video_core: Use a CV for blocking commands.

There is no need for a busy loop here. Let's just use a condition variable to save some power.

.ci/templates/build-msvc.yml

@@ -8,7 +8,7 @@ steps:
   displayName: 'Install vulkan-sdk'
 - script: python -m pip install --upgrade pip conan
   displayName: 'Install conan'
-- script: refreshenv && mkdir build && cd build && cmake -G "Visual Studio 16 2019" -A x64 --config Release -DYUZU_USE_BUNDLED_QT=1 -DYUZU_USE_QT_WEB_ENGINE=ON -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -DYUZU_ENABLE_COMPATIBILITY_REPORTING=${COMPAT} -DUSE_DISCORD_PRESENCE=ON -DENABLE_QT_TRANSLATION=ON -DDISPLAY_VERSION=${{ parameters['version'] }} .. && cd ..
+- script: refreshenv && mkdir build && cd build && cmake -G "Visual Studio 16 2019" -A x64 -DYUZU_USE_BUNDLED_QT=1 -DYUZU_USE_QT_WEB_ENGINE=ON -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -DYUZU_ENABLE_COMPATIBILITY_REPORTING=${COMPAT} -DUSE_DISCORD_PRESENCE=ON -DENABLE_QT_TRANSLATION=ON -DDISPLAY_VERSION=${{ parameters['version'] }} .. && cmake --install . --config Release && cd ..
   displayName: 'Configure CMake'
 - task: MSBuild@1
   displayName: 'Build'

.gitmodules

@@ -27,7 +27,7 @@
     url = https://github.com/ReinUsesLisp/sirit
 [submodule "mbedtls"]
     path = externals/mbedtls
-    url = https://github.com/DarkLordZach/mbedtls
+    url = https://github.com/yuzu-emu/mbedtls
 [submodule "libzip"]
     path = externals/libzip/libzip
     url = https://github.com/nih-at/libzip.git

externals/CMakeLists.txt

@@ -64,8 +64,8 @@ endif()
 add_subdirectory(sirit)
 
 # libzip
-find_package(Libzip 1.5)
-if (NOT LIBZIP_FOUND)
+find_package(libzip 1.5)
+if (NOT libzip_FOUND)
     message(STATUS "libzip 1.5 or newer not found, falling back to externals")
     add_subdirectory(libzip EXCLUDE_FROM_ALL)
 endif()

externals/find-modules/FindLibzip.cmake

@@ -1,72 +0,0 @@
-
-find_package(PkgConfig QUIET)
-pkg_check_modules(PC_LIBZIP QUIET libzip)
-
-find_path(LIBZIP_INCLUDE_DIR
-  NAMES zip.h
-  PATHS ${PC_LIBZIP_INCLUDE_DIRS}
-  "$ENV{LIB_DIR}/include"
-  "$ENV{INCLUDE}"
-  /usr/local/include
-  /usr/include
-)
-find_path(LIBZIP_INCLUDE_DIR_ZIPCONF
-  NAMES zipconf.h
-  HINTS ${PC_LIBZIP_INCLUDE_DIRS}
-  "$ENV{LIB_DIR}/include"
-  "$ENV{LIB_DIR}/lib/libzip/include"
-  "$ENV{LIB}/lib/libzip/include"
-  /usr/local/lib/libzip/include
-  /usr/lib/libzip/include
-  /usr/local/include
-  /usr/include
-  "$ENV{INCLUDE}"
-)
-find_library(LIBZIP_LIBRARY
-  NAMES zip
-  PATHS ${PC_LIBZIP_LIBRARY_DIRS}
-  "$ENV{LIB_DIR}/lib" "$ENV{LIB}" /usr/local/lib /usr/lib
-)
-
-if (LIBZIP_INCLUDE_DIR_ZIPCONF)
-  FILE(READ "${LIBZIP_INCLUDE_DIR_ZIPCONF}/zipconf.h" _LIBZIP_VERSION_CONTENTS)
-  if (_LIBZIP_VERSION_CONTENTS)
-    STRING(REGEX REPLACE ".*#define LIBZIP_VERSION \"([0-9.]+)\".*" "\\1" LIBZIP_VERSION "${_LIBZIP_VERSION_CONTENTS}")
-  endif()
-  unset(_LIBZIP_VERSION_CONTENTS)
-endif()
-
-set(LIBZIP_VERSION ${LIBZIP_VERSION} CACHE STRING "Version number of libzip")
-
-include(FindPackageHandleStandardArgs)
-find_package_handle_standard_args(Libzip
-  FOUND_VAR LIBZIP_FOUND
-  REQUIRED_VARS
-    LIBZIP_LIBRARY
-    LIBZIP_INCLUDE_DIR
-    LIBZIP_INCLUDE_DIR_ZIPCONF
-    LIBZIP_VERSION
-  VERSION_VAR LIBZIP_VERSION
-)
-
-if(LIBZIP_FOUND)
-  set(LIBZIP_LIBRARIES ${LIBZIP_LIBRARY})
-  set(LIBZIP_INCLUDE_DIRS ${LIBZIP_INCLUDE_DIR})
-  set(LIBZIP_DEFINITIONS ${PC_LIBZIP_CFLAGS_OTHER})
-endif()
-
-if(LIBZIP_FOUND AND NOT TARGET libzip::libzip)
-  add_library(libzip::libzip UNKNOWN IMPORTED)
-  set_target_properties(libzip::libzip PROPERTIES
-    IMPORTED_LOCATION "${LIBZIP_LIBRARY}"
-    INTERFACE_COMPILE_OPTIONS "${PC_LIBZIP_CFLAGS_OTHER}"
-    INTERFACE_INCLUDE_DIRECTORIES "${LIBZIP_INCLUDE_DIR}"
-  )
-endif()
-
-mark_as_advanced(
-    LIBZIP_INCLUDE_DIR
-    LIBZIP_INCLUDE_DIR_ZIPCONF
-    LIBZIP_LIBRARY
-    LIBZIP_VERSION
-)

externals/find-modules/Findlibzip.cmake

@@ -0,0 +1,72 @@
+
+find_package(PkgConfig QUIET)
+pkg_check_modules(PC_libzip QUIET libzip)
+
+find_path(libzip_INCLUDE_DIR
+  NAMES zip.h
+  PATHS ${PC_libzip_INCLUDE_DIRS}
+  "$ENV{LIB_DIR}/include"
+  "$ENV{INCLUDE}"
+  /usr/local/include
+  /usr/include
+)
+find_path(libzip_INCLUDE_DIR_ZIPCONF
+  NAMES zipconf.h
+  HINTS ${PC_libzip_INCLUDE_DIRS}
+  "$ENV{LIB_DIR}/include"
+  "$ENV{LIB_DIR}/lib/libzip/include"
+  "$ENV{LIB}/lib/libzip/include"
+  /usr/local/lib/libzip/include
+  /usr/lib/libzip/include
+  /usr/local/include
+  /usr/include
+  "$ENV{INCLUDE}"
+)
+find_library(libzip_LIBRARY
+  NAMES zip
+  PATHS ${PC_libzip_LIBRARY_DIRS}
+  "$ENV{LIB_DIR}/lib" "$ENV{LIB}" /usr/local/lib /usr/lib
+)
+
+if (libzip_INCLUDE_DIR_ZIPCONF)
+  FILE(READ "${libzip_INCLUDE_DIR_ZIPCONF}/zipconf.h" _libzip_VERSION_CONTENTS)
+  if (_libzip_VERSION_CONTENTS)
+    STRING(REGEX REPLACE ".*#define LIBZIP_VERSION \"([0-9.]+)\".*" "\\1" libzip_VERSION "${_libzip_VERSION_CONTENTS}")
+  endif()
+  unset(_libzip_VERSION_CONTENTS)
+endif()
+
+set(libzip_VERSION ${libzip_VERSION} CACHE STRING "Version number of libzip")
+
+include(FindPackageHandleStandardArgs)
+find_package_handle_standard_args(libzip
+  FOUND_VAR libzip_FOUND
+  REQUIRED_VARS
+    libzip_LIBRARY
+    libzip_INCLUDE_DIR
+    libzip_INCLUDE_DIR_ZIPCONF
+    libzip_VERSION
+  VERSION_VAR libzip_VERSION
+)
+
+if(libzip_FOUND)
+  set(libzip_LIBRARIES ${libzip_LIBRARY})
+  set(libzip_INCLUDE_DIRS ${libzip_INCLUDE_DIR})
+  set(libzip_DEFINITIONS ${PC_libzip_CFLAGS_OTHER})
+endif()
+
+if(libzip_FOUND AND NOT TARGET libzip::libzip)
+  add_library(libzip::libzip UNKNOWN IMPORTED)
+  set_target_properties(libzip::libzip PROPERTIES
+    IMPORTED_LOCATION "${libzip_LIBRARY}"
+    INTERFACE_COMPILE_OPTIONS "${PC_libzip_CFLAGS_OTHER}"
+    INTERFACE_INCLUDE_DIRECTORIES "${libzip_INCLUDE_DIR}"
+  )
+endif()
+
+mark_as_advanced(
+    libzip_INCLUDE_DIR
+    libzip_INCLUDE_DIR_ZIPCONF
+    libzip_LIBRARY
+    libzip_VERSION
+)

externals/libusb/CMakeLists.txt

@@ -1,3 +1,8 @@
+# Ensure libusb compiles with UTF-8 encoding on MSVC
+if(MSVC)
+    add_compile_options(/utf-8)
+endif()
+
 add_library(usb STATIC EXCLUDE_FROM_ALL
     libusb/libusb/core.c
     libusb/libusb/core.c

src/CMakeLists.txt

@@ -27,6 +27,7 @@ if (MSVC)
     # /Zo                 - Enhanced debug info for optimized builds
     # /permissive-        - Enables stricter C++ standards conformance checks
     # /EHsc               - C++-only exception handling semantics
+    # /utf-8              - Set source and execution character sets to UTF-8
     # /volatile:iso       - Use strict standards-compliant volatile semantics.
     # /Zc:externConstexpr - Allow extern constexpr variables to have external linkage, like the standard mandates
     # /Zc:inline          - Let codegen omit inline functions in object files
@@ -38,6 +39,7 @@ if (MSVC)
         /permissive-
         /EHsc
         /std:c++latest
+        /utf-8
         /volatile:iso
         /Zc:externConstexpr
         /Zc:inline

src/common/CMakeLists.txt

@@ -97,6 +97,7 @@ add_custom_command(OUTPUT scm_rev.cpp
 add_library(common STATIC
     algorithm.h
     alignment.h
+    assert.cpp
     assert.h
     atomic_ops.h
     detached_tasks.cpp

src/common/assert.cpp

@@ -0,0 +1,11 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "common/assert.h"
+
+#include "common/common_funcs.h"
+
+void assert_handle_failure() {
+    Crash();
+}

src/common/assert.h

@@ -4,10 +4,13 @@
 
 #pragma once
 
-#include <cstdlib>
-#include "common/common_funcs.h"
 #include "common/logging/log.h"
 
+// Sometimes we want to try to continue even after hitting an assert.
+// However touching this file yields a global recompilation as this header is included almost
+// everywhere. So let's just move the handling of the failed assert to a single cpp file.
+void assert_handle_failure();
+
 // For asserts we'd like to keep all the junk executed when an assert happens away from the
 // important code in the function. One way of doing this is to put all the relevant code inside a
 // lambda and force the compiler to not inline it. Unfortunately, MSVC seems to have no syntax to
@@ -17,15 +20,14 @@
 // enough for our purposes.
 template <typename Fn>
 #if defined(_MSC_VER)
-[[msvc::noinline, noreturn]]
+[[msvc::noinline]]
 #elif defined(__GNUC__)
-[[gnu::cold, gnu::noinline, noreturn]]
+[[gnu::cold, gnu::noinline]]
 #endif
 static void
 assert_noinline_call(const Fn& fn) {
     fn();
-    Crash();
-    exit(1); // Keeps GCC's mouth shut about this actually returning
+    assert_handle_failure();
 }
 
 #define ASSERT(_a_)                                                                                \

src/common/cityhash.h

@@ -61,6 +61,7 @@
 
 #pragma once
 
+#include <cstddef>
 #include "common/common_types.h"
 
 namespace Common {

src/common/common_funcs.h

@@ -52,9 +52,13 @@ __declspec(dllimport) void __stdcall DebugBreak(void);
 // Generic function to get last error message.
 // Call directly after the command or use the error num.
 // This function might change the error code.
-// Defined in Misc.cpp.
+// Defined in misc.cpp.
 [[nodiscard]] std::string GetLastErrorMsg();
 
+// Like GetLastErrorMsg(), but passing an explicit error code.
+// Defined in misc.cpp.
+[[nodiscard]] std::string NativeErrorToString(int e);
+
 #define DECLARE_ENUM_FLAG_OPERATORS(type)                                                          \
     [[nodiscard]] constexpr type operator|(type a, type b) noexcept {                              \
         using T = std::underlying_type_t<type>;                                                    \

src/common/fiber.cpp

@@ -11,7 +11,7 @@
 
 namespace Common {
 
-constexpr std::size_t default_stack_size = 256 * 1024;
+constexpr std::size_t default_stack_size = 512 * 1024;
 
 struct Fiber::FiberImpl {
     FiberImpl() : stack{default_stack_size}, rewind_stack{default_stack_size} {}
@@ -24,7 +24,7 @@ struct Fiber::FiberImpl {
     std::function<void(void*)> rewind_point;
     void* rewind_parameter{};
     void* start_parameter{};
-    Fiber* previous_fiber;
+    std::shared_ptr<Fiber> previous_fiber;
     bool is_thread_fiber{};
     bool released{};
 
@@ -47,7 +47,7 @@ void Fiber::Start(boost::context::detail::transfer_t& transfer) {
     ASSERT(impl->previous_fiber != nullptr);
     impl->previous_fiber->impl->context = transfer.fctx;
     impl->previous_fiber->impl->guard.unlock();
-    impl->previous_fiber = nullptr;
+    impl->previous_fiber.reset();
     impl->entry_point(impl->start_parameter);
     UNREACHABLE();
 }
@@ -116,20 +116,23 @@ void Fiber::Rewind() {
     boost::context::detail::jump_fcontext(impl->rewind_context, this);
 }
 
-void Fiber::YieldTo(Fiber* from, Fiber* to) {
-    ASSERT_MSG(from != nullptr, "Yielding fiber is null!");
-    ASSERT_MSG(to != nullptr, "Next fiber is null!");
-    to->impl->guard.lock();
-    to->impl->previous_fiber = from;
-    auto transfer = boost::context::detail::jump_fcontext(to->impl->context, to);
-    ASSERT(from->impl->previous_fiber != nullptr);
-    from->impl->previous_fiber->impl->context = transfer.fctx;
-    from->impl->previous_fiber->impl->guard.unlock();
-    from->impl->previous_fiber = nullptr;
+void Fiber::YieldTo(std::weak_ptr<Fiber> weak_from, Fiber& to) {
+    to.impl->guard.lock();
+    to.impl->previous_fiber = weak_from.lock();
+
+    auto transfer = boost::context::detail::jump_fcontext(to.impl->context, &to);
+
+    // "from" might no longer be valid if the thread was killed
+    if (auto from = weak_from.lock()) {
+        ASSERT(from->impl->previous_fiber != nullptr);
+        from->impl->previous_fiber->impl->context = transfer.fctx;
+        from->impl->previous_fiber->impl->guard.unlock();
+        from->impl->previous_fiber.reset();
+    }
 }
 
-std::unique_ptr<Fiber> Fiber::ThreadToFiber() {
-    std::unique_ptr<Fiber> fiber = std::unique_ptr<Fiber>{new Fiber()};
+std::shared_ptr<Fiber> Fiber::ThreadToFiber() {
+    std::shared_ptr<Fiber> fiber = std::shared_ptr<Fiber>{new Fiber()};
     fiber->impl->guard.lock();
     fiber->impl->is_thread_fiber = true;
     return fiber;

src/common/fiber.h

@@ -41,8 +41,8 @@ public:
 
     /// Yields control from Fiber 'from' to Fiber 'to'
     /// Fiber 'from' must be the currently running fiber.
-    static void YieldTo(Fiber* from, Fiber* to);
-    [[nodiscard]] static std::unique_ptr<Fiber> ThreadToFiber();
+    static void YieldTo(std::weak_ptr<Fiber> weak_from, Fiber& to);
+    [[nodiscard]] static std::shared_ptr<Fiber> ThreadToFiber();
 
     void SetRewindPoint(std::function<void(void*)>&& rewind_func, void* rewind_param);
 

src/common/misc.cpp

@@ -12,27 +12,41 @@
 
 #include "common/common_funcs.h"
 
-// Generic function to get last error message.
-// Call directly after the command or use the error num.
-// This function might change the error code.
-std::string GetLastErrorMsg() {
-    static constexpr std::size_t buff_size = 255;
-    char err_str[buff_size];
-
+std::string NativeErrorToString(int e) {
 #ifdef _WIN32
-    FormatMessageA(FORMAT_MESSAGE_FROM_SYSTEM, nullptr, GetLastError(),
-                   MAKELANGID(LANG_NEUTRAL, SUBLANG_DEFAULT), err_str, buff_size, nullptr);
-    return std::string(err_str, buff_size);
-#elif defined(__GLIBC__) && (_GNU_SOURCE || (_POSIX_C_SOURCE < 200112L && _XOPEN_SOURCE < 600))
+    LPSTR err_str;
+
+    DWORD res = FormatMessageA(FORMAT_MESSAGE_FROM_SYSTEM | FORMAT_MESSAGE_ALLOCATE_BUFFER |
+                                   FORMAT_MESSAGE_IGNORE_INSERTS,
+                               nullptr, e, MAKELANGID(LANG_NEUTRAL, SUBLANG_DEFAULT),
+                               reinterpret_cast<LPSTR>(&err_str), 1, nullptr);
+    if (!res) {
+        return "(FormatMessageA failed to format error)";
+    }
+    std::string ret(err_str);
+    LocalFree(err_str);
+    return ret;
+#else
+    char err_str[255];
+#if defined(__GLIBC__) && (_GNU_SOURCE || (_POSIX_C_SOURCE < 200112L && _XOPEN_SOURCE < 600))
     // Thread safe (GNU-specific)
-    const char* str = strerror_r(errno, err_str, buff_size);
+    const char* str = strerror_r(e, err_str, sizeof(err_str));
     return std::string(str);
 #else
     // Thread safe (XSI-compliant)
-    const int success = strerror_r(errno, err_str, buff_size);
-    if (success != 0) {
-        return {};
+    int second_err = strerror_r(e, err_str, sizeof(err_str));
+    if (second_err != 0) {
+        return "(strerror_r failed to format error)";
     }
     return std::string(err_str);
+#endif // GLIBC etc.
+#endif // _WIN32
+}
+
+std::string GetLastErrorMsg() {
+#ifdef _WIN32
+    return NativeErrorToString(GetLastError());
+#else
+    return NativeErrorToString(errno);
 #endif
 }

src/common/threadsafe_queue.h

@@ -83,11 +83,15 @@ public:
         return true;
     }
 
-    T PopWait() {
+    void Wait() {
         if (Empty()) {
             std::unique_lock lock{cv_mutex};
             cv.wait(lock, [this]() { return !Empty(); });
         }
+    }
+
+    T PopWait() {
+        Wait();
         T t;
         Pop(t);
         return t;
@@ -156,6 +160,10 @@ public:
         return spsc_queue.Pop(t);
     }
 
+    void Wait() {
+        spsc_queue.Wait();
+    }
+
     T PopWait() {
         return spsc_queue.PopWait();
     }

src/core/arm/dynarmic/arm_dynarmic_32.cpp

@@ -114,18 +114,17 @@ public:
     static constexpr u64 minimum_run_cycles = 1000U;
 };
 
-std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable& page_table,
-                                                             std::size_t address_space_bits) const {
+std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable* page_table) const {
     Dynarmic::A32::UserConfig config;
     config.callbacks = cb.get();
-    // TODO(bunnei): Implement page table for 32-bit
-    // config.page_table = &page_table.pointers;
     config.coprocessors[15] = cp15;
     config.define_unpredictable_behaviour = true;
     static constexpr std::size_t PAGE_BITS = 12;
     static constexpr std::size_t NUM_PAGE_TABLE_ENTRIES = 1 << (32 - PAGE_BITS);
-    config.page_table = reinterpret_cast<std::array<std::uint8_t*, NUM_PAGE_TABLE_ENTRIES>*>(
-        page_table.pointers.data());
+    if (page_table) {
+        config.page_table = reinterpret_cast<std::array<std::uint8_t*, NUM_PAGE_TABLE_ENTRIES>*>(
+            page_table->pointers.data());
+    }
     config.absolute_offset_page_table = true;
     config.page_table_pointer_mask_bits = Common::PageTable::ATTRIBUTE_BITS;
     config.detect_misaligned_access_via_page_table = 16 | 32 | 64 | 128;
@@ -138,6 +137,10 @@ std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable&
     // Timing
     config.wall_clock_cntpct = uses_wall_clock;
 
+    // Code cache size
+    config.code_cache_size = 512 * 1024 * 1024;
+    config.far_code_offset = 256 * 1024 * 1024;
+
     // Safe optimizations
     if (Settings::values.cpu_accuracy == Settings::CPUAccuracy::DebugMode) {
         if (!Settings::values.cpuopt_page_tables) {
@@ -201,7 +204,8 @@ ARM_Dynarmic_32::ARM_Dynarmic_32(System& system, CPUInterrupts& interrupt_handle
     : ARM_Interface{system, interrupt_handlers, uses_wall_clock},
       cb(std::make_unique<DynarmicCallbacks32>(*this)),
       cp15(std::make_shared<DynarmicCP15>(*this)), core_index{core_index},
-      exclusive_monitor{dynamic_cast<DynarmicExclusiveMonitor&>(exclusive_monitor)} {}
+      exclusive_monitor{dynamic_cast<DynarmicExclusiveMonitor&>(exclusive_monitor)},
+      jit(MakeJit(nullptr)) {}
 
 ARM_Dynarmic_32::~ARM_Dynarmic_32() = default;
 
@@ -256,9 +260,6 @@ void ARM_Dynarmic_32::ChangeProcessorID(std::size_t new_core_id) {
 }
 
 void ARM_Dynarmic_32::SaveContext(ThreadContext32& ctx) {
-    if (!jit) {
-        return;
-    }
     Dynarmic::A32::Context context;
     jit->SaveContext(context);
     ctx.cpu_registers = context.Regs();
@@ -268,9 +269,6 @@ void ARM_Dynarmic_32::SaveContext(ThreadContext32& ctx) {
 }
 
 void ARM_Dynarmic_32::LoadContext(const ThreadContext32& ctx) {
-    if (!jit) {
-        return;
-    }
     Dynarmic::A32::Context context;
     context.Regs() = ctx.cpu_registers;
     context.ExtRegs() = ctx.extension_registers;
@@ -284,35 +282,31 @@ void ARM_Dynarmic_32::PrepareReschedule() {
 }
 
 void ARM_Dynarmic_32::ClearInstructionCache() {
-    if (!jit) {
-        return;
-    }
     jit->ClearCache();
 }
 
 void ARM_Dynarmic_32::InvalidateCacheRange(VAddr addr, std::size_t size) {
-    if (!jit) {
-        return;
-    }
     jit->InvalidateCacheRange(static_cast<u32>(addr), size);
 }
 
 void ARM_Dynarmic_32::ClearExclusiveState() {
-    if (!jit) {
-        return;
-    }
     jit->ClearExclusiveState();
 }
 
 void ARM_Dynarmic_32::PageTableChanged(Common::PageTable& page_table,
                                        std::size_t new_address_space_size_in_bits) {
+    ThreadContext32 ctx{};
+    SaveContext(ctx);
+
     auto key = std::make_pair(&page_table, new_address_space_size_in_bits);
     auto iter = jit_cache.find(key);
     if (iter != jit_cache.end()) {
         jit = iter->second;
+        LoadContext(ctx);
         return;
     }
-    jit = MakeJit(page_table, new_address_space_size_in_bits);
+    jit = MakeJit(&page_table);
+    LoadContext(ctx);
     jit_cache.emplace(key, jit);
 }
 

src/core/arm/dynarmic/arm_dynarmic_32.h

@@ -68,8 +68,7 @@ public:
                           std::size_t new_address_space_size_in_bits) override;
 
 private:
-    std::shared_ptr<Dynarmic::A32::Jit> MakeJit(Common::PageTable& page_table,
-                                                std::size_t address_space_bits) const;
+    std::shared_ptr<Dynarmic::A32::Jit> MakeJit(Common::PageTable* page_table) const;
 
     using JitCacheKey = std::pair<Common::PageTable*, std::size_t>;
     using JitCacheType =
@@ -80,10 +79,10 @@ private:
 
     std::unique_ptr<DynarmicCallbacks32> cb;
     JitCacheType jit_cache;
-    std::shared_ptr<Dynarmic::A32::Jit> jit;
     std::shared_ptr<DynarmicCP15> cp15;
     std::size_t core_index;
     DynarmicExclusiveMonitor& exclusive_monitor;
+    std::shared_ptr<Dynarmic::A32::Jit> jit;
 };
 
 } // namespace Core

src/core/arm/dynarmic/arm_dynarmic_64.cpp

@@ -142,7 +142,7 @@ public:
     static constexpr u64 minimum_run_cycles = 1000U;
 };
 
-std::shared_ptr<Dynarmic::A64::Jit> ARM_Dynarmic_64::MakeJit(Common::PageTable& page_table,
+std::shared_ptr<Dynarmic::A64::Jit> ARM_Dynarmic_64::MakeJit(Common::PageTable* page_table,
                                                              std::size_t address_space_bits) const {
     Dynarmic::A64::UserConfig config;
 
@@ -150,13 +150,15 @@ std::shared_ptr<Dynarmic::A64::Jit> ARM_Dynarmic_64::MakeJit(Common::PageTable&
     config.callbacks = cb.get();
 
     // Memory
-    config.page_table = reinterpret_cast<void**>(page_table.pointers.data());
-    config.page_table_address_space_bits = address_space_bits;
-    config.page_table_pointer_mask_bits = Common::PageTable::ATTRIBUTE_BITS;
-    config.silently_mirror_page_table = false;
-    config.absolute_offset_page_table = true;
-    config.detect_misaligned_access_via_page_table = 16 | 32 | 64 | 128;
-    config.only_detect_misalignment_via_page_table_on_page_boundary = true;
+    if (page_table) {
+        config.page_table = reinterpret_cast<void**>(page_table->pointers.data());
+        config.page_table_address_space_bits = address_space_bits;
+        config.page_table_pointer_mask_bits = Common::PageTable::ATTRIBUTE_BITS;
+        config.silently_mirror_page_table = false;
+        config.absolute_offset_page_table = true;
+        config.detect_misaligned_access_via_page_table = 16 | 32 | 64 | 128;
+        config.only_detect_misalignment_via_page_table_on_page_boundary = true;
+    }
 
     // Multi-process state
     config.processor_id = core_index;
@@ -175,6 +177,10 @@ std::shared_ptr<Dynarmic::A64::Jit> ARM_Dynarmic_64::MakeJit(Common::PageTable&
     // Timing
     config.wall_clock_cntpct = uses_wall_clock;
 
+    // Code cache size
+    config.code_cache_size = 512 * 1024 * 1024;
+    config.far_code_offset = 256 * 1024 * 1024;
+
     // Safe optimizations
     if (Settings::values.cpu_accuracy == Settings::CPUAccuracy::DebugMode) {
         if (!Settings::values.cpuopt_page_tables) {
@@ -237,7 +243,8 @@ ARM_Dynarmic_64::ARM_Dynarmic_64(System& system, CPUInterrupts& interrupt_handle
                                  std::size_t core_index)
     : ARM_Interface{system, interrupt_handlers, uses_wall_clock},
       cb(std::make_unique<DynarmicCallbacks64>(*this)), core_index{core_index},
-      exclusive_monitor{dynamic_cast<DynarmicExclusiveMonitor&>(exclusive_monitor)} {}
+      exclusive_monitor{dynamic_cast<DynarmicExclusiveMonitor&>(exclusive_monitor)},
+      jit(MakeJit(nullptr, 48)) {}
 
 ARM_Dynarmic_64::~ARM_Dynarmic_64() = default;
 
@@ -294,9 +301,6 @@ void ARM_Dynarmic_64::ChangeProcessorID(std::size_t new_core_id) {
 }
 
 void ARM_Dynarmic_64::SaveContext(ThreadContext64& ctx) {
-    if (!jit) {
-        return;
-    }
     ctx.cpu_registers = jit->GetRegisters();
     ctx.sp = jit->GetSP();
     ctx.pc = jit->GetPC();
@@ -308,9 +312,6 @@ void ARM_Dynarmic_64::SaveContext(ThreadContext64& ctx) {
 }
 
 void ARM_Dynarmic_64::LoadContext(const ThreadContext64& ctx) {
-    if (!jit) {
-        return;
-    }
     jit->SetRegisters(ctx.cpu_registers);
     jit->SetSP(ctx.sp);
     jit->SetPC(ctx.pc);
@@ -326,35 +327,31 @@ void ARM_Dynarmic_64::PrepareReschedule() {
 }
 
 void ARM_Dynarmic_64::ClearInstructionCache() {
-    if (!jit) {
-        return;
-    }
     jit->ClearCache();
 }
 
 void ARM_Dynarmic_64::InvalidateCacheRange(VAddr addr, std::size_t size) {
-    if (!jit) {
-        return;
-    }
     jit->InvalidateCacheRange(addr, size);
 }
 
 void ARM_Dynarmic_64::ClearExclusiveState() {
-    if (!jit) {
-        return;
-    }
     jit->ClearExclusiveState();
 }
 
 void ARM_Dynarmic_64::PageTableChanged(Common::PageTable& page_table,
                                        std::size_t new_address_space_size_in_bits) {
+    ThreadContext64 ctx{};
+    SaveContext(ctx);
+
     auto key = std::make_pair(&page_table, new_address_space_size_in_bits);
     auto iter = jit_cache.find(key);
     if (iter != jit_cache.end()) {
         jit = iter->second;
+        LoadContext(ctx);
         return;
     }
-    jit = MakeJit(page_table, new_address_space_size_in_bits);
+    jit = MakeJit(&page_table, new_address_space_size_in_bits);
+    LoadContext(ctx);
     jit_cache.emplace(key, jit);
 }
 

src/core/arm/dynarmic/arm_dynarmic_64.h

@@ -61,7 +61,7 @@ public:
                           std::size_t new_address_space_size_in_bits) override;
 
 private:
-    std::shared_ptr<Dynarmic::A64::Jit> MakeJit(Common::PageTable& page_table,
+    std::shared_ptr<Dynarmic::A64::Jit> MakeJit(Common::PageTable* page_table,
                                                 std::size_t address_space_bits) const;
 
     using JitCacheKey = std::pair<Common::PageTable*, std::size_t>;
@@ -71,10 +71,11 @@ private:
     friend class DynarmicCallbacks64;
     std::unique_ptr<DynarmicCallbacks64> cb;
     JitCacheType jit_cache;
-    std::shared_ptr<Dynarmic::A64::Jit> jit;
 
     std::size_t core_index;
     DynarmicExclusiveMonitor& exclusive_monitor;
+
+    std::shared_ptr<Dynarmic::A64::Jit> jit;
 };
 
 } // namespace Core

src/core/core.cpp

@@ -296,31 +296,20 @@ struct System::Impl {
         exit_lock = false;
 
         if (gpu_core) {
-            gpu_core->WaitIdle();
+            gpu_core->ShutDown();
         }
 
-        // Shutdown emulation session
         services.reset();
         service_manager.reset();
         cheat_engine.reset();
         telemetry_session.reset();
-
-        // Close all CPU/threading state
         cpu_manager.Shutdown();
-
-        // Release the Time Manager's resources
         time_manager.Shutdown();
-
-        // Shutdown kernel and core timing
         core_timing.Shutdown();
-        kernel.Shutdown();
-
-        // Close app loader
         app_loader.reset();
         gpu_core.reset();
         perf_stats.reset();
-
-        // Clear all applets
+        kernel.Shutdown();
         applet_manager.ClearAll();
 
         LOG_DEBUG(Core, "Shutdown OK");

src/core/cpu_manager.cpp

@@ -111,7 +111,7 @@ void CpuManager::MultiCoreRunGuestThread() {
     auto& kernel = system.Kernel();
     kernel.CurrentScheduler()->OnThreadStart();
     auto* thread = kernel.CurrentScheduler()->GetCurrentThread();
-    auto host_context = thread->GetHostContext();
+    auto& host_context = thread->GetHostContext();
     host_context->SetRewindPoint(GuestRewindFunction, this);
     MultiCoreRunGuestLoop();
 }
@@ -148,8 +148,7 @@ void CpuManager::MultiCoreRunSuspendThread() {
         auto core = kernel.GetCurrentHostThreadID();
         auto& scheduler = *kernel.CurrentScheduler();
         Kernel::KThread* current_thread = scheduler.GetCurrentThread();
-        Common::Fiber::YieldTo(current_thread->GetHostContext(),
-                               core_data[core].host_context.get());
+        Common::Fiber::YieldTo(current_thread->GetHostContext(), *core_data[core].host_context);
         ASSERT(scheduler.ContextSwitchPending());
         ASSERT(core == kernel.GetCurrentHostThreadID());
         scheduler.RescheduleCurrentCore();
@@ -202,7 +201,7 @@ void CpuManager::SingleCoreRunGuestThread() {
     auto& kernel = system.Kernel();
     kernel.CurrentScheduler()->OnThreadStart();
     auto* thread = kernel.CurrentScheduler()->GetCurrentThread();
-    auto host_context = thread->GetHostContext();
+    auto& host_context = thread->GetHostContext();
     host_context->SetRewindPoint(GuestRewindFunction, this);
     SingleCoreRunGuestLoop();
 }
@@ -246,7 +245,7 @@ void CpuManager::SingleCoreRunSuspendThread() {
         auto core = kernel.GetCurrentHostThreadID();
         auto& scheduler = *kernel.CurrentScheduler();
         Kernel::KThread* current_thread = scheduler.GetCurrentThread();
-        Common::Fiber::YieldTo(current_thread->GetHostContext(), core_data[0].host_context.get());
+        Common::Fiber::YieldTo(current_thread->GetHostContext(), *core_data[0].host_context);
         ASSERT(scheduler.ContextSwitchPending());
         ASSERT(core == kernel.GetCurrentHostThreadID());
         scheduler.RescheduleCurrentCore();
@@ -272,7 +271,7 @@ void CpuManager::PreemptSingleCore(bool from_running_enviroment) {
         scheduler.Unload(scheduler.GetCurrentThread());
 
         auto& next_scheduler = kernel.Scheduler(current_core);
-        Common::Fiber::YieldTo(current_thread->GetHostContext(), next_scheduler.ControlContext());
+        Common::Fiber::YieldTo(current_thread->GetHostContext(), *next_scheduler.ControlContext());
     }
 
     // May have changed scheduler
@@ -364,7 +363,7 @@ void CpuManager::RunThread(std::size_t core) {
 
         auto current_thread = system.Kernel().CurrentScheduler()->GetCurrentThread();
         data.is_running = true;
-        Common::Fiber::YieldTo(data.host_context.get(), current_thread->GetHostContext());
+        Common::Fiber::YieldTo(data.host_context, *current_thread->GetHostContext());
         data.is_running = false;
         data.is_paused = true;
         data.exit_barrier->Wait();

src/core/cpu_manager.h

@@ -83,7 +83,7 @@ private:
     void RunThread(std::size_t core);
 
     struct CoreData {
-        std::unique_ptr<Common::Fiber> host_context;
+        std::shared_ptr<Common::Fiber> host_context;
         std::unique_ptr<Common::Event> enter_barrier;
         std::unique_ptr<Common::Event> exit_barrier;
         std::atomic<bool> is_running;

src/core/crypto/aes_util.cpp

@@ -105,8 +105,6 @@ void AESCipher<Key, KeySize>::Transcode(const u8* src, std::size_t size, u8* des
             }
         }
     }
-
-    mbedtls_cipher_finish(context, nullptr, nullptr);
 }
 
 template <typename Key, std::size_t KeySize>

src/core/file_sys/system_archive/ng_word.cpp

@@ -14,7 +14,7 @@ namespace NgWord1Data {
 constexpr std::size_t NUMBER_WORD_TXT_FILES = 0x10;
 
 // Should this archive replacement mysteriously not work on a future game, consider updating.
-constexpr std::array<u8, 4> VERSION_DAT{0x0, 0x0, 0x0, 0x19}; // 5.1.0 System Version
+constexpr std::array<u8, 4> VERSION_DAT{0x0, 0x0, 0x0, 0x20}; // 11.0.1 System Version
 
 constexpr std::array<u8, 30> WORD_TXT{
     0xFE, 0xFF, 0x00, 0x5E, 0x00, 0x76, 0x00, 0x65, 0x00, 0x72, 0x00, 0x79, 0x00, 0x62, 0x00,
@@ -43,7 +43,7 @@ namespace NgWord2Data {
 constexpr std::size_t NUMBER_AC_NX_FILES = 0x10;
 
 // Should this archive replacement mysteriously not work on a future game, consider updating.
-constexpr std::array<u8, 4> VERSION_DAT{0x0, 0x0, 0x0, 0x15}; // 5.1.0 System Version
+constexpr std::array<u8, 4> VERSION_DAT{0x0, 0x0, 0x0, 0x1A}; // 11.0.1 System Version
 
 constexpr std::array<u8, 0x2C> AC_NX_DATA{
     0x1F, 0x8B, 0x08, 0x08, 0xD5, 0x2C, 0x09, 0x5C, 0x04, 0x00, 0x61, 0x63, 0x72, 0x61, 0x77,

src/core/file_sys/system_archive/system_version.cpp

@@ -14,15 +14,15 @@ namespace SystemVersionData {
 
 constexpr u8 VERSION_MAJOR = 11;
 constexpr u8 VERSION_MINOR = 0;
-constexpr u8 VERSION_MICRO = 0;
+constexpr u8 VERSION_MICRO = 1;
 
-constexpr u8 REVISION_MAJOR = 5;
+constexpr u8 REVISION_MAJOR = 1;
 constexpr u8 REVISION_MINOR = 0;
 
 constexpr char PLATFORM_STRING[] = "NX";
-constexpr char VERSION_HASH[] = "34197eba8810e2edd5e9dfcfbde7b340882e856d";
-constexpr char DISPLAY_VERSION[] = "11.0.0";
-constexpr char DISPLAY_TITLE[] = "NintendoSDK Firmware for NX 11.0.0-5.0";
+constexpr char VERSION_HASH[] = "69103fcb2004dace877094c2f8c29e6113be5dbf";
+constexpr char DISPLAY_VERSION[] = "11.0.1";
+constexpr char DISPLAY_TITLE[] = "NintendoSDK Firmware for NX 11.0.1-1.0";
 
 } // namespace SystemVersionData
 

src/core/hle/kernel/k_scheduler.cpp

@@ -608,7 +608,7 @@ void KScheduler::YieldToAnyThread(KernelCore& kernel) {
 }
 
 KScheduler::KScheduler(Core::System& system, s32 core_id) : system(system), core_id(core_id) {
-    switch_fiber = std::make_unique<Common::Fiber>(OnSwitch, this);
+    switch_fiber = std::make_shared<Common::Fiber>(OnSwitch, this);
     state.needs_scheduling.store(true);
     state.interrupt_task_thread_runnable = false;
     state.should_count_idle = false;
@@ -726,15 +726,15 @@ void KScheduler::ScheduleImpl() {
     // Save context for previous thread
     Unload(previous_thread);
 
-    Common::Fiber* old_context;
+    std::shared_ptr<Common::Fiber>* old_context;
     if (previous_thread != nullptr) {
-        old_context = previous_thread->GetHostContext();
+        old_context = &previous_thread->GetHostContext();
     } else {
-        old_context = idle_thread->GetHostContext();
+        old_context = &idle_thread->GetHostContext();
     }
     guard.unlock();
 
-    Common::Fiber::YieldTo(old_context, switch_fiber.get());
+    Common::Fiber::YieldTo(*old_context, *switch_fiber);
     /// When a thread wakes up, the scheduler may have changed to other in another core.
     auto& next_scheduler = *system.Kernel().CurrentScheduler();
     next_scheduler.SwitchContextStep2();
@@ -769,13 +769,8 @@ void KScheduler::SwitchToCurrent() {
                     break;
                 }
             }
-            Common::Fiber* next_context;
-            if (next_thread != nullptr) {
-                next_context = next_thread->GetHostContext();
-            } else {
-                next_context = idle_thread->GetHostContext();
-            }
-            Common::Fiber::YieldTo(switch_fiber.get(), next_context);
+            auto thread = next_thread ? next_thread : idle_thread;
+            Common::Fiber::YieldTo(switch_fiber, *thread->GetHostContext());
         } while (!is_switch_pending());
     }
 }
@@ -800,9 +795,9 @@ void KScheduler::Initialize() {
     std::string name = "Idle Thread Id:" + std::to_string(core_id);
     std::function<void(void*)> init_func = Core::CpuManager::GetIdleThreadStartFunc();
     void* init_func_parameter = system.GetCpuManager().GetStartFuncParamater();
-    auto thread_res = KThread::Create(system, ThreadType::Main, name, 0,
-                                      KThread::IdleThreadPriority, 0, static_cast<u32>(core_id), 0,
-                                      nullptr, std::move(init_func), init_func_parameter);
+    auto thread_res = KThread::CreateThread(
+        system, ThreadType::Main, name, 0, KThread::IdleThreadPriority, 0,
+        static_cast<u32>(core_id), 0, nullptr, std::move(init_func), init_func_parameter);
     idle_thread = thread_res.Unwrap().get();
 }
 

src/core/hle/kernel/k_scheduler.h

@@ -68,12 +68,12 @@ public:
 
     void OnThreadStart();
 
-    [[nodiscard]] Common::Fiber* ControlContext() {
-        return switch_fiber.get();
+    [[nodiscard]] std::shared_ptr<Common::Fiber>& ControlContext() {
+        return switch_fiber;
     }
 
-    [[nodiscard]] const Common::Fiber* ControlContext() const {
-        return switch_fiber.get();
+    [[nodiscard]] const std::shared_ptr<Common::Fiber>& ControlContext() const {
+        return switch_fiber;
     }
 
     [[nodiscard]] u64 UpdateHighestPriorityThread(KThread* highest_thread);
@@ -178,7 +178,7 @@ private:
 
     KThread* idle_thread;
 
-    std::unique_ptr<Common::Fiber> switch_fiber{};
+    std::shared_ptr<Common::Fiber> switch_fiber{};
 
     struct SchedulingState {
         std::atomic<bool> needs_scheduling;

src/core/hle/kernel/k_thread.cpp

@@ -991,22 +991,15 @@ void KThread::SetState(ThreadState state) {
     }
 }
 
-ResultVal<std::shared_ptr<KThread>> KThread::Create(Core::System& system, ThreadType type_flags,
-                                                    std::string name, VAddr entry_point,
-                                                    u32 priority, u64 arg, s32 processor_id,
-                                                    VAddr stack_top, Process* owner_process) {
-    std::function<void(void*)> init_func = Core::CpuManager::GetGuestThreadStartFunc();
-    void* init_func_parameter = system.GetCpuManager().GetStartFuncParamater();
-    return Create(system, type_flags, name, entry_point, priority, arg, processor_id, stack_top,
-                  owner_process, std::move(init_func), init_func_parameter);
+std::shared_ptr<Common::Fiber>& KThread::GetHostContext() {
+    return host_context;
 }
 
-ResultVal<std::shared_ptr<KThread>> KThread::Create(Core::System& system, ThreadType type_flags,
-                                                    std::string name, VAddr entry_point,
-                                                    u32 priority, u64 arg, s32 processor_id,
-                                                    VAddr stack_top, Process* owner_process,
-                                                    std::function<void(void*)>&& thread_start_func,
-                                                    void* thread_start_parameter) {
+ResultVal<std::shared_ptr<KThread>> KThread::CreateThread(Core::System& system,
+                                                          ThreadType type_flags, std::string name,
+                                                          VAddr entry_point, u32 priority, u64 arg,
+                                                          s32 processor_id, VAddr stack_top,
+                                                          Process* owner_process) {
     auto& kernel = system.Kernel();
 
     std::shared_ptr<KThread> thread = std::make_shared<KThread>(kernel);
@@ -1023,12 +1016,35 @@ ResultVal<std::shared_ptr<KThread>> KThread::Create(Core::System& system, Thread
     auto& scheduler = kernel.GlobalSchedulerContext();
     scheduler.AddThread(thread);
 
-    thread->host_context =
-        std::make_unique<Common::Fiber>(std::move(thread_start_func), thread_start_parameter);
-
     return MakeResult<std::shared_ptr<KThread>>(std::move(thread));
 }
 
+ResultVal<std::shared_ptr<KThread>> KThread::CreateThread(
+    Core::System& system, ThreadType type_flags, std::string name, VAddr entry_point, u32 priority,
+    u64 arg, s32 processor_id, VAddr stack_top, Process* owner_process,
+    std::function<void(void*)>&& thread_start_func, void* thread_start_parameter) {
+    auto thread_result = CreateThread(system, type_flags, name, entry_point, priority, arg,
+                                      processor_id, stack_top, owner_process);
+
+    if (thread_result.Succeeded()) {
+        (*thread_result)->host_context =
+            std::make_shared<Common::Fiber>(std::move(thread_start_func), thread_start_parameter);
+    }
+
+    return thread_result;
+}
+
+ResultVal<std::shared_ptr<KThread>> KThread::CreateUserThread(
+    Core::System& system, ThreadType type_flags, std::string name, VAddr entry_point, u32 priority,
+    u64 arg, s32 processor_id, VAddr stack_top, Process* owner_process) {
+    std::function<void(void*)> init_func = Core::CpuManager::GetGuestThreadStartFunc();
+
+    void* init_func_parameter = system.GetCpuManager().GetStartFuncParamater();
+
+    return CreateThread(system, type_flags, name, entry_point, priority, arg, processor_id,
+                        stack_top, owner_process, std::move(init_func), init_func_parameter);
+}
+
 KThread* GetCurrentThreadPointer(KernelCore& kernel) {
     return kernel.GetCurrentEmuThread();
 }

src/core/hle/kernel/k_thread.h

@@ -116,7 +116,7 @@ public:
     using WaiterList = boost::intrusive::list<KThread>;
 
     /**
-     * Creates and returns a new thread. The new thread is immediately scheduled
+     * Creates and returns a new thread.
      * @param system The instance of the whole system
      * @param name The friendly name desired for the thread
      * @param entry_point The address at which the thread should start execution
@@ -127,12 +127,12 @@ public:
      * @param owner_process The parent process for the thread, if null, it's a kernel thread
      * @return A shared pointer to the newly created thread
      */
-    [[nodiscard]] static ResultVal<std::shared_ptr<KThread>> Create(
+    [[nodiscard]] static ResultVal<std::shared_ptr<KThread>> CreateThread(
         Core::System& system, ThreadType type_flags, std::string name, VAddr entry_point,
         u32 priority, u64 arg, s32 processor_id, VAddr stack_top, Process* owner_process);
 
     /**
-     * Creates and returns a new thread. The new thread is immediately scheduled
+     * Creates and returns a new thread, with a specified entry point.
      * @param system The instance of the whole system
      * @param name The friendly name desired for the thread
      * @param entry_point The address at which the thread should start execution
@@ -145,11 +145,27 @@ public:
      * @param thread_start_parameter The parameter which will passed to host context on init
      * @return A shared pointer to the newly created thread
      */
-    [[nodiscard]] static ResultVal<std::shared_ptr<KThread>> Create(
+    [[nodiscard]] static ResultVal<std::shared_ptr<KThread>> CreateThread(
         Core::System& system, ThreadType type_flags, std::string name, VAddr entry_point,
         u32 priority, u64 arg, s32 processor_id, VAddr stack_top, Process* owner_process,
         std::function<void(void*)>&& thread_start_func, void* thread_start_parameter);
 
+    /**
+     * Creates and returns a new thread for the emulated "user" process.
+     * @param system The instance of the whole system
+     * @param name The friendly name desired for the thread
+     * @param entry_point The address at which the thread should start execution
+     * @param priority The thread's priority
+     * @param arg User data to pass to the thread
+     * @param processor_id The ID(s) of the processors on which the thread is desired to be run
+     * @param stack_top The address of the thread's stack top
+     * @param owner_process The parent process for the thread, if null, it's a kernel thread
+     * @return A shared pointer to the newly created thread
+     */
+    [[nodiscard]] static ResultVal<std::shared_ptr<KThread>> CreateUserThread(
+        Core::System& system, ThreadType type_flags, std::string name, VAddr entry_point,
+        u32 priority, u64 arg, s32 processor_id, VAddr stack_top, Process* owner_process);
+
     [[nodiscard]] std::string GetName() const override {
         return name;
     }
@@ -293,13 +309,7 @@ public:
         return thread_context_64;
     }
 
-    [[nodiscard]] Common::Fiber* GetHostContext() {
-        return host_context.get();
-    }
-
-    [[nodiscard]] const Common::Fiber* GetHostContext() const {
-        return host_context.get();
-    }
+    [[nodiscard]] std::shared_ptr<Common::Fiber>& GetHostContext();
 
     [[nodiscard]] ThreadState GetState() const {
         return thread_state & ThreadState::Mask;
@@ -725,7 +735,7 @@ private:
     Common::SpinLock context_guard{};
 
     // For emulation
-    std::unique_ptr<Common::Fiber> host_context{};
+    std::shared_ptr<Common::Fiber> host_context{};
 
     // For debugging
     std::vector<KSynchronizationObject*> wait_objects_for_debugging;

src/core/hle/kernel/kernel.cpp

@@ -68,9 +68,9 @@ struct KernelCore::Impl {
         InitializePhysicalCores();
         InitializeSystemResourceLimit(kernel, system);
         InitializeMemoryLayout();
-        InitializePreemption(kernel);
         InitializeSchedulers();
         InitializeSuspendThreads();
+        InitializePreemption(kernel);
     }
 
     void InitializeCores() {
@@ -181,9 +181,9 @@ struct KernelCore::Impl {
             std::string name = "Suspend Thread Id:" + std::to_string(i);
             std::function<void(void*)> init_func = Core::CpuManager::GetSuspendThreadStartFunc();
             void* init_func_parameter = system.GetCpuManager().GetStartFuncParamater();
-            auto thread_res = KThread::Create(system, ThreadType::HighPriority, std::move(name), 0,
-                                              0, 0, static_cast<u32>(i), 0, nullptr,
-                                              std::move(init_func), init_func_parameter);
+            auto thread_res = KThread::CreateThread(
+                system, ThreadType::HighPriority, std::move(name), 0, 0, 0, static_cast<u32>(i), 0,
+                nullptr, std::move(init_func), init_func_parameter);
 
             suspend_threads[i] = std::move(thread_res).Unwrap();
         }
@@ -221,10 +221,9 @@ struct KernelCore::Impl {
     // Gets the dummy KThread for the caller, allocating a new one if this is the first time
     KThread* GetHostDummyThread() {
         const thread_local auto thread =
-            KThread::Create(
+            KThread::CreateThread(
                 system, ThreadType::Main, fmt::format("DummyThread:{}", GetHostThreadId()), 0,
-                KThread::DefaultThreadPriority, 0, static_cast<u32>(3), 0, nullptr,
-                []([[maybe_unused]] void* arg) { UNREACHABLE(); }, nullptr)
+                KThread::DefaultThreadPriority, 0, static_cast<u32>(3), 0, nullptr)
                 .Unwrap();
         return thread.get();
     }

src/core/hle/kernel/process.cpp

@@ -40,8 +40,9 @@ namespace {
 void SetupMainThread(Core::System& system, Process& owner_process, u32 priority, VAddr stack_top) {
     const VAddr entry_point = owner_process.PageTable().GetCodeRegionStart();
     ASSERT(owner_process.GetResourceLimit()->Reserve(LimitableResource::Threads, 1));
-    auto thread_res = KThread::Create(system, ThreadType::User, "main", entry_point, priority, 0,
-                                      owner_process.GetIdealCoreId(), stack_top, &owner_process);
+    auto thread_res =
+        KThread::CreateUserThread(system, ThreadType::User, "main", entry_point, priority, 0,
+                                  owner_process.GetIdealCoreId(), stack_top, &owner_process);
 
     std::shared_ptr<KThread> thread = std::move(thread_res).Unwrap();
 

src/core/hle/kernel/process_capability.cpp

@@ -281,11 +281,6 @@ ResultCode ProcessCapabilities::HandleSyscallFlags(u32& set_svc_bits, u32 flags)
             continue;
         }
 
-        if (svc_number >= svc_capabilities.size()) {
-            LOG_ERROR(Kernel, "Process svc capability is out of range! svc_number={}", svc_number);
-            return ResultOutOfRange;
-        }
-
         svc_capabilities[svc_number] = true;
     }
 

src/core/hle/kernel/process_capability.h

@@ -68,7 +68,7 @@ enum class ProgramType {
 class ProcessCapabilities {
 public:
     using InterruptCapabilities = std::bitset<1024>;
-    using SyscallCapabilities = std::bitset<128>;
+    using SyscallCapabilities = std::bitset<192>;
 
     ProcessCapabilities() = default;
     ProcessCapabilities(const ProcessCapabilities&) = delete;

src/core/hle/kernel/svc.cpp

@@ -1532,8 +1532,9 @@ static ResultCode CreateThread(Core::System& system, Handle* out_handle, VAddr e
     std::shared_ptr<KThread> thread;
     {
         KScopedLightLock lk{process.GetStateLock()};
-        CASCADE_RESULT(thread, KThread::Create(system, ThreadType::User, "", entry_point, priority,
-                                               arg, core_id, stack_bottom, &process));
+        CASCADE_RESULT(thread,
+                       KThread::CreateUserThread(system, ThreadType::User, "", entry_point,
+                                                 priority, arg, core_id, stack_bottom, &process));
     }
 
     const auto new_thread_handle = process.GetHandleTable().Create(thread);
@@ -2626,7 +2627,8 @@ void Call(Core::System& system, u32 immediate) {
     kernel.ExitSVCProfile();
 
     if (!thread->IsCallingSvc()) {
-        thread->GetHostContext()->Rewind();
+        auto* host_context = thread->GetHostContext().get();
+        host_context->Rewind();
     }
 
     system.EnterDynarmicProfile();

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

@@ -295,7 +295,7 @@ ISelfController::ISelfController(Core::System& system_, NVFlinger::NVFlinger& nv
         {80, nullptr, "SetWirelessPriorityMode"},
         {90, &ISelfController::GetAccumulatedSuspendedTickValue, "GetAccumulatedSuspendedTickValue"},
         {91, &ISelfController::GetAccumulatedSuspendedTickChangedEvent, "GetAccumulatedSuspendedTickChangedEvent"},
-        {100, nullptr, "SetAlbumImageTakenNotificationEnabled"},
+        {100, &ISelfController::SetAlbumImageTakenNotificationEnabled, "SetAlbumImageTakenNotificationEnabled"},
         {110, nullptr, "SetApplicationAlbumUserData"},
         {1000, nullptr, "GetDebugStorageChannel"},
     };
@@ -560,6 +560,21 @@ void ISelfController::GetAccumulatedSuspendedTickChangedEvent(Kernel::HLERequest
     rb.PushCopyObjects(accumulated_suspended_tick_changed_event->GetReadableEvent());
 }
 
+void ISelfController::SetAlbumImageTakenNotificationEnabled(Kernel::HLERequestContext& ctx) {
+    IPC::RequestParser rp{ctx};
+
+    // This service call sets an internal flag whether a notification is shown when an image is
+    // captured. Currently we do not support capturing images via the capture button, so this can be
+    // stubbed for now.
+    const bool album_image_taken_notification_enabled = rp.Pop<bool>();
+
+    LOG_WARNING(Service_AM, "(STUBBED) called. album_image_taken_notification_enabled={}",
+                album_image_taken_notification_enabled);
+
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(RESULT_SUCCESS);
+}
+
 AppletMessageQueue::AppletMessageQueue(Kernel::KernelCore& kernel) {
     on_new_message = Kernel::KEvent::Create(kernel, "AMMessageQueue:OnMessageReceived");
     on_new_message->Initialize();

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

@@ -146,6 +146,7 @@ private:
     void IsAutoSleepDisabled(Kernel::HLERequestContext& ctx);
     void GetAccumulatedSuspendedTickValue(Kernel::HLERequestContext& ctx);
     void GetAccumulatedSuspendedTickChangedEvent(Kernel::HLERequestContext& ctx);
+    void SetAlbumImageTakenNotificationEnabled(Kernel::HLERequestContext& ctx);
 
     enum class ScreenshotPermission : u32 {
         Inherit = 0,

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

@@ -297,6 +297,10 @@ HwOpus::HwOpus(Core::System& system_) : ServiceFramework{system_, "hwopus"} {
         {1, &HwOpus::GetWorkBufferSize, "GetWorkBufferSize"},
         {2, nullptr, "OpenOpusDecoderForMultiStream"},
         {3, nullptr, "GetWorkBufferSizeForMultiStream"},
+        {4, nullptr, "OpenHardwareOpusDecoderEx"},
+        {5, nullptr, "GetWorkBufferSizeEx"},
+        {6, nullptr, "OpenHardwareOpusDecoderForMultiStreamEx"},
+        {7, nullptr, "GetWorkBufferSizeForMultiStreamEx"},
     };
     RegisterHandlers(functions);
 }

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

@@ -133,7 +133,7 @@ private:
     void GetBlockedUserListIds(Kernel::HLERequestContext& ctx) {
         // This is safe to stub, as there should be no adverse consequences from reporting no
         // blocked users.
-        LOG_WARNING(Service_ACC, "(STUBBED) called");
+        LOG_WARNING(Service_Friend, "(STUBBED) called");
         IPC::ResponseBuilder rb{ctx, 3};
         rb.Push(RESULT_SUCCESS);
         rb.Push<u32>(0); // Indicates there are no blocked users
@@ -141,14 +141,14 @@ private:
 
     void DeclareCloseOnlinePlaySession(Kernel::HLERequestContext& ctx) {
         // Stub used by Splatoon 2
-        LOG_WARNING(Service_ACC, "(STUBBED) called");
+        LOG_WARNING(Service_Friend, "(STUBBED) called");
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
     }
 
     void UpdateUserPresence(Kernel::HLERequestContext& ctx) {
         // Stub used by Retro City Rampage
-        LOG_WARNING(Service_ACC, "(STUBBED) called");
+        LOG_WARNING(Service_Friend, "(STUBBED) called");
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
     }
@@ -159,7 +159,7 @@ private:
         const auto uuid = rp.PopRaw<Common::UUID>();
         [[maybe_unused]] const auto filter = rp.PopRaw<SizedFriendFilter>();
         const auto pid = rp.Pop<u64>();
-        LOG_WARNING(Service_ACC, "(STUBBED) called, offset={}, uuid={}, pid={}", friend_offset,
+        LOG_WARNING(Service_Friend, "(STUBBED) called, offset={}, uuid={}, pid={}", friend_offset,
                     uuid.Format(), pid);
 
         IPC::ResponseBuilder rb{ctx, 3};
@@ -191,7 +191,7 @@ public:
 
 private:
     void GetEvent(Kernel::HLERequestContext& ctx) {
-        LOG_DEBUG(Service_ACC, "called");
+        LOG_DEBUG(Service_Friend, "called");
 
         IPC::ResponseBuilder rb{ctx, 2, 1};
         rb.Push(RESULT_SUCCESS);
@@ -199,7 +199,7 @@ private:
     }
 
     void Clear(Kernel::HLERequestContext& ctx) {
-        LOG_DEBUG(Service_ACC, "called");
+        LOG_DEBUG(Service_Friend, "called");
         while (!notifications.empty()) {
             notifications.pop();
         }
@@ -210,10 +210,10 @@ private:
     }
 
     void Pop(Kernel::HLERequestContext& ctx) {
-        LOG_DEBUG(Service_ACC, "called");
+        LOG_DEBUG(Service_Friend, "called");
 
         if (notifications.empty()) {
-            LOG_ERROR(Service_ACC, "No notifications in queue!");
+            LOG_ERROR(Service_Friend, "No notifications in queue!");
             IPC::ResponseBuilder rb{ctx, 2};
             rb.Push(ERR_NO_NOTIFICATIONS);
             return;
@@ -231,7 +231,8 @@ private:
             break;
         default:
             // HOS seems not have an error case for an unknown notification
-            LOG_WARNING(Service_ACC, "Unknown notification {:08X}", notification.notification_type);
+            LOG_WARNING(Service_Friend, "Unknown notification {:08X}",
+                        notification.notification_type);
             break;
         }
 
@@ -269,14 +270,14 @@ void Module::Interface::CreateFriendService(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
     rb.PushIpcInterface<IFriendService>(system);
-    LOG_DEBUG(Service_ACC, "called");
+    LOG_DEBUG(Service_Friend, "called");
 }
 
 void Module::Interface::CreateNotificationService(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
     auto uuid = rp.PopRaw<Common::UUID>();
 
-    LOG_DEBUG(Service_ACC, "called, uuid={}", uuid.Format());
+    LOG_DEBUG(Service_Friend, "called, uuid={}", uuid.Format());
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);

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

@@ -4,16 +4,27 @@
 
 #include <cstring>
 #include "common/common_types.h"
+#include "common/logging/log.h"
 #include "core/core_timing.h"
+#include "core/frontend/emu_window.h"
 #include "core/hle/service/hid/controllers/gesture.h"
+#include "core/settings.h"
 
 namespace Service::HID {
 constexpr std::size_t SHARED_MEMORY_OFFSET = 0x3BA00;
+constexpr f32 angle_threshold = 0.08f;
+constexpr f32 pinch_threshold = 100.0f;
 
 Controller_Gesture::Controller_Gesture(Core::System& system) : ControllerBase(system) {}
 Controller_Gesture::~Controller_Gesture() = default;
 
-void Controller_Gesture::OnInit() {}
+void Controller_Gesture::OnInit() {
+    for (std::size_t id = 0; id < MAX_FINGERS; ++id) {
+        mouse_finger_id[id] = MAX_POINTS;
+        keyboard_finger_id[id] = MAX_POINTS;
+        udp_finger_id[id] = MAX_POINTS;
+    }
+}
 
 void Controller_Gesture::OnRelease() {}
 
@@ -35,10 +46,157 @@ void Controller_Gesture::OnUpdate(const Core::Timing::CoreTiming& core_timing, u
 
     cur_entry.sampling_number = last_entry.sampling_number + 1;
     cur_entry.sampling_number2 = cur_entry.sampling_number;
-    // TODO(ogniK): Update gesture states
+
+    // TODO(german77): Implement all gesture types
+
+    const Input::TouchStatus& mouse_status = touch_mouse_device->GetStatus();
+    const Input::TouchStatus& udp_status = touch_udp_device->GetStatus();
+    for (std::size_t id = 0; id < mouse_status.size(); ++id) {
+        mouse_finger_id[id] = UpdateTouchInputEvent(mouse_status[id], mouse_finger_id[id]);
+        udp_finger_id[id] = UpdateTouchInputEvent(udp_status[id], udp_finger_id[id]);
+    }
+
+    if (Settings::values.use_touch_from_button) {
+        const Input::TouchStatus& keyboard_status = touch_btn_device->GetStatus();
+        for (std::size_t id = 0; id < mouse_status.size(); ++id) {
+            keyboard_finger_id[id] =
+                UpdateTouchInputEvent(keyboard_status[id], keyboard_finger_id[id]);
+        }
+    }
+
+    TouchType type = TouchType::Idle;
+    Attribute attributes{};
+    GestureProperties gesture = GetGestureProperties();
+    if (last_gesture.active_points != gesture.active_points) {
+        ++last_gesture.detection_count;
+    }
+    if (gesture.active_points > 0) {
+        if (last_gesture.active_points == 0) {
+            attributes.is_new_touch.Assign(true);
+            last_gesture.average_distance = gesture.average_distance;
+            last_gesture.angle = gesture.angle;
+        }
+
+        type = TouchType::Touch;
+        if (gesture.mid_point.x != last_entry.x || gesture.mid_point.y != last_entry.y) {
+            type = TouchType::Pan;
+        }
+        if (std::abs(gesture.average_distance - last_gesture.average_distance) > pinch_threshold) {
+            type = TouchType::Pinch;
+        }
+        if (std::abs(gesture.angle - last_gesture.angle) > angle_threshold) {
+            type = TouchType::Rotate;
+        }
+
+        cur_entry.delta_x = gesture.mid_point.x - last_entry.x;
+        cur_entry.delta_y = gesture.mid_point.y - last_entry.y;
+        // TODO: Find how velocities are calculated
+        cur_entry.vel_x = static_cast<float>(cur_entry.delta_x) * 150.1f;
+        cur_entry.vel_y = static_cast<float>(cur_entry.delta_y) * 150.1f;
+
+        // Slowdown the rate of change for less flapping
+        last_gesture.average_distance =
+            (last_gesture.average_distance * 0.9f) + (gesture.average_distance * 0.1f);
+        last_gesture.angle = (last_gesture.angle * 0.9f) + (gesture.angle * 0.1f);
+
+    } else {
+        cur_entry.delta_x = 0;
+        cur_entry.delta_y = 0;
+        cur_entry.vel_x = 0;
+        cur_entry.vel_y = 0;
+    }
+    last_gesture.active_points = gesture.active_points;
+    cur_entry.detection_count = last_gesture.detection_count;
+    cur_entry.type = type;
+    cur_entry.attributes = attributes;
+    cur_entry.x = gesture.mid_point.x;
+    cur_entry.y = gesture.mid_point.y;
+    cur_entry.point_count = static_cast<s32>(gesture.active_points);
+    for (size_t id = 0; id < MAX_POINTS; id++) {
+        cur_entry.points[id].x = gesture.points[id].x;
+        cur_entry.points[id].y = gesture.points[id].y;
+    }
+    cur_entry.rotation_angle = 0;
+    cur_entry.scale = 0;
 
     std::memcpy(data + SHARED_MEMORY_OFFSET, &shared_memory, sizeof(SharedMemory));
 }
 
-void Controller_Gesture::OnLoadInputDevices() {}
+void Controller_Gesture::OnLoadInputDevices() {
+    touch_mouse_device = Input::CreateDevice<Input::TouchDevice>("engine:emu_window");
+    touch_udp_device = Input::CreateDevice<Input::TouchDevice>("engine:cemuhookudp");
+    touch_btn_device = Input::CreateDevice<Input::TouchDevice>("engine:touch_from_button");
+}
+
+std::optional<std::size_t> Controller_Gesture::GetUnusedFingerID() const {
+    std::size_t first_free_id = 0;
+    while (first_free_id < MAX_POINTS) {
+        if (!fingers[first_free_id].pressed) {
+            return first_free_id;
+        } else {
+            first_free_id++;
+        }
+    }
+    return std::nullopt;
+}
+
+std::size_t Controller_Gesture::UpdateTouchInputEvent(
+    const std::tuple<float, float, bool>& touch_input, std::size_t finger_id) {
+    const auto& [x, y, pressed] = touch_input;
+    if (finger_id > MAX_POINTS) {
+        LOG_ERROR(Service_HID, "Invalid finger id {}", finger_id);
+        return MAX_POINTS;
+    }
+    if (pressed) {
+        if (finger_id == MAX_POINTS) {
+            const auto first_free_id = GetUnusedFingerID();
+            if (!first_free_id) {
+                // Invalid finger id do nothing
+                return MAX_POINTS;
+            }
+            finger_id = first_free_id.value();
+            fingers[finger_id].pressed = true;
+        }
+        fingers[finger_id].x = x;
+        fingers[finger_id].y = y;
+        return finger_id;
+    }
+
+    if (finger_id != MAX_POINTS) {
+        fingers[finger_id].pressed = false;
+    }
+
+    return MAX_POINTS;
+}
+
+Controller_Gesture::GestureProperties Controller_Gesture::GetGestureProperties() {
+    GestureProperties gesture;
+    std::array<Finger, MAX_POINTS> active_fingers;
+    const auto end_iter = std::copy_if(fingers.begin(), fingers.end(), active_fingers.begin(),
+                                       [](const auto& finger) { return finger.pressed; });
+    gesture.active_points =
+        static_cast<std::size_t>(std::distance(active_fingers.begin(), end_iter));
+
+    for (size_t id = 0; id < gesture.active_points; ++id) {
+        gesture.points[id].x =
+            static_cast<int>(active_fingers[id].x * Layout::ScreenUndocked::Width);
+        gesture.points[id].y =
+            static_cast<int>(active_fingers[id].y * Layout::ScreenUndocked::Height);
+        gesture.mid_point.x += static_cast<int>(gesture.points[id].x / gesture.active_points);
+        gesture.mid_point.y += static_cast<int>(gesture.points[id].y / gesture.active_points);
+    }
+
+    for (size_t id = 0; id < gesture.active_points; ++id) {
+        const double distance =
+            std::pow(static_cast<float>(gesture.mid_point.x - gesture.points[id].x), 2) +
+            std::pow(static_cast<float>(gesture.mid_point.y - gesture.points[id].y), 2);
+        gesture.average_distance +=
+            static_cast<float>(distance) / static_cast<float>(gesture.active_points);
+    }
+
+    gesture.angle = std::atan2(static_cast<float>(gesture.mid_point.y - gesture.points[0].y),
+                               static_cast<float>(gesture.mid_point.x - gesture.points[0].x));
+    return gesture;
+}
+
 } // namespace Service::HID

src/core/hle/service/hid/controllers/gesture.h

@@ -5,8 +5,10 @@
 #pragma once
 
 #include <array>
+#include "common/bit_field.h"
 #include "common/common_types.h"
 #include "common/swap.h"
+#include "core/frontend/input.h"
 #include "core/hle/service/hid/controllers/controller_base.h"
 
 namespace Service::HID {
@@ -28,29 +30,64 @@ public:
     void OnLoadInputDevices() override;
 
 private:
-    struct Locations {
+    static constexpr size_t MAX_FINGERS = 16;
+    static constexpr size_t MAX_POINTS = 4;
+
+    enum class TouchType : u32 {
+        Idle,     // Nothing touching the screen
+        Complete, // Unknown. End of touch?
+        Cancel,   // Never triggered
+        Touch,    // Pressing without movement
+        Press,    // Never triggered
+        Tap,      // Fast press then release
+        Pan,      // All points moving together across the screen
+        Swipe,    // Fast press movement and release of a single point
+        Pinch,    // All points moving away/closer to the midpoint
+        Rotate,   // All points rotating from the midpoint
+    };
+
+    enum class Direction : u32 {
+        None,
+        Left,
+        Up,
+        Right,
+        Down,
+    };
+
+    struct Attribute {
+        union {
+            u32_le raw{};
+
+            BitField<0, 1, u32> is_new_touch;
+            BitField<1, 1, u32> is_double_tap;
+        };
+    };
+    static_assert(sizeof(Attribute) == 4, "Attribute is an invalid size");
+
+    struct Points {
         s32_le x;
         s32_le y;
     };
+    static_assert(sizeof(Points) == 8, "Points is an invalid size");
 
     struct GestureState {
         s64_le sampling_number;
         s64_le sampling_number2;
 
         s64_le detection_count;
-        s32_le type;
-        s32_le dir;
+        TouchType type;
+        Direction dir;
         s32_le x;
         s32_le y;
         s32_le delta_x;
         s32_le delta_y;
         f32 vel_x;
         f32 vel_y;
-        s32_le attributes;
-        f32 scale;
-        f32 rotation;
-        s32_le location_count;
-        std::array<Locations, 4> locations;
+        Attribute attributes;
+        u32 scale;
+        u32 rotation_angle;
+        s32_le point_count;
+        std::array<Points, 4> points;
     };
     static_assert(sizeof(GestureState) == 0x68, "GestureState is an invalid size");
 
@@ -58,6 +95,45 @@ private:
         CommonHeader header;
         std::array<GestureState, 17> gesture_states;
     };
+    static_assert(sizeof(SharedMemory) == 0x708, "SharedMemory is an invalid size");
+
+    struct Finger {
+        f32 x{};
+        f32 y{};
+        bool pressed{};
+    };
+
+    struct GestureProperties {
+        std::array<Points, MAX_POINTS> points{};
+        std::size_t active_points{};
+        Points mid_point{};
+        s64_le detection_count{};
+        u64_le delta_time{};
+        float average_distance{};
+        float angle{};
+    };
+
+    // Returns an unused finger id, if there is no fingers avaliable MAX_FINGERS will be returned
+    std::optional<size_t> GetUnusedFingerID() const;
+
+    /** If the touch is new it tries to assing a new finger id, if there is no fingers avaliable no
+     * changes will be made. Updates the coordinates if the finger id it's already set. If the touch
+     * ends delays the output by one frame to set the end_touch flag before finally freeing the
+     * finger id */
+    size_t UpdateTouchInputEvent(const std::tuple<float, float, bool>& touch_input,
+                                 size_t finger_id);
+
+    // Returns the average distance, angle and middle point of the active fingers
+    GestureProperties GetGestureProperties();
+
     SharedMemory shared_memory{};
+    std::unique_ptr<Input::TouchDevice> touch_mouse_device;
+    std::unique_ptr<Input::TouchDevice> touch_udp_device;
+    std::unique_ptr<Input::TouchDevice> touch_btn_device;
+    std::array<size_t, MAX_FINGERS> mouse_finger_id;
+    std::array<size_t, MAX_FINGERS> keyboard_finger_id;
+    std::array<size_t, MAX_FINGERS> udp_finger_id;
+    std::array<Finger, MAX_POINTS> fingers;
+    GestureProperties last_gesture;
 };
 } // namespace Service::HID

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

@@ -413,12 +413,16 @@ void Controller_NPad::RequestPadStateUpdate(u32 npad_id) {
         lstick_entry.y = static_cast<s32>(stick_l_y_f * HID_JOYSTICK_MAX);
     }
 
-    if (controller_type == NPadControllerType::JoyLeft ||
-        controller_type == NPadControllerType::JoyRight) {
+    if (controller_type == NPadControllerType::JoyLeft) {
         pad_state.left_sl.Assign(button_state[SL - BUTTON_HID_BEGIN]->GetStatus());
         pad_state.left_sr.Assign(button_state[SR - BUTTON_HID_BEGIN]->GetStatus());
     }
 
+    if (controller_type == NPadControllerType::JoyRight) {
+        pad_state.right_sl.Assign(button_state[SL - BUTTON_HID_BEGIN]->GetStatus());
+        pad_state.right_sr.Assign(button_state[SR - BUTTON_HID_BEGIN]->GetStatus());
+    }
+
     if (controller_type == NPadControllerType::GameCube) {
         trigger_entry.l_analog = static_cast<s32>(
             button_state[ZL - BUTTON_HID_BEGIN]->GetStatus() ? HID_TRIGGER_MAX : 0);

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

@@ -5,6 +5,7 @@
 #include <algorithm>
 #include <cstring>
 #include "common/common_types.h"
+#include "common/logging/log.h"
 #include "core/core_timing.h"
 #include "core/frontend/emu_window.h"
 #include "core/frontend/input.h"
@@ -118,6 +119,10 @@ std::optional<std::size_t> Controller_Touchscreen::GetUnusedFingerID() const {
 std::size_t Controller_Touchscreen::UpdateTouchInputEvent(
     const std::tuple<float, float, bool>& touch_input, std::size_t finger_id) {
     const auto& [x, y, pressed] = touch_input;
+    if (finger_id > MAX_FINGERS) {
+        LOG_ERROR(Service_HID, "Invalid finger id {}", finger_id);
+        return MAX_FINGERS;
+    }
     if (pressed) {
         Attributes attribute{};
         if (finger_id == MAX_FINGERS) {

src/core/hle/service/ldn/ldn.cpp

@@ -156,7 +156,7 @@ public:
         is_initialized = true;
 
         IPC::ResponseBuilder rb{ctx, 2};
-        rb.Push(RESULT_SUCCESS);
+        rb.Push(ERROR_DISABLED);
     }
 
 private:

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

@@ -31,7 +31,7 @@ public:
      * @param output A buffer where the output data will be written to.
      * @returns The result code of the ioctl.
      */
-    virtual NvResult Ioctl1(Ioctl command, const std::vector<u8>& input,
+    virtual NvResult Ioctl1(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
                             std::vector<u8>& output) = 0;
 
     /**
@@ -42,7 +42,7 @@ public:
      * @param output A buffer where the output data will be written to.
      * @returns The result code of the ioctl.
      */
-    virtual NvResult Ioctl2(Ioctl command, const std::vector<u8>& input,
+    virtual NvResult Ioctl2(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
                             const std::vector<u8>& inline_input, std::vector<u8>& output) = 0;
 
     /**
@@ -53,8 +53,20 @@ public:
      * @param inline_output A buffer where the inlined output data will be written to.
      * @returns The result code of the ioctl.
      */
-    virtual NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
-                            std::vector<u8>& inline_output) = 0;
+    virtual NvResult Ioctl3(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                            std::vector<u8>& output, std::vector<u8>& inline_output) = 0;
+
+    /**
+     * Called once a device is openned
+     * @param fd The device fd
+     */
+    virtual void OnOpen(DeviceFD fd) = 0;
+
+    /**
+     * Called once a device is closed
+     * @param fd The device fd
+     */
+    virtual void OnClose(DeviceFD fd) = 0;
 
 protected:
     Core::System& system;

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

@@ -18,24 +18,27 @@ nvdisp_disp0::nvdisp_disp0(Core::System& system, std::shared_ptr<nvmap> nvmap_de
     : nvdevice(system), nvmap_dev(std::move(nvmap_dev)) {}
 nvdisp_disp0 ::~nvdisp_disp0() = default;
 
-NvResult nvdisp_disp0::Ioctl1(Ioctl command, const std::vector<u8>& input,
+NvResult nvdisp_disp0::Ioctl1(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
                               std::vector<u8>& output) {
     UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
     return NvResult::NotImplemented;
 }
 
-NvResult nvdisp_disp0::Ioctl2(Ioctl command, const std::vector<u8>& input,
+NvResult nvdisp_disp0::Ioctl2(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
                               const std::vector<u8>& inline_input, std::vector<u8>& output) {
     UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
     return NvResult::NotImplemented;
 }
 
-NvResult nvdisp_disp0::Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
-                              std::vector<u8>& inline_output) {
+NvResult nvdisp_disp0::Ioctl3(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                              std::vector<u8>& output, std::vector<u8>& inline_output) {
     UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
     return NvResult::NotImplemented;
 }
 
+void nvdisp_disp0::OnOpen(DeviceFD fd) {}
+void nvdisp_disp0::OnClose(DeviceFD fd) {}
+
 void nvdisp_disp0::flip(u32 buffer_handle, u32 offset, u32 format, u32 width, u32 height,
                         u32 stride, NVFlinger::BufferQueue::BufferTransformFlags transform,
                         const Common::Rectangle<int>& crop_rect) {

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

@@ -20,11 +20,15 @@ public:
     explicit nvdisp_disp0(Core::System& system, std::shared_ptr<nvmap> nvmap_dev);
     ~nvdisp_disp0() override;
 
-    NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override;
-    NvResult Ioctl2(Ioctl command, const std::vector<u8>& input,
+    NvResult Ioctl1(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                    std::vector<u8>& output) override;
+    NvResult Ioctl2(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
                     const std::vector<u8>& inline_input, std::vector<u8>& output) override;
-    NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
-                    std::vector<u8>& inline_output) override;
+    NvResult Ioctl3(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                    std::vector<u8>& output, std::vector<u8>& inline_output) override;
+
+    void OnOpen(DeviceFD fd) override;
+    void OnClose(DeviceFD fd) 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,

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

@@ -21,7 +21,7 @@ nvhost_as_gpu::nvhost_as_gpu(Core::System& system, std::shared_ptr<nvmap> nvmap_
     : nvdevice(system), nvmap_dev(std::move(nvmap_dev)) {}
 nvhost_as_gpu::~nvhost_as_gpu() = default;
 
-NvResult nvhost_as_gpu::Ioctl1(Ioctl command, const std::vector<u8>& input,
+NvResult nvhost_as_gpu::Ioctl1(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
                                std::vector<u8>& output) {
     switch (command.group) {
     case 'A':
@@ -39,7 +39,7 @@ NvResult nvhost_as_gpu::Ioctl1(Ioctl command, const std::vector<u8>& input,
         case 0x8:
             return GetVARegions(input, output);
         case 0x9:
-            return InitalizeEx(input, output);
+            return AllocAsEx(input, output);
         case 0x14:
             return Remap(input, output);
         default:
@@ -54,14 +54,14 @@ NvResult nvhost_as_gpu::Ioctl1(Ioctl command, const std::vector<u8>& input,
     return NvResult::NotImplemented;
 }
 
-NvResult nvhost_as_gpu::Ioctl2(Ioctl command, const std::vector<u8>& input,
+NvResult nvhost_as_gpu::Ioctl2(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
                                const std::vector<u8>& inline_input, std::vector<u8>& output) {
     UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
     return NvResult::NotImplemented;
 }
 
-NvResult nvhost_as_gpu::Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
-                               std::vector<u8>& inline_output) {
+NvResult nvhost_as_gpu::Ioctl3(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                               std::vector<u8>& output, std::vector<u8>& inline_output) {
     switch (command.group) {
     case 'A':
         switch (command.cmd) {
@@ -78,11 +78,19 @@ NvResult nvhost_as_gpu::Ioctl3(Ioctl command, const std::vector<u8>& input, std:
     return NvResult::NotImplemented;
 }
 
-NvResult nvhost_as_gpu::InitalizeEx(const std::vector<u8>& input, std::vector<u8>& output) {
-    IoctlInitalizeEx params{};
+void nvhost_as_gpu::OnOpen(DeviceFD fd) {}
+void nvhost_as_gpu::OnClose(DeviceFD fd) {}
+
+NvResult nvhost_as_gpu::AllocAsEx(const std::vector<u8>& input, std::vector<u8>& output) {
+    IoctlAllocAsEx params{};
     std::memcpy(&params, input.data(), input.size());
 
     LOG_WARNING(Service_NVDRV, "(STUBBED) called, big_page_size=0x{:X}", params.big_page_size);
+    if (params.big_page_size == 0) {
+        params.big_page_size = DEFAULT_BIG_PAGE_SIZE;
+    }
+
+    big_page_size = params.big_page_size;
 
     return NvResult::Success;
 }
@@ -276,13 +284,18 @@ NvResult nvhost_as_gpu::GetVARegions(const std::vector<u8>& input, std::vector<u
                 params.buf_size);
 
     params.buf_size = 0x30;
-    params.regions[0].offset = 0x04000000;
-    params.regions[0].page_size = 0x1000;
-    params.regions[0].pages = 0x3fbfff;
 
-    params.regions[1].offset = 0x04000000;
-    params.regions[1].page_size = 0x10000;
-    params.regions[1].pages = 0x1bffff;
+    params.small = IoctlVaRegion{
+        .offset = 0x04000000,
+        .page_size = DEFAULT_SMALL_PAGE_SIZE,
+        .pages = 0x3fbfff,
+    };
+
+    params.big = IoctlVaRegion{
+        .offset = 0x04000000,
+        .page_size = big_page_size,
+        .pages = 0x1bffff,
+    };
 
     // TODO(ogniK): This probably can stay stubbed but should add support way way later
 
@@ -299,18 +312,25 @@ NvResult nvhost_as_gpu::GetVARegions(const std::vector<u8>& input, std::vector<u
                 params.buf_size);
 
     params.buf_size = 0x30;
-    params.regions[0].offset = 0x04000000;
-    params.regions[0].page_size = 0x1000;
-    params.regions[0].pages = 0x3fbfff;
 
-    params.regions[1].offset = 0x04000000;
-    params.regions[1].page_size = 0x10000;
-    params.regions[1].pages = 0x1bffff;
+    params.small = IoctlVaRegion{
+        .offset = 0x04000000,
+        .page_size = 0x1000,
+        .pages = 0x3fbfff,
+    };
+
+    params.big = IoctlVaRegion{
+        .offset = 0x04000000,
+        .page_size = big_page_size,
+        .pages = 0x1bffff,
+    };
 
     // TODO(ogniK): This probably can stay stubbed but should add support way way later
 
     std::memcpy(output.data(), &params, output.size());
-    std::memcpy(inline_output.data(), &params.regions, inline_output.size());
+    std::memcpy(inline_output.data(), &params.small, sizeof(IoctlVaRegion));
+    std::memcpy(inline_output.data() + sizeof(IoctlVaRegion), &params.big, sizeof(IoctlVaRegion));
+
     return NvResult::Success;
 }
 

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

@@ -16,6 +16,9 @@
 
 namespace Service::Nvidia::Devices {
 
+constexpr u32 DEFAULT_BIG_PAGE_SIZE = 1 << 16;
+constexpr u32 DEFAULT_SMALL_PAGE_SIZE = 1 << 12;
+
 class nvmap;
 
 enum class AddressSpaceFlags : u32 {
@@ -30,11 +33,15 @@ public:
     explicit nvhost_as_gpu(Core::System& system, std::shared_ptr<nvmap> nvmap_dev);
     ~nvhost_as_gpu() override;
 
-    NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override;
-    NvResult Ioctl2(Ioctl command, const std::vector<u8>& input,
+    NvResult Ioctl1(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                    std::vector<u8>& output) override;
+    NvResult Ioctl2(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
                     const std::vector<u8>& inline_input, std::vector<u8>& output) override;
-    NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
-                    std::vector<u8>& inline_output) override;
+    NvResult Ioctl3(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                    std::vector<u8>& output, std::vector<u8>& inline_output) override;
+
+    void OnOpen(DeviceFD fd) override;
+    void OnClose(DeviceFD fd) override;
 
 private:
     class BufferMap final {
@@ -76,16 +83,16 @@ private:
         bool is_allocated{};
     };
 
-    struct IoctlInitalizeEx {
-        u32_le big_page_size{}; // depends on GPU's available_big_page_sizes; 0=default
-        s32_le as_fd{};         // ignored; passes 0
-        u32_le flags{};         // passes 0
-        u32_le reserved{};      // ignored; passes 0
-        u64_le unk0{};
-        u64_le unk1{};
-        u64_le unk2{};
+    struct IoctlAllocAsEx {
+        u32_le flags{}; // usually passes 1
+        s32_le as_fd{}; // ignored; passes 0
+        u32_le big_page_size{};
+        u32_le reserved{}; // ignored; passes 0
+        u64_le va_range_start{};
+        u64_le va_range_end{};
+        u64_le va_range_split{};
     };
-    static_assert(sizeof(IoctlInitalizeEx) == 40, "IoctlInitalizeEx is incorrect size");
+    static_assert(sizeof(IoctlAllocAsEx) == 40, "IoctlAllocAsEx is incorrect size");
 
     struct IoctlAllocSpace {
         u32_le pages{};
@@ -149,14 +156,16 @@ private:
         u64_le buf_addr{}; // (contained output user ptr on linux, ignored)
         u32_le buf_size{}; // forced to 2*sizeof(struct va_region)
         u32_le reserved{};
-        IoctlVaRegion regions[2]{};
+        IoctlVaRegion small{};
+        IoctlVaRegion big{};
     };
     static_assert(sizeof(IoctlGetVaRegions) == 16 + sizeof(IoctlVaRegion) * 2,
                   "IoctlGetVaRegions is incorrect size");
 
     s32 channel{};
+    u32 big_page_size{DEFAULT_BIG_PAGE_SIZE};
 
-    NvResult InitalizeEx(const std::vector<u8>& input, std::vector<u8>& output);
+    NvResult AllocAsEx(const std::vector<u8>& input, std::vector<u8>& output);
     NvResult AllocateSpace(const std::vector<u8>& input, std::vector<u8>& output);
     NvResult Remap(const std::vector<u8>& input, std::vector<u8>& output);
     NvResult MapBufferEx(const std::vector<u8>& input, std::vector<u8>& output);

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

@@ -20,7 +20,8 @@ nvhost_ctrl::nvhost_ctrl(Core::System& system, EventInterface& events_interface,
     : nvdevice(system), events_interface{events_interface}, syncpoint_manager{syncpoint_manager} {}
 nvhost_ctrl::~nvhost_ctrl() = default;
 
-NvResult nvhost_ctrl::Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) {
+NvResult nvhost_ctrl::Ioctl1(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                             std::vector<u8>& output) {
     switch (command.group) {
     case 0x0:
         switch (command.cmd) {
@@ -46,18 +47,21 @@ NvResult nvhost_ctrl::Ioctl1(Ioctl command, const std::vector<u8>& input, std::v
     return NvResult::NotImplemented;
 }
 
-NvResult nvhost_ctrl::Ioctl2(Ioctl command, const std::vector<u8>& input,
+NvResult nvhost_ctrl::Ioctl2(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
                              const std::vector<u8>& inline_input, std::vector<u8>& output) {
     UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
     return NvResult::NotImplemented;
 }
 
-NvResult nvhost_ctrl::Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
-                             std::vector<u8>& inline_outpu) {
+NvResult nvhost_ctrl::Ioctl3(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                             std::vector<u8>& output, std::vector<u8>& inline_outpu) {
     UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
     return NvResult::NotImplemented;
 }
 
+void nvhost_ctrl::OnOpen(DeviceFD fd) {}
+void nvhost_ctrl::OnClose(DeviceFD fd) {}
+
 NvResult nvhost_ctrl::NvOsGetConfigU32(const std::vector<u8>& input, std::vector<u8>& output) {
     IocGetConfigParams params{};
     std::memcpy(&params, input.data(), sizeof(params));

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

@@ -18,11 +18,15 @@ public:
                          SyncpointManager& syncpoint_manager);
     ~nvhost_ctrl() override;
 
-    NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override;
-    NvResult Ioctl2(Ioctl command, const std::vector<u8>& input,
+    NvResult Ioctl1(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                    std::vector<u8>& output) override;
+    NvResult Ioctl2(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
                     const std::vector<u8>& inline_input, std::vector<u8>& output) override;
-    NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
-                    std::vector<u8>& inline_output) override;
+    NvResult Ioctl3(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                    std::vector<u8>& output, std::vector<u8>& inline_output) override;
+
+    void OnOpen(DeviceFD fd) override;
+    void OnClose(DeviceFD fd) override;
 
 private:
     struct IocSyncptReadParams {

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

@@ -15,7 +15,7 @@ namespace Service::Nvidia::Devices {
 nvhost_ctrl_gpu::nvhost_ctrl_gpu(Core::System& system) : nvdevice(system) {}
 nvhost_ctrl_gpu::~nvhost_ctrl_gpu() = default;
 
-NvResult nvhost_ctrl_gpu::Ioctl1(Ioctl command, const std::vector<u8>& input,
+NvResult nvhost_ctrl_gpu::Ioctl1(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
                                  std::vector<u8>& output) {
     switch (command.group) {
     case 'G':
@@ -47,13 +47,13 @@ NvResult nvhost_ctrl_gpu::Ioctl1(Ioctl command, const std::vector<u8>& input,
     return NvResult::NotImplemented;
 }
 
-NvResult nvhost_ctrl_gpu::Ioctl2(Ioctl command, const std::vector<u8>& input,
+NvResult nvhost_ctrl_gpu::Ioctl2(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
                                  const std::vector<u8>& inline_input, std::vector<u8>& output) {
     UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
     return NvResult::NotImplemented;
 }
 
-NvResult nvhost_ctrl_gpu::Ioctl3(Ioctl command, const std::vector<u8>& input,
+NvResult nvhost_ctrl_gpu::Ioctl3(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
                                  std::vector<u8>& output, std::vector<u8>& inline_output) {
     switch (command.group) {
     case 'G':
@@ -73,6 +73,9 @@ NvResult nvhost_ctrl_gpu::Ioctl3(Ioctl command, const std::vector<u8>& input,
     return NvResult::NotImplemented;
 }
 
+void nvhost_ctrl_gpu::OnOpen(DeviceFD fd) {}
+void nvhost_ctrl_gpu::OnClose(DeviceFD fd) {}
+
 NvResult nvhost_ctrl_gpu::GetCharacteristics(const std::vector<u8>& input,
                                              std::vector<u8>& output) {
     LOG_DEBUG(Service_NVDRV, "called");

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

@@ -16,11 +16,15 @@ public:
     explicit nvhost_ctrl_gpu(Core::System& system);
     ~nvhost_ctrl_gpu() override;
 
-    NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override;
-    NvResult Ioctl2(Ioctl command, const std::vector<u8>& input,
+    NvResult Ioctl1(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                    std::vector<u8>& output) override;
+    NvResult Ioctl2(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
                     const std::vector<u8>& inline_input, std::vector<u8>& output) override;
-    NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
-                    std::vector<u8>& inline_output) override;
+    NvResult Ioctl3(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                    std::vector<u8>& output, std::vector<u8>& inline_output) override;
+
+    void OnOpen(DeviceFD fd) override;
+    void OnClose(DeviceFD fd) override;
 
 private:
     struct IoctlGpuCharacteristics {

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

@@ -23,7 +23,8 @@ nvhost_gpu::nvhost_gpu(Core::System& system, std::shared_ptr<nvmap> nvmap_dev,
 
 nvhost_gpu::~nvhost_gpu() = default;
 
-NvResult nvhost_gpu::Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) {
+NvResult nvhost_gpu::Ioctl1(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                            std::vector<u8>& output) {
     switch (command.group) {
     case 0x0:
         switch (command.cmd) {
@@ -74,7 +75,7 @@ NvResult nvhost_gpu::Ioctl1(Ioctl command, const std::vector<u8>& input, std::ve
     return NvResult::NotImplemented;
 };
 
-NvResult nvhost_gpu::Ioctl2(Ioctl command, const std::vector<u8>& input,
+NvResult nvhost_gpu::Ioctl2(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
                             const std::vector<u8>& inline_input, std::vector<u8>& output) {
     switch (command.group) {
     case 'H':
@@ -88,12 +89,15 @@ NvResult nvhost_gpu::Ioctl2(Ioctl command, const std::vector<u8>& input,
     return NvResult::NotImplemented;
 }
 
-NvResult nvhost_gpu::Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
-                            std::vector<u8>& inline_output) {
+NvResult nvhost_gpu::Ioctl3(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                            std::vector<u8>& output, std::vector<u8>& inline_output) {
     UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
     return NvResult::NotImplemented;
 }
 
+void nvhost_gpu::OnOpen(DeviceFD fd) {}
+void nvhost_gpu::OnClose(DeviceFD fd) {}
+
 NvResult nvhost_gpu::SetNVMAPfd(const std::vector<u8>& input, std::vector<u8>& output) {
     IoctlSetNvmapFD params{};
     std::memcpy(&params, input.data(), input.size());

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

@@ -26,11 +26,15 @@ public:
                         SyncpointManager& syncpoint_manager);
     ~nvhost_gpu() override;
 
-    NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override;
-    NvResult Ioctl2(Ioctl command, const std::vector<u8>& input,
+    NvResult Ioctl1(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                    std::vector<u8>& output) override;
+    NvResult Ioctl2(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
                     const std::vector<u8>& inline_input, std::vector<u8>& output) override;
-    NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
-                    std::vector<u8>& inline_output) override;
+    NvResult Ioctl3(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                    std::vector<u8>& output, std::vector<u8>& inline_output) override;
+
+    void OnOpen(DeviceFD fd) override;
+    void OnClose(DeviceFD fd) override;
 
 private:
     enum class CtxObjects : u32_le {

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

@@ -16,7 +16,7 @@ nvhost_nvdec::nvhost_nvdec(Core::System& system, std::shared_ptr<nvmap> nvmap_de
     : nvhost_nvdec_common(system, std::move(nvmap_dev), syncpoint_manager) {}
 nvhost_nvdec::~nvhost_nvdec() = default;
 
-NvResult nvhost_nvdec::Ioctl1(Ioctl command, const std::vector<u8>& input,
+NvResult nvhost_nvdec::Ioctl1(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
                               std::vector<u8>& output) {
     switch (command.group) {
     case 0x0:
@@ -57,16 +57,19 @@ NvResult nvhost_nvdec::Ioctl1(Ioctl command, const std::vector<u8>& input,
     return NvResult::NotImplemented;
 }
 
-NvResult nvhost_nvdec::Ioctl2(Ioctl command, const std::vector<u8>& input,
+NvResult nvhost_nvdec::Ioctl2(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
                               const std::vector<u8>& inline_input, std::vector<u8>& output) {
     UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
     return NvResult::NotImplemented;
 }
 
-NvResult nvhost_nvdec::Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
-                              std::vector<u8>& inline_output) {
+NvResult nvhost_nvdec::Ioctl3(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                              std::vector<u8>& output, std::vector<u8>& inline_output) {
     UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
     return NvResult::NotImplemented;
 }
 
+void nvhost_nvdec::OnOpen(DeviceFD fd) {}
+void nvhost_nvdec::OnClose(DeviceFD fd) {}
+
 } // namespace Service::Nvidia::Devices

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

@@ -15,11 +15,15 @@ public:
                           SyncpointManager& syncpoint_manager);
     ~nvhost_nvdec() override;
 
-    NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override;
-    NvResult Ioctl2(Ioctl command, const std::vector<u8>& input,
+    NvResult Ioctl1(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                    std::vector<u8>& output) override;
+    NvResult Ioctl2(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
                     const std::vector<u8>& inline_input, std::vector<u8>& output) override;
-    NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
-                    std::vector<u8>& inline_output) override;
+    NvResult Ioctl3(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                    std::vector<u8>& output, std::vector<u8>& inline_output) override;
+
+    void OnOpen(DeviceFD fd) override;
+    void OnClose(DeviceFD fd) override;
 };
 
 } // namespace Service::Nvidia::Devices

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

@@ -13,7 +13,7 @@ namespace Service::Nvidia::Devices {
 nvhost_nvjpg::nvhost_nvjpg(Core::System& system) : nvdevice(system) {}
 nvhost_nvjpg::~nvhost_nvjpg() = default;
 
-NvResult nvhost_nvjpg::Ioctl1(Ioctl command, const std::vector<u8>& input,
+NvResult nvhost_nvjpg::Ioctl1(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
                               std::vector<u8>& output) {
     switch (command.group) {
     case 'H':
@@ -32,18 +32,21 @@ NvResult nvhost_nvjpg::Ioctl1(Ioctl command, const std::vector<u8>& input,
     return NvResult::NotImplemented;
 }
 
-NvResult nvhost_nvjpg::Ioctl2(Ioctl command, const std::vector<u8>& input,
+NvResult nvhost_nvjpg::Ioctl2(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
                               const std::vector<u8>& inline_input, std::vector<u8>& output) {
     UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
     return NvResult::NotImplemented;
 }
 
-NvResult nvhost_nvjpg::Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
-                              std::vector<u8>& inline_output) {
+NvResult nvhost_nvjpg::Ioctl3(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                              std::vector<u8>& output, std::vector<u8>& inline_output) {
     UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
     return NvResult::NotImplemented;
 }
 
+void nvhost_nvjpg::OnOpen(DeviceFD fd) {}
+void nvhost_nvjpg::OnClose(DeviceFD fd) {}
+
 NvResult nvhost_nvjpg::SetNVMAPfd(const std::vector<u8>& input, std::vector<u8>& output) {
     IoctlSetNvmapFD params{};
     std::memcpy(&params, input.data(), input.size());

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

@@ -16,11 +16,15 @@ public:
     explicit nvhost_nvjpg(Core::System& system);
     ~nvhost_nvjpg() override;
 
-    NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override;
-    NvResult Ioctl2(Ioctl command, const std::vector<u8>& input,
+    NvResult Ioctl1(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                    std::vector<u8>& output) override;
+    NvResult Ioctl2(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
                     const std::vector<u8>& inline_input, std::vector<u8>& output) override;
-    NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
-                    std::vector<u8>& inline_output) override;
+    NvResult Ioctl3(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                    std::vector<u8>& output, std::vector<u8>& inline_output) override;
+
+    void OnOpen(DeviceFD fd) override;
+    void OnClose(DeviceFD fd) override;
 
 private:
     struct IoctlSetNvmapFD {

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

@@ -16,7 +16,8 @@ nvhost_vic::nvhost_vic(Core::System& system, std::shared_ptr<nvmap> nvmap_dev,
 
 nvhost_vic::~nvhost_vic() = default;
 
-NvResult nvhost_vic::Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) {
+NvResult nvhost_vic::Ioctl1(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                            std::vector<u8>& output) {
     switch (command.group) {
     case 0x0:
         switch (command.cmd) {
@@ -55,16 +56,19 @@ NvResult nvhost_vic::Ioctl1(Ioctl command, const std::vector<u8>& input, std::ve
     return NvResult::NotImplemented;
 }
 
-NvResult nvhost_vic::Ioctl2(Ioctl command, const std::vector<u8>& input,
+NvResult nvhost_vic::Ioctl2(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
                             const std::vector<u8>& inline_input, std::vector<u8>& output) {
     UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
     return NvResult::NotImplemented;
 }
 
-NvResult nvhost_vic::Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
-                            std::vector<u8>& inline_output) {
+NvResult nvhost_vic::Ioctl3(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                            std::vector<u8>& output, std::vector<u8>& inline_output) {
     UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
     return NvResult::NotImplemented;
 }
 
+void nvhost_vic::OnOpen(DeviceFD fd) {}
+void nvhost_vic::OnClose(DeviceFD fd) {}
+
 } // namespace Service::Nvidia::Devices

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

@@ -14,10 +14,14 @@ public:
                         SyncpointManager& syncpoint_manager);
     ~nvhost_vic();
 
-    NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override;
-    NvResult Ioctl2(Ioctl command, const std::vector<u8>& input,
+    NvResult Ioctl1(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                    std::vector<u8>& output) override;
+    NvResult Ioctl2(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
                     const std::vector<u8>& inline_input, std::vector<u8>& output) override;
-    NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
-                    std::vector<u8>& inline_output) override;
+    NvResult Ioctl3(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                    std::vector<u8>& output, std::vector<u8>& inline_output) override;
+
+    void OnOpen(DeviceFD fd) override;
+    void OnClose(DeviceFD fd) override;
 };
 } // namespace Service::Nvidia::Devices

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

@@ -19,7 +19,8 @@ nvmap::nvmap(Core::System& system) : nvdevice(system) {
 
 nvmap::~nvmap() = default;
 
-NvResult nvmap::Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) {
+NvResult nvmap::Ioctl1(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                       std::vector<u8>& output) {
     switch (command.group) {
     case 0x1:
         switch (command.cmd) {
@@ -47,18 +48,21 @@ NvResult nvmap::Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<
     return NvResult::NotImplemented;
 }
 
-NvResult nvmap::Ioctl2(Ioctl command, const std::vector<u8>& input,
+NvResult nvmap::Ioctl2(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
                        const std::vector<u8>& inline_input, std::vector<u8>& output) {
     UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
     return NvResult::NotImplemented;
 }
 
-NvResult nvmap::Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
-                       std::vector<u8>& inline_output) {
+NvResult nvmap::Ioctl3(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                       std::vector<u8>& output, std::vector<u8>& inline_output) {
     UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw);
     return NvResult::NotImplemented;
 }
 
+void nvmap::OnOpen(DeviceFD fd) {}
+void nvmap::OnClose(DeviceFD fd) {}
+
 VAddr nvmap::GetObjectAddress(u32 handle) const {
     auto object = GetObject(handle);
     ASSERT(object);

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

@@ -19,11 +19,15 @@ public:
     explicit nvmap(Core::System& system);
     ~nvmap() override;
 
-    NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override;
-    NvResult Ioctl2(Ioctl command, const std::vector<u8>& input,
+    NvResult Ioctl1(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                    std::vector<u8>& output) override;
+    NvResult Ioctl2(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
                     const std::vector<u8>& inline_input, std::vector<u8>& output) override;
-    NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output,
-                    std::vector<u8>& inline_output) override;
+    NvResult Ioctl3(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
+                    std::vector<u8>& output, std::vector<u8>& inline_output) override;
+
+    void OnOpen(DeviceFD fd) override;
+    void OnClose(DeviceFD fd) override;
 
     /// Returns the allocated address of an nvmap object given its handle.
     VAddr GetObjectAddress(u32 handle) const;

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

@@ -89,6 +89,8 @@ DeviceFD Module::Open(const std::string& device_name) {
     auto device = devices[device_name];
     const DeviceFD fd = next_fd++;
 
+    device->OnOpen(fd);
+
     open_files[fd] = std::move(device);
 
     return fd;
@@ -108,7 +110,7 @@ NvResult Module::Ioctl1(DeviceFD fd, Ioctl command, const std::vector<u8>& input
         return NvResult::NotImplemented;
     }
 
-    return itr->second->Ioctl1(command, input, output);
+    return itr->second->Ioctl1(fd, command, input, output);
 }
 
 NvResult Module::Ioctl2(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
@@ -125,7 +127,7 @@ NvResult Module::Ioctl2(DeviceFD fd, Ioctl command, const std::vector<u8>& input
         return NvResult::NotImplemented;
     }
 
-    return itr->second->Ioctl2(command, input, inline_input, output);
+    return itr->second->Ioctl2(fd, command, input, inline_input, output);
 }
 
 NvResult Module::Ioctl3(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
@@ -142,7 +144,7 @@ NvResult Module::Ioctl3(DeviceFD fd, Ioctl command, const std::vector<u8>& input
         return NvResult::NotImplemented;
     }
 
-    return itr->second->Ioctl3(command, input, output, inline_output);
+    return itr->second->Ioctl3(fd, command, input, output, inline_output);
 }
 
 NvResult Module::Close(DeviceFD fd) {
@@ -158,6 +160,8 @@ NvResult Module::Close(DeviceFD fd) {
         return NvResult::NotImplemented;
     }
 
+    itr->second->OnClose(fd);
+
     open_files.erase(itr);
 
     return NvResult::Success;

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

@@ -42,7 +42,9 @@ void BSD::PollWork::Execute(BSD* bsd) {
 }
 
 void BSD::PollWork::Response(Kernel::HLERequestContext& ctx) {
-    ctx.WriteBuffer(write_buffer);
+    if (write_buffer.size() > 0) {
+        ctx.WriteBuffer(write_buffer);
+    }
 
     IPC::ResponseBuilder rb{ctx, 4};
     rb.Push(RESULT_SUCCESS);
@@ -55,7 +57,9 @@ void BSD::AcceptWork::Execute(BSD* bsd) {
 }
 
 void BSD::AcceptWork::Response(Kernel::HLERequestContext& ctx) {
-    ctx.WriteBuffer(write_buffer);
+    if (write_buffer.size() > 0) {
+        ctx.WriteBuffer(write_buffer);
+    }
 
     IPC::ResponseBuilder rb{ctx, 5};
     rb.Push(RESULT_SUCCESS);

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

@@ -43,6 +43,11 @@ void InstallInterfaces(SM::ServiceManager& service_manager, Core::System& system
     auto module = std::make_shared<Module>();
     std::make_shared<CSRNG>(system, module)->InstallAsService(service_manager);
     std::make_shared<SPL>(system, module)->InstallAsService(service_manager);
+    std::make_shared<SPL_MIG>(system, module)->InstallAsService(service_manager);
+    std::make_shared<SPL_FS>(system, module)->InstallAsService(service_manager);
+    std::make_shared<SPL_SSL>(system, module)->InstallAsService(service_manager);
+    std::make_shared<SPL_ES>(system, module)->InstallAsService(service_manager);
+    std::make_shared<SPL_MANU>(system, module)->InstallAsService(service_manager);
 }
 
 } // namespace Service::SPL

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

@@ -8,6 +8,79 @@ namespace Service::SPL {
 
 SPL::SPL(Core::System& system_, std::shared_ptr<Module> module_)
     : Interface(system_, std::move(module_), "spl:") {
+    // clang-format off
+    static const FunctionInfo functions[] = {
+        {0, nullptr, "GetConfig"},
+        {1, nullptr, "ModularExponentiate"},
+        {5, nullptr, "SetConfig"},
+        {7, &SPL::GetRandomBytes, "GetRandomBytes"},
+        {11, nullptr, "IsDevelopment"},
+        {24, nullptr, "SetBootReason"},
+        {25, nullptr, "GetBootReason"},
+    };
+    // clang-format on
+
+    RegisterHandlers(functions);
+}
+
+SPL_MIG::SPL_MIG(Core::System& system_, std::shared_ptr<Module> module_)
+    : Interface(system_, std::move(module_), "spl:mig") {
+    // clang-format off
+    static const FunctionInfo functions[] = {
+        {0, nullptr, "GetConfig"},
+        {1, nullptr, "ModularExponentiate"},
+        {2, nullptr, "GenerateAesKek"},
+        {3, nullptr, "LoadAesKey"},
+        {4, nullptr, "GenerateAesKey"},
+        {5, nullptr, "SetConfig"},
+        {7, &SPL::GetRandomBytes, "GenerateRandomBytes"},
+        {11, nullptr, "IsDevelopment"},
+        {14, nullptr, "DecryptAesKey"},
+        {15, nullptr, "CryptAesCtr"},
+        {16, nullptr, "ComputeCmac"},
+        {21, nullptr, "AllocateAesKeyslot"},
+        {22, nullptr, "DeallocateAesKeySlot"},
+        {23, nullptr, "GetAesKeyslotAvailableEvent"},
+        {24, nullptr, "SetBootReason"},
+        {25, nullptr, "GetBootReason"},
+    };
+    // clang-format on
+
+    RegisterHandlers(functions);
+}
+
+SPL_FS::SPL_FS(Core::System& system_, std::shared_ptr<Module> module_)
+    : Interface(system_, std::move(module_), "spl:fs") {
+    // clang-format off
+    static const FunctionInfo functions[] = {
+        {0, nullptr, "GetConfig"},
+        {1, nullptr, "ModularExponentiate"},
+        {2, nullptr, "GenerateAesKek"},
+        {3, nullptr, "LoadAesKey"},
+        {4, nullptr, "GenerateAesKey"},
+        {5, nullptr, "SetConfig"},
+        {7, &SPL::GetRandomBytes, "GenerateRandomBytes"},
+        {11, nullptr, "IsDevelopment"},
+        {12, nullptr, "GenerateSpecificAesKey"},
+        {14, nullptr, "DecryptAesKey"},
+        {15, nullptr, "CryptAesCtr"},
+        {16, nullptr, "ComputeCmac"},
+        {19, nullptr, "LoadTitleKey"},
+        {21, nullptr, "AllocateAesKeyslot"},
+        {22, nullptr, "DeallocateAesKeySlot"},
+        {23, nullptr, "GetAesKeyslotAvailableEvent"},
+        {24, nullptr, "SetBootReason"},
+        {25, nullptr, "GetBootReason"},
+        {31, nullptr, "GetPackage2Hash"},
+    };
+    // clang-format on
+
+    RegisterHandlers(functions);
+}
+
+SPL_SSL::SPL_SSL(Core::System& system_, std::shared_ptr<Module> module_)
+    : Interface(system_, std::move(module_), "spl:ssl") {
+    // clang-format off
     static const FunctionInfo functions[] = {
         {0, nullptr, "GetConfig"},
         {1, nullptr, "ModularExponentiate"},
@@ -16,18 +89,11 @@ SPL::SPL(Core::System& system_, std::shared_ptr<Module> module_)
         {4, nullptr, "GenerateAesKey"},
         {5, nullptr, "SetConfig"},
         {7, &SPL::GetRandomBytes, "GetRandomBytes"},
-        {9, nullptr, "ImportLotusKey"},
-        {10, nullptr, "DecryptLotusMessage"},
         {11, nullptr, "IsDevelopment"},
-        {12, nullptr, "GenerateSpecificAesKey"},
         {13, nullptr, "DecryptDeviceUniqueData"},
         {14, nullptr, "DecryptAesKey"},
         {15, nullptr, "CryptAesCtr"},
         {16, nullptr, "ComputeCmac"},
-        {17, nullptr, "ImportEsKey"},
-        {18, nullptr, "UnwrapTitleKey"},
-        {19, nullptr, "LoadTitleKey"},
-        {20, nullptr, "PrepareEsCommonKey"},
         {21, nullptr, "AllocateAesKeyslot"},
         {22, nullptr, "DeallocateAesKeySlot"},
         {23, nullptr, "GetAesKeyslotAvailableEvent"},
@@ -35,15 +101,83 @@ SPL::SPL(Core::System& system_, std::shared_ptr<Module> module_)
         {25, nullptr, "GetBootReason"},
         {26, nullptr, "DecryptAndStoreSslClientCertKey"},
         {27, nullptr, "ModularExponentiateWithSslClientCertKey"},
+    };
+    // clang-format on
+
+    RegisterHandlers(functions);
+}
+
+SPL_ES::SPL_ES(Core::System& system_, std::shared_ptr<Module> module_)
+    : Interface(system_, std::move(module_), "spl:es") {
+    // clang-format off
+    static const FunctionInfo functions[] = {
+        {0, nullptr, "GetConfig"},
+        {1, nullptr, "ModularExponentiate"},
+        {2, nullptr, "GenerateAesKek"},
+        {3, nullptr, "LoadAesKey"},
+        {4, nullptr, "GenerateAesKey"},
+        {5, nullptr, "SetConfig"},
+        {7, &SPL::GetRandomBytes, "GenerateRandomBytes"},
+        {11, nullptr, "IsDevelopment"},
+        {13, nullptr, "DecryptDeviceUniqueData"},
+        {14, nullptr, "DecryptAesKey"},
+        {15, nullptr, "CryptAesCtr"},
+        {16, nullptr, "ComputeCmac"},
+        {18, nullptr, "UnwrapTitleKey"},
+        {20, nullptr, "PrepareEsCommonKey"},
+        {21, nullptr, "AllocateAesKeyslot"},
+        {22, nullptr, "DeallocateAesKeySlot"},
+        {23, nullptr, "GetAesKeyslotAvailableEvent"},
+        {24, nullptr, "SetBootReason"},
+        {25, nullptr, "GetBootReason"},
         {28, nullptr, "DecryptAndStoreDrmDeviceCertKey"},
         {29, nullptr, "ModularExponentiateWithDrmDeviceCertKey"},
-        {30, nullptr, "ReencryptDeviceUniqueData "},
-        {31, nullptr, "PrepareEsArchiveKey"}, // This is also GetPackage2Hash?
+        {31, nullptr, "PrepareEsArchiveKey"},
         {32, nullptr, "LoadPreparedAesKey"},
     };
+    // clang-format on
+
+    RegisterHandlers(functions);
+}
+
+SPL_MANU::SPL_MANU(Core::System& system_, std::shared_ptr<Module> module_)
+    : Interface(system_, std::move(module_), "spl:manu") {
+    // clang-format off
+    static const FunctionInfo functions[] = {
+        {0, nullptr, "GetConfig"},
+        {1, nullptr, "ModularExponentiate"},
+        {2, nullptr, "GenerateAesKek"},
+        {3, nullptr, "LoadAesKey"},
+        {4, nullptr, "GenerateAesKey"},
+        {5, nullptr, "SetConfig"},
+        {7, &SPL::GetRandomBytes, "GetRandomBytes"},
+        {11, nullptr, "IsDevelopment"},
+        {13, nullptr, "DecryptDeviceUniqueData"},
+        {14, nullptr, "DecryptAesKey"},
+        {15, nullptr, "CryptAesCtr"},
+        {16, nullptr, "ComputeCmac"},
+        {21, nullptr, "AllocateAesKeyslot"},
+        {22, nullptr, "DeallocateAesKeySlot"},
+        {23, nullptr, "GetAesKeyslotAvailableEvent"},
+        {24, nullptr, "SetBootReason"},
+        {25, nullptr, "GetBootReason"},
+        {30, nullptr, "ReencryptDeviceUniqueData"},
+    };
+    // clang-format on
+
     RegisterHandlers(functions);
 }
 
 SPL::~SPL() = default;
 
+SPL_MIG::~SPL_MIG() = default;
+
+SPL_FS::~SPL_FS() = default;
+
+SPL_SSL::~SPL_SSL() = default;
+
+SPL_ES::~SPL_ES() = default;
+
+SPL_MANU::~SPL_MANU() = default;
+
 } // namespace Service::SPL

src/core/hle/service/spl/spl.h

@@ -18,4 +18,34 @@ public:
     ~SPL() override;
 };
 
+class SPL_MIG final : public Module::Interface {
+public:
+    explicit SPL_MIG(Core::System& system_, std::shared_ptr<Module> module_);
+    ~SPL_MIG() override;
+};
+
+class SPL_FS final : public Module::Interface {
+public:
+    explicit SPL_FS(Core::System& system_, std::shared_ptr<Module> module_);
+    ~SPL_FS() override;
+};
+
+class SPL_SSL final : public Module::Interface {
+public:
+    explicit SPL_SSL(Core::System& system_, std::shared_ptr<Module> module_);
+    ~SPL_SSL() override;
+};
+
+class SPL_ES final : public Module::Interface {
+public:
+    explicit SPL_ES(Core::System& system_, std::shared_ptr<Module> module_);
+    ~SPL_ES() override;
+};
+
+class SPL_MANU final : public Module::Interface {
+public:
+    explicit SPL_MANU(Core::System& system_, std::shared_ptr<Module> module_);
+    ~SPL_MANU() override;
+};
+
 } // namespace Service::SPL

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

@@ -140,6 +140,8 @@ ResultCode Module::Interface::GetClockSnapshotFromSystemClockContextInternal(
 
     const auto current_time_point{
         time_manager.GetStandardSteadyClockCore().GetCurrentTimePoint(system)};
+    clock_snapshot.steady_clock_time_point = current_time_point;
+
     if (const ResultCode result{Clock::ClockSnapshot::GetCurrentTime(
             clock_snapshot.user_time, current_time_point, clock_snapshot.user_context)};
         result != RESULT_SUCCESS) {
@@ -341,12 +343,18 @@ void Module::Interface::CalculateStandardUserSystemClockDifferenceByUser(
 void Module::Interface::CalculateSpanBetween(Kernel::HLERequestContext& ctx) {
     LOG_DEBUG(Service_Time, "called");
 
-    IPC::RequestParser rp{ctx};
-    const auto snapshot_a = rp.PopRaw<Clock::ClockSnapshot>();
-    const auto snapshot_b = rp.PopRaw<Clock::ClockSnapshot>();
+    Clock::ClockSnapshot snapshot_a;
+    Clock::ClockSnapshot snapshot_b;
+
+    const auto snapshot_a_data = ctx.ReadBuffer(0);
+    const auto snapshot_b_data = ctx.ReadBuffer(1);
+
+    std::memcpy(&snapshot_a, snapshot_a_data.data(), sizeof(Clock::ClockSnapshot));
+    std::memcpy(&snapshot_b, snapshot_b_data.data(), sizeof(Clock::ClockSnapshot));
 
     Clock::TimeSpanType time_span_type{};
     s64 span{};
+
     if (const ResultCode result{snapshot_a.steady_clock_time_point.GetSpanBetween(
             snapshot_b.steady_clock_time_point, span)};
         result != RESULT_SUCCESS) {

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

@@ -1217,6 +1217,32 @@ private:
         }
     }
 
+    void GetIndirectLayerImageMap(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        const auto width = rp.Pop<s64>();
+        const auto height = rp.Pop<s64>();
+        const auto indirect_layer_consumer_handle = rp.Pop<u64>();
+        const auto applet_resource_user_id = rp.Pop<u64>();
+
+        LOG_WARNING(Service_VI,
+                    "(STUBBED) called, width={}, height={}, indirect_layer_consumer_handle={}, "
+                    "applet_resource_user_id={}",
+                    width, height, indirect_layer_consumer_handle, applet_resource_user_id);
+
+        std::vector<u8> out_buffer(0x46);
+        ctx.WriteBuffer(out_buffer);
+
+        // TODO: Figure out what these are
+
+        constexpr s64 unknown_result_1 = 0;
+        constexpr s64 unknown_result_2 = 0;
+
+        IPC::ResponseBuilder rb{ctx, 6};
+        rb.Push(unknown_result_1);
+        rb.Push(unknown_result_2);
+        rb.Push(RESULT_SUCCESS);
+    }
+
     void GetIndirectLayerImageRequiredMemoryInfo(Kernel::HLERequestContext& ctx) {
         IPC::RequestParser rp{ctx};
         const auto width = rp.Pop<u64>();
@@ -1276,7 +1302,7 @@ IApplicationDisplayService::IApplicationDisplayService(Core::System& system_,
         {2031, &IApplicationDisplayService::DestroyStrayLayer, "DestroyStrayLayer"},
         {2101, &IApplicationDisplayService::SetLayerScalingMode, "SetLayerScalingMode"},
         {2102, &IApplicationDisplayService::ConvertScalingMode, "ConvertScalingMode"},
-        {2450, nullptr, "GetIndirectLayerImageMap"},
+        {2450, &IApplicationDisplayService::GetIndirectLayerImageMap, "GetIndirectLayerImageMap"},
         {2451, nullptr, "GetIndirectLayerImageCropMap"},
         {2460, &IApplicationDisplayService::GetIndirectLayerImageRequiredMemoryInfo,
          "GetIndirectLayerImageRequiredMemoryInfo"},

src/core/network/network.cpp

@@ -7,6 +7,7 @@
 #include <limits>
 #include <utility>
 #include <vector>
+#include "common/common_funcs.h"
 
 #ifdef _WIN32
 #define _WINSOCK_DEPRECATED_NO_WARNINGS // gethostname
@@ -90,15 +91,36 @@ LINGER MakeLinger(bool enable, u32 linger_value) {
     return value;
 }
 
-int LastError() {
-    return WSAGetLastError();
-}
-
 bool EnableNonBlock(SOCKET fd, bool enable) {
     u_long value = enable ? 1 : 0;
     return ioctlsocket(fd, FIONBIO, &value) != SOCKET_ERROR;
 }
 
+Errno TranslateNativeError(int e) {
+    switch (e) {
+    case WSAEBADF:
+        return Errno::BADF;
+    case WSAEINVAL:
+        return Errno::INVAL;
+    case WSAEMFILE:
+        return Errno::MFILE;
+    case WSAENOTCONN:
+        return Errno::NOTCONN;
+    case WSAEWOULDBLOCK:
+        return Errno::AGAIN;
+    case WSAECONNREFUSED:
+        return Errno::CONNREFUSED;
+    case WSAEHOSTUNREACH:
+        return Errno::HOSTUNREACH;
+    case WSAENETDOWN:
+        return Errno::NETDOWN;
+    case WSAENETUNREACH:
+        return Errno::NETUNREACH;
+    default:
+        return Errno::OTHER;
+    }
+}
+
 #elif YUZU_UNIX // ^ _WIN32 v YUZU_UNIX
 
 using SOCKET = int;
@@ -108,9 +130,6 @@ using ULONG = u64;
 constexpr SOCKET INVALID_SOCKET = -1;
 constexpr SOCKET SOCKET_ERROR = -1;
 
-constexpr int WSAEWOULDBLOCK = EAGAIN;
-constexpr int WSAENOTCONN = ENOTCONN;
-
 constexpr int SD_RECEIVE = SHUT_RD;
 constexpr int SD_SEND = SHUT_WR;
 constexpr int SD_BOTH = SHUT_RDWR;
@@ -162,10 +181,6 @@ linger MakeLinger(bool enable, u32 linger_value) {
     return value;
 }
 
-int LastError() {
-    return errno;
-}
-
 bool EnableNonBlock(int fd, bool enable) {
     int flags = fcntl(fd, F_GETFD);
     if (flags == -1) {
@@ -179,8 +194,43 @@ bool EnableNonBlock(int fd, bool enable) {
     return fcntl(fd, F_SETFD, flags) == 0;
 }
 
+Errno TranslateNativeError(int e) {
+    switch (e) {
+    case EBADF:
+        return Errno::BADF;
+    case EINVAL:
+        return Errno::INVAL;
+    case EMFILE:
+        return Errno::MFILE;
+    case ENOTCONN:
+        return Errno::NOTCONN;
+    case EAGAIN:
+        return Errno::AGAIN;
+    case ECONNREFUSED:
+        return Errno::CONNREFUSED;
+    case EHOSTUNREACH:
+        return Errno::HOSTUNREACH;
+    case ENETDOWN:
+        return Errno::NETDOWN;
+    case ENETUNREACH:
+        return Errno::NETUNREACH;
+    default:
+        return Errno::OTHER;
+    }
+}
+
 #endif
 
+Errno GetAndLogLastError() {
+#ifdef _WIN32
+    int e = WSAGetLastError();
+#else
+    int e = errno;
+#endif
+    LOG_ERROR(Network, "Socket operation error: {}", NativeErrorToString(e));
+    return TranslateNativeError(e);
+}
+
 int TranslateDomain(Domain domain) {
     switch (domain) {
     case Domain::INET:
@@ -290,9 +340,7 @@ Errno SetSockOpt(SOCKET fd, int option, T value) {
     if (result != SOCKET_ERROR) {
         return Errno::SUCCESS;
     }
-    const int ec = LastError();
-    UNREACHABLE_MSG("Unhandled host socket error={}", ec);
-    return Errno::SUCCESS;
+    return GetAndLogLastError();
 }
 
 } // Anonymous namespace
@@ -308,14 +356,12 @@ NetworkInstance::~NetworkInstance() {
 std::pair<IPv4Address, Errno> GetHostIPv4Address() {
     std::array<char, 256> name{};
     if (gethostname(name.data(), static_cast<int>(name.size()) - 1) == SOCKET_ERROR) {
-        UNIMPLEMENTED_MSG("Unhandled gethostname error");
-        return {IPv4Address{}, Errno::SUCCESS};
+        return {IPv4Address{}, GetAndLogLastError()};
     }
 
     hostent* const ent = gethostbyname(name.data());
     if (!ent) {
-        UNIMPLEMENTED_MSG("Unhandled gethostbyname error");
-        return {IPv4Address{}, Errno::SUCCESS};
+        return {IPv4Address{}, GetAndLogLastError()};
     }
     if (ent->h_addr_list == nullptr) {
         UNIMPLEMENTED_MSG("No addr provided in hostent->h_addr_list");
@@ -359,9 +405,7 @@ std::pair<s32, Errno> Poll(std::vector<PollFD>& pollfds, s32 timeout) {
 
     ASSERT(result == SOCKET_ERROR);
 
-    const int ec = LastError();
-    UNREACHABLE_MSG("Unhandled host socket error={}", ec);
-    return {-1, Errno::SUCCESS};
+    return {-1, GetAndLogLastError()};
 }
 
 Socket::~Socket() {
@@ -380,9 +424,7 @@ Errno Socket::Initialize(Domain domain, Type type, Protocol protocol) {
         return Errno::SUCCESS;
     }
 
-    const int ec = LastError();
-    UNREACHABLE_MSG("Unhandled host socket error={}", ec);
-    return Errno::SUCCESS;
+    return GetAndLogLastError();
 }
 
 std::pair<Socket::AcceptResult, Errno> Socket::Accept() {
@@ -391,9 +433,7 @@ std::pair<Socket::AcceptResult, Errno> Socket::Accept() {
     const SOCKET new_socket = accept(fd, &addr, &addrlen);
 
     if (new_socket == INVALID_SOCKET) {
-        const int ec = LastError();
-        UNREACHABLE_MSG("Unhandled host socket error={}", ec);
-        return {AcceptResult{}, Errno::SUCCESS};
+        return {AcceptResult{}, GetAndLogLastError()};
     }
 
     AcceptResult result;
@@ -412,23 +452,14 @@ Errno Socket::Connect(SockAddrIn addr_in) {
         return Errno::SUCCESS;
     }
 
-    switch (const int ec = LastError()) {
-    case WSAEWOULDBLOCK:
-        LOG_DEBUG(Service, "EAGAIN generated");
-        return Errno::AGAIN;
-    default:
-        UNREACHABLE_MSG("Unhandled host socket error={}", ec);
-        return Errno::SUCCESS;
-    }
+    return GetAndLogLastError();
 }
 
 std::pair<SockAddrIn, Errno> Socket::GetPeerName() {
     sockaddr addr;
     socklen_t addrlen = sizeof(addr);
     if (getpeername(fd, &addr, &addrlen) == SOCKET_ERROR) {
-        const int ec = LastError();
-        UNREACHABLE_MSG("Unhandled host socket error={}", ec);
-        return {SockAddrIn{}, Errno::SUCCESS};
+        return {SockAddrIn{}, GetAndLogLastError()};
     }
 
     ASSERT(addrlen == sizeof(sockaddr_in));
@@ -439,9 +470,7 @@ std::pair<SockAddrIn, Errno> Socket::GetSockName() {
     sockaddr addr;
     socklen_t addrlen = sizeof(addr);
     if (getsockname(fd, &addr, &addrlen) == SOCKET_ERROR) {
-        const int ec = LastError();
-        UNREACHABLE_MSG("Unhandled host socket error={}", ec);
-        return {SockAddrIn{}, Errno::SUCCESS};
+        return {SockAddrIn{}, GetAndLogLastError()};
     }
 
     ASSERT(addrlen == sizeof(sockaddr_in));
@@ -454,9 +483,7 @@ Errno Socket::Bind(SockAddrIn addr) {
         return Errno::SUCCESS;
     }
 
-    const int ec = LastError();
-    UNREACHABLE_MSG("Unhandled host socket error={}", ec);
-    return Errno::SUCCESS;
+    return GetAndLogLastError();
 }
 
 Errno Socket::Listen(s32 backlog) {
@@ -464,9 +491,7 @@ Errno Socket::Listen(s32 backlog) {
         return Errno::SUCCESS;
     }
 
-    const int ec = LastError();
-    UNREACHABLE_MSG("Unhandled host socket error={}", ec);
-    return Errno::SUCCESS;
+    return GetAndLogLastError();
 }
 
 Errno Socket::Shutdown(ShutdownHow how) {
@@ -489,14 +514,7 @@ Errno Socket::Shutdown(ShutdownHow how) {
         return Errno::SUCCESS;
     }
 
-    switch (const int ec = LastError()) {
-    case WSAENOTCONN:
-        LOG_ERROR(Service, "ENOTCONN generated");
-        return Errno::NOTCONN;
-    default:
-        UNREACHABLE_MSG("Unhandled host socket error={}", ec);
-        return Errno::SUCCESS;
-    }
+    return GetAndLogLastError();
 }
 
 std::pair<s32, Errno> Socket::Recv(int flags, std::vector<u8>& message) {
@@ -509,17 +527,7 @@ std::pair<s32, Errno> Socket::Recv(int flags, std::vector<u8>& message) {
         return {static_cast<s32>(result), Errno::SUCCESS};
     }
 
-    switch (const int ec = LastError()) {
-    case WSAEWOULDBLOCK:
-        LOG_DEBUG(Service, "EAGAIN generated");
-        return {-1, Errno::AGAIN};
-    case WSAENOTCONN:
-        LOG_ERROR(Service, "ENOTCONN generated");
-        return {-1, Errno::NOTCONN};
-    default:
-        UNREACHABLE_MSG("Unhandled host socket error={}", ec);
-        return {0, Errno::SUCCESS};
-    }
+    return {-1, GetAndLogLastError()};
 }
 
 std::pair<s32, Errno> Socket::RecvFrom(int flags, std::vector<u8>& message, SockAddrIn* addr) {
@@ -541,17 +549,7 @@ std::pair<s32, Errno> Socket::RecvFrom(int flags, std::vector<u8>& message, Sock
         return {static_cast<s32>(result), Errno::SUCCESS};
     }
 
-    switch (const int ec = LastError()) {
-    case WSAEWOULDBLOCK:
-        LOG_DEBUG(Service, "EAGAIN generated");
-        return {-1, Errno::AGAIN};
-    case WSAENOTCONN:
-        LOG_ERROR(Service, "ENOTCONN generated");
-        return {-1, Errno::NOTCONN};
-    default:
-        UNREACHABLE_MSG("Unhandled host socket error={}", ec);
-        return {-1, Errno::SUCCESS};
-    }
+    return {-1, GetAndLogLastError()};
 }
 
 std::pair<s32, Errno> Socket::Send(const std::vector<u8>& message, int flags) {
@@ -564,18 +562,7 @@ std::pair<s32, Errno> Socket::Send(const std::vector<u8>& message, int flags) {
         return {static_cast<s32>(result), Errno::SUCCESS};
     }
 
-    const int ec = LastError();
-    switch (ec) {
-    case WSAEWOULDBLOCK:
-        LOG_DEBUG(Service, "EAGAIN generated");
-        return {-1, Errno::AGAIN};
-    case WSAENOTCONN:
-        LOG_ERROR(Service, "ENOTCONN generated");
-        return {-1, Errno::NOTCONN};
-    default:
-        UNREACHABLE_MSG("Unhandled host socket error={}", ec);
-        return {-1, Errno::SUCCESS};
-    }
+    return {-1, GetAndLogLastError()};
 }
 
 std::pair<s32, Errno> Socket::SendTo(u32 flags, const std::vector<u8>& message,
@@ -597,9 +584,7 @@ std::pair<s32, Errno> Socket::SendTo(u32 flags, const std::vector<u8>& message,
         return {static_cast<s32>(result), Errno::SUCCESS};
     }
 
-    const int ec = LastError();
-    UNREACHABLE_MSG("Unhandled host socket error={}", ec);
-    return {-1, Errno::SUCCESS};
+    return {-1, GetAndLogLastError()};
 }
 
 Errno Socket::Close() {
@@ -642,9 +627,7 @@ Errno Socket::SetNonBlock(bool enable) {
     if (EnableNonBlock(fd, enable)) {
         return Errno::SUCCESS;
     }
-    const int ec = LastError();
-    UNREACHABLE_MSG("Unhandled host socket error={}", ec);
-    return Errno::SUCCESS;
+    return GetAndLogLastError();
 }
 
 bool Socket::IsOpened() const {

src/core/network/network.h

@@ -7,6 +7,7 @@
 #include <array>
 #include <utility>
 
+#include "common/common_funcs.h"
 #include "common/common_types.h"
 
 namespace Network {
@@ -21,6 +22,11 @@ enum class Errno {
     MFILE,
     NOTCONN,
     AGAIN,
+    CONNREFUSED,
+    HOSTUNREACH,
+    NETDOWN,
+    NETUNREACH,
+    OTHER,
 };
 
 /// Address families

src/input_common/keyboard.cpp

@@ -12,20 +12,39 @@ namespace InputCommon {
 
 class KeyButton final : public Input::ButtonDevice {
 public:
-    explicit KeyButton(std::shared_ptr<KeyButtonList> key_button_list_)
-        : key_button_list(std::move(key_button_list_)) {}
+    explicit KeyButton(std::shared_ptr<KeyButtonList> key_button_list_, bool toggle_)
+        : key_button_list(std::move(key_button_list_)), toggle(toggle_) {}
 
     ~KeyButton() override;
 
     bool GetStatus() const override {
+        if (toggle) {
+            return toggled_status.load(std::memory_order_relaxed);
+        }
         return status.load();
     }
 
+    void ToggleButton() {
+        if (lock) {
+            return;
+        }
+        lock = true;
+        const bool old_toggle_status = toggled_status.load();
+        toggled_status.store(!old_toggle_status);
+    }
+
+    void UnlockButton() {
+        lock = false;
+    }
+
     friend class KeyButtonList;
 
 private:
     std::shared_ptr<KeyButtonList> key_button_list;
     std::atomic<bool> status{false};
+    std::atomic<bool> toggled_status{false};
+    bool lock{false};
+    const bool toggle;
 };
 
 struct KeyButtonPair {
@@ -51,6 +70,11 @@ public:
         for (const KeyButtonPair& pair : list) {
             if (pair.key_code == key_code) {
                 pair.key_button->status.store(pressed);
+                if (pressed) {
+                    pair.key_button->ToggleButton();
+                } else {
+                    pair.key_button->UnlockButton();
+                }
             }
         }
     }
@@ -75,7 +99,8 @@ KeyButton::~KeyButton() {
 
 std::unique_ptr<Input::ButtonDevice> Keyboard::Create(const Common::ParamPackage& params) {
     const int key_code = params.Get("code", 0);
-    std::unique_ptr<KeyButton> button = std::make_unique<KeyButton>(key_button_list);
+    const bool toggle = params.Get("toggle", false);
+    std::unique_ptr<KeyButton> button = std::make_unique<KeyButton>(key_button_list, toggle);
     key_button_list->AddKeyButton(key_code, button.get());
     return button;
 }

src/input_common/mouse/mouse_input.cpp

@@ -59,7 +59,7 @@ void Mouse::UpdateYuzuSettings() {
     });
 }
 
-void Mouse::PressButton(int x, int y, int button_) {
+void Mouse::PressButton(int x, int y, MouseButton button_) {
     const auto button_index = static_cast<std::size_t>(button_);
     if (button_index >= mouse_info.size()) {
         return;
@@ -67,7 +67,7 @@ void Mouse::PressButton(int x, int y, int button_) {
 
     const auto button = 1U << button_index;
     buttons |= static_cast<u16>(button);
-    last_button = static_cast<MouseButton>(button_index);
+    last_button = button_;
 
     mouse_info[button_index].mouse_origin = Common::MakeVec(x, y);
     mouse_info[button_index].last_mouse_position = Common::MakeVec(x, y);
@@ -129,7 +129,7 @@ void Mouse::MouseMove(int x, int y, int center_x, int center_y) {
     }
 }
 
-void Mouse::ReleaseButton(int button_) {
+void Mouse::ReleaseButton(MouseButton button_) {
     const auto button_index = static_cast<std::size_t>(button_);
     if (button_index >= mouse_info.size()) {
         return;
@@ -152,11 +152,52 @@ void Mouse::BeginConfiguration() {
 
 void Mouse::EndConfiguration() {
     buttons = 0;
+    for (MouseInfo& info : mouse_info) {
+        info.tilt_speed = 0;
+        info.data.pressed = false;
+        info.data.axis = {0, 0};
+    }
     last_button = MouseButton::Undefined;
     mouse_queue.Clear();
     configuring = false;
 }
 
+bool Mouse::ToggleButton(std::size_t button_) {
+    if (button_ >= mouse_info.size()) {
+        return false;
+    }
+    const auto button = 1U << button_;
+    const bool button_state = (toggle_buttons & button) != 0;
+    const bool button_lock = (lock_buttons & button) != 0;
+
+    if (button_lock) {
+        return button_state;
+    }
+
+    lock_buttons |= static_cast<u16>(button);
+
+    if (button_state) {
+        toggle_buttons &= static_cast<u16>(0xFF - button);
+    } else {
+        toggle_buttons |= static_cast<u16>(button);
+    }
+
+    return !button_state;
+}
+
+bool Mouse::UnlockButton(std::size_t button_) {
+    if (button_ >= mouse_info.size()) {
+        return false;
+    }
+
+    const auto button = 1U << button_;
+    const bool button_state = (toggle_buttons & button) != 0;
+
+    lock_buttons &= static_cast<u16>(0xFF - button);
+
+    return button_state;
+}
+
 Common::SPSCQueue<MouseStatus>& Mouse::GetMouseQueue() {
     return mouse_queue;
 }

src/input_common/mouse/mouse_input.h

@@ -18,10 +18,12 @@ namespace MouseInput {
 
 enum class MouseButton {
     Left,
-    Wheel,
     Right,
-    Forward,
+    Wheel,
     Backward,
+    Forward,
+    Task,
+    Extra,
     Undefined,
 };
 
@@ -51,7 +53,7 @@ public:
      * @param y the y-coordinate of the cursor
      * @param button_ the button pressed
      */
-    void PressButton(int x, int y, int button_);
+    void PressButton(int x, int y, MouseButton button_);
 
     /**
      * Signals that mouse has moved.
@@ -65,7 +67,10 @@ public:
     /**
      * Signals that a motion sensor tilt has ended.
      */
-    void ReleaseButton(int button_);
+    void ReleaseButton(MouseButton button_);
+
+    [[nodiscard]] bool ToggleButton(std::size_t button_);
+    [[nodiscard]] bool UnlockButton(std::size_t button_);
 
     [[nodiscard]] Common::SPSCQueue<MouseStatus>& GetMouseQueue();
     [[nodiscard]] const Common::SPSCQueue<MouseStatus>& GetMouseQueue() const;
@@ -92,9 +97,11 @@ private:
     };
 
     u16 buttons{};
+    u16 toggle_buttons{};
+    u16 lock_buttons{};
     std::thread update_thread;
     MouseButton last_button{MouseButton::Undefined};
-    std::array<MouseInfo, 5> mouse_info;
+    std::array<MouseInfo, 7> mouse_info;
     Common::SPSCQueue<MouseStatus> mouse_queue;
     bool configuring{false};
     bool update_thread_running{true};

src/input_common/mouse/mouse_poller.cpp

@@ -14,16 +14,25 @@ namespace InputCommon {
 
 class MouseButton final : public Input::ButtonDevice {
 public:
-    explicit MouseButton(u32 button_, const MouseInput::Mouse* mouse_input_)
-        : button(button_), mouse_input(mouse_input_) {}
+    explicit MouseButton(u32 button_, bool toggle_, MouseInput::Mouse* mouse_input_)
+        : button(button_), toggle(toggle_), mouse_input(mouse_input_) {}
 
     bool GetStatus() const override {
-        return mouse_input->GetMouseState(button).pressed;
+        const bool button_state = mouse_input->GetMouseState(button).pressed;
+        if (!toggle) {
+            return button_state;
+        }
+
+        if (button_state) {
+            return mouse_input->ToggleButton(button);
+        }
+        return mouse_input->UnlockButton(button);
     }
 
 private:
     const u32 button;
-    const MouseInput::Mouse* mouse_input;
+    const bool toggle;
+    MouseInput::Mouse* mouse_input;
 };
 
 MouseButtonFactory::MouseButtonFactory(std::shared_ptr<MouseInput::Mouse> mouse_input_)
@@ -32,8 +41,9 @@ MouseButtonFactory::MouseButtonFactory(std::shared_ptr<MouseInput::Mouse> mouse_
 std::unique_ptr<Input::ButtonDevice> MouseButtonFactory::Create(
     const Common::ParamPackage& params) {
     const auto button_id = params.Get("button", 0);
+    const auto toggle = params.Get("toggle", false);
 
-    return std::make_unique<MouseButton>(button_id, mouse_input.get());
+    return std::make_unique<MouseButton>(button_id, toggle, mouse_input.get());
 }
 
 Common::ParamPackage MouseButtonFactory::GetNextInput() const {

src/input_common/udp/client.cpp

@@ -5,6 +5,7 @@
 #include <chrono>
 #include <cstring>
 #include <functional>
+#include <random>
 #include <thread>
 #include <boost/asio.hpp>
 #include "common/logging/log.h"
@@ -26,11 +27,9 @@ class Socket {
 public:
     using clock = std::chrono::system_clock;
 
-    explicit Socket(const std::string& host, u16 port, std::size_t pad_index_, u32 client_id_,
-                    SocketCallback callback_)
+    explicit Socket(const std::string& host, u16 port, SocketCallback callback_)
         : callback(std::move(callback_)), timer(io_service),
-          socket(io_service, udp::endpoint(udp::v4(), 0)), client_id(client_id_),
-          pad_index(pad_index_) {
+          socket(io_service, udp::endpoint(udp::v4(), 0)), client_id(GenerateRandomClientId()) {
         boost::system::error_code ec{};
         auto ipv4 = boost::asio::ip::make_address_v4(host, ec);
         if (ec.value() != boost::system::errc::success) {
@@ -63,6 +62,11 @@ public:
     }
 
 private:
+    u32 GenerateRandomClientId() const {
+        std::random_device device;
+        return device();
+    }
+
     void HandleReceive(const boost::system::error_code&, std::size_t bytes_transferred) {
         if (auto type = Response::Validate(receive_buffer.data(), bytes_transferred)) {
             switch (*type) {
@@ -93,15 +97,15 @@ private:
     void HandleSend(const boost::system::error_code&) {
         boost::system::error_code _ignored{};
         // Send a request for getting port info for the pad
-        const Request::PortInfo port_info{1, {static_cast<u8>(pad_index), 0, 0, 0}};
+        const Request::PortInfo port_info{4, {0, 1, 2, 3}};
         const auto port_message = Request::Create(port_info, client_id);
         std::memcpy(&send_buffer1, &port_message, PORT_INFO_SIZE);
         socket.send_to(boost::asio::buffer(send_buffer1), send_endpoint, {}, _ignored);
 
         // Send a request for getting pad data for the pad
         const Request::PadData pad_data{
-            Request::PadData::Flags::Id,
-            static_cast<u8>(pad_index),
+            Request::PadData::Flags::AllPorts,
+            0,
             EMPTY_MAC_ADDRESS,
         };
         const auto pad_message = Request::Create(pad_data, client_id);
@@ -115,8 +119,7 @@ private:
     boost::asio::basic_waitable_timer<clock> timer;
     udp::socket socket;
 
-    u32 client_id{};
-    std::size_t pad_index{};
+    const u32 client_id;
 
     static constexpr std::size_t PORT_INFO_SIZE = sizeof(Message<Request::PortInfo>);
     static constexpr std::size_t PAD_DATA_SIZE = sizeof(Message<Request::PadData>);
@@ -144,34 +147,32 @@ Client::~Client() {
     Reset();
 }
 
-Client::ClientData::ClientData() = default;
+Client::ClientConnection::ClientConnection() = default;
 
-Client::ClientData::~ClientData() = default;
+Client::ClientConnection::~ClientConnection() = default;
 
 std::vector<Common::ParamPackage> Client::GetInputDevices() const {
     std::vector<Common::ParamPackage> devices;
-    for (std::size_t client = 0; client < clients.size(); client++) {
-        if (!DeviceConnected(client)) {
+    for (std::size_t pad = 0; pad < pads.size(); pad++) {
+        if (!DeviceConnected(pad)) {
             continue;
         }
-        std::string name = fmt::format("UDP Controller {}", client);
+        std::string name = fmt::format("UDP Controller {}", pad);
         devices.emplace_back(Common::ParamPackage{
             {"class", "cemuhookudp"},
             {"display", std::move(name)},
-            {"port", std::to_string(client)},
+            {"port", std::to_string(pad)},
         });
     }
     return devices;
 }
 
-bool Client::DeviceConnected(std::size_t client) const {
+bool Client::DeviceConnected(std::size_t pad) const {
     // Use last timestamp to detect if the socket has stopped sending data
     const auto now = std::chrono::steady_clock::now();
-    const auto time_difference =
-        static_cast<u64>(std::chrono::duration_cast<std::chrono::milliseconds>(
-                             now - clients[client].last_motion_update)
-                             .count());
-    return time_difference < 1000 && clients[client].active == 1;
+    const auto time_difference = static_cast<u64>(
+        std::chrono::duration_cast<std::chrono::milliseconds>(now - pads[pad].last_update).count());
+    return time_difference < 1000 && pads[pad].connected;
 }
 
 void Client::ReloadSockets() {
@@ -196,25 +197,21 @@ void Client::ReloadSockets() {
             continue;
         }
 
-        for (std::size_t pad = 0; pad < 4; ++pad) {
-            const std::size_t client_number =
-                GetClientNumber(udp_input_address, udp_input_port, pad);
-            if (client_number != MAX_UDP_CLIENTS) {
-                LOG_ERROR(Input, "Duplicated UDP servers found");
-                continue;
-            }
-            StartCommunication(client++, udp_input_address, udp_input_port, pad, 24872);
+        const std::size_t client_number = GetClientNumber(udp_input_address, udp_input_port);
+        if (client_number != MAX_UDP_CLIENTS) {
+            LOG_ERROR(Input, "Duplicated UDP servers found");
+            continue;
         }
+        StartCommunication(client++, udp_input_address, udp_input_port);
     }
 }
 
-std::size_t Client::GetClientNumber(std::string_view host, u16 port, std::size_t pad) const {
+std::size_t Client::GetClientNumber(std::string_view host, u16 port) const {
     for (std::size_t client = 0; client < clients.size(); client++) {
         if (clients[client].active == -1) {
             continue;
         }
-        if (clients[client].host == host && clients[client].port == port &&
-            clients[client].pad_index == pad) {
+        if (clients[client].host == host && clients[client].port == port) {
             return client;
         }
     }
@@ -230,69 +227,75 @@ void Client::OnPortInfo([[maybe_unused]] Response::PortInfo data) {
 }
 
 void Client::OnPadData(Response::PadData data, std::size_t client) {
-    // Accept packets only for the correct pad
-    if (static_cast<u8>(clients[client].pad_index) != data.info.id) {
+    const std::size_t pad_index = (client * PADS_PER_CLIENT) + data.info.id;
+
+    if (pad_index >= pads.size()) {
+        LOG_ERROR(Input, "Invalid pad id {}", data.info.id);
         return;
     }
 
     LOG_TRACE(Input, "PadData packet received");
-    if (data.packet_counter == clients[client].packet_sequence) {
+    if (data.packet_counter == pads[pad_index].packet_sequence) {
         LOG_WARNING(
             Input,
             "PadData packet dropped because its stale info. Current count: {} Packet count: {}",
-            clients[client].packet_sequence, data.packet_counter);
+            pads[pad_index].packet_sequence, data.packet_counter);
+        pads[pad_index].connected = false;
         return;
     }
-    clients[client].active = static_cast<s8>(data.info.is_pad_active);
-    clients[client].packet_sequence = data.packet_counter;
+
+    clients[client].active = 1;
+    pads[pad_index].connected = true;
+    pads[pad_index].packet_sequence = data.packet_counter;
+
     const auto now = std::chrono::steady_clock::now();
-    const auto time_difference =
-        static_cast<u64>(std::chrono::duration_cast<std::chrono::microseconds>(
-                             now - clients[client].last_motion_update)
-                             .count());
-    clients[client].last_motion_update = now;
+    const auto time_difference = static_cast<u64>(
+        std::chrono::duration_cast<std::chrono::microseconds>(now - pads[pad_index].last_update)
+            .count());
+    pads[pad_index].last_update = now;
+
     const Common::Vec3f raw_gyroscope = {data.gyro.pitch, data.gyro.roll, -data.gyro.yaw};
-    clients[client].motion.SetAcceleration({data.accel.x, -data.accel.z, data.accel.y});
+    pads[pad_index].motion.SetAcceleration({data.accel.x, -data.accel.z, data.accel.y});
     // Gyroscope values are not it the correct scale from better joy.
     // Dividing by 312 allows us to make one full turn = 1 turn
     // This must be a configurable valued called sensitivity
-    clients[client].motion.SetGyroscope(raw_gyroscope / 312.0f);
-    clients[client].motion.UpdateRotation(time_difference);
-    clients[client].motion.UpdateOrientation(time_difference);
+    pads[pad_index].motion.SetGyroscope(raw_gyroscope / 312.0f);
+    pads[pad_index].motion.UpdateRotation(time_difference);
+    pads[pad_index].motion.UpdateOrientation(time_difference);
 
     {
-        std::lock_guard guard(clients[client].status.update_mutex);
-        clients[client].status.motion_status = clients[client].motion.GetMotion();
+        std::lock_guard guard(pads[pad_index].status.update_mutex);
+        pads[pad_index].status.motion_status = pads[pad_index].motion.GetMotion();
 
         for (std::size_t id = 0; id < data.touch.size(); ++id) {
             UpdateTouchInput(data.touch[id], client, id);
         }
 
         if (configuring) {
-            const Common::Vec3f gyroscope = clients[client].motion.GetGyroscope();
-            const Common::Vec3f accelerometer = clients[client].motion.GetAcceleration();
-            UpdateYuzuSettings(client, accelerometer, gyroscope);
+            const Common::Vec3f gyroscope = pads[pad_index].motion.GetGyroscope();
+            const Common::Vec3f accelerometer = pads[pad_index].motion.GetAcceleration();
+            UpdateYuzuSettings(client, data.info.id, accelerometer, gyroscope);
         }
     }
 }
 
-void Client::StartCommunication(std::size_t client, const std::string& host, u16 port,
-                                std::size_t pad_index, u32 client_id) {
+void Client::StartCommunication(std::size_t client, const std::string& host, u16 port) {
     SocketCallback callback{[this](Response::Version version) { OnVersion(version); },
                             [this](Response::PortInfo info) { OnPortInfo(info); },
                             [this, client](Response::PadData data) { OnPadData(data, client); }};
-    LOG_INFO(Input, "Starting communication with UDP input server on {}:{}:{}", host, port,
-             pad_index);
+    LOG_INFO(Input, "Starting communication with UDP input server on {}:{}", host, port);
     clients[client].host = host;
     clients[client].port = port;
-    clients[client].pad_index = pad_index;
     clients[client].active = 0;
-    clients[client].socket = std::make_unique<Socket>(host, port, pad_index, client_id, callback);
+    clients[client].socket = std::make_unique<Socket>(host, port, callback);
     clients[client].thread = std::thread{SocketLoop, clients[client].socket.get()};
+
     // Set motion parameters
     // SetGyroThreshold value should be dependent on GyroscopeZeroDriftMode
     // Real HW values are unknown, 0.0001 is an approximate to Standard
-    clients[client].motion.SetGyroThreshold(0.0001f);
+    for (std::size_t pad = 0; pad < PADS_PER_CLIENT; pad++) {
+        pads[client * PADS_PER_CLIENT + pad].motion.SetGyroThreshold(0.0001f);
+    }
 }
 
 void Client::Reset() {
@@ -305,8 +308,8 @@ void Client::Reset() {
     }
 }
 
-void Client::UpdateYuzuSettings(std::size_t client, const Common::Vec3<float>& acc,
-                                const Common::Vec3<float>& gyro) {
+void Client::UpdateYuzuSettings(std::size_t client, std::size_t pad_index,
+                                const Common::Vec3<float>& acc, const Common::Vec3<float>& gyro) {
     if (gyro.Length() > 0.2f) {
         LOG_DEBUG(Input, "UDP Controller {}: gyro=({}, {}, {}), accel=({}, {}, {})", client,
                   gyro[0], gyro[1], gyro[2], acc[0], acc[1], acc[2]);
@@ -314,7 +317,7 @@ void Client::UpdateYuzuSettings(std::size_t client, const Common::Vec3<float>& a
     UDPPadStatus pad{
         .host = clients[client].host,
         .port = clients[client].port,
-        .pad_index = clients[client].pad_index,
+        .pad_index = pad_index,
     };
     for (std::size_t i = 0; i < 3; ++i) {
         if (gyro[i] > 5.0f || gyro[i] < -5.0f) {
@@ -385,19 +388,19 @@ void Client::EndConfiguration() {
 }
 
 DeviceStatus& Client::GetPadState(const std::string& host, u16 port, std::size_t pad) {
-    const std::size_t client_number = GetClientNumber(host, port, pad);
-    if (client_number == MAX_UDP_CLIENTS) {
-        return clients[0].status;
+    const std::size_t client_number = GetClientNumber(host, port);
+    if (client_number == MAX_UDP_CLIENTS || pad >= PADS_PER_CLIENT) {
+        return pads[0].status;
     }
-    return clients[client_number].status;
+    return pads[(client_number * PADS_PER_CLIENT) + pad].status;
 }
 
 const DeviceStatus& Client::GetPadState(const std::string& host, u16 port, std::size_t pad) const {
-    const std::size_t client_number = GetClientNumber(host, port, pad);
-    if (client_number == MAX_UDP_CLIENTS) {
-        return clients[0].status;
+    const std::size_t client_number = GetClientNumber(host, port);
+    if (client_number == MAX_UDP_CLIENTS || pad >= PADS_PER_CLIENT) {
+        return pads[0].status;
     }
-    return clients[client_number].status;
+    return pads[(client_number * PADS_PER_CLIENT) + pad].status;
 }
 
 Input::TouchStatus& Client::GetTouchState() {
@@ -416,7 +419,7 @@ const Common::SPSCQueue<UDPPadStatus>& Client::GetPadQueue() const {
     return pad_queue;
 }
 
-void TestCommunication(const std::string& host, u16 port, std::size_t pad_index, u32 client_id,
+void TestCommunication(const std::string& host, u16 port,
                        const std::function<void()>& success_callback,
                        const std::function<void()>& failure_callback) {
     std::thread([=] {
@@ -426,9 +429,10 @@ void TestCommunication(const std::string& host, u16 port, std::size_t pad_index,
             .port_info = [](Response::PortInfo) {},
             .pad_data = [&](Response::PadData) { success_event.Set(); },
         };
-        Socket socket{host, port, pad_index, client_id, std::move(callback)};
+        Socket socket{host, port, std::move(callback)};
         std::thread worker_thread{SocketLoop, &socket};
-        const bool result = success_event.WaitFor(std::chrono::seconds(5));
+        const bool result =
+            success_event.WaitUntil(std::chrono::steady_clock::now() + std::chrono::seconds(10));
         socket.Stop();
         worker_thread.join();
         if (result) {
@@ -440,8 +444,7 @@ void TestCommunication(const std::string& host, u16 port, std::size_t pad_index,
 }
 
 CalibrationConfigurationJob::CalibrationConfigurationJob(
-    const std::string& host, u16 port, std::size_t pad_index, u32 client_id,
-    std::function<void(Status)> status_callback,
+    const std::string& host, u16 port, std::function<void(Status)> status_callback,
     std::function<void(u16, u16, u16, u16)> data_callback) {
 
     std::thread([=, this] {
@@ -485,7 +488,7 @@ CalibrationConfigurationJob::CalibrationConfigurationJob(
                                         complete_event.Set();
                                     }
                                 }};
-        Socket socket{host, port, pad_index, client_id, std::move(callback)};
+        Socket socket{host, port, std::move(callback)};
         std::thread worker_thread{SocketLoop, &socket};
         complete_event.Wait();
         socket.Stop();

src/input_common/udp/client.h

@@ -84,7 +84,7 @@ public:
 
     std::vector<Common::ParamPackage> GetInputDevices() const;
 
-    bool DeviceConnected(std::size_t client) const;
+    bool DeviceConnected(std::size_t pad) const;
     void ReloadSockets();
 
     Common::SPSCQueue<UDPPadStatus>& GetPadQueue();
@@ -97,38 +97,40 @@ public:
     const Input::TouchStatus& GetTouchState() const;
 
 private:
-    struct ClientData {
-        ClientData();
-        ~ClientData();
-
-        std::string host{"127.0.0.1"};
-        u16 port{26760};
+    struct PadData {
         std::size_t pad_index{};
-        std::unique_ptr<Socket> socket;
+        bool connected{};
         DeviceStatus status;
-        std::thread thread;
         u64 packet_sequence{};
-        s8 active{-1};
 
         // Realtime values
         // motion is initalized with PID values for drift correction on joycons
         InputCommon::MotionInput motion{0.3f, 0.005f, 0.0f};
-        std::chrono::time_point<std::chrono::steady_clock> last_motion_update;
+        std::chrono::time_point<std::chrono::steady_clock> last_update;
+    };
+
+    struct ClientConnection {
+        ClientConnection();
+        ~ClientConnection();
+        std::string host{"127.0.0.1"};
+        u16 port{26760};
+        s8 active{-1};
+        std::unique_ptr<Socket> socket;
+        std::thread thread;
     };
 
     // For shutting down, clear all data, join all threads, release usb
     void Reset();
 
     // Translates configuration to client number
-    std::size_t GetClientNumber(std::string_view host, u16 port, std::size_t pad) const;
+    std::size_t GetClientNumber(std::string_view host, u16 port) const;
 
     void OnVersion(Response::Version);
     void OnPortInfo(Response::PortInfo);
     void OnPadData(Response::PadData, std::size_t client);
-    void StartCommunication(std::size_t client, const std::string& host, u16 port,
-                            std::size_t pad_index, u32 client_id);
-    void UpdateYuzuSettings(std::size_t client, const Common::Vec3<float>& acc,
-                            const Common::Vec3<float>& gyro);
+    void StartCommunication(std::size_t client, const std::string& host, u16 port);
+    void UpdateYuzuSettings(std::size_t client, std::size_t pad_index,
+                            const Common::Vec3<float>& acc, const Common::Vec3<float>& gyro);
 
     // Returns an unused finger id, if there is no fingers available std::nullopt will be
     // returned
@@ -140,10 +142,12 @@ private:
     bool configuring = false;
 
     // Allocate clients for 8 udp servers
-    static constexpr std::size_t MAX_UDP_CLIENTS = 4 * 8;
+    static constexpr std::size_t MAX_UDP_CLIENTS = 8;
+    static constexpr std::size_t PADS_PER_CLIENT = 4;
     // Each client can have up 2 touch inputs
     static constexpr std::size_t MAX_TOUCH_FINGERS = MAX_UDP_CLIENTS * 2;
-    std::array<ClientData, MAX_UDP_CLIENTS> clients{};
+    std::array<PadData, MAX_UDP_CLIENTS * PADS_PER_CLIENT> pads{};
+    std::array<ClientConnection, MAX_UDP_CLIENTS> clients{};
     Common::SPSCQueue<UDPPadStatus> pad_queue{};
     Input::TouchStatus touch_status{};
     std::array<std::size_t, MAX_TOUCH_FINGERS> finger_id{};
@@ -164,8 +168,8 @@ public:
      * @param status_callback Callback for job status updates
      * @param data_callback Called when calibration data is ready
      */
-    explicit CalibrationConfigurationJob(const std::string& host, u16 port, std::size_t pad_index,
-                                         u32 client_id, std::function<void(Status)> status_callback,
+    explicit CalibrationConfigurationJob(const std::string& host, u16 port,
+                                         std::function<void(Status)> status_callback,
                                          std::function<void(u16, u16, u16, u16)> data_callback);
     ~CalibrationConfigurationJob();
     void Stop();
@@ -174,7 +178,7 @@ private:
     Common::Event complete_event;
 };
 
-void TestCommunication(const std::string& host, u16 port, std::size_t pad_index, u32 client_id,
+void TestCommunication(const std::string& host, u16 port,
                        const std::function<void()>& success_callback,
                        const std::function<void()>& failure_callback);
 

src/tests/CMakeLists.txt

@@ -5,6 +5,7 @@ add_executable(tests
     common/param_package.cpp
     common/ring_buffer.cpp
     core/core_timing.cpp
+    core/network/network.cpp
     tests.cpp
     video_core/buffer_base.cpp
 )

src/tests/common/fibers.cpp

@@ -67,15 +67,16 @@ void TestControl1::DoWork() {
         value++;
     }
     results[id] = value;
-    Fiber::YieldTo(work_fibers[id].get(), thread_fibers[id].get());
+    Fiber::YieldTo(work_fibers[id], *thread_fibers[id]);
 }
 
 void TestControl1::ExecuteThread(u32 id) {
     thread_ids.Register(id);
-    thread_fibers[id] = Fiber::ThreadToFiber();
+    auto thread_fiber = Fiber::ThreadToFiber();
+    thread_fibers[id] = thread_fiber;
     work_fibers[id] = std::make_shared<Fiber>(std::function<void(void*)>{WorkControl1}, this);
     items[id] = rand() % 256;
-    Fiber::YieldTo(thread_fibers[id].get(), work_fibers[id].get());
+    Fiber::YieldTo(thread_fibers[id], *work_fibers[id]);
     thread_fibers[id]->Exit();
 }
 
@@ -116,11 +117,11 @@ public:
         for (u32 i = 0; i < 12000; i++) {
             value1 += i;
         }
-        Fiber::YieldTo(fiber1.get(), fiber3.get());
+        Fiber::YieldTo(fiber1, *fiber3);
         const u32 id = thread_ids.Get();
         assert1 = id == 1;
         value2 += 5000;
-        Fiber::YieldTo(fiber1.get(), thread_fibers[id].get());
+        Fiber::YieldTo(fiber1, *thread_fibers[id]);
     }
 
     void DoWork2() {
@@ -128,7 +129,7 @@ public:
             ;
         value2 = 2000;
         trap = false;
-        Fiber::YieldTo(fiber2.get(), fiber1.get());
+        Fiber::YieldTo(fiber2, *fiber1);
         assert3 = false;
     }
 
@@ -136,19 +137,19 @@ public:
         const u32 id = thread_ids.Get();
         assert2 = id == 0;
         value1 += 1000;
-        Fiber::YieldTo(fiber3.get(), thread_fibers[id].get());
+        Fiber::YieldTo(fiber3, *thread_fibers[id]);
     }
 
     void ExecuteThread(u32 id);
 
     void CallFiber1() {
         const u32 id = thread_ids.Get();
-        Fiber::YieldTo(thread_fibers[id].get(), fiber1.get());
+        Fiber::YieldTo(thread_fibers[id], *fiber1);
     }
 
     void CallFiber2() {
         const u32 id = thread_ids.Get();
-        Fiber::YieldTo(thread_fibers[id].get(), fiber2.get());
+        Fiber::YieldTo(thread_fibers[id], *fiber2);
     }
 
     void Exit();
@@ -184,7 +185,8 @@ static void WorkControl2_3(void* control) {
 
 void TestControl2::ExecuteThread(u32 id) {
     thread_ids.Register(id);
-    thread_fibers[id] = Fiber::ThreadToFiber();
+    auto thread_fiber = Fiber::ThreadToFiber();
+    thread_fibers[id] = thread_fiber;
 }
 
 void TestControl2::Exit() {
@@ -239,23 +241,23 @@ public:
 
     void DoWork1() {
         value1 += 1;
-        Fiber::YieldTo(fiber1.get(), fiber2.get());
+        Fiber::YieldTo(fiber1, *fiber2);
         const u32 id = thread_ids.Get();
         value3 += 1;
-        Fiber::YieldTo(fiber1.get(), thread_fibers[id].get());
+        Fiber::YieldTo(fiber1, *thread_fibers[id]);
     }
 
     void DoWork2() {
         value2 += 1;
         const u32 id = thread_ids.Get();
-        Fiber::YieldTo(fiber2.get(), thread_fibers[id].get());
+        Fiber::YieldTo(fiber2, *thread_fibers[id]);
     }
 
     void ExecuteThread(u32 id);
 
     void CallFiber1() {
         const u32 id = thread_ids.Get();
-        Fiber::YieldTo(thread_fibers[id].get(), fiber1.get());
+        Fiber::YieldTo(thread_fibers[id], *fiber1);
     }
 
     void Exit();
@@ -264,7 +266,7 @@ public:
     u32 value2{};
     u32 value3{};
     ThreadIds thread_ids;
-    std::vector<std::unique_ptr<Common::Fiber>> thread_fibers;
+    std::vector<std::shared_ptr<Common::Fiber>> thread_fibers;
     std::shared_ptr<Common::Fiber> fiber1;
     std::shared_ptr<Common::Fiber> fiber2;
 };
@@ -281,7 +283,8 @@ static void WorkControl3_2(void* control) {
 
 void TestControl3::ExecuteThread(u32 id) {
     thread_ids.Register(id);
-    thread_fibers[id] = Fiber::ThreadToFiber();
+    auto thread_fiber = Fiber::ThreadToFiber();
+    thread_fibers[id] = thread_fiber;
 }
 
 void TestControl3::Exit() {
@@ -329,7 +332,7 @@ public:
 
     void Execute() {
         thread_fiber = Fiber::ThreadToFiber();
-        Fiber::YieldTo(thread_fiber.get(), fiber1.get());
+        Fiber::YieldTo(thread_fiber, *fiber1);
         thread_fiber->Exit();
     }
 
@@ -337,7 +340,7 @@ public:
         fiber1->SetRewindPoint(std::function<void(void*)>{WorkControl4}, this);
         if (rewinded) {
             goal_reached = true;
-            Fiber::YieldTo(fiber1.get(), thread_fiber.get());
+            Fiber::YieldTo(fiber1, *thread_fiber);
         }
         rewinded = true;
         fiber1->Rewind();

src/tests/core/network/network.cpp

@@ -0,0 +1,28 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <catch2/catch.hpp>
+
+#include "core/network/network.h"
+#include "core/network/sockets.h"
+
+TEST_CASE("Network::Errors", "[core]") {
+    Network::NetworkInstance network_instance; // initialize network
+
+    Network::Socket socks[2];
+    for (Network::Socket& sock : socks) {
+        REQUIRE(sock.Initialize(Network::Domain::INET, Network::Type::STREAM,
+                                Network::Protocol::TCP) == Network::Errno::SUCCESS);
+    }
+
+    Network::SockAddrIn addr{
+        Network::Domain::INET,
+        {127, 0, 0, 1},
+        1, // hopefully nobody running this test has something listening on port 1
+    };
+    REQUIRE(socks[0].Connect(addr) == Network::Errno::CONNREFUSED);
+
+    std::vector<u8> message{1, 2, 3, 4};
+    REQUIRE(socks[1].Recv(0, message).second == Network::Errno::NOTCONN);
+}

src/video_core/CMakeLists.txt

@@ -236,7 +236,6 @@ add_library(video_core STATIC
     texture_cache/types.h
     texture_cache/util.cpp
     texture_cache/util.h
-    textures/astc.cpp
     textures/astc.h
     textures/decoders.cpp
     textures/decoders.h

src/video_core/buffer_cache/buffer_cache.h

@@ -9,6 +9,7 @@
 #include <deque>
 #include <memory>
 #include <mutex>
+#include <numeric>
 #include <span>
 #include <unordered_map>
 #include <vector>
@@ -91,7 +92,7 @@ class BufferCache {
     };
 
 public:
-    static constexpr u32 SKIP_CACHE_SIZE = 4096;
+    static constexpr u32 DEFAULT_SKIP_CACHE_SIZE = 4096;
 
     explicit BufferCache(VideoCore::RasterizerInterface& rasterizer_,
                          Tegra::Engines::Maxwell3D& maxwell3d_,
@@ -240,9 +241,9 @@ private:
     template <bool insert>
     void ChangeRegister(BufferId buffer_id);
 
-    void SynchronizeBuffer(Buffer& buffer, VAddr cpu_addr, u32 size);
+    bool SynchronizeBuffer(Buffer& buffer, VAddr cpu_addr, u32 size);
 
-    void SynchronizeBufferImpl(Buffer& buffer, VAddr cpu_addr, u32 size);
+    bool SynchronizeBufferImpl(Buffer& buffer, VAddr cpu_addr, u32 size);
 
     void UploadMemory(Buffer& buffer, u64 total_size_bytes, u64 largest_copy,
                       std::span<BufferCopy> copies);
@@ -297,6 +298,11 @@ private:
 
     std::array<u32, NUM_STAGES> fast_bound_uniform_buffers{};
 
+    std::array<u32, 16> uniform_cache_hits{};
+    std::array<u32, 16> uniform_cache_shots{};
+
+    u32 uniform_buffer_skip_cache_size = DEFAULT_SKIP_CACHE_SIZE;
+
     bool has_deleted_buffers = false;
 
     std::conditional_t<HAS_PERSISTENT_UNIFORM_BUFFER_BINDINGS, std::array<u32, NUM_STAGES>, Empty>
@@ -328,6 +334,19 @@ BufferCache<P>::BufferCache(VideoCore::RasterizerInterface& rasterizer_,
 
 template <class P>
 void BufferCache<P>::TickFrame() {
+    // Calculate hits and shots and move hit bits to the right
+    const u32 hits = std::reduce(uniform_cache_hits.begin(), uniform_cache_hits.end());
+    const u32 shots = std::reduce(uniform_cache_shots.begin(), uniform_cache_shots.end());
+    std::copy_n(uniform_cache_hits.begin(), uniform_cache_hits.size() - 1,
+                uniform_cache_hits.begin() + 1);
+    std::copy_n(uniform_cache_shots.begin(), uniform_cache_shots.size() - 1,
+                uniform_cache_shots.begin() + 1);
+    uniform_cache_hits[0] = 0;
+    uniform_cache_shots[0] = 0;
+
+    const bool skip_preferred = hits * 256 < shots * 251;
+    uniform_buffer_skip_cache_size = skip_preferred ? DEFAULT_SKIP_CACHE_SIZE : 0;
+
     delayed_destruction_ring.Tick();
 }
 
@@ -671,7 +690,7 @@ void BufferCache<P>::BindHostGraphicsUniformBuffer(size_t stage, u32 index, u32
     const VAddr cpu_addr = binding.cpu_addr;
     const u32 size = binding.size;
     Buffer& buffer = slot_buffers[binding.buffer_id];
-    if (size <= SKIP_CACHE_SIZE && !buffer.IsRegionGpuModified(cpu_addr, size)) {
+    if (size <= uniform_buffer_skip_cache_size && !buffer.IsRegionGpuModified(cpu_addr, size)) {
         if constexpr (IS_OPENGL) {
             if (runtime.HasFastBufferSubData()) {
                 // Fast path for Nvidia
@@ -692,7 +711,12 @@ void BufferCache<P>::BindHostGraphicsUniformBuffer(size_t stage, u32 index, u32
         return;
     }
     // Classic cached path
-    SynchronizeBuffer(buffer, cpu_addr, size);
+    const bool sync_cached = SynchronizeBuffer(buffer, cpu_addr, size);
+    if (sync_cached) {
+        ++uniform_cache_hits[0];
+    }
+    ++uniform_cache_shots[0];
+
     if (!needs_bind && !HasFastUniformBufferBound(stage, binding_index)) {
         // Skip binding if it's not needed and if the bound buffer is not the fast version
         // This exists to avoid instances where the fast buffer is bound and a GPU write happens
@@ -1106,15 +1130,15 @@ void BufferCache<P>::ChangeRegister(BufferId buffer_id) {
 }
 
 template <class P>
-void BufferCache<P>::SynchronizeBuffer(Buffer& buffer, VAddr cpu_addr, u32 size) {
+bool BufferCache<P>::SynchronizeBuffer(Buffer& buffer, VAddr cpu_addr, u32 size) {
     if (buffer.CpuAddr() == 0) {
-        return;
+        return true;
     }
-    SynchronizeBufferImpl(buffer, cpu_addr, size);
+    return SynchronizeBufferImpl(buffer, cpu_addr, size);
 }
 
 template <class P>
-void BufferCache<P>::SynchronizeBufferImpl(Buffer& buffer, VAddr cpu_addr, u32 size) {
+bool BufferCache<P>::SynchronizeBufferImpl(Buffer& buffer, VAddr cpu_addr, u32 size) {
     boost::container::small_vector<BufferCopy, 4> copies;
     u64 total_size_bytes = 0;
     u64 largest_copy = 0;
@@ -1128,10 +1152,11 @@ void BufferCache<P>::SynchronizeBufferImpl(Buffer& buffer, VAddr cpu_addr, u32 s
         largest_copy = std::max(largest_copy, range_size);
     });
     if (total_size_bytes == 0) {
-        return;
+        return true;
     }
     const std::span<BufferCopy> copies_span(copies.data(), copies.size());
     UploadMemory(buffer, total_size_bytes, largest_copy, copies_span);
+    return false;
 }
 
 template <class P>

src/video_core/command_classes/codecs/vp9.cpp

@@ -2,8 +2,9 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include <cstring> // for std::memcpy
+#include <algorithm> // for std::copy
 #include <numeric>
+#include "common/assert.h"
 #include "video_core/command_classes/codecs/vp9.h"
 #include "video_core/gpu.h"
 #include "video_core/memory_manager.h"
@@ -362,7 +363,8 @@ Vp9PictureInfo VP9::GetVp9PictureInfo(const NvdecCommon::NvdecRegisters& state)
     // surface_luma_offset[0:3] contains the address of the reference frame offsets in the following
     // order: last, golden, altref, current. It may be worthwhile to track the updates done here
     // to avoid buffering frame data needed for reference frame updating in the header composition.
-    std::memcpy(vp9_info.frame_offsets.data(), state.surface_luma_offset.data(), 4 * sizeof(u64));
+    std::copy(state.surface_luma_offset.begin(), state.surface_luma_offset.begin() + 4,
+              vp9_info.frame_offsets.begin());
 
     return vp9_info;
 }
@@ -821,11 +823,11 @@ const std::vector<u8>& VP9::ComposeFrameHeader(const NvdecCommon::NvdecRegisters
 
     // Write headers and frame to buffer
     frame.resize(uncompressed_header.size() + compressed_header.size() + bitstream.size());
-    std::memcpy(frame.data(), uncompressed_header.data(), uncompressed_header.size());
-    std::memcpy(frame.data() + uncompressed_header.size(), compressed_header.data(),
-                compressed_header.size());
-    std::memcpy(frame.data() + uncompressed_header.size() + compressed_header.size(),
-                bitstream.data(), bitstream.size());
+    std::copy(uncompressed_header.begin(), uncompressed_header.end(), frame.begin());
+    std::copy(compressed_header.begin(), compressed_header.end(),
+              frame.begin() + uncompressed_header.size());
+    std::copy(bitstream.begin(), bitstream.end(),
+              frame.begin() + uncompressed_header.size() + compressed_header.size());
 
     // keep track of frame number
     current_frame_number++;

src/video_core/compatible_formats.cpp

@@ -48,6 +48,15 @@ constexpr std::array VIEW_CLASS_32_BITS{
     PixelFormat::A2B10G10R10_UINT,
 };
 
+constexpr std::array VIEW_CLASS_32_BITS_NO_BGR{
+    PixelFormat::R16G16_FLOAT,      PixelFormat::B10G11R11_FLOAT,  PixelFormat::R32_FLOAT,
+    PixelFormat::A2B10G10R10_UNORM, PixelFormat::R16G16_UINT,      PixelFormat::R32_UINT,
+    PixelFormat::R16G16_SINT,       PixelFormat::R32_SINT,         PixelFormat::A8B8G8R8_UNORM,
+    PixelFormat::R16G16_UNORM,      PixelFormat::A8B8G8R8_SNORM,   PixelFormat::R16G16_SNORM,
+    PixelFormat::A8B8G8R8_SRGB,     PixelFormat::E5B9G9R9_FLOAT,   PixelFormat::A8B8G8R8_UINT,
+    PixelFormat::A8B8G8R8_SINT,     PixelFormat::A2B10G10R10_UINT,
+};
+
 // TODO: How should we handle 24 bits?
 
 constexpr std::array VIEW_CLASS_16_BITS{
@@ -205,7 +214,6 @@ constexpr Table MakeViewTable() {
     EnableRange(view, VIEW_CLASS_128_BITS);
     EnableRange(view, VIEW_CLASS_96_BITS);
     EnableRange(view, VIEW_CLASS_64_BITS);
-    EnableRange(view, VIEW_CLASS_32_BITS);
     EnableRange(view, VIEW_CLASS_16_BITS);
     EnableRange(view, VIEW_CLASS_8_BITS);
     EnableRange(view, VIEW_CLASS_RGTC1_RED);
@@ -231,20 +239,47 @@ constexpr Table MakeCopyTable() {
     EnableRange(copy, COPY_CLASS_64_BITS);
     return copy;
 }
+
+constexpr Table MakeNativeBgrViewTable() {
+    Table copy = MakeViewTable();
+    EnableRange(copy, VIEW_CLASS_32_BITS);
+    return copy;
+}
+
+constexpr Table MakeNonNativeBgrViewTable() {
+    Table copy = MakeViewTable();
+    EnableRange(copy, VIEW_CLASS_32_BITS_NO_BGR);
+    return copy;
+}
+
+constexpr Table MakeNativeBgrCopyTable() {
+    Table copy = MakeCopyTable();
+    EnableRange(copy, VIEW_CLASS_32_BITS);
+    return copy;
+}
+
+constexpr Table MakeNonNativeBgrCopyTable() {
+    Table copy = MakeCopyTable();
+    EnableRange(copy, VIEW_CLASS_32_BITS);
+    return copy;
+}
 } // Anonymous namespace
 
-bool IsViewCompatible(PixelFormat format_a, PixelFormat format_b, bool broken_views) {
+bool IsViewCompatible(PixelFormat format_a, PixelFormat format_b, bool broken_views,
+                      bool native_bgr) {
     if (broken_views) {
         // If format views are broken, only accept formats that are identical.
         return format_a == format_b;
     }
-    static constexpr Table TABLE = MakeViewTable();
-    return IsSupported(TABLE, format_a, format_b);
+    static constexpr Table BGR_TABLE = MakeNativeBgrViewTable();
+    static constexpr Table NO_BGR_TABLE = MakeNonNativeBgrViewTable();
+    return IsSupported(native_bgr ? BGR_TABLE : NO_BGR_TABLE, format_a, format_b);
 }
 
-bool IsCopyCompatible(PixelFormat format_a, PixelFormat format_b) {
-    static constexpr Table TABLE = MakeCopyTable();
-    return IsSupported(TABLE, format_a, format_b);
+bool IsCopyCompatible(PixelFormat format_a, PixelFormat format_b, bool native_bgr) {
+    static constexpr Table BGR_TABLE = MakeNativeBgrCopyTable();
+    static constexpr Table NO_BGR_TABLE = MakeNonNativeBgrCopyTable();
+    return IsSupported(native_bgr ? BGR_TABLE : NO_BGR_TABLE, format_a, format_b);
 }
 
 } // namespace VideoCore::Surface

src/video_core/compatible_formats.h

@@ -8,8 +8,9 @@
 
 namespace VideoCore::Surface {
 
-bool IsViewCompatible(PixelFormat format_a, PixelFormat format_b, bool broken_views);
+bool IsViewCompatible(PixelFormat format_a, PixelFormat format_b, bool broken_views,
+                      bool native_bgr);
 
-bool IsCopyCompatible(PixelFormat format_a, PixelFormat format_b);
+bool IsCopyCompatible(PixelFormat format_a, PixelFormat format_b, bool native_bgr);
 
 } // namespace VideoCore::Surface

src/video_core/gpu.cpp

@@ -459,7 +459,7 @@ void GPU::ProcessSemaphoreAcquire() {
 }
 
 void GPU::Start() {
-    gpu_thread.StartThread(*renderer, renderer->Context(), *dma_pusher, *cdma_pusher);
+    gpu_thread.StartThread(*renderer, renderer->Context(), *dma_pusher);
     cpu_context = renderer->GetRenderWindow().CreateSharedContext();
     cpu_context->MakeCurrent();
 }
@@ -517,8 +517,8 @@ void GPU::TriggerCpuInterrupt(const u32 syncpoint_id, const u32 value) const {
     interrupt_manager.GPUInterruptSyncpt(syncpoint_id, value);
 }
 
-void GPU::WaitIdle() const {
-    gpu_thread.WaitIdle();
+void GPU::ShutDown() {
+    gpu_thread.ShutDown();
 }
 
 void GPU::OnCommandListEnd() {

src/video_core/gpu.h

@@ -219,8 +219,8 @@ public:
         return *shader_notify;
     }
 
-    // Waits for the GPU to finish working
-    void WaitIdle() const;
+    // Stops the GPU execution and waits for the GPU to finish working
+    void ShutDown();
 
     /// Allows the CPU/NvFlinger to wait on the GPU before presenting a frame.
     void WaitFence(u32 syncpoint_id, u32 value);

src/video_core/gpu_thread.cpp

@@ -19,7 +19,7 @@ namespace VideoCommon::GPUThread {
 /// Runs the GPU thread
 static void RunThread(Core::System& system, VideoCore::RendererBase& renderer,
                       Core::Frontend::GraphicsContext& context, Tegra::DmaPusher& dma_pusher,
-                      SynchState& state, Tegra::CDmaPusher& cdma_pusher) {
+                      SynchState& state) {
     std::string name = "yuzu:GPU";
     MicroProfileOnThreadCreate(name.c_str());
     SCOPE_EXIT({ MicroProfileOnThreadExit(); });
@@ -29,8 +29,7 @@ static void RunThread(Core::System& system, VideoCore::RendererBase& renderer,
     system.RegisterHostThread();
 
     // Wait for first GPU command before acquiring the window context
-    while (state.queue.Empty())
-        ;
+    state.queue.Wait();
 
     // If emulation was stopped during disk shader loading, abort before trying to acquire context
     if (!state.is_running) {
@@ -46,9 +45,6 @@ static void RunThread(Core::System& system, VideoCore::RendererBase& renderer,
         if (auto* submit_list = std::get_if<SubmitListCommand>(&next.data)) {
             dma_pusher.Push(std::move(submit_list->entries));
             dma_pusher.DispatchCalls();
-        } else if (auto* command_list = std::get_if<SubmitChCommandEntries>(&next.data)) {
-            // NVDEC
-            cdma_pusher.ProcessEntries(std::move(command_list->entries));
         } else if (const auto* data = std::get_if<SwapBuffersCommand>(&next.data)) {
             renderer.SwapBuffers(data->framebuffer ? &*data->framebuffer : nullptr);
         } else if (std::holds_alternative<OnCommandListEndCommand>(next.data)) {
@@ -60,11 +56,17 @@ static void RunThread(Core::System& system, VideoCore::RendererBase& renderer,
         } else if (const auto* invalidate = std::get_if<InvalidateRegionCommand>(&next.data)) {
             rasterizer->OnCPUWrite(invalidate->addr, invalidate->size);
         } else if (std::holds_alternative<EndProcessingCommand>(next.data)) {
-            return;
+            ASSERT(state.is_running == false);
         } else {
             UNREACHABLE();
         }
         state.signaled_fence.store(next.fence);
+        if (next.block) {
+            // We have to lock the write_lock to ensure that the condition_variable wait not get a
+            // race between the check and the lock itself.
+            std::lock_guard lk(state.write_lock);
+            state.cv.notify_all();
+        }
     }
 }
 
@@ -72,31 +74,21 @@ ThreadManager::ThreadManager(Core::System& system_, bool is_async_)
     : system{system_}, is_async{is_async_} {}
 
 ThreadManager::~ThreadManager() {
-    if (!thread.joinable()) {
-        return;
-    }
-
-    // Notify GPU thread that a shutdown is pending
-    PushCommand(EndProcessingCommand());
-    thread.join();
+    ShutDown();
 }
 
 void ThreadManager::StartThread(VideoCore::RendererBase& renderer,
                                 Core::Frontend::GraphicsContext& context,
-                                Tegra::DmaPusher& dma_pusher, Tegra::CDmaPusher& cdma_pusher) {
+                                Tegra::DmaPusher& dma_pusher) {
     rasterizer = renderer.ReadRasterizer();
     thread = std::thread(RunThread, std::ref(system), std::ref(renderer), std::ref(context),
-                         std::ref(dma_pusher), std::ref(state), std::ref(cdma_pusher));
+                         std::ref(dma_pusher), std::ref(state));
 }
 
 void ThreadManager::SubmitList(Tegra::CommandList&& entries) {
     PushCommand(SubmitListCommand(std::move(entries)));
 }
 
-void ThreadManager::SubmitCommandBuffer(Tegra::ChCommandHeaderList&& entries) {
-    PushCommand(SubmitChCommandEntries(std::move(entries)));
-}
-
 void ThreadManager::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
     PushCommand(SwapBuffersCommand(framebuffer ? std::make_optional(*framebuffer) : std::nullopt));
 }
@@ -119,9 +111,8 @@ void ThreadManager::FlushRegion(VAddr addr, u64 size) {
     case Settings::GPUAccuracy::Extreme: {
         auto& gpu = system.GPU();
         u64 fence = gpu.RequestFlush(addr, size);
-        PushCommand(GPUTickCommand());
-        while (fence > gpu.CurrentFlushRequestFence()) {
-        }
+        PushCommand(GPUTickCommand(), true);
+        ASSERT(fence <= gpu.CurrentFlushRequestFence());
         break;
     }
     default:
@@ -138,23 +129,45 @@ void ThreadManager::FlushAndInvalidateRegion(VAddr addr, u64 size) {
     rasterizer->OnCPUWrite(addr, size);
 }
 
-void ThreadManager::WaitIdle() const {
-    while (state.last_fence > state.signaled_fence.load(std::memory_order_relaxed) &&
-           system.IsPoweredOn()) {
+void ThreadManager::ShutDown() {
+    if (!state.is_running) {
+        return;
     }
+
+    {
+        std::lock_guard lk(state.write_lock);
+        state.is_running = false;
+        state.cv.notify_all();
+    }
+
+    if (!thread.joinable()) {
+        return;
+    }
+
+    // Notify GPU thread that a shutdown is pending
+    PushCommand(EndProcessingCommand());
+    thread.join();
 }
 
 void ThreadManager::OnCommandListEnd() {
     PushCommand(OnCommandListEndCommand());
 }
 
-u64 ThreadManager::PushCommand(CommandData&& command_data) {
-    const u64 fence{++state.last_fence};
-    state.queue.Push(CommandDataContainer(std::move(command_data), fence));
-
+u64 ThreadManager::PushCommand(CommandData&& command_data, bool block) {
     if (!is_async) {
         // In synchronous GPU mode, block the caller until the command has executed
-        WaitIdle();
+        block = true;
+    }
+
+    std::unique_lock lk(state.write_lock);
+    const u64 fence{++state.last_fence};
+    state.queue.Push(CommandDataContainer(std::move(command_data), fence, block));
+
+    if (block) {
+        state.cv.wait(lk, [this, fence] {
+            return fence <= state.signaled_fence.load(std::memory_order_relaxed) ||
+                   !state.is_running;
+        });
     }
 
     return fence;

src/video_core/gpu_thread.h

@@ -43,14 +43,6 @@ struct SubmitListCommand final {
     Tegra::CommandList entries;
 };
 
-/// Command to signal to the GPU thread that a cdma command list is ready for processing
-struct SubmitChCommandEntries final {
-    explicit SubmitChCommandEntries(Tegra::ChCommandHeaderList&& entries_)
-        : entries{std::move(entries_)} {}
-
-    Tegra::ChCommandHeaderList entries;
-};
-
 /// Command to signal to the GPU thread that a swap buffers is pending
 struct SwapBuffersCommand final {
     explicit SwapBuffersCommand(std::optional<const Tegra::FramebufferConfig> framebuffer_)
@@ -91,28 +83,31 @@ struct OnCommandListEndCommand final {};
 struct GPUTickCommand final {};
 
 using CommandData =
-    std::variant<EndProcessingCommand, SubmitListCommand, SubmitChCommandEntries,
-                 SwapBuffersCommand, FlushRegionCommand, InvalidateRegionCommand,
-                 FlushAndInvalidateRegionCommand, OnCommandListEndCommand, GPUTickCommand>;
+    std::variant<EndProcessingCommand, SubmitListCommand, SwapBuffersCommand, FlushRegionCommand,
+                 InvalidateRegionCommand, FlushAndInvalidateRegionCommand, OnCommandListEndCommand,
+                 GPUTickCommand>;
 
 struct CommandDataContainer {
     CommandDataContainer() = default;
 
-    explicit CommandDataContainer(CommandData&& data_, u64 next_fence_)
-        : data{std::move(data_)}, fence{next_fence_} {}
+    explicit CommandDataContainer(CommandData&& data_, u64 next_fence_, bool block_)
+        : data{std::move(data_)}, fence{next_fence_}, block(block_) {}
 
     CommandData data;
     u64 fence{};
+    bool block{};
 };
 
 /// Struct used to synchronize the GPU thread
 struct SynchState final {
     std::atomic_bool is_running{true};
 
-    using CommandQueue = Common::MPSCQueue<CommandDataContainer>;
+    using CommandQueue = Common::SPSCQueue<CommandDataContainer>;
+    std::mutex write_lock;
     CommandQueue queue;
     u64 last_fence{};
     std::atomic<u64> signaled_fence{};
+    std::condition_variable cv;
 };
 
 /// Class used to manage the GPU thread
@@ -123,14 +118,11 @@ public:
 
     /// Creates and starts the GPU thread.
     void StartThread(VideoCore::RendererBase& renderer, Core::Frontend::GraphicsContext& context,
-                     Tegra::DmaPusher& dma_pusher, Tegra::CDmaPusher& cdma_pusher);
+                     Tegra::DmaPusher& dma_pusher);
 
     /// Push GPU command entries to be processed
     void SubmitList(Tegra::CommandList&& entries);
 
-    /// Push GPU CDMA command buffer entries to be processed
-    void SubmitCommandBuffer(Tegra::ChCommandHeaderList&& entries);
-
     /// Swap buffers (render frame)
     void SwapBuffers(const Tegra::FramebufferConfig* framebuffer);
 
@@ -143,14 +135,14 @@ public:
     /// Notify rasterizer that any caches of the specified region should be flushed and invalidated
     void FlushAndInvalidateRegion(VAddr addr, u64 size);
 
-    // Wait until the gpu thread is idle.
-    void WaitIdle() const;
+    // Stops the GPU execution and waits for the GPU to finish working
+    void ShutDown();
 
     void OnCommandListEnd();
 
 private:
     /// Pushes a command to be executed by the GPU thread
-    u64 PushCommand(CommandData&& command_data);
+    u64 PushCommand(CommandData&& command_data, bool block = false);
 
     Core::System& system;
     const bool is_async;

src/video_core/host_shaders/CMakeLists.txt

@@ -1,10 +1,12 @@
 set(SHADER_FILES
+    astc_decoder.comp
     block_linear_unswizzle_2d.comp
     block_linear_unswizzle_3d.comp
     convert_depth_to_float.frag
     convert_float_to_depth.frag
     full_screen_triangle.vert
     opengl_copy_bc4.comp
+    opengl_copy_bgra.comp
     opengl_present.frag
     opengl_present.vert
     pitch_unswizzle.comp

src/video_core/host_shaders/StringShaderHeader.cmake

@@ -6,7 +6,27 @@ get_filename_component(CONTENTS_NAME ${SOURCE_FILE} NAME)
 string(REPLACE "." "_" CONTENTS_NAME ${CONTENTS_NAME})
 string(TOUPPER ${CONTENTS_NAME} CONTENTS_NAME)
 
-file(READ ${SOURCE_FILE} CONTENTS)
+FILE(READ ${SOURCE_FILE} line_contents)
+
+# Replace double quotes with single quotes,
+# as double quotes will be used to wrap the lines
+STRING(REGEX REPLACE "\"" "'" line_contents "${line_contents}")
+
+# CMake separates list elements with semicolons, but semicolons
+# are used extensively in the shader code.
+# Replace with a temporary marker, to be reverted later.
+STRING(REGEX REPLACE ";" "{{SEMICOLON}}" line_contents "${line_contents}")
+
+# Make every line an individual element in the CMake list.
+STRING(REGEX REPLACE "\n" ";" line_contents "${line_contents}")
+
+# Build the shader string, wrapping each line in double quotes.
+foreach(line IN LISTS line_contents)
+    string(CONCAT CONTENTS "${CONTENTS}" \"${line}\\n\"\n)
+endforeach()
+
+# Revert the original semicolons in the source.
+STRING(REGEX REPLACE "{{SEMICOLON}}" ";" CONTENTS "${CONTENTS}")
 
 get_filename_component(OUTPUT_DIR ${HEADER_FILE} DIRECTORY)
 make_directory(${OUTPUT_DIR})

src/video_core/host_shaders/opengl_copy_bgra.comp

@@ -0,0 +1,15 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#version 430 core
+
+layout (local_size_x = 4, local_size_y = 4) in;
+
+layout(binding = 0, rgba8) readonly uniform image2DArray bgr_input;
+layout(binding = 1, rgba8) writeonly uniform image2DArray bgr_output;
+
+void main() {
+    vec4 color = imageLoad(bgr_input, ivec3(gl_GlobalInvocationID));
+    imageStore(bgr_output, ivec3(gl_GlobalInvocationID), color.bgra);
+}

src/video_core/host_shaders/source_shader.h.in

@@ -4,6 +4,8 @@
 
 namespace HostShaders {
 
-constexpr std::string_view @CONTENTS_NAME@ = R"(@CONTENTS@)";
+constexpr std::string_view @CONTENTS_NAME@ = {
+@CONTENTS@
+};
 
 } // namespace HostShaders

src/video_core/rasterizer_accelerated.cpp

@@ -2,63 +2,43 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include <mutex>
-
-#include <boost/icl/interval_map.hpp>
-#include <boost/range/iterator_range.hpp>
-
 #include "common/assert.h"
 #include "common/common_types.h"
+#include "common/div_ceil.h"
 #include "core/memory.h"
 #include "video_core/rasterizer_accelerated.h"
 
 namespace VideoCore {
 
-namespace {
-
-template <typename Map, typename Interval>
-constexpr auto RangeFromInterval(Map& map, const Interval& interval) {
-    return boost::make_iterator_range(map.equal_range(interval));
-}
-
-} // Anonymous namespace
-
 RasterizerAccelerated::RasterizerAccelerated(Core::Memory::Memory& cpu_memory_)
     : cpu_memory{cpu_memory_} {}
 
 RasterizerAccelerated::~RasterizerAccelerated() = default;
 
 void RasterizerAccelerated::UpdatePagesCachedCount(VAddr addr, u64 size, int delta) {
-    std::lock_guard lock{pages_mutex};
-    const u64 page_start{addr >> Core::Memory::PAGE_BITS};
-    const u64 page_end{(addr + size + Core::Memory::PAGE_SIZE - 1) >> Core::Memory::PAGE_BITS};
+    const auto page_end = Common::DivCeil(addr + size, Core::Memory::PAGE_SIZE);
+    for (auto page = addr >> Core::Memory::PAGE_BITS; page != page_end; ++page) {
+        auto& count = cached_pages.at(page >> 3).Count(page);
 
-    // Interval maps will erase segments if count reaches 0, so if delta is negative we have to
-    // subtract after iterating
-    const auto pages_interval = CachedPageMap::interval_type::right_open(page_start, page_end);
-    if (delta > 0) {
-        cached_pages.add({pages_interval, delta});
-    }
-
-    for (const auto& pair : RangeFromInterval(cached_pages, pages_interval)) {
-        const auto interval = pair.first & pages_interval;
-        const int count = pair.second;
-
-        const VAddr interval_start_addr = boost::icl::first(interval) << Core::Memory::PAGE_BITS;
-        const VAddr interval_end_addr = boost::icl::last_next(interval) << Core::Memory::PAGE_BITS;
-        const u64 interval_size = interval_end_addr - interval_start_addr;
-
-        if (delta > 0 && count == delta) {
-            cpu_memory.RasterizerMarkRegionCached(interval_start_addr, interval_size, true);
-        } else if (delta < 0 && count == -delta) {
-            cpu_memory.RasterizerMarkRegionCached(interval_start_addr, interval_size, false);
+        if (delta > 0) {
+            ASSERT_MSG(count < UINT8_MAX, "Count may overflow!");
+        } else if (delta < 0) {
+            ASSERT_MSG(count > 0, "Count may underflow!");
         } else {
-            ASSERT(count >= 0);
+            ASSERT_MSG(true, "Delta must be non-zero!");
         }
-    }
 
-    if (delta < 0) {
-        cached_pages.add({pages_interval, delta});
+        // Adds or subtracts 1, as count is a unsigned 8-bit value
+        count += static_cast<u8>(delta);
+
+        // Assume delta is either -1 or 1
+        if (count == 0) {
+            cpu_memory.RasterizerMarkRegionCached(page << Core::Memory::PAGE_BITS,
+                                                  Core::Memory::PAGE_SIZE, false);
+        } else if (count == 1 && delta > 0) {
+            cpu_memory.RasterizerMarkRegionCached(page << Core::Memory::PAGE_BITS,
+                                                  Core::Memory::PAGE_SIZE, true);
+        }
     }
 }
 

src/video_core/rasterizer_accelerated.h

@@ -4,9 +4,8 @@
 
 #pragma once
 
-#include <mutex>
-
-#include <boost/icl/interval_map.hpp>
+#include <array>
+#include <atomic>
 
 #include "common/common_types.h"
 #include "video_core/rasterizer_interface.h"
@@ -26,10 +25,24 @@ public:
     void UpdatePagesCachedCount(VAddr addr, u64 size, int delta) override;
 
 private:
-    using CachedPageMap = boost::icl::interval_map<u64, int>;
-    CachedPageMap cached_pages;
-    std::mutex pages_mutex;
+    class CacheEntry final {
+    public:
+        CacheEntry() = default;
 
+        std::atomic_uint8_t& Count(std::size_t page) {
+            return values[page & 7];
+        }
+
+        const std::atomic_uint8_t& Count(std::size_t page) const {
+            return values[page & 7];
+        }
+
+    private:
+        std::array<std::atomic_uint8_t, 8> values{};
+    };
+    static_assert(sizeof(CacheEntry) == 8, "CacheEntry should be 8 bytes!");
+
+    std::array<CacheEntry, 0x800000> cached_pages;
     Core::Memory::Memory& cpu_memory;
 };
 

src/video_core/renderer_opengl/gl_buffer_cache.cpp

@@ -73,7 +73,8 @@ BufferCacheRuntime::BufferCacheRuntime(const Device& device_)
     for (auto& stage_uniforms : fast_uniforms) {
         for (OGLBuffer& buffer : stage_uniforms) {
             buffer.Create();
-            glNamedBufferData(buffer.handle, BufferCache::SKIP_CACHE_SIZE, nullptr, GL_STREAM_DRAW);
+            glNamedBufferData(buffer.handle, BufferCache::DEFAULT_SKIP_CACHE_SIZE, nullptr,
+                              GL_STREAM_DRAW);
         }
     }
     for (auto& stage_uniforms : copy_uniforms) {