core: Fix scheduler-shutdown related crash

Service/AOC: stub ListAddOnContent function

.travis.yml

@@ -3,16 +3,11 @@ matrix:
   include:
     - os: linux
       env: NAME="clang-format"
+      sudo: required
       dist: trusty
-      addons:
-        apt:
-          sources:
-            - sourceline: 'deb http://apt.llvm.org/trusty/ llvm-toolchain-trusty-6.0 main'
-              key_url: 'https://apt.llvm.org/llvm-snapshot.gpg.key'
-            - sourceline: 'deb http://ppa.launchpad.net/ubuntu-toolchain-r/test/ubuntu trusty main'
-          packages:
-            - clang-format-6.0
-      script: "./.travis/clang-format/script.sh"
+      services: docker
+      install: "./.travis/clang-format/deps.sh"
+      script: "./.travis/clang-format/build.sh"
     - os: linux
       env: NAME="linux build"
       sudo: required

.travis/clang-format/build.sh

@@ -0,0 +1,3 @@
+#!/bin/bash -ex
+
+docker run -v $(pwd):/yuzu ubuntu:18.04 /bin/bash -ex /yuzu/.travis/clang-format/docker.sh

.travis/clang-format/deps.sh

@@ -0,0 +1,3 @@
+#!/bin/sh -ex
+
+docker pull ubuntu:18.04

.travis/clang-format/docker.sh

@@ -0,0 +1,8 @@
+#!/bin/bash -ex
+
+apt-get update
+apt-get install -y clang-format-6.0
+
+# Run clang-format
+cd /yuzu
+./.travis/clang-format/script.sh

.travis/macos/build.sh

@@ -2,7 +2,7 @@
 
 set -o pipefail
 
-export MACOSX_DEPLOYMENT_TARGET=10.9
+export MACOSX_DEPLOYMENT_TARGET=10.12
 export Qt5_DIR=$(brew --prefix)/opt/qt5
 export UNICORNDIR=$(pwd)/externals/unicorn
 

src/common/logging/backend.cpp

@@ -35,11 +35,14 @@ namespace Log {
     SUB(Service, ACC)                                                                              \
     SUB(Service, Audio)                                                                            \
     SUB(Service, AM)                                                                               \
+    SUB(Service, AOC)                                                                              \
     SUB(Service, APM)                                                                              \
+    SUB(Service, Friend)                                                                           \
     SUB(Service, FS)                                                                               \
     SUB(Service, HID)                                                                              \
     SUB(Service, LM)                                                                               \
     SUB(Service, NIFM)                                                                             \
+    SUB(Service, NS)                                                                               \
     SUB(Service, NVDRV)                                                                            \
     SUB(Service, PCTL)                                                                             \
     SUB(Service, SET)                                                                              \

src/common/logging/log.h

@@ -51,12 +51,15 @@ enum class Class : ClassType {
                        ///  should have its own subclass.
     Service_ACC,       ///< The ACC (Accounts) service
     Service_AM,        ///< The AM (Applet manager) service
+    Service_AOC,       ///< The AOC (AddOn Content) service
     Service_APM,       ///< The APM (Performance) service
     Service_Audio,     ///< The Audio (Audio control) service
+    Service_Friend,    ///< The friend service
     Service_FS,        ///< The FS (Filesystem) service
     Service_HID,       ///< The HID (Human interface device) service
     Service_LM,        ///< The LM (Logger) service
     Service_NIFM,      ///< The NIFM (Network interface) service
+    Service_NS,        ///< The NS services
     Service_NVDRV,     ///< The NVDRV (Nvidia driver) service
     Service_PCTL,      ///< The PCTL (Parental control) service
     Service_SET,       ///< The SET (Settings) service

src/core/CMakeLists.txt

@@ -28,8 +28,6 @@ add_library(core STATIC
     hle/config_mem.h
     hle/ipc.h
     hle/ipc_helpers.h
-    hle/kernel/address_arbiter.cpp
-    hle/kernel/address_arbiter.h
     hle/kernel/client_port.cpp
     hle/kernel/client_port.h
     hle/kernel/client_session.cpp
@@ -55,6 +53,8 @@ add_library(core STATIC
     hle/kernel/process.h
     hle/kernel/resource_limit.cpp
     hle/kernel/resource_limit.h
+    hle/kernel/scheduler.cpp
+    hle/kernel/scheduler.h
     hle/kernel/server_port.cpp
     hle/kernel/server_port.h
     hle/kernel/server_session.cpp
@@ -92,6 +92,8 @@ add_library(core STATIC
     hle/service/aoc/aoc_u.h
     hle/service/apm/apm.cpp
     hle/service/apm/apm.h
+    hle/service/apm/interface.cpp
+    hle/service/apm/interface.h
     hle/service/audio/audio.cpp
     hle/service/audio/audio.h
     hle/service/audio/audin_u.cpp
@@ -110,6 +112,10 @@ add_library(core STATIC
     hle/service/filesystem/filesystem.h
     hle/service/filesystem/fsp_srv.cpp
     hle/service/filesystem/fsp_srv.h
+    hle/service/friend/friend.cpp
+    hle/service/friend/friend.h
+    hle/service/friend/friend_a.cpp
+    hle/service/friend/friend_a.h
     hle/service/hid/hid.cpp
     hle/service/hid/hid.h
     hle/service/lm/lm.cpp
@@ -122,6 +128,10 @@ add_library(core STATIC
     hle/service/nifm/nifm_s.h
     hle/service/nifm/nifm_u.cpp
     hle/service/nifm/nifm_u.h
+    hle/service/ns/ns.cpp
+    hle/service/ns/ns.h
+    hle/service/ns/pl_u.cpp
+    hle/service/ns/pl_u.h
     hle/service/nvdrv/devices/nvdevice.h
     hle/service/nvdrv/devices/nvdisp_disp0.cpp
     hle/service/nvdrv/devices/nvdisp_disp0.h

src/core/arm/dynarmic/arm_dynarmic.cpp

@@ -8,9 +8,12 @@
 #include <dynarmic/A64/config.h>
 #include "core/arm/dynarmic/arm_dynarmic.h"
 #include "core/core_timing.h"
+#include "core/hle/kernel/memory.h"
 #include "core/hle/kernel/svc.h"
 #include "core/memory.h"
 
+using Vector = Dynarmic::A64::Vector;
+
 class ARM_Dynarmic_Callbacks : public Dynarmic::A64::UserCallbacks {
 public:
     explicit ARM_Dynarmic_Callbacks(ARM_Dynarmic& parent) : parent(parent) {}
@@ -28,6 +31,9 @@ public:
     u64 MemoryRead64(u64 vaddr) override {
         return Memory::Read64(vaddr);
     }
+    Vector MemoryRead128(u64 vaddr) override {
+        return {Memory::Read64(vaddr), Memory::Read64(vaddr + 8)};
+    }
 
     void MemoryWrite8(u64 vaddr, u8 value) override {
         Memory::Write8(vaddr, value);
@@ -41,6 +47,10 @@ public:
     void MemoryWrite64(u64 vaddr, u64 value) override {
         Memory::Write64(vaddr, value);
     }
+    void MemoryWrite128(u64 vaddr, Vector value) override {
+        Memory::Write64(vaddr, value[0]);
+        Memory::Write64(vaddr + 8, value[1]);
+    }
 
     void InterpreterFallback(u64 pc, size_t num_instructions) override {
         ARM_Interface::ThreadContext ctx;
@@ -52,12 +62,12 @@ public:
         num_interpreted_instructions += num_instructions;
     }
 
-    void ExceptionRaised(u64 pc, Dynarmic::A64::Exception /*exception*/) override {
-        ASSERT_MSG(false, "ExceptionRaised(%" PRIx64 ")", pc);
+    void ExceptionRaised(u64 pc, Dynarmic::A64::Exception exception) override {
+        ASSERT_MSG(false, "ExceptionRaised(exception = %zu, pc = %" PRIx64 ")",
+                   static_cast<size_t>(exception), pc);
     }
 
     void CallSVC(u32 swi) override {
-        printf("svc %x\n", swi);
         Kernel::CallSVC(swi);
     }
 
@@ -75,12 +85,24 @@ public:
     ARM_Dynarmic& parent;
     size_t ticks_remaining = 0;
     size_t num_interpreted_instructions = 0;
-    u64 tpidrr0_el0 = 0;
+    u64 tpidrro_el0 = 0;
 };
 
+std::unique_ptr<Dynarmic::A64::Jit> MakeJit(const std::unique_ptr<ARM_Dynarmic_Callbacks>& cb) {
+    const auto page_table = Kernel::g_current_process->vm_manager.page_table.pointers.data();
+
+    Dynarmic::A64::UserConfig config;
+    config.callbacks = cb.get();
+    config.tpidrro_el0 = &cb->tpidrro_el0;
+    config.dczid_el0 = 4;
+    config.page_table = reinterpret_cast<void**>(page_table);
+    config.page_table_address_space_bits = Memory::ADDRESS_SPACE_BITS;
+    config.silently_mirror_page_table = false;
+    return std::make_unique<Dynarmic::A64::Jit>(config);
+}
+
 ARM_Dynarmic::ARM_Dynarmic()
-    : cb(std::make_unique<ARM_Dynarmic_Callbacks>(*this)),
-      jit(Dynarmic::A64::UserConfig{cb.get()}) {
+    : cb(std::make_unique<ARM_Dynarmic_Callbacks>(*this)), jit(MakeJit(cb)) {
     ARM_Interface::ThreadContext ctx;
     inner_unicorn.SaveContext(ctx);
     LoadContext(ctx);
@@ -94,27 +116,27 @@ void ARM_Dynarmic::MapBackingMemory(u64 address, size_t size, u8* memory,
 }
 
 void ARM_Dynarmic::SetPC(u64 pc) {
-    jit.SetPC(pc);
+    jit->SetPC(pc);
 }
 
 u64 ARM_Dynarmic::GetPC() const {
-    return jit.GetPC();
+    return jit->GetPC();
 }
 
 u64 ARM_Dynarmic::GetReg(int index) const {
-    return jit.GetRegister(index);
+    return jit->GetRegister(index);
 }
 
 void ARM_Dynarmic::SetReg(int index, u64 value) {
-    jit.SetRegister(index, value);
+    jit->SetRegister(index, value);
 }
 
 u128 ARM_Dynarmic::GetExtReg(int index) const {
-    return jit.GetVector(index);
+    return jit->GetVector(index);
 }
 
 void ARM_Dynarmic::SetExtReg(int index, u128 value) {
-    jit.SetVector(index, value);
+    jit->SetVector(index, value);
 }
 
 u32 ARM_Dynarmic::GetVFPReg(int /*index*/) const {
@@ -127,58 +149,58 @@ void ARM_Dynarmic::SetVFPReg(int /*index*/, u32 /*value*/) {
 }
 
 u32 ARM_Dynarmic::GetCPSR() const {
-    return jit.GetPstate();
+    return jit->GetPstate();
 }
 
 void ARM_Dynarmic::SetCPSR(u32 cpsr) {
-    jit.SetPstate(cpsr);
+    jit->SetPstate(cpsr);
 }
 
 u64 ARM_Dynarmic::GetTlsAddress() const {
-    return cb->tpidrr0_el0;
+    return cb->tpidrro_el0;
 }
 
 void ARM_Dynarmic::SetTlsAddress(u64 address) {
-    cb->tpidrr0_el0 = address;
+    cb->tpidrro_el0 = address;
 }
 
 void ARM_Dynarmic::ExecuteInstructions(int num_instructions) {
     cb->ticks_remaining = num_instructions;
-    jit.Run();
+    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();
-    ctx.pc = jit.GetPC();
-    ctx.cpsr = jit.GetPstate();
-    ctx.fpu_registers = jit.GetVectors();
-    ctx.fpscr = jit.GetFpcr();
-    ctx.tls_address = cb->tpidrr0_el0;
+    ctx.cpu_registers = jit->GetRegisters();
+    ctx.sp = jit->GetSP();
+    ctx.pc = jit->GetPC();
+    ctx.cpsr = jit->GetPstate();
+    ctx.fpu_registers = jit->GetVectors();
+    ctx.fpscr = jit->GetFpcr();
+    ctx.tls_address = cb->tpidrro_el0;
 }
 
 void ARM_Dynarmic::LoadContext(const ARM_Interface::ThreadContext& ctx) {
-    jit.SetRegisters(ctx.cpu_registers);
-    jit.SetSP(ctx.sp);
-    jit.SetPC(ctx.pc);
-    jit.SetPstate(static_cast<u32>(ctx.cpsr));
-    jit.SetVectors(ctx.fpu_registers);
-    jit.SetFpcr(static_cast<u32>(ctx.fpscr));
-    cb->tpidrr0_el0 = ctx.tls_address;
+    jit->SetRegisters(ctx.cpu_registers);
+    jit->SetSP(ctx.sp);
+    jit->SetPC(ctx.pc);
+    jit->SetPstate(static_cast<u32>(ctx.cpsr));
+    jit->SetVectors(ctx.fpu_registers);
+    jit->SetFpcr(static_cast<u32>(ctx.fpscr));
+    cb->tpidrro_el0 = ctx.tls_address;
 }
 
 void ARM_Dynarmic::PrepareReschedule() {
-    if (jit.IsExecuting()) {
-        jit.HaltExecution();
+    if (jit->IsExecuting()) {
+        jit->HaltExecution();
     }
 }
 
 void ARM_Dynarmic::ClearInstructionCache() {
-    jit.ClearCache();
+    jit->ClearCache();
 }
 
 void ARM_Dynarmic::PageTableChanged() {
-    UNIMPLEMENTED();
+    jit = MakeJit(cb);
 }

src/core/arm/dynarmic/arm_dynarmic.h

@@ -45,6 +45,6 @@ public:
 private:
     friend class ARM_Dynarmic_Callbacks;
     std::unique_ptr<ARM_Dynarmic_Callbacks> cb;
-    Dynarmic::A64::Jit jit;
+    std::unique_ptr<Dynarmic::A64::Jit> jit;
     ARM_Unicorn inner_unicorn;
 };

src/core/core.cpp

@@ -133,32 +133,36 @@ void System::Reschedule() {
     }
 
     reschedule_pending = false;
-    Kernel::Reschedule();
+    Core::System::GetInstance().Scheduler().Reschedule();
 }
 
 System::ResultStatus System::Init(EmuWindow* emu_window, u32 system_mode) {
     LOG_DEBUG(HW_Memory, "initialized OK");
 
+    CoreTiming::Init();
+
     switch (Settings::values.cpu_core) {
     case Settings::CpuCore::Unicorn:
-        cpu_core = std::make_unique<ARM_Unicorn>();
+        cpu_core = std::make_shared<ARM_Unicorn>();
         break;
     case Settings::CpuCore::Dynarmic:
     default:
 #ifdef ARCHITECTURE_x86_64
-        cpu_core = std::make_unique<ARM_Dynarmic>();
+        cpu_core = std::make_shared<ARM_Dynarmic>();
 #else
-        cpu_core = std::make_unique<ARM_Unicorn>();
+        cpu_core = std::make_shared<ARM_Unicorn>();
         LOG_WARNING(Core, "CPU JIT requested, but Dynarmic not available");
 #endif
         break;
     }
 
+    gpu_core = std::make_unique<Tegra::GPU>();
+
     telemetry_session = std::make_unique<Core::TelemetrySession>();
 
-    CoreTiming::Init();
     HW::Init();
     Kernel::Init(system_mode);
+    scheduler = std::make_unique<Kernel::Scheduler>(cpu_core.get());
     Service::Init();
     GDBStub::Init();
 
@@ -186,15 +190,18 @@ void System::Shutdown() {
                          perf_results.frametime * 1000.0);
 
     // Shutdown emulation session
-    GDBStub::Shutdown();
     VideoCore::Shutdown();
+    GDBStub::Shutdown();
     Service::Shutdown();
+    scheduler = nullptr;
     Kernel::Shutdown();
     HW::Shutdown();
-    CoreTiming::Shutdown();
-    cpu_core = nullptr;
-    app_loader = nullptr;
     telemetry_session = nullptr;
+    gpu_core = nullptr;
+    cpu_core = nullptr;
+    CoreTiming::Shutdown();
+
+    app_loader = nullptr;
 
     LOG_DEBUG(Core, "Shutdown OK");
 }

src/core/core.h

@@ -7,10 +7,12 @@
 #include <memory>
 #include <string>
 #include "common/common_types.h"
+#include "core/hle/kernel/scheduler.h"
 #include "core/loader/loader.h"
 #include "core/memory.h"
 #include "core/perf_stats.h"
 #include "core/telemetry_session.h"
+#include "video_core/gpu.h"
 
 class EmuWindow;
 class ARM_Interface;
@@ -102,6 +104,14 @@ public:
         return *cpu_core;
     }
 
+    Tegra::GPU& GPU() {
+        return *gpu_core;
+    }
+
+    Kernel::Scheduler& Scheduler() {
+        return *scheduler;
+    }
+
     PerfStats perf_stats;
     FrameLimiter frame_limiter;
 
@@ -135,8 +145,9 @@ private:
     /// AppLoader used to load the current executing application
     std::unique_ptr<Loader::AppLoader> app_loader;
 
-    ///< ARM11 CPU core
-    std::unique_ptr<ARM_Interface> cpu_core;
+    std::shared_ptr<ARM_Interface> cpu_core;
+    std::unique_ptr<Kernel::Scheduler> scheduler;
+    std::unique_ptr<Tegra::GPU> gpu_core;
 
     /// When true, signals that a reschedule should happen
     bool reschedule_pending{};

src/core/gdbstub/gdbstub.cpp

@@ -6,6 +6,7 @@
 
 #include <algorithm>
 #include <atomic>
+#include <cinttypes>
 #include <climits>
 #include <csignal>
 #include <cstdarg>
@@ -360,8 +361,9 @@ static void RemoveBreakpoint(BreakpointType type, PAddr addr) {
 
     auto bp = p.find(static_cast<u64>(addr));
     if (bp != p.end()) {
-        LOG_DEBUG(Debug_GDBStub, "gdb: removed a breakpoint: %08x bytes at %08x of type %d\n",
-                  bp->second.len, bp->second.addr, type);
+        LOG_DEBUG(Debug_GDBStub,
+                  "gdb: removed a breakpoint: %016" PRIx64 " bytes at %016" PRIx64 " of type %d\n",
+                  bp->second.len, bp->second.addr, static_cast<int>(type));
         p.erase(static_cast<u64>(addr));
     }
 }
@@ -407,8 +409,9 @@ bool CheckBreakpoint(PAddr addr, BreakpointType type) {
 
         if (bp->second.active && (addr >= bp->second.addr && addr < bp->second.addr + len)) {
             LOG_DEBUG(Debug_GDBStub,
-                      "Found breakpoint type %d @ %08x, range: %08x - %08x (%d bytes)\n", type,
-                      addr, bp->second.addr, bp->second.addr + len, len);
+                      "Found breakpoint type %d @ %016" PRIx64 ", range: %016" PRIx64
+                      " - %016" PRIx64 " (%" PRIx64 " bytes)\n",
+                      static_cast<int>(type), addr, bp->second.addr, bp->second.addr + len, len);
             return true;
         }
     }
@@ -778,8 +781,8 @@ static bool CommitBreakpoint(BreakpointType type, PAddr addr, u64 len) {
     breakpoint.len = len;
     p.insert({addr, breakpoint});
 
-    LOG_DEBUG(Debug_GDBStub, "gdb: added %d breakpoint: %08x bytes at %08x\n", type, breakpoint.len,
-              breakpoint.addr);
+    LOG_DEBUG(Debug_GDBStub, "gdb: added %d breakpoint: %016" PRIx64 " bytes at %016" PRIx64 "\n",
+              static_cast<int>(type), breakpoint.len, breakpoint.addr);
 
     return true;
 }

src/core/hle/ipc.h

@@ -91,6 +91,10 @@ struct BufferDescriptorX {
         address |= static_cast<VAddr>(address_bits_36_38) << 36;
         return address;
     }
+
+    u64 Size() const {
+        return static_cast<u64>(size);
+    }
 };
 static_assert(sizeof(BufferDescriptorX) == 8, "BufferDescriptorX size is incorrect");
 

src/core/hle/kernel/address_arbiter.cpp

@@ -1,91 +0,0 @@
-// Copyright 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include "common/common_types.h"
-#include "common/logging/log.h"
-#include "core/hle/kernel/address_arbiter.h"
-#include "core/hle/kernel/errors.h"
-#include "core/hle/kernel/thread.h"
-#include "core/memory.h"
-
-////////////////////////////////////////////////////////////////////////////////////////////////////
-// Kernel namespace
-
-namespace Kernel {
-
-AddressArbiter::AddressArbiter() {}
-AddressArbiter::~AddressArbiter() {}
-
-SharedPtr<AddressArbiter> AddressArbiter::Create(std::string name) {
-    SharedPtr<AddressArbiter> address_arbiter(new AddressArbiter);
-
-    address_arbiter->name = std::move(name);
-
-    return address_arbiter;
-}
-
-ResultCode AddressArbiter::ArbitrateAddress(ArbitrationType type, VAddr address, s32 value,
-                                            u64 nanoseconds) {
-    switch (type) {
-
-    // Signal thread(s) waiting for arbitrate address...
-    case ArbitrationType::Signal:
-        // Negative value means resume all threads
-        if (value < 0) {
-            ArbitrateAllThreads(address);
-        } else {
-            // Resume first N threads
-            for (int i = 0; i < value; i++)
-                ArbitrateHighestPriorityThread(address);
-        }
-        break;
-
-    // Wait current thread (acquire the arbiter)...
-    case ArbitrationType::WaitIfLessThan:
-        if ((s32)Memory::Read32(address) < value) {
-            Kernel::WaitCurrentThread_ArbitrateAddress(address);
-        }
-        break;
-    case ArbitrationType::WaitIfLessThanWithTimeout:
-        if ((s32)Memory::Read32(address) < value) {
-            Kernel::WaitCurrentThread_ArbitrateAddress(address);
-            GetCurrentThread()->WakeAfterDelay(nanoseconds);
-        }
-        break;
-    case ArbitrationType::DecrementAndWaitIfLessThan: {
-        s32 memory_value = Memory::Read32(address);
-        if (memory_value < value) {
-            // Only change the memory value if the thread should wait
-            Memory::Write32(address, (s32)memory_value - 1);
-            Kernel::WaitCurrentThread_ArbitrateAddress(address);
-        }
-        break;
-    }
-    case ArbitrationType::DecrementAndWaitIfLessThanWithTimeout: {
-        s32 memory_value = Memory::Read32(address);
-        if (memory_value < value) {
-            // Only change the memory value if the thread should wait
-            Memory::Write32(address, (s32)memory_value - 1);
-            Kernel::WaitCurrentThread_ArbitrateAddress(address);
-            GetCurrentThread()->WakeAfterDelay(nanoseconds);
-        }
-        break;
-    }
-
-    default:
-        LOG_ERROR(Kernel, "unknown type=%d", type);
-        return ERR_INVALID_ENUM_VALUE_FND;
-    }
-
-    // The calls that use a timeout seem to always return a Timeout error even if they did not put
-    // the thread to sleep
-    if (type == ArbitrationType::WaitIfLessThanWithTimeout ||
-        type == ArbitrationType::DecrementAndWaitIfLessThanWithTimeout) {
-
-        return RESULT_TIMEOUT;
-    }
-    return RESULT_SUCCESS;
-}
-
-} // namespace Kernel

src/core/hle/kernel/address_arbiter.h

@@ -1,60 +0,0 @@
-// Copyright 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include "common/common_types.h"
-#include "core/hle/kernel/kernel.h"
-#include "core/hle/result.h"
-
-// Address arbiters are an underlying kernel synchronization object that can be created/used via
-// supervisor calls (SVCs). They function as sort of a global lock. Typically, games/other CTR
-// applications use them as an underlying mechanism to implement thread-safe barriers, events, and
-// semphores.
-
-////////////////////////////////////////////////////////////////////////////////////////////////////
-// Kernel namespace
-
-namespace Kernel {
-
-enum class ArbitrationType : u32 {
-    Signal,
-    WaitIfLessThan,
-    DecrementAndWaitIfLessThan,
-    WaitIfLessThanWithTimeout,
-    DecrementAndWaitIfLessThanWithTimeout,
-};
-
-class AddressArbiter final : public Object {
-public:
-    /**
-     * Creates an address arbiter.
-     *
-     * @param name Optional name used for debugging.
-     * @returns The created AddressArbiter.
-     */
-    static SharedPtr<AddressArbiter> Create(std::string name = "Unknown");
-
-    std::string GetTypeName() const override {
-        return "Arbiter";
-    }
-    std::string GetName() const override {
-        return name;
-    }
-
-    static const HandleType HANDLE_TYPE = HandleType::AddressArbiter;
-    HandleType GetHandleType() const override {
-        return HANDLE_TYPE;
-    }
-
-    std::string name; ///< Name of address arbiter object (optional)
-
-    ResultCode ArbitrateAddress(ArbitrationType type, VAddr address, s32 value, u64 nanoseconds);
-
-private:
-    AddressArbiter();
-    ~AddressArbiter() override;
-};
-
-} // namespace Kernel

src/core/hle/kernel/hle_ipc.cpp

@@ -12,6 +12,7 @@
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/server_session.h"
+#include "core/memory.h"
 
 namespace Kernel {
 
@@ -210,4 +211,98 @@ ResultCode HLERequestContext::WriteToOutgoingCommandBuffer(u32_le* dst_cmdbuf, P
     return RESULT_SUCCESS;
 }
 
+std::vector<u8> HLERequestContext::ReadBuffer() const {
+    std::vector<u8> buffer;
+    const bool is_buffer_a{BufferDescriptorA().size() && BufferDescriptorA()[0].Size()};
+
+    if (is_buffer_a) {
+        buffer.resize(BufferDescriptorA()[0].Size());
+        Memory::ReadBlock(BufferDescriptorA()[0].Address(), buffer.data(), buffer.size());
+    } else {
+        buffer.resize(BufferDescriptorX()[0].Size());
+        Memory::ReadBlock(BufferDescriptorX()[0].Address(), buffer.data(), buffer.size());
+    }
+
+    return buffer;
+}
+
+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);
+
+    if (is_buffer_b) {
+        Memory::WriteBlock(BufferDescriptorB()[0].Address(), buffer, size);
+    } else {
+        Memory::WriteBlock(BufferDescriptorC()[0].Address(), buffer, size);
+    }
+
+    return size;
+}
+
+size_t HLERequestContext::WriteBuffer(const std::vector<u8>& buffer) const {
+    return WriteBuffer(buffer.data(), buffer.size());
+}
+
+size_t HLERequestContext::GetReadBufferSize() const {
+    const bool is_buffer_a{BufferDescriptorA().size() && BufferDescriptorA()[0].Size()};
+    return is_buffer_a ? BufferDescriptorA()[0].Size() : BufferDescriptorX()[0].Size();
+}
+
+size_t HLERequestContext::GetWriteBufferSize() const {
+    const bool is_buffer_b{BufferDescriptorB().size() && BufferDescriptorB()[0].Size()};
+    return is_buffer_b ? BufferDescriptorB()[0].Size() : BufferDescriptorC()[0].Size();
+}
+
+std::string HLERequestContext::Description() const {
+    if (!command_header) {
+        return "No command header available";
+    }
+    std::ostringstream s;
+    s << "IPC::CommandHeader: Type:" << static_cast<u32>(command_header->type.Value());
+    s << ", X(Pointer):" << command_header->num_buf_x_descriptors;
+    if (command_header->num_buf_x_descriptors) {
+        s << '[';
+        for (u64 i = 0; i < command_header->num_buf_x_descriptors; ++i) {
+            s << "0x" << std::hex << BufferDescriptorX()[i].Size();
+            if (i < command_header->num_buf_x_descriptors - 1)
+                s << ", ";
+        }
+        s << ']';
+    }
+    s << ", A(Send):" << command_header->num_buf_a_descriptors;
+    if (command_header->num_buf_a_descriptors) {
+        s << '[';
+        for (u64 i = 0; i < command_header->num_buf_a_descriptors; ++i) {
+            s << "0x" << std::hex << BufferDescriptorA()[i].Size();
+            if (i < command_header->num_buf_a_descriptors - 1)
+                s << ", ";
+        }
+        s << ']';
+    }
+    s << ", B(Receive):" << command_header->num_buf_b_descriptors;
+    if (command_header->num_buf_b_descriptors) {
+        s << '[';
+        for (u64 i = 0; i < command_header->num_buf_b_descriptors; ++i) {
+            s << "0x" << std::hex << BufferDescriptorB()[i].Size();
+            if (i < command_header->num_buf_b_descriptors - 1)
+                s << ", ";
+        }
+        s << ']';
+    }
+    s << ", C(ReceiveList):" << BufferDescriptorC().size();
+    if (!BufferDescriptorC().empty()) {
+        s << '[';
+        for (u64 i = 0; i < BufferDescriptorC().size(); ++i) {
+            s << "0x" << std::hex << BufferDescriptorC()[i].Size();
+            if (i < BufferDescriptorC().size() - 1)
+                s << ", ";
+        }
+        s << ']';
+    }
+    s << ", data_size:" << command_header->data_size.Value();
+
+    return s.str();
+}
+
 } // namespace Kernel

src/core/hle/kernel/hle_ipc.h

@@ -143,6 +143,21 @@ public:
         return domain_message_header;
     }
 
+    /// Helper function to read a buffer using the appropriate buffer descriptor
+    std::vector<u8> ReadBuffer() const;
+
+    /// Helper function to write a buffer using the appropriate buffer descriptor
+    size_t WriteBuffer(const void* buffer, size_t size) const;
+
+    /// Helper function to write a buffer using the appropriate buffer descriptor
+    size_t WriteBuffer(const std::vector<u8>& buffer) const;
+
+    /// Helper function to get the size of the input buffer
+    size_t GetReadBufferSize() const;
+
+    /// Helper function to get the size of the output buffer
+    size_t GetWriteBufferSize() const;
+
     template <typename T>
     SharedPtr<T> GetCopyObject(size_t index) {
         ASSERT(index < copy_objects.size());
@@ -187,6 +202,8 @@ public:
         return domain_objects.size();
     }
 
+    std::string Description() const;
+
 private:
     std::array<u32, IPC::COMMAND_BUFFER_LENGTH> cmd_buf;
     SharedPtr<Kernel::ServerSession> server_session;

src/core/hle/kernel/scheduler.cpp

@@ -0,0 +1,134 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/core_timing.h"
+#include "core/hle/kernel/process.h"
+#include "core/hle/kernel/scheduler.h"
+
+namespace Kernel {
+
+Scheduler::Scheduler(ARM_Interface* cpu_core) : cpu_core(cpu_core) {}
+
+Scheduler::~Scheduler() {
+    for (auto& thread : thread_list) {
+        thread->Stop();
+    }
+}
+
+bool Scheduler::HaveReadyThreads() {
+    return ready_queue.get_first() != nullptr;
+}
+
+Thread* Scheduler::GetCurrentThread() const {
+    return current_thread.get();
+}
+
+Thread* Scheduler::PopNextReadyThread() {
+    Thread* next = nullptr;
+    Thread* thread = GetCurrentThread();
+
+    if (thread && thread->status == THREADSTATUS_RUNNING) {
+        // We have to do better than the current thread.
+        // This call returns null when that's not possible.
+        next = ready_queue.pop_first_better(thread->current_priority);
+        if (!next) {
+            // Otherwise just keep going with the current thread
+            next = thread;
+        }
+    } else {
+        next = ready_queue.pop_first();
+    }
+
+    return next;
+}
+
+void Scheduler::SwitchContext(Thread* new_thread) {
+    Thread* previous_thread = GetCurrentThread();
+
+    // Save context for previous thread
+    if (previous_thread) {
+        previous_thread->last_running_ticks = CoreTiming::GetTicks();
+        cpu_core->SaveContext(previous_thread->context);
+
+        if (previous_thread->status == THREADSTATUS_RUNNING) {
+            // This is only the case when a reschedule is triggered without the current thread
+            // yielding execution (i.e. an event triggered, system core time-sliced, etc)
+            ready_queue.push_front(previous_thread->current_priority, previous_thread);
+            previous_thread->status = THREADSTATUS_READY;
+        }
+    }
+
+    // Load context of new thread
+    if (new_thread) {
+        ASSERT_MSG(new_thread->status == THREADSTATUS_READY,
+                   "Thread must be ready to become running.");
+
+        // Cancel any outstanding wakeup events for this thread
+        new_thread->CancelWakeupTimer();
+
+        auto previous_process = Kernel::g_current_process;
+
+        current_thread = new_thread;
+
+        ready_queue.remove(new_thread->current_priority, new_thread);
+        new_thread->status = THREADSTATUS_RUNNING;
+
+        if (previous_process != current_thread->owner_process) {
+            Kernel::g_current_process = current_thread->owner_process;
+            SetCurrentPageTable(&Kernel::g_current_process->vm_manager.page_table);
+        }
+
+        cpu_core->LoadContext(new_thread->context);
+        cpu_core->SetTlsAddress(new_thread->GetTLSAddress());
+    } else {
+        current_thread = nullptr;
+        // Note: We do not reset the current process and current page table when idling because
+        // technically we haven't changed processes, our threads are just paused.
+    }
+}
+
+void Scheduler::Reschedule() {
+    Thread* cur = GetCurrentThread();
+    Thread* next = PopNextReadyThread();
+
+    if (cur && next) {
+        LOG_TRACE(Kernel, "context switch %u -> %u", cur->GetObjectId(), next->GetObjectId());
+    } else if (cur) {
+        LOG_TRACE(Kernel, "context switch %u -> idle", cur->GetObjectId());
+    } else if (next) {
+        LOG_TRACE(Kernel, "context switch idle -> %u", next->GetObjectId());
+    }
+
+    SwitchContext(next);
+}
+
+void Scheduler::AddThread(SharedPtr<Thread> thread, u32 priority) {
+    thread_list.push_back(thread);
+    ready_queue.prepare(priority);
+}
+
+void Scheduler::RemoveThread(Thread* thread) {
+    thread_list.erase(std::remove(thread_list.begin(), thread_list.end(), thread),
+                      thread_list.end());
+}
+
+void Scheduler::ScheduleThread(Thread* thread, u32 priority) {
+    ASSERT(thread->status == THREADSTATUS_READY);
+    ready_queue.push_back(priority, thread);
+}
+
+void Scheduler::UnscheduleThread(Thread* thread, u32 priority) {
+    ASSERT(thread->status == THREADSTATUS_READY);
+    ready_queue.remove(priority, thread);
+}
+
+void Scheduler::SetThreadPriority(Thread* thread, u32 priority) {
+    // If thread was ready, adjust queues
+    if (thread->status == THREADSTATUS_READY)
+        ready_queue.move(thread, thread->current_priority, priority);
+    else
+        ready_queue.prepare(priority);
+}
+
+} // namespace Kernel

src/core/hle/kernel/scheduler.h

@@ -0,0 +1,73 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <vector>
+#include "common/common_types.h"
+#include "common/thread_queue_list.h"
+#include "core/arm/arm_interface.h"
+#include "core/hle/kernel/thread.h"
+
+namespace Kernel {
+
+class Scheduler final {
+public:
+    explicit Scheduler(ARM_Interface* cpu_core);
+    ~Scheduler();
+
+    /// Returns whether there are any threads that are ready to run.
+    bool HaveReadyThreads();
+
+    /// Reschedules to the next available thread (call after current thread is suspended)
+    void Reschedule();
+
+    /// Gets the current running thread
+    Thread* GetCurrentThread() const;
+
+    /// Adds a new thread to the scheduler
+    void AddThread(SharedPtr<Thread> thread, u32 priority);
+
+    /// Removes a thread from the scheduler
+    void RemoveThread(Thread* thread);
+
+    /// Schedules a thread that has become "ready"
+    void ScheduleThread(Thread* thread, u32 priority);
+
+    /// Unschedules a thread that was already scheduled
+    void UnscheduleThread(Thread* thread, u32 priority);
+
+    /// Sets the priority of a thread in the scheduler
+    void SetThreadPriority(Thread* thread, u32 priority);
+
+    /// Returns a list of all threads managed by the scheduler
+    const std::vector<SharedPtr<Thread>>& GetThreadList() const {
+        return thread_list;
+    }
+
+private:
+    /**
+     * Pops and returns the next thread from the thread queue
+     * @return A pointer to the next ready thread
+     */
+    Thread* PopNextReadyThread();
+
+    /**
+     * Switches the CPU's active thread context to that of the specified thread
+     * @param new_thread The thread to switch to
+     */
+    void SwitchContext(Thread* new_thread);
+
+    /// Lists all thread ids that aren't deleted/etc.
+    std::vector<SharedPtr<Thread>> thread_list;
+
+    /// Lists only ready thread ids.
+    Common::ThreadQueueList<Thread*, THREADPRIO_LOWEST + 1> ready_queue;
+
+    SharedPtr<Thread> current_thread = nullptr;
+
+    ARM_Interface* cpu_core;
+};
+
+} // namespace Kernel

src/core/hle/kernel/server_session.cpp

@@ -57,6 +57,33 @@ void ServerSession::Acquire(Thread* thread) {
     pending_requesting_threads.pop_back();
 }
 
+ResultCode ServerSession::HandleDomainSyncRequest(Kernel::HLERequestContext& context) {
+    auto& domain_message_header = context.GetDomainMessageHeader();
+    if (domain_message_header) {
+        // If there is a DomainMessageHeader, then this is CommandType "Request"
+        const u32 object_id{context.GetDomainMessageHeader()->object_id};
+        switch (domain_message_header->command) {
+        case IPC::DomainMessageHeader::CommandType::SendMessage:
+            return domain_request_handlers[object_id - 1]->HandleSyncRequest(context);
+
+        case IPC::DomainMessageHeader::CommandType::CloseVirtualHandle: {
+            LOG_DEBUG(IPC, "CloseVirtualHandle, object_id=0x%08X", object_id);
+
+            domain_request_handlers[object_id - 1] = nullptr;
+
+            IPC::ResponseBuilder rb{context, 2};
+            rb.Push(RESULT_SUCCESS);
+            return RESULT_SUCCESS;
+        }
+        }
+
+        LOG_CRITICAL(IPC, "Unknown domain command=%d", domain_message_header->command.Value());
+        ASSERT(false);
+    }
+
+    return RESULT_SUCCESS;
+}
+
 ResultCode ServerSession::HandleSyncRequest(SharedPtr<Thread> thread) {
     // The ServerSession received a sync request, this means that there's new data available
     // from its ClientSession, so wake up any threads that may be waiting on a svcReplyAndReceive or
@@ -67,46 +94,39 @@ ResultCode ServerSession::HandleSyncRequest(SharedPtr<Thread> thread) {
     context.PopulateFromIncomingCommandBuffer(cmd_buf, *Kernel::g_current_process,
                                               Kernel::g_handle_table);
 
-    // If the session has been converted to a domain, handle the doomain request
+    ResultCode result = RESULT_SUCCESS;
+    // If the session has been converted to a domain, handle the domain request
     if (IsDomain()) {
-        auto& domain_message_header = context.GetDomainMessageHeader();
-        if (domain_message_header) {
-            // If there is a DomainMessageHeader, then this is CommandType "Request"
-            const u32 object_id{context.GetDomainMessageHeader()->object_id};
-            switch (domain_message_header->command) {
-            case IPC::DomainMessageHeader::CommandType::SendMessage:
-                return domain_request_handlers[object_id - 1]->HandleSyncRequest(context);
-
-            case IPC::DomainMessageHeader::CommandType::CloseVirtualHandle: {
-                LOG_DEBUG(IPC, "CloseVirtualHandle, object_id=0x%08X", object_id);
-
-                domain_request_handlers[object_id - 1] = nullptr;
-
-                IPC::ResponseBuilder rb{context, 2};
-                rb.Push(RESULT_SUCCESS);
-                return RESULT_SUCCESS;
-            }
-            }
-
-            LOG_CRITICAL(IPC, "Unknown domain command=%d", domain_message_header->command.Value());
-            ASSERT(false);
-        }
+        result = HandleDomainSyncRequest(context);
         // If there is no domain header, the regular session handler is used
+    } else if (hle_handler != nullptr) {
+        // If this ServerSession has an associated HLE handler, forward the request to it.
+        result = hle_handler->HandleSyncRequest(context);
     }
 
-    // If this ServerSession has an associated HLE handler, forward the request to it.
-    ResultCode result{RESULT_SUCCESS};
-    if (hle_handler != nullptr) {
-        // Attempt to translate the incoming request's command buffer.
-        ResultCode translate_result = TranslateHLERequest(this);
-        if (translate_result.IsError())
-            return translate_result;
+    if (thread->status == THREADSTATUS_RUNNING) {
+        // Put the thread to sleep until the server replies, it will be awoken in
+        // svcReplyAndReceive for LLE servers.
+        thread->status = THREADSTATUS_WAIT_IPC;
 
-        result = hle_handler->HandleSyncRequest(context);
-    } else {
-        // Add the thread to the list of threads that have issued a sync request with this
-        // server.
-        pending_requesting_threads.push_back(std::move(thread));
+        if (hle_handler != nullptr) {
+            // For HLE services, we put the request threads to sleep for a short duration to
+            // simulate IPC overhead, but only if the HLE handler didn't put the thread to sleep for
+            // other reasons like an async callback. The IPC overhead is needed to prevent
+            // starvation when a thread only does sync requests to HLE services while a
+            // lower-priority thread is waiting to run.
+
+            // This delay was approximated in a homebrew application by measuring the average time
+            // it takes for svcSendSyncRequest to return when performing the SetLcdForceBlack IPC
+            // request to the GSP:GPU service in a n3DS with firmware 11.6. The measured values have
+            // a high variance and vary between models.
+            static constexpr u64 IPCDelayNanoseconds = 39000;
+            thread->WakeAfterDelay(IPCDelayNanoseconds);
+        } else {
+            // Add the thread to the list of threads that have issued a sync request with this
+            // server.
+            pending_requesting_threads.push_back(std::move(thread));
+        }
     }
 
     // If this ServerSession does not have an HLE implementation, just wake up the threads waiting
@@ -140,9 +160,4 @@ ServerSession::SessionPair ServerSession::CreateSessionPair(const std::string& n
 
     return std::make_tuple(std::move(server_session), std::move(client_session));
 }
-
-ResultCode TranslateHLERequest(ServerSession* server_session) {
-    // TODO(Subv): Implement this function once multiple concurrent processes are supported.
-    return RESULT_SUCCESS;
-}
 } // namespace Kernel

src/core/hle/kernel/server_session.h

@@ -21,6 +21,7 @@ class ServerSession;
 class Session;
 class SessionRequestHandler;
 class Thread;
+class HLERequestContext;
 
 /**
  * Kernel object representing the server endpoint of an IPC session. Sessions are the basic CTR-OS
@@ -116,17 +117,12 @@ private:
      */
     static ResultVal<SharedPtr<ServerSession>> Create(std::string name = "Unknown");
 
+    /// Handles a SyncRequest to a domain, forwarding the request to the proper object or closing an
+    /// object handle.
+    ResultCode HandleDomainSyncRequest(Kernel::HLERequestContext& context);
+
     /// When set to True, converts the session to a domain at the end of the command
     bool convert_to_domain{};
 };
 
-/**
- * Performs command buffer translation for an HLE IPC request.
- * The command buffer from the ServerSession thread's TLS is copied into a
- * buffer and all descriptors in the buffer are processed.
- * TODO(Subv): Implement this function, currently we do not support multiple processes running at
- * once, but once that is implemented we'll need to properly translate all descriptors
- * in the command buffer.
- */
-ResultCode TranslateHLERequest(ServerSession* server_session);
 } // namespace Kernel

src/core/hle/kernel/shared_memory.cpp

@@ -111,13 +111,6 @@ ResultCode SharedMemory::Map(Process* target_process, VAddr address, MemoryPermi
         return ERR_INVALID_COMBINATION;
     }
 
-    // Heap-backed memory blocks can not be mapped with other_permissions = DontCare
-    if (base_address != 0 && other_permissions == MemoryPermission::DontCare) {
-        LOG_ERROR(Kernel, "cannot map id=%u, address=0x%llx name=%s, permissions don't match",
-                  GetObjectId(), address, name.c_str());
-        return ERR_INVALID_COMBINATION;
-    }
-
     // 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)) {
@@ -126,12 +119,6 @@ ResultCode SharedMemory::Map(Process* target_process, VAddr address, MemoryPermi
         return ERR_WRONG_PERMISSION;
     }
 
-    // TODO(Subv): Check for the Shared Device Mem flag in the creator process.
-    /*if (was_created_with_shared_device_mem && address != 0) {
-        return ResultCode(ErrorDescription::InvalidCombination, ErrorModule::OS,
-    ErrorSummary::InvalidArgument, ErrorLevel::Usage);
-    }*/
-
     // TODO(Subv): The same process that created a SharedMemory object
     // can not map it in its own address space unless it was created with addr=0, result 0xD900182C.
 

src/core/hle/kernel/svc.cpp

@@ -483,7 +483,7 @@ static void ExitProcess() {
     g_current_process->status = ProcessStatus::Exited;
 
     // Stop all the process threads that are currently waiting for objects.
-    auto& thread_list = GetThreadList();
+    auto& thread_list = Core::System::GetInstance().Scheduler().GetThreadList();
     for (auto& thread : thread_list) {
         if (thread->owner_process != g_current_process)
             continue;
@@ -585,7 +585,7 @@ static void SleepThread(s64 nanoseconds) {
 
     // Don't attempt to yield execution if there are no available threads to run,
     // this way we avoid a useless reschedule to the idle thread.
-    if (nanoseconds == 0 && !HaveReadyThreads())
+    if (nanoseconds == 0 && !Core::System::GetInstance().Scheduler().HaveReadyThreads())
         return;
 
     // Sleep current thread and check for next thread to schedule

src/core/hle/kernel/thread.cpp

@@ -3,6 +3,7 @@
 // Refer to the license.txt file included.
 
 #include <algorithm>
+#include <cinttypes>
 #include <list>
 #include <vector>
 #include "common/assert.h"
@@ -40,14 +41,6 @@ void Thread::Acquire(Thread* thread) {
 //               us to simply use a pool index or similar.
 static Kernel::HandleTable wakeup_callback_handle_table;
 
-// Lists all thread ids that aren't deleted/etc.
-static std::vector<SharedPtr<Thread>> thread_list;
-
-// Lists only ready thread ids.
-static Common::ThreadQueueList<Thread*, THREADPRIO_LOWEST + 1> ready_queue;
-
-static SharedPtr<Thread> current_thread;
-
 // The first available thread id at startup
 static u32 next_thread_id;
 
@@ -62,10 +55,6 @@ inline static u32 const NewThreadId() {
 Thread::Thread() {}
 Thread::~Thread() {}
 
-Thread* GetCurrentThread() {
-    return current_thread.get();
-}
-
 /**
  * Check if the specified thread is waiting on the specified address to be arbitrated
  * @param thread The thread to test
@@ -85,7 +74,7 @@ void Thread::Stop() {
     // Clean up thread from ready queue
     // This is only needed when the thread is termintated forcefully (SVC TerminateProcess)
     if (status == THREADSTATUS_READY) {
-        ready_queue.remove(current_priority, this);
+        Core::System::GetInstance().Scheduler().UnscheduleThread(this, current_priority);
     }
 
     status = THREADSTATUS_DEAD;
@@ -108,112 +97,6 @@ void Thread::Stop() {
     Kernel::g_current_process->tls_slots[tls_page].reset(tls_slot);
 }
 
-Thread* ArbitrateHighestPriorityThread(u32 address) {
-    Thread* highest_priority_thread = nullptr;
-    u32 priority = THREADPRIO_LOWEST;
-
-    // Iterate through threads, find highest priority thread that is waiting to be arbitrated...
-    for (auto& thread : thread_list) {
-        if (!CheckWait_AddressArbiter(thread.get(), address))
-            continue;
-
-        if (thread == nullptr)
-            continue;
-
-        if (thread->current_priority <= priority) {
-            highest_priority_thread = thread.get();
-            priority = thread->current_priority;
-        }
-    }
-
-    // If a thread was arbitrated, resume it
-    if (nullptr != highest_priority_thread) {
-        highest_priority_thread->ResumeFromWait();
-    }
-
-    return highest_priority_thread;
-}
-
-void ArbitrateAllThreads(u32 address) {
-    // Resume all threads found to be waiting on the address
-    for (auto& thread : thread_list) {
-        if (CheckWait_AddressArbiter(thread.get(), address))
-            thread->ResumeFromWait();
-    }
-}
-
-/**
- * Switches the CPU's active thread context to that of the specified thread
- * @param new_thread The thread to switch to
- */
-static void SwitchContext(Thread* new_thread) {
-    Thread* previous_thread = GetCurrentThread();
-
-    // Save context for previous thread
-    if (previous_thread) {
-        previous_thread->last_running_ticks = CoreTiming::GetTicks();
-        Core::CPU().SaveContext(previous_thread->context);
-
-        if (previous_thread->status == THREADSTATUS_RUNNING) {
-            // This is only the case when a reschedule is triggered without the current thread
-            // yielding execution (i.e. an event triggered, system core time-sliced, etc)
-            ready_queue.push_front(previous_thread->current_priority, previous_thread);
-            previous_thread->status = THREADSTATUS_READY;
-        }
-    }
-
-    // Load context of new thread
-    if (new_thread) {
-        ASSERT_MSG(new_thread->status == THREADSTATUS_READY,
-                   "Thread must be ready to become running.");
-
-        // Cancel any outstanding wakeup events for this thread
-        CoreTiming::UnscheduleEvent(ThreadWakeupEventType, new_thread->callback_handle);
-
-        auto previous_process = Kernel::g_current_process;
-
-        current_thread = new_thread;
-
-        ready_queue.remove(new_thread->current_priority, new_thread);
-        new_thread->status = THREADSTATUS_RUNNING;
-
-        if (previous_process != current_thread->owner_process) {
-            Kernel::g_current_process = current_thread->owner_process;
-            SetCurrentPageTable(&Kernel::g_current_process->vm_manager.page_table);
-        }
-
-        Core::CPU().LoadContext(new_thread->context);
-        Core::CPU().SetTlsAddress(new_thread->GetTLSAddress());
-    } else {
-        current_thread = nullptr;
-        // Note: We do not reset the current process and current page table when idling because
-        // technically we haven't changed processes, our threads are just paused.
-    }
-}
-
-/**
- * Pops and returns the next thread from the thread queue
- * @return A pointer to the next ready thread
- */
-static Thread* PopNextReadyThread() {
-    Thread* next;
-    Thread* thread = GetCurrentThread();
-
-    if (thread && thread->status == THREADSTATUS_RUNNING) {
-        // We have to do better than the current thread.
-        // This call returns null when that's not possible.
-        next = ready_queue.pop_first_better(thread->current_priority);
-        if (!next) {
-            // Otherwise just keep going with the current thread
-            next = thread;
-        }
-    } else {
-        next = ready_queue.pop_first();
-    }
-
-    return next;
-}
-
 void WaitCurrentThread_Sleep() {
     Thread* thread = GetCurrentThread();
     thread->status = THREADSTATUS_WAIT_SLEEP;
@@ -228,8 +111,7 @@ void WaitCurrentThread_ArbitrateAddress(VAddr wait_address) {
 void ExitCurrentThread() {
     Thread* thread = GetCurrentThread();
     thread->Stop();
-    thread_list.erase(std::remove(thread_list.begin(), thread_list.end(), thread),
-                      thread_list.end());
+    Core::System::GetInstance().Scheduler().RemoveThread(thread);
 }
 
 /**
@@ -283,6 +165,7 @@ void Thread::ResumeFromWait() {
     case THREADSTATUS_WAIT_SYNCH_ANY:
     case THREADSTATUS_WAIT_ARB:
     case THREADSTATUS_WAIT_SLEEP:
+    case THREADSTATUS_WAIT_IPC:
         break;
 
     case THREADSTATUS_READY:
@@ -306,31 +189,11 @@ void Thread::ResumeFromWait() {
 
     wakeup_callback = nullptr;
 
-    ready_queue.push_back(current_priority, this);
     status = THREADSTATUS_READY;
+    Core::System::GetInstance().Scheduler().ScheduleThread(this, current_priority);
     Core::System::GetInstance().PrepareReschedule();
 }
 
-/**
- * Prints the thread queue for debugging purposes
- */
-static void DebugThreadQueue() {
-    Thread* thread = GetCurrentThread();
-    if (!thread) {
-        LOG_DEBUG(Kernel, "Current: NO CURRENT THREAD");
-    } else {
-        LOG_DEBUG(Kernel, "0x%02X %u (current)", thread->current_priority,
-                  GetCurrentThread()->GetObjectId());
-    }
-
-    for (auto& t : thread_list) {
-        u32 priority = ready_queue.contains(t.get());
-        if (priority != -1) {
-            LOG_DEBUG(Kernel, "0x%02X %u", priority, t->GetObjectId());
-        }
-    }
-}
-
 /**
  * Finds a free location for the TLS section of a thread.
  * @param tls_slots The TLS page array of the thread's owner process.
@@ -379,7 +242,7 @@ ResultVal<SharedPtr<Thread>> Thread::Create(std::string name, VAddr entry_point,
                                             SharedPtr<Process> owner_process) {
     // Check if priority is in ranged. Lowest priority -> highest priority id.
     if (priority > THREADPRIO_LOWEST) {
-        LOG_ERROR(Kernel_SVC, "Invalid thread priority: %d", priority);
+        LOG_ERROR(Kernel_SVC, "Invalid thread priority: %u", priority);
         return ERR_OUT_OF_RANGE;
     }
 
@@ -391,15 +254,14 @@ ResultVal<SharedPtr<Thread>> Thread::Create(std::string name, VAddr entry_point,
     // TODO(yuriks): Other checks, returning 0xD9001BEA
 
     if (!Memory::IsValidVirtualAddress(*owner_process, entry_point)) {
-        LOG_ERROR(Kernel_SVC, "(name=%s): invalid entry %08x", name.c_str(), entry_point);
+        LOG_ERROR(Kernel_SVC, "(name=%s): invalid entry %016" PRIx64, name.c_str(), entry_point);
         // TODO (bunnei): Find the correct error code to use here
         return ResultCode(-1);
     }
 
     SharedPtr<Thread> thread(new Thread);
 
-    thread_list.push_back(thread);
-    ready_queue.prepare(priority);
+    Core::System::GetInstance().Scheduler().AddThread(thread, priority);
 
     thread->thread_id = NewThreadId();
     thread->status = THREADSTATUS_DORMANT;
@@ -470,12 +332,7 @@ 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.");
-    // If thread was ready, adjust queues
-    if (status == THREADSTATUS_READY)
-        ready_queue.move(this, current_priority, priority);
-    else
-        ready_queue.prepare(priority);
-
+    Core::System::GetInstance().Scheduler().SetThreadPriority(this, priority);
     nominal_priority = current_priority = priority;
 }
 
@@ -489,11 +346,7 @@ void Thread::UpdatePriority() {
 }
 
 void Thread::BoostPriority(u32 priority) {
-    // If thread was ready, adjust queues
-    if (status == THREADSTATUS_READY)
-        ready_queue.move(this, current_priority, priority);
-    else
-        ready_queue.prepare(priority);
+    Core::System::GetInstance().Scheduler().SetThreadPriority(this, priority);
     current_priority = priority;
 }
 
@@ -519,25 +372,6 @@ SharedPtr<Thread> SetupMainThread(VAddr entry_point, u32 priority,
     return thread;
 }
 
-bool HaveReadyThreads() {
-    return ready_queue.get_first() != nullptr;
-}
-
-void Reschedule() {
-    Thread* cur = GetCurrentThread();
-    Thread* next = PopNextReadyThread();
-
-    if (cur && next) {
-        LOG_TRACE(Kernel, "context switch %u -> %u", cur->GetObjectId(), next->GetObjectId());
-    } else if (cur) {
-        LOG_TRACE(Kernel, "context switch %u -> idle", cur->GetObjectId());
-    } else if (next) {
-        LOG_TRACE(Kernel, "context switch idle -> %u", next->GetObjectId());
-    }
-
-    SwitchContext(next);
-}
-
 void Thread::SetWaitSynchronizationResult(ResultCode result) {
     context.cpu_registers[0] = result.raw;
 }
@@ -560,25 +394,18 @@ VAddr Thread::GetCommandBufferAddress() const {
 
 ////////////////////////////////////////////////////////////////////////////////////////////////////
 
+/**
+ * Gets the current thread
+ */
+Thread* GetCurrentThread() {
+    return Core::System::GetInstance().Scheduler().GetCurrentThread();
+}
+
 void ThreadingInit() {
     ThreadWakeupEventType = CoreTiming::RegisterEvent("ThreadWakeupCallback", ThreadWakeupCallback);
-
-    current_thread = nullptr;
     next_thread_id = 1;
 }
 
-void ThreadingShutdown() {
-    current_thread = nullptr;
-
-    for (auto& t : thread_list) {
-        t->Stop();
-    }
-    thread_list.clear();
-    ready_queue.clear();
-}
-
-const std::vector<SharedPtr<Thread>>& GetThreadList() {
-    return thread_list;
-}
+void ThreadingShutdown() {}
 
 } // namespace Kernel

src/core/hle/kernel/thread.h

@@ -40,6 +40,7 @@ enum ThreadStatus {
     THREADSTATUS_READY,          ///< Ready to run
     THREADSTATUS_WAIT_ARB,       ///< Waiting on an address arbiter
     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
     THREADSTATUS_WAIT_SYNCH_ALL, ///< Waiting due to WaitSynchronizationN with wait_all = true
     THREADSTATUS_DORMANT,        ///< Created but not yet made ready
@@ -248,28 +249,6 @@ private:
 SharedPtr<Thread> SetupMainThread(VAddr entry_point, u32 priority,
                                   SharedPtr<Process> owner_process);
 
-/**
- * Returns whether there are any threads that are ready to run.
- */
-bool HaveReadyThreads();
-
-/**
- * Reschedules to the next available thread (call after current thread is suspended)
- */
-void Reschedule();
-
-/**
- * Arbitrate the highest priority thread that is waiting
- * @param address The address for which waiting threads should be arbitrated
- */
-Thread* ArbitrateHighestPriorityThread(VAddr address);
-
-/**
- * Arbitrate all threads currently waiting.
- * @param address The address for which waiting threads should be arbitrated
- */
-void ArbitrateAllThreads(VAddr address);
-
 /**
  * Gets the current thread
  */
@@ -301,9 +280,4 @@ void ThreadingInit();
  */
 void ThreadingShutdown();
 
-/**
- * Get a const reference to the thread list for debug use
- */
-const std::vector<SharedPtr<Thread>>& GetThreadList();
-
 } // namespace Kernel

src/core/hle/kernel/vm_manager.cpp

@@ -2,6 +2,7 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <cinttypes>
 #include <iterator>
 #include "common/assert.h"
 #include "common/logging/log.h"
@@ -206,7 +207,8 @@ void VMManager::RefreshMemoryBlockMappings(const std::vector<u8>* block) {
 void VMManager::LogLayout(Log::Level log_level) const {
     for (const auto& p : vma_map) {
         const VirtualMemoryArea& vma = p.second;
-        LOG_GENERIC(Log::Class::Kernel, log_level, "%08X - %08X  size: %8X %c%c%c %s", vma.base,
+        LOG_GENERIC(Log::Class::Kernel, log_level,
+                    "%016" PRIx64 " - %016" PRIx64 "  size: %16" PRIx64 " %c%c%c %s", vma.base,
                     vma.base + vma.size, vma.size,
                     (u8)vma.permissions & (u8)VMAPermission::Read ? 'R' : '-',
                     (u8)vma.permissions & (u8)VMAPermission::Write ? 'W' : '-',
@@ -222,8 +224,8 @@ VMManager::VMAIter VMManager::StripIterConstness(const VMAHandle& iter) {
 }
 
 ResultVal<VMManager::VMAIter> VMManager::CarveVMA(VAddr base, u64 size) {
-    ASSERT_MSG((size & Memory::PAGE_MASK) == 0, "non-page aligned size: 0x%8X", size);
-    ASSERT_MSG((base & Memory::PAGE_MASK) == 0, "non-page aligned base: 0x%08X", base);
+    ASSERT_MSG((size & Memory::PAGE_MASK) == 0, "non-page aligned size: 0x%16" PRIx64, size);
+    ASSERT_MSG((base & Memory::PAGE_MASK) == 0, "non-page aligned base: 0x%016" PRIx64, base);
 
     VMAIter vma_handle = StripIterConstness(FindVMA(base));
     if (vma_handle == vma_map.end()) {
@@ -258,8 +260,8 @@ ResultVal<VMManager::VMAIter> VMManager::CarveVMA(VAddr base, u64 size) {
 }
 
 ResultVal<VMManager::VMAIter> VMManager::CarveVMARange(VAddr target, u64 size) {
-    ASSERT_MSG((size & Memory::PAGE_MASK) == 0, "non-page aligned size: 0x%8X", size);
-    ASSERT_MSG((target & Memory::PAGE_MASK) == 0, "non-page aligned base: 0x%08X", target);
+    ASSERT_MSG((size & Memory::PAGE_MASK) == 0, "non-page aligned size: 0x%16" PRIx64, size);
+    ASSERT_MSG((target & Memory::PAGE_MASK) == 0, "non-page aligned base: 0x%016" PRIx64, target);
 
     VAddr target_end = target + size;
     ASSERT(target_end >= target);

src/core/hle/service/acc/acc_u0.cpp

@@ -9,6 +9,9 @@
 namespace Service {
 namespace Account {
 
+using Uid = std::array<u64, 2>;
+static constexpr Uid DEFAULT_USER_ID{0x10ull, 0x20ull};
+
 class IProfile final : public ServiceFramework<IProfile> {
 public:
     IProfile() : ServiceFramework("IProfile") {
@@ -61,6 +64,22 @@ void ACC_U0::GetUserExistence(Kernel::HLERequestContext& ctx) {
     rb.Push(true); // TODO: Check when this is supposed to return true and when not
 }
 
+void ACC_U0::ListAllUsers(Kernel::HLERequestContext& ctx) {
+    LOG_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};
+    rb.Push(RESULT_SUCCESS);
+}
+
+void ACC_U0::ListOpenUsers(Kernel::HLERequestContext& ctx) {
+    LOG_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};
+    rb.Push(RESULT_SUCCESS);
+}
+
 void ACC_U0::GetProfile(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
@@ -85,13 +104,14 @@ void ACC_U0::GetLastOpenedUser(Kernel::HLERequestContext& ctx) {
     LOG_WARNING(Service_ACC, "(STUBBED) called");
     IPC::ResponseBuilder rb{ctx, 6};
     rb.Push(RESULT_SUCCESS);
-    rb.Push<u64>(0x0);
-    rb.Push<u64>(0x0);
+    rb.PushRaw(DEFAULT_USER_ID);
 }
 
 ACC_U0::ACC_U0() : ServiceFramework("acc:u0") {
     static const FunctionInfo functions[] = {
         {1, &ACC_U0::GetUserExistence, "GetUserExistence"},
+        {2, &ACC_U0::ListAllUsers, "ListAllUsers"},
+        {3, &ACC_U0::ListOpenUsers, "ListOpenUsers"},
         {4, &ACC_U0::GetLastOpenedUser, "GetLastOpenedUser"},
         {5, &ACC_U0::GetProfile, "GetProfile"},
         {100, &ACC_U0::InitializeApplicationInfo, "InitializeApplicationInfo"},

src/core/hle/service/acc/acc_u0.h

@@ -28,6 +28,8 @@ public:
 
 private:
     void GetUserExistence(Kernel::HLERequestContext& ctx);
+    void ListAllUsers(Kernel::HLERequestContext& ctx);
+    void ListOpenUsers(Kernel::HLERequestContext& ctx);
     void GetLastOpenedUser(Kernel::HLERequestContext& ctx);
     void GetProfile(Kernel::HLERequestContext& ctx);
     void InitializeApplicationInfo(Kernel::HLERequestContext& ctx);

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

@@ -45,6 +45,7 @@ ISelfController::ISelfController(std::shared_ptr<NVFlinger::NVFlinger> nvflinger
     static const FunctionInfo functions[] = {
         {1, &ISelfController::LockExit, "LockExit"},
         {2, &ISelfController::UnlockExit, "UnlockExit"},
+        {9, &ISelfController::GetLibraryAppletLaunchableEvent, "GetLibraryAppletLaunchableEvent"},
         {11, &ISelfController::SetOperationModeChangedNotification,
          "SetOperationModeChangedNotification"},
         {12, &ISelfController::SetPerformanceModeChangedNotification,
@@ -55,6 +56,9 @@ ISelfController::ISelfController(std::shared_ptr<NVFlinger::NVFlinger> nvflinger
         {40, &ISelfController::CreateManagedDisplayLayer, "CreateManagedDisplayLayer"},
     };
     RegisterHandlers(functions);
+
+    launchable_event =
+        Kernel::Event::Create(Kernel::ResetType::OneShot, "ISelfController:LaunchableEvent");
 }
 
 void ISelfController::SetFocusHandlingMode(Kernel::HLERequestContext& ctx) {
@@ -132,6 +136,16 @@ void ISelfController::UnlockExit(Kernel::HLERequestContext& ctx) {
     LOG_WARNING(Service_AM, "(STUBBED) called");
 }
 
+void ISelfController::GetLibraryAppletLaunchableEvent(Kernel::HLERequestContext& ctx) {
+    launchable_event->Signal();
+
+    IPC::ResponseBuilder rb{ctx, 2, 1};
+    rb.Push(RESULT_SUCCESS);
+    rb.PushCopyObjects(launchable_event);
+
+    LOG_WARNING(Service_AM, "(STUBBED) called");
+}
+
 void ISelfController::CreateManagedDisplayLayer(Kernel::HLERequestContext& ctx) {
     // TODO(Subv): Find out how AM determines the display to use, for now just create the layer
     // in the Default display.
@@ -200,7 +214,69 @@ void ICommonStateGetter::GetPerformanceMode(Kernel::HLERequestContext& ctx) {
     LOG_WARNING(Service_AM, "(STUBBED) called");
 }
 
-ILibraryAppletCreator::ILibraryAppletCreator() : ServiceFramework("ILibraryAppletCreator") {}
+class ILibraryAppletAccessor final : public ServiceFramework<ILibraryAppletAccessor> {
+public:
+    explicit ILibraryAppletAccessor() : ServiceFramework("ILibraryAppletAccessor") {
+        static const FunctionInfo functions[] = {
+            {0, &ILibraryAppletAccessor::GetAppletStateChangedEvent, "GetAppletStateChangedEvent"},
+            {1, nullptr, "IsCompleted"},
+            {10, nullptr, "Start"},
+            {20, nullptr, "RequestExit"},
+            {25, nullptr, "Terminate"},
+            {30, nullptr, "GetResult"},
+            {50, nullptr, "SetOutOfFocusApplicationSuspendingEnabled"},
+            {100, nullptr, "PushInData"},
+            {101, nullptr, "PopOutData"},
+            {102, nullptr, "PushExtraStorage"},
+            {103, nullptr, "PushInteractiveInData"},
+            {104, nullptr, "PopInteractiveOutData"},
+            {105, nullptr, "GetPopOutDataEvent"},
+            {106, nullptr, "GetPopInteractiveOutDataEvent"},
+            {120, nullptr, "NeedsToExitProcess"},
+            {120, nullptr, "GetLibraryAppletInfo"},
+            {150, nullptr, "RequestForAppletToGetForeground"},
+            {160, nullptr, "GetIndirectLayerConsumerHandle"},
+        };
+        RegisterHandlers(functions);
+
+        state_changed_event = Kernel::Event::Create(Kernel::ResetType::OneShot,
+                                                    "ILibraryAppletAccessor:StateChangedEvent");
+    }
+
+private:
+    void GetAppletStateChangedEvent(Kernel::HLERequestContext& ctx) {
+        state_changed_event->Signal();
+
+        IPC::ResponseBuilder rb{ctx, 2, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushCopyObjects(state_changed_event);
+
+        LOG_WARNING(Service_AM, "(STUBBED) called");
+    }
+
+    Kernel::SharedPtr<Kernel::Event> state_changed_event;
+};
+
+ILibraryAppletCreator::ILibraryAppletCreator() : ServiceFramework("ILibraryAppletCreator") {
+    static const FunctionInfo functions[] = {
+        {0, &ILibraryAppletCreator::CreateLibraryApplet, "CreateLibraryApplet"},
+        {1, nullptr, "TerminateAllLibraryApplets"},
+        {2, nullptr, "AreAnyLibraryAppletsLeft"},
+        {10, nullptr, "CreateStorage"},
+        {11, nullptr, "CreateTransferMemoryStorage"},
+        {12, nullptr, "CreateHandleStorage"},
+    };
+    RegisterHandlers(functions);
+}
+
+void ILibraryAppletCreator::CreateLibraryApplet(Kernel::HLERequestContext& ctx) {
+    IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+
+    rb.Push(RESULT_SUCCESS);
+    rb.PushIpcInterface<AM::ILibraryAppletAccessor>();
+
+    LOG_DEBUG(Service_AM, "called");
+}
 
 class IStorageAccessor final : public ServiceFramework<IStorageAccessor> {
 public:
@@ -230,11 +306,11 @@ private:
 
         u64 offset = rp.Pop<u64>();
 
-        const auto& output_buffer = ctx.BufferDescriptorC()[0];
+        const size_t size{ctx.GetWriteBufferSize()};
 
-        ASSERT(offset + output_buffer.Size() <= buffer.size());
+        ASSERT(offset + size <= buffer.size());
 
-        Memory::WriteBlock(output_buffer.Address(), buffer.data() + offset, output_buffer.Size());
+        ctx.WriteBuffer(buffer.data() + offset, size);
 
         IPC::ResponseBuilder rb{ctx, 2};
 
@@ -302,8 +378,9 @@ void IApplicationFunctions::PopLaunchParameter(Kernel::HLERequestContext& ctx) {
 
 void IApplicationFunctions::EnsureSaveData(Kernel::HLERequestContext& ctx) {
     LOG_WARNING(Service, "(STUBBED) called");
-    IPC::ResponseBuilder rb{ctx, 2};
+    IPC::ResponseBuilder rb{ctx, 4};
     rb.Push(RESULT_SUCCESS);
+    rb.Push<u64>(0);
 }
 
 void IApplicationFunctions::SetTerminateResult(Kernel::HLERequestContext& ctx) {

src/core/hle/service/am/am.h

@@ -60,9 +60,11 @@ private:
     void SetOutOfFocusSuspendingEnabled(Kernel::HLERequestContext& ctx);
     void LockExit(Kernel::HLERequestContext& ctx);
     void UnlockExit(Kernel::HLERequestContext& ctx);
+    void GetLibraryAppletLaunchableEvent(Kernel::HLERequestContext& ctx);
     void CreateManagedDisplayLayer(Kernel::HLERequestContext& ctx);
 
     std::shared_ptr<NVFlinger::NVFlinger> nvflinger;
+    Kernel::SharedPtr<Kernel::Event> launchable_event;
 };
 
 class ICommonStateGetter final : public ServiceFramework<ICommonStateGetter> {
@@ -92,6 +94,9 @@ private:
 class ILibraryAppletCreator final : public ServiceFramework<ILibraryAppletCreator> {
 public:
     ILibraryAppletCreator();
+
+private:
+    void CreateLibraryApplet(Kernel::HLERequestContext& ctx);
 };
 
 class IApplicationFunctions final : public ServiceFramework<IApplicationFunctions> {

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

@@ -2,16 +2,37 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include "common/logging/log.h"
+#include "core/hle/ipc_helpers.h"
 #include "core/hle/service/aoc/aoc_u.h"
 
 namespace Service {
 namespace AOC {
 
+AOC_U::AOC_U() : ServiceFramework("aoc:u") {
+    static const FunctionInfo functions[] = {
+        {0, nullptr, "CountAddOnContentByApplicationId"},
+        {1, nullptr, "ListAddOnContentByApplicationId"},
+        {2, nullptr, "CountAddOnContent"},
+        {3, &AOC_U::ListAddOnContent, "ListAddOnContent"},
+        {4, nullptr, "GetAddOnContentBaseIdByApplicationId"},
+        {5, nullptr, "GetAddOnContentBaseId"},
+        {6, nullptr, "PrepareAddOnContentByApplicationId"},
+        {7, nullptr, "PrepareAddOnContent"},
+    };
+    RegisterHandlers(functions);
+}
+
+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");
+}
+
 void InstallInterfaces(SM::ServiceManager& service_manager) {
     std::make_shared<AOC_U>()->InstallAsService(service_manager);
 }
 
-AOC_U::AOC_U() : ServiceFramework("aoc:u") {}
-
 } // namespace AOC
 } // namespace Service

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

@@ -13,6 +13,9 @@ class AOC_U final : public ServiceFramework<AOC_U> {
 public:
     AOC_U();
     ~AOC_U() = default;
+
+private:
+    void ListAddOnContent(Kernel::HLERequestContext& ctx);
 };
 
 /// Registers all AOC services with the specified service manager.

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

@@ -5,63 +5,15 @@
 #include "common/logging/log.h"
 #include "core/hle/ipc_helpers.h"
 #include "core/hle/service/apm/apm.h"
+#include "core/hle/service/apm/interface.h"
 
 namespace Service {
 namespace APM {
 
 void InstallInterfaces(SM::ServiceManager& service_manager) {
-    std::make_shared<APM>()->InstallAsService(service_manager);
-}
-
-class ISession final : public ServiceFramework<ISession> {
-public:
-    ISession() : ServiceFramework("ISession") {
-        static const FunctionInfo functions[] = {
-            {0, &ISession::SetPerformanceConfiguration, "SetPerformanceConfiguration"},
-            {1, &ISession::GetPerformanceConfiguration, "GetPerformanceConfiguration"},
-        };
-        RegisterHandlers(functions);
-    }
-
-private:
-    void SetPerformanceConfiguration(Kernel::HLERequestContext& ctx) {
-        IPC::RequestParser rp{ctx};
-
-        auto mode = static_cast<PerformanceMode>(rp.Pop<u32>());
-        u32 config = rp.Pop<u32>();
-
-        IPC::ResponseBuilder rb{ctx, 2};
-        rb.Push(RESULT_SUCCESS);
-
-        LOG_WARNING(Service_APM, "(STUBBED) called mode=%u config=%u", static_cast<u32>(mode),
-                    config);
-    }
-
-    void GetPerformanceConfiguration(Kernel::HLERequestContext& ctx) {
-        IPC::RequestParser rp{ctx};
-
-        auto mode = static_cast<PerformanceMode>(rp.Pop<u32>());
-
-        IPC::ResponseBuilder rb{ctx, 3};
-        rb.Push(RESULT_SUCCESS);
-        rb.Push<u32>(0); // Performance configuration
-
-        LOG_WARNING(Service_APM, "(STUBBED) called mode=%u", static_cast<u32>(mode));
-    }
-};
-
-APM::APM() : ServiceFramework("apm") {
-    static const FunctionInfo functions[] = {
-        {0x00000000, &APM::OpenSession, "OpenSession"},
-        {0x00000001, nullptr, "GetPerformanceMode"},
-    };
-    RegisterHandlers(functions);
-}
-
-void APM::OpenSession(Kernel::HLERequestContext& ctx) {
-    IPC::ResponseBuilder rb{ctx, 2, 0, 1};
-    rb.Push(RESULT_SUCCESS);
-    rb.PushIpcInterface<ISession>();
+    auto module_ = std::make_shared<Module>();
+    std::make_shared<APM>(module_, "apm")->InstallAsService(service_manager);
+    std::make_shared<APM>(module_, "apm:p")->InstallAsService(service_manager);
 }
 
 } // namespace APM

src/core/hle/service/apm/apm.h

@@ -14,13 +14,10 @@ enum class PerformanceMode : u8 {
     Docked = 1,
 };
 
-class APM final : public ServiceFramework<APM> {
+class Module final {
 public:
-    APM();
-    ~APM() = default;
-
-private:
-    void OpenSession(Kernel::HLERequestContext& ctx);
+    Module() = default;
+    ~Module() = default;
 };
 
 /// Registers all AM services with the specified service manager.

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

@@ -0,0 +1,66 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "common/logging/log.h"
+#include "core/hle/ipc_helpers.h"
+#include "core/hle/service/apm/apm.h"
+#include "core/hle/service/apm/interface.h"
+
+namespace Service {
+namespace APM {
+
+class ISession final : public ServiceFramework<ISession> {
+public:
+    ISession() : ServiceFramework("ISession") {
+        static const FunctionInfo functions[] = {
+            {0, &ISession::SetPerformanceConfiguration, "SetPerformanceConfiguration"},
+            {1, &ISession::GetPerformanceConfiguration, "GetPerformanceConfiguration"},
+        };
+        RegisterHandlers(functions);
+    }
+
+private:
+    void SetPerformanceConfiguration(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+
+        auto mode = static_cast<PerformanceMode>(rp.Pop<u32>());
+        u32 config = rp.Pop<u32>();
+
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+
+        LOG_WARNING(Service_APM, "(STUBBED) called mode=%u config=%u", static_cast<u32>(mode),
+                    config);
+    }
+
+    void GetPerformanceConfiguration(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+
+        auto mode = static_cast<PerformanceMode>(rp.Pop<u32>());
+
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.Push<u32>(0); // Performance configuration
+
+        LOG_WARNING(Service_APM, "(STUBBED) called mode=%u", static_cast<u32>(mode));
+    }
+};
+
+APM::APM(std::shared_ptr<Module> apm, const char* name)
+    : ServiceFramework(name), apm(std::move(apm)) {
+    static const FunctionInfo functions[] = {
+        {0, &APM::OpenSession, "OpenSession"},
+        {1, nullptr, "GetPerformanceMode"},
+    };
+    RegisterHandlers(functions);
+}
+
+void APM::OpenSession(Kernel::HLERequestContext& ctx) {
+    IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+    rb.Push(RESULT_SUCCESS);
+    rb.PushIpcInterface<ISession>();
+}
+
+} // namespace APM
+} // namespace Service

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

@@ -0,0 +1,27 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "core/hle/service/service.h"
+
+namespace Service {
+namespace APM {
+
+class APM final : public ServiceFramework<APM> {
+public:
+    APM(std::shared_ptr<Module> apm, const char* name);
+    ~APM() = default;
+
+private:
+    void OpenSession(Kernel::HLERequestContext& ctx);
+
+    std::shared_ptr<Module> apm;
+};
+
+/// Registers all AM services with the specified service manager.
+void InstallInterfaces(SM::ServiceManager& service_manager);
+
+} // namespace APM
+} // namespace Service

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

@@ -99,8 +99,6 @@ private:
     void GetReleasedAudioOutBuffer_1(Kernel::HLERequestContext& ctx) {
         LOG_WARNING(Service_Audio, "(STUBBED) called");
 
-        const auto& buffer = ctx.BufferDescriptorB()[0];
-
         // TODO(st4rk): This is how libtransistor currently implements the
         // GetReleasedAudioOutBuffer, it should return the key (a VAddr) to the app and this address
         // is used to know which buffer should be filled with data and send again to the service
@@ -112,7 +110,7 @@ private:
             queue_keys.pop_back();
         }
 
-        Memory::WriteBlock(buffer.Address(), &key, sizeof(u64));
+        ctx.WriteBuffer(&key, sizeof(u64));
 
         IPC::ResponseBuilder rb{ctx, 3};
         rb.Push(RESULT_SUCCESS);
@@ -158,10 +156,8 @@ void AudOutU::ListAudioOuts(Kernel::HLERequestContext& ctx) {
     LOG_WARNING(Service_Audio, "(STUBBED) called");
     IPC::RequestParser rp{ctx};
 
-    auto& buffer = ctx.BufferDescriptorB()[0];
     const std::string audio_interface = "AudioInterface";
-
-    Memory::WriteBlock(buffer.Address(), &audio_interface[0], audio_interface.size());
+    ctx.WriteBuffer(audio_interface.c_str(), audio_interface.size());
 
     IPC::ResponseBuilder rb = rp.MakeBuilder(3, 0, 0);
 

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

@@ -3,13 +3,18 @@
 // Refer to the license.txt file included.
 
 #include "common/logging/log.h"
+#include "core/core_timing.h"
 #include "core/hle/ipc_helpers.h"
+#include "core/hle/kernel/event.h"
 #include "core/hle/kernel/hle_ipc.h"
 #include "core/hle/service/audio/audren_u.h"
 
 namespace Service {
 namespace Audio {
 
+/// TODO(bunnei): Find a proper value for the audio_ticks
+constexpr u64 audio_ticks{static_cast<u64>(BASE_CLOCK_RATE / 200)};
+
 class IAudioRenderer final : public ServiceFramework<IAudioRenderer> {
 public:
     IAudioRenderer() : ServiceFramework("IAudioRenderer") {
@@ -18,27 +23,167 @@ public:
             {0x1, nullptr, "GetAudioRendererSampleCount"},
             {0x2, nullptr, "GetAudioRendererMixBufferCount"},
             {0x3, nullptr, "GetAudioRendererState"},
-            {0x4, nullptr, "RequestUpdateAudioRenderer"},
-            {0x5, nullptr, "StartAudioRenderer"},
-            {0x6, nullptr, "StopAudioRenderer"},
-            {0x7, nullptr, "QuerySystemEvent"},
+            {0x4, &IAudioRenderer::RequestUpdateAudioRenderer, "RequestUpdateAudioRenderer"},
+            {0x5, &IAudioRenderer::StartAudioRenderer, "StartAudioRenderer"},
+            {0x6, &IAudioRenderer::StopAudioRenderer, "StopAudioRenderer"},
+            {0x7, &IAudioRenderer::QuerySystemEvent, "QuerySystemEvent"},
             {0x8, nullptr, "SetAudioRendererRenderingTimeLimit"},
             {0x9, nullptr, "GetAudioRendererRenderingTimeLimit"},
         };
         RegisterHandlers(functions);
+
+        system_event =
+            Kernel::Event::Create(Kernel::ResetType::OneShot, "IAudioRenderer:SystemEvent");
+
+        // Register event callback to update the Audio Buffer
+        audio_event = CoreTiming::RegisterEvent(
+            "IAudioRenderer::UpdateAudioCallback", [this](u64 userdata, int cycles_late) {
+                UpdateAudioCallback();
+                CoreTiming::ScheduleEvent(audio_ticks - cycles_late, audio_event);
+            });
+
+        // Start the audio event
+        CoreTiming::ScheduleEvent(audio_ticks, audio_event);
     }
     ~IAudioRenderer() = default;
+
+private:
+    void UpdateAudioCallback() {
+        system_event->Signal();
+    }
+
+    void RequestUpdateAudioRenderer(Kernel::HLERequestContext& ctx) {
+        LOG_DEBUG(Service_Audio, "%s", ctx.Description().c_str());
+        AudioRendererResponseData response_data{};
+
+        response_data.section_0_size =
+            response_data.state_entries.size() * sizeof(AudioRendererStateEntry);
+        response_data.section_1_size = response_data.section_1.size();
+        response_data.section_2_size = response_data.section_2.size();
+        response_data.section_3_size = response_data.section_3.size();
+        response_data.section_4_size = response_data.section_4.size();
+        response_data.section_5_size = response_data.section_5.size();
+        response_data.total_size = sizeof(AudioRendererResponseData);
+
+        for (unsigned i = 0; i < response_data.state_entries.size(); i++) {
+            // 4 = Busy and 5 = Ready?
+            response_data.state_entries[i].state = 5;
+        }
+
+        ctx.WriteBuffer(&response_data, response_data.total_size);
+
+        IPC::ResponseBuilder rb{ctx, 2};
+
+        rb.Push(RESULT_SUCCESS);
+
+        LOG_WARNING(Service_Audio, "(STUBBED) called");
+    }
+
+    void StartAudioRenderer(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2};
+
+        rb.Push(RESULT_SUCCESS);
+
+        LOG_WARNING(Service_Audio, "(STUBBED) called");
+    }
+
+    void StopAudioRenderer(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2};
+
+        rb.Push(RESULT_SUCCESS);
+
+        LOG_WARNING(Service_Audio, "(STUBBED) called");
+    }
+
+    void QuerySystemEvent(Kernel::HLERequestContext& ctx) {
+        // system_event->Signal();
+
+        IPC::ResponseBuilder rb{ctx, 2, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushCopyObjects(system_event);
+
+        LOG_WARNING(Service_Audio, "(STUBBED) called");
+    }
+
+    struct AudioRendererStateEntry {
+        u32_le state;
+        u32_le unknown_4;
+        u32_le unknown_8;
+        u32_le unknown_c;
+    };
+    static_assert(sizeof(AudioRendererStateEntry) == 0x10,
+                  "AudioRendererStateEntry has wrong size");
+
+    struct AudioRendererResponseData {
+        u32_le unknown_0;
+        u32_le section_5_size;
+        u32_le section_0_size;
+        u32_le section_1_size;
+        u32_le unknown_10;
+        u32_le section_2_size;
+        u32_le unknown_18;
+        u32_le section_3_size;
+        u32_le section_4_size;
+        u32_le unknown_24;
+        u32_le unknown_28;
+        u32_le unknown_2c;
+        u32_le unknown_30;
+        u32_le unknown_34;
+        u32_le unknown_38;
+        u32_le total_size;
+
+        std::array<AudioRendererStateEntry, 0x18e> state_entries;
+
+        std::array<u8, 0x600> section_1;
+        std::array<u8, 0xe0> section_2;
+        std::array<u8, 0x20> section_3;
+        std::array<u8, 0x10> section_4;
+        std::array<u8, 0xb0> section_5;
+    };
+    static_assert(sizeof(AudioRendererResponseData) == 0x20e0,
+                  "AudioRendererResponseData has wrong size");
+
+    /// This is used to trigger the audio event callback.
+    CoreTiming::EventType* audio_event;
+
+    Kernel::SharedPtr<Kernel::Event> system_event;
 };
 
 AudRenU::AudRenU() : ServiceFramework("audren:u") {
     static const FunctionInfo functions[] = {
-        {0x00000000, nullptr, "OpenAudioRenderer"},
-        {0x00000001, nullptr, "GetAudioRendererWorkBufferSize"},
-        {0x00000002, nullptr, "GetAudioRenderersProcessMasterVolume"},
-        {0x00000003, nullptr, "SetAudioRenderersProcessMasterVolume"},
+        {0, &AudRenU::OpenAudioRenderer, "OpenAudioRenderer"},
+        {1, &AudRenU::GetAudioRendererWorkBufferSize, "GetAudioRendererWorkBufferSize"},
+        {2, &AudRenU::GetAudioRenderersProcessMasterVolume, "GetAudioRenderersProcessMasterVolume"},
+        {3, nullptr, "SetAudioRenderersProcessMasterVolume"},
     };
     RegisterHandlers(functions);
 }
 
+void AudRenU::OpenAudioRenderer(Kernel::HLERequestContext& ctx) {
+    IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+
+    rb.Push(RESULT_SUCCESS);
+    rb.PushIpcInterface<Audio::IAudioRenderer>();
+
+    LOG_DEBUG(Service_Audio, "called");
+}
+
+void AudRenU::GetAudioRendererWorkBufferSize(Kernel::HLERequestContext& ctx) {
+    IPC::ResponseBuilder rb{ctx, 4};
+
+    rb.Push(RESULT_SUCCESS);
+    rb.Push<u64>(0x400);
+
+    LOG_WARNING(Service_Audio, "(STUBBED) called");
+}
+
+void AudRenU::GetAudioRenderersProcessMasterVolume(Kernel::HLERequestContext& ctx) {
+    IPC::ResponseBuilder rb{ctx, 2};
+
+    rb.Push(RESULT_SUCCESS);
+
+    LOG_WARNING(Service_Audio, "(STUBBED) called");
+}
+
 } // namespace Audio
 } // namespace Service

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

@@ -17,6 +17,11 @@ class AudRenU final : public ServiceFramework<AudRenU> {
 public:
     explicit AudRenU();
     ~AudRenU() = default;
+
+private:
+    void OpenAudioRenderer(Kernel::HLERequestContext& ctx);
+    void GetAudioRendererWorkBufferSize(Kernel::HLERequestContext& ctx);
+    void GetAudioRenderersProcessMasterVolume(Kernel::HLERequestContext& ctx);
 };
 
 } // namespace Audio

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

@@ -33,12 +33,10 @@ private:
         IPC::RequestParser rp{ctx};
         const s64 offset = rp.Pop<s64>();
         const s64 length = rp.Pop<s64>();
-        const auto& descriptor = ctx.BufferDescriptorB()[0];
 
         LOG_DEBUG(Service_FS, "called, offset=0x%llx, length=0x%llx", offset, length);
 
         // Error checking
-        ASSERT_MSG(length == descriptor.Size(), "unexpected size difference");
         if (length < 0) {
             IPC::ResponseBuilder rb{ctx, 2};
             rb.Push(ResultCode(ErrorModule::FS, ErrorDescription::InvalidLength));
@@ -60,7 +58,7 @@ private:
         }
 
         // Write the data to memory
-        Memory::WriteBlock(descriptor.Address(), output.data(), descriptor.Size());
+        ctx.WriteBuffer(output);
 
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
@@ -70,7 +68,9 @@ private:
 FSP_SRV::FSP_SRV() : ServiceFramework("fsp-srv") {
     static const FunctionInfo functions[] = {
         {1, &FSP_SRV::Initalize, "Initalize"},
+        {18, &FSP_SRV::MountSdCard, "MountSdCard"},
         {200, &FSP_SRV::OpenDataStorageByCurrentProcess, "OpenDataStorageByCurrentProcess"},
+        {202, nullptr, "OpenDataStorageByDataId"},
         {203, &FSP_SRV::OpenRomStorage, "OpenRomStorage"},
         {1005, &FSP_SRV::GetGlobalAccessLogMode, "GetGlobalAccessLogMode"},
     };
@@ -95,6 +95,13 @@ void FSP_SRV::Initalize(Kernel::HLERequestContext& ctx) {
     rb.Push(RESULT_SUCCESS);
 }
 
+void FSP_SRV::MountSdCard(Kernel::HLERequestContext& ctx) {
+    LOG_WARNING(Service_FS, "(STUBBED) called");
+
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(RESULT_SUCCESS);
+}
+
 void FSP_SRV::GetGlobalAccessLogMode(Kernel::HLERequestContext& ctx) {
     LOG_WARNING(Service_FS, "(STUBBED) called");
 

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

@@ -23,6 +23,7 @@ private:
     void TryLoadRomFS();
 
     void Initalize(Kernel::HLERequestContext& ctx);
+    void MountSdCard(Kernel::HLERequestContext& ctx);
     void GetGlobalAccessLogMode(Kernel::HLERequestContext& ctx);
     void OpenDataStorageByCurrentProcess(Kernel::HLERequestContext& ctx);
     void OpenRomStorage(Kernel::HLERequestContext& ctx);

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

@@ -0,0 +1,28 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "common/logging/log.h"
+#include "core/hle/ipc_helpers.h"
+#include "core/hle/service/friend/friend.h"
+#include "core/hle/service/friend/friend_a.h"
+
+namespace Service {
+namespace Friend {
+
+void Module::Interface::Unknown(Kernel::HLERequestContext& ctx) {
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(RESULT_SUCCESS);
+    LOG_WARNING(Service_Friend, "(STUBBED) called");
+}
+
+Module::Interface::Interface(std::shared_ptr<Module> module, const char* name)
+    : ServiceFramework(name), module(std::move(module)) {}
+
+void InstallInterfaces(SM::ServiceManager& service_manager) {
+    auto module = std::make_shared<Module>();
+    std::make_shared<Friend_A>(module)->InstallAsService(service_manager);
+}
+
+} // namespace Friend
+} // namespace Service

src/core/hle/service/friend/friend.h

@@ -0,0 +1,29 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "core/hle/service/service.h"
+
+namespace Service {
+namespace Friend {
+
+class Module final {
+public:
+    class Interface : public ServiceFramework<Interface> {
+    public:
+        Interface(std::shared_ptr<Module> module, const char* name);
+
+        void Unknown(Kernel::HLERequestContext& ctx);
+
+    protected:
+        std::shared_ptr<Module> module;
+    };
+};
+
+/// Registers all Friend services with the specified service manager.
+void InstallInterfaces(SM::ServiceManager& service_manager);
+
+} // namespace Friend
+} // namespace Service

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

@@ -0,0 +1,19 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/hle/service/friend/friend_a.h"
+
+namespace Service {
+namespace Friend {
+
+Friend_A::Friend_A(std::shared_ptr<Module> module)
+    : Module::Interface(std::move(module), "friend:a") {
+    static const FunctionInfo functions[] = {
+        {0, &Friend_A::Unknown, "Unknown"},
+    };
+    RegisterHandlers(functions);
+}
+
+} // namespace Friend
+} // namespace Service

src/core/hle/service/friend/friend_a.h

@@ -0,0 +1,18 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "core/hle/service/friend/friend.h"
+
+namespace Service {
+namespace Friend {
+
+class Friend_A final : public Module::Interface {
+public:
+    explicit Friend_A(std::shared_ptr<Module> module);
+};
+
+} // namespace Friend
+} // namespace Service

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

@@ -9,6 +9,7 @@
 #include "core/hle/ipc_helpers.h"
 #include "core/hle/kernel/client_port.h"
 #include "core/hle/kernel/client_session.h"
+#include "core/hle/kernel/event.h"
 #include "core/hle/kernel/shared_memory.h"
 #include "core/hle/service/hid/hid.h"
 #include "core/hle/service/service.h"
@@ -175,21 +176,37 @@ public:
             {0, &Hid::CreateAppletResource, "CreateAppletResource"},
             {1, &Hid::ActivateDebugPad, "ActivateDebugPad"},
             {11, &Hid::ActivateTouchScreen, "ActivateTouchScreen"},
+            {21, &Hid::ActivateMouse, "ActivateMouse"},
+            {31, &Hid::ActivateKeyboard, "ActivateKeyboard"},
             {66, &Hid::StartSixAxisSensor, "StartSixAxisSensor"},
+            {79, &Hid::SetGyroscopeZeroDriftMode, "SetGyroscopeZeroDriftMode"},
             {100, &Hid::SetSupportedNpadStyleSet, "SetSupportedNpadStyleSet"},
             {102, &Hid::SetSupportedNpadIdType, "SetSupportedNpadIdType"},
             {103, &Hid::ActivateNpad, "ActivateNpad"},
+            {106, &Hid::AcquireNpadStyleSetUpdateEventHandle,
+             "AcquireNpadStyleSetUpdateEventHandle"},
             {120, &Hid::SetNpadJoyHoldType, "SetNpadJoyHoldType"},
+            {121, &Hid::GetNpadJoyHoldType, "GetNpadJoyHoldType"},
+            {122, &Hid::SetNpadJoyAssignmentModeSingleByDefault,
+             "SetNpadJoyAssignmentModeSingleByDefault"},
             {124, nullptr, "SetNpadJoyAssignmentModeDual"},
             {128, &Hid::SetNpadHandheldActivationMode, "SetNpadHandheldActivationMode"},
+            {200, &Hid::GetVibrationDeviceInfo, "GetVibrationDeviceInfo"},
+            {201, &Hid::SendVibrationValue, "SendVibrationValue"},
+            {202, &Hid::GetActualVibrationValue, "GetActualVibrationValue"},
             {203, &Hid::CreateActiveVibrationDeviceList, "CreateActiveVibrationDeviceList"},
+            {206, &Hid::SendVibrationValues, "SendVibrationValues"},
         };
         RegisterHandlers(functions);
+
+        event = Kernel::Event::Create(Kernel::ResetType::OneShot, "hid:EventHandle");
     }
     ~Hid() = default;
 
 private:
     std::shared_ptr<IAppletResource> applet_resource;
+    u32 joy_hold_type{0};
+    Kernel::SharedPtr<Kernel::Event> event;
 
     void CreateAppletResource(Kernel::HLERequestContext& ctx) {
         if (applet_resource == nullptr) {
@@ -214,12 +231,30 @@ private:
         LOG_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");
+    }
+
+    void ActivateKeyboard(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+        LOG_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");
     }
 
+    void SetGyroscopeZeroDriftMode(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+        LOG_WARNING(Service_HID, "(STUBBED) called");
+    }
+
     void SetSupportedNpadStyleSet(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
@@ -238,24 +273,69 @@ private:
         LOG_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");
+    }
+
     void SetNpadJoyHoldType(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
         LOG_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");
+    }
+
+    void SetNpadJoyAssignmentModeSingleByDefault(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+        LOG_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");
+    }
+
+    void GetActualVibrationValue(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+        LOG_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");
     }
 
+    void GetVibrationDeviceInfo(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 4};
+        rb.Push(RESULT_SUCCESS);
+        rb.Push<u64>(0);
+        LOG_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");
     }
+
+    void SendVibrationValues(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+        LOG_WARNING(Service_HID, "(STUBBED) called");
+    }
 };
 
 void ReloadInputDevices() {}

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

@@ -2,6 +2,7 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <sstream>
 #include <string>
 #include "common/logging/log.h"
 #include "core/hle/ipc_helpers.h"
@@ -28,19 +29,28 @@ private:
             IsHead = 1,
             IsTail = 2,
         };
+        enum Severity : u32_le {
+            Trace,
+            Info,
+            Warning,
+            Error,
+            Critical,
+        };
 
         u64_le pid;
         u64_le threadContext;
         union {
             BitField<0, 16, Flags> flags;
-            BitField<16, 8, u32_le> severity;
+            BitField<16, 8, Severity> severity;
             BitField<24, 8, u32_le> verbosity;
         };
         u32_le payload_size;
 
-        /// Returns true if this is part of a single log message
-        bool IsSingleMessage() const {
-            return (flags & Flags::IsHead) && (flags & Flags::IsTail);
+        bool IsHeadLog() const {
+            return flags & Flags::IsHead;
+        }
+        bool IsTailLog() const {
+            return flags & Flags::IsTail;
         }
     };
     static_assert(sizeof(MessageHeader) == 0x18, "MessageHeader is incorrect size");
@@ -57,7 +67,7 @@ private:
     };
 
     /**
-     * LM::Initialize service function
+     * LM::Log service function
      *  Inputs:
      *      0: 0x00000000
      *  Outputs:
@@ -75,9 +85,9 @@ private:
         Memory::ReadBlock(addr, &header, sizeof(MessageHeader));
         addr += sizeof(MessageHeader);
 
-        if (!header.IsSingleMessage()) {
-            LOG_WARNING(Service_LM, "Multi message logs are unimplemeneted");
-            return;
+        if (header.IsHeadLog()) {
+            log_stream.str("");
+            log_stream.clear();
         }
 
         // Parse out log metadata
@@ -85,7 +95,7 @@ private:
         std::string message, filename, function;
         while (addr < end_addr) {
             const Field field{static_cast<Field>(Memory::Read8(addr++))};
-            size_t length{Memory::Read8(addr++)};
+            const size_t length{Memory::Read8(addr++)};
 
             if (static_cast<Field>(Memory::Read8(addr)) == Field::Skip) {
                 ++addr;
@@ -110,28 +120,47 @@ private:
         }
 
         // Empty log - nothing to do here
-        if (message.empty()) {
+        if (log_stream.str().empty() && message.empty()) {
             return;
         }
 
         // Format a nicely printable string out of the log metadata
-        std::string output;
-        if (filename.size()) {
-            output += filename + ':';
+        if (!filename.empty()) {
+            log_stream << filename << ':';
         }
-        if (function.size()) {
-            output += function + ':';
+        if (!function.empty()) {
+            log_stream << function << ':';
         }
         if (line) {
-            output += std::to_string(line) + ':';
+            log_stream << std::to_string(line) << ':';
         }
-        if (output.length() > 0 && output.back() == ':') {
-            output += ' ';
+        if (log_stream.str().length() > 0 && log_stream.str().back() == ':') {
+            log_stream << ' ';
         }
-        output += message;
+        log_stream << message;
 
-        LOG_INFO(Debug_Emulated, "%s", output.c_str());
+        if (header.IsTailLog()) {
+            switch (header.severity) {
+            case MessageHeader::Severity::Trace:
+                LOG_TRACE(Debug_Emulated, "%s", log_stream.str().c_str());
+                break;
+            case MessageHeader::Severity::Info:
+                LOG_INFO(Debug_Emulated, "%s", log_stream.str().c_str());
+                break;
+            case MessageHeader::Severity::Warning:
+                LOG_WARNING(Debug_Emulated, "%s", log_stream.str().c_str());
+                break;
+            case MessageHeader::Severity::Error:
+                LOG_ERROR(Debug_Emulated, "%s", log_stream.str().c_str());
+                break;
+            case MessageHeader::Severity::Critical:
+                LOG_CRITICAL(Debug_Emulated, "%s", log_stream.str().c_str());
+                break;
+            }
+        }
     }
+
+    std::ostringstream log_stream;
 };
 
 void InstallInterfaces(SM::ServiceManager& service_manager) {

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

@@ -0,0 +1,16 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/hle/service/ns/ns.h"
+#include "core/hle/service/ns/pl_u.h"
+
+namespace Service {
+namespace NS {
+
+void InstallInterfaces(SM::ServiceManager& service_manager) {
+    std::make_shared<PL_U>()->InstallAsService(service_manager);
+}
+
+} // namespace NS
+} // namespace Service

src/core/hle/service/ns/ns.h

@@ -0,0 +1,16 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "core/hle/service/service.h"
+
+namespace Service {
+namespace NS {
+
+/// Registers all NS services with the specified service manager.
+void InstallInterfaces(SM::ServiceManager& service_manager);
+
+} // namespace NS
+} // namespace Service

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

@@ -0,0 +1,111 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "common/common_paths.h"
+#include "common/file_util.h"
+#include "core/hle/ipc_helpers.h"
+#include "core/hle/service/ns/pl_u.h"
+
+namespace Service {
+namespace NS {
+
+struct FontRegion {
+    u32 offset;
+    u32 size;
+};
+
+// The below data is specific to shared font data dumped from Switch on f/w 2.2
+// Virtual address and offsets/sizes likely will vary by dump
+static constexpr VAddr SHARED_FONT_MEM_VADDR{0x00000009d3016000ULL};
+static constexpr u64 SHARED_FONT_MEM_SIZE{0x1100000};
+static constexpr std::array<FontRegion, 6> SHARED_FONT_REGIONS{
+    FontRegion{0x00000008, 0x001fe764}, FontRegion{0x001fe774, 0x00773e58},
+    FontRegion{0x009725d4, 0x0001aca8}, FontRegion{0x0098d284, 0x00369cec},
+    FontRegion{0x00cf6f78, 0x0039b858}, FontRegion{0x010927d8, 0x00019e80},
+};
+
+enum class LoadState : u32 {
+    Loading = 0,
+    Done = 1,
+};
+
+PL_U::PL_U() : ServiceFramework("pl:u") {
+    static const FunctionInfo functions[] = {
+        {1, &PL_U::GetLoadState, "GetLoadState"},
+        {2, &PL_U::GetSize, "GetSize"},
+        {3, &PL_U::GetSharedMemoryAddressOffset, "GetSharedMemoryAddressOffset"},
+        {4, &PL_U::GetSharedMemoryNativeHandle, "GetSharedMemoryNativeHandle"}};
+    RegisterHandlers(functions);
+
+    // Attempt to load shared font data from disk
+    const std::string filepath{FileUtil::GetUserPath(D_SYSDATA_IDX) + SHARED_FONT};
+    FileUtil::CreateFullPath(filepath); // Create path if not already created
+    FileUtil::IOFile file(filepath, "rb");
+
+    if (file.IsOpen()) {
+        // Read shared font data
+        ASSERT(file.GetSize() == SHARED_FONT_MEM_SIZE);
+        shared_font = std::make_shared<std::vector<u8>>(static_cast<size_t>(file.GetSize()));
+        file.ReadBytes(shared_font->data(), shared_font->size());
+    } else {
+        LOG_WARNING(Service_NS, "Unable to load shared font: %s", filepath.c_str());
+    }
+}
+
+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);
+    IPC::ResponseBuilder rb{ctx, 3};
+    rb.Push(RESULT_SUCCESS);
+    rb.Push<u32>(static_cast<u32>(LoadState::Done));
+}
+
+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);
+    IPC::ResponseBuilder rb{ctx, 3};
+    rb.Push(RESULT_SUCCESS);
+    rb.Push<u32>(SHARED_FONT_REGIONS[font_id].size);
+}
+
+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);
+    IPC::ResponseBuilder rb{ctx, 3};
+    rb.Push(RESULT_SUCCESS);
+    rb.Push<u32>(SHARED_FONT_REGIONS[font_id].offset);
+}
+
+void PL_U::GetSharedMemoryNativeHandle(Kernel::HLERequestContext& ctx) {
+    if (shared_font != nullptr) {
+        // TODO(bunnei): This is a less-than-ideal solution to load a RAM dump of the Switch shared
+        // font data. This (likely) relies on exact address, size, and offsets from the original
+        // dump. In the future, we need to replace this with a more robust solution.
+
+        // Map backing memory for the font data
+        Kernel::g_current_process->vm_manager.MapMemoryBlock(SHARED_FONT_MEM_VADDR, shared_font, 0,
+                                                             SHARED_FONT_MEM_SIZE,
+                                                             Kernel::MemoryState::Shared);
+
+        // Create shared font memory object
+        shared_font_mem = Kernel::SharedMemory::Create(
+            Kernel::g_current_process, SHARED_FONT_MEM_SIZE, Kernel::MemoryPermission::ReadWrite,
+            Kernel::MemoryPermission::Read, SHARED_FONT_MEM_VADDR, Kernel::MemoryRegion::BASE,
+            "PL_U:shared_font_mem");
+    }
+
+    LOG_DEBUG(Service_NS, "called");
+    IPC::ResponseBuilder rb{ctx, 2, 1};
+    rb.Push(RESULT_SUCCESS);
+    rb.PushCopyObjects(shared_font_mem);
+}
+
+} // namespace NS
+} // namespace Service

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

@@ -0,0 +1,33 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <memory>
+#include "core/hle/kernel/shared_memory.h"
+#include "core/hle/service/service.h"
+
+namespace Service {
+namespace NS {
+
+class PL_U final : public ServiceFramework<PL_U> {
+public:
+    PL_U();
+    ~PL_U() = default;
+
+private:
+    void GetLoadState(Kernel::HLERequestContext& ctx);
+    void GetSize(Kernel::HLERequestContext& ctx);
+    void GetSharedMemoryAddressOffset(Kernel::HLERequestContext& ctx);
+    void GetSharedMemoryNativeHandle(Kernel::HLERequestContext& ctx);
+
+    /// Handle to shared memory region designated for a shared font
+    Kernel::SharedPtr<Kernel::SharedMemory> shared_font_mem;
+
+    /// Backing memory for the shared font data
+    std::shared_ptr<std::vector<u8>> shared_font;
+};
+
+} // namespace NS
+} // namespace Service

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

@@ -20,15 +20,17 @@ u32 nvdisp_disp0::ioctl(Ioctl command, const std::vector<u8>& input, std::vector
 }
 
 void nvdisp_disp0::flip(u32 buffer_handle, u32 offset, u32 format, u32 width, u32 height,
-                        u32 stride) {
+                        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);
 
     using PixelFormat = RendererBase::FramebufferInfo::PixelFormat;
+    using Flags = NVFlinger::BufferQueue::BufferTransformFlags;
+    const bool flip_vertical = static_cast<u32>(transform) & static_cast<u32>(Flags::FlipV);
     const RendererBase::FramebufferInfo framebuffer_info{
-        addr, offset, width, height, stride, static_cast<PixelFormat>(format)};
+        addr, offset, width, height, stride, static_cast<PixelFormat>(format), flip_vertical};
 
     Core::System::GetInstance().perf_stats.EndGameFrame();
     VideoCore::g_renderer->SwapBuffers(framebuffer_info);

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

@@ -8,6 +8,7 @@
 #include <vector>
 #include "common/common_types.h"
 #include "core/hle/service/nvdrv/devices/nvdevice.h"
+#include "core/hle/service/nvflinger/buffer_queue.h"
 
 namespace Service {
 namespace Nvidia {
@@ -23,7 +24,8 @@ public:
     u32 ioctl(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override;
 
     /// Performs a screen flip, drawing the buffer pointed to by the handle.
-    void flip(u32 buffer_handle, u32 offset, u32 format, u32 width, u32 height, u32 stride);
+    void flip(u32 buffer_handle, u32 offset, u32 format, u32 width, u32 height, u32 stride,
+              NVFlinger::BufferQueue::BufferTransformFlags transform);
 
 private:
     std::shared_ptr<nvmap> nvmap_dev;

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

@@ -2,17 +2,20 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <cinttypes>
 #include "common/assert.h"
 #include "common/logging/log.h"
+#include "core/core.h"
 #include "core/hle/service/nvdrv/devices/nvhost_as_gpu.h"
+#include "core/hle/service/nvdrv/devices/nvmap.h"
 
 namespace Service {
 namespace Nvidia {
 namespace 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%llx, output_size=0x%llx",
-              command, input.size(), output.size());
+    LOG_DEBUG(Service_NVDRV, "called, command=0x%08x, input_size=0x%zx, output_size=0x%zx",
+              command.raw, input.size(), output.size());
 
     switch (static_cast<IoctlCommand>(command.raw)) {
     case IoctlCommand::IocInitalizeExCommand:
@@ -40,9 +43,17 @@ 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_WARNING(Service_NVDRV, "(STUBBED) called, pages=%x, page_size=%x, flags=%x", params.pages,
-                params.page_size, params.flags);
-    params.offset = 0xdeadbeef; // TODO(ogniK): Actually allocate space and give a real offset
+    LOG_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)};
+    if (params.flags & 1) {
+        params.offset = gpu.memory_manager->AllocateSpace(params.offset, size, 1);
+    } else {
+        params.offset = gpu.memory_manager->AllocateSpace(size, params.align);
+    }
+
     std::memcpy(output.data(), &params, output.size());
     return 0;
 }
@@ -51,12 +62,27 @@ u32 nvhost_as_gpu::MapBufferEx(const std::vector<u8>& input, std::vector<u8>& ou
     IoctlMapBufferEx params{};
     std::memcpy(&params, input.data(), input.size());
 
-    LOG_WARNING(Service_NVDRV,
-                "(STUBBED) called, flags=%x, nvmap_handle=%x, buffer_offset=%lx, mapping_size=%lx, "
-                "offset=%lx",
-                params.flags, params.nvmap_handle, params.buffer_offset, params.mapping_size,
-                params.offset);
-    params.offset = 0x0; // TODO(ogniK): Actually map and give a real offset
+    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);
+
+    if (!params.nvmap_handle) {
+        return 0;
+    }
+
+    auto object = nvmap_dev->GetObject(params.nvmap_handle);
+    ASSERT(object);
+
+    auto& gpu = Core::System::GetInstance().GPU();
+
+    if (params.flags & 1) {
+        params.offset = gpu.memory_manager->MapBufferEx(object->addr, params.offset, object->size);
+    } else {
+        params.offset = gpu.memory_manager->MapBufferEx(object->addr, object->size);
+    }
+
     std::memcpy(output.data(), &params, output.size());
     return 0;
 }
@@ -73,8 +99,8 @@ 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=%lx, buf_size=%x", params.buf_addr,
-                params.buf_size);
+    LOG_WARNING(Service_NVDRV, "(STUBBED) called, buf_addr=%" PRIu64 ", 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

@@ -4,6 +4,8 @@
 
 #pragma once
 
+#include <memory>
+#include <utility>
 #include <vector>
 #include "common/common_types.h"
 #include "common/swap.h"
@@ -13,9 +15,11 @@ namespace Service {
 namespace Nvidia {
 namespace Devices {
 
+class nvmap;
+
 class nvhost_as_gpu final : public nvdevice {
 public:
-    nvhost_as_gpu() = default;
+    nvhost_as_gpu(std::shared_ptr<nvmap> nvmap_dev) : nvmap_dev(std::move(nvmap_dev)) {}
     ~nvhost_as_gpu() override = default;
 
     u32 ioctl(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override;
@@ -92,6 +96,8 @@ private:
     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);
+
+    std::shared_ptr<nvmap> nvmap_dev;
 };
 
 } // namespace Devices

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

@@ -11,12 +11,14 @@ namespace Nvidia {
 namespace 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%lx, output_size=0x%lx", command,
-              input.size(), output.size());
+    LOG_DEBUG(Service_NVDRV, "called, command=0x%08x, input_size=0x%zx, output_size=0x%zx",
+              command.raw, input.size(), output.size());
 
     switch (static_cast<IoctlCommand>(command.raw)) {
     case IoctlCommand::IocGetConfigCommand:
         return NvOsGetConfigU32(input, output);
+    case IoctlCommand::IocCtrlEventWaitCommand:
+        return IocCtrlEventWait(input, output);
     }
     UNIMPLEMENTED();
     return 0;
@@ -45,6 +47,18 @@ u32 nvhost_ctrl::NvOsGetConfigU32(const std::vector<u8>& input, std::vector<u8>&
     return 0;
 }
 
+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);
+
+    // TODO(Subv): Implement actual syncpt waiting.
+    params.value = 0;
+    std::memcpy(output.data(), &params, sizeof(params));
+    return 0;
+}
+
 } // namespace Devices
 } // namespace Nvidia
 } // namespace Service

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

@@ -31,6 +31,7 @@ private:
         IocModuleRegRDWRCommand = 0xC008010E,
         IocSyncptWaitexCommand = 0xC0100019,
         IocSyncptReadMaxCommand = 0xC008001A,
+        IocCtrlEventWaitCommand = 0xC010001D,
         IocGetConfigCommand = 0xC183001B,
     };
 
@@ -41,7 +42,17 @@ private:
     };
     static_assert(sizeof(IocGetConfigParams) == 387, "IocGetConfigParams is incorrect size");
 
+    struct IocCtrlEventWaitParams {
+        u32_le syncpt_id;
+        u32_le threshold;
+        s32_le timeout;
+        u32_le value;
+    };
+    static_assert(sizeof(IocCtrlEventWaitParams) == 16, "IocCtrlEventWaitParams is incorrect size");
+
     u32 NvOsGetConfigU32(const std::vector<u8>& input, std::vector<u8>& output);
+
+    u32 IocCtrlEventWait(const std::vector<u8>& input, std::vector<u8>& output);
 };
 
 } // namespace Devices

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

@@ -2,6 +2,7 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <cinttypes>
 #include "common/assert.h"
 #include "common/logging/log.h"
 #include "core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.h"
@@ -11,8 +12,8 @@ namespace Nvidia {
 namespace 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%llx, output_size=0x%llx",
-              command, input.size(), output.size());
+    LOG_DEBUG(Service_NVDRV, "called, command=0x%08x, input_size=0x%zx, output_size=0x%zx",
+              command.raw, input.size(), output.size());
 
     switch (static_cast<IoctlCommand>(command.raw)) {
     case IoctlCommand::IocGetCharacteristicsCommand:
@@ -78,7 +79,7 @@ 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%lx",
+    LOG_WARNING(Service_NVDRV, "(STUBBED) called, mask=0x%x, mask_buf_addr=0x%" PRIx64,
                 params.mask_buf_size, params.mask_buf_addr);
     std::memcpy(output.data(), &params, sizeof(params));
     return 0;
@@ -104,8 +105,20 @@ 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_WARNING(Service_NVDRV, "(STUBBED) called");
-    std::memset(output.data(), 0, output.size());
+    LOG_DEBUG(Service_NVDRV, "called");
+    IoctlNvgpuGpuZcullGetInfoArgs params{};
+    std::memcpy(&params, input.data(), input.size());
+    params.width_align_pixels = 0x20;
+    params.height_align_pixels = 0x20;
+    params.pixel_squares_by_aliquots = 0x400;
+    params.aliquot_total = 0x800;
+    params.region_byte_multiplier = 0x20;
+    params.region_header_size = 0x20;
+    params.subregion_header_size = 0xc0;
+    params.subregion_width_align_pixels = 0x20;
+    params.subregion_height_align_pixels = 0x40;
+    params.subregion_count = 0x10;
+    std::memcpy(output.data(), &params, output.size());
     return 0;
 }
 

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

@@ -2,9 +2,11 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <cinttypes>
 #include <map>
 #include "common/assert.h"
 #include "common/logging/log.h"
+#include "core/core.h"
 #include "core/hle/service/nvdrv/devices/nvhost_gpu.h"
 
 namespace Service {
@@ -12,8 +14,8 @@ namespace Nvidia {
 namespace 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%llx, output_size=0x%llx",
-              command, input.size(), output.size());
+    LOG_DEBUG(Service_NVDRV, "called, command=0x%08x, input_size=0x%zx, output_size=0x%zx",
+              command.raw, input.size(), output.size());
 
     switch (static_cast<IoctlCommand>(command.raw)) {
     case IoctlCommand::IocSetNVMAPfdCommand:
@@ -73,7 +75,8 @@ 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=%lx, mode=%x", zcull_params.gpu_va, zcull_params.mode);
+    LOG_DEBUG(Service_NVDRV, "called, gpu_va=%" PRIx64 ", mode=%x", zcull_params.gpu_va,
+              zcull_params.mode);
     std::memcpy(output.data(), &zcull_params, output.size());
     return 0;
 }
@@ -81,8 +84,8 @@ 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=%lx, size=%lx, mem=%x", params.offset,
-                params.size, params.mem);
+    LOG_WARNING(Service_NVDRV, "(STUBBED) called, offset=%" PRIx64 ", size=%" PRIx64 ", mem=%x",
+                params.offset, params.size, params.mem);
     std::memcpy(output.data(), &params, output.size());
     return 0;
 }
@@ -122,7 +125,7 @@ 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=%lx, num_entries=%x, flags=%x",
+    LOG_WARNING(Service_NVDRV, "(STUBBED) called, gpfifo=%" PRIx64 ", num_entries=%x, flags=%x",
                 params.gpfifo, params.num_entries, params.flags);
 
     auto entries = std::vector<IoctlGpfifoEntry>();
@@ -131,7 +134,7 @@ u32 nvhost_gpu::SubmitGPFIFO(const std::vector<u8>& input, std::vector<u8>& outp
                 params.num_entries * sizeof(IoctlGpfifoEntry));
     for (auto entry : entries) {
         VAddr va_addr = entry.Address();
-        // TODO(ogniK): Process these
+        Core::System::GetInstance().GPU().ProcessCommandList(va_addr, entry.sz);
     }
     params.fence_out.id = 0;
     params.fence_out.value = 0;

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

@@ -4,6 +4,7 @@
 
 #pragma once
 
+#include <memory>
 #include <vector>
 #include "common/common_types.h"
 #include "common/swap.h"
@@ -12,12 +13,14 @@
 namespace Service {
 namespace Nvidia {
 namespace Devices {
+
+class nvmap;
 constexpr u32 NVGPU_IOCTL_MAGIC('H');
 constexpr u32 NVGPU_IOCTL_CHANNEL_SUBMIT_GPFIFO(0x8);
 
 class nvhost_gpu final : public nvdevice {
 public:
-    nvhost_gpu() = default;
+    nvhost_gpu(std::shared_ptr<nvmap> nvmap_dev) : nvmap_dev(std::move(nvmap_dev)) {}
     ~nvhost_gpu() override = default;
 
     u32 ioctl(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override;
@@ -132,6 +135,8 @@ private:
     u32 AllocGPFIFOEx2(const std::vector<u8>& input, std::vector<u8>& output);
     u32 AllocateObjectContext(const std::vector<u8>& input, std::vector<u8>& output);
     u32 SubmitGPFIFO(const std::vector<u8>& input, std::vector<u8>& output);
+
+    std::shared_ptr<nvmap> nvmap_dev;
 };
 
 } // namespace Devices

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

@@ -3,6 +3,7 @@
 // Refer to the license.txt file included.
 
 #include <algorithm>
+#include <cinttypes>
 
 #include "common/assert.h"
 #include "common/logging/log.h"
@@ -13,10 +14,8 @@ namespace Nvidia {
 namespace Devices {
 
 VAddr nvmap::GetObjectAddress(u32 handle) const {
-    auto itr = handles.find(handle);
-    ASSERT(itr != handles.end());
-
-    auto object = itr->second;
+    auto object = GetObject(handle);
+    ASSERT(object);
     ASSERT(object->status == Object::Status::Allocated);
     return object->addr;
 }
@@ -52,7 +51,7 @@ u32 nvmap::IocCreate(const std::vector<u8>& input, std::vector<u8>& output) {
     u32 handle = next_handle++;
     handles[handle] = std::move(object);
 
-    LOG_WARNING(Service_NVDRV, "(STUBBED) size 0x%08X", params.size);
+    LOG_DEBUG(Service_NVDRV, "size=0x%08X", params.size);
 
     params.handle = handle;
 
@@ -64,17 +63,16 @@ u32 nvmap::IocAlloc(const std::vector<u8>& input, std::vector<u8>& output) {
     IocAllocParams params;
     std::memcpy(&params, input.data(), sizeof(params));
 
-    auto itr = handles.find(params.handle);
-    ASSERT(itr != handles.end());
+    auto object = GetObject(params.handle);
+    ASSERT(object);
 
-    auto object = itr->second;
     object->flags = params.flags;
     object->align = params.align;
     object->kind = params.kind;
     object->addr = params.addr;
     object->status = Object::Status::Allocated;
 
-    LOG_WARNING(Service_NVDRV, "(STUBBED) Allocated address 0x%llx", params.addr);
+    LOG_DEBUG(Service_NVDRV, "called, addr=0x%" PRIx64, params.addr);
 
     std::memcpy(output.data(), &params, sizeof(params));
     return 0;
@@ -86,10 +84,10 @@ u32 nvmap::IocGetId(const std::vector<u8>& input, std::vector<u8>& output) {
 
     LOG_WARNING(Service_NVDRV, "called");
 
-    auto itr = handles.find(params.handle);
-    ASSERT(itr != handles.end());
+    auto object = GetObject(params.handle);
+    ASSERT(object);
 
-    params.id = itr->second->id;
+    params.id = object->id;
 
     std::memcpy(output.data(), &params, sizeof(params));
     return 0;
@@ -105,11 +103,8 @@ u32 nvmap::IocFromId(const std::vector<u8>& input, std::vector<u8>& output) {
                             [&](const auto& entry) { return entry.second->id == params.id; });
     ASSERT(itr != handles.end());
 
-    // Make a new handle for the object
-    u32 handle = next_handle++;
-    handles[handle] = itr->second;
-
-    params.handle = handle;
+    // Return the existing handle instead of creating a new one.
+    params.handle = itr->first;
 
     std::memcpy(output.data(), &params, sizeof(params));
     return 0;
@@ -123,10 +118,8 @@ u32 nvmap::IocParam(const std::vector<u8>& input, std::vector<u8>& output) {
 
     LOG_WARNING(Service_NVDRV, "(STUBBED) called type=%u", params.type);
 
-    auto itr = handles.find(params.handle);
-    ASSERT(itr != handles.end());
-
-    auto object = itr->second;
+    auto object = GetObject(params.handle);
+    ASSERT(object);
     ASSERT(object->status == Object::Status::Allocated);
 
     switch (static_cast<ParamTypes>(params.type)) {

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

@@ -26,8 +26,7 @@ public:
 
     u32 ioctl(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override;
 
-private:
-    // Represents an nvmap object.
+    /// Represents an nvmap object.
     struct Object {
         enum class Status { Created, Allocated };
         u32 id;
@@ -39,10 +38,19 @@ private:
         Status status;
     };
 
+    std::shared_ptr<Object> GetObject(u32 handle) const {
+        auto itr = handles.find(handle);
+        if (itr != handles.end()) {
+            return itr->second;
+        }
+        return {};
+    }
+
+private:
     /// Id to use for the next handle that is created.
     u32 next_handle = 1;
 
-    // Id to use for the next object that is created.
+    /// Id to use for the next object that is created.
     u32 next_id = 1;
 
     /// Mapping of currently allocated handles to the objects they represent.

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

@@ -2,6 +2,7 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <cinttypes>
 #include "common/logging/log.h"
 #include "core/hle/ipc_helpers.h"
 #include "core/hle/kernel/event.h"
@@ -14,9 +15,8 @@ namespace Nvidia {
 void NVDRV::Open(Kernel::HLERequestContext& ctx) {
     LOG_DEBUG(Service_NVDRV, "called");
 
-    auto buffer = ctx.BufferDescriptorA()[0];
-
-    std::string device_name = Memory::ReadCString(buffer.Address(), buffer.Size());
+    const auto& buffer = ctx.ReadBuffer();
+    std::string device_name(buffer.begin(), buffer.end());
 
     u32 fd = nvdrv->Open(device_name);
     IPC::ResponseBuilder rb{ctx, 4};
@@ -32,25 +32,13 @@ void NVDRV::Ioctl(Kernel::HLERequestContext& ctx) {
     u32 fd = rp.Pop<u32>();
     u32 command = rp.Pop<u32>();
 
+    std::vector<u8> output(ctx.GetWriteBufferSize());
+
     IPC::ResponseBuilder rb{ctx, 3};
     rb.Push(RESULT_SUCCESS);
-    if (ctx.BufferDescriptorA()[0].Size() != 0) {
-        auto input_buffer = ctx.BufferDescriptorA()[0];
-        auto output_buffer = ctx.BufferDescriptorB()[0];
-        std::vector<u8> input(input_buffer.Size());
-        std::vector<u8> output(output_buffer.Size());
-        Memory::ReadBlock(input_buffer.Address(), input.data(), input_buffer.Size());
-        rb.Push(nvdrv->Ioctl(fd, command, input, output));
-        Memory::WriteBlock(output_buffer.Address(), output.data(), output_buffer.Size());
-    } else {
-        auto input_buffer = ctx.BufferDescriptorX()[0];
-        auto output_buffer = ctx.BufferDescriptorC()[0];
-        std::vector<u8> input(input_buffer.size);
-        std::vector<u8> output(output_buffer.size);
-        Memory::ReadBlock(input_buffer.Address(), input.data(), input_buffer.size);
-        rb.Push(nvdrv->Ioctl(fd, command, input, output));
-        Memory::WriteBlock(output_buffer.Address(), output.data(), output_buffer.size);
-    }
+    rb.Push(nvdrv->Ioctl(fd, command, ctx.ReadBuffer(), output));
+
+    ctx.WriteBuffer(output);
 }
 
 void NVDRV::Close(Kernel::HLERequestContext& ctx) {
@@ -78,18 +66,17 @@ void NVDRV::QueryEvent(Kernel::HLERequestContext& ctx) {
     u32 event_id = rp.Pop<u32>();
     LOG_WARNING(Service_NVDRV, "(STUBBED) called, fd=%x, event_id=%x", fd, event_id);
 
-    IPC::ResponseBuilder rb{ctx, 2, 1};
+    IPC::ResponseBuilder rb{ctx, 3, 1};
     rb.Push(RESULT_SUCCESS);
-    auto event = Kernel::Event::Create(Kernel::ResetType::Pulse, "NVEvent");
-    event->Signal();
-    rb.PushCopyObjects(event);
+    rb.PushCopyObjects(query_event);
+    rb.Push<u32>(0);
 }
 
 void NVDRV::SetClientPID(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
     pid = rp.Pop<u64>();
 
-    LOG_WARNING(Service_NVDRV, "(STUBBED) called, pid=0x%lx", pid);
+    LOG_WARNING(Service_NVDRV, "(STUBBED) called, pid=0x%" PRIx64, pid);
     IPC::ResponseBuilder rb{ctx, 3};
     rb.Push(RESULT_SUCCESS);
     rb.Push<u32>(0);
@@ -113,6 +100,8 @@ NVDRV::NVDRV(std::shared_ptr<Module> nvdrv, const char* name)
         {13, &NVDRV::FinishInitialize, "FinishInitialize"},
     };
     RegisterHandlers(functions);
+
+    query_event = Kernel::Event::Create(Kernel::ResetType::OneShot, "NVDRV::query_event");
 }
 
 } // namespace Nvidia

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

@@ -6,6 +6,7 @@
 
 #include <memory>
 #include <string>
+#include "core/hle/kernel/event.h"
 #include "core/hle/service/nvdrv/nvdrv.h"
 #include "core/hle/service/service.h"
 
@@ -29,6 +30,8 @@ private:
     std::shared_ptr<Module> nvdrv;
 
     u64 pid{};
+
+    Kernel::SharedPtr<Kernel::Event> query_event;
 };
 
 } // namespace Nvidia

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

@@ -31,12 +31,12 @@ void InstallInterfaces(SM::ServiceManager& service_manager) {
 
 Module::Module() {
     auto nvmap_dev = std::make_shared<Devices::nvmap>();
-    devices["/dev/nvhost-as-gpu"] = std::make_shared<Devices::nvhost_as_gpu>();
+    devices["/dev/nvhost-as-gpu"] = std::make_shared<Devices::nvhost_as_gpu>(nvmap_dev);
+    devices["/dev/nvhost-gpu"] = std::make_shared<Devices::nvhost_gpu>(nvmap_dev);
     devices["/dev/nvhost-ctrl-gpu"] = std::make_shared<Devices::nvhost_ctrl_gpu>();
     devices["/dev/nvmap"] = nvmap_dev;
     devices["/dev/nvdisp_disp0"] = std::make_shared<Devices::nvdisp_disp0>(nvmap_dev);
     devices["/dev/nvhost-ctrl"] = std::make_shared<Devices::nvhost_ctrl>();
-    devices["/dev/nvhost-gpu"] = std::make_shared<Devices::nvhost_gpu>();
 }
 
 u32 Module::Open(std::string device_name) {

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

@@ -17,6 +17,13 @@ namespace Devices {
 class nvdevice;
 }
 
+struct IoctlFence {
+    u32 id;
+    u32 value;
+};
+
+static_assert(sizeof(IoctlFence) == 8, "IoctlFence has wrong size");
+
 class Module final {
 public:
     Module();

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

@@ -40,7 +40,11 @@ u32 BufferQueue::DequeueBuffer(u32 pixel_format, u32 width, u32 height) {
         return igbp_buffer.format == pixel_format && igbp_buffer.width == width &&
                igbp_buffer.height == height;
     });
-    ASSERT(itr != queue.end());
+    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();
+    }
 
     itr->status = Buffer::Status::Dequeued;
     return itr->slot;
@@ -54,12 +58,13 @@ const IGBPBuffer& BufferQueue::RequestBuffer(u32 slot) const {
     return itr->igbp_buffer;
 }
 
-void BufferQueue::QueueBuffer(u32 slot) {
+void BufferQueue::QueueBuffer(u32 slot, BufferTransformFlags transform) {
     auto itr = std::find_if(queue.begin(), queue.end(),
                             [&](const Buffer& buffer) { return buffer.slot == slot; });
     ASSERT(itr != queue.end());
     ASSERT(itr->status == Buffer::Status::Dequeued);
     itr->status = Buffer::Status::Queued;
+    itr->transform = transform;
 }
 
 boost::optional<const BufferQueue::Buffer&> BufferQueue::AcquireBuffer() {

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

@@ -46,18 +46,32 @@ public:
     BufferQueue(u32 id, u64 layer_id);
     ~BufferQueue() = default;
 
+    enum class BufferTransformFlags : u32 {
+        /// Flip source image horizontally (around the vertical axis)
+        FlipH = 0x01,
+        /// Flip source image vertically (around the horizontal axis)
+        FlipV = 0x02,
+        /// Rotate source image 90 degrees clockwise
+        Rotate90 = 0x04,
+        /// Rotate source image 180 degrees
+        Roate180 = 0x03,
+        /// Rotate source image 270 degrees clockwise
+        Roate270 = 0x07,
+    };
+
     struct Buffer {
         enum class Status { Free = 0, Queued = 1, Dequeued = 2, Acquired = 3 };
 
         u32 slot;
         Status status = Status::Free;
         IGBPBuffer igbp_buffer;
+        BufferTransformFlags transform;
     };
 
     void SetPreallocatedBuffer(u32 slot, IGBPBuffer& buffer);
     u32 DequeueBuffer(u32 pixel_format, u32 width, u32 height);
     const IGBPBuffer& RequestBuffer(u32 slot) const;
-    void QueueBuffer(u32 slot);
+    void QueueBuffer(u32 slot, BufferTransformFlags transform);
     boost::optional<const Buffer&> AcquireBuffer();
     void ReleaseBuffer(u32 slot);
     u32 Query(QueryType type);

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

@@ -6,6 +6,7 @@
 
 #include "common/alignment.h"
 #include "common/scope_exit.h"
+#include "core/core.h"
 #include "core/core_timing.h"
 #include "core/hle/service/nvdrv/devices/nvdisp_disp0.h"
 #include "core/hle/service/nvdrv/nvdrv.h"
@@ -129,6 +130,7 @@ void NVFlinger::Compose() {
 
         if (buffer == boost::none) {
             // There was no queued buffer to draw, render previous frame
+            Core::System::GetInstance().perf_stats.EndGameFrame();
             VideoCore::g_renderer->SwapBuffers({});
             continue;
         }
@@ -145,7 +147,7 @@ void NVFlinger::Compose() {
         ASSERT(nvdisp);
 
         nvdisp->flip(igbp_buffer.gpu_buffer_id, igbp_buffer.offset, igbp_buffer.format,
-                     igbp_buffer.width, igbp_buffer.height, igbp_buffer.stride);
+                     igbp_buffer.width, igbp_buffer.height, igbp_buffer.stride, buffer->transform);
 
         buffer_queue->ReleaseBuffer(buffer->slot);
     }

src/core/hle/service/service.cpp

@@ -20,9 +20,11 @@
 #include "core/hle/service/apm/apm.h"
 #include "core/hle/service/audio/audio.h"
 #include "core/hle/service/filesystem/filesystem.h"
+#include "core/hle/service/friend/friend.h"
 #include "core/hle/service/hid/hid.h"
 #include "core/hle/service/lm/lm.h"
 #include "core/hle/service/nifm/nifm.h"
+#include "core/hle/service/ns/ns.h"
 #include "core/hle/service/nvdrv/nvdrv.h"
 #include "core/hle/service/pctl/pctl.h"
 #include "core/hle/service/service.h"
@@ -179,9 +181,11 @@ void Init() {
     APM::InstallInterfaces(*SM::g_service_manager);
     Audio::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);
     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);

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

@@ -17,9 +17,7 @@ void SET::GetAvailableLanguageCodes(Kernel::HLERequestContext& ctx) {
     u32 id = rp.Pop<u32>();
     constexpr std::array<u8, 13> lang_codes{};
 
-    const auto& output_buffer = ctx.BufferDescriptorC()[0];
-
-    Memory::WriteBlock(output_buffer.Address(), lang_codes.data(), lang_codes.size());
+    ctx.WriteBuffer(lang_codes.data(), lang_codes.size());
 
     IPC::ResponseBuilder rb{ctx, 2};
 

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

@@ -68,8 +68,15 @@ public:
     ITimeZoneService() : ServiceFramework("ITimeZoneService") {
         static const FunctionInfo functions[] = {
             {0, &ITimeZoneService::GetDeviceLocationName, "GetDeviceLocationName"},
+            {1, nullptr, "SetDeviceLocationName"},
             {2, &ITimeZoneService::GetTotalLocationNameCount, "GetTotalLocationNameCount"},
+            {3, nullptr, "LoadLocationNameList"},
+            {4, &ITimeZoneService::LoadTimeZoneRule, "LoadTimeZoneRule"},
+            {5, nullptr, "GetTimeZoneRuleVersion"},
+            {100, nullptr, "ToCalendarTime"},
             {101, &ITimeZoneService::ToCalendarTimeWithMyRule, "ToCalendarTimeWithMyRule"},
+            {200, nullptr, "ToPosixTime"},
+            {201, nullptr, "ToPosixTimeWithMyRule"},
         };
         RegisterHandlers(functions);
     }
@@ -90,6 +97,12 @@ private:
         rb.Push<u32>(0);
     }
 
+    void LoadTimeZoneRule(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_Time, "(STUBBED) called");
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+    }
+
     void ToCalendarTimeWithMyRule(Kernel::HLERequestContext& ctx) {
         IPC::RequestParser rp{ctx};
         u64 posix_time = rp.Pop<u64>();

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

@@ -3,11 +3,12 @@
 // Refer to the license.txt file included.
 
 #include <algorithm>
-
+#include <array>
 #include "common/alignment.h"
 #include "common/scope_exit.h"
 #include "core/core_timing.h"
 #include "core/hle/ipc_helpers.h"
+#include "core/hle/service/nvdrv/nvdrv.h"
 #include "core/hle/service/nvflinger/buffer_queue.h"
 #include "core/hle/service/vi/vi.h"
 #include "core/hle/service/vi/vi_m.h"
@@ -19,6 +20,15 @@
 namespace Service {
 namespace VI {
 
+struct DisplayInfo {
+    char display_name[0x40]{"Default"};
+    u64 unknown_1{1};
+    u64 unknown_2{1};
+    u64 width{1920};
+    u64 height{1080};
+};
+static_assert(sizeof(DisplayInfo) == 0x60, "DisplayInfo has wrong size");
+
 class Parcel {
 public:
     // This default size was chosen arbitrarily.
@@ -29,6 +39,7 @@ public:
 
     template <typename T>
     T Read() {
+        ASSERT(read_index + sizeof(T) <= buffer.size());
         T val;
         std::memcpy(&val, buffer.data() + read_index, sizeof(T));
         read_index += sizeof(T);
@@ -38,6 +49,7 @@ public:
 
     template <typename T>
     T ReadUnaligned() {
+        ASSERT(read_index + sizeof(T) <= buffer.size());
         T val;
         std::memcpy(&val, buffer.data() + read_index, sizeof(T));
         read_index += sizeof(T);
@@ -45,6 +57,7 @@ public:
     }
 
     std::vector<u8> ReadBlock(size_t length) {
+        ASSERT(read_index + length <= buffer.size());
         const u8* const begin = buffer.data() + read_index;
         const u8* const end = begin + length;
         std::vector<u8> data(begin, end);
@@ -77,7 +90,18 @@ public:
         write_index = Common::AlignUp(write_index, 4);
     }
 
+    template <typename T>
+    void WriteObject(const T& val) {
+        u32_le size = static_cast<u32>(sizeof(val));
+        Write(size);
+        // TODO(Subv): Support file descriptors.
+        Write<u32_le>(0); // Fd count.
+        Write(val);
+    }
+
     void Deserialize() {
+        ASSERT(buffer.size() > sizeof(Header));
+
         Header header{};
         std::memcpy(&header, buffer.data(), sizeof(Header));
 
@@ -92,8 +116,10 @@ public:
         SerializeData();
 
         Header header{};
-        header.data_offset = sizeof(Header);
         header.data_size = static_cast<u32_le>(write_index - sizeof(Header));
+        header.data_offset = sizeof(Header);
+        header.objects_size = 4;
+        header.objects_offset = sizeof(Header) + header.data_size;
         std::memcpy(buffer.data(), &header, sizeof(Header));
 
         return buffer;
@@ -133,11 +159,11 @@ protected:
 private:
     struct Data {
         u32_le magic = 2;
-        u32_le process_id;
+        u32_le process_id = 1;
         u32_le id;
-        INSERT_PADDING_BYTES(0xC);
+        INSERT_PADDING_WORDS(3);
         std::array<u8, 8> dispdrv = {'d', 'i', 's', 'p', 'd', 'r', 'v', '\0'};
-        INSERT_PADDING_BYTES(8);
+        INSERT_PADDING_WORDS(2);
     };
     static_assert(sizeof(Data) == 0x28, "ParcelData has wrong size");
 
@@ -202,7 +228,6 @@ public:
     void DeserializeData() override {
         std::u16string token = ReadInterfaceToken();
         data = Read<Data>();
-        ASSERT(data.graphic_buffer_length == sizeof(NVFlinger::IGBPBuffer));
         buffer = Read<NVFlinger::IGBPBuffer>();
     }
 
@@ -252,6 +277,12 @@ public:
     Data data;
 };
 
+struct BufferProducerFence {
+    u32 is_valid;
+    std::array<Nvidia::IoctlFence, 4> fences;
+};
+static_assert(sizeof(BufferProducerFence) == 36, "BufferProducerFence has wrong size");
+
 class IGBPDequeueBufferResponseParcel : public Parcel {
 public:
     explicit IGBPDequeueBufferResponseParcel(u32 slot) : Parcel(), slot(slot) {}
@@ -259,10 +290,15 @@ public:
 
 protected:
     void SerializeData() override {
-        Write(slot);
         // TODO(Subv): Find out how this Fence is used.
-        std::array<u32_le, 11> fence = {};
-        Write(fence);
+        BufferProducerFence fence = {};
+        fence.is_valid = 1;
+        for (auto& fence_ : fence.fences)
+            fence_.id = -1;
+
+        Write(slot);
+        Write<u32_le>(1);
+        WriteObject(fence);
         Write<u32_le>(0);
     }
 
@@ -292,14 +328,10 @@ public:
 
 protected:
     void SerializeData() override {
-        // TODO(Subv): Find out what this all means
+        // TODO(Subv): Figure out what this value means, writing non-zero here will make libnx try
+        // to read an IGBPBuffer object from the parcel.
         Write<u32_le>(1);
-
-        Write<u32_le>(sizeof(NVFlinger::IGBPBuffer));
-        Write<u32_le>(0); // Unknown
-
-        Write(buffer);
-
+        WriteObject(buffer);
         Write<u32_le>(0);
     }
 
@@ -318,13 +350,29 @@ public:
         data = Read<Data>();
     }
 
+    struct Fence {
+        u32_le id;
+        u32_le value;
+    };
+    static_assert(sizeof(Fence) == 8, "Fence has wrong size");
+
     struct Data {
         u32_le slot;
-        INSERT_PADDING_WORDS(2);
+        INSERT_PADDING_WORDS(3);
         u32_le timestamp;
-        INSERT_PADDING_WORDS(20);
+        s32_le is_auto_timestamp;
+        s32_le crop_left;
+        s32_le crop_top;
+        s32_le crop_right;
+        s32_le crop_bottom;
+        s32_le scaling_mode;
+        NVFlinger::BufferQueue::BufferTransformFlags transform;
+        u32_le sticky_transform;
+        INSERT_PADDING_WORDS(2);
+        u32_le fence_is_valid;
+        std::array<Fence, 2> fences;
     };
-    static_assert(sizeof(Data) == 96, "ParcelData has wrong size");
+    static_assert(sizeof(Data) == 80, "ParcelData has wrong size");
 
     Data data;
 };
@@ -392,7 +440,7 @@ public:
             {0, &IHOSBinderDriver::TransactParcel, "TransactParcel"},
             {1, &IHOSBinderDriver::AdjustRefcount, "AdjustRefcount"},
             {2, &IHOSBinderDriver::GetNativeHandle, "GetNativeHandle"},
-            {3, nullptr, "TransactParcelAuto"},
+            {3, &IHOSBinderDriver::TransactParcel, "TransactParcelAuto"},
         };
         RegisterHandlers(functions);
     }
@@ -421,68 +469,53 @@ private:
         u32 id = rp.Pop<u32>();
         auto transaction = static_cast<TransactionId>(rp.Pop<u32>());
         u32 flags = rp.Pop<u32>();
-
-        auto& input_buffer = ctx.BufferDescriptorA()[0];
-        std::vector<u8> input_data(input_buffer.Size());
-        Memory::ReadBlock(input_buffer.Address(), input_data.data(), input_buffer.Size());
-
-        auto& output_buffer = ctx.BufferDescriptorB()[0];
-
         auto buffer_queue = nv_flinger->GetBufferQueue(id);
-        LOG_WARNING(Service_VI, "(STUBBED) called, transaction=%x", transaction);
+
+        LOG_DEBUG(Service_VI, "called, transaction=%x", transaction);
+
         if (transaction == TransactionId::Connect) {
-            IGBPConnectRequestParcel request{input_data};
+            IGBPConnectRequestParcel request{ctx.ReadBuffer()};
             IGBPConnectResponseParcel response{1280, 720};
-            auto response_buffer = response.Serialize();
-            Memory::WriteBlock(output_buffer.Address(), response_buffer.data(),
-                               output_buffer.Size());
+            ctx.WriteBuffer(response.Serialize());
         } else if (transaction == TransactionId::SetPreallocatedBuffer) {
-            IGBPSetPreallocatedBufferRequestParcel request{input_data};
+            IGBPSetPreallocatedBufferRequestParcel request{ctx.ReadBuffer()};
 
             buffer_queue->SetPreallocatedBuffer(request.data.slot, request.buffer);
 
             IGBPSetPreallocatedBufferResponseParcel response{};
-            auto response_buffer = response.Serialize();
-            Memory::WriteBlock(output_buffer.Address(), response_buffer.data(),
-                               output_buffer.Size());
+            ctx.WriteBuffer(response.Serialize());
         } else if (transaction == TransactionId::DequeueBuffer) {
-            IGBPDequeueBufferRequestParcel request{input_data};
+            IGBPDequeueBufferRequestParcel request{ctx.ReadBuffer()};
 
             u32 slot = buffer_queue->DequeueBuffer(request.data.pixel_format, request.data.width,
                                                    request.data.height);
 
             IGBPDequeueBufferResponseParcel response{slot};
-            auto response_buffer = response.Serialize();
-            Memory::WriteBlock(output_buffer.Address(), response_buffer.data(),
-                               output_buffer.Size());
+            ctx.WriteBuffer(response.Serialize());
         } else if (transaction == TransactionId::RequestBuffer) {
-            IGBPRequestBufferRequestParcel request{input_data};
+            IGBPRequestBufferRequestParcel request{ctx.ReadBuffer()};
 
             auto& buffer = buffer_queue->RequestBuffer(request.slot);
 
             IGBPRequestBufferResponseParcel response{buffer};
-            auto response_buffer = response.Serialize();
-            Memory::WriteBlock(output_buffer.Address(), response_buffer.data(),
-                               output_buffer.Size());
+            ctx.WriteBuffer(response.Serialize());
         } else if (transaction == TransactionId::QueueBuffer) {
-            IGBPQueueBufferRequestParcel request{input_data};
+            IGBPQueueBufferRequestParcel request{ctx.ReadBuffer()};
 
-            buffer_queue->QueueBuffer(request.data.slot);
+            buffer_queue->QueueBuffer(request.data.slot, request.data.transform);
 
             IGBPQueueBufferResponseParcel response{1280, 720};
-            auto response_buffer = response.Serialize();
-            Memory::WriteBlock(output_buffer.Address(), response_buffer.data(),
-                               output_buffer.Size());
+            ctx.WriteBuffer(response.Serialize());
         } else if (transaction == TransactionId::Query) {
-            IGBPQueryRequestParcel request{input_data};
+            IGBPQueryRequestParcel request{ctx.ReadBuffer()};
 
             u32 value =
                 buffer_queue->Query(static_cast<NVFlinger::BufferQueue::QueryType>(request.type));
 
             IGBPQueryResponseParcel response{value};
-            auto response_buffer = response.Serialize();
-            Memory::WriteBlock(output_buffer.Address(), response_buffer.data(),
-                               output_buffer.Size());
+            ctx.WriteBuffer(response.Serialize());
+        } else if (transaction == TransactionId::CancelBuffer) {
+            LOG_WARNING(Service_VI, "(STUBBED) called, transaction=CancelBuffer");
         } else {
             ASSERT_MSG(false, "Unimplemented");
         }
@@ -518,7 +551,7 @@ private:
     }
 
     std::shared_ptr<NVFlinger::NVFlinger> nv_flinger;
-};
+}; // namespace VI
 
 class ISystemDisplayService final : public ServiceFramework<ISystemDisplayService> {
 public:
@@ -648,12 +681,12 @@ void IApplicationDisplayService::CloseDisplay(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
     u64 display_id = rp.Pop<u64>();
 
-    IPC::ResponseBuilder rb = rp.MakeBuilder(4, 0, 0);
+    IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
     rb.Push(RESULT_SUCCESS);
 }
 
 void IApplicationDisplayService::OpenLayer(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_VI, "(STUBBED) called");
+    LOG_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');
@@ -663,43 +696,33 @@ void IApplicationDisplayService::OpenLayer(Kernel::HLERequestContext& ctx) {
     u64 layer_id = rp.Pop<u64>();
     u64 aruid = rp.Pop<u64>();
 
-    auto& buffer = ctx.BufferDescriptorB()[0];
-
     u64 display_id = nv_flinger->OpenDisplay(display_name);
     u32 buffer_queue_id = nv_flinger->GetBufferQueueId(display_id, layer_id);
 
     NativeWindow native_window{buffer_queue_id};
-    auto data = native_window.Serialize();
-    Memory::WriteBlock(buffer.Address(), data.data(), data.size());
-
     IPC::ResponseBuilder rb = rp.MakeBuilder(4, 0, 0);
     rb.Push(RESULT_SUCCESS);
-    rb.Push<u64>(data.size());
+    rb.Push<u64>(ctx.WriteBuffer(native_window.Serialize()));
 }
 
 void IApplicationDisplayService::CreateStrayLayer(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service, "(STUBBED) called");
+    LOG_DEBUG(Service_VI, "called");
 
     IPC::RequestParser rp{ctx};
     u32 flags = rp.Pop<u32>();
     rp.Pop<u32>(); // padding
     u64 display_id = rp.Pop<u64>();
 
-    auto& buffer = ctx.BufferDescriptorB()[0];
-
     // TODO(Subv): What's the difference between a Stray and a Managed layer?
 
     u64 layer_id = nv_flinger->CreateLayer(display_id);
     u32 buffer_queue_id = nv_flinger->GetBufferQueueId(display_id, layer_id);
 
     NativeWindow native_window{buffer_queue_id};
-    auto data = native_window.Serialize();
-    Memory::WriteBlock(buffer.Address(), data.data(), data.size());
-
     IPC::ResponseBuilder rb = rp.MakeBuilder(6, 0, 0);
     rb.Push(RESULT_SUCCESS);
     rb.Push(layer_id);
-    rb.Push<u64>(data.size());
+    rb.Push<u64>(ctx.WriteBuffer(native_window.Serialize()));
 }
 
 void IApplicationDisplayService::DestroyStrayLayer(Kernel::HLERequestContext& ctx) {
@@ -722,6 +745,16 @@ void IApplicationDisplayService::SetLayerScalingMode(Kernel::HLERequestContext&
     rb.Push(RESULT_SUCCESS);
 }
 
+void IApplicationDisplayService::ListDisplays(Kernel::HLERequestContext& ctx) {
+    IPC::RequestParser rp{ctx};
+    DisplayInfo display_info;
+    ctx.WriteBuffer(&display_info, sizeof(DisplayInfo));
+    IPC::ResponseBuilder rb = rp.MakeBuilder(4, 0, 0);
+    rb.Push(RESULT_SUCCESS);
+    rb.Push<u64>(1);
+    LOG_WARNING(Service_VI, "(STUBBED) called");
+}
+
 void IApplicationDisplayService::GetDisplayVsyncEvent(Kernel::HLERequestContext& ctx) {
     LOG_WARNING(Service_VI, "(STUBBED) called");
     IPC::RequestParser rp{ctx};
@@ -743,7 +776,7 @@ IApplicationDisplayService::IApplicationDisplayService(
         {102, &IApplicationDisplayService::GetManagerDisplayService, "GetManagerDisplayService"},
         {103, &IApplicationDisplayService::GetIndirectDisplayTransactionService,
          "GetIndirectDisplayTransactionService"},
-        {1000, nullptr, "ListDisplays"},
+        {1000, &IApplicationDisplayService::ListDisplays, "ListDisplays"},
         {1010, &IApplicationDisplayService::OpenDisplay, "OpenDisplay"},
         {1020, &IApplicationDisplayService::CloseDisplay, "CloseDisplay"},
         {2101, &IApplicationDisplayService::SetLayerScalingMode, "SetLayerScalingMode"},

src/core/hle/service/vi/vi.h

@@ -30,6 +30,7 @@ private:
     void OpenDisplay(Kernel::HLERequestContext& ctx);
     void CloseDisplay(Kernel::HLERequestContext& ctx);
     void SetLayerScalingMode(Kernel::HLERequestContext& ctx);
+    void ListDisplays(Kernel::HLERequestContext& ctx);
     void OpenLayer(Kernel::HLERequestContext& ctx);
     void CreateStrayLayer(Kernel::HLERequestContext& ctx);
     void DestroyStrayLayer(Kernel::HLERequestContext& ctx);

src/core/loader/deconstructed_rom_directory.cpp

@@ -2,6 +2,7 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <cinttypes>
 #include "common/common_funcs.h"
 #include "common/common_paths.h"
 #include "common/file_util.h"
@@ -116,7 +117,7 @@ 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%llx", module, load_addr);
+            LOG_DEBUG(Loader, "loaded module %s @ 0x%" PRIx64, module, load_addr);
         } else {
             next_load_addr = load_addr;
         }
@@ -158,8 +159,8 @@ ResultStatus AppLoader_DeconstructedRomDirectory::ReadRomFS(
     offset = 0;
     size = romfs_file->GetSize();
 
-    LOG_DEBUG(Loader, "RomFS offset:           0x%08X", offset);
-    LOG_DEBUG(Loader, "RomFS size:             0x%08X", size);
+    LOG_DEBUG(Loader, "RomFS offset:           0x%016" PRIX64, offset);
+    LOG_DEBUG(Loader, "RomFS size:             0x%016" PRIX64, size);
 
     // Reset read pointer
     file.Seek(0, SEEK_SET);

src/core/loader/nso.cpp

@@ -2,6 +2,7 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <cinttypes>
 #include <vector>
 #include <lz4.h>
 #include "common/common_funcs.h"
@@ -82,7 +83,7 @@ static std::vector<u8> ReadSegment(FileUtil::IOFile& file, const NsoSegmentHeade
         reinterpret_cast<char*>(uncompressed_data.data()), compressed_size, header.size);
 
     ASSERT_MSG(bytes_uncompressed == header.size && bytes_uncompressed == uncompressed_data.size(),
-               "%d != %d != %d", bytes_uncompressed, header.size, uncompressed_data.size());
+               "%d != %u != %zu", bytes_uncompressed, header.size, uncompressed_data.size());
 
     return uncompressed_data;
 }
@@ -158,7 +159,8 @@ ResultStatus AppLoader_NSO::Load(Kernel::SharedPtr<Kernel::Process>& process) {
 
     // Load module
     LoadModule(filepath, Memory::PROCESS_IMAGE_VADDR);
-    LOG_DEBUG(Loader, "loaded module %s @ 0x%llx", filepath.c_str(), Memory::PROCESS_IMAGE_VADDR);
+    LOG_DEBUG(Loader, "loaded module %s @ 0x%" PRIx64, filepath.c_str(),
+              Memory::PROCESS_IMAGE_VADDR);
 
     process->svc_access_mask.set();
     process->address_mappings = default_address_mappings;

src/core/memory.cpp

@@ -4,6 +4,7 @@
 
 #include <algorithm>
 #include <array>
+#include <cinttypes>
 #include <cstring>
 #include <boost/optional.hpp>
 #include "common/assert.h"
@@ -38,12 +39,12 @@ PageTable* GetCurrentPageTable() {
 }
 
 static void MapPages(PageTable& page_table, VAddr base, u64 size, u8* memory, PageType type) {
-    LOG_DEBUG(HW_Memory, "Mapping %p onto %08X-%08X", memory, base * PAGE_SIZE,
+    LOG_DEBUG(HW_Memory, "Mapping %p onto %016" PRIX64 "-%016" PRIX64, memory, base * PAGE_SIZE,
               (base + size) * PAGE_SIZE);
 
     VAddr end = base + size;
     while (base != end) {
-        ASSERT_MSG(base < PAGE_TABLE_NUM_ENTRIES, "out of range mapping at %08X", base);
+        ASSERT_MSG(base < PAGE_TABLE_NUM_ENTRIES, "out of range mapping at %016" PRIX64, base);
 
         page_table.attributes[base] = type;
         page_table.pointers[base] = memory;
@@ -55,14 +56,14 @@ static void MapPages(PageTable& page_table, VAddr base, u64 size, u8* memory, Pa
 }
 
 void MapMemoryRegion(PageTable& page_table, VAddr base, u64 size, u8* target) {
-    ASSERT_MSG((size & PAGE_MASK) == 0, "non-page aligned size: %08X", size);
-    ASSERT_MSG((base & PAGE_MASK) == 0, "non-page aligned base: %08X", base);
+    ASSERT_MSG((size & PAGE_MASK) == 0, "non-page aligned size: %016" PRIX64, size);
+    ASSERT_MSG((base & PAGE_MASK) == 0, "non-page aligned base: %016" PRIX64, base);
     MapPages(page_table, base / PAGE_SIZE, size / PAGE_SIZE, target, PageType::Memory);
 }
 
 void MapIoRegion(PageTable& page_table, VAddr base, u64 size, MemoryHookPointer mmio_handler) {
-    ASSERT_MSG((size & PAGE_MASK) == 0, "non-page aligned size: %08X", size);
-    ASSERT_MSG((base & PAGE_MASK) == 0, "non-page aligned base: %08X", base);
+    ASSERT_MSG((size & PAGE_MASK) == 0, "non-page aligned size: %016" PRIX64, size);
+    ASSERT_MSG((base & PAGE_MASK) == 0, "non-page aligned base: %016" PRIX64, base);
     MapPages(page_table, base / PAGE_SIZE, size / PAGE_SIZE, nullptr, PageType::Special);
 
     auto interval = boost::icl::discrete_interval<VAddr>::closed(base, base + size - 1);
@@ -71,8 +72,8 @@ void MapIoRegion(PageTable& page_table, VAddr base, u64 size, MemoryHookPointer
 }
 
 void UnmapRegion(PageTable& page_table, VAddr base, u64 size) {
-    ASSERT_MSG((size & PAGE_MASK) == 0, "non-page aligned size: %08X", size);
-    ASSERT_MSG((base & PAGE_MASK) == 0, "non-page aligned base: %08X", base);
+    ASSERT_MSG((size & PAGE_MASK) == 0, "non-page aligned size: %016" PRIX64, size);
+    ASSERT_MSG((base & PAGE_MASK) == 0, "non-page aligned base: %016" PRIX64, base);
     MapPages(page_table, base / PAGE_SIZE, size / PAGE_SIZE, nullptr, PageType::Unmapped);
 
     auto interval = boost::icl::discrete_interval<VAddr>::closed(base, base + size - 1);
@@ -120,7 +121,7 @@ T Read(const VAddr vaddr) {
     const PageType type = current_page_table->attributes[vaddr >> PAGE_BITS];
     switch (type) {
     case PageType::Unmapped:
-        LOG_ERROR(HW_Memory, "unmapped Read%lu @ 0x%016llX", sizeof(T) * 8, vaddr);
+        LOG_ERROR(HW_Memory, "unmapped Read%zu @ 0x%016" PRIX64, sizeof(T) * 8, vaddr);
         return 0;
     case PageType::Special: {
         if (auto result = ReadSpecial<T>(vaddr))
@@ -129,7 +130,7 @@ T Read(const VAddr vaddr) {
     }
     case PageType::Memory: {
         const u8* page_pointer = current_page_table->pointers[vaddr >> PAGE_BITS];
-        ASSERT_MSG(page_pointer, "Mapped memory page without a pointer @ %08X", vaddr);
+        ASSERT_MSG(page_pointer, "Mapped memory page without a pointer @ %016" PRIX64, vaddr);
 
         T value;
         std::memcpy(&value, &page_pointer[vaddr & PAGE_MASK], sizeof(T));
@@ -148,8 +149,8 @@ void Write(const VAddr vaddr, const T data) {
     const PageType type = current_page_table->attributes[vaddr >> PAGE_BITS];
     switch (type) {
     case PageType::Unmapped:
-        LOG_ERROR(HW_Memory, "unmapped Write%lu 0x%08X @ 0x%08X", sizeof(data) * 8, (u32)data,
-                  vaddr);
+        LOG_ERROR(HW_Memory, "unmapped Write%zu 0x%08X @ 0x%016" PRIX64, sizeof(data) * 8,
+                  static_cast<u32>(data), vaddr);
         return;
     case PageType::Special: {
         if (WriteSpecial<T>(vaddr, data))
@@ -158,7 +159,7 @@ void Write(const VAddr vaddr, const T data) {
     }
     case PageType::Memory: {
         u8* page_pointer = current_page_table->pointers[vaddr >> PAGE_BITS];
-        ASSERT_MSG(page_pointer, "Mapped memory page without a pointer @ %08X", vaddr);
+        ASSERT_MSG(page_pointer, "Mapped memory page without a pointer @ %016" PRIX64, vaddr);
         std::memcpy(&page_pointer[vaddr & PAGE_MASK], &data, sizeof(T));
         return;
     }
@@ -203,7 +204,7 @@ u8* GetPointer(const VAddr vaddr) {
         return page_pointer + (vaddr & PAGE_MASK);
     }
 
-    LOG_ERROR(HW_Memory, "unknown GetPointer @ 0x%08x", vaddr);
+    LOG_ERROR(HW_Memory, "unknown GetPointer @ 0x%016" PRIx64, vaddr);
     return nullptr;
 }
 
@@ -241,12 +242,13 @@ u8* GetPhysicalPointer(PAddr address) {
         });
 
     if (area == std::end(memory_areas)) {
-        LOG_ERROR(HW_Memory, "unknown GetPhysicalPointer @ 0x%08X", address);
+        LOG_ERROR(HW_Memory, "unknown GetPhysicalPointer @ 0x%016" PRIX64, address);
         return nullptr;
     }
 
     if (area->paddr_base == IO_AREA_PADDR) {
-        LOG_ERROR(HW_Memory, "MMIO mappings are not supported yet. phys_addr=0x%08X", address);
+        LOG_ERROR(HW_Memory, "MMIO mappings are not supported yet. phys_addr=0x%016" PRIX64,
+                  address);
         return nullptr;
     }
 
@@ -321,7 +323,9 @@ void ReadBlock(const Kernel::Process& process, const VAddr src_addr, void* dest_
 
         switch (page_table.attributes[page_index]) {
         case PageType::Unmapped:
-            LOG_ERROR(HW_Memory, "unmapped ReadBlock @ 0x%08X (start address = 0xllx, size = %zu)",
+            LOG_ERROR(HW_Memory,
+                      "unmapped ReadBlock @ 0x%016" PRIX64 " (start address = 0x%" PRIx64
+                      ", size = %zu)",
                       current_vaddr, src_addr, size);
             std::memset(dest_buffer, 0, copy_amount);
             break;
@@ -393,7 +397,8 @@ void WriteBlock(const Kernel::Process& process, const VAddr dest_addr, const voi
         switch (page_table.attributes[page_index]) {
         case PageType::Unmapped:
             LOG_ERROR(HW_Memory,
-                      "unmapped WriteBlock @ 0x%08X (start address = 0x%08X, size = %zu)",
+                      "unmapped WriteBlock @ 0x%016" PRIX64 " (start address = 0x%016" PRIX64
+                      ", size = %zu)",
                       current_vaddr, dest_addr, size);
             break;
         case PageType::Special:
@@ -437,7 +442,9 @@ void ZeroBlock(const VAddr dest_addr, const size_t size) {
 
         switch (current_page_table->attributes[page_index]) {
         case PageType::Unmapped:
-            LOG_ERROR(HW_Memory, "unmapped ZeroBlock @ 0x%08X (start address = 0x%08X, size = %zu)",
+            LOG_ERROR(HW_Memory,
+                      "unmapped ZeroBlock @ 0x%016" PRIX64 " (start address = 0x%016" PRIX64
+                      ", size = %zu)",
                       current_vaddr, dest_addr, size);
             break;
         case PageType::Special:
@@ -474,7 +481,9 @@ void CopyBlock(VAddr dest_addr, VAddr src_addr, const size_t size) {
 
         switch (current_page_table->attributes[page_index]) {
         case PageType::Unmapped:
-            LOG_ERROR(HW_Memory, "unmapped CopyBlock @ 0x%08X (start address = 0x%08X, size = %zu)",
+            LOG_ERROR(HW_Memory,
+                      "unmapped CopyBlock @ 0x%016" PRIX64 " (start address = 0x%016" PRIX64
+                      ", size = %zu)",
                       current_vaddr, src_addr, size);
             ZeroBlock(dest_addr, copy_amount);
             break;
@@ -599,7 +608,7 @@ boost::optional<PAddr> TryVirtualToPhysicalAddress(const VAddr addr) {
 PAddr VirtualToPhysicalAddress(const VAddr addr) {
     auto paddr = TryVirtualToPhysicalAddress(addr);
     if (!paddr) {
-        LOG_ERROR(HW_Memory, "Unknown virtual address @ 0x%08X", addr);
+        LOG_ERROR(HW_Memory, "Unknown virtual address @ 0x%016" PRIX64, addr);
         // To help with debugging, set bit on address so that it's obviously invalid.
         return addr | 0x80000000;
     }

src/core/memory.h

@@ -25,10 +25,11 @@ namespace Memory {
  * Page size used by the ARM architecture. This is the smallest granularity with which memory can
  * be mapped.
  */
-const int PAGE_BITS = 12;
-const u64 PAGE_SIZE = 1 << PAGE_BITS;
-const u64 PAGE_MASK = PAGE_SIZE - 1;
-const size_t PAGE_TABLE_NUM_ENTRIES = 1ULL << (36 - PAGE_BITS);
+constexpr size_t PAGE_BITS = 12;
+constexpr u64 PAGE_SIZE = 1 << PAGE_BITS;
+constexpr u64 PAGE_MASK = PAGE_SIZE - 1;
+constexpr size_t ADDRESS_SPACE_BITS = 36;
+constexpr size_t PAGE_TABLE_NUM_ENTRIES = 1ULL << (ADDRESS_SPACE_BITS - PAGE_BITS);
 
 enum class PageType : u8 {
     /// Page is unmapped and should cause an access error.

src/video_core/CMakeLists.txt

@@ -1,4 +1,15 @@
 add_library(video_core STATIC
+    command_processor.cpp
+    command_processor.h
+    engines/fermi_2d.cpp
+    engines/fermi_2d.h
+    engines/maxwell_3d.cpp
+    engines/maxwell_3d.h
+    engines/maxwell_compute.cpp
+    engines/maxwell_compute.h
+    gpu.h
+    memory_manager.cpp
+    memory_manager.h
     renderer_base.cpp
     renderer_base.h
     renderer_opengl/gl_resource_manager.h

src/video_core/command_processor.cpp

@@ -0,0 +1,119 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <array>
+#include <cstddef>
+#include <memory>
+#include <utility>
+#include "common/assert.h"
+#include "common/logging/log.h"
+#include "common/microprofile.h"
+#include "common/vector_math.h"
+#include "core/memory.h"
+#include "core/tracer/recorder.h"
+#include "video_core/command_processor.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/gpu.h"
+#include "video_core/renderer_base.h"
+#include "video_core/video_core.h"
+
+namespace Tegra {
+
+enum class BufferMethods {
+    BindObject = 0,
+    CountBufferMethods = 0x100,
+};
+
+void GPU::WriteReg(u32 method, u32 subchannel, u32 value) {
+    LOG_WARNING(HW_GPU, "Processing method %08X on subchannel %u value %08X", method, subchannel,
+                value);
+
+    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);
+        ASSERT(bound_engines.find(subchannel) == bound_engines.end());
+        bound_engines[subchannel] = static_cast<EngineID>(value);
+        return;
+    }
+
+    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");
+        return;
+    }
+
+    ASSERT(bound_engines.find(subchannel) != bound_engines.end());
+
+    const EngineID engine = bound_engines[subchannel];
+
+    switch (engine) {
+    case EngineID::FERMI_TWOD_A:
+        fermi_2d->WriteReg(method, value);
+        break;
+    case EngineID::MAXWELL_B:
+        maxwell_3d->WriteReg(method, value);
+        break;
+    case EngineID::MAXWELL_COMPUTE_B:
+        maxwell_compute->WriteReg(method, value);
+        break;
+    default:
+        UNIMPLEMENTED();
+    }
+}
+
+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 CommandHeader header = {Memory::Read32(current_addr)};
+        current_addr += sizeof(u32);
+
+        switch (header.mode.Value()) {
+        case SubmissionMode::IncreasingOld:
+        case SubmissionMode::Increasing: {
+            // Increase the method value with each argument.
+            for (unsigned i = 0; i < header.arg_count; ++i) {
+                WriteReg(header.method + i, header.subchannel, Memory::Read32(current_addr));
+                current_addr += sizeof(u32);
+            }
+            break;
+        }
+        case SubmissionMode::NonIncreasingOld:
+        case SubmissionMode::NonIncreasing: {
+            // Use the same method value for all arguments.
+            for (unsigned i = 0; i < header.arg_count; ++i) {
+                WriteReg(header.method, header.subchannel, Memory::Read32(current_addr));
+                current_addr += sizeof(u32);
+            }
+            break;
+        }
+        case SubmissionMode::IncreaseOnce: {
+            ASSERT(header.arg_count.Value() >= 1);
+            // Use the original method for the first argument and then the next method for all other
+            // arguments.
+            WriteReg(header.method, header.subchannel, Memory::Read32(current_addr));
+            current_addr += sizeof(u32);
+            // Use the same method value for all arguments.
+            for (unsigned i = 1; i < header.arg_count; ++i) {
+                WriteReg(header.method + 1, header.subchannel, Memory::Read32(current_addr));
+                current_addr += sizeof(u32);
+            }
+            break;
+        }
+        case SubmissionMode::Inline: {
+            // The register value is stored in the bits 16-28 as an immediate
+            WriteReg(header.method, header.subchannel, header.inline_data);
+            break;
+        }
+        default:
+            UNIMPLEMENTED();
+        }
+    }
+}
+
+} // namespace Tegra

src/video_core/command_processor.h

@@ -0,0 +1,39 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <type_traits>
+#include "common/bit_field.h"
+#include "common/common_types.h"
+
+namespace Tegra {
+
+enum class SubmissionMode : u32 {
+    IncreasingOld = 0,
+    Increasing = 1,
+    NonIncreasingOld = 2,
+    NonIncreasing = 3,
+    Inline = 4,
+    IncreaseOnce = 5
+};
+
+union CommandHeader {
+    u32 hex;
+
+    BitField<0, 13, u32> method;
+    BitField<13, 3, u32> subchannel;
+
+    BitField<16, 13, u32> arg_count;
+    BitField<16, 13, u32> inline_data;
+
+    BitField<29, 3, SubmissionMode> mode;
+};
+static_assert(std::is_standard_layout<CommandHeader>::value == true,
+              "CommandHeader does not use standard layout");
+static_assert(sizeof(CommandHeader) == sizeof(u32), "CommandHeader has incorrect size!");
+
+void ProcessCommandList(VAddr address, u32 size);
+
+} // namespace Tegra

src/video_core/engines/fermi_2d.cpp

@@ -0,0 +1,13 @@
+// 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"
+
+namespace Tegra {
+namespace Engines {
+
+void Fermi2D::WriteReg(u32 method, u32 value) {}
+
+} // namespace Engines
+} // namespace Tegra

src/video_core/engines/fermi_2d.h

@@ -0,0 +1,22 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "common/common_types.h"
+
+namespace Tegra {
+namespace Engines {
+
+class Fermi2D final {
+public:
+    Fermi2D() = default;
+    ~Fermi2D() = default;
+
+    /// Write the value to the register identified by method.
+    void WriteReg(u32 method, u32 value);
+};
+
+} // namespace Engines
+} // namespace Tegra

src/video_core/engines/maxwell_3d.cpp

@@ -0,0 +1,51 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "common/assert.h"
+#include "video_core/engines/maxwell_3d.h"
+
+namespace Tegra {
+namespace Engines {
+
+Maxwell3D::Maxwell3D(MemoryManager& memory_manager) : memory_manager(memory_manager) {}
+
+void Maxwell3D::WriteReg(u32 method, u32 value) {
+    ASSERT_MSG(method < Regs::NUM_REGS,
+               "Invalid Maxwell3D register, increase the size of the Regs structure");
+
+    regs.reg_array[method] = value;
+
+#define MAXWELL3D_REG_INDEX(field_name) (offsetof(Regs, field_name) / sizeof(u32))
+
+    switch (method) {
+    case MAXWELL3D_REG_INDEX(query.query_get): {
+        ProcessQueryGet();
+        break;
+    }
+    default:
+        break;
+    }
+
+#undef MAXWELL3D_REG_INDEX
+}
+
+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);
+
+    switch (regs.query.query_get.mode) {
+    case Regs::QueryMode::Write: {
+        // Write the current query sequence to the sequence address.
+        u32 sequence = regs.query.query_sequence;
+        Memory::Write32(address, sequence);
+        break;
+    }
+    default:
+        UNIMPLEMENTED_MSG("Query mode %u not implemented", regs.query.query_get.mode.Value());
+    }
+}
+} // namespace Engines
+} // namespace Tegra

src/video_core/engines/maxwell_3d.h

@@ -0,0 +1,76 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "common/bit_field.h"
+#include "common/common_funcs.h"
+#include "common/common_types.h"
+#include "video_core/memory_manager.h"
+
+namespace Tegra {
+namespace Engines {
+
+class Maxwell3D final {
+public:
+    explicit Maxwell3D(MemoryManager& memory_manager);
+    ~Maxwell3D() = default;
+
+    /// Write the value to the register identified by method.
+    void WriteReg(u32 method, u32 value);
+
+    /// 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;
+
+        enum class QueryMode : u32 {
+            Write = 0,
+            Sync = 1,
+        };
+
+        union {
+            struct {
+                INSERT_PADDING_WORDS(0x6C0);
+                struct {
+                    u32 query_address_high;
+                    u32 query_address_low;
+                    u32 query_sequence;
+                    union {
+                        u32 raw;
+                        BitField<0, 2, QueryMode> mode;
+                        BitField<4, 1, u32> fence;
+                        BitField<12, 4, u32> unit;
+                    } query_get;
+
+                    GPUVAddr QueryAddress() const {
+                        return static_cast<GPUVAddr>(
+                            (static_cast<GPUVAddr>(query_address_high) << 32) | query_address_low);
+                    }
+                } query;
+                INSERT_PADDING_WORDS(0x772);
+            };
+            std::array<u32, NUM_REGS> reg_array;
+        };
+    } regs{};
+
+    static_assert(sizeof(Regs) == Regs::NUM_REGS * sizeof(u32), "Maxwell3D Regs has wrong size");
+
+private:
+    /// Handles a write to the QUERY_GET register.
+    void ProcessQueryGet();
+
+    MemoryManager& memory_manager;
+};
+
+#define ASSERT_REG_POSITION(field_name, position)                                                  \
+    static_assert(offsetof(Maxwell3D::Regs, field_name) == position * 4,                           \
+                  "Field " #field_name " has invalid position")
+
+ASSERT_REG_POSITION(query, 0x6C0);
+
+#undef ASSERT_REG_POSITION
+
+} // namespace Engines
+} // namespace Tegra

src/video_core/engines/maxwell_compute.cpp

@@ -0,0 +1,13 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "video_core/engines/maxwell_compute.h"
+
+namespace Tegra {
+namespace Engines {
+
+void MaxwellCompute::WriteReg(u32 method, u32 value) {}
+
+} // namespace Engines
+} // namespace Tegra

src/video_core/engines/maxwell_compute.h

@@ -0,0 +1,22 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "common/common_types.h"
+
+namespace Tegra {
+namespace Engines {
+
+class MaxwellCompute final {
+public:
+    MaxwellCompute() = default;
+    ~MaxwellCompute() = default;
+
+    /// Write the value to the register identified by method.
+    void WriteReg(u32 method, u32 value);
+};
+
+} // namespace Engines
+} // namespace Tegra

src/video_core/gpu.h

@@ -0,0 +1,55 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <memory>
+#include <unordered_map>
+#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 {
+
+enum class EngineID {
+    FERMI_TWOD_A = 0x902D, // 2D Engine
+    MAXWELL_B = 0xB197,    // 3D Engine
+    MAXWELL_COMPUTE_B = 0xB1C0,
+    KEPLER_INLINE_TO_MEMORY_B = 0xA140,
+    MAXWELL_DMA_COPY_A = 0xB0B5,
+};
+
+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;
+
+    /// Processes a command list stored at the specified address in GPU memory.
+    void ProcessCommandList(GPUVAddr address, u32 size);
+
+    std::unique_ptr<MemoryManager> memory_manager;
+
+private:
+    /// Writes a single register in the engine bound to the specified subchannel
+    void WriteReg(u32 method, u32 subchannel, u32 value);
+
+    /// Mapping of command subchannels to their bound engine ids.
+    std::unordered_map<u32, EngineID> bound_engines;
+
+    /// 3D engine
+    std::unique_ptr<Engines::Maxwell3D> maxwell_3d;
+    /// 2D engine
+    std::unique_ptr<Engines::Fermi2D> fermi_2d;
+    /// Compute engine
+    std::unique_ptr<Engines::MaxwellCompute> maxwell_compute;
+};
+
+} // namespace Tegra

src/video_core/memory_manager.cpp

@@ -0,0 +1,110 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#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);
+
+    for (u64 offset = 0; offset < size; offset += Memory::PAGE_SIZE) {
+        PageSlot(*paddr + offset) = static_cast<u64>(PageStatus::Allocated);
+    }
+
+    return *paddr;
+}
+
+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);
+        }
+    }
+
+    for (u64 offset = 0; offset < size; offset += Memory::PAGE_SIZE) {
+        PageSlot(paddr + offset) = static_cast<u64>(PageStatus::Allocated);
+    }
+
+    return paddr;
+}
+
+PAddr MemoryManager::MapBufferEx(VAddr vaddr, u64 size) {
+    vaddr &= ~Memory::PAGE_MASK;
+
+    boost::optional<PAddr> paddr = FindFreeBlock(size);
+    ASSERT(paddr);
+
+    for (u64 offset = 0; offset < size; offset += Memory::PAGE_SIZE) {
+        PageSlot(*paddr + offset) = vaddr + offset;
+    }
+
+    return *paddr;
+}
+
+PAddr MemoryManager::MapBufferEx(VAddr vaddr, PAddr paddr, u64 size) {
+    vaddr &= ~Memory::PAGE_MASK;
+    paddr &= ~Memory::PAGE_MASK;
+
+    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 += Memory::PAGE_SIZE) {
+        PageSlot(paddr + offset) = vaddr + offset;
+    }
+
+    return paddr;
+}
+
+boost::optional<PAddr> MemoryManager::FindFreeBlock(u64 size, u64 align) {
+    PAddr paddr{};
+    u64 free_space{};
+    align = (align + Memory::PAGE_MASK) & ~Memory::PAGE_MASK;
+
+    while (paddr + free_space < MAX_ADDRESS) {
+        if (!IsPageMapped(paddr + free_space)) {
+            free_space += Memory::PAGE_SIZE;
+            if (free_space >= size) {
+                return paddr;
+            }
+        } else {
+            paddr += free_space + Memory::PAGE_SIZE;
+            free_space = 0;
+            const u64 remainder{paddr % align};
+            if (!remainder) {
+                paddr = (paddr - remainder) + align;
+            }
+        }
+    }
+
+    return {};
+}
+
+VAddr MemoryManager::PhysicalToVirtualAddress(PAddr paddr) {
+    VAddr base_addr = PageSlot(paddr);
+    ASSERT(base_addr != static_cast<u64>(PageStatus::Unmapped));
+    return base_addr + (paddr & Memory::PAGE_MASK);
+}
+
+bool MemoryManager::IsPageMapped(PAddr paddr) {
+    return PageSlot(paddr) != static_cast<u64>(PageStatus::Unmapped);
+}
+
+VAddr& MemoryManager::PageSlot(PAddr paddr) {
+    auto& block = page_table[(paddr >> (Memory::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];
+}
+
+} // namespace Tegra

src/video_core/memory_manager.h

@@ -0,0 +1,49 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include <memory>
+#include "common/common_types.h"
+#include "core/memory.h"
+
+namespace Tegra {
+
+/// Virtual addresses in the GPU's memory map are 64 bit.
+using GPUVAddr = u64;
+
+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);
+
+private:
+    boost::optional<PAddr> FindFreeBlock(u64 size, u64 align = 1);
+    bool IsPageMapped(PAddr paddr);
+    VAddr& PageSlot(PAddr paddr);
+
+    enum class PageStatus : u64 {
+        Unmapped = 0xFFFFFFFFFFFFFFFFULL,
+        Allocated = 0xFFFFFFFFFFFFFFFEULL,
+    };
+
+    static constexpr u64 MAX_ADDRESS{0x10000000000ULL};
+    static constexpr u64 PAGE_TABLE_BITS{14};
+    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};
+    static constexpr u64 PAGE_BLOCK_SIZE{1 << PAGE_BLOCK_BITS};
+    static constexpr u64 PAGE_BLOCK_MASK{PAGE_BLOCK_SIZE - 1};
+
+    using PageBlock = std::array<VAddr, PAGE_BLOCK_SIZE>;
+    std::array<std::unique_ptr<PageBlock>, PAGE_TABLE_SIZE> page_table{};
+};
+
+} // namespace Tegra

src/video_core/renderer_base.h

@@ -43,6 +43,7 @@ public:
         u32 height;
         u32 stride;
         PixelFormat pixel_format;
+        bool flip_vertical;
     };
 
     virtual ~RendererBase() {}

src/video_core/renderer_opengl/renderer_opengl.cpp

@@ -262,6 +262,8 @@ void RendererOpenGL::LoadFBToScreenInfo(const FramebufferInfo& framebuffer_info,
     // only allows rows to have a memory alignement of 4.
     ASSERT(framebuffer_info.stride % 4 == 0);
 
+    framebuffer_flip_vertical = framebuffer_info.flip_vertical;
+
     // Reset the screen info's display texture to its own permanent texture
     screen_info.display_texture = screen_info.texture.resource.handle;
     screen_info.display_texcoords = MathUtil::Rectangle<float>(0.f, 0.f, 1.f, 1.f);
@@ -401,13 +403,15 @@ void RendererOpenGL::ConfigureFramebufferTexture(TextureInfo& texture,
 
 void RendererOpenGL::DrawSingleScreen(const ScreenInfo& screen_info, float x, float y, float w,
                                       float h) {
-    auto& texcoords = screen_info.display_texcoords;
+    const auto& texcoords = screen_info.display_texcoords;
+    const auto& left = framebuffer_flip_vertical ? texcoords.right : texcoords.left;
+    const auto& right = framebuffer_flip_vertical ? texcoords.left : texcoords.right;
 
     std::array<ScreenRectVertex, 4> vertices = {{
-        ScreenRectVertex(x, y, texcoords.top, texcoords.right),
-        ScreenRectVertex(x + w, y, texcoords.bottom, texcoords.right),
-        ScreenRectVertex(x, y + h, texcoords.top, texcoords.left),
-        ScreenRectVertex(x + w, y + h, texcoords.bottom, texcoords.left),
+        ScreenRectVertex(x, y, texcoords.top, right),
+        ScreenRectVertex(x + w, y, texcoords.bottom, right),
+        ScreenRectVertex(x, y + h, texcoords.top, left),
+        ScreenRectVertex(x + w, y + h, texcoords.bottom, left),
     }};
 
     state.texture_units[0].texture_2d = screen_info.display_texture;

src/video_core/renderer_opengl/renderer_opengl.h

@@ -86,4 +86,7 @@ private:
     // Shader attribute input indices
     GLuint attrib_position;
     GLuint attrib_tex_coord;
+
+    /// Flips the framebuffer vertically when true
+    bool framebuffer_flip_vertical;
 };

src/yuzu/debugger/wait_tree.cpp

@@ -5,6 +5,7 @@
 #include "yuzu/debugger/wait_tree.h"
 #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/mutex.h"
@@ -50,7 +51,7 @@ std::size_t WaitTreeItem::Row() const {
 }
 
 std::vector<std::unique_ptr<WaitTreeThread>> WaitTreeItem::MakeThreadItemList() {
-    const auto& threads = Kernel::GetThreadList();
+    const auto& threads = Core::System::GetInstance().Scheduler().GetThreadList();
     std::vector<std::unique_ptr<WaitTreeThread>> item_list;
     item_list.reserve(threads.size());
     for (std::size_t i = 0; i < threads.size(); ++i) {
@@ -168,7 +169,7 @@ QString WaitTreeThread::GetText() const {
     }
     QString pc_info = tr(" PC = 0x%1 LR = 0x%2")
                           .arg(thread.context.pc, 8, 16, QLatin1Char('0'))
-                          .arg(thread.context.cpu_registers[31], 8, 16, QLatin1Char('0'));
+                          .arg(thread.context.cpu_registers[30], 8, 16, QLatin1Char('0'));
     return WaitTreeWaitObject::GetText() + pc_info + " (" + status + ") ";
 }