex array data
-    const u64 data_size{vertex_array_limit.LimitAddress() - vertex_array.StartAddress() + 1};
-    const VAddr data_addr{memory_manager->PhysicalToVirtualAddress(vertex_array.StartAddress())};
-    res_cache.FlushRegion(data_addr, data_size, nullptr);
-    Memory::ReadBlock(data_addr, array_ptr, data_size);
-
-    array_ptr += data_size;
-    buffer_offset += data_size;
+    return {array_ptr, buffer_offset};
 }
 
-void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset, size_t ptr_pos) {
+void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset) {
     // Helper function for uploading uniform data
     const auto copy_buffer = [&](GLuint handle, GLintptr offset, GLsizeiptr size) {
         if (has_ARB_direct_state_access) {
@@ -191,8 +206,6 @@ void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset, size
     u32 current_constbuffer_bindpoint = 0;
 
     for (unsigned index = 1; index < Maxwell::MaxShaderProgram; ++index) {
-        ptr_pos += sizeof(GLShader::MaxwellUniformData);
-
         auto& shader_config = gpu.regs.shader_config[index];
         const Maxwell::ShaderProgram program{static_cast<Maxwell::ShaderProgram>(index)};
 
@@ -206,18 +219,21 @@ void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset, size
         }
 
         // Upload uniform data as one UBO per stage
-        const GLintptr ubo_offset = buffer_offset + static_cast<GLintptr>(ptr_pos);
+        const GLintptr ubo_offset = buffer_offset;
         copy_buffer(uniform_buffers[stage].handle, ubo_offset,
                     sizeof(GLShader::MaxwellUniformData));
         GLShader::MaxwellUniformData* ub_ptr =
-            reinterpret_cast<GLShader::MaxwellUniformData*>(&buffer_ptr[ptr_pos]);
+            reinterpret_cast<GLShader::MaxwellUniformData*>(buffer_ptr);
         ub_ptr->SetFromRegs(gpu.state.shader_stages[stage]);
 
+        buffer_ptr += sizeof(GLShader::MaxwellUniformData);
+        buffer_offset += sizeof(GLShader::MaxwellUniformData);
+
         // Fetch program code from memory
         GLShader::ProgramCode program_code;
         const u64 gpu_address{gpu.regs.code_address.CodeAddress() + shader_config.offset};
-        const VAddr cpu_address{gpu.memory_manager.PhysicalToVirtualAddress(gpu_address)};
-        Memory::ReadBlock(cpu_address, program_code.data(), program_code.size() * sizeof(u64));
+        const boost::optional<VAddr> cpu_address{gpu.memory_manager.GpuToCpuAddress(gpu_address)};
+        Memory::ReadBlock(*cpu_address, program_code.data(), program_code.size() * sizeof(u64));
         GLShader::ShaderSetup setup{std::move(program_code)};
 
         GLShader::ShaderEntries shader_resources;
@@ -236,8 +252,8 @@ void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset, size
             break;
         }
         default:
-            LOG_CRITICAL(HW_GPU, "Unimplemented shader index=%d, enable=%d, offset=0x%08X", index,
-                         shader_config.enable.Value(), shader_config.offset);
+            NGLOG_CRITICAL(HW_GPU, "Unimplemented shader index={}, enable={}, offset={:#010X}",
+                           index, shader_config.enable.Value(), shader_config.offset);
             UNREACHABLE();
         }
 
@@ -253,6 +269,24 @@ void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset, size
     shader_program_manager->UseTrivialGeometryShader();
 }
 
+size_t RasterizerOpenGL::CalculateVertexArraysSize() const {
+    const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
+
+    size_t size = 0;
+    for (u32 index = 0; index < Maxwell::NumVertexArrays; ++index) {
+        if (!regs.vertex_array[index].IsEnabled())
+            continue;
+
+        const Tegra::GPUVAddr start = regs.vertex_array[index].StartAddress();
+        const Tegra::GPUVAddr end = regs.vertex_array_limit[index].LimitAddress();
+
+        ASSERT(end > start);
+        size += end - start + 1;
+    }
+
+    return size;
+}
+
 bool RasterizerOpenGL::AccelerateDrawBatch(bool is_indexed) {
     accelerate_draw = is_indexed ? AccelDraw::Indexed : AccelDraw::Arrays;
     DrawArrays();
@@ -330,44 +364,49 @@ void RasterizerOpenGL::DrawArrays() {
     const u64 index_buffer_size{regs.index_array.count * regs.index_array.FormatSizeInBytes()};
     const unsigned vertex_num{is_indexed ? regs.index_array.count : regs.vertex_buffer.count};
 
-    // TODO(bunnei): Add support for 1+ vertex arrays
-    vs_input_size = vertex_num * regs.vertex_array[0].stride;
-
     state.draw.vertex_buffer = stream_buffer->GetHandle();
     state.Apply();
 
-    size_t buffer_size = static_cast<size_t>(vs_input_size);
+    size_t buffer_size = CalculateVertexArraysSize();
+
     if (is_indexed) {
-        buffer_size = Common::AlignUp(buffer_size, 4) + index_buffer_size;
+        buffer_size = Common::AlignUp<size_t>(buffer_size, 4) + index_buffer_size;
     }
 
     // Uniform space for the 5 shader stages
-    buffer_size += sizeof(GLShader::MaxwellUniformData) * Maxwell::MaxShaderStage;
+    buffer_size = Common::AlignUp<size_t>(buffer_size, 4) +
+                  sizeof(GLShader::MaxwellUniformData) * Maxwell::MaxShaderStage;
 
-    size_t ptr_pos = 0;
     u8* buffer_ptr;
     GLintptr buffer_offset;
     std::tie(buffer_ptr, buffer_offset) =
         stream_buffer->Map(static_cast<GLsizeiptr>(buffer_size), 4);
 
-    SetupVertexArray(buffer_ptr, buffer_offset);
-    ptr_pos += vs_input_size;
+    u8* offseted_buffer;
+    std::tie(offseted_buffer, buffer_offset) = SetupVertexArrays(buffer_ptr, buffer_offset);
+
+    offseted_buffer =
+        reinterpret_cast<u8*>(Common::AlignUp(reinterpret_cast<size_t>(offseted_buffer), 4));
+    buffer_offset = Common::AlignUp<size_t>(buffer_offset, 4);
 
     // If indexed mode, copy the index buffer
     GLintptr index_buffer_offset = 0;
     if (is_indexed) {
-        ptr_pos = Common::AlignUp(ptr_pos, 4);
-
         const auto& memory_manager = Core::System().GetInstance().GPU().memory_manager;
-        const VAddr index_data_addr{
-            memory_manager->PhysicalToVirtualAddress(regs.index_array.StartAddress())};
-        Memory::ReadBlock(index_data_addr, &buffer_ptr[ptr_pos], index_buffer_size);
+        const boost::optional<VAddr> index_data_addr{
+            memory_manager->GpuToCpuAddress(regs.index_array.StartAddress())};
+        Memory::ReadBlock(*index_data_addr, offseted_buffer, index_buffer_size);
 
-        index_buffer_offset = buffer_offset + static_cast<GLintptr>(ptr_pos);
-        ptr_pos += index_buffer_size;
+        index_buffer_offset = buffer_offset;
+        offseted_buffer += index_buffer_size;
+        buffer_offset += index_buffer_size;
     }
 
-    SetupShaders(buffer_ptr, buffer_offset, ptr_pos);
+    offseted_buffer =
+        reinterpret_cast<u8*>(Common::AlignUp(reinterpret_cast<size_t>(offseted_buffer), 4));
+    buffer_offset = Common::AlignUp<size_t>(buffer_offset, 4);
+
+    SetupShaders(offseted_buffer, buffer_offset);
 
     stream_buffer->Unmap();
 
@@ -479,17 +518,17 @@ void RasterizerOpenGL::FlushAll() {
     res_cache.FlushAll();
 }
 
-void RasterizerOpenGL::FlushRegion(VAddr addr, u64 size) {
+void RasterizerOpenGL::FlushRegion(Tegra::GPUVAddr addr, u64 size) {
     MICROPROFILE_SCOPE(OpenGL_CacheManagement);
     res_cache.FlushRegion(addr, size);
 }
 
-void RasterizerOpenGL::InvalidateRegion(VAddr addr, u64 size) {
+void RasterizerOpenGL::InvalidateRegion(Tegra::GPUVAddr addr, u64 size) {
     MICROPROFILE_SCOPE(OpenGL_CacheManagement);
     res_cache.InvalidateRegion(addr, size, nullptr);
 }
 
-void RasterizerOpenGL::FlushAndInvalidateRegion(VAddr addr, u64 size) {
+void RasterizerOpenGL::FlushAndInvalidateRegion(Tegra::GPUVAddr addr, u64 size) {
     MICROPROFILE_SCOPE(OpenGL_CacheManagement);
     res_cache.FlushRegion(addr, size);
     res_cache.InvalidateRegion(addr, size, nullptr);
@@ -520,7 +559,8 @@ bool RasterizerOpenGL::AccelerateDisplay(const Tegra::FramebufferConfig& framebu
     MICROPROFILE_SCOPE(OpenGL_CacheManagement);
 
     SurfaceParams src_params;
-    src_params.addr = framebuffer_addr;
+    src_params.cpu_addr = framebuffer_addr;
+    src_params.addr = res_cache.TryFindFramebufferGpuAddress(framebuffer_addr).get_value_or(0);
     src_params.width = std::min(framebuffer.width, pixel_stride);
     src_params.height = framebuffer.height;
     src_params.stride = pixel_stride;
@@ -619,9 +659,9 @@ u32 RasterizerOpenGL::SetupConstBuffers(Maxwell::ShaderStage stage, GLuint progr
         buffer_draw_state.enabled = true;
         buffer_draw_state.bindpoint = current_bindpoint + bindpoint;
 
-        VAddr addr = gpu.memory_manager->PhysicalToVirtualAddress(buffer.address);
+        boost::optional<VAddr> addr = gpu.memory_manager->GpuToCpuAddress(buffer.address);
         std::vector<u8> data(used_buffer.GetSize() * sizeof(float));
-        Memory::ReadBlock(addr, data.data(), data.size());
+        Memory::ReadBlock(*addr, data.data(), data.size());
 
         glBindBuffer(GL_SHADER_STORAGE_BUFFER, buffer_draw_state.ssbo);
         glBufferData(GL_SHADER_STORAGE_BUFFER, data.size(), data.data(), GL_DYNAMIC_DRAW);

src/video_core/renderer_opengl/gl_rasterizer.h

@@ -6,16 +6,12 @@
 
 #include <array>
 #include <cstddef>
-#include <cstring>
 #include <memory>
-#include <unordered_map>
 #include <vector>
 #include <glad/glad.h>
-#include "common/bit_field.h"
 #include "common/common_types.h"
-#include "common/hash.h"
-#include "common/vector_math.h"
 #include "video_core/engines/maxwell_3d.h"
+#include "video_core/memory_manager.h"
 #include "video_core/rasterizer_interface.h"
 #include "video_core/renderer_opengl/gl_rasterizer_cache.h"
 #include "video_core/renderer_opengl/gl_resource_manager.h"
@@ -34,9 +30,9 @@ public:
     void DrawArrays() override;
     void NotifyMaxwellRegisterChanged(u32 method) override;
     void FlushAll() override;
-    void FlushRegion(VAddr addr, u64 size) override;
-    void InvalidateRegion(VAddr addr, u64 size) override;
-    void FlushAndInvalidateRegion(VAddr addr, u64 size) override;
+    void FlushRegion(Tegra::GPUVAddr addr, u64 size) override;
+    void InvalidateRegion(Tegra::GPUVAddr addr, u64 size) override;
+    void FlushAndInvalidateRegion(Tegra::GPUVAddr addr, u64 size) override;
     bool AccelerateDisplayTransfer(const void* config) override;
     bool AccelerateTextureCopy(const void* config) override;
     bool AccelerateFill(const void* config) override;
@@ -153,13 +149,13 @@ private:
     static constexpr size_t STREAM_BUFFER_SIZE = 4 * 1024 * 1024;
     std::unique_ptr<OGLStreamBuffer> stream_buffer;
 
-    GLsizeiptr vs_input_size;
+    size_t CalculateVertexArraysSize() const;
 
-    void SetupVertexArray(u8* array_ptr, GLintptr buffer_offset);
+    std::pair<u8*, GLintptr> SetupVertexArrays(u8* array_ptr, GLintptr buffer_offset);
 
     std::array<OGLBuffer, Tegra::Engines::Maxwell3D::Regs::MaxShaderStage> uniform_buffers;
 
-    void SetupShaders(u8* buffer_ptr, GLintptr buffer_offset, size_t ptr_pos);
+    void SetupShaders(u8* buffer_ptr, GLintptr buffer_offset);
 
     enum class AccelDraw { Disabled, Arrays, Indexed };
     AccelDraw accelerate_draw;

src/video_core/renderer_opengl/gl_rasterizer_cache.cpp

@@ -7,7 +7,6 @@
 #include <cstring>
 #include <iterator>
 #include <memory>
-#include <unordered_set>
 #include <utility>
 #include <vector>
 #include <boost/optional.hpp>
@@ -20,7 +19,6 @@
 #include "common/math_util.h"
 #include "common/microprofile.h"
 #include "common/scope_exit.h"
-#include "common/vector_math.h"
 #include "core/core.h"
 #include "core/frontend/emu_window.h"
 #include "core/hle/kernel/process.h"
@@ -43,17 +41,15 @@ struct FormatTuple {
     GLenum format;
     GLenum type;
     bool compressed;
-    // How many pixels in the original texture are equivalent to one pixel in the compressed
-    // texture.
-    u32 compression_factor;
 };
 
 static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_format_tuples = {{
-    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV, false, 1},                     // ABGR8
-    {GL_RGB, GL_RGB, GL_UNSIGNED_SHORT_5_6_5_REV, false, 1},                        // B5G6R5
-    {GL_COMPRESSED_RGB_S3TC_DXT1_EXT, GL_RGB, GL_UNSIGNED_INT_8_8_8_8, true, 16},   // DXT1
-    {GL_COMPRESSED_RGBA_S3TC_DXT3_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, true, 16}, // DXT23
-    {GL_COMPRESSED_RGBA_S3TC_DXT5_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, true, 16}, // DXT45
+    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV, false},                    // ABGR8
+    {GL_RGB, GL_RGB, GL_UNSIGNED_SHORT_5_6_5_REV, false},                       // B5G6R5
+    {GL_RGB10_A2, GL_RGBA, GL_UNSIGNED_INT_2_10_10_10_REV, false},              // A2B10G10R10
+    {GL_COMPRESSED_RGB_S3TC_DXT1_EXT, GL_RGB, GL_UNSIGNED_INT_8_8_8_8, true},   // DXT1
+    {GL_COMPRESSED_RGBA_S3TC_DXT3_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, true}, // DXT23
+    {GL_COMPRESSED_RGBA_S3TC_DXT5_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, true}, // DXT45
 }};
 
 static const FormatTuple& GetFormatTuple(PixelFormat pixel_format, ComponentType component_type) {
@@ -84,38 +80,44 @@ static u16 GetResolutionScaleFactor() {
 }
 
 template <bool morton_to_gl, PixelFormat format>
-void MortonCopy(u32 stride, u32 block_height, u32 height, u8* gl_buffer, VAddr base, VAddr start,
-                VAddr end) {
-    constexpr u32 bytes_per_pixel = SurfaceParams::GetFormatBpp(format) / 8;
+void MortonCopy(u32 stride, u32 block_height, u32 height, u8* gl_buffer, Tegra::GPUVAddr base,
+                Tegra::GPUVAddr start, Tegra::GPUVAddr end) {
+    constexpr u32 bytes_per_pixel = SurfaceParams::GetFormatBpp(format) / CHAR_BIT;
     constexpr u32 gl_bytes_per_pixel = CachedSurface::GetGLBytesPerPixel(format);
+    const auto& gpu = Core::System::GetInstance().GPU();
 
     if (morton_to_gl) {
         auto data = Tegra::Texture::UnswizzleTexture(
-            base, SurfaceParams::TextureFormatFromPixelFormat(format), stride, height,
-            block_height);
+            *gpu.memory_manager->GpuToCpuAddress(base),
+            SurfaceParams::TextureFormatFromPixelFormat(format), stride, height, block_height);
         std::memcpy(gl_buffer, data.data(), data.size());
     } else {
         // TODO(bunnei): Assumes the default rendering GOB size of 16 (128 lines). We should check
         // the configuration for this and perform more generic un/swizzle
-        LOG_WARNING(Render_OpenGL, "need to use correct swizzle/GOB parameters!");
-        VideoCore::MortonCopyPixels128(stride, height, bytes_per_pixel, gl_bytes_per_pixel,
-                                       Memory::GetPointer(base), gl_buffer, morton_to_gl);
+        NGLOG_WARNING(Render_OpenGL, "need to use correct swizzle/GOB parameters!");
+        VideoCore::MortonCopyPixels128(
+            stride, height, bytes_per_pixel, gl_bytes_per_pixel,
+            Memory::GetPointer(*gpu.memory_manager->GpuToCpuAddress(base)), gl_buffer,
+            morton_to_gl);
     }
 }
 
-static constexpr std::array<void (*)(u32, u32, u32, u8*, VAddr, VAddr, VAddr),
+static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr, Tegra::GPUVAddr,
+                                     Tegra::GPUVAddr),
                             SurfaceParams::MaxPixelFormat>
     morton_to_gl_fns = {
-        MortonCopy<true, PixelFormat::ABGR8>, MortonCopy<true, PixelFormat::B5G6R5>,
-        MortonCopy<true, PixelFormat::DXT1>,  MortonCopy<true, PixelFormat::DXT23>,
-        MortonCopy<true, PixelFormat::DXT45>,
+        MortonCopy<true, PixelFormat::ABGR8>,       MortonCopy<true, PixelFormat::B5G6R5>,
+        MortonCopy<true, PixelFormat::A2B10G10R10>, MortonCopy<true, PixelFormat::DXT1>,
+        MortonCopy<true, PixelFormat::DXT23>,       MortonCopy<true, PixelFormat::DXT45>,
 };
 
-static constexpr std::array<void (*)(u32, u32, u32, u8*, VAddr, VAddr, VAddr),
+static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr, Tegra::GPUVAddr,
+                                     Tegra::GPUVAddr),
                             SurfaceParams::MaxPixelFormat>
     gl_to_morton_fns = {
         MortonCopy<false, PixelFormat::ABGR8>,
         MortonCopy<false, PixelFormat::B5G6R5>,
+        MortonCopy<false, PixelFormat::A2B10G10R10>,
         // TODO(Subv): Swizzling the DXT1/DXT23/DXT45 formats is not yet supported
         nullptr,
         nullptr,
@@ -219,9 +221,9 @@ SurfaceParams SurfaceParams::FromInterval(SurfaceInterval interval) const {
     SurfaceParams params = *this;
     const u32 tiled_size = is_tiled ? 8 : 1;
     const u64 stride_tiled_bytes = BytesInPixels(stride * tiled_size);
-    VAddr aligned_start =
+    Tegra::GPUVAddr aligned_start =
         addr + Common::AlignDown(boost::icl::first(interval) - addr, stride_tiled_bytes);
-    VAddr aligned_end =
+    Tegra::GPUVAddr aligned_end =
         addr + Common::AlignUp(boost::icl::last_next(interval) - addr, stride_tiled_bytes);
 
     if (aligned_end - aligned_start > stride_tiled_bytes) {
@@ -342,6 +344,13 @@ bool SurfaceParams::CanTexCopy(const SurfaceParams& texcopy_params) const {
     return FromInterval(texcopy_params.GetInterval()).GetInterval() == texcopy_params.GetInterval();
 }
 
+VAddr SurfaceParams::GetCpuAddr() const {
+    // When this function is used, only cpu_addr or (GPU) addr should be set, not both
+    ASSERT(!(cpu_addr && addr));
+    const auto& gpu = Core::System::GetInstance().GPU();
+    return cpu_addr.get_value_or(*gpu.memory_manager->GpuToCpuAddress(addr));
+}
+
 bool CachedSurface::CanFill(const SurfaceParams& dest_surface,
                             SurfaceInterval fill_interval) const {
     if (type == SurfaceType::Fill && IsRegionValid(fill_interval) &&
@@ -349,9 +358,9 @@ bool CachedSurface::CanFill(const SurfaceParams& dest_surface,
         boost::icl::last_next(fill_interval) <= end && // dest_surface is within our fill range
         dest_surface.FromInterval(fill_interval).GetInterval() ==
             fill_interval) { // make sure interval is a rectangle in dest surface
-        if (fill_size * 8 != dest_surface.GetFormatBpp()) {
+        if (fill_size * CHAR_BIT != dest_surface.GetFormatBpp()) {
             // Check if bits repeat for our fill_size
-            const u32 dest_bytes_per_pixel = std::max(dest_surface.GetFormatBpp() / 8, 1u);
+            const u32 dest_bytes_per_pixel = std::max(dest_surface.GetFormatBpp() / CHAR_BIT, 1u);
             std::vector<u8> fill_test(fill_size * dest_bytes_per_pixel);
 
             for (u32 i = 0; i < dest_bytes_per_pixel; ++i)
@@ -456,15 +465,15 @@ void RasterizerCacheOpenGL::CopySurface(const Surface& src_surface, const Surfac
 }
 
 MICROPROFILE_DEFINE(OpenGL_SurfaceLoad, "OpenGL", "Surface Load", MP_RGB(128, 64, 192));
-void CachedSurface::LoadGLBuffer(VAddr load_start, VAddr load_end) {
+void CachedSurface::LoadGLBuffer(Tegra::GPUVAddr load_start, Tegra::GPUVAddr load_end) {
     ASSERT(type != SurfaceType::Fill);
 
-    u8* const texture_src_data = Memory::GetPointer(addr);
+    u8* const texture_src_data = Memory::GetPointer(GetCpuAddr());
     if (texture_src_data == nullptr)
         return;
 
     if (gl_buffer == nullptr) {
-        gl_buffer_size = width * height * GetGLBytesPerPixel(pixel_format);
+        gl_buffer_size = GetActualWidth() * GetActualHeight() * GetGLBytesPerPixel(pixel_format);
         gl_buffer.reset(new u8[gl_buffer_size]);
     }
 
@@ -479,14 +488,15 @@ void CachedSurface::LoadGLBuffer(VAddr load_start, VAddr load_end) {
         std::memcpy(&gl_buffer[start_offset], texture_src_data + start_offset,
                     bytes_per_pixel * width * height);
     } else {
-        morton_to_gl_fns[static_cast<size_t>(pixel_format)](
-            stride, block_height, height, &gl_buffer[0], addr, load_start, load_end);
+        morton_to_gl_fns[static_cast<size_t>(pixel_format)](GetActualWidth(), block_height,
+                                                            GetActualHeight(), &gl_buffer[0], addr,
+                                                            load_start, load_end);
     }
 }
 
 MICROPROFILE_DEFINE(OpenGL_SurfaceFlush, "OpenGL", "Surface Flush", MP_RGB(128, 192, 64));
-void CachedSurface::FlushGLBuffer(VAddr flush_start, VAddr flush_end) {
-    u8* const dst_buffer = Memory::GetPointer(addr);
+void CachedSurface::FlushGLBuffer(Tegra::GPUVAddr flush_start, Tegra::GPUVAddr flush_end) {
+    u8* const dst_buffer = Memory::GetPointer(GetCpuAddr());
     if (dst_buffer == nullptr)
         return;
 
@@ -536,7 +546,8 @@ void CachedSurface::UploadGLTexture(const MathUtil::Rectangle<u32>& rect, GLuint
 
     MICROPROFILE_SCOPE(OpenGL_TextureUL);
 
-    ASSERT(gl_buffer_size == width * height * GetGLBytesPerPixel(pixel_format));
+    ASSERT(gl_buffer_size ==
+           GetActualWidth() * GetActualHeight() * GetGLBytesPerPixel(pixel_format));
 
     // Load data from memory to the surface
     GLint x0 = static_cast<GLint>(rect.left);
@@ -571,11 +582,9 @@ void CachedSurface::UploadGLTexture(const MathUtil::Rectangle<u32>& rect, GLuint
     glActiveTexture(GL_TEXTURE0);
     if (tuple.compressed) {
         glCompressedTexImage2D(GL_TEXTURE_2D, 0, tuple.internal_format,
-                               static_cast<GLsizei>(rect.GetWidth()),
-                               static_cast<GLsizei>(rect.GetHeight()), 0,
-                               rect.GetWidth() * rect.GetHeight() *
-                                   GetGLBytesPerPixel(pixel_format) / tuple.compression_factor,
-                               &gl_buffer[buffer_offset]);
+                               static_cast<GLsizei>(rect.GetWidth() * GetCompresssionFactor()),
+                               static_cast<GLsizei>(rect.GetHeight() * GetCompresssionFactor()), 0,
+                               size, &gl_buffer[buffer_offset]);
     } else {
         glTexSubImage2D(GL_TEXTURE_2D, 0, x0, y0, static_cast<GLsizei>(rect.GetWidth()),
                         static_cast<GLsizei>(rect.GetHeight()), tuple.format, tuple.type,
@@ -672,7 +681,8 @@ void CachedSurface::DownloadGLTexture(const MathUtil::Rectangle<u32>& rect, GLui
     glPixelStorei(GL_PACK_ROW_LENGTH, 0);
 }
 
-enum MatchFlags {
+enum class MatchFlags {
+    None = 0,
     Invalid = 1,      // Flag that can be applied to other match types, invalid matches require
                       // validation before they can be used
     Exact = 1 << 1,   // Surfaces perfectly match
@@ -686,6 +696,10 @@ constexpr MatchFlags operator|(MatchFlags lhs, MatchFlags rhs) {
     return static_cast<MatchFlags>(static_cast<int>(lhs) | static_cast<int>(rhs));
 }
 
+constexpr MatchFlags operator&(MatchFlags lhs, MatchFlags rhs) {
+    return static_cast<MatchFlags>(static_cast<int>(lhs) & static_cast<int>(rhs));
+}
+
 /// Get the best surface match (and its match type) for the given flags
 template <MatchFlags find_flags>
 Surface FindMatch(const SurfaceCache& surface_cache, const SurfaceParams& params,
@@ -703,15 +717,15 @@ Surface FindMatch(const SurfaceCache& surface_cache, const SurfaceParams& params
                                          : (params.res_scale <= surface->res_scale);
             // validity will be checked in GetCopyableInterval
             bool is_valid =
-                find_flags & MatchFlags::Copy
+                (find_flags & MatchFlags::Copy) != MatchFlags::None
                     ? true
                     : surface->IsRegionValid(validate_interval.value_or(params.GetInterval()));
 
-            if (!(find_flags & MatchFlags::Invalid) && !is_valid)
+            if ((find_flags & MatchFlags::Invalid) == MatchFlags::None && !is_valid)
                 continue;
 
             auto IsMatch_Helper = [&](auto check_type, auto match_fn) {
-                if (!(find_flags & check_type))
+                if ((find_flags & check_type) == MatchFlags::None)
                     return;
 
                 bool matched;
@@ -940,6 +954,33 @@ Surface RasterizerCacheOpenGL::GetSurface(const SurfaceParams& params, ScaleMatc
     return surface;
 }
 
+boost::optional<Tegra::GPUVAddr> RasterizerCacheOpenGL::TryFindFramebufferGpuAddress(
+    VAddr cpu_addr) const {
+    // Tries to find the GPU address of a framebuffer based on the CPU address. This is because
+    // final output framebuffers are specified by CPU address, but internally our GPU cache uses GPU
+    // addresses. We iterate through all cached framebuffers, and compare their starting CPU address
+    // to the one provided. This is obviously not great, and won't work if the framebuffer overlaps
+    // surfaces.
+
+    std::vector<Tegra::GPUVAddr> gpu_addresses;
+    for (const auto& pair : surface_cache) {
+        for (const auto& surface : pair.second) {
+            const VAddr surface_cpu_addr = surface->GetCpuAddr();
+            if (cpu_addr >= surface_cpu_addr && cpu_addr < (surface_cpu_addr + surface->size)) {
+                ASSERT_MSG(cpu_addr == surface_cpu_addr, "overlapping surfaces are unsupported");
+                gpu_addresses.push_back(surface->addr);
+            }
+        }
+    }
+
+    if (gpu_addresses.empty()) {
+        return {};
+    }
+
+    ASSERT_MSG(gpu_addresses.size() == 1, ">1 surface is unsupported");
+    return gpu_addresses[0];
+}
+
 SurfaceRect_Tuple RasterizerCacheOpenGL::GetSurfaceSubRect(const SurfaceParams& params,
                                                            ScaleMatch match_res_scale,
                                                            bool load_if_create) {
@@ -1023,11 +1064,11 @@ Surface RasterizerCacheOpenGL::GetTextureSurface(const Tegra::Texture::FullTextu
     auto& gpu = Core::System::GetInstance().GPU();
 
     SurfaceParams params;
-    params.addr = gpu.memory_manager->PhysicalToVirtualAddress(config.tic.Address());
-    params.width = config.tic.Width();
-    params.height = config.tic.Height();
+    params.addr = config.tic.Address();
     params.is_tiled = config.tic.IsTiled();
     params.pixel_format = SurfaceParams::PixelFormatFromTextureFormat(config.tic.format);
+    params.width = config.tic.Width() / params.GetCompresssionFactor();
+    params.height = config.tic.Height() / params.GetCompresssionFactor();
 
     // TODO(Subv): Different types per component are not supported.
     ASSERT(config.tic.r_type.Value() == config.tic.g_type.Value() &&
@@ -1040,7 +1081,7 @@ Surface RasterizerCacheOpenGL::GetTextureSurface(const Tegra::Texture::FullTextu
         params.block_height = config.tic.BlockHeight();
     } else {
         // Use the texture-provided stride value if the texture isn't tiled.
-        params.stride = params.PixelsInBytes(config.tic.Pitch());
+        params.stride = static_cast<u32>(params.PixelsInBytes(config.tic.Pitch()));
     }
 
     params.UpdateParams();
@@ -1068,11 +1109,10 @@ Surface RasterizerCacheOpenGL::GetTextureSurface(const Tegra::Texture::FullTextu
 SurfaceSurfaceRect_Tuple RasterizerCacheOpenGL::GetFramebufferSurfaces(
     bool using_color_fb, bool using_depth_fb, const MathUtil::Rectangle<s32>& viewport) {
     const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
-    const auto& memory_manager = Core::System().GetInstance().GPU().memory_manager;
     const auto& config = regs.rt[0];
 
     // TODO(bunnei): This is hard corded to use just the first render buffer
-    LOG_WARNING(Render_OpenGL, "hard-coded for render target 0!");
+    NGLOG_WARNING(Render_OpenGL, "hard-coded for render target 0!");
 
     // update resolution_scale_factor and reset cache if changed
     // TODO (bunnei): This code was ported as-is from Citra, and is technically not thread-safe. We
@@ -1101,7 +1141,7 @@ SurfaceSurfaceRect_Tuple RasterizerCacheOpenGL::GetFramebufferSurfaces(
     color_params.block_height = Tegra::Texture::TICEntry::DefaultBlockHeight;
     SurfaceParams depth_params = color_params;
 
-    color_params.addr = memory_manager->PhysicalToVirtualAddress(config.Address());
+    color_params.addr = config.Address();
     color_params.pixel_format = SurfaceParams::PixelFormatFromRenderTargetFormat(config.format);
     color_params.component_type = SurfaceParams::ComponentTypeFromRenderTarget(config.format);
     color_params.UpdateParams();
@@ -1117,8 +1157,8 @@ SurfaceSurfaceRect_Tuple RasterizerCacheOpenGL::GetFramebufferSurfaces(
     // Make sure that framebuffers don't overlap if both color and depth are being used
     if (using_color_fb && using_depth_fb &&
         boost::icl::length(color_vp_interval & depth_vp_interval)) {
-        LOG_CRITICAL(Render_OpenGL, "Color and depth framebuffer memory regions overlap; "
-                                    "overlapping framebuffers not supported!");
+        NGLOG_CRITICAL(Render_OpenGL, "Color and depth framebuffer memory regions overlap; "
+                                      "overlapping framebuffers not supported!");
         using_depth_fb = false;
     }
 
@@ -1217,7 +1257,8 @@ void RasterizerCacheOpenGL::DuplicateSurface(const Surface& src_surface,
     }
 }
 
-void RasterizerCacheOpenGL::ValidateSurface(const Surface& surface, VAddr addr, u64 size) {
+void RasterizerCacheOpenGL::ValidateSurface(const Surface& surface, Tegra::GPUVAddr addr,
+                                            u64 size) {
     if (size == 0)
         return;
 
@@ -1256,7 +1297,7 @@ void RasterizerCacheOpenGL::ValidateSurface(const Surface& surface, VAddr addr,
     }
 }
 
-void RasterizerCacheOpenGL::FlushRegion(VAddr addr, u64 size, Surface flush_surface) {
+void RasterizerCacheOpenGL::FlushRegion(Tegra::GPUVAddr addr, u64 size, Surface flush_surface) {
     if (size == 0)
         return;
 
@@ -1292,7 +1333,8 @@ void RasterizerCacheOpenGL::FlushAll() {
     FlushRegion(0, Kernel::VMManager::MAX_ADDRESS);
 }
 
-void RasterizerCacheOpenGL::InvalidateRegion(VAddr addr, u64 size, const Surface& region_owner) {
+void RasterizerCacheOpenGL::InvalidateRegion(Tegra::GPUVAddr addr, u64 size,
+                                             const Surface& region_owner) {
     if (size == 0)
         return;
 
@@ -1385,10 +1427,10 @@ void RasterizerCacheOpenGL::UnregisterSurface(const Surface& surface) {
     surface_cache.subtract({surface->GetInterval(), SurfaceSet{surface}});
 }
 
-void RasterizerCacheOpenGL::UpdatePagesCachedCount(VAddr addr, u64 size, int delta) {
-    const u64 num_pages =
-        ((addr + size - 1) >> Memory::PAGE_BITS) - (addr >> Memory::PAGE_BITS) + 1;
-    const u64 page_start = addr >> Memory::PAGE_BITS;
+void RasterizerCacheOpenGL::UpdatePagesCachedCount(Tegra::GPUVAddr addr, u64 size, int delta) {
+    const u64 num_pages = ((addr + size - 1) >> Tegra::MemoryManager::PAGE_BITS) -
+                          (addr >> Tegra::MemoryManager::PAGE_BITS) + 1;
+    const u64 page_start = addr >> Tegra::MemoryManager::PAGE_BITS;
     const u64 page_end = page_start + num_pages;
 
     // Interval maps will erase segments if count reaches 0, so if delta is negative we have to
@@ -1401,8 +1443,10 @@ void RasterizerCacheOpenGL::UpdatePagesCachedCount(VAddr addr, u64 size, int del
         const auto interval = pair.first & pages_interval;
         const int count = pair.second;
 
-        const VAddr interval_start_addr = boost::icl::first(interval) << Memory::PAGE_BITS;
-        const VAddr interval_end_addr = boost::icl::last_next(interval) << Memory::PAGE_BITS;
+        const Tegra::GPUVAddr interval_start_addr = boost::icl::first(interval)
+                                                    << Tegra::MemoryManager::PAGE_BITS;
+        const Tegra::GPUVAddr interval_end_addr = boost::icl::last_next(interval)
+                                                  << Tegra::MemoryManager::PAGE_BITS;
         const u64 interval_size = interval_end_addr - interval_start_addr;
 
         if (delta > 0 && count == delta)

src/video_core/renderer_opengl/gl_rasterizer_cache.h

@@ -17,12 +17,14 @@
 #ifdef __GNUC__
 #pragma GCC diagnostic pop
 #endif
+#include <boost/optional.hpp>
 #include <glad/glad.h>
 #include "common/assert.h"
 #include "common/common_funcs.h"
 #include "common/common_types.h"
 #include "common/math_util.h"
 #include "video_core/gpu.h"
+#include "video_core/memory_manager.h"
 #include "video_core/renderer_opengl/gl_resource_manager.h"
 #include "video_core/textures/texture.h"
 
@@ -30,9 +32,9 @@ struct CachedSurface;
 using Surface = std::shared_ptr<CachedSurface>;
 using SurfaceSet = std::set<Surface>;
 
-using SurfaceRegions = boost::icl::interval_set<VAddr>;
-using SurfaceMap = boost::icl::interval_map<VAddr, Surface>;
-using SurfaceCache = boost::icl::interval_map<VAddr, SurfaceSet>;
+using SurfaceRegions = boost::icl::interval_set<Tegra::GPUVAddr>;
+using SurfaceMap = boost::icl::interval_map<Tegra::GPUVAddr, Surface>;
+using SurfaceCache = boost::icl::interval_map<Tegra::GPUVAddr, SurfaceSet>;
 
 using SurfaceInterval = SurfaceCache::interval_type;
 static_assert(std::is_same<SurfaceRegions::interval_type, SurfaceCache::interval_type>() &&
@@ -54,9 +56,10 @@ struct SurfaceParams {
     enum class PixelFormat {
         ABGR8 = 0,
         B5G6R5 = 1,
-        DXT1 = 2,
-        DXT23 = 3,
-        DXT45 = 4,
+        A2B10G10R10 = 2,
+        DXT1 = 3,
+        DXT23 = 4,
+        DXT45 = 5,
 
         Max,
         Invalid = 255,
@@ -81,22 +84,49 @@ struct SurfaceParams {
         Invalid = 4,
     };
 
-    static constexpr unsigned int GetFormatBpp(PixelFormat format) {
+    /**
+     * Gets the compression factor for the specified PixelFormat. This applies to just the
+     * "compressed width" and "compressed height", not the overall compression factor of a
+     * compressed image. This is used for maintaining proper surface sizes for compressed texture
+     * formats.
+     */
+    static constexpr u32 GetCompresssionFactor(PixelFormat format) {
         if (format == PixelFormat::Invalid)
             return 0;
 
-        constexpr std::array<unsigned int, MaxPixelFormat> bpp_table = {
+        constexpr std::array<u32, MaxPixelFormat> compression_factor_table = {{
+            1, // ABGR8
+            1, // B5G6R5
+            1, // A2B10G10R10
+            4, // DXT1
+            4, // DXT23
+            4, // DXT45
+        }};
+
+        ASSERT(static_cast<size_t>(format) < compression_factor_table.size());
+        return compression_factor_table[static_cast<size_t>(format)];
+    }
+    u32 GetCompresssionFactor() const {
+        return GetCompresssionFactor(pixel_format);
+    }
+
+    static constexpr u32 GetFormatBpp(PixelFormat format) {
+        if (format == PixelFormat::Invalid)
+            return 0;
+
+        constexpr std::array<u32, MaxPixelFormat> bpp_table = {{
             32,  // ABGR8
             16,  // B5G6R5
+            32,  // A2B10G10R10
             64,  // DXT1
             128, // DXT23
             128, // DXT45
-        };
+        }};
 
         ASSERT(static_cast<size_t>(format) < bpp_table.size());
         return bpp_table[static_cast<size_t>(format)];
     }
-    unsigned int GetFormatBpp() const {
+    u32 GetFormatBpp() const {
         return GetFormatBpp(pixel_format);
     }
 
@@ -104,6 +134,8 @@ struct SurfaceParams {
         switch (format) {
         case Tegra::RenderTargetFormat::RGBA8_UNORM:
             return PixelFormat::ABGR8;
+        case Tegra::RenderTargetFormat::RGB10_A2_UNORM:
+            return PixelFormat::A2B10G10R10;
         default:
             NGLOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
             UNREACHABLE();
@@ -127,6 +159,8 @@ struct SurfaceParams {
             return PixelFormat::ABGR8;
         case Tegra::Texture::TextureFormat::B5G6R5:
             return PixelFormat::B5G6R5;
+        case Tegra::Texture::TextureFormat::A2B10G10R10:
+            return PixelFormat::A2B10G10R10;
         case Tegra::Texture::TextureFormat::DXT1:
             return PixelFormat::DXT1;
         case Tegra::Texture::TextureFormat::DXT23:
@@ -146,6 +180,8 @@ struct SurfaceParams {
             return Tegra::Texture::TextureFormat::A8R8G8B8;
         case PixelFormat::B5G6R5:
             return Tegra::Texture::TextureFormat::B5G6R5;
+        case PixelFormat::A2B10G10R10:
+            return Tegra::Texture::TextureFormat::A2B10G10R10;
         case PixelFormat::DXT1:
             return Tegra::Texture::TextureFormat::DXT1;
         case PixelFormat::DXT23:
@@ -245,6 +281,24 @@ struct SurfaceParams {
     // Returns the region of the biggest valid rectange within interval
     SurfaceInterval GetCopyableInterval(const Surface& src_surface) const;
 
+    /**
+     * Gets the actual width (in pixels) of the surface. This is provided because `width` is used
+     * for tracking the surface region in memory, which may be compressed for certain formats. In
+     * this scenario, `width` is actually the compressed width.
+     */
+    u32 GetActualWidth() const {
+        return width * GetCompresssionFactor();
+    }
+
+    /**
+     * Gets the actual height (in pixels) of the surface. This is provided because `height` is used
+     * for tracking the surface region in memory, which may be compressed for certain formats. In
+     * this scenario, `height` is actually the compressed height.
+     */
+    u32 GetActualHeight() const {
+        return height * GetCompresssionFactor();
+    }
+
     u32 GetScaledWidth() const {
         return width * res_scale;
     }
@@ -269,6 +323,8 @@ struct SurfaceParams {
         return pixels * GetFormatBpp(pixel_format) / CHAR_BIT;
     }
 
+    VAddr GetCpuAddr() const;
+
     bool ExactMatch(const SurfaceParams& other_surface) const;
     bool CanSubRect(const SurfaceParams& sub_surface) const;
     bool CanExpand(const SurfaceParams& expanded_surface) const;
@@ -277,8 +333,9 @@ struct SurfaceParams {
     MathUtil::Rectangle<u32> GetSubRect(const SurfaceParams& sub_surface) const;
     MathUtil::Rectangle<u32> GetScaledSubRect(const SurfaceParams& sub_surface) const;
 
-    VAddr addr = 0;
-    VAddr end = 0;
+    Tegra::GPUVAddr addr = 0;
+    Tegra::GPUVAddr end = 0;
+    boost::optional<VAddr> cpu_addr;
     u64 size = 0;
 
     u32 width = 0;
@@ -317,15 +374,15 @@ struct CachedSurface : SurfaceParams {
         if (format == PixelFormat::Invalid)
             return 0;
 
-        return SurfaceParams::GetFormatBpp(format) / 8;
+        return SurfaceParams::GetFormatBpp(format) / CHAR_BIT;
     }
 
     std::unique_ptr<u8[]> gl_buffer;
     size_t gl_buffer_size = 0;
 
     // Read/Write data in Switch memory to/from gl_buffer
-    void LoadGLBuffer(VAddr load_start, VAddr load_end);
-    void FlushGLBuffer(VAddr flush_start, VAddr flush_end);
+    void LoadGLBuffer(Tegra::GPUVAddr load_start, Tegra::GPUVAddr load_end);
+    void FlushGLBuffer(Tegra::GPUVAddr flush_start, Tegra::GPUVAddr flush_end);
 
     // Upload/Download data in gl_buffer in/to this surface's texture
     void UploadGLTexture(const MathUtil::Rectangle<u32>& rect, GLuint read_fb_handle,
@@ -354,6 +411,9 @@ public:
     Surface GetSurface(const SurfaceParams& params, ScaleMatch match_res_scale,
                        bool load_if_create);
 
+    /// Tries to find a framebuffer GPU address based on the provided CPU address
+    boost::optional<Tegra::GPUVAddr> TryFindFramebufferGpuAddress(VAddr cpu_addr) const;
+
     /// Attempt to find a subrect (resolution scaled) of a surface, otherwise loads a texture from
     /// Switch memory to OpenGL and caches it (if not already cached)
     SurfaceRect_Tuple GetSurfaceSubRect(const SurfaceParams& params, ScaleMatch match_res_scale,
@@ -373,10 +433,10 @@ public:
     SurfaceRect_Tuple GetTexCopySurface(const SurfaceParams& params);
 
     /// Write any cached resources overlapping the region back to memory (if dirty)
-    void FlushRegion(VAddr addr, u64 size, Surface flush_surface = nullptr);
+    void FlushRegion(Tegra::GPUVAddr addr, u64 size, Surface flush_surface = nullptr);
 
     /// Mark region as being invalidated by region_owner (nullptr if Switch memory)
-    void InvalidateRegion(VAddr addr, u64 size, const Surface& region_owner);
+    void InvalidateRegion(Tegra::GPUVAddr addr, u64 size, const Surface& region_owner);
 
     /// Flush all cached resources tracked by this cache manager
     void FlushAll();
@@ -385,7 +445,7 @@ private:
     void DuplicateSurface(const Surface& src_surface, const Surface& dest_surface);
 
     /// Update surface's texture for given region when necessary
-    void ValidateSurface(const Surface& surface, VAddr addr, u64 size);
+    void ValidateSurface(const Surface& surface, Tegra::GPUVAddr addr, u64 size);
 
     /// Create a new surface
     Surface CreateSurface(const SurfaceParams& params);
@@ -397,7 +457,7 @@ private:
     void UnregisterSurface(const Surface& surface);
 
     /// Increase/decrease the number of surface in pages touching the specified region
-    void UpdatePagesCachedCount(VAddr addr, u64 size, int delta);
+    void UpdatePagesCachedCount(Tegra::GPUVAddr addr, u64 size, int delta);
 
     SurfaceCache surface_cache;
     PageMap cached_pages;

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -97,11 +97,12 @@ private:
             return exit_method;
 
         for (u32 offset = begin; offset != end && offset != PROGRAM_END; ++offset) {
-            const Instruction instr = {program_code[offset]};
-            switch (instr.opcode.EffectiveOpCode()) {
-            case OpCode::Id::EXIT: {
-                return exit_method = ExitMethod::AlwaysEnd;
-            }
+            if (const auto opcode = OpCode::Decode({program_code[offset]})) {
+                switch (opcode->GetId()) {
+                case OpCode::Id::EXIT: {
+                    return exit_method = ExitMethod::AlwaysEnd;
+                }
+                }
             }
         }
         return exit_method = ExitMethod::AlwaysReturn;
@@ -220,6 +221,8 @@ private:
 
     /// Generates code representing a temporary (GPR) register.
     std::string GetRegister(const Register& reg, unsigned elem = 0) {
+        if (reg == Register::ZeroIndex)
+            return "0";
         if (stage == Maxwell3D::Regs::ShaderStage::Fragment && reg < 4) {
             // GPRs 0-3 are output color for the fragment shader
             return std::string{"color."} + "rgba"[(reg + elem) & 3];
@@ -276,6 +279,52 @@ private:
         shader.AddLine(dest + " = " + src + ";");
     }
 
+    /*
+     * Writes code that assigns a predicate boolean variable.
+     * @param pred The id of the predicate to write to.
+     * @param value The expression value to assign to the predicate.
+     */
+    void SetPredicate(u64 pred, const std::string& value) {
+        using Tegra::Shader::Pred;
+        // Can't assign to the constant predicate.
+        ASSERT(pred != static_cast<u64>(Pred::UnusedIndex));
+
+        std::string variable = 'p' + std::to_string(pred);
+        shader.AddLine(variable + " = " + value + ';');
+        declr_predicates.insert(std::move(variable));
+    }
+
+    /*
+     * Returns the condition to use in the 'if' for a predicated instruction.
+     * @param instr Instruction to generate the if condition for.
+     * @returns string containing the predicate condition.
+     */
+    std::string GetPredicateCondition(Instruction instr) const {
+        using Tegra::Shader::Pred;
+        ASSERT(instr.pred.pred_index != static_cast<u64>(Pred::UnusedIndex));
+
+        std::string variable =
+            'p' + std::to_string(static_cast<u64>(instr.pred.pred_index.Value()));
+
+        if (instr.negate_pred) {
+            return "!(" + variable + ')';
+        }
+
+        return variable;
+    }
+
+    /*
+     * Returns whether the instruction at the specified offset is a 'sched' instruction.
+     * Sched instructions always appear before a sequence of 3 instructions.
+     */
+    bool IsSchedInstruction(u32 offset) const {
+        // sched instructions appear once every 4 instructions.
+        static constexpr size_t SchedPeriod = 4;
+        u32 absolute_offset = offset - main_offset;
+
+        return (absolute_offset % SchedPeriod) == 0;
+    }
+
     /**
      * Compiles a single instruction from Tegra to GLSL.
      * @param offset the offset of the Tegra shader instruction.
@@ -283,11 +332,33 @@ private:
      * + 1. If the current instruction always terminates the program, returns PROGRAM_END.
      */
     u32 CompileInstr(u32 offset) {
+        // Ignore sched instructions when generating code.
+        if (IsSchedInstruction(offset)) {
+            return offset + 1;
+        }
+
         const Instruction instr = {program_code[offset]};
+        const auto opcode = OpCode::Decode(instr);
 
-        shader.AddLine("// " + std::to_string(offset) + ": " + OpCode::GetInfo(instr.opcode).name);
+        // Decoding failure
+        if (!opcode) {
+            NGLOG_CRITICAL(HW_GPU, "Unhandled instruction: {0:x}", instr.value);
+            UNREACHABLE();
+        }
 
-        switch (OpCode::GetInfo(instr.opcode).type) {
+        shader.AddLine("// " + std::to_string(offset) + ": " + opcode->GetName());
+
+        using Tegra::Shader::Pred;
+        ASSERT_MSG(instr.pred.full_pred != Pred::NeverExecute,
+                   "NeverExecute predicate not implemented");
+
+        if (instr.pred.pred_index != static_cast<u64>(Pred::UnusedIndex)) {
+            shader.AddLine("if (" + GetPredicateCondition(instr) + ')');
+            shader.AddLine('{');
+            ++shader.scope;
+        }
+
+        switch (opcode->GetType()) {
         case OpCode::Type::Arithmetic: {
             std::string dest = GetRegister(instr.gpr0);
             std::string op_a = instr.alu.negate_a ? "-" : "";
@@ -312,7 +383,7 @@ private:
                 op_b = "abs(" + op_b + ")";
             }
 
-            switch (instr.opcode.EffectiveOpCode()) {
+            switch (opcode->GetId()) {
             case OpCode::Id::FMUL_C:
             case OpCode::Id::FMUL_R:
             case OpCode::Id::FMUL_IMM: {
@@ -354,16 +425,18 @@ private:
                     SetDest(0, dest, "min(" + op_a + "," + op_b + ")", 1, 1, instr.alu.abs_d);
                     break;
                 default:
-                    NGLOG_CRITICAL(HW_GPU, "Unhandled MUFU sub op: {}",
+                    NGLOG_CRITICAL(HW_GPU, "Unhandled MUFU sub op: {0:x}",
                                    static_cast<unsigned>(instr.sub_op.Value()));
                     UNREACHABLE();
                 }
                 break;
             }
+            case OpCode::Id::RRO: {
+                NGLOG_DEBUG(HW_GPU, "Skipping RRO instruction");
+                break;
+            }
             default: {
-                NGLOG_CRITICAL(HW_GPU, "Unhandled arithmetic instruction: {} ({}): {}",
-                               static_cast<unsigned>(instr.opcode.EffectiveOpCode()),
-                               OpCode::GetInfo(instr.opcode).name, instr.hex);
+                NGLOG_CRITICAL(HW_GPU, "Unhandled arithmetic instruction: {}", opcode->GetName());
                 UNREACHABLE();
             }
             }
@@ -375,7 +448,7 @@ private:
             std::string op_b = instr.ffma.negate_b ? "-" : "";
             std::string op_c = instr.ffma.negate_c ? "-" : "";
 
-            switch (instr.opcode.EffectiveOpCode()) {
+            switch (opcode->GetId()) {
             case OpCode::Id::FFMA_CR: {
                 op_b += GetUniform(instr.uniform);
                 op_c += GetRegister(instr.gpr39);
@@ -397,9 +470,7 @@ private:
                 break;
             }
             default: {
-                NGLOG_CRITICAL(HW_GPU, "Unhandled FFMA instruction: {} ({}): {}",
-                               static_cast<unsigned>(instr.opcode.EffectiveOpCode()),
-                               OpCode::GetInfo(instr.opcode).name, instr.hex);
+                NGLOG_CRITICAL(HW_GPU, "Unhandled FFMA instruction: {}", opcode->GetName());
                 UNREACHABLE();
             }
             }
@@ -411,7 +482,7 @@ private:
             std::string gpr0 = GetRegister(instr.gpr0);
             const Attribute::Index attribute = instr.attribute.fmt20.index;
 
-            switch (instr.opcode.EffectiveOpCode()) {
+            switch (opcode->GetId()) {
             case OpCode::Id::LD_A: {
                 ASSERT_MSG(instr.attribute.fmt20.size == 0, "untested");
                 SetDest(instr.attribute.fmt20.element, gpr0, GetInputAttribute(attribute), 1, 4);
@@ -442,22 +513,76 @@ private:
                 break;
             }
             default: {
-                NGLOG_CRITICAL(HW_GPU, "Unhandled memory instruction: {} ({}): {}",
-                               static_cast<unsigned>(instr.opcode.EffectiveOpCode()),
-                               OpCode::GetInfo(instr.opcode).name, instr.hex);
+                NGLOG_CRITICAL(HW_GPU, "Unhandled memory instruction: {}", opcode->GetName());
                 UNREACHABLE();
             }
             }
             break;
         }
+        case OpCode::Type::FloatPredicate: {
+            std::string op_a = instr.fsetp.neg_a ? "-" : "";
+            op_a += GetRegister(instr.gpr8);
 
+            if (instr.fsetp.abs_a) {
+                op_a = "abs(" + op_a + ')';
+            }
+
+            std::string op_b{};
+
+            if (instr.is_b_imm) {
+                if (instr.fsetp.neg_b) {
+                    // Only the immediate version of fsetp has a neg_b bit.
+                    op_b += '-';
+                }
+                op_b += '(' + GetImmediate19(instr) + ')';
+            } else {
+                if (instr.is_b_gpr) {
+                    op_b += GetRegister(instr.gpr20);
+                } else {
+                    op_b += GetUniform(instr.uniform);
+                }
+            }
+
+            if (instr.fsetp.abs_b) {
+                op_b = "abs(" + op_b + ')';
+            }
+
+            using Tegra::Shader::Pred;
+            ASSERT_MSG(instr.fsetp.pred0 == static_cast<u64>(Pred::UnusedIndex) &&
+                           instr.fsetp.pred39 == static_cast<u64>(Pred::UnusedIndex),
+                       "Compound predicates are not implemented");
+
+            // We can't use the constant predicate as destination.
+            ASSERT(instr.fsetp.pred3 != static_cast<u64>(Pred::UnusedIndex));
+
+            using Tegra::Shader::PredCondition;
+            switch (instr.fsetp.cond) {
+            case PredCondition::LessThan:
+                SetPredicate(instr.fsetp.pred3, '(' + op_a + ") < (" + op_b + ')');
+                break;
+            case PredCondition::Equal:
+                SetPredicate(instr.fsetp.pred3, '(' + op_a + ") == (" + op_b + ')');
+                break;
+            default:
+                NGLOG_CRITICAL(HW_GPU, "Unhandled predicate condition: {} (a: {}, b: {})",
+                               static_cast<unsigned>(instr.fsetp.cond.Value()), op_a, op_b);
+                UNREACHABLE();
+            }
+            break;
+        }
         default: {
-            switch (instr.opcode.EffectiveOpCode()) {
+            switch (opcode->GetId()) {
             case OpCode::Id::EXIT: {
+                ASSERT_MSG(instr.pred.pred_index == static_cast<u64>(Pred::UnusedIndex),
+                           "Predicated exits not implemented");
                 shader.AddLine("return true;");
                 offset = PROGRAM_END - 1;
                 break;
             }
+            case OpCode::Id::KIL: {
+                shader.AddLine("discard;");
+                break;
+            }
             case OpCode::Id::IPA: {
                 const auto& attribute = instr.attribute.fmt28;
                 std::string dest = GetRegister(instr.gpr0);
@@ -465,9 +590,7 @@ private:
                 break;
             }
             default: {
-                NGLOG_CRITICAL(HW_GPU, "Unhandled instruction: {} ({}): {}",
-                               static_cast<unsigned>(instr.opcode.EffectiveOpCode()),
-                               OpCode::GetInfo(instr.opcode).name, instr.hex);
+                NGLOG_CRITICAL(HW_GPU, "Unhandled instruction: {}", opcode->GetName());
                 UNREACHABLE();
             }
             }
@@ -476,6 +599,12 @@ private:
         }
         }
 
+        // Close the predicate condition scope.
+        if (instr.pred.pred_index != static_cast<u64>(Pred::UnusedIndex)) {
+            --shader.scope;
+            shader.AddLine('}');
+        }
+
         return offset + 1;
     }
 
@@ -605,6 +734,12 @@ private:
             declarations.AddNewLine();
             ++const_buffer_layout;
         }
+
+        declarations.AddNewLine();
+        for (const auto& pred : declr_predicates) {
+            declarations.AddLine("bool " + pred + " = false;");
+        }
+        declarations.AddNewLine();
     }
 
 private:
@@ -618,6 +753,7 @@ private:
 
     // Declarations
     std::set<std::string> declr_register;
+    std::set<std::string> declr_predicates;
     std::set<Attribute::Index> declr_input_attribute;
     std::set<Attribute::Index> declr_output_attribute;
     std::array<ConstBufferEntry, Maxwell3D::Regs::MaxConstBuffers> declr_const_buffers;

src/video_core/renderer_opengl/renderer_opengl.cpp

@@ -9,13 +9,10 @@
 #include <memory>
 #include <glad/glad.h>
 #include "common/assert.h"
-#include "common/bit_field.h"
 #include "common/logging/log.h"
 #include "core/core.h"
 #include "core/core_timing.h"
 #include "core/frontend/emu_window.h"
-#include "core/hw/hw.h"
-#include "core/hw/lcd.h"
 #include "core/memory.h"
 #include "core/settings.h"
 #include "core/tracer/recorder.h"
@@ -155,7 +152,8 @@ void RendererOpenGL::LoadFBToScreenInfo(const Tegra::FramebufferConfig& framebuf
         screen_info.display_texture = screen_info.texture.resource.handle;
         screen_info.display_texcoords = MathUtil::Rectangle<float>(0.f, 0.f, 1.f, 1.f);
 
-        Rasterizer()->FlushRegion(framebuffer_addr, size_in_bytes);
+        Memory::RasterizerFlushVirtualRegion(framebuffer_addr, size_in_bytes,
+                                             Memory::FlushMode::Flush);
 
         VideoCore::MortonCopyPixels128(framebuffer.width, framebuffer.height, bytes_per_pixel, 4,
                                        Memory::GetPointer(framebuffer_addr),
@@ -272,10 +270,9 @@ void RendererOpenGL::ConfigureFramebufferTexture(TextureInfo& texture,
     GLint internal_format;
     switch (framebuffer.pixel_format) {
     case Tegra::FramebufferConfig::PixelFormat::ABGR8:
-        // Use RGBA8 and swap in the fragment shader
         internal_format = GL_RGBA;
         texture.gl_format = GL_RGBA;
-        texture.gl_type = GL_UNSIGNED_INT_8_8_8_8;
+        texture.gl_type = GL_UNSIGNED_INT_8_8_8_8_REV;
         gl_framebuffer_data.resize(texture.width * texture.height * 4);
         break;
     default:
@@ -298,17 +295,18 @@ void RendererOpenGL::DrawScreenTriangles(const ScreenInfo& screen_info, float x,
     const auto& texcoords = screen_info.display_texcoords;
     auto left = texcoords.left;
     auto right = texcoords.right;
-    if (framebuffer_transform_flags != Tegra::FramebufferConfig::TransformFlags::Unset)
+    if (framebuffer_transform_flags != Tegra::FramebufferConfig::TransformFlags::Unset) {
         if (framebuffer_transform_flags == Tegra::FramebufferConfig::TransformFlags::FlipV) {
             // Flip the framebuffer vertically
             left = texcoords.right;
             right = texcoords.left;
         } else {
             // Other transformations are unsupported
-            LOG_CRITICAL(Render_OpenGL, "Unsupported framebuffer_transform_flags=%d",
-                         framebuffer_transform_flags);
+            NGLOG_CRITICAL(Render_OpenGL, "Unsupported framebuffer_transform_flags={}",
+                           static_cast<u32>(framebuffer_transform_flags));
             UNIMPLEMENTED();
         }
+    }
 
     std::array<ScreenRectVertex, 4> vertices = {{
         ScreenRectVertex(x, y, texcoords.top, left),
@@ -430,9 +428,9 @@ bool RendererOpenGL::Init() {
     const char* gpu_vendor{reinterpret_cast<char const*>(glGetString(GL_VENDOR))};
     const char* gpu_model{reinterpret_cast<char const*>(glGetString(GL_RENDERER))};
 
-    LOG_INFO(Render_OpenGL, "GL_VERSION: %s", gl_version);
-    LOG_INFO(Render_OpenGL, "GL_VENDOR: %s", gpu_vendor);
-    LOG_INFO(Render_OpenGL, "GL_RENDERER: %s", gpu_model);
+    NGLOG_INFO(Render_OpenGL, "GL_VERSION: {}", gl_version);
+    NGLOG_INFO(Render_OpenGL, "GL_VENDOR: {}", gpu_vendor);
+    NGLOG_INFO(Render_OpenGL, "GL_RENDERER: {}", gpu_model);
 
     Core::Telemetry().AddField(Telemetry::FieldType::UserSystem, "GPU_Vendor", gpu_vendor);
     Core::Telemetry().AddField(Telemetry::FieldType::UserSystem, "GPU_Model", gpu_model);

src/video_core/textures/decoders.cpp

@@ -4,6 +4,7 @@
 
 #include <cstring>
 #include "common/assert.h"
+#include "core/memory.h"
 #include "video_core/textures/decoders.h"
 #include "video_core/textures/texture.h"
 
@@ -26,9 +27,8 @@ static u32 GetSwizzleOffset(u32 x, u32 y, u32 image_width, u32 bytes_per_pixel,
     return address;
 }
 
-static void CopySwizzledData(u32 width, u32 height, u32 bytes_per_pixel, u32 out_bytes_per_pixel,
-                             u8* swizzled_data, u8* unswizzled_data, bool unswizzle,
-                             u32 block_height) {
+void CopySwizzledData(u32 width, u32 height, u32 bytes_per_pixel, u32 out_bytes_per_pixel,
+                      u8* swizzled_data, u8* unswizzled_data, bool unswizzle, u32 block_height) {
     u8* data_ptrs[2];
     for (unsigned y = 0; y < height; ++y) {
         for (unsigned x = 0; x < width; ++x) {
@@ -53,6 +53,7 @@ u32 BytesPerPixel(TextureFormat format) {
         // In this case a 'pixel' actually refers to a 4x4 tile.
         return 16;
     case TextureFormat::A8R8G8B8:
+    case TextureFormat::A2B10G10R10:
         return 4;
     case TextureFormat::B5G6R5:
         return 2;
@@ -78,6 +79,7 @@ std::vector<u8> UnswizzleTexture(VAddr address, TextureFormat format, u32 width,
                          unswizzled_data.data(), true, block_height);
         break;
     case TextureFormat::A8R8G8B8:
+    case TextureFormat::A2B10G10R10:
     case TextureFormat::B5G6R5:
         CopySwizzledData(width, height, bytes_per_pixel, bytes_per_pixel, data,
                          unswizzled_data.data(), true, block_height);
@@ -100,6 +102,7 @@ std::vector<u8> DecodeTexture(const std::vector<u8>& texture_data, TextureFormat
     case TextureFormat::DXT23:
     case TextureFormat::DXT45:
     case TextureFormat::A8R8G8B8:
+    case TextureFormat::A2B10G10R10:
     case TextureFormat::B5G6R5:
         // TODO(Subv): For the time being just forward the same data without any decoding.
         rgba_data = texture_data;

src/video_core/textures/decoders.h

@@ -17,6 +17,10 @@ namespace Texture {
 std::vector<u8> UnswizzleTexture(VAddr address, TextureFormat format, u32 width, u32 height,
                                  u32 block_height = TICEntry::DefaultBlockHeight);
 
+/// Copies texture data from a buffer and performs swizzling/unswizzling as necessary.
+void CopySwizzledData(u32 width, u32 height, u32 bytes_per_pixel, u32 out_bytes_per_pixel,
+                      u8* swizzled_data, u8* unswizzled_data, bool unswizzle, u32 block_height);
+
 /**
  * Decodes an unswizzled texture into a A8R8G8B8 texture.
  */

src/video_core/textures/texture.h

@@ -15,6 +15,7 @@ namespace Texture {
 
 enum class TextureFormat : u32 {
     A8R8G8B8 = 0x8,
+    A2B10G10R10 = 0x9,
     B5G6R5 = 0x15,
     DXT1 = 0x24,
     DXT23 = 0x25,

src/video_core/video_core.cpp

@@ -24,9 +24,9 @@ bool Init(EmuWindow* emu_window) {
     g_renderer = std::make_unique<RendererOpenGL>();
     g_renderer->SetWindow(g_emu_window);
     if (g_renderer->Init()) {
-        LOG_DEBUG(Render, "initialized OK");
+        NGLOG_DEBUG(Render, "initialized OK");
     } else {
-        LOG_CRITICAL(Render, "initialization failed !");
+        NGLOG_CRITICAL(Render, "initialization failed !");
         return false;
     }
     return true;
@@ -36,7 +36,7 @@ bool Init(EmuWindow* emu_window) {
 void Shutdown() {
     g_renderer.reset();
 
-    LOG_DEBUG(Render, "shutdown OK");
+    NGLOG_DEBUG(Render, "shutdown OK");
 }
 
 } // namespace VideoCore

src/yuzu/debugger/graphics/graphics_surface.cpp

@@ -25,6 +25,8 @@ static Tegra::Texture::TextureFormat ConvertToTextureFormat(
     switch (render_target_format) {
     case Tegra::RenderTargetFormat::RGBA8_UNORM:
         return Tegra::Texture::TextureFormat::A8R8G8B8;
+    case Tegra::RenderTargetFormat::RGB10_A2_UNORM:
+        return Tegra::Texture::TextureFormat::A2B10G10R10;
     default:
         UNIMPLEMENTED_MSG("Unimplemented RT format");
     }
@@ -376,10 +378,10 @@ void GraphicsSurfaceWidget::OnUpdate() {
     // TODO: Implement a good way to visualize alpha components!
 
     QImage decoded_image(surface_width, surface_height, QImage::Format_ARGB32);
-    VAddr address = gpu.memory_manager->PhysicalToVirtualAddress(surface_address);
+    boost::optional<VAddr> address = gpu.memory_manager->GpuToCpuAddress(surface_address);
 
     auto unswizzled_data =
-        Tegra::Texture::UnswizzleTexture(address, surface_format, surface_width, surface_height);
+        Tegra::Texture::UnswizzleTexture(*address, surface_format, surface_width, surface_height);
 
     auto texture_data = Tegra::Texture::DecodeTexture(unswizzled_data, surface_format,
                                                       surface_width, surface_height);
@@ -435,9 +437,9 @@ void GraphicsSurfaceWidget::SaveSurface() {
             pixmap->save(&file, "PNG");
     } else if (selectedFilter == bin_filter) {
         auto& gpu = Core::System::GetInstance().GPU();
-        VAddr address = gpu.memory_manager->PhysicalToVirtualAddress(surface_address);
+        boost::optional<VAddr> address = gpu.memory_manager->GpuToCpuAddress(surface_address);
 
-        const u8* buffer = Memory::GetPointer(address);
+        const u8* buffer = Memory::GetPointer(*address);
         ASSERT_MSG(buffer != nullptr, "Memory not accessible");
 
         QFile file(filename);

src/yuzu/debugger/wait_tree.cpp

@@ -6,8 +6,8 @@
 #include "yuzu/util/util.h"
 
 #include "core/core.h"
-#include "core/hle/kernel/condition_variable.h"
 #include "core/hle/kernel/event.h"
+#include "core/hle/kernel/handle_table.h"
 #include "core/hle/kernel/mutex.h"
 #include "core/hle/kernel/thread.h"
 #include "core/hle/kernel/timer.h"
@@ -67,6 +67,29 @@ QString WaitTreeText::GetText() const {
     return text;
 }
 
+WaitTreeMutexInfo::WaitTreeMutexInfo(VAddr mutex_address) : mutex_address(mutex_address) {
+    mutex_value = Memory::Read32(mutex_address);
+    owner_handle = static_cast<Kernel::Handle>(mutex_value & Kernel::Mutex::MutexOwnerMask);
+    owner = Kernel::g_handle_table.Get<Kernel::Thread>(owner_handle);
+}
+
+QString WaitTreeMutexInfo::GetText() const {
+    return tr("waiting for mutex 0x%1").arg(mutex_address, 16, 16, QLatin1Char('0'));
+}
+
+std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeMutexInfo::GetChildren() const {
+    std::vector<std::unique_ptr<WaitTreeItem>> list;
+
+    bool has_waiters = (mutex_value & Kernel::Mutex::MutexHasWaitersFlag) != 0;
+
+    list.push_back(std::make_unique<WaitTreeText>(tr("has waiters: %1").arg(has_waiters)));
+    list.push_back(std::make_unique<WaitTreeText>(
+        tr("owner handle: 0x%1").arg(owner_handle, 8, 16, QLatin1Char('0'))));
+    if (owner != nullptr)
+        list.push_back(std::make_unique<WaitTreeThread>(*owner));
+    return list;
+}
+
 WaitTreeWaitObject::WaitTreeWaitObject(const Kernel::WaitObject& o) : object(o) {}
 
 bool WaitTreeExpandableItem::IsExpandable() const {
@@ -84,11 +107,6 @@ std::unique_ptr<WaitTreeWaitObject> WaitTreeWaitObject::make(const Kernel::WaitO
     switch (object.GetHandleType()) {
     case Kernel::HandleType::Event:
         return std::make_unique<WaitTreeEvent>(static_cast<const Kernel::Event&>(object));
-    case Kernel::HandleType::Mutex:
-        return std::make_unique<WaitTreeMutex>(static_cast<const Kernel::Mutex&>(object));
-    case Kernel::HandleType::ConditionVariable:
-        return std::make_unique<WaitTreeConditionVariable>(
-            static_cast<const Kernel::ConditionVariable&>(object));
     case Kernel::HandleType::Timer:
         return std::make_unique<WaitTreeTimer>(static_cast<const Kernel::Timer&>(object));
     case Kernel::HandleType::Thread:
@@ -160,6 +178,9 @@ QString WaitTreeThread::GetText() const {
     case THREADSTATUS_WAIT_SYNCH_ANY:
         status = tr("waiting for objects");
         break;
+    case THREADSTATUS_WAIT_MUTEX:
+        status = tr("waiting for mutex");
+        break;
     case THREADSTATUS_DORMANT:
         status = tr("dormant");
         break;
@@ -186,6 +207,7 @@ QColor WaitTreeThread::GetColor() const {
         return QColor(Qt::GlobalColor::darkYellow);
     case THREADSTATUS_WAIT_SYNCH_ALL:
     case THREADSTATUS_WAIT_SYNCH_ANY:
+    case THREADSTATUS_WAIT_MUTEX:
         return QColor(Qt::GlobalColor::red);
     case THREADSTATUS_DORMANT:
         return QColor(Qt::GlobalColor::darkCyan);
@@ -225,11 +247,11 @@ std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeThread::GetChildren() const {
     list.push_back(std::make_unique<WaitTreeText>(
         tr("last running ticks = %1").arg(thread.last_running_ticks)));
 
-    if (thread.held_mutexes.empty()) {
-        list.push_back(std::make_unique<WaitTreeText>(tr("not holding mutex")));
-    } else {
-        list.push_back(std::make_unique<WaitTreeMutexList>(thread.held_mutexes));
-    }
+    if (thread.mutex_wait_address != 0)
+        list.push_back(std::make_unique<WaitTreeMutexInfo>(thread.mutex_wait_address));
+    else
+        list.push_back(std::make_unique<WaitTreeText>(tr("not waiting for mutex")));
+
     if (thread.status == THREADSTATUS_WAIT_SYNCH_ANY ||
         thread.status == THREADSTATUS_WAIT_SYNCH_ALL) {
         list.push_back(std::make_unique<WaitTreeObjectList>(thread.wait_objects,
@@ -250,33 +272,6 @@ std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeEvent::GetChildren() const {
     return list;
 }
 
-WaitTreeMutex::WaitTreeMutex(const Kernel::Mutex& object) : WaitTreeWaitObject(object) {}
-
-std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeMutex::GetChildren() const {
-    std::vector<std::unique_ptr<WaitTreeItem>> list(WaitTreeWaitObject::GetChildren());
-
-    const auto& mutex = static_cast<const Kernel::Mutex&>(object);
-    if (mutex.GetHasWaiters()) {
-        list.push_back(std::make_unique<WaitTreeText>(tr("locked by thread:")));
-        list.push_back(std::make_unique<WaitTreeThread>(*mutex.GetHoldingThread()));
-    } else {
-        list.push_back(std::make_unique<WaitTreeText>(tr("free")));
-    }
-    return list;
-}
-
-WaitTreeConditionVariable::WaitTreeConditionVariable(const Kernel::ConditionVariable& object)
-    : WaitTreeWaitObject(object) {}
-
-std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeConditionVariable::GetChildren() const {
-    std::vector<std::unique_ptr<WaitTreeItem>> list(WaitTreeWaitObject::GetChildren());
-
-    const auto& condition_variable = static_cast<const Kernel::ConditionVariable&>(object);
-    list.push_back(std::make_unique<WaitTreeText>(
-        tr("available count = %1").arg(condition_variable.GetAvailableCount())));
-    return list;
-}
-
 WaitTreeTimer::WaitTreeTimer(const Kernel::Timer& object) : WaitTreeWaitObject(object) {}
 
 std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeTimer::GetChildren() const {
@@ -293,21 +288,6 @@ std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeTimer::GetChildren() const {
     return list;
 }
 
-WaitTreeMutexList::WaitTreeMutexList(
-    const boost::container::flat_set<Kernel::SharedPtr<Kernel::Mutex>>& list)
-    : mutex_list(list) {}
-
-QString WaitTreeMutexList::GetText() const {
-    return tr("holding mutexes");
-}
-
-std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeMutexList::GetChildren() const {
-    std::vector<std::unique_ptr<WaitTreeItem>> list(mutex_list.size());
-    std::transform(mutex_list.begin(), mutex_list.end(), list.begin(),
-                   [](const auto& t) { return std::make_unique<WaitTreeMutex>(*t); });
-    return list;
-}
-
 WaitTreeThreadList::WaitTreeThreadList(const std::vector<Kernel::SharedPtr<Kernel::Thread>>& list)
     : thread_list(list) {}
 

src/yuzu/debugger/wait_tree.h

@@ -16,8 +16,6 @@ class EmuThread;
 namespace Kernel {
 class WaitObject;
 class Event;
-class Mutex;
-class ConditionVariable;
 class Thread;
 class Timer;
 } // namespace Kernel
@@ -61,6 +59,20 @@ public:
     bool IsExpandable() const override;
 };
 
+class WaitTreeMutexInfo : public WaitTreeExpandableItem {
+    Q_OBJECT
+public:
+    explicit WaitTreeMutexInfo(VAddr mutex_address);
+    QString GetText() const override;
+    std::vector<std::unique_ptr<WaitTreeItem>> GetChildren() const override;
+
+private:
+    VAddr mutex_address;
+    u32 mutex_value;
+    Kernel::Handle owner_handle;
+    Kernel::SharedPtr<Kernel::Thread> owner;
+};
+
 class WaitTreeWaitObject : public WaitTreeExpandableItem {
     Q_OBJECT
 public:
@@ -104,20 +116,6 @@ public:
     std::vector<std::unique_ptr<WaitTreeItem>> GetChildren() const override;
 };
 
-class WaitTreeMutex : public WaitTreeWaitObject {
-    Q_OBJECT
-public:
-    explicit WaitTreeMutex(const Kernel::Mutex& object);
-    std::vector<std::unique_ptr<WaitTreeItem>> GetChildren() const override;
-};
-
-class WaitTreeConditionVariable : public WaitTreeWaitObject {
-    Q_OBJECT
-public:
-    explicit WaitTreeConditionVariable(const Kernel::ConditionVariable& object);
-    std::vector<std::unique_ptr<WaitTreeItem>> GetChildren() const override;
-};
-
 class WaitTreeTimer : public WaitTreeWaitObject {
     Q_OBJECT
 public:
@@ -125,19 +123,6 @@ public:
     std::vector<std::unique_ptr<WaitTreeItem>> GetChildren() const override;
 };
 
-class WaitTreeMutexList : public WaitTreeExpandableItem {
-    Q_OBJECT
-public:
-    explicit WaitTreeMutexList(
-        const boost::container::flat_set<Kernel::SharedPtr<Kernel::Mutex>>& list);
-
-    QString GetText() const override;
-    std::vector<std::unique_ptr<WaitTreeItem>> GetChildren() const override;
-
-private:
-    const boost::container::flat_set<Kernel::SharedPtr<Kernel::Mutex>>& mutex_list;
-};
-
 class WaitTreeThreadList : public WaitTreeExpandableItem {
     Q_OBJECT
 public:

src/yuzu_cmd/emu_window/emu_window_sdl2.cpp

@@ -56,7 +56,28 @@ void EmuWindow_SDL2::OnResize() {
     UpdateCurrentFramebufferLayout(width, height);
 }
 
-EmuWindow_SDL2::EmuWindow_SDL2() {
+void EmuWindow_SDL2::Fullscreen() {
+    if (SDL_SetWindowFullscreen(render_window, SDL_WINDOW_FULLSCREEN) == 0) {
+        return;
+    }
+
+    NGLOG_ERROR(Frontend, "Fullscreening failed: {}", SDL_GetError());
+
+    // Try a different fullscreening method
+    NGLOG_INFO(Frontend, "Attempting to use borderless fullscreen...");
+    if (SDL_SetWindowFullscreen(render_window, SDL_WINDOW_FULLSCREEN_DESKTOP) == 0) {
+        return;
+    }
+
+    NGLOG_ERROR(Frontend, "Borderless fullscreening failed: {}", SDL_GetError());
+
+    // Fallback algorithm: Maximise window.
+    // Works on all systems (unless something is seriously wrong), so no fallback for this one.
+    NGLOG_INFO(Frontend, "Falling back on a maximised window...");
+    SDL_MaximizeWindow(render_window);
+}
+
+EmuWindow_SDL2::EmuWindow_SDL2(bool fullscreen) {
     InputCommon::Init();
 
     SDL_SetMainReady();
@@ -90,6 +111,10 @@ EmuWindow_SDL2::EmuWindow_SDL2() {
         exit(1);
     }
 
+    if (fullscreen) {
+        Fullscreen();
+    }
+
     gl_context = SDL_GL_CreateContext(render_window);
 
     if (gl_context == nullptr) {

src/yuzu_cmd/emu_window/emu_window_sdl2.h

@@ -12,7 +12,7 @@ struct SDL_Window;
 
 class EmuWindow_SDL2 : public EmuWindow {
 public:
-    EmuWindow_SDL2();
+    explicit EmuWindow_SDL2(bool fullscreen);
     ~EmuWindow_SDL2();
 
     /// Swap buffers to display the next frame
@@ -43,6 +43,9 @@ private:
     /// Called by PollEvents when any event that may cause the window to be resized occurs
     void OnResize();
 
+    /// Called when user passes the fullscreen parameter flag
+    void Fullscreen();
+
     /// Called when a configuration change affects the minimal size of the window
     void OnMinimalClientAreaChangeRequest(
         const std::pair<unsigned, unsigned>& minimal_size) override;

src/yuzu_cmd/yuzu.cpp

@@ -50,6 +50,7 @@ static void PrintHelp(const char* argv0) {
     std::cout << "Usage: " << argv0
               << " [options] <filename>\n"
                  "-g, --gdbport=NUMBER  Enable gdb stub on port NUMBER\n"
+                 "-f, --fullscreen     Start in fullscreen mode\n"
                  "-h, --help            Display this help and exit\n"
                  "-v, --version         Output version information and exit\n";
 }
@@ -76,15 +77,18 @@ int main(int argc, char** argv) {
 #endif
     std::string filepath;
 
+    bool fullscreen = false;
+
     static struct option long_options[] = {
         {"gdbport", required_argument, 0, 'g'},
+        {"fullscreen", no_argument, 0, 'f'},
         {"help", no_argument, 0, 'h'},
         {"version", no_argument, 0, 'v'},
         {0, 0, 0, 0},
     };
 
     while (optind < argc) {
-        char arg = getopt_long(argc, argv, "g:hv", long_options, &option_index);
+        char arg = getopt_long(argc, argv, "g:fhv", long_options, &option_index);
         if (arg != -1) {
             switch (arg) {
             case 'g':
@@ -98,6 +102,10 @@ int main(int argc, char** argv) {
                     exit(1);
                 }
                 break;
+            case 'f':
+                fullscreen = true;
+                NGLOG_INFO(Frontend, "Starting in fullscreen mode...");
+                break;
             case 'h':
                 PrintHelp(argv[0]);
                 return 0;
@@ -137,7 +145,7 @@ int main(int argc, char** argv) {
     Settings::values.use_gdbstub = use_gdbstub;
     Settings::Apply();
 
-    std::unique_ptr<EmuWindow_SDL2> emu_window{std::make_unique<EmuWindow_SDL2>()};
+    std::unique_ptr<EmuWindow_SDL2> emu_window{std::make_unique<EmuWindow_SDL2>(fullscreen)};
 
     Core::System& system{Core::System::GetInstance()};