"Merge Tagged PR 1012"

Kernel: Correct behavior of Condition Variables to be more similar to real hardware.

.ci/scripts/windows/upload.ps1

@@ -1,11 +1,12 @@
 param($BUILD_NAME)
 
-$GITDATE = $(git show -s --date=short --format='%ad') -replace "-",""
+$GITDATE = $(git show -s --date=short --format='%ad') -replace "-", ""
 $GITREV = $(git show -s --format='%h')
 
 if ("$BUILD_NAME" -eq "mainline") {
     $RELEASE_DIST = "yuzu-windows-msvc"
-} else {
+}
+else {
     $RELEASE_DIST = "yuzu-windows-msvc-$BUILD_NAME"
 }
 
@@ -14,7 +15,9 @@ $MSVC_BUILD_PDB = "yuzu-windows-msvc-$GITDATE-$GITREV-debugsymbols.zip" -replace
 $MSVC_SEVENZIP = "yuzu-windows-msvc-$GITDATE-$GITREV.7z" -replace " ", ""
 $MSVC_TAR = "yuzu-windows-msvc-$GITDATE-$GITREV.tar" -replace " ", ""
 $MSVC_TARXZ = "yuzu-windows-msvc-$GITDATE-$GITREV.tar.xz" -replace " ", ""
-$MSVC_SOURCE = "yuzu-windows-msvc-source-$GITDATE-$GITREV.tar.xz" -replace " ", ""
+$MSVC_SOURCE = "yuzu-windows-msvc-source-$GITDATE-$GITREV" -replace " ", ""
+$MSVC_SOURCE_TAR = "$MSVC_SOURCE.tar"
+$MSVC_SOURCE_TARXZ = "$MSVC_SOURCE_TAR.xz"
 
 $env:BUILD_ZIP = $MSVC_BUILD_ZIP
 $env:BUILD_SYMBOLS = $MSVC_BUILD_PDB
@@ -22,21 +25,33 @@ $env:BUILD_UPDATE = $MSVC_SEVENZIP
 
 $BUILD_DIR = ".\build\bin\Release"
 
+# Upload debugging symbols
 mkdir pdb
 Get-ChildItem "$BUILD_DIR\" -Recurse -Filter "*.pdb" | Copy-Item -destination .\pdb
 7z a -tzip $MSVC_BUILD_PDB .\pdb\*.pdb
 rm "$BUILD_DIR\*.pdb"
+
+# Create artifact directories
 mkdir $RELEASE_DIST
+mkdir $MSVC_SOURCE
 mkdir "artifacts"
 
+# Build a tar.xz for the source of the release
+Copy-Item .\license.txt -Destination $MSVC_SOURCE
+Copy-Item .\README.md -Destination $MSVC_SOURCE
+Copy-Item .\src -Recurse -Destination $MSVC_SOURCE
+Copy-Item .\externals -Recurse -Destination $MSVC_SOURCE
+Copy-Item .\dist -Recurse -Destination $MSVC_SOURCE
+Copy-Item .\CMakeModules -Recurse -Destination $MSVC_SOURCE
+7z a -r -ttar $MSVC_SOURCE_TAR $MSVC_SOURCE
+7z a -r -txz $MSVC_SOURCE_TARXZ $MSVC_SOURCE_TAR
+
+# Build the final release artifacts
+Copy-Item $MSVC_SOURCE_TARXZ -Destination $RELEASE_DIST
 Copy-Item "$BUILD_DIR\*" -Destination $RELEASE_DIST -Recurse
 rm "$RELEASE_DIST\*.exe"
 Get-ChildItem "$BUILD_DIR" -Recurse -Filter "yuzu*.exe" | Copy-Item -destination $RELEASE_DIST
 Get-ChildItem "$BUILD_DIR" -Recurse -Filter "QtWebEngineProcess*.exe" | Copy-Item -destination $RELEASE_DIST
-Copy-Item .\license.txt -Destination $RELEASE_DIST
-Copy-Item .\README.md -Destination $RELEASE_DIST
-Copy-Item .\src -Recurse -Destination $RELEASE_DIST
-Copy-Item .\externals -Recurse -Destination $RELEASE_DIST
 7z a -tzip $MSVC_BUILD_ZIP $RELEASE_DIST\*
 7z a $MSVC_SEVENZIP $RELEASE_DIST
 

CONTRIBUTING.md

@@ -1 +1 @@
-**The Contributor's Guide has moved to [the Yuzu wiki](https://github.com/yuzu-emu/yuzu/wiki/Contributing).**
+**The Contributor's Guide has moved to [the yuzu wiki](https://github.com/yuzu-emu/yuzu/wiki/Contributing).**

externals/microprofile/microprofile.h

@@ -814,7 +814,7 @@ struct MicroProfile
 
 inline int MicroProfileLogType(MicroProfileLogEntry Index)
 {
-    return ((MP_LOG_BEGIN_MASK & Index)>>62) & 0x3;
+    return (int)(((MP_LOG_BEGIN_MASK & Index)>>62) & 0x3ULL);
 }
 
 inline uint64_t MicroProfileLogTimerIndex(MicroProfileLogEntry Index)
@@ -861,12 +861,12 @@ T MicroProfileMax(T a, T b)
 
 inline int64_t MicroProfileMsToTick(float fMs, int64_t nTicksPerSecond)
 {
-    return (int64_t)(fMs*0.001f*nTicksPerSecond);
+    return (int64_t)(fMs*0.001f*(float)nTicksPerSecond);
 }
 
 inline float MicroProfileTickToMsMultiplier(int64_t nTicksPerSecond)
 {
-    return 1000.f / nTicksPerSecond;
+    return 1000.f / (float)nTicksPerSecond;
 }
 
 inline uint16_t MicroProfileGetGroupIndex(MicroProfileToken t)

src/common/assert.h

@@ -28,18 +28,14 @@ __declspec(noinline, noreturn)
 }
 
 #define ASSERT(_a_)                                                                                \
-    do                                                                                             \
-        if (!(_a_)) {                                                                              \
-            assert_noinline_call([] { LOG_CRITICAL(Debug, "Assertion Failed!"); });                \
-        }                                                                                          \
-    while (0)
+    if (!(_a_)) {                                                                                  \
+        LOG_CRITICAL(Debug, "Assertion Failed!");                                                  \
+    }
 
 #define ASSERT_MSG(_a_, ...)                                                                       \
-    do                                                                                             \
-        if (!(_a_)) {                                                                              \
-            assert_noinline_call([&] { LOG_CRITICAL(Debug, "Assertion Failed!\n" __VA_ARGS__); }); \
-        }                                                                                          \
-    while (0)
+    if (!(_a_)) {                                                                                  \
+        LOG_CRITICAL(Debug, "Assertion Failed! " __VA_ARGS__);                                     \
+    }
 
 #define UNREACHABLE() ASSERT_MSG(false, "Unreachable code!")
 #define UNREACHABLE_MSG(...) ASSERT_MSG(false, __VA_ARGS__)

src/common/common_funcs.h

@@ -19,13 +19,15 @@
 
 /// Helper macros to insert unused bytes or words to properly align structs. These values will be
 /// zero-initialized.
-#define INSERT_PADDING_BYTES(num_bytes) std::array<u8, num_bytes> CONCAT2(pad, __LINE__){};
-#define INSERT_PADDING_WORDS(num_words) std::array<u32, num_words> CONCAT2(pad, __LINE__){};
+#define INSERT_PADDING_BYTES(num_bytes)                                                            \
+    std::array<u8, num_bytes> CONCAT2(pad, __LINE__) {}
+#define INSERT_PADDING_WORDS(num_words)                                                            \
+    std::array<u32, num_words> CONCAT2(pad, __LINE__) {}
 
 /// These are similar to the INSERT_PADDING_* macros, but are needed for padding unions. This is
 /// because unions can only be initialized by one member.
-#define INSERT_UNION_PADDING_BYTES(num_bytes) std::array<u8, num_bytes> CONCAT2(pad, __LINE__);
-#define INSERT_UNION_PADDING_WORDS(num_words) std::array<u32, num_words> CONCAT2(pad, __LINE__);
+#define INSERT_UNION_PADDING_BYTES(num_bytes) std::array<u8, num_bytes> CONCAT2(pad, __LINE__)
+#define INSERT_UNION_PADDING_WORDS(num_words) std::array<u32, num_words> CONCAT2(pad, __LINE__)
 
 #ifndef _MSC_VER
 
@@ -56,7 +58,7 @@ std::string GetLastErrorMsg();
 namespace Common {
 
 constexpr u32 MakeMagic(char a, char b, char c, char d) {
-    return a | b << 8 | c << 16 | d << 24;
+    return u32(a) | u32(b) << 8 | u32(c) << 16 | u32(d) << 24;
 }
 
 } // namespace Common

src/common/hash.h

@@ -35,41 +35,6 @@ static inline u64 ComputeStructHash64(const T& data) {
     return ComputeHash64(&data, sizeof(data));
 }
 
-/// A helper template that ensures the padding in a struct is initialized by memsetting to 0.
-template <typename T>
-struct HashableStruct {
-    // In addition to being trivially copyable, T must also have a trivial default constructor,
-    // because any member initialization would be overridden by memset
-    static_assert(std::is_trivial_v<T>, "Type passed to HashableStruct must be trivial");
-    /*
-     * We use a union because "implicitly-defined copy/move constructor for a union X copies the
-     * object representation of X." and "implicitly-defined copy assignment operator for a union X
-     * copies the object representation (3.9) of X." = Bytewise copy instead of memberwise copy.
-     * This is important because the padding bytes are included in the hash and comparison between
-     * objects.
-     */
-    union {
-        T state;
-    };
-
-    HashableStruct() {
-        // Memset structure to zero padding bits, so that they will be deterministic when hashing
-        std::memset(&state, 0, sizeof(T));
-    }
-
-    bool operator==(const HashableStruct<T>& o) const {
-        return std::memcmp(&state, &o.state, sizeof(T)) == 0;
-    };
-
-    bool operator!=(const HashableStruct<T>& o) const {
-        return !(*this == o);
-    };
-
-    std::size_t Hash() const {
-        return Common::ComputeStructHash64(state);
-    }
-};
-
 struct PairHash {
     template <class T1, class T2>
     std::size_t operator()(const std::pair<T1, T2>& pair) const noexcept {

src/common/logging/backend.cpp

@@ -272,8 +272,10 @@ const char* GetLogClassName(Class log_class) {
 #undef CLS
 #undef SUB
     case Class::Count:
-        UNREACHABLE();
+        break;
     }
+    UNREACHABLE();
+    return "Invalid";
 }
 
 const char* GetLevelName(Level log_level) {
@@ -288,9 +290,11 @@ const char* GetLevelName(Level log_level) {
         LVL(Error);
         LVL(Critical);
     case Level::Count:
-        UNREACHABLE();
+        break;
     }
 #undef LVL
+    UNREACHABLE();
+    return "Invalid";
 }
 
 void SetGlobalFilter(const Filter& filter) {

src/core/CMakeLists.txt

@@ -522,6 +522,23 @@ add_library(core STATIC
     tools/freezer.h
 )
 
+if (MSVC)
+    target_compile_options(core PRIVATE
+        # 'expression' : signed/unsigned mismatch
+        /we4018
+        # 'argument' : conversion from 'type1' to 'type2', possible loss of data (floating-point)
+        /we4244
+        # 'conversion' : conversion from 'type1' to 'type2', signed/unsigned mismatch
+        /we4245
+        # 'operator': conversion from 'type1:field_bits' to 'type2:field_bits', possible loss of data
+        /we4254
+        # 'var' : conversion from 'size_t' to 'type', possible loss of data
+        /we4267
+        # 'context' : truncation from 'type1' to 'type2'
+        /we4305
+    )
+endif()
+
 create_target_directory_groups(core)
 
 target_link_libraries(core PUBLIC common PRIVATE audio_core video_core)

src/core/arm/dynarmic/arm_dynarmic.cpp

@@ -67,7 +67,7 @@ public:
         ARM_Interface::ThreadContext ctx;
         parent.SaveContext(ctx);
         parent.inner_unicorn.LoadContext(ctx);
-        parent.inner_unicorn.ExecuteInstructions(static_cast<int>(num_instructions));
+        parent.inner_unicorn.ExecuteInstructions(num_instructions);
         parent.inner_unicorn.SaveContext(ctx);
         parent.LoadContext(ctx);
         num_interpreted_instructions += num_instructions;

src/core/arm/unicorn/arm_unicorn.cpp

@@ -67,10 +67,11 @@ ARM_Unicorn::ARM_Unicorn(System& system) : system{system} {
     CHECKED(uc_reg_write(uc, UC_ARM64_REG_CPACR_EL1, &fpv));
 
     uc_hook hook{};
-    CHECKED(uc_hook_add(uc, &hook, UC_HOOK_INTR, (void*)InterruptHook, this, 0, -1));
-    CHECKED(uc_hook_add(uc, &hook, UC_HOOK_MEM_INVALID, (void*)UnmappedMemoryHook, &system, 0, -1));
+    CHECKED(uc_hook_add(uc, &hook, UC_HOOK_INTR, (void*)InterruptHook, this, 0, UINT64_MAX));
+    CHECKED(uc_hook_add(uc, &hook, UC_HOOK_MEM_INVALID, (void*)UnmappedMemoryHook, &system, 0,
+                        UINT64_MAX));
     if (GDBStub::IsServerEnabled()) {
-        CHECKED(uc_hook_add(uc, &hook, UC_HOOK_CODE, (void*)CodeHook, this, 0, -1));
+        CHECKED(uc_hook_add(uc, &hook, UC_HOOK_CODE, (void*)CodeHook, this, 0, UINT64_MAX));
         last_bkpt_hit = false;
     }
 }
@@ -154,9 +155,10 @@ void ARM_Unicorn::SetTPIDR_EL0(u64 value) {
 
 void ARM_Unicorn::Run() {
     if (GDBStub::IsServerEnabled()) {
-        ExecuteInstructions(std::max(4000000, 0));
+        ExecuteInstructions(std::max(4000000U, 0U));
     } else {
-        ExecuteInstructions(std::max(system.CoreTiming().GetDowncount(), s64{0}));
+        ExecuteInstructions(
+            std::max(std::size_t(system.CoreTiming().GetDowncount()), std::size_t{0}));
     }
 }
 
@@ -166,7 +168,7 @@ void ARM_Unicorn::Step() {
 
 MICROPROFILE_DEFINE(ARM_Jit_Unicorn, "ARM JIT", "Unicorn", MP_RGB(255, 64, 64));
 
-void ARM_Unicorn::ExecuteInstructions(int num_instructions) {
+void ARM_Unicorn::ExecuteInstructions(std::size_t num_instructions) {
     MICROPROFILE_SCOPE(ARM_Jit_Unicorn);
     CHECKED(uc_emu_start(uc, GetPC(), 1ULL << 63, 0, num_instructions));
     system.CoreTiming().AddTicks(num_instructions);

src/core/arm/unicorn/arm_unicorn.h

@@ -34,7 +34,7 @@ public:
     void LoadContext(const ThreadContext& ctx) override;
     void PrepareReschedule() override;
     void ClearExclusiveState() override;
-    void ExecuteInstructions(int num_instructions);
+    void ExecuteInstructions(std::size_t num_instructions);
     void Run() override;
     void Step() override;
     void ClearInstructionCache() override;

src/core/crypto/key_manager.cpp

@@ -22,6 +22,7 @@
 #include "common/file_util.h"
 #include "common/hex_util.h"
 #include "common/logging/log.h"
+#include "common/string_util.h"
 #include "core/core.h"
 #include "core/crypto/aes_util.h"
 #include "core/crypto/key_manager.h"
@@ -378,8 +379,9 @@ std::vector<Ticket> GetTicketblob(const FileUtil::IOFile& ticket_save) {
 template <size_t size>
 static std::array<u8, size> operator^(const std::array<u8, size>& lhs,
                                       const std::array<u8, size>& rhs) {
-    std::array<u8, size> out{};
-    std::transform(lhs.begin(), lhs.end(), rhs.begin(), out.begin(), std::bit_xor<>());
+    std::array<u8, size> out;
+    std::transform(lhs.begin(), lhs.end(), rhs.begin(), out.begin(),
+                   [](u8 lhs, u8 rhs) { return u8(lhs ^ rhs); });
     return out;
 }
 
@@ -396,7 +398,7 @@ static std::array<u8, target_size> MGF1(const std::array<u8, in_size>& seed) {
     while (out.size() < target_size) {
         out.resize(out.size() + 0x20);
         seed_exp[in_size + 3] = static_cast<u8>(i);
-        mbedtls_sha256(seed_exp.data(), seed_exp.size(), out.data() + out.size() - 0x20, 0);
+        mbedtls_sha256_ret(seed_exp.data(), seed_exp.size(), out.data() + out.size() - 0x20, 0);
         ++i;
     }
 
@@ -538,7 +540,7 @@ void KeyManager::LoadFromFile(const std::string& filename, bool is_title_keys) {
             Key128 key = Common::HexStringToArray<16>(out[1]);
             s128_keys[{S128KeyType::Titlekey, rights_id[1], rights_id[0]}] = key;
         } else {
-            std::transform(out[0].begin(), out[0].end(), out[0].begin(), ::tolower);
+            out[0] = Common::ToLower(out[0]);
             if (s128_file_id.find(out[0]) != s128_file_id.end()) {
                 const auto index = s128_file_id.at(out[0]);
                 Key128 key = Common::HexStringToArray<16>(out[1]);
@@ -668,23 +670,27 @@ void KeyManager::WriteKeyToFile(KeyCategory category, std::string_view keyname,
                                 const std::array<u8, Size>& key) {
     const std::string yuzu_keys_dir = FileUtil::GetUserPath(FileUtil::UserPath::KeysDir);
     std::string filename = "title.keys_autogenerated";
-    if (category == KeyCategory::Standard)
+    if (category == KeyCategory::Standard) {
         filename = dev_mode ? "dev.keys_autogenerated" : "prod.keys_autogenerated";
-    else if (category == KeyCategory::Console)
+    } else if (category == KeyCategory::Console) {
         filename = "console.keys_autogenerated";
-    const auto add_info_text = !FileUtil::Exists(yuzu_keys_dir + DIR_SEP + filename);
-    FileUtil::CreateFullPath(yuzu_keys_dir + DIR_SEP + filename);
-    std::ofstream file(yuzu_keys_dir + DIR_SEP + filename, std::ios::app);
-    if (!file.is_open())
-        return;
-    if (add_info_text) {
-        file
-            << "# This file is autogenerated by Yuzu\n"
-            << "# It serves to store keys that were automatically generated from the normal keys\n"
-            << "# If you are experiencing issues involving keys, it may help to delete this file\n";
     }
 
-    file << fmt::format("\n{} = {}", keyname, Common::HexToString(key));
+    const auto path = yuzu_keys_dir + DIR_SEP + filename;
+    const auto add_info_text = !FileUtil::Exists(path);
+    FileUtil::CreateFullPath(path);
+    FileUtil::IOFile file{path, "a"};
+    if (!file.IsOpen()) {
+        return;
+    }
+    if (add_info_text) {
+        file.WriteString(
+            "# This file is autogenerated by Yuzu\n"
+            "# It serves to store keys that were automatically generated from the normal keys\n"
+            "# If you are experiencing issues involving keys, it may help to delete this file\n");
+    }
+
+    file.WriteString(fmt::format("\n{} = {}", keyname, Common::HexToString(key)));
     AttemptLoadKeyFile(yuzu_keys_dir, yuzu_keys_dir, filename, category == KeyCategory::Title);
 }
 
@@ -944,12 +950,10 @@ void KeyManager::DeriveETicket(PartitionDataManager& data) {
         return;
     }
 
-    Key128 rsa_oaep_kek{};
-    std::transform(seed3.begin(), seed3.end(), mask0.begin(), rsa_oaep_kek.begin(),
-                   std::bit_xor<>());
-
-    if (rsa_oaep_kek == Key128{})
+    const Key128 rsa_oaep_kek = seed3 ^ mask0;
+    if (rsa_oaep_kek == Key128{}) {
         return;
+    }
 
     SetKey(S128KeyType::Source, rsa_oaep_kek,
            static_cast<u64>(SourceKeyType::RSAOaepKekGeneration));

src/core/crypto/partition_data_manager.cpp

@@ -161,7 +161,7 @@ std::array<u8, key_size> FindKeyFromHex(const std::vector<u8>& binary,
 
     std::array<u8, 0x20> temp{};
     for (size_t i = 0; i < binary.size() - key_size; ++i) {
-        mbedtls_sha256(binary.data() + i, key_size, temp.data(), 0);
+        mbedtls_sha256_ret(binary.data() + i, key_size, temp.data(), 0);
 
         if (temp != hash)
             continue;
@@ -189,7 +189,7 @@ static std::array<Key128, 0x20> FindEncryptedMasterKeyFromHex(const std::vector<
     AESCipher<Key128> cipher(key, Mode::ECB);
     for (size_t i = 0; i < binary.size() - 0x10; ++i) {
         cipher.Transcode(binary.data() + i, dec_temp.size(), dec_temp.data(), Op::Decrypt);
-        mbedtls_sha256(dec_temp.data(), dec_temp.size(), temp.data(), 0);
+        mbedtls_sha256_ret(dec_temp.data(), dec_temp.size(), temp.data(), 0);
 
         for (size_t k = 0; k < out.size(); ++k) {
             if (temp == master_key_hashes[k]) {
@@ -204,11 +204,12 @@ static std::array<Key128, 0x20> FindEncryptedMasterKeyFromHex(const std::vector<
 
 FileSys::VirtualFile FindFileInDirWithNames(const FileSys::VirtualDir& dir,
                                             const std::string& name) {
-    auto upper = name;
-    std::transform(upper.begin(), upper.end(), upper.begin(), [](u8 c) { return std::toupper(c); });
+    const auto upper = Common::ToUpper(name);
+
     for (const auto& fname : {name, name + ".bin", upper, upper + ".BIN"}) {
-        if (dir->GetFile(fname) != nullptr)
+        if (dir->GetFile(fname) != nullptr) {
             return dir->GetFile(fname);
+        }
     }
 
     return nullptr;

src/core/file_sys/kernel_executable.cpp

@@ -147,7 +147,7 @@ std::vector<u32> KIP::GetKernelCapabilities() const {
 }
 
 s32 KIP::GetMainThreadPriority() const {
-    return header.main_thread_priority;
+    return static_cast<s32>(header.main_thread_priority);
 }
 
 u32 KIP::GetMainThreadStackSize() const {

src/core/file_sys/program_metadata.cpp

@@ -52,14 +52,14 @@ Loader::ResultStatus ProgramMetadata::Load(VirtualFile file) {
 }
 
 void ProgramMetadata::LoadManual(bool is_64_bit, ProgramAddressSpaceType address_space,
-                                 u8 main_thread_prio, u8 main_thread_core,
+                                 s32 main_thread_prio, u32 main_thread_core,
                                  u32 main_thread_stack_size, u64 title_id,
                                  u64 filesystem_permissions,
                                  KernelCapabilityDescriptors capabilities) {
     npdm_header.has_64_bit_instructions.Assign(is_64_bit);
     npdm_header.address_space_type.Assign(address_space);
-    npdm_header.main_thread_priority = main_thread_prio;
-    npdm_header.main_thread_cpu = main_thread_core;
+    npdm_header.main_thread_priority = static_cast<u8>(main_thread_prio);
+    npdm_header.main_thread_cpu = static_cast<u8>(main_thread_core);
     npdm_header.main_stack_size = main_thread_stack_size;
     aci_header.title_id = title_id;
     aci_file_access.permissions = filesystem_permissions;

src/core/file_sys/program_metadata.h

@@ -47,8 +47,8 @@ public:
     Loader::ResultStatus Load(VirtualFile file);
 
     // Load from parameters instead of NPDM file, used for KIP
-    void LoadManual(bool is_64_bit, ProgramAddressSpaceType address_space, u8 main_thread_prio,
-                    u8 main_thread_core, u32 main_thread_stack_size, u64 title_id,
+    void LoadManual(bool is_64_bit, ProgramAddressSpaceType address_space, s32 main_thread_prio,
+                    u32 main_thread_core, u32 main_thread_stack_size, u64 title_id,
                     u64 filesystem_permissions, KernelCapabilityDescriptors capabilities);
 
     bool Is64BitProgram() const;

src/core/file_sys/registered_cache.cpp

@@ -62,7 +62,7 @@ static std::string GetRelativePathFromNcaID(const std::array<u8, 16>& nca_id, bo
                            Common::HexToString(nca_id, second_hex_upper));
 
     Core::Crypto::SHA256Hash hash{};
-    mbedtls_sha256(nca_id.data(), nca_id.size(), hash.data(), 0);
+    mbedtls_sha256_ret(nca_id.data(), nca_id.size(), hash.data(), 0);
     return fmt::format(cnmt_suffix ? "/000000{:02X}/{}.cnmt.nca" : "/000000{:02X}/{}.nca", hash[0],
                        Common::HexToString(nca_id, second_hex_upper));
 }
@@ -141,7 +141,7 @@ bool PlaceholderCache::Create(const NcaID& id, u64 size) const {
     }
 
     Core::Crypto::SHA256Hash hash{};
-    mbedtls_sha256(id.data(), id.size(), hash.data(), 0);
+    mbedtls_sha256_ret(id.data(), id.size(), hash.data(), 0);
     const auto dirname = fmt::format("000000{:02X}", hash[0]);
 
     const auto dir2 = GetOrCreateDirectoryRelative(dir, dirname);
@@ -165,7 +165,7 @@ bool PlaceholderCache::Delete(const NcaID& id) const {
     }
 
     Core::Crypto::SHA256Hash hash{};
-    mbedtls_sha256(id.data(), id.size(), hash.data(), 0);
+    mbedtls_sha256_ret(id.data(), id.size(), hash.data(), 0);
     const auto dirname = fmt::format("000000{:02X}", hash[0]);
 
     const auto dir2 = GetOrCreateDirectoryRelative(dir, dirname);
@@ -603,7 +603,7 @@ InstallResult RegisteredCache::InstallEntry(const NCA& nca, TitleType type,
     OptionalHeader opt_header{0, 0};
     ContentRecord c_rec{{}, {}, {}, GetCRTypeFromNCAType(nca.GetType()), {}};
     const auto& data = nca.GetBaseFile()->ReadBytes(0x100000);
-    mbedtls_sha256(data.data(), data.size(), c_rec.hash.data(), 0);
+    mbedtls_sha256_ret(data.data(), data.size(), c_rec.hash.data(), 0);
     memcpy(&c_rec.nca_id, &c_rec.hash, 16);
     const CNMT new_cnmt(header, opt_header, {c_rec}, {});
     if (!RawInstallYuzuMeta(new_cnmt))
@@ -626,7 +626,7 @@ InstallResult RegisteredCache::RawInstallNCA(const NCA& nca, const VfsCopyFuncti
         id = *override_id;
     } else {
         const auto& data = in->ReadBytes(0x100000);
-        mbedtls_sha256(data.data(), data.size(), hash.data(), 0);
+        mbedtls_sha256_ret(data.data(), data.size(), hash.data(), 0);
         memcpy(id.data(), hash.data(), 16);
     }
 

src/core/file_sys/romfs_factory.cpp

@@ -71,12 +71,12 @@ ResultVal<VirtualFile> RomFSFactory::Open(u64 title_id, StorageId storage,
 
     if (res == nullptr) {
         // TODO(DarkLordZach): Find the right error code to use here
-        return ResultCode(-1);
+        return RESULT_UNKNOWN;
     }
     const auto romfs = res->GetRomFS();
     if (romfs == nullptr) {
         // TODO(DarkLordZach): Find the right error code to use here
-        return ResultCode(-1);
+        return RESULT_UNKNOWN;
     }
     return MakeResult<VirtualFile>(romfs);
 }

src/core/file_sys/savedata_factory.cpp

@@ -90,7 +90,7 @@ ResultVal<VirtualDir> SaveDataFactory::Create(SaveDataSpaceId space,
     // Return an error if the save data doesn't actually exist.
     if (out == nullptr) {
         // TODO(DarkLordZach): Find out correct error code.
-        return ResultCode(-1);
+        return RESULT_UNKNOWN;
     }
 
     return MakeResult<VirtualDir>(std::move(out));
@@ -111,7 +111,7 @@ ResultVal<VirtualDir> SaveDataFactory::Open(SaveDataSpaceId space,
     // Return an error if the save data doesn't actually exist.
     if (out == nullptr) {
         // TODO(Subv): Find out correct error code.
-        return ResultCode(-1);
+        return RESULT_UNKNOWN;
     }
 
     return MakeResult<VirtualDir>(std::move(out));

src/core/file_sys/vfs_libzip.cpp

@@ -27,7 +27,7 @@ VirtualDir ExtractZIP(VirtualFile file) {
 
     std::shared_ptr<VectorVfsDirectory> out = std::make_shared<VectorVfsDirectory>();
 
-    const auto num_entries = zip_get_num_entries(zip.get(), 0);
+    const auto num_entries = static_cast<std::size_t>(zip_get_num_entries(zip.get(), 0));
 
     zip_stat_t stat{};
     zip_stat_init(&stat);

src/core/file_sys/xts_archive.cpp

@@ -7,12 +7,13 @@
 #include <cstring>
 #include <regex>
 #include <string>
+
 #include <mbedtls/md.h>
 #include <mbedtls/sha256.h>
-#include "common/assert.h"
+
 #include "common/file_util.h"
 #include "common/hex_util.h"
-#include "common/logging/log.h"
+#include "common/string_util.h"
 #include "core/crypto/aes_util.h"
 #include "core/crypto/xts_encryption_layer.h"
 #include "core/file_sys/partition_filesystem.h"
@@ -53,18 +54,15 @@ NAX::NAX(VirtualFile file_) : header(std::make_unique<NAXHeader>()), file(std::m
         return;
     }
 
-    std::string two_dir = match[1];
-    std::string nca_id = match[2];
-    std::transform(two_dir.begin(), two_dir.end(), two_dir.begin(), ::toupper);
-    std::transform(nca_id.begin(), nca_id.end(), nca_id.begin(), ::tolower);
-
+    const std::string two_dir = Common::ToUpper(match[1]);
+    const std::string nca_id = Common::ToLower(match[2]);
     status = Parse(fmt::format("/registered/{}/{}.nca", two_dir, nca_id));
 }
 
 NAX::NAX(VirtualFile file_, std::array<u8, 0x10> nca_id)
     : header(std::make_unique<NAXHeader>()), file(std::move(file_)) {
     Core::Crypto::SHA256Hash hash{};
-    mbedtls_sha256(nca_id.data(), nca_id.size(), hash.data(), 0);
+    mbedtls_sha256_ret(nca_id.data(), nca_id.size(), hash.data(), 0);
     status = Parse(fmt::format("/registered/000000{:02X}/{}.nca", hash[0],
                                Common::HexToString(nca_id, false)));
 }
@@ -93,8 +91,7 @@ Loader::ResultStatus NAX::Parse(std::string_view path) {
     std::size_t i = 0;
     for (; i < sd_keys.size(); ++i) {
         std::array<Core::Crypto::Key128, 2> nax_keys{};
-        if (!CalculateHMAC256(nax_keys.data(), sd_keys[i].data(), 0x10, std::string(path).c_str(),
-                              path.size())) {
+        if (!CalculateHMAC256(nax_keys.data(), sd_keys[i].data(), 0x10, path.data(), path.size())) {
             return Loader::ResultStatus::ErrorNAXKeyHMACFailed;
         }
 

src/core/gdbstub/gdbstub.cpp

@@ -468,7 +468,8 @@ static u8 ReadByte() {
 
 /// Calculate the checksum of the current command buffer.
 static u8 CalculateChecksum(const u8* buffer, std::size_t length) {
-    return static_cast<u8>(std::accumulate(buffer, buffer + length, 0, std::plus<u8>()));
+    return static_cast<u8>(std::accumulate(buffer, buffer + length, u8{0},
+                                           [](u8 lhs, u8 rhs) { return u8(lhs + rhs); }));
 }
 
 /**

src/core/hle/kernel/kernel.cpp

@@ -64,8 +64,11 @@ static void ThreadWakeupCallback(u64 thread_handle, [[maybe_unused]] s64 cycles_
     } else if (thread->GetStatus() == ThreadStatus::WaitMutex ||
                thread->GetStatus() == ThreadStatus::WaitCondVar) {
         thread->SetMutexWaitAddress(0);
-        thread->SetCondVarWaitAddress(0);
         thread->SetWaitHandle(0);
+        if (thread->GetStatus() == ThreadStatus::WaitCondVar) {
+            thread->GetOwnerProcess()->RemoveConditionVariableThread(thread);
+            thread->SetCondVarWaitAddress(0);
+        }
 
         auto* const lock_owner = thread->GetLockOwner();
         // Threads waking up by timeout from WaitProcessWideKey do not perform priority inheritance

src/core/hle/kernel/mutex.cpp

@@ -7,6 +7,7 @@
 
 #include "common/assert.h"
 #include "core/core.h"
+#include "core/core_cpu.h"
 #include "core/hle/kernel/errors.h"
 #include "core/hle/kernel/handle_table.h"
 #include "core/hle/kernel/kernel.h"
@@ -78,7 +79,7 @@ ResultCode Mutex::TryAcquire(VAddr address, Handle holding_thread_handle,
     // thread.
     ASSERT(requesting_thread == current_thread);
 
-    const u32 addr_value = Memory::Read32(address);
+    u32 addr_value = Memory::Read32(address);
 
     // If the mutex isn't being held, just return success.
     if (addr_value != (holding_thread_handle | Mutex::MutexHasWaitersFlag)) {
@@ -89,6 +90,20 @@ ResultCode Mutex::TryAcquire(VAddr address, Handle holding_thread_handle,
         return ERR_INVALID_HANDLE;
     }
 
+    // This a workaround where an unknown bug writes the mutex value to give ownership to a cond var
+    // waiting thread.
+    if (holding_thread->GetStatus() == ThreadStatus::WaitCondVar) {
+        if (holding_thread->GetMutexWaitAddress() == address) {
+            Release(address, holding_thread.get());
+            addr_value = Memory::Read32(address);
+            if (addr_value == 0)
+                return RESULT_SUCCESS;
+            else {
+                holding_thread = handle_table.Get<Thread>(addr_value & Mutex::MutexOwnerMask);
+            }
+        }
+    }
+
     // Wait until the mutex is released
     current_thread->SetMutexWaitAddress(address);
     current_thread->SetWaitHandle(requesting_thread_handle);
@@ -104,14 +119,13 @@ ResultCode Mutex::TryAcquire(VAddr address, Handle holding_thread_handle,
     return RESULT_SUCCESS;
 }
 
-ResultCode Mutex::Release(VAddr address) {
+ResultCode Mutex::Release(VAddr address, Thread* holding_thread) {
     // The mutex address must be 4-byte aligned
     if ((address % sizeof(u32)) != 0) {
         return ERR_INVALID_ADDRESS;
     }
 
-    auto* const current_thread = system.CurrentScheduler().GetCurrentThread();
-    auto [thread, num_waiters] = GetHighestPriorityMutexWaitingThread(current_thread, address);
+    auto [thread, num_waiters] = GetHighestPriorityMutexWaitingThread(holding_thread, address);
 
     // There are no more threads waiting for the mutex, release it completely.
     if (thread == nullptr) {
@@ -120,7 +134,7 @@ ResultCode Mutex::Release(VAddr address) {
     }
 
     // Transfer the ownership of the mutex from the previous owner to the new one.
-    TransferMutexOwnership(address, current_thread, thread);
+    TransferMutexOwnership(address, holding_thread, thread);
 
     u32 mutex_value = thread->GetWaitHandle();
 
@@ -141,7 +155,10 @@ ResultCode Mutex::Release(VAddr address) {
     thread->SetWaitHandle(0);
     thread->SetWaitSynchronizationResult(RESULT_SUCCESS);
 
-    system.PrepareReschedule();
+    if (thread->GetProcessorID() >= 0)
+        system.CpuCore(thread->GetProcessorID()).PrepareReschedule();
+    if (holding_thread->GetProcessorID() >= 0)
+        system.CpuCore(holding_thread->GetProcessorID()).PrepareReschedule();
 
     return RESULT_SUCCESS;
 }

src/core/hle/kernel/mutex.h

@@ -29,7 +29,7 @@ public:
                           Handle requesting_thread_handle);
 
     /// Releases the mutex at the specified address.
-    ResultCode Release(VAddr address);
+    ResultCode Release(VAddr address, Thread* holding_thread);
 
 private:
     Core::System& system;

src/core/hle/kernel/process.cpp

@@ -142,6 +142,48 @@ u64 Process::GetTotalPhysicalMemoryUsedWithoutSystemResource() const {
     return GetTotalPhysicalMemoryUsed() - GetSystemResourceUsage();
 }
 
+void Process::InsertConditionVariableThread(SharedPtr<Thread> thread) {
+    VAddr cond_var_addr = thread->GetCondVarWaitAddress();
+    std::list<SharedPtr<Thread>>& thread_list = cond_var_threads[cond_var_addr];
+    auto it = thread_list.begin();
+    while (it != thread_list.end()) {
+        const SharedPtr<Thread> current_thread = *it;
+        if (current_thread->GetPriority() > thread->GetPriority()) {
+            thread_list.insert(it, thread);
+            return;
+        }
+        ++it;
+    }
+    thread_list.push_back(thread);
+}
+
+void Process::RemoveConditionVariableThread(SharedPtr<Thread> thread) {
+    VAddr cond_var_addr = thread->GetCondVarWaitAddress();
+    std::list<SharedPtr<Thread>>& thread_list = cond_var_threads[cond_var_addr];
+    auto it = thread_list.begin();
+    while (it != thread_list.end()) {
+        const SharedPtr<Thread> current_thread = *it;
+        if (current_thread.get() == thread.get()) {
+            thread_list.erase(it);
+            return;
+        }
+        ++it;
+    }
+    UNREACHABLE();
+}
+
+std::vector<SharedPtr<Thread>> Process::GetConditionVariableThreads(const VAddr cond_var_addr) {
+    std::vector<SharedPtr<Thread>> result{};
+    std::list<SharedPtr<Thread>>& thread_list = cond_var_threads[cond_var_addr];
+    auto it = thread_list.begin();
+    while (it != thread_list.end()) {
+        SharedPtr<Thread> current_thread = *it;
+        result.push_back(current_thread);
+        ++it;
+    }
+    return result;
+}
+
 void Process::RegisterThread(const Thread* thread) {
     thread_list.push_back(thread);
 }

src/core/hle/kernel/process.h

@@ -8,6 +8,7 @@
 #include <cstddef>
 #include <list>
 #include <string>
+#include <unordered_map>
 #include <vector>
 #include "common/common_types.h"
 #include "core/hle/kernel/address_arbiter.h"
@@ -232,6 +233,15 @@ public:
         return thread_list;
     }
 
+    /// Insert a thread into the condition variable wait container
+    void InsertConditionVariableThread(SharedPtr<Thread> thread);
+
+    /// Remove a thread from the condition variable wait container
+    void RemoveConditionVariableThread(SharedPtr<Thread> thread);
+
+    /// Obtain all condition variable threads waiting for some address
+    std::vector<SharedPtr<Thread>> GetConditionVariableThreads(VAddr cond_var_addr);
+
     /// Registers a thread as being created under this process,
     /// adding it to this process' thread list.
     void RegisterThread(const Thread* thread);
@@ -375,6 +385,9 @@ private:
     /// List of threads that are running with this process as their owner.
     std::list<const Thread*> thread_list;
 
+    /// List of threads waiting for a condition variable
+    std::unordered_map<VAddr, std::list<SharedPtr<Thread>>> cond_var_threads;
+
     /// System context
     Core::System& system;
 

src/core/hle/kernel/scheduler.cpp

@@ -35,12 +35,12 @@ void GlobalScheduler::RemoveThread(const Thread* thread) {
                       thread_list.end());
 }
 
-void GlobalScheduler::UnloadThread(s32 core) {
+void GlobalScheduler::UnloadThread(std::size_t core) {
     Scheduler& sched = system.Scheduler(core);
     sched.UnloadThread();
 }
 
-void GlobalScheduler::SelectThread(u32 core) {
+void GlobalScheduler::SelectThread(std::size_t core) {
     const auto update_thread = [](Thread* thread, Scheduler& sched) {
         if (thread != sched.selected_thread) {
             if (thread == nullptr) {
@@ -77,9 +77,9 @@ void GlobalScheduler::SelectThread(u32 core) {
     // if we got a suggested thread, select it, else do a second pass.
     if (winner && winner->GetPriority() > 2) {
         if (winner->IsRunning()) {
-            UnloadThread(winner->GetProcessorID());
+            UnloadThread(static_cast<u32>(winner->GetProcessorID()));
         }
-        TransferToCore(winner->GetPriority(), core, winner);
+        TransferToCore(winner->GetPriority(), static_cast<s32>(core), winner);
         update_thread(winner, sched);
         return;
     }
@@ -91,9 +91,9 @@ void GlobalScheduler::SelectThread(u32 core) {
             Thread* thread_on_core = scheduled_queue[src_core].front();
             Thread* to_change = *it;
             if (thread_on_core->IsRunning() || to_change->IsRunning()) {
-                UnloadThread(src_core);
+                UnloadThread(static_cast<u32>(src_core));
             }
-            TransferToCore(thread_on_core->GetPriority(), core, thread_on_core);
+            TransferToCore(thread_on_core->GetPriority(), static_cast<s32>(core), thread_on_core);
             current_thread = thread_on_core;
             break;
         }
@@ -154,9 +154,9 @@ bool GlobalScheduler::YieldThreadAndBalanceLoad(Thread* yielding_thread) {
     if (winner != nullptr) {
         if (winner != yielding_thread) {
             if (winner->IsRunning()) {
-                UnloadThread(winner->GetProcessorID());
+                UnloadThread(static_cast<u32>(winner->GetProcessorID()));
             }
-            TransferToCore(winner->GetPriority(), core_id, winner);
+            TransferToCore(winner->GetPriority(), s32(core_id), winner);
         }
     } else {
         winner = next_thread;
@@ -196,9 +196,9 @@ bool GlobalScheduler::YieldThreadAndWaitForLoadBalancing(Thread* yielding_thread
         if (winner != nullptr) {
             if (winner != yielding_thread) {
                 if (winner->IsRunning()) {
-                    UnloadThread(winner->GetProcessorID());
+                    UnloadThread(static_cast<u32>(winner->GetProcessorID()));
                 }
-                TransferToCore(winner->GetPriority(), core_id, winner);
+                TransferToCore(winner->GetPriority(), static_cast<s32>(core_id), winner);
             }
         } else {
             winner = yielding_thread;
@@ -248,7 +248,7 @@ void GlobalScheduler::PreemptThreads() {
 
         if (winner != nullptr) {
             if (winner->IsRunning()) {
-                UnloadThread(winner->GetProcessorID());
+                UnloadThread(static_cast<u32>(winner->GetProcessorID()));
             }
             TransferToCore(winner->GetPriority(), s32(core_id), winner);
             current_thread =
@@ -281,7 +281,7 @@ void GlobalScheduler::PreemptThreads() {
 
             if (winner != nullptr) {
                 if (winner->IsRunning()) {
-                    UnloadThread(winner->GetProcessorID());
+                    UnloadThread(static_cast<u32>(winner->GetProcessorID()));
                 }
                 TransferToCore(winner->GetPriority(), s32(core_id), winner);
                 current_thread = winner;
@@ -292,30 +292,30 @@ void GlobalScheduler::PreemptThreads() {
     }
 }
 
-void GlobalScheduler::Suggest(u32 priority, u32 core, Thread* thread) {
+void GlobalScheduler::Suggest(u32 priority, std::size_t core, Thread* thread) {
     suggested_queue[core].add(thread, priority);
 }
 
-void GlobalScheduler::Unsuggest(u32 priority, u32 core, Thread* thread) {
+void GlobalScheduler::Unsuggest(u32 priority, std::size_t core, Thread* thread) {
     suggested_queue[core].remove(thread, priority);
 }
 
-void GlobalScheduler::Schedule(u32 priority, u32 core, Thread* thread) {
+void GlobalScheduler::Schedule(u32 priority, std::size_t core, Thread* thread) {
     ASSERT_MSG(thread->GetProcessorID() == s32(core), "Thread must be assigned to this core.");
     scheduled_queue[core].add(thread, priority);
 }
 
-void GlobalScheduler::SchedulePrepend(u32 priority, u32 core, Thread* thread) {
+void GlobalScheduler::SchedulePrepend(u32 priority, std::size_t core, Thread* thread) {
     ASSERT_MSG(thread->GetProcessorID() == s32(core), "Thread must be assigned to this core.");
     scheduled_queue[core].add(thread, priority, false);
 }
 
-void GlobalScheduler::Reschedule(u32 priority, u32 core, Thread* thread) {
+void GlobalScheduler::Reschedule(u32 priority, std::size_t core, Thread* thread) {
     scheduled_queue[core].remove(thread, priority);
     scheduled_queue[core].add(thread, priority);
 }
 
-void GlobalScheduler::Unschedule(u32 priority, u32 core, Thread* thread) {
+void GlobalScheduler::Unschedule(u32 priority, std::size_t core, Thread* thread) {
     scheduled_queue[core].remove(thread, priority);
 }
 
@@ -327,14 +327,14 @@ void GlobalScheduler::TransferToCore(u32 priority, s32 destination_core, Thread*
     }
     thread->SetProcessorID(destination_core);
     if (source_core >= 0) {
-        Unschedule(priority, source_core, thread);
+        Unschedule(priority, static_cast<u32>(source_core), thread);
     }
     if (destination_core >= 0) {
-        Unsuggest(priority, destination_core, thread);
-        Schedule(priority, destination_core, thread);
+        Unsuggest(priority, static_cast<u32>(destination_core), thread);
+        Schedule(priority, static_cast<u32>(destination_core), thread);
     }
     if (source_core >= 0) {
-        Suggest(priority, source_core, thread);
+        Suggest(priority, static_cast<u32>(source_core), thread);
     }
 }
 
@@ -357,7 +357,7 @@ void GlobalScheduler::Shutdown() {
     thread_list.clear();
 }
 
-Scheduler::Scheduler(Core::System& system, Core::ARM_Interface& cpu_core, u32 core_id)
+Scheduler::Scheduler(Core::System& system, Core::ARM_Interface& cpu_core, std::size_t core_id)
     : system(system), cpu_core(cpu_core), core_id(core_id) {}
 
 Scheduler::~Scheduler() = default;

src/core/hle/kernel/scheduler.h

@@ -42,41 +42,34 @@ public:
      * Add a thread to the suggested queue of a cpu core. Suggested threads may be
      * picked if no thread is scheduled to run on the core.
      */
-    void Suggest(u32 priority, u32 core, Thread* thread);
+    void Suggest(u32 priority, std::size_t core, Thread* thread);
 
     /**
      * Remove a thread to the suggested queue of a cpu core. Suggested threads may be
      * picked if no thread is scheduled to run on the core.
      */
-    void Unsuggest(u32 priority, u32 core, Thread* thread);
+    void Unsuggest(u32 priority, std::size_t core, Thread* thread);
 
     /**
      * Add a thread to the scheduling queue of a cpu core. The thread is added at the
      * back the queue in its priority level.
      */
-    void Schedule(u32 priority, u32 core, Thread* thread);
+    void Schedule(u32 priority, std::size_t core, Thread* thread);
 
     /**
      * Add a thread to the scheduling queue of a cpu core. The thread is added at the
      * front the queue in its priority level.
      */
-    void SchedulePrepend(u32 priority, u32 core, Thread* thread);
+    void SchedulePrepend(u32 priority, std::size_t core, Thread* thread);
 
     /// Reschedule an already scheduled thread based on a new priority
-    void Reschedule(u32 priority, u32 core, Thread* thread);
+    void Reschedule(u32 priority, std::size_t core, Thread* thread);
 
     /// Unschedules a thread.
-    void Unschedule(u32 priority, u32 core, Thread* thread);
-
-    /**
-     * Transfers a thread into an specific core. If the destination_core is -1
-     * it will be unscheduled from its source code and added into its suggested
-     * queue.
-     */
-    void TransferToCore(u32 priority, s32 destination_core, Thread* thread);
+    void Unschedule(u32 priority, std::size_t core, Thread* thread);
 
     /// Selects a core and forces it to unload its current thread's context
-    void UnloadThread(s32 core);
+    void UnloadThread(std::size_t core);
 
     /**
      * Takes care of selecting the new scheduled thread in three steps:
@@ -90,9 +83,9 @@ public:
      * 3. Third is no suggested thread is found, we do a second pass and pick a running
      *    thread in another core and swap it with its current thread.
      */
-    void SelectThread(u32 core);
+    void SelectThread(std::size_t core);
 
-    bool HaveReadyThreads(u32 core_id) const {
+    bool HaveReadyThreads(std::size_t core_id) const {
         return !scheduled_queue[core_id].empty();
     }
 
@@ -145,6 +138,13 @@ public:
     void Shutdown();
 
 private:
+    /**
+     * Transfers a thread into an specific core. If the destination_core is -1
+     * it will be unscheduled from its source code and added into its suggested
+     * queue.
+     */
+    void TransferToCore(u32 priority, s32 destination_core, Thread* thread);
+
     bool AskForReselectionOrMarkRedundant(Thread* current_thread, const Thread* winner);
 
     static constexpr u32 min_regular_priority = 2;
@@ -163,7 +163,7 @@ private:
 
 class Scheduler final {
 public:
-    explicit Scheduler(Core::System& system, Core::ARM_Interface& cpu_core, u32 core_id);
+    explicit Scheduler(Core::System& system, Core::ARM_Interface& cpu_core, std::size_t core_id);
     ~Scheduler();
 
     /// Returns whether there are any threads that are ready to run.
@@ -220,7 +220,7 @@ private:
     Core::ARM_Interface& cpu_core;
     u64 last_context_switch_time = 0;
     u64 idle_selection_count = 0;
-    const u32 core_id;
+    const std::size_t core_id;
 
     bool is_context_switch_pending = false;
 };

src/core/hle/kernel/svc.cpp

@@ -505,6 +505,11 @@ static ResultCode WaitSynchronization(Core::System& system, Handle* index, VAddr
         return RESULT_TIMEOUT;
     }
 
+    if (thread->IsSyncCancelled()) {
+        thread->SetSyncCancelled(false);
+        return ERR_SYNCHRONIZATION_CANCELED;
+    }
+
     for (auto& object : objects) {
         object->AddWaitingThread(thread);
     }
@@ -578,7 +583,8 @@ static ResultCode ArbitrateUnlock(Core::System& system, VAddr mutex_addr) {
     }
 
     auto* const current_process = system.Kernel().CurrentProcess();
-    return current_process->GetMutex().Release(mutex_addr);
+    return current_process->GetMutex().Release(mutex_addr,
+                                               system.CurrentScheduler().GetCurrentThread());
 }
 
 enum class BreakType : u32 {
@@ -1615,17 +1621,20 @@ static ResultCode WaitProcessWideKeyAtomic(Core::System& system, VAddr mutex_add
     SharedPtr<Thread> thread = handle_table.Get<Thread>(thread_handle);
     ASSERT(thread);
 
-    const auto release_result = current_process->GetMutex().Release(mutex_addr);
+    SharedPtr<Thread> current_thread = system.CurrentScheduler().GetCurrentThread();
+
+    const auto release_result =
+        current_process->GetMutex().Release(mutex_addr, current_thread.get());
     if (release_result.IsError()) {
         return release_result;
     }
 
-    SharedPtr<Thread> current_thread = system.CurrentScheduler().GetCurrentThread();
     current_thread->SetCondVarWaitAddress(condition_variable_addr);
     current_thread->SetMutexWaitAddress(mutex_addr);
     current_thread->SetWaitHandle(thread_handle);
     current_thread->SetStatus(ThreadStatus::WaitCondVar);
     current_thread->InvalidateWakeupCallback();
+    current_process->InsertConditionVariableThread(current_thread);
 
     current_thread->WakeAfterDelay(nano_seconds);
 
@@ -1644,38 +1653,23 @@ static ResultCode SignalProcessWideKey(Core::System& system, VAddr condition_var
     ASSERT(condition_variable_addr == Common::AlignDown(condition_variable_addr, 4));
 
     // Retrieve a list of all threads that are waiting for this condition variable.
-    std::vector<SharedPtr<Thread>> waiting_threads;
-    const auto& scheduler = system.GlobalScheduler();
-    const auto& thread_list = scheduler.GetThreadList();
+    auto* const current_process = system.Kernel().CurrentProcess();
+    std::vector<SharedPtr<Thread>> waiting_threads =
+        current_process->GetConditionVariableThreads(condition_variable_addr);
 
-    for (const auto& thread : thread_list) {
-        if (thread->GetCondVarWaitAddress() == condition_variable_addr) {
-            waiting_threads.push_back(thread);
-        }
-    }
-
-    // Sort them by priority, such that the highest priority ones come first.
-    std::sort(waiting_threads.begin(), waiting_threads.end(),
-              [](const SharedPtr<Thread>& lhs, const SharedPtr<Thread>& rhs) {
-                  return lhs->GetPriority() < rhs->GetPriority();
-              });
-
-    // Only process up to 'target' threads, unless 'target' is -1, in which case process
+    // Only process up to 'target' threads, unless 'target' is less equal 0, in which case process
     // them all.
     std::size_t last = waiting_threads.size();
-    if (target != -1)
+    if (target > 0)
         last = std::min(waiting_threads.size(), static_cast<std::size_t>(target));
 
-    // If there are no threads waiting on this condition variable, just exit
-    if (last == 0)
-        return RESULT_SUCCESS;
-
     for (std::size_t index = 0; index < last; ++index) {
         auto& thread = waiting_threads[index];
 
         ASSERT(thread->GetCondVarWaitAddress() == condition_variable_addr);
 
         // liberate Cond Var Thread.
+        current_process->RemoveConditionVariableThread(thread);
         thread->SetCondVarWaitAddress(0);
 
         const std::size_t current_core = system.CurrentCoreIndex();

src/core/hle/kernel/thread.cpp

@@ -77,18 +77,6 @@ void Thread::CancelWakeupTimer() {
                                                              callback_handle);
 }
 
-static std::optional<s32> GetNextProcessorId(u64 mask) {
-    for (s32 index = 0; index < Core::NUM_CPU_CORES; ++index) {
-        if (mask & (1ULL << index)) {
-            if (!Core::System::GetInstance().Scheduler(index).GetCurrentThread()) {
-                // Core is enabled and not running any threads, use this one
-                return index;
-            }
-        }
-    }
-    return {};
-}
-
 void Thread::ResumeFromWait() {
     ASSERT_MSG(wait_objects.empty(), "Thread is waking up while waiting for objects");
 
@@ -132,8 +120,11 @@ void Thread::ResumeFromWait() {
 }
 
 void Thread::CancelWait() {
-    ASSERT(GetStatus() == ThreadStatus::WaitSynch);
-    ClearWaitObjects();
+    if (GetSchedulingStatus() != ThreadSchedStatus::Paused) {
+        is_sync_cancelled = true;
+        return;
+    }
+    is_sync_cancelled = false;
     SetWaitSynchronizationResult(ERR_SYNCHRONIZATION_CANCELED);
     ResumeFromWait();
 }
@@ -173,7 +164,7 @@ ResultVal<SharedPtr<Thread>> Thread::Create(KernelCore& kernel, std::string name
     if (!Memory::IsValidVirtualAddress(owner_process, entry_point)) {
         LOG_ERROR(Kernel_SVC, "(name={}): invalid entry {:016X}", name, entry_point);
         // TODO (bunnei): Find the correct error code to use here
-        return ResultCode(-1);
+        return RESULT_UNKNOWN;
     }
 
     auto& system = Core::System::GetInstance();
@@ -318,8 +309,16 @@ void Thread::UpdatePriority() {
         return;
     }
 
+    if (GetStatus() == ThreadStatus::WaitCondVar) {
+        owner_process->RemoveConditionVariableThread(this);
+    }
+
     SetCurrentPriority(new_priority);
 
+    if (GetStatus() == ThreadStatus::WaitCondVar) {
+        owner_process->InsertConditionVariableThread(this);
+    }
+
     if (!lock_owner) {
         return;
     }
@@ -401,7 +400,7 @@ void Thread::SetCurrentPriority(u32 new_priority) {
 
 ResultCode Thread::SetCoreAndAffinityMask(s32 new_core, u64 new_affinity_mask) {
     const auto HighestSetCore = [](u64 mask, u32 max_cores) {
-        for (s32 core = max_cores - 1; core >= 0; core--) {
+        for (s32 core = static_cast<s32>(max_cores - 1); core >= 0; core--) {
             if (((mask >> core) & 1) != 0) {
                 return core;
             }
@@ -425,7 +424,7 @@ ResultCode Thread::SetCoreAndAffinityMask(s32 new_core, u64 new_affinity_mask) {
         if (old_affinity_mask != new_affinity_mask) {
             const s32 old_core = processor_id;
             if (processor_id >= 0 && ((affinity_mask >> processor_id) & 1) == 0) {
-                if (ideal_core < 0) {
+                if (static_cast<s32>(ideal_core) < 0) {
                     processor_id = HighestSetCore(affinity_mask, GlobalScheduler::NUM_CPU_CORES);
                 } else {
                     processor_id = ideal_core;
@@ -447,23 +446,23 @@ void Thread::AdjustSchedulingOnStatus(u32 old_flags) {
         ThreadSchedStatus::Runnable) {
         // In this case the thread was running, now it's pausing/exitting
         if (processor_id >= 0) {
-            scheduler.Unschedule(current_priority, processor_id, this);
+            scheduler.Unschedule(current_priority, static_cast<u32>(processor_id), this);
         }
 
-        for (s32 core = 0; core < GlobalScheduler::NUM_CPU_CORES; core++) {
-            if (core != processor_id && ((affinity_mask >> core) & 1) != 0) {
-                scheduler.Unsuggest(current_priority, static_cast<u32>(core), this);
+        for (u32 core = 0; core < GlobalScheduler::NUM_CPU_CORES; core++) {
+            if (core != static_cast<u32>(processor_id) && ((affinity_mask >> core) & 1) != 0) {
+                scheduler.Unsuggest(current_priority, core, this);
             }
         }
     } else if (GetSchedulingStatus() == ThreadSchedStatus::Runnable) {
         // The thread is now set to running from being stopped
         if (processor_id >= 0) {
-            scheduler.Schedule(current_priority, processor_id, this);
+            scheduler.Schedule(current_priority, static_cast<u32>(processor_id), this);
         }
 
-        for (s32 core = 0; core < GlobalScheduler::NUM_CPU_CORES; core++) {
-            if (core != processor_id && ((affinity_mask >> core) & 1) != 0) {
-                scheduler.Suggest(current_priority, static_cast<u32>(core), this);
+        for (u32 core = 0; core < GlobalScheduler::NUM_CPU_CORES; core++) {
+            if (core != static_cast<u32>(processor_id) && ((affinity_mask >> core) & 1) != 0) {
+                scheduler.Suggest(current_priority, core, this);
             }
         }
     }
@@ -477,11 +476,11 @@ void Thread::AdjustSchedulingOnPriority(u32 old_priority) {
     }
     auto& scheduler = Core::System::GetInstance().GlobalScheduler();
     if (processor_id >= 0) {
-        scheduler.Unschedule(old_priority, processor_id, this);
+        scheduler.Unschedule(old_priority, static_cast<u32>(processor_id), this);
     }
 
     for (u32 core = 0; core < GlobalScheduler::NUM_CPU_CORES; core++) {
-        if (core != processor_id && ((affinity_mask >> core) & 1) != 0) {
+        if (core != static_cast<u32>(processor_id) && ((affinity_mask >> core) & 1) != 0) {
             scheduler.Unsuggest(old_priority, core, this);
         }
     }
@@ -491,14 +490,14 @@ void Thread::AdjustSchedulingOnPriority(u32 old_priority) {
 
     if (processor_id >= 0) {
         if (current_thread == this) {
-            scheduler.SchedulePrepend(current_priority, processor_id, this);
+            scheduler.SchedulePrepend(current_priority, static_cast<u32>(processor_id), this);
         } else {
-            scheduler.Schedule(current_priority, processor_id, this);
+            scheduler.Schedule(current_priority, static_cast<u32>(processor_id), this);
         }
     }
 
     for (u32 core = 0; core < GlobalScheduler::NUM_CPU_CORES; core++) {
-        if (core != processor_id && ((affinity_mask >> core) & 1) != 0) {
+        if (core != static_cast<u32>(processor_id) && ((affinity_mask >> core) & 1) != 0) {
             scheduler.Suggest(current_priority, core, this);
         }
     }
@@ -515,7 +514,7 @@ void Thread::AdjustSchedulingOnAffinity(u64 old_affinity_mask, s32 old_core) {
 
     for (u32 core = 0; core < GlobalScheduler::NUM_CPU_CORES; core++) {
         if (((old_affinity_mask >> core) & 1) != 0) {
-            if (core == old_core) {
+            if (core == static_cast<u32>(old_core)) {
                 scheduler.Unschedule(current_priority, core, this);
             } else {
                 scheduler.Unsuggest(current_priority, core, this);
@@ -525,7 +524,7 @@ void Thread::AdjustSchedulingOnAffinity(u64 old_affinity_mask, s32 old_core) {
 
     for (u32 core = 0; core < GlobalScheduler::NUM_CPU_CORES; core++) {
         if (((affinity_mask >> core) & 1) != 0) {
-            if (core == processor_id) {
+            if (core == static_cast<u32>(processor_id)) {
                 scheduler.Schedule(current_priority, core, this);
             } else {
                 scheduler.Suggest(current_priority, core, this);

src/core/hle/kernel/thread.h

@@ -440,6 +440,14 @@ public:
         is_running = value;
     }
 
+    bool IsSyncCancelled() const {
+        return is_sync_cancelled;
+    }
+
+    void SetSyncCancelled(bool value) {
+        is_sync_cancelled = value;
+    }
+
 private:
     explicit Thread(KernelCore& kernel);
     ~Thread() override;
@@ -524,6 +532,7 @@ private:
 
     u32 scheduling_state = 0;
     bool is_running = false;
+    bool is_sync_cancelled = false;
 
     std::string name;
 };

src/core/hle/kernel/transfer_memory.cpp

@@ -23,6 +23,8 @@ SharedPtr<TransferMemory> TransferMemory::Create(KernelCore& kernel, VAddr base_
     transfer_memory->owner_permissions = permissions;
     transfer_memory->owner_process = kernel.CurrentProcess();
 
+    transfer_memory->MapMemory(base_address, size, permissions);
+
     return transfer_memory;
 }
 

src/core/hle/kernel/vm_manager.cpp

@@ -167,7 +167,7 @@ ResultVal<VAddr> VMManager::FindFreeRegion(VAddr begin, VAddr end, u64 size) con
 
     if (vma_handle == vma_map.cend()) {
         // TODO(Subv): Find the correct error code here.
-        return ResultCode(-1);
+        return RESULT_UNKNOWN;
     }
 
     const VAddr target = std::max(begin, vma_handle->second.base);

src/core/hle/kernel/wait_object.cpp

@@ -51,17 +51,8 @@ SharedPtr<Thread> WaitObject::GetHighestPriorityReadyThread() const {
         if (ShouldWait(thread.get()))
             continue;
 
-        // A thread is ready to run if it's either in ThreadStatus::WaitSynch
-        // and the rest of the objects it is waiting on are ready.
-        bool ready_to_run = true;
-        if (thread_status == ThreadStatus::WaitSynch) {
-            ready_to_run = thread->AllWaitObjectsReady();
-        }
-
-        if (ready_to_run) {
-            candidate = thread.get();
-            candidate_priority = thread->GetPriority();
-        }
+        candidate = thread.get();
+        candidate_priority = thread->GetPriority();
     }
 
     return candidate;

src/core/hle/result.h

@@ -146,6 +146,14 @@ constexpr bool operator!=(const ResultCode& a, const ResultCode& b) {
 /// The default success `ResultCode`.
 constexpr ResultCode RESULT_SUCCESS(0);
 
+/**
+ * Placeholder result code used for unknown error codes.
+ *
+ * @note This should only be used when a particular error code
+ *       is not known yet.
+ */
+constexpr ResultCode RESULT_UNKNOWN(UINT32_MAX);
+
 /**
  * This is an optional value type. It holds a `ResultCode` and, if that code is a success code,
  * also holds a result of type `T`. If the code is an error code then trying to access the inner
@@ -183,7 +191,7 @@ class ResultVal {
 public:
     /// Constructs an empty `ResultVal` with the given error code. The code must not be a success
     /// code.
-    ResultVal(ResultCode error_code = ResultCode(-1)) : result_code(error_code) {
+    ResultVal(ResultCode error_code = RESULT_UNKNOWN) : result_code(error_code) {
         ASSERT(error_code.IsError());
     }
 

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

@@ -84,7 +84,7 @@ protected:
             LOG_ERROR(Service_ACC, "Failed to get profile base and data for user={}",
                       user_id.Format());
             IPC::ResponseBuilder rb{ctx, 2};
-            rb.Push(ResultCode(-1)); // TODO(ogniK): Get actual error code
+            rb.Push(RESULT_UNKNOWN); // TODO(ogniK): Get actual error code
         }
     }
 
@@ -98,7 +98,7 @@ protected:
         } else {
             LOG_ERROR(Service_ACC, "Failed to get profile base for user={}", user_id.Format());
             IPC::ResponseBuilder rb{ctx, 2};
-            rb.Push(ResultCode(-1)); // TODO(ogniK): Get actual error code
+            rb.Push(RESULT_UNKNOWN); // TODO(ogniK): Get actual error code
         }
     }
 
@@ -442,7 +442,7 @@ void Module::Interface::TrySelectUserWithoutInteraction(Kernel::HLERequestContex
     const auto user_list = profile_manager->GetAllUsers();
     if (std::all_of(user_list.begin(), user_list.end(),
                     [](const auto& user) { return user.uuid == Common::INVALID_UUID; })) {
-        rb.Push(ResultCode(-1)); // TODO(ogniK): Find the correct error code
+        rb.Push(RESULT_UNKNOWN); // TODO(ogniK): Find the correct error code
         rb.PushRaw<u128>(Common::INVALID_UUID);
         return;
     }

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

@@ -31,8 +31,8 @@ struct ProfileDataRaw {
 static_assert(sizeof(ProfileDataRaw) == 0x650, "ProfileDataRaw has incorrect size.");
 
 // TODO(ogniK): Get actual error codes
-constexpr ResultCode ERROR_TOO_MANY_USERS(ErrorModule::Account, -1);
-constexpr ResultCode ERROR_USER_ALREADY_EXISTS(ErrorModule::Account, -2);
+constexpr ResultCode ERROR_TOO_MANY_USERS(ErrorModule::Account, u32(-1));
+constexpr ResultCode ERROR_USER_ALREADY_EXISTS(ErrorModule::Account, u32(-2));
 constexpr ResultCode ERROR_ARGUMENT_IS_NULL(ErrorModule::Account, 20);
 
 constexpr char ACC_SAVE_AVATORS_BASE_PATH[] = "/system/save/8000000000000010/su/avators/";

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

@@ -847,17 +847,16 @@ private:
     void PopInteractiveOutData(Kernel::HLERequestContext& ctx) {
         LOG_DEBUG(Service_AM, "called");
 
-        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
-
         const auto storage = applet->GetBroker().PopInteractiveDataToGame();
         if (storage == nullptr) {
             LOG_ERROR(Service_AM,
                       "storage is a nullptr. There is no data in the current interactive channel");
-
+            IPC::ResponseBuilder rb{ctx, 2};
             rb.Push(ERR_NO_DATA_IN_CHANNEL);
             return;
         }
 
+        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
         rb.PushIpcInterface<IStorage>(std::move(*storage));
     }
@@ -991,7 +990,7 @@ void ILibraryAppletCreator::CreateLibraryApplet(Kernel::HLERequestContext& ctx)
         LOG_ERROR(Service_AM, "Applet doesn't exist! applet_id={}", static_cast<u32>(applet_id));
 
         IPC::ResponseBuilder rb{ctx, 2};
-        rb.Push(ResultCode(-1));
+        rb.Push(RESULT_UNKNOWN);
         return;
     }
 
@@ -1027,7 +1026,7 @@ void ILibraryAppletCreator::CreateTransferMemoryStorage(Kernel::HLERequestContex
     if (transfer_mem == nullptr) {
         LOG_ERROR(Service_AM, "shared_mem is a nullpr for handle={:08X}", handle);
         IPC::ResponseBuilder rb{ctx, 2};
-        rb.Push(ResultCode(-1));
+        rb.Push(RESULT_UNKNOWN);
         return;
     }
 
@@ -1076,7 +1075,8 @@ IApplicationFunctions::IApplicationFunctions(Core::System& system_)
         {100, &IApplicationFunctions::InitializeApplicationCopyrightFrameBuffer, "InitializeApplicationCopyrightFrameBuffer"},
         {101, &IApplicationFunctions::SetApplicationCopyrightImage, "SetApplicationCopyrightImage"},
         {102, &IApplicationFunctions::SetApplicationCopyrightVisibility, "SetApplicationCopyrightVisibility"},
-        {110, nullptr, "QueryApplicationPlayStatistics"},
+        {110, &IApplicationFunctions::QueryApplicationPlayStatistics, "QueryApplicationPlayStatistics"},
+        {111, &IApplicationFunctions::QueryApplicationPlayStatisticsByUid, "QueryApplicationPlayStatisticsByUid"},
         {120, nullptr, "ExecuteProgram"},
         {121, nullptr, "ClearUserChannel"},
         {122, nullptr, "UnpopToUserChannel"},
@@ -1335,12 +1335,16 @@ void IApplicationFunctions::GetPseudoDeviceId(Kernel::HLERequestContext& ctx) {
 }
 
 void IApplicationFunctions::ExtendSaveData(Kernel::HLERequestContext& ctx) {
+    struct Parameters {
+        FileSys::SaveDataType type;
+        u128 user_id;
+        u64 new_normal_size;
+        u64 new_journal_size;
+    };
+    static_assert(sizeof(Parameters) == 40);
+
     IPC::RequestParser rp{ctx};
-    const auto type{rp.PopRaw<FileSys::SaveDataType>()};
-    rp.Skip(1, false);
-    const auto user_id{rp.PopRaw<u128>()};
-    const auto new_normal_size{rp.PopRaw<u64>()};
-    const auto new_journal_size{rp.PopRaw<u64>()};
+    const auto [type, user_id, new_normal_size, new_journal_size] = rp.PopRaw<Parameters>();
 
     LOG_DEBUG(Service_AM,
               "called with type={:02X}, user_id={:016X}{:016X}, new_normal={:016X}, "
@@ -1359,10 +1363,14 @@ void IApplicationFunctions::ExtendSaveData(Kernel::HLERequestContext& ctx) {
 }
 
 void IApplicationFunctions::GetSaveDataSize(Kernel::HLERequestContext& ctx) {
+    struct Parameters {
+        FileSys::SaveDataType type;
+        u128 user_id;
+    };
+    static_assert(sizeof(Parameters) == 24);
+
     IPC::RequestParser rp{ctx};
-    const auto type{rp.PopRaw<FileSys::SaveDataType>()};
-    rp.Skip(1, false);
-    const auto user_id{rp.PopRaw<u128>()};
+    const auto [type, user_id] = rp.PopRaw<Parameters>();
 
     LOG_DEBUG(Service_AM, "called with type={:02X}, user_id={:016X}{:016X}", static_cast<u8>(type),
               user_id[1], user_id[0]);
@@ -1376,6 +1384,22 @@ void IApplicationFunctions::GetSaveDataSize(Kernel::HLERequestContext& ctx) {
     rb.Push(size.journal);
 }
 
+void IApplicationFunctions::QueryApplicationPlayStatistics(Kernel::HLERequestContext& ctx) {
+    LOG_WARNING(Service_AM, "(STUBBED) called");
+
+    IPC::ResponseBuilder rb{ctx, 3};
+    rb.Push(RESULT_SUCCESS);
+    rb.Push<u32>(0);
+}
+
+void IApplicationFunctions::QueryApplicationPlayStatisticsByUid(Kernel::HLERequestContext& ctx) {
+    LOG_WARNING(Service_AM, "(STUBBED) called");
+
+    IPC::ResponseBuilder rb{ctx, 3};
+    rb.Push(RESULT_SUCCESS);
+    rb.Push<u32>(0);
+}
+
 void IApplicationFunctions::GetGpuErrorDetectedSystemEvent(Kernel::HLERequestContext& ctx) {
     LOG_WARNING(Service_AM, "(STUBBED) called");
 

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

@@ -255,6 +255,8 @@ private:
     void InitializeApplicationCopyrightFrameBuffer(Kernel::HLERequestContext& ctx);
     void SetApplicationCopyrightImage(Kernel::HLERequestContext& ctx);
     void SetApplicationCopyrightVisibility(Kernel::HLERequestContext& ctx);
+    void QueryApplicationPlayStatistics(Kernel::HLERequestContext& ctx);
+    void QueryApplicationPlayStatisticsByUid(Kernel::HLERequestContext& ctx);
     void GetGpuErrorDetectedSystemEvent(Kernel::HLERequestContext& ctx);
 
     bool launch_popped_application_specific = false;

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

@@ -88,6 +88,7 @@ std::unique_ptr<IStorage> AppletDataBroker::PopInteractiveDataToApplet() {
 
 void AppletDataBroker::PushNormalDataFromGame(IStorage storage) {
     in_channel.push_back(std::make_unique<IStorage>(storage));
+    pop_out_data_event.writable->Clear();
 }
 
 void AppletDataBroker::PushNormalDataFromApplet(IStorage storage) {
@@ -97,6 +98,7 @@ void AppletDataBroker::PushNormalDataFromApplet(IStorage storage) {
 
 void AppletDataBroker::PushInteractiveDataFromGame(IStorage storage) {
     in_interactive_channel.push_back(std::make_unique<IStorage>(storage));
+    pop_interactive_out_data_event.writable->Clear();
 }
 
 void AppletDataBroker::PushInteractiveDataFromApplet(IStorage storage) {

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

@@ -91,6 +91,7 @@ void SoftwareKeyboard::ExecuteInteractive() {
 
     if (status == INTERACTIVE_STATUS_OK) {
         complete = true;
+        broker.SignalStateChanged();
     } else {
         std::array<char16_t, SWKBD_OUTPUT_INTERACTIVE_BUFFER_SIZE / 2 - 2> string;
         std::memcpy(string.data(), data.data() + 4, string.size() * 2);

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

@@ -337,7 +337,7 @@ void WebBrowser::ExecuteInternal() {
 void WebBrowser::InitializeShop() {
     if (frontend_e_commerce == nullptr) {
         LOG_ERROR(Service_AM, "Missing ECommerce Applet frontend!");
-        status = ResultCode(-1);
+        status = RESULT_UNKNOWN;
         return;
     }
 
@@ -353,7 +353,7 @@ void WebBrowser::InitializeShop() {
 
     if (url == args.end()) {
         LOG_ERROR(Service_AM, "Missing EShop Arguments URL for initialization!");
-        status = ResultCode(-1);
+        status = RESULT_UNKNOWN;
         return;
     }
 
@@ -366,7 +366,7 @@ void WebBrowser::InitializeShop() {
     // Less is missing info, More is malformed
     if (split_query.size() != 2) {
         LOG_ERROR(Service_AM, "EShop Arguments has more than one question mark, malformed");
-        status = ResultCode(-1);
+        status = RESULT_UNKNOWN;
         return;
     }
 
@@ -390,7 +390,7 @@ void WebBrowser::InitializeShop() {
 
     if (scene == shop_query.end()) {
         LOG_ERROR(Service_AM, "No scene parameter was passed via shop query!");
-        status = ResultCode(-1);
+        status = RESULT_UNKNOWN;
         return;
     }
 
@@ -406,7 +406,7 @@ void WebBrowser::InitializeShop() {
     const auto target = target_map.find(scene->second);
     if (target == target_map.end()) {
         LOG_ERROR(Service_AM, "Scene for shop query is invalid! (scene={})", scene->second);
-        status = ResultCode(-1);
+        status = RESULT_UNKNOWN;
         return;
     }
 
@@ -427,7 +427,7 @@ void WebBrowser::InitializeOffline() {
     if (args.find(WebArgTLVType::DocumentPath) == args.end() ||
         args.find(WebArgTLVType::DocumentKind) == args.end() ||
         args.find(WebArgTLVType::ApplicationID) == args.end()) {
-        status = ResultCode(-1);
+        status = RESULT_UNKNOWN;
         LOG_ERROR(Service_AM, "Missing necessary parameters for initialization!");
     }
 
@@ -476,7 +476,7 @@ void WebBrowser::InitializeOffline() {
 
     offline_romfs = GetApplicationRomFS(system, title_id, type);
     if (offline_romfs == nullptr) {
-        status = ResultCode(-1);
+        status = RESULT_UNKNOWN;
         LOG_ERROR(Service_AM, "Failed to find offline data for request!");
     }
 
@@ -496,7 +496,7 @@ void WebBrowser::ExecuteShop() {
     const auto check_optional_parameter = [this](const auto& p) {
         if (!p.has_value()) {
             LOG_ERROR(Service_AM, "Missing one or more necessary parameters for execution!");
-            status = ResultCode(-1);
+            status = RESULT_UNKNOWN;
             return false;
         }
 

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

@@ -131,7 +131,7 @@ void AOC_U::ListAddOnContent(Kernel::HLERequestContext& ctx) {
     if (out.size() < offset) {
         IPC::ResponseBuilder rb{ctx, 2};
         // TODO(DarkLordZach): Find the correct error code.
-        rb.Push(ResultCode(-1));
+        rb.Push(RESULT_UNKNOWN);
         return;
     }
 

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

@@ -80,7 +80,7 @@ private:
             LOG_ERROR(Audio, "Failed to decode opus data");
             IPC::ResponseBuilder rb{ctx, 2};
             // TODO(ogniK): Use correct error code
-            rb.Push(ResultCode(-1));
+            rb.Push(RESULT_UNKNOWN);
             return;
         }
 
@@ -278,7 +278,7 @@ void HwOpus::OpenOpusDecoder(Kernel::HLERequestContext& ctx) {
         LOG_ERROR(Audio, "Failed to create Opus decoder (error={}).", error);
         IPC::ResponseBuilder rb{ctx, 2};
         // TODO(ogniK): Use correct error code
-        rb.Push(ResultCode(-1));
+        rb.Push(RESULT_UNKNOWN);
         return;
     }
 

src/core/hle/service/bcat/backend/boxcat.cpp

@@ -114,7 +114,7 @@ void HandleDownloadDisplayResult(const AM::Applets::AppletManager& applet_manage
 
     const auto& frontend{applet_manager.GetAppletFrontendSet()};
     frontend.error->ShowCustomErrorText(
-        ResultCode(-1), "There was an error while attempting to use Boxcat.",
+        RESULT_UNKNOWN, "There was an error while attempting to use Boxcat.",
         DOWNLOAD_RESULT_LOG_MESSAGES[static_cast<std::size_t>(res)], [] {});
 }
 
@@ -255,7 +255,7 @@ private:
     using Digest = std::array<u8, 0x20>;
     static Digest DigestFile(std::vector<u8> bytes) {
         Digest out{};
-        mbedtls_sha256(bytes.data(), bytes.size(), out.data(), 0);
+        mbedtls_sha256_ret(bytes.data(), bytes.size(), out.data(), 0);
         return out;
     }
 

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

@@ -46,7 +46,7 @@ u64 GetCurrentBuildID(const Core::System::CurrentBuildProcessID& id) {
 BCATDigest DigestFile(const FileSys::VirtualFile& file) {
     BCATDigest out{};
     const auto bytes = file->ReadAllBytes();
-    mbedtls_md5(bytes.data(), bytes.size(), out.data());
+    mbedtls_md5_ret(bytes.data(), bytes.size(), out.data());
     return out;
 }
 

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

@@ -40,7 +40,10 @@ static FileSys::VirtualDir GetDirectoryRelativeWrapped(FileSys::VirtualDir base,
     if (dir_name.empty() || dir_name == "." || dir_name == "/" || dir_name == "\\")
         return base;
 
-    return base->GetDirectoryRelative(dir_name);
+    const auto res = base->GetDirectoryRelative(dir_name);
+    if (res == nullptr)
+        return base->CreateDirectoryRelative(dir_name);
+    return res;
 }
 
 VfsDirectoryServiceWrapper::VfsDirectoryServiceWrapper(FileSys::VirtualDir backing_)
@@ -58,11 +61,11 @@ ResultCode VfsDirectoryServiceWrapper::CreateFile(const std::string& path_, u64
     auto file = dir->CreateFile(FileUtil::GetFilename(path));
     if (file == nullptr) {
         // TODO(DarkLordZach): Find a better error code for this
-        return ResultCode(-1);
+        return RESULT_UNKNOWN;
     }
     if (!file->Resize(size)) {
         // TODO(DarkLordZach): Find a better error code for this
-        return ResultCode(-1);
+        return RESULT_UNKNOWN;
     }
     return RESULT_SUCCESS;
 }
@@ -80,7 +83,7 @@ ResultCode VfsDirectoryServiceWrapper::DeleteFile(const std::string& path_) cons
     }
     if (!dir->DeleteFile(FileUtil::GetFilename(path))) {
         // TODO(DarkLordZach): Find a better error code for this
-        return ResultCode(-1);
+        return RESULT_UNKNOWN;
     }
 
     return RESULT_SUCCESS;
@@ -94,7 +97,7 @@ ResultCode VfsDirectoryServiceWrapper::CreateDirectory(const std::string& path_)
     auto new_dir = dir->CreateSubdirectory(FileUtil::GetFilename(path));
     if (new_dir == nullptr) {
         // TODO(DarkLordZach): Find a better error code for this
-        return ResultCode(-1);
+        return RESULT_UNKNOWN;
     }
     return RESULT_SUCCESS;
 }
@@ -104,7 +107,7 @@ ResultCode VfsDirectoryServiceWrapper::DeleteDirectory(const std::string& path_)
     auto dir = GetDirectoryRelativeWrapped(backing, FileUtil::GetParentPath(path));
     if (!dir->DeleteSubdirectory(FileUtil::GetFilename(path))) {
         // TODO(DarkLordZach): Find a better error code for this
-        return ResultCode(-1);
+        return RESULT_UNKNOWN;
     }
     return RESULT_SUCCESS;
 }
@@ -114,7 +117,7 @@ ResultCode VfsDirectoryServiceWrapper::DeleteDirectoryRecursively(const std::str
     auto dir = GetDirectoryRelativeWrapped(backing, FileUtil::GetParentPath(path));
     if (!dir->DeleteSubdirectoryRecursive(FileUtil::GetFilename(path))) {
         // TODO(DarkLordZach): Find a better error code for this
-        return ResultCode(-1);
+        return RESULT_UNKNOWN;
     }
     return RESULT_SUCCESS;
 }
@@ -125,7 +128,7 @@ ResultCode VfsDirectoryServiceWrapper::CleanDirectoryRecursively(const std::stri
 
     if (!dir->CleanSubdirectoryRecursive(FileUtil::GetFilename(sanitized_path))) {
         // TODO(DarkLordZach): Find a better error code for this
-        return ResultCode(-1);
+        return RESULT_UNKNOWN;
     }
 
     return RESULT_SUCCESS;
@@ -142,7 +145,7 @@ ResultCode VfsDirectoryServiceWrapper::RenameFile(const std::string& src_path_,
             return FileSys::ERROR_PATH_NOT_FOUND;
         if (!src->Rename(FileUtil::GetFilename(dest_path))) {
             // TODO(DarkLordZach): Find a better error code for this
-            return ResultCode(-1);
+            return RESULT_UNKNOWN;
         }
         return RESULT_SUCCESS;
     }
@@ -160,7 +163,7 @@ ResultCode VfsDirectoryServiceWrapper::RenameFile(const std::string& src_path_,
 
     if (!src->GetContainingDirectory()->DeleteFile(FileUtil::GetFilename(src_path))) {
         // TODO(DarkLordZach): Find a better error code for this
-        return ResultCode(-1);
+        return RESULT_UNKNOWN;
     }
 
     return RESULT_SUCCESS;
@@ -177,7 +180,7 @@ ResultCode VfsDirectoryServiceWrapper::RenameDirectory(const std::string& src_pa
             return FileSys::ERROR_PATH_NOT_FOUND;
         if (!src->Rename(FileUtil::GetFilename(dest_path))) {
             // TODO(DarkLordZach): Find a better error code for this
-            return ResultCode(-1);
+            return RESULT_UNKNOWN;
         }
         return RESULT_SUCCESS;
     }
@@ -189,7 +192,7 @@ ResultCode VfsDirectoryServiceWrapper::RenameDirectory(const std::string& src_pa
                src_path, dest_path);
 
     // TODO(DarkLordZach): Find a better error code for this
-    return ResultCode(-1);
+    return RESULT_UNKNOWN;
 }
 
 ResultVal<FileSys::VirtualFile> VfsDirectoryServiceWrapper::OpenFile(const std::string& path_,
@@ -287,7 +290,7 @@ ResultVal<FileSys::VirtualFile> FileSystemController::OpenRomFSCurrentProcess()
 
     if (romfs_factory == nullptr) {
         // TODO(bunnei): Find a better error code for this
-        return ResultCode(-1);
+        return RESULT_UNKNOWN;
     }
 
     return romfs_factory->OpenCurrentProcess(system.CurrentProcess()->GetTitleID());
@@ -300,7 +303,7 @@ ResultVal<FileSys::VirtualFile> FileSystemController::OpenRomFS(
 
     if (romfs_factory == nullptr) {
         // TODO(bunnei): Find a better error code for this
-        return ResultCode(-1);
+        return RESULT_UNKNOWN;
     }
 
     return romfs_factory->Open(title_id, storage_id, type);

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

@@ -785,7 +785,7 @@ void FSP_SRV::OpenFileSystemWithPatch(Kernel::HLERequestContext& ctx) {
                 static_cast<u8>(type), title_id);
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 0};
-    rb.Push(ResultCode(-1));
+    rb.Push(RESULT_UNKNOWN);
 }
 
 void FSP_SRV::OpenSdCardFileSystem(Kernel::HLERequestContext& ctx) {
@@ -891,7 +891,7 @@ void FSP_SRV::OpenDataStorageByCurrentProcess(Kernel::HLERequestContext& ctx) {
         // TODO (bunnei): Find the right error code to use here
         LOG_CRITICAL(Service_FS, "no file system interface available!");
         IPC::ResponseBuilder rb{ctx, 2};
-        rb.Push(ResultCode(-1));
+        rb.Push(RESULT_UNKNOWN);
         return;
     }
 
@@ -928,7 +928,7 @@ void FSP_SRV::OpenDataStorageByDataId(Kernel::HLERequestContext& ctx) {
                   "could not open data storage with title_id={:016X}, storage_id={:02X}", title_id,
                   static_cast<u8>(storage_id));
         IPC::ResponseBuilder rb{ctx, 2};
-        rb.Push(ResultCode(-1));
+        rb.Push(RESULT_UNKNOWN);
         return;
     }
 

src/core/hle/service/ldr/ldr.cpp

@@ -294,7 +294,7 @@ public:
         Memory::ReadBlock(nro_address, nro_data.data(), nro_size);
 
         SHA256Hash hash{};
-        mbedtls_sha256(nro_data.data(), nro_data.size(), hash.data(), 0);
+        mbedtls_sha256_ret(nro_data.data(), nro_data.size(), hash.data(), 0);
 
         // NRO Hash is already loaded
         if (std::any_of(nro.begin(), nro.end(), [&hash](const std::pair<VAddr, NROInfo>& info) {

src/core/hle/service/mii/mii.cpp

@@ -242,7 +242,7 @@ private:
         const auto index = db.IndexOf(uuid);
         if (index > MAX_MIIS) {
             // TODO(DarkLordZach): Find a better error code
-            rb.Push(ResultCode(-1));
+            rb.Push(RESULT_UNKNOWN);
             rb.Push(index);
         } else {
             rb.Push(RESULT_SUCCESS);
@@ -268,7 +268,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2};
         // TODO(DarkLordZach): Find a better error code
-        rb.Push(success ? RESULT_SUCCESS : ResultCode(-1));
+        rb.Push(success ? RESULT_SUCCESS : RESULT_UNKNOWN);
     }
 
     void AddOrReplace(Kernel::HLERequestContext& ctx) {
@@ -282,7 +282,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2};
         // TODO(DarkLordZach): Find a better error code
-        rb.Push(success ? RESULT_SUCCESS : ResultCode(-1));
+        rb.Push(success ? RESULT_SUCCESS : RESULT_UNKNOWN);
     }
 
     void Delete(Kernel::HLERequestContext& ctx) {

src/core/hle/service/nfp/nfp.cpp

@@ -16,10 +16,7 @@
 #include "core/hle/service/nfp/nfp_user.h"
 
 namespace Service::NFP {
-
 namespace ErrCodes {
-[[maybe_unused]] constexpr ResultCode ERR_TAG_FAILED(ErrorModule::NFP,
-                                                     -1); // TODO(ogniK): Find the actual error code
 constexpr ResultCode ERR_NO_APPLICATION_AREA(ErrorModule::NFP, 152);
 } // namespace ErrCodes
 

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

@@ -9,6 +9,7 @@
 #include "core/hle/kernel/writable_event.h"
 #include "core/hle/service/nifm/nifm.h"
 #include "core/hle/service/service.h"
+#include "core/settings.h"
 
 namespace Service::NIFM {
 
@@ -86,7 +87,12 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 3};
         rb.Push(RESULT_SUCCESS);
-        rb.PushEnum(RequestState::Connected);
+
+        if (Settings::values.bcat_backend == "none") {
+            rb.PushEnum(RequestState::NotSubmitted);
+        } else {
+            rb.PushEnum(RequestState::Connected);
+        }
     }
 
     void GetResult(Kernel::HLERequestContext& ctx) {
@@ -194,14 +200,22 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 3};
         rb.Push(RESULT_SUCCESS);
-        rb.Push<u8>(1);
+        if (Settings::values.bcat_backend == "none") {
+            rb.Push<u8>(0);
+        } else {
+            rb.Push<u8>(1);
+        }
     }
     void IsAnyInternetRequestAccepted(Kernel::HLERequestContext& ctx) {
         LOG_WARNING(Service_NIFM, "(STUBBED) called");
 
         IPC::ResponseBuilder rb{ctx, 3};
         rb.Push(RESULT_SUCCESS);
-        rb.Push<u8>(1);
+        if (Settings::values.bcat_backend == "none") {
+            rb.Push<u8>(0);
+        } else {
+            rb.Push<u8>(1);
+        }
     }
     Core::System& system;
 };

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

@@ -271,7 +271,7 @@ void IApplicationManagerInterface::GetApplicationControlData(Kernel::HLERequestC
                       "output buffer is too small! (actual={:016X}, expected_min=0x4000)", size);
             IPC::ResponseBuilder rb{ctx, 2};
             // TODO(DarkLordZach): Find a better error code for this.
-            rb.Push(ResultCode(-1));
+            rb.Push(RESULT_UNKNOWN);
             return;
         }
 
@@ -291,7 +291,7 @@ void IApplicationManagerInterface::GetApplicationControlData(Kernel::HLERequestC
                       0x4000 + control.second->GetSize());
             IPC::ResponseBuilder rb{ctx, 2};
             // TODO(DarkLordZach): Find a better error code for this.
-            rb.Push(ResultCode(-1));
+            rb.Push(RESULT_UNKNOWN);
             return;
         }
 

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

@@ -97,7 +97,7 @@ void EncryptSharedFont(const std::vector<u32>& input, std::vector<u8>& output,
     const auto key = Common::swap32(EXPECTED_RESULT ^ EXPECTED_MAGIC);
     std::vector<u32> transformed_font(input.size() + 2);
     transformed_font[0] = Common::swap32(EXPECTED_MAGIC);
-    transformed_font[1] = Common::swap32(input.size() * sizeof(u32)) ^ key;
+    transformed_font[1] = Common::swap32(static_cast<u32>(input.size() * sizeof(u32))) ^ key;
     std::transform(input.begin(), input.end(), transformed_font.begin() + 2,
                    [key](u32 in) { return in ^ key; });
     std::memcpy(output.data() + offset, transformed_font.data(),

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

@@ -22,6 +22,18 @@ u32 nvhost_nvdec::ioctl(Ioctl command, const std::vector<u8>& input, const std::
     switch (static_cast<IoctlCommand>(command.raw)) {
     case IoctlCommand::IocSetNVMAPfdCommand:
         return SetNVMAPfd(input, output);
+    case IoctlCommand::IocSubmit:
+        return Submit(input, output);
+    case IoctlCommand::IocGetSyncpoint:
+        return GetSyncpoint(input, output);
+    case IoctlCommand::IocGetWaitbase:
+        return GetWaitbase(input, output);
+    case IoctlCommand::IocMapBuffer:
+        return MapBuffer(input, output);
+    case IoctlCommand::IocMapBufferEx:
+        return MapBufferEx(input, output);
+    case IoctlCommand::IocUnmapBufferEx:
+        return UnmapBufferEx(input, output);
     }
 
     UNIMPLEMENTED_MSG("Unimplemented ioctl");
@@ -30,11 +42,67 @@ u32 nvhost_nvdec::ioctl(Ioctl command, const std::vector<u8>& input, const std::
 
 u32 nvhost_nvdec::SetNVMAPfd(const std::vector<u8>& input, std::vector<u8>& output) {
     IoctlSetNvmapFD params{};
-    std::memcpy(&params, input.data(), input.size());
+    std::memcpy(&params, input.data(), sizeof(IoctlSetNvmapFD));
     LOG_DEBUG(Service_NVDRV, "called, fd={}", params.nvmap_fd);
 
     nvmap_fd = params.nvmap_fd;
     return 0;
 }
 
+u32 nvhost_nvdec::Submit(const std::vector<u8>& input, std::vector<u8>& output) {
+    IoctlSubmit params{};
+    std::memcpy(&params, input.data(), sizeof(IoctlSubmit));
+    LOG_WARNING(Service_NVDRV, "(STUBBED) called");
+    std::memcpy(output.data(), &params, sizeof(IoctlSubmit));
+    return 0;
+}
+
+u32 nvhost_nvdec::GetSyncpoint(const std::vector<u8>& input, std::vector<u8>& output) {
+    IoctlGetSyncpoint params{};
+    std::memcpy(&params, input.data(), sizeof(IoctlGetSyncpoint));
+    LOG_INFO(Service_NVDRV, "called, unknown=0x{:X}", params.unknown);
+    params.value = 0; // Seems to be hard coded at 0
+    std::memcpy(output.data(), &params, sizeof(IoctlGetSyncpoint));
+    return 0;
+}
+
+u32 nvhost_nvdec::GetWaitbase(const std::vector<u8>& input, std::vector<u8>& output) {
+    IoctlGetWaitbase params{};
+    std::memcpy(&params, input.data(), sizeof(IoctlGetWaitbase));
+    LOG_INFO(Service_NVDRV, "called, unknown=0x{:X}", params.unknown);
+    params.value = 0; // Seems to be hard coded at 0
+    std::memcpy(output.data(), &params, sizeof(IoctlGetWaitbase));
+    return 0;
+}
+
+u32 nvhost_nvdec::MapBuffer(const std::vector<u8>& input, std::vector<u8>& output) {
+    IoctlMapBuffer params{};
+    std::memcpy(&params, input.data(), sizeof(IoctlMapBuffer));
+    LOG_WARNING(Service_NVDRV, "(STUBBED) called with address={:08X}{:08X}", params.address_2,
+                params.address_1);
+    params.address_1 = 0;
+    params.address_2 = 0;
+    std::memcpy(output.data(), &params, sizeof(IoctlMapBuffer));
+    return 0;
+}
+
+u32 nvhost_nvdec::MapBufferEx(const std::vector<u8>& input, std::vector<u8>& output) {
+    IoctlMapBufferEx params{};
+    std::memcpy(&params, input.data(), sizeof(IoctlMapBufferEx));
+    LOG_WARNING(Service_NVDRV, "(STUBBED) called with address={:08X}{:08X}", params.address_2,
+                params.address_1);
+    params.address_1 = 0;
+    params.address_2 = 0;
+    std::memcpy(output.data(), &params, sizeof(IoctlMapBufferEx));
+    return 0;
+}
+
+u32 nvhost_nvdec::UnmapBufferEx(const std::vector<u8>& input, std::vector<u8>& output) {
+    IoctlUnmapBufferEx params{};
+    std::memcpy(&params, input.data(), sizeof(IoctlUnmapBufferEx));
+    LOG_WARNING(Service_NVDRV, "(STUBBED) called");
+    std::memcpy(output.data(), &params, sizeof(IoctlUnmapBufferEx));
+    return 0;
+}
+
 } // namespace Service::Nvidia::Devices

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

@@ -23,16 +23,66 @@ public:
 private:
     enum class IoctlCommand : u32_le {
         IocSetNVMAPfdCommand = 0x40044801,
+        IocSubmit = 0xC0400001,
+        IocGetSyncpoint = 0xC0080002,
+        IocGetWaitbase = 0xC0080003,
+        IocMapBuffer = 0xC01C0009,
+        IocMapBufferEx = 0xC0A40009,
+        IocUnmapBufferEx = 0xC0A4000A,
     };
 
     struct IoctlSetNvmapFD {
         u32_le nvmap_fd;
     };
-    static_assert(sizeof(IoctlSetNvmapFD) == 4, "IoctlSetNvmapFD is incorrect size");
+    static_assert(sizeof(IoctlSetNvmapFD) == 0x4, "IoctlSetNvmapFD is incorrect size");
+
+    struct IoctlSubmit {
+        INSERT_PADDING_BYTES(0x40); // TODO(DarkLordZach): RE this structure
+    };
+    static_assert(sizeof(IoctlSubmit) == 0x40, "IoctlSubmit has incorrect size");
+
+    struct IoctlGetSyncpoint {
+        u32 unknown; // seems to be ignored? Nintendo added this
+        u32 value;
+    };
+    static_assert(sizeof(IoctlGetSyncpoint) == 0x08, "IoctlGetSyncpoint has incorrect size");
+
+    struct IoctlGetWaitbase {
+        u32 unknown; // seems to be ignored? Nintendo added this
+        u32 value;
+    };
+    static_assert(sizeof(IoctlGetWaitbase) == 0x08, "IoctlGetWaitbase has incorrect size");
+
+    struct IoctlMapBuffer {
+        u32 unknown;
+        u32 address_1;
+        u32 address_2;
+        INSERT_PADDING_BYTES(0x10); // TODO(DarkLordZach): RE this structure
+    };
+    static_assert(sizeof(IoctlMapBuffer) == 0x1C, "IoctlMapBuffer is incorrect size");
+
+    struct IoctlMapBufferEx {
+        u32 unknown;
+        u32 address_1;
+        u32 address_2;
+        INSERT_PADDING_BYTES(0x98); // TODO(DarkLordZach): RE this structure
+    };
+    static_assert(sizeof(IoctlMapBufferEx) == 0xA4, "IoctlMapBufferEx has incorrect size");
+
+    struct IoctlUnmapBufferEx {
+        INSERT_PADDING_BYTES(0xA4); // TODO(DarkLordZach): RE this structure
+    };
+    static_assert(sizeof(IoctlUnmapBufferEx) == 0xA4, "IoctlUnmapBufferEx has incorrect size");
 
     u32_le nvmap_fd{};
 
     u32 SetNVMAPfd(const std::vector<u8>& input, std::vector<u8>& output);
+    u32 Submit(const std::vector<u8>& input, std::vector<u8>& output);
+    u32 GetSyncpoint(const std::vector<u8>& input, std::vector<u8>& output);
+    u32 GetWaitbase(const std::vector<u8>& input, std::vector<u8>& output);
+    u32 MapBuffer(const std::vector<u8>& input, std::vector<u8>& output);
+    u32 MapBufferEx(const std::vector<u8>& input, std::vector<u8>& output);
+    u32 UnmapBufferEx(const std::vector<u8>& input, std::vector<u8>& output);
 };
 
 } // namespace Service::Nvidia::Devices

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

@@ -22,6 +22,18 @@ u32 nvhost_vic::ioctl(Ioctl command, const std::vector<u8>& input, const std::ve
     switch (static_cast<IoctlCommand>(command.raw)) {
     case IoctlCommand::IocSetNVMAPfdCommand:
         return SetNVMAPfd(input, output);
+    case IoctlCommand::IocSubmit:
+        return Submit(input, output);
+    case IoctlCommand::IocGetSyncpoint:
+        return GetSyncpoint(input, output);
+    case IoctlCommand::IocGetWaitbase:
+        return GetWaitbase(input, output);
+    case IoctlCommand::IocMapBuffer:
+        return MapBuffer(input, output);
+    case IoctlCommand::IocMapBufferEx:
+        return MapBuffer(input, output);
+    case IoctlCommand::IocUnmapBufferEx:
+        return UnmapBufferEx(input, output);
     }
 
     UNIMPLEMENTED_MSG("Unimplemented ioctl");
@@ -30,11 +42,67 @@ u32 nvhost_vic::ioctl(Ioctl command, const std::vector<u8>& input, const std::ve
 
 u32 nvhost_vic::SetNVMAPfd(const std::vector<u8>& input, std::vector<u8>& output) {
     IoctlSetNvmapFD params{};
-    std::memcpy(&params, input.data(), input.size());
+    std::memcpy(&params, input.data(), sizeof(IoctlSetNvmapFD));
     LOG_DEBUG(Service_NVDRV, "called, fd={}", params.nvmap_fd);
 
     nvmap_fd = params.nvmap_fd;
     return 0;
 }
 
+u32 nvhost_vic::Submit(const std::vector<u8>& input, std::vector<u8>& output) {
+    IoctlSubmit params{};
+    std::memcpy(&params, input.data(), sizeof(IoctlSubmit));
+    LOG_WARNING(Service_NVDRV, "(STUBBED) called");
+    std::memcpy(output.data(), &params, sizeof(IoctlSubmit));
+    return 0;
+}
+
+u32 nvhost_vic::GetSyncpoint(const std::vector<u8>& input, std::vector<u8>& output) {
+    IoctlGetSyncpoint params{};
+    std::memcpy(&params, input.data(), sizeof(IoctlGetSyncpoint));
+    LOG_INFO(Service_NVDRV, "called, unknown=0x{:X}", params.unknown);
+    params.value = 0; // Seems to be hard coded at 0
+    std::memcpy(output.data(), &params, sizeof(IoctlGetSyncpoint));
+    return 0;
+}
+
+u32 nvhost_vic::GetWaitbase(const std::vector<u8>& input, std::vector<u8>& output) {
+    IoctlGetWaitbase params{};
+    std::memcpy(&params, input.data(), sizeof(IoctlGetWaitbase));
+    LOG_INFO(Service_NVDRV, "called, unknown=0x{:X}", params.unknown);
+    params.value = 0; // Seems to be hard coded at 0
+    std::memcpy(output.data(), &params, sizeof(IoctlGetWaitbase));
+    return 0;
+}
+
+u32 nvhost_vic::MapBuffer(const std::vector<u8>& input, std::vector<u8>& output) {
+    IoctlMapBuffer params{};
+    std::memcpy(&params, input.data(), sizeof(IoctlMapBuffer));
+    LOG_WARNING(Service_NVDRV, "(STUBBED) called with address={:08X}{:08X}", params.address_2,
+                params.address_1);
+    params.address_1 = 0;
+    params.address_2 = 0;
+    std::memcpy(output.data(), &params, sizeof(IoctlMapBuffer));
+    return 0;
+}
+
+u32 nvhost_vic::MapBufferEx(const std::vector<u8>& input, std::vector<u8>& output) {
+    IoctlMapBufferEx params{};
+    std::memcpy(&params, input.data(), sizeof(IoctlMapBufferEx));
+    LOG_WARNING(Service_NVDRV, "(STUBBED) called with address={:08X}{:08X}", params.address_2,
+                params.address_1);
+    params.address_1 = 0;
+    params.address_2 = 0;
+    std::memcpy(output.data(), &params, sizeof(IoctlMapBufferEx));
+    return 0;
+}
+
+u32 nvhost_vic::UnmapBufferEx(const std::vector<u8>& input, std::vector<u8>& output) {
+    IoctlUnmapBufferEx params{};
+    std::memcpy(&params, input.data(), sizeof(IoctlUnmapBufferEx));
+    LOG_WARNING(Service_NVDRV, "(STUBBED) called");
+    std::memcpy(output.data(), &params, sizeof(IoctlUnmapBufferEx));
+    return 0;
+}
+
 } // namespace Service::Nvidia::Devices

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

@@ -23,6 +23,12 @@ public:
 private:
     enum class IoctlCommand : u32_le {
         IocSetNVMAPfdCommand = 0x40044801,
+        IocSubmit = 0xC0400001,
+        IocGetSyncpoint = 0xC0080002,
+        IocGetWaitbase = 0xC0080003,
+        IocMapBuffer = 0xC01C0009,
+        IocMapBufferEx = 0xC03C0009,
+        IocUnmapBufferEx = 0xC03C000A,
     };
 
     struct IoctlSetNvmapFD {
@@ -30,9 +36,53 @@ private:
     };
     static_assert(sizeof(IoctlSetNvmapFD) == 4, "IoctlSetNvmapFD is incorrect size");
 
+    struct IoctlSubmit {
+        INSERT_PADDING_BYTES(0x40); // TODO(DarkLordZach): RE this structure
+    };
+    static_assert(sizeof(IoctlSubmit) == 0x40, "IoctlSubmit is incorrect size");
+
+    struct IoctlGetSyncpoint {
+        u32 unknown; // seems to be ignored? Nintendo added this
+        u32 value;
+    };
+    static_assert(sizeof(IoctlGetSyncpoint) == 0x8, "IoctlGetSyncpoint is incorrect size");
+
+    struct IoctlGetWaitbase {
+        u32 unknown; // seems to be ignored? Nintendo added this
+        u32 value;
+    };
+    static_assert(sizeof(IoctlGetWaitbase) == 0x8, "IoctlGetWaitbase is incorrect size");
+
+    struct IoctlMapBuffer {
+        u32 unknown;
+        u32 address_1;
+        u32 address_2;
+        INSERT_PADDING_BYTES(0x10); // TODO(DarkLordZach): RE this structure
+    };
+    static_assert(sizeof(IoctlMapBuffer) == 0x1C, "IoctlMapBuffer is incorrect size");
+
+    struct IoctlMapBufferEx {
+        u32 unknown;
+        u32 address_1;
+        u32 address_2;
+        INSERT_PADDING_BYTES(0x30); // TODO(DarkLordZach): RE this structure
+    };
+    static_assert(sizeof(IoctlMapBufferEx) == 0x3C, "IoctlMapBufferEx is incorrect size");
+
+    struct IoctlUnmapBufferEx {
+        INSERT_PADDING_BYTES(0x3C); // TODO(DarkLordZach): RE this structure
+    };
+    static_assert(sizeof(IoctlUnmapBufferEx) == 0x3C, "IoctlUnmapBufferEx is incorrect size");
+
     u32_le nvmap_fd{};
 
     u32 SetNVMAPfd(const std::vector<u8>& input, std::vector<u8>& output);
+    u32 Submit(const std::vector<u8>& input, std::vector<u8>& output);
+    u32 GetSyncpoint(const std::vector<u8>& input, std::vector<u8>& output);
+    u32 GetWaitbase(const std::vector<u8>& input, std::vector<u8>& output);
+    u32 MapBuffer(const std::vector<u8>& input, std::vector<u8>& output);
+    u32 MapBufferEx(const std::vector<u8>& input, std::vector<u8>& output);
+    u32 UnmapBufferEx(const std::vector<u8>& input, std::vector<u8>& output);
 };
 
 } // namespace Service::Nvidia::Devices

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

@@ -34,12 +34,12 @@ static void PosixToCalendar(u64 posix_time, CalendarTime& calendar_time,
         additional_info = {};
         return;
     }
-    calendar_time.year = tm->tm_year + 1900;
-    calendar_time.month = tm->tm_mon + 1;
-    calendar_time.day = tm->tm_mday;
-    calendar_time.hour = tm->tm_hour;
-    calendar_time.minute = tm->tm_min;
-    calendar_time.second = tm->tm_sec;
+    calendar_time.year = static_cast<u16_le>(tm->tm_year + 1900);
+    calendar_time.month = static_cast<u8>(tm->tm_mon + 1);
+    calendar_time.day = static_cast<u8>(tm->tm_mday);
+    calendar_time.hour = static_cast<u8>(tm->tm_hour);
+    calendar_time.minute = static_cast<u8>(tm->tm_min);
+    calendar_time.second = static_cast<u8>(tm->tm_sec);
 
     additional_info.day_of_week = tm->tm_wday;
     additional_info.day_of_year = tm->tm_yday;
@@ -322,7 +322,7 @@ void Module::Interface::GetClockSnapshot(Kernel::HLERequestContext& ctx) {
     if (tm == nullptr) {
         LOG_ERROR(Service_Time, "tm is a nullptr");
         IPC::ResponseBuilder rb{ctx, 2};
-        rb.Push(ResultCode(-1)); // TODO(ogniK): Find appropriate error code
+        rb.Push(RESULT_UNKNOWN); // TODO(ogniK): Find appropriate error code
         return;
     }
 
@@ -331,12 +331,12 @@ void Module::Interface::GetClockSnapshot(Kernel::HLERequestContext& ctx) {
     const SteadyClockTimePoint steady_clock_time_point{static_cast<u64_le>(ms.count() / 1000), {}};
 
     CalendarTime calendar_time{};
-    calendar_time.year = tm->tm_year + 1900;
-    calendar_time.month = tm->tm_mon + 1;
-    calendar_time.day = tm->tm_mday;
-    calendar_time.hour = tm->tm_hour;
-    calendar_time.minute = tm->tm_min;
-    calendar_time.second = tm->tm_sec;
+    calendar_time.year = static_cast<u16_le>(tm->tm_year + 1900);
+    calendar_time.month = static_cast<u8>(tm->tm_mon + 1);
+    calendar_time.day = static_cast<u8>(tm->tm_mday);
+    calendar_time.hour = static_cast<u8>(tm->tm_hour);
+    calendar_time.minute = static_cast<u8>(tm->tm_min);
+    calendar_time.second = static_cast<u8>(tm->tm_sec);
 
     ClockSnapshot clock_snapshot{};
     clock_snapshot.system_posix_time = time_since_epoch;

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

@@ -541,7 +541,7 @@ private:
             } else {
                 // Wait the current thread until a buffer becomes available
                 ctx.SleepClientThread(
-                    "IHOSBinderDriver::DequeueBuffer", -1,
+                    "IHOSBinderDriver::DequeueBuffer", UINT64_MAX,
                     [=](Kernel::SharedPtr<Kernel::Thread> thread, Kernel::HLERequestContext& ctx,
                         Kernel::ThreadWakeupReason reason) {
                         // Repeat TransactParcel DequeueBuffer when a buffer is available

src/core/perf_stats.cpp

@@ -81,7 +81,7 @@ double PerfStats::GetMeanFrametime() {
         return 0;
     }
     const double sum = std::accumulate(perf_history.begin() + IgnoreFrames,
-                                       perf_history.begin() + current_index, 0);
+                                       perf_history.begin() + current_index, 0.0);
     return sum / (current_index - IgnoreFrames);
 }
 

src/video_core/CMakeLists.txt

@@ -22,6 +22,7 @@ add_library(video_core STATIC
     engines/maxwell_dma.h
     engines/shader_bytecode.h
     engines/shader_header.h
+    engines/shader_type.h
     gpu.cpp
     gpu.h
     gpu_asynch.cpp
@@ -127,6 +128,8 @@ add_library(video_core STATIC
     shader/track.cpp
     surface.cpp
     surface.h
+    texture_cache/format_lookup_table.cpp
+    texture_cache/format_lookup_table.h
     texture_cache/surface_base.cpp
     texture_cache/surface_base.h
     texture_cache/surface_params.cpp
@@ -180,3 +183,9 @@ target_link_libraries(video_core PRIVATE glad)
 if (ENABLE_VULKAN)
     target_link_libraries(video_core PRIVATE sirit)
 endif()
+
+if (MSVC)
+    target_compile_options(video_core PRIVATE /we4267)
+else()
+    target_compile_options(video_core PRIVATE -Werror=conversion -Wno-error=sign-conversion)
+endif()

src/video_core/engines/const_buffer_engine_interface.h

@@ -8,19 +8,11 @@
 #include "common/bit_field.h"
 #include "common/common_types.h"
 #include "video_core/engines/shader_bytecode.h"
+#include "video_core/engines/shader_type.h"
 #include "video_core/textures/texture.h"
 
 namespace Tegra::Engines {
 
-enum class ShaderType : u32 {
-    Vertex = 0,
-    TesselationControl = 1,
-    TesselationEval = 2,
-    Geometry = 3,
-    Fragment = 4,
-    Compute = 5,
-};
-
 struct SamplerDescriptor {
     union {
         BitField<0, 20, Tegra::Shader::TextureType> texture_type;

src/video_core/engines/kepler_compute.cpp

@@ -8,6 +8,7 @@
 #include "core/core.h"
 #include "video_core/engines/kepler_compute.h"
 #include "video_core/engines/maxwell_3d.h"
+#include "video_core/engines/shader_type.h"
 #include "video_core/memory_manager.h"
 #include "video_core/rasterizer_interface.h"
 #include "video_core/renderer_base.h"

src/video_core/engines/kepler_compute.h

@@ -12,6 +12,7 @@
 #include "common/common_types.h"
 #include "video_core/engines/const_buffer_engine_interface.h"
 #include "video_core/engines/engine_upload.h"
+#include "video_core/engines/shader_type.h"
 #include "video_core/gpu.h"
 #include "video_core/textures/texture.h"
 
@@ -140,7 +141,7 @@ public:
 
         INSERT_PADDING_WORDS(0x3);
 
-        BitField<0, 16, u32> shared_alloc;
+        BitField<0, 18, u32> shared_alloc;
 
         BitField<16, 16, u32> block_dim_x;
         union {
@@ -178,7 +179,12 @@ public:
             BitField<24, 5, u32> gpr_alloc;
         };
 
-        INSERT_PADDING_WORDS(0x11);
+        union {
+            BitField<0, 20, u32> local_crs_alloc;
+            BitField<24, 5, u32> sass_version;
+        };
+
+        INSERT_PADDING_WORDS(0x10);
     } launch_description{};
 
     struct {

src/video_core/engines/maxwell_3d.cpp

@@ -9,6 +9,7 @@
 #include "core/core_timing.h"
 #include "video_core/debug_utils/debug_utils.h"
 #include "video_core/engines/maxwell_3d.h"
+#include "video_core/engines/shader_type.h"
 #include "video_core/memory_manager.h"
 #include "video_core/rasterizer_interface.h"
 #include "video_core/textures/texture.h"
@@ -261,7 +262,8 @@ void Maxwell3D::CallMacroMethod(u32 method, std::size_t num_parameters, const u3
     executing_macro = 0;
 
     // Lookup the macro offset
-    const u32 entry = ((method - MacroRegistersStart) >> 1) % macro_positions.size();
+    const u32 entry =
+        ((method - MacroRegistersStart) >> 1) % static_cast<u32>(macro_positions.size());
 
     // Execute the current macro.
     macro_interpreter.Execute(macro_positions[entry], num_parameters, parameters);
@@ -367,24 +369,24 @@ void Maxwell3D::CallMethod(const GPU::MethodCall& method_call) {
         StartCBData(method);
         break;
     }
-    case MAXWELL3D_REG_INDEX(cb_bind[0].raw_config): {
-        ProcessCBBind(Regs::ShaderStage::Vertex);
+    case MAXWELL3D_REG_INDEX(cb_bind[0]): {
+        ProcessCBBind(0);
         break;
     }
-    case MAXWELL3D_REG_INDEX(cb_bind[1].raw_config): {
-        ProcessCBBind(Regs::ShaderStage::TesselationControl);
+    case MAXWELL3D_REG_INDEX(cb_bind[1]): {
+        ProcessCBBind(1);
         break;
     }
-    case MAXWELL3D_REG_INDEX(cb_bind[2].raw_config): {
-        ProcessCBBind(Regs::ShaderStage::TesselationEval);
+    case MAXWELL3D_REG_INDEX(cb_bind[2]): {
+        ProcessCBBind(2);
         break;
     }
-    case MAXWELL3D_REG_INDEX(cb_bind[3].raw_config): {
-        ProcessCBBind(Regs::ShaderStage::Geometry);
+    case MAXWELL3D_REG_INDEX(cb_bind[3]): {
+        ProcessCBBind(3);
         break;
     }
-    case MAXWELL3D_REG_INDEX(cb_bind[4].raw_config): {
-        ProcessCBBind(Regs::ShaderStage::Fragment);
+    case MAXWELL3D_REG_INDEX(cb_bind[4]): {
+        ProcessCBBind(4);
         break;
     }
     case MAXWELL3D_REG_INDEX(draw.vertex_end_gl): {
@@ -686,10 +688,10 @@ void Maxwell3D::DrawArrays() {
     }
 }
 
-void Maxwell3D::ProcessCBBind(Regs::ShaderStage stage) {
+void Maxwell3D::ProcessCBBind(std::size_t stage_index) {
     // Bind the buffer currently in CB_ADDRESS to the specified index in the desired shader stage.
-    auto& shader = state.shader_stages[static_cast<std::size_t>(stage)];
-    auto& bind_data = regs.cb_bind[static_cast<std::size_t>(stage)];
+    auto& shader = state.shader_stages[stage_index];
+    auto& bind_data = regs.cb_bind[stage_index];
 
     ASSERT(bind_data.index < Regs::MaxConstBuffers);
     auto& buffer = shader.const_buffers[bind_data.index];
@@ -741,14 +743,6 @@ Texture::TICEntry Maxwell3D::GetTICEntry(u32 tic_index) const {
     Texture::TICEntry tic_entry;
     memory_manager.ReadBlockUnsafe(tic_address_gpu, &tic_entry, sizeof(Texture::TICEntry));
 
-    [[maybe_unused]] const auto r_type{tic_entry.r_type.Value()};
-    [[maybe_unused]] const auto g_type{tic_entry.g_type.Value()};
-    [[maybe_unused]] const auto b_type{tic_entry.b_type.Value()};
-    [[maybe_unused]] const auto a_type{tic_entry.a_type.Value()};
-
-    // TODO(Subv): Different data types for separate components are not supported
-    DEBUG_ASSERT(r_type == g_type && r_type == b_type && r_type == a_type);
-
     return tic_entry;
 }
 
@@ -764,9 +758,9 @@ Texture::FullTextureInfo Maxwell3D::GetTextureInfo(Texture::TextureHandle tex_ha
     return Texture::FullTextureInfo{GetTICEntry(tex_handle.tic_id), GetTSCEntry(tex_handle.tsc_id)};
 }
 
-Texture::FullTextureInfo Maxwell3D::GetStageTexture(Regs::ShaderStage stage,
-                                                    std::size_t offset) const {
-    const auto& shader = state.shader_stages[static_cast<std::size_t>(stage)];
+Texture::FullTextureInfo Maxwell3D::GetStageTexture(ShaderType stage, std::size_t offset) const {
+    const auto stage_index = static_cast<std::size_t>(stage);
+    const auto& shader = state.shader_stages[stage_index];
     const auto& tex_info_buffer = shader.const_buffers[regs.tex_cb_index];
     ASSERT(tex_info_buffer.enabled && tex_info_buffer.address != 0);
 

src/video_core/engines/maxwell_3d.h

@@ -18,6 +18,7 @@
 #include "video_core/engines/const_buffer_engine_interface.h"
 #include "video_core/engines/const_buffer_info.h"
 #include "video_core/engines/engine_upload.h"
+#include "video_core/engines/shader_type.h"
 #include "video_core/gpu.h"
 #include "video_core/macro_interpreter.h"
 #include "video_core/textures/texture.h"
@@ -62,7 +63,6 @@ public:
         static constexpr std::size_t NumVertexArrays = 32;
         static constexpr std::size_t NumVertexAttributes = 32;
         static constexpr std::size_t NumVaryings = 31;
-        static constexpr std::size_t NumTextureSamplers = 32;
         static constexpr std::size_t NumImages = 8; // TODO(Rodrigo): Investigate this number
         static constexpr std::size_t NumClipDistances = 8;
         static constexpr std::size_t MaxShaderProgram = 6;
@@ -130,14 +130,6 @@ public:
             Fragment = 5,
         };
 
-        enum class ShaderStage : u32 {
-            Vertex = 0,
-            TesselationControl = 1,
-            TesselationEval = 2,
-            Geometry = 3,
-            Fragment = 4,
-        };
-
         struct VertexAttribute {
             enum class Size : u32 {
                 Invalid = 0x0,
@@ -1254,7 +1246,7 @@ public:
     Texture::FullTextureInfo GetTextureInfo(Texture::TextureHandle tex_handle) const;
 
     /// Returns the texture information for a specific texture in a specific shader stage.
-    Texture::FullTextureInfo GetStageTexture(Regs::ShaderStage stage, std::size_t offset) const;
+    Texture::FullTextureInfo GetStageTexture(ShaderType stage, std::size_t offset) const;
 
     u32 AccessConstBuffer32(ShaderType stage, u64 const_buffer, u64 offset) const override;
 
@@ -1376,7 +1368,7 @@ private:
     void FinishCBData();
 
     /// Handles a write to the CB_BIND register.
-    void ProcessCBBind(Regs::ShaderStage stage);
+    void ProcessCBBind(std::size_t stage_index);
 
     /// Handles a write to the VERTEX_END_GL register, triggering a draw.
     void DrawArrays();

src/video_core/engines/shader_bytecode.h

@@ -615,6 +615,14 @@ union Instruction {
         BitField<34, 13, u64> mask_imm;
     } shfl;
 
+    union {
+        BitField<44, 1, u64> ftz;
+        BitField<39, 2, u64> tab5cb8_2;
+        BitField<38, 1, u64> ndv;
+        BitField<47, 1, u64> cc;
+        BitField<28, 8, u64> swizzle;
+    } fswzadd;
+
     union {
         BitField<8, 8, Register> gpr;
         BitField<20, 24, s64> offset;
@@ -1478,7 +1486,8 @@ union Instruction {
             u32 value = static_cast<u32>(target);
             // The branch offset is relative to the next instruction and is stored in bytes, so
             // divide it by the size of an instruction and add 1 to it.
-            return static_cast<s32>((value ^ mask) - mask) / sizeof(Instruction) + 1;
+            return static_cast<s32>((value ^ mask) - mask) / static_cast<s32>(sizeof(Instruction)) +
+                   1;
         }
     } bra;
 
@@ -1492,7 +1501,8 @@ union Instruction {
             u32 value = static_cast<u32>(target);
             // The branch offset is relative to the next instruction and is stored in bytes, so
             // divide it by the size of an instruction and add 1 to it.
-            return static_cast<s32>((value ^ mask) - mask) / sizeof(Instruction) + 1;
+            return static_cast<s32>((value ^ mask) - mask) / static_cast<s32>(sizeof(Instruction)) +
+                   1;
         }
     } brx;
 
@@ -1590,6 +1600,7 @@ public:
         DEPBAR,
         VOTE,
         SHFL,
+        FSWZADD,
         BFE_C,
         BFE_R,
         BFE_IMM,
@@ -1851,11 +1862,11 @@ private:
                 const std::size_t bit_position = opcode_bitsize - i - 1;
                 switch (bitstring[i]) {
                 case '0':
-                    mask |= 1 << bit_position;
+                    mask |= static_cast<u16>(1U << bit_position);
                     break;
                 case '1':
-                    expect |= 1 << bit_position;
-                    mask |= 1 << bit_position;
+                    expect |= static_cast<u16>(1U << bit_position);
+                    mask |= static_cast<u16>(1U << bit_position);
                     break;
                 default:
                     // Ignore
@@ -1888,6 +1899,7 @@ private:
             INST("1111000011110---", Id::DEPBAR, Type::Synch, "DEPBAR"),
             INST("0101000011011---", Id::VOTE, Type::Warp, "VOTE"),
             INST("1110111100010---", Id::SHFL, Type::Warp, "SHFL"),
+            INST("0101000011111---", Id::FSWZADD, Type::Warp, "FSWZADD"),
             INST("1110111111011---", Id::LD_A, Type::Memory, "LD_A"),
             INST("1110111101001---", Id::LD_S, Type::Memory, "LD_S"),
             INST("1110111101000---", Id::LD_L, Type::Memory, "LD_L"),

src/video_core/engines/shader_type.h

@@ -0,0 +1,21 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "common/common_types.h"
+
+namespace Tegra::Engines {
+
+enum class ShaderType : u32 {
+    Vertex = 0,
+    TesselationControl = 1,
+    TesselationEval = 2,
+    Geometry = 3,
+    Fragment = 4,
+    Compute = 5,
+};
+static constexpr std::size_t MaxShaderTypes = 6;
+
+} // namespace Tegra::Engines

src/video_core/renderer_opengl/gl_device.cpp

@@ -5,7 +5,9 @@
 #include <algorithm>
 #include <array>
 #include <cstddef>
+#include <optional>
 #include <vector>
+
 #include <glad/glad.h>
 
 #include "common/logging/log.h"
@@ -17,6 +19,30 @@ namespace OpenGL {
 
 namespace {
 
+// One uniform block is reserved for emulation purposes
+constexpr u32 ReservedUniformBlocks = 1;
+
+constexpr u32 NumStages = 5;
+
+constexpr std::array LimitUBOs = {GL_MAX_VERTEX_UNIFORM_BLOCKS, GL_MAX_TESS_CONTROL_UNIFORM_BLOCKS,
+                                  GL_MAX_TESS_EVALUATION_UNIFORM_BLOCKS,
+                                  GL_MAX_GEOMETRY_UNIFORM_BLOCKS, GL_MAX_FRAGMENT_UNIFORM_BLOCKS};
+
+constexpr std::array LimitSSBOs = {
+    GL_MAX_VERTEX_SHADER_STORAGE_BLOCKS, GL_MAX_TESS_CONTROL_SHADER_STORAGE_BLOCKS,
+    GL_MAX_TESS_EVALUATION_SHADER_STORAGE_BLOCKS, GL_MAX_GEOMETRY_SHADER_STORAGE_BLOCKS,
+    GL_MAX_FRAGMENT_SHADER_STORAGE_BLOCKS};
+
+constexpr std::array LimitSamplers = {
+    GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS, GL_MAX_TESS_CONTROL_TEXTURE_IMAGE_UNITS,
+    GL_MAX_TESS_EVALUATION_TEXTURE_IMAGE_UNITS, GL_MAX_GEOMETRY_TEXTURE_IMAGE_UNITS,
+    GL_MAX_TEXTURE_IMAGE_UNITS};
+
+constexpr std::array LimitImages = {GL_MAX_VERTEX_IMAGE_UNIFORMS,
+                                    GL_MAX_TESS_CONTROL_IMAGE_UNIFORMS,
+                                    GL_MAX_TESS_EVALUATION_IMAGE_UNIFORMS,
+                                    GL_MAX_GEOMETRY_IMAGE_UNIFORMS, GL_MAX_FRAGMENT_IMAGE_UNIFORMS};
+
 template <typename T>
 T GetInteger(GLenum pname) {
     GLint temporary;
@@ -48,13 +74,70 @@ bool HasExtension(const std::vector<std::string_view>& images, std::string_view
     return std::find(images.begin(), images.end(), extension) != images.end();
 }
 
+u32 Extract(u32& base, u32& num, u32 amount, std::optional<GLenum> limit = {}) {
+    ASSERT(num >= amount);
+    if (limit) {
+        amount = std::min(amount, GetInteger<u32>(*limit));
+    }
+    num -= amount;
+    return std::exchange(base, base + amount);
+}
+
+std::array<Device::BaseBindings, Tegra::Engines::MaxShaderTypes> BuildBaseBindings() noexcept {
+    std::array<Device::BaseBindings, Tegra::Engines::MaxShaderTypes> bindings;
+
+    static std::array<std::size_t, 5> stage_swizzle = {0, 1, 2, 3, 4};
+    const u32 total_ubos = GetInteger<u32>(GL_MAX_UNIFORM_BUFFER_BINDINGS);
+    const u32 total_ssbos = GetInteger<u32>(GL_MAX_SHADER_STORAGE_BUFFER_BINDINGS);
+    const u32 total_samplers = GetInteger<u32>(GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS);
+
+    u32 num_ubos = total_ubos - ReservedUniformBlocks;
+    u32 num_ssbos = total_ssbos;
+    u32 num_samplers = total_samplers;
+
+    u32 base_ubo = ReservedUniformBlocks;
+    u32 base_ssbo = 0;
+    u32 base_samplers = 0;
+
+    for (std::size_t i = 0; i < NumStages; ++i) {
+        const std::size_t stage = stage_swizzle[i];
+        bindings[stage] = {
+            Extract(base_ubo, num_ubos, total_ubos / NumStages, LimitUBOs[stage]),
+            Extract(base_ssbo, num_ssbos, total_ssbos / NumStages, LimitSSBOs[stage]),
+            Extract(base_samplers, num_samplers, total_samplers / NumStages, LimitSamplers[stage])};
+    }
+
+    u32 num_images = GetInteger<u32>(GL_MAX_IMAGE_UNITS);
+    u32 base_images = 0;
+
+    // Reserve more image bindings on fragment and vertex stages.
+    bindings[4].image =
+        Extract(base_images, num_images, num_images / NumStages + 2, LimitImages[4]);
+    bindings[0].image =
+        Extract(base_images, num_images, num_images / NumStages + 1, LimitImages[0]);
+
+    // Reserve the other image bindings.
+    const u32 total_extracted_images = num_images / (NumStages - 2);
+    for (std::size_t i = 2; i < NumStages; ++i) {
+        const std::size_t stage = stage_swizzle[i];
+        bindings[stage].image =
+            Extract(base_images, num_images, total_extracted_images, LimitImages[stage]);
+    }
+
+    // Compute doesn't care about any of this.
+    bindings[5] = {0, 0, 0, 0};
+
+    return bindings;
+}
+
 } // Anonymous namespace
 
-Device::Device() {
+Device::Device() : base_bindings{BuildBaseBindings()} {
     const std::string_view vendor = reinterpret_cast<const char*>(glGetString(GL_VENDOR));
     const std::vector extensions = GetExtensions();
 
     const bool is_nvidia = vendor == "NVIDIA Corporation";
+    const bool is_intel = vendor == "Intel";
 
     uniform_buffer_alignment = GetInteger<std::size_t>(GL_UNIFORM_BUFFER_OFFSET_ALIGNMENT);
     shader_storage_alignment = GetInteger<std::size_t>(GL_SHADER_STORAGE_BUFFER_OFFSET_ALIGNMENT);
@@ -62,11 +145,13 @@ Device::Device() {
     max_varyings = GetInteger<u32>(GL_MAX_VARYING_VECTORS);
     has_warp_intrinsics = GLAD_GL_NV_gpu_shader5 && GLAD_GL_NV_shader_thread_group &&
                           GLAD_GL_NV_shader_thread_shuffle;
+    has_shader_ballot = GLAD_GL_ARB_shader_ballot;
     has_vertex_viewport_layer = GLAD_GL_ARB_shader_viewport_layer_array;
     has_image_load_formatted = HasExtension(extensions, "GL_EXT_shader_image_load_formatted");
     has_variable_aoffi = TestVariableAoffi();
     has_component_indexing_bug = TestComponentIndexingBug();
     has_precise_bug = TestPreciseBug();
+    has_broken_compute = is_intel;
     has_fast_buffer_sub_data = is_nvidia;
 
     LOG_INFO(Render_OpenGL, "Renderer_VariableAOFFI: {}", has_variable_aoffi);
@@ -79,10 +164,12 @@ Device::Device(std::nullptr_t) {
     max_vertex_attributes = 16;
     max_varyings = 15;
     has_warp_intrinsics = true;
+    has_shader_ballot = true;
     has_vertex_viewport_layer = true;
     has_image_load_formatted = true;
     has_variable_aoffi = true;
     has_component_indexing_bug = false;
+    has_broken_compute = false;
     has_precise_bug = false;
 }
 

src/video_core/renderer_opengl/gl_device.h

@@ -6,14 +6,32 @@
 
 #include <cstddef>
 #include "common/common_types.h"
+#include "video_core/engines/shader_type.h"
 
 namespace OpenGL {
 
-class Device {
+static constexpr u32 EmulationUniformBlockBinding = 0;
+
+class Device final {
 public:
+    struct BaseBindings final {
+        u32 uniform_buffer{};
+        u32 shader_storage_buffer{};
+        u32 sampler{};
+        u32 image{};
+    };
+
     explicit Device();
     explicit Device(std::nullptr_t);
 
+    const BaseBindings& GetBaseBindings(std::size_t stage_index) const noexcept {
+        return base_bindings[stage_index];
+    }
+
+    const BaseBindings& GetBaseBindings(Tegra::Engines::ShaderType shader_type) const noexcept {
+        return GetBaseBindings(static_cast<std::size_t>(shader_type));
+    }
+
     std::size_t GetUniformBufferAlignment() const {
         return uniform_buffer_alignment;
     }
@@ -34,6 +52,10 @@ public:
         return has_warp_intrinsics;
     }
 
+    bool HasShaderBallot() const {
+        return has_shader_ballot;
+    }
+
     bool HasVertexViewportLayer() const {
         return has_vertex_viewport_layer;
     }
@@ -54,6 +76,10 @@ public:
         return has_precise_bug;
     }
 
+    bool HasBrokenCompute() const {
+        return has_broken_compute;
+    }
+
     bool HasFastBufferSubData() const {
         return has_fast_buffer_sub_data;
     }
@@ -63,16 +89,19 @@ private:
     static bool TestComponentIndexingBug();
     static bool TestPreciseBug();
 
+    std::array<BaseBindings, Tegra::Engines::MaxShaderTypes> base_bindings;
     std::size_t uniform_buffer_alignment{};
     std::size_t shader_storage_alignment{};
     u32 max_vertex_attributes{};
     u32 max_varyings{};
     bool has_warp_intrinsics{};
+    bool has_shader_ballot{};
     bool has_vertex_viewport_layer{};
     bool has_image_load_formatted{};
     bool has_variable_aoffi{};
     bool has_component_indexing_bug{};
     bool has_precise_bug{};
+    bool has_broken_compute{};
     bool has_fast_buffer_sub_data{};
 };
 

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -22,6 +22,7 @@
 #include "core/settings.h"
 #include "video_core/engines/kepler_compute.h"
 #include "video_core/engines/maxwell_3d.h"
+#include "video_core/engines/shader_type.h"
 #include "video_core/memory_manager.h"
 #include "video_core/renderer_opengl/gl_rasterizer.h"
 #include "video_core/renderer_opengl/gl_shader_cache.h"
@@ -49,8 +50,25 @@ MICROPROFILE_DEFINE(OpenGL_Blits, "OpenGL", "Blits", MP_RGB(128, 128, 192));
 MICROPROFILE_DEFINE(OpenGL_CacheManagement, "OpenGL", "Cache Mgmt", MP_RGB(100, 255, 100));
 MICROPROFILE_DEFINE(OpenGL_PrimitiveAssembly, "OpenGL", "Prim Asmbl", MP_RGB(255, 100, 100));
 
-static std::size_t GetConstBufferSize(const Tegra::Engines::ConstBufferInfo& buffer,
-                                      const GLShader::ConstBufferEntry& entry) {
+namespace {
+
+template <typename Engine, typename Entry>
+Tegra::Texture::FullTextureInfo GetTextureInfo(const Engine& engine, const Entry& entry,
+                                               Tegra::Engines::ShaderType shader_type) {
+    if (entry.IsBindless()) {
+        const Tegra::Texture::TextureHandle tex_handle =
+            engine.AccessConstBuffer32(shader_type, entry.GetBuffer(), entry.GetOffset());
+        return engine.GetTextureInfo(tex_handle);
+    }
+    if constexpr (std::is_same_v<Engine, Tegra::Engines::Maxwell3D>) {
+        return engine.GetStageTexture(shader_type, entry.GetOffset());
+    } else {
+        return engine.GetTexture(entry.GetOffset());
+    }
+}
+
+std::size_t GetConstBufferSize(const Tegra::Engines::ConstBufferInfo& buffer,
+                               const GLShader::ConstBufferEntry& entry) {
     if (!entry.IsIndirect()) {
         return entry.GetSize();
     }
@@ -64,6 +82,8 @@ static std::size_t GetConstBufferSize(const Tegra::Engines::ConstBufferInfo& buf
     return buffer.size;
 }
 
+} // Anonymous namespace
+
 RasterizerOpenGL::RasterizerOpenGL(Core::System& system, Core::Frontend::EmuWindow& emu_window,
                                    ScreenInfo& info)
     : texture_cache{system, *this, device}, shader_cache{*this, system, emu_window, device},
@@ -238,12 +258,11 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
     MICROPROFILE_SCOPE(OpenGL_Shader);
     auto& gpu = system.GPU().Maxwell3D();
 
-    BaseBindings base_bindings;
     std::array<bool, Maxwell::NumClipDistances> clip_distances{};
 
     for (std::size_t index = 0; index < Maxwell::MaxShaderProgram; ++index) {
         const auto& shader_config = gpu.regs.shader_config[index];
-        const Maxwell::ShaderProgram program{static_cast<Maxwell::ShaderProgram>(index)};
+        const auto program{static_cast<Maxwell::ShaderProgram>(index)};
 
         // Skip stages that are not enabled
         if (!gpu.regs.IsShaderConfigEnabled(index)) {
@@ -257,25 +276,17 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
             continue;
         }
 
-        const std::size_t stage{index == 0 ? 0 : index - 1}; // Stage indices are 0 - 5
-
-        GLShader::MaxwellUniformData ubo{};
-        ubo.SetFromRegs(gpu, stage);
-        const auto [buffer, offset] =
-            buffer_cache.UploadHostMemory(&ubo, sizeof(ubo), device.GetUniformBufferAlignment());
-
-        // Bind the emulation info buffer
-        bind_ubo_pushbuffer.Push(buffer, offset, static_cast<GLsizeiptr>(sizeof(ubo)));
-
         Shader shader{shader_cache.GetStageProgram(program)};
 
-        const auto stage_enum = static_cast<Maxwell::ShaderStage>(stage);
-        SetupDrawConstBuffers(stage_enum, shader);
-        SetupDrawGlobalMemory(stage_enum, shader);
-        const auto texture_buffer_usage{SetupDrawTextures(stage_enum, shader, base_bindings)};
+        // Stage indices are 0 - 5
+        const std::size_t stage = index == 0 ? 0 : index - 1;
+        SetupDrawConstBuffers(stage, shader);
+        SetupDrawGlobalMemory(stage, shader);
+        SetupDrawTextures(stage, shader);
+        SetupDrawImages(stage, shader);
 
-        const ProgramVariant variant{base_bindings, primitive_mode, texture_buffer_usage};
-        const auto [program_handle, next_bindings] = shader->GetProgramHandle(variant);
+        const ProgramVariant variant(primitive_mode);
+        const auto program_handle = shader->GetHandle(variant);
 
         switch (program) {
         case Maxwell::ShaderProgram::VertexA:
@@ -304,10 +315,8 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
         // When VertexA is enabled, we have dual vertex shaders
         if (program == Maxwell::ShaderProgram::VertexA) {
             // VertexB was combined with VertexA, so we skip the VertexB iteration
-            index++;
+            ++index;
         }
-
-        base_bindings = next_bindings;
     }
 
     SyncClipEnabled(clip_distances);
@@ -375,7 +384,7 @@ void RasterizerOpenGL::ConfigureFramebuffers() {
         fbkey.color_attachments[index] = GL_COLOR_ATTACHMENT0 + regs.rt_control.GetMap(index);
         fbkey.colors[index] = std::move(color_surface);
     }
-    fbkey.colors_count = regs.rt_control.count;
+    fbkey.colors_count = static_cast<u16>(regs.rt_control.count);
 
     if (depth_surface) {
         // Assume that a surface will be written to if it is used as a framebuffer, even if
@@ -592,8 +601,16 @@ void RasterizerOpenGL::DrawPrelude() {
     index_buffer_offset = SetupIndexBuffer();
 
     // Prepare packed bindings.
-    bind_ubo_pushbuffer.Setup(0);
-    bind_ssbo_pushbuffer.Setup(0);
+    bind_ubo_pushbuffer.Setup();
+    bind_ssbo_pushbuffer.Setup();
+
+    // Setup emulation uniform buffer.
+    GLShader::MaxwellUniformData ubo;
+    ubo.SetFromRegs(gpu);
+    const auto [buffer, offset] =
+        buffer_cache.UploadHostMemory(&ubo, sizeof(ubo), device.GetUniformBufferAlignment());
+    bind_ubo_pushbuffer.Push(EmulationUniformBlockBinding, buffer, offset,
+                             static_cast<GLsizeiptr>(sizeof(ubo)));
 
     // Setup shaders and their used resources.
     texture_cache.GuardSamplers(true);
@@ -726,19 +743,21 @@ bool RasterizerOpenGL::DrawMultiBatch(bool is_indexed) {
 }
 
 void RasterizerOpenGL::DispatchCompute(GPUVAddr code_addr) {
-    if (!GLAD_GL_ARB_compute_variable_group_size) {
-        LOG_ERROR(Render_OpenGL, "Compute is currently not supported on this device due to the "
-                                 "lack of GL_ARB_compute_variable_group_size");
+    if (device.HasBrokenCompute()) {
         return;
     }
 
+    buffer_cache.Acquire();
+
     auto kernel = shader_cache.GetComputeKernel(code_addr);
-    ProgramVariant variant;
-    variant.texture_buffer_usage = SetupComputeTextures(kernel);
+    SetupComputeTextures(kernel);
     SetupComputeImages(kernel);
 
-    const auto [program, next_bindings] = kernel->GetProgramHandle(variant);
-    state.draw.shader_program = program;
+    const auto& launch_desc = system.GPU().KeplerCompute().launch_description;
+    const ProgramVariant variant(launch_desc.block_dim_x, launch_desc.block_dim_y,
+                                 launch_desc.block_dim_z, launch_desc.shared_alloc,
+                                 launch_desc.local_pos_alloc);
+    state.draw.shader_program = kernel->GetHandle(variant);
     state.draw.program_pipeline = 0;
 
     const std::size_t buffer_size =
@@ -746,8 +765,8 @@ void RasterizerOpenGL::DispatchCompute(GPUVAddr code_addr) {
         (Maxwell::MaxConstBufferSize + device.GetUniformBufferAlignment());
     buffer_cache.Map(buffer_size);
 
-    bind_ubo_pushbuffer.Setup(0);
-    bind_ssbo_pushbuffer.Setup(0);
+    bind_ubo_pushbuffer.Setup();
+    bind_ssbo_pushbuffer.Setup();
 
     SetupComputeConstBuffers(kernel);
     SetupComputeGlobalMemory(kernel);
@@ -762,10 +781,7 @@ void RasterizerOpenGL::DispatchCompute(GPUVAddr code_addr) {
     state.ApplyShaderProgram();
     state.ApplyProgramPipeline();
 
-    const auto& launch_desc = system.GPU().KeplerCompute().launch_description;
-    glDispatchComputeGroupSizeARB(launch_desc.grid_dim_x, launch_desc.grid_dim_y,
-                                  launch_desc.grid_dim_z, launch_desc.block_dim_x,
-                                  launch_desc.block_dim_y, launch_desc.block_dim_z);
+    glDispatchCompute(launch_desc.grid_dim_x, launch_desc.grid_dim_y, launch_desc.grid_dim_z);
 }
 
 void RasterizerOpenGL::FlushAll() {}
@@ -834,7 +850,7 @@ bool RasterizerOpenGL::AccelerateDisplay(const Tegra::FramebufferConfig& config,
     ASSERT_MSG(params.height == config.height, "Framebuffer height is different");
 
     if (params.pixel_format != pixel_format) {
-        LOG_WARNING(Render_OpenGL, "Framebuffer pixel_format is different");
+        LOG_DEBUG(Render_OpenGL, "Framebuffer pixel_format is different");
     }
 
     screen_info.display_texture = surface->GetTexture();
@@ -843,20 +859,23 @@ bool RasterizerOpenGL::AccelerateDisplay(const Tegra::FramebufferConfig& config,
     return true;
 }
 
-void RasterizerOpenGL::SetupDrawConstBuffers(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage,
-                                             const Shader& shader) {
+void RasterizerOpenGL::SetupDrawConstBuffers(std::size_t stage_index, const Shader& shader) {
     MICROPROFILE_SCOPE(OpenGL_UBO);
     const auto& stages = system.GPU().Maxwell3D().state.shader_stages;
-    const auto& shader_stage = stages[static_cast<std::size_t>(stage)];
+    const auto& shader_stage = stages[stage_index];
+
+    u32 binding = device.GetBaseBindings(stage_index).uniform_buffer;
     for (const auto& entry : shader->GetShaderEntries().const_buffers) {
         const auto& buffer = shader_stage.const_buffers[entry.GetIndex()];
-        SetupConstBuffer(buffer, entry);
+        SetupConstBuffer(binding++, buffer, entry);
     }
 }
 
 void RasterizerOpenGL::SetupComputeConstBuffers(const Shader& kernel) {
     MICROPROFILE_SCOPE(OpenGL_UBO);
     const auto& launch_desc = system.GPU().KeplerCompute().launch_description;
+
+    u32 binding = 0;
     for (const auto& entry : kernel->GetShaderEntries().const_buffers) {
         const auto& config = launch_desc.const_buffer_config[entry.GetIndex()];
         const std::bitset<8> mask = launch_desc.const_buffer_enable_mask.Value();
@@ -864,15 +883,16 @@ void RasterizerOpenGL::SetupComputeConstBuffers(const Shader& kernel) {
         buffer.address = config.Address();
         buffer.size = config.size;
         buffer.enabled = mask[entry.GetIndex()];
-        SetupConstBuffer(buffer, entry);
+        SetupConstBuffer(binding++, buffer, entry);
     }
 }
 
-void RasterizerOpenGL::SetupConstBuffer(const Tegra::Engines::ConstBufferInfo& buffer,
+void RasterizerOpenGL::SetupConstBuffer(u32 binding, const Tegra::Engines::ConstBufferInfo& buffer,
                                         const GLShader::ConstBufferEntry& entry) {
     if (!buffer.enabled) {
         // Set values to zero to unbind buffers
-        bind_ubo_pushbuffer.Push(buffer_cache.GetEmptyBuffer(sizeof(float)), 0, sizeof(float));
+        bind_ubo_pushbuffer.Push(binding, buffer_cache.GetEmptyBuffer(sizeof(float)), 0,
+                                 sizeof(float));
         return;
     }
 
@@ -883,19 +903,20 @@ void RasterizerOpenGL::SetupConstBuffer(const Tegra::Engines::ConstBufferInfo& b
     const auto alignment = device.GetUniformBufferAlignment();
     const auto [cbuf, offset] = buffer_cache.UploadMemory(buffer.address, size, alignment, false,
                                                           device.HasFastBufferSubData());
-    bind_ubo_pushbuffer.Push(cbuf, offset, size);
+    bind_ubo_pushbuffer.Push(binding, cbuf, offset, size);
 }
 
-void RasterizerOpenGL::SetupDrawGlobalMemory(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage,
-                                             const Shader& shader) {
+void RasterizerOpenGL::SetupDrawGlobalMemory(std::size_t stage_index, const Shader& shader) {
     auto& gpu{system.GPU()};
     auto& memory_manager{gpu.MemoryManager()};
-    const auto cbufs{gpu.Maxwell3D().state.shader_stages[static_cast<std::size_t>(stage)]};
+    const auto cbufs{gpu.Maxwell3D().state.shader_stages[stage_index]};
+
+    u32 binding = device.GetBaseBindings(stage_index).shader_storage_buffer;
     for (const auto& entry : shader->GetShaderEntries().global_memory_entries) {
         const auto addr{cbufs.const_buffers[entry.GetCbufIndex()].address + entry.GetCbufOffset()};
         const auto gpu_addr{memory_manager.Read<u64>(addr)};
         const auto size{memory_manager.Read<u32>(addr + 8)};
-        SetupGlobalMemory(entry, gpu_addr, size);
+        SetupGlobalMemory(binding++, entry, gpu_addr, size);
     }
 }
 
@@ -903,120 +924,82 @@ void RasterizerOpenGL::SetupComputeGlobalMemory(const Shader& kernel) {
     auto& gpu{system.GPU()};
     auto& memory_manager{gpu.MemoryManager()};
     const auto cbufs{gpu.KeplerCompute().launch_description.const_buffer_config};
+
+    u32 binding = 0;
     for (const auto& entry : kernel->GetShaderEntries().global_memory_entries) {
         const auto addr{cbufs[entry.GetCbufIndex()].Address() + entry.GetCbufOffset()};
         const auto gpu_addr{memory_manager.Read<u64>(addr)};
         const auto size{memory_manager.Read<u32>(addr + 8)};
-        SetupGlobalMemory(entry, gpu_addr, size);
+        SetupGlobalMemory(binding++, entry, gpu_addr, size);
     }
 }
 
-void RasterizerOpenGL::SetupGlobalMemory(const GLShader::GlobalMemoryEntry& entry,
+void RasterizerOpenGL::SetupGlobalMemory(u32 binding, const GLShader::GlobalMemoryEntry& entry,
                                          GPUVAddr gpu_addr, std::size_t size) {
     const auto alignment{device.GetShaderStorageBufferAlignment()};
     const auto [ssbo, buffer_offset] =
         buffer_cache.UploadMemory(gpu_addr, size, alignment, entry.IsWritten());
-    bind_ssbo_pushbuffer.Push(ssbo, buffer_offset, static_cast<GLsizeiptr>(size));
+    bind_ssbo_pushbuffer.Push(binding, ssbo, buffer_offset, static_cast<GLsizeiptr>(size));
 }
 
-TextureBufferUsage RasterizerOpenGL::SetupDrawTextures(Maxwell::ShaderStage stage,
-                                                       const Shader& shader,
-                                                       BaseBindings base_bindings) {
+void RasterizerOpenGL::SetupDrawTextures(std::size_t stage_index, const Shader& shader) {
     MICROPROFILE_SCOPE(OpenGL_Texture);
-    const auto& gpu = system.GPU();
-    const auto& maxwell3d = gpu.Maxwell3D();
-    const auto& entries = shader->GetShaderEntries().samplers;
-
-    ASSERT_MSG(base_bindings.sampler + entries.size() <= std::size(state.textures),
-               "Exceeded the number of active textures.");
-
-    TextureBufferUsage texture_buffer_usage{0};
-
-    for (u32 bindpoint = 0; bindpoint < entries.size(); ++bindpoint) {
-        const auto& entry = entries[bindpoint];
-        const auto texture = [&] {
-            if (!entry.IsBindless()) {
-                return maxwell3d.GetStageTexture(stage, entry.GetOffset());
-            }
-            const auto shader_type = static_cast<Tegra::Engines::ShaderType>(stage);
-            const Tegra::Texture::TextureHandle tex_handle =
-                maxwell3d.AccessConstBuffer32(shader_type, entry.GetBuffer(), entry.GetOffset());
-            return maxwell3d.GetTextureInfo(tex_handle);
-        }();
-
-        if (SetupTexture(base_bindings.sampler + bindpoint, texture, entry)) {
-            texture_buffer_usage.set(bindpoint);
-        }
+    const auto& maxwell3d = system.GPU().Maxwell3D();
+    u32 binding = device.GetBaseBindings(stage_index).sampler;
+    for (const auto& entry : shader->GetShaderEntries().samplers) {
+        const auto shader_type = static_cast<Tegra::Engines::ShaderType>(stage_index);
+        const auto texture = GetTextureInfo(maxwell3d, entry, shader_type);
+        SetupTexture(binding++, texture, entry);
     }
-
-    return texture_buffer_usage;
 }
 
-TextureBufferUsage RasterizerOpenGL::SetupComputeTextures(const Shader& kernel) {
+void RasterizerOpenGL::SetupComputeTextures(const Shader& kernel) {
     MICROPROFILE_SCOPE(OpenGL_Texture);
     const auto& compute = system.GPU().KeplerCompute();
-    const auto& entries = kernel->GetShaderEntries().samplers;
-
-    ASSERT_MSG(entries.size() <= std::size(state.textures),
-               "Exceeded the number of active textures.");
-
-    TextureBufferUsage texture_buffer_usage{0};
-
-    for (u32 bindpoint = 0; bindpoint < entries.size(); ++bindpoint) {
-        const auto& entry = entries[bindpoint];
-        const auto texture = [&] {
-            if (!entry.IsBindless()) {
-                return compute.GetTexture(entry.GetOffset());
-            }
-            const Tegra::Texture::TextureHandle tex_handle = compute.AccessConstBuffer32(
-                Tegra::Engines::ShaderType::Compute, entry.GetBuffer(), entry.GetOffset());
-            return compute.GetTextureInfo(tex_handle);
-        }();
-
-        if (SetupTexture(bindpoint, texture, entry)) {
-            texture_buffer_usage.set(bindpoint);
-        }
+    u32 binding = 0;
+    for (const auto& entry : kernel->GetShaderEntries().samplers) {
+        const auto texture = GetTextureInfo(compute, entry, Tegra::Engines::ShaderType::Compute);
+        SetupTexture(binding++, texture, entry);
     }
-
-    return texture_buffer_usage;
 }
 
-bool RasterizerOpenGL::SetupTexture(u32 binding, const Tegra::Texture::FullTextureInfo& texture,
+void RasterizerOpenGL::SetupTexture(u32 binding, const Tegra::Texture::FullTextureInfo& texture,
                                     const GLShader::SamplerEntry& entry) {
-    state.samplers[binding] = sampler_cache.GetSampler(texture.tsc);
-
     const auto view = texture_cache.GetTextureSurface(texture.tic, entry);
     if (!view) {
         // Can occur when texture addr is null or its memory is unmapped/invalid
+        state.samplers[binding] = 0;
         state.textures[binding] = 0;
-        return false;
+        return;
     }
     state.textures[binding] = view->GetTexture();
 
     if (view->GetSurfaceParams().IsBuffer()) {
-        return true;
+        return;
     }
+    state.samplers[binding] = sampler_cache.GetSampler(texture.tsc);
 
     // Apply swizzle to textures that are not buffers.
     view->ApplySwizzle(texture.tic.x_source, texture.tic.y_source, texture.tic.z_source,
                        texture.tic.w_source);
-    return false;
+}
+
+void RasterizerOpenGL::SetupDrawImages(std::size_t stage_index, const Shader& shader) {
+    const auto& maxwell3d = system.GPU().Maxwell3D();
+    u32 binding = device.GetBaseBindings(stage_index).image;
+    for (const auto& entry : shader->GetShaderEntries().images) {
+        const auto shader_type = static_cast<Tegra::Engines::ShaderType>(stage_index);
+        const auto tic = GetTextureInfo(maxwell3d, entry, shader_type).tic;
+        SetupImage(binding++, tic, entry);
+    }
 }
 
 void RasterizerOpenGL::SetupComputeImages(const Shader& shader) {
     const auto& compute = system.GPU().KeplerCompute();
-    const auto& entries = shader->GetShaderEntries().images;
-    for (u32 bindpoint = 0; bindpoint < entries.size(); ++bindpoint) {
-        const auto& entry = entries[bindpoint];
-        const auto tic = [&] {
-            if (!entry.IsBindless()) {
-                return compute.GetTexture(entry.GetOffset()).tic;
-            }
-            const Tegra::Texture::TextureHandle tex_handle = compute.AccessConstBuffer32(
-                Tegra::Engines::ShaderType::Compute, entry.GetBuffer(), entry.GetOffset());
-            return compute.GetTextureInfo(tex_handle).tic;
-        }();
-        SetupImage(bindpoint, tic, entry);
+    u32 binding = 0;
+    for (const auto& entry : shader->GetShaderEntries().images) {
+        const auto tic = GetTextureInfo(compute, entry, Tegra::Engines::ShaderType::Compute).tic;
+        SetupImage(binding++, tic, entry);
     }
 }
 
@@ -1055,6 +1038,15 @@ void RasterizerOpenGL::SyncViewport(OpenGLState& current_state) {
     }
     state.depth_clamp.far_plane = regs.view_volume_clip_control.depth_clamp_far != 0;
     state.depth_clamp.near_plane = regs.view_volume_clip_control.depth_clamp_near != 0;
+
+    bool flip_y = false;
+    if (regs.viewport_transform[0].scale_y < 0.0) {
+        flip_y = !flip_y;
+    }
+    if (regs.screen_y_control.y_negate != 0) {
+        flip_y = !flip_y;
+    }
+    state.clip_control.origin = flip_y ? GL_UPPER_LEFT : GL_LOWER_LEFT;
 }
 
 void RasterizerOpenGL::SyncClipEnabled(
@@ -1077,28 +1069,14 @@ void RasterizerOpenGL::SyncClipCoef() {
 }
 
 void RasterizerOpenGL::SyncCullMode() {
-    auto& maxwell3d = system.GPU().Maxwell3D();
-
-    const auto& regs = maxwell3d.regs;
+    const auto& regs = system.GPU().Maxwell3D().regs;
 
     state.cull.enabled = regs.cull.enabled != 0;
     if (state.cull.enabled) {
-        state.cull.front_face = MaxwellToGL::FrontFace(regs.cull.front_face);
         state.cull.mode = MaxwellToGL::CullFace(regs.cull.cull_face);
-
-        const bool flip_triangles{regs.screen_y_control.triangle_rast_flip == 0 ||
-                                  regs.viewport_transform[0].scale_y < 0.0f};
-
-        // If the GPU is configured to flip the rasterized triangles, then we need to flip the
-        // notion of front and back. Note: We flip the triangles when the value of the register is 0
-        // because OpenGL already does it for us.
-        if (flip_triangles) {
-            if (state.cull.front_face == GL_CCW)
-                state.cull.front_face = GL_CW;
-            else if (state.cull.front_face == GL_CW)
-                state.cull.front_face = GL_CCW;
-        }
     }
+
+    state.cull.front_face = MaxwellToGL::FrontFace(regs.cull.front_face);
 }
 
 void RasterizerOpenGL::SyncPrimitiveRestart() {

src/video_core/renderer_opengl/gl_rasterizer.h

@@ -83,42 +83,41 @@ private:
                                    bool using_depth_fb, bool using_stencil_fb);
 
     /// Configures the current constbuffers to use for the draw command.
-    void SetupDrawConstBuffers(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage,
-                               const Shader& shader);
+    void SetupDrawConstBuffers(std::size_t stage_index, const Shader& shader);
 
     /// Configures the current constbuffers to use for the kernel invocation.
     void SetupComputeConstBuffers(const Shader& kernel);
 
     /// Configures a constant buffer.
-    void SetupConstBuffer(const Tegra::Engines::ConstBufferInfo& buffer,
+    void SetupConstBuffer(u32 binding, const Tegra::Engines::ConstBufferInfo& buffer,
                           const GLShader::ConstBufferEntry& entry);
 
     /// Configures the current global memory entries to use for the draw command.
-    void SetupDrawGlobalMemory(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage,
-                               const Shader& shader);
+    void SetupDrawGlobalMemory(std::size_t stage_index, const Shader& shader);
 
     /// Configures the current global memory entries to use for the kernel invocation.
     void SetupComputeGlobalMemory(const Shader& kernel);
 
     /// Configures a constant buffer.
-    void SetupGlobalMemory(const GLShader::GlobalMemoryEntry& entry, GPUVAddr gpu_addr,
+    void SetupGlobalMemory(u32 binding, const GLShader::GlobalMemoryEntry& entry, GPUVAddr gpu_addr,
                            std::size_t size);
 
     /// Syncs all the state, shaders, render targets and textures setting before a draw call.
     void DrawPrelude();
 
-    /// Configures the current textures to use for the draw command. Returns shaders texture buffer
-    /// usage.
-    TextureBufferUsage SetupDrawTextures(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage,
-                                         const Shader& shader, BaseBindings base_bindings);
+    /// Configures the current textures to use for the draw command.
+    void SetupDrawTextures(std::size_t stage_index, const Shader& shader);
 
-    /// Configures the textures used in a compute shader. Returns texture buffer usage.
-    TextureBufferUsage SetupComputeTextures(const Shader& kernel);
+    /// Configures the textures used in a compute shader.
+    void SetupComputeTextures(const Shader& kernel);
 
-    /// Configures a texture. Returns true when the texture is a texture buffer.
-    bool SetupTexture(u32 binding, const Tegra::Texture::FullTextureInfo& texture,
+    /// Configures a texture.
+    void SetupTexture(u32 binding, const Tegra::Texture::FullTextureInfo& texture,
                       const GLShader::SamplerEntry& entry);
 
+    /// Configures images in a graphics shader.
+    void SetupDrawImages(std::size_t stage_index, const Shader& shader);
+
     /// Configures images in a compute shader.
     void SetupComputeImages(const Shader& shader);
 

src/video_core/renderer_opengl/gl_shader_cache.cpp

@@ -8,12 +8,15 @@
 #include <thread>
 #include <unordered_set>
 #include <boost/functional/hash.hpp>
+#include "common/alignment.h"
 #include "common/assert.h"
+#include "common/logging/log.h"
 #include "common/scope_exit.h"
 #include "core/core.h"
 #include "core/frontend/emu_window.h"
 #include "video_core/engines/kepler_compute.h"
 #include "video_core/engines/maxwell_3d.h"
+#include "video_core/engines/shader_type.h"
 #include "video_core/memory_manager.h"
 #include "video_core/renderer_opengl/gl_rasterizer.h"
 #include "video_core/renderer_opengl/gl_shader_cache.h"
@@ -82,28 +85,26 @@ std::size_t CalculateProgramSize(const GLShader::ProgramCode& program) {
 /// Gets the shader program code from memory for the specified address
 ProgramCode GetShaderCode(Tegra::MemoryManager& memory_manager, const GPUVAddr gpu_addr,
                           const u8* host_ptr) {
-    ProgramCode program_code(VideoCommon::Shader::MAX_PROGRAM_LENGTH);
+    ProgramCode 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::fill(code.begin(), code.end(), 0);
+        return code;
     });
-    memory_manager.ReadBlockUnsafe(gpu_addr, program_code.data(),
-                                   program_code.size() * sizeof(u64));
-    program_code.resize(CalculateProgramSize(program_code));
-    return program_code;
+    memory_manager.ReadBlockUnsafe(gpu_addr, code.data(), code.size() * sizeof(u64));
+    code.resize(CalculateProgramSize(code));
+    return code;
 }
 
 /// Gets the shader type from a Maxwell program type
-constexpr GLenum GetShaderType(ProgramType program_type) {
-    switch (program_type) {
-    case ProgramType::VertexA:
-    case ProgramType::VertexB:
+constexpr GLenum GetGLShaderType(ShaderType shader_type) {
+    switch (shader_type) {
+    case ShaderType::Vertex:
         return GL_VERTEX_SHADER;
-    case ProgramType::Geometry:
+    case ShaderType::Geometry:
         return GL_GEOMETRY_SHADER;
-    case ProgramType::Fragment:
+    case ShaderType::Fragment:
         return GL_FRAGMENT_SHADER;
-    case ProgramType::Compute:
+    case ShaderType::Compute:
         return GL_COMPUTE_SHADER;
     default:
         return GL_NONE;
@@ -133,30 +134,11 @@ constexpr std::tuple<const char*, const char*, u32> GetPrimitiveDescription(GLen
     }
 }
 
-ProgramType GetProgramType(Maxwell::ShaderProgram program) {
-    switch (program) {
-    case Maxwell::ShaderProgram::VertexA:
-        return ProgramType::VertexA;
-    case Maxwell::ShaderProgram::VertexB:
-        return ProgramType::VertexB;
-    case Maxwell::ShaderProgram::TesselationControl:
-        return ProgramType::TessellationControl;
-    case Maxwell::ShaderProgram::TesselationEval:
-        return ProgramType::TessellationEval;
-    case Maxwell::ShaderProgram::Geometry:
-        return ProgramType::Geometry;
-    case Maxwell::ShaderProgram::Fragment:
-        return ProgramType::Fragment;
-    }
-    UNREACHABLE();
-    return {};
-}
-
 /// Hashes one (or two) program streams
-u64 GetUniqueIdentifier(ProgramType program_type, const ProgramCode& code,
+u64 GetUniqueIdentifier(ShaderType shader_type, bool is_a, const ProgramCode& code,
                         const ProgramCode& code_b) {
     u64 unique_identifier = boost::hash_value(code);
-    if (program_type == ProgramType::VertexA) {
+    if (is_a) {
         // VertexA programs include two programs
         boost::hash_combine(unique_identifier, boost::hash_value(code_b));
     }
@@ -164,79 +146,74 @@ u64 GetUniqueIdentifier(ProgramType program_type, const ProgramCode& code,
 }
 
 /// Creates an unspecialized program from code streams
-std::string GenerateGLSL(const Device& device, ProgramType program_type, const ShaderIR& ir,
+std::string GenerateGLSL(const Device& device, ShaderType shader_type, const ShaderIR& ir,
                          const std::optional<ShaderIR>& ir_b) {
-    switch (program_type) {
-    case ProgramType::VertexA:
-    case ProgramType::VertexB:
+    switch (shader_type) {
+    case ShaderType::Vertex:
         return GLShader::GenerateVertexShader(device, ir, ir_b ? &*ir_b : nullptr);
-    case ProgramType::Geometry:
+    case ShaderType::Geometry:
         return GLShader::GenerateGeometryShader(device, ir);
-    case ProgramType::Fragment:
+    case ShaderType::Fragment:
         return GLShader::GenerateFragmentShader(device, ir);
-    case ProgramType::Compute:
+    case ShaderType::Compute:
         return GLShader::GenerateComputeShader(device, ir);
     default:
-        UNIMPLEMENTED_MSG("Unimplemented program_type={}", static_cast<u32>(program_type));
+        UNIMPLEMENTED_MSG("Unimplemented shader_type={}", static_cast<u32>(shader_type));
         return {};
     }
 }
 
-constexpr const char* GetProgramTypeName(ProgramType program_type) {
-    switch (program_type) {
-    case ProgramType::VertexA:
-    case ProgramType::VertexB:
+constexpr const char* GetShaderTypeName(ShaderType shader_type) {
+    switch (shader_type) {
+    case ShaderType::Vertex:
         return "VS";
-    case ProgramType::TessellationControl:
-        return "TCS";
-    case ProgramType::TessellationEval:
-        return "TES";
-    case ProgramType::Geometry:
+    case ShaderType::TesselationControl:
+        return "HS";
+    case ShaderType::TesselationEval:
+        return "DS";
+    case ShaderType::Geometry:
         return "GS";
-    case ProgramType::Fragment:
+    case ShaderType::Fragment:
         return "FS";
-    case ProgramType::Compute:
+    case ShaderType::Compute:
         return "CS";
     }
     return "UNK";
 }
 
-Tegra::Engines::ShaderType GetEnginesShaderType(ProgramType program_type) {
+constexpr ShaderType GetShaderType(Maxwell::ShaderProgram program_type) {
     switch (program_type) {
-    case ProgramType::VertexA:
-    case ProgramType::VertexB:
-        return Tegra::Engines::ShaderType::Vertex;
-    case ProgramType::TessellationControl:
-        return Tegra::Engines::ShaderType::TesselationControl;
-    case ProgramType::TessellationEval:
-        return Tegra::Engines::ShaderType::TesselationEval;
-    case ProgramType::Geometry:
-        return Tegra::Engines::ShaderType::Geometry;
-    case ProgramType::Fragment:
-        return Tegra::Engines::ShaderType::Fragment;
-    case ProgramType::Compute:
-        return Tegra::Engines::ShaderType::Compute;
+    case Maxwell::ShaderProgram::VertexA:
+    case Maxwell::ShaderProgram::VertexB:
+        return ShaderType::Vertex;
+    case Maxwell::ShaderProgram::TesselationControl:
+        return ShaderType::TesselationControl;
+    case Maxwell::ShaderProgram::TesselationEval:
+        return ShaderType::TesselationEval;
+    case Maxwell::ShaderProgram::Geometry:
+        return ShaderType::Geometry;
+    case Maxwell::ShaderProgram::Fragment:
+        return ShaderType::Fragment;
     }
-    UNREACHABLE();
     return {};
 }
 
-std::string GetShaderId(u64 unique_identifier, ProgramType program_type) {
-    return fmt::format("{}{:016X}", GetProgramTypeName(program_type), unique_identifier);
+std::string GetShaderId(u64 unique_identifier, ShaderType shader_type) {
+    return fmt::format("{}{:016X}", GetShaderTypeName(shader_type), unique_identifier);
 }
 
-Tegra::Engines::ConstBufferEngineInterface& GetConstBufferEngineInterface(
-    Core::System& system, ProgramType program_type) {
-    if (program_type == ProgramType::Compute) {
+Tegra::Engines::ConstBufferEngineInterface& GetConstBufferEngineInterface(Core::System& system,
+                                                                          ShaderType shader_type) {
+    if (shader_type == ShaderType::Compute) {
         return system.GPU().KeplerCompute();
     } else {
         return system.GPU().Maxwell3D();
     }
 }
 
-std::unique_ptr<ConstBufferLocker> MakeLocker(Core::System& system, ProgramType program_type) {
-    return std::make_unique<ConstBufferLocker>(GetEnginesShaderType(program_type),
-                                               GetConstBufferEngineInterface(system, program_type));
+std::unique_ptr<ConstBufferLocker> MakeLocker(Core::System& system, ShaderType shader_type) {
+    return std::make_unique<ConstBufferLocker>(shader_type,
+                                               GetConstBufferEngineInterface(system, shader_type));
 }
 
 void FillLocker(ConstBufferLocker& locker, const ShaderDiskCacheUsage& usage) {
@@ -253,88 +230,71 @@ void FillLocker(ConstBufferLocker& locker, const ShaderDiskCacheUsage& usage) {
     }
 }
 
-CachedProgram BuildShader(const Device& device, u64 unique_identifier, ProgramType program_type,
-                          const ProgramCode& program_code, const ProgramCode& program_code_b,
-                          const ProgramVariant& variant, ConstBufferLocker& locker,
+CachedProgram BuildShader(const Device& device, u64 unique_identifier, ShaderType shader_type,
+                          const ProgramCode& code, const ProgramCode& code_b,
+                          ConstBufferLocker& locker, const ProgramVariant& variant,
                           bool hint_retrievable = false) {
-    LOG_INFO(Render_OpenGL, "called. {}", GetShaderId(unique_identifier, program_type));
+    LOG_INFO(Render_OpenGL, "called. {}", GetShaderId(unique_identifier, shader_type));
 
-    const bool is_compute = program_type == ProgramType::Compute;
+    const bool is_compute = shader_type == ShaderType::Compute;
     const u32 main_offset = is_compute ? KERNEL_MAIN_OFFSET : STAGE_MAIN_OFFSET;
-    const ShaderIR ir(program_code, main_offset, COMPILER_SETTINGS, locker);
+    const ShaderIR ir(code, main_offset, COMPILER_SETTINGS, locker);
     std::optional<ShaderIR> ir_b;
-    if (!program_code_b.empty()) {
-        ir_b.emplace(program_code_b, main_offset, COMPILER_SETTINGS, locker);
+    if (!code_b.empty()) {
+        ir_b.emplace(code_b, main_offset, COMPILER_SETTINGS, locker);
     }
     const auto entries = GLShader::GetEntries(ir);
 
-    auto base_bindings{variant.base_bindings};
-    const auto primitive_mode{variant.primitive_mode};
-    const auto texture_buffer_usage{variant.texture_buffer_usage};
-
     std::string source = fmt::format(R"(// {}
 #version 430 core
 #extension GL_ARB_separate_shader_objects : enable
-#extension GL_ARB_shader_viewport_layer_array : enable
-#extension GL_EXT_shader_image_load_formatted : enable
-#extension GL_NV_gpu_shader5 : enable
-#extension GL_NV_shader_thread_group : enable
-#extension GL_NV_shader_thread_shuffle : enable
 )",
-                                     GetShaderId(unique_identifier, program_type));
-    if (is_compute) {
-        source += "#extension GL_ARB_compute_variable_group_size : require\n";
+                                     GetShaderId(unique_identifier, shader_type));
+    if (device.HasShaderBallot()) {
+        source += "#extension GL_ARB_shader_ballot : require\n";
+    }
+    if (device.HasVertexViewportLayer()) {
+        source += "#extension GL_ARB_shader_viewport_layer_array : require\n";
+    }
+    if (device.HasImageLoadFormatted()) {
+        source += "#extension GL_EXT_shader_image_load_formatted : require\n";
+    }
+    if (device.HasWarpIntrinsics()) {
+        source += "#extension GL_NV_gpu_shader5 : require\n"
+                  "#extension GL_NV_shader_thread_group : require\n"
+                  "#extension GL_NV_shader_thread_shuffle : require\n";
     }
     source += '\n';
 
-    if (!is_compute) {
-        source += fmt::format("#define EMULATION_UBO_BINDING {}\n", base_bindings.cbuf++);
-    }
-
-    for (const auto& cbuf : entries.const_buffers) {
-        source +=
-            fmt::format("#define CBUF_BINDING_{} {}\n", cbuf.GetIndex(), base_bindings.cbuf++);
-    }
-    for (const auto& gmem : entries.global_memory_entries) {
-        source += fmt::format("#define GMEM_BINDING_{}_{} {}\n", gmem.GetCbufIndex(),
-                              gmem.GetCbufOffset(), base_bindings.gmem++);
-    }
-    for (const auto& sampler : entries.samplers) {
-        source += fmt::format("#define SAMPLER_BINDING_{} {}\n", sampler.GetIndex(),
-                              base_bindings.sampler++);
-    }
-    for (const auto& image : entries.images) {
-        source +=
-            fmt::format("#define IMAGE_BINDING_{} {}\n", image.GetIndex(), base_bindings.image++);
-    }
-
-    // Transform 1D textures to texture samplers by declaring its preprocessor macros.
-    for (std::size_t i = 0; i < texture_buffer_usage.size(); ++i) {
-        if (!texture_buffer_usage.test(i)) {
-            continue;
-        }
-        source += fmt::format("#define SAMPLER_{}_IS_BUFFER\n", i);
-    }
-    if (texture_buffer_usage.any()) {
-        source += '\n';
-    }
-
-    if (program_type == ProgramType::Geometry) {
+    if (shader_type == ShaderType::Geometry) {
         const auto [glsl_topology, debug_name, max_vertices] =
-            GetPrimitiveDescription(primitive_mode);
+            GetPrimitiveDescription(variant.primitive_mode);
 
-        source += "layout (" + std::string(glsl_topology) + ") in;\n\n";
-        source += "#define MAX_VERTEX_INPUT " + std::to_string(max_vertices) + '\n';
+        source += fmt::format("layout ({}) in;\n\n", glsl_topology);
+        source += fmt::format("#define MAX_VERTEX_INPUT {}\n", max_vertices);
     }
-    if (program_type == ProgramType::Compute) {
-        source += "layout (local_size_variable) in;\n";
+    if (shader_type == ShaderType::Compute) {
+        source +=
+            fmt::format("layout (local_size_x = {}, local_size_y = {}, local_size_z = {}) in;\n",
+                        variant.block_x, variant.block_y, variant.block_z);
+
+        if (variant.shared_memory_size > 0) {
+            // TODO(Rodrigo): We should divide by four here, but having a larger shared memory pool
+            // avoids out of bound stores. Find out why shared memory size is being invalid.
+            source += fmt::format("shared uint smem[{}];", variant.shared_memory_size);
+        }
+
+        if (variant.local_memory_size > 0) {
+            source += fmt::format("#define LOCAL_MEMORY_SIZE {}",
+                                  Common::AlignUp(variant.local_memory_size, 4) / 4);
+        }
     }
 
     source += '\n';
-    source += GenerateGLSL(device, program_type, ir, ir_b);
+    source += GenerateGLSL(device, shader_type, ir, ir_b);
 
     OGLShader shader;
-    shader.Create(source.c_str(), GetShaderType(program_type));
+    shader.Create(source.c_str(), GetGLShaderType(shader_type));
 
     auto program = std::make_shared<OGLProgram>();
     program->Create(true, hint_retrievable, shader.handle);
@@ -357,18 +317,16 @@ std::unordered_set<GLenum> GetSupportedFormats() {
 
 } // Anonymous namespace
 
-CachedShader::CachedShader(const ShaderParameters& params, ProgramType program_type,
-                           GLShader::ShaderEntries entries, ProgramCode program_code,
-                           ProgramCode program_code_b)
-    : RasterizerCacheObject{params.host_ptr}, system{params.system},
-      disk_cache{params.disk_cache}, device{params.device}, cpu_addr{params.cpu_addr},
-      unique_identifier{params.unique_identifier}, program_type{program_type}, entries{entries},
-      program_code{std::move(program_code)}, program_code_b{std::move(program_code_b)} {
+CachedShader::CachedShader(const ShaderParameters& params, ShaderType shader_type,
+                           GLShader::ShaderEntries entries, ProgramCode code, ProgramCode code_b)
+    : RasterizerCacheObject{params.host_ptr}, system{params.system}, disk_cache{params.disk_cache},
+      device{params.device}, cpu_addr{params.cpu_addr}, unique_identifier{params.unique_identifier},
+      shader_type{shader_type}, entries{entries}, code{std::move(code)}, code_b{std::move(code_b)} {
     if (!params.precompiled_variants) {
         return;
     }
     for (const auto& pair : *params.precompiled_variants) {
-        auto locker = MakeLocker(system, program_type);
+        auto locker = MakeLocker(system, shader_type);
         const auto& usage = pair->first;
         FillLocker(*locker, usage);
 
@@ -389,94 +347,83 @@ CachedShader::CachedShader(const ShaderParameters& params, ProgramType program_t
 }
 
 Shader CachedShader::CreateStageFromMemory(const ShaderParameters& params,
-                                           Maxwell::ShaderProgram program_type,
-                                           ProgramCode program_code, ProgramCode program_code_b) {
-    params.disk_cache.SaveRaw(ShaderDiskCacheRaw(
-        params.unique_identifier, GetProgramType(program_type), program_code, program_code_b));
+                                           Maxwell::ShaderProgram program_type, ProgramCode code,
+                                           ProgramCode code_b) {
+    const auto shader_type = GetShaderType(program_type);
+    params.disk_cache.SaveRaw(
+        ShaderDiskCacheRaw(params.unique_identifier, shader_type, code, code_b));
 
-    ConstBufferLocker locker(GetEnginesShaderType(GetProgramType(program_type)),
-                             params.system.GPU().Maxwell3D());
-    const ShaderIR ir(program_code, STAGE_MAIN_OFFSET, COMPILER_SETTINGS, locker);
+    ConstBufferLocker locker(shader_type, params.system.GPU().Maxwell3D());
+    const ShaderIR ir(code, STAGE_MAIN_OFFSET, COMPILER_SETTINGS, locker);
     // TODO(Rodrigo): Handle VertexA shaders
     // std::optional<ShaderIR> ir_b;
-    // if (!program_code_b.empty()) {
-    //     ir_b.emplace(program_code_b, STAGE_MAIN_OFFSET);
+    // if (!code_b.empty()) {
+    //     ir_b.emplace(code_b, STAGE_MAIN_OFFSET);
     // }
-    return std::shared_ptr<CachedShader>(
-        new CachedShader(params, GetProgramType(program_type), GLShader::GetEntries(ir),
-                         std::move(program_code), std::move(program_code_b)));
+    return std::shared_ptr<CachedShader>(new CachedShader(
+        params, shader_type, GLShader::GetEntries(ir), std::move(code), std::move(code_b)));
 }
 
 Shader CachedShader::CreateKernelFromMemory(const ShaderParameters& params, ProgramCode code) {
     params.disk_cache.SaveRaw(
-        ShaderDiskCacheRaw(params.unique_identifier, ProgramType::Compute, code));
+        ShaderDiskCacheRaw(params.unique_identifier, ShaderType::Compute, code));
 
     ConstBufferLocker locker(Tegra::Engines::ShaderType::Compute,
                              params.system.GPU().KeplerCompute());
     const ShaderIR ir(code, KERNEL_MAIN_OFFSET, COMPILER_SETTINGS, locker);
     return std::shared_ptr<CachedShader>(new CachedShader(
-        params, ProgramType::Compute, GLShader::GetEntries(ir), std::move(code), {}));
+        params, ShaderType::Compute, GLShader::GetEntries(ir), std::move(code), {}));
 }
 
 Shader CachedShader::CreateFromCache(const ShaderParameters& params,
                                      const UnspecializedShader& unspecialized) {
-    return std::shared_ptr<CachedShader>(new CachedShader(params, unspecialized.program_type,
+    return std::shared_ptr<CachedShader>(new CachedShader(params, unspecialized.type,
                                                           unspecialized.entries, unspecialized.code,
                                                           unspecialized.code_b));
 }
 
-std::tuple<GLuint, BaseBindings> CachedShader::GetProgramHandle(const ProgramVariant& variant) {
-    UpdateVariant();
+GLuint CachedShader::GetHandle(const ProgramVariant& variant) {
+    EnsureValidLockerVariant();
 
-    const auto [entry, is_cache_miss] = curr_variant->programs.try_emplace(variant);
+    const auto [entry, is_cache_miss] = curr_locker_variant->programs.try_emplace(variant);
     auto& program = entry->second;
-    if (is_cache_miss) {
-        program = BuildShader(device, unique_identifier, program_type, program_code, program_code_b,
-                              variant, *curr_variant->locker);
-        disk_cache.SaveUsage(GetUsage(variant, *curr_variant->locker));
-
-        LabelGLObject(GL_PROGRAM, program->handle, cpu_addr);
+    if (!is_cache_miss) {
+        return program->handle;
     }
 
-    auto base_bindings = variant.base_bindings;
-    base_bindings.cbuf += static_cast<u32>(entries.const_buffers.size());
-    if (program_type != ProgramType::Compute) {
-        base_bindings.cbuf += STAGE_RESERVED_UBOS;
-    }
-    base_bindings.gmem += static_cast<u32>(entries.global_memory_entries.size());
-    base_bindings.sampler += static_cast<u32>(entries.samplers.size());
+    program = BuildShader(device, unique_identifier, shader_type, code, code_b,
+                          *curr_locker_variant->locker, variant);
+    disk_cache.SaveUsage(GetUsage(variant, *curr_locker_variant->locker));
 
-    return {program->handle, base_bindings};
+    LabelGLObject(GL_PROGRAM, program->handle, cpu_addr);
+    return program->handle;
 }
 
-void CachedShader::UpdateVariant() {
-    if (curr_variant && !curr_variant->locker->IsConsistent()) {
-        curr_variant = nullptr;
+bool CachedShader::EnsureValidLockerVariant() {
+    const auto previous_variant = curr_locker_variant;
+    if (curr_locker_variant && !curr_locker_variant->locker->IsConsistent()) {
+        curr_locker_variant = nullptr;
     }
-    if (!curr_variant) {
+    if (!curr_locker_variant) {
         for (auto& variant : locker_variants) {
             if (variant->locker->IsConsistent()) {
-                curr_variant = variant.get();
+                curr_locker_variant = variant.get();
             }
         }
     }
-    if (!curr_variant) {
+    if (!curr_locker_variant) {
         auto& new_variant = locker_variants.emplace_back();
         new_variant = std::make_unique<LockerVariant>();
-        new_variant->locker = MakeLocker(system, program_type);
-        curr_variant = new_variant.get();
+        new_variant->locker = MakeLocker(system, shader_type);
+        curr_locker_variant = new_variant.get();
     }
+    return previous_variant == curr_locker_variant;
 }
 
 ShaderDiskCacheUsage CachedShader::GetUsage(const ProgramVariant& variant,
                                             const ConstBufferLocker& locker) const {
-    ShaderDiskCacheUsage usage;
-    usage.unique_identifier = unique_identifier;
-    usage.variant = variant;
-    usage.keys = locker.GetKeys();
-    usage.bound_samplers = locker.GetBoundSamplers();
-    usage.bindless_samplers = locker.GetBindlessSamplers();
-    return usage;
+    return ShaderDiskCacheUsage{unique_identifier, variant, locker.GetKeys(),
+                                locker.GetBoundSamplers(), locker.GetBindlessSamplers()};
 }
 
 ShaderCacheOpenGL::ShaderCacheOpenGL(RasterizerOpenGL& rasterizer, Core::System& system,
@@ -535,11 +482,12 @@ void ShaderCacheOpenGL::LoadDiskCache(const std::atomic_bool& stop_loading,
                 }
             }
             if (!shader) {
-                auto locker{MakeLocker(system, unspecialized.program_type)};
+                auto locker{MakeLocker(system, unspecialized.type)};
                 FillLocker(*locker, usage);
-                shader = BuildShader(device, usage.unique_identifier, unspecialized.program_type,
-                                     unspecialized.code, unspecialized.code_b, usage.variant,
-                                     *locker, true);
+
+                shader = BuildShader(device, usage.unique_identifier, unspecialized.type,
+                                     unspecialized.code, unspecialized.code_b, *locker,
+                                     usage.variant, true);
             }
 
             std::scoped_lock lock{mutex};
@@ -642,7 +590,7 @@ bool ShaderCacheOpenGL::GenerateUnspecializedShaders(
         const auto& raw{raws[i]};
         const u64 unique_identifier{raw.GetUniqueIdentifier()};
         const u64 calculated_hash{
-            GetUniqueIdentifier(raw.GetProgramType(), raw.GetProgramCode(), raw.GetProgramCodeB())};
+            GetUniqueIdentifier(raw.GetType(), raw.HasProgramA(), raw.GetCode(), raw.GetCodeB())};
         if (unique_identifier != calculated_hash) {
             LOG_ERROR(Render_OpenGL,
                       "Invalid hash in entry={:016x} (obtained hash={:016x}) - "
@@ -653,9 +601,9 @@ bool ShaderCacheOpenGL::GenerateUnspecializedShaders(
         }
 
         const u32 main_offset =
-            raw.GetProgramType() == ProgramType::Compute ? KERNEL_MAIN_OFFSET : STAGE_MAIN_OFFSET;
-        ConstBufferLocker locker(GetEnginesShaderType(raw.GetProgramType()));
-        const ShaderIR ir(raw.GetProgramCode(), main_offset, COMPILER_SETTINGS, locker);
+            raw.GetType() == ShaderType::Compute ? KERNEL_MAIN_OFFSET : STAGE_MAIN_OFFSET;
+        ConstBufferLocker locker(raw.GetType());
+        const ShaderIR ir(raw.GetCode(), main_offset, COMPILER_SETTINGS, locker);
         // TODO(Rodrigo): Handle VertexA shaders
         // std::optional<ShaderIR> ir_b;
         // if (raw.HasProgramA()) {
@@ -664,9 +612,9 @@ bool ShaderCacheOpenGL::GenerateUnspecializedShaders(
 
         UnspecializedShader unspecialized;
         unspecialized.entries = GLShader::GetEntries(ir);
-        unspecialized.program_type = raw.GetProgramType();
-        unspecialized.code = raw.GetProgramCode();
-        unspecialized.code_b = raw.GetProgramCodeB();
+        unspecialized.type = raw.GetType();
+        unspecialized.code = raw.GetCode();
+        unspecialized.code_b = raw.GetCodeB();
         unspecialized_shaders.emplace(raw.GetUniqueIdentifier(), unspecialized);
 
         if (callback) {
@@ -699,7 +647,8 @@ Shader ShaderCacheOpenGL::GetStageProgram(Maxwell::ShaderProgram program) {
         code_b = GetShaderCode(memory_manager, address_b, memory_manager.GetPointer(address_b));
     }
 
-    const auto unique_identifier = GetUniqueIdentifier(GetProgramType(program), code, code_b);
+    const auto unique_identifier = GetUniqueIdentifier(
+        GetShaderType(program), program == Maxwell::ShaderProgram::VertexA, code, code_b);
     const auto precompiled_variants = GetPrecompiledVariants(unique_identifier);
     const auto cpu_addr{*memory_manager.GpuToCpuAddress(address)};
     const ShaderParameters params{system,   disk_cache, precompiled_variants, device,
@@ -727,7 +676,7 @@ Shader ShaderCacheOpenGL::GetComputeKernel(GPUVAddr code_addr) {
 
     // No kernel found - create a new one
     auto code{GetShaderCode(memory_manager, code_addr, host_ptr)};
-    const auto unique_identifier{GetUniqueIdentifier(ProgramType::Compute, code, {})};
+    const auto unique_identifier{GetUniqueIdentifier(ShaderType::Compute, false, code, {})};
     const auto precompiled_variants = GetPrecompiledVariants(unique_identifier);
     const auto cpu_addr{*memory_manager.GpuToCpuAddress(code_addr)};
     const ShaderParameters params{system,   disk_cache, precompiled_variants, device,

src/video_core/renderer_opengl/gl_shader_cache.h

@@ -17,6 +17,7 @@
 #include <glad/glad.h>
 
 #include "common/common_types.h"
+#include "video_core/engines/shader_type.h"
 #include "video_core/rasterizer_cache.h"
 #include "video_core/renderer_opengl/gl_resource_manager.h"
 #include "video_core/renderer_opengl/gl_shader_decompiler.h"
@@ -47,7 +48,7 @@ using PrecompiledVariants = std::vector<PrecompiledPrograms::iterator>;
 
 struct UnspecializedShader {
     GLShader::ShaderEntries entries;
-    ProgramType program_type;
+    Tegra::Engines::ShaderType type;
     ProgramCode code;
     ProgramCode code_b;
 };
@@ -77,7 +78,7 @@ public:
     }
 
     std::size_t GetSizeInBytes() const override {
-        return program_code.size() * sizeof(u64);
+        return code.size() * sizeof(u64);
     }
 
     /// Gets the shader entries for the shader
@@ -86,7 +87,7 @@ public:
     }
 
     /// Gets the GL program handle for the shader
-    std::tuple<GLuint, BaseBindings> GetProgramHandle(const ProgramVariant& variant);
+    GLuint GetHandle(const ProgramVariant& variant);
 
 private:
     struct LockerVariant {
@@ -94,11 +95,11 @@ private:
         std::unordered_map<ProgramVariant, CachedProgram> programs;
     };
 
-    explicit CachedShader(const ShaderParameters& params, ProgramType program_type,
+    explicit CachedShader(const ShaderParameters& params, Tegra::Engines::ShaderType shader_type,
                           GLShader::ShaderEntries entries, ProgramCode program_code,
                           ProgramCode program_code_b);
 
-    void UpdateVariant();
+    bool EnsureValidLockerVariant();
 
     ShaderDiskCacheUsage GetUsage(const ProgramVariant& variant,
                                   const VideoCommon::Shader::ConstBufferLocker& locker) const;
@@ -110,14 +111,14 @@ private:
     VAddr cpu_addr{};
 
     u64 unique_identifier{};
-    ProgramType program_type{};
+    Tegra::Engines::ShaderType shader_type{};
 
     GLShader::ShaderEntries entries;
 
-    ProgramCode program_code;
-    ProgramCode program_code_b;
+    ProgramCode code;
+    ProgramCode code_b;
 
-    LockerVariant* curr_variant = nullptr;
+    LockerVariant* curr_locker_variant = nullptr;
     std::vector<std::unique_ptr<LockerVariant>> locker_variants;
 };
 

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -16,6 +16,7 @@
 #include "common/common_types.h"
 #include "common/logging/log.h"
 #include "video_core/engines/maxwell_3d.h"
+#include "video_core/engines/shader_type.h"
 #include "video_core/renderer_opengl/gl_device.h"
 #include "video_core/renderer_opengl/gl_rasterizer.h"
 #include "video_core/renderer_opengl/gl_shader_decompiler.h"
@@ -27,6 +28,7 @@ namespace OpenGL::GLShader {
 
 namespace {
 
+using Tegra::Engines::ShaderType;
 using Tegra::Shader::Attribute;
 using Tegra::Shader::AttributeUse;
 using Tegra::Shader::Header;
@@ -41,6 +43,9 @@ using namespace VideoCommon::Shader;
 using Maxwell = Tegra::Engines::Maxwell3D::Regs;
 using Operation = const OperationNode&;
 
+class ASTDecompiler;
+class ExprDecompiler;
+
 enum class Type { Void, Bool, Bool2, Float, Int, Uint, HalfFloat };
 
 struct TextureAoffi {};
@@ -223,7 +228,7 @@ private:
     Type type{};
 };
 
-constexpr const char* GetTypeString(Type type) {
+const char* GetTypeString(Type type) {
     switch (type) {
     case Type::Bool:
         return "bool";
@@ -243,7 +248,7 @@ constexpr const char* GetTypeString(Type type) {
     }
 }
 
-constexpr const char* GetImageTypeDeclaration(Tegra::Shader::ImageType image_type) {
+const char* GetImageTypeDeclaration(Tegra::Shader::ImageType image_type) {
     switch (image_type) {
     case Tegra::Shader::ImageType::Texture1D:
         return "1D";
@@ -331,16 +336,13 @@ std::string FlowStackTopName(MetaStackClass stack) {
     return fmt::format("{}_flow_stack_top", GetFlowStackPrefix(stack));
 }
 
-constexpr bool IsVertexShader(ProgramType stage) {
-    return stage == ProgramType::VertexA || stage == ProgramType::VertexB;
+[[deprecated]] constexpr bool IsVertexShader(ShaderType stage) {
+    return stage == ShaderType::Vertex;
 }
 
-class ASTDecompiler;
-class ExprDecompiler;
-
 class GLSLDecompiler final {
 public:
-    explicit GLSLDecompiler(const Device& device, const ShaderIR& ir, ProgramType stage,
+    explicit GLSLDecompiler(const Device& device, const ShaderIR& ir, ShaderType stage,
                             std::string suffix)
         : device{device}, ir{ir}, stage{stage}, suffix{suffix}, header{ir.GetHeader()} {}
 
@@ -427,7 +429,7 @@ private:
     }
 
     void DeclareGeometry() {
-        if (stage != ProgramType::Geometry) {
+        if (stage != ShaderType::Geometry) {
             return;
         }
 
@@ -510,10 +512,14 @@ private:
     }
 
     void DeclareLocalMemory() {
-        // TODO(Rodrigo): Unstub kernel local memory size and pass it from a register at
-        // specialization time.
-        const u64 local_memory_size =
-            stage == ProgramType::Compute ? 0x400 : header.GetLocalMemorySize();
+        if (stage == ShaderType::Compute) {
+            code.AddLine("#ifdef LOCAL_MEMORY_SIZE");
+            code.AddLine("uint {}[LOCAL_MEMORY_SIZE];", GetLocalMemory());
+            code.AddLine("#endif");
+            return;
+        }
+
+        const u64 local_memory_size = header.GetLocalMemorySize();
         if (local_memory_size == 0) {
             return;
         }
@@ -522,13 +528,6 @@ private:
         code.AddNewLine();
     }
 
-    void DeclareSharedMemory() {
-        if (stage != ProgramType::Compute) {
-            return;
-        }
-        code.AddLine("shared uint {}[];", GetSharedMemory());
-    }
-
     void DeclareInternalFlags() {
         for (u32 flag = 0; flag < static_cast<u32>(InternalFlag::Amount); flag++) {
             const auto flag_code = static_cast<InternalFlag>(flag);
@@ -578,12 +577,12 @@ private:
         const u32 location{GetGenericAttributeIndex(index)};
 
         std::string name{GetInputAttribute(index)};
-        if (stage == ProgramType::Geometry) {
+        if (stage == ShaderType::Geometry) {
             name = "gs_" + name + "[]";
         }
 
         std::string suffix;
-        if (stage == ProgramType::Fragment) {
+        if (stage == ShaderType::Fragment) {
             const auto input_mode{header.ps.GetAttributeUse(location)};
             if (skip_unused && input_mode == AttributeUse::Unused) {
                 return;
@@ -595,7 +594,7 @@ private:
     }
 
     void DeclareOutputAttributes() {
-        if (ir.HasPhysicalAttributes() && stage != ProgramType::Fragment) {
+        if (ir.HasPhysicalAttributes() && stage != ShaderType::Fragment) {
             for (u32 i = 0; i < GetNumPhysicalVaryings(); ++i) {
                 DeclareOutputAttribute(ToGenericAttribute(i));
             }
@@ -620,9 +619,9 @@ private:
     }
 
     void DeclareConstantBuffers() {
-        for (const auto& entry : ir.GetConstantBuffers()) {
-            const auto [index, size] = entry;
-            code.AddLine("layout (std140, binding = CBUF_BINDING_{}) uniform {} {{", index,
+        u32 binding = device.GetBaseBindings(stage).uniform_buffer;
+        for (const auto& [index, cbuf] : ir.GetConstantBuffers()) {
+            code.AddLine("layout (std140, binding = {}) uniform {} {{", binding++,
                          GetConstBufferBlock(index));
             code.AddLine("    uvec4 {}[{}];", GetConstBuffer(index), MAX_CONSTBUFFER_ELEMENTS);
             code.AddLine("}};");
@@ -631,9 +630,8 @@ private:
     }
 
     void DeclareGlobalMemory() {
-        for (const auto& gmem : ir.GetGlobalMemory()) {
-            const auto& [base, usage] = gmem;
-
+        u32 binding = device.GetBaseBindings(stage).shader_storage_buffer;
+        for (const auto& [base, usage] : ir.GetGlobalMemory()) {
             // Since we don't know how the shader will use the shader, hint the driver to disable as
             // much optimizations as possible
             std::string qualifier = "coherent volatile";
@@ -643,8 +641,8 @@ private:
                 qualifier += " writeonly";
             }
 
-            code.AddLine("layout (std430, binding = GMEM_BINDING_{}_{}) {} buffer {} {{",
-                         base.cbuf_index, base.cbuf_offset, qualifier, GetGlobalMemoryBlock(base));
+            code.AddLine("layout (std430, binding = {}) {} buffer {} {{", binding++, qualifier,
+                         GetGlobalMemoryBlock(base));
             code.AddLine("    uint {}[];", GetGlobalMemory(base));
             code.AddLine("}};");
             code.AddNewLine();
@@ -652,15 +650,17 @@ private:
     }
 
     void DeclareSamplers() {
-        const auto& samplers = ir.GetSamplers();
-        for (const auto& sampler : samplers) {
-            const std::string name{GetSampler(sampler)};
-            const std::string description{"layout (binding = SAMPLER_BINDING_" +
-                                          std::to_string(sampler.GetIndex()) + ") uniform"};
+        u32 binding = device.GetBaseBindings(stage).sampler;
+        for (const auto& sampler : ir.GetSamplers()) {
+            const std::string name = GetSampler(sampler);
+            const std::string description = fmt::format("layout (binding = {}) uniform", binding++);
+
             std::string sampler_type = [&]() {
+                if (sampler.IsBuffer()) {
+                    return "samplerBuffer";
+                }
                 switch (sampler.GetType()) {
                 case Tegra::Shader::TextureType::Texture1D:
-                    // Special cased, read below.
                     return "sampler1D";
                 case Tegra::Shader::TextureType::Texture2D:
                     return "sampler2D";
@@ -680,21 +680,9 @@ private:
                 sampler_type += "Shadow";
             }
 
-            if (sampler.GetType() == Tegra::Shader::TextureType::Texture1D) {
-                // 1D textures can be aliased to texture buffers, hide the declarations behind a
-                // preprocessor flag and use one or the other from the GPU state. This has to be
-                // done because shaders don't have enough information to determine the texture type.
-                EmitIfdefIsBuffer(sampler);
-                code.AddLine("{} samplerBuffer {};", description, name);
-                code.AddLine("#else");
-                code.AddLine("{} {} {};", description, sampler_type, name);
-                code.AddLine("#endif");
-            } else {
-                // The other texture types (2D, 3D and cubes) don't have this issue.
-                code.AddLine("{} {} {};", description, sampler_type, name);
-            }
+            code.AddLine("{} {} {};", description, sampler_type, name);
         }
-        if (!samplers.empty()) {
+        if (!ir.GetSamplers().empty()) {
             code.AddNewLine();
         }
     }
@@ -717,7 +705,7 @@ private:
                 constexpr u32 element_stride = 4;
                 const u32 address{generic_base + index * generic_stride + element * element_stride};
 
-                const bool declared = stage != ProgramType::Fragment ||
+                const bool declared = stage != ShaderType::Fragment ||
                                       header.ps.GetAttributeUse(index) != AttributeUse::Unused;
                 const std::string value =
                     declared ? ReadAttribute(attribute, element).AsFloat() : "0.0f";
@@ -734,8 +722,8 @@ private:
     }
 
     void DeclareImages() {
-        const auto& images{ir.GetImages()};
-        for (const auto& image : images) {
+        u32 binding = device.GetBaseBindings(stage).image;
+        for (const auto& image : ir.GetImages()) {
             std::string qualifier = "coherent volatile";
             if (image.IsRead() && !image.IsWritten()) {
                 qualifier += " readonly";
@@ -745,10 +733,10 @@ private:
 
             const char* format = image.IsAtomic() ? "r32ui, " : "";
             const char* type_declaration = GetImageTypeDeclaration(image.GetType());
-            code.AddLine("layout ({}binding = IMAGE_BINDING_{}) {} uniform uimage{} {};", format,
-                         image.GetIndex(), qualifier, type_declaration, GetImage(image));
+            code.AddLine("layout ({}binding = {}) {} uniform uimage{} {};", format, binding++,
+                         qualifier, type_declaration, GetImage(image));
         }
-        if (!images.empty()) {
+        if (!ir.GetImages().empty()) {
             code.AddNewLine();
         }
     }
@@ -809,7 +797,7 @@ private:
         }
 
         if (const auto abuf = std::get_if<AbufNode>(&*node)) {
-            UNIMPLEMENTED_IF_MSG(abuf->IsPhysicalBuffer() && stage == ProgramType::Geometry,
+            UNIMPLEMENTED_IF_MSG(abuf->IsPhysicalBuffer() && stage == ShaderType::Geometry,
                                  "Physical attributes in geometry shaders are not implemented");
             if (abuf->IsPhysicalBuffer()) {
                 return {fmt::format("ReadPhysicalAttribute({})",
@@ -868,18 +856,13 @@ private:
         }
 
         if (const auto lmem = std::get_if<LmemNode>(&*node)) {
-            if (stage == ProgramType::Compute) {
-                LOG_WARNING(Render_OpenGL, "Local memory is stubbed on compute shaders");
-            }
             return {
                 fmt::format("{}[{} >> 2]", GetLocalMemory(), Visit(lmem->GetAddress()).AsUint()),
                 Type::Uint};
         }
 
         if (const auto smem = std::get_if<SmemNode>(&*node)) {
-            return {
-                fmt::format("{}[{} >> 2]", GetSharedMemory(), Visit(smem->GetAddress()).AsUint()),
-                Type::Uint};
+            return {fmt::format("smem[{} >> 2]", Visit(smem->GetAddress()).AsUint()), Type::Uint};
         }
 
         if (const auto internal_flag = std::get_if<InternalFlagNode>(&*node)) {
@@ -909,7 +892,7 @@ private:
 
     Expression ReadAttribute(Attribute::Index attribute, u32 element, const Node& buffer = {}) {
         const auto GeometryPass = [&](std::string_view name) {
-            if (stage == ProgramType::Geometry && buffer) {
+            if (stage == ShaderType::Geometry && buffer) {
                 // TODO(Rodrigo): Guard geometry inputs against out of bound reads. Some games
                 // set an 0x80000000 index for those and the shader fails to build. Find out why
                 // this happens and what's its intent.
@@ -921,11 +904,11 @@ private:
         switch (attribute) {
         case Attribute::Index::Position:
             switch (stage) {
-            case ProgramType::Geometry:
+            case ShaderType::Geometry:
                 return {fmt::format("gl_in[{}].gl_Position{}", Visit(buffer).AsUint(),
                                     GetSwizzle(element)),
                         Type::Float};
-            case ProgramType::Fragment:
+            case ShaderType::Fragment:
                 return {element == 3 ? "1.0f" : ("gl_FragCoord"s + GetSwizzle(element)),
                         Type::Float};
             default:
@@ -959,7 +942,7 @@ private:
             return {"0", Type::Int};
         case Attribute::Index::FrontFacing:
             // TODO(Subv): Find out what the values are for the other elements.
-            ASSERT(stage == ProgramType::Fragment);
+            ASSERT(stage == ShaderType::Fragment);
             switch (element) {
             case 3:
                 return {"(gl_FrontFacing ? -1 : 0)", Type::Int};
@@ -985,7 +968,7 @@ private:
         // be found in fragment shaders, so we disable precise there. There are vertex shaders that
         // also fail to build but nobody seems to care about those.
         // Note: Only bugged drivers will skip precise.
-        const bool disable_precise = device.HasPreciseBug() && stage == ProgramType::Fragment;
+        const bool disable_precise = device.HasPreciseBug() && stage == ShaderType::Fragment;
 
         std::string temporary = code.GenerateTemporary();
         code.AddLine("{}{} {} = {};", disable_precise ? "" : "precise ", GetTypeString(type),
@@ -1247,17 +1230,12 @@ private:
             }
             target = std::move(*output);
         } else if (const auto lmem = std::get_if<LmemNode>(&*dest)) {
-            if (stage == ProgramType::Compute) {
-                LOG_WARNING(Render_OpenGL, "Local memory is stubbed on compute shaders");
-            }
             target = {
                 fmt::format("{}[{} >> 2]", GetLocalMemory(), Visit(lmem->GetAddress()).AsUint()),
                 Type::Uint};
         } else if (const auto smem = std::get_if<SmemNode>(&*dest)) {
-            ASSERT(stage == ProgramType::Compute);
-            target = {
-                fmt::format("{}[{} >> 2]", GetSharedMemory(), Visit(smem->GetAddress()).AsUint()),
-                Type::Uint};
+            ASSERT(stage == ShaderType::Compute);
+            target = {fmt::format("smem[{} >> 2]", Visit(smem->GetAddress()).AsUint()), Type::Uint};
         } else if (const auto gmem = std::get_if<GmemNode>(&*dest)) {
             const std::string real = Visit(gmem->GetRealAddress()).AsUint();
             const std::string base = Visit(gmem->GetBaseAddress()).AsUint();
@@ -1379,6 +1357,26 @@ private:
         return GenerateUnary(operation, "float", Type::Float, type);
     }
 
+    Expression FSwizzleAdd(Operation operation) {
+        const std::string op_a = VisitOperand(operation, 0).AsFloat();
+        const std::string op_b = VisitOperand(operation, 1).AsFloat();
+
+        if (!device.HasShaderBallot()) {
+            LOG_ERROR(Render_OpenGL, "Shader ballot is unavailable but required by the shader");
+            return {fmt::format("{} + {}", op_a, op_b), Type::Float};
+        }
+
+        const std::string instr_mask = VisitOperand(operation, 2).AsUint();
+        const std::string mask = code.GenerateTemporary();
+        code.AddLine("uint {} = ({} >> ((gl_SubGroupInvocationARB & 3) << 1)) & 3;", mask,
+                     instr_mask);
+
+        const std::string modifier_a = fmt::format("fswzadd_modifiers_a[{}]", mask);
+        const std::string modifier_b = fmt::format("fswzadd_modifiers_b[{}]", mask);
+        return {fmt::format("(({} * {}) + ({} * {}))", op_a, modifier_a, op_b, modifier_b),
+                Type::Float};
+    }
+
     Expression ICastFloat(Operation operation) {
         return GenerateUnary(operation, "int", Type::Int, Type::Float);
     }
@@ -1729,27 +1727,14 @@ private:
                 expr += ", ";
         }
 
-        // Store a copy of the expression without the lod to be used with texture buffers
-        std::string expr_buffer = expr;
-
-        if (meta->lod) {
+        if (meta->lod && !meta->sampler.IsBuffer()) {
             expr += ", ";
             expr += Visit(meta->lod).AsInt();
         }
         expr += ')';
         expr += GetSwizzle(meta->element);
 
-        expr_buffer += ')';
-        expr_buffer += GetSwizzle(meta->element);
-
-        const std::string tmp{code.GenerateTemporary()};
-        EmitIfdefIsBuffer(meta->sampler);
-        code.AddLine("float {} = {};", tmp, expr_buffer);
-        code.AddLine("#else");
-        code.AddLine("float {} = {};", tmp, expr);
-        code.AddLine("#endif");
-
-        return {tmp, Type::Float};
+        return {std::move(expr), Type::Float};
     }
 
     Expression ImageLoad(Operation operation) {
@@ -1817,7 +1802,7 @@ private:
     }
 
     void PreExit() {
-        if (stage != ProgramType::Fragment) {
+        if (stage != ShaderType::Fragment) {
             return;
         }
         const auto& used_registers = ir.GetRegisters();
@@ -1870,27 +1855,21 @@ private:
     }
 
     Expression EmitVertex(Operation operation) {
-        ASSERT_MSG(stage == ProgramType::Geometry,
+        ASSERT_MSG(stage == ShaderType::Geometry,
                    "EmitVertex is expected to be used in a geometry shader.");
-
-        // If a geometry shader is attached, it will always flip (it's the last stage before
-        // fragment). For more info about flipping, refer to gl_shader_gen.cpp.
-        code.AddLine("gl_Position.xy *= viewport_flip.xy;");
         code.AddLine("EmitVertex();");
         return {};
     }
 
     Expression EndPrimitive(Operation operation) {
-        ASSERT_MSG(stage == ProgramType::Geometry,
+        ASSERT_MSG(stage == ShaderType::Geometry,
                    "EndPrimitive is expected to be used in a geometry shader.");
-
         code.AddLine("EndPrimitive();");
         return {};
     }
 
     Expression YNegate(Operation operation) {
-        // Config pack's third value is Y_NEGATE's state.
-        return {"config_pack[2]", Type::Uint};
+        return {"y_direction", Type::Float};
     }
 
     template <u32 element>
@@ -1942,34 +1921,24 @@ private:
         return Vote(operation, "allThreadsEqualNV");
     }
 
-    template <const std::string_view& func>
-    Expression Shuffle(Operation operation) {
-        const std::string value = VisitOperand(operation, 0).AsFloat();
-        if (!device.HasWarpIntrinsics()) {
-            LOG_ERROR(Render_OpenGL, "Nvidia shuffle intrinsics are required by this shader");
-            // On a "single-thread" device we are either on the same thread or out of bounds. Both
-            // cases return the passed value.
-            return {value, Type::Float};
+    Expression ThreadId(Operation operation) {
+        if (!device.HasShaderBallot()) {
+            LOG_ERROR(Render_OpenGL, "Shader ballot is unavailable but required by the shader");
+            return {"0U", Type::Uint};
+        }
+        return {"gl_SubGroupInvocationARB", Type::Uint};
+    }
+
+    Expression ShuffleIndexed(Operation operation) {
+        std::string value = VisitOperand(operation, 0).AsFloat();
+
+        if (!device.HasShaderBallot()) {
+            LOG_ERROR(Render_OpenGL, "Shader ballot is unavailable but required by the shader");
+            return {std::move(value), Type::Float};
         }
 
         const std::string index = VisitOperand(operation, 1).AsUint();
-        const std::string width = VisitOperand(operation, 2).AsUint();
-        return {fmt::format("{}({}, {}, {})", func, value, index, width), Type::Float};
-    }
-
-    template <const std::string_view& func>
-    Expression InRangeShuffle(Operation operation) {
-        const std::string index = VisitOperand(operation, 0).AsUint();
-        const std::string width = VisitOperand(operation, 1).AsUint();
-        if (!device.HasWarpIntrinsics()) {
-            // On a "single-thread" device we are only in bounds when the requested index is 0.
-            return {fmt::format("({} == 0U)", index), Type::Bool};
-        }
-
-        const std::string in_range = code.GenerateTemporary();
-        code.AddLine("bool {};", in_range);
-        code.AddLine("{}(0U, {}, {}, {});", func, index, width, in_range);
-        return {in_range, Type::Bool};
+        return {fmt::format("readInvocationARB({}, {})", value, index), Type::Float};
     }
 
     struct Func final {
@@ -1981,11 +1950,6 @@ private:
         static constexpr std::string_view Or = "Or";
         static constexpr std::string_view Xor = "Xor";
         static constexpr std::string_view Exchange = "Exchange";
-
-        static constexpr std::string_view ShuffleIndexed = "shuffleNV";
-        static constexpr std::string_view ShuffleUp = "shuffleUpNV";
-        static constexpr std::string_view ShuffleDown = "shuffleDownNV";
-        static constexpr std::string_view ShuffleButterfly = "shuffleXorNV";
     };
 
     static constexpr std::array operation_decompilers = {
@@ -2016,6 +1980,7 @@ private:
         &GLSLDecompiler::FTrunc,
         &GLSLDecompiler::FCastInteger<Type::Int>,
         &GLSLDecompiler::FCastInteger<Type::Uint>,
+        &GLSLDecompiler::FSwizzleAdd,
 
         &GLSLDecompiler::Add<Type::Int>,
         &GLSLDecompiler::Mul<Type::Int>,
@@ -2151,15 +2116,8 @@ private:
         &GLSLDecompiler::VoteAny,
         &GLSLDecompiler::VoteEqual,
 
-        &GLSLDecompiler::Shuffle<Func::ShuffleIndexed>,
-        &GLSLDecompiler::Shuffle<Func::ShuffleUp>,
-        &GLSLDecompiler::Shuffle<Func::ShuffleDown>,
-        &GLSLDecompiler::Shuffle<Func::ShuffleButterfly>,
-
-        &GLSLDecompiler::InRangeShuffle<Func::ShuffleIndexed>,
-        &GLSLDecompiler::InRangeShuffle<Func::ShuffleUp>,
-        &GLSLDecompiler::InRangeShuffle<Func::ShuffleDown>,
-        &GLSLDecompiler::InRangeShuffle<Func::ShuffleButterfly>,
+        &GLSLDecompiler::ThreadId,
+        &GLSLDecompiler::ShuffleIndexed,
     };
     static_assert(operation_decompilers.size() == static_cast<std::size_t>(OperationCode::Amount));
 
@@ -2200,10 +2158,6 @@ private:
         return "lmem_" + suffix;
     }
 
-    std::string GetSharedMemory() const {
-        return fmt::format("smem_{}", suffix);
-    }
-
     std::string GetInternalFlag(InternalFlag flag) const {
         constexpr std::array InternalFlagNames = {"zero_flag", "sign_flag", "carry_flag",
                                                   "overflow_flag"};
@@ -2221,10 +2175,6 @@ private:
         return GetDeclarationWithSuffix(static_cast<u32>(image.GetIndex()), "image");
     }
 
-    void EmitIfdefIsBuffer(const Sampler& sampler) {
-        code.AddLine("#ifdef SAMPLER_{}_IS_BUFFER", sampler.GetIndex());
-    }
-
     std::string GetDeclarationWithSuffix(u32 index, std::string_view name) const {
         return fmt::format("{}_{}_{}", name, index, suffix);
     }
@@ -2243,7 +2193,7 @@ private:
 
     const Device& device;
     const ShaderIR& ir;
-    const ProgramType stage;
+    const ShaderType stage;
     const std::string suffix;
     const Header header;
 
@@ -2492,10 +2442,13 @@ bvec2 HalfFloatNanComparison(bvec2 comparison, vec2 pair1, vec2 pair2) {
     bvec2 is_nan2 = isnan(pair2);
     return bvec2(comparison.x || is_nan1.x || is_nan2.x, comparison.y || is_nan1.y || is_nan2.y);
 }
+
+const float fswzadd_modifiers_a[] = float[4](-1.0f,  1.0f, -1.0f,  0.0f );
+const float fswzadd_modifiers_b[] = float[4](-1.0f, -1.0f,  1.0f, -1.0f );
 )";
 }
 
-std::string Decompile(const Device& device, const ShaderIR& ir, ProgramType stage,
+std::string Decompile(const Device& device, const ShaderIR& ir, ShaderType stage,
                       const std::string& suffix) {
     GLSLDecompiler decompiler(device, ir, stage, suffix);
     decompiler.Decompile();

src/video_core/renderer_opengl/gl_shader_decompiler.h

@@ -10,6 +10,7 @@
 #include <vector>
 #include "common/common_types.h"
 #include "video_core/engines/maxwell_3d.h"
+#include "video_core/engines/shader_type.h"
 #include "video_core/shader/shader_ir.h"
 
 namespace VideoCommon::Shader {
@@ -17,20 +18,8 @@ class ShaderIR;
 }
 
 namespace OpenGL {
-
 class Device;
-
-enum class ProgramType : u32 {
-    VertexA = 0,
-    VertexB = 1,
-    TessellationControl = 2,
-    TessellationEval = 3,
-    Geometry = 4,
-    Fragment = 5,
-    Compute = 6
-};
-
-} // namespace OpenGL
+}
 
 namespace OpenGL::GLShader {
 
@@ -94,6 +83,6 @@ ShaderEntries GetEntries(const VideoCommon::Shader::ShaderIR& ir);
 std::string GetCommonDeclarations();
 
 std::string Decompile(const Device& device, const VideoCommon::Shader::ShaderIR& ir,
-                      ProgramType stage, const std::string& suffix);
+                      Tegra::Engines::ShaderType stage, const std::string& suffix);
 
 } // namespace OpenGL::GLShader

src/video_core/renderer_opengl/gl_shader_disk_cache.cpp

@@ -3,6 +3,7 @@
 // Refer to the license.txt file included.
 
 #include <cstring>
+
 #include <fmt/format.h>
 
 #include "common/assert.h"
@@ -12,39 +13,22 @@
 #include "common/logging/log.h"
 #include "common/scm_rev.h"
 #include "common/zstd_compression.h"
-
 #include "core/core.h"
 #include "core/hle/kernel/process.h"
 #include "core/settings.h"
-
+#include "video_core/engines/shader_type.h"
 #include "video_core/renderer_opengl/gl_shader_cache.h"
 #include "video_core/renderer_opengl/gl_shader_disk_cache.h"
 
 namespace OpenGL {
 
+using Tegra::Engines::ShaderType;
 using VideoCommon::Shader::BindlessSamplerMap;
 using VideoCommon::Shader::BoundSamplerMap;
 using VideoCommon::Shader::KeyMap;
 
 namespace {
 
-struct ConstBufferKey {
-    u32 cbuf;
-    u32 offset;
-    u32 value;
-};
-
-struct BoundSamplerKey {
-    u32 offset;
-    Tegra::Engines::SamplerDescriptor sampler;
-};
-
-struct BindlessSamplerKey {
-    u32 cbuf;
-    u32 offset;
-    Tegra::Engines::SamplerDescriptor sampler;
-};
-
 using ShaderCacheVersionHash = std::array<u8, 64>;
 
 enum class TransferableEntryKind : u32 {
@@ -52,10 +36,27 @@ enum class TransferableEntryKind : u32 {
     Usage,
 };
 
-constexpr u32 NativeVersion = 5;
+struct ConstBufferKey {
+    u32 cbuf{};
+    u32 offset{};
+    u32 value{};
+};
+
+struct BoundSamplerKey {
+    u32 offset{};
+    Tegra::Engines::SamplerDescriptor sampler{};
+};
+
+struct BindlessSamplerKey {
+    u32 cbuf{};
+    u32 offset{};
+    Tegra::Engines::SamplerDescriptor sampler{};
+};
+
+constexpr u32 NativeVersion = 11;
 
 // Making sure sizes doesn't change by accident
-static_assert(sizeof(BaseBindings) == 16);
+static_assert(sizeof(ProgramVariant) == 20);
 
 ShaderCacheVersionHash GetShaderCacheVersionHash() {
     ShaderCacheVersionHash hash{};
@@ -66,10 +67,10 @@ ShaderCacheVersionHash GetShaderCacheVersionHash() {
 
 } // Anonymous namespace
 
-ShaderDiskCacheRaw::ShaderDiskCacheRaw(u64 unique_identifier, ProgramType program_type,
-                                       ProgramCode program_code, ProgramCode program_code_b)
-    : unique_identifier{unique_identifier}, program_type{program_type},
-      program_code{std::move(program_code)}, program_code_b{std::move(program_code_b)} {}
+ShaderDiskCacheRaw::ShaderDiskCacheRaw(u64 unique_identifier, ShaderType type, ProgramCode code,
+                                       ProgramCode code_b)
+    : unique_identifier{unique_identifier}, type{type}, code{std::move(code)}, code_b{std::move(
+                                                                                   code_b)} {}
 
 ShaderDiskCacheRaw::ShaderDiskCacheRaw() = default;
 
@@ -77,42 +78,39 @@ ShaderDiskCacheRaw::~ShaderDiskCacheRaw() = default;
 
 bool ShaderDiskCacheRaw::Load(FileUtil::IOFile& file) {
     if (file.ReadBytes(&unique_identifier, sizeof(u64)) != sizeof(u64) ||
-        file.ReadBytes(&program_type, sizeof(u32)) != sizeof(u32)) {
+        file.ReadBytes(&type, sizeof(u32)) != sizeof(u32)) {
         return false;
     }
-    u32 program_code_size{};
-    u32 program_code_size_b{};
-    if (file.ReadBytes(&program_code_size, sizeof(u32)) != sizeof(u32) ||
-        file.ReadBytes(&program_code_size_b, sizeof(u32)) != sizeof(u32)) {
+    u32 code_size{};
+    u32 code_size_b{};
+    if (file.ReadBytes(&code_size, sizeof(u32)) != sizeof(u32) ||
+        file.ReadBytes(&code_size_b, sizeof(u32)) != sizeof(u32)) {
         return false;
     }
 
-    program_code.resize(program_code_size);
-    program_code_b.resize(program_code_size_b);
+    code.resize(code_size);
+    code_b.resize(code_size_b);
 
-    if (file.ReadArray(program_code.data(), program_code_size) != program_code_size)
+    if (file.ReadArray(code.data(), code_size) != code_size)
         return false;
 
-    if (HasProgramA() &&
-        file.ReadArray(program_code_b.data(), program_code_size_b) != program_code_size_b) {
+    if (HasProgramA() && file.ReadArray(code_b.data(), code_size_b) != code_size_b) {
         return false;
     }
     return true;
 }
 
 bool ShaderDiskCacheRaw::Save(FileUtil::IOFile& file) const {
-    if (file.WriteObject(unique_identifier) != 1 ||
-        file.WriteObject(static_cast<u32>(program_type)) != 1 ||
-        file.WriteObject(static_cast<u32>(program_code.size())) != 1 ||
-        file.WriteObject(static_cast<u32>(program_code_b.size())) != 1) {
+    if (file.WriteObject(unique_identifier) != 1 || file.WriteObject(static_cast<u32>(type)) != 1 ||
+        file.WriteObject(static_cast<u32>(code.size())) != 1 ||
+        file.WriteObject(static_cast<u32>(code_b.size())) != 1) {
         return false;
     }
 
-    if (file.WriteArray(program_code.data(), program_code.size()) != program_code.size())
+    if (file.WriteArray(code.data(), code.size()) != code.size())
         return false;
 
-    if (HasProgramA() &&
-        file.WriteArray(program_code_b.data(), program_code_b.size()) != program_code_b.size()) {
+    if (HasProgramA() && file.WriteArray(code_b.data(), code_b.size()) != code_b.size()) {
         return false;
     }
     return true;

src/video_core/renderer_opengl/gl_shader_disk_cache.h

@@ -4,7 +4,6 @@
 
 #pragma once
 
-#include <bitset>
 #include <optional>
 #include <string>
 #include <tuple>
@@ -19,6 +18,7 @@
 #include "common/assert.h"
 #include "common/common_types.h"
 #include "core/file_sys/vfs_vector.h"
+#include "video_core/engines/shader_type.h"
 #include "video_core/renderer_opengl/gl_shader_gen.h"
 #include "video_core/shader/const_buffer_locker.h"
 
@@ -37,42 +37,42 @@ struct ShaderDiskCacheDump;
 
 using ProgramCode = std::vector<u64>;
 using ShaderDumpsMap = std::unordered_map<ShaderDiskCacheUsage, ShaderDiskCacheDump>;
-using TextureBufferUsage = std::bitset<64>;
 
-/// Allocated bindings used by an OpenGL shader program
-struct BaseBindings {
-    u32 cbuf{};
-    u32 gmem{};
-    u32 sampler{};
-    u32 image{};
+/// Describes the different variants a program can be compiled with.
+struct ProgramVariant final {
+    ProgramVariant() = default;
 
-    bool operator==(const BaseBindings& rhs) const {
-        return std::tie(cbuf, gmem, sampler, image) ==
-               std::tie(rhs.cbuf, rhs.gmem, rhs.sampler, rhs.image);
-    }
+    /// Graphics constructor.
+    explicit constexpr ProgramVariant(GLenum primitive_mode) noexcept
+        : primitive_mode{primitive_mode} {}
 
-    bool operator!=(const BaseBindings& rhs) const {
-        return !operator==(rhs);
-    }
-};
-static_assert(std::is_trivially_copyable_v<BaseBindings>);
+    /// Compute constructor.
+    explicit constexpr ProgramVariant(u32 block_x, u32 block_y, u32 block_z, u32 shared_memory_size,
+                                      u32 local_memory_size) noexcept
+        : block_x{block_x}, block_y{static_cast<u16>(block_y)}, block_z{static_cast<u16>(block_z)},
+          shared_memory_size{shared_memory_size}, local_memory_size{local_memory_size} {}
 
-/// Describes the different variants a single program can be compiled.
-struct ProgramVariant {
-    BaseBindings base_bindings;
+    // Graphics specific parameters.
     GLenum primitive_mode{};
-    TextureBufferUsage texture_buffer_usage{};
 
-    bool operator==(const ProgramVariant& rhs) const {
-        return std::tie(base_bindings, primitive_mode, texture_buffer_usage) ==
-               std::tie(rhs.base_bindings, rhs.primitive_mode, rhs.texture_buffer_usage);
+    // Compute specific parameters.
+    u32 block_x{};
+    u16 block_y{};
+    u16 block_z{};
+    u32 shared_memory_size{};
+    u32 local_memory_size{};
+
+    bool operator==(const ProgramVariant& rhs) const noexcept {
+        return std::tie(primitive_mode, block_x, block_y, block_z, shared_memory_size,
+                        local_memory_size) == std::tie(rhs.primitive_mode, rhs.block_x, rhs.block_y,
+                                                       rhs.block_z, rhs.shared_memory_size,
+                                                       rhs.local_memory_size);
     }
 
-    bool operator!=(const ProgramVariant& rhs) const {
+    bool operator!=(const ProgramVariant& rhs) const noexcept {
         return !operator==(rhs);
     }
 };
-
 static_assert(std::is_trivially_copyable_v<ProgramVariant>);
 
 /// Describes how a shader is used.
@@ -98,22 +98,15 @@ struct ShaderDiskCacheUsage {
 
 namespace std {
 
-template <>
-struct hash<OpenGL::BaseBindings> {
-    std::size_t operator()(const OpenGL::BaseBindings& bindings) const noexcept {
-        return static_cast<std::size_t>(bindings.cbuf) ^
-               (static_cast<std::size_t>(bindings.gmem) << 8) ^
-               (static_cast<std::size_t>(bindings.sampler) << 16) ^
-               (static_cast<std::size_t>(bindings.image) << 24);
-    }
-};
-
 template <>
 struct hash<OpenGL::ProgramVariant> {
     std::size_t operator()(const OpenGL::ProgramVariant& variant) const noexcept {
-        return std::hash<OpenGL::BaseBindings>()(variant.base_bindings) ^
-               std::hash<OpenGL::TextureBufferUsage>()(variant.texture_buffer_usage) ^
-               (static_cast<std::size_t>(variant.primitive_mode) << 6);
+        return (static_cast<std::size_t>(variant.primitive_mode) << 6) ^
+               static_cast<std::size_t>(variant.block_x) ^
+               (static_cast<std::size_t>(variant.block_y) << 32) ^
+               (static_cast<std::size_t>(variant.block_z) << 48) ^
+               (static_cast<std::size_t>(variant.shared_memory_size) << 16) ^
+               (static_cast<std::size_t>(variant.local_memory_size) << 36);
     }
 };
 
@@ -121,7 +114,7 @@ template <>
 struct hash<OpenGL::ShaderDiskCacheUsage> {
     std::size_t operator()(const OpenGL::ShaderDiskCacheUsage& usage) const noexcept {
         return static_cast<std::size_t>(usage.unique_identifier) ^
-               std::hash<OpenGL::ProgramVariant>()(usage.variant);
+               std::hash<OpenGL::ProgramVariant>{}(usage.variant);
     }
 };
 
@@ -132,8 +125,8 @@ namespace OpenGL {
 /// Describes a shader how it's used by the guest GPU
 class ShaderDiskCacheRaw {
 public:
-    explicit ShaderDiskCacheRaw(u64 unique_identifier, ProgramType program_type,
-                                ProgramCode program_code, ProgramCode program_code_b = {});
+    explicit ShaderDiskCacheRaw(u64 unique_identifier, Tegra::Engines::ShaderType type,
+                                ProgramCode code, ProgramCode code_b = {});
     ShaderDiskCacheRaw();
     ~ShaderDiskCacheRaw();
 
@@ -146,27 +139,26 @@ public:
     }
 
     bool HasProgramA() const {
-        return program_type == ProgramType::VertexA;
+        return !code.empty() && !code_b.empty();
     }
 
-    ProgramType GetProgramType() const {
-        return program_type;
+    Tegra::Engines::ShaderType GetType() const {
+        return type;
     }
 
-    const ProgramCode& GetProgramCode() const {
-        return program_code;
+    const ProgramCode& GetCode() const {
+        return code;
     }
 
-    const ProgramCode& GetProgramCodeB() const {
-        return program_code_b;
+    const ProgramCode& GetCodeB() const {
+        return code_b;
     }
 
 private:
     u64 unique_identifier{};
-    ProgramType program_type{};
-
-    ProgramCode program_code;
-    ProgramCode program_code_b;
+    Tegra::Engines::ShaderType type{};
+    ProgramCode code;
+    ProgramCode code_b;
 };
 
 /// Contains an OpenGL dumped binary program

src/video_core/renderer_opengl/gl_shader_gen.cpp

@@ -2,8 +2,13 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <string>
+
 #include <fmt/format.h>
+
 #include "video_core/engines/maxwell_3d.h"
+#include "video_core/engines/shader_type.h"
+#include "video_core/renderer_opengl/gl_device.h"
 #include "video_core/renderer_opengl/gl_shader_decompiler.h"
 #include "video_core/renderer_opengl/gl_shader_gen.h"
 #include "video_core/shader/shader_ir.h"
@@ -11,6 +16,7 @@
 namespace OpenGL::GLShader {
 
 using Tegra::Engines::Maxwell3D;
+using Tegra::Engines::ShaderType;
 using VideoCommon::Shader::CompileDepth;
 using VideoCommon::Shader::CompilerSettings;
 using VideoCommon::Shader::ProgramCode;
@@ -18,53 +24,40 @@ using VideoCommon::Shader::ShaderIR;
 
 std::string GenerateVertexShader(const Device& device, const ShaderIR& ir, const ShaderIR* ir_b) {
     std::string out = GetCommonDeclarations();
-    out += R"(
-layout (std140, binding = EMULATION_UBO_BINDING) uniform vs_config {
-    vec4 viewport_flip;
-    uvec4 config_pack; // instance_id, flip_stage, y_direction, padding
-};
+    out += fmt::format(R"(
+layout (std140, binding = {}) uniform vs_config {{
+    float y_direction;
+}};
 
-)";
-    const auto stage = ir_b ? ProgramType::VertexA : ProgramType::VertexB;
-    out += Decompile(device, ir, stage, "vertex");
+)",
+                       EmulationUniformBlockBinding);
+    out += Decompile(device, ir, ShaderType::Vertex, "vertex");
     if (ir_b) {
-        out += Decompile(device, *ir_b, ProgramType::VertexB, "vertex_b");
+        out += Decompile(device, *ir_b, ShaderType::Vertex, "vertex_b");
     }
 
     out += R"(
 void main() {
+    gl_Position = vec4(0.0f, 0.0f, 0.0f, 1.0f);
     execute_vertex();
 )";
-
     if (ir_b) {
         out += "    execute_vertex_b();";
     }
-
-    out += R"(
-
-    // Set Position Y direction
-    gl_Position.y *= utof(config_pack[2]);
-    // Check if the flip stage is VertexB
-    // Config pack's second value is flip_stage
-    if (config_pack[1] == 1) {
-        // Viewport can be flipped, which is unsupported by glViewport
-        gl_Position.xy *= viewport_flip.xy;
-    }
-}
-)";
+    out += "}\n";
     return out;
 }
 
 std::string GenerateGeometryShader(const Device& device, const ShaderIR& ir) {
     std::string out = GetCommonDeclarations();
-    out += R"(
-layout (std140, binding = EMULATION_UBO_BINDING) uniform gs_config {
-    vec4 viewport_flip;
-    uvec4 config_pack; // instance_id, flip_stage, y_direction, padding
-};
+    out += fmt::format(R"(
+layout (std140, binding = {}) uniform gs_config {{
+    float y_direction;
+}};
 
-)";
-    out += Decompile(device, ir, ProgramType::Geometry, "geometry");
+)",
+                       EmulationUniformBlockBinding);
+    out += Decompile(device, ir, ShaderType::Geometry, "geometry");
 
     out += R"(
 void main() {
@@ -76,7 +69,7 @@ void main() {
 
 std::string GenerateFragmentShader(const Device& device, const ShaderIR& ir) {
     std::string out = GetCommonDeclarations();
-    out += R"(
+    out += fmt::format(R"(
 layout (location = 0) out vec4 FragColor0;
 layout (location = 1) out vec4 FragColor1;
 layout (location = 2) out vec4 FragColor2;
@@ -86,13 +79,13 @@ layout (location = 5) out vec4 FragColor5;
 layout (location = 6) out vec4 FragColor6;
 layout (location = 7) out vec4 FragColor7;
 
-layout (std140, binding = EMULATION_UBO_BINDING) uniform fs_config {
-    vec4 viewport_flip;
-    uvec4 config_pack; // instance_id, flip_stage, y_direction, padding
-};
+layout (std140, binding = {}) uniform fs_config {{
+    float y_direction;
+}};
 
-)";
-    out += Decompile(device, ir, ProgramType::Fragment, "fragment");
+)",
+                       EmulationUniformBlockBinding);
+    out += Decompile(device, ir, ShaderType::Fragment, "fragment");
 
     out += R"(
 void main() {
@@ -104,7 +97,7 @@ void main() {
 
 std::string GenerateComputeShader(const Device& device, const ShaderIR& ir) {
     std::string out = GetCommonDeclarations();
-    out += Decompile(device, ir, ProgramType::Compute, "compute");
+    out += Decompile(device, ir, ShaderType::Compute, "compute");
     out += R"(
 void main() {
     execute_compute();

src/video_core/renderer_opengl/gl_shader_manager.cpp

@@ -40,27 +40,11 @@ void ProgramManager::UpdatePipeline() {
     old_state = current_state;
 }
 
-void MaxwellUniformData::SetFromRegs(const Maxwell3D& maxwell, std::size_t shader_stage) {
+void MaxwellUniformData::SetFromRegs(const Maxwell3D& maxwell) {
     const auto& regs = maxwell.regs;
-    const auto& state = maxwell.state;
-
-    // TODO(bunnei): Support more than one viewport
-    viewport_flip[0] = regs.viewport_transform[0].scale_x < 0.0 ? -1.0f : 1.0f;
-    viewport_flip[1] = regs.viewport_transform[0].scale_y < 0.0 ? -1.0f : 1.0f;
-
-    instance_id = state.current_instance;
-
-    // Assign in which stage the position has to be flipped
-    // (the last stage before the fragment shader).
-    constexpr u32 geometry_index = static_cast<u32>(Maxwell3D::Regs::ShaderProgram::Geometry);
-    if (maxwell.regs.shader_config[geometry_index].enable) {
-        flip_stage = geometry_index;
-    } else {
-        flip_stage = static_cast<u32>(Maxwell3D::Regs::ShaderProgram::VertexB);
-    }
 
     // Y_NEGATE controls what value S2R returns for the Y_DIRECTION system value.
-    y_direction = regs.screen_y_control.y_negate == 0 ? 1.f : -1.f;
+    y_direction = regs.screen_y_control.y_negate == 0 ? 1.0f : -1.0f;
 }
 
 } // namespace OpenGL::GLShader

src/video_core/renderer_opengl/gl_shader_manager.h

@@ -18,17 +18,12 @@ namespace OpenGL::GLShader {
 /// @note Always keep a vec4 at the end. The GL spec is not clear whether the alignment at
 ///       the end of a uniform block is included in UNIFORM_BLOCK_DATA_SIZE or not.
 ///       Not following that rule will cause problems on some AMD drivers.
-struct MaxwellUniformData {
-    void SetFromRegs(const Tegra::Engines::Maxwell3D& maxwell, std::size_t shader_stage);
+struct alignas(16) MaxwellUniformData {
+    void SetFromRegs(const Tegra::Engines::Maxwell3D& maxwell);
 
-    alignas(16) GLvec4 viewport_flip;
-    struct alignas(16) {
-        GLuint instance_id;
-        GLuint flip_stage;
-        GLfloat y_direction;
-    };
+    GLfloat y_direction;
 };
-static_assert(sizeof(MaxwellUniformData) == 32, "MaxwellUniformData structure size is incorrect");
+static_assert(sizeof(MaxwellUniformData) == 16, "MaxwellUniformData structure size is incorrect");
 static_assert(sizeof(MaxwellUniformData) < 16384,
               "MaxwellUniformData structure must be less than 16kb as per the OpenGL spec");
 

src/video_core/renderer_opengl/gl_state.cpp

@@ -410,15 +410,31 @@ void OpenGLState::ApplyAlphaTest() {
     }
 }
 
+void OpenGLState::ApplyClipControl() {
+    if (UpdateValue(cur_state.clip_control.origin, clip_control.origin)) {
+        glClipControl(clip_control.origin, GL_NEGATIVE_ONE_TO_ONE);
+    }
+}
+
 void OpenGLState::ApplyTextures() {
-    if (const auto update = UpdateArray(cur_state.textures, textures)) {
-        glBindTextures(update->first, update->second, textures.data() + update->first);
+    const std::size_t size = std::size(textures);
+    for (std::size_t i = 0; i < size; ++i) {
+        if (UpdateValue(cur_state.textures[i], textures[i])) {
+            // BindTextureUnit doesn't support binding null textures, skip those binds.
+            // TODO(Rodrigo): Stop using null textures
+            if (textures[i] != 0) {
+                glBindTextureUnit(static_cast<GLuint>(i), textures[i]);
+            }
+        }
     }
 }
 
 void OpenGLState::ApplySamplers() {
-    if (const auto update = UpdateArray(cur_state.samplers, samplers)) {
-        glBindSamplers(update->first, update->second, samplers.data() + update->first);
+    const std::size_t size = std::size(samplers);
+    for (std::size_t i = 0; i < size; ++i) {
+        if (UpdateValue(cur_state.samplers[i], samplers[i])) {
+            glBindSampler(static_cast<GLuint>(i), samplers[i]);
+        }
     }
 }
 
@@ -453,6 +469,7 @@ void OpenGLState::Apply() {
     ApplyImages();
     ApplyPolygonOffset();
     ApplyAlphaTest();
+    ApplyClipControl();
 }
 
 void OpenGLState::EmulateViewportWithScissor() {

src/video_core/renderer_opengl/gl_state.h

@@ -96,9 +96,11 @@ public:
         GLenum operation = GL_COPY;
     } logic_op;
 
-    std::array<GLuint, Tegra::Engines::Maxwell3D::Regs::NumTextureSamplers> textures = {};
-    std::array<GLuint, Tegra::Engines::Maxwell3D::Regs::NumTextureSamplers> samplers = {};
-    std::array<GLuint, Tegra::Engines::Maxwell3D::Regs::NumImages> images = {};
+    static constexpr std::size_t NumSamplers = 32 * 5;
+    static constexpr std::size_t NumImages = 8 * 5;
+    std::array<GLuint, NumSamplers> textures = {};
+    std::array<GLuint, NumSamplers> samplers = {};
+    std::array<GLuint, NumImages> images = {};
 
     struct {
         GLuint read_framebuffer = 0; // GL_READ_FRAMEBUFFER_BINDING
@@ -146,6 +148,10 @@ public:
 
     std::array<bool, 8> clip_distance = {}; // GL_CLIP_DISTANCE
 
+    struct {
+        GLenum origin = GL_LOWER_LEFT;
+    } clip_control;
+
     OpenGLState();
 
     /// Get the currently active OpenGL state
@@ -182,6 +188,7 @@ public:
     void ApplyDepthClamp();
     void ApplyPolygonOffset();
     void ApplyAlphaTest();
+    void ApplyClipControl();
 
     /// Resets any references to the given resource
     OpenGLState& UnbindTexture(GLuint handle);

src/video_core/renderer_opengl/gl_texture_cache.cpp

@@ -23,7 +23,6 @@ namespace OpenGL {
 using Tegra::Texture::SwizzleSource;
 using VideoCore::MortonSwizzleMode;
 
-using VideoCore::Surface::ComponentType;
 using VideoCore::Surface::PixelFormat;
 using VideoCore::Surface::SurfaceCompression;
 using VideoCore::Surface::SurfaceTarget;
@@ -40,114 +39,95 @@ struct FormatTuple {
     GLint internal_format;
     GLenum format;
     GLenum type;
-    ComponentType component_type;
     bool compressed;
 };
 
 constexpr std::array<FormatTuple, VideoCore::Surface::MaxPixelFormat> tex_format_tuples = {{
-    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV, ComponentType::UNorm, false}, // ABGR8U
-    {GL_RGBA8, GL_RGBA, GL_BYTE, ComponentType::SNorm, false},                     // ABGR8S
-    {GL_RGBA8UI, GL_RGBA_INTEGER, GL_UNSIGNED_BYTE, ComponentType::UInt, false},   // ABGR8UI
-    {GL_RGB565, GL_RGB, GL_UNSIGNED_SHORT_5_6_5_REV, ComponentType::UNorm, false}, // B5G6R5U
-    {GL_RGB10_A2, GL_RGBA, GL_UNSIGNED_INT_2_10_10_10_REV, ComponentType::UNorm,
-     false}, // A2B10G10R10U
-    {GL_RGB5_A1, GL_RGBA, GL_UNSIGNED_SHORT_1_5_5_5_REV, ComponentType::UNorm, false}, // A1B5G5R5U
-    {GL_R8, GL_RED, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},                    // R8U
-    {GL_R8UI, GL_RED_INTEGER, GL_UNSIGNED_BYTE, ComponentType::UInt, false},           // R8UI
-    {GL_RGBA16F, GL_RGBA, GL_HALF_FLOAT, ComponentType::Float, false},                 // RGBA16F
-    {GL_RGBA16, GL_RGBA, GL_UNSIGNED_SHORT, ComponentType::UNorm, false},              // RGBA16U
-    {GL_RGBA16UI, GL_RGBA_INTEGER, GL_UNSIGNED_SHORT, ComponentType::UInt, false},     // RGBA16UI
-    {GL_R11F_G11F_B10F, GL_RGB, GL_UNSIGNED_INT_10F_11F_11F_REV, ComponentType::Float,
-     false},                                                                     // R11FG11FB10F
-    {GL_RGBA32UI, GL_RGBA_INTEGER, GL_UNSIGNED_INT, ComponentType::UInt, false}, // RGBA32UI
-    {GL_COMPRESSED_RGBA_S3TC_DXT1_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
-     true}, // DXT1
-    {GL_COMPRESSED_RGBA_S3TC_DXT3_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
-     true}, // DXT23
-    {GL_COMPRESSED_RGBA_S3TC_DXT5_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
-     true},                                                                                 // DXT45
-    {GL_COMPRESSED_RED_RGTC1, GL_RED, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm, true}, // DXN1
-    {GL_COMPRESSED_RG_RGTC2, GL_RG, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
-     true},                                                                     // DXN2UNORM
-    {GL_COMPRESSED_SIGNED_RG_RGTC2, GL_RG, GL_INT, ComponentType::SNorm, true}, // DXN2SNORM
-    {GL_COMPRESSED_RGBA_BPTC_UNORM, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
-     true}, // BC7U
-    {GL_COMPRESSED_RGB_BPTC_UNSIGNED_FLOAT, GL_RGB, GL_UNSIGNED_INT_8_8_8_8, ComponentType::Float,
-     true}, // BC6H_UF16
-    {GL_COMPRESSED_RGB_BPTC_SIGNED_FLOAT, GL_RGB, GL_UNSIGNED_INT_8_8_8_8, ComponentType::Float,
-     true},                                                                    // BC6H_SF16
-    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},        // ASTC_2D_4X4
-    {GL_RGBA8, GL_BGRA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},        // BGRA8
-    {GL_RGBA32F, GL_RGBA, GL_FLOAT, ComponentType::Float, false},              // RGBA32F
-    {GL_RG32F, GL_RG, GL_FLOAT, ComponentType::Float, false},                  // RG32F
-    {GL_R32F, GL_RED, GL_FLOAT, ComponentType::Float, false},                  // R32F
-    {GL_R16F, GL_RED, GL_HALF_FLOAT, ComponentType::Float, false},             // R16F
-    {GL_R16, GL_RED, GL_UNSIGNED_SHORT, ComponentType::UNorm, false},          // R16U
-    {GL_R16_SNORM, GL_RED, GL_SHORT, ComponentType::SNorm, false},             // R16S
-    {GL_R16UI, GL_RED_INTEGER, GL_UNSIGNED_SHORT, ComponentType::UInt, false}, // R16UI
-    {GL_R16I, GL_RED_INTEGER, GL_SHORT, ComponentType::SInt, false},           // R16I
-    {GL_RG16, GL_RG, GL_UNSIGNED_SHORT, ComponentType::UNorm, false},          // RG16
-    {GL_RG16F, GL_RG, GL_HALF_FLOAT, ComponentType::Float, false},             // RG16F
-    {GL_RG16UI, GL_RG_INTEGER, GL_UNSIGNED_SHORT, ComponentType::UInt, false}, // RG16UI
-    {GL_RG16I, GL_RG_INTEGER, GL_SHORT, ComponentType::SInt, false},           // RG16I
-    {GL_RG16_SNORM, GL_RG, GL_SHORT, ComponentType::SNorm, false},             // RG16S
-    {GL_RGB32F, GL_RGB, GL_FLOAT, ComponentType::Float, false},                // RGB32F
-    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV, ComponentType::UNorm,
-     false},                                                                   // RGBA8_SRGB
-    {GL_RG8, GL_RG, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},            // RG8U
-    {GL_RG8, GL_RG, GL_BYTE, ComponentType::SNorm, false},                     // RG8S
-    {GL_RG32UI, GL_RG_INTEGER, GL_UNSIGNED_INT, ComponentType::UInt, false},   // RG32UI
-    {GL_RGB16F, GL_RGBA16, GL_HALF_FLOAT, ComponentType::Float, false},        // RGBX16F
-    {GL_R32UI, GL_RED_INTEGER, GL_UNSIGNED_INT, ComponentType::UInt, false},   // R32UI
-    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},        // ASTC_2D_8X8
-    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},        // ASTC_2D_8X5
-    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},        // ASTC_2D_5X4
-    {GL_SRGB8_ALPHA8, GL_BGRA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // BGRA8
+    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV, false},                        // ABGR8U
+    {GL_RGBA8, GL_RGBA, GL_BYTE, false},                                            // ABGR8S
+    {GL_RGBA8UI, GL_RGBA_INTEGER, GL_UNSIGNED_BYTE, false},                         // ABGR8UI
+    {GL_RGB565, GL_RGB, GL_UNSIGNED_SHORT_5_6_5_REV, false},                        // B5G6R5U
+    {GL_RGB10_A2, GL_RGBA, GL_UNSIGNED_INT_2_10_10_10_REV, false},                  // A2B10G10R10U
+    {GL_RGB5_A1, GL_RGBA, GL_UNSIGNED_SHORT_1_5_5_5_REV, false},                    // A1B5G5R5U
+    {GL_R8, GL_RED, GL_UNSIGNED_BYTE, false},                                       // R8U
+    {GL_R8UI, GL_RED_INTEGER, GL_UNSIGNED_BYTE, false},                             // R8UI
+    {GL_RGBA16F, GL_RGBA, GL_HALF_FLOAT, false},                                    // RGBA16F
+    {GL_RGBA16, GL_RGBA, GL_UNSIGNED_SHORT, false},                                 // RGBA16U
+    {GL_RGBA16UI, GL_RGBA_INTEGER, GL_UNSIGNED_SHORT, false},                       // RGBA16UI
+    {GL_R11F_G11F_B10F, GL_RGB, GL_UNSIGNED_INT_10F_11F_11F_REV, false},            // R11FG11FB10F
+    {GL_RGBA32UI, GL_RGBA_INTEGER, GL_UNSIGNED_INT, false},                         // RGBA32UI
+    {GL_COMPRESSED_RGBA_S3TC_DXT1_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, true},     // DXT1
+    {GL_COMPRESSED_RGBA_S3TC_DXT3_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, true},     // DXT23
+    {GL_COMPRESSED_RGBA_S3TC_DXT5_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, true},     // DXT45
+    {GL_COMPRESSED_RED_RGTC1, GL_RED, GL_UNSIGNED_INT_8_8_8_8, true},               // DXN1
+    {GL_COMPRESSED_RG_RGTC2, GL_RG, GL_UNSIGNED_INT_8_8_8_8, true},                 // DXN2UNORM
+    {GL_COMPRESSED_SIGNED_RG_RGTC2, GL_RG, GL_INT, true},                           // DXN2SNORM
+    {GL_COMPRESSED_RGBA_BPTC_UNORM, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, true},        // BC7U
+    {GL_COMPRESSED_RGB_BPTC_UNSIGNED_FLOAT, GL_RGB, GL_UNSIGNED_INT_8_8_8_8, true}, // BC6H_UF16
+    {GL_COMPRESSED_RGB_BPTC_SIGNED_FLOAT, GL_RGB, GL_UNSIGNED_INT_8_8_8_8, true},   // BC6H_SF16
+    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, false},                                   // ASTC_2D_4X4
+    {GL_RGBA8, GL_BGRA, GL_UNSIGNED_BYTE, false},                                   // BGRA8
+    {GL_RGBA32F, GL_RGBA, GL_FLOAT, false},                                         // RGBA32F
+    {GL_RG32F, GL_RG, GL_FLOAT, false},                                             // RG32F
+    {GL_R32F, GL_RED, GL_FLOAT, false},                                             // R32F
+    {GL_R16F, GL_RED, GL_HALF_FLOAT, false},                                        // R16F
+    {GL_R16, GL_RED, GL_UNSIGNED_SHORT, false},                                     // R16U
+    {GL_R16_SNORM, GL_RED, GL_SHORT, false},                                        // R16S
+    {GL_R16UI, GL_RED_INTEGER, GL_UNSIGNED_SHORT, false},                           // R16UI
+    {GL_R16I, GL_RED_INTEGER, GL_SHORT, false},                                     // R16I
+    {GL_RG16, GL_RG, GL_UNSIGNED_SHORT, false},                                     // RG16
+    {GL_RG16F, GL_RG, GL_HALF_FLOAT, false},                                        // RG16F
+    {GL_RG16UI, GL_RG_INTEGER, GL_UNSIGNED_SHORT, false},                           // RG16UI
+    {GL_RG16I, GL_RG_INTEGER, GL_SHORT, false},                                     // RG16I
+    {GL_RG16_SNORM, GL_RG, GL_SHORT, false},                                        // RG16S
+    {GL_RGB32F, GL_RGB, GL_FLOAT, false},                                           // RGB32F
+    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV, false},                 // RGBA8_SRGB
+    {GL_RG8, GL_RG, GL_UNSIGNED_BYTE, false},                                       // RG8U
+    {GL_RG8, GL_RG, GL_BYTE, false},                                                // RG8S
+    {GL_RG32UI, GL_RG_INTEGER, GL_UNSIGNED_INT, false},                             // RG32UI
+    {GL_RGB16F, GL_RGBA16, GL_HALF_FLOAT, false},                                   // RGBX16F
+    {GL_R32UI, GL_RED_INTEGER, GL_UNSIGNED_INT, false},                             // R32UI
+    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, false},                                   // ASTC_2D_8X8
+    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, false},                                   // ASTC_2D_8X5
+    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, false},                                   // ASTC_2D_5X4
+    {GL_SRGB8_ALPHA8, GL_BGRA, GL_UNSIGNED_BYTE, false},                            // BGRA8
     // Compressed sRGB formats
-    {GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT1_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
-     true}, // DXT1_SRGB
-    {GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT3_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
-     true}, // DXT23_SRGB
-    {GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
-     true}, // DXT45_SRGB
-    {GL_COMPRESSED_SRGB_ALPHA_BPTC_UNORM, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
-     true},                                                                          // BC7U_SRGB
-    {GL_RGBA4, GL_RGBA, GL_UNSIGNED_SHORT_4_4_4_4_REV, ComponentType::UNorm, false}, // R4G4B4A4U
-    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_4X4_SRGB
-    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_8X8_SRGB
-    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_8X5_SRGB
-    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_5X4_SRGB
-    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},        // ASTC_2D_5X5
-    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_5X5_SRGB
-    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},        // ASTC_2D_10X8
-    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_10X8_SRGB
-    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},        // ASTC_2D_6X6
-    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_6X6_SRGB
-    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},        // ASTC_2D_10X10
-    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_10X10_SRGB
-    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},        // ASTC_2D_12X12
-    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_12X12_SRGB
-    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},        // ASTC_2D_8X6
-    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_8X6_SRGB
-    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},        // ASTC_2D_6X5
-    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_6X5_SRGB
-    {GL_RGB9_E5, GL_RGB, GL_UNSIGNED_INT_5_9_9_9_REV, ComponentType::Float, false}, // E5B9G9R9F
+    {GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT1_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, true}, // DXT1_SRGB
+    {GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT3_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, true}, // DXT23_SRGB
+    {GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, true}, // DXT45_SRGB
+    {GL_COMPRESSED_SRGB_ALPHA_BPTC_UNORM, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, true},    // BC7U_SRGB
+    {GL_RGBA4, GL_RGBA, GL_UNSIGNED_SHORT_4_4_4_4_REV, false},                        // R4G4B4A4U
+    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, false},      // ASTC_2D_4X4_SRGB
+    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, false},      // ASTC_2D_8X8_SRGB
+    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, false},      // ASTC_2D_8X5_SRGB
+    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, false},      // ASTC_2D_5X4_SRGB
+    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, false},             // ASTC_2D_5X5
+    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, false},      // ASTC_2D_5X5_SRGB
+    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, false},             // ASTC_2D_10X8
+    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, false},      // ASTC_2D_10X8_SRGB
+    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, false},             // ASTC_2D_6X6
+    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, false},      // ASTC_2D_6X6_SRGB
+    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, false},             // ASTC_2D_10X10
+    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, false},      // ASTC_2D_10X10_SRGB
+    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, false},             // ASTC_2D_12X12
+    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, false},      // ASTC_2D_12X12_SRGB
+    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, false},             // ASTC_2D_8X6
+    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, false},      // ASTC_2D_8X6_SRGB
+    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, false},             // ASTC_2D_6X5
+    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, false},      // ASTC_2D_6X5_SRGB
+    {GL_RGB9_E5, GL_RGB, GL_UNSIGNED_INT_5_9_9_9_REV, false}, // E5B9G9R9F
 
     // Depth formats
-    {GL_DEPTH_COMPONENT32F, GL_DEPTH_COMPONENT, GL_FLOAT, ComponentType::Float, false}, // Z32F
-    {GL_DEPTH_COMPONENT16, GL_DEPTH_COMPONENT, GL_UNSIGNED_SHORT, ComponentType::UNorm,
-     false}, // Z16
+    {GL_DEPTH_COMPONENT32F, GL_DEPTH_COMPONENT, GL_FLOAT, false},         // Z32F
+    {GL_DEPTH_COMPONENT16, GL_DEPTH_COMPONENT, GL_UNSIGNED_SHORT, false}, // Z16
 
     // DepthStencil formats
-    {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8, ComponentType::UNorm,
-     false}, // Z24S8
-    {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8, ComponentType::UNorm,
-     false}, // S8Z24
-    {GL_DEPTH32F_STENCIL8, GL_DEPTH_STENCIL, GL_FLOAT_32_UNSIGNED_INT_24_8_REV,
-     ComponentType::Float, false}, // Z32FS8
+    {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8, false},               // Z24S8
+    {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8, false},               // S8Z24
+    {GL_DEPTH32F_STENCIL8, GL_DEPTH_STENCIL, GL_FLOAT_32_UNSIGNED_INT_24_8_REV, false}, // Z32FS8
 }};
 
-const FormatTuple& GetFormatTuple(PixelFormat pixel_format, ComponentType component_type) {
+const FormatTuple& GetFormatTuple(PixelFormat pixel_format) {
     ASSERT(static_cast<std::size_t>(pixel_format) < tex_format_tuples.size());
     const auto& format{tex_format_tuples[static_cast<std::size_t>(pixel_format)]};
     return format;
@@ -249,7 +229,7 @@ OGLTexture CreateTexture(const SurfaceParams& params, GLenum target, GLenum inte
 
 CachedSurface::CachedSurface(const GPUVAddr gpu_addr, const SurfaceParams& params)
     : VideoCommon::SurfaceBase<View>(gpu_addr, params) {
-    const auto& tuple{GetFormatTuple(params.pixel_format, params.component_type)};
+    const auto& tuple{GetFormatTuple(params.pixel_format)};
     internal_format = tuple.internal_format;
     format = tuple.format;
     type = tuple.type;
@@ -451,8 +431,7 @@ OGLTextureView CachedSurfaceView::CreateTextureView() const {
     texture_view.Create();
 
     const GLuint handle{texture_view.handle};
-    const FormatTuple& tuple{
-        GetFormatTuple(owner_params.pixel_format, owner_params.component_type)};
+    const FormatTuple& tuple{GetFormatTuple(owner_params.pixel_format)};
 
     glTextureView(handle, target, surface.texture.handle, tuple.internal_format, params.base_level,
                   params.num_levels, params.base_layer, params.num_layers);
@@ -562,8 +541,8 @@ void TextureCacheOpenGL::BufferCopy(Surface& src_surface, Surface& dst_surface)
     const auto& dst_params = dst_surface->GetSurfaceParams();
     UNIMPLEMENTED_IF(src_params.num_levels > 1 || dst_params.num_levels > 1);
 
-    const auto source_format = GetFormatTuple(src_params.pixel_format, src_params.component_type);
-    const auto dest_format = GetFormatTuple(dst_params.pixel_format, dst_params.component_type);
+    const auto source_format = GetFormatTuple(src_params.pixel_format);
+    const auto dest_format = GetFormatTuple(dst_params.pixel_format);
 
     const std::size_t source_size = src_surface->GetHostSizeInBytes();
     const std::size_t dest_size = dst_surface->GetHostSizeInBytes();

src/video_core/renderer_opengl/renderer_opengl.cpp

@@ -323,10 +323,12 @@ void RendererOpenGL::DrawScreenTriangles(const ScreenInfo& screen_info, float x,
     // (e.g. handheld mode) on a 1920x1080 framebuffer.
     f32 scale_u = 1.f, scale_v = 1.f;
     if (framebuffer_crop_rect.GetWidth() > 0) {
-        scale_u = static_cast<f32>(framebuffer_crop_rect.GetWidth()) / screen_info.texture.width;
+        scale_u = static_cast<f32>(framebuffer_crop_rect.GetWidth()) /
+                  static_cast<f32>(screen_info.texture.width);
     }
     if (framebuffer_crop_rect.GetHeight() > 0) {
-        scale_v = static_cast<f32>(framebuffer_crop_rect.GetHeight()) / screen_info.texture.height;
+        scale_v = static_cast<f32>(framebuffer_crop_rect.GetHeight()) /
+                  static_cast<f32>(screen_info.texture.height);
     }
 
     std::array<ScreenRectVertex, 4> vertices = {{

src/video_core/renderer_opengl/utils.cpp

@@ -3,7 +3,10 @@
 // Refer to the license.txt file included.
 
 #include <string>
+#include <vector>
+
 #include <fmt/format.h>
+
 #include <glad/glad.h>
 
 #include "common/assert.h"
@@ -48,34 +51,19 @@ BindBuffersRangePushBuffer::BindBuffersRangePushBuffer(GLenum target) : target{t
 
 BindBuffersRangePushBuffer::~BindBuffersRangePushBuffer() = default;
 
-void BindBuffersRangePushBuffer::Setup(GLuint first_) {
-    first = first_;
-    buffer_pointers.clear();
-    offsets.clear();
-    sizes.clear();
+void BindBuffersRangePushBuffer::Setup() {
+    entries.clear();
 }
 
-void BindBuffersRangePushBuffer::Push(const GLuint* buffer, GLintptr offset, GLsizeiptr size) {
-    buffer_pointers.push_back(buffer);
-    offsets.push_back(offset);
-    sizes.push_back(size);
+void BindBuffersRangePushBuffer::Push(GLuint binding, const GLuint* buffer, GLintptr offset,
+                                      GLsizeiptr size) {
+    entries.push_back(Entry{binding, buffer, offset, size});
 }
 
 void BindBuffersRangePushBuffer::Bind() {
-    // Ensure sizes are valid.
-    const std::size_t count{buffer_pointers.size()};
-    DEBUG_ASSERT(count == offsets.size() && count == sizes.size());
-    if (count == 0) {
-        return;
+    for (const Entry& entry : entries) {
+        glBindBufferRange(target, entry.binding, *entry.buffer, entry.offset, entry.size);
     }
-
-    // Dereference buffers.
-    buffers.resize(count);
-    std::transform(buffer_pointers.begin(), buffer_pointers.end(), buffers.begin(),
-                   [](const GLuint* pointer) { return *pointer; });
-
-    glBindBuffersRange(target, first, static_cast<GLsizei>(count), buffers.data(), offsets.data(),
-                       sizes.data());
 }
 
 void LabelGLObject(GLenum identifier, GLuint handle, VAddr addr, std::string_view extra_info) {

src/video_core/renderer_opengl/utils.h

@@ -43,20 +43,22 @@ public:
     explicit BindBuffersRangePushBuffer(GLenum target);
     ~BindBuffersRangePushBuffer();
 
-    void Setup(GLuint first_);
+    void Setup();
 
-    void Push(const GLuint* buffer, GLintptr offset, GLsizeiptr size);
+    void Push(GLuint binding, const GLuint* buffer, GLintptr offset, GLsizeiptr size);
 
     void Bind();
 
 private:
-    GLenum target{};
-    GLuint first{};
-    std::vector<const GLuint*> buffer_pointers;
+    struct Entry {
+        GLuint binding;
+        const GLuint* buffer;
+        GLintptr offset;
+        GLsizeiptr size;
+    };
 
-    std::vector<GLuint> buffers;
-    std::vector<GLintptr> offsets;
-    std::vector<GLsizeiptr> sizes;
+    GLenum target;
+    std::vector<Entry> entries;
 };
 
 void LabelGLObject(GLenum identifier, GLuint handle, VAddr addr, std::string_view extra_info = {});

src/video_core/renderer_vulkan/maxwell_to_vk.cpp

@@ -13,6 +13,8 @@
 
 namespace Vulkan::MaxwellToVK {
 
+using Maxwell = Tegra::Engines::Maxwell3D::Regs;
+
 namespace Sampler {
 
 vk::Filter Filter(Tegra::Texture::TextureFilter filter) {
@@ -95,83 +97,82 @@ vk::CompareOp DepthCompareFunction(Tegra::Texture::DepthCompareFunc depth_compar
 } // namespace Sampler
 
 struct FormatTuple {
-    vk::Format format;            ///< Vulkan format
-    ComponentType component_type; ///< Abstracted component type
-    bool attachable;              ///< True when this format can be used as an attachment
+    vk::Format format; ///< Vulkan format
+    bool attachable;   ///< True when this format can be used as an attachment
 };
 
 static constexpr std::array<FormatTuple, VideoCore::Surface::MaxPixelFormat> tex_format_tuples = {{
-    {vk::Format::eA8B8G8R8UnormPack32, ComponentType::UNorm, true},    // ABGR8U
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // ABGR8S
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // ABGR8UI
-    {vk::Format::eB5G6R5UnormPack16, ComponentType::UNorm, false},     // B5G6R5U
-    {vk::Format::eA2B10G10R10UnormPack32, ComponentType::UNorm, true}, // A2B10G10R10U
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // A1B5G5R5U
-    {vk::Format::eR8Unorm, ComponentType::UNorm, true},                // R8U
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // R8UI
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // RGBA16F
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // RGBA16U
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // RGBA16UI
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // R11FG11FB10F
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // RGBA32UI
-    {vk::Format::eBc1RgbaUnormBlock, ComponentType::UNorm, false},     // DXT1
-    {vk::Format::eBc2UnormBlock, ComponentType::UNorm, false},         // DXT23
-    {vk::Format::eBc3UnormBlock, ComponentType::UNorm, false},         // DXT45
-    {vk::Format::eBc4UnormBlock, ComponentType::UNorm, false},         // DXN1
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // DXN2UNORM
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // DXN2SNORM
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // BC7U
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // BC6H_UF16
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // BC6H_SF16
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // ASTC_2D_4X4
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // BGRA8
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // RGBA32F
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // RG32F
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // R32F
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // R16F
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // R16U
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // R16S
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // R16UI
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // R16I
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // RG16
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // RG16F
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // RG16UI
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // RG16I
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // RG16S
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // RGB32F
-    {vk::Format::eA8B8G8R8SrgbPack32, ComponentType::UNorm, true},     // RGBA8_SRGB
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // RG8U
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // RG8S
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // RG32UI
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // RGBX16F
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // R32UI
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // ASTC_2D_8X8
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // ASTC_2D_8X5
-    {vk::Format::eUndefined, ComponentType::Invalid, false},           // ASTC_2D_5X4
+    {vk::Format::eA8B8G8R8UnormPack32, true},    // ABGR8U
+    {vk::Format::eUndefined, false},             // ABGR8S
+    {vk::Format::eUndefined, false},             // ABGR8UI
+    {vk::Format::eB5G6R5UnormPack16, false},     // B5G6R5U
+    {vk::Format::eA2B10G10R10UnormPack32, true}, // A2B10G10R10U
+    {vk::Format::eUndefined, false},             // A1B5G5R5U
+    {vk::Format::eR8Unorm, true},                // R8U
+    {vk::Format::eUndefined, false},             // R8UI
+    {vk::Format::eUndefined, false},             // RGBA16F
+    {vk::Format::eUndefined, false},             // RGBA16U
+    {vk::Format::eUndefined, false},             // RGBA16UI
+    {vk::Format::eUndefined, false},             // R11FG11FB10F
+    {vk::Format::eUndefined, false},             // RGBA32UI
+    {vk::Format::eBc1RgbaUnormBlock, false},     // DXT1
+    {vk::Format::eBc2UnormBlock, false},         // DXT23
+    {vk::Format::eBc3UnormBlock, false},         // DXT45
+    {vk::Format::eBc4UnormBlock, false},         // DXN1
+    {vk::Format::eUndefined, false},             // DXN2UNORM
+    {vk::Format::eUndefined, false},             // DXN2SNORM
+    {vk::Format::eUndefined, false},             // BC7U
+    {vk::Format::eUndefined, false},             // BC6H_UF16
+    {vk::Format::eUndefined, false},             // BC6H_SF16
+    {vk::Format::eUndefined, false},             // ASTC_2D_4X4
+    {vk::Format::eUndefined, false},             // BGRA8
+    {vk::Format::eUndefined, false},             // RGBA32F
+    {vk::Format::eUndefined, false},             // RG32F
+    {vk::Format::eUndefined, false},             // R32F
+    {vk::Format::eUndefined, false},             // R16F
+    {vk::Format::eUndefined, false},             // R16U
+    {vk::Format::eUndefined, false},             // R16S
+    {vk::Format::eUndefined, false},             // R16UI
+    {vk::Format::eUndefined, false},             // R16I
+    {vk::Format::eUndefined, false},             // RG16
+    {vk::Format::eUndefined, false},             // RG16F
+    {vk::Format::eUndefined, false},             // RG16UI
+    {vk::Format::eUndefined, false},             // RG16I
+    {vk::Format::eUndefined, false},             // RG16S
+    {vk::Format::eUndefined, false},             // RGB32F
+    {vk::Format::eA8B8G8R8SrgbPack32, true},     // RGBA8_SRGB
+    {vk::Format::eUndefined, false},             // RG8U
+    {vk::Format::eUndefined, false},             // RG8S
+    {vk::Format::eUndefined, false},             // RG32UI
+    {vk::Format::eUndefined, false},             // RGBX16F
+    {vk::Format::eUndefined, false},             // R32UI
+    {vk::Format::eUndefined, false},             // ASTC_2D_8X8
+    {vk::Format::eUndefined, false},             // ASTC_2D_8X5
+    {vk::Format::eUndefined, false},             // ASTC_2D_5X4
 
     // Compressed sRGB formats
-    {vk::Format::eUndefined, ComponentType::Invalid, false}, // BGRA8_SRGB
-    {vk::Format::eUndefined, ComponentType::Invalid, false}, // DXT1_SRGB
-    {vk::Format::eUndefined, ComponentType::Invalid, false}, // DXT23_SRGB
-    {vk::Format::eUndefined, ComponentType::Invalid, false}, // DXT45_SRGB
-    {vk::Format::eUndefined, ComponentType::Invalid, false}, // BC7U_SRGB
-    {vk::Format::eUndefined, ComponentType::Invalid, false}, // ASTC_2D_4X4_SRGB
-    {vk::Format::eUndefined, ComponentType::Invalid, false}, // ASTC_2D_8X8_SRGB
-    {vk::Format::eUndefined, ComponentType::Invalid, false}, // ASTC_2D_8X5_SRGB
-    {vk::Format::eUndefined, ComponentType::Invalid, false}, // ASTC_2D_5X4_SRGB
-    {vk::Format::eUndefined, ComponentType::Invalid, false}, // ASTC_2D_5X5
-    {vk::Format::eUndefined, ComponentType::Invalid, false}, // ASTC_2D_5X5_SRGB
-    {vk::Format::eUndefined, ComponentType::Invalid, false}, // ASTC_2D_10X8
-    {vk::Format::eUndefined, ComponentType::Invalid, false}, // ASTC_2D_10X8_SRGB
+    {vk::Format::eUndefined, false}, // BGRA8_SRGB
+    {vk::Format::eUndefined, false}, // DXT1_SRGB
+    {vk::Format::eUndefined, false}, // DXT23_SRGB
+    {vk::Format::eUndefined, false}, // DXT45_SRGB
+    {vk::Format::eUndefined, false}, // BC7U_SRGB
+    {vk::Format::eUndefined, false}, // ASTC_2D_4X4_SRGB
+    {vk::Format::eUndefined, false}, // ASTC_2D_8X8_SRGB
+    {vk::Format::eUndefined, false}, // ASTC_2D_8X5_SRGB
+    {vk::Format::eUndefined, false}, // ASTC_2D_5X4_SRGB
+    {vk::Format::eUndefined, false}, // ASTC_2D_5X5
+    {vk::Format::eUndefined, false}, // ASTC_2D_5X5_SRGB
+    {vk::Format::eUndefined, false}, // ASTC_2D_10X8
+    {vk::Format::eUndefined, false}, // ASTC_2D_10X8_SRGB
 
     // Depth formats
-    {vk::Format::eD32Sfloat, ComponentType::Float, true}, // Z32F
-    {vk::Format::eD16Unorm, ComponentType::UNorm, true},  // Z16
+    {vk::Format::eD32Sfloat, true}, // Z32F
+    {vk::Format::eD16Unorm, true},  // Z16
 
     // DepthStencil formats
-    {vk::Format::eD24UnormS8Uint, ComponentType::UNorm, true}, // Z24S8
-    {vk::Format::eD24UnormS8Uint, ComponentType::UNorm, true}, // S8Z24 (emulated)
-    {vk::Format::eUndefined, ComponentType::Invalid, false},   // Z32FS8
+    {vk::Format::eD24UnormS8Uint, true}, // Z24S8
+    {vk::Format::eD24UnormS8Uint, true}, // S8Z24 (emulated)
+    {vk::Format::eUndefined, false},     // Z32FS8
 }};
 
 static constexpr bool IsZetaFormat(PixelFormat pixel_format) {
@@ -180,14 +181,13 @@ static constexpr bool IsZetaFormat(PixelFormat pixel_format) {
 }
 
 std::pair<vk::Format, bool> SurfaceFormat(const VKDevice& device, FormatType format_type,
-                                          PixelFormat pixel_format, ComponentType component_type) {
+                                          PixelFormat pixel_format) {
     ASSERT(static_cast<std::size_t>(pixel_format) < tex_format_tuples.size());
 
     const auto tuple = tex_format_tuples[static_cast<u32>(pixel_format)];
     UNIMPLEMENTED_IF_MSG(tuple.format == vk::Format::eUndefined,
-                         "Unimplemented texture format with pixel format={} and component type={}",
-                         static_cast<u32>(pixel_format), static_cast<u32>(component_type));
-    ASSERT_MSG(component_type == tuple.component_type, "Component type mismatch");
+                         "Unimplemented texture format with pixel format={}",
+                         static_cast<u32>(pixel_format));
 
     auto usage = vk::FormatFeatureFlagBits::eSampledImage |
                  vk::FormatFeatureFlagBits::eTransferDst | vk::FormatFeatureFlagBits::eTransferSrc;
@@ -198,17 +198,17 @@ std::pair<vk::Format, bool> SurfaceFormat(const VKDevice& device, FormatType for
     return {device.GetSupportedFormat(tuple.format, usage, format_type), tuple.attachable};
 }
 
-vk::ShaderStageFlagBits ShaderStage(Maxwell::ShaderStage stage) {
+vk::ShaderStageFlagBits ShaderStage(Tegra::Engines::ShaderType stage) {
     switch (stage) {
-    case Maxwell::ShaderStage::Vertex:
+    case Tegra::Engines::ShaderType::Vertex:
         return vk::ShaderStageFlagBits::eVertex;
-    case Maxwell::ShaderStage::TesselationControl:
+    case Tegra::Engines::ShaderType::TesselationControl:
         return vk::ShaderStageFlagBits::eTessellationControl;
-    case Maxwell::ShaderStage::TesselationEval:
+    case Tegra::Engines::ShaderType::TesselationEval:
         return vk::ShaderStageFlagBits::eTessellationEvaluation;
-    case Maxwell::ShaderStage::Geometry:
+    case Tegra::Engines::ShaderType::Geometry:
         return vk::ShaderStageFlagBits::eGeometry;
-    case Maxwell::ShaderStage::Fragment:
+    case Tegra::Engines::ShaderType::Fragment:
         return vk::ShaderStageFlagBits::eFragment;
     }
     UNIMPLEMENTED_MSG("Unimplemented shader stage={}", static_cast<u32>(stage));

src/video_core/renderer_vulkan/maxwell_to_vk.h

@@ -16,7 +16,6 @@ namespace Vulkan::MaxwellToVK {
 
 using Maxwell = Tegra::Engines::Maxwell3D::Regs;
 using PixelFormat = VideoCore::Surface::PixelFormat;
-using ComponentType = VideoCore::Surface::ComponentType;
 
 namespace Sampler {
 
@@ -31,9 +30,9 @@ vk::CompareOp DepthCompareFunction(Tegra::Texture::DepthCompareFunc depth_compar
 } // namespace Sampler
 
 std::pair<vk::Format, bool> SurfaceFormat(const VKDevice& device, FormatType format_type,
-                                          PixelFormat pixel_format, ComponentType component_type);
+                                          PixelFormat pixel_format);
 
-vk::ShaderStageFlagBits ShaderStage(Maxwell::ShaderStage stage);
+vk::ShaderStageFlagBits ShaderStage(Tegra::Engines::ShaderType stage);
 
 vk::PrimitiveTopology PrimitiveTopology(Maxwell::PrimitiveTopology topology);
 

src/video_core/renderer_vulkan/vk_shader_decompiler.cpp

@@ -17,6 +17,7 @@
 #include "video_core/engines/maxwell_3d.h"
 #include "video_core/engines/shader_bytecode.h"
 #include "video_core/engines/shader_header.h"
+#include "video_core/engines/shader_type.h"
 #include "video_core/renderer_vulkan/vk_device.h"
 #include "video_core/renderer_vulkan/vk_shader_decompiler.h"
 #include "video_core/shader/node.h"
@@ -25,13 +26,13 @@
 namespace Vulkan::VKShader {
 
 using Sirit::Id;
+using Tegra::Engines::ShaderType;
 using Tegra::Shader::Attribute;
 using Tegra::Shader::AttributeUse;
 using Tegra::Shader::Register;
 using namespace VideoCommon::Shader;
 
 using Maxwell = Tegra::Engines::Maxwell3D::Regs;
-using ShaderStage = Tegra::Engines::Maxwell3D::Regs::ShaderStage;
 using Operation = const OperationNode&;
 
 // TODO(Rodrigo): Use rasterizer's value
@@ -93,7 +94,7 @@ class ExprDecompiler;
 
 class SPIRVDecompiler : public Sirit::Module {
 public:
-    explicit SPIRVDecompiler(const VKDevice& device, const ShaderIR& ir, ShaderStage stage)
+    explicit SPIRVDecompiler(const VKDevice& device, const ShaderIR& ir, ShaderType stage)
         : Module(0x00010300), device{device}, ir{ir}, stage{stage}, header{ir.GetHeader()} {
         AddCapability(spv::Capability::Shader);
         AddExtension("SPV_KHR_storage_buffer_storage_class");
@@ -256,21 +257,21 @@ private:
     }
 
     void DeclareVertex() {
-        if (stage != ShaderStage::Vertex)
+        if (stage != ShaderType::Vertex)
             return;
 
         DeclareVertexRedeclarations();
     }
 
     void DeclareGeometry() {
-        if (stage != ShaderStage::Geometry)
+        if (stage != ShaderType::Geometry)
             return;
 
         UNIMPLEMENTED();
     }
 
     void DeclareFragment() {
-        if (stage != ShaderStage::Fragment)
+        if (stage != ShaderType::Fragment)
             return;
 
         for (u32 rt = 0; rt < static_cast<u32>(frag_colors.size()); ++rt) {
@@ -354,7 +355,7 @@ private:
                 continue;
             }
 
-            UNIMPLEMENTED_IF(stage == ShaderStage::Geometry);
+            UNIMPLEMENTED_IF(stage == ShaderType::Geometry);
 
             const u32 location = GetGenericAttributeLocation(index);
             const Id id = OpVariable(t_in_float4, spv::StorageClass::Input);
@@ -364,7 +365,7 @@ private:
 
             Decorate(id, spv::Decoration::Location, location);
 
-            if (stage != ShaderStage::Fragment) {
+            if (stage != ShaderType::Fragment) {
                 continue;
             }
             switch (header.ps.GetAttributeUse(location)) {
@@ -548,7 +549,7 @@ private:
 
             switch (attribute) {
             case Attribute::Index::Position:
-                if (stage != ShaderStage::Fragment) {
+                if (stage != ShaderType::Fragment) {
                     UNIMPLEMENTED();
                     break;
                 } else {
@@ -561,7 +562,7 @@ private:
                 // TODO(Subv): Find out what the values are for the first two elements when inside a
                 // vertex shader, and what's the value of the fourth element when inside a Tess Eval
                 // shader.
-                ASSERT(stage == ShaderStage::Vertex);
+                ASSERT(stage == ShaderType::Vertex);
                 switch (element) {
                 case 2:
                     return BitcastFrom<Type::Uint>(Emit(OpLoad(t_uint, instance_index)));
@@ -572,7 +573,7 @@ private:
                 return Constant(t_float, 0);
             case Attribute::Index::FrontFacing:
                 // TODO(Subv): Find out what the values are for the other elements.
-                ASSERT(stage == ShaderStage::Fragment);
+                ASSERT(stage == ShaderType::Fragment);
                 if (element == 3) {
                     const Id is_front_facing = Emit(OpLoad(t_bool, front_facing));
                     const Id true_value =
@@ -783,6 +784,11 @@ private:
         return {};
     }
 
+    Id FSwizzleAdd(Operation operation) {
+        UNIMPLEMENTED();
+        return {};
+    }
+
     Id HNegate(Operation operation) {
         UNIMPLEMENTED();
         return {};
@@ -1070,7 +1076,7 @@ private:
 
     Id PreExit() {
         switch (stage) {
-        case ShaderStage::Vertex: {
+        case ShaderType::Vertex: {
             // TODO(Rodrigo): We should use VK_EXT_depth_range_unrestricted instead, but it doesn't
             // seem to be working on Nvidia's drivers and Intel (mesa and blob) doesn't support it.
             const Id z_pointer = AccessElement(t_out_float, per_vertex, position_index, 2u);
@@ -1080,7 +1086,7 @@ private:
             Emit(OpStore(z_pointer, depth));
             break;
         }
-        case ShaderStage::Fragment: {
+        case ShaderType::Fragment: {
             const auto SafeGetRegister = [&](u32 reg) {
                 // TODO(Rodrigo): Replace with contains once C++20 releases
                 if (const auto it = registers.find(reg); it != registers.end()) {
@@ -1195,46 +1201,16 @@ private:
         return {};
     }
 
+    Id ThreadId(Operation) {
+        UNIMPLEMENTED();
+        return {};
+    }
+
     Id ShuffleIndexed(Operation) {
         UNIMPLEMENTED();
         return {};
     }
 
-    Id ShuffleUp(Operation) {
-        UNIMPLEMENTED();
-        return {};
-    }
-
-    Id ShuffleDown(Operation) {
-        UNIMPLEMENTED();
-        return {};
-    }
-
-    Id ShuffleButterfly(Operation) {
-        UNIMPLEMENTED();
-        return {};
-    }
-
-    Id InRangeShuffleIndexed(Operation) {
-        UNIMPLEMENTED();
-        return {};
-    }
-
-    Id InRangeShuffleUp(Operation) {
-        UNIMPLEMENTED();
-        return {};
-    }
-
-    Id InRangeShuffleDown(Operation) {
-        UNIMPLEMENTED();
-        return {};
-    }
-
-    Id InRangeShuffleButterfly(Operation) {
-        UNIMPLEMENTED();
-        return {};
-    }
-
     Id DeclareBuiltIn(spv::BuiltIn builtin, spv::StorageClass storage, Id type,
                       const std::string& name) {
         const Id id = OpVariable(type, storage);
@@ -1393,6 +1369,7 @@ private:
         &SPIRVDecompiler::Unary<&Module::OpTrunc, Type::Float>,
         &SPIRVDecompiler::Unary<&Module::OpConvertSToF, Type::Float, Type::Int>,
         &SPIRVDecompiler::Unary<&Module::OpConvertUToF, Type::Float, Type::Uint>,
+        &SPIRVDecompiler::FSwizzleAdd,
 
         &SPIRVDecompiler::Binary<&Module::OpIAdd, Type::Int>,
         &SPIRVDecompiler::Binary<&Module::OpIMul, Type::Int>,
@@ -1528,21 +1505,14 @@ private:
         &SPIRVDecompiler::VoteAny,
         &SPIRVDecompiler::VoteEqual,
 
+        &SPIRVDecompiler::ThreadId,
         &SPIRVDecompiler::ShuffleIndexed,
-        &SPIRVDecompiler::ShuffleUp,
-        &SPIRVDecompiler::ShuffleDown,
-        &SPIRVDecompiler::ShuffleButterfly,
-
-        &SPIRVDecompiler::InRangeShuffleIndexed,
-        &SPIRVDecompiler::InRangeShuffleUp,
-        &SPIRVDecompiler::InRangeShuffleDown,
-        &SPIRVDecompiler::InRangeShuffleButterfly,
     };
     static_assert(operation_decompilers.size() == static_cast<std::size_t>(OperationCode::Amount));
 
     const VKDevice& device;
     const ShaderIR& ir;
-    const ShaderStage stage;
+    const ShaderType stage;
     const Tegra::Shader::Header header;
     u64 conditional_nest_count{};
     u64 inside_branch{};
@@ -1874,7 +1844,7 @@ void SPIRVDecompiler::DecompileAST() {
 }
 
 DecompilerResult Decompile(const VKDevice& device, const VideoCommon::Shader::ShaderIR& ir,
-                           Maxwell::ShaderStage stage) {
+                           ShaderType stage) {
     auto decompiler = std::make_unique<SPIRVDecompiler>(device, ir, stage);
     decompiler->Decompile();
     return {std::move(decompiler), decompiler->GetShaderEntries()};

src/video_core/renderer_vulkan/vk_shader_decompiler.h

@@ -79,6 +79,6 @@ struct ShaderEntries {
 using DecompilerResult = std::pair<std::unique_ptr<Sirit::Module>, ShaderEntries>;
 
 DecompilerResult Decompile(const VKDevice& device, const VideoCommon::Shader::ShaderIR& ir,
-                           Maxwell::ShaderStage stage);
+                           Tegra::Engines::ShaderType stage);
 
 } // namespace Vulkan::VKShader

src/video_core/shader/const_buffer_locker.cpp

@@ -9,6 +9,7 @@
 #include "common/assert.h"
 #include "common/common_types.h"
 #include "video_core/engines/maxwell_3d.h"
+#include "video_core/engines/shader_type.h"
 #include "video_core/shader/const_buffer_locker.h"
 
 namespace VideoCommon::Shader {

src/video_core/shader/const_buffer_locker.h

@@ -8,6 +8,7 @@
 #include "common/common_types.h"
 #include "common/hash.h"
 #include "video_core/engines/const_buffer_engine_interface.h"
+#include "video_core/engines/shader_type.h"
 
 namespace VideoCommon::Shader {
 
@@ -20,7 +21,7 @@ using BindlessSamplerMap =
  * The ConstBufferLocker is a class use to interface the 3D and compute engines with the shader
  * compiler. with it, the shader can obtain required data from GPU state and store it for disk
  * shader compilation.
- **/
+ */
 class ConstBufferLocker {
 public:
     explicit ConstBufferLocker(Tegra::Engines::ShaderType shader_stage);

src/video_core/shader/decode.cpp

@@ -154,10 +154,10 @@ void ShaderIR::Decode() {
         LOG_CRITICAL(HW_GPU, "Unknown decompilation mode!");
         [[fallthrough]];
     case CompileDepth::BruteForce: {
+        const auto shader_end = static_cast<u32>(program_code.size());
         coverage_begin = main_offset;
-        const std::size_t shader_end = program_code.size();
         coverage_end = shader_end;
-        for (u32 label = main_offset; label < shader_end; label++) {
+        for (u32 label = main_offset; label < shader_end; ++label) {
             basic_blocks.insert({label, DecodeRange(label, label + 1)});
         }
         break;

src/video_core/shader/decode/other.cpp

@@ -256,7 +256,7 @@ u32 ShaderIR::DecodeOther(NodeBlock& bb, u32 pc) {
         break;
     }
     case OpCode::Id::DEPBAR: {
-        LOG_WARNING(HW_GPU, "DEPBAR instruction is stubbed");
+        LOG_DEBUG(HW_GPU, "DEPBAR instruction is stubbed");
         break;
     }
     default:

src/video_core/shader/decode/texture.cpp

@@ -128,8 +128,8 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
         }
         const Node component = Immediate(static_cast<u32>(instr.tld4s.component));
 
-        const auto& sampler =
-            GetSampler(instr.sampler, {{TextureType::Texture2D, false, depth_compare}});
+        const SamplerInfo info{TextureType::Texture2D, false, depth_compare};
+        const auto& sampler = GetSampler(instr.sampler, info);
 
         Node4 values;
         for (u32 element = 0; element < values.size(); ++element) {
@@ -149,7 +149,7 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
         // Sadly, not all texture instructions specify the type of texture their sampler
         // uses. This must be fixed at a later instance.
         const auto& sampler =
-            is_bindless ? GetBindlessSampler(instr.gpr8, {}) : GetSampler(instr.sampler, {});
+            is_bindless ? GetBindlessSampler(instr.gpr8) : GetSampler(instr.sampler);
 
         u32 indexer = 0;
         switch (instr.txq.query_type) {
@@ -185,8 +185,7 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
         auto texture_type = instr.tmml.texture_type.Value();
         const bool is_array = instr.tmml.array != 0;
         const auto& sampler =
-            is_bindless ? GetBindlessSampler(instr.gpr20, {{texture_type, is_array, false}})
-                        : GetSampler(instr.sampler, {{texture_type, is_array, false}});
+            is_bindless ? GetBindlessSampler(instr.gpr20) : GetSampler(instr.sampler);
 
         std::vector<Node> coords;
 
@@ -254,67 +253,50 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
     return pc;
 }
 
+ShaderIR::SamplerInfo ShaderIR::GetSamplerInfo(std::optional<SamplerInfo> sampler_info, u32 offset,
+                                               std::optional<u32> buffer) {
+    if (sampler_info) {
+        return *sampler_info;
+    }
+    const auto sampler =
+        buffer ? locker.ObtainBindlessSampler(*buffer, offset) : locker.ObtainBoundSampler(offset);
+    if (!sampler) {
+        LOG_WARNING(HW_GPU, "Unknown sampler info");
+        return SamplerInfo{TextureType::Texture2D, false, false, false};
+    }
+    return SamplerInfo{sampler->texture_type, sampler->is_array != 0, sampler->is_shadow != 0,
+                       sampler->is_buffer != 0};
+}
+
 const Sampler& ShaderIR::GetSampler(const Tegra::Shader::Sampler& sampler,
                                     std::optional<SamplerInfo> sampler_info) {
     const auto offset = static_cast<u32>(sampler.index.Value());
-
-    TextureType type;
-    bool is_array;
-    bool is_shadow;
-    if (sampler_info) {
-        type = sampler_info->type;
-        is_array = sampler_info->is_array;
-        is_shadow = sampler_info->is_shadow;
-    } else if (const auto sampler = locker.ObtainBoundSampler(offset)) {
-        type = sampler->texture_type.Value();
-        is_array = sampler->is_array.Value() != 0;
-        is_shadow = sampler->is_shadow.Value() != 0;
-    } else {
-        LOG_WARNING(HW_GPU, "Unknown sampler info");
-        type = TextureType::Texture2D;
-        is_array = false;
-        is_shadow = false;
-    }
+    const auto info = GetSamplerInfo(sampler_info, offset);
 
     // If this sampler has already been used, return the existing mapping.
     const auto it =
         std::find_if(used_samplers.begin(), used_samplers.end(),
                      [offset](const Sampler& entry) { return entry.GetOffset() == offset; });
     if (it != used_samplers.end()) {
-        ASSERT(!it->IsBindless() && it->GetType() == type && it->IsArray() == is_array &&
-               it->IsShadow() == is_shadow);
+        ASSERT(!it->IsBindless() && it->GetType() == info.type && it->IsArray() == info.is_array &&
+               it->IsShadow() == info.is_shadow && it->IsBuffer() == info.is_buffer);
         return *it;
     }
 
     // Otherwise create a new mapping for this sampler
     const auto next_index = static_cast<u32>(used_samplers.size());
-    return used_samplers.emplace_back(Sampler(next_index, offset, type, is_array, is_shadow));
+    return used_samplers.emplace_back(next_index, offset, info.type, info.is_array, info.is_shadow,
+                                      info.is_buffer);
 }
 
-const Sampler& ShaderIR::GetBindlessSampler(const Tegra::Shader::Register& reg,
+const Sampler& ShaderIR::GetBindlessSampler(Tegra::Shader::Register reg,
                                             std::optional<SamplerInfo> sampler_info) {
     const Node sampler_register = GetRegister(reg);
     const auto [base_sampler, buffer, offset] =
         TrackCbuf(sampler_register, global_code, static_cast<s64>(global_code.size()));
     ASSERT(base_sampler != nullptr);
 
-    TextureType type;
-    bool is_array;
-    bool is_shadow;
-    if (sampler_info) {
-        type = sampler_info->type;
-        is_array = sampler_info->is_array;
-        is_shadow = sampler_info->is_shadow;
-    } else if (const auto sampler = locker.ObtainBindlessSampler(buffer, offset)) {
-        type = sampler->texture_type.Value();
-        is_array = sampler->is_array.Value() != 0;
-        is_shadow = sampler->is_shadow.Value() != 0;
-    } else {
-        LOG_WARNING(HW_GPU, "Unknown sampler info");
-        type = TextureType::Texture2D;
-        is_array = false;
-        is_shadow = false;
-    }
+    const auto info = GetSamplerInfo(sampler_info, offset, buffer);
 
     // If this sampler has already been used, return the existing mapping.
     const auto it =
@@ -323,15 +305,15 @@ const Sampler& ShaderIR::GetBindlessSampler(const Tegra::Shader::Register& reg,
                          return entry.GetBuffer() == buffer && entry.GetOffset() == offset;
                      });
     if (it != used_samplers.end()) {
-        ASSERT(it->IsBindless() && it->GetType() == type && it->IsArray() == is_array &&
-               it->IsShadow() == is_shadow);
+        ASSERT(it->IsBindless() && it->GetType() == info.type && it->IsArray() == info.is_array &&
+               it->IsShadow() == info.is_shadow);
         return *it;
     }
 
     // Otherwise create a new mapping for this sampler
     const auto next_index = static_cast<u32>(used_samplers.size());
-    return used_samplers.emplace_back(
-        Sampler(next_index, offset, buffer, type, is_array, is_shadow));
+    return used_samplers.emplace_back(next_index, offset, buffer, info.type, info.is_array,
+                                      info.is_shadow, info.is_buffer);
 }
 
 void ShaderIR::WriteTexInstructionFloat(NodeBlock& bb, Instruction instr, const Node4& components) {
@@ -416,17 +398,16 @@ Node4 ShaderIR::GetTextureCode(Instruction instr, TextureType texture_type,
                              (texture_type == TextureType::TextureCube && is_array && is_shadow),
                          "This method is not supported.");
 
+    const SamplerInfo info{texture_type, is_array, is_shadow, false};
     const auto& sampler =
-        is_bindless ? GetBindlessSampler(*bindless_reg, {{texture_type, is_array, is_shadow}})
-                    : GetSampler(instr.sampler, {{texture_type, is_array, is_shadow}});
+        is_bindless ? GetBindlessSampler(*bindless_reg, info) : GetSampler(instr.sampler, info);
 
     const bool lod_needed = process_mode == TextureProcessMode::LZ ||
                             process_mode == TextureProcessMode::LL ||
                             process_mode == TextureProcessMode::LLA;
 
-    // LOD selection (either via bias or explicit textureLod) not
-    // supported in GL for sampler2DArrayShadow and
-    // samplerCubeArrayShadow.
+    // LOD selection (either via bias or explicit textureLod) not supported in GL for
+    // sampler2DArrayShadow and samplerCubeArrayShadow.
     const bool gl_lod_supported =
         !((texture_type == Tegra::Shader::TextureType::Texture2D && is_array && is_shadow) ||
           (texture_type == Tegra::Shader::TextureType::TextureCube && is_array && is_shadow));
@@ -436,8 +417,8 @@ Node4 ShaderIR::GetTextureCode(Instruction instr, TextureType texture_type,
 
     UNIMPLEMENTED_IF(process_mode != TextureProcessMode::None && !gl_lod_supported);
 
-    Node bias = {};
-    Node lod = {};
+    Node bias;
+    Node lod;
     if (process_mode != TextureProcessMode::None && gl_lod_supported) {
         switch (process_mode) {
         case TextureProcessMode::LZ:
@@ -573,10 +554,9 @@ Node4 ShaderIR::GetTld4Code(Instruction instr, TextureType texture_type, bool de
 
     u64 parameter_register = instr.gpr20.Value();
 
-    const auto& sampler =
-        is_bindless
-            ? GetBindlessSampler(parameter_register++, {{texture_type, is_array, depth_compare}})
-            : GetSampler(instr.sampler, {{texture_type, is_array, depth_compare}});
+    const SamplerInfo info{texture_type, is_array, depth_compare, false};
+    const auto& sampler = is_bindless ? GetBindlessSampler(parameter_register++, info)
+                                      : GetSampler(instr.sampler, info);
 
     std::vector<Node> aoffi;
     if (is_aoffi) {
@@ -623,7 +603,7 @@ Node4 ShaderIR::GetTldCode(Tegra::Shader::Instruction instr) {
     // const Node aoffi_register{is_aoffi ? GetRegister(gpr20_cursor++) : nullptr};
     // const Node multisample{is_multisample ? GetRegister(gpr20_cursor++) : nullptr};
 
-    const auto& sampler = GetSampler(instr.sampler, {{texture_type, is_array, false}});
+    const auto& sampler = GetSampler(instr.sampler);
 
     Node4 values;
     for (u32 element = 0; element < values.size(); ++element) {
@@ -636,6 +616,8 @@ Node4 ShaderIR::GetTldCode(Tegra::Shader::Instruction instr) {
 }
 
 Node4 ShaderIR::GetTldsCode(Instruction instr, TextureType texture_type, bool is_array) {
+    const auto& sampler = GetSampler(instr.sampler);
+
     const std::size_t type_coord_count = GetCoordCount(texture_type);
     const bool lod_enabled = instr.tlds.GetTextureProcessMode() == TextureProcessMode::LL;
 
@@ -659,7 +641,14 @@ Node4 ShaderIR::GetTldsCode(Instruction instr, TextureType texture_type, bool is
     // When lod is used always is in gpr20
     const Node lod = lod_enabled ? GetRegister(instr.gpr20) : Immediate(0);
 
-    const auto& sampler = GetSampler(instr.sampler, {{texture_type, is_array, false}});
+    // Fill empty entries from the guest sampler.
+    const std::size_t entry_coord_count = GetCoordCount(sampler.GetType());
+    if (type_coord_count != entry_coord_count) {
+        LOG_WARNING(HW_GPU, "Bound and built texture types mismatch");
+    }
+    for (std::size_t i = type_coord_count; i < entry_coord_count; ++i) {
+        coords.push_back(GetRegister(Register::ZeroIndex));
+    }
 
     Node4 values;
     for (u32 element = 0; element < values.size(); ++element) {

src/video_core/shader/decode/warp.cpp

@@ -17,6 +17,7 @@ using Tegra::Shader::ShuffleOperation;
 using Tegra::Shader::VoteOperation;
 
 namespace {
+
 OperationCode GetOperationCode(VoteOperation vote_op) {
     switch (vote_op) {
     case VoteOperation::All:
@@ -30,6 +31,7 @@ OperationCode GetOperationCode(VoteOperation vote_op) {
         return OperationCode::VoteAll;
     }
 }
+
 } // Anonymous namespace
 
 u32 ShaderIR::DecodeWarp(NodeBlock& bb, u32 pc) {
@@ -46,50 +48,59 @@ u32 ShaderIR::DecodeWarp(NodeBlock& bb, u32 pc) {
         break;
     }
     case OpCode::Id::SHFL: {
-        Node width = [this, instr] {
-            Node mask = instr.shfl.is_mask_imm ? Immediate(static_cast<u32>(instr.shfl.mask_imm))
-                                               : GetRegister(instr.gpr39);
-
-            // Convert the obscure SHFL mask back into GL_NV_shader_thread_shuffle's width. This has
-            // been done reversing Nvidia's math. It won't work on all cases due to SHFL having
-            // different parameters that don't properly map to GLSL's interface, but it should work
-            // for cases emitted by Nvidia's compiler.
-            if (instr.shfl.operation == ShuffleOperation::Up) {
-                return Operation(
-                    OperationCode::ILogicalShiftRight,
-                    Operation(OperationCode::IAdd, std::move(mask), Immediate(-0x2000)),
-                    Immediate(8));
-            } else {
-                return Operation(OperationCode::ILogicalShiftRight,
-                                 Operation(OperationCode::IAdd, Immediate(0x201F),
-                                           Operation(OperationCode::INegate, std::move(mask))),
-                                 Immediate(8));
-            }
-        }();
-
-        const auto [operation, in_range] = [instr]() -> std::pair<OperationCode, OperationCode> {
-            switch (instr.shfl.operation) {
-            case ShuffleOperation::Idx:
-                return {OperationCode::ShuffleIndexed, OperationCode::InRangeShuffleIndexed};
-            case ShuffleOperation::Up:
-                return {OperationCode::ShuffleUp, OperationCode::InRangeShuffleUp};
-            case ShuffleOperation::Down:
-                return {OperationCode::ShuffleDown, OperationCode::InRangeShuffleDown};
-            case ShuffleOperation::Bfly:
-                return {OperationCode::ShuffleButterfly, OperationCode::InRangeShuffleButterfly};
-            }
-            UNREACHABLE_MSG("Invalid SHFL operation: {}",
-                            static_cast<u64>(instr.shfl.operation.Value()));
-            return {};
-        }();
-
-        // Setting the predicate before the register is intentional to avoid overwriting.
+        Node mask = instr.shfl.is_mask_imm ? Immediate(static_cast<u32>(instr.shfl.mask_imm))
+                                           : GetRegister(instr.gpr39);
         Node index = instr.shfl.is_index_imm ? Immediate(static_cast<u32>(instr.shfl.index_imm))
                                              : GetRegister(instr.gpr20);
-        SetPredicate(bb, instr.shfl.pred48, Operation(in_range, index, width));
+
+        Node thread_id = Operation(OperationCode::ThreadId);
+        Node clamp = Operation(OperationCode::IBitwiseAnd, mask, Immediate(0x1FU));
+        Node seg_mask = BitfieldExtract(mask, 8, 16);
+
+        Node neg_seg_mask = Operation(OperationCode::IBitwiseNot, seg_mask);
+        Node min_thread_id = Operation(OperationCode::IBitwiseAnd, thread_id, seg_mask);
+        Node max_thread_id = Operation(OperationCode::IBitwiseOr, min_thread_id,
+                                       Operation(OperationCode::IBitwiseAnd, clamp, neg_seg_mask));
+
+        Node src_thread_id = [instr, index, neg_seg_mask, min_thread_id, thread_id] {
+            switch (instr.shfl.operation) {
+            case ShuffleOperation::Idx:
+                return Operation(OperationCode::IBitwiseOr,
+                                 Operation(OperationCode::IBitwiseAnd, index, neg_seg_mask),
+                                 min_thread_id);
+            case ShuffleOperation::Down:
+                return Operation(OperationCode::IAdd, thread_id, index);
+            case ShuffleOperation::Up:
+                return Operation(OperationCode::IAdd, thread_id,
+                                 Operation(OperationCode::INegate, index));
+            case ShuffleOperation::Bfly:
+                return Operation(OperationCode::IBitwiseXor, thread_id, index);
+            }
+            UNREACHABLE();
+            return Immediate(0U);
+        }();
+
+        Node in_bounds = [instr, src_thread_id, min_thread_id, max_thread_id] {
+            if (instr.shfl.operation == ShuffleOperation::Up) {
+                return Operation(OperationCode::LogicalIGreaterEqual, src_thread_id, min_thread_id);
+            } else {
+                return Operation(OperationCode::LogicalILessEqual, src_thread_id, max_thread_id);
+            }
+        }();
+
+        SetPredicate(bb, instr.shfl.pred48, in_bounds);
         SetRegister(
             bb, instr.gpr0,
-            Operation(operation, GetRegister(instr.gpr8), std::move(index), std::move(width)));
+            Operation(OperationCode::ShuffleIndexed, GetRegister(instr.gpr8), src_thread_id));
+        break;
+    }
+    case OpCode::Id::FSWZADD: {
+        UNIMPLEMENTED_IF(instr.fswzadd.ndv);
+
+        Node op_a = GetRegister(instr.gpr8);
+        Node op_b = GetRegister(instr.gpr20);
+        Node mask = Immediate(static_cast<u32>(instr.fswzadd.swizzle));
+        SetRegister(bb, instr.gpr0, Operation(OperationCode::FSwizzleAdd, op_a, op_b, mask));
         break;
     }
     default: