Clang Fixes

GPU: Added TIC and TSC registers to the Maxwell3D register structure.

src/core/arm/arm_interface.h

@@ -25,19 +25,11 @@ public:
         VAddr tls_address;
     };
 
-    /**
-     * Runs the CPU for the given number of instructions
-     * @param num_instructions Number of instructions to run
-     */
-    void Run(int num_instructions) {
-        ExecuteInstructions(num_instructions);
-        this->num_instructions += num_instructions;
-    }
+    /// Runs the CPU until an event happens
+    virtual void Run() = 0;
 
     /// Step CPU by one instruction
-    void Step() {
-        Run(1);
-    }
+    virtual void Step() = 0;
 
     /// Maps a backing memory region for the CPU
     virtual void MapBackingMemory(VAddr address, size_t size, u8* memory,
@@ -126,19 +118,4 @@ public:
 
     /// Prepare core for thread reschedule (if needed to correctly handle state)
     virtual void PrepareReschedule() = 0;
-
-    /// Getter for num_instructions
-    u64 GetNumInstructions() const {
-        return num_instructions;
-    }
-
-protected:
-    /**
-     * Executes the given number of instructions
-     * @param num_instructions Number of instructions to executes
-     */
-    virtual void ExecuteInstructions(int num_instructions) = 0;
-
-private:
-    u64 num_instructions = 0; ///< Number of instructions executed
 };

src/core/arm/dynarmic/arm_dynarmic.cpp

@@ -122,11 +122,22 @@ std::unique_ptr<Dynarmic::A64::Jit> MakeJit(const std::unique_ptr<ARM_Dynarmic_C
     return std::make_unique<Dynarmic::A64::Jit>(config);
 }
 
+void ARM_Dynarmic::Run() {
+    ASSERT(Memory::GetCurrentPageTable() == current_page_table);
+
+    jit->Run();
+}
+
+void ARM_Dynarmic::Step() {
+    cb->InterpreterFallback(jit->GetPC(), 1);
+}
+
 ARM_Dynarmic::ARM_Dynarmic()
     : cb(std::make_unique<ARM_Dynarmic_Callbacks>(*this)), jit(MakeJit(cb)) {
     ARM_Interface::ThreadContext ctx;
     inner_unicorn.SaveContext(ctx);
     LoadContext(ctx);
+    PageTableChanged();
 }
 
 ARM_Dynarmic::~ARM_Dynarmic() = default;
@@ -189,13 +200,6 @@ void ARM_Dynarmic::SetTlsAddress(u64 address) {
     cb->tpidrro_el0 = address;
 }
 
-void ARM_Dynarmic::ExecuteInstructions(int num_instructions) {
-    cb->ticks_remaining = num_instructions;
-    jit->Run();
-    CoreTiming::AddTicks(num_instructions - cb->num_interpreted_instructions);
-    cb->num_interpreted_instructions = 0;
-}
-
 void ARM_Dynarmic::SaveContext(ARM_Interface::ThreadContext& ctx) {
     ctx.cpu_registers = jit->GetRegisters();
     ctx.sp = jit->GetSP();
@@ -228,4 +232,5 @@ void ARM_Dynarmic::ClearInstructionCache() {
 
 void ARM_Dynarmic::PageTableChanged() {
     jit = MakeJit(cb);
+    current_page_table = Memory::GetCurrentPageTable();
 }

src/core/arm/dynarmic/arm_dynarmic.h

@@ -29,6 +29,8 @@ public:
     u32 GetVFPReg(int index) const override;
     void SetVFPReg(int index, u32 value) override;
     u32 GetCPSR() const override;
+    void Run() override;
+    void Step() override;
     void SetCPSR(u32 cpsr) override;
     VAddr GetTlsAddress() const override;
     void SetTlsAddress(VAddr address) override;
@@ -37,7 +39,6 @@ public:
     void LoadContext(const ThreadContext& ctx) override;
 
     void PrepareReschedule() override;
-    void ExecuteInstructions(int num_instructions) override;
 
     void ClearInstructionCache() override;
     void PageTableChanged() override;
@@ -47,4 +48,6 @@ private:
     std::unique_ptr<ARM_Dynarmic_Callbacks> cb;
     std::unique_ptr<Dynarmic::A64::Jit> jit;
     ARM_Unicorn inner_unicorn;
+
+    Memory::PageTable* current_page_table = nullptr;
 };

src/core/arm/unicorn/arm_unicorn.cpp

@@ -2,6 +2,7 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <algorithm>
 #include <unicorn/arm64.h>
 #include "common/assert.h"
 #include "common/microprofile.h"
@@ -52,7 +53,7 @@ static bool UnmappedMemoryHook(uc_engine* uc, uc_mem_type type, u64 addr, int si
                                void* user_data) {
     ARM_Interface::ThreadContext ctx{};
     Core::CPU().SaveContext(ctx);
-    ASSERT_MSG(false, "Attempted to read from unmapped memory: 0x%llx, pc=0x%llx, lr=0x%llx", addr,
+    ASSERT_MSG(false, "Attempted to read from unmapped memory: 0x%lx, pc=0x%lx, lr=0x%lx", addr,
                ctx.pc, ctx.cpu_registers[30]);
     return {};
 }
@@ -153,6 +154,14 @@ void ARM_Unicorn::SetTlsAddress(VAddr base) {
     CHECKED(uc_reg_write(uc, UC_ARM64_REG_TPIDRRO_EL0, &base));
 }
 
+void ARM_Unicorn::Run() {
+    ExecuteInstructions(std::max(CoreTiming::GetDowncount(), 0));
+}
+
+void ARM_Unicorn::Step() {
+    ExecuteInstructions(1);
+}
+
 MICROPROFILE_DEFINE(ARM_Jit, "ARM JIT", "ARM JIT", MP_RGB(255, 64, 64));
 
 void ARM_Unicorn::ExecuteInstructions(int num_instructions) {

src/core/arm/unicorn/arm_unicorn.h

@@ -30,7 +30,9 @@ public:
     void SaveContext(ThreadContext& ctx) override;
     void LoadContext(const ThreadContext& ctx) override;
     void PrepareReschedule() override;
-    void ExecuteInstructions(int num_instructions) override;
+    void ExecuteInstructions(int num_instructions);
+    void Run() override;
+    void Step() override;
     void ClearInstructionCache() override;
     void PageTableChanged() override{};
 

src/core/core.cpp

@@ -26,7 +26,7 @@ namespace Core {
 
 /*static*/ System System::s_instance;
 
-System::ResultStatus System::RunLoop(int tight_loop) {
+System::ResultStatus System::RunLoop(bool tight_loop) {
     status = ResultStatus::Success;
     if (!cpu_core) {
         return ResultStatus::ErrorNotInitialized;
@@ -40,7 +40,7 @@ System::ResultStatus System::RunLoop(int tight_loop) {
         if (GDBStub::GetCpuHaltFlag()) {
             if (GDBStub::GetCpuStepFlag()) {
                 GDBStub::SetCpuStepFlag(false);
-                tight_loop = 1;
+                tight_loop = false;
             } else {
                 return ResultStatus::Success;
             }
@@ -56,7 +56,11 @@ System::ResultStatus System::RunLoop(int tight_loop) {
         PrepareReschedule();
     } else {
         CoreTiming::Advance();
-        cpu_core->Run(tight_loop);
+        if (tight_loop) {
+            cpu_core->Run();
+        } else {
+            cpu_core->Step();
+        }
     }
 
     HW::Update();
@@ -66,7 +70,7 @@ System::ResultStatus System::RunLoop(int tight_loop) {
 }
 
 System::ResultStatus System::SingleStep() {
-    return RunLoop(1);
+    return RunLoop(false);
 }
 
 System::ResultStatus System::Load(EmuWindow* emu_window, const std::string& filepath) {
@@ -95,14 +99,15 @@ System::ResultStatus System::Load(EmuWindow* emu_window, const std::string& file
 
     ResultStatus init_result{Init(emu_window, system_mode.first.get())};
     if (init_result != ResultStatus::Success) {
-        LOG_CRITICAL(Core, "Failed to initialize system (Error %i)!", init_result);
+        LOG_CRITICAL(Core, "Failed to initialize system (Error %i)!",
+                     static_cast<int>(init_result));
         System::Shutdown();
         return init_result;
     }
 
     const Loader::ResultStatus load_result{app_loader->Load(current_process)};
     if (Loader::ResultStatus::Success != load_result) {
-        LOG_CRITICAL(Core, "Failed to load ROM (Error %i)!", load_result);
+        LOG_CRITICAL(Core, "Failed to load ROM (Error %i)!", static_cast<int>(load_result));
         System::Shutdown();
 
         switch (load_result) {

src/core/core.h

@@ -53,10 +53,10 @@ public:
      * is not required to do a full dispatch with each instruction. NOTE: the number of instructions
      * requested is not guaranteed to run, as this will be interrupted preemptively if a hardware
      * update is requested (e.g. on a thread switch).
-     * @param tight_loop Number of instructions to execute.
+     * @param tight_loop If false, the CPU single-steps.
      * @return Result status, indicating whether or not the operation succeeded.
      */
-    ResultStatus RunLoop(int tight_loop = 100000);
+    ResultStatus RunLoop(bool tight_loop = true);
 
     /**
      * Step the CPU one instruction

src/core/gdbstub/gdbstub.cpp

@@ -693,7 +693,7 @@ static void ReadMemory() {
     u64 len =
         HexToLong(start_offset, static_cast<u64>((command_buffer + command_length) - start_offset));
 
-    LOG_DEBUG(Debug_GDBStub, "gdb: addr: %016llx len: %016llx\n", addr, len);
+    LOG_DEBUG(Debug_GDBStub, "gdb: addr: %016lx len: %016lx\n", addr, len);
 
     if (len * 2 > sizeof(reply)) {
         SendReply("E01");

src/core/hle/kernel/hle_ipc.cpp

@@ -7,6 +7,7 @@
 #include "common/common_funcs.h"
 #include "common/common_types.h"
 #include "core/hle/ipc_helpers.h"
+#include "core/hle/kernel/event.h"
 #include "core/hle/kernel/handle_table.h"
 #include "core/hle/kernel/hle_ipc.h"
 #include "core/hle/kernel/kernel.h"
@@ -26,6 +27,32 @@ void SessionRequestHandler::ClientDisconnected(SharedPtr<ServerSession> server_s
     boost::range::remove_erase(connected_sessions, server_session);
 }
 
+SharedPtr<Event> HLERequestContext::SleepClientThread(SharedPtr<Thread> thread,
+                                                      const std::string& reason, u64 timeout,
+                                                      WakeupCallback&& callback) {
+
+    // Put the client thread to sleep until the wait event is signaled or the timeout expires.
+    thread->wakeup_callback =
+        [context = *this, callback](ThreadWakeupReason reason, SharedPtr<Thread> thread,
+                                    SharedPtr<WaitObject> object, size_t index) mutable -> bool {
+        ASSERT(thread->status == THREADSTATUS_WAIT_HLE_EVENT);
+        callback(thread, context, reason);
+        context.WriteToOutgoingCommandBuffer(*thread);
+        return true;
+    };
+
+    auto event = Kernel::Event::Create(Kernel::ResetType::OneShot, "HLE Pause Event: " + reason);
+    thread->status = THREADSTATUS_WAIT_HLE_EVENT;
+    thread->wait_objects = {event};
+    event->AddWaitingThread(thread);
+
+    if (timeout > 0) {
+        thread->WakeAfterDelay(timeout);
+    }
+
+    return event;
+}
+
 HLERequestContext::HLERequestContext(SharedPtr<Kernel::ServerSession> server_session)
     : server_session(std::move(server_session)) {
     cmd_buf[0] = 0;
@@ -35,7 +62,7 @@ HLERequestContext::~HLERequestContext() = default;
 
 void HLERequestContext::ParseCommandBuffer(u32_le* src_cmdbuf, bool incoming) {
     IPC::RequestParser rp(src_cmdbuf);
-    command_header = std::make_unique<IPC::CommandHeader>(rp.PopRaw<IPC::CommandHeader>());
+    command_header = std::make_shared<IPC::CommandHeader>(rp.PopRaw<IPC::CommandHeader>());
 
     if (command_header->type == IPC::CommandType::Close) {
         // Close does not populate the rest of the IPC header
@@ -45,7 +72,7 @@ void HLERequestContext::ParseCommandBuffer(u32_le* src_cmdbuf, bool incoming) {
     // If handle descriptor is present, add size of it
     if (command_header->enable_handle_descriptor) {
         handle_descriptor_header =
-            std::make_unique<IPC::HandleDescriptorHeader>(rp.PopRaw<IPC::HandleDescriptorHeader>());
+            std::make_shared<IPC::HandleDescriptorHeader>(rp.PopRaw<IPC::HandleDescriptorHeader>());
         if (handle_descriptor_header->send_current_pid) {
             rp.Skip(2, false);
         }
@@ -88,7 +115,7 @@ void HLERequestContext::ParseCommandBuffer(u32_le* src_cmdbuf, bool incoming) {
         // All outgoing domain messages have the domain header, if only incoming has it
         if (incoming || domain_message_header) {
             domain_message_header =
-                std::make_unique<IPC::DomainMessageHeader>(rp.PopRaw<IPC::DomainMessageHeader>());
+                std::make_shared<IPC::DomainMessageHeader>(rp.PopRaw<IPC::DomainMessageHeader>());
         } else {
             if (Session()->IsDomain())
                 LOG_WARNING(IPC, "Domain request has no DomainMessageHeader!");
@@ -96,7 +123,7 @@ void HLERequestContext::ParseCommandBuffer(u32_le* src_cmdbuf, bool incoming) {
     }
 
     data_payload_header =
-        std::make_unique<IPC::DataPayloadHeader>(rp.PopRaw<IPC::DataPayloadHeader>());
+        std::make_shared<IPC::DataPayloadHeader>(rp.PopRaw<IPC::DataPayloadHeader>());
 
     data_payload_offset = rp.GetCurrentOffset();
 
@@ -159,8 +186,11 @@ ResultCode HLERequestContext::PopulateFromIncomingCommandBuffer(u32_le* src_cmdb
     return RESULT_SUCCESS;
 }
 
-ResultCode HLERequestContext::WriteToOutgoingCommandBuffer(u32_le* dst_cmdbuf, Process& dst_process,
-                                                           HandleTable& dst_table) {
+ResultCode HLERequestContext::WriteToOutgoingCommandBuffer(Thread& thread) {
+    std::array<u32, IPC::COMMAND_BUFFER_LENGTH> dst_cmdbuf;
+    Memory::ReadBlock(*thread.owner_process, thread.GetTLSAddress(), dst_cmdbuf.data(),
+                      dst_cmdbuf.size() * sizeof(u32));
+
     // The header was already built in the internal command buffer. Attempt to parse it to verify
     // the integrity and then copy it over to the target command buffer.
     ParseCommandBuffer(cmd_buf.data(), false);
@@ -171,7 +201,7 @@ ResultCode HLERequestContext::WriteToOutgoingCommandBuffer(u32_le* dst_cmdbuf, P
     if (domain_message_header)
         size -= sizeof(IPC::DomainMessageHeader) / sizeof(u32);
 
-    std::copy_n(cmd_buf.begin(), size, dst_cmdbuf);
+    std::copy_n(cmd_buf.begin(), size, dst_cmdbuf.data());
 
     if (command_header->enable_handle_descriptor) {
         ASSERT_MSG(!move_objects.empty() || !copy_objects.empty(),
@@ -213,6 +243,11 @@ ResultCode HLERequestContext::WriteToOutgoingCommandBuffer(u32_le* dst_cmdbuf, P
             dst_cmdbuf[domain_offset++] = static_cast<u32_le>(request_handlers.size());
         }
     }
+
+    // Copy the translated command buffer back into the thread's command buffer area.
+    Memory::WriteBlock(*thread.owner_process, thread.GetTLSAddress(), dst_cmdbuf.data(),
+                       dst_cmdbuf.size() * sizeof(u32));
+
     return RESULT_SUCCESS;
 }
 
@@ -234,7 +269,7 @@ std::vector<u8> HLERequestContext::ReadBuffer() const {
 size_t HLERequestContext::WriteBuffer(const void* buffer, size_t size) const {
     const bool is_buffer_b{BufferDescriptorB().size() && BufferDescriptorB()[0].Size()};
 
-    ASSERT_MSG(size <= GetWriteBufferSize(), "Size %d is too big", size);
+    ASSERT_MSG(size <= GetWriteBufferSize(), "Size %lx is too big", size);
 
     if (is_buffer_b) {
         Memory::WriteBlock(BufferDescriptorB()[0].Address(), buffer, size);

src/core/hle/kernel/hle_ipc.h

@@ -6,6 +6,7 @@
 
 #include <array>
 #include <memory>
+#include <string>
 #include <vector>
 #include <boost/container/small_vector.hpp>
 #include "common/common_types.h"
@@ -13,6 +14,7 @@
 #include "core/hle/ipc.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/server_session.h"
+#include "core/hle/kernel/thread.h"
 
 namespace Service {
 class ServiceFrameworkBase;
@@ -24,6 +26,7 @@ class Domain;
 class HandleTable;
 class HLERequestContext;
 class Process;
+class Event;
 
 /**
  * Interface implemented by HLE Session handlers.
@@ -102,14 +105,31 @@ public:
         return server_session;
     }
 
+    using WakeupCallback = std::function<void(SharedPtr<Thread> thread, HLERequestContext& context,
+                                              ThreadWakeupReason reason)>;
+
+    /**
+     * Puts the specified guest thread to sleep until the returned event is signaled or until the
+     * specified timeout expires.
+     * @param thread Thread to be put to sleep.
+     * @param reason Reason for pausing the thread, to be used for debugging purposes.
+     * @param timeout Timeout in nanoseconds after which the thread will be awoken and the callback
+     * invoked with a Timeout reason.
+     * @param callback Callback to be invoked when the thread is resumed. This callback must write
+     * the entire command response once again, regardless of the state of it before this function
+     * was called.
+     * @returns Event that when signaled will resume the thread and call the callback function.
+     */
+    SharedPtr<Event> SleepClientThread(SharedPtr<Thread> thread, const std::string& reason,
+                                       u64 timeout, WakeupCallback&& callback);
+
     void ParseCommandBuffer(u32_le* src_cmdbuf, bool incoming);
 
     /// Populates this context with data from the requesting process/thread.
     ResultCode PopulateFromIncomingCommandBuffer(u32_le* src_cmdbuf, Process& src_process,
                                                  HandleTable& src_table);
     /// Writes data from this context back to the requesting process/thread.
-    ResultCode WriteToOutgoingCommandBuffer(u32_le* dst_cmdbuf, Process& dst_process,
-                                            HandleTable& dst_table);
+    ResultCode WriteToOutgoingCommandBuffer(Thread& thread);
 
     u32_le GetCommand() const {
         return command;
@@ -139,7 +159,7 @@ public:
         return buffer_c_desciptors;
     }
 
-    const std::unique_ptr<IPC::DomainMessageHeader>& GetDomainMessageHeader() const {
+    const std::shared_ptr<IPC::DomainMessageHeader>& GetDomainMessageHeader() const {
         return domain_message_header;
     }
 
@@ -212,10 +232,10 @@ private:
     boost::container::small_vector<SharedPtr<Object>, 8> copy_objects;
     boost::container::small_vector<std::shared_ptr<SessionRequestHandler>, 8> domain_objects;
 
-    std::unique_ptr<IPC::CommandHeader> command_header;
-    std::unique_ptr<IPC::HandleDescriptorHeader> handle_descriptor_header;
-    std::unique_ptr<IPC::DataPayloadHeader> data_payload_header;
-    std::unique_ptr<IPC::DomainMessageHeader> domain_message_header;
+    std::shared_ptr<IPC::CommandHeader> command_header;
+    std::shared_ptr<IPC::HandleDescriptorHeader> handle_descriptor_header;
+    std::shared_ptr<IPC::DataPayloadHeader> data_payload_header;
+    std::shared_ptr<IPC::DomainMessageHeader> domain_message_header;
     std::vector<IPC::BufferDescriptorX> buffer_x_desciptors;
     std::vector<IPC::BufferDescriptorABW> buffer_a_desciptors;
     std::vector<IPC::BufferDescriptorABW> buffer_b_desciptors;

src/core/hle/kernel/object_address_table.cpp

@@ -10,12 +10,12 @@ namespace Kernel {
 ObjectAddressTable g_object_address_table;
 
 void ObjectAddressTable::Insert(VAddr addr, SharedPtr<Object> obj) {
-    ASSERT_MSG(objects.find(addr) == objects.end(), "Object already exists with addr=0x%llx", addr);
+    ASSERT_MSG(objects.find(addr) == objects.end(), "Object already exists with addr=0x%lx", addr);
     objects[addr] = obj;
 }
 
 void ObjectAddressTable::Close(VAddr addr) {
-    ASSERT_MSG(objects.find(addr) != objects.end(), "Object does not exist with addr=0x%llx", addr);
+    ASSERT_MSG(objects.find(addr) != objects.end(), "Object does not exist with addr=0x%lx", addr);
     objects.erase(addr);
 }
 

src/core/hle/kernel/server_session.cpp

@@ -78,7 +78,8 @@ ResultCode ServerSession::HandleDomainSyncRequest(Kernel::HLERequestContext& con
         }
         }
 
-        LOG_CRITICAL(IPC, "Unknown domain command=%d", domain_message_header->command.Value());
+        LOG_CRITICAL(IPC, "Unknown domain command=%d",
+                     static_cast<int>(domain_message_header->command.Value()));
         ASSERT(false);
     }
 

src/core/hle/kernel/shared_memory.cpp

@@ -107,7 +107,7 @@ ResultCode SharedMemory::Map(Process* target_process, VAddr address, MemoryPermi
 
     // Error out if the requested permissions don't match what the creator process allows.
     if (static_cast<u32>(permissions) & ~static_cast<u32>(own_other_permissions)) {
-        LOG_ERROR(Kernel, "cannot map id=%u, address=0x%llx name=%s, permissions don't match",
+        LOG_ERROR(Kernel, "cannot map id=%u, address=0x%lx name=%s, permissions don't match",
                   GetObjectId(), address, name.c_str());
         return ERR_INVALID_COMBINATION;
     }
@@ -115,7 +115,7 @@ ResultCode SharedMemory::Map(Process* target_process, VAddr address, MemoryPermi
     // Error out if the provided permissions are not compatible with what the creator process needs.
     if (other_permissions != MemoryPermission::DontCare &&
         static_cast<u32>(this->permissions) & ~static_cast<u32>(other_permissions)) {
-        LOG_ERROR(Kernel, "cannot map id=%u, address=0x%llx name=%s, permissions don't match",
+        LOG_ERROR(Kernel, "cannot map id=%u, address=0x%lx name=%s, permissions don't match",
                   GetObjectId(), address, name.c_str());
         return ERR_WRONG_PERMISSION;
     }
@@ -126,7 +126,7 @@ ResultCode SharedMemory::Map(Process* target_process, VAddr address, MemoryPermi
     if (address != 0) {
         // TODO(shinyquagsire23): Check for virtual/mappable memory here too?
         if (address >= Memory::HEAP_VADDR && address < Memory::HEAP_VADDR_END) {
-            LOG_ERROR(Kernel, "cannot map id=%u, address=0x%llx name=%s, invalid address",
+            LOG_ERROR(Kernel, "cannot map id=%u, address=0x%lx name=%s, invalid address",
                       GetObjectId(), address, name.c_str());
             return ERR_INVALID_ADDRESS;
         }
@@ -143,10 +143,9 @@ ResultCode SharedMemory::Map(Process* target_process, VAddr address, MemoryPermi
     auto result = target_process->vm_manager.MapMemoryBlock(
         target_address, backing_block, backing_block_offset, size, MemoryState::Shared);
     if (result.Failed()) {
-        LOG_ERROR(
-            Kernel,
-            "cannot map id=%u, target_address=0x%llx name=%s, error mapping to virtual memory",
-            GetObjectId(), target_address, name.c_str());
+        LOG_ERROR(Kernel,
+                  "cannot map id=%u, target_address=0x%lx name=%s, error mapping to virtual memory",
+                  GetObjectId(), target_address, name.c_str());
         return result.Code();
     }
 

src/core/hle/kernel/svc.cpp

@@ -39,7 +39,7 @@ static ResultCode SetHeapSize(VAddr* heap_addr, u64 heap_size) {
 }
 
 static ResultCode SetMemoryAttribute(VAddr addr, u64 size, u32 state0, u32 state1) {
-    LOG_WARNING(Kernel_SVC, "(STUBBED) called, addr=0x%llx", addr);
+    LOG_WARNING(Kernel_SVC, "(STUBBED) called, addr=0x%lx", addr);
     return RESULT_SUCCESS;
 }
 
@@ -750,7 +750,7 @@ static ResultCode ResetSignal(Handle handle) {
 
 /// Creates a TransferMemory object
 static ResultCode CreateTransferMemory(Handle* handle, VAddr addr, u64 size, u32 permissions) {
-    LOG_WARNING(Kernel_SVC, "(STUBBED) called addr=0x%llx, size=0x%llx, perms=%08X", addr, size,
+    LOG_WARNING(Kernel_SVC, "(STUBBED) called addr=0x%lx, size=0x%lx, perms=%08X", addr, size,
                 permissions);
     *handle = 0;
     return RESULT_SUCCESS;

src/core/hle/kernel/thread.cpp

@@ -55,16 +55,6 @@ inline static u32 const NewThreadId() {
 Thread::Thread() {}
 Thread::~Thread() {}
 
-/**
- * Check if the specified thread is waiting on the specified address to be arbitrated
- * @param thread The thread to test
- * @param wait_address The address to test against
- * @return True if the thread is waiting, false otherwise
- */
-static bool CheckWait_AddressArbiter(const Thread* thread, VAddr wait_address) {
-    return thread->status == THREADSTATUS_WAIT_ARB && wait_address == thread->wait_address;
-}
-
 void Thread::Stop() {
     // Cancel any outstanding wakeup events for this thread
     CoreTiming::UnscheduleEvent(ThreadWakeupEventType, callback_handle);
@@ -102,12 +92,6 @@ void WaitCurrentThread_Sleep() {
     thread->status = THREADSTATUS_WAIT_SLEEP;
 }
 
-void WaitCurrentThread_ArbitrateAddress(VAddr wait_address) {
-    Thread* thread = GetCurrentThread();
-    thread->wait_address = wait_address;
-    thread->status = THREADSTATUS_WAIT_ARB;
-}
-
 void ExitCurrentThread() {
     Thread* thread = GetCurrentThread();
     thread->Stop();
@@ -129,7 +113,8 @@ static void ThreadWakeupCallback(u64 thread_handle, int cycles_late) {
     bool resume = true;
 
     if (thread->status == THREADSTATUS_WAIT_SYNCH_ANY ||
-        thread->status == THREADSTATUS_WAIT_SYNCH_ALL || thread->status == THREADSTATUS_WAIT_ARB) {
+        thread->status == THREADSTATUS_WAIT_SYNCH_ALL ||
+        thread->status == THREADSTATUS_WAIT_HLE_EVENT) {
 
         // Remove the thread from each of its waiting objects' waitlists
         for (auto& object : thread->wait_objects)
@@ -163,7 +148,7 @@ void Thread::ResumeFromWait() {
     switch (status) {
     case THREADSTATUS_WAIT_SYNCH_ALL:
     case THREADSTATUS_WAIT_SYNCH_ANY:
-    case THREADSTATUS_WAIT_ARB:
+    case THREADSTATUS_WAIT_HLE_EVENT:
     case THREADSTATUS_WAIT_SLEEP:
     case THREADSTATUS_WAIT_IPC:
         break;

src/core/hle/kernel/thread.h

@@ -38,7 +38,7 @@ enum ThreadProcessorId : s32 {
 enum ThreadStatus {
     THREADSTATUS_RUNNING,        ///< Currently running
     THREADSTATUS_READY,          ///< Ready to run
-    THREADSTATUS_WAIT_ARB,       ///< Waiting on an address arbiter
+    THREADSTATUS_WAIT_HLE_EVENT, ///< Waiting for hle event to finish
     THREADSTATUS_WAIT_SLEEP,     ///< Waiting due to a SleepThread SVC
     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

src/core/hle/kernel/wait_object.cpp

@@ -39,7 +39,8 @@ SharedPtr<Thread> WaitObject::GetHighestPriorityReadyThread() {
     for (const auto& thread : waiting_threads) {
         // The list of waiting threads must not contain threads that are not waiting to be awakened.
         ASSERT_MSG(thread->status == THREADSTATUS_WAIT_SYNCH_ANY ||
-                       thread->status == THREADSTATUS_WAIT_SYNCH_ALL,
+                       thread->status == THREADSTATUS_WAIT_SYNCH_ALL ||
+                       thread->status == THREADSTATUS_WAIT_HLE_EVENT,
                    "Inconsistent thread statuses in waiting_threads");
 
         if (thread->current_priority >= candidate_priority)

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%llx, length=0x%llx", offset, length);
+        LOG_DEBUG(Service_FS, "called, offset=0x%ld, length=0x%ld", offset, length);
 
         // Error checking
         if (length < 0) {
@@ -86,7 +86,7 @@ private:
         const s64 offset = rp.Pop<s64>();
         const s64 length = rp.Pop<s64>();
 
-        LOG_DEBUG(Service_FS, "called, offset=0x%llx, length=0x%llx", offset, length);
+        LOG_DEBUG(Service_FS, "called, offset=0x%ld, length=0x%ld", offset, length);
 
         // Error checking
         if (length < 0) {
@@ -123,7 +123,7 @@ private:
         const s64 offset = rp.Pop<s64>();
         const s64 length = rp.Pop<s64>();
 
-        LOG_DEBUG(Service_FS, "called, offset=0x%llx, length=0x%llx", offset, length);
+        LOG_DEBUG(Service_FS, "called, offset=0x%ld, length=0x%ld", offset, length);
 
         // Error checking
         if (length < 0) {

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

@@ -23,8 +23,8 @@ void nvdisp_disp0::flip(u32 buffer_handle, u32 offset, u32 format, u32 width, u3
                         u32 stride, NVFlinger::BufferQueue::BufferTransformFlags transform) {
     VAddr addr = nvmap_dev->GetObjectAddress(buffer_handle);
     LOG_WARNING(Service,
-                "Drawing from address %llx offset %08X Width %u Height %u Stride %u Format %u",
-                addr, offset, width, height, stride, format);
+                "Drawing from address %lx offset %08X Width %u Height %u Stride %u Format %u", addr,
+                offset, width, height, stride, format);
 
     using PixelFormat = RendererBase::FramebufferInfo::PixelFormat;
     using Flags = NVFlinger::BufferQueue::BufferTransformFlags;

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

@@ -26,24 +26,30 @@ void BufferQueue::SetPreallocatedBuffer(u32 slot, IGBPBuffer& igbp_buffer) {
     LOG_WARNING(Service, "Adding graphics buffer %u", slot);
 
     queue.emplace_back(buffer);
+
+    if (buffer_wait_event) {
+        buffer_wait_event->Signal();
+    }
 }
 
-u32 BufferQueue::DequeueBuffer(u32 pixel_format, u32 width, u32 height) {
+boost::optional<u32> BufferQueue::DequeueBuffer(u32 width, u32 height) {
     auto itr = std::find_if(queue.begin(), queue.end(), [&](const Buffer& buffer) {
         // Only consider free buffers. Buffers become free once again after they've been Acquired
         // and Released by the compositor, see the NVFlinger::Compose method.
-        if (buffer.status != Buffer::Status::Free)
+        if (buffer.status != Buffer::Status::Free) {
             return false;
+        }
 
         // Make sure that the parameters match.
         return buffer.igbp_buffer.width == width && buffer.igbp_buffer.height == height;
     });
+
     if (itr == queue.end()) {
-        LOG_CRITICAL(Service_NVDRV, "no free buffers for pixel_format=%d, width=%d, height=%d",
-                     pixel_format, width, height);
-        itr = queue.begin();
+        return boost::none;
     }
 
+    buffer_wait_event = nullptr;
+
     itr->status = Buffer::Status::Dequeued;
     return itr->slot;
 }
@@ -81,6 +87,10 @@ void BufferQueue::ReleaseBuffer(u32 slot) {
     ASSERT(itr != queue.end());
     ASSERT(itr->status == Buffer::Status::Acquired);
     itr->status = Buffer::Status::Free;
+
+    if (buffer_wait_event) {
+        buffer_wait_event->Signal();
+    }
 }
 
 u32 BufferQueue::Query(QueryType type) {
@@ -96,5 +106,10 @@ u32 BufferQueue::Query(QueryType type) {
     return 0;
 }
 
+void BufferQueue::SetBufferWaitEvent(Kernel::SharedPtr<Kernel::Event>&& wait_event) {
+    ASSERT_MSG(!buffer_wait_event, "buffer_wait_event only supports a single waiting thread!");
+    buffer_wait_event = std::move(wait_event);
+}
+
 } // namespace NVFlinger
 } // namespace Service

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

@@ -69,12 +69,13 @@ public:
     };
 
     void SetPreallocatedBuffer(u32 slot, IGBPBuffer& buffer);
-    u32 DequeueBuffer(u32 pixel_format, u32 width, u32 height);
+    boost::optional<u32> DequeueBuffer(u32 width, u32 height);
     const IGBPBuffer& RequestBuffer(u32 slot) const;
     void QueueBuffer(u32 slot, BufferTransformFlags transform);
     boost::optional<const Buffer&> AcquireBuffer();
     void ReleaseBuffer(u32 slot);
     u32 Query(QueryType type);
+    void SetBufferWaitEvent(Kernel::SharedPtr<Kernel::Event>&& wait_event);
 
     u32 GetId() const {
         return id;
@@ -90,6 +91,9 @@ private:
 
     std::vector<Buffer> queue;
     Kernel::SharedPtr<Kernel::Event> native_handle;
+
+    /// Used to signal waiting thread when no buffers are available
+    Kernel::SharedPtr<Kernel::Event> buffer_wait_event;
 };
 
 } // namespace NVFlinger

src/core/hle/service/service.cpp

@@ -149,11 +149,10 @@ ResultCode ServiceFrameworkBase::HandleSyncRequest(Kernel::HLERequestContext& co
         break;
     }
     default:
-        UNIMPLEMENTED_MSG("command_type=%d", context.GetCommandType());
+        UNIMPLEMENTED_MSG("command_type=%d", static_cast<int>(context.GetCommandType()));
     }
 
-    u32* cmd_buf = (u32*)Memory::GetPointer(Kernel::GetCurrentThread()->GetTLSAddress());
-    context.WriteToOutgoingCommandBuffer(cmd_buf, *Core::CurrentProcess(), Kernel::g_handle_table);
+    context.WriteToOutgoingCommandBuffer(*Kernel::GetCurrentThread());
 
     return RESULT_SUCCESS;
 }

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

@@ -107,7 +107,7 @@ private:
         IPC::RequestParser rp{ctx};
         u64 posix_time = rp.Pop<u64>();
 
-        LOG_WARNING(Service_Time, "(STUBBED) called, posix_time=0x%016llX", posix_time);
+        LOG_WARNING(Service_Time, "(STUBBED) called, posix_time=0x%016lX", posix_time);
 
         CalendarTime calendar_time{2018, 1, 1, 0, 0, 0};
         CalendarAdditionalInfo additional_info{};

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

@@ -471,7 +471,7 @@ private:
         u32 flags = rp.Pop<u32>();
         auto buffer_queue = nv_flinger->GetBufferQueue(id);
 
-        LOG_DEBUG(Service_VI, "called, transaction=%x", transaction);
+        LOG_DEBUG(Service_VI, "called, transaction=%x", static_cast<u32>(transaction));
 
         if (transaction == TransactionId::Connect) {
             IGBPConnectRequestParcel request{ctx.ReadBuffer()};
@@ -486,12 +486,30 @@ private:
             ctx.WriteBuffer(response.Serialize());
         } else if (transaction == TransactionId::DequeueBuffer) {
             IGBPDequeueBufferRequestParcel request{ctx.ReadBuffer()};
+            const u32 width{request.data.width};
+            const u32 height{request.data.height};
+            boost::optional<u32> slot = buffer_queue->DequeueBuffer(width, height);
 
-            u32 slot = buffer_queue->DequeueBuffer(request.data.pixel_format, request.data.width,
-                                                   request.data.height);
-
-            IGBPDequeueBufferResponseParcel response{slot};
-            ctx.WriteBuffer(response.Serialize());
+            if (slot != boost::none) {
+                // Buffer is available
+                IGBPDequeueBufferResponseParcel response{*slot};
+                ctx.WriteBuffer(response.Serialize());
+            } else {
+                // Wait the current thread until a buffer becomes available
+                auto wait_event = ctx.SleepClientThread(
+                    Kernel::GetCurrentThread(), "IHOSBinderDriver::DequeueBuffer", -1,
+                    [=](Kernel::SharedPtr<Kernel::Thread> thread, Kernel::HLERequestContext& ctx,
+                        ThreadWakeupReason reason) {
+                        // Repeat TransactParcel DequeueBuffer when a buffer is available
+                        auto buffer_queue = nv_flinger->GetBufferQueue(id);
+                        boost::optional<u32> slot = buffer_queue->DequeueBuffer(width, height);
+                        IGBPDequeueBufferResponseParcel response{*slot};
+                        ctx.WriteBuffer(response.Serialize());
+                        IPC::ResponseBuilder rb{ctx, 2};
+                        rb.Push(RESULT_SUCCESS);
+                    });
+                buffer_queue->SetBufferWaitEvent(std::move(wait_event));
+            }
         } else if (transaction == TransactionId::RequestBuffer) {
             IGBPRequestBufferRequestParcel request{ctx.ReadBuffer()};
 

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", rela.type);
+            LOG_CRITICAL(Loader, "Unknown relocation type: %d", static_cast<int>(rela.type));
             break;
         }
     }

src/video_core/CMakeLists.txt

@@ -7,6 +7,7 @@ add_library(video_core STATIC
     engines/maxwell_3d.h
     engines/maxwell_compute.cpp
     engines/maxwell_compute.h
+    gpu.cpp
     gpu.h
     memory_manager.cpp
     memory_manager.h

src/video_core/engines/maxwell_3d.cpp

@@ -12,7 +12,9 @@ namespace Engines {
 constexpr u32 MacroRegistersStart = 0xE00;
 
 const std::unordered_map<u32, Maxwell3D::MethodInfo> Maxwell3D::method_handlers = {
+    {0xE1A, {"BindTextureInfoBuffer", 1, &Maxwell3D::BindTextureInfoBuffer}},
     {0xE24, {"SetShader", 5, &Maxwell3D::SetShader}},
+    {0xE2A, {"BindStorageBuffer", 1, &Maxwell3D::BindStorageBuffer}},
 };
 
 Maxwell3D::Maxwell3D(MemoryManager& memory_manager) : memory_manager(memory_manager) {}
@@ -83,6 +85,25 @@ void Maxwell3D::WriteReg(u32 method, u32 value, u32 remaining_params) {
         ASSERT_MSG(regs.code_address.CodeAddress() == 0, "Unexpected CODE_ADDRESS register value.");
         break;
     }
+    case MAXWELL3D_REG_INDEX(const_buffer.cb_data[0]):
+    case MAXWELL3D_REG_INDEX(const_buffer.cb_data[1]):
+    case MAXWELL3D_REG_INDEX(const_buffer.cb_data[2]):
+    case MAXWELL3D_REG_INDEX(const_buffer.cb_data[3]):
+    case MAXWELL3D_REG_INDEX(const_buffer.cb_data[4]):
+    case MAXWELL3D_REG_INDEX(const_buffer.cb_data[5]):
+    case MAXWELL3D_REG_INDEX(const_buffer.cb_data[6]):
+    case MAXWELL3D_REG_INDEX(const_buffer.cb_data[7]):
+    case MAXWELL3D_REG_INDEX(const_buffer.cb_data[8]):
+    case MAXWELL3D_REG_INDEX(const_buffer.cb_data[9]):
+    case MAXWELL3D_REG_INDEX(const_buffer.cb_data[10]):
+    case MAXWELL3D_REG_INDEX(const_buffer.cb_data[11]):
+    case MAXWELL3D_REG_INDEX(const_buffer.cb_data[12]):
+    case MAXWELL3D_REG_INDEX(const_buffer.cb_data[13]):
+    case MAXWELL3D_REG_INDEX(const_buffer.cb_data[14]):
+    case MAXWELL3D_REG_INDEX(const_buffer.cb_data[15]): {
+        ProcessCBData(value);
+        break;
+    }
     case MAXWELL3D_REG_INDEX(cb_bind[0].raw_config): {
         ProcessCBBind(Regs::ShaderStage::Vertex);
         break;
@@ -132,7 +153,8 @@ void Maxwell3D::ProcessQueryGet() {
         break;
     }
     default:
-        UNIMPLEMENTED_MSG("Query mode %u not implemented", regs.query.query_get.mode.Value());
+        UNIMPLEMENTED_MSG("Query mode %u not implemented",
+                          static_cast<u32>(regs.query.query_get.mode.Value()));
     }
 }
 
@@ -140,6 +162,23 @@ void Maxwell3D::DrawArrays() {
     LOG_WARNING(HW_GPU, "Game requested a DrawArrays, ignoring");
 }
 
+void Maxwell3D::BindTextureInfoBuffer(const std::vector<u32>& parameters) {
+    /**
+     * Parameters description:
+     * [0] = Shader stage, usually 4 for FragmentShader
+     */
+
+    u32 stage = parameters[0];
+
+    // Perform the same operations as the real macro code.
+    GPUVAddr address = static_cast<GPUVAddr>(regs.tex_info_buffers.address[stage]) << 8;
+    u32 size = regs.tex_info_buffers.size[stage];
+
+    regs.const_buffer.cb_size = size;
+    regs.const_buffer.cb_address_high = address >> 32;
+    regs.const_buffer.cb_address_low = address & 0xFFFFFFFF;
+}
+
 void Maxwell3D::SetShader(const std::vector<u32>& parameters) {
     /**
      * Parameters description:
@@ -181,6 +220,26 @@ void Maxwell3D::SetShader(const std::vector<u32>& parameters) {
     ProcessCBBind(shader_stage);
 }
 
+void Maxwell3D::BindStorageBuffer(const std::vector<u32>& parameters) {
+    /**
+     * Parameters description:
+     * [0] = Buffer offset >> 2
+     */
+
+    u32 buffer_offset = parameters[0] << 2;
+
+    // Perform the same operations as the real macro code.
+    // Note: This value is hardcoded in the macro's code.
+    static constexpr u32 DefaultCBSize = 0x5F00;
+    regs.const_buffer.cb_size = DefaultCBSize;
+
+    GPUVAddr address = regs.ssbo_info.BufferAddress();
+    regs.const_buffer.cb_address_high = address >> 32;
+    regs.const_buffer.cb_address_low = address & 0xFFFFFFFF;
+
+    regs.const_buffer.cb_pos = buffer_offset;
+}
+
 void Maxwell3D::ProcessCBBind(Regs::ShaderStage stage) {
     // Bind the buffer currently in CB_ADDRESS to the specified index in the desired shader stage.
     auto& shader = state.shader_stages[static_cast<size_t>(stage)];
@@ -194,5 +253,22 @@ void Maxwell3D::ProcessCBBind(Regs::ShaderStage stage) {
     buffer.size = regs.const_buffer.cb_size;
 }
 
+void Maxwell3D::ProcessCBData(u32 value) {
+    // Write the input value to the current const buffer at the current position.
+    GPUVAddr buffer_address = regs.const_buffer.BufferAddress();
+    ASSERT(buffer_address != 0);
+
+    // 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);
+
+    Memory::Write32(address, value);
+
+    // Increment the current buffer position.
+    regs.const_buffer.cb_pos = regs.const_buffer.cb_pos + 4;
+}
+
 } // namespace Engines
 } // namespace Tegra

src/video_core/engines/maxwell_3d.h

@@ -62,7 +62,34 @@ public:
 
         union {
             struct {
-                INSERT_PADDING_WORDS(0x582);
+                INSERT_PADDING_WORDS(0x557);
+
+                struct {
+                    u32 tsc_address_high;
+                    u32 tsc_address_low;
+                    u32 tsc_limit;
+
+                    GPUVAddr TSCAddress() const {
+                        return static_cast<GPUVAddr>(
+                            (static_cast<GPUVAddr>(tsc_address_high) << 32) | tsc_address_low);
+                    }
+                } tsc;
+
+                INSERT_PADDING_WORDS(0x3);
+
+                struct {
+                    u32 tic_address_high;
+                    u32 tic_address_low;
+                    u32 tic_limit;
+
+                    GPUVAddr TICAddress() const {
+                        return static_cast<GPUVAddr>(
+                            (static_cast<GPUVAddr>(tic_address_high) << 32) | tic_address_low);
+                    }
+                } tic;
+
+                INSERT_PADDING_WORDS(0x22);
+
                 struct {
                     u32 code_address_high;
                     u32 code_address_low;
@@ -166,7 +193,26 @@ public:
 
                 u32 tex_cb_index;
 
-                INSERT_PADDING_WORDS(0x4B3);
+                INSERT_PADDING_WORDS(0x395);
+
+                struct {
+                    /// Compressed address of a buffer that holds information about bound SSBOs.
+                    /// This address is usually bound to c0 in the shaders.
+                    u32 buffer_address;
+
+                    GPUVAddr BufferAddress() const {
+                        return static_cast<GPUVAddr>(buffer_address) << 8;
+                    }
+                } ssbo_info;
+
+                INSERT_PADDING_WORDS(0x11);
+
+                struct {
+                    u32 address[MaxShaderStage];
+                    u32 size[MaxShaderStage];
+                } tex_info_buffers;
+
+                INSERT_PADDING_WORDS(0x102);
             };
             std::array<u32, NUM_REGS> reg_array;
         };
@@ -218,6 +264,9 @@ private:
     /// Handles a write to the QUERY_GET register.
     void ProcessQueryGet();
 
+    /// Handles a write to the CB_DATA[i] register.
+    void ProcessCBData(u32 value);
+
     /// Handles a write to the CB_BIND register.
     void ProcessCBBind(Regs::ShaderStage stage);
 
@@ -225,7 +274,9 @@ private:
     void DrawArrays();
 
     /// Method call handlers
+    void BindTextureInfoBuffer(const std::vector<u32>& parameters);
     void SetShader(const std::vector<u32>& parameters);
+    void BindStorageBuffer(const std::vector<u32>& parameters);
 
     struct MethodInfo {
         const char* name;
@@ -240,6 +291,8 @@ private:
     static_assert(offsetof(Maxwell3D::Regs, field_name) == position * 4,                           \
                   "Field " #field_name " has invalid position")
 
+ASSERT_REG_POSITION(tsc, 0x557);
+ASSERT_REG_POSITION(tic, 0x55D);
 ASSERT_REG_POSITION(code_address, 0x582);
 ASSERT_REG_POSITION(draw, 0x585);
 ASSERT_REG_POSITION(query, 0x6C0);
@@ -249,6 +302,9 @@ ASSERT_REG_POSITION(shader_config[0], 0x800);
 ASSERT_REG_POSITION(const_buffer, 0x8E0);
 ASSERT_REG_POSITION(cb_bind[0], 0x904);
 ASSERT_REG_POSITION(tex_cb_index, 0x982);
+ASSERT_REG_POSITION(ssbo_info, 0xD18);
+ASSERT_REG_POSITION(tex_info_buffers.address[0], 0xD2A);
+ASSERT_REG_POSITION(tex_info_buffers.size[0], 0xD2F);
 
 #undef ASSERT_REG_POSITION
 

src/video_core/gpu.cpp

@@ -0,0 +1,21 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "video_core/engines/fermi_2d.h"
+#include "video_core/engines/maxwell_3d.h"
+#include "video_core/engines/maxwell_compute.h"
+#include "video_core/gpu.h"
+
+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>();
+    maxwell_compute = std::make_unique<Engines::MaxwellCompute>();
+}
+
+GPU::~GPU() = default;
+
+} // namespace Tegra

src/video_core/gpu.h

@@ -8,13 +8,16 @@
 #include <unordered_map>
 #include <vector>
 #include "common/common_types.h"
-#include "video_core/engines/fermi_2d.h"
-#include "video_core/engines/maxwell_3d.h"
-#include "video_core/engines/maxwell_compute.h"
 #include "video_core/memory_manager.h"
 
 namespace Tegra {
 
+namespace Engines {
+class Fermi2D;
+class Maxwell3D;
+class MaxwellCompute;
+} // namespace Engines
+
 enum class EngineID {
     FERMI_TWOD_A = 0x902D, // 2D Engine
     MAXWELL_B = 0xB197,    // 3D Engine
@@ -25,13 +28,8 @@ enum class EngineID {
 
 class GPU final {
 public:
-    GPU() {
-        memory_manager = std::make_unique<MemoryManager>();
-        maxwell_3d = std::make_unique<Engines::Maxwell3D>(*memory_manager);
-        fermi_2d = std::make_unique<Engines::Fermi2D>();
-        maxwell_compute = std::make_unique<Engines::MaxwellCompute>();
-    }
-    ~GPU() = default;
+    GPU();
+    ~GPU();
 
     /// Processes a command list stored at the specified address in GPU memory.
     void ProcessCommandList(GPUVAddr address, u32 size);

src/yuzu/debugger/wait_tree.cpp

@@ -150,8 +150,8 @@ QString WaitTreeThread::GetText() const {
     case THREADSTATUS_READY:
         status = tr("ready");
         break;
-    case THREADSTATUS_WAIT_ARB:
-        status = tr("waiting for address 0x%1").arg(thread.wait_address, 8, 16, QLatin1Char('0'));
+    case THREADSTATUS_WAIT_HLE_EVENT:
+        status = tr("waiting for HLE return");
         break;
     case THREADSTATUS_WAIT_SLEEP:
         status = tr("sleeping");
@@ -180,7 +180,7 @@ QColor WaitTreeThread::GetColor() const {
         return QColor(Qt::GlobalColor::darkGreen);
     case THREADSTATUS_READY:
         return QColor(Qt::GlobalColor::darkBlue);
-    case THREADSTATUS_WAIT_ARB:
+    case THREADSTATUS_WAIT_HLE_EVENT:
         return QColor(Qt::GlobalColor::darkRed);
     case THREADSTATUS_WAIT_SLEEP:
         return QColor(Qt::GlobalColor::darkYellow);