added a toggle to force 30fps mode

shader_cache: Permit a Null Shader in case of a bad host_ptr.

.gitmodules

@@ -40,3 +40,6 @@
 [submodule "Vulkan-Headers"]
     path = externals/Vulkan-Headers
     url = https://github.com/KhronosGroup/Vulkan-Headers.git
+[submodule "externals/zstd"]
+    path = externals/zstd
+    url = https://github.com/facebook/zstd

CMakeLists.txt

@@ -309,7 +309,7 @@ if (CLANG_FORMAT)
     set(CCOMMENT "Running clang format against all the .h and .cpp files in src/")
     if (WIN32)
         add_custom_target(clang-format
-            COMMAND powershell.exe -Command "Get-ChildItem ${SRCS}/* -Include *.cpp,*.h -Recurse | Foreach {${CLANG_FORMAT} -i $_.fullname}"
+            COMMAND powershell.exe -Command "Get-ChildItem '${SRCS}/*' -Include *.cpp,*.h -Recurse | Foreach {&'${CLANG_FORMAT}' -i $_.fullname}"
             COMMENT ${CCOMMENT})
     elseif(MINGW)
         add_custom_target(clang-format

externals/CMakeLists.txt

@@ -49,6 +49,10 @@ add_subdirectory(open_source_archives EXCLUDE_FROM_ALL)
 add_library(unicorn-headers INTERFACE)
 target_include_directories(unicorn-headers INTERFACE ./unicorn/include)
 
+# Zstandard
+add_subdirectory(zstd/build/cmake EXCLUDE_FROM_ALL)
+target_include_directories(libzstd_static INTERFACE ./zstd/lib)
+
 # SoundTouch
 add_subdirectory(soundtouch)
 

src/common/CMakeLists.txt

@@ -125,6 +125,8 @@ add_library(common STATIC
     uint128.h
     vector_math.h
     web_result.h
+    zstd_compression.cpp
+    zstd_compression.h
 )
 
 if(ARCHITECTURE_x86_64)
@@ -138,4 +140,4 @@ endif()
 create_target_directory_groups(common)
 
 target_link_libraries(common PUBLIC Boost::boost fmt microprofile)
-target_link_libraries(common PRIVATE lz4_static)
+target_link_libraries(common PRIVATE lz4_static libzstd_static)

src/common/assert.h

@@ -57,3 +57,21 @@ __declspec(noinline, noreturn)
 
 #define UNIMPLEMENTED_IF(cond) ASSERT_MSG(!(cond), "Unimplemented code!")
 #define UNIMPLEMENTED_IF_MSG(cond, ...) ASSERT_MSG(!(cond), __VA_ARGS__)
+
+// If the assert is ignored, execute _b_
+#define ASSERT_OR_EXECUTE(_a_, _b_)                                                                \
+    do {                                                                                           \
+        ASSERT(_a_);                                                                               \
+        if (!(_a_)) {                                                                              \
+            _b_                                                                                    \
+        }                                                                                          \
+    } while (0)
+
+// If the assert is ignored, execute _b_
+#define ASSERT_OR_EXECUTE_MSG(_a_, _b_, ...)                                                       \
+    do {                                                                                           \
+        ASSERT_MSG(_a_, __VA_ARGS__);                                                              \
+        if (!(_a_)) {                                                                              \
+            _b_                                                                                    \
+        }                                                                                          \
+    } while (0)

src/common/bit_util.h

@@ -32,7 +32,7 @@ inline u32 CountLeadingZeroes32(u32 value) {
     return 32;
 }
 
-inline u64 CountLeadingZeroes64(u64 value) {
+inline u32 CountLeadingZeroes64(u64 value) {
     unsigned long leading_zero = 0;
 
     if (_BitScanReverse64(&leading_zero, value) != 0) {
@@ -47,15 +47,15 @@ inline u32 CountLeadingZeroes32(u32 value) {
         return 32;
     }
 
-    return __builtin_clz(value);
+    return static_cast<u32>(__builtin_clz(value));
 }
 
-inline u64 CountLeadingZeroes64(u64 value) {
+inline u32 CountLeadingZeroes64(u64 value) {
     if (value == 0) {
         return 64;
     }
 
-    return __builtin_clzll(value);
+    return static_cast<u32>(__builtin_clzll(value));
 }
 #endif
 
@@ -70,7 +70,7 @@ inline u32 CountTrailingZeroes32(u32 value) {
     return 32;
 }
 
-inline u64 CountTrailingZeroes64(u64 value) {
+inline u32 CountTrailingZeroes64(u64 value) {
     unsigned long trailing_zero = 0;
 
     if (_BitScanForward64(&trailing_zero, value) != 0) {
@@ -85,15 +85,15 @@ inline u32 CountTrailingZeroes32(u32 value) {
         return 32;
     }
 
-    return __builtin_ctz(value);
+    return static_cast<u32>(__builtin_ctz(value));
 }
 
-inline u64 CountTrailingZeroes64(u64 value) {
+inline u32 CountTrailingZeroes64(u64 value) {
     if (value == 0) {
         return 64;
     }
 
-    return __builtin_ctzll(value);
+    return static_cast<u32>(__builtin_ctzll(value));
 }
 #endif
 

src/common/multi_level_queue.h

@@ -72,7 +72,7 @@ public:
                 u64 prios = mlq.used_priorities;
                 prios &= ~((1ULL << (current_priority + 1)) - 1);
                 if (prios == 0) {
-                    current_priority = mlq.depth();
+                    current_priority = static_cast<u32>(mlq.depth());
                 } else {
                     current_priority = CountTrailingZeroes64(prios);
                     it = GetBeginItForPrio();

src/common/zstd_compression.cpp

@@ -0,0 +1,53 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <algorithm>
+#include <zstd.h>
+
+#include "common/assert.h"
+#include "common/zstd_compression.h"
+
+namespace Common::Compression {
+
+std::vector<u8> CompressDataZSTD(const u8* source, std::size_t source_size, s32 compression_level) {
+    compression_level = std::clamp(compression_level, 1, ZSTD_maxCLevel());
+
+    const std::size_t max_compressed_size = ZSTD_compressBound(source_size);
+    std::vector<u8> compressed(max_compressed_size);
+
+    const std::size_t compressed_size =
+        ZSTD_compress(compressed.data(), compressed.size(), source, source_size, compression_level);
+
+    if (ZSTD_isError(compressed_size)) {
+        // Compression failed
+        return {};
+    }
+
+    compressed.resize(compressed_size);
+
+    return compressed;
+}
+
+std::vector<u8> CompressDataZSTDDefault(const u8* source, std::size_t source_size) {
+    return CompressDataZSTD(source, source_size, ZSTD_CLEVEL_DEFAULT);
+}
+
+std::vector<u8> DecompressDataZSTD(const std::vector<u8>& compressed) {
+    const std::size_t decompressed_size =
+        ZSTD_getDecompressedSize(compressed.data(), compressed.size());
+    std::vector<u8> decompressed(decompressed_size);
+
+    const std::size_t uncompressed_result_size = ZSTD_decompress(
+        decompressed.data(), decompressed.size(), compressed.data(), compressed.size());
+
+    if (decompressed_size != uncompressed_result_size || ZSTD_isError(uncompressed_result_size)) {
+        // Decompression failed
+        return {};
+    }
+    return decompressed;
+}
+
+} // namespace Common::Compression

src/common/zstd_compression.h

@@ -0,0 +1,42 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <vector>
+
+#include "common/common_types.h"
+
+namespace Common::Compression {
+
+/**
+ * Compresses a source memory region with Zstandard and returns the compressed data in a vector.
+ *
+ * @param source the uncompressed source memory region.
+ * @param source_size the size in bytes of the uncompressed source memory region.
+ * @param compression_level the used compression level. Should be between 1 and 22.
+ *
+ * @return the compressed data.
+ */
+std::vector<u8> CompressDataZSTD(const u8* source, std::size_t source_size, s32 compression_level);
+
+/**
+ * Compresses a source memory region with Zstandard with the default compression level and returns
+ * the compressed data in a vector.
+ *
+ * @param source the uncompressed source memory region.
+ * @param source_size the size in bytes of the uncompressed source memory region.
+ *
+ * @return the compressed data.
+ */
+std::vector<u8> CompressDataZSTDDefault(const u8* source, std::size_t source_size);
+
+/**
+ * Decompresses a source memory region with Zstandard and returns the uncompressed data in a vector.
+ *
+ * @param compressed the compressed source memory region.
+ *
+ * @return the decompressed data.
+ */
+std::vector<u8> DecompressDataZSTD(const std::vector<u8>& compressed);
+
+} // namespace Common::Compression

src/core/arm/dynarmic/arm_dynarmic.cpp

@@ -26,7 +26,6 @@ using Vector = Dynarmic::A64::Vector;
 class ARM_Dynarmic_Callbacks : public Dynarmic::A64::UserCallbacks {
 public:
     explicit ARM_Dynarmic_Callbacks(ARM_Dynarmic& parent) : parent(parent) {}
-    ~ARM_Dynarmic_Callbacks() = default;
 
     u8 MemoryRead8(u64 vaddr) override {
         return Memory::Read8(vaddr);
@@ -164,7 +163,6 @@ MICROPROFILE_DEFINE(ARM_Jit_Dynarmic, "ARM JIT", "Dynarmic", MP_RGB(255, 64, 64)
 
 void ARM_Dynarmic::Run() {
     MICROPROFILE_SCOPE(ARM_Jit_Dynarmic);
-    ASSERT(Memory::GetCurrentPageTable() == current_page_table);
 
     jit->Run();
 }
@@ -279,7 +277,6 @@ void ARM_Dynarmic::ClearExclusiveState() {
 
 void ARM_Dynarmic::PageTableChanged() {
     jit = MakeJit();
-    current_page_table = Memory::GetCurrentPageTable();
 }
 
 DynarmicExclusiveMonitor::DynarmicExclusiveMonitor(std::size_t core_count) : monitor(core_count) {}

src/core/arm/dynarmic/arm_dynarmic.h

@@ -12,10 +12,6 @@
 #include "core/arm/exclusive_monitor.h"
 #include "core/arm/unicorn/arm_unicorn.h"
 
-namespace Common {
-struct PageTable;
-}
-
 namespace Core::Timing {
 class CoreTiming;
 }
@@ -29,7 +25,7 @@ class ARM_Dynarmic final : public ARM_Interface {
 public:
     ARM_Dynarmic(Timing::CoreTiming& core_timing, ExclusiveMonitor& exclusive_monitor,
                  std::size_t core_index);
-    ~ARM_Dynarmic();
+    ~ARM_Dynarmic() override;
 
     void MapBackingMemory(VAddr address, std::size_t size, u8* memory,
                           Kernel::VMAPermission perms) override;
@@ -69,14 +65,12 @@ private:
     std::size_t core_index;
     Timing::CoreTiming& core_timing;
     DynarmicExclusiveMonitor& exclusive_monitor;
-
-    Common::PageTable* current_page_table = nullptr;
 };
 
 class DynarmicExclusiveMonitor final : public ExclusiveMonitor {
 public:
     explicit DynarmicExclusiveMonitor(std::size_t core_count);
-    ~DynarmicExclusiveMonitor();
+    ~DynarmicExclusiveMonitor() override;
 
     void SetExclusive(std::size_t core_index, VAddr addr) override;
     void ClearExclusive() override;

src/core/arm/unicorn/arm_unicorn.cpp

@@ -192,12 +192,13 @@ void ARM_Unicorn::ExecuteInstructions(int num_instructions) {
     CHECKED(uc_emu_start(uc, GetPC(), 1ULL << 63, 0, num_instructions));
     core_timing.AddTicks(num_instructions);
     if (GDBStub::IsServerEnabled()) {
-        if (last_bkpt_hit) {
+        if (last_bkpt_hit && last_bkpt.type == GDBStub::BreakpointType::Execute) {
             uc_reg_write(uc, UC_ARM64_REG_PC, &last_bkpt.address);
         }
+
         Kernel::Thread* thread = Kernel::GetCurrentThread();
         SaveContext(thread->GetContext());
-        if (last_bkpt_hit || GDBStub::GetCpuStepFlag()) {
+        if (last_bkpt_hit || GDBStub::IsMemoryBreak() || GDBStub::GetCpuStepFlag()) {
             last_bkpt_hit = false;
             GDBStub::Break();
             GDBStub::SendTrap(thread, 5);

src/core/arm/unicorn/arm_unicorn.h

@@ -18,7 +18,7 @@ namespace Core {
 class ARM_Unicorn final : public ARM_Interface {
 public:
     explicit ARM_Unicorn(Timing::CoreTiming& core_timing);
-    ~ARM_Unicorn();
+    ~ARM_Unicorn() override;
 
     void MapBackingMemory(VAddr address, std::size_t size, u8* memory,
                           Kernel::VMAPermission perms) override;
@@ -50,7 +50,7 @@ private:
     uc_engine* uc{};
     Timing::CoreTiming& core_timing;
     GDBStub::BreakpointAddress last_bkpt{};
-    bool last_bkpt_hit;
+    bool last_bkpt_hit = false;
 };
 
 } // namespace Core

src/core/file_sys/control_metadata.cpp

@@ -67,7 +67,7 @@ std::string NACP::GetDeveloperName(Language language) const {
 }
 
 u64 NACP::GetTitleId() const {
-    return raw.title_id;
+    return raw.save_data_owner_id;
 }
 
 u64 NACP::GetDLCBaseTitleId() const {
@@ -80,11 +80,11 @@ std::string NACP::GetVersionString() const {
 }
 
 u64 NACP::GetDefaultNormalSaveSize() const {
-    return raw.normal_save_data_size;
+    return raw.user_account_save_data_size;
 }
 
 u64 NACP::GetDefaultJournalSaveSize() const {
-    return raw.journal_sava_data_size;
+    return raw.user_account_save_data_journal_size;
 }
 
 std::vector<u8> NACP::GetRawBytes() const {

src/core/file_sys/control_metadata.h

@@ -38,23 +38,35 @@ struct RawNACP {
     u8 video_capture_mode;
     bool data_loss_confirmation;
     INSERT_PADDING_BYTES(1);
-    u64_le title_id;
+    u64_le presence_group_id;
     std::array<u8, 0x20> rating_age;
     std::array<char, 0x10> version_string;
     u64_le dlc_base_title_id;
-    u64_le title_id_2;
-    u64_le normal_save_data_size;
-    u64_le journal_sava_data_size;
-    INSERT_PADDING_BYTES(0x18);
-    u64_le product_code;
+    u64_le save_data_owner_id;
+    u64_le user_account_save_data_size;
+    u64_le user_account_save_data_journal_size;
+    u64_le device_save_data_size;
+    u64_le device_save_data_journal_size;
+    u64_le bcat_delivery_cache_storage_size;
+    char application_error_code_category[8];
     std::array<u64_le, 0x8> local_communication;
     u8 logo_type;
     u8 logo_handling;
     bool runtime_add_on_content_install;
     INSERT_PADDING_BYTES(5);
-    u64_le title_id_update;
-    std::array<u8, 0x40> bcat_passphrase;
-    INSERT_PADDING_BYTES(0xEC0);
+    u64_le seed_for_pseudo_device_id;
+    std::array<u8, 0x41> bcat_passphrase;
+    INSERT_PADDING_BYTES(7);
+    u64_le user_account_save_data_max_size;
+    u64_le user_account_save_data_max_journal_size;
+    u64_le device_save_data_max_size;
+    u64_le device_save_data_max_journal_size;
+    u64_le temporary_storage_size;
+    u64_le cache_storage_size;
+    u64_le cache_storage_journal_size;
+    u64_le cache_storage_data_and_journal_max_size;
+    u64_le cache_storage_max_index;
+    INSERT_PADDING_BYTES(0xE70);
 };
 static_assert(sizeof(RawNACP) == 0x4000, "RawNACP has incorrect size.");
 

src/core/file_sys/fsmitm_romfsbuild.cpp

@@ -23,6 +23,7 @@
  */
 
 #include <cstring>
+#include <string_view>
 #include "common/alignment.h"
 #include "common/assert.h"
 #include "core/file_sys/fsmitm_romfsbuild.h"
@@ -97,7 +98,8 @@ struct RomFSBuildFileContext {
     VirtualFile source;
 };
 
-static u32 romfs_calc_path_hash(u32 parent, std::string path, u32 start, std::size_t path_len) {
+static u32 romfs_calc_path_hash(u32 parent, std::string_view path, u32 start,
+                                std::size_t path_len) {
     u32 hash = parent ^ 123456789;
     for (u32 i = 0; i < path_len; i++) {
         hash = (hash >> 5) | (hash << 27);

src/core/file_sys/nca_metadata.cpp

@@ -10,14 +10,6 @@
 
 namespace FileSys {
 
-bool operator>=(TitleType lhs, TitleType rhs) {
-    return static_cast<std::size_t>(lhs) >= static_cast<std::size_t>(rhs);
-}
-
-bool operator<=(TitleType lhs, TitleType rhs) {
-    return static_cast<std::size_t>(lhs) <= static_cast<std::size_t>(rhs);
-}
-
 CNMT::CNMT(VirtualFile file) {
     if (file->ReadObject(&header) != sizeof(CNMTHeader))
         return;

src/core/file_sys/nca_metadata.h

@@ -29,9 +29,6 @@ enum class TitleType : u8 {
     DeltaTitle = 0x83,
 };
 
-bool operator>=(TitleType lhs, TitleType rhs);
-bool operator<=(TitleType lhs, TitleType rhs);
-
 enum class ContentRecordType : u8 {
     Meta = 0,
     Program = 1,

src/core/file_sys/program_metadata.cpp

@@ -3,7 +3,6 @@
 // Refer to the license.txt file included.
 
 #include <cstddef>
-#include <cstring>
 #include <vector>
 
 #include "common/logging/log.h"
@@ -17,28 +16,30 @@ ProgramMetadata::ProgramMetadata() = default;
 ProgramMetadata::~ProgramMetadata() = default;
 
 Loader::ResultStatus ProgramMetadata::Load(VirtualFile file) {
-    std::size_t total_size = static_cast<std::size_t>(file->GetSize());
-    if (total_size < sizeof(Header))
+    const std::size_t total_size = file->GetSize();
+    if (total_size < sizeof(Header)) {
         return Loader::ResultStatus::ErrorBadNPDMHeader;
+    }
 
-    // TODO(DarkLordZach): Use ReadObject when Header/AcidHeader becomes trivially copyable.
-    std::vector<u8> npdm_header_data = file->ReadBytes(sizeof(Header));
-    if (sizeof(Header) != npdm_header_data.size())
+    if (sizeof(Header) != file->ReadObject(&npdm_header)) {
         return Loader::ResultStatus::ErrorBadNPDMHeader;
-    std::memcpy(&npdm_header, npdm_header_data.data(), sizeof(Header));
+    }
 
-    std::vector<u8> acid_header_data = file->ReadBytes(sizeof(AcidHeader), npdm_header.acid_offset);
-    if (sizeof(AcidHeader) != acid_header_data.size())
+    if (sizeof(AcidHeader) != file->ReadObject(&acid_header, npdm_header.acid_offset)) {
         return Loader::ResultStatus::ErrorBadACIDHeader;
-    std::memcpy(&acid_header, acid_header_data.data(), sizeof(AcidHeader));
+    }
 
-    if (sizeof(AciHeader) != file->ReadObject(&aci_header, npdm_header.aci_offset))
+    if (sizeof(AciHeader) != file->ReadObject(&aci_header, npdm_header.aci_offset)) {
         return Loader::ResultStatus::ErrorBadACIHeader;
+    }
 
-    if (sizeof(FileAccessControl) != file->ReadObject(&acid_file_access, acid_header.fac_offset))
+    if (sizeof(FileAccessControl) != file->ReadObject(&acid_file_access, acid_header.fac_offset)) {
         return Loader::ResultStatus::ErrorBadFileAccessControl;
-    if (sizeof(FileAccessHeader) != file->ReadObject(&aci_file_access, aci_header.fah_offset))
+    }
+
+    if (sizeof(FileAccessHeader) != file->ReadObject(&aci_file_access, aci_header.fah_offset)) {
         return Loader::ResultStatus::ErrorBadFileAccessHeader;
+    }
 
     aci_kernel_capabilities.resize(aci_header.kac_size / sizeof(u32));
     const u64 read_size = aci_header.kac_size;

src/core/file_sys/program_metadata.h

@@ -58,7 +58,6 @@ public:
     void Print() const;
 
 private:
-    // TODO(DarkLordZach): BitField is not trivially copyable.
     struct Header {
         std::array<char, 4> magic;
         std::array<u8, 8> reserved;
@@ -85,7 +84,6 @@ private:
 
     static_assert(sizeof(Header) == 0x80, "NPDM header structure size is wrong");
 
-    // TODO(DarkLordZach): BitField is not trivially copyable.
     struct AcidHeader {
         std::array<u8, 0x100> signature;
         std::array<u8, 0x100> nca_modulus;

src/core/file_sys/savedata_factory.cpp

@@ -16,8 +16,10 @@ namespace FileSys {
 constexpr char SAVE_DATA_SIZE_FILENAME[] = ".yuzu_save_size";
 
 std::string SaveDataDescriptor::DebugInfo() const {
-    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);
+    return fmt::format("[type={:02X}, title_id={:016X}, user_id={:016X}{:016X}, save_id={:016X}, "
+                       "rank={}, index={}]",
+                       static_cast<u8>(type), title_id, user_id[1], user_id[0], save_id,
+                       static_cast<u8>(rank), index);
 }
 
 SaveDataFactory::SaveDataFactory(VirtualDir save_directory) : dir(std::move(save_directory)) {
@@ -28,7 +30,7 @@ SaveDataFactory::SaveDataFactory(VirtualDir save_directory) : dir(std::move(save
 
 SaveDataFactory::~SaveDataFactory() = default;
 
-ResultVal<VirtualDir> SaveDataFactory::Open(SaveDataSpaceId space, SaveDataDescriptor meta) {
+ResultVal<VirtualDir> SaveDataFactory::Open(SaveDataSpaceId space, const SaveDataDescriptor& meta) {
     if (meta.type == SaveDataType::SystemSaveData || meta.type == SaveDataType::SaveData) {
         if (meta.zero_1 != 0) {
             LOG_WARNING(Service_FS,

src/core/file_sys/savedata_factory.h

@@ -32,12 +32,19 @@ enum class SaveDataType : u8 {
     CacheStorage = 5,
 };
 
+enum class SaveDataRank : u8 {
+    Primary,
+    Secondary,
+};
+
 struct SaveDataDescriptor {
     u64_le title_id;
     u128 user_id;
     u64_le save_id;
     SaveDataType type;
-    INSERT_PADDING_BYTES(7);
+    SaveDataRank rank;
+    u16_le index;
+    INSERT_PADDING_BYTES(4);
     u64_le zero_1;
     u64_le zero_2;
     u64_le zero_3;
@@ -57,7 +64,7 @@ public:
     explicit SaveDataFactory(VirtualDir dir);
     ~SaveDataFactory();
 
-    ResultVal<VirtualDir> Open(SaveDataSpaceId space, SaveDataDescriptor meta);
+    ResultVal<VirtualDir> Open(SaveDataSpaceId space, const SaveDataDescriptor& meta);
 
     VirtualDir GetSaveDataSpaceDirectory(SaveDataSpaceId space) const;
 

src/core/gdbstub/gdbstub.cpp

@@ -1030,7 +1030,7 @@ static void Step() {
 
 /// Tell the CPU if we hit a memory breakpoint.
 bool IsMemoryBreak() {
-    if (IsConnected()) {
+    if (!IsConnected()) {
         return false;
     }
 

src/core/hle/ipc_helpers.h

@@ -139,10 +139,8 @@ public:
             context->AddDomainObject(std::move(iface));
         } else {
             auto& kernel = Core::System::GetInstance().Kernel();
-            auto sessions =
+            auto [server, client] =
                 Kernel::ServerSession::CreateSessionPair(kernel, iface->GetServiceName());
-            auto server = std::get<Kernel::SharedPtr<Kernel::ServerSession>>(sessions);
-            auto client = std::get<Kernel::SharedPtr<Kernel::ClientSession>>(sessions);
             iface->ClientConnected(server);
             context->AddMoveObject(std::move(client));
         }

src/core/hle/kernel/client_port.cpp

@@ -2,8 +2,6 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include <tuple>
-
 #include "core/hle/kernel/client_port.h"
 #include "core/hle/kernel/client_session.h"
 #include "core/hle/kernel/errors.h"
@@ -31,18 +29,18 @@ ResultVal<SharedPtr<ClientSession>> ClientPort::Connect() {
     active_sessions++;
 
     // Create a new session pair, let the created sessions inherit the parent port's HLE handler.
-    auto sessions = ServerSession::CreateSessionPair(kernel, server_port->GetName(), this);
+    auto [server, client] = ServerSession::CreateSessionPair(kernel, server_port->GetName(), this);
 
     if (server_port->HasHLEHandler()) {
-        server_port->GetHLEHandler()->ClientConnected(std::get<SharedPtr<ServerSession>>(sessions));
+        server_port->GetHLEHandler()->ClientConnected(server);
     } else {
-        server_port->AppendPendingSession(std::get<SharedPtr<ServerSession>>(sessions));
+        server_port->AppendPendingSession(server);
     }
 
     // Wake the threads waiting on the ServerPort
     server_port->WakeupAllWaitingThreads();
 
-    return MakeResult(std::get<SharedPtr<ClientSession>>(sessions));
+    return MakeResult(client);
 }
 
 void ClientPort::ConnectionClosed() {

src/core/hle/kernel/kernel.cpp

@@ -21,6 +21,7 @@
 #include "core/hle/kernel/thread.h"
 #include "core/hle/lock.h"
 #include "core/hle/result.h"
+#include "core/memory.h"
 
 namespace Kernel {
 
@@ -181,6 +182,7 @@ void KernelCore::AppendNewProcess(SharedPtr<Process> process) {
 
 void KernelCore::MakeCurrentProcess(Process* process) {
     impl->current_process = process;
+    Memory::SetCurrentPageTable(&process->VMManager().page_table);
 }
 
 Process* KernelCore::CurrentProcess() {

src/core/hle/kernel/process.cpp

@@ -32,9 +32,6 @@ namespace {
  * @param priority The priority to give the main thread
  */
 void SetupMainThread(Process& owner_process, KernelCore& kernel, VAddr entry_point, u32 priority) {
-    // Setup page table so we can write to memory
-    Memory::SetCurrentPageTable(&owner_process.VMManager().page_table);
-
     // Initialize new "main" thread
     const VAddr stack_top = owner_process.VMManager().GetTLSIORegionEndAddress();
     auto thread_res = Thread::Create(kernel, "main", entry_point, priority, 0,

src/core/hle/kernel/scheduler.cpp

@@ -101,7 +101,6 @@ void Scheduler::SwitchContext(Thread* new_thread) {
         auto* const thread_owner_process = current_thread->GetOwnerProcess();
         if (previous_process != thread_owner_process) {
             system.Kernel().MakeCurrentProcess(thread_owner_process);
-            Memory::SetCurrentPageTable(&thread_owner_process->VMManager().page_table);
         }
 
         cpu_core.LoadContext(new_thread->GetContext());

src/core/hle/kernel/server_port.cpp

@@ -39,9 +39,8 @@ void ServerPort::Acquire(Thread* thread) {
     ASSERT_MSG(!ShouldWait(thread), "object unavailable!");
 }
 
-std::tuple<SharedPtr<ServerPort>, SharedPtr<ClientPort>> ServerPort::CreatePortPair(
-    KernelCore& kernel, u32 max_sessions, std::string name) {
-
+ServerPort::PortPair ServerPort::CreatePortPair(KernelCore& kernel, u32 max_sessions,
+                                                std::string name) {
     SharedPtr<ServerPort> server_port(new ServerPort(kernel));
     SharedPtr<ClientPort> client_port(new ClientPort(kernel));
 
@@ -51,7 +50,7 @@ std::tuple<SharedPtr<ServerPort>, SharedPtr<ClientPort>> ServerPort::CreatePortP
     client_port->max_sessions = max_sessions;
     client_port->active_sessions = 0;
 
-    return std::make_tuple(std::move(server_port), std::move(client_port));
+    return std::make_pair(std::move(server_port), std::move(client_port));
 }
 
 } // namespace Kernel

src/core/hle/kernel/server_port.h

@@ -6,7 +6,7 @@
 
 #include <memory>
 #include <string>
-#include <tuple>
+#include <utility>
 #include <vector>
 #include "common/common_types.h"
 #include "core/hle/kernel/object.h"
@@ -23,6 +23,7 @@ class SessionRequestHandler;
 class ServerPort final : public WaitObject {
 public:
     using HLEHandler = std::shared_ptr<SessionRequestHandler>;
+    using PortPair = std::pair<SharedPtr<ServerPort>, SharedPtr<ClientPort>>;
 
     /**
      * Creates a pair of ServerPort and an associated ClientPort.
@@ -32,8 +33,8 @@ public:
      * @param name Optional name of the ports
      * @return The created port tuple
      */
-    static std::tuple<SharedPtr<ServerPort>, SharedPtr<ClientPort>> CreatePortPair(
-        KernelCore& kernel, u32 max_sessions, std::string name = "UnknownPort");
+    static PortPair CreatePortPair(KernelCore& kernel, u32 max_sessions,
+                                   std::string name = "UnknownPort");
 
     std::string GetTypeName() const override {
         return "ServerPort";

src/core/hle/kernel/server_session.cpp

@@ -204,6 +204,6 @@ ServerSession::SessionPair ServerSession::CreateSessionPair(KernelCore& kernel,
     client_session->parent = parent;
     server_session->parent = parent;
 
-    return std::make_tuple(std::move(server_session), std::move(client_session));
+    return std::make_pair(std::move(server_session), std::move(client_session));
 }
 } // namespace Kernel

src/core/hle/kernel/server_session.h

@@ -6,6 +6,7 @@
 
 #include <memory>
 #include <string>
+#include <utility>
 #include <vector>
 
 #include "core/hle/kernel/object.h"
@@ -41,6 +42,10 @@ public:
         return "ServerSession";
     }
 
+    std::string GetName() const override {
+        return name;
+    }
+
     static const HandleType HANDLE_TYPE = HandleType::ServerSession;
     HandleType GetHandleType() const override {
         return HANDLE_TYPE;
@@ -54,7 +59,7 @@ public:
         return parent.get();
     }
 
-    using SessionPair = std::tuple<SharedPtr<ServerSession>, SharedPtr<ClientSession>>;
+    using SessionPair = std::pair<SharedPtr<ServerSession>, SharedPtr<ClientSession>>;
 
     /**
      * Creates a pair of ServerSession and an associated ClientSession.

src/core/hle/kernel/svc.cpp

@@ -1339,6 +1339,20 @@ static ResultCode WaitProcessWideKeyAtomic(VAddr mutex_addr, VAddr condition_var
         "called mutex_addr={:X}, condition_variable_addr={:X}, thread_handle=0x{:08X}, timeout={}",
         mutex_addr, condition_variable_addr, thread_handle, nano_seconds);
 
+    if (Memory::IsKernelVirtualAddress(mutex_addr)) {
+        LOG_ERROR(
+            Kernel_SVC,
+            "Given mutex address must not be within the kernel address space. address=0x{:016X}",
+            mutex_addr);
+        return ERR_INVALID_ADDRESS_STATE;
+    }
+
+    if (!Common::IsWordAligned(mutex_addr)) {
+        LOG_ERROR(Kernel_SVC, "Given mutex address must be word-aligned. address=0x{:016X}",
+                  mutex_addr);
+        return ERR_INVALID_ADDRESS;
+    }
+
     auto* const current_process = Core::System::GetInstance().Kernel().CurrentProcess();
     const auto& handle_table = current_process->GetHandleTable();
     SharedPtr<Thread> thread = handle_table.Get<Thread>(thread_handle);

src/core/hle/result.h

@@ -119,10 +119,6 @@ union ResultCode {
     BitField<0, 9, ErrorModule> module;
     BitField<9, 13, u32> description;
 
-    // The last bit of `level` is checked by apps and the kernel to determine if a result code is an
-    // error
-    BitField<31, 1, u32> is_error;
-
     constexpr explicit ResultCode(u32 raw) : raw(raw) {}
 
     constexpr ResultCode(ErrorModule module_, u32 description_)

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

@@ -363,8 +363,6 @@ void ISelfController::SetPerformanceModeChangedNotification(Kernel::HLERequestCo
 void ISelfController::SetFocusHandlingMode(Kernel::HLERequestContext& ctx) {
     // Takes 3 input u8s with each field located immediately after the previous
     // u8, these are bool flags. No output.
-    LOG_WARNING(Service_AM, "(STUBBED) called");
-
     IPC::RequestParser rp{ctx};
 
     struct FocusHandlingModeParams {
@@ -372,7 +370,10 @@ void ISelfController::SetFocusHandlingMode(Kernel::HLERequestContext& ctx) {
         u8 unknown1;
         u8 unknown2;
     };
-    auto flags = rp.PopRaw<FocusHandlingModeParams>();
+    const auto flags = rp.PopRaw<FocusHandlingModeParams>();
+
+    LOG_WARNING(Service_AM, "(STUBBED) called. unknown0={}, unknown1={}, unknown2={}",
+                flags.unknown0, flags.unknown1, flags.unknown2);
 
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);

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

@@ -2,9 +2,6 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include "common/logging/log.h"
-#include "core/hle/ipc_helpers.h"
-#include "core/hle/kernel/hle_ipc.h"
 #include "core/hle/service/audio/audin_u.h"
 
 namespace Service::Audio {
@@ -33,7 +30,6 @@ public:
 
         RegisterHandlers(functions);
     }
-    ~IAudioIn() = default;
 };
 
 AudInU::AudInU() : ServiceFramework("audin:u") {

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

@@ -150,7 +150,6 @@ private:
     void GetReleasedAudioOutBufferImpl(Kernel::HLERequestContext& ctx) {
         LOG_DEBUG(Service_Audio, "called {}", ctx.Description());
 
-        IPC::RequestParser rp{ctx};
         const u64 max_count{ctx.GetWriteBufferSize() / sizeof(u64)};
         const auto released_buffers{audio_core.GetTagsAndReleaseBuffers(stream, max_count)};
 
@@ -194,12 +193,9 @@ private:
 void AudOutU::ListAudioOutsImpl(Kernel::HLERequestContext& ctx) {
     LOG_DEBUG(Service_Audio, "called");
 
-    IPC::RequestParser rp{ctx};
-
     ctx.WriteBuffer(DefaultDevice);
 
     IPC::ResponseBuilder rb{ctx, 3};
-
     rb.Push(RESULT_SUCCESS);
     rb.Push<u32>(1); // Amount of audio devices
 }

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

@@ -2,9 +2,6 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include "common/logging/log.h"
-#include "core/hle/ipc_helpers.h"
-#include "core/hle/kernel/hle_ipc.h"
 #include "core/hle/service/audio/audrec_u.h"
 
 namespace Service::Audio {
@@ -30,7 +27,6 @@ public:
 
         RegisterHandlers(functions);
     }
-    ~IFinalOutputRecorder() = default;
 };
 
 AudRecU::AudRecU() : ServiceFramework("audrec:u") {

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

@@ -10,6 +10,7 @@
 #include "common/alignment.h"
 #include "common/common_funcs.h"
 #include "common/logging/log.h"
+#include "common/string_util.h"
 #include "core/core.h"
 #include "core/hle/ipc_helpers.h"
 #include "core/hle/kernel/hle_ipc.h"
@@ -184,7 +185,6 @@ public:
 private:
     void ListAudioDeviceName(Kernel::HLERequestContext& ctx) {
         LOG_WARNING(Service_Audio, "(STUBBED) called");
-        IPC::RequestParser rp{ctx};
 
         constexpr std::array<char, 15> audio_interface{{"AudioInterface"}};
         ctx.WriteBuffer(audio_interface);
@@ -195,13 +195,13 @@ private:
     }
 
     void SetAudioDeviceOutputVolume(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_Audio, "(STUBBED) called");
-
         IPC::RequestParser rp{ctx};
-        f32 volume = static_cast<f32>(rp.Pop<u32>());
+        const f32 volume = rp.Pop<f32>();
 
-        auto file_buffer = ctx.ReadBuffer();
-        auto end = std::find(file_buffer.begin(), file_buffer.end(), '\0');
+        const auto device_name_buffer = ctx.ReadBuffer();
+        const std::string name = Common::StringFromBuffer(device_name_buffer);
+
+        LOG_WARNING(Service_Audio, "(STUBBED) called. name={}, volume={}", name, volume);
 
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
@@ -209,7 +209,6 @@ private:
 
     void GetActiveAudioDeviceName(Kernel::HLERequestContext& ctx) {
         LOG_WARNING(Service_Audio, "(STUBBED) called");
-        IPC::RequestParser rp{ctx};
 
         constexpr std::array<char, 12> audio_interface{{"AudioDevice"}};
         ctx.WriteBuffer(audio_interface);

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

@@ -197,13 +197,16 @@ ResultCode VfsDirectoryServiceWrapper::RenameDirectory(const std::string& src_pa
 
 ResultVal<FileSys::VirtualFile> VfsDirectoryServiceWrapper::OpenFile(const std::string& path_,
                                                                      FileSys::Mode mode) const {
-    std::string path(FileUtil::SanitizePath(path_));
-    auto npath = path;
-    while (npath.size() > 0 && (npath[0] == '/' || npath[0] == '\\'))
-        npath = npath.substr(1);
+    const std::string path(FileUtil::SanitizePath(path_));
+    std::string_view npath = path;
+    while (!npath.empty() && (npath[0] == '/' || npath[0] == '\\')) {
+        npath.remove_prefix(1);
+    }
+
     auto file = backing->GetFileRelative(npath);
-    if (file == nullptr)
+    if (file == nullptr) {
         return FileSys::ERROR_PATH_NOT_FOUND;
+    }
 
     if (mode == FileSys::Mode::Append) {
         return MakeResult<FileSys::VirtualFile>(
@@ -319,15 +322,15 @@ ResultVal<FileSys::VirtualFile> OpenRomFS(u64 title_id, FileSys::StorageId stora
 }
 
 ResultVal<FileSys::VirtualDir> OpenSaveData(FileSys::SaveDataSpaceId space,
-                                            FileSys::SaveDataDescriptor save_struct) {
+                                            const FileSys::SaveDataDescriptor& descriptor) {
     LOG_TRACE(Service_FS, "Opening Save Data for space_id={:01X}, save_struct={}",
-              static_cast<u8>(space), save_struct.DebugInfo());
+              static_cast<u8>(space), descriptor.DebugInfo());
 
     if (save_data_factory == nullptr) {
         return FileSys::ERROR_ENTITY_NOT_FOUND;
     }
 
-    return save_data_factory->Open(space, save_struct);
+    return save_data_factory->Open(space, descriptor);
 }
 
 ResultVal<FileSys::VirtualDir> OpenSaveDataSpace(FileSys::SaveDataSpaceId space) {

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

@@ -46,7 +46,7 @@ ResultVal<FileSys::VirtualFile> OpenRomFSCurrentProcess();
 ResultVal<FileSys::VirtualFile> OpenRomFS(u64 title_id, FileSys::StorageId storage_id,
                                           FileSys::ContentRecordType type);
 ResultVal<FileSys::VirtualDir> OpenSaveData(FileSys::SaveDataSpaceId space,
-                                            FileSys::SaveDataDescriptor save_struct);
+                                            const FileSys::SaveDataDescriptor& descriptor);
 ResultVal<FileSys::VirtualDir> OpenSaveDataSpace(FileSys::SaveDataSpaceId space);
 ResultVal<FileSys::VirtualDir> OpenSDMC();
 

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

@@ -315,61 +315,53 @@ public:
     void CreateFile(Kernel::HLERequestContext& ctx) {
         IPC::RequestParser rp{ctx};
 
-        auto file_buffer = ctx.ReadBuffer();
-        std::string name = Common::StringFromBuffer(file_buffer);
+        const auto file_buffer = ctx.ReadBuffer();
+        const std::string name = Common::StringFromBuffer(file_buffer);
 
-        u64 mode = rp.Pop<u64>();
-        u32 size = rp.Pop<u32>();
+        const u64 mode = rp.Pop<u64>();
+        const u32 size = rp.Pop<u32>();
 
-        LOG_DEBUG(Service_FS, "called file {} mode 0x{:X} size 0x{:08X}", name, mode, size);
+        LOG_DEBUG(Service_FS, "called. file={}, mode=0x{:X}, size=0x{:08X}", name, mode, size);
 
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(backend.CreateFile(name, size));
     }
 
     void DeleteFile(Kernel::HLERequestContext& ctx) {
-        IPC::RequestParser rp{ctx};
+        const auto file_buffer = ctx.ReadBuffer();
+        const std::string name = Common::StringFromBuffer(file_buffer);
 
-        auto file_buffer = ctx.ReadBuffer();
-        std::string name = Common::StringFromBuffer(file_buffer);
-
-        LOG_DEBUG(Service_FS, "called file {}", name);
+        LOG_DEBUG(Service_FS, "called. file={}", name);
 
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(backend.DeleteFile(name));
     }
 
     void CreateDirectory(Kernel::HLERequestContext& ctx) {
-        IPC::RequestParser rp{ctx};
+        const auto file_buffer = ctx.ReadBuffer();
+        const std::string name = Common::StringFromBuffer(file_buffer);
 
-        auto file_buffer = ctx.ReadBuffer();
-        std::string name = Common::StringFromBuffer(file_buffer);
-
-        LOG_DEBUG(Service_FS, "called directory {}", name);
+        LOG_DEBUG(Service_FS, "called. directory={}", name);
 
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(backend.CreateDirectory(name));
     }
 
     void DeleteDirectory(Kernel::HLERequestContext& ctx) {
-        const IPC::RequestParser rp{ctx};
-
         const auto file_buffer = ctx.ReadBuffer();
-        std::string name = Common::StringFromBuffer(file_buffer);
+        const std::string name = Common::StringFromBuffer(file_buffer);
 
-        LOG_DEBUG(Service_FS, "called directory {}", name);
+        LOG_DEBUG(Service_FS, "called. directory={}", name);
 
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(backend.DeleteDirectory(name));
     }
 
     void DeleteDirectoryRecursively(Kernel::HLERequestContext& ctx) {
-        const IPC::RequestParser rp{ctx};
-
         const auto file_buffer = ctx.ReadBuffer();
-        std::string name = Common::StringFromBuffer(file_buffer);
+        const std::string name = Common::StringFromBuffer(file_buffer);
 
-        LOG_DEBUG(Service_FS, "called directory {}", name);
+        LOG_DEBUG(Service_FS, "called. directory={}", name);
 
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(backend.DeleteDirectoryRecursively(name));
@@ -386,18 +378,16 @@ public:
     }
 
     void RenameFile(Kernel::HLERequestContext& ctx) {
-        IPC::RequestParser rp{ctx};
-
         std::vector<u8> buffer;
         buffer.resize(ctx.BufferDescriptorX()[0].Size());
         Memory::ReadBlock(ctx.BufferDescriptorX()[0].Address(), buffer.data(), buffer.size());
-        std::string src_name = Common::StringFromBuffer(buffer);
+        const std::string src_name = Common::StringFromBuffer(buffer);
 
         buffer.resize(ctx.BufferDescriptorX()[1].Size());
         Memory::ReadBlock(ctx.BufferDescriptorX()[1].Address(), buffer.data(), buffer.size());
-        std::string dst_name = Common::StringFromBuffer(buffer);
+        const std::string dst_name = Common::StringFromBuffer(buffer);
 
-        LOG_DEBUG(Service_FS, "called file '{}' to file '{}'", src_name, dst_name);
+        LOG_DEBUG(Service_FS, "called. file '{}' to file '{}'", src_name, dst_name);
 
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(backend.RenameFile(src_name, dst_name));
@@ -406,12 +396,12 @@ public:
     void OpenFile(Kernel::HLERequestContext& ctx) {
         IPC::RequestParser rp{ctx};
 
-        auto file_buffer = ctx.ReadBuffer();
-        std::string name = Common::StringFromBuffer(file_buffer);
+        const auto file_buffer = ctx.ReadBuffer();
+        const std::string name = Common::StringFromBuffer(file_buffer);
 
-        auto mode = static_cast<FileSys::Mode>(rp.Pop<u32>());
+        const auto mode = static_cast<FileSys::Mode>(rp.Pop<u32>());
 
-        LOG_DEBUG(Service_FS, "called file {} mode {}", name, static_cast<u32>(mode));
+        LOG_DEBUG(Service_FS, "called. file={}, mode={}", name, static_cast<u32>(mode));
 
         auto result = backend.OpenFile(name, mode);
         if (result.Failed()) {
@@ -430,13 +420,13 @@ public:
     void OpenDirectory(Kernel::HLERequestContext& ctx) {
         IPC::RequestParser rp{ctx};
 
-        auto file_buffer = ctx.ReadBuffer();
-        std::string name = Common::StringFromBuffer(file_buffer);
+        const auto file_buffer = ctx.ReadBuffer();
+        const std::string name = Common::StringFromBuffer(file_buffer);
 
         // TODO(Subv): Implement this filter.
-        u32 filter_flags = rp.Pop<u32>();
+        const u32 filter_flags = rp.Pop<u32>();
 
-        LOG_DEBUG(Service_FS, "called directory {} filter {}", name, filter_flags);
+        LOG_DEBUG(Service_FS, "called. directory={}, filter={}", name, filter_flags);
 
         auto result = backend.OpenDirectory(name);
         if (result.Failed()) {
@@ -453,12 +443,10 @@ public:
     }
 
     void GetEntryType(Kernel::HLERequestContext& ctx) {
-        IPC::RequestParser rp{ctx};
+        const auto file_buffer = ctx.ReadBuffer();
+        const std::string name = Common::StringFromBuffer(file_buffer);
 
-        auto file_buffer = ctx.ReadBuffer();
-        std::string name = Common::StringFromBuffer(file_buffer);
-
-        LOG_DEBUG(Service_FS, "called file {}", name);
+        LOG_DEBUG(Service_FS, "called. file={}", name);
 
         auto result = backend.GetEntryType(name);
         if (result.Failed()) {
@@ -616,7 +604,9 @@ private:
         u64_le save_id;
         u64_le title_id;
         u64_le save_image_size;
-        INSERT_PADDING_BYTES(0x28);
+        u16_le index;
+        FileSys::SaveDataRank rank;
+        INSERT_PADDING_BYTES(0x25);
     };
     static_assert(sizeof(SaveDataInfo) == 0x60, "SaveDataInfo has incorrect size.");
 
@@ -779,16 +769,17 @@ void FSP_SRV::CreateSaveDataFileSystem(Kernel::HLERequestContext& ctx) {
 }
 
 void FSP_SRV::OpenSaveDataFileSystem(Kernel::HLERequestContext& ctx) {
+    LOG_INFO(Service_FS, "called.");
+
+    struct Parameters {
+        FileSys::SaveDataSpaceId save_data_space_id;
+        FileSys::SaveDataDescriptor descriptor;
+    };
+
     IPC::RequestParser rp{ctx};
+    const auto parameters = rp.PopRaw<Parameters>();
 
-    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 dir = OpenSaveData(space_id, save_struct);
-
+    auto dir = OpenSaveData(parameters.save_data_space_id, parameters.descriptor);
     if (dir.Failed()) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 0};
         rb.Push(FileSys::ERROR_ENTITY_NOT_FOUND);

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

@@ -18,7 +18,7 @@ class nvmap;
 class nvdisp_disp0 final : public nvdevice {
 public:
     explicit nvdisp_disp0(std::shared_ptr<nvmap> nvmap_dev);
-    ~nvdisp_disp0();
+    ~nvdisp_disp0() override;
 
     u32 ioctl(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override;
 

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

@@ -17,7 +17,7 @@ namespace Service::Nvidia {
 class NVDRV final : public ServiceFramework<NVDRV> {
 public:
     NVDRV(std::shared_ptr<Module> nvdrv, const char* name);
-    ~NVDRV();
+    ~NVDRV() override;
 
 private:
     void Open(Kernel::HLERequestContext& ctx);

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

@@ -11,7 +11,7 @@ namespace Service::Nvidia {
 class NVMEMP final : public ServiceFramework<NVMEMP> {
 public:
     NVMEMP();
-    ~NVMEMP();
+    ~NVMEMP() override;
 
 private:
     void Cmd0(Kernel::HLERequestContext& ctx);

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

@@ -21,12 +21,13 @@
 #include "core/hle/service/vi/display/vi_display.h"
 #include "core/hle/service/vi/layer/vi_layer.h"
 #include "core/perf_stats.h"
+#include "core/settings.h"
 #include "video_core/renderer_base.h"
 
 namespace Service::NVFlinger {
 
-constexpr std::size_t SCREEN_REFRESH_RATE = 60;
-constexpr s64 frame_ticks = static_cast<s64>(Core::Timing::BASE_CLOCK_RATE / SCREEN_REFRESH_RATE);
+constexpr s64 frame_ticks = static_cast<s64>(Core::Timing::BASE_CLOCK_RATE / 60);
+constexpr s64 frame_ticks_30fps = static_cast<s64>(Core::Timing::BASE_CLOCK_RATE / 30);
 
 NVFlinger::NVFlinger(Core::Timing::CoreTiming& core_timing) : core_timing{core_timing} {
     displays.emplace_back(0, "Default");
@@ -36,13 +37,15 @@ NVFlinger::NVFlinger(Core::Timing::CoreTiming& core_timing) : core_timing{core_t
     displays.emplace_back(4, "Null");
 
     // Schedule the screen composition events
-    composition_event =
-        core_timing.RegisterEvent("ScreenComposition", [this](u64 userdata, s64 cycles_late) {
+    const auto ticks = Settings::values.force_30fps_mode ? frame_ticks_30fps : frame_ticks;
+
+    composition_event = core_timing.RegisterEvent(
+        "ScreenComposition", [this, ticks](u64 userdata, s64 cycles_late) {
             Compose();
-            this->core_timing.ScheduleEvent(frame_ticks - cycles_late, composition_event);
+            this->core_timing.ScheduleEvent(ticks - cycles_late, composition_event);
         });
 
-    core_timing.ScheduleEvent(frame_ticks, composition_event);
+    core_timing.ScheduleEvent(ticks, composition_event);
 }
 
 NVFlinger::~NVFlinger() {
@@ -62,6 +65,7 @@ std::optional<u64> NVFlinger::OpenDisplay(std::string_view name) {
     const auto itr =
         std::find_if(displays.begin(), displays.end(),
                      [&](const VI::Display& display) { return display.GetName() == name; });
+
     if (itr == displays.end()) {
         return {};
     }

src/core/hle/service/service.h

@@ -90,7 +90,7 @@ private:
                            Kernel::HLERequestContext& ctx);
 
     ServiceFrameworkBase(const char* service_name, u32 max_sessions, InvokerFn* handler_invoker);
-    ~ServiceFrameworkBase();
+    ~ServiceFrameworkBase() override;
 
     void RegisterHandlersBase(const FunctionInfoBase* functions, std::size_t n);
     void ReportUnimplementedFunction(Kernel::HLERequestContext& ctx, const FunctionInfoBase* info);

src/core/hle/service/set/set_cal.h

@@ -11,7 +11,7 @@ namespace Service::Set {
 class SET_CAL final : public ServiceFramework<SET_CAL> {
 public:
     explicit SET_CAL();
-    ~SET_CAL();
+    ~SET_CAL() override;
 };
 
 } // namespace Service::Set

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

@@ -8,12 +8,20 @@
 namespace Service::Sockets {
 
 void SFDNSRES::GetAddrInfo(Kernel::HLERequestContext& ctx) {
-    IPC::RequestParser rp{ctx};
+    struct Parameters {
+        u8 use_nsd_resolve;
+        u32 unknown;
+        u64 process_id;
+    };
 
-    LOG_WARNING(Service, "(STUBBED) called");
+    IPC::RequestParser rp{ctx};
+    const auto parameters = rp.PopRaw<Parameters>();
+
+    LOG_WARNING(Service,
+                "(STUBBED) called. use_nsd_resolve={}, unknown=0x{:08X}, process_id=0x{:016X}",
+                parameters.use_nsd_resolve, parameters.unknown, parameters.process_id);
 
     IPC::ResponseBuilder rb{ctx, 2};
-
     rb.Push(RESULT_SUCCESS);
 }
 

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

@@ -26,9 +26,7 @@ Module::Interface::~Interface() = default;
 void Module::Interface::GetRandomBytes(Kernel::HLERequestContext& ctx) {
     LOG_DEBUG(Service_SPL, "called");
 
-    IPC::RequestParser rp{ctx};
-
-    std::size_t size = ctx.GetWriteBufferSize();
+    const std::size_t size = ctx.GetWriteBufferSize();
 
     std::uniform_int_distribution<u16> distribution(0, std::numeric_limits<u8>::max());
     std::vector<u8> data(size);

src/core/hle/service/ssl/ssl.cpp

@@ -64,13 +64,19 @@ public:
         };
         RegisterHandlers(functions);
     }
-    ~ISslContext() = default;
 
 private:
     void SetOption(Kernel::HLERequestContext& ctx) {
-        LOG_WARNING(Service_SSL, "(STUBBED) called");
+        struct Parameters {
+            u8 enable;
+            u32 option;
+        };
 
         IPC::RequestParser rp{ctx};
+        const auto parameters = rp.PopRaw<Parameters>();
+
+        LOG_WARNING(Service_SSL, "(STUBBED) called. enable={}, option={}", parameters.enable,
+                    parameters.option);
 
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);

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

@@ -498,7 +498,6 @@ public:
         };
         RegisterHandlers(functions);
     }
-    ~IHOSBinderDriver() = default;
 
 private:
     enum class TransactionId {
@@ -692,7 +691,6 @@ public:
         };
         RegisterHandlers(functions);
     }
-    ~ISystemDisplayService() = default;
 
 private:
     void SetLayerZ(Kernel::HLERequestContext& ctx) {
@@ -818,7 +816,6 @@ public:
         };
         RegisterHandlers(functions);
     }
-    ~IManagerDisplayService() = default;
 
 private:
     void CloseDisplay(Kernel::HLERequestContext& ctx) {
@@ -884,7 +881,6 @@ private:
 class IApplicationDisplayService final : public ServiceFramework<IApplicationDisplayService> {
 public:
     explicit IApplicationDisplayService(std::shared_ptr<NVFlinger::NVFlinger> nv_flinger);
-    ~IApplicationDisplayService() = default;
 
 private:
     enum class ConvertedScaleMode : u64 {
@@ -1037,7 +1033,6 @@ private:
     void ListDisplays(Kernel::HLERequestContext& ctx) {
         LOG_WARNING(Service_VI, "(STUBBED) called");
 
-        IPC::RequestParser rp{ctx};
         DisplayInfo display_info;
         display_info.width *= static_cast<u64>(Settings::values.resolution_factor);
         display_info.height *= static_cast<u64>(Settings::values.resolution_factor);

src/core/loader/xci.h

@@ -22,7 +22,7 @@ class AppLoader_NCA;
 class AppLoader_XCI final : public AppLoader {
 public:
     explicit AppLoader_XCI(FileSys::VirtualFile file);
-    ~AppLoader_XCI();
+    ~AppLoader_XCI() override;
 
     /**
      * Returns the type of the file

src/core/memory.cpp

@@ -38,10 +38,6 @@ void SetCurrentPageTable(Common::PageTable* page_table) {
     }
 }
 
-Common::PageTable* GetCurrentPageTable() {
-    return current_page_table;
-}
-
 static void MapPages(Common::PageTable& page_table, VAddr base, u64 size, u8* memory,
                      Common::PageType type) {
     LOG_DEBUG(HW_Memory, "Mapping {} onto {:016X}-{:016X}", fmt::ptr(memory), base * PAGE_SIZE,

src/core/memory.h

@@ -28,16 +28,6 @@ constexpr u64 PAGE_MASK = PAGE_SIZE - 1;
 
 /// Virtual user-space memory regions
 enum : VAddr {
-    /// Read-only page containing kernel and system configuration values.
-    CONFIG_MEMORY_VADDR = 0x1FF80000,
-    CONFIG_MEMORY_SIZE = 0x00001000,
-    CONFIG_MEMORY_VADDR_END = CONFIG_MEMORY_VADDR + CONFIG_MEMORY_SIZE,
-
-    /// Usually read-only page containing mostly values read from hardware.
-    SHARED_PAGE_VADDR = 0x1FF81000,
-    SHARED_PAGE_SIZE = 0x00001000,
-    SHARED_PAGE_VADDR_END = SHARED_PAGE_VADDR + SHARED_PAGE_SIZE,
-
     /// TLS (Thread-Local Storage) related.
     TLS_ENTRY_SIZE = 0x200,
 
@@ -50,9 +40,8 @@ enum : VAddr {
     KERNEL_REGION_END = KERNEL_REGION_VADDR + KERNEL_REGION_SIZE,
 };
 
-/// Currently active page table
+/// Changes the currently active page table.
 void SetCurrentPageTable(Common::PageTable* page_table);
-Common::PageTable* GetCurrentPageTable();
 
 /// Determines if the given VAddr is valid for the specified process.
 bool IsValidVirtualAddress(const Kernel::Process& process, VAddr vaddr);

src/core/settings.h

@@ -393,6 +393,7 @@ struct Values {
     bool use_disk_shader_cache;
     bool use_accurate_gpu_emulation;
     bool use_asynchronous_gpu_emulation;
+    bool force_30fps_mode;
 
     float bg_red;
     float bg_green;

src/tests/core/arm/arm_test_common.cpp

@@ -17,7 +17,6 @@ TestEnvironment::TestEnvironment(bool mutable_memory_)
     : mutable_memory(mutable_memory_),
       test_memory(std::make_shared<TestMemory>(this)), kernel{Core::System::GetInstance()} {
     auto process = Kernel::Process::Create(Core::System::GetInstance(), "");
-    kernel.MakeCurrentProcess(process.get());
     page_table = &process->VMManager().page_table;
 
     std::fill(page_table->pointers.begin(), page_table->pointers.end(), nullptr);
@@ -28,7 +27,7 @@ TestEnvironment::TestEnvironment(bool mutable_memory_)
     Memory::MapIoRegion(*page_table, 0x00000000, 0x80000000, test_memory);
     Memory::MapIoRegion(*page_table, 0x80000000, 0x80000000, test_memory);
 
-    Memory::SetCurrentPageTable(page_table);
+    kernel.MakeCurrentProcess(process.get());
 }
 
 TestEnvironment::~TestEnvironment() {

src/video_core/engines/fermi_2d.cpp

@@ -6,12 +6,13 @@
 #include "common/logging/log.h"
 #include "common/math_util.h"
 #include "video_core/engines/fermi_2d.h"
+#include "video_core/memory_manager.h"
 #include "video_core/rasterizer_interface.h"
 
 namespace Tegra::Engines {
 
 Fermi2D::Fermi2D(VideoCore::RasterizerInterface& rasterizer, MemoryManager& memory_manager)
-    : memory_manager(memory_manager), rasterizer{rasterizer} {}
+    : rasterizer{rasterizer}, memory_manager{memory_manager} {}
 
 void Fermi2D::CallMethod(const GPU::MethodCall& method_call) {
     ASSERT_MSG(method_call.method < Regs::NUM_REGS,

src/video_core/engines/fermi_2d.h

@@ -10,7 +10,10 @@
 #include "common/common_funcs.h"
 #include "common/common_types.h"
 #include "video_core/gpu.h"
-#include "video_core/memory_manager.h"
+
+namespace Tegra {
+class MemoryManager;
+}
 
 namespace VideoCore {
 class RasterizerInterface;
@@ -115,10 +118,9 @@ public:
         };
     } regs{};
 
-    MemoryManager& memory_manager;
-
 private:
     VideoCore::RasterizerInterface& rasterizer;
+    MemoryManager& memory_manager;
 
     /// Performs the copy from the source surface to the destination surface as configured in the
     /// registers.

src/video_core/engines/kepler_compute.h

@@ -9,7 +9,10 @@
 #include "common/common_funcs.h"
 #include "common/common_types.h"
 #include "video_core/gpu.h"
-#include "video_core/memory_manager.h"
+
+namespace Tegra {
+class MemoryManager;
+}
 
 namespace Tegra::Engines {
 
@@ -40,10 +43,11 @@ public:
     static_assert(sizeof(Regs) == Regs::NUM_REGS * sizeof(u32),
                   "KeplerCompute Regs has wrong size");
 
-    MemoryManager& memory_manager;
-
     /// Write the value to the register identified by method.
     void CallMethod(const GPU::MethodCall& method_call);
+
+private:
+    MemoryManager& memory_manager;
 };
 
 #define ASSERT_REG_POSITION(field_name, position)                                                  \

src/video_core/engines/kepler_memory.cpp

@@ -5,9 +5,9 @@
 #include "common/assert.h"
 #include "common/logging/log.h"
 #include "core/core.h"
-#include "core/memory.h"
 #include "video_core/engines/kepler_memory.h"
 #include "video_core/engines/maxwell_3d.h"
+#include "video_core/memory_manager.h"
 #include "video_core/rasterizer_interface.h"
 #include "video_core/renderer_base.h"
 
@@ -15,7 +15,7 @@ namespace Tegra::Engines {
 
 KeplerMemory::KeplerMemory(Core::System& system, VideoCore::RasterizerInterface& rasterizer,
                            MemoryManager& memory_manager)
-    : system{system}, memory_manager(memory_manager), rasterizer{rasterizer} {}
+    : system{system}, rasterizer{rasterizer}, memory_manager{memory_manager} {}
 
 KeplerMemory::~KeplerMemory() = default;
 

src/video_core/engines/kepler_memory.h

@@ -10,12 +10,15 @@
 #include "common/common_funcs.h"
 #include "common/common_types.h"
 #include "video_core/gpu.h"
-#include "video_core/memory_manager.h"
 
 namespace Core {
 class System;
 }
 
+namespace Tegra {
+class MemoryManager;
+}
+
 namespace VideoCore {
 class RasterizerInterface;
 }
@@ -82,8 +85,8 @@ public:
 
 private:
     Core::System& system;
-    MemoryManager& memory_manager;
     VideoCore::RasterizerInterface& rasterizer;
+    MemoryManager& memory_manager;
 
     void ProcessData(u32 data);
 };

src/video_core/engines/maxwell_3d.cpp

@@ -7,11 +7,10 @@
 #include "common/assert.h"
 #include "core/core.h"
 #include "core/core_timing.h"
-#include "core/memory.h"
 #include "video_core/debug_utils/debug_utils.h"
 #include "video_core/engines/maxwell_3d.h"
+#include "video_core/memory_manager.h"
 #include "video_core/rasterizer_interface.h"
-#include "video_core/renderer_base.h"
 #include "video_core/textures/texture.h"
 
 namespace Tegra::Engines {
@@ -21,8 +20,8 @@ constexpr u32 MacroRegistersStart = 0xE00;
 
 Maxwell3D::Maxwell3D(Core::System& system, VideoCore::RasterizerInterface& rasterizer,
                      MemoryManager& memory_manager)
-    : memory_manager(memory_manager), system{system}, rasterizer{rasterizer},
-      macro_interpreter(*this) {
+    : system{system}, rasterizer{rasterizer}, memory_manager{memory_manager}, macro_interpreter{
+                                                                                  *this} {
     InitializeRegisterDefaults();
 }
 
@@ -250,6 +249,10 @@ void Maxwell3D::CallMethod(const GPU::MethodCall& method_call) {
         ProcessQueryGet();
         break;
     }
+    case MAXWELL3D_REG_INDEX(sync_info): {
+        ProcessSyncPoint();
+        break;
+    }
     default:
         break;
     }
@@ -327,6 +330,14 @@ void Maxwell3D::ProcessQueryGet() {
     }
 }
 
+void Maxwell3D::ProcessSyncPoint() {
+    const u32 sync_point = regs.sync_info.sync_point.Value();
+    const u32 increment = regs.sync_info.increment.Value();
+    const u32 cache_flush = regs.sync_info.unknown.Value();
+    LOG_DEBUG(HW_GPU, "Syncpoint set {}, increment: {}, unk: {}", sync_point, increment,
+              cache_flush);
+}
+
 void Maxwell3D::DrawArrays() {
     LOG_DEBUG(HW_GPU, "called, topology={}, count={}", static_cast<u32>(regs.draw.topology.Value()),
               regs.vertex_buffer.count);

src/video_core/engines/maxwell_3d.h

@@ -16,13 +16,16 @@
 #include "common/math_util.h"
 #include "video_core/gpu.h"
 #include "video_core/macro_interpreter.h"
-#include "video_core/memory_manager.h"
 #include "video_core/textures/texture.h"
 
 namespace Core {
 class System;
 }
 
+namespace Tegra {
+class MemoryManager;
+}
+
 namespace VideoCore {
 class RasterizerInterface;
 }
@@ -576,7 +579,17 @@ public:
                     u32 bind;
                 } macros;
 
-                INSERT_PADDING_WORDS(0x188);
+                INSERT_PADDING_WORDS(0x69);
+
+                struct {
+                    union {
+                        BitField<0, 16, u32> sync_point;
+                        BitField<16, 1, u32> unknown;
+                        BitField<20, 1, u32> increment;
+                    };
+                } sync_info;
+
+                INSERT_PADDING_WORDS(0x11E);
 
                 u32 tfb_enabled;
 
@@ -1093,7 +1106,6 @@ public:
     };
 
     State state{};
-    MemoryManager& memory_manager;
 
     struct DirtyFlags {
         std::bitset<8> color_buffer{0xFF};
@@ -1141,6 +1153,8 @@ private:
 
     VideoCore::RasterizerInterface& rasterizer;
 
+    MemoryManager& memory_manager;
+
     /// Start offsets of each macro in macro_memory
     std::unordered_map<u32, u32> macro_offsets;
 
@@ -1180,6 +1194,9 @@ private:
     /// Handles a write to the QUERY_GET register.
     void ProcessQueryGet();
 
+    /// Handles writes to syncing register.
+    void ProcessSyncPoint();
+
     /// Handles a write to the CB_DATA[i] register.
     void ProcessCBData(u32 value);
 
@@ -1195,6 +1212,7 @@ private:
                   "Field " #field_name " has invalid position")
 
 ASSERT_REG_POSITION(macros, 0x45);
+ASSERT_REG_POSITION(sync_info, 0xB2);
 ASSERT_REG_POSITION(tfb_enabled, 0x1D1);
 ASSERT_REG_POSITION(rt, 0x200);
 ASSERT_REG_POSITION(viewport_transform, 0x280);

src/video_core/engines/maxwell_dma.cpp

@@ -5,9 +5,9 @@
 #include "common/assert.h"
 #include "common/logging/log.h"
 #include "core/core.h"
-#include "core/memory.h"
 #include "video_core/engines/maxwell_3d.h"
 #include "video_core/engines/maxwell_dma.h"
+#include "video_core/memory_manager.h"
 #include "video_core/rasterizer_interface.h"
 #include "video_core/renderer_base.h"
 #include "video_core/textures/decoders.h"
@@ -16,7 +16,7 @@ namespace Tegra::Engines {
 
 MaxwellDMA::MaxwellDMA(Core::System& system, VideoCore::RasterizerInterface& rasterizer,
                        MemoryManager& memory_manager)
-    : memory_manager(memory_manager), system{system}, rasterizer{rasterizer} {}
+    : system{system}, rasterizer{rasterizer}, memory_manager{memory_manager} {}
 
 void MaxwellDMA::CallMethod(const GPU::MethodCall& method_call) {
     ASSERT_MSG(method_call.method < Regs::NUM_REGS,

src/video_core/engines/maxwell_dma.h

@@ -10,12 +10,15 @@
 #include "common/common_funcs.h"
 #include "common/common_types.h"
 #include "video_core/gpu.h"
-#include "video_core/memory_manager.h"
 
 namespace Core {
 class System;
 }
 
+namespace Tegra {
+class MemoryManager;
+}
+
 namespace VideoCore {
 class RasterizerInterface;
 }
@@ -139,13 +142,13 @@ public:
         };
     } regs{};
 
-    MemoryManager& memory_manager;
-
 private:
     Core::System& system;
 
     VideoCore::RasterizerInterface& rasterizer;
 
+    MemoryManager& memory_manager;
+
     /// Performs the copy from the source buffer to the destination buffer as configured in the
     /// registers.
     void HandleCopy();

src/video_core/gpu.cpp

@@ -31,7 +31,7 @@ u32 FramebufferConfig::BytesPerPixel(PixelFormat format) {
 
 GPU::GPU(Core::System& system, VideoCore::RendererBase& renderer) : renderer{renderer} {
     auto& rasterizer{renderer.Rasterizer()};
-    memory_manager = std::make_unique<Tegra::MemoryManager>();
+    memory_manager = std::make_unique<Tegra::MemoryManager>(rasterizer);
     dma_pusher = std::make_unique<Tegra::DmaPusher>(*this);
     maxwell_3d = std::make_unique<Engines::Maxwell3D>(system, rasterizer, *memory_manager);
     fermi_2d = std::make_unique<Engines::Fermi2D>(rasterizer, *memory_manager);

src/video_core/gpu_asynch.cpp

@@ -9,7 +9,7 @@
 namespace VideoCommon {
 
 GPUAsynch::GPUAsynch(Core::System& system, VideoCore::RendererBase& renderer)
-    : Tegra::GPU(system, renderer), gpu_thread{renderer, *dma_pusher} {}
+    : Tegra::GPU(system, renderer), gpu_thread{system, renderer, *dma_pusher} {}
 
 GPUAsynch::~GPUAsynch() = default;
 

src/video_core/gpu_thread.cpp

@@ -4,6 +4,9 @@
 
 #include "common/assert.h"
 #include "common/microprofile.h"
+#include "core/core.h"
+#include "core/core_timing.h"
+#include "core/core_timing_util.h"
 #include "core/frontend/scope_acquire_window_context.h"
 #include "video_core/dma_pusher.h"
 #include "video_core/gpu.h"
@@ -36,7 +39,6 @@ static void RunThread(VideoCore::RendererBase& renderer, Tegra::DmaPusher& dma_p
                 dma_pusher.Push(std::move(submit_list->entries));
                 dma_pusher.DispatchCalls();
             } else if (const auto data = std::get_if<SwapBuffersCommand>(&next.data)) {
-                state.DecrementFramesCounter();
                 renderer.SwapBuffers(std::move(data->framebuffer));
             } else if (const auto data = std::get_if<FlushRegionCommand>(&next.data)) {
                 renderer.Rasterizer().FlushRegion(data->addr, data->size);
@@ -47,13 +49,18 @@ static void RunThread(VideoCore::RendererBase& renderer, Tegra::DmaPusher& dma_p
             } else {
                 UNREACHABLE();
             }
+            state.signaled_fence = next.fence;
+            state.TrySynchronize();
         }
     }
 }
 
-ThreadManager::ThreadManager(VideoCore::RendererBase& renderer, Tegra::DmaPusher& dma_pusher)
-    : renderer{renderer}, thread{RunThread, std::ref(renderer), std::ref(dma_pusher),
-                                 std::ref(state)} {}
+ThreadManager::ThreadManager(Core::System& system, VideoCore::RendererBase& renderer,
+                             Tegra::DmaPusher& dma_pusher)
+    : system{system}, thread{RunThread, std::ref(renderer), std::ref(dma_pusher), std::ref(state)} {
+    synchronization_event = system.CoreTiming().RegisterEvent(
+        "GPUThreadSynch", [this](u64 fence, s64) { state.WaitForSynchronization(fence); });
+}
 
 ThreadManager::~ThreadManager() {
     // Notify GPU thread that a shutdown is pending
@@ -62,14 +69,14 @@ ThreadManager::~ThreadManager() {
 }
 
 void ThreadManager::SubmitList(Tegra::CommandList&& entries) {
-    PushCommand(SubmitListCommand(std::move(entries)));
+    const u64 fence{PushCommand(SubmitListCommand(std::move(entries)))};
+    const s64 synchronization_ticks{Core::Timing::usToCycles(9000)};
+    system.CoreTiming().ScheduleEvent(synchronization_ticks, synchronization_event, fence);
 }
 
 void ThreadManager::SwapBuffers(
     std::optional<std::reference_wrapper<const Tegra::FramebufferConfig>> framebuffer) {
-    state.IncrementFramesCounter();
     PushCommand(SwapBuffersCommand(std::move(framebuffer)));
-    state.WaitForFrames();
 }
 
 void ThreadManager::FlushRegion(CacheAddr addr, u64 size) {
@@ -79,7 +86,7 @@ void ThreadManager::FlushRegion(CacheAddr addr, u64 size) {
 void ThreadManager::InvalidateRegion(CacheAddr addr, u64 size) {
     if (state.queue.Empty()) {
         // It's quicker to invalidate a single region on the CPU if the queue is already empty
-        renderer.Rasterizer().InvalidateRegion(addr, size);
+        system.Renderer().Rasterizer().InvalidateRegion(addr, size);
     } else {
         PushCommand(InvalidateRegionCommand(addr, size));
     }
@@ -90,9 +97,25 @@ void ThreadManager::FlushAndInvalidateRegion(CacheAddr addr, u64 size) {
     InvalidateRegion(addr, size);
 }
 
-void ThreadManager::PushCommand(CommandData&& command_data) {
-    state.queue.Push(CommandDataContainer(std::move(command_data)));
+u64 ThreadManager::PushCommand(CommandData&& command_data) {
+    const u64 fence{++state.last_fence};
+    state.queue.Push(CommandDataContainer(std::move(command_data), fence));
     state.SignalCommands();
+    return fence;
+}
+
+MICROPROFILE_DEFINE(GPU_wait, "GPU", "Wait for the GPU", MP_RGB(128, 128, 192));
+void SynchState::WaitForSynchronization(u64 fence) {
+    if (signaled_fence >= fence) {
+        return;
+    }
+
+    // Wait for the GPU to be idle (all commands to be executed)
+    {
+        MICROPROFILE_SCOPE(GPU_wait);
+        std::unique_lock<std::mutex> lock{synchronization_mutex};
+        synchronization_condition.wait(lock, [this, fence] { return signaled_fence >= fence; });
+    }
 }
 
 } // namespace VideoCommon::GPUThread

src/video_core/gpu_thread.h

@@ -19,9 +19,12 @@ struct FramebufferConfig;
 class DmaPusher;
 } // namespace Tegra
 
-namespace VideoCore {
-class RendererBase;
-} // namespace VideoCore
+namespace Core {
+class System;
+namespace Timing {
+struct EventType;
+} // namespace Timing
+} // namespace Core
 
 namespace VideoCommon::GPUThread {
 
@@ -75,63 +78,47 @@ using CommandData =
 struct CommandDataContainer {
     CommandDataContainer() = default;
 
-    CommandDataContainer(CommandData&& data) : data{std::move(data)} {}
+    CommandDataContainer(CommandData&& data, u64 next_fence)
+        : data{std::move(data)}, fence{next_fence} {}
 
     CommandDataContainer& operator=(const CommandDataContainer& t) {
         data = std::move(t.data);
+        fence = t.fence;
         return *this;
     }
 
     CommandData data;
+    u64 fence{};
 };
 
 /// Struct used to synchronize the GPU thread
 struct SynchState final {
     std::atomic_bool is_running{true};
     std::atomic_int queued_frame_count{};
-    std::mutex frames_mutex;
+    std::mutex synchronization_mutex;
     std::mutex commands_mutex;
     std::condition_variable commands_condition;
-    std::condition_variable frames_condition;
+    std::condition_variable synchronization_condition;
 
-    void IncrementFramesCounter() {
-        std::lock_guard lock{frames_mutex};
-        ++queued_frame_count;
+    /// Returns true if the gap in GPU commands is small enough that we can consider the CPU and GPU
+    /// synchronized. This is entirely empirical.
+    bool IsSynchronized() const {
+        constexpr std::size_t max_queue_gap{5};
+        return queue.Size() <= max_queue_gap;
     }
 
-    void DecrementFramesCounter() {
-        {
-            std::lock_guard lock{frames_mutex};
-            --queued_frame_count;
-
-            if (queued_frame_count) {
-                return;
-            }
-        }
-        frames_condition.notify_one();
-    }
-
-    void WaitForFrames() {
-        {
-            std::lock_guard lock{frames_mutex};
-            if (!queued_frame_count) {
-                return;
-            }
-        }
-
-        // Wait for the GPU to be idle (all commands to be executed)
-        {
-            std::unique_lock lock{frames_mutex};
-            frames_condition.wait(lock, [this] { return !queued_frame_count; });
+    void TrySynchronize() {
+        if (IsSynchronized()) {
+            std::lock_guard<std::mutex> lock{synchronization_mutex};
+            synchronization_condition.notify_one();
         }
     }
 
+    void WaitForSynchronization(u64 fence);
+
     void SignalCommands() {
-        {
-            std::unique_lock lock{commands_mutex};
-            if (queue.Empty()) {
-                return;
-            }
+        if (queue.Empty()) {
+            return;
         }
 
         commands_condition.notify_one();
@@ -144,12 +131,15 @@ struct SynchState final {
 
     using CommandQueue = Common::SPSCQueue<CommandDataContainer>;
     CommandQueue queue;
+    u64 last_fence{};
+    std::atomic<u64> signaled_fence{};
 };
 
 /// Class used to manage the GPU thread
 class ThreadManager final {
 public:
-    explicit ThreadManager(VideoCore::RendererBase& renderer, Tegra::DmaPusher& dma_pusher);
+    explicit ThreadManager(Core::System& system, VideoCore::RendererBase& renderer,
+                           Tegra::DmaPusher& dma_pusher);
     ~ThreadManager();
 
     /// Push GPU command entries to be processed
@@ -170,11 +160,12 @@ public:
 
 private:
     /// Pushes a command to be executed by the GPU thread
-    void PushCommand(CommandData&& command_data);
+    u64 PushCommand(CommandData&& command_data);
 
 private:
     SynchState state;
-    VideoCore::RendererBase& renderer;
+    Core::System& system;
+    Core::Timing::EventType* synchronization_event{};
     std::thread thread;
     std::thread::id thread_id;
 };

src/video_core/macro_interpreter.cpp

@@ -223,27 +223,21 @@ void MacroInterpreter::ProcessResult(ResultOperation operation, u32 reg, u32 res
 }
 
 u32 MacroInterpreter::FetchParameter() {
-    ASSERT(next_parameter_index < parameters.size());
-    return parameters[next_parameter_index++];
+    return parameters.at(next_parameter_index++);
 }
 
 u32 MacroInterpreter::GetRegister(u32 register_id) const {
-    // Register 0 is supposed to always return 0.
-    if (register_id == 0)
-        return 0;
-
-    ASSERT(register_id < registers.size());
-    return registers[register_id];
+    return registers.at(register_id);
 }
 
 void MacroInterpreter::SetRegister(u32 register_id, u32 value) {
-    // Register 0 is supposed to always return 0. NOP is implemented as a store to the zero
-    // register.
-    if (register_id == 0)
+    // Register 0 is hardwired as the zero register.
+    // Ensure no writes to it actually occur.
+    if (register_id == 0) {
         return;
+    }
 
-    ASSERT(register_id < registers.size());
-    registers[register_id] = value;
+    registers.at(register_id) = value;
 }
 
 void MacroInterpreter::SetMethodAddress(u32 address) {

src/video_core/memory_manager.cpp

@@ -5,16 +5,13 @@
 #include "common/alignment.h"
 #include "common/assert.h"
 #include "common/logging/log.h"
-#include "core/core.h"
 #include "core/memory.h"
-#include "video_core/gpu.h"
 #include "video_core/memory_manager.h"
 #include "video_core/rasterizer_interface.h"
-#include "video_core/renderer_base.h"
 
 namespace Tegra {
 
-MemoryManager::MemoryManager() {
+MemoryManager::MemoryManager(VideoCore::RasterizerInterface& rasterizer) : rasterizer{rasterizer} {
     std::fill(page_table.pointers.begin(), page_table.pointers.end(), nullptr);
     std::fill(page_table.attributes.begin(), page_table.attributes.end(),
               Common::PageType::Unmapped);
@@ -70,23 +67,23 @@ GPUVAddr MemoryManager::UnmapBuffer(GPUVAddr gpu_addr, u64 size) {
     const u64 aligned_size{Common::AlignUp(size, page_size)};
     const CacheAddr cache_addr{ToCacheAddr(GetPointer(gpu_addr))};
 
-    Core::System::GetInstance().Renderer().Rasterizer().FlushAndInvalidateRegion(cache_addr,
-                                                                                 aligned_size);
+    rasterizer.FlushAndInvalidateRegion(cache_addr, aligned_size);
     UnmapRange(gpu_addr, aligned_size);
 
     return gpu_addr;
 }
 
-GPUVAddr MemoryManager::FindFreeRegion(GPUVAddr region_start, u64 size) {
+GPUVAddr MemoryManager::FindFreeRegion(GPUVAddr region_start, u64 size) const {
     // Find the first Free VMA.
-    const VMAHandle vma_handle{std::find_if(vma_map.begin(), vma_map.end(), [&](const auto& vma) {
-        if (vma.second.type != VirtualMemoryArea::Type::Unmapped) {
-            return false;
-        }
+    const VMAHandle vma_handle{
+        std::find_if(vma_map.begin(), vma_map.end(), [region_start, size](const auto& vma) {
+            if (vma.second.type != VirtualMemoryArea::Type::Unmapped) {
+                return false;
+            }
 
-        const VAddr vma_end{vma.second.base + vma.second.size};
-        return vma_end > region_start && vma_end >= region_start + size;
-    })};
+            const VAddr vma_end{vma.second.base + vma.second.size};
+            return vma_end > region_start && vma_end >= region_start + size;
+        })};
 
     if (vma_handle == vma_map.end()) {
         return {};
@@ -99,12 +96,12 @@ bool MemoryManager::IsAddressValid(GPUVAddr addr) const {
     return (addr >> page_bits) < page_table.pointers.size();
 }
 
-std::optional<VAddr> MemoryManager::GpuToCpuAddress(GPUVAddr addr) {
+std::optional<VAddr> MemoryManager::GpuToCpuAddress(GPUVAddr addr) const {
     if (!IsAddressValid(addr)) {
         return {};
     }
 
-    VAddr cpu_addr{page_table.backing_addr[addr >> page_bits]};
+    const VAddr cpu_addr{page_table.backing_addr[addr >> page_bits]};
     if (cpu_addr) {
         return cpu_addr + (addr & page_mask);
     }
@@ -113,7 +110,7 @@ std::optional<VAddr> MemoryManager::GpuToCpuAddress(GPUVAddr addr) {
 }
 
 template <typename T>
-T MemoryManager::Read(GPUVAddr addr) {
+T MemoryManager::Read(GPUVAddr addr) const {
     if (!IsAddressValid(addr)) {
         return {};
     }
@@ -165,10 +162,10 @@ void MemoryManager::Write(GPUVAddr addr, T data) {
     }
 }
 
-template u8 MemoryManager::Read<u8>(GPUVAddr addr);
-template u16 MemoryManager::Read<u16>(GPUVAddr addr);
-template u32 MemoryManager::Read<u32>(GPUVAddr addr);
-template u64 MemoryManager::Read<u64>(GPUVAddr addr);
+template u8 MemoryManager::Read<u8>(GPUVAddr addr) const;
+template u16 MemoryManager::Read<u16>(GPUVAddr addr) const;
+template u32 MemoryManager::Read<u32>(GPUVAddr addr) const;
+template u64 MemoryManager::Read<u64>(GPUVAddr addr) const;
 template void MemoryManager::Write<u8>(GPUVAddr addr, u8 data);
 template void MemoryManager::Write<u16>(GPUVAddr addr, u16 data);
 template void MemoryManager::Write<u32>(GPUVAddr addr, u32 data);
@@ -179,8 +176,8 @@ u8* MemoryManager::GetPointer(GPUVAddr addr) {
         return {};
     }
 
-    u8* page_pointer{page_table.pointers[addr >> page_bits]};
-    if (page_pointer) {
+    u8* const page_pointer{page_table.pointers[addr >> page_bits]};
+    if (page_pointer != nullptr) {
         return page_pointer + (addr & page_mask);
     }
 
@@ -188,15 +185,100 @@ u8* MemoryManager::GetPointer(GPUVAddr addr) {
     return {};
 }
 
-void MemoryManager::ReadBlock(GPUVAddr src_addr, void* dest_buffer, std::size_t size) {
-    std::memcpy(dest_buffer, GetPointer(src_addr), size);
+const u8* MemoryManager::GetPointer(GPUVAddr addr) const {
+    if (!IsAddressValid(addr)) {
+        return {};
+    }
+
+    const u8* const page_pointer{page_table.pointers[addr >> page_bits]};
+    if (page_pointer != nullptr) {
+        return page_pointer + (addr & page_mask);
+    }
+
+    LOG_ERROR(HW_GPU, "Unknown GetPointer @ 0x{:016X}", addr);
+    return {};
 }
+
+void MemoryManager::ReadBlock(GPUVAddr src_addr, void* dest_buffer, std::size_t size) const {
+    std::size_t remaining_size{size};
+    std::size_t page_index{src_addr >> page_bits};
+    std::size_t page_offset{src_addr & page_mask};
+
+    while (remaining_size > 0) {
+        const std::size_t copy_amount{
+            std::min(static_cast<std::size_t>(page_size) - page_offset, remaining_size)};
+
+        switch (page_table.attributes[page_index]) {
+        case Common::PageType::Memory: {
+            const u8* src_ptr{page_table.pointers[page_index] + page_offset};
+            rasterizer.FlushRegion(ToCacheAddr(src_ptr), copy_amount);
+            std::memcpy(dest_buffer, src_ptr, copy_amount);
+            break;
+        }
+        default:
+            UNREACHABLE();
+        }
+
+        page_index++;
+        page_offset = 0;
+        dest_buffer = static_cast<u8*>(dest_buffer) + copy_amount;
+        remaining_size -= copy_amount;
+    }
+}
+
 void MemoryManager::WriteBlock(GPUVAddr dest_addr, const void* src_buffer, std::size_t size) {
-    std::memcpy(GetPointer(dest_addr), src_buffer, size);
+    std::size_t remaining_size{size};
+    std::size_t page_index{dest_addr >> page_bits};
+    std::size_t page_offset{dest_addr & page_mask};
+
+    while (remaining_size > 0) {
+        const std::size_t copy_amount{
+            std::min(static_cast<std::size_t>(page_size) - page_offset, remaining_size)};
+
+        switch (page_table.attributes[page_index]) {
+        case Common::PageType::Memory: {
+            u8* dest_ptr{page_table.pointers[page_index] + page_offset};
+            rasterizer.InvalidateRegion(ToCacheAddr(dest_ptr), copy_amount);
+            std::memcpy(dest_ptr, src_buffer, copy_amount);
+            break;
+        }
+        default:
+            UNREACHABLE();
+        }
+
+        page_index++;
+        page_offset = 0;
+        src_buffer = static_cast<const u8*>(src_buffer) + copy_amount;
+        remaining_size -= copy_amount;
+    }
 }
 
 void MemoryManager::CopyBlock(GPUVAddr dest_addr, GPUVAddr src_addr, std::size_t size) {
-    std::memcpy(GetPointer(dest_addr), GetPointer(src_addr), size);
+    std::size_t remaining_size{size};
+    std::size_t page_index{src_addr >> page_bits};
+    std::size_t page_offset{src_addr & page_mask};
+
+    while (remaining_size > 0) {
+        const std::size_t copy_amount{
+            std::min(static_cast<std::size_t>(page_size) - page_offset, remaining_size)};
+
+        switch (page_table.attributes[page_index]) {
+        case Common::PageType::Memory: {
+            const u8* src_ptr{page_table.pointers[page_index] + page_offset};
+            rasterizer.FlushRegion(ToCacheAddr(src_ptr), copy_amount);
+            WriteBlock(dest_addr, src_ptr, copy_amount);
+            break;
+        }
+        default:
+            UNREACHABLE();
+        }
+
+        page_index++;
+        page_offset = 0;
+        dest_addr += static_cast<VAddr>(copy_amount);
+        src_addr += static_cast<VAddr>(copy_amount);
+        remaining_size -= copy_amount;
+    }
 }
 
 void MemoryManager::MapPages(GPUVAddr base, u64 size, u8* memory, Common::PageType type,
@@ -336,7 +418,7 @@ MemoryManager::VMAIter MemoryManager::CarveVMA(GPUVAddr base, u64 size) {
     const VirtualMemoryArea& vma{vma_handle->second};
     if (vma.type == VirtualMemoryArea::Type::Mapped) {
         // Region is already allocated
-        return {};
+        return vma_handle;
     }
 
     const VAddr start_in_vma{base - vma.base};

src/video_core/memory_manager.h

@@ -10,6 +10,10 @@
 #include "common/common_types.h"
 #include "common/page_table.h"
 
+namespace VideoCore {
+class RasterizerInterface;
+}
+
 namespace Tegra {
 
 /**
@@ -43,24 +47,25 @@ struct VirtualMemoryArea {
 
 class MemoryManager final {
 public:
-    MemoryManager();
+    MemoryManager(VideoCore::RasterizerInterface& rasterizer);
 
     GPUVAddr AllocateSpace(u64 size, u64 align);
     GPUVAddr AllocateSpace(GPUVAddr addr, u64 size, u64 align);
     GPUVAddr MapBufferEx(VAddr cpu_addr, u64 size);
     GPUVAddr MapBufferEx(VAddr cpu_addr, GPUVAddr addr, u64 size);
     GPUVAddr UnmapBuffer(GPUVAddr addr, u64 size);
-    std::optional<VAddr> GpuToCpuAddress(GPUVAddr addr);
+    std::optional<VAddr> GpuToCpuAddress(GPUVAddr addr) const;
 
     template <typename T>
-    T Read(GPUVAddr addr);
+    T Read(GPUVAddr addr) const;
 
     template <typename T>
     void Write(GPUVAddr addr, T data);
 
     u8* GetPointer(GPUVAddr addr);
+    const u8* GetPointer(GPUVAddr addr) const;
 
-    void ReadBlock(GPUVAddr src_addr, void* dest_buffer, std::size_t size);
+    void ReadBlock(GPUVAddr src_addr, void* dest_buffer, std::size_t size) const;
     void WriteBlock(GPUVAddr dest_addr, const void* src_buffer, std::size_t size);
     void CopyBlock(GPUVAddr dest_addr, GPUVAddr src_addr, std::size_t size);
 
@@ -127,7 +132,7 @@ private:
     void UpdatePageTableForVMA(const VirtualMemoryArea& vma);
 
     /// Finds a free (unmapped region) of the specified size starting at the specified address.
-    GPUVAddr FindFreeRegion(GPUVAddr region_start, u64 size);
+    GPUVAddr FindFreeRegion(GPUVAddr region_start, u64 size) const;
 
 private:
     static constexpr u64 page_bits{16};
@@ -143,6 +148,7 @@ private:
 
     Common::PageTable page_table{page_bits};
     VMAMap vma_map;
+    VideoCore::RasterizerInterface& rasterizer;
 };
 
 } // namespace Tegra

src/video_core/renderer_opengl/gl_buffer_cache.cpp

@@ -7,7 +7,6 @@
 
 #include "common/alignment.h"
 #include "core/core.h"
-#include "core/memory.h"
 #include "video_core/renderer_opengl/gl_buffer_cache.h"
 #include "video_core/renderer_opengl/gl_rasterizer.h"
 

src/video_core/renderer_opengl/gl_global_cache.cpp

@@ -4,7 +4,6 @@
 
 #include <glad/glad.h>
 
-#include "common/assert.h"
 #include "common/logging/log.h"
 #include "core/core.h"
 #include "video_core/renderer_opengl/gl_global_cache.h"

src/video_core/renderer_opengl/gl_primitive_assembler.cpp

@@ -7,7 +7,7 @@
 #include "common/assert.h"
 #include "common/common_types.h"
 #include "core/core.h"
-#include "core/memory.h"
+#include "video_core/memory_manager.h"
 #include "video_core/renderer_opengl/gl_buffer_cache.h"
 #include "video_core/renderer_opengl/gl_primitive_assembler.h"
 

src/video_core/renderer_opengl/gl_primitive_assembler.h

@@ -4,11 +4,9 @@
 
 #pragma once
 
-#include <vector>
 #include <glad/glad.h>
 
 #include "common/common_types.h"
-#include "video_core/memory_manager.h"
 
 namespace OpenGL {
 

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -17,7 +17,6 @@
 #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"
@@ -26,7 +25,6 @@
 #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"
 
 namespace OpenGL {
 

src/video_core/renderer_opengl/gl_rasterizer.h

@@ -12,15 +12,12 @@
 #include <optional>
 #include <tuple>
 #include <utility>
-#include <vector>
 
 #include <boost/icl/interval_map.hpp>
-#include <boost/range/iterator_range.hpp>
 #include <glad/glad.h>
 
 #include "common/common_types.h"
 #include "video_core/engines/maxwell_3d.h"
-#include "video_core/memory_manager.h"
 #include "video_core/rasterizer_cache.h"
 #include "video_core/rasterizer_interface.h"
 #include "video_core/renderer_opengl/gl_buffer_cache.h"
@@ -29,10 +26,8 @@
 #include "video_core/renderer_opengl/gl_rasterizer_cache.h"
 #include "video_core/renderer_opengl/gl_resource_manager.h"
 #include "video_core/renderer_opengl/gl_shader_cache.h"
-#include "video_core/renderer_opengl/gl_shader_gen.h"
 #include "video_core/renderer_opengl/gl_shader_manager.h"
 #include "video_core/renderer_opengl/gl_state.h"
-#include "video_core/renderer_opengl/gl_stream_buffer.h"
 
 namespace Core {
 class System;

src/video_core/renderer_opengl/gl_rasterizer_cache.cpp

@@ -13,7 +13,6 @@
 #include "common/scope_exit.h"
 #include "core/core.h"
 #include "core/hle/kernel/process.h"
-#include "core/memory.h"
 #include "core/settings.h"
 #include "video_core/engines/maxwell_3d.h"
 #include "video_core/morton.h"

src/video_core/renderer_opengl/gl_rasterizer_cache.h

@@ -5,10 +5,9 @@
 #pragma once
 
 #include <array>
-#include <map>
 #include <memory>
 #include <string>
-#include <unordered_set>
+#include <tuple>
 #include <vector>
 
 #include "common/alignment.h"

src/video_core/renderer_opengl/gl_shader_cache.cpp

@@ -39,6 +39,10 @@ GPUVAddr GetShaderAddress(Maxwell::ShaderProgram program) {
 /// Gets the shader program code from memory for the specified address
 ProgramCode GetShaderCode(const u8* host_ptr) {
     ProgramCode program_code(VideoCommon::Shader::MAX_PROGRAM_LENGTH);
+    ASSERT_OR_EXECUTE(host_ptr != nullptr, {
+        std::fill(program_code.begin(), program_code.end(), 0);
+        return program_code;
+    });
     std::memcpy(program_code.data(), host_ptr, program_code.size() * sizeof(u64));
     return program_code;
 }

src/video_core/renderer_opengl/gl_shader_cache.h

@@ -5,21 +5,20 @@
 #pragma once
 
 #include <array>
+#include <atomic>
 #include <memory>
 #include <set>
 #include <tuple>
 #include <unordered_map>
+#include <vector>
 
 #include <glad/glad.h>
 
-#include "common/assert.h"
 #include "common/common_types.h"
 #include "video_core/rasterizer_cache.h"
-#include "video_core/renderer_base.h"
 #include "video_core/renderer_opengl/gl_resource_manager.h"
 #include "video_core/renderer_opengl/gl_shader_decompiler.h"
 #include "video_core/renderer_opengl/gl_shader_disk_cache.h"
-#include "video_core/renderer_opengl/gl_shader_gen.h"
 
 namespace Core {
 class System;

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -21,6 +21,8 @@
 
 namespace OpenGL::GLShader {
 
+namespace {
+
 using Tegra::Shader::Attribute;
 using Tegra::Shader::AttributeUse;
 using Tegra::Shader::Header;
@@ -34,14 +36,18 @@ using Maxwell = Tegra::Engines::Maxwell3D::Regs;
 using ShaderStage = Tegra::Engines::Maxwell3D::Regs::ShaderStage;
 using Operation = const OperationNode&;
 
+enum class Type { Bool, Bool2, Float, Int, Uint, HalfFloat };
+
+struct TextureAoffi {};
+using TextureArgument = std::pair<Type, Node>;
+using TextureIR = std::variant<TextureAoffi, TextureArgument>;
+
 enum : u32 { POSITION_VARYING_LOCATION = 0, GENERIC_VARYING_START_LOCATION = 1 };
 constexpr u32 MAX_CONSTBUFFER_ELEMENTS =
     static_cast<u32>(RasterizerOpenGL::MaxConstbufferSize) / (4 * sizeof(float));
 constexpr u32 MAX_GLOBALMEMORY_ELEMENTS =
     static_cast<u32>(RasterizerOpenGL::MaxGlobalMemorySize) / sizeof(float);
 
-enum class Type { Bool, Bool2, Float, Int, Uint, HalfFloat };
-
 class ShaderWriter {
 public:
     void AddExpression(std::string_view text) {
@@ -69,10 +75,10 @@ public:
         shader_source += '\n';
     }
 
-    std::string GenerateTemporal() {
-        std::string temporal = "tmp";
-        temporal += std::to_string(temporal_index++);
-        return temporal;
+    std::string GenerateTemporary() {
+        std::string temporary = "tmp";
+        temporary += std::to_string(temporary_index++);
+        return temporary;
     }
 
     std::string GetResult() {
@@ -87,11 +93,11 @@ private:
     }
 
     std::string shader_source;
-    u32 temporal_index = 1;
+    u32 temporary_index = 1;
 };
 
 /// Generates code to use for a swizzle operation.
-static std::string GetSwizzle(u32 elem) {
+std::string GetSwizzle(u32 elem) {
     ASSERT(elem <= 3);
     std::string swizzle = ".";
     swizzle += "xyzw"[elem];
@@ -99,7 +105,7 @@ static std::string GetSwizzle(u32 elem) {
 }
 
 /// Translate topology
-static std::string GetTopologyName(Tegra::Shader::OutputTopology topology) {
+std::string GetTopologyName(Tegra::Shader::OutputTopology topology) {
     switch (topology) {
     case Tegra::Shader::OutputTopology::PointList:
         return "points";
@@ -114,7 +120,7 @@ static std::string GetTopologyName(Tegra::Shader::OutputTopology topology) {
 }
 
 /// Returns true if an object has to be treated as precise
-static bool IsPrecise(Operation operand) {
+bool IsPrecise(Operation operand) {
     const auto& meta = operand.GetMeta();
 
     if (const auto arithmetic = std::get_if<MetaArithmetic>(&meta)) {
@@ -126,7 +132,7 @@ static bool IsPrecise(Operation operand) {
     return false;
 }
 
-static bool IsPrecise(Node node) {
+bool IsPrecise(Node node) {
     if (const auto operation = std::get_if<OperationNode>(node)) {
         return IsPrecise(*operation);
     }
@@ -426,9 +432,14 @@ private:
     std::string Visit(Node node) {
         if (const auto operation = std::get_if<OperationNode>(node)) {
             const auto operation_index = static_cast<std::size_t>(operation->GetCode());
+            if (operation_index >= operation_decompilers.size()) {
+                UNREACHABLE_MSG("Out of bounds operation: {}", operation_index);
+                return {};
+            }
             const auto decompiler = operation_decompilers[operation_index];
             if (decompiler == nullptr) {
-                UNREACHABLE_MSG("Operation decompiler {} not defined", operation_index);
+                UNREACHABLE_MSG("Undefined operation: {}", operation_index);
+                return {};
             }
             return (this->*decompiler)(*operation);
 
@@ -540,7 +551,7 @@ private:
 
             } else if (std::holds_alternative<OperationNode>(*offset)) {
                 // Indirect access
-                const std::string final_offset = code.GenerateTemporal();
+                const std::string final_offset = code.GenerateTemporary();
                 code.AddLine("uint " + final_offset + " = (ftou(" + Visit(offset) + ") / 4) & " +
                              std::to_string(MAX_CONSTBUFFER_ELEMENTS - 1) + ';');
                 return fmt::format("{}[{} / 4][{} % 4]", GetConstBuffer(cbuf->GetIndex()),
@@ -587,9 +598,9 @@ private:
         // There's a bug in NVidia's proprietary drivers that makes precise fail on fragment shaders
         const std::string precise = stage != ShaderStage::Fragment ? "precise " : "";
 
-        const std::string temporal = code.GenerateTemporal();
-        code.AddLine(precise + "float " + temporal + " = " + value + ';');
-        return temporal;
+        const std::string temporary = code.GenerateTemporary();
+        code.AddLine(precise + "float " + temporary + " = " + value + ';');
+        return temporary;
     }
 
     std::string VisitOperand(Operation operation, std::size_t operand_index) {
@@ -601,9 +612,9 @@ private:
             return Visit(operand);
         }
 
-        const std::string temporal = code.GenerateTemporal();
-        code.AddLine("float " + temporal + " = " + Visit(operand) + ';');
-        return temporal;
+        const std::string temporary = code.GenerateTemporary();
+        code.AddLine("float " + temporary + " = " + Visit(operand) + ';');
+        return temporary;
     }
 
     std::string VisitOperand(Operation operation, std::size_t operand_index, Type type) {
@@ -718,8 +729,8 @@ private:
                                                          result_type));
     }
 
-    std::string GenerateTexture(Operation operation, const std::string& func,
-                                const std::vector<std::pair<Type, Node>>& extras) {
+    std::string GenerateTexture(Operation operation, const std::string& function_suffix,
+                                const std::vector<TextureIR>& extras) {
         constexpr std::array<const char*, 4> coord_constructors = {"float", "vec2", "vec3", "vec4"};
 
         const auto meta = std::get_if<MetaTexture>(&operation.GetMeta());
@@ -729,11 +740,11 @@ private:
         const bool has_array = meta->sampler.IsArray();
         const bool has_shadow = meta->sampler.IsShadow();
 
-        std::string expr = func;
-        expr += '(';
-        expr += GetSampler(meta->sampler);
-        expr += ", ";
-
+        std::string expr = "texture" + function_suffix;
+        if (!meta->aoffi.empty()) {
+            expr += "Offset";
+        }
+        expr += '(' + GetSampler(meta->sampler) + ", ";
         expr += coord_constructors.at(count + (has_array ? 1 : 0) + (has_shadow ? 1 : 0) - 1);
         expr += '(';
         for (std::size_t i = 0; i < count; ++i) {
@@ -751,38 +762,76 @@ private:
         }
         expr += ')';
 
-        for (const auto& extra_pair : extras) {
-            const auto [type, operand] = extra_pair;
-            if (operand == nullptr) {
-                continue;
-            }
-            expr += ", ";
-
-            switch (type) {
-            case Type::Int:
-                if (const auto immediate = std::get_if<ImmediateNode>(operand)) {
-                    // Inline the string as an immediate integer in GLSL (some extra arguments are
-                    // required to be constant)
-                    expr += std::to_string(static_cast<s32>(immediate->GetValue()));
-                } else {
-                    expr += "ftoi(" + Visit(operand) + ')';
-                }
-                break;
-            case Type::Float:
-                expr += Visit(operand);
-                break;
-            default: {
-                const auto type_int = static_cast<u32>(type);
-                UNIMPLEMENTED_MSG("Unimplemented extra type={}", type_int);
-                expr += '0';
-                break;
-            }
+        for (const auto& variant : extras) {
+            if (const auto argument = std::get_if<TextureArgument>(&variant)) {
+                expr += GenerateTextureArgument(*argument);
+            } else if (std::get_if<TextureAoffi>(&variant)) {
+                expr += GenerateTextureAoffi(meta->aoffi);
+            } else {
+                UNREACHABLE();
             }
         }
 
         return expr + ')';
     }
 
+    std::string GenerateTextureArgument(TextureArgument argument) {
+        const auto [type, operand] = argument;
+        if (operand == nullptr) {
+            return {};
+        }
+
+        std::string expr = ", ";
+        switch (type) {
+        case Type::Int:
+            if (const auto immediate = std::get_if<ImmediateNode>(operand)) {
+                // Inline the string as an immediate integer in GLSL (some extra arguments are
+                // required to be constant)
+                expr += std::to_string(static_cast<s32>(immediate->GetValue()));
+            } else {
+                expr += "ftoi(" + Visit(operand) + ')';
+            }
+            break;
+        case Type::Float:
+            expr += Visit(operand);
+            break;
+        default: {
+            const auto type_int = static_cast<u32>(type);
+            UNIMPLEMENTED_MSG("Unimplemented extra type={}", type_int);
+            expr += '0';
+            break;
+        }
+        }
+        return expr;
+    }
+
+    std::string GenerateTextureAoffi(const std::vector<Node>& aoffi) {
+        if (aoffi.empty()) {
+            return {};
+        }
+        constexpr std::array<const char*, 3> coord_constructors = {"int", "ivec2", "ivec3"};
+        std::string expr = ", ";
+        expr += coord_constructors.at(aoffi.size() - 1);
+        expr += '(';
+
+        for (std::size_t index = 0; index < aoffi.size(); ++index) {
+            const auto operand{aoffi.at(index)};
+            if (const auto immediate = std::get_if<ImmediateNode>(operand)) {
+                // Inline the string as an immediate integer in GLSL (AOFFI arguments are required
+                // to be constant by the standard).
+                expr += std::to_string(static_cast<s32>(immediate->GetValue()));
+            } else {
+                expr += "ftoi(" + Visit(operand) + ')';
+            }
+            if (index + 1 < aoffi.size()) {
+                expr += ", ";
+            }
+        }
+        expr += ')';
+
+        return expr;
+    }
+
     std::string Assign(Operation operation) {
         const Node dest = operation[0];
         const Node src = operation[1];
@@ -1159,7 +1208,8 @@ private:
         const auto meta = std::get_if<MetaTexture>(&operation.GetMeta());
         ASSERT(meta);
 
-        std::string expr = GenerateTexture(operation, "texture", {{Type::Float, meta->bias}});
+        std::string expr = GenerateTexture(
+            operation, "", {TextureAoffi{}, TextureArgument{Type::Float, meta->bias}});
         if (meta->sampler.IsShadow()) {
             expr = "vec4(" + expr + ')';
         }
@@ -1170,7 +1220,8 @@ private:
         const auto meta = std::get_if<MetaTexture>(&operation.GetMeta());
         ASSERT(meta);
 
-        std::string expr = GenerateTexture(operation, "textureLod", {{Type::Float, meta->lod}});
+        std::string expr = GenerateTexture(
+            operation, "Lod", {TextureArgument{Type::Float, meta->lod}, TextureAoffi{}});
         if (meta->sampler.IsShadow()) {
             expr = "vec4(" + expr + ')';
         }
@@ -1182,7 +1233,8 @@ private:
         ASSERT(meta);
 
         const auto type = meta->sampler.IsShadow() ? Type::Float : Type::Int;
-        return GenerateTexture(operation, "textureGather", {{type, meta->component}}) +
+        return GenerateTexture(operation, "Gather",
+                               {TextureArgument{type, meta->component}, TextureAoffi{}}) +
                GetSwizzle(meta->element);
     }
 
@@ -1196,11 +1248,12 @@ private:
         switch (meta->element) {
         case 0:
         case 1:
-            return "textureSize(" + sampler + ", " + lod + ')' + GetSwizzle(meta->element);
+            return "itof(int(textureSize(" + sampler + ", " + lod + ')' +
+                   GetSwizzle(meta->element) + "))";
         case 2:
             return "0";
         case 3:
-            return "textureQueryLevels(" + sampler + ')';
+            return "itof(textureQueryLevels(" + sampler + "))";
         }
         UNREACHABLE();
         return "0";
@@ -1211,8 +1264,8 @@ private:
         ASSERT(meta);
 
         if (meta->element < 2) {
-            return "itof(int((" + GenerateTexture(operation, "textureQueryLod", {}) +
-                   " * vec2(256))" + GetSwizzle(meta->element) + "))";
+            return "itof(int((" + GenerateTexture(operation, "QueryLod", {}) + " * vec2(256))" +
+                   GetSwizzle(meta->element) + "))";
         }
         return "0";
     }
@@ -1565,6 +1618,8 @@ private:
     ShaderWriter code;
 };
 
+} // Anonymous namespace
+
 std::string GetCommonDeclarations() {
     const auto cbuf = std::to_string(MAX_CONSTBUFFER_ELEMENTS);
     const auto gmem = std::to_string(MAX_GLOBALMEMORY_ELEMENTS);

src/video_core/renderer_opengl/gl_shader_decompiler.h

@@ -5,7 +5,6 @@
 #pragma once
 
 #include <array>
-#include <set>
 #include <string>
 #include <utility>
 #include <vector>

src/video_core/renderer_opengl/gl_shader_disk_cache.cpp

@@ -10,8 +10,8 @@
 #include "common/common_types.h"
 #include "common/file_util.h"
 #include "common/logging/log.h"
-#include "common/lz4_compression.h"
 #include "common/scm_rev.h"
+#include "common/zstd_compression.h"
 
 #include "core/core.h"
 #include "core/hle/kernel/process.h"
@@ -259,7 +259,7 @@ ShaderDiskCacheOpenGL::LoadPrecompiledFile(FileUtil::IOFile& file) {
                 return {};
             }
 
-            dump.binary = Common::Compression::DecompressDataLZ4(compressed_binary, binary_length);
+            dump.binary = Common::Compression::DecompressDataZSTD(compressed_binary);
             if (dump.binary.empty()) {
                 return {};
             }
@@ -288,7 +288,7 @@ std::optional<ShaderDiskCacheDecompiled> ShaderDiskCacheOpenGL::LoadDecompiledEn
         return {};
     }
 
-    const std::vector<u8> code = Common::Compression::DecompressDataLZ4(compressed_code, code_size);
+    const std::vector<u8> code = Common::Compression::DecompressDataZSTD(compressed_code);
     if (code.empty()) {
         return {};
     }
@@ -474,8 +474,8 @@ void ShaderDiskCacheOpenGL::SaveDecompiled(u64 unique_identifier, const std::str
     if (!IsUsable())
         return;
 
-    const std::vector<u8> compressed_code{Common::Compression::CompressDataLZ4HC(
-        reinterpret_cast<const u8*>(code.data()), code.size(), 9)};
+    const std::vector<u8> compressed_code{Common::Compression::CompressDataZSTDDefault(
+        reinterpret_cast<const u8*>(code.data()), code.size())};
     if (compressed_code.empty()) {
         LOG_ERROR(Render_OpenGL, "Failed to compress GLSL code - skipping shader {:016x}",
                   unique_identifier);
@@ -506,7 +506,7 @@ void ShaderDiskCacheOpenGL::SaveDump(const ShaderDiskCacheUsage& usage, GLuint p
     glGetProgramBinary(program, binary_length, nullptr, &binary_format, binary.data());
 
     const std::vector<u8> compressed_binary =
-        Common::Compression::CompressDataLZ4HC(binary.data(), binary.size(), 9);
+        Common::Compression::CompressDataZSTDDefault(binary.data(), binary.size());
 
     if (compressed_binary.empty()) {
         LOG_ERROR(Render_OpenGL, "Failed to compress binary program in shader={:016x}",

src/video_core/renderer_opengl/gl_shader_gen.cpp

@@ -3,7 +3,6 @@
 // Refer to the license.txt file included.
 
 #include <fmt/format.h>
-#include "common/assert.h"
 #include "video_core/engines/maxwell_3d.h"
 #include "video_core/renderer_opengl/gl_shader_decompiler.h"
 #include "video_core/renderer_opengl/gl_shader_gen.h"

src/video_core/renderer_opengl/gl_shader_gen.h

@@ -4,12 +4,9 @@
 
 #pragma once
 
-#include <array>
-#include <string>
 #include <vector>
 
 #include "common/common_types.h"
-#include "video_core/engines/shader_bytecode.h"
 #include "video_core/renderer_opengl/gl_shader_decompiler.h"
 #include "video_core/shader/shader_ir.h"
 

src/video_core/renderer_opengl/gl_shader_manager.h

@@ -62,7 +62,6 @@ public:
         UpdatePipeline();
         state.draw.shader_program = 0;
         state.draw.program_pipeline = pipeline.handle;
-        state.geometry_shaders.enabled = (gs != 0);
     }
 
 private:

src/video_core/renderer_opengl/gl_state.cpp

@@ -10,16 +10,62 @@
 
 namespace OpenGL {
 
-OpenGLState OpenGLState::cur_state;
+using Maxwell = Tegra::Engines::Maxwell3D::Regs;
 
+OpenGLState OpenGLState::cur_state;
 bool OpenGLState::s_rgb_used;
 
+namespace {
+
+template <typename T>
+bool UpdateValue(T& current_value, const T new_value) {
+    const bool changed = current_value != new_value;
+    current_value = new_value;
+    return changed;
+}
+
+template <typename T1, typename T2>
+bool UpdateTie(T1 current_value, const T2 new_value) {
+    const bool changed = current_value != new_value;
+    current_value = new_value;
+    return changed;
+}
+
+void Enable(GLenum cap, bool enable) {
+    if (enable) {
+        glEnable(cap);
+    } else {
+        glDisable(cap);
+    }
+}
+
+void Enable(GLenum cap, GLuint index, bool enable) {
+    if (enable) {
+        glEnablei(cap, index);
+    } else {
+        glDisablei(cap, index);
+    }
+}
+
+void Enable(GLenum cap, bool& current_value, bool new_value) {
+    if (UpdateValue(current_value, new_value))
+        Enable(cap, new_value);
+}
+
+void Enable(GLenum cap, GLuint index, bool& current_value, bool new_value) {
+    if (UpdateValue(current_value, new_value))
+        Enable(cap, index, new_value);
+}
+
+} // namespace
+
 OpenGLState::OpenGLState() {
     // These all match default OpenGL values
-    geometry_shaders.enabled = false;
     framebuffer_srgb.enabled = false;
+
     multisample_control.alpha_to_coverage = false;
     multisample_control.alpha_to_one = false;
+
     cull.enabled = false;
     cull.mode = GL_BACK;
     cull.front_face = GL_CCW;
@@ -30,14 +76,15 @@ OpenGLState::OpenGLState() {
 
     primitive_restart.enabled = false;
     primitive_restart.index = 0;
+
     for (auto& item : color_mask) {
         item.red_enabled = GL_TRUE;
         item.green_enabled = GL_TRUE;
         item.blue_enabled = GL_TRUE;
         item.alpha_enabled = GL_TRUE;
     }
-    stencil.test_enabled = false;
-    auto reset_stencil = [](auto& config) {
+
+    const auto ResetStencil = [](auto& config) {
         config.test_func = GL_ALWAYS;
         config.test_ref = 0;
         config.test_mask = 0xFFFFFFFF;
@@ -46,8 +93,10 @@ OpenGLState::OpenGLState() {
         config.action_depth_pass = GL_KEEP;
         config.action_stencil_fail = GL_KEEP;
     };
-    reset_stencil(stencil.front);
-    reset_stencil(stencil.back);
+    stencil.test_enabled = false;
+    ResetStencil(stencil.front);
+    ResetStencil(stencil.back);
+
     for (auto& item : viewports) {
         item.x = 0;
         item.y = 0;
@@ -61,6 +110,7 @@ OpenGLState::OpenGLState() {
         item.scissor.width = 0;
         item.scissor.height = 0;
     }
+
     for (auto& item : blend) {
         item.enabled = true;
         item.rgb_equation = GL_FUNC_ADD;
@@ -70,11 +120,14 @@ OpenGLState::OpenGLState() {
         item.src_a_func = GL_ONE;
         item.dst_a_func = GL_ZERO;
     }
+
     independant_blend.enabled = false;
+
     blend_color.red = 0.0f;
     blend_color.green = 0.0f;
     blend_color.blue = 0.0f;
     blend_color.alpha = 0.0f;
+
     logic_op.enabled = false;
     logic_op.operation = GL_COPY;
 
@@ -91,9 +144,12 @@ OpenGLState::OpenGLState() {
     clip_distance = {};
 
     point.size = 1;
+
     fragment_color_clamp.enabled = false;
+
     depth_clamp.far_plane = false;
     depth_clamp.near_plane = false;
+
     polygon_offset.fill_enable = false;
     polygon_offset.line_enable = false;
     polygon_offset.point_enable = false;
@@ -103,132 +159,380 @@ OpenGLState::OpenGLState() {
 }
 
 void OpenGLState::ApplyDefaultState() {
+    glEnable(GL_BLEND);
     glDisable(GL_FRAMEBUFFER_SRGB);
     glDisable(GL_CULL_FACE);
     glDisable(GL_DEPTH_TEST);
     glDisable(GL_PRIMITIVE_RESTART);
     glDisable(GL_STENCIL_TEST);
-    glEnable(GL_BLEND);
     glDisable(GL_COLOR_LOGIC_OP);
     glDisable(GL_SCISSOR_TEST);
 }
 
+void OpenGLState::ApplyFramebufferState() const {
+    if (UpdateValue(cur_state.draw.read_framebuffer, draw.read_framebuffer)) {
+        glBindFramebuffer(GL_READ_FRAMEBUFFER, draw.read_framebuffer);
+    }
+    if (UpdateValue(cur_state.draw.draw_framebuffer, draw.draw_framebuffer)) {
+        glBindFramebuffer(GL_DRAW_FRAMEBUFFER, draw.draw_framebuffer);
+    }
+}
+
+void OpenGLState::ApplyVertexArrayState() const {
+    if (UpdateValue(cur_state.draw.vertex_array, draw.vertex_array)) {
+        glBindVertexArray(draw.vertex_array);
+    }
+}
+
+void OpenGLState::ApplyShaderProgram() const {
+    if (UpdateValue(cur_state.draw.shader_program, draw.shader_program)) {
+        glUseProgram(draw.shader_program);
+    }
+}
+
+void OpenGLState::ApplyProgramPipeline() const {
+    if (UpdateValue(cur_state.draw.program_pipeline, draw.program_pipeline)) {
+        glBindProgramPipeline(draw.program_pipeline);
+    }
+}
+
+void OpenGLState::ApplyClipDistances() const {
+    for (std::size_t i = 0; i < clip_distance.size(); ++i) {
+        Enable(GL_CLIP_DISTANCE0 + static_cast<GLenum>(i), cur_state.clip_distance[i],
+               clip_distance[i]);
+    }
+}
+
+void OpenGLState::ApplyPointSize() const {
+    if (UpdateValue(cur_state.point.size, point.size)) {
+        glPointSize(point.size);
+    }
+}
+
+void OpenGLState::ApplyFragmentColorClamp() const {
+    if (UpdateValue(cur_state.fragment_color_clamp.enabled, fragment_color_clamp.enabled)) {
+        glClampColor(GL_CLAMP_FRAGMENT_COLOR_ARB,
+                     fragment_color_clamp.enabled ? GL_TRUE : GL_FALSE);
+    }
+}
+
+void OpenGLState::ApplyMultisample() const {
+    Enable(GL_SAMPLE_ALPHA_TO_COVERAGE, cur_state.multisample_control.alpha_to_coverage,
+           multisample_control.alpha_to_coverage);
+    Enable(GL_SAMPLE_ALPHA_TO_ONE, cur_state.multisample_control.alpha_to_one,
+           multisample_control.alpha_to_one);
+}
+
+void OpenGLState::ApplyDepthClamp() const {
+    if (depth_clamp.far_plane == cur_state.depth_clamp.far_plane &&
+        depth_clamp.near_plane == cur_state.depth_clamp.near_plane) {
+        return;
+    }
+    cur_state.depth_clamp = depth_clamp;
+
+    UNIMPLEMENTED_IF_MSG(depth_clamp.far_plane != depth_clamp.near_plane,
+                         "Unimplemented Depth Clamp Separation!");
+
+    Enable(GL_DEPTH_CLAMP, depth_clamp.far_plane || depth_clamp.near_plane);
+}
+
 void OpenGLState::ApplySRgb() const {
-    if (framebuffer_srgb.enabled != cur_state.framebuffer_srgb.enabled) {
-        if (framebuffer_srgb.enabled) {
-            // Track if sRGB is used
-            s_rgb_used = true;
-            glEnable(GL_FRAMEBUFFER_SRGB);
-        } else {
-            glDisable(GL_FRAMEBUFFER_SRGB);
-        }
+    if (cur_state.framebuffer_srgb.enabled == framebuffer_srgb.enabled)
+        return;
+    cur_state.framebuffer_srgb.enabled = framebuffer_srgb.enabled;
+    if (framebuffer_srgb.enabled) {
+        // Track if sRGB is used
+        s_rgb_used = true;
+        glEnable(GL_FRAMEBUFFER_SRGB);
+    } else {
+        glDisable(GL_FRAMEBUFFER_SRGB);
     }
 }
 
 void OpenGLState::ApplyCulling() const {
-    if (cull.enabled != cur_state.cull.enabled) {
-        if (cull.enabled) {
-            glEnable(GL_CULL_FACE);
-        } else {
-            glDisable(GL_CULL_FACE);
-        }
-    }
+    Enable(GL_CULL_FACE, cur_state.cull.enabled, cull.enabled);
 
-    if (cull.mode != cur_state.cull.mode) {
+    if (UpdateValue(cur_state.cull.mode, cull.mode)) {
         glCullFace(cull.mode);
     }
 
-    if (cull.front_face != cur_state.cull.front_face) {
+    if (UpdateValue(cur_state.cull.front_face, cull.front_face)) {
         glFrontFace(cull.front_face);
     }
 }
 
 void OpenGLState::ApplyColorMask() const {
-    if (independant_blend.enabled) {
-        for (size_t i = 0; i < Tegra::Engines::Maxwell3D::Regs::NumRenderTargets; i++) {
-            const auto& updated = color_mask[i];
-            const auto& current = cur_state.color_mask[i];
-            if (updated.red_enabled != current.red_enabled ||
-                updated.green_enabled != current.green_enabled ||
-                updated.blue_enabled != current.blue_enabled ||
-                updated.alpha_enabled != current.alpha_enabled) {
-                glColorMaski(static_cast<GLuint>(i), updated.red_enabled, updated.green_enabled,
-                             updated.blue_enabled, updated.alpha_enabled);
-            }
-        }
-    } else {
-        const auto& updated = color_mask[0];
-        const auto& current = cur_state.color_mask[0];
+    for (std::size_t i = 0; i < Maxwell::NumRenderTargets; ++i) {
+        const auto& updated = color_mask[i];
+        auto& current = cur_state.color_mask[i];
         if (updated.red_enabled != current.red_enabled ||
             updated.green_enabled != current.green_enabled ||
             updated.blue_enabled != current.blue_enabled ||
             updated.alpha_enabled != current.alpha_enabled) {
-            glColorMask(updated.red_enabled, updated.green_enabled, updated.blue_enabled,
-                        updated.alpha_enabled);
+            current = updated;
+            glColorMaski(static_cast<GLuint>(i), updated.red_enabled, updated.green_enabled,
+                         updated.blue_enabled, updated.alpha_enabled);
         }
     }
 }
 
 void OpenGLState::ApplyDepth() const {
-    if (depth.test_enabled != cur_state.depth.test_enabled) {
-        if (depth.test_enabled) {
-            glEnable(GL_DEPTH_TEST);
-        } else {
-            glDisable(GL_DEPTH_TEST);
-        }
-    }
+    Enable(GL_DEPTH_TEST, cur_state.depth.test_enabled, depth.test_enabled);
 
-    if (depth.test_func != cur_state.depth.test_func) {
+    if (cur_state.depth.test_func != depth.test_func) {
+        cur_state.depth.test_func = depth.test_func;
         glDepthFunc(depth.test_func);
     }
 
-    if (depth.write_mask != cur_state.depth.write_mask) {
+    if (cur_state.depth.write_mask != depth.write_mask) {
+        cur_state.depth.write_mask = depth.write_mask;
         glDepthMask(depth.write_mask);
     }
 }
 
 void OpenGLState::ApplyPrimitiveRestart() const {
-    if (primitive_restart.enabled != cur_state.primitive_restart.enabled) {
-        if (primitive_restart.enabled) {
-            glEnable(GL_PRIMITIVE_RESTART);
-        } else {
-            glDisable(GL_PRIMITIVE_RESTART);
-        }
-    }
+    Enable(GL_PRIMITIVE_RESTART, cur_state.primitive_restart.enabled, primitive_restart.enabled);
 
-    if (primitive_restart.index != cur_state.primitive_restart.index) {
+    if (cur_state.primitive_restart.index != primitive_restart.index) {
+        cur_state.primitive_restart.index = primitive_restart.index;
         glPrimitiveRestartIndex(primitive_restart.index);
     }
 }
 
 void OpenGLState::ApplyStencilTest() const {
-    if (stencil.test_enabled != cur_state.stencil.test_enabled) {
-        if (stencil.test_enabled) {
-            glEnable(GL_STENCIL_TEST);
-        } else {
-            glDisable(GL_STENCIL_TEST);
-        }
-    }
+    Enable(GL_STENCIL_TEST, cur_state.stencil.test_enabled, stencil.test_enabled);
 
-    const auto ConfigStencil = [](GLenum face, const auto& config, const auto& prev_config) {
-        if (config.test_func != prev_config.test_func || config.test_ref != prev_config.test_ref ||
-            config.test_mask != prev_config.test_mask) {
+    const auto ConfigStencil = [](GLenum face, const auto& config, auto& current) {
+        if (current.test_func != config.test_func || current.test_ref != config.test_ref ||
+            current.test_mask != config.test_mask) {
+            current.test_func = config.test_func;
+            current.test_ref = config.test_ref;
+            current.test_mask = config.test_mask;
             glStencilFuncSeparate(face, config.test_func, config.test_ref, config.test_mask);
         }
-        if (config.action_depth_fail != prev_config.action_depth_fail ||
-            config.action_depth_pass != prev_config.action_depth_pass ||
-            config.action_stencil_fail != prev_config.action_stencil_fail) {
+        if (current.action_depth_fail != config.action_depth_fail ||
+            current.action_depth_pass != config.action_depth_pass ||
+            current.action_stencil_fail != config.action_stencil_fail) {
+            current.action_depth_fail = config.action_depth_fail;
+            current.action_depth_pass = config.action_depth_pass;
+            current.action_stencil_fail = config.action_stencil_fail;
             glStencilOpSeparate(face, config.action_stencil_fail, config.action_depth_fail,
                                 config.action_depth_pass);
         }
-        if (config.write_mask != prev_config.write_mask) {
+        if (current.write_mask != config.write_mask) {
+            current.write_mask = config.write_mask;
             glStencilMaskSeparate(face, config.write_mask);
         }
     };
     ConfigStencil(GL_FRONT, stencil.front, cur_state.stencil.front);
     ConfigStencil(GL_BACK, stencil.back, cur_state.stencil.back);
 }
-// Viewport does not affects glClearBuffer so emulate viewport using scissor test
+
+void OpenGLState::ApplyViewport() const {
+    for (GLuint i = 0; i < static_cast<GLuint>(Maxwell::NumViewports); ++i) {
+        const auto& updated = viewports[i];
+        auto& current = cur_state.viewports[i];
+
+        if (current.x != updated.x || current.y != updated.y || current.width != updated.width ||
+            current.height != updated.height) {
+            current.x = updated.x;
+            current.y = updated.y;
+            current.width = updated.width;
+            current.height = updated.height;
+            glViewportIndexedf(i, static_cast<GLfloat>(updated.x), static_cast<GLfloat>(updated.y),
+                               static_cast<GLfloat>(updated.width),
+                               static_cast<GLfloat>(updated.height));
+        }
+        if (current.depth_range_near != updated.depth_range_near ||
+            current.depth_range_far != updated.depth_range_far) {
+            current.depth_range_near = updated.depth_range_near;
+            current.depth_range_far = updated.depth_range_far;
+            glDepthRangeIndexed(i, updated.depth_range_near, updated.depth_range_far);
+        }
+
+        Enable(GL_SCISSOR_TEST, i, current.scissor.enabled, updated.scissor.enabled);
+
+        if (current.scissor.x != updated.scissor.x || current.scissor.y != updated.scissor.y ||
+            current.scissor.width != updated.scissor.width ||
+            current.scissor.height != updated.scissor.height) {
+            current.scissor.x = updated.scissor.x;
+            current.scissor.y = updated.scissor.y;
+            current.scissor.width = updated.scissor.width;
+            current.scissor.height = updated.scissor.height;
+            glScissorIndexed(i, updated.scissor.x, updated.scissor.y, updated.scissor.width,
+                             updated.scissor.height);
+        }
+    }
+}
+
+void OpenGLState::ApplyGlobalBlending() const {
+    const Blend& updated = blend[0];
+    Blend& current = cur_state.blend[0];
+
+    Enable(GL_BLEND, current.enabled, updated.enabled);
+
+    if (current.src_rgb_func != updated.src_rgb_func ||
+        current.dst_rgb_func != updated.dst_rgb_func || current.src_a_func != updated.src_a_func ||
+        current.dst_a_func != updated.dst_a_func) {
+        current.src_rgb_func = updated.src_rgb_func;
+        current.dst_rgb_func = updated.dst_rgb_func;
+        current.src_a_func = updated.src_a_func;
+        current.dst_a_func = updated.dst_a_func;
+        glBlendFuncSeparate(updated.src_rgb_func, updated.dst_rgb_func, updated.src_a_func,
+                            updated.dst_a_func);
+    }
+
+    if (current.rgb_equation != updated.rgb_equation || current.a_equation != updated.a_equation) {
+        current.rgb_equation = updated.rgb_equation;
+        current.a_equation = updated.a_equation;
+        glBlendEquationSeparate(updated.rgb_equation, updated.a_equation);
+    }
+}
+
+void OpenGLState::ApplyTargetBlending(std::size_t target, bool force) const {
+    const Blend& updated = blend[target];
+    Blend& current = cur_state.blend[target];
+
+    if (current.enabled != updated.enabled || force) {
+        current.enabled = updated.enabled;
+        Enable(GL_BLEND, static_cast<GLuint>(target), updated.enabled);
+    }
+
+    if (UpdateTie(std::tie(current.src_rgb_func, current.dst_rgb_func, current.src_a_func,
+                           current.dst_a_func),
+                  std::tie(updated.src_rgb_func, updated.dst_rgb_func, updated.src_a_func,
+                           updated.dst_a_func))) {
+        glBlendFuncSeparatei(static_cast<GLuint>(target), updated.src_rgb_func,
+                             updated.dst_rgb_func, updated.src_a_func, updated.dst_a_func);
+    }
+
+    if (UpdateTie(std::tie(current.rgb_equation, current.a_equation),
+                  std::tie(updated.rgb_equation, updated.a_equation))) {
+        glBlendEquationSeparatei(static_cast<GLuint>(target), updated.rgb_equation,
+                                 updated.a_equation);
+    }
+}
+
+void OpenGLState::ApplyBlending() const {
+    if (independant_blend.enabled) {
+        const bool force = independant_blend.enabled != cur_state.independant_blend.enabled;
+        for (std::size_t target = 0; target < Maxwell::NumRenderTargets; ++target) {
+            ApplyTargetBlending(target, force);
+        }
+    } else {
+        ApplyGlobalBlending();
+    }
+    cur_state.independant_blend.enabled = independant_blend.enabled;
+
+    if (UpdateTie(
+            std::tie(cur_state.blend_color.red, cur_state.blend_color.green,
+                     cur_state.blend_color.blue, cur_state.blend_color.alpha),
+            std::tie(blend_color.red, blend_color.green, blend_color.blue, blend_color.alpha))) {
+        glBlendColor(blend_color.red, blend_color.green, blend_color.blue, blend_color.alpha);
+    }
+}
+
+void OpenGLState::ApplyLogicOp() const {
+    Enable(GL_COLOR_LOGIC_OP, cur_state.logic_op.enabled, logic_op.enabled);
+
+    if (UpdateValue(cur_state.logic_op.operation, logic_op.operation)) {
+        glLogicOp(logic_op.operation);
+    }
+}
+
+void OpenGLState::ApplyPolygonOffset() const {
+    Enable(GL_POLYGON_OFFSET_FILL, cur_state.polygon_offset.fill_enable,
+           polygon_offset.fill_enable);
+    Enable(GL_POLYGON_OFFSET_LINE, cur_state.polygon_offset.line_enable,
+           polygon_offset.line_enable);
+    Enable(GL_POLYGON_OFFSET_POINT, cur_state.polygon_offset.point_enable,
+           polygon_offset.point_enable);
+
+    if (UpdateTie(std::tie(cur_state.polygon_offset.factor, cur_state.polygon_offset.units,
+                           cur_state.polygon_offset.clamp),
+                  std::tie(polygon_offset.factor, polygon_offset.units, polygon_offset.clamp))) {
+        if (GLAD_GL_EXT_polygon_offset_clamp && polygon_offset.clamp != 0) {
+            glPolygonOffsetClamp(polygon_offset.factor, polygon_offset.units, polygon_offset.clamp);
+        } else {
+            UNIMPLEMENTED_IF_MSG(polygon_offset.clamp != 0,
+                                 "Unimplemented Depth polygon offset clamp.");
+            glPolygonOffset(polygon_offset.factor, polygon_offset.units);
+        }
+    }
+}
+
+void OpenGLState::ApplyTextures() const {
+    bool has_delta{};
+    std::size_t first{};
+    std::size_t last{};
+    std::array<GLuint, Maxwell::NumTextureSamplers> textures;
+
+    for (std::size_t i = 0; i < std::size(texture_units); ++i) {
+        const auto& texture_unit = texture_units[i];
+        auto& cur_state_texture_unit = cur_state.texture_units[i];
+        textures[i] = texture_unit.texture;
+        if (cur_state_texture_unit.texture == textures[i])
+            continue;
+        cur_state_texture_unit.texture = textures[i];
+        if (!has_delta) {
+            first = i;
+            has_delta = true;
+        }
+        last = i;
+    }
+    if (has_delta) {
+        glBindTextures(static_cast<GLuint>(first), static_cast<GLsizei>(last - first + 1),
+                       textures.data() + first);
+    }
+}
+
+void OpenGLState::ApplySamplers() const {
+    bool has_delta{};
+    std::size_t first{};
+    std::size_t last{};
+    std::array<GLuint, Maxwell::NumTextureSamplers> samplers;
+
+    for (std::size_t i = 0; i < std::size(samplers); ++i) {
+        if (cur_state.texture_units[i].sampler == texture_units[i].sampler)
+            continue;
+        cur_state.texture_units[i].sampler = texture_units[i].sampler;
+        samplers[i] = texture_units[i].sampler;
+        if (!has_delta) {
+            first = i;
+            has_delta = true;
+        }
+        last = i;
+    }
+    if (has_delta) {
+        glBindSamplers(static_cast<GLuint>(first), static_cast<GLsizei>(last - first + 1),
+                       samplers.data() + first);
+    }
+}
+
+void OpenGLState::Apply() const {
+    ApplyFramebufferState();
+    ApplyVertexArrayState();
+    ApplyShaderProgram();
+    ApplyProgramPipeline();
+    ApplyClipDistances();
+    ApplyPointSize();
+    ApplyFragmentColorClamp();
+    ApplyMultisample();
+    ApplyDepthClamp();
+    ApplyColorMask();
+    ApplyViewport();
+    ApplyStencilTest();
+    ApplySRgb();
+    ApplyCulling();
+    ApplyDepth();
+    ApplyPrimitiveRestart();
+    ApplyBlending();
+    ApplyLogicOp();
+    ApplyTextures();
+    ApplySamplers();
+    ApplyPolygonOffset();
+}
+
 void OpenGLState::EmulateViewportWithScissor() {
     auto& current = viewports[0];
     if (current.scissor.enabled) {
@@ -251,332 +555,6 @@ void OpenGLState::EmulateViewportWithScissor() {
     }
 }
 
-void OpenGLState::ApplyViewport() const {
-    if (geometry_shaders.enabled) {
-        for (GLuint i = 0; i < static_cast<GLuint>(Tegra::Engines::Maxwell3D::Regs::NumViewports);
-             i++) {
-            const auto& current = cur_state.viewports[i];
-            const auto& updated = viewports[i];
-            if (updated.x != current.x || updated.y != current.y ||
-                updated.width != current.width || updated.height != current.height) {
-                glViewportIndexedf(
-                    i, static_cast<GLfloat>(updated.x), static_cast<GLfloat>(updated.y),
-                    static_cast<GLfloat>(updated.width), static_cast<GLfloat>(updated.height));
-            }
-            if (updated.depth_range_near != current.depth_range_near ||
-                updated.depth_range_far != current.depth_range_far) {
-                glDepthRangeIndexed(i, updated.depth_range_near, updated.depth_range_far);
-            }
-
-            if (updated.scissor.enabled != current.scissor.enabled) {
-                if (updated.scissor.enabled) {
-                    glEnablei(GL_SCISSOR_TEST, i);
-                } else {
-                    glDisablei(GL_SCISSOR_TEST, i);
-                }
-            }
-
-            if (updated.scissor.x != current.scissor.x || updated.scissor.y != current.scissor.y ||
-                updated.scissor.width != current.scissor.width ||
-                updated.scissor.height != current.scissor.height) {
-                glScissorIndexed(i, updated.scissor.x, updated.scissor.y, updated.scissor.width,
-                                 updated.scissor.height);
-            }
-        }
-    } else {
-        const auto& current = cur_state.viewports[0];
-        const auto& updated = viewports[0];
-        if (updated.x != current.x || updated.y != current.y || updated.width != current.width ||
-            updated.height != current.height) {
-            glViewport(updated.x, updated.y, updated.width, updated.height);
-        }
-
-        if (updated.depth_range_near != current.depth_range_near ||
-            updated.depth_range_far != current.depth_range_far) {
-            glDepthRange(updated.depth_range_near, updated.depth_range_far);
-        }
-
-        if (updated.scissor.enabled != current.scissor.enabled) {
-            if (updated.scissor.enabled) {
-                glEnable(GL_SCISSOR_TEST);
-            } else {
-                glDisable(GL_SCISSOR_TEST);
-            }
-        }
-
-        if (updated.scissor.x != current.scissor.x || updated.scissor.y != current.scissor.y ||
-            updated.scissor.width != current.scissor.width ||
-            updated.scissor.height != current.scissor.height) {
-            glScissor(updated.scissor.x, updated.scissor.y, updated.scissor.width,
-                      updated.scissor.height);
-        }
-    }
-}
-
-void OpenGLState::ApplyGlobalBlending() const {
-    const Blend& current = cur_state.blend[0];
-    const Blend& updated = blend[0];
-    if (updated.enabled != current.enabled) {
-        if (updated.enabled) {
-            glEnable(GL_BLEND);
-        } else {
-            glDisable(GL_BLEND);
-        }
-    }
-    if (!updated.enabled) {
-        return;
-    }
-    if (updated.src_rgb_func != current.src_rgb_func ||
-        updated.dst_rgb_func != current.dst_rgb_func || updated.src_a_func != current.src_a_func ||
-        updated.dst_a_func != current.dst_a_func) {
-        glBlendFuncSeparate(updated.src_rgb_func, updated.dst_rgb_func, updated.src_a_func,
-                            updated.dst_a_func);
-    }
-
-    if (updated.rgb_equation != current.rgb_equation || updated.a_equation != current.a_equation) {
-        glBlendEquationSeparate(updated.rgb_equation, updated.a_equation);
-    }
-}
-
-void OpenGLState::ApplyTargetBlending(std::size_t target, bool force) const {
-    const Blend& updated = blend[target];
-    const Blend& current = cur_state.blend[target];
-    if (updated.enabled != current.enabled || force) {
-        if (updated.enabled) {
-            glEnablei(GL_BLEND, static_cast<GLuint>(target));
-        } else {
-            glDisablei(GL_BLEND, static_cast<GLuint>(target));
-        }
-    }
-
-    if (updated.src_rgb_func != current.src_rgb_func ||
-        updated.dst_rgb_func != current.dst_rgb_func || updated.src_a_func != current.src_a_func ||
-        updated.dst_a_func != current.dst_a_func) {
-        glBlendFuncSeparatei(static_cast<GLuint>(target), updated.src_rgb_func,
-                             updated.dst_rgb_func, updated.src_a_func, updated.dst_a_func);
-    }
-
-    if (updated.rgb_equation != current.rgb_equation || updated.a_equation != current.a_equation) {
-        glBlendEquationSeparatei(static_cast<GLuint>(target), updated.rgb_equation,
-                                 updated.a_equation);
-    }
-}
-
-void OpenGLState::ApplyBlending() const {
-    if (independant_blend.enabled) {
-        for (size_t i = 0; i < Tegra::Engines::Maxwell3D::Regs::NumRenderTargets; i++) {
-            ApplyTargetBlending(i,
-                                independant_blend.enabled != cur_state.independant_blend.enabled);
-        }
-    } else {
-        ApplyGlobalBlending();
-    }
-    if (blend_color.red != cur_state.blend_color.red ||
-        blend_color.green != cur_state.blend_color.green ||
-        blend_color.blue != cur_state.blend_color.blue ||
-        blend_color.alpha != cur_state.blend_color.alpha) {
-        glBlendColor(blend_color.red, blend_color.green, blend_color.blue, blend_color.alpha);
-    }
-}
-
-void OpenGLState::ApplyLogicOp() const {
-    if (logic_op.enabled != cur_state.logic_op.enabled) {
-        if (logic_op.enabled) {
-            glEnable(GL_COLOR_LOGIC_OP);
-        } else {
-            glDisable(GL_COLOR_LOGIC_OP);
-        }
-    }
-
-    if (logic_op.operation != cur_state.logic_op.operation) {
-        glLogicOp(logic_op.operation);
-    }
-}
-
-void OpenGLState::ApplyPolygonOffset() const {
-    const bool fill_enable_changed =
-        polygon_offset.fill_enable != cur_state.polygon_offset.fill_enable;
-    const bool line_enable_changed =
-        polygon_offset.line_enable != cur_state.polygon_offset.line_enable;
-    const bool point_enable_changed =
-        polygon_offset.point_enable != cur_state.polygon_offset.point_enable;
-    const bool factor_changed = polygon_offset.factor != cur_state.polygon_offset.factor;
-    const bool units_changed = polygon_offset.units != cur_state.polygon_offset.units;
-    const bool clamp_changed = polygon_offset.clamp != cur_state.polygon_offset.clamp;
-
-    if (fill_enable_changed) {
-        if (polygon_offset.fill_enable) {
-            glEnable(GL_POLYGON_OFFSET_FILL);
-        } else {
-            glDisable(GL_POLYGON_OFFSET_FILL);
-        }
-    }
-
-    if (line_enable_changed) {
-        if (polygon_offset.line_enable) {
-            glEnable(GL_POLYGON_OFFSET_LINE);
-        } else {
-            glDisable(GL_POLYGON_OFFSET_LINE);
-        }
-    }
-
-    if (point_enable_changed) {
-        if (polygon_offset.point_enable) {
-            glEnable(GL_POLYGON_OFFSET_POINT);
-        } else {
-            glDisable(GL_POLYGON_OFFSET_POINT);
-        }
-    }
-
-    if (factor_changed || units_changed || clamp_changed) {
-        if (GLAD_GL_EXT_polygon_offset_clamp && polygon_offset.clamp != 0) {
-            glPolygonOffsetClamp(polygon_offset.factor, polygon_offset.units, polygon_offset.clamp);
-        } else {
-            glPolygonOffset(polygon_offset.factor, polygon_offset.units);
-            UNIMPLEMENTED_IF_MSG(polygon_offset.clamp != 0,
-                                 "Unimplemented Depth polygon offset clamp.");
-        }
-    }
-}
-
-void OpenGLState::ApplyTextures() const {
-    bool has_delta{};
-    std::size_t first{};
-    std::size_t last{};
-    std::array<GLuint, Tegra::Engines::Maxwell3D::Regs::NumTextureSamplers> textures;
-
-    for (std::size_t i = 0; i < std::size(texture_units); ++i) {
-        const auto& texture_unit = texture_units[i];
-        const auto& cur_state_texture_unit = cur_state.texture_units[i];
-        textures[i] = texture_unit.texture;
-
-        if (textures[i] != cur_state_texture_unit.texture) {
-            if (!has_delta) {
-                first = i;
-                has_delta = true;
-            }
-            last = i;
-        }
-    }
-
-    if (has_delta) {
-        glBindTextures(static_cast<GLuint>(first), static_cast<GLsizei>(last - first + 1),
-                       textures.data() + first);
-    }
-}
-
-void OpenGLState::ApplySamplers() const {
-    bool has_delta{};
-    std::size_t first{};
-    std::size_t last{};
-    std::array<GLuint, Tegra::Engines::Maxwell3D::Regs::NumTextureSamplers> samplers;
-    for (std::size_t i = 0; i < std::size(samplers); ++i) {
-        samplers[i] = texture_units[i].sampler;
-        if (samplers[i] != cur_state.texture_units[i].sampler) {
-            if (!has_delta) {
-                first = i;
-                has_delta = true;
-            }
-            last = i;
-        }
-    }
-    if (has_delta) {
-        glBindSamplers(static_cast<GLuint>(first), static_cast<GLsizei>(last - first + 1),
-                       samplers.data() + first);
-    }
-}
-
-void OpenGLState::ApplyFramebufferState() const {
-    if (draw.read_framebuffer != cur_state.draw.read_framebuffer) {
-        glBindFramebuffer(GL_READ_FRAMEBUFFER, draw.read_framebuffer);
-    }
-    if (draw.draw_framebuffer != cur_state.draw.draw_framebuffer) {
-        glBindFramebuffer(GL_DRAW_FRAMEBUFFER, draw.draw_framebuffer);
-    }
-}
-
-void OpenGLState::ApplyVertexArrayState() const {
-    if (draw.vertex_array != cur_state.draw.vertex_array) {
-        glBindVertexArray(draw.vertex_array);
-    }
-}
-
-void OpenGLState::ApplyDepthClamp() const {
-    if (depth_clamp.far_plane == cur_state.depth_clamp.far_plane &&
-        depth_clamp.near_plane == cur_state.depth_clamp.near_plane) {
-        return;
-    }
-    UNIMPLEMENTED_IF_MSG(depth_clamp.far_plane != depth_clamp.near_plane,
-                         "Unimplemented Depth Clamp Separation!");
-
-    if (depth_clamp.far_plane || depth_clamp.near_plane) {
-        glEnable(GL_DEPTH_CLAMP);
-    } else {
-        glDisable(GL_DEPTH_CLAMP);
-    }
-}
-
-void OpenGLState::Apply() const {
-    ApplyFramebufferState();
-    ApplyVertexArrayState();
-
-    // Shader program
-    if (draw.shader_program != cur_state.draw.shader_program) {
-        glUseProgram(draw.shader_program);
-    }
-
-    // Program pipeline
-    if (draw.program_pipeline != cur_state.draw.program_pipeline) {
-        glBindProgramPipeline(draw.program_pipeline);
-    }
-    // Clip distance
-    for (std::size_t i = 0; i < clip_distance.size(); ++i) {
-        if (clip_distance[i] != cur_state.clip_distance[i]) {
-            if (clip_distance[i]) {
-                glEnable(GL_CLIP_DISTANCE0 + static_cast<GLenum>(i));
-            } else {
-                glDisable(GL_CLIP_DISTANCE0 + static_cast<GLenum>(i));
-            }
-        }
-    }
-    // Point
-    if (point.size != cur_state.point.size) {
-        glPointSize(point.size);
-    }
-    if (fragment_color_clamp.enabled != cur_state.fragment_color_clamp.enabled) {
-        glClampColor(GL_CLAMP_FRAGMENT_COLOR_ARB,
-                     fragment_color_clamp.enabled ? GL_TRUE : GL_FALSE);
-    }
-    if (multisample_control.alpha_to_coverage != cur_state.multisample_control.alpha_to_coverage) {
-        if (multisample_control.alpha_to_coverage) {
-            glEnable(GL_SAMPLE_ALPHA_TO_COVERAGE);
-        } else {
-            glDisable(GL_SAMPLE_ALPHA_TO_COVERAGE);
-        }
-    }
-    if (multisample_control.alpha_to_one != cur_state.multisample_control.alpha_to_one) {
-        if (multisample_control.alpha_to_one) {
-            glEnable(GL_SAMPLE_ALPHA_TO_ONE);
-        } else {
-            glDisable(GL_SAMPLE_ALPHA_TO_ONE);
-        }
-    }
-    ApplyDepthClamp();
-    ApplyColorMask();
-    ApplyViewport();
-    ApplyStencilTest();
-    ApplySRgb();
-    ApplyCulling();
-    ApplyDepth();
-    ApplyPrimitiveRestart();
-    ApplyBlending();
-    ApplyLogicOp();
-    ApplyTextures();
-    ApplySamplers();
-    ApplyPolygonOffset();
-    cur_state = *this;
-}
-
 OpenGLState& OpenGLState::UnbindTexture(GLuint handle) {
     for (auto& unit : texture_units) {
         if (unit.texture == handle) {

src/video_core/renderer_opengl/gl_state.h

@@ -53,10 +53,6 @@ public:
         bool near_plane;
     } depth_clamp; // GL_DEPTH_CLAMP
 
-    struct {
-        bool enabled; // viewports arrays are only supported when geometry shaders are enabled.
-    } geometry_shaders;
-
     struct {
         bool enabled;      // GL_CULL_FACE
         GLenum mode;       // GL_CULL_FACE_MODE
@@ -184,34 +180,26 @@ public:
     static OpenGLState GetCurState() {
         return cur_state;
     }
+
     static bool GetsRGBUsed() {
         return s_rgb_used;
     }
+
     static void ClearsRGBUsed() {
         s_rgb_used = false;
     }
+
     /// Apply this state as the current OpenGL state
     void Apply() const;
-    /// Apply only the state affecting the framebuffer
-    void ApplyFramebufferState() const;
-    /// Apply only the state affecting the vertex array
-    void ApplyVertexArrayState() const;
-    /// Set the initial OpenGL state
-    static void ApplyDefaultState();
-    /// Resets any references to the given resource
-    OpenGLState& UnbindTexture(GLuint handle);
-    OpenGLState& ResetSampler(GLuint handle);
-    OpenGLState& ResetProgram(GLuint handle);
-    OpenGLState& ResetPipeline(GLuint handle);
-    OpenGLState& ResetVertexArray(GLuint handle);
-    OpenGLState& ResetFramebuffer(GLuint handle);
-    void EmulateViewportWithScissor();
 
-private:
-    static OpenGLState cur_state;
-    // Workaround for sRGB problems caused by
-    // QT not supporting srgb output
-    static bool s_rgb_used;
+    void ApplyFramebufferState() const;
+    void ApplyVertexArrayState() const;
+    void ApplyShaderProgram() const;
+    void ApplyProgramPipeline() const;
+    void ApplyClipDistances() const;
+    void ApplyPointSize() const;
+    void ApplyFragmentColorClamp() const;
+    void ApplyMultisample() const;
     void ApplySRgb() const;
     void ApplyCulling() const;
     void ApplyColorMask() const;
@@ -227,6 +215,26 @@ private:
     void ApplySamplers() const;
     void ApplyDepthClamp() const;
     void ApplyPolygonOffset() const;
+
+    /// Set the initial OpenGL state
+    static void ApplyDefaultState();
+
+    /// Resets any references to the given resource
+    OpenGLState& UnbindTexture(GLuint handle);
+    OpenGLState& ResetSampler(GLuint handle);
+    OpenGLState& ResetProgram(GLuint handle);
+    OpenGLState& ResetPipeline(GLuint handle);
+    OpenGLState& ResetVertexArray(GLuint handle);
+    OpenGLState& ResetFramebuffer(GLuint handle);
+
+    /// Viewport does not affects glClearBuffer so emulate viewport using scissor test
+    void EmulateViewportWithScissor();
+
+private:
+    static OpenGLState cur_state;
+
+    // Workaround for sRGB problems caused by QT not supporting srgb output
+    static bool s_rgb_used;
 };
 
 } // namespace OpenGL

src/video_core/renderer_opengl/renderer_opengl.cpp

@@ -5,7 +5,6 @@
 #include <algorithm>
 #include <cstddef>
 #include <cstdlib>
-#include <cstring>
 #include <memory>
 #include <glad/glad.h>
 #include "common/assert.h"

src/video_core/renderer_vulkan/vk_buffer_cache.cpp

@@ -10,6 +10,7 @@
 #include "common/alignment.h"
 #include "common/assert.h"
 #include "core/memory.h"
+#include "video_core/memory_manager.h"
 #include "video_core/renderer_vulkan/declarations.h"
 #include "video_core/renderer_vulkan/vk_buffer_cache.h"
 #include "video_core/renderer_vulkan/vk_scheduler.h"

src/video_core/shader/decode/texture.cpp

@@ -7,7 +7,9 @@
 #include <fmt/format.h>
 
 #include "common/assert.h"
+#include "common/bit_field.h"
 #include "common/common_types.h"
+#include "common/logging/log.h"
 #include "video_core/engines/shader_bytecode.h"
 #include "video_core/shader/shader_ir.h"
 
@@ -41,19 +43,18 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
 
     switch (opcode->get().GetId()) {
     case OpCode::Id::TEX: {
-        UNIMPLEMENTED_IF_MSG(instr.tex.UsesMiscMode(TextureMiscMode::AOFFI),
-                             "AOFFI is not implemented");
-
         if (instr.tex.UsesMiscMode(TextureMiscMode::NODEP)) {
             LOG_WARNING(HW_GPU, "TEX.NODEP implementation is incomplete");
         }
 
         const TextureType texture_type{instr.tex.texture_type};
         const bool is_array = instr.tex.array != 0;
+        const bool is_aoffi = instr.tex.UsesMiscMode(TextureMiscMode::AOFFI);
         const bool depth_compare = instr.tex.UsesMiscMode(TextureMiscMode::DC);
         const auto process_mode = instr.tex.GetTextureProcessMode();
         WriteTexInstructionFloat(
-            bb, instr, GetTexCode(instr, texture_type, process_mode, depth_compare, is_array));
+            bb, instr,
+            GetTexCode(instr, texture_type, process_mode, depth_compare, is_array, is_aoffi));
         break;
     }
     case OpCode::Id::TEXS: {
@@ -78,8 +79,6 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
     }
     case OpCode::Id::TLD4: {
         ASSERT(instr.tld4.array == 0);
-        UNIMPLEMENTED_IF_MSG(instr.tld4.UsesMiscMode(TextureMiscMode::AOFFI),
-                             "AOFFI is not implemented");
         UNIMPLEMENTED_IF_MSG(instr.tld4.UsesMiscMode(TextureMiscMode::NDV),
                              "NDV is not implemented");
         UNIMPLEMENTED_IF_MSG(instr.tld4.UsesMiscMode(TextureMiscMode::PTP),
@@ -92,8 +91,9 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
         const auto texture_type = instr.tld4.texture_type.Value();
         const bool depth_compare = instr.tld4.UsesMiscMode(TextureMiscMode::DC);
         const bool is_array = instr.tld4.array != 0;
-        WriteTexInstructionFloat(bb, instr,
-                                 GetTld4Code(instr, texture_type, depth_compare, is_array));
+        const bool is_aoffi = instr.tld4.UsesMiscMode(TextureMiscMode::AOFFI);
+        WriteTexInstructionFloat(
+            bb, instr, GetTld4Code(instr, texture_type, depth_compare, is_array, is_aoffi));
         break;
     }
     case OpCode::Id::TLD4S: {
@@ -127,7 +127,7 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
         Node4 values;
         for (u32 element = 0; element < values.size(); ++element) {
             auto coords_copy = coords;
-            MetaTexture meta{sampler, {}, {}, {}, {}, component, element};
+            MetaTexture meta{sampler, {}, {}, {}, {}, {}, component, element};
             values[element] = Operation(OperationCode::TextureGather, meta, std::move(coords_copy));
         }
 
@@ -152,7 +152,7 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
                 if (!instr.txq.IsComponentEnabled(element)) {
                     continue;
                 }
-                MetaTexture meta{sampler, {}, {}, {}, {}, {}, element};
+                MetaTexture meta{sampler, {}, {}, {}, {}, {}, {}, element};
                 const Node value =
                     Operation(OperationCode::TextureQueryDimensions, meta, GetRegister(instr.gpr8));
                 SetTemporal(bb, indexer++, value);
@@ -202,7 +202,7 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
 
         for (u32 element = 0; element < 2; ++element) {
             auto params = coords;
-            MetaTexture meta{sampler, {}, {}, {}, {}, {}, element};
+            MetaTexture meta{sampler, {}, {}, {}, {}, {}, {}, element};
             const Node value = Operation(OperationCode::TextureQueryLod, meta, std::move(params));
             SetTemporal(bb, element, value);
         }
@@ -325,7 +325,8 @@ void ShaderIR::WriteTexsInstructionHalfFloat(NodeBlock& bb, Instruction instr,
 
 Node4 ShaderIR::GetTextureCode(Instruction instr, TextureType texture_type,
                                TextureProcessMode process_mode, std::vector<Node> coords,
-                               Node array, Node depth_compare, u32 bias_offset) {
+                               Node array, Node depth_compare, u32 bias_offset,
+                               std::vector<Node> aoffi) {
     const bool is_array = array;
     const bool is_shadow = depth_compare;
 
@@ -374,7 +375,7 @@ Node4 ShaderIR::GetTextureCode(Instruction instr, TextureType texture_type,
     Node4 values;
     for (u32 element = 0; element < values.size(); ++element) {
         auto copy_coords = coords;
-        MetaTexture meta{sampler, array, depth_compare, bias, lod, {}, element};
+        MetaTexture meta{sampler, array, depth_compare, aoffi, bias, lod, {}, element};
         values[element] = Operation(read_method, meta, std::move(copy_coords));
     }
 
@@ -382,9 +383,15 @@ Node4 ShaderIR::GetTextureCode(Instruction instr, TextureType texture_type,
 }
 
 Node4 ShaderIR::GetTexCode(Instruction instr, TextureType texture_type,
-                           TextureProcessMode process_mode, bool depth_compare, bool is_array) {
-    const bool lod_bias_enabled =
-        (process_mode != TextureProcessMode::None && process_mode != TextureProcessMode::LZ);
+                           TextureProcessMode process_mode, bool depth_compare, bool is_array,
+                           bool is_aoffi) {
+    const bool lod_bias_enabled{
+        (process_mode != TextureProcessMode::None && process_mode != TextureProcessMode::LZ)};
+
+    u64 parameter_register = instr.gpr20.Value();
+    if (lod_bias_enabled) {
+        ++parameter_register;
+    }
 
     const auto [coord_count, total_coord_count] = ValidateAndGetCoordinateElement(
         texture_type, depth_compare, is_array, lod_bias_enabled, 4, 5);
@@ -404,15 +411,19 @@ Node4 ShaderIR::GetTexCode(Instruction instr, TextureType texture_type,
 
     const Node array = is_array ? GetRegister(array_register) : nullptr;
 
+    std::vector<Node> aoffi;
+    if (is_aoffi) {
+        aoffi = GetAoffiCoordinates(GetRegister(parameter_register++), coord_count, false);
+    }
+
     Node dc{};
     if (depth_compare) {
         // Depth is always stored in the register signaled by gpr20 or in the next register if lod
         // or bias are used
-        const u64 depth_register = instr.gpr20.Value() + (lod_bias_enabled ? 1 : 0);
-        dc = GetRegister(depth_register);
+        dc = GetRegister(parameter_register++);
     }
 
-    return GetTextureCode(instr, texture_type, process_mode, coords, array, dc, 0);
+    return GetTextureCode(instr, texture_type, process_mode, coords, array, dc, 0, aoffi);
 }
 
 Node4 ShaderIR::GetTexsCode(Instruction instr, TextureType texture_type,
@@ -448,11 +459,11 @@ Node4 ShaderIR::GetTexsCode(Instruction instr, TextureType texture_type,
         dc = GetRegister(depth_register);
     }
 
-    return GetTextureCode(instr, texture_type, process_mode, coords, array, dc, bias_offset);
+    return GetTextureCode(instr, texture_type, process_mode, coords, array, dc, bias_offset, {});
 }
 
 Node4 ShaderIR::GetTld4Code(Instruction instr, TextureType texture_type, bool depth_compare,
-                            bool is_array) {
+                            bool is_array, bool is_aoffi) {
     const std::size_t coord_count = GetCoordCount(texture_type);
     const std::size_t total_coord_count = coord_count + (is_array ? 1 : 0);
     const std::size_t total_reg_count = total_coord_count + (depth_compare ? 1 : 0);
@@ -463,15 +474,27 @@ Node4 ShaderIR::GetTld4Code(Instruction instr, TextureType texture_type, bool de
     const u64 coord_register = array_register + (is_array ? 1 : 0);
 
     std::vector<Node> coords;
-    for (size_t i = 0; i < coord_count; ++i)
+    for (std::size_t i = 0; i < coord_count; ++i) {
         coords.push_back(GetRegister(coord_register + i));
+    }
+
+    u64 parameter_register = instr.gpr20.Value();
+    std::vector<Node> aoffi;
+    if (is_aoffi) {
+        aoffi = GetAoffiCoordinates(GetRegister(parameter_register++), coord_count, true);
+    }
+
+    Node dc{};
+    if (depth_compare) {
+        dc = GetRegister(parameter_register++);
+    }
 
     const auto& sampler = GetSampler(instr.sampler, texture_type, is_array, depth_compare);
 
     Node4 values;
     for (u32 element = 0; element < values.size(); ++element) {
         auto coords_copy = coords;
-        MetaTexture meta{sampler, GetRegister(array_register), {}, {}, {}, {}, element};
+        MetaTexture meta{sampler, GetRegister(array_register), dc, aoffi, {}, {}, {}, element};
         values[element] = Operation(OperationCode::TextureGather, meta, std::move(coords_copy));
     }
 
@@ -507,7 +530,7 @@ Node4 ShaderIR::GetTldsCode(Instruction instr, TextureType texture_type, bool is
     Node4 values;
     for (u32 element = 0; element < values.size(); ++element) {
         auto coords_copy = coords;
-        MetaTexture meta{sampler, array, {}, {}, lod, {}, element};
+        MetaTexture meta{sampler, array, {}, {}, {}, lod, {}, element};
         values[element] = Operation(OperationCode::TexelFetch, meta, std::move(coords_copy));
     }
     return values;
@@ -531,4 +554,45 @@ std::tuple<std::size_t, std::size_t> ShaderIR::ValidateAndGetCoordinateElement(
     return {coord_count, total_coord_count};
 }
 
-} // namespace VideoCommon::Shader
+std::vector<Node> ShaderIR::GetAoffiCoordinates(Node aoffi_reg, std::size_t coord_count,
+                                                bool is_tld4) {
+    const auto [coord_offsets, size, wrap_value,
+                diff_value] = [is_tld4]() -> std::tuple<std::array<u32, 3>, u32, s32, s32> {
+        if (is_tld4) {
+            return {{0, 8, 16}, 6, 32, 64};
+        } else {
+            return {{0, 4, 8}, 4, 8, 16};
+        }
+    }();
+    const u32 mask = (1U << size) - 1;
+
+    std::vector<Node> aoffi;
+    aoffi.reserve(coord_count);
+
+    const auto aoffi_immediate{
+        TrackImmediate(aoffi_reg, global_code, static_cast<s64>(global_code.size()))};
+    if (!aoffi_immediate) {
+        // Variable access, not supported on AMD.
+        LOG_WARNING(HW_GPU,
+                    "AOFFI constant folding failed, some hardware might have graphical issues");
+        for (std::size_t coord = 0; coord < coord_count; ++coord) {
+            const Node value = BitfieldExtract(aoffi_reg, coord_offsets.at(coord), size);
+            const Node condition =
+                Operation(OperationCode::LogicalIGreaterEqual, value, Immediate(wrap_value));
+            const Node negative = Operation(OperationCode::IAdd, value, Immediate(-diff_value));
+            aoffi.push_back(Operation(OperationCode::Select, condition, negative, value));
+        }
+        return aoffi;
+    }
+
+    for (std::size_t coord = 0; coord < coord_count; ++coord) {
+        s32 value = (*aoffi_immediate >> coord_offsets.at(coord)) & mask;
+        if (value >= wrap_value) {
+            value -= diff_value;
+        }
+        aoffi.push_back(Immediate(value));
+    }
+    return aoffi;
+}
+
+} // namespace VideoCommon::Shader

src/video_core/shader/shader_ir.h

@@ -7,6 +7,7 @@
 #include <array>
 #include <cstring>
 #include <map>
+#include <optional>
 #include <set>
 #include <string>
 #include <tuple>
@@ -290,6 +291,7 @@ struct MetaTexture {
     const Sampler& sampler;
     Node array{};
     Node depth_compare{};
+    std::vector<Node> aoffi;
     Node bias{};
     Node lod{};
     Node component{};
@@ -741,14 +743,14 @@ private:
 
     Node4 GetTexCode(Tegra::Shader::Instruction instr, Tegra::Shader::TextureType texture_type,
                      Tegra::Shader::TextureProcessMode process_mode, bool depth_compare,
-                     bool is_array);
+                     bool is_array, bool is_aoffi);
 
     Node4 GetTexsCode(Tegra::Shader::Instruction instr, Tegra::Shader::TextureType texture_type,
                       Tegra::Shader::TextureProcessMode process_mode, bool depth_compare,
                       bool is_array);
 
     Node4 GetTld4Code(Tegra::Shader::Instruction instr, Tegra::Shader::TextureType texture_type,
-                      bool depth_compare, bool is_array);
+                      bool depth_compare, bool is_array, bool is_aoffi);
 
     Node4 GetTldsCode(Tegra::Shader::Instruction instr, Tegra::Shader::TextureType texture_type,
                       bool is_array);
@@ -757,9 +759,11 @@ private:
         Tegra::Shader::TextureType texture_type, bool depth_compare, bool is_array,
         bool lod_bias_enabled, std::size_t max_coords, std::size_t max_inputs);
 
+    std::vector<Node> GetAoffiCoordinates(Node aoffi_reg, std::size_t coord_count, bool is_tld4);
+
     Node4 GetTextureCode(Tegra::Shader::Instruction instr, Tegra::Shader::TextureType texture_type,
                          Tegra::Shader::TextureProcessMode process_mode, std::vector<Node> coords,
-                         Node array, Node depth_compare, u32 bias_offset);
+                         Node array, Node depth_compare, u32 bias_offset, std::vector<Node> aoffi);
 
     Node GetVideoOperand(Node op, bool is_chunk, bool is_signed, Tegra::Shader::VideoType type,
                          u64 byte_height);
@@ -773,6 +777,8 @@ private:
 
     Node TrackCbuf(Node tracked, const NodeBlock& code, s64 cursor);
 
+    std::optional<u32> TrackImmediate(Node tracked, const NodeBlock& code, s64 cursor);
+
     std::pair<Node, s64> TrackRegister(const GprNode* tracked, const NodeBlock& code, s64 cursor);
 
     template <typename... T>

src/video_core/shader/track.cpp

@@ -6,6 +6,7 @@
 #include <utility>
 #include <variant>
 
+#include "common/common_types.h"
 #include "video_core/shader/shader_ir.h"
 
 namespace VideoCommon::Shader {
@@ -14,7 +15,7 @@ namespace {
 std::pair<Node, s64> FindOperation(const NodeBlock& code, s64 cursor,
                                    OperationCode operation_code) {
     for (; cursor >= 0; --cursor) {
-        const Node node = code[cursor];
+        const Node node = code.at(cursor);
         if (const auto operation = std::get_if<OperationNode>(node)) {
             if (operation->GetCode() == operation_code)
                 return {node, cursor};
@@ -64,6 +65,20 @@ Node ShaderIR::TrackCbuf(Node tracked, const NodeBlock& code, s64 cursor) {
     return nullptr;
 }
 
+std::optional<u32> ShaderIR::TrackImmediate(Node tracked, const NodeBlock& code, s64 cursor) {
+    // Reduce the cursor in one to avoid infinite loops when the instruction sets the same register
+    // that it uses as operand
+    const auto [found, found_cursor] =
+        TrackRegister(&std::get<GprNode>(*tracked), code, cursor - 1);
+    if (!found) {
+        return {};
+    }
+    if (const auto immediate = std::get_if<ImmediateNode>(found)) {
+        return immediate->GetValue();
+    }
+    return {};
+}
+
 std::pair<Node, s64> ShaderIR::TrackRegister(const GprNode* tracked, const NodeBlock& code,
                                              s64 cursor) {
     for (; cursor >= 0; --cursor) {

src/yuzu/configuration/config.cpp

@@ -373,6 +373,7 @@ void Config::ReadValues() {
         ReadSetting("use_accurate_gpu_emulation", false).toBool();
     Settings::values.use_asynchronous_gpu_emulation =
         ReadSetting("use_asynchronous_gpu_emulation", false).toBool();
+    Settings::values.force_30fps_mode = ReadSetting("force_30fps_mode", false).toBool();
 
     Settings::values.bg_red = ReadSetting("bg_red", 0.0).toFloat();
     Settings::values.bg_green = ReadSetting("bg_green", 0.0).toFloat();
@@ -648,6 +649,7 @@ void Config::SaveValues() {
     WriteSetting("use_accurate_gpu_emulation", Settings::values.use_accurate_gpu_emulation, false);
     WriteSetting("use_asynchronous_gpu_emulation", Settings::values.use_asynchronous_gpu_emulation,
                  false);
+    WriteSetting("force_30fps_mode", Settings::values.force_30fps_mode, false);
 
     // Cast to double because Qt's written float values are not human-readable
     WriteSetting("bg_red", (double)Settings::values.bg_red, 0.0);

src/yuzu/configuration/configure_graphics.cpp

@@ -77,6 +77,8 @@ void ConfigureGraphics::setConfiguration() {
     ui->use_accurate_gpu_emulation->setChecked(Settings::values.use_accurate_gpu_emulation);
     ui->use_asynchronous_gpu_emulation->setEnabled(!Core::System::GetInstance().IsPoweredOn());
     ui->use_asynchronous_gpu_emulation->setChecked(Settings::values.use_asynchronous_gpu_emulation);
+    ui->force_30fps_mode->setEnabled(!Core::System::GetInstance().IsPoweredOn());
+    ui->force_30fps_mode->setChecked(Settings::values.force_30fps_mode);
     UpdateBackgroundColorButton(QColor::fromRgbF(Settings::values.bg_red, Settings::values.bg_green,
                                                  Settings::values.bg_blue));
 }
@@ -90,6 +92,7 @@ void ConfigureGraphics::applyConfiguration() {
     Settings::values.use_accurate_gpu_emulation = ui->use_accurate_gpu_emulation->isChecked();
     Settings::values.use_asynchronous_gpu_emulation =
         ui->use_asynchronous_gpu_emulation->isChecked();
+    Settings::values.force_30fps_mode = ui->force_30fps_mode->isChecked();
     Settings::values.bg_red = static_cast<float>(bg_color.redF());
     Settings::values.bg_green = static_cast<float>(bg_color.greenF());
     Settings::values.bg_blue = static_cast<float>(bg_color.blueF());

src/yuzu/configuration/configure_graphics.ui

@@ -70,6 +70,13 @@
           </property>
          </widget>
         </item>
+        <item>
+         <widget class="QCheckBox" name="force_30fps_mode">
+          <property name="text">
+           <string>Force 30 FPS mode</string>
+          </property>
+         </widget>
+        </item>
         <item>
          <layout class="QHBoxLayout" name="horizontalLayout">
           <item>