Update README.md

Basic documentation edits in readme file.

.gitmodules

@@ -16,6 +16,9 @@
 [submodule "libressl"]
     path = externals/libressl
     url = https://github.com/citra-emu/ext-libressl-portable.git
+[submodule "libusb"]
+	path = externals/libusb/libusb
+	url = https://github.com/libusb/libusb.git
 [submodule "discord-rpc"]
     path = externals/discord-rpc
     url = https://github.com/discordapp/discord-rpc.git
@@ -34,9 +37,6 @@
 [submodule "xbyak"]
     path = externals/xbyak
     url = https://github.com/herumi/xbyak.git
-[submodule "externals/libusb"]
-	path = externals/libusb
-	url = https://github.com/ameerj/libusb
 [submodule "opus"]
 	path = externals/opus/opus
 	url = https://github.com/xiph/opus.git

.lgtm.yml

@@ -0,0 +1,13 @@
+path_classifiers:
+  library: "externals"
+extraction:
+  cpp:
+    prepare:
+      packages:
+      - "libsdl2-dev"
+      - "qtmultimedia5-dev"
+      - "clang-format-10"
+      - "libtbb-dev"
+      - "libjack-jackd2-dev"
+      - "doxygen"
+      - "graphviz"

README.md

@@ -4,15 +4,15 @@ yuzu emulator
 [![Azure Mainline CI Build Status](https://dev.azure.com/yuzu-emu/yuzu/_apis/build/status/yuzu%20mainline?branchName=master)](https://dev.azure.com/yuzu-emu/yuzu/)
 [![Discord](https://img.shields.io/discord/398318088170242053?color=%237289DA&label=yuzu&logo=discord&logoColor=white)](https://discord.com/invite/u77vRWY)
 
-yuzu is an experimental open-source emulator for the Nintendo Switch from the creators of [Citra](https://citra-emu.org/).
+Yuzu is an experimental open-source emulator for the Nintendo Switch from the creators of [Citra](https://citra-emu.org/).
 
 It is written in C++ with portability in mind, with builds actively maintained for Windows and Linux. The emulator is capable of running several commercial games.
 
-yuzu only emulates a subset of Switch hardware and therefore most commercial games **do not** run at full speed or are not fully functional.
+Yuzu only emulates a subset of Switch hardware and therefore most commercial games **do not** run at full speed or are not fully functional.
 
 Do you want to check which games are compatible and which ones are not? Please visit our [Compatibility page](https://yuzu-emu.org/game/)!
 
-yuzu is licensed under the GPLv2 (or any later version). Refer to the license.txt file included.
+Yuzu is licensed under the GPLv2 (or any later version). Refer to the license.txt file included.
 
 Check out our [website](https://yuzu-emu.org/)!
 
@@ -22,9 +22,9 @@ For development discussion, please join us on [Discord](https://discord.com/invi
 
 Most of the development happens on GitHub. It's also where [our central repository](https://github.com/yuzu-emu/yuzu) is hosted.
 
-If you want to contribute please take a look at the [Contributor's Guide](https://github.com/yuzu-emu/yuzu/wiki/Contributing) and [Developer Information](https://github.com/yuzu-emu/yuzu/wiki/Developer-Information). You should also contact any of the developers on Discord in order to know about the current state of the emulator.
+If you want to contribute please take a look at the [Contributor's Guide](https://github.com/yuzu-emu/yuzu/wiki/Contributing) and [Developer Information](https://github.com/yuzu-emu/yuzu/wiki/Developer-Information). You may also contact any of the developers on Discord in order to know about the current state of the emulator.
 
-If you want to contribute to the user interface translation, please check out the [yuzu project on transifex](https://www.transifex.com/yuzu-emulator/yuzu). We centralize translation work there, and periodically upstream translations.
+If you want to contribute to the user interface translation, please check out the [yuzu project on transifex](https://www.transifex.com/yuzu-emulator/yuzu). We centralize translation work there and periodically upstream translations.
 
 ### Building
 
@@ -41,4 +41,4 @@ We happily accept monetary donations or donated games and hardware. Please see o
 * Software licenses (e.g. Visual Studio, IDA Pro, etc.)
 * Additional hardware (e.g. GPUs as-needed to improve rendering support, other peripherals to add support for, etc.)
 
-We also more than gladly accept used Switch consoles, preferably ones with firmware 3.0.0 or lower! If you would like to give yours away, don't hesitate to join our [Discord](https://discord.gg/VXqngT3) and talk to bunnei. You may also contact: donations@yuzu-emu.org.
+We also more than gladly accept used Switch consoles, preferably ones with firmware 3.0.0 or lower! If you would like to give yours away, don't hesitate to join our [Discord](https://discord.gg/VXqngT3) and talk to Bunnei. You may also contact: donations@yuzu-emu.org.

dist/qt_themes/qdarkstyle/style.qss

@@ -1371,3 +1371,8 @@ QGroupBox#vibrationGroup::title {
 	padding-left: 1px;
 	padding-right: 1px;
 }
+
+/* touchscreen mapping widget */
+TouchScreenPreview {
+    qproperty-dotHighlightColor: #3daee9;
+}

externals/libusb/CMakeLists.txt

@@ -0,0 +1,150 @@
+add_library(usb STATIC EXCLUDE_FROM_ALL
+    libusb/libusb/core.c
+    libusb/libusb/core.c
+    libusb/libusb/descriptor.c
+    libusb/libusb/hotplug.c
+    libusb/libusb/io.c
+    libusb/libusb/strerror.c
+    libusb/libusb/sync.c
+)
+set_target_properties(usb PROPERTIES VERSION 1.0.23)
+if(WIN32)
+    target_include_directories(usb
+        BEFORE
+        PUBLIC
+          libusb/libusb
+
+        PRIVATE
+          "${CMAKE_CURRENT_BINARY_DIR}"
+    )
+
+    if (NOT MINGW)
+        target_include_directories(usb BEFORE PRIVATE libusb/msvc)
+    endif()
+
+    # Works around other libraries providing their own definition of USB GUIDs (e.g. SDL2)
+    target_compile_definitions(usb PRIVATE "-DGUID_DEVINTERFACE_USB_DEVICE=(GUID){ 0xA5DCBF10, 0x6530, 0x11D2, {0x90, 0x1F, 0x00, 0xC0, 0x4F, 0xB9, 0x51, 0xED}}")
+else()
+target_include_directories(usb
+    # turns out other projects also have "config.h", so make sure the
+    # LibUSB one comes first
+    BEFORE
+
+    PUBLIC
+      libusb/libusb
+
+    PRIVATE
+      "${CMAKE_CURRENT_BINARY_DIR}"
+)
+endif()
+
+if(WIN32 OR CYGWIN)
+    target_sources(usb PRIVATE
+      libusb/libusb/os/threads_windows.c
+      libusb/libusb/os/windows_winusb.c
+      libusb/libusb/os/windows_usbdk.c
+      libusb/libusb/os/windows_nt_common.c
+    )
+    set(OS_WINDOWS TRUE)
+elseif(APPLE)
+    target_sources(usb PRIVATE
+        libusb/libusb/os/darwin_usb.c
+    )
+    find_library(COREFOUNDATION_LIBRARY CoreFoundation)
+    find_library(IOKIT_LIBRARY IOKit)
+    find_library(OBJC_LIBRARY objc)
+    target_link_libraries(usb PRIVATE
+        ${COREFOUNDATION_LIBRARY}
+        ${IOKIT_LIBRARY}
+        ${OBJC_LIBRARY}
+    )
+    set(OS_DARWIN TRUE)
+elseif(ANDROID)
+    target_sources(usb PRIVATE
+        libusb/libusb/os/linux_usbfs.c
+        libusb/libusb/os/linux_netlink.c
+    )
+    find_library(LOG_LIBRARY log)
+    target_link_libraries(usb PRIVATE ${LOG_LIBRARY})
+    set(OS_LINUX TRUE)
+elseif(${CMAKE_SYSTEM_NAME} MATCHES "Linux")
+    target_sources(usb PRIVATE
+        libusb/libusb/os/linux_usbfs.c
+    )
+    find_package(Libudev)
+    if(LIBUDEV_FOUND)
+        target_sources(usb PRIVATE
+            libusb/libusb/os/linux_udev.c
+        )
+        target_link_libraries(usb PRIVATE "${LIBUDEV_LIBRARIES}")
+        target_include_directories(usb PRIVATE "${LIBUDEV_INCLUDE_DIR}")
+        set(HAVE_LIBUDEV TRUE)
+        set(USE_UDEV TRUE)
+    else()
+        target_sources(usb PRIVATE
+            libusb/libusb/os/linux_netlink.c
+        )
+    endif()
+    set(OS_LINUX TRUE)
+elseif(${CMAKE_SYSTEM_NAME} MATCHES "NetBSD")
+    target_sources(usb PRIVATE
+        libusb/libusb/os/netbsd_usb.c
+    )
+    set(OS_NETBSD TRUE)
+elseif(${CMAKE_SYSTEM_NAME} MATCHES "OpenBSD")
+    target_sources(usb PRIVATE
+        libusb/libusb/os/openbsd_usb.c
+    )
+    set(OS_OPENBSD TRUE)
+endif()
+
+if(UNIX)
+    target_sources(usb PRIVATE
+        libusb/libusb/os/poll_posix.c
+        libusb/libusb/os/threads_posix.c
+    )
+    find_package(Threads REQUIRED)
+    if(THREADS_HAVE_PTHREAD_ARG)
+      target_compile_options(usb PUBLIC "-pthread")
+    endif()
+    if(CMAKE_THREAD_LIBS_INIT)
+      target_link_libraries(usb PRIVATE "${CMAKE_THREAD_LIBS_INIT}")
+    endif()
+    set(THREADS_POSIX TRUE)
+elseif(WIN32)
+    target_sources(usb PRIVATE
+        libusb/libusb/os/poll_windows.c
+        libusb/libusb/os/threads_windows.c
+    )
+endif()
+
+include(CheckFunctionExists)
+include(CheckIncludeFiles)
+include(CheckTypeSize)
+check_include_files(asm/types.h HAVE_ASM_TYPES_H)
+check_function_exists(gettimeofday HAVE_GETTIMEOFDAY)
+check_include_files(linux/filter.h HAVE_LINUX_FILTER_H)
+check_include_files(linux/netlink.h HAVE_LINUX_NETLINK_H)
+check_include_files(poll.h HAVE_POLL_H)
+check_include_files(signal.h HAVE_SIGNAL_H)
+check_include_files(strings.h HAVE_STRINGS_H)
+check_type_size("struct timespec" STRUCT_TIMESPEC)
+check_function_exists(syslog HAVE_SYSLOG_FUNC)
+check_include_files(syslog.h HAVE_SYSLOG_H)
+check_include_files(sys/socket.h HAVE_SYS_SOCKET_H)
+check_include_files(sys/time.h HAVE_SYS_TIME_H)
+check_include_files(sys/types.h HAVE_SYS_TYPES_H)
+
+set(CMAKE_EXTRA_INCLUDE_FILES poll.h)
+check_type_size("nfds_t" nfds_t)
+unset(CMAKE_EXTRA_INCLUDE_FILES)
+if(HAVE_NFDS_T)
+    set(POLL_NFDS_TYPE "nfds_t")
+else()
+    set(POLL_NFDS_TYPE "unsigned int")
+endif()
+
+check_include_files(sys/timerfd.h USBI_TIMERFD_AVAILABLE)
+
+
+configure_file(config.h.in config.h)

externals/libusb/config.h.in

@@ -0,0 +1,90 @@
+/* Default visibility */
+#if defined(__GNUC__) || defined(__clang__)
+  #define DEFAULT_VISIBILITY __attribute__((visibility("default")))
+#elif defined(_MSC_VER)
+  #define DEFAULT_VISIBILITY __declspec(dllexport)
+#endif
+
+/* Start with debug message logging enabled */
+#undef ENABLE_DEBUG_LOGGING
+
+/* Message logging */
+#undef ENABLE_LOGGING
+
+/* Define to 1 if you have the <asm/types.h> header file. */
+#cmakedefine HAVE_ASM_TYPES_H 1
+
+/* Define to 1 if you have the `gettimeofday' function. */
+#cmakedefine HAVE_GETTIMEOFDAY 1
+
+/* Define to 1 if you have the `udev' library (-ludev). */
+#cmakedefine HAVE_LIBUDEV 1
+
+/* Define to 1 if you have the <linux/filter.h> header file. */
+#cmakedefine HAVE_LINUX_FILTER_H 1
+
+/* Define to 1 if you have the <linux/netlink.h> header file. */
+#cmakedefine HAVE_LINUX_NETLINK_H 1
+
+/* Define to 1 if you have the <poll.h> header file. */
+#cmakedefine HAVE_POLL_H 1
+
+/* Define to 1 if you have the <signal.h> header file. */
+#cmakedefine HAVE_SIGNAL_H 1
+
+/* Define to 1 if you have the <strings.h> header file. */
+#cmakedefine HAVE_STRINGS_H 1
+
+/* Define to 1 if the system has the type `struct timespec'. */
+#cmakedefine HAVE_STRUCT_TIMESPEC 1
+
+/* syslog() function available */
+#cmakedefine HAVE_SYSLOG_FUNC 1
+
+/* Define to 1 if you have the <syslog.h> header file. */
+#cmakedefine HAVE_SYSLOG_H 1
+
+/* Define to 1 if you have the <sys/socket.h> header file. */
+#cmakedefine HAVE_SYS_SOCKET_H 1
+
+/* Define to 1 if you have the <sys/time.h> header file. */
+#cmakedefine HAVE_SYS_TIME_H 1
+
+/* Define to 1 if you have the <sys/types.h> header file. */
+#cmakedefine HAVE_SYS_TYPES_H 1
+
+/* Darwin backend */
+#cmakedefine OS_DARWIN 1
+
+/* Linux backend */
+#cmakedefine OS_LINUX 1
+
+/* NetBSD backend */
+#cmakedefine OS_NETBSD 1
+
+/* OpenBSD backend */
+#cmakedefine OS_OPENBSD 1
+
+/* Windows backend */
+#cmakedefine OS_WINDOWS 1
+
+/* type of second poll() argument */
+#define POLL_NFDS_TYPE @POLL_NFDS_TYPE@
+
+/* Use POSIX Threads */
+#cmakedefine THREADS_POSIX
+
+/* timerfd headers available */
+#cmakedefine USBI_TIMERFD_AVAILABLE 1
+
+/* Enable output to system log */
+#define USE_SYSTEM_LOGGING_FACILITY 1
+
+/* Use udev for device enumeration/hotplug */
+#cmakedefine USE_UDEV 1
+
+/* Use GNU extensions */
+#define _GNU_SOURCE
+
+/* Oldest Windows version supported */
+#define WINVER 0x0501

externals/microprofile/microprofile.h

@@ -1037,7 +1037,7 @@ static void MicroProfileCreateThreadLogKey()
 #else
 MP_THREAD_LOCAL MicroProfileThreadLog* g_MicroProfileThreadLog = 0;
 #endif
-static bool g_bUseLock = false; /// This is used because windows does not support using mutexes under dll init(which is where global initialization is handled)
+static std::atomic<bool> g_bUseLock{false}; /// This is used because windows does not support using mutexes under dll init(which is where global initialization is handled)
 
 
 MICROPROFILE_DEFINE(g_MicroProfileFlip, "MicroProfile", "MicroProfileFlip", 0x3355ee);

src/common/common_funcs.h

@@ -64,14 +64,20 @@ __declspec(dllimport) void __stdcall DebugBreak(void);
         using T = std::underlying_type_t<type>;                                                    \
         return static_cast<type>(static_cast<T>(a) & static_cast<T>(b));                           \
     }                                                                                              \
-    constexpr type& operator|=(type& a, type b) noexcept {                                         \
+    [[nodiscard]] constexpr type operator^(type a, type b) noexcept {                              \
         using T = std::underlying_type_t<type>;                                                    \
-        a = static_cast<type>(static_cast<T>(a) | static_cast<T>(b));                              \
+        return static_cast<type>(static_cast<T>(a) ^ static_cast<T>(b));                           \
+    }                                                                                              \
+    constexpr type& operator|=(type& a, type b) noexcept {                                         \
+        a = a | b;                                                                                 \
         return a;                                                                                  \
     }                                                                                              \
     constexpr type& operator&=(type& a, type b) noexcept {                                         \
-        using T = std::underlying_type_t<type>;                                                    \
-        a = static_cast<type>(static_cast<T>(a) & static_cast<T>(b));                              \
+        a = a & b;                                                                                 \
+        return a;                                                                                  \
+    }                                                                                              \
+    constexpr type& operator^=(type& a, type b) noexcept {                                         \
+        a = a ^ b;                                                                                 \
         return a;                                                                                  \
     }                                                                                              \
     [[nodiscard]] constexpr type operator~(type key) noexcept {                                    \

src/common/math_util.h

@@ -9,7 +9,7 @@
 
 namespace Common {
 
-constexpr float PI = 3.14159265f;
+constexpr float PI = 3.1415926535f;
 
 template <class T>
 struct Rectangle {

src/common/quaternion.h

@@ -36,6 +36,36 @@ public:
         T length = std::sqrt(xyz.Length2() + w * w);
         return {xyz / length, w / length};
     }
+
+    [[nodiscard]] std::array<decltype(-T{}), 16> ToMatrix() const {
+        const T x2 = xyz[0] * xyz[0];
+        const T y2 = xyz[1] * xyz[1];
+        const T z2 = xyz[2] * xyz[2];
+
+        const T xy = xyz[0] * xyz[1];
+        const T wz = w * xyz[2];
+        const T xz = xyz[0] * xyz[2];
+        const T wy = w * xyz[1];
+        const T yz = xyz[1] * xyz[2];
+        const T wx = w * xyz[0];
+
+        return {1.0f - 2.0f * (y2 + z2),
+                2.0f * (xy + wz),
+                2.0f * (xz - wy),
+                0.0f,
+                2.0f * (xy - wz),
+                1.0f - 2.0f * (x2 + z2),
+                2.0f * (yz + wx),
+                0.0f,
+                2.0f * (xz + wy),
+                2.0f * (yz - wx),
+                1.0f - 2.0f * (x2 + y2),
+                0.0f,
+                0.0f,
+                0.0f,
+                0.0f,
+                1.0f};
+    }
 };
 
 template <typename T>

src/common/thread.cpp

@@ -2,6 +2,8 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include "common/common_funcs.h"
+#include "common/logging/log.h"
 #include "common/thread.h"
 #ifdef __APPLE__
 #include <mach/mach.h>
@@ -19,6 +21,8 @@
 #include <unistd.h>
 #endif
 
+#include <string>
+
 #ifdef __FreeBSD__
 #define cpu_set_t cpuset_t
 #endif
@@ -110,6 +114,14 @@ void SetCurrentThreadName(const char* name) {
     pthread_set_name_np(pthread_self(), name);
 #elif defined(__NetBSD__)
     pthread_setname_np(pthread_self(), "%s", (void*)name);
+#elif defined(__linux__)
+    // Linux limits thread names to 15 characters and will outright reject any
+    // attempt to set a longer name with ERANGE.
+    std::string truncated(name, std::min(strlen(name), static_cast<size_t>(15)));
+    if (int e = pthread_setname_np(pthread_self(), truncated.c_str())) {
+        errno = e;
+        LOG_ERROR(Common, "Failed to set thread name to '{}': {}", truncated, GetLastErrorMsg());
+    }
 #else
     pthread_setname_np(pthread_self(), name);
 #endif

src/common/thread.h

@@ -4,6 +4,7 @@
 
 #pragma once
 
+#include <atomic>
 #include <chrono>
 #include <condition_variable>
 #include <cstddef>
@@ -25,13 +26,13 @@ public:
 
     void Wait() {
         std::unique_lock lk{mutex};
-        condvar.wait(lk, [&] { return is_set; });
+        condvar.wait(lk, [&] { return is_set.load(); });
         is_set = false;
     }
 
     bool WaitFor(const std::chrono::nanoseconds& time) {
         std::unique_lock lk{mutex};
-        if (!condvar.wait_for(lk, time, [this] { return is_set; }))
+        if (!condvar.wait_for(lk, time, [this] { return is_set.load(); }))
             return false;
         is_set = false;
         return true;
@@ -40,7 +41,7 @@ public:
     template <class Clock, class Duration>
     bool WaitUntil(const std::chrono::time_point<Clock, Duration>& time) {
         std::unique_lock lk{mutex};
-        if (!condvar.wait_until(lk, time, [this] { return is_set; }))
+        if (!condvar.wait_until(lk, time, [this] { return is_set.load(); }))
             return false;
         is_set = false;
         return true;
@@ -54,9 +55,9 @@ public:
     }
 
 private:
-    bool is_set = false;
     std::condition_variable condvar;
     std::mutex mutex;
+    std::atomic_bool is_set{false};
 };
 
 class Barrier {

src/common/x64/xbyak_abi.h

@@ -11,7 +11,7 @@
 
 namespace Common::X64 {
 
-inline std::size_t RegToIndex(const Xbyak::Reg& reg) {
+constexpr std::size_t RegToIndex(const Xbyak::Reg& reg) {
     using Kind = Xbyak::Reg::Kind;
     ASSERT_MSG((reg.getKind() & (Kind::REG | Kind::XMM)) != 0,
                "RegSet only support GPRs and XMM registers.");
@@ -19,17 +19,17 @@ inline std::size_t RegToIndex(const Xbyak::Reg& reg) {
     return reg.getIdx() + (reg.getKind() == Kind::REG ? 0 : 16);
 }
 
-inline Xbyak::Reg64 IndexToReg64(std::size_t reg_index) {
+constexpr Xbyak::Reg64 IndexToReg64(std::size_t reg_index) {
     ASSERT(reg_index < 16);
     return Xbyak::Reg64(static_cast<int>(reg_index));
 }
 
-inline Xbyak::Xmm IndexToXmm(std::size_t reg_index) {
+constexpr Xbyak::Xmm IndexToXmm(std::size_t reg_index) {
     ASSERT(reg_index >= 16 && reg_index < 32);
     return Xbyak::Xmm(static_cast<int>(reg_index - 16));
 }
 
-inline Xbyak::Reg IndexToReg(std::size_t reg_index) {
+constexpr Xbyak::Reg IndexToReg(std::size_t reg_index) {
     if (reg_index < 16) {
         return IndexToReg64(reg_index);
     } else {
@@ -45,17 +45,17 @@ inline std::bitset<32> BuildRegSet(std::initializer_list<Xbyak::Reg> regs) {
     return bits;
 }
 
-const std::bitset<32> ABI_ALL_GPRS(0x0000FFFF);
-const std::bitset<32> ABI_ALL_XMMS(0xFFFF0000);
+constexpr inline std::bitset<32> ABI_ALL_GPRS(0x0000FFFF);
+constexpr inline std::bitset<32> ABI_ALL_XMMS(0xFFFF0000);
 
 #ifdef _WIN32
 
 // Microsoft x64 ABI
-const Xbyak::Reg ABI_RETURN = Xbyak::util::rax;
-const Xbyak::Reg ABI_PARAM1 = Xbyak::util::rcx;
-const Xbyak::Reg ABI_PARAM2 = Xbyak::util::rdx;
-const Xbyak::Reg ABI_PARAM3 = Xbyak::util::r8;
-const Xbyak::Reg ABI_PARAM4 = Xbyak::util::r9;
+constexpr inline Xbyak::Reg ABI_RETURN = Xbyak::util::rax;
+constexpr inline Xbyak::Reg ABI_PARAM1 = Xbyak::util::rcx;
+constexpr inline Xbyak::Reg ABI_PARAM2 = Xbyak::util::rdx;
+constexpr inline Xbyak::Reg ABI_PARAM3 = Xbyak::util::r8;
+constexpr inline Xbyak::Reg ABI_PARAM4 = Xbyak::util::r9;
 
 const std::bitset<32> ABI_ALL_CALLER_SAVED = BuildRegSet({
     // GPRs
@@ -102,11 +102,11 @@ constexpr size_t ABI_SHADOW_SPACE = 0x20;
 #else
 
 // System V x86-64 ABI
-const Xbyak::Reg ABI_RETURN = Xbyak::util::rax;
-const Xbyak::Reg ABI_PARAM1 = Xbyak::util::rdi;
-const Xbyak::Reg ABI_PARAM2 = Xbyak::util::rsi;
-const Xbyak::Reg ABI_PARAM3 = Xbyak::util::rdx;
-const Xbyak::Reg ABI_PARAM4 = Xbyak::util::rcx;
+constexpr inline Xbyak::Reg ABI_RETURN = Xbyak::util::rax;
+constexpr inline Xbyak::Reg ABI_PARAM1 = Xbyak::util::rdi;
+constexpr inline Xbyak::Reg ABI_PARAM2 = Xbyak::util::rsi;
+constexpr inline Xbyak::Reg ABI_PARAM3 = Xbyak::util::rdx;
+constexpr inline Xbyak::Reg ABI_PARAM4 = Xbyak::util::rcx;
 
 const std::bitset<32> ABI_ALL_CALLER_SAVED = BuildRegSet({
     // GPRs

src/core/CMakeLists.txt

@@ -491,6 +491,7 @@ add_library(core STATIC
     hle/service/sm/controller.h
     hle/service/sm/sm.cpp
     hle/service/sm/sm.h
+    hle/service/sockets/blocking_worker.h
     hle/service/sockets/bsd.cpp
     hle/service/sockets/bsd.h
     hle/service/sockets/ethc.cpp
@@ -501,6 +502,8 @@ add_library(core STATIC
     hle/service/sockets/sfdnsres.h
     hle/service/sockets/sockets.cpp
     hle/service/sockets/sockets.h
+    hle/service/sockets/sockets_translate.cpp
+    hle/service/sockets/sockets_translate.h
     hle/service/spl/csrng.cpp
     hle/service/spl/csrng.h
     hle/service/spl/module.cpp

src/core/core.cpp

@@ -188,7 +188,6 @@ struct System::Impl {
         if (!gpu_core) {
             return ResultStatus::ErrorVideoCore;
         }
-        gpu_core->Renderer().Rasterizer().SetupDirtyFlags();
 
         is_powered_on = true;
         exit_lock = false;

src/core/cpu_manager.cpp

@@ -328,7 +328,7 @@ void CpuManager::RunThread(std::size_t core) {
     system.RegisterCoreThread(core);
     std::string name;
     if (is_multicore) {
-        name = "yuzu:CoreCPUThread_" + std::to_string(core);
+        name = "yuzu:CPUCore_" + std::to_string(core);
     } else {
         name = "yuzu:CPUThread";
     }

src/core/crypto/partition_data_manager.cpp

@@ -26,6 +26,7 @@
 #include "core/file_sys/vfs.h"
 #include "core/file_sys/vfs_offset.h"
 #include "core/file_sys/vfs_vector.h"
+#include "core/loader/loader.h"
 
 using Common::AsArray;
 

src/core/file_sys/bis_factory.cpp

@@ -8,7 +8,6 @@
 #include "core/file_sys/bis_factory.h"
 #include "core/file_sys/mode.h"
 #include "core/file_sys/registered_cache.h"
-#include "core/settings.h"
 
 namespace FileSys {
 

src/core/file_sys/bis_factory.h

@@ -6,7 +6,8 @@
 
 #include <memory>
 
-#include "core/file_sys/vfs.h"
+#include "common/common_types.h"
+#include "core/file_sys/vfs_types.h"
 
 namespace FileSys {
 

src/core/file_sys/card_image.cpp

@@ -8,11 +8,11 @@
 #include <fmt/ostream.h>
 
 #include "common/logging/log.h"
+#include "core/crypto/key_manager.h"
 #include "core/file_sys/card_image.h"
 #include "core/file_sys/content_archive.h"
 #include "core/file_sys/nca_metadata.h"
 #include "core/file_sys/partition_filesystem.h"
-#include "core/file_sys/romfs.h"
 #include "core/file_sys/submission_package.h"
 #include "core/file_sys/vfs_concat.h"
 #include "core/file_sys/vfs_offset.h"
@@ -31,7 +31,8 @@ constexpr std::array partition_names{
 
 XCI::XCI(VirtualFile file_)
     : file(std::move(file_)), program_nca_status{Loader::ResultStatus::ErrorXCIMissingProgramNCA},
-      partitions(partition_names.size()), partitions_raw(partition_names.size()) {
+      partitions(partition_names.size()),
+      partitions_raw(partition_names.size()), keys{Core::Crypto::KeyManager::Instance()} {
     if (file->ReadObject(&header) != sizeof(GamecardHeader)) {
         status = Loader::ResultStatus::ErrorBadXCIHeader;
         return;

src/core/file_sys/card_image.h

@@ -9,9 +9,12 @@
 #include <vector>
 #include "common/common_types.h"
 #include "common/swap.h"
-#include "core/crypto/key_manager.h"
 #include "core/file_sys/vfs.h"
 
+namespace Core::Crypto {
+class KeyManager;
+}
+
 namespace Loader {
 enum class ResultStatus : u16;
 }
@@ -140,6 +143,6 @@ private:
 
     u64 update_normal_partition_end;
 
-    Core::Crypto::KeyManager& keys = Core::Crypto::KeyManager::Instance();
+    Core::Crypto::KeyManager& keys;
 };
 } // namespace FileSys

src/core/file_sys/content_archive.cpp

@@ -10,10 +10,10 @@
 #include "common/logging/log.h"
 #include "core/crypto/aes_util.h"
 #include "core/crypto/ctr_encryption_layer.h"
+#include "core/crypto/key_manager.h"
 #include "core/file_sys/content_archive.h"
 #include "core/file_sys/nca_patch.h"
 #include "core/file_sys/partition_filesystem.h"
-#include "core/file_sys/romfs.h"
 #include "core/file_sys/vfs_offset.h"
 #include "core/loader/loader.h"
 
@@ -119,7 +119,8 @@ static bool IsValidNCA(const NCAHeader& header) {
 }
 
 NCA::NCA(VirtualFile file_, VirtualFile bktr_base_romfs_, u64 bktr_base_ivfc_offset)
-    : file(std::move(file_)), bktr_base_romfs(std::move(bktr_base_romfs_)) {
+    : file(std::move(file_)),
+      bktr_base_romfs(std::move(bktr_base_romfs_)), keys{Core::Crypto::KeyManager::Instance()} {
     if (file == nullptr) {
         status = Loader::ResultStatus::ErrorNullFile;
         return;

src/core/file_sys/content_archive.h

@@ -158,7 +158,7 @@ private:
     bool encrypted = false;
     bool is_update = false;
 
-    Core::Crypto::KeyManager& keys = Core::Crypto::KeyManager::Instance();
+    Core::Crypto::KeyManager& keys;
 };
 
 } // namespace FileSys

src/core/file_sys/control_metadata.cpp

@@ -5,6 +5,7 @@
 #include "common/string_util.h"
 #include "common/swap.h"
 #include "core/file_sys/control_metadata.h"
+#include "core/file_sys/vfs.h"
 
 namespace FileSys {
 

src/core/file_sys/control_metadata.h

@@ -10,7 +10,7 @@
 #include "common/common_funcs.h"
 #include "common/common_types.h"
 #include "common/swap.h"
-#include "core/file_sys/vfs.h"
+#include "core/file_sys/vfs_types.h"
 
 namespace FileSys {
 

src/core/file_sys/kernel_executable.cpp

@@ -2,9 +2,12 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <cstring>
+
 #include "common/string_util.h"
 #include "core/file_sys/kernel_executable.h"
 #include "core/file_sys/vfs_offset.h"
+#include "core/loader/loader.h"
 
 namespace FileSys {
 

src/core/file_sys/kernel_executable.h

@@ -4,10 +4,17 @@
 
 #pragma once
 
+#include <array>
+#include <vector>
+
 #include "common/common_funcs.h"
+#include "common/common_types.h"
 #include "common/swap.h"
 #include "core/file_sys/vfs_types.h"
-#include "core/loader/loader.h"
+
+namespace Loader {
+enum class ResultStatus : u16;
+}
 
 namespace FileSys {
 

src/core/file_sys/nca_metadata.cpp

@@ -7,6 +7,7 @@
 #include "common/logging/log.h"
 #include "common/swap.h"
 #include "core/file_sys/nca_metadata.h"
+#include "core/file_sys/vfs.h"
 
 namespace FileSys {
 

src/core/file_sys/nca_metadata.h

@@ -10,7 +10,7 @@
 #include "common/common_funcs.h"
 #include "common/common_types.h"
 #include "common/swap.h"
-#include "core/file_sys/vfs.h"
+#include "core/file_sys/vfs_types.h"
 
 namespace FileSys {
 class CNMT;

src/core/file_sys/partition_filesystem.cpp

@@ -21,7 +21,7 @@ bool PartitionFilesystem::Header::HasValidMagicValue() const {
            magic == Common::MakeMagic('P', 'F', 'S', '0');
 }
 
-PartitionFilesystem::PartitionFilesystem(std::shared_ptr<VfsFile> file) {
+PartitionFilesystem::PartitionFilesystem(VirtualFile file) {
     // At least be as large as the header
     if (file->GetSize() < sizeof(Header)) {
         status = Loader::ResultStatus::ErrorBadPFSHeader;
@@ -89,11 +89,11 @@ std::map<std::string, u64> PartitionFilesystem::GetFileSizes() const {
     return sizes;
 }
 
-std::vector<std::shared_ptr<VfsFile>> PartitionFilesystem::GetFiles() const {
+std::vector<VirtualFile> PartitionFilesystem::GetFiles() const {
     return pfs_files;
 }
 
-std::vector<std::shared_ptr<VfsDirectory>> PartitionFilesystem::GetSubdirectories() const {
+std::vector<VirtualDir> PartitionFilesystem::GetSubdirectories() const {
     return {};
 }
 
@@ -101,7 +101,7 @@ std::string PartitionFilesystem::GetName() const {
     return is_hfs ? "HFS0" : "PFS0";
 }
 
-std::shared_ptr<VfsDirectory> PartitionFilesystem::GetParentDirectory() const {
+VirtualDir PartitionFilesystem::GetParentDirectory() const {
     // TODO(DarkLordZach): Add support for nested containers.
     return nullptr;
 }

src/core/file_sys/partition_filesystem.h

@@ -24,7 +24,7 @@ namespace FileSys {
  */
 class PartitionFilesystem : public ReadOnlyVfsDirectory {
 public:
-    explicit PartitionFilesystem(std::shared_ptr<VfsFile> file);
+    explicit PartitionFilesystem(VirtualFile file);
     ~PartitionFilesystem() override;
 
     Loader::ResultStatus GetStatus() const;
@@ -32,10 +32,10 @@ public:
     std::map<std::string, u64> GetFileOffsets() const;
     std::map<std::string, u64> GetFileSizes() const;
 
-    std::vector<std::shared_ptr<VfsFile>> GetFiles() const override;
-    std::vector<std::shared_ptr<VfsDirectory>> GetSubdirectories() const override;
+    std::vector<VirtualFile> GetFiles() const override;
+    std::vector<VirtualDir> GetSubdirectories() const override;
     std::string GetName() const override;
-    std::shared_ptr<VfsDirectory> GetParentDirectory() const override;
+    VirtualDir GetParentDirectory() const override;
     void PrintDebugInfo() const;
 
 private:

src/core/file_sys/patch_manager.cpp

@@ -49,8 +49,7 @@ std::string FormatTitleVersion(u32 version, TitleVersionFormat format) {
     return fmt::format("v{}.{}.{}", bytes[3], bytes[2], bytes[1]);
 }
 
-std::shared_ptr<VfsDirectory> FindSubdirectoryCaseless(const std::shared_ptr<VfsDirectory> dir,
-                                                       std::string_view name) {
+VirtualDir FindSubdirectoryCaseless(const VirtualDir dir, std::string_view name) {
 #ifdef _WIN32
     return dir->GetSubdirectory(name);
 #else

src/core/file_sys/patch_manager.h

@@ -6,10 +6,11 @@
 
 #include <map>
 #include <memory>
+#include <optional>
 #include <string>
 #include "common/common_types.h"
 #include "core/file_sys/nca_metadata.h"
-#include "core/file_sys/vfs.h"
+#include "core/file_sys/vfs_types.h"
 #include "core/memory/dmnt_cheat_types.h"
 
 namespace Core {
@@ -31,8 +32,7 @@ std::string FormatTitleVersion(u32 version,
 
 // Returns a directory with name matching name case-insensitive. Returns nullptr if directory
 // doesn't have a directory with name.
-std::shared_ptr<VfsDirectory> FindSubdirectoryCaseless(const std::shared_ptr<VfsDirectory> dir,
-                                                       std::string_view name);
+VirtualDir FindSubdirectoryCaseless(VirtualDir dir, std::string_view name);
 
 // A centralized class to manage patches to games.
 class PatchManager {

src/core/file_sys/program_metadata.cpp

@@ -7,6 +7,7 @@
 
 #include "common/logging/log.h"
 #include "core/file_sys/program_metadata.h"
+#include "core/file_sys/vfs.h"
 #include "core/loader/loader.h"
 
 namespace FileSys {

src/core/file_sys/program_metadata.h

@@ -9,7 +9,7 @@
 #include "common/bit_field.h"
 #include "common/common_types.h"
 #include "common/swap.h"
-#include "core/file_sys/vfs.h"
+#include "core/file_sys/vfs_types.h"
 
 namespace Loader {
 enum class ResultStatus : u16;

src/core/file_sys/romfs.h

@@ -4,7 +4,6 @@
 
 #pragma once
 
-#include <array>
 #include "core/file_sys/vfs.h"
 
 namespace FileSys {

src/core/file_sys/sdmc_factory.cpp

@@ -5,8 +5,8 @@
 #include <memory>
 #include "core/file_sys/registered_cache.h"
 #include "core/file_sys/sdmc_factory.h"
+#include "core/file_sys/vfs.h"
 #include "core/file_sys/xts_archive.h"
-#include "core/settings.h"
 
 namespace FileSys {
 

src/core/file_sys/sdmc_factory.h

@@ -5,7 +5,7 @@
 #pragma once
 
 #include <memory>
-#include "core/file_sys/vfs.h"
+#include "core/file_sys/vfs_types.h"
 #include "core/hle/result.h"
 
 namespace FileSys {

src/core/file_sys/submission_package.cpp

@@ -54,7 +54,7 @@ void SetTicketKeys(const std::vector<VirtualFile>& files) {
 
 NSP::NSP(VirtualFile file_)
     : file(std::move(file_)), status{Loader::ResultStatus::Success},
-      pfs(std::make_shared<PartitionFilesystem>(file)) {
+      pfs(std::make_shared<PartitionFilesystem>(file)), keys{Core::Crypto::KeyManager::Instance()} {
     if (pfs->GetStatus() != Loader::ResultStatus::Success) {
         status = pfs->GetStatus();
         return;

src/core/file_sys/submission_package.h

@@ -10,6 +10,10 @@
 #include "common/common_types.h"
 #include "core/file_sys/vfs.h"
 
+namespace Core::Crypto {
+class KeyManager;
+}
+
 namespace Loader {
 enum class ResultStatus : u16;
 }
@@ -73,7 +77,7 @@ private:
     std::map<u64, std::map<std::pair<TitleType, ContentRecordType>, std::shared_ptr<NCA>>> ncas;
     std::vector<VirtualFile> ticket_files;
 
-    Core::Crypto::KeyManager& keys = Core::Crypto::KeyManager::Instance();
+    Core::Crypto::KeyManager& keys;
 
     VirtualFile romfs;
     VirtualDir exefs;

src/core/file_sys/xts_archive.cpp

@@ -15,8 +15,9 @@
 #include "common/hex_util.h"
 #include "common/string_util.h"
 #include "core/crypto/aes_util.h"
+#include "core/crypto/key_manager.h"
 #include "core/crypto/xts_encryption_layer.h"
-#include "core/file_sys/partition_filesystem.h"
+#include "core/file_sys/content_archive.h"
 #include "core/file_sys/vfs_offset.h"
 #include "core/file_sys/xts_archive.h"
 #include "core/loader/loader.h"
@@ -43,7 +44,9 @@ static bool CalculateHMAC256(Destination* out, const SourceKey* key, std::size_t
     return true;
 }
 
-NAX::NAX(VirtualFile file_) : header(std::make_unique<NAXHeader>()), file(std::move(file_)) {
+NAX::NAX(VirtualFile file_)
+    : header(std::make_unique<NAXHeader>()),
+      file(std::move(file_)), keys{Core::Crypto::KeyManager::Instance()} {
     std::string path = Common::FS::SanitizePath(file->GetFullPath());
     static const std::regex nax_path_regex("/registered/(000000[0-9A-F]{2})/([0-9A-F]{32})\\.nca",
                                            std::regex_constants::ECMAScript |
@@ -60,7 +63,8 @@ NAX::NAX(VirtualFile file_) : header(std::make_unique<NAXHeader>()), file(std::m
 }
 
 NAX::NAX(VirtualFile file_, std::array<u8, 0x10> nca_id)
-    : header(std::make_unique<NAXHeader>()), file(std::move(file_)) {
+    : header(std::make_unique<NAXHeader>()),
+      file(std::move(file_)), keys{Core::Crypto::KeyManager::Instance()} {
     Core::Crypto::SHA256Hash hash{};
     mbedtls_sha256_ret(nca_id.data(), nca_id.size(), hash.data(), 0);
     status = Parse(fmt::format("/registered/000000{:02X}/{}.nca", hash[0],

src/core/file_sys/xts_archive.h

@@ -9,12 +9,16 @@
 #include "common/common_types.h"
 #include "common/swap.h"
 #include "core/crypto/key_manager.h"
-#include "core/file_sys/content_archive.h"
 #include "core/file_sys/vfs.h"
-#include "core/loader/loader.h"
+
+namespace Loader {
+enum class ResultStatus : u16;
+}
 
 namespace FileSys {
 
+class NCA;
+
 struct NAXHeader {
     std::array<u8, 0x20> hmac;
     u64_le magic;
@@ -62,6 +66,6 @@ private:
 
     VirtualFile dec_file;
 
-    Core::Crypto::KeyManager& keys = Core::Crypto::KeyManager::Instance();
+    Core::Crypto::KeyManager& keys;
 };
 } // namespace FileSys

src/core/frontend/framebuffer_layout.h

@@ -4,6 +4,7 @@
 
 #pragma once
 
+#include "common/common_types.h"
 #include "common/math_util.h"
 
 namespace Layout {

src/core/hle/kernel/kernel.cpp

@@ -219,6 +219,7 @@ struct KernelCore::Impl {
                 return static_cast<u32>(system.GetCpuManager().CurrentCore());
             }
         }
+        std::unique_lock lock{register_thread_mutex};
         const auto it = host_thread_ids.find(this_id);
         if (it == host_thread_ids.end()) {
             return Core::INVALID_HOST_THREAD_ID;
@@ -324,7 +325,7 @@ struct KernelCore::Impl {
     std::unordered_map<std::thread::id, u32> host_thread_ids;
     u32 registered_thread_ids{Core::Hardware::NUM_CPU_CORES};
     std::bitset<Core::Hardware::NUM_CPU_CORES> registered_core_threads;
-    std::mutex register_thread_mutex;
+    mutable std::mutex register_thread_mutex;
 
     // Kernel memory management
     std::unique_ptr<Memory::MemoryManager> memory_manager;

src/core/hle/kernel/scheduler.cpp

@@ -756,7 +756,11 @@ void Scheduler::SwitchToCurrent() {
             current_thread = selected_thread;
             is_context_switch_pending = false;
         }
-        while (!is_context_switch_pending) {
+        const auto is_switch_pending = [this] {
+            std::scoped_lock lock{guard};
+            return is_context_switch_pending;
+        };
+        do {
             if (current_thread != nullptr && !current_thread->IsHLEThread()) {
                 current_thread->context_guard.lock();
                 if (!current_thread->IsRunnable()) {
@@ -775,7 +779,7 @@ void Scheduler::SwitchToCurrent() {
                 next_context = &idle_thread->GetHostContext();
             }
             Common::Fiber::YieldTo(switch_fiber, *next_context);
-        }
+        } while (!is_switch_pending());
     }
 }
 

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

@@ -844,8 +844,7 @@ void FSP_SRV::OpenSaveDataFileSystem(Kernel::HLERequestContext& ctx) {
         return;
     }
 
-    FileSys::StorageId id;
-
+    FileSys::StorageId id{};
     switch (parameters.space_id) {
     case FileSys::SaveDataSpaceId::NandUser:
         id = FileSys::StorageId::NandUser;
@@ -857,6 +856,10 @@ void FSP_SRV::OpenSaveDataFileSystem(Kernel::HLERequestContext& ctx) {
     case FileSys::SaveDataSpaceId::NandSystem:
         id = FileSys::StorageId::NandSystem;
         break;
+    case FileSys::SaveDataSpaceId::TemporaryStorage:
+    case FileSys::SaveDataSpaceId::ProperSystem:
+    case FileSys::SaveDataSpaceId::SafeMode:
+        UNREACHABLE();
     }
 
     auto filesystem =
@@ -902,7 +905,14 @@ void FSP_SRV::ReadSaveDataFileSystemExtraDataWithMaskBySaveDataAttribute(
     // Stub this to None for now, backend needs an impl to read/write the SaveDataExtraData
     constexpr auto flags = static_cast<u32>(FileSys::SaveDataFlags::None);
 
-    LOG_WARNING(Service_FS, "(STUBBED) called, flags={}", flags);
+    LOG_WARNING(Service_FS,
+                "(STUBBED) called, flags={}, space_id={}, attribute.title_id={:016X}\n"
+                "attribute.user_id={:016X}{:016X}, attribute.save_id={:016X}\n"
+                "attribute.type={}, attribute.rank={}, attribute.index={}",
+                flags, static_cast<u32>(parameters.space_id), parameters.attribute.title_id,
+                parameters.attribute.user_id[1], parameters.attribute.user_id[0],
+                parameters.attribute.save_id, static_cast<u32>(parameters.attribute.type),
+                static_cast<u32>(parameters.attribute.rank), parameters.attribute.index);
 
     IPC::ResponseBuilder rb{ctx, 3};
     rb.Push(RESULT_SUCCESS);

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

@@ -574,6 +574,22 @@ Controller_NPad::GyroscopeZeroDriftMode Controller_NPad::GetGyroscopeZeroDriftMo
     return gyroscope_zero_drift_mode;
 }
 
+void Controller_NPad::MergeSingleJoyAsDualJoy(u32 npad_id_1, u32 npad_id_2) {
+    const auto npad_index_1 = NPadIdToIndex(npad_id_1);
+    const auto npad_index_2 = NPadIdToIndex(npad_id_2);
+
+    // If the controllers at both npad indices form a pair of left and right joycons, merge them.
+    // Otherwise, do nothing.
+    if ((connected_controllers[npad_index_1].type == NPadControllerType::JoyLeft &&
+         connected_controllers[npad_index_2].type == NPadControllerType::JoyRight) ||
+        (connected_controllers[npad_index_2].type == NPadControllerType::JoyLeft &&
+         connected_controllers[npad_index_1].type == NPadControllerType::JoyRight)) {
+        // Disconnect the joycon at the second id and connect the dual joycon at the first index.
+        DisconnectNPad(npad_id_2);
+        AddNewControllerAt(NPadControllerType::JoyDual, npad_index_1);
+    }
+}
+
 void Controller_NPad::StartLRAssignmentMode() {
     // Nothing internally is used for lr assignment mode. Since we have the ability to set the
     // controller types from boot, it doesn't really matter about showing a selection screen

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

@@ -134,6 +134,7 @@ public:
     void ConnectAllDisconnectedControllers();
     void ClearAllControllers();
 
+    void MergeSingleJoyAsDualJoy(u32 npad_id_1, u32 npad_id_2);
     void StartLRAssignmentMode();
     void StopLRAssignmentMode();
     bool SwapNpadAssignment(u32 npad_id_1, u32 npad_id_2);

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

@@ -40,9 +40,14 @@ void Controller_Touchscreen::OnUpdate(const Core::Timing::CoreTiming& core_timin
     cur_entry.sampling_number = last_entry.sampling_number + 1;
     cur_entry.sampling_number2 = cur_entry.sampling_number;
 
-    const auto [x, y, pressed] = touch_device->GetStatus();
+    bool pressed = false;
+    float x, y;
+    std::tie(x, y, pressed) = touch_device->GetStatus();
     auto& touch_entry = cur_entry.states[0];
     touch_entry.attribute.raw = 0;
+    if (!pressed && touch_btn_device) {
+        std::tie(x, y, pressed) = touch_btn_device->GetStatus();
+    }
     if (pressed && Settings::values.touchscreen.enabled) {
         touch_entry.x = static_cast<u16>(x * Layout::ScreenUndocked::Width);
         touch_entry.y = static_cast<u16>(y * Layout::ScreenUndocked::Height);
@@ -63,5 +68,10 @@ void Controller_Touchscreen::OnUpdate(const Core::Timing::CoreTiming& core_timin
 
 void Controller_Touchscreen::OnLoadInputDevices() {
     touch_device = Input::CreateDevice<Input::TouchDevice>(Settings::values.touchscreen.device);
+    if (Settings::values.use_touch_from_button) {
+        touch_btn_device = Input::CreateDevice<Input::TouchDevice>("engine:touch_from_button");
+    } else {
+        touch_btn_device.reset();
+    }
 }
 } // namespace Service::HID

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

@@ -68,6 +68,7 @@ private:
                   "TouchScreenSharedMemory is an invalid size");
     TouchScreenSharedMemory shared_memory{};
     std::unique_ptr<Input::TouchDevice> touch_device;
+    std::unique_ptr<Input::TouchDevice> touch_btn_device;
     s64_le last_touch{};
 };
 } // namespace Service::HID

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

@@ -671,13 +671,15 @@ void Hid::SetNpadJoyAssignmentModeDual(Kernel::HLERequestContext& ctx) {
 
 void Hid::MergeSingleJoyAsDualJoy(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
-    const auto unknown_1{rp.Pop<u32>()};
-    const auto unknown_2{rp.Pop<u32>()};
+    const auto npad_id_1{rp.Pop<u32>()};
+    const auto npad_id_2{rp.Pop<u32>()};
     const auto applet_resource_user_id{rp.Pop<u64>()};
 
-    LOG_WARNING(Service_HID,
-                "(STUBBED) called, unknown_1={}, unknown_2={}, applet_resource_user_id={}",
-                unknown_1, unknown_2, applet_resource_user_id);
+    LOG_DEBUG(Service_HID, "called, npad_id_1={}, npad_id_2={}, applet_resource_user_id={}",
+              npad_id_1, npad_id_2, applet_resource_user_id);
+
+    auto& controller = applet_resource->GetController<Controller_NPad>(HidController::NPad);
+    controller.MergeSingleJoyAsDualJoy(npad_id_1, npad_id_2);
 
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);

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

@@ -9,6 +9,7 @@
 #include "core/hle/kernel/writable_event.h"
 #include "core/hle/service/nifm/nifm.h"
 #include "core/hle/service/service.h"
+#include "core/network/network.h"
 #include "core/settings.h"
 
 namespace Service::NIFM {
@@ -174,6 +175,16 @@ private:
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
     }
+    void GetCurrentIpAddress(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_NIFM, "(STUBBED) called");
+
+        const auto [ipv4, error] = Network::GetHostIPv4Address();
+        UNIMPLEMENTED_IF(error != Network::Errno::SUCCESS);
+
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushRaw(ipv4);
+    }
     void CreateTemporaryNetworkProfile(Kernel::HLERequestContext& ctx) {
         LOG_DEBUG(Service_NIFM, "called");
 
@@ -235,7 +246,7 @@ IGeneralService::IGeneralService(Core::System& system)
         {9, nullptr, "SetNetworkProfile"},
         {10, &IGeneralService::RemoveNetworkProfile, "RemoveNetworkProfile"},
         {11, nullptr, "GetScanDataOld"},
-        {12, nullptr, "GetCurrentIpAddress"},
+        {12, &IGeneralService::GetCurrentIpAddress, "GetCurrentIpAddress"},
         {13, nullptr, "GetCurrentAccessPointOld"},
         {14, &IGeneralService::CreateTemporaryNetworkProfile, "CreateTemporaryNetworkProfile"},
         {15, nullptr, "GetCurrentIpConfigInfo"},

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

@@ -5,6 +5,7 @@
 #include "common/logging/log.h"
 #include "core/file_sys/control_metadata.h"
 #include "core/file_sys/patch_manager.h"
+#include "core/file_sys/vfs.h"
 #include "core/hle/ipc_helpers.h"
 #include "core/hle/kernel/hle_ipc.h"
 #include "core/hle/service/ns/errors.h"

src/core/hle/service/service.cpp

@@ -246,7 +246,7 @@ void Init(std::shared_ptr<SM::ServiceManager>& sm, Core::System& system) {
     PSC::InstallInterfaces(*sm);
     PSM::InstallInterfaces(*sm);
     Set::InstallInterfaces(*sm);
-    Sockets::InstallInterfaces(*sm);
+    Sockets::InstallInterfaces(*sm, system);
     SPL::InstallInterfaces(*sm);
     SSL::InstallInterfaces(*sm);
     Time::InstallInterfaces(system);

src/core/hle/service/sockets/blocking_worker.h

@@ -0,0 +1,162 @@
+// Copyright 2020 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <atomic>
+#include <memory>
+#include <string>
+#include <string_view>
+#include <thread>
+#include <variant>
+#include <vector>
+
+#include <fmt/format.h>
+
+#include "common/assert.h"
+#include "common/microprofile.h"
+#include "common/thread.h"
+#include "core/core.h"
+#include "core/hle/kernel/hle_ipc.h"
+#include "core/hle/kernel/kernel.h"
+#include "core/hle/kernel/thread.h"
+#include "core/hle/kernel/writable_event.h"
+
+namespace Service::Sockets {
+
+/**
+ * Worker abstraction to execute blocking calls on host without blocking the guest thread
+ *
+ * @tparam Service  Service where the work is executed
+ * @tparam ...Types Types of work to execute
+ */
+template <class Service, class... Types>
+class BlockingWorker {
+    using This = BlockingWorker<Service, Types...>;
+    using WorkVariant = std::variant<std::monostate, Types...>;
+
+public:
+    /// Create a new worker
+    static std::unique_ptr<This> Create(Core::System& system, Service* service,
+                                        std::string_view name) {
+        return std::unique_ptr<This>(new This(system, service, name));
+    }
+
+    ~BlockingWorker() {
+        while (!is_available.load(std::memory_order_relaxed)) {
+            // Busy wait until work is finished
+            std::this_thread::yield();
+        }
+        // Monostate means to exit the thread
+        work = std::monostate{};
+        work_event.Set();
+        thread.join();
+    }
+
+    /**
+     * Try to capture the worker to send work after a success
+     * @returns True when the worker has been successfully captured
+     */
+    bool TryCapture() {
+        bool expected = true;
+        return is_available.compare_exchange_weak(expected, false, std::memory_order_relaxed,
+                                                  std::memory_order_relaxed);
+    }
+
+    /**
+     * Send work to this worker abstraction
+     * @see TryCapture must be called before attempting to call this function
+     */
+    template <class Work>
+    void SendWork(Work new_work) {
+        ASSERT_MSG(!is_available, "Trying to send work on a worker that's not captured");
+        work = std::move(new_work);
+        work_event.Set();
+    }
+
+    /// Generate a callback for @see SleepClientThread
+    template <class Work>
+    auto Callback() {
+        return [this](std::shared_ptr<Kernel::Thread>, Kernel::HLERequestContext& ctx,
+                      Kernel::ThreadWakeupReason reason) {
+            ASSERT(reason == Kernel::ThreadWakeupReason::Signal);
+            std::get<Work>(work).Response(ctx);
+            is_available.store(true);
+        };
+    }
+
+    /// Get kernel event that will be signalled by the worker when the host operation finishes
+    std::shared_ptr<Kernel::WritableEvent> KernelEvent() const {
+        return kernel_event;
+    }
+
+private:
+    explicit BlockingWorker(Core::System& system, Service* service, std::string_view name) {
+        auto pair = Kernel::WritableEvent::CreateEventPair(system.Kernel(), std::string(name));
+        kernel_event = std::move(pair.writable);
+        thread = std::thread([this, &system, service, name] { Run(system, service, name); });
+    }
+
+    void Run(Core::System& system, Service* service, std::string_view name) {
+        system.RegisterHostThread();
+
+        const std::string thread_name = fmt::format("yuzu:{}", name);
+        MicroProfileOnThreadCreate(thread_name.c_str());
+        Common::SetCurrentThreadName(thread_name.c_str());
+
+        bool keep_running = true;
+        while (keep_running) {
+            work_event.Wait();
+
+            const auto visit_fn = [service, &keep_running](auto&& w) {
+                using T = std::decay_t<decltype(w)>;
+                if constexpr (std::is_same_v<T, std::monostate>) {
+                    keep_running = false;
+                } else {
+                    w.Execute(service);
+                }
+            };
+            std::visit(visit_fn, work);
+
+            kernel_event->Signal();
+        }
+    }
+
+    std::thread thread;
+    WorkVariant work;
+    Common::Event work_event;
+    std::shared_ptr<Kernel::WritableEvent> kernel_event;
+    std::atomic_bool is_available{true};
+};
+
+template <class Service, class... Types>
+class BlockingWorkerPool {
+    using Worker = BlockingWorker<Service, Types...>;
+
+public:
+    explicit BlockingWorkerPool(Core::System& system_, Service* service_)
+        : system{system_}, service{service_} {}
+
+    /// Returns a captured worker thread, creating new ones if necessary
+    Worker* CaptureWorker() {
+        for (auto& worker : workers) {
+            if (worker->TryCapture()) {
+                return worker.get();
+            }
+        }
+        auto new_worker = Worker::Create(system, service, fmt::format("BSD:{}", workers.size()));
+        [[maybe_unused]] const bool success = new_worker->TryCapture();
+        ASSERT(success);
+
+        return workers.emplace_back(std::move(new_worker)).get();
+    }
+
+private:
+    Core::System& system;
+    Service* const service;
+
+    std::vector<std::unique_ptr<Worker>> workers;
+};
+
+} // namespace Service::Sockets

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

@@ -2,18 +2,138 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <array>
+#include <memory>
+#include <string>
+#include <utility>
+#include <vector>
+
+#include <fmt/format.h>
+
+#include "common/microprofile.h"
+#include "common/thread.h"
 #include "core/hle/ipc_helpers.h"
+#include "core/hle/kernel/thread.h"
 #include "core/hle/service/sockets/bsd.h"
+#include "core/hle/service/sockets/sockets_translate.h"
+#include "core/network/network.h"
+#include "core/network/sockets.h"
 
 namespace Service::Sockets {
 
+namespace {
+
+bool IsConnectionBased(Type type) {
+    switch (type) {
+    case Type::STREAM:
+        return true;
+    case Type::DGRAM:
+        return false;
+    default:
+        UNIMPLEMENTED_MSG("Unimplemented type={}", static_cast<int>(type));
+        return false;
+    }
+}
+
+} // Anonymous namespace
+
+void BSD::PollWork::Execute(BSD* bsd) {
+    std::tie(ret, bsd_errno) = bsd->PollImpl(write_buffer, read_buffer, nfds, timeout);
+}
+
+void BSD::PollWork::Response(Kernel::HLERequestContext& ctx) {
+    ctx.WriteBuffer(write_buffer);
+
+    IPC::ResponseBuilder rb{ctx, 4};
+    rb.Push(RESULT_SUCCESS);
+    rb.Push<s32>(ret);
+    rb.PushEnum(bsd_errno);
+}
+
+void BSD::AcceptWork::Execute(BSD* bsd) {
+    std::tie(ret, bsd_errno) = bsd->AcceptImpl(fd, write_buffer);
+}
+
+void BSD::AcceptWork::Response(Kernel::HLERequestContext& ctx) {
+    ctx.WriteBuffer(write_buffer);
+
+    IPC::ResponseBuilder rb{ctx, 5};
+    rb.Push(RESULT_SUCCESS);
+    rb.Push<s32>(ret);
+    rb.PushEnum(bsd_errno);
+    rb.Push<u32>(static_cast<u32>(write_buffer.size()));
+}
+
+void BSD::ConnectWork::Execute(BSD* bsd) {
+    bsd_errno = bsd->ConnectImpl(fd, addr);
+}
+
+void BSD::ConnectWork::Response(Kernel::HLERequestContext& ctx) {
+    IPC::ResponseBuilder rb{ctx, 4};
+    rb.Push(RESULT_SUCCESS);
+    rb.Push<s32>(bsd_errno == Errno::SUCCESS ? 0 : -1);
+    rb.PushEnum(bsd_errno);
+}
+
+void BSD::RecvWork::Execute(BSD* bsd) {
+    std::tie(ret, bsd_errno) = bsd->RecvImpl(fd, flags, message);
+}
+
+void BSD::RecvWork::Response(Kernel::HLERequestContext& ctx) {
+    ctx.WriteBuffer(message);
+
+    IPC::ResponseBuilder rb{ctx, 4};
+    rb.Push(RESULT_SUCCESS);
+    rb.Push<s32>(ret);
+    rb.PushEnum(bsd_errno);
+}
+
+void BSD::RecvFromWork::Execute(BSD* bsd) {
+    std::tie(ret, bsd_errno) = bsd->RecvFromImpl(fd, flags, message, addr);
+}
+
+void BSD::RecvFromWork::Response(Kernel::HLERequestContext& ctx) {
+    ctx.WriteBuffer(message, 0);
+    if (!addr.empty()) {
+        ctx.WriteBuffer(addr, 1);
+    }
+
+    IPC::ResponseBuilder rb{ctx, 5};
+    rb.Push(RESULT_SUCCESS);
+    rb.Push<s32>(ret);
+    rb.PushEnum(bsd_errno);
+    rb.Push<u32>(static_cast<u32>(addr.size()));
+}
+
+void BSD::SendWork::Execute(BSD* bsd) {
+    std::tie(ret, bsd_errno) = bsd->SendImpl(fd, flags, message);
+}
+
+void BSD::SendWork::Response(Kernel::HLERequestContext& ctx) {
+    IPC::ResponseBuilder rb{ctx, 4};
+    rb.Push(RESULT_SUCCESS);
+    rb.Push<s32>(ret);
+    rb.PushEnum(bsd_errno);
+}
+
+void BSD::SendToWork::Execute(BSD* bsd) {
+    std::tie(ret, bsd_errno) = bsd->SendToImpl(fd, flags, message, addr);
+}
+
+void BSD::SendToWork::Response(Kernel::HLERequestContext& ctx) {
+    IPC::ResponseBuilder rb{ctx, 4};
+    rb.Push(RESULT_SUCCESS);
+    rb.Push<s32>(ret);
+    rb.PushEnum(bsd_errno);
+}
+
 void BSD::RegisterClient(Kernel::HLERequestContext& ctx) {
     LOG_WARNING(Service, "(STUBBED) called");
 
     IPC::ResponseBuilder rb{ctx, 3};
 
     rb.Push(RESULT_SUCCESS);
-    rb.Push<u32>(0); // bsd errno
+    rb.Push<s32>(0); // bsd errno
 }
 
 void BSD::StartMonitoring(Kernel::HLERequestContext& ctx) {
@@ -26,20 +146,19 @@ void BSD::StartMonitoring(Kernel::HLERequestContext& ctx) {
 
 void BSD::Socket(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
+    const u32 domain = rp.Pop<u32>();
+    const u32 type = rp.Pop<u32>();
+    const u32 protocol = rp.Pop<u32>();
 
-    u32 domain = rp.Pop<u32>();
-    u32 type = rp.Pop<u32>();
-    u32 protocol = rp.Pop<u32>();
+    LOG_DEBUG(Service, "called. domain={} type={} protocol={}", domain, type, protocol);
 
-    LOG_WARNING(Service, "(STUBBED) called domain={} type={} protocol={}", domain, type, protocol);
-
-    u32 fd = next_fd++;
+    const auto [fd, bsd_errno] = SocketImpl(static_cast<Domain>(domain), static_cast<Type>(type),
+                                            static_cast<Protocol>(protocol));
 
     IPC::ResponseBuilder rb{ctx, 4};
-
     rb.Push(RESULT_SUCCESS);
-    rb.Push<u32>(fd);
-    rb.Push<u32>(0); // bsd errno
+    rb.Push<s32>(fd);
+    rb.PushEnum(bsd_errno);
 }
 
 void BSD::Select(Kernel::HLERequestContext& ctx) {
@@ -52,67 +171,663 @@ void BSD::Select(Kernel::HLERequestContext& ctx) {
     rb.Push<u32>(0); // bsd errno
 }
 
+void BSD::Poll(Kernel::HLERequestContext& ctx) {
+    IPC::RequestParser rp{ctx};
+    const s32 nfds = rp.Pop<s32>();
+    const s32 timeout = rp.Pop<s32>();
+
+    LOG_DEBUG(Service, "called. nfds={} timeout={}", nfds, timeout);
+
+    ExecuteWork(ctx, "BSD:Poll", timeout != 0,
+                PollWork{
+                    .nfds = nfds,
+                    .timeout = timeout,
+                    .read_buffer = ctx.ReadBuffer(),
+                    .write_buffer = std::vector<u8>(ctx.GetWriteBufferSize()),
+                });
+}
+
+void BSD::Accept(Kernel::HLERequestContext& ctx) {
+    IPC::RequestParser rp{ctx};
+    const s32 fd = rp.Pop<s32>();
+
+    LOG_DEBUG(Service, "called. fd={}", fd);
+
+    ExecuteWork(ctx, "BSD:Accept", IsBlockingSocket(fd),
+                AcceptWork{
+                    .fd = fd,
+                    .write_buffer = std::vector<u8>(ctx.GetWriteBufferSize()),
+                });
+}
+
 void BSD::Bind(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service, "(STUBBED) called");
+    IPC::RequestParser rp{ctx};
+    const s32 fd = rp.Pop<s32>();
 
-    IPC::ResponseBuilder rb{ctx, 4};
+    LOG_DEBUG(Service, "called. fd={} addrlen={}", fd, ctx.GetReadBufferSize());
 
-    rb.Push(RESULT_SUCCESS);
-    rb.Push<u32>(0); // ret
-    rb.Push<u32>(0); // bsd errno
+    BuildErrnoResponse(ctx, BindImpl(fd, ctx.ReadBuffer()));
 }
 
 void BSD::Connect(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service, "(STUBBED) called");
+    IPC::RequestParser rp{ctx};
+    const s32 fd = rp.Pop<s32>();
 
-    IPC::ResponseBuilder rb{ctx, 4};
+    LOG_DEBUG(Service, "called. fd={} addrlen={}", fd, ctx.GetReadBufferSize());
 
+    ExecuteWork(ctx, "BSD:Connect", IsBlockingSocket(fd),
+                ConnectWork{
+                    .fd = fd,
+                    .addr = ctx.ReadBuffer(),
+                });
+}
+
+void BSD::GetPeerName(Kernel::HLERequestContext& ctx) {
+    IPC::RequestParser rp{ctx};
+    const s32 fd = rp.Pop<s32>();
+
+    LOG_DEBUG(Service, "called. fd={}", fd);
+
+    std::vector<u8> write_buffer(ctx.GetWriteBufferSize());
+    const Errno bsd_errno = GetPeerNameImpl(fd, write_buffer);
+
+    ctx.WriteBuffer(write_buffer);
+
+    IPC::ResponseBuilder rb{ctx, 5};
     rb.Push(RESULT_SUCCESS);
-    rb.Push<u32>(0); // ret
-    rb.Push<u32>(0); // bsd errno
+    rb.Push<s32>(bsd_errno != Errno::SUCCESS ? -1 : 0);
+    rb.PushEnum(bsd_errno);
+    rb.Push<u32>(static_cast<u32>(write_buffer.size()));
+}
+
+void BSD::GetSockName(Kernel::HLERequestContext& ctx) {
+    IPC::RequestParser rp{ctx};
+    const s32 fd = rp.Pop<s32>();
+
+    LOG_DEBUG(Service, "called. fd={}", fd);
+
+    std::vector<u8> write_buffer(ctx.GetWriteBufferSize());
+    const Errno bsd_errno = GetSockNameImpl(fd, write_buffer);
+
+    ctx.WriteBuffer(write_buffer);
+
+    IPC::ResponseBuilder rb{ctx, 5};
+    rb.Push(RESULT_SUCCESS);
+    rb.Push<s32>(bsd_errno != Errno::SUCCESS ? -1 : 0);
+    rb.PushEnum(bsd_errno);
+    rb.Push<u32>(static_cast<u32>(write_buffer.size()));
 }
 
 void BSD::Listen(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service, "(STUBBED) called");
+    IPC::RequestParser rp{ctx};
+    const s32 fd = rp.Pop<s32>();
+    const s32 backlog = rp.Pop<s32>();
+
+    LOG_DEBUG(Service, "called. fd={} backlog={}", fd, backlog);
+
+    BuildErrnoResponse(ctx, ListenImpl(fd, backlog));
+}
+
+void BSD::Fcntl(Kernel::HLERequestContext& ctx) {
+    IPC::RequestParser rp{ctx};
+    const s32 fd = rp.Pop<s32>();
+    const s32 cmd = rp.Pop<s32>();
+    const s32 arg = rp.Pop<s32>();
+
+    LOG_DEBUG(Service, "called. fd={} cmd={} arg={}", fd, cmd, arg);
+
+    const auto [ret, bsd_errno] = FcntlImpl(fd, static_cast<FcntlCmd>(cmd), arg);
 
     IPC::ResponseBuilder rb{ctx, 4};
-
     rb.Push(RESULT_SUCCESS);
-    rb.Push<u32>(0); // ret
-    rb.Push<u32>(0); // bsd errno
+    rb.Push<s32>(ret);
+    rb.PushEnum(bsd_errno);
 }
 
 void BSD::SetSockOpt(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service, "(STUBBED) called");
+    IPC::RequestParser rp{ctx};
 
-    IPC::ResponseBuilder rb{ctx, 4};
+    const s32 fd = rp.Pop<s32>();
+    const u32 level = rp.Pop<u32>();
+    const OptName optname = static_cast<OptName>(rp.Pop<u32>());
 
-    rb.Push(RESULT_SUCCESS);
-    rb.Push<u32>(0); // ret
-    rb.Push<u32>(0); // bsd errno
+    const std::vector<u8> buffer = ctx.ReadBuffer();
+    const u8* optval = buffer.empty() ? nullptr : buffer.data();
+    size_t optlen = buffer.size();
+
+    std::array<u64, 2> values;
+    if ((optname == OptName::SNDTIMEO || optname == OptName::RCVTIMEO) && buffer.size() == 8) {
+        std::memcpy(values.data(), buffer.data(), sizeof(values));
+        optlen = sizeof(values);
+        optval = reinterpret_cast<const u8*>(values.data());
+    }
+
+    LOG_DEBUG(Service, "called. fd={} level={} optname=0x{:x} optlen={}", fd, level,
+              static_cast<u32>(optname), optlen);
+
+    BuildErrnoResponse(ctx, SetSockOptImpl(fd, level, optname, optlen, optval));
+}
+
+void BSD::Shutdown(Kernel::HLERequestContext& ctx) {
+    IPC::RequestParser rp{ctx};
+
+    const s32 fd = rp.Pop<s32>();
+    const s32 how = rp.Pop<s32>();
+
+    LOG_DEBUG(Service, "called. fd={} how={}", fd, how);
+
+    BuildErrnoResponse(ctx, ShutdownImpl(fd, how));
+}
+
+void BSD::Recv(Kernel::HLERequestContext& ctx) {
+    IPC::RequestParser rp{ctx};
+
+    const s32 fd = rp.Pop<s32>();
+    const u32 flags = rp.Pop<u32>();
+
+    LOG_DEBUG(Service, "called. fd={} flags=0x{:x} len={}", fd, flags, ctx.GetWriteBufferSize());
+
+    ExecuteWork(ctx, "BSD:Recv", IsBlockingSocket(fd),
+                RecvWork{
+                    .fd = fd,
+                    .flags = flags,
+                    .message = std::vector<u8>(ctx.GetWriteBufferSize()),
+                });
+}
+
+void BSD::RecvFrom(Kernel::HLERequestContext& ctx) {
+    IPC::RequestParser rp{ctx};
+
+    const s32 fd = rp.Pop<s32>();
+    const u32 flags = rp.Pop<u32>();
+
+    LOG_DEBUG(Service, "called. fd={} flags=0x{:x} len={} addrlen={}", fd, flags,
+              ctx.GetWriteBufferSize(0), ctx.GetWriteBufferSize(1));
+
+    ExecuteWork(ctx, "BSD:RecvFrom", IsBlockingSocket(fd),
+                RecvFromWork{
+                    .fd = fd,
+                    .flags = flags,
+                    .message = std::vector<u8>(ctx.GetWriteBufferSize(0)),
+                    .addr = std::vector<u8>(ctx.GetWriteBufferSize(1)),
+                });
+}
+
+void BSD::Send(Kernel::HLERequestContext& ctx) {
+    IPC::RequestParser rp{ctx};
+
+    const s32 fd = rp.Pop<s32>();
+    const u32 flags = rp.Pop<u32>();
+
+    LOG_DEBUG(Service, "called. fd={} flags=0x{:x} len={}", fd, flags, ctx.GetReadBufferSize());
+
+    ExecuteWork(ctx, "BSD:Send", IsBlockingSocket(fd),
+                SendWork{
+                    .fd = fd,
+                    .flags = flags,
+                    .message = ctx.ReadBuffer(),
+                });
 }
 
 void BSD::SendTo(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service, "(STUBBED) called");
+    IPC::RequestParser rp{ctx};
+    const s32 fd = rp.Pop<s32>();
+    const u32 flags = rp.Pop<u32>();
 
-    IPC::ResponseBuilder rb{ctx, 4};
+    LOG_DEBUG(Service, "called. fd={} flags=0x{} len={} addrlen={}", fd, flags,
+              ctx.GetReadBufferSize(0), ctx.GetReadBufferSize(1));
 
-    rb.Push(RESULT_SUCCESS);
-    rb.Push<u32>(0); // ret
-    rb.Push<u32>(0); // bsd errno
+    ExecuteWork(ctx, "BSD:SendTo", IsBlockingSocket(fd),
+                SendToWork{
+                    .fd = fd,
+                    .flags = flags,
+                    .message = ctx.ReadBuffer(0),
+                    .addr = ctx.ReadBuffer(1),
+                });
+}
+
+void BSD::Write(Kernel::HLERequestContext& ctx) {
+    IPC::RequestParser rp{ctx};
+    const s32 fd = rp.Pop<s32>();
+
+    LOG_DEBUG(Service, "called. fd={} len={}", fd, ctx.GetReadBufferSize());
+
+    ExecuteWork(ctx, "BSD:Write", IsBlockingSocket(fd),
+                SendWork{
+                    .fd = fd,
+                    .flags = 0,
+                    .message = ctx.ReadBuffer(),
+                });
 }
 
 void BSD::Close(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service, "(STUBBED) called");
+    IPC::RequestParser rp{ctx};
+    const s32 fd = rp.Pop<s32>();
 
+    LOG_DEBUG(Service, "called. fd={}", fd);
+
+    BuildErrnoResponse(ctx, CloseImpl(fd));
+}
+
+template <typename Work>
+void BSD::ExecuteWork(Kernel::HLERequestContext& ctx, std::string_view sleep_reason,
+                      bool is_blocking, Work work) {
+    if (!is_blocking) {
+        work.Execute(this);
+        work.Response(ctx);
+        return;
+    }
+
+    // Signal a dummy response to make IPC validation happy
+    // This will be overwritten by the SleepClientThread callback
+    work.Response(ctx);
+
+    auto worker = worker_pool.CaptureWorker();
+
+    ctx.SleepClientThread(std::string(sleep_reason), std::numeric_limits<u64>::max(),
+                          worker->Callback<Work>(), worker->KernelEvent());
+
+    worker->SendWork(std::move(work));
+}
+
+std::pair<s32, Errno> BSD::SocketImpl(Domain domain, Type type, Protocol protocol) {
+    if (type == Type::SEQPACKET) {
+        UNIMPLEMENTED_MSG("SOCK_SEQPACKET errno management");
+    } else if (type == Type::RAW && (domain != Domain::INET || protocol != Protocol::ICMP)) {
+        UNIMPLEMENTED_MSG("SOCK_RAW errno management");
+    }
+
+    [[maybe_unused]] const bool unk_flag = (static_cast<u32>(type) & 0x20000000) != 0;
+    UNIMPLEMENTED_IF_MSG(unk_flag, "Unknown flag in type");
+    type = static_cast<Type>(static_cast<u32>(type) & ~0x20000000);
+
+    const s32 fd = FindFreeFileDescriptorHandle();
+    if (fd < 0) {
+        LOG_ERROR(Service, "No more file descriptors available");
+        return {-1, Errno::MFILE};
+    }
+
+    FileDescriptor& descriptor = file_descriptors[fd].emplace();
+    // ENONMEM might be thrown here
+
+    LOG_INFO(Service, "New socket fd={}", fd);
+
+    descriptor.socket = std::make_unique<Network::Socket>();
+    descriptor.socket->Initialize(Translate(domain), Translate(type), Translate(type, protocol));
+    descriptor.is_connection_based = IsConnectionBased(type);
+
+    return {fd, Errno::SUCCESS};
+}
+
+std::pair<s32, Errno> BSD::PollImpl(std::vector<u8>& write_buffer, std::vector<u8> read_buffer,
+                                    s32 nfds, s32 timeout) {
+    if (write_buffer.size() < nfds * sizeof(PollFD)) {
+        return {-1, Errno::INVAL};
+    }
+
+    if (nfds == 0) {
+        // When no entries are provided, -1 is returned with errno zero
+        return {-1, Errno::SUCCESS};
+    }
+
+    const size_t length = std::min(read_buffer.size(), write_buffer.size());
+    std::vector<PollFD> fds(nfds);
+    std::memcpy(fds.data(), read_buffer.data(), length);
+
+    if (timeout >= 0) {
+        const s64 seconds = timeout / 1000;
+        const u64 nanoseconds = 1'000'000 * (static_cast<u64>(timeout) % 1000);
+
+        if (seconds < 0) {
+            return {-1, Errno::INVAL};
+        }
+        if (nanoseconds > 999'999'999) {
+            return {-1, Errno::INVAL};
+        }
+    } else if (timeout != -1) {
+        return {-1, Errno::INVAL};
+    }
+
+    for (PollFD& pollfd : fds) {
+        ASSERT(pollfd.revents == 0);
+
+        if (pollfd.fd > MAX_FD || pollfd.fd < 0) {
+            LOG_ERROR(Service, "File descriptor handle={} is invalid", pollfd.fd);
+            pollfd.revents = 0;
+            return {0, Errno::SUCCESS};
+        }
+
+        std::optional<FileDescriptor>& descriptor = file_descriptors[pollfd.fd];
+        if (!descriptor) {
+            LOG_ERROR(Service, "File descriptor handle={} is not allocated", pollfd.fd);
+            pollfd.revents = POLL_NVAL;
+            return {0, Errno::SUCCESS};
+        }
+    }
+
+    std::vector<Network::PollFD> host_pollfds(fds.size());
+    std::transform(fds.begin(), fds.end(), host_pollfds.begin(), [this](PollFD pollfd) {
+        Network::PollFD result;
+        result.socket = file_descriptors[pollfd.fd]->socket.get();
+        result.events = TranslatePollEventsToHost(pollfd.events);
+        result.revents = 0;
+        return result;
+    });
+
+    const auto result = Network::Poll(host_pollfds, timeout);
+
+    const size_t num = host_pollfds.size();
+    for (size_t i = 0; i < num; ++i) {
+        fds[i].revents = TranslatePollEventsToGuest(host_pollfds[i].revents);
+    }
+    std::memcpy(write_buffer.data(), fds.data(), length);
+
+    return Translate(result);
+}
+
+std::pair<s32, Errno> BSD::AcceptImpl(s32 fd, std::vector<u8>& write_buffer) {
+    if (!IsFileDescriptorValid(fd)) {
+        return {-1, Errno::BADF};
+    }
+
+    const s32 new_fd = FindFreeFileDescriptorHandle();
+    if (new_fd < 0) {
+        LOG_ERROR(Service, "No more file descriptors available");
+        return {-1, Errno::MFILE};
+    }
+
+    FileDescriptor& descriptor = *file_descriptors[fd];
+    auto [result, bsd_errno] = descriptor.socket->Accept();
+    if (bsd_errno != Network::Errno::SUCCESS) {
+        return {-1, Translate(bsd_errno)};
+    }
+
+    FileDescriptor& new_descriptor = file_descriptors[new_fd].emplace();
+    new_descriptor.socket = std::move(result.socket);
+    new_descriptor.is_connection_based = descriptor.is_connection_based;
+
+    ASSERT(write_buffer.size() == sizeof(SockAddrIn));
+    const SockAddrIn guest_addr_in = Translate(result.sockaddr_in);
+    std::memcpy(write_buffer.data(), &guest_addr_in, sizeof(guest_addr_in));
+
+    return {new_fd, Errno::SUCCESS};
+}
+
+Errno BSD::BindImpl(s32 fd, const std::vector<u8>& addr) {
+    if (!IsFileDescriptorValid(fd)) {
+        return Errno::BADF;
+    }
+    ASSERT(addr.size() == sizeof(SockAddrIn));
+    SockAddrIn addr_in;
+    std::memcpy(&addr_in, addr.data(), sizeof(addr_in));
+
+    return Translate(file_descriptors[fd]->socket->Bind(Translate(addr_in)));
+}
+
+Errno BSD::ConnectImpl(s32 fd, const std::vector<u8>& addr) {
+    if (!IsFileDescriptorValid(fd)) {
+        return Errno::BADF;
+    }
+
+    UNIMPLEMENTED_IF(addr.size() != sizeof(SockAddrIn));
+    SockAddrIn addr_in;
+    std::memcpy(&addr_in, addr.data(), sizeof(addr_in));
+
+    return Translate(file_descriptors[fd]->socket->Connect(Translate(addr_in)));
+}
+
+Errno BSD::GetPeerNameImpl(s32 fd, std::vector<u8>& write_buffer) {
+    if (!IsFileDescriptorValid(fd)) {
+        return Errno::BADF;
+    }
+
+    const auto [addr_in, bsd_errno] = file_descriptors[fd]->socket->GetPeerName();
+    if (bsd_errno != Network::Errno::SUCCESS) {
+        return Translate(bsd_errno);
+    }
+    const SockAddrIn guest_addrin = Translate(addr_in);
+
+    ASSERT(write_buffer.size() == sizeof(guest_addrin));
+    std::memcpy(write_buffer.data(), &guest_addrin, sizeof(guest_addrin));
+    return Translate(bsd_errno);
+}
+
+Errno BSD::GetSockNameImpl(s32 fd, std::vector<u8>& write_buffer) {
+    if (!IsFileDescriptorValid(fd)) {
+        return Errno::BADF;
+    }
+
+    const auto [addr_in, bsd_errno] = file_descriptors[fd]->socket->GetSockName();
+    if (bsd_errno != Network::Errno::SUCCESS) {
+        return Translate(bsd_errno);
+    }
+    const SockAddrIn guest_addrin = Translate(addr_in);
+
+    ASSERT(write_buffer.size() == sizeof(guest_addrin));
+    std::memcpy(write_buffer.data(), &guest_addrin, sizeof(guest_addrin));
+    return Translate(bsd_errno);
+}
+
+Errno BSD::ListenImpl(s32 fd, s32 backlog) {
+    if (!IsFileDescriptorValid(fd)) {
+        return Errno::BADF;
+    }
+    return Translate(file_descriptors[fd]->socket->Listen(backlog));
+}
+
+std::pair<s32, Errno> BSD::FcntlImpl(s32 fd, FcntlCmd cmd, s32 arg) {
+    if (!IsFileDescriptorValid(fd)) {
+        return {-1, Errno::BADF};
+    }
+
+    FileDescriptor& descriptor = *file_descriptors[fd];
+
+    switch (cmd) {
+    case FcntlCmd::GETFL:
+        ASSERT(arg == 0);
+        return {descriptor.flags, Errno::SUCCESS};
+    case FcntlCmd::SETFL: {
+        const bool enable = (arg & FLAG_O_NONBLOCK) != 0;
+        const Errno bsd_errno = Translate(descriptor.socket->SetNonBlock(enable));
+        if (bsd_errno != Errno::SUCCESS) {
+            return {-1, bsd_errno};
+        }
+        descriptor.flags = arg;
+        return {0, Errno::SUCCESS};
+    }
+    default:
+        UNIMPLEMENTED_MSG("Unimplemented cmd={}", static_cast<int>(cmd));
+        return {-1, Errno::SUCCESS};
+    }
+}
+
+Errno BSD::SetSockOptImpl(s32 fd, u32 level, OptName optname, size_t optlen, const void* optval) {
+    UNIMPLEMENTED_IF(level != 0xffff); // SOL_SOCKET
+
+    if (!IsFileDescriptorValid(fd)) {
+        return Errno::BADF;
+    }
+
+    Network::Socket* const socket = file_descriptors[fd]->socket.get();
+
+    if (optname == OptName::LINGER) {
+        ASSERT(optlen == sizeof(Linger));
+        Linger linger;
+        std::memcpy(&linger, optval, sizeof(linger));
+        ASSERT(linger.onoff == 0 || linger.onoff == 1);
+
+        return Translate(socket->SetLinger(linger.onoff != 0, linger.linger));
+    }
+
+    ASSERT(optlen == sizeof(u32));
+    u32 value;
+    std::memcpy(&value, optval, sizeof(value));
+
+    switch (optname) {
+    case OptName::REUSEADDR:
+        ASSERT(value == 0 || value == 1);
+        return Translate(socket->SetReuseAddr(value != 0));
+    case OptName::BROADCAST:
+        ASSERT(value == 0 || value == 1);
+        return Translate(socket->SetBroadcast(value != 0));
+    case OptName::SNDBUF:
+        return Translate(socket->SetSndBuf(value));
+    case OptName::RCVBUF:
+        return Translate(socket->SetRcvBuf(value));
+    case OptName::SNDTIMEO:
+        return Translate(socket->SetSndTimeo(value));
+    case OptName::RCVTIMEO:
+        return Translate(socket->SetRcvTimeo(value));
+    default:
+        UNIMPLEMENTED_MSG("Unimplemented optname={}", static_cast<int>(optname));
+        return Errno::SUCCESS;
+    }
+}
+
+Errno BSD::ShutdownImpl(s32 fd, s32 how) {
+    if (!IsFileDescriptorValid(fd)) {
+        return Errno::BADF;
+    }
+    const Network::ShutdownHow host_how = Translate(static_cast<ShutdownHow>(how));
+    return Translate(file_descriptors[fd]->socket->Shutdown(host_how));
+}
+
+std::pair<s32, Errno> BSD::RecvImpl(s32 fd, u32 flags, std::vector<u8>& message) {
+    if (!IsFileDescriptorValid(fd)) {
+        return {-1, Errno::BADF};
+    }
+    return Translate(file_descriptors[fd]->socket->Recv(flags, message));
+}
+
+std::pair<s32, Errno> BSD::RecvFromImpl(s32 fd, u32 flags, std::vector<u8>& message,
+                                        std::vector<u8>& addr) {
+    if (!IsFileDescriptorValid(fd)) {
+        return {-1, Errno::BADF};
+    }
+
+    FileDescriptor& descriptor = *file_descriptors[fd];
+
+    Network::SockAddrIn addr_in{};
+    Network::SockAddrIn* p_addr_in = nullptr;
+    if (descriptor.is_connection_based) {
+        // Connection based file descriptors (e.g. TCP) zero addr
+        addr.clear();
+    } else {
+        p_addr_in = &addr_in;
+    }
+
+    // Apply flags
+    if ((flags & FLAG_MSG_DONTWAIT) != 0) {
+        flags &= ~FLAG_MSG_DONTWAIT;
+        if ((descriptor.flags & FLAG_O_NONBLOCK) == 0) {
+            descriptor.socket->SetNonBlock(true);
+        }
+    }
+
+    const auto [ret, bsd_errno] = Translate(descriptor.socket->RecvFrom(flags, message, p_addr_in));
+
+    // Restore original state
+    if ((descriptor.flags & FLAG_O_NONBLOCK) == 0) {
+        descriptor.socket->SetNonBlock(false);
+    }
+
+    if (p_addr_in) {
+        if (ret < 0) {
+            addr.clear();
+        } else {
+            ASSERT(addr.size() == sizeof(SockAddrIn));
+            const SockAddrIn result = Translate(addr_in);
+            std::memcpy(addr.data(), &result, sizeof(result));
+        }
+    }
+
+    return {ret, bsd_errno};
+}
+
+std::pair<s32, Errno> BSD::SendImpl(s32 fd, u32 flags, const std::vector<u8>& message) {
+    if (!IsFileDescriptorValid(fd)) {
+        return {-1, Errno::BADF};
+    }
+    return Translate(file_descriptors[fd]->socket->Send(message, flags));
+}
+
+std::pair<s32, Errno> BSD::SendToImpl(s32 fd, u32 flags, const std::vector<u8>& message,
+                                      const std::vector<u8>& addr) {
+    if (!IsFileDescriptorValid(fd)) {
+        return {-1, Errno::BADF};
+    }
+
+    Network::SockAddrIn addr_in;
+    Network::SockAddrIn* p_addr_in = nullptr;
+    if (!addr.empty()) {
+        ASSERT(addr.size() == sizeof(SockAddrIn));
+        SockAddrIn guest_addr_in;
+        std::memcpy(&guest_addr_in, addr.data(), sizeof(guest_addr_in));
+        addr_in = Translate(guest_addr_in);
+    }
+
+    return Translate(file_descriptors[fd]->socket->SendTo(flags, message, p_addr_in));
+}
+
+Errno BSD::CloseImpl(s32 fd) {
+    if (!IsFileDescriptorValid(fd)) {
+        return Errno::BADF;
+    }
+
+    const Errno bsd_errno = Translate(file_descriptors[fd]->socket->Close());
+    if (bsd_errno != Errno::SUCCESS) {
+        return bsd_errno;
+    }
+
+    LOG_INFO(Service, "Close socket fd={}", fd);
+
+    file_descriptors[fd].reset();
+    return bsd_errno;
+}
+
+s32 BSD::FindFreeFileDescriptorHandle() noexcept {
+    for (s32 fd = 0; fd < static_cast<s32>(file_descriptors.size()); ++fd) {
+        if (!file_descriptors[fd]) {
+            return fd;
+        }
+    }
+    return -1;
+}
+
+bool BSD::IsFileDescriptorValid(s32 fd) const noexcept {
+    if (fd > MAX_FD || fd < 0) {
+        LOG_ERROR(Service, "Invalid file descriptor handle={}", fd);
+        return false;
+    }
+    if (!file_descriptors[fd]) {
+        LOG_ERROR(Service, "File descriptor handle={} is not allocated", fd);
+        return false;
+    }
+    return true;
+}
+
+bool BSD::IsBlockingSocket(s32 fd) const noexcept {
+    // Inform invalid sockets as non-blocking
+    // This way we avoid using a worker thread as it will fail without blocking host
+    if (fd > MAX_FD || fd < 0) {
+        return false;
+    }
+    if (!file_descriptors[fd]) {
+        return false;
+    }
+    return (file_descriptors[fd]->flags & FLAG_O_NONBLOCK) != 0;
+}
+
+void BSD::BuildErrnoResponse(Kernel::HLERequestContext& ctx, Errno bsd_errno) const noexcept {
     IPC::ResponseBuilder rb{ctx, 4};
 
     rb.Push(RESULT_SUCCESS);
-    rb.Push<u32>(0); // ret
-    rb.Push<u32>(0); // bsd errno
+    rb.Push<s32>(bsd_errno == Errno::SUCCESS ? 0 : -1);
+    rb.PushEnum(bsd_errno);
 }
 
-BSD::BSD(const char* name) : ServiceFramework(name) {
+BSD::BSD(Core::System& system, const char* name)
+    : ServiceFramework(name), worker_pool{system, this} {
     // clang-format off
     static const FunctionInfo functions[] = {
         {0, &BSD::RegisterClient, "RegisterClient"},
@@ -121,25 +836,25 @@ BSD::BSD(const char* name) : ServiceFramework(name) {
         {3, nullptr, "SocketExempt"},
         {4, nullptr, "Open"},
         {5, &BSD::Select, "Select"},
-        {6, nullptr, "Poll"},
+        {6, &BSD::Poll, "Poll"},
         {7, nullptr, "Sysctl"},
-        {8, nullptr, "Recv"},
-        {9, nullptr, "RecvFrom"},
-        {10, nullptr, "Send"},
+        {8, &BSD::Recv, "Recv"},
+        {9, &BSD::RecvFrom, "RecvFrom"},
+        {10, &BSD::Send, "Send"},
         {11, &BSD::SendTo, "SendTo"},
-        {12, nullptr, "Accept"},
+        {12, &BSD::Accept, "Accept"},
         {13, &BSD::Bind, "Bind"},
         {14, &BSD::Connect, "Connect"},
-        {15, nullptr, "GetPeerName"},
-        {16, nullptr, "GetSockName"},
+        {15, &BSD::GetPeerName, "GetPeerName"},
+        {16, &BSD::GetSockName, "GetSockName"},
         {17, nullptr, "GetSockOpt"},
         {18, &BSD::Listen, "Listen"},
         {19, nullptr, "Ioctl"},
-        {20, nullptr, "Fcntl"},
+        {20, &BSD::Fcntl, "Fcntl"},
         {21, &BSD::SetSockOpt, "SetSockOpt"},
-        {22, nullptr, "Shutdown"},
+        {22, &BSD::Shutdown, "Shutdown"},
         {23, nullptr, "ShutdownAllSockets"},
-        {24, nullptr, "Write"},
+        {24, &BSD::Write, "Write"},
         {25, nullptr, "Read"},
         {26, &BSD::Close, "Close"},
         {27, nullptr, "DuplicateSocket"},

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

@@ -4,30 +4,174 @@
 
 #pragma once
 
+#include <memory>
+#include <string_view>
+#include <vector>
+
+#include "common/common_types.h"
 #include "core/hle/kernel/hle_ipc.h"
 #include "core/hle/service/service.h"
+#include "core/hle/service/sockets/blocking_worker.h"
+#include "core/hle/service/sockets/sockets.h"
+
+namespace Core {
+class System;
+}
+
+namespace Network {
+class Socket;
+}
 
 namespace Service::Sockets {
 
 class BSD final : public ServiceFramework<BSD> {
 public:
-    explicit BSD(const char* name);
+    explicit BSD(Core::System& system, const char* name);
     ~BSD() override;
 
 private:
+    /// Maximum number of file descriptors
+    static constexpr size_t MAX_FD = 128;
+
+    struct FileDescriptor {
+        std::unique_ptr<Network::Socket> socket;
+        s32 flags = 0;
+        bool is_connection_based = false;
+    };
+
+    struct PollWork {
+        void Execute(BSD* bsd);
+        void Response(Kernel::HLERequestContext& ctx);
+
+        s32 nfds;
+        s32 timeout;
+        std::vector<u8> read_buffer;
+        std::vector<u8> write_buffer;
+        s32 ret{};
+        Errno bsd_errno{};
+    };
+
+    struct AcceptWork {
+        void Execute(BSD* bsd);
+        void Response(Kernel::HLERequestContext& ctx);
+
+        s32 fd;
+        std::vector<u8> write_buffer;
+        s32 ret{};
+        Errno bsd_errno{};
+    };
+
+    struct ConnectWork {
+        void Execute(BSD* bsd);
+        void Response(Kernel::HLERequestContext& ctx);
+
+        s32 fd;
+        std::vector<u8> addr;
+        Errno bsd_errno{};
+    };
+
+    struct RecvWork {
+        void Execute(BSD* bsd);
+        void Response(Kernel::HLERequestContext& ctx);
+
+        s32 fd;
+        u32 flags;
+        std::vector<u8> message;
+        s32 ret{};
+        Errno bsd_errno{};
+    };
+
+    struct RecvFromWork {
+        void Execute(BSD* bsd);
+        void Response(Kernel::HLERequestContext& ctx);
+
+        s32 fd;
+        u32 flags;
+        std::vector<u8> message;
+        std::vector<u8> addr;
+        s32 ret{};
+        Errno bsd_errno{};
+    };
+
+    struct SendWork {
+        void Execute(BSD* bsd);
+        void Response(Kernel::HLERequestContext& ctx);
+
+        s32 fd;
+        u32 flags;
+        std::vector<u8> message;
+        s32 ret{};
+        Errno bsd_errno{};
+    };
+
+    struct SendToWork {
+        void Execute(BSD* bsd);
+        void Response(Kernel::HLERequestContext& ctx);
+
+        s32 fd;
+        u32 flags;
+        std::vector<u8> message;
+        std::vector<u8> addr;
+        s32 ret{};
+        Errno bsd_errno{};
+    };
+
     void RegisterClient(Kernel::HLERequestContext& ctx);
     void StartMonitoring(Kernel::HLERequestContext& ctx);
     void Socket(Kernel::HLERequestContext& ctx);
     void Select(Kernel::HLERequestContext& ctx);
+    void Poll(Kernel::HLERequestContext& ctx);
+    void Accept(Kernel::HLERequestContext& ctx);
     void Bind(Kernel::HLERequestContext& ctx);
     void Connect(Kernel::HLERequestContext& ctx);
+    void GetPeerName(Kernel::HLERequestContext& ctx);
+    void GetSockName(Kernel::HLERequestContext& ctx);
     void Listen(Kernel::HLERequestContext& ctx);
+    void Fcntl(Kernel::HLERequestContext& ctx);
     void SetSockOpt(Kernel::HLERequestContext& ctx);
+    void Shutdown(Kernel::HLERequestContext& ctx);
+    void Recv(Kernel::HLERequestContext& ctx);
+    void RecvFrom(Kernel::HLERequestContext& ctx);
+    void Send(Kernel::HLERequestContext& ctx);
     void SendTo(Kernel::HLERequestContext& ctx);
+    void Write(Kernel::HLERequestContext& ctx);
     void Close(Kernel::HLERequestContext& ctx);
 
-    /// Id to use for the next open file descriptor.
-    u32 next_fd = 1;
+    template <typename Work>
+    void ExecuteWork(Kernel::HLERequestContext& ctx, std::string_view sleep_reason,
+                     bool is_blocking, Work work);
+
+    std::pair<s32, Errno> SocketImpl(Domain domain, Type type, Protocol protocol);
+    std::pair<s32, Errno> PollImpl(std::vector<u8>& write_buffer, std::vector<u8> read_buffer,
+                                   s32 nfds, s32 timeout);
+    std::pair<s32, Errno> AcceptImpl(s32 fd, std::vector<u8>& write_buffer);
+    Errno BindImpl(s32 fd, const std::vector<u8>& addr);
+    Errno ConnectImpl(s32 fd, const std::vector<u8>& addr);
+    Errno GetPeerNameImpl(s32 fd, std::vector<u8>& write_buffer);
+    Errno GetSockNameImpl(s32 fd, std::vector<u8>& write_buffer);
+    Errno ListenImpl(s32 fd, s32 backlog);
+    std::pair<s32, Errno> FcntlImpl(s32 fd, FcntlCmd cmd, s32 arg);
+    Errno SetSockOptImpl(s32 fd, u32 level, OptName optname, size_t optlen, const void* optval);
+    Errno ShutdownImpl(s32 fd, s32 how);
+    std::pair<s32, Errno> RecvImpl(s32 fd, u32 flags, std::vector<u8>& message);
+    std::pair<s32, Errno> RecvFromImpl(s32 fd, u32 flags, std::vector<u8>& message,
+                                       std::vector<u8>& addr);
+    std::pair<s32, Errno> SendImpl(s32 fd, u32 flags, const std::vector<u8>& message);
+    std::pair<s32, Errno> SendToImpl(s32 fd, u32 flags, const std::vector<u8>& message,
+                                     const std::vector<u8>& addr);
+    Errno CloseImpl(s32 fd);
+
+    s32 FindFreeFileDescriptorHandle() noexcept;
+    bool IsFileDescriptorValid(s32 fd) const noexcept;
+    bool IsBlockingSocket(s32 fd) const noexcept;
+
+    void BuildErrnoResponse(Kernel::HLERequestContext& ctx, Errno bsd_errno) const noexcept;
+
+    std::array<std::optional<FileDescriptor>, MAX_FD> file_descriptors;
+
+    BlockingWorkerPool<BSD, PollWork, AcceptWork, ConnectWork, RecvWork, RecvFromWork, SendWork,
+                       SendToWork>
+        worker_pool;
 };
 
 class BSDCFG final : public ServiceFramework<BSDCFG> {

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

@@ -10,9 +10,9 @@
 
 namespace Service::Sockets {
 
-void InstallInterfaces(SM::ServiceManager& service_manager) {
-    std::make_shared<BSD>("bsd:s")->InstallAsService(service_manager);
-    std::make_shared<BSD>("bsd:u")->InstallAsService(service_manager);
+void InstallInterfaces(SM::ServiceManager& service_manager, Core::System& system) {
+    std::make_shared<BSD>(system, "bsd:s")->InstallAsService(service_manager);
+    std::make_shared<BSD>(system, "bsd:u")->InstallAsService(service_manager);
     std::make_shared<BSDCFG>()->InstallAsService(service_manager);
 
     std::make_shared<ETHC_C>()->InstallAsService(service_manager);

src/core/hle/service/sockets/sockets.h

@@ -4,11 +4,94 @@
 
 #pragma once
 
+#include "common/common_types.h"
 #include "core/hle/service/service.h"
 
+namespace Core {
+class System;
+}
+
 namespace Service::Sockets {
 
+enum class Errno : u32 {
+    SUCCESS = 0,
+    BADF = 9,
+    AGAIN = 11,
+    INVAL = 22,
+    MFILE = 24,
+    NOTCONN = 107,
+};
+
+enum class Domain : u32 {
+    INET = 2,
+};
+
+enum class Type : u32 {
+    STREAM = 1,
+    DGRAM = 2,
+    RAW = 3,
+    SEQPACKET = 5,
+};
+
+enum class Protocol : u32 {
+    UNSPECIFIED = 0,
+    ICMP = 1,
+    TCP = 6,
+    UDP = 17,
+};
+
+enum class OptName : u32 {
+    REUSEADDR = 0x4,
+    BROADCAST = 0x20,
+    LINGER = 0x80,
+    SNDBUF = 0x1001,
+    RCVBUF = 0x1002,
+    SNDTIMEO = 0x1005,
+    RCVTIMEO = 0x1006,
+};
+
+enum class ShutdownHow : s32 {
+    RD = 0,
+    WR = 1,
+    RDWR = 2,
+};
+
+enum class FcntlCmd : s32 {
+    GETFL = 3,
+    SETFL = 4,
+};
+
+struct SockAddrIn {
+    u8 len;
+    u8 family;
+    u16 portno;
+    std::array<u8, 4> ip;
+    std::array<u8, 8> zeroes;
+};
+
+struct PollFD {
+    s32 fd;
+    u16 events;
+    u16 revents;
+};
+
+struct Linger {
+    u32 onoff;
+    u32 linger;
+};
+
+constexpr u16 POLL_IN = 0x01;
+constexpr u16 POLL_PRI = 0x02;
+constexpr u16 POLL_OUT = 0x04;
+constexpr u16 POLL_ERR = 0x08;
+constexpr u16 POLL_HUP = 0x10;
+constexpr u16 POLL_NVAL = 0x20;
+
+constexpr u32 FLAG_MSG_DONTWAIT = 0x80;
+
+constexpr u32 FLAG_O_NONBLOCK = 0x800;
+
 /// Registers all Sockets services with the specified service manager.
-void InstallInterfaces(SM::ServiceManager& service_manager);
+void InstallInterfaces(SM::ServiceManager& service_manager, Core::System& system);
 
 } // namespace Service::Sockets

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

@@ -0,0 +1,165 @@
+// Copyright 2020 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <utility>
+
+#include "common/assert.h"
+#include "common/common_types.h"
+#include "core/hle/service/sockets/sockets.h"
+#include "core/hle/service/sockets/sockets_translate.h"
+#include "core/network/network.h"
+
+namespace Service::Sockets {
+
+Errno Translate(Network::Errno value) {
+    switch (value) {
+    case Network::Errno::SUCCESS:
+        return Errno::SUCCESS;
+    case Network::Errno::BADF:
+        return Errno::BADF;
+    case Network::Errno::AGAIN:
+        return Errno::AGAIN;
+    case Network::Errno::INVAL:
+        return Errno::INVAL;
+    case Network::Errno::MFILE:
+        return Errno::MFILE;
+    case Network::Errno::NOTCONN:
+        return Errno::NOTCONN;
+    default:
+        UNIMPLEMENTED_MSG("Unimplemented errno={}", static_cast<int>(value));
+        return Errno::SUCCESS;
+    }
+}
+
+std::pair<s32, Errno> Translate(std::pair<s32, Network::Errno> value) {
+    return {value.first, Translate(value.second)};
+}
+
+Network::Domain Translate(Domain domain) {
+    switch (domain) {
+    case Domain::INET:
+        return Network::Domain::INET;
+    default:
+        UNIMPLEMENTED_MSG("Unimplemented domain={}", static_cast<int>(domain));
+        return {};
+    }
+}
+
+Domain Translate(Network::Domain domain) {
+    switch (domain) {
+    case Network::Domain::INET:
+        return Domain::INET;
+    default:
+        UNIMPLEMENTED_MSG("Unimplemented domain={}", static_cast<int>(domain));
+        return {};
+    }
+}
+
+Network::Type Translate(Type type) {
+    switch (type) {
+    case Type::STREAM:
+        return Network::Type::STREAM;
+    case Type::DGRAM:
+        return Network::Type::DGRAM;
+    default:
+        UNIMPLEMENTED_MSG("Unimplemented type={}", static_cast<int>(type));
+    }
+}
+
+Network::Protocol Translate(Type type, Protocol protocol) {
+    switch (protocol) {
+    case Protocol::UNSPECIFIED:
+        LOG_WARNING(Service, "Unspecified protocol, assuming protocol from type");
+        switch (type) {
+        case Type::DGRAM:
+            return Network::Protocol::UDP;
+        case Type::STREAM:
+            return Network::Protocol::TCP;
+        default:
+            return Network::Protocol::TCP;
+        }
+    case Protocol::TCP:
+        return Network::Protocol::TCP;
+    case Protocol::UDP:
+        return Network::Protocol::UDP;
+    default:
+        UNIMPLEMENTED_MSG("Unimplemented protocol={}", static_cast<int>(protocol));
+        return Network::Protocol::TCP;
+    }
+}
+
+u16 TranslatePollEventsToHost(u16 flags) {
+    u16 result = 0;
+    const auto translate = [&result, &flags](u16 from, u16 to) {
+        if ((flags & from) != 0) {
+            flags &= ~from;
+            result |= to;
+        }
+    };
+    translate(POLL_IN, Network::POLL_IN);
+    translate(POLL_PRI, Network::POLL_PRI);
+    translate(POLL_OUT, Network::POLL_OUT);
+    translate(POLL_ERR, Network::POLL_ERR);
+    translate(POLL_HUP, Network::POLL_HUP);
+    translate(POLL_NVAL, Network::POLL_NVAL);
+
+    UNIMPLEMENTED_IF_MSG(flags != 0, "Unimplemented flags={}", flags);
+    return result;
+}
+
+u16 TranslatePollEventsToGuest(u16 flags) {
+    u16 result = 0;
+    const auto translate = [&result, &flags](u16 from, u16 to) {
+        if ((flags & from) != 0) {
+            flags &= ~from;
+            result |= to;
+        }
+    };
+
+    translate(Network::POLL_IN, POLL_IN);
+    translate(Network::POLL_PRI, POLL_PRI);
+    translate(Network::POLL_OUT, POLL_OUT);
+    translate(Network::POLL_ERR, POLL_ERR);
+    translate(Network::POLL_HUP, POLL_HUP);
+    translate(Network::POLL_NVAL, POLL_NVAL);
+
+    UNIMPLEMENTED_IF_MSG(flags != 0, "Unimplemented flags={}", flags);
+    return result;
+}
+
+Network::SockAddrIn Translate(SockAddrIn value) {
+    ASSERT(value.len == 0 || value.len == sizeof(value));
+
+    Network::SockAddrIn result;
+    result.family = Translate(static_cast<Domain>(value.family));
+    result.ip = value.ip;
+    result.portno = value.portno >> 8 | value.portno << 8;
+    return result;
+}
+
+SockAddrIn Translate(Network::SockAddrIn value) {
+    SockAddrIn result;
+    result.len = sizeof(result);
+    result.family = static_cast<u8>(Translate(value.family));
+    result.portno = value.portno >> 8 | value.portno << 8;
+    result.ip = value.ip;
+    result.zeroes = {};
+    return result;
+}
+
+Network::ShutdownHow Translate(ShutdownHow how) {
+    switch (how) {
+    case ShutdownHow::RD:
+        return Network::ShutdownHow::RD;
+    case ShutdownHow::WR:
+        return Network::ShutdownHow::WR;
+    case ShutdownHow::RDWR:
+        return Network::ShutdownHow::RDWR;
+    default:
+        UNIMPLEMENTED_MSG("Unimplemented how={}", static_cast<int>(how));
+        return {};
+    }
+}
+
+} // namespace Service::Sockets

src/core/hle/service/sockets/sockets_translate.h

@@ -0,0 +1,48 @@
+// Copyright 2020 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <utility>
+
+#include "common/common_types.h"
+#include "core/hle/service/sockets/sockets.h"
+#include "core/network/network.h"
+
+namespace Service::Sockets {
+
+/// Translate abstract errno to guest errno
+Errno Translate(Network::Errno value);
+
+/// Translate abstract return value errno pair to guest return value errno pair
+std::pair<s32, Errno> Translate(std::pair<s32, Network::Errno> value);
+
+/// Translate guest domain to abstract domain
+Network::Domain Translate(Domain domain);
+
+/// Translate abstract domain to guest domain
+Domain Translate(Network::Domain domain);
+
+/// Translate guest type to abstract type
+Network::Type Translate(Type type);
+
+/// Translate guest protocol to abstract protocol
+Network::Protocol Translate(Type type, Protocol protocol);
+
+/// Translate abstract poll event flags to guest poll event flags
+u16 TranslatePollEventsToHost(u16 flags);
+
+/// Translate guest poll event flags to abstract poll event flags
+u16 TranslatePollEventsToGuest(u16 flags);
+
+/// Translate guest socket address structure to abstract socket address structure
+Network::SockAddrIn Translate(SockAddrIn value);
+
+/// Translate abstract socket address structure to guest socket address structure
+SockAddrIn Translate(Network::SockAddrIn value);
+
+/// Translate guest shutdown mode to abstract shutdown mode
+Network::ShutdownHow Translate(ShutdownHow how);
+
+} // namespace Service::Sockets

src/core/settings.h

@@ -67,6 +67,11 @@ private:
     Type local{};
 };
 
+struct TouchFromButtonMap {
+    std::string name;
+    std::vector<std::string> buttons;
+};
+
 struct Values {
     // Audio
     std::string audio_device_id;
@@ -145,15 +150,18 @@ struct Values {
     ButtonsRaw debug_pad_buttons;
     AnalogsRaw debug_pad_analogs;
 
-    std::string motion_device;
-
     bool vibration_enabled;
 
+    std::string motion_device;
+    std::string touch_device;
     TouchscreenInput touchscreen;
     std::atomic_bool is_device_reload_pending{true};
+    bool use_touch_from_button;
+    int touch_from_button_map_index;
     std::string udp_input_address;
     u16 udp_input_port;
     u8 udp_pad_index;
+    std::vector<TouchFromButtonMap> touch_from_button_maps;
 
     // Data Storage
     bool use_virtual_sd;

src/input_common/CMakeLists.txt

@@ -7,8 +7,12 @@ add_library(input_common STATIC
     main.h
     motion_emu.cpp
     motion_emu.h
+    motion_input.cpp
+    motion_input.h
     settings.cpp
     settings.h
+    touch_from_button.cpp
+    touch_from_button.h
     gcadapter/gc_adapter.cpp
     gcadapter/gc_adapter.h
     gcadapter/gc_poller.cpp

src/input_common/gcadapter/gc_adapter.cpp

@@ -283,7 +283,7 @@ void Adapter::Reset() {
     }
 }
 
-bool Adapter::DeviceConnected(std::size_t port) {
+bool Adapter::DeviceConnected(std::size_t port) const {
     return adapter_controllers_status[port] != ControllerTypes::None;
 }
 

src/input_common/gcadapter/gc_adapter.h

@@ -76,7 +76,7 @@ public:
     void EndConfiguration();
 
     /// Returns true if there is a device connected to port
-    bool DeviceConnected(std::size_t port);
+    bool DeviceConnected(std::size_t port) const;
 
     std::array<Common::SPSCQueue<GCPadStatus>, 4>& GetPadQueue();
     const std::array<Common::SPSCQueue<GCPadStatus>, 4>& GetPadQueue() const;

src/input_common/gcadapter/gc_poller.cpp

@@ -15,7 +15,7 @@ namespace InputCommon {
 
 class GCButton final : public Input::ButtonDevice {
 public:
-    explicit GCButton(int port_, int button_, GCAdapter::Adapter* adapter)
+    explicit GCButton(int port_, int button_, const GCAdapter::Adapter* adapter)
         : port(port_), button(button_), gcadapter(adapter) {}
 
     ~GCButton() override;
@@ -30,15 +30,16 @@ public:
 private:
     const int port;
     const int button;
-    GCAdapter::Adapter* gcadapter;
+    const GCAdapter::Adapter* gcadapter;
 };
 
 class GCAxisButton final : public Input::ButtonDevice {
 public:
     explicit GCAxisButton(int port_, int axis_, float threshold_, bool trigger_if_greater_,
-                          GCAdapter::Adapter* adapter)
+                          const GCAdapter::Adapter* adapter)
         : port(port_), axis(axis_), threshold(threshold_), trigger_if_greater(trigger_if_greater_),
-          gcadapter(adapter), origin_value(adapter->GetOriginValue(port_, axis_)) {}
+          gcadapter(adapter),
+          origin_value(static_cast<float>(adapter->GetOriginValue(port_, axis_))) {}
 
     bool GetStatus() const override {
         if (gcadapter->DeviceConnected(port)) {
@@ -59,7 +60,7 @@ private:
     const int axis;
     float threshold;
     bool trigger_if_greater;
-    GCAdapter::Adapter* gcadapter;
+    const GCAdapter::Adapter* gcadapter;
     const float origin_value;
 };
 
@@ -148,11 +149,12 @@ void GCButtonFactory::EndConfiguration() {
 
 class GCAnalog final : public Input::AnalogDevice {
 public:
-    GCAnalog(int port_, int axis_x_, int axis_y_, float deadzone_, GCAdapter::Adapter* adapter,
-             float range_)
+    GCAnalog(int port_, int axis_x_, int axis_y_, float deadzone_,
+             const GCAdapter::Adapter* adapter, float range_)
         : port(port_), axis_x(axis_x_), axis_y(axis_y_), deadzone(deadzone_), gcadapter(adapter),
-          origin_value_x(adapter->GetOriginValue(port_, axis_x_)),
-          origin_value_y(adapter->GetOriginValue(port_, axis_y_)), range(range_) {}
+          origin_value_x(static_cast<float>(adapter->GetOriginValue(port_, axis_x_))),
+          origin_value_y(static_cast<float>(adapter->GetOriginValue(port_, axis_y_))),
+          range(range_) {}
 
     float GetAxis(int axis) const {
         if (gcadapter->DeviceConnected(port)) {
@@ -210,7 +212,7 @@ private:
     const int axis_x;
     const int axis_y;
     const float deadzone;
-    GCAdapter::Adapter* gcadapter;
+    const GCAdapter::Adapter* gcadapter;
     const float origin_value_x;
     const float origin_value_y;
     const float range;

src/input_common/main.cpp

@@ -11,6 +11,7 @@
 #include "input_common/keyboard.h"
 #include "input_common/main.h"
 #include "input_common/motion_emu.h"
+#include "input_common/touch_from_button.h"
 #include "input_common/udp/udp.h"
 #ifdef HAVE_SDL2
 #include "input_common/sdl/sdl.h"
@@ -32,6 +33,8 @@ struct InputSubsystem::Impl {
                                                     std::make_shared<AnalogFromButton>());
         motion_emu = std::make_shared<MotionEmu>();
         Input::RegisterFactory<Input::MotionDevice>("motion_emu", motion_emu);
+        Input::RegisterFactory<Input::TouchDevice>("touch_from_button",
+                                                   std::make_shared<TouchFromButtonFactory>());
 
 #ifdef HAVE_SDL2
         sdl = SDL::Init();
@@ -46,6 +49,7 @@ struct InputSubsystem::Impl {
         Input::UnregisterFactory<Input::AnalogDevice>("analog_from_button");
         Input::UnregisterFactory<Input::MotionDevice>("motion_emu");
         motion_emu.reset();
+        Input::UnregisterFactory<Input::TouchDevice>("touch_from_button");
 #ifdef HAVE_SDL2
         sdl.reset();
 #endif
@@ -171,6 +175,13 @@ const GCButtonFactory* InputSubsystem::GetGCButtons() const {
     return impl->gcbuttons.get();
 }
 
+void InputSubsystem::ReloadInputDevices() {
+    if (!impl->udp) {
+        return;
+    }
+    impl->udp->ReloadUDPClient();
+}
+
 std::vector<std::unique_ptr<Polling::DevicePoller>> InputSubsystem::GetPollers(
     Polling::DeviceType type) const {
 #ifdef HAVE_SDL2

src/input_common/main.h

@@ -8,13 +8,19 @@
 #include <string>
 #include <unordered_map>
 #include <vector>
-#include "input_common/gcadapter/gc_poller.h"
-#include "input_common/settings.h"
 
 namespace Common {
 class ParamPackage;
 }
 
+namespace Settings::NativeAnalog {
+enum Values : int;
+}
+
+namespace Settings::NativeButton {
+enum Values : int;
+}
+
 namespace InputCommon {
 namespace Polling {
 
@@ -40,9 +46,6 @@ public:
      */
     virtual Common::ParamPackage GetNextInput() = 0;
 };
-
-// Get all DevicePoller from all backends for a specific device type
-std::vector<std::unique_ptr<DevicePoller>> GetPollers(DeviceType type);
 } // namespace Polling
 
 class GCAnalogFactory;
@@ -112,6 +115,9 @@ public:
     /// Retrieves the underlying GameCube button handler.
     [[nodiscard]] const GCButtonFactory* GetGCButtons() const;
 
+    /// Reloads the input devices
+    void ReloadInputDevices();
+
     /// Get all DevicePoller from all backends for a specific device type
     [[nodiscard]] std::vector<std::unique_ptr<Polling::DevicePoller>> GetPollers(
         Polling::DeviceType type) const;

src/input_common/motion_input.cpp

@@ -0,0 +1,181 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included
+
+#include "common/math_util.h"
+#include "input_common/motion_input.h"
+
+namespace InputCommon {
+
+MotionInput::MotionInput(f32 new_kp, f32 new_ki, f32 new_kd)
+    : kp(new_kp), ki(new_ki), kd(new_kd), quat{{0, 0, -1}, 0} {}
+
+void MotionInput::SetAcceleration(const Common::Vec3f& acceleration) {
+    accel = acceleration;
+}
+
+void MotionInput::SetGyroscope(const Common::Vec3f& gyroscope) {
+    gyro = gyroscope - gyro_drift;
+    if (gyro.Length2() < gyro_threshold) {
+        gyro = {};
+    }
+}
+
+void MotionInput::SetQuaternion(const Common::Quaternion<f32>& quaternion) {
+    quat = quaternion;
+}
+
+void MotionInput::SetGyroDrift(const Common::Vec3f& drift) {
+    gyro_drift = drift;
+}
+
+void MotionInput::SetGyroThreshold(f32 threshold) {
+    gyro_threshold = threshold;
+}
+
+void MotionInput::EnableReset(bool reset) {
+    reset_enabled = reset;
+}
+
+void MotionInput::ResetRotations() {
+    rotations = {};
+}
+
+bool MotionInput::IsMoving(f32 sensitivity) const {
+    return gyro.Length() >= sensitivity || accel.Length() <= 0.9f || accel.Length() >= 1.1f;
+}
+
+bool MotionInput::IsCalibrated(f32 sensitivity) const {
+    return real_error.Length() < sensitivity;
+}
+
+void MotionInput::UpdateRotation(u64 elapsed_time) {
+    const f32 sample_period = elapsed_time / 1000000.0f;
+    if (sample_period > 0.1f) {
+        return;
+    }
+    rotations += gyro * sample_period;
+}
+
+void MotionInput::UpdateOrientation(u64 elapsed_time) {
+    if (!IsCalibrated(0.1f)) {
+        ResetOrientation();
+    }
+    // Short name local variable for readability
+    f32 q1 = quat.w;
+    f32 q2 = quat.xyz[0];
+    f32 q3 = quat.xyz[1];
+    f32 q4 = quat.xyz[2];
+    const f32 sample_period = elapsed_time / 1000000.0f;
+
+    // ignore invalid elapsed time
+    if (sample_period > 0.1f) {
+        return;
+    }
+
+    const auto normal_accel = accel.Normalized();
+    auto rad_gyro = gyro * Common::PI * 2;
+    const f32 swap = rad_gyro.x;
+    rad_gyro.x = rad_gyro.y;
+    rad_gyro.y = -swap;
+    rad_gyro.z = -rad_gyro.z;
+
+    // Ignore drift correction if acceleration is not reliable
+    if (accel.Length() >= 0.75f && accel.Length() <= 1.25f) {
+        const f32 ax = -normal_accel.x;
+        const f32 ay = normal_accel.y;
+        const f32 az = -normal_accel.z;
+
+        // Estimated direction of gravity
+        const f32 vx = 2.0f * (q2 * q4 - q1 * q3);
+        const f32 vy = 2.0f * (q1 * q2 + q3 * q4);
+        const f32 vz = q1 * q1 - q2 * q2 - q3 * q3 + q4 * q4;
+
+        // Error is cross product between estimated direction and measured direction of gravity
+        const Common::Vec3f new_real_error = {az * vx - ax * vz, ay * vz - az * vy,
+                                              ax * vy - ay * vx};
+
+        derivative_error = new_real_error - real_error;
+        real_error = new_real_error;
+
+        // Prevent integral windup
+        if (ki != 0.0f && !IsCalibrated(0.05f)) {
+            integral_error += real_error;
+        } else {
+            integral_error = {};
+        }
+
+        // Apply feedback terms
+        rad_gyro += kp * real_error;
+        rad_gyro += ki * integral_error;
+        rad_gyro += kd * derivative_error;
+    }
+
+    const f32 gx = rad_gyro.y;
+    const f32 gy = rad_gyro.x;
+    const f32 gz = rad_gyro.z;
+
+    // Integrate rate of change of quaternion
+    const f32 pa = q2;
+    const f32 pb = q3;
+    const f32 pc = q4;
+    q1 = q1 + (-q2 * gx - q3 * gy - q4 * gz) * (0.5f * sample_period);
+    q2 = pa + (q1 * gx + pb * gz - pc * gy) * (0.5f * sample_period);
+    q3 = pb + (q1 * gy - pa * gz + pc * gx) * (0.5f * sample_period);
+    q4 = pc + (q1 * gz + pa * gy - pb * gx) * (0.5f * sample_period);
+
+    quat.w = q1;
+    quat.xyz[0] = q2;
+    quat.xyz[1] = q3;
+    quat.xyz[2] = q4;
+    quat = quat.Normalized();
+}
+
+std::array<Common::Vec3f, 3> MotionInput::GetOrientation() const {
+    const Common::Quaternion<float> quad{
+        .xyz = {-quat.xyz[1], -quat.xyz[0], -quat.w},
+        .w = -quat.xyz[2],
+    };
+    const std::array<float, 16> matrix4x4 = quad.ToMatrix();
+
+    return {Common::Vec3f(matrix4x4[0], matrix4x4[1], -matrix4x4[2]),
+            Common::Vec3f(matrix4x4[4], matrix4x4[5], -matrix4x4[6]),
+            Common::Vec3f(-matrix4x4[8], -matrix4x4[9], matrix4x4[10])};
+}
+
+Common::Vec3f MotionInput::GetAcceleration() const {
+    return accel;
+}
+
+Common::Vec3f MotionInput::GetGyroscope() const {
+    return gyro;
+}
+
+Common::Quaternion<f32> MotionInput::GetQuaternion() const {
+    return quat;
+}
+
+Common::Vec3f MotionInput::GetRotations() const {
+    return rotations;
+}
+
+void MotionInput::ResetOrientation() {
+    if (!reset_enabled) {
+        return;
+    }
+    if (!IsMoving(0.5f) && accel.z <= -0.9f) {
+        ++reset_counter;
+        if (reset_counter > 900) {
+            // TODO: calculate quaternion from gravity vector
+            quat.w = 0;
+            quat.xyz[0] = 0;
+            quat.xyz[1] = 0;
+            quat.xyz[2] = -1;
+            integral_error = {};
+            reset_counter = 0;
+        }
+    } else {
+        reset_counter = 0;
+    }
+}
+} // namespace InputCommon

src/input_common/motion_input.h

@@ -0,0 +1,68 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included
+
+#pragma once
+
+#include "common/common_types.h"
+#include "common/quaternion.h"
+#include "common/vector_math.h"
+
+namespace InputCommon {
+
+class MotionInput {
+public:
+    MotionInput(f32 new_kp, f32 new_ki, f32 new_kd);
+
+    MotionInput(const MotionInput&) = default;
+    MotionInput& operator=(const MotionInput&) = default;
+
+    MotionInput(MotionInput&&) = default;
+    MotionInput& operator=(MotionInput&&) = default;
+
+    void SetAcceleration(const Common::Vec3f& acceleration);
+    void SetGyroscope(const Common::Vec3f& acceleration);
+    void SetQuaternion(const Common::Quaternion<f32>& quaternion);
+    void SetGyroDrift(const Common::Vec3f& drift);
+    void SetGyroThreshold(f32 threshold);
+
+    void EnableReset(bool reset);
+    void ResetRotations();
+
+    void UpdateRotation(u64 elapsed_time);
+    void UpdateOrientation(u64 elapsed_time);
+
+    std::array<Common::Vec3f, 3> GetOrientation() const;
+    Common::Vec3f GetAcceleration() const;
+    Common::Vec3f GetGyroscope() const;
+    Common::Vec3f GetRotations() const;
+    Common::Quaternion<f32> GetQuaternion() const;
+
+    bool IsMoving(f32 sensitivity) const;
+    bool IsCalibrated(f32 sensitivity) const;
+
+private:
+    void ResetOrientation();
+
+    // PID constants
+    const f32 kp;
+    const f32 ki;
+    const f32 kd;
+
+    // PID errors
+    Common::Vec3f real_error;
+    Common::Vec3f integral_error;
+    Common::Vec3f derivative_error;
+
+    Common::Quaternion<f32> quat;
+    Common::Vec3f rotations;
+    Common::Vec3f accel;
+    Common::Vec3f gyro;
+    Common::Vec3f gyro_drift;
+
+    f32 gyro_threshold = 0.0f;
+    u32 reset_counter = 0;
+    bool reset_enabled = true;
+};
+
+} // namespace InputCommon

src/input_common/sdl/sdl_impl.cpp

@@ -3,6 +3,7 @@
 // Refer to the license.txt file included.
 
 #include <algorithm>
+#include <array>
 #include <atomic>
 #include <cmath>
 #include <functional>
@@ -17,11 +18,11 @@
 #include <vector>
 #include <SDL.h>
 #include "common/logging/log.h"
-#include "common/math_util.h"
 #include "common/param_package.h"
 #include "common/threadsafe_queue.h"
 #include "core/frontend/input.h"
 #include "input_common/sdl/sdl_impl.h"
+#include "input_common/settings.h"
 
 namespace InputCommon::SDL {
 
@@ -358,7 +359,7 @@ public:
             return std::make_tuple(x / r * (r - deadzone) / (1 - deadzone),
                                    y / r * (r - deadzone) / (1 - deadzone));
         }
-        return std::make_tuple<float, float>(0.0f, 0.0f);
+        return {};
     }
 
     bool GetAnalogDirectionStatus(Input::AnalogDirection direction) const override {
@@ -574,10 +575,10 @@ std::vector<Common::ParamPackage> SDLState::GetInputDevices() {
 
 namespace {
 Common::ParamPackage BuildAnalogParamPackageForButton(int port, std::string guid, u8 axis,
-                                                      float value = 0.1) {
+                                                      float value = 0.1f) {
     Common::ParamPackage params({{"engine", "sdl"}});
     params.Set("port", port);
-    params.Set("guid", guid);
+    params.Set("guid", std::move(guid));
     params.Set("axis", axis);
     if (value > 0) {
         params.Set("direction", "+");
@@ -592,7 +593,7 @@ Common::ParamPackage BuildAnalogParamPackageForButton(int port, std::string guid
 Common::ParamPackage BuildButtonParamPackageForButton(int port, std::string guid, u8 button) {
     Common::ParamPackage params({{"engine", "sdl"}});
     params.Set("port", port);
-    params.Set("guid", guid);
+    params.Set("guid", std::move(guid));
     params.Set("button", button);
     return params;
 }
@@ -601,7 +602,7 @@ Common::ParamPackage BuildHatParamPackageForButton(int port, std::string guid, u
     Common::ParamPackage params({{"engine", "sdl"}});
 
     params.Set("port", port);
-    params.Set("guid", guid);
+    params.Set("guid", std::move(guid));
     params.Set("hat", hat);
     switch (value) {
     case SDL_HAT_UP:
@@ -670,55 +671,62 @@ Common::ParamPackage BuildParamPackageForAnalog(int port, const std::string& gui
 } // Anonymous namespace
 
 ButtonMapping SDLState::GetButtonMappingForDevice(const Common::ParamPackage& params) {
-    // This list is missing ZL/ZR since those are not considered buttons in SDL GameController.
-    // We will add those afterwards
-    // This list also excludes Screenshot since theres not really a mapping for that
-    std::unordered_map<Settings::NativeButton::Values, SDL_GameControllerButton>
-        switch_to_sdl_button = {
-            {Settings::NativeButton::A, SDL_CONTROLLER_BUTTON_B},
-            {Settings::NativeButton::B, SDL_CONTROLLER_BUTTON_A},
-            {Settings::NativeButton::X, SDL_CONTROLLER_BUTTON_Y},
-            {Settings::NativeButton::Y, SDL_CONTROLLER_BUTTON_X},
-            {Settings::NativeButton::LStick, SDL_CONTROLLER_BUTTON_LEFTSTICK},
-            {Settings::NativeButton::RStick, SDL_CONTROLLER_BUTTON_RIGHTSTICK},
-            {Settings::NativeButton::L, SDL_CONTROLLER_BUTTON_LEFTSHOULDER},
-            {Settings::NativeButton::R, SDL_CONTROLLER_BUTTON_RIGHTSHOULDER},
-            {Settings::NativeButton::Plus, SDL_CONTROLLER_BUTTON_START},
-            {Settings::NativeButton::Minus, SDL_CONTROLLER_BUTTON_BACK},
-            {Settings::NativeButton::DLeft, SDL_CONTROLLER_BUTTON_DPAD_LEFT},
-            {Settings::NativeButton::DUp, SDL_CONTROLLER_BUTTON_DPAD_UP},
-            {Settings::NativeButton::DRight, SDL_CONTROLLER_BUTTON_DPAD_RIGHT},
-            {Settings::NativeButton::DDown, SDL_CONTROLLER_BUTTON_DPAD_DOWN},
-            {Settings::NativeButton::SL, SDL_CONTROLLER_BUTTON_LEFTSHOULDER},
-            {Settings::NativeButton::SR, SDL_CONTROLLER_BUTTON_RIGHTSHOULDER},
-            {Settings::NativeButton::Home, SDL_CONTROLLER_BUTTON_GUIDE},
-        };
     if (!params.Has("guid") || !params.Has("port")) {
         return {};
     }
     const auto joystick = GetSDLJoystickByGUID(params.Get("guid", ""), params.Get("port", 0));
-    auto controller = joystick->GetSDLGameController();
-    if (!controller) {
+    auto* controller = joystick->GetSDLGameController();
+    if (controller == nullptr) {
         return {};
     }
 
-    ButtonMapping mapping{};
-    for (const auto& [switch_button, sdl_button] : switch_to_sdl_button) {
-        const auto& binding = SDL_GameControllerGetBindForButton(controller, sdl_button);
-        mapping[switch_button] =
-            BuildParamPackageForBinding(joystick->GetPort(), joystick->GetGUID(), binding);
-    }
+    // This list is missing ZL/ZR since those are not considered buttons in SDL GameController.
+    // We will add those afterwards
+    // This list also excludes Screenshot since theres not really a mapping for that
+    using ButtonBindings =
+        std::array<std::pair<Settings::NativeButton::Values, SDL_GameControllerButton>, 17>;
+    static constexpr ButtonBindings switch_to_sdl_button{{
+        {Settings::NativeButton::A, SDL_CONTROLLER_BUTTON_B},
+        {Settings::NativeButton::B, SDL_CONTROLLER_BUTTON_A},
+        {Settings::NativeButton::X, SDL_CONTROLLER_BUTTON_Y},
+        {Settings::NativeButton::Y, SDL_CONTROLLER_BUTTON_X},
+        {Settings::NativeButton::LStick, SDL_CONTROLLER_BUTTON_LEFTSTICK},
+        {Settings::NativeButton::RStick, SDL_CONTROLLER_BUTTON_RIGHTSTICK},
+        {Settings::NativeButton::L, SDL_CONTROLLER_BUTTON_LEFTSHOULDER},
+        {Settings::NativeButton::R, SDL_CONTROLLER_BUTTON_RIGHTSHOULDER},
+        {Settings::NativeButton::Plus, SDL_CONTROLLER_BUTTON_START},
+        {Settings::NativeButton::Minus, SDL_CONTROLLER_BUTTON_BACK},
+        {Settings::NativeButton::DLeft, SDL_CONTROLLER_BUTTON_DPAD_LEFT},
+        {Settings::NativeButton::DUp, SDL_CONTROLLER_BUTTON_DPAD_UP},
+        {Settings::NativeButton::DRight, SDL_CONTROLLER_BUTTON_DPAD_RIGHT},
+        {Settings::NativeButton::DDown, SDL_CONTROLLER_BUTTON_DPAD_DOWN},
+        {Settings::NativeButton::SL, SDL_CONTROLLER_BUTTON_LEFTSHOULDER},
+        {Settings::NativeButton::SR, SDL_CONTROLLER_BUTTON_RIGHTSHOULDER},
+        {Settings::NativeButton::Home, SDL_CONTROLLER_BUTTON_GUIDE},
+    }};
 
     // Add the missing bindings for ZL/ZR
-    std::unordered_map<Settings::NativeButton::Values, SDL_GameControllerAxis> switch_to_sdl_axis =
-        {
-            {Settings::NativeButton::ZL, SDL_CONTROLLER_AXIS_TRIGGERLEFT},
-            {Settings::NativeButton::ZR, SDL_CONTROLLER_AXIS_TRIGGERRIGHT},
-        };
+    using ZBindings =
+        std::array<std::pair<Settings::NativeButton::Values, SDL_GameControllerAxis>, 2>;
+    static constexpr ZBindings switch_to_sdl_axis{{
+        {Settings::NativeButton::ZL, SDL_CONTROLLER_AXIS_TRIGGERLEFT},
+        {Settings::NativeButton::ZR, SDL_CONTROLLER_AXIS_TRIGGERRIGHT},
+    }};
+
+    ButtonMapping mapping;
+    mapping.reserve(switch_to_sdl_button.size() + switch_to_sdl_axis.size());
+
+    for (const auto& [switch_button, sdl_button] : switch_to_sdl_button) {
+        const auto& binding = SDL_GameControllerGetBindForButton(controller, sdl_button);
+        mapping.insert_or_assign(
+            switch_button,
+            BuildParamPackageForBinding(joystick->GetPort(), joystick->GetGUID(), binding));
+    }
     for (const auto& [switch_button, sdl_axis] : switch_to_sdl_axis) {
         const auto& binding = SDL_GameControllerGetBindForAxis(controller, sdl_axis);
-        mapping[switch_button] =
-            BuildParamPackageForBinding(joystick->GetPort(), joystick->GetGUID(), binding);
+        mapping.insert_or_assign(
+            switch_button,
+            BuildParamPackageForBinding(joystick->GetPort(), joystick->GetGUID(), binding));
     }
 
     return mapping;
@@ -729,8 +737,8 @@ AnalogMapping SDLState::GetAnalogMappingForDevice(const Common::ParamPackage& pa
         return {};
     }
     const auto joystick = GetSDLJoystickByGUID(params.Get("guid", ""), params.Get("port", 0));
-    auto controller = joystick->GetSDLGameController();
-    if (!controller) {
+    auto* controller = joystick->GetSDLGameController();
+    if (controller == nullptr) {
         return {};
     }
 
@@ -739,16 +747,18 @@ AnalogMapping SDLState::GetAnalogMappingForDevice(const Common::ParamPackage& pa
         SDL_GameControllerGetBindForAxis(controller, SDL_CONTROLLER_AXIS_LEFTX);
     const auto& binding_left_y =
         SDL_GameControllerGetBindForAxis(controller, SDL_CONTROLLER_AXIS_LEFTY);
-    mapping[Settings::NativeAnalog::LStick] =
-        BuildParamPackageForAnalog(joystick->GetPort(), joystick->GetGUID(),
-                                   binding_left_x.value.axis, binding_left_y.value.axis);
+    mapping.insert_or_assign(Settings::NativeAnalog::LStick,
+                             BuildParamPackageForAnalog(joystick->GetPort(), joystick->GetGUID(),
+                                                        binding_left_x.value.axis,
+                                                        binding_left_y.value.axis));
     const auto& binding_right_x =
         SDL_GameControllerGetBindForAxis(controller, SDL_CONTROLLER_AXIS_RIGHTX);
     const auto& binding_right_y =
         SDL_GameControllerGetBindForAxis(controller, SDL_CONTROLLER_AXIS_RIGHTY);
-    mapping[Settings::NativeAnalog::RStick] =
-        BuildParamPackageForAnalog(joystick->GetPort(), joystick->GetGUID(),
-                                   binding_right_x.value.axis, binding_right_y.value.axis);
+    mapping.insert_or_assign(Settings::NativeAnalog::RStick,
+                             BuildParamPackageForAnalog(joystick->GetPort(), joystick->GetGUID(),
+                                                        binding_right_x.value.axis,
+                                                        binding_right_y.value.axis));
     return mapping;
 }
 
@@ -784,7 +794,7 @@ public:
         }
         return {};
     }
-    std::optional<Common::ParamPackage> FromEvent(const SDL_Event& event) {
+    [[nodiscard]] std::optional<Common::ParamPackage> FromEvent(const SDL_Event& event) const {
         switch (event.type) {
         case SDL_JOYAXISMOTION:
             if (std::abs(event.jaxis.value / 32767.0) < 0.5) {
@@ -795,7 +805,7 @@ public:
         case SDL_JOYHATMOTION:
             return {SDLEventToButtonParamPackage(state, event)};
         }
-        return {};
+        return std::nullopt;
     }
 };
 

src/input_common/settings.h

@@ -10,7 +10,7 @@
 
 namespace Settings {
 namespace NativeButton {
-enum Values {
+enum Values : int {
     A,
     B,
     X,
@@ -52,7 +52,7 @@ extern const std::array<const char*, NumButtons> mapping;
 } // namespace NativeButton
 
 namespace NativeAnalog {
-enum Values {
+enum Values : int {
     LStick,
     RStick,
 

src/input_common/touch_from_button.cpp

@@ -0,0 +1,50 @@
+// Copyright 2020 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/frontend/framebuffer_layout.h"
+#include "core/settings.h"
+#include "input_common/touch_from_button.h"
+
+namespace InputCommon {
+
+class TouchFromButtonDevice final : public Input::TouchDevice {
+public:
+    TouchFromButtonDevice() {
+        for (const auto& config_entry :
+             Settings::values.touch_from_button_maps[Settings::values.touch_from_button_map_index]
+                 .buttons) {
+            const Common::ParamPackage package{config_entry};
+            map.emplace_back(
+                Input::CreateDevice<Input::ButtonDevice>(config_entry),
+                std::clamp(package.Get("x", 0), 0, static_cast<int>(Layout::ScreenUndocked::Width)),
+                std::clamp(package.Get("y", 0), 0,
+                           static_cast<int>(Layout::ScreenUndocked::Height)));
+        }
+    }
+
+    std::tuple<float, float, bool> GetStatus() const override {
+        for (const auto& m : map) {
+            const bool state = std::get<0>(m)->GetStatus();
+            if (state) {
+                const float x = static_cast<float>(std::get<1>(m)) /
+                                static_cast<int>(Layout::ScreenUndocked::Width);
+                const float y = static_cast<float>(std::get<2>(m)) /
+                                static_cast<int>(Layout::ScreenUndocked::Height);
+                return {x, y, true};
+            }
+        }
+        return {};
+    }
+
+private:
+    // A vector of the mapped button, its x and its y-coordinate
+    std::vector<std::tuple<std::unique_ptr<Input::ButtonDevice>, int, int>> map;
+};
+
+std::unique_ptr<Input::TouchDevice> TouchFromButtonFactory::Create(
+    const Common::ParamPackage& params) {
+    return std::make_unique<TouchFromButtonDevice>();
+}
+
+} // namespace InputCommon

src/input_common/touch_from_button.h

@@ -0,0 +1,23 @@
+// Copyright 2020 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <memory>
+#include "core/frontend/input.h"
+
+namespace InputCommon {
+
+/**
+ * A touch device factory that takes a list of button devices and combines them into a touch device.
+ */
+class TouchFromButtonFactory final : public Input::Factory<Input::TouchDevice> {
+public:
+    /**
+     * Creates a touch device from a list of button devices
+     */
+    std::unique_ptr<Input::TouchDevice> Create(const Common::ParamPackage& params) override;
+};
+
+} // namespace InputCommon

src/video_core/buffer_cache/buffer_cache.h

@@ -51,46 +51,43 @@ public:
                             bool is_written = false, bool use_fast_cbuf = false) {
         std::lock_guard lock{mutex};
 
-        auto& memory_manager = system.GPU().MemoryManager();
-        const std::optional<VAddr> cpu_addr_opt = memory_manager.GpuToCpuAddress(gpu_addr);
-        if (!cpu_addr_opt) {
+        const std::optional<VAddr> cpu_addr = gpu_memory.GpuToCpuAddress(gpu_addr);
+        if (!cpu_addr) {
             return GetEmptyBuffer(size);
         }
-        const VAddr cpu_addr = *cpu_addr_opt;
 
         // Cache management is a big overhead, so only cache entries with a given size.
         // TODO: Figure out which size is the best for given games.
         constexpr std::size_t max_stream_size = 0x800;
         if (use_fast_cbuf || size < max_stream_size) {
-            if (!is_written && !IsRegionWritten(cpu_addr, cpu_addr + size - 1)) {
-                const bool is_granular = memory_manager.IsGranularRange(gpu_addr, size);
+            if (!is_written && !IsRegionWritten(*cpu_addr, *cpu_addr + size - 1)) {
+                const bool is_granular = gpu_memory.IsGranularRange(gpu_addr, size);
                 if (use_fast_cbuf) {
                     u8* dest;
                     if (is_granular) {
-                        dest = memory_manager.GetPointer(gpu_addr);
+                        dest = gpu_memory.GetPointer(gpu_addr);
                     } else {
                         staging_buffer.resize(size);
                         dest = staging_buffer.data();
-                        memory_manager.ReadBlockUnsafe(gpu_addr, dest, size);
+                        gpu_memory.ReadBlockUnsafe(gpu_addr, dest, size);
                     }
                     return ConstBufferUpload(dest, size);
                 }
                 if (is_granular) {
-                    u8* const host_ptr = memory_manager.GetPointer(gpu_addr);
+                    u8* const host_ptr = gpu_memory.GetPointer(gpu_addr);
                     return StreamBufferUpload(size, alignment, [host_ptr, size](u8* dest) {
                         std::memcpy(dest, host_ptr, size);
                     });
                 } else {
-                    return StreamBufferUpload(
-                        size, alignment, [&memory_manager, gpu_addr, size](u8* dest) {
-                            memory_manager.ReadBlockUnsafe(gpu_addr, dest, size);
-                        });
+                    return StreamBufferUpload(size, alignment, [this, gpu_addr, size](u8* dest) {
+                        gpu_memory.ReadBlockUnsafe(gpu_addr, dest, size);
+                    });
                 }
             }
         }
 
-        Buffer* const block = GetBlock(cpu_addr, size);
-        MapInterval* const map = MapAddress(block, gpu_addr, cpu_addr, size);
+        Buffer* const block = GetBlock(*cpu_addr, size);
+        MapInterval* const map = MapAddress(block, gpu_addr, *cpu_addr, size);
         if (!map) {
             return GetEmptyBuffer(size);
         }
@@ -106,7 +103,7 @@ public:
             }
         }
 
-        return BufferInfo{block->Handle(), block->Offset(cpu_addr), block->Address()};
+        return BufferInfo{block->Handle(), block->Offset(*cpu_addr), block->Address()};
     }
 
     /// Uploads from a host memory. Returns the OpenGL buffer where it's located and its offset.
@@ -262,9 +259,11 @@ public:
     virtual BufferInfo GetEmptyBuffer(std::size_t size) = 0;
 
 protected:
-    explicit BufferCache(VideoCore::RasterizerInterface& rasterizer, Core::System& system,
-                         std::unique_ptr<StreamBuffer> stream_buffer)
-        : rasterizer{rasterizer}, system{system}, stream_buffer{std::move(stream_buffer)} {}
+    explicit BufferCache(VideoCore::RasterizerInterface& rasterizer_,
+                         Tegra::MemoryManager& gpu_memory_, Core::Memory::Memory& cpu_memory_,
+                         std::unique_ptr<StreamBuffer> stream_buffer_)
+        : rasterizer{rasterizer_}, gpu_memory{gpu_memory_}, cpu_memory{cpu_memory_},
+          stream_buffer{std::move(stream_buffer_)}, stream_buffer_handle{stream_buffer->Handle()} {}
 
     ~BufferCache() = default;
 
@@ -326,14 +325,13 @@ private:
     MapInterval* MapAddress(Buffer* block, GPUVAddr gpu_addr, VAddr cpu_addr, std::size_t size) {
         const VectorMapInterval overlaps = GetMapsInRange(cpu_addr, size);
         if (overlaps.empty()) {
-            auto& memory_manager = system.GPU().MemoryManager();
             const VAddr cpu_addr_end = cpu_addr + size;
-            if (memory_manager.IsGranularRange(gpu_addr, size)) {
-                u8* host_ptr = memory_manager.GetPointer(gpu_addr);
+            if (gpu_memory.IsGranularRange(gpu_addr, size)) {
+                u8* const host_ptr = gpu_memory.GetPointer(gpu_addr);
                 block->Upload(block->Offset(cpu_addr), size, host_ptr);
             } else {
                 staging_buffer.resize(size);
-                memory_manager.ReadBlockUnsafe(gpu_addr, staging_buffer.data(), size);
+                gpu_memory.ReadBlockUnsafe(gpu_addr, staging_buffer.data(), size);
                 block->Upload(block->Offset(cpu_addr), size, staging_buffer.data());
             }
             return Register(MapInterval(cpu_addr, cpu_addr_end, gpu_addr));
@@ -392,7 +390,7 @@ private:
                 continue;
             }
             staging_buffer.resize(size);
-            system.Memory().ReadBlockUnsafe(interval.lower(), staging_buffer.data(), size);
+            cpu_memory.ReadBlockUnsafe(interval.lower(), staging_buffer.data(), size);
             block->Upload(block->Offset(interval.lower()), size, staging_buffer.data());
         }
     }
@@ -431,7 +429,7 @@ private:
         const std::size_t size = map->end - map->start;
         staging_buffer.resize(size);
         block->Download(block->Offset(map->start), size, staging_buffer.data());
-        system.Memory().WriteBlockUnsafe(map->start, staging_buffer.data(), size);
+        cpu_memory.WriteBlockUnsafe(map->start, staging_buffer.data(), size);
         map->MarkAsModified(false, 0);
     }
 
@@ -567,7 +565,8 @@ private:
     }
 
     VideoCore::RasterizerInterface& rasterizer;
-    Core::System& system;
+    Tegra::MemoryManager& gpu_memory;
+    Core::Memory::Memory& cpu_memory;
 
     std::unique_ptr<StreamBuffer> stream_buffer;
     BufferType stream_buffer_handle;

src/video_core/fence_manager.h

@@ -74,8 +74,6 @@ public:
     }
 
     void WaitPendingFences() {
-        auto& gpu{system.GPU()};
-        auto& memory_manager{gpu.MemoryManager()};
         while (!fences.empty()) {
             TFence& current_fence = fences.front();
             if (ShouldWait()) {
@@ -83,8 +81,8 @@ public:
             }
             PopAsyncFlushes();
             if (current_fence->IsSemaphore()) {
-                memory_manager.template Write<u32>(current_fence->GetAddress(),
-                                                   current_fence->GetPayload());
+                gpu_memory.template Write<u32>(current_fence->GetAddress(),
+                                               current_fence->GetPayload());
             } else {
                 gpu.IncrementSyncPoint(current_fence->GetPayload());
             }
@@ -93,13 +91,13 @@ public:
     }
 
 protected:
-    FenceManager(Core::System& system, VideoCore::RasterizerInterface& rasterizer,
-                 TTextureCache& texture_cache, TTBufferCache& buffer_cache,
-                 TQueryCache& query_cache)
-        : system{system}, rasterizer{rasterizer}, texture_cache{texture_cache},
-          buffer_cache{buffer_cache}, query_cache{query_cache} {}
+    explicit FenceManager(VideoCore::RasterizerInterface& rasterizer_, Tegra::GPU& gpu_,
+                          TTextureCache& texture_cache_, TTBufferCache& buffer_cache_,
+                          TQueryCache& query_cache_)
+        : rasterizer{rasterizer_}, gpu{gpu_}, gpu_memory{gpu.MemoryManager()},
+          texture_cache{texture_cache_}, buffer_cache{buffer_cache_}, query_cache{query_cache_} {}
 
-    virtual ~FenceManager() {}
+    virtual ~FenceManager() = default;
 
     /// Creates a Sync Point Fence Interface, does not create a backend fence if 'is_stubbed' is
     /// true
@@ -113,16 +111,15 @@ protected:
     /// Waits until a fence has been signalled by the host GPU.
     virtual void WaitFence(TFence& fence) = 0;
 
-    Core::System& system;
     VideoCore::RasterizerInterface& rasterizer;
+    Tegra::GPU& gpu;
+    Tegra::MemoryManager& gpu_memory;
     TTextureCache& texture_cache;
     TTBufferCache& buffer_cache;
     TQueryCache& query_cache;
 
 private:
     void TryReleasePendingFences() {
-        auto& gpu{system.GPU()};
-        auto& memory_manager{gpu.MemoryManager()};
         while (!fences.empty()) {
             TFence& current_fence = fences.front();
             if (ShouldWait() && !IsFenceSignaled(current_fence)) {
@@ -130,8 +127,8 @@ private:
             }
             PopAsyncFlushes();
             if (current_fence->IsSemaphore()) {
-                memory_manager.template Write<u32>(current_fence->GetAddress(),
-                                                   current_fence->GetPayload());
+                gpu_memory.template Write<u32>(current_fence->GetAddress(),
+                                               current_fence->GetPayload());
             } else {
                 gpu.IncrementSyncPoint(current_fence->GetPayload());
             }

src/video_core/gpu.cpp

@@ -28,8 +28,8 @@ namespace Tegra {
 MICROPROFILE_DEFINE(GPU_wait, "GPU", "Wait for the GPU", MP_RGB(128, 128, 192));
 
 GPU::GPU(Core::System& system_, bool is_async_)
-    : system{system_}, dma_pusher{std::make_unique<Tegra::DmaPusher>(system, *this)},
-      memory_manager{std::make_unique<Tegra::MemoryManager>(system)},
+    : system{system_}, memory_manager{std::make_unique<Tegra::MemoryManager>(system)},
+      dma_pusher{std::make_unique<Tegra::DmaPusher>(system, *this)},
       maxwell_3d{std::make_unique<Engines::Maxwell3D>(system, *memory_manager)},
       fermi_2d{std::make_unique<Engines::Fermi2D>()},
       kepler_compute{std::make_unique<Engines::KeplerCompute>(system, *memory_manager)},

src/video_core/gpu.h

@@ -347,12 +347,11 @@ private:
 
 protected:
     Core::System& system;
+    std::unique_ptr<Tegra::MemoryManager> memory_manager;
     std::unique_ptr<Tegra::DmaPusher> dma_pusher;
     std::unique_ptr<VideoCore::RendererBase> renderer;
 
 private:
-    std::unique_ptr<Tegra::MemoryManager> memory_manager;
-
     /// Mapping of command subchannels to their bound engine ids
     std::array<EngineID, 8> bound_engines = {};
     /// 3D engine

src/video_core/macro/macro_jit_x64.cpp

@@ -14,11 +14,11 @@ MICROPROFILE_DEFINE(MacroJitCompile, "GPU", "Compile macro JIT", MP_RGB(173, 255
 MICROPROFILE_DEFINE(MacroJitExecute, "GPU", "Execute macro JIT", MP_RGB(255, 255, 0));
 
 namespace Tegra {
-static const Xbyak::Reg64 STATE = Xbyak::util::rbx;
-static const Xbyak::Reg32 RESULT = Xbyak::util::ebp;
-static const Xbyak::Reg64 PARAMETERS = Xbyak::util::r12;
-static const Xbyak::Reg32 METHOD_ADDRESS = Xbyak::util::r14d;
-static const Xbyak::Reg64 BRANCH_HOLDER = Xbyak::util::r15;
+constexpr Xbyak::Reg64 STATE = Xbyak::util::rbx;
+constexpr Xbyak::Reg32 RESULT = Xbyak::util::ebp;
+constexpr Xbyak::Reg64 PARAMETERS = Xbyak::util::r12;
+constexpr Xbyak::Reg32 METHOD_ADDRESS = Xbyak::util::r14d;
+constexpr Xbyak::Reg64 BRANCH_HOLDER = Xbyak::util::r15;
 
 static const std::bitset<32> PERSISTENT_REGISTERS = Common::X64::BuildRegSet({
     STATE,

src/video_core/query_cache.h

@@ -95,10 +95,12 @@ template <class QueryCache, class CachedQuery, class CounterStream, class HostCo
           class QueryPool>
 class QueryCacheBase {
 public:
-    explicit QueryCacheBase(Core::System& system, VideoCore::RasterizerInterface& rasterizer)
-        : system{system}, rasterizer{rasterizer}, streams{{CounterStream{
-                                                      static_cast<QueryCache&>(*this),
-                                                      VideoCore::QueryType::SamplesPassed}}} {}
+    explicit QueryCacheBase(VideoCore::RasterizerInterface& rasterizer_,
+                            Tegra::Engines::Maxwell3D& maxwell3d_,
+                            Tegra::MemoryManager& gpu_memory_)
+        : rasterizer{rasterizer_}, maxwell3d{maxwell3d_},
+          gpu_memory{gpu_memory_}, streams{{CounterStream{static_cast<QueryCache&>(*this),
+                                                          VideoCore::QueryType::SamplesPassed}}} {}
 
     void InvalidateRegion(VAddr addr, std::size_t size) {
         std::unique_lock lock{mutex};
@@ -118,29 +120,27 @@ public:
      */
     void Query(GPUVAddr gpu_addr, VideoCore::QueryType type, std::optional<u64> timestamp) {
         std::unique_lock lock{mutex};
-        auto& memory_manager = system.GPU().MemoryManager();
-        const std::optional<VAddr> cpu_addr_opt = memory_manager.GpuToCpuAddress(gpu_addr);
-        ASSERT(cpu_addr_opt);
-        VAddr cpu_addr = *cpu_addr_opt;
+        const std::optional<VAddr> cpu_addr = gpu_memory.GpuToCpuAddress(gpu_addr);
+        ASSERT(cpu_addr);
 
-        CachedQuery* query = TryGet(cpu_addr);
+        CachedQuery* query = TryGet(*cpu_addr);
         if (!query) {
-            ASSERT_OR_EXECUTE(cpu_addr_opt, return;);
-            const auto host_ptr = memory_manager.GetPointer(gpu_addr);
+            ASSERT_OR_EXECUTE(cpu_addr, return;);
+            u8* const host_ptr = gpu_memory.GetPointer(gpu_addr);
 
-            query = Register(type, cpu_addr, host_ptr, timestamp.has_value());
+            query = Register(type, *cpu_addr, host_ptr, timestamp.has_value());
         }
 
         query->BindCounter(Stream(type).Current(), timestamp);
         if (Settings::values.use_asynchronous_gpu_emulation.GetValue()) {
-            AsyncFlushQuery(cpu_addr);
+            AsyncFlushQuery(*cpu_addr);
         }
     }
 
     /// Updates counters from GPU state. Expected to be called once per draw, clear or dispatch.
     void UpdateCounters() {
         std::unique_lock lock{mutex};
-        const auto& regs = system.GPU().Maxwell3D().regs;
+        const auto& regs = maxwell3d.regs;
         Stream(VideoCore::QueryType::SamplesPassed).Update(regs.samplecnt_enable);
     }
 
@@ -270,8 +270,9 @@ private:
     static constexpr std::uintptr_t PAGE_SIZE = 4096;
     static constexpr unsigned PAGE_BITS = 12;
 
-    Core::System& system;
     VideoCore::RasterizerInterface& rasterizer;
+    Tegra::Engines::Maxwell3D& maxwell3d;
+    Tegra::MemoryManager& gpu_memory;
 
     std::recursive_mutex mutex;
 

src/video_core/rasterizer_interface.h

@@ -106,11 +106,8 @@ public:
     virtual void UpdatePagesCachedCount(VAddr addr, u64 size, int delta) {}
 
     /// Initialize disk cached resources for the game being emulated
-    virtual void LoadDiskResources(const std::atomic_bool& stop_loading = false,
-                                   const DiskResourceLoadCallback& callback = {}) {}
-
-    /// Initializes renderer dirty flags
-    virtual void SetupDirtyFlags() {}
+    virtual void LoadDiskResources(u64 title_id, const std::atomic_bool& stop_loading,
+                                   const DiskResourceLoadCallback& callback) {}
 
     /// Grant access to the Guest Driver Profile for recording/obtaining info on the guest driver.
     GuestDriverProfile& AccessGuestDriverProfile() {

src/video_core/renderer_opengl/gl_buffer_cache.cpp

@@ -59,9 +59,10 @@ void Buffer::CopyFrom(const Buffer& src, std::size_t src_offset, std::size_t dst
                              static_cast<GLintptr>(dst_offset), static_cast<GLsizeiptr>(size));
 }
 
-OGLBufferCache::OGLBufferCache(RasterizerOpenGL& rasterizer, Core::System& system,
+OGLBufferCache::OGLBufferCache(VideoCore::RasterizerInterface& rasterizer,
+                               Tegra::MemoryManager& gpu_memory, Core::Memory::Memory& cpu_memory,
                                const Device& device_, std::size_t stream_size)
-    : GenericBufferCache{rasterizer, system,
+    : GenericBufferCache{rasterizer, gpu_memory, cpu_memory,
                          std::make_unique<OGLStreamBuffer>(device_, stream_size, true)},
       device{device_} {
     if (!device.HasFastBufferSubData()) {

src/video_core/renderer_opengl/gl_buffer_cache.h

@@ -52,7 +52,8 @@ private:
 using GenericBufferCache = VideoCommon::BufferCache<Buffer, GLuint, OGLStreamBuffer>;
 class OGLBufferCache final : public GenericBufferCache {
 public:
-    explicit OGLBufferCache(RasterizerOpenGL& rasterizer, Core::System& system,
+    explicit OGLBufferCache(VideoCore::RasterizerInterface& rasterizer,
+                            Tegra::MemoryManager& gpu_memory, Core::Memory::Memory& cpu_memory,
                             const Device& device, std::size_t stream_size);
     ~OGLBufferCache();
 

src/video_core/renderer_opengl/gl_fence_manager.cpp

@@ -45,11 +45,10 @@ void GLInnerFence::Wait() {
     glClientWaitSync(sync_object.handle, 0, GL_TIMEOUT_IGNORED);
 }
 
-FenceManagerOpenGL::FenceManagerOpenGL(Core::System& system,
-                                       VideoCore::RasterizerInterface& rasterizer,
+FenceManagerOpenGL::FenceManagerOpenGL(VideoCore::RasterizerInterface& rasterizer, Tegra::GPU& gpu,
                                        TextureCacheOpenGL& texture_cache,
                                        OGLBufferCache& buffer_cache, QueryCache& query_cache)
-    : GenericFenceManager(system, rasterizer, texture_cache, buffer_cache, query_cache) {}
+    : GenericFenceManager{rasterizer, gpu, texture_cache, buffer_cache, query_cache} {}
 
 Fence FenceManagerOpenGL::CreateFence(u32 value, bool is_stubbed) {
     return std::make_shared<GLInnerFence>(value, is_stubbed);

src/video_core/renderer_opengl/gl_fence_manager.h

@@ -37,9 +37,9 @@ using GenericFenceManager =
 
 class FenceManagerOpenGL final : public GenericFenceManager {
 public:
-    FenceManagerOpenGL(Core::System& system, VideoCore::RasterizerInterface& rasterizer,
-                       TextureCacheOpenGL& texture_cache, OGLBufferCache& buffer_cache,
-                       QueryCache& query_cache);
+    explicit FenceManagerOpenGL(VideoCore::RasterizerInterface& rasterizer, Tegra::GPU& gpu,
+                                TextureCacheOpenGL& texture_cache, OGLBufferCache& buffer_cache,
+                                QueryCache& query_cache);
 
 protected:
     Fence CreateFence(u32 value, bool is_stubbed) override;

src/video_core/renderer_opengl/gl_query_cache.cpp

@@ -30,12 +30,13 @@ constexpr GLenum GetTarget(VideoCore::QueryType type) {
 
 } // Anonymous namespace
 
-QueryCache::QueryCache(Core::System& system, RasterizerOpenGL& gl_rasterizer)
+QueryCache::QueryCache(RasterizerOpenGL& rasterizer, Tegra::Engines::Maxwell3D& maxwell3d,
+                       Tegra::MemoryManager& gpu_memory)
     : VideoCommon::QueryCacheBase<
           QueryCache, CachedQuery, CounterStream, HostCounter,
-          std::vector<OGLQuery>>{system,
-                                 static_cast<VideoCore::RasterizerInterface&>(gl_rasterizer)},
-      gl_rasterizer{gl_rasterizer} {}
+          std::vector<OGLQuery>>{static_cast<VideoCore::RasterizerInterface&>(rasterizer),
+                                 maxwell3d, gpu_memory},
+      gl_rasterizer{rasterizer} {}
 
 QueryCache::~QueryCache() = default;
 

src/video_core/renderer_opengl/gl_query_cache.h

@@ -29,7 +29,8 @@ using CounterStream = VideoCommon::CounterStreamBase<QueryCache, HostCounter>;
 class QueryCache final : public VideoCommon::QueryCacheBase<QueryCache, CachedQuery, CounterStream,
                                                             HostCounter, std::vector<OGLQuery>> {
 public:
-    explicit QueryCache(Core::System& system, RasterizerOpenGL& rasterizer);
+    explicit QueryCache(RasterizerOpenGL& rasterizer, Tegra::Engines::Maxwell3D& maxwell3d,
+                        Tegra::MemoryManager& gpu_memory);
     ~QueryCache();
 
     OGLQuery AllocateQuery(VideoCore::QueryType type);

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -153,16 +153,19 @@ void UpdateBindlessPointers(GLenum target, GLuint64EXT* pointers, std::size_t nu
 
 } // Anonymous namespace
 
-RasterizerOpenGL::RasterizerOpenGL(Core::System& system, Core::Frontend::EmuWindow& emu_window,
-                                   const Device& device, ScreenInfo& info,
-                                   ProgramManager& program_manager, StateTracker& state_tracker)
-    : RasterizerAccelerated{system.Memory()}, device{device}, texture_cache{system, *this, device,
-                                                                            state_tracker},
-      shader_cache{*this, system, emu_window, device}, query_cache{system, *this},
-      buffer_cache{*this, system, device, STREAM_BUFFER_SIZE},
-      fence_manager{system, *this, texture_cache, buffer_cache, query_cache}, system{system},
-      screen_info{info}, program_manager{program_manager}, state_tracker{state_tracker},
-      async_shaders{emu_window} {
+RasterizerOpenGL::RasterizerOpenGL(Core::Frontend::EmuWindow& emu_window, Tegra::GPU& gpu_,
+                                   Core::Memory::Memory& cpu_memory, const Device& device_,
+                                   ScreenInfo& screen_info_, ProgramManager& program_manager_,
+                                   StateTracker& state_tracker_)
+    : RasterizerAccelerated{cpu_memory}, gpu(gpu_), maxwell3d(gpu.Maxwell3D()),
+      kepler_compute(gpu.KeplerCompute()), gpu_memory(gpu.MemoryManager()), device(device_),
+      screen_info(screen_info_), program_manager(program_manager_), state_tracker(state_tracker_),
+      texture_cache(*this, maxwell3d, gpu_memory, device, state_tracker),
+      shader_cache(*this, emu_window, gpu, maxwell3d, kepler_compute, gpu_memory, device),
+      query_cache(*this, maxwell3d, gpu_memory),
+      buffer_cache(*this, gpu_memory, cpu_memory, device, STREAM_BUFFER_SIZE),
+      fence_manager(*this, gpu, texture_cache, buffer_cache, query_cache),
+      async_shaders(emu_window) {
     CheckExtensions();
 
     unified_uniform_buffer.Create();
@@ -196,8 +199,7 @@ void RasterizerOpenGL::CheckExtensions() {
 }
 
 void RasterizerOpenGL::SetupVertexFormat() {
-    auto& gpu = system.GPU().Maxwell3D();
-    auto& flags = gpu.dirty.flags;
+    auto& flags = maxwell3d.dirty.flags;
     if (!flags[Dirty::VertexFormats]) {
         return;
     }
@@ -217,7 +219,7 @@ void RasterizerOpenGL::SetupVertexFormat() {
         }
         flags[Dirty::VertexFormat0 + index] = false;
 
-        const auto attrib = gpu.regs.vertex_attrib_format[index];
+        const auto attrib = maxwell3d.regs.vertex_attrib_format[index];
         const auto gl_index = static_cast<GLuint>(index);
 
         // Disable constant attributes.
@@ -241,8 +243,7 @@ void RasterizerOpenGL::SetupVertexFormat() {
 }
 
 void RasterizerOpenGL::SetupVertexBuffer() {
-    auto& gpu = system.GPU().Maxwell3D();
-    auto& flags = gpu.dirty.flags;
+    auto& flags = maxwell3d.dirty.flags;
     if (!flags[Dirty::VertexBuffers]) {
         return;
     }
@@ -253,7 +254,7 @@ void RasterizerOpenGL::SetupVertexBuffer() {
     const bool use_unified_memory = device.HasVertexBufferUnifiedMemory();
 
     // Upload all guest vertex arrays sequentially to our buffer
-    const auto& regs = gpu.regs;
+    const auto& regs = maxwell3d.regs;
     for (std::size_t index = 0; index < NUM_SUPPORTED_VERTEX_BINDINGS; ++index) {
         if (!flags[Dirty::VertexBuffer0 + index]) {
             continue;
@@ -290,14 +291,13 @@ void RasterizerOpenGL::SetupVertexBuffer() {
 }
 
 void RasterizerOpenGL::SetupVertexInstances() {
-    auto& gpu = system.GPU().Maxwell3D();
-    auto& flags = gpu.dirty.flags;
+    auto& flags = maxwell3d.dirty.flags;
     if (!flags[Dirty::VertexInstances]) {
         return;
     }
     flags[Dirty::VertexInstances] = false;
 
-    const auto& regs = gpu.regs;
+    const auto& regs = maxwell3d.regs;
     for (std::size_t index = 0; index < NUM_SUPPORTED_VERTEX_ATTRIBUTES; ++index) {
         if (!flags[Dirty::VertexInstance0 + index]) {
             continue;
@@ -313,7 +313,7 @@ void RasterizerOpenGL::SetupVertexInstances() {
 
 GLintptr RasterizerOpenGL::SetupIndexBuffer() {
     MICROPROFILE_SCOPE(OpenGL_Index);
-    const auto& regs = system.GPU().Maxwell3D().regs;
+    const auto& regs = maxwell3d.regs;
     const std::size_t size = CalculateIndexBufferSize();
     const auto info = buffer_cache.UploadMemory(regs.index_array.IndexStart(), size);
     glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, info.handle);
@@ -322,15 +322,14 @@ GLintptr RasterizerOpenGL::SetupIndexBuffer() {
 
 void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
     MICROPROFILE_SCOPE(OpenGL_Shader);
-    auto& gpu = system.GPU().Maxwell3D();
     u32 clip_distances = 0;
 
     for (std::size_t index = 0; index < Maxwell::MaxShaderProgram; ++index) {
-        const auto& shader_config = gpu.regs.shader_config[index];
+        const auto& shader_config = maxwell3d.regs.shader_config[index];
         const auto program{static_cast<Maxwell::ShaderProgram>(index)};
 
         // Skip stages that are not enabled
-        if (!gpu.regs.IsShaderConfigEnabled(index)) {
+        if (!maxwell3d.regs.IsShaderConfigEnabled(index)) {
             switch (program) {
             case Maxwell::ShaderProgram::Geometry:
                 program_manager.UseGeometryShader(0);
@@ -391,11 +390,11 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
     }
 
     SyncClipEnabled(clip_distances);
-    gpu.dirty.flags[Dirty::Shaders] = false;
+    maxwell3d.dirty.flags[Dirty::Shaders] = false;
 }
 
 std::size_t RasterizerOpenGL::CalculateVertexArraysSize() const {
-    const auto& regs = system.GPU().Maxwell3D().regs;
+    const auto& regs = maxwell3d.regs;
 
     std::size_t size = 0;
     for (u32 index = 0; index < Maxwell::NumVertexArrays; ++index) {
@@ -413,34 +412,27 @@ std::size_t RasterizerOpenGL::CalculateVertexArraysSize() const {
 }
 
 std::size_t RasterizerOpenGL::CalculateIndexBufferSize() const {
-    const auto& regs = system.GPU().Maxwell3D().regs;
-
-    return static_cast<std::size_t>(regs.index_array.count) *
-           static_cast<std::size_t>(regs.index_array.FormatSizeInBytes());
+    return static_cast<std::size_t>(maxwell3d.regs.index_array.count) *
+           static_cast<std::size_t>(maxwell3d.regs.index_array.FormatSizeInBytes());
 }
 
-void RasterizerOpenGL::LoadDiskResources(const std::atomic_bool& stop_loading,
+void RasterizerOpenGL::LoadDiskResources(u64 title_id, const std::atomic_bool& stop_loading,
                                          const VideoCore::DiskResourceLoadCallback& callback) {
-    shader_cache.LoadDiskCache(stop_loading, callback);
-}
-
-void RasterizerOpenGL::SetupDirtyFlags() {
-    state_tracker.Initialize();
+    shader_cache.LoadDiskCache(title_id, stop_loading, callback);
 }
 
 void RasterizerOpenGL::ConfigureFramebuffers() {
     MICROPROFILE_SCOPE(OpenGL_Framebuffer);
-    auto& gpu = system.GPU().Maxwell3D();
-    if (!gpu.dirty.flags[VideoCommon::Dirty::RenderTargets]) {
+    if (!maxwell3d.dirty.flags[VideoCommon::Dirty::RenderTargets]) {
         return;
     }
-    gpu.dirty.flags[VideoCommon::Dirty::RenderTargets] = false;
+    maxwell3d.dirty.flags[VideoCommon::Dirty::RenderTargets] = false;
 
     texture_cache.GuardRenderTargets(true);
 
     View depth_surface = texture_cache.GetDepthBufferSurface(true);
 
-    const auto& regs = gpu.regs;
+    const auto& regs = maxwell3d.regs;
     UNIMPLEMENTED_IF(regs.rt_separate_frag_data == 0);
 
     // Bind the framebuffer surfaces
@@ -472,8 +464,7 @@ void RasterizerOpenGL::ConfigureFramebuffers() {
 }
 
 void RasterizerOpenGL::ConfigureClearFramebuffer(bool using_color, bool using_depth_stencil) {
-    auto& gpu = system.GPU().Maxwell3D();
-    const auto& regs = gpu.regs;
+    const auto& regs = maxwell3d.regs;
 
     texture_cache.GuardRenderTargets(true);
     View color_surface;
@@ -523,12 +514,11 @@ void RasterizerOpenGL::ConfigureClearFramebuffer(bool using_color, bool using_de
 }
 
 void RasterizerOpenGL::Clear() {
-    const auto& gpu = system.GPU().Maxwell3D();
-    if (!gpu.ShouldExecute()) {
+    if (!maxwell3d.ShouldExecute()) {
         return;
     }
 
-    const auto& regs = gpu.regs;
+    const auto& regs = maxwell3d.regs;
     bool use_color{};
     bool use_depth{};
     bool use_stencil{};
@@ -593,7 +583,6 @@ void RasterizerOpenGL::Clear() {
 
 void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) {
     MICROPROFILE_SCOPE(OpenGL_Drawing);
-    auto& gpu = system.GPU().Maxwell3D();
 
     query_cache.UpdateCounters();
 
@@ -641,7 +630,7 @@ void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) {
 
     if (invalidated) {
         // When the stream buffer has been invalidated, we have to consider vertex buffers as dirty
-        auto& dirty = gpu.dirty.flags;
+        auto& dirty = maxwell3d.dirty.flags;
         dirty[Dirty::VertexBuffers] = true;
         for (int index = Dirty::VertexBuffer0; index <= Dirty::VertexBuffer31; ++index) {
             dirty[index] = true;
@@ -662,7 +651,7 @@ void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) {
     // Setup emulation uniform buffer.
     if (!device.UseAssemblyShaders()) {
         MaxwellUniformData ubo;
-        ubo.SetFromRegs(gpu);
+        ubo.SetFromRegs(maxwell3d);
         const auto info =
             buffer_cache.UploadHostMemory(&ubo, sizeof(ubo), device.GetUniformBufferAlignment());
         glBindBufferRange(GL_UNIFORM_BUFFER, EmulationUniformBlockBinding, info.handle, info.offset,
@@ -671,7 +660,7 @@ void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) {
 
     // Setup shaders and their used resources.
     texture_cache.GuardSamplers(true);
-    const GLenum primitive_mode = MaxwellToGL::PrimitiveTopology(gpu.regs.draw.topology);
+    const GLenum primitive_mode = MaxwellToGL::PrimitiveTopology(maxwell3d.regs.draw.topology);
     SetupShaders(primitive_mode);
     texture_cache.GuardSamplers(false);
 
@@ -688,14 +677,14 @@ void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) {
 
     BeginTransformFeedback(primitive_mode);
 
-    const GLuint base_instance = static_cast<GLuint>(gpu.regs.vb_base_instance);
+    const GLuint base_instance = static_cast<GLuint>(maxwell3d.regs.vb_base_instance);
     const GLsizei num_instances =
-        static_cast<GLsizei>(is_instanced ? gpu.mme_draw.instance_count : 1);
+        static_cast<GLsizei>(is_instanced ? maxwell3d.mme_draw.instance_count : 1);
     if (is_indexed) {
-        const GLint base_vertex = static_cast<GLint>(gpu.regs.vb_element_base);
-        const GLsizei num_vertices = static_cast<GLsizei>(gpu.regs.index_array.count);
+        const GLint base_vertex = static_cast<GLint>(maxwell3d.regs.vb_element_base);
+        const GLsizei num_vertices = static_cast<GLsizei>(maxwell3d.regs.index_array.count);
         const GLvoid* offset = reinterpret_cast<const GLvoid*>(index_buffer_offset);
-        const GLenum format = MaxwellToGL::IndexFormat(gpu.regs.index_array.format);
+        const GLenum format = MaxwellToGL::IndexFormat(maxwell3d.regs.index_array.format);
         if (num_instances == 1 && base_instance == 0 && base_vertex == 0) {
             glDrawElements(primitive_mode, num_vertices, format, offset);
         } else if (num_instances == 1 && base_instance == 0) {
@@ -714,8 +703,8 @@ void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) {
                                                           base_instance);
         }
     } else {
-        const GLint base_vertex = static_cast<GLint>(gpu.regs.vertex_buffer.first);
-        const GLsizei num_vertices = static_cast<GLsizei>(gpu.regs.vertex_buffer.count);
+        const GLint base_vertex = static_cast<GLint>(maxwell3d.regs.vertex_buffer.first);
+        const GLsizei num_vertices = static_cast<GLsizei>(maxwell3d.regs.vertex_buffer.count);
         if (num_instances == 1 && base_instance == 0) {
             glDrawArrays(primitive_mode, base_vertex, num_vertices);
         } else if (base_instance == 0) {
@@ -730,7 +719,7 @@ void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) {
 
     ++num_queued_commands;
 
-    system.GPU().TickWork();
+    gpu.TickWork();
 }
 
 void RasterizerOpenGL::DispatchCompute(GPUVAddr code_addr) {
@@ -753,7 +742,8 @@ void RasterizerOpenGL::DispatchCompute(GPUVAddr code_addr) {
 
     buffer_cache.Unmap();
 
-    const auto& launch_desc = system.GPU().KeplerCompute().launch_description;
+    const auto& launch_desc = kepler_compute.launch_description;
+    program_manager.BindCompute(kernel->GetHandle());
     glDispatchCompute(launch_desc.grid_dim_x, launch_desc.grid_dim_y, launch_desc.grid_dim_z);
     ++num_queued_commands;
 }
@@ -815,17 +805,14 @@ void RasterizerOpenGL::SyncGuestHost() {
 }
 
 void RasterizerOpenGL::SignalSemaphore(GPUVAddr addr, u32 value) {
-    auto& gpu{system.GPU()};
     if (!gpu.IsAsync()) {
-        auto& memory_manager{gpu.MemoryManager()};
-        memory_manager.Write<u32>(addr, value);
+        gpu_memory.Write<u32>(addr, value);
         return;
     }
     fence_manager.SignalSemaphore(addr, value);
 }
 
 void RasterizerOpenGL::SignalSyncPoint(u32 value) {
-    auto& gpu{system.GPU()};
     if (!gpu.IsAsync()) {
         gpu.IncrementSyncPoint(value);
         return;
@@ -834,7 +821,6 @@ void RasterizerOpenGL::SignalSyncPoint(u32 value) {
 }
 
 void RasterizerOpenGL::ReleaseFences() {
-    auto& gpu{system.GPU()};
     if (!gpu.IsAsync()) {
         return;
     }
@@ -920,7 +906,7 @@ void RasterizerOpenGL::SetupDrawConstBuffers(std::size_t stage_index, Shader* sh
         GL_FRAGMENT_PROGRAM_PARAMETER_BUFFER_NV};
 
     MICROPROFILE_SCOPE(OpenGL_UBO);
-    const auto& stages = system.GPU().Maxwell3D().state.shader_stages;
+    const auto& stages = maxwell3d.state.shader_stages;
     const auto& shader_stage = stages[stage_index];
     const auto& entries = shader->GetEntries();
     const bool use_unified = entries.use_unified_uniforms;
@@ -945,7 +931,7 @@ void RasterizerOpenGL::SetupDrawConstBuffers(std::size_t stage_index, Shader* sh
 
 void RasterizerOpenGL::SetupComputeConstBuffers(Shader* kernel) {
     MICROPROFILE_SCOPE(OpenGL_UBO);
-    const auto& launch_desc = system.GPU().KeplerCompute().launch_description;
+    const auto& launch_desc = kepler_compute.launch_description;
     const auto& entries = kernel->GetEntries();
     const bool use_unified = entries.use_unified_uniforms;
 
@@ -1018,9 +1004,7 @@ void RasterizerOpenGL::SetupDrawGlobalMemory(std::size_t stage_index, Shader* sh
         GL_GEOMETRY_PROGRAM_NV, GL_FRAGMENT_PROGRAM_NV,
     };
 
-    auto& gpu{system.GPU()};
-    auto& memory_manager{gpu.MemoryManager()};
-    const auto& cbufs{gpu.Maxwell3D().state.shader_stages[stage_index]};
+    const auto& cbufs{maxwell3d.state.shader_stages[stage_index]};
     const auto& entries{shader->GetEntries().global_memory_entries};
 
     std::array<GLuint64EXT, 32> pointers;
@@ -1030,8 +1014,8 @@ void RasterizerOpenGL::SetupDrawGlobalMemory(std::size_t stage_index, Shader* sh
     u32 binding = assembly_shaders ? 0 : device.GetBaseBindings(stage_index).shader_storage_buffer;
     for (const auto& entry : entries) {
         const GPUVAddr addr{cbufs.const_buffers[entry.cbuf_index].address + entry.cbuf_offset};
-        const GPUVAddr gpu_addr{memory_manager.Read<u64>(addr)};
-        const u32 size{memory_manager.Read<u32>(addr + 8)};
+        const GPUVAddr gpu_addr{gpu_memory.Read<u64>(addr)};
+        const u32 size{gpu_memory.Read<u32>(addr + 8)};
         SetupGlobalMemory(binding, entry, gpu_addr, size, &pointers[binding]);
         ++binding;
     }
@@ -1041,9 +1025,7 @@ void RasterizerOpenGL::SetupDrawGlobalMemory(std::size_t stage_index, Shader* sh
 }
 
 void RasterizerOpenGL::SetupComputeGlobalMemory(Shader* kernel) {
-    auto& gpu{system.GPU()};
-    auto& memory_manager{gpu.MemoryManager()};
-    const auto& cbufs{gpu.KeplerCompute().launch_description.const_buffer_config};
+    const auto& cbufs{kepler_compute.launch_description.const_buffer_config};
     const auto& entries{kernel->GetEntries().global_memory_entries};
 
     std::array<GLuint64EXT, 32> pointers;
@@ -1052,8 +1034,8 @@ void RasterizerOpenGL::SetupComputeGlobalMemory(Shader* kernel) {
     u32 binding = 0;
     for (const auto& entry : entries) {
         const GPUVAddr addr{cbufs[entry.cbuf_index].Address() + entry.cbuf_offset};
-        const GPUVAddr gpu_addr{memory_manager.Read<u64>(addr)};
-        const u32 size{memory_manager.Read<u32>(addr + 8)};
+        const GPUVAddr gpu_addr{gpu_memory.Read<u64>(addr)};
+        const u32 size{gpu_memory.Read<u32>(addr + 8)};
         SetupGlobalMemory(binding, entry, gpu_addr, size, &pointers[binding]);
         ++binding;
     }
@@ -1077,7 +1059,6 @@ void RasterizerOpenGL::SetupGlobalMemory(u32 binding, const GlobalMemoryEntry& e
 
 void RasterizerOpenGL::SetupDrawTextures(std::size_t stage_index, Shader* shader) {
     MICROPROFILE_SCOPE(OpenGL_Texture);
-    const auto& maxwell3d = system.GPU().Maxwell3D();
     u32 binding = device.GetBaseBindings(stage_index).sampler;
     for (const auto& entry : shader->GetEntries().samplers) {
         const auto shader_type = static_cast<ShaderType>(stage_index);
@@ -1090,11 +1071,10 @@ void RasterizerOpenGL::SetupDrawTextures(std::size_t stage_index, Shader* shader
 
 void RasterizerOpenGL::SetupComputeTextures(Shader* kernel) {
     MICROPROFILE_SCOPE(OpenGL_Texture);
-    const auto& compute = system.GPU().KeplerCompute();
     u32 binding = 0;
     for (const auto& entry : kernel->GetEntries().samplers) {
         for (std::size_t i = 0; i < entry.size; ++i) {
-            const auto texture = GetTextureInfo(compute, entry, ShaderType::Compute, i);
+            const auto texture = GetTextureInfo(kepler_compute, entry, ShaderType::Compute, i);
             SetupTexture(binding++, texture, entry);
         }
     }
@@ -1118,20 +1098,18 @@ void RasterizerOpenGL::SetupTexture(u32 binding, const Tegra::Texture::FullTextu
 }
 
 void RasterizerOpenGL::SetupDrawImages(std::size_t stage_index, Shader* shader) {
-    const auto& maxwell3d = system.GPU().Maxwell3D();
     u32 binding = device.GetBaseBindings(stage_index).image;
     for (const auto& entry : shader->GetEntries().images) {
-        const auto shader_type = static_cast<Tegra::Engines::ShaderType>(stage_index);
+        const auto shader_type = static_cast<ShaderType>(stage_index);
         const auto tic = GetTextureInfo(maxwell3d, entry, shader_type).tic;
         SetupImage(binding++, tic, entry);
     }
 }
 
 void RasterizerOpenGL::SetupComputeImages(Shader* shader) {
-    const auto& compute = system.GPU().KeplerCompute();
     u32 binding = 0;
     for (const auto& entry : shader->GetEntries().images) {
-        const auto tic = GetTextureInfo(compute, entry, Tegra::Engines::ShaderType::Compute).tic;
+        const auto tic = GetTextureInfo(kepler_compute, entry, ShaderType::Compute).tic;
         SetupImage(binding++, tic, entry);
     }
 }
@@ -1151,9 +1129,8 @@ void RasterizerOpenGL::SetupImage(u32 binding, const Tegra::Texture::TICEntry& t
 }
 
 void RasterizerOpenGL::SyncViewport() {
-    auto& gpu = system.GPU().Maxwell3D();
-    auto& flags = gpu.dirty.flags;
-    const auto& regs = gpu.regs;
+    auto& flags = maxwell3d.dirty.flags;
+    const auto& regs = maxwell3d.regs;
 
     const bool dirty_viewport = flags[Dirty::Viewports];
     const bool dirty_clip_control = flags[Dirty::ClipControl];
@@ -1225,25 +1202,23 @@ void RasterizerOpenGL::SyncViewport() {
 }
 
 void RasterizerOpenGL::SyncDepthClamp() {
-    auto& gpu = system.GPU().Maxwell3D();
-    auto& flags = gpu.dirty.flags;
+    auto& flags = maxwell3d.dirty.flags;
     if (!flags[Dirty::DepthClampEnabled]) {
         return;
     }
     flags[Dirty::DepthClampEnabled] = false;
 
-    oglEnable(GL_DEPTH_CLAMP, gpu.regs.view_volume_clip_control.depth_clamp_disabled == 0);
+    oglEnable(GL_DEPTH_CLAMP, maxwell3d.regs.view_volume_clip_control.depth_clamp_disabled == 0);
 }
 
 void RasterizerOpenGL::SyncClipEnabled(u32 clip_mask) {
-    auto& gpu = system.GPU().Maxwell3D();
-    auto& flags = gpu.dirty.flags;
+    auto& flags = maxwell3d.dirty.flags;
     if (!flags[Dirty::ClipDistances] && !flags[Dirty::Shaders]) {
         return;
     }
     flags[Dirty::ClipDistances] = false;
 
-    clip_mask &= gpu.regs.clip_distance_enabled;
+    clip_mask &= maxwell3d.regs.clip_distance_enabled;
     if (clip_mask == last_clip_distance_mask) {
         return;
     }
@@ -1259,9 +1234,8 @@ void RasterizerOpenGL::SyncClipCoef() {
 }
 
 void RasterizerOpenGL::SyncCullMode() {
-    auto& gpu = system.GPU().Maxwell3D();
-    auto& flags = gpu.dirty.flags;
-    const auto& regs = gpu.regs;
+    auto& flags = maxwell3d.dirty.flags;
+    const auto& regs = maxwell3d.regs;
 
     if (flags[Dirty::CullTest]) {
         flags[Dirty::CullTest] = false;
@@ -1276,26 +1250,24 @@ void RasterizerOpenGL::SyncCullMode() {
 }
 
 void RasterizerOpenGL::SyncPrimitiveRestart() {
-    auto& gpu = system.GPU().Maxwell3D();
-    auto& flags = gpu.dirty.flags;
+    auto& flags = maxwell3d.dirty.flags;
     if (!flags[Dirty::PrimitiveRestart]) {
         return;
     }
     flags[Dirty::PrimitiveRestart] = false;
 
-    if (gpu.regs.primitive_restart.enabled) {
+    if (maxwell3d.regs.primitive_restart.enabled) {
         glEnable(GL_PRIMITIVE_RESTART);
-        glPrimitiveRestartIndex(gpu.regs.primitive_restart.index);
+        glPrimitiveRestartIndex(maxwell3d.regs.primitive_restart.index);
     } else {
         glDisable(GL_PRIMITIVE_RESTART);
     }
 }
 
 void RasterizerOpenGL::SyncDepthTestState() {
-    auto& gpu = system.GPU().Maxwell3D();
-    auto& flags = gpu.dirty.flags;
+    auto& flags = maxwell3d.dirty.flags;
+    const auto& regs = maxwell3d.regs;
 
-    const auto& regs = gpu.regs;
     if (flags[Dirty::DepthMask]) {
         flags[Dirty::DepthMask] = false;
         glDepthMask(regs.depth_write_enabled ? GL_TRUE : GL_FALSE);
@@ -1313,14 +1285,13 @@ void RasterizerOpenGL::SyncDepthTestState() {
 }
 
 void RasterizerOpenGL::SyncStencilTestState() {
-    auto& gpu = system.GPU().Maxwell3D();
-    auto& flags = gpu.dirty.flags;
+    auto& flags = maxwell3d.dirty.flags;
     if (!flags[Dirty::StencilTest]) {
         return;
     }
     flags[Dirty::StencilTest] = false;
 
-    const auto& regs = gpu.regs;
+    const auto& regs = maxwell3d.regs;
     oglEnable(GL_STENCIL_TEST, regs.stencil_enable);
 
     glStencilFuncSeparate(GL_FRONT, MaxwellToGL::ComparisonOp(regs.stencil_front_func_func),
@@ -1345,25 +1316,24 @@ void RasterizerOpenGL::SyncStencilTestState() {
 }
 
 void RasterizerOpenGL::SyncRasterizeEnable() {
-    auto& gpu = system.GPU().Maxwell3D();
-    auto& flags = gpu.dirty.flags;
+    auto& flags = maxwell3d.dirty.flags;
     if (!flags[Dirty::RasterizeEnable]) {
         return;
     }
     flags[Dirty::RasterizeEnable] = false;
 
-    oglEnable(GL_RASTERIZER_DISCARD, gpu.regs.rasterize_enable == 0);
+    oglEnable(GL_RASTERIZER_DISCARD, maxwell3d.regs.rasterize_enable == 0);
 }
 
 void RasterizerOpenGL::SyncPolygonModes() {
-    auto& gpu = system.GPU().Maxwell3D();
-    auto& flags = gpu.dirty.flags;
+    auto& flags = maxwell3d.dirty.flags;
     if (!flags[Dirty::PolygonModes]) {
         return;
     }
     flags[Dirty::PolygonModes] = false;
 
-    if (gpu.regs.fill_rectangle) {
+    const auto& regs = maxwell3d.regs;
+    if (regs.fill_rectangle) {
         if (!GLAD_GL_NV_fill_rectangle) {
             LOG_ERROR(Render_OpenGL, "GL_NV_fill_rectangle used and not supported");
             glPolygonMode(GL_FRONT_AND_BACK, GL_FILL);
@@ -1376,27 +1346,26 @@ void RasterizerOpenGL::SyncPolygonModes() {
         return;
     }
 
-    if (gpu.regs.polygon_mode_front == gpu.regs.polygon_mode_back) {
+    if (regs.polygon_mode_front == regs.polygon_mode_back) {
         flags[Dirty::PolygonModeFront] = false;
         flags[Dirty::PolygonModeBack] = false;
-        glPolygonMode(GL_FRONT_AND_BACK, MaxwellToGL::PolygonMode(gpu.regs.polygon_mode_front));
+        glPolygonMode(GL_FRONT_AND_BACK, MaxwellToGL::PolygonMode(regs.polygon_mode_front));
         return;
     }
 
     if (flags[Dirty::PolygonModeFront]) {
         flags[Dirty::PolygonModeFront] = false;
-        glPolygonMode(GL_FRONT, MaxwellToGL::PolygonMode(gpu.regs.polygon_mode_front));
+        glPolygonMode(GL_FRONT, MaxwellToGL::PolygonMode(regs.polygon_mode_front));
     }
 
     if (flags[Dirty::PolygonModeBack]) {
         flags[Dirty::PolygonModeBack] = false;
-        glPolygonMode(GL_BACK, MaxwellToGL::PolygonMode(gpu.regs.polygon_mode_back));
+        glPolygonMode(GL_BACK, MaxwellToGL::PolygonMode(regs.polygon_mode_back));
     }
 }
 
 void RasterizerOpenGL::SyncColorMask() {
-    auto& gpu = system.GPU().Maxwell3D();
-    auto& flags = gpu.dirty.flags;
+    auto& flags = maxwell3d.dirty.flags;
     if (!flags[Dirty::ColorMasks]) {
         return;
     }
@@ -1405,7 +1374,7 @@ void RasterizerOpenGL::SyncColorMask() {
     const bool force = flags[Dirty::ColorMaskCommon];
     flags[Dirty::ColorMaskCommon] = false;
 
-    const auto& regs = gpu.regs;
+    const auto& regs = maxwell3d.regs;
     if (regs.color_mask_common) {
         if (!force && !flags[Dirty::ColorMask0]) {
             return;
@@ -1430,33 +1399,30 @@ void RasterizerOpenGL::SyncColorMask() {
 }
 
 void RasterizerOpenGL::SyncMultiSampleState() {
-    auto& gpu = system.GPU().Maxwell3D();
-    auto& flags = gpu.dirty.flags;
+    auto& flags = maxwell3d.dirty.flags;
     if (!flags[Dirty::MultisampleControl]) {
         return;
     }
     flags[Dirty::MultisampleControl] = false;
 
-    const auto& regs = system.GPU().Maxwell3D().regs;
+    const auto& regs = maxwell3d.regs;
     oglEnable(GL_SAMPLE_ALPHA_TO_COVERAGE, regs.multisample_control.alpha_to_coverage);
     oglEnable(GL_SAMPLE_ALPHA_TO_ONE, regs.multisample_control.alpha_to_one);
 }
 
 void RasterizerOpenGL::SyncFragmentColorClampState() {
-    auto& gpu = system.GPU().Maxwell3D();
-    auto& flags = gpu.dirty.flags;
+    auto& flags = maxwell3d.dirty.flags;
     if (!flags[Dirty::FragmentClampColor]) {
         return;
     }
     flags[Dirty::FragmentClampColor] = false;
 
-    glClampColor(GL_CLAMP_FRAGMENT_COLOR, gpu.regs.frag_color_clamp ? GL_TRUE : GL_FALSE);
+    glClampColor(GL_CLAMP_FRAGMENT_COLOR, maxwell3d.regs.frag_color_clamp ? GL_TRUE : GL_FALSE);
 }
 
 void RasterizerOpenGL::SyncBlendState() {
-    auto& gpu = system.GPU().Maxwell3D();
-    auto& flags = gpu.dirty.flags;
-    const auto& regs = gpu.regs;
+    auto& flags = maxwell3d.dirty.flags;
+    const auto& regs = maxwell3d.regs;
 
     if (flags[Dirty::BlendColor]) {
         flags[Dirty::BlendColor] = false;
@@ -1513,14 +1479,13 @@ void RasterizerOpenGL::SyncBlendState() {
 }
 
 void RasterizerOpenGL::SyncLogicOpState() {
-    auto& gpu = system.GPU().Maxwell3D();
-    auto& flags = gpu.dirty.flags;
+    auto& flags = maxwell3d.dirty.flags;
     if (!flags[Dirty::LogicOp]) {
         return;
     }
     flags[Dirty::LogicOp] = false;
 
-    const auto& regs = gpu.regs;
+    const auto& regs = maxwell3d.regs;
     if (regs.logic_op.enable) {
         glEnable(GL_COLOR_LOGIC_OP);
         glLogicOp(MaxwellToGL::LogicOp(regs.logic_op.operation));
@@ -1530,14 +1495,13 @@ void RasterizerOpenGL::SyncLogicOpState() {
 }
 
 void RasterizerOpenGL::SyncScissorTest() {
-    auto& gpu = system.GPU().Maxwell3D();
-    auto& flags = gpu.dirty.flags;
+    auto& flags = maxwell3d.dirty.flags;
     if (!flags[Dirty::Scissors]) {
         return;
     }
     flags[Dirty::Scissors] = false;
 
-    const auto& regs = gpu.regs;
+    const auto& regs = maxwell3d.regs;
     for (std::size_t index = 0; index < Maxwell::NumViewports; ++index) {
         if (!flags[Dirty::Scissor0 + index]) {
             continue;
@@ -1556,16 +1520,15 @@ void RasterizerOpenGL::SyncScissorTest() {
 }
 
 void RasterizerOpenGL::SyncPointState() {
-    auto& gpu = system.GPU().Maxwell3D();
-    auto& flags = gpu.dirty.flags;
+    auto& flags = maxwell3d.dirty.flags;
     if (!flags[Dirty::PointSize]) {
         return;
     }
     flags[Dirty::PointSize] = false;
 
-    oglEnable(GL_POINT_SPRITE, gpu.regs.point_sprite_enable);
+    oglEnable(GL_POINT_SPRITE, maxwell3d.regs.point_sprite_enable);
 
-    if (gpu.regs.vp_point_size.enable) {
+    if (maxwell3d.regs.vp_point_size.enable) {
         // By definition of GL_POINT_SIZE, it only matters if GL_PROGRAM_POINT_SIZE is disabled.
         glEnable(GL_PROGRAM_POINT_SIZE);
         return;
@@ -1573,32 +1536,30 @@ void RasterizerOpenGL::SyncPointState() {
 
     // Limit the point size to 1 since nouveau sometimes sets a point size of 0 (and that's invalid
     // in OpenGL).
-    glPointSize(std::max(1.0f, gpu.regs.point_size));
+    glPointSize(std::max(1.0f, maxwell3d.regs.point_size));
     glDisable(GL_PROGRAM_POINT_SIZE);
 }
 
 void RasterizerOpenGL::SyncLineState() {
-    auto& gpu = system.GPU().Maxwell3D();
-    auto& flags = gpu.dirty.flags;
+    auto& flags = maxwell3d.dirty.flags;
     if (!flags[Dirty::LineWidth]) {
         return;
     }
     flags[Dirty::LineWidth] = false;
 
-    const auto& regs = gpu.regs;
+    const auto& regs = maxwell3d.regs;
     oglEnable(GL_LINE_SMOOTH, regs.line_smooth_enable);
     glLineWidth(regs.line_smooth_enable ? regs.line_width_smooth : regs.line_width_aliased);
 }
 
 void RasterizerOpenGL::SyncPolygonOffset() {
-    auto& gpu = system.GPU().Maxwell3D();
-    auto& flags = gpu.dirty.flags;
+    auto& flags = maxwell3d.dirty.flags;
     if (!flags[Dirty::PolygonOffset]) {
         return;
     }
     flags[Dirty::PolygonOffset] = false;
 
-    const auto& regs = gpu.regs;
+    const auto& regs = maxwell3d.regs;
     oglEnable(GL_POLYGON_OFFSET_FILL, regs.polygon_offset_fill_enable);
     oglEnable(GL_POLYGON_OFFSET_LINE, regs.polygon_offset_line_enable);
     oglEnable(GL_POLYGON_OFFSET_POINT, regs.polygon_offset_point_enable);
@@ -1612,14 +1573,13 @@ void RasterizerOpenGL::SyncPolygonOffset() {
 }
 
 void RasterizerOpenGL::SyncAlphaTest() {
-    auto& gpu = system.GPU().Maxwell3D();
-    auto& flags = gpu.dirty.flags;
+    auto& flags = maxwell3d.dirty.flags;
     if (!flags[Dirty::AlphaTest]) {
         return;
     }
     flags[Dirty::AlphaTest] = false;
 
-    const auto& regs = gpu.regs;
+    const auto& regs = maxwell3d.regs;
     if (regs.alpha_test_enabled && regs.rt_control.count > 1) {
         LOG_WARNING(Render_OpenGL, "Alpha testing with more than one render target is not tested");
     }
@@ -1633,20 +1593,19 @@ void RasterizerOpenGL::SyncAlphaTest() {
 }
 
 void RasterizerOpenGL::SyncFramebufferSRGB() {
-    auto& gpu = system.GPU().Maxwell3D();
-    auto& flags = gpu.dirty.flags;
+    auto& flags = maxwell3d.dirty.flags;
     if (!flags[Dirty::FramebufferSRGB]) {
         return;
     }
     flags[Dirty::FramebufferSRGB] = false;
 
-    oglEnable(GL_FRAMEBUFFER_SRGB, gpu.regs.framebuffer_srgb);
+    oglEnable(GL_FRAMEBUFFER_SRGB, maxwell3d.regs.framebuffer_srgb);
 }
 
 void RasterizerOpenGL::SyncTransformFeedback() {
     // TODO(Rodrigo): Inject SKIP_COMPONENTS*_NV when required. An unimplemented message will signal
     // when this is required.
-    const auto& regs = system.GPU().Maxwell3D().regs;
+    const auto& regs = maxwell3d.regs;
 
     static constexpr std::size_t STRIDE = 3;
     std::array<GLint, 128 * STRIDE * Maxwell::NumTransformFeedbackBuffers> attribs;
@@ -1698,7 +1657,7 @@ void RasterizerOpenGL::SyncTransformFeedback() {
 }
 
 void RasterizerOpenGL::BeginTransformFeedback(GLenum primitive_mode) {
-    const auto& regs = system.GPU().Maxwell3D().regs;
+    const auto& regs = maxwell3d.regs;
     if (regs.tfb_enabled == 0) {
         return;
     }
@@ -1741,7 +1700,7 @@ void RasterizerOpenGL::BeginTransformFeedback(GLenum primitive_mode) {
 }
 
 void RasterizerOpenGL::EndTransformFeedback() {
-    const auto& regs = system.GPU().Maxwell3D().regs;
+    const auto& regs = maxwell3d.regs;
     if (regs.tfb_enabled == 0) {
         return;
     }

src/video_core/renderer_opengl/gl_rasterizer.h

@@ -36,8 +36,8 @@
 #include "video_core/shader/async_shaders.h"
 #include "video_core/textures/texture.h"
 
-namespace Core {
-class System;
+namespace Core::Memory {
+class Memory;
 }
 
 namespace Core::Frontend {
@@ -55,9 +55,10 @@ struct DrawParameters;
 
 class RasterizerOpenGL : public VideoCore::RasterizerAccelerated {
 public:
-    explicit RasterizerOpenGL(Core::System& system, Core::Frontend::EmuWindow& emu_window,
-                              const Device& device, ScreenInfo& info,
-                              ProgramManager& program_manager, StateTracker& state_tracker);
+    explicit RasterizerOpenGL(Core::Frontend::EmuWindow& emu_window, Tegra::GPU& gpu,
+                              Core::Memory::Memory& cpu_memory, const Device& device,
+                              ScreenInfo& screen_info, ProgramManager& program_manager,
+                              StateTracker& state_tracker);
     ~RasterizerOpenGL() override;
 
     void Draw(bool is_indexed, bool is_instanced) override;
@@ -83,9 +84,8 @@ public:
                                const Tegra::Engines::Fermi2D::Config& copy_config) override;
     bool AccelerateDisplay(const Tegra::FramebufferConfig& config, VAddr framebuffer_addr,
                            u32 pixel_stride) override;
-    void LoadDiskResources(const std::atomic_bool& stop_loading,
+    void LoadDiskResources(u64 title_id, const std::atomic_bool& stop_loading,
                            const VideoCore::DiskResourceLoadCallback& callback) override;
-    void SetupDirtyFlags() override;
 
     /// Returns true when there are commands queued to the OpenGL server.
     bool AnyCommandQueued() const {
@@ -237,7 +237,15 @@ private:
 
     void SetupShaders(GLenum primitive_mode);
 
+    Tegra::GPU& gpu;
+    Tegra::Engines::Maxwell3D& maxwell3d;
+    Tegra::Engines::KeplerCompute& kepler_compute;
+    Tegra::MemoryManager& gpu_memory;
+
     const Device& device;
+    ScreenInfo& screen_info;
+    ProgramManager& program_manager;
+    StateTracker& state_tracker;
 
     TextureCacheOpenGL texture_cache;
     ShaderCacheOpenGL shader_cache;
@@ -247,10 +255,6 @@ private:
     OGLBufferCache buffer_cache;
     FenceManagerOpenGL fence_manager;
 
-    Core::System& system;
-    ScreenInfo& screen_info;
-    ProgramManager& program_manager;
-    StateTracker& state_tracker;
     VideoCommon::Shader::AsyncShaders async_shaders;
 
     static constexpr std::size_t STREAM_BUFFER_SIZE = 128 * 1024 * 1024;

src/video_core/renderer_opengl/gl_shader_cache.cpp

@@ -239,12 +239,11 @@ std::unique_ptr<Shader> Shader::CreateStageFromMemory(
     ProgramCode code_b, VideoCommon::Shader::AsyncShaders& async_shaders, VAddr cpu_addr) {
     const auto shader_type = GetShaderType(program_type);
 
-    auto& gpu = params.system.GPU();
+    auto& gpu = params.gpu;
     gpu.ShaderNotify().MarkSharderBuilding();
 
     auto registry = std::make_shared<Registry>(shader_type, gpu.Maxwell3D());
-    if (!async_shaders.IsShaderAsync(params.system.GPU()) ||
-        !params.device.UseAsynchronousShaders()) {
+    if (!async_shaders.IsShaderAsync(gpu) || !params.device.UseAsynchronousShaders()) {
         const ShaderIR ir(code, STAGE_MAIN_OFFSET, COMPILER_SETTINGS, *registry);
         // TODO(Rodrigo): Handle VertexA shaders
         // std::optional<ShaderIR> ir_b;
@@ -287,11 +286,10 @@ std::unique_ptr<Shader> Shader::CreateStageFromMemory(
 
 std::unique_ptr<Shader> Shader::CreateKernelFromMemory(const ShaderParameters& params,
                                                        ProgramCode code) {
-    auto& gpu = params.system.GPU();
+    auto& gpu = params.gpu;
     gpu.ShaderNotify().MarkSharderBuilding();
 
-    auto& engine = gpu.KeplerCompute();
-    auto registry = std::make_shared<Registry>(ShaderType::Compute, engine);
+    auto registry = std::make_shared<Registry>(ShaderType::Compute, params.engine);
     const ShaderIR ir(code, KERNEL_MAIN_OFFSET, COMPILER_SETTINGS, *registry);
     const u64 uid = params.unique_identifier;
     auto program = BuildShader(params.device, ShaderType::Compute, uid, ir, *registry);
@@ -320,15 +318,20 @@ std::unique_ptr<Shader> Shader::CreateFromCache(const ShaderParameters& params,
         precompiled_shader.registry, precompiled_shader.entries, precompiled_shader.program));
 }
 
-ShaderCacheOpenGL::ShaderCacheOpenGL(RasterizerOpenGL& rasterizer, Core::System& system,
-                                     Core::Frontend::EmuWindow& emu_window, const Device& device)
-    : VideoCommon::ShaderCache<Shader>{rasterizer}, system{system},
-      emu_window{emu_window}, device{device}, disk_cache{system} {}
+ShaderCacheOpenGL::ShaderCacheOpenGL(RasterizerOpenGL& rasterizer,
+                                     Core::Frontend::EmuWindow& emu_window_, Tegra::GPU& gpu_,
+                                     Tegra::Engines::Maxwell3D& maxwell3d_,
+                                     Tegra::Engines::KeplerCompute& kepler_compute_,
+                                     Tegra::MemoryManager& gpu_memory_, const Device& device_)
+    : VideoCommon::ShaderCache<Shader>{rasterizer}, emu_window{emu_window_}, gpu{gpu_},
+      gpu_memory{gpu_memory_}, maxwell3d{maxwell3d_},
+      kepler_compute{kepler_compute_}, device{device_} {}
 
 ShaderCacheOpenGL::~ShaderCacheOpenGL() = default;
 
-void ShaderCacheOpenGL::LoadDiskCache(const std::atomic_bool& stop_loading,
+void ShaderCacheOpenGL::LoadDiskCache(u64 title_id, const std::atomic_bool& stop_loading,
                                       const VideoCore::DiskResourceLoadCallback& callback) {
+    disk_cache.BindTitleID(title_id);
     const std::optional transferable = disk_cache.LoadTransferable();
     if (!transferable) {
         return;
@@ -481,21 +484,19 @@ ProgramSharedPtr ShaderCacheOpenGL::GeneratePrecompiledProgram(
 
 Shader* ShaderCacheOpenGL::GetStageProgram(Maxwell::ShaderProgram program,
                                            VideoCommon::Shader::AsyncShaders& async_shaders) {
-    if (!system.GPU().Maxwell3D().dirty.flags[Dirty::Shaders]) {
+    if (!maxwell3d.dirty.flags[Dirty::Shaders]) {
         auto* last_shader = last_shaders[static_cast<std::size_t>(program)];
         if (last_shader->IsBuilt()) {
             return last_shader;
         }
     }
 
-    auto& memory_manager{system.GPU().MemoryManager()};
-    const GPUVAddr address{GetShaderAddress(system, program)};
+    const GPUVAddr address{GetShaderAddress(maxwell3d, program)};
 
     if (device.UseAsynchronousShaders() && async_shaders.HasCompletedWork()) {
         auto completed_work = async_shaders.GetCompletedWork();
         for (auto& work : completed_work) {
             Shader* shader = TryGet(work.cpu_address);
-            auto& gpu = system.GPU();
             gpu.ShaderNotify().MarkShaderComplete();
             if (shader == nullptr) {
                 continue;
@@ -507,14 +508,13 @@ Shader* ShaderCacheOpenGL::GetStageProgram(Maxwell::ShaderProgram program,
                 shader->AsyncGLASMBuilt(std::move(work.program.glasm));
             }
 
+            auto& registry = shader->GetRegistry();
+
             ShaderDiskCacheEntry entry;
             entry.type = work.shader_type;
             entry.code = std::move(work.code);
             entry.code_b = std::move(work.code_b);
             entry.unique_identifier = work.uid;
-
-            auto& registry = shader->GetRegistry();
-
             entry.bound_buffer = registry.GetBoundBuffer();
             entry.graphics_info = registry.GetGraphicsInfo();
             entry.keys = registry.GetKeys();
@@ -525,28 +525,28 @@ Shader* ShaderCacheOpenGL::GetStageProgram(Maxwell::ShaderProgram program,
     }
 
     // Look up shader in the cache based on address
-    const auto cpu_addr{memory_manager.GpuToCpuAddress(address)};
+    const std::optional<VAddr> cpu_addr{gpu_memory.GpuToCpuAddress(address)};
     if (Shader* const shader{cpu_addr ? TryGet(*cpu_addr) : null_shader.get()}) {
         return last_shaders[static_cast<std::size_t>(program)] = shader;
     }
 
-    const auto host_ptr{memory_manager.GetPointer(address)};
+    const u8* const host_ptr{gpu_memory.GetPointer(address)};
 
     // No shader found - create a new one
-    ProgramCode code{GetShaderCode(memory_manager, address, host_ptr, false)};
+    ProgramCode code{GetShaderCode(gpu_memory, address, host_ptr, false)};
     ProgramCode code_b;
     if (program == Maxwell::ShaderProgram::VertexA) {
-        const GPUVAddr address_b{GetShaderAddress(system, Maxwell::ShaderProgram::VertexB)};
-        const u8* host_ptr_b = memory_manager.GetPointer(address_b);
-        code_b = GetShaderCode(memory_manager, address_b, host_ptr_b, false);
+        const GPUVAddr address_b{GetShaderAddress(maxwell3d, Maxwell::ShaderProgram::VertexB)};
+        const u8* host_ptr_b = gpu_memory.GetPointer(address_b);
+        code_b = GetShaderCode(gpu_memory, address_b, host_ptr_b, false);
     }
     const std::size_t code_size = code.size() * sizeof(u64);
 
     const u64 unique_identifier = GetUniqueIdentifier(
         GetShaderType(program), program == Maxwell::ShaderProgram::VertexA, code, code_b);
 
-    const ShaderParameters params{system,    disk_cache, device,
-                                  *cpu_addr, host_ptr,   unique_identifier};
+    const ShaderParameters params{gpu,       maxwell3d, disk_cache,       device,
+                                  *cpu_addr, host_ptr,  unique_identifier};
 
     std::unique_ptr<Shader> shader;
     const auto found = runtime_cache.find(unique_identifier);
@@ -568,21 +568,20 @@ Shader* ShaderCacheOpenGL::GetStageProgram(Maxwell::ShaderProgram program,
 }
 
 Shader* ShaderCacheOpenGL::GetComputeKernel(GPUVAddr code_addr) {
-    auto& memory_manager{system.GPU().MemoryManager()};
-    const auto cpu_addr{memory_manager.GpuToCpuAddress(code_addr)};
+    const std::optional<VAddr> cpu_addr{gpu_memory.GpuToCpuAddress(code_addr)};
 
     if (Shader* const kernel = cpu_addr ? TryGet(*cpu_addr) : null_kernel.get()) {
         return kernel;
     }
 
-    const auto host_ptr{memory_manager.GetPointer(code_addr)};
     // No kernel found, create a new one
-    ProgramCode code{GetShaderCode(memory_manager, code_addr, host_ptr, true)};
+    const u8* host_ptr{gpu_memory.GetPointer(code_addr)};
+    ProgramCode code{GetShaderCode(gpu_memory, code_addr, host_ptr, true)};
     const std::size_t code_size{code.size() * sizeof(u64)};
     const u64 unique_identifier{GetUniqueIdentifier(ShaderType::Compute, false, code)};
 
-    const ShaderParameters params{system,    disk_cache, device,
-                                  *cpu_addr, host_ptr,   unique_identifier};
+    const ShaderParameters params{gpu,       kepler_compute, disk_cache,       device,
+                                  *cpu_addr, host_ptr,       unique_identifier};
 
     std::unique_ptr<Shader> kernel;
     const auto found = runtime_cache.find(unique_identifier);

src/video_core/renderer_opengl/gl_shader_cache.h

@@ -25,8 +25,8 @@
 #include "video_core/shader/shader_ir.h"
 #include "video_core/shader_cache.h"
 
-namespace Core {
-class System;
+namespace Tegra {
+class MemoryManager;
 }
 
 namespace Core::Frontend {
@@ -57,11 +57,12 @@ struct PrecompiledShader {
 };
 
 struct ShaderParameters {
-    Core::System& system;
+    Tegra::GPU& gpu;
+    Tegra::Engines::ConstBufferEngineInterface& engine;
     ShaderDiskCacheOpenGL& disk_cache;
     const Device& device;
     VAddr cpu_addr;
-    u8* host_ptr;
+    const u8* host_ptr;
     u64 unique_identifier;
 };
 
@@ -118,12 +119,14 @@ private:
 
 class ShaderCacheOpenGL final : public VideoCommon::ShaderCache<Shader> {
 public:
-    explicit ShaderCacheOpenGL(RasterizerOpenGL& rasterizer, Core::System& system,
-                               Core::Frontend::EmuWindow& emu_window, const Device& device);
+    explicit ShaderCacheOpenGL(RasterizerOpenGL& rasterizer, Core::Frontend::EmuWindow& emu_window,
+                               Tegra::GPU& gpu, Tegra::Engines::Maxwell3D& maxwell3d,
+                               Tegra::Engines::KeplerCompute& kepler_compute,
+                               Tegra::MemoryManager& gpu_memory, const Device& device);
     ~ShaderCacheOpenGL() override;
 
     /// Loads disk cache for the current game
-    void LoadDiskCache(const std::atomic_bool& stop_loading,
+    void LoadDiskCache(u64 title_id, const std::atomic_bool& stop_loading,
                        const VideoCore::DiskResourceLoadCallback& callback);
 
     /// Gets the current specified shader stage program
@@ -138,9 +141,13 @@ private:
         const ShaderDiskCacheEntry& entry, const ShaderDiskCachePrecompiled& precompiled_entry,
         const std::unordered_set<GLenum>& supported_formats);
 
-    Core::System& system;
     Core::Frontend::EmuWindow& emu_window;
+    Tegra::GPU& gpu;
+    Tegra::MemoryManager& gpu_memory;
+    Tegra::Engines::Maxwell3D& maxwell3d;
+    Tegra::Engines::KeplerCompute& kepler_compute;
     const Device& device;
+
     ShaderDiskCacheOpenGL disk_cache;
     std::unordered_map<u64, PrecompiledShader> runtime_cache;
 

src/video_core/renderer_opengl/gl_shader_disk_cache.cpp

@@ -206,13 +206,17 @@ bool ShaderDiskCacheEntry::Save(Common::FS::IOFile& file) const {
                flat_bindless_samplers.size();
 }
 
-ShaderDiskCacheOpenGL::ShaderDiskCacheOpenGL(Core::System& system) : system{system} {}
+ShaderDiskCacheOpenGL::ShaderDiskCacheOpenGL() = default;
 
 ShaderDiskCacheOpenGL::~ShaderDiskCacheOpenGL() = default;
 
+void ShaderDiskCacheOpenGL::BindTitleID(u64 title_id_) {
+    title_id = title_id_;
+}
+
 std::optional<std::vector<ShaderDiskCacheEntry>> ShaderDiskCacheOpenGL::LoadTransferable() {
     // Skip games without title id
-    const bool has_title_id = system.CurrentProcess()->GetTitleID() != 0;
+    const bool has_title_id = title_id != 0;
     if (!Settings::values.use_disk_shader_cache.GetValue() || !has_title_id) {
         return std::nullopt;
     }
@@ -474,7 +478,7 @@ std::string ShaderDiskCacheOpenGL::GetBaseDir() const {
 }
 
 std::string ShaderDiskCacheOpenGL::GetTitleID() const {
-    return fmt::format("{:016X}", system.CurrentProcess()->GetTitleID());
+    return fmt::format("{:016X}", title_id);
 }
 
 } // namespace OpenGL

src/video_core/renderer_opengl/gl_shader_disk_cache.h

@@ -21,10 +21,6 @@
 #include "video_core/engines/shader_type.h"
 #include "video_core/shader/registry.h"
 
-namespace Core {
-class System;
-}
-
 namespace Common::FS {
 class IOFile;
 }
@@ -70,9 +66,12 @@ struct ShaderDiskCachePrecompiled {
 
 class ShaderDiskCacheOpenGL {
 public:
-    explicit ShaderDiskCacheOpenGL(Core::System& system);
+    explicit ShaderDiskCacheOpenGL();
     ~ShaderDiskCacheOpenGL();
 
+    /// Binds a title ID for all future operations.
+    void BindTitleID(u64 title_id);
+
     /// Loads transferable cache. If file has a old version or on failure, it deletes the file.
     std::optional<std::vector<ShaderDiskCacheEntry>> LoadTransferable();
 
@@ -157,8 +156,6 @@ private:
         return LoadArrayFromPrecompiled(&object, 1);
     }
 
-    Core::System& system;
-
     // Stores whole precompiled cache which will be read from or saved to the precompiled chache
     // file
     FileSys::VectorVfsFile precompiled_cache_virtual_file;
@@ -168,8 +165,11 @@ private:
     // Stored transferable shaders
     std::unordered_set<u64> stored_transferable;
 
+    /// Title ID to operate on
+    u64 title_id = 0;
+
     // The cache has been loaded at boot
-    bool is_usable{};
+    bool is_usable = false;
 };
 
 } // namespace OpenGL

src/video_core/renderer_opengl/gl_state_tracker.cpp

@@ -214,10 +214,8 @@ void SetupDirtyMisc(Tables& tables) {
 
 } // Anonymous namespace
 
-StateTracker::StateTracker(Core::System& system) : system{system} {}
-
-void StateTracker::Initialize() {
-    auto& dirty = system.GPU().Maxwell3D().dirty;
+StateTracker::StateTracker(Tegra::GPU& gpu) : flags{gpu.Maxwell3D().dirty.flags} {
+    auto& dirty = gpu.Maxwell3D().dirty;
     auto& tables = dirty.tables;
     SetupDirtyRenderTargets(tables);
     SetupDirtyColorMasks(tables);

src/video_core/renderer_opengl/gl_state_tracker.h

@@ -13,8 +13,8 @@
 #include "video_core/dirty_flags.h"
 #include "video_core/engines/maxwell_3d.h"
 
-namespace Core {
-class System;
+namespace Tegra {
+class GPU;
 }
 
 namespace OpenGL {
@@ -90,9 +90,7 @@ static_assert(Last <= std::numeric_limits<u8>::max());
 
 class StateTracker {
 public:
-    explicit StateTracker(Core::System& system);
-
-    void Initialize();
+    explicit StateTracker(Tegra::GPU& gpu);
 
     void BindIndexBuffer(GLuint new_index_buffer) {
         if (index_buffer == new_index_buffer) {
@@ -103,7 +101,6 @@ public:
     }
 
     void NotifyScreenDrawVertexArray() {
-        auto& flags = system.GPU().Maxwell3D().dirty.flags;
         flags[OpenGL::Dirty::VertexFormats] = true;
         flags[OpenGL::Dirty::VertexFormat0 + 0] = true;
         flags[OpenGL::Dirty::VertexFormat0 + 1] = true;
@@ -117,98 +114,81 @@ public:
     }
 
     void NotifyPolygonModes() {
-        auto& flags = system.GPU().Maxwell3D().dirty.flags;
         flags[OpenGL::Dirty::PolygonModes] = true;
         flags[OpenGL::Dirty::PolygonModeFront] = true;
         flags[OpenGL::Dirty::PolygonModeBack] = true;
     }
 
     void NotifyViewport0() {
-        auto& flags = system.GPU().Maxwell3D().dirty.flags;
         flags[OpenGL::Dirty::Viewports] = true;
         flags[OpenGL::Dirty::Viewport0] = true;
     }
 
     void NotifyScissor0() {
-        auto& flags = system.GPU().Maxwell3D().dirty.flags;
         flags[OpenGL::Dirty::Scissors] = true;
         flags[OpenGL::Dirty::Scissor0] = true;
     }
 
     void NotifyColorMask0() {
-        auto& flags = system.GPU().Maxwell3D().dirty.flags;
         flags[OpenGL::Dirty::ColorMasks] = true;
         flags[OpenGL::Dirty::ColorMask0] = true;
     }
 
     void NotifyBlend0() {
-        auto& flags = system.GPU().Maxwell3D().dirty.flags;
         flags[OpenGL::Dirty::BlendStates] = true;
         flags[OpenGL::Dirty::BlendState0] = true;
     }
 
     void NotifyFramebuffer() {
-        auto& flags = system.GPU().Maxwell3D().dirty.flags;
         flags[VideoCommon::Dirty::RenderTargets] = true;
     }
 
     void NotifyFrontFace() {
-        auto& flags = system.GPU().Maxwell3D().dirty.flags;
         flags[OpenGL::Dirty::FrontFace] = true;
     }
 
     void NotifyCullTest() {
-        auto& flags = system.GPU().Maxwell3D().dirty.flags;
         flags[OpenGL::Dirty::CullTest] = true;
     }
 
     void NotifyDepthMask() {
-        auto& flags = system.GPU().Maxwell3D().dirty.flags;
         flags[OpenGL::Dirty::DepthMask] = true;
     }
 
     void NotifyDepthTest() {
-        auto& flags = system.GPU().Maxwell3D().dirty.flags;
         flags[OpenGL::Dirty::DepthTest] = true;
     }
 
     void NotifyStencilTest() {
-        auto& flags = system.GPU().Maxwell3D().dirty.flags;
         flags[OpenGL::Dirty::StencilTest] = true;
     }
 
     void NotifyPolygonOffset() {
-        auto& flags = system.GPU().Maxwell3D().dirty.flags;
         flags[OpenGL::Dirty::PolygonOffset] = true;
     }
 
     void NotifyRasterizeEnable() {
-        auto& flags = system.GPU().Maxwell3D().dirty.flags;
         flags[OpenGL::Dirty::RasterizeEnable] = true;
     }
 
     void NotifyFramebufferSRGB() {
-        auto& flags = system.GPU().Maxwell3D().dirty.flags;
         flags[OpenGL::Dirty::FramebufferSRGB] = true;
     }
 
     void NotifyLogicOp() {
-        auto& flags = system.GPU().Maxwell3D().dirty.flags;
         flags[OpenGL::Dirty::LogicOp] = true;
     }
 
     void NotifyClipControl() {
-        auto& flags = system.GPU().Maxwell3D().dirty.flags;
         flags[OpenGL::Dirty::ClipControl] = true;
     }
 
     void NotifyAlphaTest() {
-        auto& flags = system.GPU().Maxwell3D().dirty.flags;
         flags[OpenGL::Dirty::AlphaTest] = true;
     }
 
 private:
-    Core::System& system;
+    Tegra::Engines::Maxwell3D::DirtyState::Flags& flags;
 
     GLuint index_buffer = 0;
 };

src/video_core/renderer_opengl/gl_texture_cache.cpp

@@ -532,10 +532,12 @@ OGLTextureView CachedSurfaceView::CreateTextureView() const {
     return texture_view;
 }
 
-TextureCacheOpenGL::TextureCacheOpenGL(Core::System& system,
-                                       VideoCore::RasterizerInterface& rasterizer,
-                                       const Device& device, StateTracker& state_tracker)
-    : TextureCacheBase{system, rasterizer, device.HasASTC()}, state_tracker{state_tracker} {
+TextureCacheOpenGL::TextureCacheOpenGL(VideoCore::RasterizerInterface& rasterizer,
+                                       Tegra::Engines::Maxwell3D& maxwell3d,
+                                       Tegra::MemoryManager& gpu_memory, const Device& device,
+                                       StateTracker& state_tracker_)
+    : TextureCacheBase{rasterizer, maxwell3d, gpu_memory, device.HasASTC()}, state_tracker{
+                                                                                 state_tracker_} {
     src_framebuffer.Create();
     dst_framebuffer.Create();
 }

src/video_core/renderer_opengl/gl_texture_cache.h

@@ -129,8 +129,10 @@ private:
 
 class TextureCacheOpenGL final : public TextureCacheBase {
 public:
-    explicit TextureCacheOpenGL(Core::System& system, VideoCore::RasterizerInterface& rasterizer,
-                                const Device& device, StateTracker& state_tracker);
+    explicit TextureCacheOpenGL(VideoCore::RasterizerInterface& rasterizer,
+                                Tegra::Engines::Maxwell3D& maxwell3d,
+                                Tegra::MemoryManager& gpu_memory, const Device& device,
+                                StateTracker& state_tracker);
     ~TextureCacheOpenGL();
 
 protected: