vm_manager: Add missing commas to string literal array elements in GetMemoryStateName()

Without these, this would perform concatenation, which is definitely not
what we want here.

src/common/logging/backend.cpp

@@ -5,6 +5,7 @@
 #include <algorithm>
 #include <array>
 #include <chrono>
+#include <climits>
 #include <condition_variable>
 #include <memory>
 #include <thread>
@@ -83,8 +84,10 @@ private:
                 }
             };
             while (true) {
-                std::unique_lock<std::mutex> lock(message_mutex);
-                message_cv.wait(lock, [&] { return !running || message_queue.Pop(entry); });
+                {
+                    std::unique_lock<std::mutex> lock(message_mutex);
+                    message_cv.wait(lock, [&] { return !running || message_queue.Pop(entry); });
+                }
                 if (!running) {
                     break;
                 }
@@ -92,7 +95,7 @@ private:
             }
             // Drain the logging queue. Only writes out up to MAX_LOGS_TO_WRITE to prevent a case
             // where a system is repeatedly spamming logs even on close.
-            constexpr int MAX_LOGS_TO_WRITE = 100;
+            const int MAX_LOGS_TO_WRITE = filter.IsDebug() ? INT_MAX : 100;
             int logs_written = 0;
             while (logs_written++ < MAX_LOGS_TO_WRITE && message_queue.Pop(entry)) {
                 write_logs(entry);
@@ -282,4 +285,4 @@ void FmtLogMessageImpl(Class log_class, Level log_level, const char* filename,
 
     Impl::Instance().PushEntry(std::move(entry));
 }
-} // namespace Log
+} // namespace Log

src/common/logging/filter.cpp

@@ -94,4 +94,11 @@ bool Filter::ParseFilterRule(const std::string::const_iterator begin,
 bool Filter::CheckMessage(Class log_class, Level level) const {
     return static_cast<u8>(level) >= static_cast<u8>(class_levels[static_cast<size_t>(log_class)]);
 }
+
+bool Filter::IsDebug() const {
+    return std::any_of(class_levels.begin(), class_levels.end(), [](const Level& l) {
+        return static_cast<u8>(l) <= static_cast<u8>(Level::Debug);
+    });
+}
+
 } // namespace Log

src/common/logging/filter.h

@@ -47,6 +47,9 @@ public:
     /// Matches class/level combination against the filter, returning true if it passed.
     bool CheckMessage(Class log_class, Level level) const;
 
+    /// Returns true if any logging classes are set to debug
+    bool IsDebug() const;
+
 private:
     std::array<Level, (size_t)Class::Count> class_levels;
 };

src/common/string_util.cpp

@@ -134,7 +134,7 @@ bool SplitPath(const std::string& full_path, std::string* _pPath, std::string* _
     size_t dir_end = full_path.find_last_of("/"
 // windows needs the : included for something like just "C:" to be considered a directory
 #ifdef _WIN32
-                                            ":"
+                                            "\\:"
 #endif
     );
     if (std::string::npos == dir_end)

src/common/telemetry.h

@@ -52,27 +52,14 @@ public:
 template <typename T>
 class Field : public FieldInterface {
 public:
-    Field(FieldType type, std::string name, const T& value)
-        : name(std::move(name)), type(type), value(value) {}
-
-    Field(FieldType type, std::string name, T&& value)
+    Field(FieldType type, std::string name, T value)
         : name(std::move(name)), type(type), value(std::move(value)) {}
 
-    Field(const Field& other) : Field(other.type, other.name, other.value) {}
+    Field(const Field&) = default;
+    Field& operator=(const Field&) = default;
 
-    Field& operator=(const Field& other) {
-        type = other.type;
-        name = other.name;
-        value = other.value;
-        return *this;
-    }
-
-    Field& operator=(Field&& other) {
-        type = other.type;
-        name = std::move(other.name);
-        value = std::move(other.value);
-        return *this;
-    }
+    Field(Field&&) = default;
+    Field& operator=(Field&& other) = default;
 
     void Accept(VisitorInterface& visitor) const override;
 
@@ -94,11 +81,11 @@ public:
         return value;
     }
 
-    inline bool operator==(const Field<T>& other) {
+    bool operator==(const Field& other) const {
         return (type == other.type) && (name == other.name) && (value == other.value);
     }
 
-    inline bool operator!=(const Field<T>& other) {
+    bool operator!=(const Field& other) const {
         return !(*this == other);
     }
 

src/core/arm/arm_interface.h

@@ -116,6 +116,8 @@ public:
      */
     virtual void LoadContext(const ThreadContext& ctx) = 0;
 
+    virtual void ClearExclusiveState() = 0;
+
     /// Prepare core for thread reschedule (if needed to correctly handle state)
     virtual void PrepareReschedule() = 0;
 };

src/core/arm/dynarmic/arm_dynarmic.cpp

@@ -226,6 +226,10 @@ void ARM_Dynarmic::ClearInstructionCache() {
     jit->ClearCache();
 }
 
+void ARM_Dynarmic::ClearExclusiveState() {
+    jit->ClearExclusiveState();
+}
+
 void ARM_Dynarmic::PageTableChanged() {
     jit = MakeJit(cb);
     current_page_table = Memory::GetCurrentPageTable();

src/core/arm/dynarmic/arm_dynarmic.h

@@ -39,6 +39,7 @@ public:
     void LoadContext(const ThreadContext& ctx) override;
 
     void PrepareReschedule() override;
+    void ClearExclusiveState() override;
 
     void ClearInstructionCache() override;
     void PageTableChanged() override;

src/core/arm/unicorn/arm_unicorn.cpp

@@ -193,11 +193,11 @@ void ARM_Unicorn::ExecuteInstructions(int num_instructions) {
         }
         Kernel::Thread* thread = Kernel::GetCurrentThread();
         SaveContext(thread->context);
-        if (last_bkpt_hit) {
+        if (last_bkpt_hit || (num_instructions == 1)) {
             last_bkpt_hit = false;
             GDBStub::Break();
+            GDBStub::SendTrap(thread, 5);
         }
-        GDBStub::SendTrap(thread, 5);
     }
 }
 
@@ -263,6 +263,8 @@ void ARM_Unicorn::PrepareReschedule() {
     CHECKED(uc_emu_stop(uc));
 }
 
+void ARM_Unicorn::ClearExclusiveState() {}
+
 void ARM_Unicorn::ClearInstructionCache() {}
 
 void ARM_Unicorn::RecordBreak(GDBStub::BreakpointAddress bkpt) {

src/core/arm/unicorn/arm_unicorn.h

@@ -31,6 +31,7 @@ public:
     void SaveContext(ThreadContext& ctx) override;
     void LoadContext(const ThreadContext& ctx) override;
     void PrepareReschedule() override;
+    void ClearExclusiveState() override;
     void ExecuteInstructions(int num_instructions);
     void Run() override;
     void Step() override;

src/core/file_sys/disk_filesystem.cpp

@@ -58,11 +58,13 @@ ResultVal<std::unique_ptr<StorageBackend>> Disk_FileSystem::OpenFile(const std::
 }
 
 ResultCode Disk_FileSystem::DeleteFile(const std::string& path) const {
-    if (!FileUtil::Exists(path)) {
+    std::string full_path = base_directory + path;
+
+    if (!FileUtil::Exists(full_path)) {
         return ERROR_PATH_NOT_FOUND;
     }
 
-    FileUtil::Delete(path);
+    FileUtil::Delete(full_path);
 
     return RESULT_SUCCESS;
 }

src/core/file_sys/errors.h

@@ -12,6 +12,8 @@ namespace ErrCodes {
 enum {
     NotFound = 1,
     SaveDataNotFound = 1002,
+    SdCardNotFound = 2001,
+    RomFSNotFound = 2520,
 };
 }
 

src/core/file_sys/filesystem.h

@@ -167,35 +167,4 @@ public:
     virtual ResultVal<EntryType> GetEntryType(const std::string& path) const = 0;
 };
 
-class FileSystemFactory : NonCopyable {
-public:
-    virtual ~FileSystemFactory() {}
-
-    /**
-     * Get a descriptive name for the archive (e.g. "RomFS", "SaveData", etc.)
-     */
-    virtual std::string GetName() const = 0;
-
-    /**
-     * Tries to open the archive of this type with the specified path
-     * @param path Path to the archive
-     * @return An ArchiveBackend corresponding operating specified archive path.
-     */
-    virtual ResultVal<std::unique_ptr<FileSystemBackend>> Open(const Path& path) = 0;
-
-    /**
-     * Deletes the archive contents and then re-creates the base folder
-     * @param path Path to the archive
-     * @return ResultCode of the operation, 0 on success
-     */
-    virtual ResultCode Format(const Path& path) = 0;
-
-    /**
-     * Retrieves the format info about the archive with the specified path
-     * @param path Path to the archive
-     * @return Format information about the archive or error code
-     */
-    virtual ResultVal<ArchiveFormatInfo> GetFormatInfo(const Path& path) const = 0;
-};
-
 } // namespace FileSys

src/core/file_sys/romfs_factory.cpp

@@ -11,28 +11,17 @@
 
 namespace FileSys {
 
-RomFS_Factory::RomFS_Factory(Loader::AppLoader& app_loader) {
+RomFSFactory::RomFSFactory(Loader::AppLoader& app_loader) {
     // Load the RomFS from the app
     if (Loader::ResultStatus::Success != app_loader.ReadRomFS(romfs_file, data_offset, data_size)) {
         LOG_ERROR(Service_FS, "Unable to read RomFS!");
     }
 }
 
-ResultVal<std::unique_ptr<FileSystemBackend>> RomFS_Factory::Open(const Path& path) {
+ResultVal<std::unique_ptr<FileSystemBackend>> RomFSFactory::Open(u64 title_id) {
+    // TODO(DarkLordZach): Use title id.
     auto archive = std::make_unique<RomFS_FileSystem>(romfs_file, data_offset, data_size);
     return MakeResult<std::unique_ptr<FileSystemBackend>>(std::move(archive));
 }
 
-ResultCode RomFS_Factory::Format(const Path& path) {
-    LOG_ERROR(Service_FS, "Unimplemented Format archive {}", GetName());
-    // TODO(bunnei): Find the right error code for this
-    return ResultCode(-1);
-}
-
-ResultVal<ArchiveFormatInfo> RomFS_Factory::GetFormatInfo(const Path& path) const {
-    LOG_ERROR(Service_FS, "Unimplemented GetFormatInfo archive {}", GetName());
-    // TODO(bunnei): Find the right error code for this
-    return ResultCode(-1);
-}
-
 } // namespace FileSys

src/core/file_sys/romfs_factory.h

@@ -15,16 +15,11 @@
 namespace FileSys {
 
 /// File system interface to the RomFS archive
-class RomFS_Factory final : public FileSystemFactory {
+class RomFSFactory {
 public:
-    explicit RomFS_Factory(Loader::AppLoader& app_loader);
+    explicit RomFSFactory(Loader::AppLoader& app_loader);
 
-    std::string GetName() const override {
-        return "ArchiveFactory_RomFS";
-    }
-    ResultVal<std::unique_ptr<FileSystemBackend>> Open(const Path& path) override;
-    ResultCode Format(const Path& path) override;
-    ResultVal<ArchiveFormatInfo> GetFormatInfo(const Path& path) const override;
+    ResultVal<std::unique_ptr<FileSystemBackend>> Open(u64 title_id);
 
 private:
     std::shared_ptr<FileUtil::IOFile> romfs_file;

src/core/file_sys/savedata_factory.cpp

@@ -12,11 +12,49 @@
 
 namespace FileSys {
 
-SaveData_Factory::SaveData_Factory(std::string nand_directory)
+std::string SaveDataDescriptor::DebugInfo() {
+    return fmt::format("[type={:02X}, title_id={:016X}, user_id={:016X}{:016X}, save_id={:016X}]",
+                       static_cast<u8>(type), title_id, user_id[1], user_id[0], save_id);
+}
+
+SaveDataFactory::SaveDataFactory(std::string nand_directory)
     : nand_directory(std::move(nand_directory)) {}
 
-ResultVal<std::unique_ptr<FileSystemBackend>> SaveData_Factory::Open(const Path& path) {
-    std::string save_directory = GetFullPath();
+ResultVal<std::unique_ptr<FileSystemBackend>> SaveDataFactory::Open(SaveDataSpaceId space,
+                                                                    SaveDataDescriptor meta) {
+    if (meta.type == SaveDataType::SystemSaveData || meta.type == SaveDataType::SaveData) {
+        if (meta.zero_1 != 0) {
+            LOG_WARNING(Service_FS,
+                        "Possibly incorrect SaveDataDescriptor, type is "
+                        "SystemSaveData||SaveData but offset 0x28 is non-zero ({:016X}).",
+                        meta.zero_1);
+        }
+        if (meta.zero_2 != 0) {
+            LOG_WARNING(Service_FS,
+                        "Possibly incorrect SaveDataDescriptor, type is "
+                        "SystemSaveData||SaveData but offset 0x30 is non-zero ({:016X}).",
+                        meta.zero_2);
+        }
+        if (meta.zero_3 != 0) {
+            LOG_WARNING(Service_FS,
+                        "Possibly incorrect SaveDataDescriptor, type is "
+                        "SystemSaveData||SaveData but offset 0x38 is non-zero ({:016X}).",
+                        meta.zero_3);
+        }
+    }
+
+    if (meta.type == SaveDataType::SystemSaveData && meta.title_id != 0) {
+        LOG_WARNING(Service_FS,
+                    "Possibly incorrect SaveDataDescriptor, type is SystemSaveData but title_id is "
+                    "non-zero ({:016X}).",
+                    meta.title_id);
+    }
+
+    std::string save_directory =
+        GetFullPath(space, meta.type, meta.title_id, meta.user_id, meta.save_id);
+
+    // TODO(DarkLordZach): Try to not create when opening, there are dedicated create save methods.
+    // But, user_ids don't match so this works for now.
 
     if (!FileUtil::Exists(save_directory)) {
         // TODO(bunnei): This is a work-around to always create a save data directory if it does not
@@ -26,6 +64,12 @@ ResultVal<std::unique_ptr<FileSystemBackend>> SaveData_Factory::Open(const Path&
         FileUtil::CreateFullPath(save_directory);
     }
 
+    // TODO(DarkLordZach): For some reason, CreateFullPath doesn't create the last bit. Should be
+    // fixed with VFS.
+    if (!FileUtil::IsDirectory(save_directory)) {
+        FileUtil::CreateDir(save_directory);
+    }
+
     // Return an error if the save data doesn't actually exist.
     if (!FileUtil::IsDirectory(save_directory)) {
         // TODO(Subv): Find out correct error code.
@@ -36,28 +80,35 @@ ResultVal<std::unique_ptr<FileSystemBackend>> SaveData_Factory::Open(const Path&
     return MakeResult<std::unique_ptr<FileSystemBackend>>(std::move(archive));
 }
 
-ResultCode SaveData_Factory::Format(const Path& path) {
-    LOG_WARNING(Service_FS, "Format archive {}", GetName());
-    // Create the save data directory.
-    if (!FileUtil::CreateFullPath(GetFullPath())) {
-        // TODO(Subv): Find the correct error code.
-        return ResultCode(-1);
+std::string SaveDataFactory::GetFullPath(SaveDataSpaceId space, SaveDataType type, u64 title_id,
+                                         u128 user_id, u64 save_id) const {
+    // According to switchbrew, if a save is of type SaveData and the title id field is 0, it should
+    // be interpreted as the title id of the current process.
+    if (type == SaveDataType::SaveData && title_id == 0)
+        title_id = Core::CurrentProcess()->program_id;
+
+    std::string prefix;
+
+    switch (space) {
+    case SaveDataSpaceId::NandSystem:
+        prefix = nand_directory + "system/save/";
+        break;
+    case SaveDataSpaceId::NandUser:
+        prefix = nand_directory + "user/save/";
+        break;
+    default:
+        ASSERT_MSG(false, "Unrecognized SaveDataSpaceId: {:02X}", static_cast<u8>(space));
     }
 
-    return RESULT_SUCCESS;
-}
-
-ResultVal<ArchiveFormatInfo> SaveData_Factory::GetFormatInfo(const Path& path) const {
-    LOG_ERROR(Service_FS, "Unimplemented GetFormatInfo archive {}", GetName());
-    // TODO(bunnei): Find the right error code for this
-    return ResultCode(-1);
-}
-
-std::string SaveData_Factory::GetFullPath() const {
-    u64 title_id = Core::CurrentProcess()->program_id;
-    // TODO(Subv): Somehow obtain this value.
-    u32 user = 0;
-    return fmt::format("{}save/{:016X}/{:08X}/", nand_directory, title_id, user);
+    switch (type) {
+    case SaveDataType::SystemSaveData:
+        return fmt::format("{}{:016X}/{:016X}{:016X}", prefix, save_id, user_id[1], user_id[0]);
+    case SaveDataType::SaveData:
+        return fmt::format("{}{:016X}/{:016X}{:016X}/{:016X}", prefix, 0, user_id[1], user_id[0],
+                           title_id);
+    default:
+        ASSERT_MSG(false, "Unrecognized SaveDataType: {:02X}", static_cast<u8>(type));
+    }
 }
 
 } // namespace FileSys

src/core/file_sys/savedata_factory.h

@@ -12,22 +12,50 @@
 
 namespace FileSys {
 
-/// File system interface to the SaveData archive
-class SaveData_Factory final : public FileSystemFactory {
-public:
-    explicit SaveData_Factory(std::string nand_directory);
+enum class SaveDataSpaceId : u8 {
+    NandSystem = 0,
+    NandUser = 1,
+    SdCard = 2,
+    TemporaryStorage = 3,
+};
 
-    std::string GetName() const override {
-        return "SaveData_Factory";
-    }
-    ResultVal<std::unique_ptr<FileSystemBackend>> Open(const Path& path) override;
-    ResultCode Format(const Path& path) override;
-    ResultVal<ArchiveFormatInfo> GetFormatInfo(const Path& path) const override;
+enum class SaveDataType : u8 {
+    SystemSaveData = 0,
+    SaveData = 1,
+    BcatDeliveryCacheStorage = 2,
+    DeviceSaveData = 3,
+    TemporaryStorage = 4,
+    CacheStorage = 5,
+};
+
+struct SaveDataDescriptor {
+    u64_le title_id;
+    u128 user_id;
+    u64_le save_id;
+    SaveDataType type;
+    INSERT_PADDING_BYTES(7);
+    u64_le zero_1;
+    u64_le zero_2;
+    u64_le zero_3;
+
+    std::string DebugInfo();
+};
+static_assert(sizeof(SaveDataDescriptor) == 0x40, "SaveDataDescriptor has incorrect size.");
+
+/// File system interface to the SaveData archive
+class SaveDataFactory {
+public:
+    explicit SaveDataFactory(std::string nand_directory);
+
+    ResultVal<std::unique_ptr<FileSystemBackend>> Open(SaveDataSpaceId space,
+                                                       SaveDataDescriptor meta);
 
 private:
     std::string nand_directory;
+    std::string sd_directory;
 
-    std::string GetFullPath() const;
+    std::string GetFullPath(SaveDataSpaceId space, SaveDataType type, u64 title_id, u128 user_id,
+                            u64 save_id) const;
 };
 
 } // namespace FileSys

src/core/file_sys/sdmc_factory.cpp

@@ -12,9 +12,9 @@
 
 namespace FileSys {
 
-SDMC_Factory::SDMC_Factory(std::string sd_directory) : sd_directory(std::move(sd_directory)) {}
+SDMCFactory::SDMCFactory(std::string sd_directory) : sd_directory(std::move(sd_directory)) {}
 
-ResultVal<std::unique_ptr<FileSystemBackend>> SDMC_Factory::Open(const Path& path) {
+ResultVal<std::unique_ptr<FileSystemBackend>> SDMCFactory::Open() {
     // Create the SD Card directory if it doesn't already exist.
     if (!FileUtil::IsDirectory(sd_directory)) {
         FileUtil::CreateFullPath(sd_directory);
@@ -24,16 +24,4 @@ ResultVal<std::unique_ptr<FileSystemBackend>> SDMC_Factory::Open(const Path& pat
     return MakeResult<std::unique_ptr<FileSystemBackend>>(std::move(archive));
 }
 
-ResultCode SDMC_Factory::Format(const Path& path) {
-    LOG_ERROR(Service_FS, "Unimplemented Format archive {}", GetName());
-    // TODO(Subv): Find the right error code for this
-    return ResultCode(-1);
-}
-
-ResultVal<ArchiveFormatInfo> SDMC_Factory::GetFormatInfo(const Path& path) const {
-    LOG_ERROR(Service_FS, "Unimplemented GetFormatInfo archive {}", GetName());
-    // TODO(bunnei): Find the right error code for this
-    return ResultCode(-1);
-}
-
 } // namespace FileSys

src/core/file_sys/sdmc_factory.h

@@ -13,16 +13,11 @@
 namespace FileSys {
 
 /// File system interface to the SDCard archive
-class SDMC_Factory final : public FileSystemFactory {
+class SDMCFactory {
 public:
-    explicit SDMC_Factory(std::string sd_directory);
+    explicit SDMCFactory(std::string sd_directory);
 
-    std::string GetName() const override {
-        return "SDMC_Factory";
-    }
-    ResultVal<std::unique_ptr<FileSystemBackend>> Open(const Path& path) override;
-    ResultCode Format(const Path& path) override;
-    ResultVal<ArchiveFormatInfo> GetFormatInfo(const Path& path) const override;
+    ResultVal<std::unique_ptr<FileSystemBackend>> Open();
 
 private:
     std::string sd_directory;

src/core/gdbstub/gdbstub.cpp

@@ -61,10 +61,16 @@ const u32 SIGTERM = 15;
 const u32 MSG_WAITALL = 8;
 #endif
 
-const u32 X30_REGISTER = 30;
+const u32 LR_REGISTER = 30;
 const u32 SP_REGISTER = 31;
 const u32 PC_REGISTER = 32;
 const u32 CPSR_REGISTER = 33;
+const u32 UC_ARM64_REG_Q0 = 34;
+const u32 FPSCR_REGISTER = 66;
+
+// TODO/WiP - Used while working on support for FPU
+const u32 TODO_DUMMY_REG_997 = 997;
+const u32 TODO_DUMMY_REG_998 = 998;
 
 // For sample XML files see the GDB source /gdb/features
 // GDB also wants the l character at the start
@@ -130,6 +136,8 @@ static const char* target_xml =
     </flags>
     <reg name="cpsr" bitsize="32" type="cpsr_flags"/>
   </feature>
+  <feature name="org.gnu.gdb.aarch64.fpu">
+  </feature>
 </target>
 )";
 
@@ -144,6 +152,7 @@ static u32 latest_signal = 0;
 static bool memory_break = false;
 
 static Kernel::Thread* current_thread = nullptr;
+static u32 current_core = 0;
 
 // Binding to a port within the reserved ports range (0-1023) requires root permissions,
 // so default to a port outside of that range.
@@ -171,13 +180,34 @@ static std::map<u64, Breakpoint> breakpoints_execute;
 static std::map<u64, Breakpoint> breakpoints_read;
 static std::map<u64, Breakpoint> breakpoints_write;
 
+struct Module {
+    std::string name;
+    PAddr beg;
+    PAddr end;
+};
+
+static std::vector<Module> modules;
+
+void RegisterModule(std::string name, PAddr beg, PAddr end, bool add_elf_ext) {
+    Module module;
+    if (add_elf_ext) {
+        Common::SplitPath(name, nullptr, &module.name, nullptr);
+        module.name += ".elf";
+    } else {
+        module.name = std::move(name);
+    }
+    module.beg = beg;
+    module.end = end;
+    modules.push_back(std::move(module));
+}
+
 static Kernel::Thread* FindThreadById(int id) {
-    for (int core = 0; core < Core::NUM_CPU_CORES; core++) {
-        auto threads = Core::System::GetInstance().Scheduler(core)->GetThreadList();
-        for (auto thread : threads) {
+    for (u32 core = 0; core < Core::NUM_CPU_CORES; core++) {
+        const auto& threads = Core::System::GetInstance().Scheduler(core)->GetThreadList();
+        for (auto& thread : threads) {
             if (thread->GetThreadId() == id) {
-                current_thread = thread.get();
-                return current_thread;
+                current_core = core;
+                return thread.get();
             }
         }
     }
@@ -197,6 +227,8 @@ static u64 RegRead(int id, Kernel::Thread* thread = nullptr) {
         return thread->context.pc;
     } else if (id == CPSR_REGISTER) {
         return thread->context.cpsr;
+    } else if (id > CPSR_REGISTER && id < FPSCR_REGISTER) {
+        return thread->context.fpu_registers[id - UC_ARM64_REG_Q0][0];
     } else {
         return 0;
     }
@@ -215,6 +247,8 @@ static void RegWrite(int id, u64 val, Kernel::Thread* thread = nullptr) {
         thread->context.pc = val;
     } else if (id == CPSR_REGISTER) {
         thread->context.cpsr = val;
+    } else if (id > CPSR_REGISTER && id < FPSCR_REGISTER) {
+        thread->context.fpu_registers[id - (CPSR_REGISTER + 1)][0] = val;
     }
 }
 
@@ -534,7 +568,11 @@ static void HandleQuery() {
         SendReply("T0");
     } else if (strncmp(query, "Supported", strlen("Supported")) == 0) {
         // PacketSize needs to be large enough for target xml
-        SendReply("PacketSize=2000;qXfer:features:read+");
+        std::string buffer = "PacketSize=2000;qXfer:features:read+;qXfer:threads:read+";
+        if (!modules.empty()) {
+            buffer += ";qXfer:libraries:read+";
+        }
+        SendReply(buffer.c_str());
     } else if (strncmp(query, "Xfer:features:read:target.xml:",
                        strlen("Xfer:features:read:target.xml:")) == 0) {
         SendReply(target_xml);
@@ -543,9 +581,9 @@ static void HandleQuery() {
         SendReply(buffer.c_str());
     } else if (strncmp(query, "fThreadInfo", strlen("fThreadInfo")) == 0) {
         std::string val = "m";
-        for (int core = 0; core < Core::NUM_CPU_CORES; core++) {
-            auto threads = Core::System::GetInstance().Scheduler(core)->GetThreadList();
-            for (auto thread : threads) {
+        for (u32 core = 0; core < Core::NUM_CPU_CORES; core++) {
+            const auto& threads = Core::System::GetInstance().Scheduler(core)->GetThreadList();
+            for (const auto& thread : threads) {
                 val += fmt::format("{:x}", thread->GetThreadId());
                 val += ",";
             }
@@ -554,6 +592,31 @@ static void HandleQuery() {
         SendReply(val.c_str());
     } else if (strncmp(query, "sThreadInfo", strlen("sThreadInfo")) == 0) {
         SendReply("l");
+    } else if (strncmp(query, "Xfer:threads:read", strlen("Xfer:threads:read")) == 0) {
+        std::string buffer;
+        buffer += "l<?xml version=\"1.0\"?>";
+        buffer += "<threads>";
+        for (u32 core = 0; core < Core::NUM_CPU_CORES; core++) {
+            const auto& threads = Core::System::GetInstance().Scheduler(core)->GetThreadList();
+            for (const auto& thread : threads) {
+                buffer +=
+                    fmt::format(R"*(<thread id="{:x}" core="{:d}" name="Thread {:x}"></thread>)*",
+                                thread->GetThreadId(), core, thread->GetThreadId());
+            }
+        }
+        buffer += "</threads>";
+        SendReply(buffer.c_str());
+    } else if (strncmp(query, "Xfer:libraries:read", strlen("Xfer:libraries:read")) == 0) {
+        std::string buffer;
+        buffer += "l<?xml version=\"1.0\"?>";
+        buffer += "<library-list>";
+        for (const auto& module : modules) {
+            buffer +=
+                fmt::format(R"*("<library name = "{}"><segment address = "0x{:x}"/></library>)*",
+                            module.name, module.beg);
+        }
+        buffer += "</library-list>";
+        SendReply(buffer.c_str());
     } else {
         SendReply("");
     }
@@ -561,33 +624,27 @@ static void HandleQuery() {
 
 /// Handle set thread command from gdb client.
 static void HandleSetThread() {
-    if (memcmp(command_buffer, "Hc", 2) == 0 || memcmp(command_buffer, "Hg", 2) == 0) {
-        int thread_id = -1;
-        if (command_buffer[2] != '-') {
-            thread_id = static_cast<int>(HexToInt(
-                command_buffer + 2,
-                command_length - 2 /*strlen(reinterpret_cast<char*>(command_buffer) + 2)*/));
-        }
-        if (thread_id >= 1) {
-            current_thread = FindThreadById(thread_id);
-        }
-        if (!current_thread) {
-            thread_id = 1;
-            current_thread = FindThreadById(thread_id);
-        }
-        if (current_thread) {
-            SendReply("OK");
-            return;
-        }
+    int thread_id = -1;
+    if (command_buffer[2] != '-') {
+        thread_id = static_cast<int>(HexToInt(command_buffer + 2, command_length - 2));
+    }
+    if (thread_id >= 1) {
+        current_thread = FindThreadById(thread_id);
+    }
+    if (!current_thread) {
+        thread_id = 1;
+        current_thread = FindThreadById(thread_id);
+    }
+    if (current_thread) {
+        SendReply("OK");
+        return;
     }
     SendReply("E01");
 }
 
 /// Handle thread alive command from gdb client.
 static void HandleThreadAlive() {
-    int thread_id = static_cast<int>(
-        HexToInt(command_buffer + 1,
-                 command_length - 1 /*strlen(reinterpret_cast<char*>(command_buffer) + 1)*/));
+    int thread_id = static_cast<int>(HexToInt(command_buffer + 1, command_length - 1));
     if (thread_id == 0) {
         thread_id = 1;
     }
@@ -610,16 +667,23 @@ static void SendSignal(Kernel::Thread* thread, u32 signal, bool full = true) {
 
     latest_signal = signal;
 
-    std::string buffer;
-    if (full) {
-        buffer = fmt::format("T{:02x}{:02x}:{:016x};{:02x}:{:016x};", latest_signal, PC_REGISTER,
-                             Common::swap64(RegRead(PC_REGISTER, thread)), SP_REGISTER,
-                             Common::swap64(RegRead(SP_REGISTER, thread)));
-    } else {
-        buffer = fmt::format("T{:02x};", latest_signal);
+    if (!thread) {
+        full = false;
     }
 
-    buffer += fmt::format("thread:{:x};", thread->GetThreadId());
+    std::string buffer;
+    if (full) {
+        buffer = fmt::format("T{:02x}{:02x}:{:016x};{:02x}:{:016x};{:02x}:{:016x}", latest_signal,
+                             PC_REGISTER, Common::swap64(RegRead(PC_REGISTER, thread)), SP_REGISTER,
+                             Common::swap64(RegRead(SP_REGISTER, thread)), LR_REGISTER,
+                             Common::swap64(RegRead(LR_REGISTER, thread)));
+    } else {
+        buffer = fmt::format("T{:02x}", latest_signal);
+    }
+
+    if (thread) {
+        buffer += fmt::format(";thread:{:x};", thread->GetThreadId());
+    }
 
     SendReply(buffer.c_str());
 }
@@ -711,8 +775,12 @@ static void ReadRegister() {
         LongToGdbHex(reply, RegRead(id, current_thread));
     } else if (id == CPSR_REGISTER) {
         IntToGdbHex(reply, (u32)RegRead(id, current_thread));
+    } else if (id >= UC_ARM64_REG_Q0 && id < FPSCR_REGISTER) {
+        LongToGdbHex(reply, RegRead(id, current_thread));
+    } else if (id == FPSCR_REGISTER) {
+        LongToGdbHex(reply, RegRead(TODO_DUMMY_REG_998, current_thread));
     } else {
-        return SendReply("E01");
+        LongToGdbHex(reply, RegRead(TODO_DUMMY_REG_997, current_thread));
     }
 
     SendReply(reinterpret_cast<char*>(reply));
@@ -729,7 +797,7 @@ static void ReadRegisters() {
         LongToGdbHex(bufptr + reg * 16, RegRead(reg, current_thread));
     }
 
-    bufptr += (32 * 16);
+    bufptr += 32 * 16;
 
     LongToGdbHex(bufptr, RegRead(PC_REGISTER, current_thread));
 
@@ -739,6 +807,16 @@ static void ReadRegisters() {
 
     bufptr += 8;
 
+    for (int reg = UC_ARM64_REG_Q0; reg <= UC_ARM64_REG_Q0 + 31; reg++) {
+        LongToGdbHex(bufptr + reg * 16, RegRead(reg, current_thread));
+    }
+
+    bufptr += 32 * 32;
+
+    LongToGdbHex(bufptr, RegRead(TODO_DUMMY_REG_998, current_thread));
+
+    bufptr += 8;
+
     SendReply(reinterpret_cast<char*>(buffer));
 }
 
@@ -759,10 +837,17 @@ static void WriteRegister() {
         RegWrite(id, GdbHexToLong(buffer_ptr), current_thread);
     } else if (id == CPSR_REGISTER) {
         RegWrite(id, GdbHexToInt(buffer_ptr), current_thread);
+    } else if (id >= UC_ARM64_REG_Q0 && id < FPSCR_REGISTER) {
+        RegWrite(id, GdbHexToLong(buffer_ptr), current_thread);
+    } else if (id == FPSCR_REGISTER) {
+        RegWrite(TODO_DUMMY_REG_998, GdbHexToLong(buffer_ptr), current_thread);
     } else {
-        return SendReply("E01");
+        RegWrite(TODO_DUMMY_REG_997, GdbHexToLong(buffer_ptr), current_thread);
     }
 
+    // Update Unicorn context skipping scheduler, no running threads at this point
+    Core::System::GetInstance().ArmInterface(current_core).LoadContext(current_thread->context);
+
     SendReply("OK");
 }
 
@@ -773,18 +858,25 @@ static void WriteRegisters() {
     if (command_buffer[0] != 'G')
         return SendReply("E01");
 
-    for (int i = 0, reg = 0; reg <= CPSR_REGISTER; i++, reg++) {
+    for (int i = 0, reg = 0; reg <= FPSCR_REGISTER; i++, reg++) {
         if (reg <= SP_REGISTER) {
             RegWrite(reg, GdbHexToLong(buffer_ptr + i * 16), current_thread);
         } else if (reg == PC_REGISTER) {
             RegWrite(PC_REGISTER, GdbHexToLong(buffer_ptr + i * 16), current_thread);
         } else if (reg == CPSR_REGISTER) {
             RegWrite(CPSR_REGISTER, GdbHexToInt(buffer_ptr + i * 16), current_thread);
+        } else if (reg >= UC_ARM64_REG_Q0 && reg < FPSCR_REGISTER) {
+            RegWrite(reg, GdbHexToLong(buffer_ptr + i * 16), current_thread);
+        } else if (reg == FPSCR_REGISTER) {
+            RegWrite(TODO_DUMMY_REG_998, GdbHexToLong(buffer_ptr + i * 16), current_thread);
         } else {
             UNIMPLEMENTED();
         }
     }
 
+    // Update Unicorn context skipping scheduler, no running threads at this point
+    Core::System::GetInstance().ArmInterface(current_core).LoadContext(current_thread->context);
+
     SendReply("OK");
 }
 
@@ -806,6 +898,10 @@ static void ReadMemory() {
         SendReply("E01");
     }
 
+    if (addr < Memory::PROCESS_IMAGE_VADDR || addr >= Memory::MAP_REGION_VADDR_END) {
+        return SendReply("E00");
+    }
+
     if (!Memory::IsValidVirtualAddress(addr)) {
         return SendReply("E00");
     }
@@ -840,16 +936,18 @@ static void WriteMemory() {
 }
 
 void Break(bool is_memory_break) {
-    if (!halt_loop) {
-        halt_loop = true;
-        send_trap = true;
-    }
+    send_trap = true;
 
     memory_break = is_memory_break;
 }
 
 /// Tell the CPU that it should perform a single step.
 static void Step() {
+    if (command_length > 1) {
+        RegWrite(PC_REGISTER, GdbHexToLong(command_buffer + 1), current_thread);
+        // Update Unicorn context skipping scheduler, no running threads at this point
+        Core::System::GetInstance().ArmInterface(current_core).LoadContext(current_thread->context);
+    }
     step_loop = true;
     halt_loop = true;
     send_trap = true;
@@ -1090,6 +1188,8 @@ static void Init(u16 port) {
     breakpoints_read.clear();
     breakpoints_write.clear();
 
+    modules.clear();
+
     // Start gdb server
     LOG_INFO(Debug_GDBStub, "Starting GDB server on port {}...", port);
 
@@ -1192,8 +1292,12 @@ void SetCpuStepFlag(bool is_step) {
 
 void SendTrap(Kernel::Thread* thread, int trap) {
     if (send_trap) {
+        if (!halt_loop || current_thread == thread) {
+            current_thread = thread;
+            SendSignal(thread, trap);
+        }
+        halt_loop = true;
         send_trap = false;
-        SendSignal(thread, trap);
     }
 }
 }; // namespace GDBStub

src/core/gdbstub/gdbstub.h

@@ -6,6 +6,7 @@
 
 #pragma once
 
+#include <string>
 #include "common/common_types.h"
 #include "core/hle/kernel/thread.h"
 
@@ -51,6 +52,9 @@ bool IsServerEnabled();
 /// Returns true if there is an active socket connection.
 bool IsConnected();
 
+/// Register module.
+void RegisterModule(std::string name, PAddr beg, PAddr end, bool add_elf_ext = true);
+
 /**
  * Signal to the gdbstub server that it should halt CPU execution.
  *
@@ -80,10 +84,10 @@ BreakpointAddress GetNextBreakpointFromAddress(PAddr addr, GDBStub::BreakpointTy
  */
 bool CheckBreakpoint(PAddr addr, GDBStub::BreakpointType type);
 
-// If set to true, the CPU will halt at the beginning of the next CPU loop.
+/// If set to true, the CPU will halt at the beginning of the next CPU loop.
 bool GetCpuHaltFlag();
 
-// If set to true and the CPU is halted, the CPU will step one instruction.
+/// If set to true and the CPU is halted, the CPU will step one instruction.
 bool GetCpuStepFlag();
 
 /**

src/core/hle/kernel/address_arbiter.cpp

@@ -115,7 +115,7 @@ ResultCode ModifyByWaitingCountAndSignalToAddressIfEqual(VAddr address, s32 valu
     s32 updated_value;
     if (waiting_threads.size() == 0) {
         updated_value = value - 1;
-    } else if (num_to_wake <= 0 || waiting_threads.size() <= num_to_wake) {
+    } else if (num_to_wake <= 0 || waiting_threads.size() <= static_cast<u32>(num_to_wake)) {
         updated_value = value + 1;
     } else {
         updated_value = value;
@@ -140,7 +140,9 @@ ResultCode WaitForAddressIfLessThan(VAddr address, s32 value, s64 timeout, bool
 
     s32 cur_value = static_cast<s32>(Memory::Read32(address));
     if (cur_value < value) {
-        Memory::Write32(address, static_cast<u32>(cur_value - 1));
+        if (should_decrement) {
+            Memory::Write32(address, static_cast<u32>(cur_value - 1));
+        }
     } else {
         return ERR_INVALID_STATE;
     }

src/core/hle/kernel/hle_ipc.cpp

@@ -29,7 +29,8 @@ void SessionRequestHandler::ClientDisconnected(SharedPtr<ServerSession> server_s
 
 SharedPtr<Event> HLERequestContext::SleepClientThread(SharedPtr<Thread> thread,
                                                       const std::string& reason, u64 timeout,
-                                                      WakeupCallback&& callback) {
+                                                      WakeupCallback&& callback,
+                                                      Kernel::SharedPtr<Kernel::Event> event) {
 
     // Put the client thread to sleep until the wait event is signaled or the timeout expires.
     thread->wakeup_callback =
@@ -41,7 +42,12 @@ SharedPtr<Event> HLERequestContext::SleepClientThread(SharedPtr<Thread> thread,
         return true;
     };
 
-    auto event = Kernel::Event::Create(Kernel::ResetType::OneShot, "HLE Pause Event: " + reason);
+    if (!event) {
+        // Create event if not provided
+        event = Kernel::Event::Create(Kernel::ResetType::OneShot, "HLE Pause Event: " + reason);
+    }
+
+    event->Clear();
     thread->status = THREADSTATUS_WAIT_HLE_EVENT;
     thread->wait_objects = {event};
     event->AddWaitingThread(thread);
@@ -214,8 +220,8 @@ ResultCode HLERequestContext::WriteToOutgoingCommandBuffer(Thread& thread) {
             (sizeof(IPC::CommandHeader) + sizeof(IPC::HandleDescriptorHeader)) / sizeof(u32);
         ASSERT_MSG(!handle_descriptor_header->send_current_pid, "Sending PID is not implemented");
 
-        ASSERT_MSG(copy_objects.size() == handle_descriptor_header->num_handles_to_copy);
-        ASSERT_MSG(move_objects.size() == handle_descriptor_header->num_handles_to_move);
+        ASSERT(copy_objects.size() == handle_descriptor_header->num_handles_to_copy);
+        ASSERT(move_objects.size() == handle_descriptor_header->num_handles_to_move);
 
         // We don't make a distinction between copy and move handles when translating since HLE
         // services don't deal with handles directly. However, the guest applications might check

src/core/hle/kernel/hle_ipc.h

@@ -118,10 +118,12 @@ public:
      * @param callback Callback to be invoked when the thread is resumed. This callback must write
      * the entire command response once again, regardless of the state of it before this function
      * was called.
+     * @param event Event to use to wake up the thread. If unspecified, an event will be created.
      * @returns Event that when signaled will resume the thread and call the callback function.
      */
     SharedPtr<Event> SleepClientThread(SharedPtr<Thread> thread, const std::string& reason,
-                                       u64 timeout, WakeupCallback&& callback);
+                                       u64 timeout, WakeupCallback&& callback,
+                                       Kernel::SharedPtr<Kernel::Event> event = nullptr);
 
     void ParseCommandBuffer(u32_le* src_cmdbuf, bool incoming);
 

src/core/hle/kernel/scheduler.cpp

@@ -85,6 +85,7 @@ void Scheduler::SwitchContext(Thread* new_thread) {
 
         cpu_core->LoadContext(new_thread->context);
         cpu_core->SetTlsAddress(new_thread->GetTLSAddress());
+        cpu_core->ClearExclusiveState();
     } else {
         current_thread = nullptr;
         // Note: We do not reset the current process and current page table when idling because

src/core/hle/kernel/vm_manager.cpp

@@ -16,30 +16,20 @@
 namespace Kernel {
 
 static const char* GetMemoryStateName(MemoryState state) {
-    static const char* names[] = {
-        "Unmapped",
-        "Io",
-        "Normal",
-        "CodeStatic",
-        "CodeMutable",
-        "Heap",
-        "Shared",
-        "Unknown1"
-        "ModuleCodeStatic",
-        "ModuleCodeMutable",
-        "IpcBuffer0",
-        "Mapped",
-        "ThreadLocal",
-        "TransferMemoryIsolated",
-        "TransferMemory",
-        "ProcessMemory",
-        "Unknown2"
-        "IpcBuffer1",
-        "IpcBuffer3",
-        "KernelStack",
+    static constexpr const char* names[] = {
+        "Unmapped",         "Io",
+        "Normal",           "CodeStatic",
+        "CodeMutable",      "Heap",
+        "Shared",           "Unknown1",
+        "ModuleCodeStatic", "ModuleCodeMutable",
+        "IpcBuffer0",       "Mapped",
+        "ThreadLocal",      "TransferMemoryIsolated",
+        "TransferMemory",   "ProcessMemory",
+        "Unknown2",         "IpcBuffer1",
+        "IpcBuffer3",       "KernelStack",
     };
 
-    return names[(int)state];
+    return names[static_cast<int>(state)];
 }
 
 bool VirtualMemoryArea::CanBeMergedWith(const VirtualMemoryArea& next) const {

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

@@ -4,9 +4,11 @@
 
 #include <cinttypes>
 #include <stack>
+#include "core/core.h"
 #include "core/file_sys/filesystem.h"
 #include "core/hle/ipc_helpers.h"
 #include "core/hle/kernel/event.h"
+#include "core/hle/kernel/process.h"
 #include "core/hle/service/am/am.h"
 #include "core/hle/service/am/applet_ae.h"
 #include "core/hle/service/am/applet_oe.h"
@@ -614,25 +616,14 @@ void IApplicationFunctions::CreateApplicationAndRequestToStartForQuest(
 
 void IApplicationFunctions::EnsureSaveData(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
-    u128 uid = rp.PopRaw<u128>();
+    u128 uid = rp.PopRaw<u128>(); // What does this do?
 
     LOG_WARNING(Service, "(STUBBED) called uid = {:016X}{:016X}", uid[1], uid[0]);
 
     IPC::ResponseBuilder rb{ctx, 4};
-
-    FileSys::Path unused;
-    auto savedata = FileSystem::OpenFileSystem(FileSystem::Type::SaveData, unused);
-    if (savedata.Failed()) {
-        // Create the save data and return an error indicating that the operation was performed.
-        FileSystem::FormatFileSystem(FileSystem::Type::SaveData);
-        // TODO(Subv): Find out the correct error code for this.
-        rb.Push(ResultCode(ErrorModule::FS, 40));
-    } else {
-        rb.Push(RESULT_SUCCESS);
-    }
-
+    rb.Push(RESULT_SUCCESS);
     rb.Push<u64>(0);
-}
+} // namespace Service::AM
 
 void IApplicationFunctions::SetTerminateResult(Kernel::HLERequestContext& ctx) {
     // Takes an input u32 Result, no output.

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

@@ -4,6 +4,7 @@
 
 #include <boost/container/flat_map.hpp>
 #include "common/file_util.h"
+#include "core/file_sys/errors.h"
 #include "core/file_sys/filesystem.h"
 #include "core/file_sys/savedata_factory.h"
 #include "core/file_sys/sdmc_factory.h"
@@ -16,57 +17,77 @@ namespace Service::FileSystem {
  * Map of registered file systems, identified by type. Once an file system is registered here, it
  * is never removed until UnregisterFileSystems is called.
  */
-static boost::container::flat_map<Type, std::unique_ptr<FileSys::FileSystemFactory>> filesystem_map;
+static std::unique_ptr<FileSys::RomFSFactory> romfs_factory;
+static std::unique_ptr<FileSys::SaveDataFactory> save_data_factory;
+static std::unique_ptr<FileSys::SDMCFactory> sdmc_factory;
 
-ResultCode RegisterFileSystem(std::unique_ptr<FileSys::FileSystemFactory>&& factory, Type type) {
-    auto result = filesystem_map.emplace(type, std::move(factory));
-
-    bool inserted = result.second;
-    ASSERT_MSG(inserted, "Tried to register more than one system with same id code");
-
-    auto& filesystem = result.first->second;
-    LOG_DEBUG(Service_FS, "Registered file system {} with id code 0x{:08X}", filesystem->GetName(),
-              static_cast<u32>(type));
+ResultCode RegisterRomFS(std::unique_ptr<FileSys::RomFSFactory>&& factory) {
+    ASSERT_MSG(romfs_factory == nullptr, "Tried to register a second RomFS");
+    romfs_factory = std::move(factory);
+    LOG_DEBUG(Service_FS, "Registered RomFS");
     return RESULT_SUCCESS;
 }
 
-ResultVal<std::unique_ptr<FileSys::FileSystemBackend>> OpenFileSystem(Type type,
-                                                                      FileSys::Path& path) {
-    LOG_TRACE(Service_FS, "Opening FileSystem with type={}", static_cast<u32>(type));
-
-    auto itr = filesystem_map.find(type);
-    if (itr == filesystem_map.end()) {
-        // TODO(bunnei): Find a better error code for this
-        return ResultCode(-1);
-    }
-
-    return itr->second->Open(path);
+ResultCode RegisterSaveData(std::unique_ptr<FileSys::SaveDataFactory>&& factory) {
+    ASSERT_MSG(romfs_factory == nullptr, "Tried to register a second save data");
+    save_data_factory = std::move(factory);
+    LOG_DEBUG(Service_FS, "Registered save data");
+    return RESULT_SUCCESS;
 }
 
-ResultCode FormatFileSystem(Type type) {
-    LOG_TRACE(Service_FS, "Formatting FileSystem with type={}", static_cast<u32>(type));
+ResultCode RegisterSDMC(std::unique_ptr<FileSys::SDMCFactory>&& factory) {
+    ASSERT_MSG(sdmc_factory == nullptr, "Tried to register a second SDMC");
+    sdmc_factory = std::move(factory);
+    LOG_DEBUG(Service_FS, "Registered SDMC");
+    return RESULT_SUCCESS;
+}
 
-    auto itr = filesystem_map.find(type);
-    if (itr == filesystem_map.end()) {
+ResultVal<std::unique_ptr<FileSys::FileSystemBackend>> OpenRomFS(u64 title_id) {
+    LOG_TRACE(Service_FS, "Opening RomFS for title_id={:016X}", title_id);
+
+    if (romfs_factory == nullptr) {
         // TODO(bunnei): Find a better error code for this
         return ResultCode(-1);
     }
 
-    FileSys::Path unused;
-    return itr->second->Format(unused);
+    return romfs_factory->Open(title_id);
+}
+
+ResultVal<std::unique_ptr<FileSys::FileSystemBackend>> OpenSaveData(
+    FileSys::SaveDataSpaceId space, FileSys::SaveDataDescriptor save_struct) {
+    LOG_TRACE(Service_FS, "Opening Save Data for space_id={:01X}, save_struct={}",
+              static_cast<u8>(space), SaveStructDebugInfo(save_struct));
+
+    if (save_data_factory == nullptr) {
+        return ResultCode(ErrorModule::FS, FileSys::ErrCodes::SaveDataNotFound);
+    }
+
+    return save_data_factory->Open(space, save_struct);
+}
+
+ResultVal<std::unique_ptr<FileSys::FileSystemBackend>> OpenSDMC() {
+    LOG_TRACE(Service_FS, "Opening SDMC");
+
+    if (sdmc_factory == nullptr) {
+        return ResultCode(ErrorModule::FS, FileSys::ErrCodes::SdCardNotFound);
+    }
+
+    return sdmc_factory->Open();
 }
 
 void RegisterFileSystems() {
-    filesystem_map.clear();
+    romfs_factory = nullptr;
+    save_data_factory = nullptr;
+    sdmc_factory = nullptr;
 
     std::string nand_directory = FileUtil::GetUserPath(D_NAND_IDX);
     std::string sd_directory = FileUtil::GetUserPath(D_SDMC_IDX);
 
-    auto savedata = std::make_unique<FileSys::SaveData_Factory>(std::move(nand_directory));
-    RegisterFileSystem(std::move(savedata), Type::SaveData);
+    auto savedata = std::make_unique<FileSys::SaveDataFactory>(std::move(nand_directory));
+    save_data_factory = std::move(savedata);
 
-    auto sdcard = std::make_unique<FileSys::SDMC_Factory>(std::move(sd_directory));
-    RegisterFileSystem(std::move(sdcard), Type::SDMC);
+    auto sdcard = std::make_unique<FileSys::SDMCFactory>(std::move(sd_directory));
+    sdmc_factory = std::move(sdcard);
 }
 
 void InstallInterfaces(SM::ServiceManager& service_manager) {

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

@@ -6,12 +6,13 @@
 
 #include <memory>
 #include "common/common_types.h"
+#include "core/file_sys/romfs_factory.h"
+#include "core/file_sys/savedata_factory.h"
+#include "core/file_sys/sdmc_factory.h"
 #include "core/hle/result.h"
 
 namespace FileSys {
 class FileSystemBackend;
-class FileSystemFactory;
-class Path;
 } // namespace FileSys
 
 namespace Service {
@@ -22,35 +23,20 @@ class ServiceManager;
 
 namespace FileSystem {
 
-/// Supported FileSystem types
-enum class Type {
-    RomFS = 1,
-    SaveData = 2,
-    SDMC = 3,
-};
+ResultCode RegisterRomFS(std::unique_ptr<FileSys::RomFSFactory>&& factory);
+ResultCode RegisterSaveData(std::unique_ptr<FileSys::SaveDataFactory>&& factory);
+ResultCode RegisterSDMC(std::unique_ptr<FileSys::SDMCFactory>&& factory);
 
-/**
- * Registers a FileSystem, instances of which can later be opened using its IdCode.
- * @param factory FileSystem backend interface to use
- * @param type Type used to access this type of FileSystem
- */
-ResultCode RegisterFileSystem(std::unique_ptr<FileSys::FileSystemFactory>&& factory, Type type);
+// TODO(DarkLordZach): BIS Filesystem
+// ResultCode RegisterBIS(std::unique_ptr<FileSys::BISFactory>&& factory);
 
-/**
- * Opens a file system
- * @param type Type of the file system to open
- * @param path Path to the file system, used with Binary paths
- * @return FileSys::FileSystemBackend interface to the file system
- */
-ResultVal<std::unique_ptr<FileSys::FileSystemBackend>> OpenFileSystem(Type type,
-                                                                      FileSys::Path& path);
+ResultVal<std::unique_ptr<FileSys::FileSystemBackend>> OpenRomFS(u64 title_id);
+ResultVal<std::unique_ptr<FileSys::FileSystemBackend>> OpenSaveData(
+    FileSys::SaveDataSpaceId space, FileSys::SaveDataDescriptor save_struct);
+ResultVal<std::unique_ptr<FileSys::FileSystemBackend>> OpenSDMC();
 
-/**
- * Formats a file system
- * @param type Type of the file system to format
- * @return ResultCode of the operation
- */
-ResultCode FormatFileSystem(Type type);
+// TODO(DarkLordZach): BIS Filesystem
+// ResultVal<std::unique_ptr<FileSys::FileSystemBackend>> OpenBIS();
 
 /// Registers all Filesystem services with the specified service manager.
 void InstallInterfaces(SM::ServiceManager& service_manager);

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

@@ -13,11 +13,21 @@
 #include "core/hle/ipc_helpers.h"
 #include "core/hle/kernel/client_port.h"
 #include "core/hle/kernel/client_session.h"
+#include "core/hle/kernel/process.h"
 #include "core/hle/service/filesystem/filesystem.h"
 #include "core/hle/service/filesystem/fsp_srv.h"
 
 namespace Service::FileSystem {
 
+enum class StorageId : u8 {
+    None = 0,
+    Host = 1,
+    GameCard = 2,
+    NandSystem = 3,
+    NandUser = 4,
+    SdCard = 5
+};
+
 class IStorage final : public ServiceFramework<IStorage> {
 public:
     IStorage(std::unique_ptr<FileSys::StorageBackend>&& backend)
@@ -487,17 +497,6 @@ FSP_SRV::FSP_SRV() : ServiceFramework("fsp-srv") {
     RegisterHandlers(functions);
 }
 
-void FSP_SRV::TryLoadRomFS() {
-    if (romfs) {
-        return;
-    }
-    FileSys::Path unused;
-    auto res = OpenFileSystem(Type::RomFS, unused);
-    if (res.Succeeded()) {
-        romfs = std::move(res.Unwrap());
-    }
-}
-
 void FSP_SRV::Initialize(Kernel::HLERequestContext& ctx) {
     LOG_WARNING(Service_FS, "(STUBBED) called");
 
@@ -508,8 +507,7 @@ void FSP_SRV::Initialize(Kernel::HLERequestContext& ctx) {
 void FSP_SRV::MountSdCard(Kernel::HLERequestContext& ctx) {
     LOG_DEBUG(Service_FS, "called");
 
-    FileSys::Path unused;
-    auto filesystem = OpenFileSystem(Type::SDMC, unused).Unwrap();
+    IFileSystem filesystem(OpenSDMC().Unwrap());
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
@@ -519,23 +517,26 @@ void FSP_SRV::MountSdCard(Kernel::HLERequestContext& ctx) {
 void FSP_SRV::CreateSaveData(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
 
-    auto save_struct = rp.PopRaw<std::array<u8, 0x40>>();
+    auto save_struct = rp.PopRaw<FileSys::SaveDataDescriptor>();
     auto save_create_struct = rp.PopRaw<std::array<u8, 0x40>>();
     u128 uid = rp.PopRaw<u128>();
 
-    LOG_WARNING(Service_FS, "(STUBBED) called uid = {:016X}{:016X}", uid[1], uid[0]);
+    LOG_WARNING(Service_FS, "(STUBBED) called save_struct = {}, uid = {:016X}{:016X}",
+                save_struct.DebugInfo(), uid[1], uid[0]);
 
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
 }
 
 void FSP_SRV::MountSaveData(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_FS, "(STUBBED) called");
+    IPC::RequestParser rp{ctx};
 
-    // TODO(Subv): Read the input parameters and mount the requested savedata instead of always
-    // mounting the current process' savedata.
-    FileSys::Path unused;
-    auto filesystem = OpenFileSystem(Type::SaveData, unused);
+    auto space_id = rp.PopRaw<FileSys::SaveDataSpaceId>();
+    auto unk = rp.Pop<u32>();
+    LOG_INFO(Service_FS, "called with unknown={:08X}", unk);
+    auto save_struct = rp.PopRaw<FileSys::SaveDataDescriptor>();
+
+    auto filesystem = OpenSaveData(space_id, save_struct);
 
     if (filesystem.Failed()) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 0};
@@ -559,8 +560,8 @@ void FSP_SRV::GetGlobalAccessLogMode(Kernel::HLERequestContext& ctx) {
 void FSP_SRV::OpenDataStorageByCurrentProcess(Kernel::HLERequestContext& ctx) {
     LOG_DEBUG(Service_FS, "called");
 
-    TryLoadRomFS();
-    if (!romfs) {
+    auto romfs = OpenRomFS(Core::System::GetInstance().CurrentProcess()->program_id);
+    if (romfs.Failed()) {
         // TODO (bunnei): Find the right error code to use here
         LOG_CRITICAL(Service_FS, "no file system interface available!");
         IPC::ResponseBuilder rb{ctx, 2};
@@ -568,8 +569,8 @@ void FSP_SRV::OpenDataStorageByCurrentProcess(Kernel::HLERequestContext& ctx) {
         return;
     }
 
-    // Attempt to open a StorageBackend interface to the RomFS
-    auto storage = romfs->OpenFile({}, {});
+    auto storage = romfs.Unwrap()->OpenFile({}, {});
+
     if (storage.Failed()) {
         LOG_CRITICAL(Service_FS, "no storage interface available!");
         IPC::ResponseBuilder rb{ctx, 2};
@@ -583,8 +584,40 @@ void FSP_SRV::OpenDataStorageByCurrentProcess(Kernel::HLERequestContext& ctx) {
 }
 
 void FSP_SRV::OpenRomStorage(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_FS, "(STUBBED) called, using OpenDataStorageByCurrentProcess");
-    OpenDataStorageByCurrentProcess(ctx);
+    IPC::RequestParser rp{ctx};
+
+    auto storage_id = rp.PopRaw<StorageId>();
+    auto title_id = rp.PopRaw<u64>();
+
+    LOG_DEBUG(Service_FS, "called with storage_id={:02X}, title_id={:016X}",
+              static_cast<u8>(storage_id), title_id);
+    if (title_id != Core::System::GetInstance().CurrentProcess()->program_id) {
+        LOG_CRITICAL(
+            Service_FS,
+            "Attempting to access RomFS of another title id (current={:016X}, requested={:016X}).",
+            Core::System::GetInstance().CurrentProcess()->program_id, title_id);
+    }
+
+    auto romfs = OpenRomFS(title_id);
+    if (romfs.Failed()) {
+        LOG_CRITICAL(Service_FS, "no file system interface available!");
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(ResultCode(ErrorModule::FS, FileSys::ErrCodes::RomFSNotFound));
+        return;
+    }
+
+    auto storage = romfs.Unwrap()->OpenFile({}, {});
+
+    if (storage.Failed()) {
+        LOG_CRITICAL(Service_FS, "no storage interface available!");
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(storage.Code());
+        return;
+    }
+
+    IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+    rb.Push(RESULT_SUCCESS);
+    rb.PushIpcInterface<IStorage>(std::move(storage.Unwrap()));
 }
 
 } // namespace Service::FileSystem

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

@@ -19,8 +19,6 @@ public:
     ~FSP_SRV() = default;
 
 private:
-    void TryLoadRomFS();
-
     void Initialize(Kernel::HLERequestContext& ctx);
     void MountSdCard(Kernel::HLERequestContext& ctx);
     void CreateSaveData(Kernel::HLERequestContext& ctx);

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

@@ -18,9 +18,9 @@ namespace Service::HID {
 
 // Updating period for each HID device.
 // TODO(shinyquagsire23): These need better values.
-constexpr u64 pad_update_ticks = CoreTiming::BASE_CLOCK_RATE / 10000;
-constexpr u64 accelerometer_update_ticks = CoreTiming::BASE_CLOCK_RATE / 10000;
-constexpr u64 gyroscope_update_ticks = CoreTiming::BASE_CLOCK_RATE / 10000;
+constexpr u64 pad_update_ticks = CoreTiming::BASE_CLOCK_RATE / 100;
+constexpr u64 accelerometer_update_ticks = CoreTiming::BASE_CLOCK_RATE / 100;
+constexpr u64 gyroscope_update_ticks = CoreTiming::BASE_CLOCK_RATE / 100;
 
 class IAppletResource final : public ServiceFramework<IAppletResource> {
 public:

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

@@ -18,7 +18,8 @@ u32 nvdisp_disp0::ioctl(Ioctl command, const std::vector<u8>& input, std::vector
 }
 
 void nvdisp_disp0::flip(u32 buffer_handle, u32 offset, u32 format, u32 width, u32 height,
-                        u32 stride, NVFlinger::BufferQueue::BufferTransformFlags transform) {
+                        u32 stride, NVFlinger::BufferQueue::BufferTransformFlags transform,
+                        const MathUtil::Rectangle<int>& crop_rect) {
     VAddr addr = nvmap_dev->GetObjectAddress(buffer_handle);
     LOG_WARNING(Service,
                 "Drawing from address {:X} offset {:08X} Width {} Height {} Stride {} Format {}",
@@ -26,7 +27,8 @@ void nvdisp_disp0::flip(u32 buffer_handle, u32 offset, u32 format, u32 width, u3
 
     using PixelFormat = Tegra::FramebufferConfig::PixelFormat;
     const Tegra::FramebufferConfig framebuffer{
-        addr, offset, width, height, stride, static_cast<PixelFormat>(format), transform};
+        addr,      offset,   width, height, stride, static_cast<PixelFormat>(format),
+        transform, crop_rect};
 
     Core::System::GetInstance().perf_stats.EndGameFrame();
 

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

@@ -7,6 +7,7 @@
 #include <memory>
 #include <vector>
 #include "common/common_types.h"
+#include "common/math_util.h"
 #include "core/hle/service/nvdrv/devices/nvdevice.h"
 #include "core/hle/service/nvflinger/buffer_queue.h"
 
@@ -23,7 +24,8 @@ public:
 
     /// Performs a screen flip, drawing the buffer pointed to by the handle.
     void flip(u32 buffer_handle, u32 offset, u32 format, u32 width, u32 height, u32 stride,
-              NVFlinger::BufferQueue::BufferTransformFlags transform);
+              NVFlinger::BufferQueue::BufferTransformFlags transform,
+              const MathUtil::Rectangle<int>& crop_rect);
 
 private:
     std::shared_ptr<nvmap> nvmap_dev;

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

@@ -13,8 +13,8 @@ namespace Service {
 namespace NVFlinger {
 
 BufferQueue::BufferQueue(u32 id, u64 layer_id) : id(id), layer_id(layer_id) {
-    native_handle = Kernel::Event::Create(Kernel::ResetType::OneShot, "BufferQueue NativeHandle");
-    native_handle->Signal();
+    buffer_wait_event =
+        Kernel::Event::Create(Kernel::ResetType::Sticky, "BufferQueue NativeHandle");
 }
 
 void BufferQueue::SetPreallocatedBuffer(u32 slot, IGBPBuffer& igbp_buffer) {
@@ -26,10 +26,7 @@ void BufferQueue::SetPreallocatedBuffer(u32 slot, IGBPBuffer& igbp_buffer) {
     LOG_WARNING(Service, "Adding graphics buffer {}", slot);
 
     queue.emplace_back(buffer);
-
-    if (buffer_wait_event) {
-        buffer_wait_event->Signal();
-    }
+    buffer_wait_event->Signal();
 }
 
 boost::optional<u32> BufferQueue::DequeueBuffer(u32 width, u32 height) {
@@ -48,8 +45,6 @@ boost::optional<u32> BufferQueue::DequeueBuffer(u32 width, u32 height) {
         return boost::none;
     }
 
-    buffer_wait_event = nullptr;
-
     itr->status = Buffer::Status::Dequeued;
     return itr->slot;
 }
@@ -62,13 +57,15 @@ const IGBPBuffer& BufferQueue::RequestBuffer(u32 slot) const {
     return itr->igbp_buffer;
 }
 
-void BufferQueue::QueueBuffer(u32 slot, BufferTransformFlags transform) {
+void BufferQueue::QueueBuffer(u32 slot, BufferTransformFlags transform,
+                              const MathUtil::Rectangle<int>& crop_rect) {
     auto itr = std::find_if(queue.begin(), queue.end(),
                             [&](const Buffer& buffer) { return buffer.slot == slot; });
     ASSERT(itr != queue.end());
     ASSERT(itr->status == Buffer::Status::Dequeued);
     itr->status = Buffer::Status::Queued;
     itr->transform = transform;
+    itr->crop_rect = crop_rect;
 }
 
 boost::optional<const BufferQueue::Buffer&> BufferQueue::AcquireBuffer() {
@@ -88,9 +85,7 @@ void BufferQueue::ReleaseBuffer(u32 slot) {
     ASSERT(itr->status == Buffer::Status::Acquired);
     itr->status = Buffer::Status::Free;
 
-    if (buffer_wait_event) {
-        buffer_wait_event->Signal();
-    }
+    buffer_wait_event->Signal();
 }
 
 u32 BufferQueue::Query(QueryType type) {
@@ -106,10 +101,5 @@ u32 BufferQueue::Query(QueryType type) {
     return 0;
 }
 
-void BufferQueue::SetBufferWaitEvent(Kernel::SharedPtr<Kernel::Event>&& wait_event) {
-    ASSERT_MSG(!buffer_wait_event, "buffer_wait_event only supports a single waiting thread!");
-    buffer_wait_event = std::move(wait_event);
-}
-
 } // namespace NVFlinger
 } // namespace Service

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

@@ -6,6 +6,7 @@
 
 #include <vector>
 #include <boost/optional.hpp>
+#include "common/math_util.h"
 #include "common/swap.h"
 #include "core/hle/kernel/event.h"
 
@@ -68,23 +69,24 @@ public:
         Status status = Status::Free;
         IGBPBuffer igbp_buffer;
         BufferTransformFlags transform;
+        MathUtil::Rectangle<int> crop_rect;
     };
 
     void SetPreallocatedBuffer(u32 slot, IGBPBuffer& buffer);
     boost::optional<u32> DequeueBuffer(u32 width, u32 height);
     const IGBPBuffer& RequestBuffer(u32 slot) const;
-    void QueueBuffer(u32 slot, BufferTransformFlags transform);
+    void QueueBuffer(u32 slot, BufferTransformFlags transform,
+                     const MathUtil::Rectangle<int>& crop_rect);
     boost::optional<const Buffer&> AcquireBuffer();
     void ReleaseBuffer(u32 slot);
     u32 Query(QueryType type);
-    void SetBufferWaitEvent(Kernel::SharedPtr<Kernel::Event>&& wait_event);
 
     u32 GetId() const {
         return id;
     }
 
-    Kernel::SharedPtr<Kernel::Event> GetNativeHandle() const {
-        return native_handle;
+    Kernel::SharedPtr<Kernel::Event> GetBufferWaitEvent() const {
+        return buffer_wait_event;
     }
 
 private:
@@ -92,9 +94,6 @@ private:
     u64 layer_id;
 
     std::vector<Buffer> queue;
-    Kernel::SharedPtr<Kernel::Event> native_handle;
-
-    /// Used to signal waiting thread when no buffers are available
     Kernel::SharedPtr<Kernel::Event> buffer_wait_event;
 };
 

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

@@ -149,12 +149,10 @@ void NVFlinger::Compose() {
         ASSERT(nvdisp);
 
         nvdisp->flip(igbp_buffer.gpu_buffer_id, igbp_buffer.offset, igbp_buffer.format,
-                     igbp_buffer.width, igbp_buffer.height, igbp_buffer.stride, buffer->transform);
+                     igbp_buffer.width, igbp_buffer.height, igbp_buffer.stride, buffer->transform,
+                     buffer->crop_rect);
 
         buffer_queue->ReleaseBuffer(buffer->slot);
-
-        // TODO(Subv): Figure out when we should actually signal this event.
-        buffer_queue->GetNativeHandle()->Signal();
     }
 }
 

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

@@ -19,10 +19,9 @@ void BSD::RegisterClient(Kernel::HLERequestContext& ctx) {
 void BSD::StartMonitoring(Kernel::HLERequestContext& ctx) {
     LOG_WARNING(Service, "(STUBBED) called");
 
-    IPC::ResponseBuilder rb{ctx, 3};
+    IPC::ResponseBuilder rb{ctx, 2};
 
     rb.Push(RESULT_SUCCESS);
-    rb.Push<u32>(0); // bsd errno
 }
 
 void BSD::Socket(Kernel::HLERequestContext& ctx) {

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

@@ -7,6 +7,7 @@
 #include <memory>
 #include <boost/optional.hpp>
 #include "common/alignment.h"
+#include "common/math_util.h"
 #include "common/scope_exit.h"
 #include "core/core_timing.h"
 #include "core/hle/ipc_helpers.h"
@@ -27,8 +28,8 @@ struct DisplayInfo {
     char display_name[0x40]{"Default"};
     u64 unknown_1{1};
     u64 unknown_2{1};
-    u64 width{1920};
-    u64 height{1080};
+    u64 width{1280};
+    u64 height{720};
 };
 static_assert(sizeof(DisplayInfo) == 0x60, "DisplayInfo has wrong size");
 
@@ -327,8 +328,8 @@ public:
 
 protected:
     void SerializeData() override {
-        // TODO(Subv): Figure out what this value means, writing non-zero here will make libnx try
-        // to read an IGBPBuffer object from the parcel.
+        // TODO(Subv): Figure out what this value means, writing non-zero here will make libnx
+        // try to read an IGBPBuffer object from the parcel.
         Write<u32_le>(1);
         WriteObject(buffer);
         Write<u32_le>(0);
@@ -360,8 +361,8 @@ public:
         INSERT_PADDING_WORDS(3);
         u32_le timestamp;
         s32_le is_auto_timestamp;
-        s32_le crop_left;
         s32_le crop_top;
+        s32_le crop_left;
         s32_le crop_right;
         s32_le crop_bottom;
         s32_le scaling_mode;
@@ -370,6 +371,10 @@ public:
         INSERT_PADDING_WORDS(2);
         u32_le fence_is_valid;
         std::array<Fence, 2> fences;
+
+        MathUtil::Rectangle<int> GetCropRect() const {
+            return {crop_left, crop_top, crop_right, crop_bottom};
+        }
     };
     static_assert(sizeof(Data) == 80, "ParcelData has wrong size");
 
@@ -495,7 +500,7 @@ private:
                 ctx.WriteBuffer(response.Serialize());
             } else {
                 // Wait the current thread until a buffer becomes available
-                auto wait_event = ctx.SleepClientThread(
+                ctx.SleepClientThread(
                     Kernel::GetCurrentThread(), "IHOSBinderDriver::DequeueBuffer", -1,
                     [=](Kernel::SharedPtr<Kernel::Thread> thread, Kernel::HLERequestContext& ctx,
                         ThreadWakeupReason reason) {
@@ -506,8 +511,8 @@ private:
                         ctx.WriteBuffer(response.Serialize());
                         IPC::ResponseBuilder rb{ctx, 2};
                         rb.Push(RESULT_SUCCESS);
-                    });
-                buffer_queue->SetBufferWaitEvent(std::move(wait_event));
+                    },
+                    buffer_queue->GetBufferWaitEvent());
             }
         } else if (transaction == TransactionId::RequestBuffer) {
             IGBPRequestBufferRequestParcel request{ctx.ReadBuffer()};
@@ -519,7 +524,8 @@ private:
         } else if (transaction == TransactionId::QueueBuffer) {
             IGBPQueueBufferRequestParcel request{ctx.ReadBuffer()};
 
-            buffer_queue->QueueBuffer(request.data.slot, request.data.transform);
+            buffer_queue->QueueBuffer(request.data.slot, request.data.transform,
+                                      request.data.GetCropRect());
 
             IGBPQueueBufferResponseParcel response{1280, 720};
             ctx.WriteBuffer(response.Serialize());
@@ -532,7 +538,7 @@ private:
             IGBPQueryResponseParcel response{value};
             ctx.WriteBuffer(response.Serialize());
         } else if (transaction == TransactionId::CancelBuffer) {
-            LOG_WARNING(Service_VI, "(STUBBED) called, transaction=CancelBuffer");
+            LOG_CRITICAL(Service_VI, "(STUBBED) called, transaction=CancelBuffer");
         } else {
             ASSERT_MSG(false, "Unimplemented");
         }
@@ -565,7 +571,7 @@ private:
         LOG_WARNING(Service_VI, "(STUBBED) called id={}, unknown={:08X}", id, unknown);
         IPC::ResponseBuilder rb{ctx, 2, 1};
         rb.Push(RESULT_SUCCESS);
-        rb.PushCopyObjects(buffer_queue->GetNativeHandle());
+        rb.PushCopyObjects(buffer_queue->GetBufferWaitEvent());
     }
 
     std::shared_ptr<NVFlinger::NVFlinger> nv_flinger;

src/core/loader/deconstructed_rom_directory.cpp

@@ -9,6 +9,7 @@
 #include "common/logging/log.h"
 #include "common/string_util.h"
 #include "core/file_sys/romfs_factory.h"
+#include "core/gdbstub/gdbstub.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/resource_limit.h"
 #include "core/hle/service/filesystem/filesystem.h"
@@ -133,6 +134,8 @@ ResultStatus AppLoader_DeconstructedRomDirectory::Load(
         next_load_addr = AppLoader_NSO::LoadModule(path, load_addr);
         if (next_load_addr) {
             LOG_DEBUG(Loader, "loaded module {} @ 0x{:X}", module, load_addr);
+            // Register module with GDBStub
+            GDBStub::RegisterModule(module, load_addr, next_load_addr - 1, false);
         } else {
             next_load_addr = load_addr;
         }
@@ -151,8 +154,7 @@ ResultStatus AppLoader_DeconstructedRomDirectory::Load(
 
     // Register the RomFS if a ".romfs" file was found
     if (!filepath_romfs.empty()) {
-        Service::FileSystem::RegisterFileSystem(std::make_unique<FileSys::RomFS_Factory>(*this),
-                                                Service::FileSystem::Type::RomFS);
+        Service::FileSystem::RegisterRomFS(std::make_unique<FileSys::RomFSFactory>(*this));
     }
 
     is_loaded = true;

src/core/loader/nca.cpp

@@ -7,10 +7,12 @@
 #include "common/common_funcs.h"
 #include "common/file_util.h"
 #include "common/logging/log.h"
+#include "common/string_util.h"
 #include "common/swap.h"
 #include "core/core.h"
 #include "core/file_sys/program_metadata.h"
 #include "core/file_sys/romfs_factory.h"
+#include "core/gdbstub/gdbstub.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/resource_limit.h"
 #include "core/hle/service/filesystem/filesystem.h"
@@ -259,6 +261,8 @@ ResultStatus AppLoader_NCA::Load(Kernel::SharedPtr<Kernel::Process>& process) {
         next_load_addr = AppLoader_NSO::LoadModule(module, nca->GetExeFsFile(module), load_addr);
         if (next_load_addr) {
             LOG_DEBUG(Loader, "loaded module {} @ 0x{:X}", module, load_addr);
+            // Register module with GDBStub
+            GDBStub::RegisterModule(module, load_addr, next_load_addr - 1, false);
         } else {
             next_load_addr = load_addr;
         }
@@ -273,8 +277,7 @@ ResultStatus AppLoader_NCA::Load(Kernel::SharedPtr<Kernel::Process>& process) {
                  metadata.GetMainThreadStackSize());
 
     if (nca->GetRomFsSize() > 0)
-        Service::FileSystem::RegisterFileSystem(std::make_unique<FileSys::RomFS_Factory>(*this),
-                                                Service::FileSystem::Type::RomFS);
+        Service::FileSystem::RegisterRomFS(std::make_unique<FileSys::RomFSFactory>(*this));
 
     is_loaded = true;
     return ResultStatus::Success;

src/core/loader/nro.cpp

@@ -9,6 +9,7 @@
 #include "common/logging/log.h"
 #include "common/swap.h"
 #include "core/core.h"
+#include "core/gdbstub/gdbstub.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/resource_limit.h"
 #include "core/loader/nro.h"
@@ -115,6 +116,9 @@ bool AppLoader_NRO::LoadNro(const std::string& path, VAddr load_base) {
     codeset->memory = std::make_shared<std::vector<u8>>(std::move(program_image));
     Core::CurrentProcess()->LoadModule(codeset, load_base);
 
+    // Register module with GDBStub
+    GDBStub::RegisterModule(codeset->name, load_base, load_base);
+
     return true;
 }
 

src/core/loader/nso.cpp

@@ -10,6 +10,7 @@
 #include "common/logging/log.h"
 #include "common/swap.h"
 #include "core/core.h"
+#include "core/gdbstub/gdbstub.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/resource_limit.h"
 #include "core/loader/nso.h"
@@ -114,7 +115,7 @@ VAddr AppLoader_NSO::LoadModule(const std::string& name, const std::vector<u8>&
     std::vector<u8> program_image;
     for (int i = 0; i < nso_header.segments.size(); ++i) {
         std::vector<u8> compressed_data(nso_header.segments_compressed_size[i]);
-        for (int j = 0; j < nso_header.segments_compressed_size[i]; ++j)
+        for (auto j = 0; j < nso_header.segments_compressed_size[i]; ++j)
             compressed_data[j] = file_data[nso_header.segments[i].offset + j];
         std::vector<u8> data = DecompressSegment(compressed_data, nso_header.segments[i]);
         program_image.resize(nso_header.segments[i].location);
@@ -147,6 +148,9 @@ VAddr AppLoader_NSO::LoadModule(const std::string& name, const std::vector<u8>&
     codeset->memory = std::make_shared<std::vector<u8>>(std::move(program_image));
     Core::CurrentProcess()->LoadModule(codeset, load_base);
 
+    // Register module with GDBStub
+    GDBStub::RegisterModule(codeset->name, load_base, load_base);
+
     return load_base + image_size;
 }
 

src/video_core/engines/maxwell_3d.cpp

@@ -398,27 +398,6 @@ u32 Maxwell3D::GetRegisterValue(u32 method) const {
     return regs.reg_array[method];
 }
 
-bool Maxwell3D::IsShaderStageEnabled(Regs::ShaderStage stage) const {
-    // The Vertex stage is always enabled.
-    if (stage == Regs::ShaderStage::Vertex)
-        return true;
-
-    switch (stage) {
-    case Regs::ShaderStage::TesselationControl:
-        return regs.shader_config[static_cast<size_t>(Regs::ShaderProgram::TesselationControl)]
-                   .enable != 0;
-    case Regs::ShaderStage::TesselationEval:
-        return regs.shader_config[static_cast<size_t>(Regs::ShaderProgram::TesselationEval)]
-                   .enable != 0;
-    case Regs::ShaderStage::Geometry:
-        return regs.shader_config[static_cast<size_t>(Regs::ShaderProgram::Geometry)].enable != 0;
-    case Regs::ShaderStage::Fragment:
-        return regs.shader_config[static_cast<size_t>(Regs::ShaderProgram::Fragment)].enable != 0;
-    }
-
-    UNREACHABLE();
-}
-
 void Maxwell3D::ProcessClearBuffers() {
     ASSERT(regs.clear_buffers.R == regs.clear_buffers.G &&
            regs.clear_buffers.R == regs.clear_buffers.B &&

src/video_core/engines/maxwell_3d.h

@@ -379,6 +379,14 @@ public:
             }
         };
 
+        bool IsShaderConfigEnabled(size_t index) const {
+            // The VertexB is always enabled.
+            if (index == static_cast<size_t>(Regs::ShaderProgram::VertexB)) {
+                return true;
+            }
+            return shader_config[index].enable != 0;
+        }
+
         union {
             struct {
                 INSERT_PADDING_WORDS(0x45);
@@ -512,7 +520,18 @@ public:
                     u32 enable[NumRenderTargets];
                 } blend;
 
-                INSERT_PADDING_WORDS(0xB);
+                struct {
+                    u32 enable;
+                    u32 front_op_fail;
+                    u32 front_op_zfail;
+                    u32 front_op_zpass;
+                    u32 front_func_func;
+                    u32 front_func_ref;
+                    u32 front_func_mask;
+                    u32 front_mask;
+                } stencil;
+
+                INSERT_PADDING_WORDS(0x3);
 
                 union {
                     BitField<4, 1, u32> triangle_rast_flip;
@@ -548,7 +567,17 @@ public:
                     }
                 } tic;
 
-                INSERT_PADDING_WORDS(0x21);
+                INSERT_PADDING_WORDS(0x5);
+
+                struct {
+                    u32 enable;
+                    u32 back_op_fail;
+                    u32 back_op_zfail;
+                    u32 back_op_zpass;
+                    u32 back_func_func;
+                } stencil_two_side;
+
+                INSERT_PADDING_WORDS(0x17);
 
                 union {
                     BitField<2, 1, u32> coord_origin;
@@ -780,9 +809,6 @@ public:
     /// Returns the texture information for a specific texture in a specific shader stage.
     Texture::FullTextureInfo GetStageTexture(Regs::ShaderStage stage, size_t offset) const;
 
-    /// Returns whether the specified shader stage is enabled or not.
-    bool IsShaderStageEnabled(Regs::ShaderStage stage) const;
-
 private:
     std::unordered_map<u32, std::vector<u32>> uploaded_macros;
 
@@ -846,10 +872,12 @@ ASSERT_REG_POSITION(depth_write_enabled, 0x4BA);
 ASSERT_REG_POSITION(d3d_cull_mode, 0x4C2);
 ASSERT_REG_POSITION(depth_test_func, 0x4C3);
 ASSERT_REG_POSITION(blend, 0x4CF);
+ASSERT_REG_POSITION(stencil, 0x4E0);
 ASSERT_REG_POSITION(screen_y_control, 0x4EB);
 ASSERT_REG_POSITION(vb_element_base, 0x50D);
 ASSERT_REG_POSITION(tsc, 0x557);
 ASSERT_REG_POSITION(tic, 0x55D);
+ASSERT_REG_POSITION(stencil_two_side, 0x565);
 ASSERT_REG_POSITION(point_coord_replace, 0x581);
 ASSERT_REG_POSITION(code_address, 0x582);
 ASSERT_REG_POSITION(draw, 0x585);

src/video_core/engines/shader_bytecode.h

@@ -142,6 +142,7 @@ enum class PredCondition : u64 {
     GreaterThan = 4,
     NotEqual = 5,
     GreaterEqual = 6,
+    LessThanWithNan = 9,
     NotEqualWithNan = 13,
     // TODO(Subv): Other condition types
 };
@@ -201,6 +202,11 @@ enum class IMinMaxExchange : u64 {
     XHi = 3,
 };
 
+enum class FlowCondition : u64 {
+    Always = 0xF,
+    Fcsm_Tr = 0x1C, // TODO(bunnei): What is this used for?
+};
+
 union Instruction {
     Instruction& operator=(const Instruction& instr) {
         value = instr.value;
@@ -315,6 +321,10 @@ union Instruction {
         }
     } bfe;
 
+    union {
+        BitField<0, 5, FlowCondition> cond;
+    } flow;
+
     union {
         BitField<48, 1, u64> negate_b;
         BitField<49, 1, u64> negate_c;

src/video_core/gpu.h

@@ -67,6 +67,7 @@ struct FramebufferConfig {
 
     using TransformFlags = Service::NVFlinger::BufferQueue::BufferTransformFlags;
     TransformFlags transform_flags;
+    MathUtil::Rectangle<int> crop_rect;
 };
 
 namespace Engines {

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -15,6 +15,7 @@
 #include "common/microprofile.h"
 #include "common/scope_exit.h"
 #include "core/core.h"
+#include "core/frontend/emu_window.h"
 #include "core/hle/kernel/process.h"
 #include "core/settings.h"
 #include "video_core/engines/maxwell_3d.h"
@@ -22,6 +23,7 @@
 #include "video_core/renderer_opengl/gl_shader_gen.h"
 #include "video_core/renderer_opengl/maxwell_to_gl.h"
 #include "video_core/renderer_opengl/renderer_opengl.h"
+#include "video_core/video_core.h"
 
 using Maxwell = Tegra::Engines::Maxwell3D::Regs;
 using PixelFormat = SurfaceParams::PixelFormat;
@@ -181,6 +183,19 @@ std::pair<u8*, GLintptr> RasterizerOpenGL::SetupVertexArrays(u8* array_ptr,
     return {array_ptr, buffer_offset};
 }
 
+static GLShader::ProgramCode GetShaderProgramCode(Maxwell::ShaderProgram program) {
+    auto& gpu = Core::System().GetInstance().GPU().Maxwell3D();
+
+    // Fetch program code from memory
+    GLShader::ProgramCode program_code;
+    auto& shader_config = gpu.regs.shader_config[static_cast<size_t>(program)];
+    const u64 gpu_address{gpu.regs.code_address.CodeAddress() + shader_config.offset};
+    const boost::optional<VAddr> cpu_address{gpu.memory_manager.GpuToCpuAddress(gpu_address)};
+    Memory::ReadBlock(*cpu_address, program_code.data(), program_code.size() * sizeof(u64));
+
+    return program_code;
+}
+
 void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset) {
     // Helper function for uploading uniform data
     const auto copy_buffer = [&](GLuint handle, GLintptr offset, GLsizeiptr size) {
@@ -193,26 +208,23 @@ void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset) {
     };
 
     auto& gpu = Core::System().GetInstance().GPU().Maxwell3D();
-    ASSERT_MSG(!gpu.regs.shader_config[0].enable, "VertexA is unsupported!");
 
     // Next available bindpoints to use when uploading the const buffers and textures to the GLSL
     // shaders. The constbuffer bindpoint starts after the shader stage configuration bind points.
     u32 current_constbuffer_bindpoint = uniform_buffers.size();
     u32 current_texture_bindpoint = 0;
 
-    for (unsigned index = 1; index < Maxwell::MaxShaderProgram; ++index) {
+    for (size_t index = 0; index < Maxwell::MaxShaderProgram; ++index) {
         auto& shader_config = gpu.regs.shader_config[index];
         const Maxwell::ShaderProgram program{static_cast<Maxwell::ShaderProgram>(index)};
 
-        const auto& stage = index - 1; // Stage indices are 0 - 5
-
-        const bool is_enabled = gpu.IsShaderStageEnabled(static_cast<Maxwell::ShaderStage>(stage));
-
         // Skip stages that are not enabled
-        if (!is_enabled) {
+        if (!gpu.regs.IsShaderConfigEnabled(index)) {
             continue;
         }
 
+        const size_t stage{index == 0 ? 0 : index - 1}; // Stage indices are 0 - 5
+
         GLShader::MaxwellUniformData ubo{};
         ubo.SetFromRegs(gpu.state.shader_stages[stage]);
         std::memcpy(buffer_ptr, &ubo, sizeof(ubo));
@@ -228,16 +240,21 @@ void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset) {
         buffer_ptr += sizeof(GLShader::MaxwellUniformData);
         buffer_offset += sizeof(GLShader::MaxwellUniformData);
 
-        // Fetch program code from memory
-        GLShader::ProgramCode program_code;
-        const u64 gpu_address{gpu.regs.code_address.CodeAddress() + shader_config.offset};
-        const boost::optional<VAddr> cpu_address{gpu.memory_manager.GpuToCpuAddress(gpu_address)};
-        Memory::ReadBlock(*cpu_address, program_code.data(), program_code.size() * sizeof(u64));
-        GLShader::ShaderSetup setup{std::move(program_code)};
-
+        GLShader::ShaderSetup setup{GetShaderProgramCode(program)};
         GLShader::ShaderEntries shader_resources;
 
         switch (program) {
+        case Maxwell::ShaderProgram::VertexA: {
+            // VertexB is always enabled, so when VertexA is enabled, we have two vertex shaders.
+            // Conventional HW does not support this, so we combine VertexA and VertexB into one
+            // stage here.
+            setup.SetProgramB(GetShaderProgramCode(Maxwell::ShaderProgram::VertexB));
+            GLShader::MaxwellVSConfig vs_config{setup};
+            shader_resources =
+                shader_program_manager->UseProgrammableVertexShader(vs_config, setup);
+            break;
+        }
+
         case Maxwell::ShaderProgram::VertexB: {
             GLShader::MaxwellVSConfig vs_config{setup};
             shader_resources =
@@ -268,6 +285,12 @@ void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset) {
         current_texture_bindpoint =
             SetupTextures(static_cast<Maxwell::ShaderStage>(stage), gl_stage_program,
                           current_texture_bindpoint, shader_resources.texture_samplers);
+
+        // When VertexA is enabled, we have dual vertex shaders
+        if (program == Maxwell::ShaderProgram::VertexA) {
+            // VertexB was combined with VertexA, so we skip the VertexB iteration
+            index++;
+        }
     }
 
     shader_program_manager->UseTrivialGeometryShader();
@@ -301,9 +324,6 @@ std::pair<Surface, Surface> RasterizerOpenGL::ConfigureFramebuffers(bool using_c
                                                                     bool using_depth_fb) {
     const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
 
-    // Sync the depth test state before configuring the framebuffer surfaces.
-    SyncDepthTestState();
-
     // TODO(bunnei): Implement this
     const bool has_stencil = false;
 
@@ -368,11 +388,20 @@ void RasterizerOpenGL::Clear() {
     if (regs.clear_buffers.Z) {
         clear_mask |= GL_DEPTH_BUFFER_BIT;
         use_depth_fb = true;
+
+        // Always enable the depth write when clearing the depth buffer. The depth write mask is
+        // ignored when clearing the buffer in the Switch, but OpenGL obeys it so we set it to true.
+        state.depth.test_enabled = true;
+        state.depth.write_mask = GL_TRUE;
+        state.depth.test_func = GL_ALWAYS;
+        state.Apply();
     }
 
     if (clear_mask == 0)
         return;
 
+    ScopeAcquireGLContext acquire_context;
+
     auto [dirty_color_surface, dirty_depth_surface] =
         ConfigureFramebuffers(use_color_fb, use_depth_fb);
 
@@ -399,9 +428,12 @@ void RasterizerOpenGL::DrawArrays() {
     MICROPROFILE_SCOPE(OpenGL_Drawing);
     const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
 
+    ScopeAcquireGLContext acquire_context;
+
     auto [dirty_color_surface, dirty_depth_surface] =
         ConfigureFramebuffers(true, regs.zeta.Address() != 0);
 
+    SyncDepthTestState();
     SyncBlendState();
     SyncCullMode();
 
@@ -605,9 +637,6 @@ u32 RasterizerOpenGL::SetupConstBuffers(Maxwell::ShaderStage stage, GLuint progr
     auto& gpu = Core::System::GetInstance().GPU();
     auto& maxwell3d = gpu.Get3DEngine();
 
-    ASSERT_MSG(maxwell3d.IsShaderStageEnabled(stage),
-               "Attempted to upload constbuffer of disabled shader stage");
-
     // Reset all buffer draw state for this stage.
     for (auto& buffer : state.draw.const_buffers[static_cast<size_t>(stage)]) {
         buffer.bindpoint = 0;
@@ -674,9 +703,6 @@ u32 RasterizerOpenGL::SetupTextures(Maxwell::ShaderStage stage, GLuint program,
     auto& gpu = Core::System::GetInstance().GPU();
     auto& maxwell3d = gpu.Get3DEngine();
 
-    ASSERT_MSG(maxwell3d.IsShaderStageEnabled(stage),
-               "Attempted to upload textures of disabled shader stage");
-
     ASSERT_MSG(current_unit + entries.size() <= std::size(state.texture_units),
                "Exceeded the number of active textures.");
 

src/video_core/renderer_opengl/gl_rasterizer_cache.cpp

@@ -105,6 +105,7 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form
     {GL_COMPRESSED_RGBA_BPTC_UNORM_ARB, GL_RGB, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
      true},                                                             // BC7U
     {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_4X4
+    {GL_RG8, GL_RG, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},     // G8R8
 
     // DepthStencil formats
     {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8, ComponentType::UNorm,
@@ -112,6 +113,8 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form
     {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8, ComponentType::UNorm,
      false},                                                                            // S8Z24
     {GL_DEPTH_COMPONENT32F, GL_DEPTH_COMPONENT, GL_FLOAT, ComponentType::Float, false}, // Z32F
+    {GL_DEPTH_COMPONENT16, GL_DEPTH_COMPONENT, GL_UNSIGNED_SHORT, ComponentType::UNorm,
+     false}, // Z16
 }};
 
 static const FormatTuple& GetFormatTuple(PixelFormat pixel_format, ComponentType component_type) {
@@ -194,8 +197,9 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr),
         MortonCopy<true, PixelFormat::DXT1>,         MortonCopy<true, PixelFormat::DXT23>,
         MortonCopy<true, PixelFormat::DXT45>,        MortonCopy<true, PixelFormat::DXN1>,
         MortonCopy<true, PixelFormat::BC7U>,         MortonCopy<true, PixelFormat::ASTC_2D_4X4>,
-        MortonCopy<true, PixelFormat::Z24S8>,        MortonCopy<true, PixelFormat::S8Z24>,
-        MortonCopy<true, PixelFormat::Z32F>,
+        MortonCopy<true, PixelFormat::G8R8>,         MortonCopy<true, PixelFormat::Z24S8>,
+        MortonCopy<true, PixelFormat::S8Z24>,        MortonCopy<true, PixelFormat::Z32F>,
+        MortonCopy<true, PixelFormat::Z16>,
 };
 
 static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr),
@@ -215,10 +219,12 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr),
         nullptr,
         nullptr,
         nullptr,
-        MortonCopy<false, PixelFormat::ABGR8>,
+        nullptr,
+        MortonCopy<false, PixelFormat::G8R8>,
         MortonCopy<false, PixelFormat::Z24S8>,
         MortonCopy<false, PixelFormat::S8Z24>,
         MortonCopy<false, PixelFormat::Z32F>,
+        MortonCopy<false, PixelFormat::Z16>,
 };
 
 // Allocate an uninitialized texture of appropriate size and format for the surface
@@ -271,9 +277,10 @@ static void ConvertS8Z24ToZ24S8(std::vector<u8>& data, u32 width, u32 height) {
 
     S8Z24 input_pixel{};
     Z24S8 output_pixel{};
+    const auto bpp{CachedSurface::GetGLBytesPerPixel(PixelFormat::S8Z24)};
     for (size_t y = 0; y < height; ++y) {
         for (size_t x = 0; x < width; ++x) {
-            const size_t offset{y * width + x};
+            const size_t offset{bpp * (y * width + x)};
             std::memcpy(&input_pixel, &data[offset], sizeof(S8Z24));
             output_pixel.s8.Assign(input_pixel.s8);
             output_pixel.z24.Assign(input_pixel.z24);
@@ -281,6 +288,19 @@ static void ConvertS8Z24ToZ24S8(std::vector<u8>& data, u32 width, u32 height) {
         }
     }
 }
+
+static void ConvertG8R8ToR8G8(std::vector<u8>& data, u32 width, u32 height) {
+    const auto bpp{CachedSurface::GetGLBytesPerPixel(PixelFormat::G8R8)};
+    for (size_t y = 0; y < height; ++y) {
+        for (size_t x = 0; x < width; ++x) {
+            const size_t offset{bpp * (y * width + x)};
+            const u8 temp{data[offset]};
+            data[offset] = data[offset + 1];
+            data[offset + 1] = temp;
+        }
+    }
+}
+
 /**
  * Helper function to perform software conversion (as needed) when loading a buffer from Switch
  * memory. This is for Maxwell pixel formats that cannot be represented as-is in OpenGL or with
@@ -301,6 +321,11 @@ static void ConvertFormatAsNeeded_LoadGLBuffer(std::vector<u8>& data, PixelForma
         // Convert the S8Z24 depth format to Z24S8, as OpenGL does not support S8Z24.
         ConvertS8Z24ToZ24S8(data, width, height);
         break;
+
+    case PixelFormat::G8R8:
+        // Convert the G8R8 color format to R8G8, as OpenGL does not support G8R8.
+        ConvertG8R8ToR8G8(data, width, height);
+        break;
     }
 }
 

src/video_core/renderer_opengl/gl_rasterizer_cache.h

@@ -37,13 +37,15 @@ struct SurfaceParams {
         DXN1 = 11, // This is also known as BC4
         BC7U = 12,
         ASTC_2D_4X4 = 13,
+        G8R8 = 14,
 
         MaxColorFormat,
 
         // DepthStencil formats
-        Z24S8 = 14,
-        S8Z24 = 15,
-        Z32F = 16,
+        Z24S8 = 15,
+        S8Z24 = 16,
+        Z32F = 17,
+        Z16 = 18,
 
         MaxDepthStencilFormat,
 
@@ -95,9 +97,11 @@ struct SurfaceParams {
             4, // DXN1
             4, // BC7U
             4, // ASTC_2D_4X4
+            1, // G8R8
             1, // Z24S8
             1, // S8Z24
             1, // Z32F
+            1, // Z16
         }};
 
         ASSERT(static_cast<size_t>(format) < compression_factor_table.size());
@@ -123,9 +127,11 @@ struct SurfaceParams {
             64,  // DXN1
             128, // BC7U
             32,  // ASTC_2D_4X4
+            16,  // G8R8
             32,  // Z24S8
             32,  // S8Z24
             32,  // Z32F
+            16,  // Z16
         }};
 
         ASSERT(static_cast<size_t>(format) < bpp_table.size());
@@ -143,6 +149,8 @@ struct SurfaceParams {
             return PixelFormat::Z24S8;
         case Tegra::DepthFormat::Z32_FLOAT:
             return PixelFormat::Z32F;
+        case Tegra::DepthFormat::Z16_UNORM:
+            return PixelFormat::Z16;
         default:
             LOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
             UNREACHABLE();
@@ -181,6 +189,8 @@ struct SurfaceParams {
             return PixelFormat::A1B5G5R5;
         case Tegra::Texture::TextureFormat::R8:
             return PixelFormat::R8;
+        case Tegra::Texture::TextureFormat::G8R8:
+            return PixelFormat::G8R8;
         case Tegra::Texture::TextureFormat::R16_G16_B16_A16:
             return PixelFormat::RGBA16F;
         case Tegra::Texture::TextureFormat::BF10GF11RF11:
@@ -218,6 +228,8 @@ struct SurfaceParams {
             return Tegra::Texture::TextureFormat::A1B5G5R5;
         case PixelFormat::R8:
             return Tegra::Texture::TextureFormat::R8;
+        case PixelFormat::G8R8:
+            return Tegra::Texture::TextureFormat::G8R8;
         case PixelFormat::RGBA16F:
             return Tegra::Texture::TextureFormat::R16_G16_B16_A16;
         case PixelFormat::R11FG11FB10F:
@@ -249,6 +261,8 @@ struct SurfaceParams {
             return Tegra::DepthFormat::Z24_S8_UNORM;
         case PixelFormat::Z32F:
             return Tegra::DepthFormat::Z32_FLOAT;
+        case PixelFormat::Z16:
+            return Tegra::DepthFormat::Z16_UNORM;
         default:
             UNREACHABLE();
         }
@@ -295,6 +309,7 @@ struct SurfaceParams {
 
     static ComponentType ComponentTypeFromDepthFormat(Tegra::DepthFormat format) {
         switch (format) {
+        case Tegra::DepthFormat::Z16_UNORM:
         case Tegra::DepthFormat::S8_Z24_UNORM:
         case Tegra::DepthFormat::Z24_S8_UNORM:
             return ComponentType::UNorm;

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -42,13 +42,14 @@ enum class ExitMethod {
 struct Subroutine {
     /// Generates a name suitable for GLSL source code.
     std::string GetName() const {
-        return "sub_" + std::to_string(begin) + '_' + std::to_string(end);
+        return "sub_" + std::to_string(begin) + '_' + std::to_string(end) + '_' + suffix;
     }
 
-    u32 begin;              ///< Entry point of the subroutine.
-    u32 end;                ///< Return point of the subroutine.
-    ExitMethod exit_method; ///< Exit method of the subroutine.
-    std::set<u32> labels;   ///< Addresses refereced by JMP instructions.
+    u32 begin;                 ///< Entry point of the subroutine.
+    u32 end;                   ///< Return point of the subroutine.
+    const std::string& suffix; ///< Suffix of the shader, used to make a unique subroutine name
+    ExitMethod exit_method;    ///< Exit method of the subroutine.
+    std::set<u32> labels;      ///< Addresses refereced by JMP instructions.
 
     bool operator<(const Subroutine& rhs) const {
         return std::tie(begin, end) < std::tie(rhs.begin, rhs.end);
@@ -58,11 +59,11 @@ struct Subroutine {
 /// Analyzes shader code and produces a set of subroutines.
 class ControlFlowAnalyzer {
 public:
-    ControlFlowAnalyzer(const ProgramCode& program_code, u32 main_offset)
+    ControlFlowAnalyzer(const ProgramCode& program_code, u32 main_offset, const std::string& suffix)
         : program_code(program_code) {
 
         // Recursively finds all subroutines.
-        const Subroutine& program_main = AddSubroutine(main_offset, PROGRAM_END);
+        const Subroutine& program_main = AddSubroutine(main_offset, PROGRAM_END, suffix);
         if (program_main.exit_method != ExitMethod::AlwaysEnd)
             throw DecompileFail("Program does not always end");
     }
@@ -77,12 +78,12 @@ private:
     std::map<std::pair<u32, u32>, ExitMethod> exit_method_map;
 
     /// Adds and analyzes a new subroutine if it is not added yet.
-    const Subroutine& AddSubroutine(u32 begin, u32 end) {
-        auto iter = subroutines.find(Subroutine{begin, end});
+    const Subroutine& AddSubroutine(u32 begin, u32 end, const std::string& suffix) {
+        auto iter = subroutines.find(Subroutine{begin, end, suffix});
         if (iter != subroutines.end())
             return *iter;
 
-        Subroutine subroutine{begin, end};
+        Subroutine subroutine{begin, end, suffix};
         subroutine.exit_method = Scan(begin, end, subroutine.labels);
         if (subroutine.exit_method == ExitMethod::Undetermined)
             throw DecompileFail("Recursive function detected");
@@ -191,7 +192,8 @@ public:
         UnsignedInteger,
     };
 
-    GLSLRegister(size_t index, ShaderWriter& shader) : index{index}, shader{shader} {}
+    GLSLRegister(size_t index, ShaderWriter& shader, const std::string& suffix)
+        : index{index}, shader{shader}, suffix{suffix} {}
 
     /// Gets the GLSL type string for a register
     static std::string GetTypeString(Type type) {
@@ -216,7 +218,7 @@ public:
     /// Returns a GLSL string representing the current state of the register
     const std::string GetActiveString() {
         declr_type.insert(active_type);
-        return GetPrefixString(active_type) + std::to_string(index);
+        return GetPrefixString(active_type) + std::to_string(index) + '_' + suffix;
     }
 
     /// Returns true if the active type is a float
@@ -251,6 +253,7 @@ private:
     ShaderWriter& shader;
     Type active_type{Type::Float};
     std::set<Type> declr_type;
+    const std::string& suffix;
 };
 
 /**
@@ -262,8 +265,8 @@ private:
 class GLSLRegisterManager {
 public:
     GLSLRegisterManager(ShaderWriter& shader, ShaderWriter& declarations,
-                        const Maxwell3D::Regs::ShaderStage& stage)
-        : shader{shader}, declarations{declarations}, stage{stage} {
+                        const Maxwell3D::Regs::ShaderStage& stage, const std::string& suffix)
+        : shader{shader}, declarations{declarations}, stage{stage}, suffix{suffix} {
         BuildRegisterList();
     }
 
@@ -430,12 +433,12 @@ public:
     }
 
     /// Add declarations for registers
-    void GenerateDeclarations() {
+    void GenerateDeclarations(const std::string& suffix) {
         for (const auto& reg : regs) {
             for (const auto& type : reg.DeclaredTypes()) {
                 declarations.AddLine(GLSLRegister::GetTypeString(type) + ' ' +
-                                     GLSLRegister::GetPrefixString(type) +
-                                     std::to_string(reg.GetIndex()) + " = 0;");
+                                     reg.GetPrefixString(type) + std::to_string(reg.GetIndex()) +
+                                     '_' + suffix + " = 0;");
             }
         }
         declarations.AddNewLine();
@@ -558,7 +561,7 @@ private:
     /// Build the GLSL register list.
     void BuildRegisterList() {
         for (size_t index = 0; index < Register::NumRegisters; ++index) {
-            regs.emplace_back(index, shader);
+            regs.emplace_back(index, shader, suffix);
         }
     }
 
@@ -620,16 +623,17 @@ private:
     std::array<ConstBufferEntry, Maxwell3D::Regs::MaxConstBuffers> declr_const_buffers;
     std::vector<SamplerEntry> used_samplers;
     const Maxwell3D::Regs::ShaderStage& stage;
+    const std::string& suffix;
 };
 
 class GLSLGenerator {
 public:
     GLSLGenerator(const std::set<Subroutine>& subroutines, const ProgramCode& program_code,
-                  u32 main_offset, Maxwell3D::Regs::ShaderStage stage)
+                  u32 main_offset, Maxwell3D::Regs::ShaderStage stage, const std::string& suffix)
         : subroutines(subroutines), program_code(program_code), main_offset(main_offset),
-          stage(stage) {
+          stage(stage), suffix(suffix) {
 
-        Generate();
+        Generate(suffix);
     }
 
     std::string GetShaderCode() {
@@ -644,7 +648,7 @@ public:
 private:
     /// Gets the Subroutine object corresponding to the specified address.
     const Subroutine& GetSubroutine(u32 begin, u32 end) const {
-        auto iter = subroutines.find(Subroutine{begin, end});
+        auto iter = subroutines.find(Subroutine{begin, end, suffix});
         ASSERT(iter != subroutines.end());
         return *iter;
     }
@@ -689,7 +693,7 @@ private:
         // Can't assign to the constant predicate.
         ASSERT(pred != static_cast<u64>(Pred::UnusedIndex));
 
-        std::string variable = 'p' + std::to_string(pred);
+        std::string variable = 'p' + std::to_string(pred) + '_' + suffix;
         shader.AddLine(variable + " = " + value + ';');
         declr_predicates.insert(std::move(variable));
     }
@@ -707,7 +711,7 @@ private:
         if (index == static_cast<u64>(Pred::UnusedIndex))
             variable = "true";
         else
-            variable = 'p' + std::to_string(index);
+            variable = 'p' + std::to_string(index) + '_' + suffix;
 
         if (negate) {
             return "!(" + variable + ')';
@@ -728,10 +732,10 @@ private:
                                        const std::string& op_a, const std::string& op_b) const {
         using Tegra::Shader::PredCondition;
         static const std::unordered_map<PredCondition, const char*> PredicateComparisonStrings = {
-            {PredCondition::LessThan, "<"},         {PredCondition::Equal, "=="},
-            {PredCondition::LessEqual, "<="},       {PredCondition::GreaterThan, ">"},
-            {PredCondition::NotEqual, "!="},        {PredCondition::GreaterEqual, ">="},
-            {PredCondition::NotEqualWithNan, "!="},
+            {PredCondition::LessThan, "<"},        {PredCondition::Equal, "=="},
+            {PredCondition::LessEqual, "<="},      {PredCondition::GreaterThan, ">"},
+            {PredCondition::NotEqual, "!="},       {PredCondition::GreaterEqual, ">="},
+            {PredCondition::LessThanWithNan, "<"}, {PredCondition::NotEqualWithNan, "!="},
         };
 
         const auto& comparison{PredicateComparisonStrings.find(condition)};
@@ -739,7 +743,8 @@ private:
                    "Unknown predicate comparison operation");
 
         std::string predicate{'(' + op_a + ") " + comparison->second + " (" + op_b + ')'};
-        if (condition == PredCondition::NotEqualWithNan) {
+        if (condition == PredCondition::LessThanWithNan ||
+            condition == PredCondition::NotEqualWithNan) {
             predicate += " || isnan(" + op_a + ") || isnan(" + op_b + ')';
         }
 
@@ -1639,16 +1644,32 @@ private:
                     shader.AddLine("color.a = " + regs.GetRegisterAsFloat(3) + ';');
                 }
 
-                shader.AddLine("return true;");
-                if (instr.pred.pred_index == static_cast<u64>(Pred::UnusedIndex)) {
-                    // If this is an unconditional exit then just end processing here, otherwise
-                    // we have to account for the possibility of the condition not being met, so
-                    // continue processing the next instruction.
-                    offset = PROGRAM_END - 1;
+                switch (instr.flow.cond) {
+                case Tegra::Shader::FlowCondition::Always:
+                    shader.AddLine("return true;");
+                    if (instr.pred.pred_index == static_cast<u64>(Pred::UnusedIndex)) {
+                        // If this is an unconditional exit then just end processing here,
+                        // otherwise we have to account for the possibility of the condition
+                        // not being met, so continue processing the next instruction.
+                        offset = PROGRAM_END - 1;
+                    }
+                    break;
+
+                case Tegra::Shader::FlowCondition::Fcsm_Tr:
+                    // TODO(bunnei): What is this used for? If we assume this conditon is not
+                    // satisifed, dual vertex shaders in Farming Simulator make more sense
+                    LOG_CRITICAL(HW_GPU, "Skipping unknown FlowCondition::Fcsm_Tr");
+                    break;
+
+                default:
+                    LOG_CRITICAL(HW_GPU, "Unhandled flow condition: {}",
+                                 static_cast<u32>(instr.flow.cond.Value()));
+                    UNREACHABLE();
                 }
                 break;
             }
             case OpCode::Id::KIL: {
+                ASSERT(instr.flow.cond == Tegra::Shader::FlowCondition::Always);
                 shader.AddLine("discard;");
                 break;
             }
@@ -1669,8 +1690,9 @@ private:
                 // can ignore this when generating GLSL code.
                 break;
             }
-            case OpCode::Id::DEPBAR:
-            case OpCode::Id::SYNC: {
+            case OpCode::Id::SYNC:
+                ASSERT(instr.flow.cond == Tegra::Shader::FlowCondition::Always);
+            case OpCode::Id::DEPBAR: {
                 // TODO(Subv): Find out if we actually have to care about these instructions or if
                 // the GLSL compiler takes care of that for us.
                 LOG_WARNING(HW_GPU, "DEPBAR/SYNC instruction is stubbed");
@@ -1710,7 +1732,7 @@ private:
         return program_counter;
     }
 
-    void Generate() {
+    void Generate(const std::string& suffix) {
         // Add declarations for all subroutines
         for (const auto& subroutine : subroutines) {
             shader.AddLine("bool " + subroutine.GetName() + "();");
@@ -1718,7 +1740,7 @@ private:
         shader.AddNewLine();
 
         // Add the main entry point
-        shader.AddLine("bool exec_shader() {");
+        shader.AddLine("bool exec_" + suffix + "() {");
         ++shader.scope;
         CallSubroutine(GetSubroutine(main_offset, PROGRAM_END));
         --shader.scope;
@@ -1781,7 +1803,7 @@ private:
 
     /// Add declarations for registers
     void GenerateDeclarations() {
-        regs.GenerateDeclarations();
+        regs.GenerateDeclarations(suffix);
 
         for (const auto& pred : declr_predicates) {
             declarations.AddLine("bool " + pred + " = false;");
@@ -1794,27 +1816,30 @@ private:
     const ProgramCode& program_code;
     const u32 main_offset;
     Maxwell3D::Regs::ShaderStage stage;
+    const std::string& suffix;
 
     ShaderWriter shader;
     ShaderWriter declarations;
-    GLSLRegisterManager regs{shader, declarations, stage};
+    GLSLRegisterManager regs{shader, declarations, stage, suffix};
 
     // Declarations
     std::set<std::string> declr_predicates;
 }; // namespace Decompiler
 
 std::string GetCommonDeclarations() {
-    std::string declarations = "bool exec_shader();\n";
+    std::string declarations;
     declarations += "#define MAX_CONSTBUFFER_ELEMENTS " +
                     std::to_string(RasterizerOpenGL::MaxConstbufferSize / (sizeof(GLvec4)));
+    declarations += '\n';
     return declarations;
 }
 
 boost::optional<ProgramResult> DecompileProgram(const ProgramCode& program_code, u32 main_offset,
-                                                Maxwell3D::Regs::ShaderStage stage) {
+                                                Maxwell3D::Regs::ShaderStage stage,
+                                                const std::string& suffix) {
     try {
-        auto subroutines = ControlFlowAnalyzer(program_code, main_offset).GetSubroutines();
-        GLSLGenerator generator(subroutines, program_code, main_offset, stage);
+        auto subroutines = ControlFlowAnalyzer(program_code, main_offset, suffix).GetSubroutines();
+        GLSLGenerator generator(subroutines, program_code, main_offset, stage, suffix);
         return ProgramResult{generator.GetShaderCode(), generator.GetEntries()};
     } catch (const DecompileFail& exception) {
         LOG_ERROR(HW_GPU, "Shader decompilation failed: {}", exception.what());

src/video_core/renderer_opengl/gl_shader_decompiler.h

@@ -20,7 +20,8 @@ using Tegra::Engines::Maxwell3D;
 std::string GetCommonDeclarations();
 
 boost::optional<ProgramResult> DecompileProgram(const ProgramCode& program_code, u32 main_offset,
-                                                Maxwell3D::Regs::ShaderStage stage);
+                                                Maxwell3D::Regs::ShaderStage stage,
+                                                const std::string& suffix);
 
 } // namespace Decompiler
 } // namespace GLShader

src/video_core/renderer_opengl/gl_shader_gen.cpp

@@ -17,10 +17,17 @@ ProgramResult GenerateVertexShader(const ShaderSetup& setup, const MaxwellVSConf
     std::string out = "#version 430 core\n";
     out += "#extension GL_ARB_separate_shader_objects : enable\n\n";
     out += Decompiler::GetCommonDeclarations();
+    out += "bool exec_vertex();\n";
+
+    if (setup.IsDualProgram()) {
+        out += "bool exec_vertex_b();\n";
+    }
+
+    ProgramResult program =
+        Decompiler::DecompileProgram(setup.program.code, PROGRAM_OFFSET,
+                                     Maxwell3D::Regs::ShaderStage::Vertex, "vertex")
+            .get_value_or({});
 
-    ProgramResult program = Decompiler::DecompileProgram(setup.program_code, PROGRAM_OFFSET,
-                                                         Maxwell3D::Regs::ShaderStage::Vertex)
-                                .get_value_or({});
     out += R"(
 
 out gl_PerVertex {
@@ -34,7 +41,14 @@ layout (std140) uniform vs_config {
 };
 
 void main() {
-    exec_shader();
+    exec_vertex();
+)";
+
+    if (setup.IsDualProgram()) {
+        out += "    exec_vertex_b();";
+    }
+
+    out += R"(
 
     // Viewport can be flipped, which is unsupported by glViewport
     position.xy *= viewport_flip.xy;
@@ -44,8 +58,19 @@ void main() {
     // For now, this is here to bring order in lieu of proper emulation
     position.w = 1.0;
 }
+
 )";
+
     out += program.first;
+
+    if (setup.IsDualProgram()) {
+        ProgramResult program_b =
+            Decompiler::DecompileProgram(setup.program.code_b, PROGRAM_OFFSET,
+                                         Maxwell3D::Regs::ShaderStage::Vertex, "vertex_b")
+                .get_value_or({});
+        out += program_b.first;
+    }
+
     return {out, program.second};
 }
 
@@ -53,12 +78,13 @@ ProgramResult GenerateFragmentShader(const ShaderSetup& setup, const MaxwellFSCo
     std::string out = "#version 430 core\n";
     out += "#extension GL_ARB_separate_shader_objects : enable\n\n";
     out += Decompiler::GetCommonDeclarations();
+    out += "bool exec_fragment();\n";
 
-    ProgramResult program = Decompiler::DecompileProgram(setup.program_code, PROGRAM_OFFSET,
-                                                         Maxwell3D::Regs::ShaderStage::Fragment)
-                                .get_value_or({});
+    ProgramResult program =
+        Decompiler::DecompileProgram(setup.program.code, PROGRAM_OFFSET,
+                                     Maxwell3D::Regs::ShaderStage::Fragment, "fragment")
+            .get_value_or({});
     out += R"(
-
 in vec4 position;
 out vec4 color;
 
@@ -67,7 +93,7 @@ layout (std140) uniform fs_config {
 };
 
 void main() {
-    exec_shader();
+    exec_fragment();
 }
 
 )";

src/video_core/renderer_opengl/gl_shader_gen.h

@@ -115,21 +115,48 @@ struct ShaderEntries {
 using ProgramResult = std::pair<std::string, ShaderEntries>;
 
 struct ShaderSetup {
-    ShaderSetup(ProgramCode&& program_code) : program_code(std::move(program_code)) {}
+    ShaderSetup(const ProgramCode& program_code) {
+        program.code = program_code;
+    }
+
+    struct {
+        ProgramCode code;
+        ProgramCode code_b; // Used for dual vertex shaders
+    } program;
 
-    ProgramCode program_code;
     bool program_code_hash_dirty = true;
 
     u64 GetProgramCodeHash() {
         if (program_code_hash_dirty) {
-            program_code_hash = Common::ComputeHash64(&program_code, sizeof(program_code));
+            program_code_hash = GetNewHash();
             program_code_hash_dirty = false;
         }
         return program_code_hash;
     }
 
+    /// Used in scenarios where we have a dual vertex shaders
+    void SetProgramB(const ProgramCode& program_b) {
+        program.code_b = program_b;
+        has_program_b = true;
+    }
+
+    bool IsDualProgram() const {
+        return has_program_b;
+    }
+
 private:
+    u64 GetNewHash() const {
+        if (has_program_b) {
+            // Compute hash over dual shader programs
+            return Common::ComputeHash64(&program, sizeof(program));
+        } else {
+            // Compute hash over a single shader program
+            return Common::ComputeHash64(&program.code, program.code.size());
+        }
+    }
+
     u64 program_code_hash{};
+    bool has_program_b{};
 };
 
 struct MaxwellShaderConfigCommon {

src/video_core/renderer_opengl/renderer_opengl.cpp

@@ -92,11 +92,24 @@ static std::array<GLfloat, 3 * 2> MakeOrthographicMatrix(const float width, cons
     return matrix;
 }
 
+ScopeAcquireGLContext::ScopeAcquireGLContext() {
+    if (Settings::values.use_multi_core) {
+        VideoCore::g_emu_window->MakeCurrent();
+    }
+}
+ScopeAcquireGLContext::~ScopeAcquireGLContext() {
+    if (Settings::values.use_multi_core) {
+        VideoCore::g_emu_window->DoneCurrent();
+    }
+}
+
 RendererOpenGL::RendererOpenGL() = default;
 RendererOpenGL::~RendererOpenGL() = default;
 
 /// Swap buffers (render frame)
 void RendererOpenGL::SwapBuffers(boost::optional<const Tegra::FramebufferConfig&> framebuffer) {
+    ScopeAcquireGLContext acquire_context;
+
     Core::System::GetInstance().perf_stats.EndSystemFrame();
 
     // Maintain the rasterizer's state as a priority
@@ -141,6 +154,7 @@ void RendererOpenGL::LoadFBToScreenInfo(const Tegra::FramebufferConfig& framebuf
 
     // Framebuffer orientation handling
     framebuffer_transform_flags = framebuffer.transform_flags;
+    framebuffer_crop_rect = framebuffer.crop_rect;
 
     // Ensure no bad interactions with GL_UNPACK_ALIGNMENT, which by default
     // only allows rows to have a memory alignement of 4.
@@ -307,11 +321,24 @@ void RendererOpenGL::DrawScreenTriangles(const ScreenInfo& screen_info, float x,
         }
     }
 
+    ASSERT_MSG(framebuffer_crop_rect.top == 0, "Unimplemented");
+    ASSERT_MSG(framebuffer_crop_rect.left == 0, "Unimplemented");
+
+    // Scale the output by the crop width/height. This is commonly used with 1280x720 rendering
+    // (e.g. handheld mode) on a 1920x1080 framebuffer.
+    f32 scale_u = 1.f, scale_v = 1.f;
+    if (framebuffer_crop_rect.GetWidth() > 0) {
+        scale_u = static_cast<f32>(framebuffer_crop_rect.GetWidth()) / screen_info.texture.width;
+    }
+    if (framebuffer_crop_rect.GetHeight() > 0) {
+        scale_v = static_cast<f32>(framebuffer_crop_rect.GetHeight()) / screen_info.texture.height;
+    }
+
     std::array<ScreenRectVertex, 4> vertices = {{
-        ScreenRectVertex(x, y, texcoords.top, left),
-        ScreenRectVertex(x + w, y, texcoords.bottom, left),
-        ScreenRectVertex(x, y + h, texcoords.top, right),
-        ScreenRectVertex(x + w, y + h, texcoords.bottom, right),
+        ScreenRectVertex(x, y, texcoords.top * scale_u, left * scale_v),
+        ScreenRectVertex(x + w, y, texcoords.bottom * scale_u, left * scale_v),
+        ScreenRectVertex(x, y + h, texcoords.top * scale_u, right * scale_v),
+        ScreenRectVertex(x + w, y + h, texcoords.bottom * scale_u, right * scale_v),
     }};
 
     state.texture_units[0].texture_2d = screen_info.display_texture;
@@ -418,7 +445,7 @@ static void APIENTRY DebugHandler(GLenum source, GLenum type, GLuint id, GLenum
 
 /// Initialize the renderer
 bool RendererOpenGL::Init() {
-    render_window->MakeCurrent();
+    ScopeAcquireGLContext acquire_context;
 
     if (GLAD_GL_KHR_debug) {
         glEnable(GL_DEBUG_OUTPUT);

src/video_core/renderer_opengl/renderer_opengl.h

@@ -31,6 +31,13 @@ struct ScreenInfo {
     TextureInfo texture;
 };
 
+/// Helper class to acquire/release OpenGL context within a given scope
+class ScopeAcquireGLContext : NonCopyable {
+public:
+    ScopeAcquireGLContext();
+    ~ScopeAcquireGLContext();
+};
+
 class RendererOpenGL : public RendererBase {
 public:
     RendererOpenGL();
@@ -90,4 +97,5 @@ private:
 
     /// Used for transforming the framebuffer orientation
     Tegra::FramebufferConfig::TransformFlags framebuffer_transform_flags;
+    MathUtil::Rectangle<int> framebuffer_crop_rect;
 };

src/video_core/textures/astc.cpp

@@ -25,16 +25,15 @@
 
 class BitStream {
 public:
-    BitStream(unsigned char* ptr, int nBits = 0, int start_offset = 0)
-        : m_BitsWritten(0), m_BitsRead(0), m_NumBits(nBits), m_CurByte(ptr),
-          m_NextBit(start_offset % 8), done(false) {}
+    explicit BitStream(unsigned char* ptr, int nBits = 0, int start_offset = 0)
+        : m_NumBits(nBits), m_CurByte(ptr), m_NextBit(start_offset % 8) {}
+
+    ~BitStream() = default;
 
     int GetBitsWritten() const {
         return m_BitsWritten;
     }
 
-    ~BitStream() {}
-
     void WriteBitsR(unsigned int val, unsigned int nBits) {
         for (unsigned int i = 0; i < nBits; i++) {
             WriteBit((val >> (nBits - i - 1)) & 1);
@@ -95,33 +94,28 @@ private:
         done = done || ++m_BitsWritten >= m_NumBits;
     }
 
-    int m_BitsWritten;
+    int m_BitsWritten = 0;
     const int m_NumBits;
     unsigned char* m_CurByte;
-    int m_NextBit;
-    int m_BitsRead;
+    int m_NextBit = 0;
+    int m_BitsRead = 0;
 
-    bool done;
+    bool done = false;
 };
 
 template <typename IntType>
 class Bits {
-private:
-    const IntType& m_Bits;
-
-    // Don't copy
-    Bits() {}
-    Bits(const Bits&) {}
-    Bits& operator=(const Bits&) {}
-
 public:
-    explicit Bits(IntType& v) : m_Bits(v) {}
+    explicit Bits(const IntType& v) : m_Bits(v) {}
 
-    uint8_t operator[](uint32_t bitPos) {
+    Bits(const Bits&) = delete;
+    Bits& operator=(const Bits&) = delete;
+
+    uint8_t operator[](uint32_t bitPos) const {
         return static_cast<uint8_t>((m_Bits >> bitPos) & 1);
     }
 
-    IntType operator()(uint32_t start, uint32_t end) {
+    IntType operator()(uint32_t start, uint32_t end) const {
         if (start == end) {
             return (*this)[start];
         } else if (start > end) {
@@ -133,6 +127,9 @@ public:
         uint64_t mask = (1 << (end - start + 1)) - 1;
         return (m_Bits >> start) & mask;
     }
+
+private:
+    const IntType& m_Bits;
 };
 
 enum EIntegerEncoding { eIntegerEncoding_JustBits, eIntegerEncoding_Quint, eIntegerEncoding_Trit };
@@ -186,12 +183,12 @@ public:
         m_QuintValue = val;
     }
 
-    bool MatchesEncoding(const IntegerEncodedValue& other) {
+    bool MatchesEncoding(const IntegerEncodedValue& other) const {
         return m_Encoding == other.m_Encoding && m_NumBits == other.m_NumBits;
     }
 
     // Returns the number of bits required to encode nVals values.
-    uint32_t GetBitLength(uint32_t nVals) {
+    uint32_t GetBitLength(uint32_t nVals) const {
         uint32_t totalBits = m_NumBits * nVals;
         if (m_Encoding == eIntegerEncoding_Trit) {
             totalBits += (nVals * 8 + 4) / 5;
@@ -382,19 +379,15 @@ private:
 namespace ASTCC {
 
 struct TexelWeightParams {
-    uint32_t m_Width;
-    uint32_t m_Height;
-    bool m_bDualPlane;
-    uint32_t m_MaxWeight;
-    bool m_bError;
-    bool m_bVoidExtentLDR;
-    bool m_bVoidExtentHDR;
+    uint32_t m_Width = 0;
+    uint32_t m_Height = 0;
+    bool m_bDualPlane = false;
+    uint32_t m_MaxWeight = 0;
+    bool m_bError = false;
+    bool m_bVoidExtentLDR = false;
+    bool m_bVoidExtentHDR = false;
 
-    TexelWeightParams() {
-        memset(this, 0, sizeof(*this));
-    }
-
-    uint32_t GetPackedBitSize() {
+    uint32_t GetPackedBitSize() const {
         // How many indices do we have?
         uint32_t nIdxs = m_Height * m_Width;
         if (m_bDualPlane) {
@@ -413,7 +406,7 @@ struct TexelWeightParams {
     }
 };
 
-TexelWeightParams DecodeBlockInfo(BitStream& strm) {
+static TexelWeightParams DecodeBlockInfo(BitStream& strm) {
     TexelWeightParams params;
 
     // Read the entire block mode all at once
@@ -612,8 +605,8 @@ TexelWeightParams DecodeBlockInfo(BitStream& strm) {
     return params;
 }
 
-void FillVoidExtentLDR(BitStream& strm, uint32_t* const outBuf, uint32_t blockWidth,
-                       uint32_t blockHeight) {
+static void FillVoidExtentLDR(BitStream& strm, uint32_t* const outBuf, uint32_t blockWidth,
+                              uint32_t blockHeight) {
     // Don't actually care about the void extent, just read the bits...
     for (int i = 0; i < 4; ++i) {
         strm.ReadBits(13);
@@ -628,23 +621,25 @@ void FillVoidExtentLDR(BitStream& strm, uint32_t* const outBuf, uint32_t blockWi
     uint32_t rgba = (r >> 8) | (g & 0xFF00) | (static_cast<uint32_t>(b) & 0xFF00) << 8 |
                     (static_cast<uint32_t>(a) & 0xFF00) << 16;
 
-    for (uint32_t j = 0; j < blockHeight; j++)
+    for (uint32_t j = 0; j < blockHeight; j++) {
         for (uint32_t i = 0; i < blockWidth; i++) {
             outBuf[j * blockWidth + i] = rgba;
         }
+    }
 }
 
-void FillError(uint32_t* outBuf, uint32_t blockWidth, uint32_t blockHeight) {
-    for (uint32_t j = 0; j < blockHeight; j++)
+static void FillError(uint32_t* outBuf, uint32_t blockWidth, uint32_t blockHeight) {
+    for (uint32_t j = 0; j < blockHeight; j++) {
         for (uint32_t i = 0; i < blockWidth; i++) {
             outBuf[j * blockWidth + i] = 0xFFFF00FF;
         }
+    }
 }
 
 // Replicates low numBits such that [(toBit - 1):(toBit - 1 - fromBit)]
 // is the same as [(numBits - 1):0] and repeats all the way down.
 template <typename IntType>
-IntType Replicate(const IntType& val, uint32_t numBits, uint32_t toBit) {
+static IntType Replicate(const IntType& val, uint32_t numBits, uint32_t toBit) {
     if (numBits == 0)
         return 0;
     if (toBit == 0)
@@ -668,27 +663,15 @@ IntType Replicate(const IntType& val, uint32_t numBits, uint32_t toBit) {
 
 class Pixel {
 protected:
-    typedef int16_t ChannelType;
-    uint8_t m_BitDepth[4];
-    int16_t color[4];
+    using ChannelType = int16_t;
+    uint8_t m_BitDepth[4] = {8, 8, 8, 8};
+    int16_t color[4] = {};
 
 public:
-    Pixel() {
-        for (int i = 0; i < 4; i++) {
-            m_BitDepth[i] = 8;
-            color[i] = 0;
-        }
-    }
-
-    Pixel(ChannelType a, ChannelType r, ChannelType g, ChannelType b, unsigned bitDepth = 8) {
-        for (int i = 0; i < 4; i++)
-            m_BitDepth[i] = bitDepth;
-
-        color[0] = a;
-        color[1] = r;
-        color[2] = g;
-        color[3] = b;
-    }
+    Pixel() = default;
+    Pixel(ChannelType a, ChannelType r, ChannelType g, ChannelType b, unsigned bitDepth = 8)
+        : m_BitDepth{uint8_t(bitDepth), uint8_t(bitDepth), uint8_t(bitDepth), uint8_t(bitDepth)},
+          color{a, r, g, b} {}
 
     // Changes the depth of each pixel. This scales the values to
     // the appropriate bit depth by either truncating the least
@@ -807,8 +790,8 @@ public:
     }
 };
 
-void DecodeColorValues(uint32_t* out, uint8_t* data, uint32_t* modes, const uint32_t nPartitions,
-                       const uint32_t nBitsForColorData) {
+static void DecodeColorValues(uint32_t* out, uint8_t* data, const uint32_t* modes,
+                              const uint32_t nPartitions, const uint32_t nBitsForColorData) {
     // First figure out how many color values we have
     uint32_t nValues = 0;
     for (uint32_t i = 0; i < nPartitions; i++) {
@@ -844,8 +827,7 @@ void DecodeColorValues(uint32_t* out, uint8_t* data, uint32_t* modes, const uint
     // Once we have the decoded values, we need to dequantize them to the 0-255 range
     // This procedure is outlined in ASTC spec C.2.13
     uint32_t outIdx = 0;
-    std::vector<IntegerEncodedValue>::const_iterator itr;
-    for (itr = decodedColorValues.begin(); itr != decodedColorValues.end(); itr++) {
+    for (auto itr = decodedColorValues.begin(); itr != decodedColorValues.end(); ++itr) {
         // Have we already decoded all that we need?
         if (outIdx >= nValues) {
             break;
@@ -978,7 +960,7 @@ void DecodeColorValues(uint32_t* out, uint8_t* data, uint32_t* modes, const uint
     }
 }
 
-uint32_t UnquantizeTexelWeight(const IntegerEncodedValue& val) {
+static uint32_t UnquantizeTexelWeight(const IntegerEncodedValue& val) {
     uint32_t bitval = val.GetBitValue();
     uint32_t bitlen = val.BaseBitLength();
 
@@ -1067,17 +1049,18 @@ uint32_t UnquantizeTexelWeight(const IntegerEncodedValue& val) {
     return result;
 }
 
-void UnquantizeTexelWeights(uint32_t out[2][144], std::vector<IntegerEncodedValue>& weights,
-                            const TexelWeightParams& params, const uint32_t blockWidth,
-                            const uint32_t blockHeight) {
+static void UnquantizeTexelWeights(uint32_t out[2][144],
+                                   const std::vector<IntegerEncodedValue>& weights,
+                                   const TexelWeightParams& params, const uint32_t blockWidth,
+                                   const uint32_t blockHeight) {
     uint32_t weightIdx = 0;
     uint32_t unquantized[2][144];
-    std::vector<IntegerEncodedValue>::const_iterator itr;
-    for (itr = weights.begin(); itr != weights.end(); itr++) {
+
+    for (auto itr = weights.begin(); itr != weights.end(); ++itr) {
         unquantized[0][weightIdx] = UnquantizeTexelWeight(*itr);
 
         if (params.m_bDualPlane) {
-            itr++;
+            ++itr;
             unquantized[1][weightIdx] = UnquantizeTexelWeight(*itr);
             if (itr == weights.end()) {
                 break;
@@ -1261,8 +1244,8 @@ static inline uint32_t Select2DPartition(int32_t seed, int32_t x, int32_t y, int
 }
 
 // Section C.2.14
-void ComputeEndpoints(Pixel& ep1, Pixel& ep2, const uint32_t*& colorValues,
-                      uint32_t colorEndpointMode) {
+static void ComputeEndpoints(Pixel& ep1, Pixel& ep2, const uint32_t*& colorValues,
+                             uint32_t colorEndpointMode) {
 #define READ_UINT_VALUES(N)                                                                        \
     uint32_t v[N];                                                                                 \
     for (uint32_t i = 0; i < N; i++) {                                                             \
@@ -1382,8 +1365,8 @@ void ComputeEndpoints(Pixel& ep1, Pixel& ep2, const uint32_t*& colorValues,
 #undef READ_INT_VALUES
 }
 
-void DecompressBlock(uint8_t inBuf[16], const uint32_t blockWidth, const uint32_t blockHeight,
-                     uint32_t* outBuf) {
+static void DecompressBlock(uint8_t inBuf[16], const uint32_t blockWidth,
+                            const uint32_t blockHeight, uint32_t* outBuf) {
     BitStream strm(inBuf);
     TexelWeightParams weightParams = DecodeBlockInfo(strm);
 
@@ -1617,8 +1600,7 @@ namespace Tegra::Texture::ASTC {
 std::vector<uint8_t> Decompress(std::vector<uint8_t>& data, uint32_t width, uint32_t height,
                                 uint32_t block_width, uint32_t block_height) {
     uint32_t blockIdx = 0;
-    std::vector<uint8_t> outData;
-    outData.resize(height * width * 4);
+    std::vector<uint8_t> outData(height * width * 4);
     for (uint32_t j = 0; j < height; j += block_height) {
         for (uint32_t i = 0; i < width; i += block_width) {
 

src/video_core/textures/decoders.cpp

@@ -62,6 +62,7 @@ u32 BytesPerPixel(TextureFormat format) {
         return 4;
     case TextureFormat::A1B5G5R5:
     case TextureFormat::B5G6R5:
+    case TextureFormat::G8R8:
         return 2;
     case TextureFormat::R8:
         return 1;
@@ -77,6 +78,8 @@ u32 BytesPerPixel(TextureFormat format) {
 
 static u32 DepthBytesPerPixel(DepthFormat format) {
     switch (format) {
+    case DepthFormat::Z16_UNORM:
+        return 2;
     case DepthFormat::S8_Z24_UNORM:
     case DepthFormat::Z24_S8_UNORM:
     case DepthFormat::Z32_FLOAT:
@@ -110,6 +113,7 @@ std::vector<u8> UnswizzleTexture(VAddr address, TextureFormat format, u32 width,
     case TextureFormat::A1B5G5R5:
     case TextureFormat::B5G6R5:
     case TextureFormat::R8:
+    case TextureFormat::G8R8:
     case TextureFormat::R16_G16_B16_A16:
     case TextureFormat::R32_G32_B32_A32:
     case TextureFormat::BF10GF11RF11:
@@ -133,6 +137,7 @@ std::vector<u8> UnswizzleDepthTexture(VAddr address, DepthFormat format, u32 wid
     std::vector<u8> unswizzled_data(width * height * bytes_per_pixel);
 
     switch (format) {
+    case DepthFormat::Z16_UNORM:
     case DepthFormat::S8_Z24_UNORM:
     case DepthFormat::Z24_S8_UNORM:
     case DepthFormat::Z32_FLOAT:
@@ -164,6 +169,7 @@ std::vector<u8> DecodeTexture(const std::vector<u8>& texture_data, TextureFormat
     case TextureFormat::A1B5G5R5:
     case TextureFormat::B5G6R5:
     case TextureFormat::R8:
+    case TextureFormat::G8R8:
     case TextureFormat::BF10GF11RF11:
     case TextureFormat::R32_G32_B32_A32:
         // TODO(Subv): For the time being just forward the same data without any decoding.

src/yuzu/bootmanager.cpp

@@ -20,7 +20,10 @@
 EmuThread::EmuThread(GRenderWindow* render_window) : render_window(render_window) {}
 
 void EmuThread::run() {
-    render_window->MakeCurrent();
+    if (!Settings::values.use_multi_core) {
+        // Single core mode must acquire OpenGL context for entire emulation session
+        render_window->MakeCurrent();
+    }
 
     MicroProfileOnThreadCreate("EmuThread");
 

src/yuzu/configuration/config.cpp

@@ -97,7 +97,7 @@ void Config::ReadValues() {
     qt_config->endGroup();
 
     qt_config->beginGroup("System");
-    Settings::values.use_docked_mode = qt_config->value("use_docked_mode", true).toBool();
+    Settings::values.use_docked_mode = qt_config->value("use_docked_mode", false).toBool();
     qt_config->endGroup();
 
     qt_config->beginGroup("Miscellaneous");

src/yuzu/game_list.cpp

@@ -17,10 +17,7 @@
 #include "game_list_p.h"
 #include "ui_settings.h"
 
-GameList::SearchField::KeyReleaseEater::KeyReleaseEater(GameList* gamelist) {
-    this->gamelist = gamelist;
-    edit_filter_text_old = "";
-}
+GameList::SearchField::KeyReleaseEater::KeyReleaseEater(GameList* gamelist) : gamelist{gamelist} {}
 
 // EventFilter in order to process systemkeys while editing the searchfield
 bool GameList::SearchField::KeyReleaseEater::eventFilter(QObject* obj, QEvent* event) {
@@ -141,10 +138,12 @@ GameList::SearchField::SearchField(GameList* parent) : QWidget{parent} {
  * @param userinput String containing all words getting checked
  * @return true if the haystack contains all words of userinput
  */
-bool GameList::containsAllWords(QString haystack, QString userinput) {
-    QStringList userinput_split = userinput.split(" ", QString::SplitBehavior::SkipEmptyParts);
+bool GameList::ContainsAllWords(const QString& haystack, const QString& userinput) const {
+    const QStringList userinput_split =
+        userinput.split(' ', QString::SplitBehavior::SkipEmptyParts);
+
     return std::all_of(userinput_split.begin(), userinput_split.end(),
-                       [haystack](QString s) { return haystack.contains(s); });
+                       [&haystack](const QString& s) { return haystack.contains(s); });
 }
 
 // Event in order to filter the gamelist after editing the searchfield
@@ -178,7 +177,7 @@ void GameList::onTextChanged(const QString& newText) {
             // The search is case insensitive because of toLower()
             // I decided not to use Qt::CaseInsensitive in containsAllWords to prevent
             // multiple conversions of edit_filter_text for each game in the gamelist
-            if (containsAllWords(file_name.append(" ").append(file_title), edit_filter_text) ||
+            if (ContainsAllWords(file_name.append(' ').append(file_title), edit_filter_text) ||
                 (file_programmid.count() == 16 && edit_filter_text.contains(file_programmid))) {
                 tree_view->setRowHidden(i, root_index, false);
                 ++result_count;

src/yuzu/game_list.h

@@ -89,7 +89,7 @@ private:
 
     void PopupContextMenu(const QPoint& menu_location);
     void RefreshGameDirectory();
-    bool containsAllWords(QString haystack, QString userinput);
+    bool ContainsAllWords(const QString& haystack, const QString& userinput) const;
 
     SearchField* search_field;
     GMainWindow* main_window = nullptr;

src/yuzu/main.cpp

@@ -374,6 +374,8 @@ bool GMainWindow::LoadROM(const QString& filename) {
 
     const Core::System::ResultStatus result{system.Load(render_window, filename.toStdString())};
 
+    render_window->DoneCurrent();
+
     if (result != Core::System::ResultStatus::Success) {
         switch (result) {
         case Core::System::ResultStatus::ErrorGetLoader:
@@ -916,6 +918,7 @@ int main(int argc, char* argv[]) {
     QCoreApplication::setApplicationName("yuzu");
 
     QApplication::setAttribute(Qt::AA_X11InitThreads);
+    QApplication::setAttribute(Qt::AA_DontCheckOpenGLContextThreadAffinity);
     QApplication app(argc, argv);
 
     // Qt changes the locale and causes issues in float conversion using std::to_string() when

src/yuzu_cmd/config.cpp

@@ -110,7 +110,7 @@ void Config::ReadValues() {
         sdl2_config->GetBoolean("Data Storage", "use_virtual_sd", true);
 
     // System
-    Settings::values.use_docked_mode = sdl2_config->GetBoolean("System", "use_docked_mode", true);
+    Settings::values.use_docked_mode = sdl2_config->GetBoolean("System", "use_docked_mode", false);
 
     // Miscellaneous
     Settings::values.log_filter = sdl2_config->Get("Miscellaneous", "log_filter", "*:Trace");

src/yuzu_cmd/default_ini.h

@@ -163,7 +163,7 @@ use_virtual_sd =
 
 [System]
 # Whether the system is docked
-# 1 (default): Yes, 0: No
+# 1: Yes, 0 (default): No
 use_docked_mode =
 
 # The system region that yuzu will use during emulation

src/yuzu_cmd/yuzu.cpp

@@ -148,6 +148,11 @@ int main(int argc, char** argv) {
 
     std::unique_ptr<EmuWindow_SDL2> emu_window{std::make_unique<EmuWindow_SDL2>(fullscreen)};
 
+    if (!Settings::values.use_multi_core) {
+        // Single core mode must acquire OpenGL context for entire emulation session
+        emu_window->MakeCurrent();
+    }
+
     Core::System& system{Core::System::GetInstance()};
 
     SCOPE_EXIT({ system.Shutdown(); });