nfp: Correct erroneous sizeof expression within GetTagInfo()

The previous expression would copy sizeof(size_t) amount of bytes (8 on
a 64-bit platform) rather than the full 10 bytes comprising the uuid
member.

Given the source and destination types are the same, we can just use an
assignment here instead.

src/audio_core/cubeb_sink.cpp

@@ -121,7 +121,8 @@ CubebSink::CubebSink(std::string target_device_name) {
             const auto collection_end{collection.device + collection.count};
             const auto device{
                 std::find_if(collection.device, collection_end, [&](const cubeb_device_info& info) {
-                    return target_device_name == info.friendly_name;
+                    return info.friendly_name != nullptr &&
+                           target_device_name == info.friendly_name;
                 })};
             if (device != collection_end) {
                 output_device = device->devid;

src/audio_core/stream.cpp

@@ -11,7 +11,6 @@
 #include "audio_core/stream.h"
 #include "common/assert.h"
 #include "common/logging/log.h"
-#include "common/microprofile.h"
 #include "core/core_timing.h"
 #include "core/core_timing_util.h"
 #include "core/settings.h"
@@ -104,10 +103,7 @@ void Stream::PlayNextBuffer() {
     CoreTiming::ScheduleEventThreadsafe(GetBufferReleaseCycles(*active_buffer), release_event, {});
 }
 
-MICROPROFILE_DEFINE(AudioOutput, "Audio", "ReleaseActiveBuffer", MP_RGB(100, 100, 255));
-
 void Stream::ReleaseActiveBuffer() {
-    MICROPROFILE_SCOPE(AudioOutput);
     ASSERT(active_buffer);
     released_buffers.push(std::move(active_buffer));
     release_callback();

src/audio_core/time_stretch.cpp

@@ -32,10 +32,10 @@ std::size_t TimeStretcher::Process(const s16* in, std::size_t num_in, s16* out,
     // We were given actual_samples number of samples, and num_samples were requested from us.
     double current_ratio = static_cast<double>(num_in) / static_cast<double>(num_out);
 
-    const double max_latency = 1.0; // seconds
+    const double max_latency = 0.25; // seconds
     const double max_backlog = m_sample_rate * max_latency;
     const double backlog_fullness = m_sound_touch.numSamples() / max_backlog;
-    if (backlog_fullness > 5.0) {
+    if (backlog_fullness > 4.0) {
         // Too many samples in backlog: Don't push anymore on
         num_in = 0;
     }
@@ -49,7 +49,7 @@ std::size_t TimeStretcher::Process(const s16* in, std::size_t num_in, s16* out,
 
     // This low-pass filter smoothes out variance in the calculated stretch ratio.
     // The time-scale determines how responsive this filter is.
-    constexpr double lpf_time_scale = 2.0; // seconds
+    constexpr double lpf_time_scale = 0.712; // seconds
     const double lpf_gain = 1.0 - std::exp(-time_delta / lpf_time_scale);
     m_stretch_ratio += lpf_gain * (current_ratio - m_stretch_ratio);
 

src/common/logging/backend.cpp

@@ -12,7 +12,8 @@
 #include <thread>
 #include <vector>
 #ifdef _WIN32
-#include <share.h> // For _SH_DENYWR
+#include <share.h>   // For _SH_DENYWR
+#include <windows.h> // For OutputDebugStringA
 #else
 #define _SH_DENYWR 0
 #endif
@@ -139,12 +140,18 @@ void FileBackend::Write(const Entry& entry) {
     if (!file.IsOpen() || bytes_written > MAX_BYTES_WRITTEN) {
         return;
     }
-    bytes_written += file.WriteString(FormatLogMessage(entry) + '\n');
+    bytes_written += file.WriteString(FormatLogMessage(entry).append(1, '\n'));
     if (entry.log_level >= Level::Error) {
         file.Flush();
     }
 }
 
+void DebuggerBackend::Write(const Entry& entry) {
+#ifdef _WIN32
+    ::OutputDebugStringA(FormatLogMessage(entry).append(1, '\n').c_str());
+#endif
+}
+
 /// Macro listing all log classes. Code should define CLS and SUB as desired before invoking this.
 #define ALL_LOG_CLASSES()                                                                          \
     CLS(Log)                                                                                       \

src/common/logging/backend.h

@@ -103,6 +103,20 @@ private:
     std::size_t bytes_written;
 };
 
+/**
+ * Backend that writes to Visual Studio's output window
+ */
+class DebuggerBackend : public Backend {
+public:
+    static const char* Name() {
+        return "debugger";
+    }
+    const char* GetName() const override {
+        return Name();
+    }
+    void Write(const Entry& entry) override;
+};
+
 void AddBackend(std::unique_ptr<Backend> backend);
 
 void RemoveBackend(std::string_view backend_name);

src/common/telemetry.h

@@ -153,6 +153,7 @@ struct VisitorInterface : NonCopyable {
 
     /// Completion method, called once all fields have been visited
     virtual void Complete() = 0;
+    virtual bool SubmitTestcase() = 0;
 };
 
 /**
@@ -178,6 +179,9 @@ struct NullVisitor : public VisitorInterface {
     void Visit(const Field<std::chrono::microseconds>& /*field*/) override {}
 
     void Complete() override {}
+    bool SubmitTestcase() override {
+        return false;
+    }
 };
 
 /// Appends build-specific information to the given FieldCollection,

src/core/core.cpp

@@ -185,7 +185,7 @@ struct System::Impl {
             LOG_CRITICAL(Core, "Failed to obtain loader for {}!", filepath);
             return ResultStatus::ErrorGetLoader;
         }
-        std::pair<boost::optional<u32>, Loader::ResultStatus> system_mode =
+        std::pair<std::optional<u32>, Loader::ResultStatus> system_mode =
             app_loader->LoadKernelSystemMode();
 
         if (system_mode.second != Loader::ResultStatus::Success) {
@@ -312,6 +312,10 @@ Cpu& System::CurrentCpuCore() {
     return impl->CurrentCpuCore();
 }
 
+const Cpu& System::CurrentCpuCore() const {
+    return impl->CurrentCpuCore();
+}
+
 System::ResultStatus System::RunLoop(bool tight_loop) {
     return impl->RunLoop(tight_loop);
 }
@@ -342,7 +346,11 @@ PerfStatsResults System::GetAndResetPerfStats() {
     return impl->GetAndResetPerfStats();
 }
 
-Core::TelemetrySession& System::TelemetrySession() const {
+TelemetrySession& System::TelemetrySession() {
+    return *impl->telemetry_session;
+}
+
+const TelemetrySession& System::TelemetrySession() const {
     return *impl->telemetry_session;
 }
 
@@ -350,7 +358,11 @@ ARM_Interface& System::CurrentArmInterface() {
     return CurrentCpuCore().ArmInterface();
 }
 
-std::size_t System::CurrentCoreIndex() {
+const ARM_Interface& System::CurrentArmInterface() const {
+    return CurrentCpuCore().ArmInterface();
+}
+
+std::size_t System::CurrentCoreIndex() const {
     return CurrentCpuCore().CoreIndex();
 }
 
@@ -358,6 +370,10 @@ Kernel::Scheduler& System::CurrentScheduler() {
     return CurrentCpuCore().Scheduler();
 }
 
+const Kernel::Scheduler& System::CurrentScheduler() const {
+    return CurrentCpuCore().Scheduler();
+}
+
 Kernel::Scheduler& System::Scheduler(std::size_t core_index) {
     return CpuCore(core_index).Scheduler();
 }
@@ -378,6 +394,10 @@ ARM_Interface& System::ArmInterface(std::size_t core_index) {
     return CpuCore(core_index).ArmInterface();
 }
 
+const ARM_Interface& System::ArmInterface(std::size_t core_index) const {
+    return CpuCore(core_index).ArmInterface();
+}
+
 Cpu& System::CpuCore(std::size_t core_index) {
     ASSERT(core_index < NUM_CPU_CORES);
     return *impl->cpu_cores[core_index];
@@ -392,6 +412,10 @@ ExclusiveMonitor& System::Monitor() {
     return *impl->cpu_exclusive_monitor;
 }
 
+const ExclusiveMonitor& System::Monitor() const {
+    return *impl->cpu_exclusive_monitor;
+}
+
 Tegra::GPU& System::GPU() {
     return *impl->gpu_core;
 }

src/core/core.h

@@ -129,11 +129,11 @@ public:
      */
     bool IsPoweredOn() const;
 
-    /**
-     * Returns a reference to the telemetry session for this emulation session.
-     * @returns Reference to the telemetry session.
-     */
-    Core::TelemetrySession& TelemetrySession() const;
+    /// Gets a reference to the telemetry session for this emulation session.
+    Core::TelemetrySession& TelemetrySession();
+
+    /// Gets a reference to the telemetry session for this emulation session.
+    const Core::TelemetrySession& TelemetrySession() const;
 
     /// Prepare the core emulation for a reschedule
     void PrepareReschedule();
@@ -144,24 +144,36 @@ public:
     /// Gets an ARM interface to the CPU core that is currently running
     ARM_Interface& CurrentArmInterface();
 
+    /// Gets an ARM interface to the CPU core that is currently running
+    const ARM_Interface& CurrentArmInterface() const;
+
     /// Gets the index of the currently running CPU core
-    std::size_t CurrentCoreIndex();
+    std::size_t CurrentCoreIndex() const;
 
     /// Gets the scheduler for the CPU core that is currently running
     Kernel::Scheduler& CurrentScheduler();
 
-    /// Gets an ARM interface to the CPU core with the specified index
+    /// Gets the scheduler for the CPU core that is currently running
+    const Kernel::Scheduler& CurrentScheduler() const;
+
+    /// Gets a reference to an ARM interface for the CPU core with the specified index
     ARM_Interface& ArmInterface(std::size_t core_index);
 
+    /// Gets a const reference to an ARM interface from the CPU core with the specified index
+    const ARM_Interface& ArmInterface(std::size_t core_index) const;
+
     /// Gets a CPU interface to the CPU core with the specified index
     Cpu& CpuCore(std::size_t core_index);
 
     /// Gets a CPU interface to the CPU core with the specified index
     const Cpu& CpuCore(std::size_t core_index) const;
 
-    /// Gets the exclusive monitor
+    /// Gets a reference to the exclusive monitor
     ExclusiveMonitor& Monitor();
 
+    /// Gets a constant reference to the exclusive monitor
+    const ExclusiveMonitor& Monitor() const;
+
     /// Gets a mutable reference to the GPU interface
     Tegra::GPU& GPU();
 
@@ -230,6 +242,9 @@ private:
     /// Returns the currently running CPU core
     Cpu& CurrentCpuCore();
 
+    /// Returns the currently running CPU core
+    const Cpu& CurrentCpuCore() const;
+
     /**
      * Initialize the emulated system.
      * @param emu_window Reference to the host-system window used for video output and keyboard

src/core/crypto/key_manager.cpp

@@ -141,28 +141,28 @@ Key128 DeriveKeyblobMACKey(const Key128& keyblob_key, const Key128& mac_source)
     return mac_key;
 }
 
-boost::optional<Key128> DeriveSDSeed() {
+std::optional<Key128> DeriveSDSeed() {
     const FileUtil::IOFile save_43(FileUtil::GetUserPath(FileUtil::UserPath::NANDDir) +
                                        "/system/save/8000000000000043",
                                    "rb+");
     if (!save_43.IsOpen())
-        return boost::none;
+        return {};
 
     const FileUtil::IOFile sd_private(
         FileUtil::GetUserPath(FileUtil::UserPath::SDMCDir) + "/Nintendo/Contents/private", "rb+");
     if (!sd_private.IsOpen())
-        return boost::none;
+        return {};
 
     std::array<u8, 0x10> private_seed{};
     if (sd_private.ReadBytes(private_seed.data(), private_seed.size()) != private_seed.size()) {
-        return boost::none;
+        return {};
     }
 
     std::array<u8, 0x10> buffer{};
     std::size_t offset = 0;
     for (; offset + 0x10 < save_43.GetSize(); ++offset) {
         if (!save_43.Seek(offset, SEEK_SET)) {
-            return boost::none;
+            return {};
         }
 
         save_43.ReadBytes(buffer.data(), buffer.size());
@@ -172,12 +172,12 @@ boost::optional<Key128> DeriveSDSeed() {
     }
 
     if (!save_43.Seek(offset + 0x10, SEEK_SET)) {
-        return boost::none;
+        return {};
     }
 
     Key128 seed{};
     if (save_43.ReadBytes(seed.data(), seed.size()) != seed.size()) {
-        return boost::none;
+        return {};
     }
     return seed;
 }
@@ -291,26 +291,26 @@ static std::array<u8, target_size> MGF1(const std::array<u8, in_size>& seed) {
 }
 
 template <size_t size>
-static boost::optional<u64> FindTicketOffset(const std::array<u8, size>& data) {
+static std::optional<u64> FindTicketOffset(const std::array<u8, size>& data) {
     u64 offset = 0;
     for (size_t i = 0x20; i < data.size() - 0x10; ++i) {
         if (data[i] == 0x1) {
             offset = i + 1;
             break;
         } else if (data[i] != 0x0) {
-            return boost::none;
+            return {};
         }
     }
 
     return offset;
 }
 
-boost::optional<std::pair<Key128, Key128>> ParseTicket(const TicketRaw& ticket,
-                                                       const RSAKeyPair<2048>& key) {
+std::optional<std::pair<Key128, Key128>> ParseTicket(const TicketRaw& ticket,
+                                                     const RSAKeyPair<2048>& key) {
     u32 cert_authority;
     std::memcpy(&cert_authority, ticket.data() + 0x140, sizeof(cert_authority));
     if (cert_authority == 0)
-        return boost::none;
+        return {};
     if (cert_authority != Common::MakeMagic('R', 'o', 'o', 't')) {
         LOG_INFO(Crypto,
                  "Attempting to parse ticket with non-standard certificate authority {:08X}.",
@@ -321,7 +321,7 @@ boost::optional<std::pair<Key128, Key128>> ParseTicket(const TicketRaw& ticket,
     std::memcpy(rights_id.data(), ticket.data() + 0x2A0, sizeof(Key128));
 
     if (rights_id == Key128{})
-        return boost::none;
+        return {};
 
     Key128 key_temp{};
 
@@ -356,17 +356,17 @@ boost::optional<std::pair<Key128, Key128>> ParseTicket(const TicketRaw& ticket,
     std::memcpy(m_2.data(), rsa_step.data() + 0x21, m_2.size());
 
     if (m_0 != 0)
-        return boost::none;
+        return {};
 
     m_1 = m_1 ^ MGF1<0x20>(m_2);
     m_2 = m_2 ^ MGF1<0xDF>(m_1);
 
     const auto offset = FindTicketOffset(m_2);
-    if (offset == boost::none)
-        return boost::none;
-    ASSERT(offset.get() > 0);
+    if (!offset)
+        return {};
+    ASSERT(*offset > 0);
 
-    std::memcpy(key_temp.data(), m_2.data() + offset.get(), key_temp.size());
+    std::memcpy(key_temp.data(), m_2.data() + *offset, key_temp.size());
 
     return std::make_pair(rights_id, key_temp);
 }
@@ -395,7 +395,7 @@ static bool ValidCryptoRevisionString(std::string_view base, size_t begin, size_
     if (base.size() < begin + length)
         return false;
     return std::all_of(base.begin() + begin, base.begin() + begin + length,
-                       [](u8 c) { return std::isdigit(c); });
+                       [](u8 c) { return std::isxdigit(c); });
 }
 
 void KeyManager::LoadFromFile(const std::string& filename, bool is_title_keys) {
@@ -661,8 +661,8 @@ void KeyManager::DeriveSDSeedLazy() {
         return;
 
     const auto res = DeriveSDSeed();
-    if (res != boost::none)
-        SetKey(S128KeyType::SDSeed, res.get());
+    if (res)
+        SetKey(S128KeyType::SDSeed, *res);
 }
 
 static Key128 CalculateCMAC(const u8* source, size_t size, const Key128& key) {
@@ -889,9 +889,9 @@ void KeyManager::DeriveETicket(PartitionDataManager& data) {
 
     for (const auto& raw : res) {
         const auto pair = ParseTicket(raw, rsa_key);
-        if (pair == boost::none)
+        if (!pair)
             continue;
-        const auto& [rid, key] = pair.value();
+        const auto& [rid, key] = *pair;
         u128 rights_id;
         std::memcpy(rights_id.data(), rid.data(), rid.size());
         SetKey(S128KeyType::Titlekey, key, rights_id[1], rights_id[0]);

src/core/crypto/key_manager.h

@@ -6,9 +6,10 @@
 
 #include <array>
 #include <map>
+#include <optional>
 #include <string>
+
 #include <boost/container/flat_map.hpp>
-#include <boost/optional.hpp>
 #include <fmt/format.h>
 #include "common/common_types.h"
 #include "core/crypto/partition_data_manager.h"
@@ -191,14 +192,14 @@ Key128 DeriveMasterKey(const std::array<u8, 0x90>& keyblob, const Key128& master
 std::array<u8, 0x90> DecryptKeyblob(const std::array<u8, 0xB0>& encrypted_keyblob,
                                     const Key128& key);
 
-boost::optional<Key128> DeriveSDSeed();
+std::optional<Key128> DeriveSDSeed();
 Loader::ResultStatus DeriveSDKeys(std::array<Key256, 2>& sd_keys, KeyManager& keys);
 
 std::vector<TicketRaw> GetTicketblob(const FileUtil::IOFile& ticket_save);
 
 // Returns a pair of {rights_id, titlekey}. Fails if the ticket has no certificate authority (offset
 // 0x140-0x144 is zero)
-boost::optional<std::pair<Key128, Key128>> ParseTicket(
-    const TicketRaw& ticket, const RSAKeyPair<2048>& eticket_extended_key);
+std::optional<std::pair<Key128, Key128>> ParseTicket(const TicketRaw& ticket,
+                                                     const RSAKeyPair<2048>& eticket_extended_key);
 
 } // namespace Core::Crypto

src/core/file_sys/content_archive.cpp

@@ -4,10 +4,9 @@
 
 #include <algorithm>
 #include <cstring>
+#include <optional>
 #include <utility>
 
-#include <boost/optional.hpp>
-
 #include "common/logging/log.h"
 #include "core/crypto/aes_util.h"
 #include "core/crypto/ctr_encryption_layer.h"
@@ -306,18 +305,18 @@ bool NCA::ReadRomFSSection(const NCASectionHeader& section, const NCASectionTabl
         subsection_buckets.back().entries.push_back({section.bktr.relocation.offset, {0}, ctr_low});
         subsection_buckets.back().entries.push_back({size, {0}, 0});
 
-        boost::optional<Core::Crypto::Key128> key = boost::none;
+        std::optional<Core::Crypto::Key128> key = {};
         if (encrypted) {
             if (has_rights_id) {
                 status = Loader::ResultStatus::Success;
                 key = GetTitlekey();
-                if (key == boost::none) {
+                if (!key) {
                     status = Loader::ResultStatus::ErrorMissingTitlekey;
                     return false;
                 }
             } else {
                 key = GetKeyAreaKey(NCASectionCryptoType::BKTR);
-                if (key == boost::none) {
+                if (!key) {
                     status = Loader::ResultStatus::ErrorMissingKeyAreaKey;
                     return false;
                 }
@@ -332,7 +331,7 @@ bool NCA::ReadRomFSSection(const NCASectionHeader& section, const NCASectionTabl
         auto bktr = std::make_shared<BKTR>(
             bktr_base_romfs, std::make_shared<OffsetVfsFile>(file, romfs_size, base_offset),
             relocation_block, relocation_buckets, subsection_block, subsection_buckets, encrypted,
-            encrypted ? key.get() : Core::Crypto::Key128{}, base_offset, bktr_base_ivfc_offset,
+            encrypted ? *key : Core::Crypto::Key128{}, base_offset, bktr_base_ivfc_offset,
             section.raw.section_ctr);
 
         // BKTR applies to entire IVFC, so make an offset version to level 6
@@ -388,11 +387,11 @@ u8 NCA::GetCryptoRevision() const {
     return master_key_id;
 }
 
-boost::optional<Core::Crypto::Key128> NCA::GetKeyAreaKey(NCASectionCryptoType type) const {
+std::optional<Core::Crypto::Key128> NCA::GetKeyAreaKey(NCASectionCryptoType type) const {
     const auto master_key_id = GetCryptoRevision();
 
     if (!keys.HasKey(Core::Crypto::S128KeyType::KeyArea, master_key_id, header.key_index))
-        return boost::none;
+        return {};
 
     std::vector<u8> key_area(header.key_area.begin(), header.key_area.end());
     Core::Crypto::AESCipher<Core::Crypto::Key128> cipher(
@@ -416,25 +415,25 @@ boost::optional<Core::Crypto::Key128> NCA::GetKeyAreaKey(NCASectionCryptoType ty
     return out;
 }
 
-boost::optional<Core::Crypto::Key128> NCA::GetTitlekey() {
+std::optional<Core::Crypto::Key128> NCA::GetTitlekey() {
     const auto master_key_id = GetCryptoRevision();
 
     u128 rights_id{};
     memcpy(rights_id.data(), header.rights_id.data(), 16);
     if (rights_id == u128{}) {
         status = Loader::ResultStatus::ErrorInvalidRightsID;
-        return boost::none;
+        return {};
     }
 
     auto titlekey = keys.GetKey(Core::Crypto::S128KeyType::Titlekey, rights_id[1], rights_id[0]);
     if (titlekey == Core::Crypto::Key128{}) {
         status = Loader::ResultStatus::ErrorMissingTitlekey;
-        return boost::none;
+        return {};
     }
 
     if (!keys.HasKey(Core::Crypto::S128KeyType::Titlekek, master_key_id)) {
         status = Loader::ResultStatus::ErrorMissingTitlekek;
-        return boost::none;
+        return {};
     }
 
     Core::Crypto::AESCipher<Core::Crypto::Key128> cipher(
@@ -458,25 +457,25 @@ VirtualFile NCA::Decrypt(const NCASectionHeader& s_header, VirtualFile in, u64 s
     case NCASectionCryptoType::BKTR:
         LOG_DEBUG(Crypto, "called with mode=CTR, starting_offset={:016X}", starting_offset);
         {
-            boost::optional<Core::Crypto::Key128> key = boost::none;
+            std::optional<Core::Crypto::Key128> key = {};
             if (has_rights_id) {
                 status = Loader::ResultStatus::Success;
                 key = GetTitlekey();
-                if (key == boost::none) {
+                if (!key) {
                     if (status == Loader::ResultStatus::Success)
                         status = Loader::ResultStatus::ErrorMissingTitlekey;
                     return nullptr;
                 }
             } else {
                 key = GetKeyAreaKey(NCASectionCryptoType::CTR);
-                if (key == boost::none) {
+                if (!key) {
                     status = Loader::ResultStatus::ErrorMissingKeyAreaKey;
                     return nullptr;
                 }
             }
 
-            auto out = std::make_shared<Core::Crypto::CTREncryptionLayer>(
-                std::move(in), key.value(), starting_offset);
+            auto out = std::make_shared<Core::Crypto::CTREncryptionLayer>(std::move(in), *key,
+                                                                          starting_offset);
             std::vector<u8> iv(16);
             for (u8 i = 0; i < 8; ++i)
                 iv[i] = s_header.raw.section_ctr[0x8 - i - 1];

src/core/file_sys/content_archive.h

@@ -6,9 +6,10 @@
 
 #include <array>
 #include <memory>
+#include <optional>
 #include <string>
 #include <vector>
-#include <boost/optional.hpp>
+
 #include "common/common_funcs.h"
 #include "common/common_types.h"
 #include "common/swap.h"
@@ -111,8 +112,8 @@ private:
     bool ReadPFS0Section(const NCASectionHeader& section, const NCASectionTableEntry& entry);
 
     u8 GetCryptoRevision() const;
-    boost::optional<Core::Crypto::Key128> GetKeyAreaKey(NCASectionCryptoType type) const;
-    boost::optional<Core::Crypto::Key128> GetTitlekey();
+    std::optional<Core::Crypto::Key128> GetKeyAreaKey(NCASectionCryptoType type) const;
+    std::optional<Core::Crypto::Key128> GetTitlekey();
     VirtualFile Decrypt(const NCASectionHeader& header, VirtualFile in, u64 starting_offset);
 
     std::vector<VirtualDir> dirs;

src/core/file_sys/fsmitm_romfsbuild.h

@@ -27,7 +27,6 @@
 #include <map>
 #include <memory>
 #include <string>
-#include <boost/detail/container_fwd.hpp>
 #include "common/common_types.h"
 #include "core/file_sys/vfs.h"
 

src/core/file_sys/ips_layer.cpp

@@ -103,12 +103,12 @@ VirtualFile PatchIPS(const VirtualFile& in, const VirtualFile& ips) {
             offset += sizeof(u16);
 
             const auto data = ips->ReadByte(offset++);
-            if (data == boost::none)
+            if (!data)
                 return nullptr;
 
             if (real_offset + rle_size > in_data.size())
                 rle_size = static_cast<u16>(in_data.size() - real_offset);
-            std::memset(in_data.data() + real_offset, data.get(), rle_size);
+            std::memset(in_data.data() + real_offset, *data, rle_size);
         } else { // Standard Patch
             auto read = data_size;
             if (real_offset + read > in_data.size())

src/core/file_sys/patch_manager.cpp

@@ -61,13 +61,12 @@ VirtualDir PatchManager::PatchExeFS(VirtualDir exefs) const {
     // Game Updates
     const auto update_tid = GetUpdateTitleID(title_id);
     const auto update = installed->GetEntry(update_tid, ContentRecordType::Program);
-    if (update != nullptr) {
-        if (update->GetStatus() == Loader::ResultStatus::ErrorMissingBKTRBaseRomFS &&
-            update->GetExeFS() != nullptr) {
-            LOG_INFO(Loader, "    ExeFS: Update ({}) applied successfully",
-                     FormatTitleVersion(installed->GetEntryVersion(update_tid).get_value_or(0)));
-            exefs = update->GetExeFS();
-        }
+
+    if (update != nullptr && update->GetExeFS() != nullptr &&
+        update->GetStatus() == Loader::ResultStatus::ErrorMissingBKTRBaseRomFS) {
+        LOG_INFO(Loader, "    ExeFS: Update ({}) applied successfully",
+                 FormatTitleVersion(installed->GetEntryVersion(update_tid).value_or(0)));
+        exefs = update->GetExeFS();
     }
 
     return exefs;
@@ -237,7 +236,7 @@ VirtualFile PatchManager::PatchRomFS(VirtualFile romfs, u64 ivfc_offset, Content
         if (new_nca->GetStatus() == Loader::ResultStatus::Success &&
             new_nca->GetRomFS() != nullptr) {
             LOG_INFO(Loader, "    RomFS: Update ({}) applied successfully",
-                     FormatTitleVersion(installed->GetEntryVersion(update_tid).get_value_or(0)));
+                     FormatTitleVersion(installed->GetEntryVersion(update_tid).value_or(0)));
             romfs = new_nca->GetRomFS();
         }
     } else if (update_raw != nullptr) {
@@ -281,12 +280,11 @@ std::map<std::string, std::string, std::less<>> PatchManager::GetPatchVersionNam
     } else {
         if (installed->HasEntry(update_tid, ContentRecordType::Program)) {
             const auto meta_ver = installed->GetEntryVersion(update_tid);
-            if (meta_ver == boost::none || meta_ver.get() == 0) {
+            if (meta_ver.value_or(0) == 0) {
                 out.insert_or_assign("Update", "");
             } else {
                 out.insert_or_assign(
-                    "Update",
-                    FormatTitleVersion(meta_ver.get(), TitleVersionFormat::ThreeElements));
+                    "Update", FormatTitleVersion(*meta_ver, TitleVersionFormat::ThreeElements));
             }
         } else if (update_raw != nullptr) {
             out.insert_or_assign("Update", "PACKED");

src/core/file_sys/registered_cache.cpp

@@ -159,28 +159,28 @@ VirtualFile RegisteredCache::GetFileAtID(NcaID id) const {
     return file;
 }
 
-static boost::optional<NcaID> CheckMapForContentRecord(
+static std::optional<NcaID> CheckMapForContentRecord(
     const boost::container::flat_map<u64, CNMT>& map, u64 title_id, ContentRecordType type) {
     if (map.find(title_id) == map.end())
-        return boost::none;
+        return {};
 
     const auto& cnmt = map.at(title_id);
 
     const auto iter = std::find_if(cnmt.GetContentRecords().begin(), cnmt.GetContentRecords().end(),
                                    [type](const ContentRecord& rec) { return rec.type == type; });
     if (iter == cnmt.GetContentRecords().end())
-        return boost::none;
+        return {};
 
-    return boost::make_optional(iter->nca_id);
+    return std::make_optional(iter->nca_id);
 }
 
-boost::optional<NcaID> RegisteredCache::GetNcaIDFromMetadata(u64 title_id,
-                                                             ContentRecordType type) const {
+std::optional<NcaID> RegisteredCache::GetNcaIDFromMetadata(u64 title_id,
+                                                           ContentRecordType type) const {
     if (type == ContentRecordType::Meta && meta_id.find(title_id) != meta_id.end())
         return meta_id.at(title_id);
 
     const auto res1 = CheckMapForContentRecord(yuzu_meta, title_id, type);
-    if (res1 != boost::none)
+    if (res1)
         return res1;
     return CheckMapForContentRecord(meta, title_id, type);
 }
@@ -283,17 +283,14 @@ bool RegisteredCache::HasEntry(RegisteredCacheEntry entry) const {
 
 VirtualFile RegisteredCache::GetEntryUnparsed(u64 title_id, ContentRecordType type) const {
     const auto id = GetNcaIDFromMetadata(title_id, type);
-    if (id == boost::none)
-        return nullptr;
-
-    return GetFileAtID(id.get());
+    return id ? GetFileAtID(*id) : nullptr;
 }
 
 VirtualFile RegisteredCache::GetEntryUnparsed(RegisteredCacheEntry entry) const {
     return GetEntryUnparsed(entry.title_id, entry.type);
 }
 
-boost::optional<u32> RegisteredCache::GetEntryVersion(u64 title_id) const {
+std::optional<u32> RegisteredCache::GetEntryVersion(u64 title_id) const {
     const auto meta_iter = meta.find(title_id);
     if (meta_iter != meta.end())
         return meta_iter->second.GetTitleVersion();
@@ -302,15 +299,12 @@ boost::optional<u32> RegisteredCache::GetEntryVersion(u64 title_id) const {
     if (yuzu_meta_iter != yuzu_meta.end())
         return yuzu_meta_iter->second.GetTitleVersion();
 
-    return boost::none;
+    return {};
 }
 
 VirtualFile RegisteredCache::GetEntryRaw(u64 title_id, ContentRecordType type) const {
     const auto id = GetNcaIDFromMetadata(title_id, type);
-    if (id == boost::none)
-        return nullptr;
-
-    return parser(GetFileAtID(id.get()), id.get());
+    return id ? parser(GetFileAtID(*id), *id) : nullptr;
 }
 
 VirtualFile RegisteredCache::GetEntryRaw(RegisteredCacheEntry entry) const {
@@ -364,8 +358,8 @@ std::vector<RegisteredCacheEntry> RegisteredCache::ListEntries() const {
 }
 
 std::vector<RegisteredCacheEntry> RegisteredCache::ListEntriesFilter(
-    boost::optional<TitleType> title_type, boost::optional<ContentRecordType> record_type,
-    boost::optional<u64> title_id) const {
+    std::optional<TitleType> title_type, std::optional<ContentRecordType> record_type,
+    std::optional<u64> title_id) const {
     std::vector<RegisteredCacheEntry> out;
     IterateAllMetadata<RegisteredCacheEntry>(
         out,
@@ -373,11 +367,11 @@ std::vector<RegisteredCacheEntry> RegisteredCache::ListEntriesFilter(
             return RegisteredCacheEntry{c.GetTitleID(), r.type};
         },
         [&title_type, &record_type, &title_id](const CNMT& c, const ContentRecord& r) {
-            if (title_type != boost::none && title_type.get() != c.GetType())
+            if (title_type && *title_type != c.GetType())
                 return false;
-            if (record_type != boost::none && record_type.get() != r.type)
+            if (record_type && *record_type != r.type)
                 return false;
-            if (title_id != boost::none && title_id.get() != c.GetTitleID())
+            if (title_id && *title_id != c.GetTitleID())
                 return false;
             return true;
         });
@@ -459,7 +453,7 @@ InstallResult RegisteredCache::InstallEntry(std::shared_ptr<NCA> nca, TitleType
 
 InstallResult RegisteredCache::RawInstallNCA(std::shared_ptr<NCA> nca, const VfsCopyFunction& copy,
                                              bool overwrite_if_exists,
-                                             boost::optional<NcaID> override_id) {
+                                             std::optional<NcaID> override_id) {
     const auto in = nca->GetBaseFile();
     Core::Crypto::SHA256Hash hash{};
 
@@ -468,12 +462,12 @@ InstallResult RegisteredCache::RawInstallNCA(std::shared_ptr<NCA> nca, const Vfs
     // game is massive), we're going to cheat and only hash the first MB of the NCA.
     // Also, for XCIs the NcaID matters, so if the override id isn't none, use that.
     NcaID id{};
-    if (override_id == boost::none) {
+    if (override_id) {
+        id = *override_id;
+    } else {
         const auto& data = in->ReadBytes(0x100000);
         mbedtls_sha256(data.data(), data.size(), hash.data(), 0);
         memcpy(id.data(), hash.data(), 16);
-    } else {
-        id = override_id.get();
     }
 
     std::string path = GetRelativePathFromNcaID(id, false, true);
@@ -543,14 +537,14 @@ bool RegisteredCacheUnion::HasEntry(RegisteredCacheEntry entry) const {
     return HasEntry(entry.title_id, entry.type);
 }
 
-boost::optional<u32> RegisteredCacheUnion::GetEntryVersion(u64 title_id) const {
+std::optional<u32> RegisteredCacheUnion::GetEntryVersion(u64 title_id) const {
     for (const auto& c : caches) {
         const auto res = c->GetEntryVersion(title_id);
-        if (res != boost::none)
+        if (res)
             return res;
     }
 
-    return boost::none;
+    return {};
 }
 
 VirtualFile RegisteredCacheUnion::GetEntryUnparsed(u64 title_id, ContentRecordType type) const {
@@ -609,8 +603,8 @@ std::vector<RegisteredCacheEntry> RegisteredCacheUnion::ListEntries() const {
 }
 
 std::vector<RegisteredCacheEntry> RegisteredCacheUnion::ListEntriesFilter(
-    boost::optional<TitleType> title_type, boost::optional<ContentRecordType> record_type,
-    boost::optional<u64> title_id) const {
+    std::optional<TitleType> title_type, std::optional<ContentRecordType> record_type,
+    std::optional<u64> title_id) const {
     std::vector<RegisteredCacheEntry> out;
     for (const auto& c : caches) {
         c->IterateAllMetadata<RegisteredCacheEntry>(
@@ -619,11 +613,11 @@ std::vector<RegisteredCacheEntry> RegisteredCacheUnion::ListEntriesFilter(
                 return RegisteredCacheEntry{c.GetTitleID(), r.type};
             },
             [&title_type, &record_type, &title_id](const CNMT& c, const ContentRecord& r) {
-                if (title_type != boost::none && title_type.get() != c.GetType())
+                if (title_type && *title_type != c.GetType())
                     return false;
-                if (record_type != boost::none && record_type.get() != r.type)
+                if (record_type && *record_type != r.type)
                     return false;
-                if (title_id != boost::none && title_id.get() != c.GetTitleID())
+                if (title_id && *title_id != c.GetTitleID())
                     return false;
                 return true;
             });

src/core/file_sys/registered_cache.h

@@ -84,7 +84,7 @@ public:
     bool HasEntry(u64 title_id, ContentRecordType type) const;
     bool HasEntry(RegisteredCacheEntry entry) const;
 
-    boost::optional<u32> GetEntryVersion(u64 title_id) const;
+    std::optional<u32> GetEntryVersion(u64 title_id) const;
 
     VirtualFile GetEntryUnparsed(u64 title_id, ContentRecordType type) const;
     VirtualFile GetEntryUnparsed(RegisteredCacheEntry entry) const;
@@ -96,11 +96,10 @@ public:
     std::unique_ptr<NCA> GetEntry(RegisteredCacheEntry entry) const;
 
     std::vector<RegisteredCacheEntry> ListEntries() const;
-    // If a parameter is not boost::none, it will be filtered for from all entries.
+    // If a parameter is not std::nullopt, it will be filtered for from all entries.
     std::vector<RegisteredCacheEntry> ListEntriesFilter(
-        boost::optional<TitleType> title_type = boost::none,
-        boost::optional<ContentRecordType> record_type = boost::none,
-        boost::optional<u64> title_id = boost::none) const;
+        std::optional<TitleType> title_type = {}, std::optional<ContentRecordType> record_type = {},
+        std::optional<u64> title_id = {}) const;
 
     // Raw copies all the ncas from the xci/nsp to the csache. Does some quick checks to make sure
     // there is a meta NCA and all of them are accessible.
@@ -125,12 +124,11 @@ private:
     std::vector<NcaID> AccumulateFiles() const;
     void ProcessFiles(const std::vector<NcaID>& ids);
     void AccumulateYuzuMeta();
-    boost::optional<NcaID> GetNcaIDFromMetadata(u64 title_id, ContentRecordType type) const;
+    std::optional<NcaID> GetNcaIDFromMetadata(u64 title_id, ContentRecordType type) const;
     VirtualFile GetFileAtID(NcaID id) const;
     VirtualFile OpenFileOrDirectoryConcat(const VirtualDir& dir, std::string_view path) const;
     InstallResult RawInstallNCA(std::shared_ptr<NCA> nca, const VfsCopyFunction& copy,
-                                bool overwrite_if_exists,
-                                boost::optional<NcaID> override_id = boost::none);
+                                bool overwrite_if_exists, std::optional<NcaID> override_id = {});
     bool RawInstallYuzuMeta(const CNMT& cnmt);
 
     VirtualDir dir;
@@ -153,7 +151,7 @@ public:
     bool HasEntry(u64 title_id, ContentRecordType type) const;
     bool HasEntry(RegisteredCacheEntry entry) const;
 
-    boost::optional<u32> GetEntryVersion(u64 title_id) const;
+    std::optional<u32> GetEntryVersion(u64 title_id) const;
 
     VirtualFile GetEntryUnparsed(u64 title_id, ContentRecordType type) const;
     VirtualFile GetEntryUnparsed(RegisteredCacheEntry entry) const;
@@ -165,11 +163,10 @@ public:
     std::unique_ptr<NCA> GetEntry(RegisteredCacheEntry entry) const;
 
     std::vector<RegisteredCacheEntry> ListEntries() const;
-    // If a parameter is not boost::none, it will be filtered for from all entries.
+    // If a parameter is not std::nullopt, it will be filtered for from all entries.
     std::vector<RegisteredCacheEntry> ListEntriesFilter(
-        boost::optional<TitleType> title_type = boost::none,
-        boost::optional<ContentRecordType> record_type = boost::none,
-        boost::optional<u64> title_id = boost::none) const;
+        std::optional<TitleType> title_type = {}, std::optional<ContentRecordType> record_type = {},
+        std::optional<u64> title_id = {}) const;
 
 private:
     std::vector<RegisteredCache*> caches;

src/core/file_sys/vfs.cpp

@@ -167,13 +167,13 @@ std::string VfsFile::GetExtension() const {
 
 VfsDirectory::~VfsDirectory() = default;
 
-boost::optional<u8> VfsFile::ReadByte(std::size_t offset) const {
+std::optional<u8> VfsFile::ReadByte(std::size_t offset) const {
     u8 out{};
     std::size_t size = Read(&out, 1, offset);
     if (size == 1)
         return out;
 
-    return boost::none;
+    return {};
 }
 
 std::vector<u8> VfsFile::ReadBytes(std::size_t size, std::size_t offset) const {

src/core/file_sys/vfs.h

@@ -4,13 +4,15 @@
 
 #pragma once
 
+#include <functional>
 #include <map>
 #include <memory>
+#include <optional>
 #include <string>
 #include <string_view>
 #include <type_traits>
 #include <vector>
-#include <boost/optional.hpp>
+
 #include "common/common_types.h"
 #include "core/file_sys/vfs_types.h"
 
@@ -103,8 +105,8 @@ public:
     // into file. Returns number of bytes successfully written.
     virtual std::size_t Write(const u8* data, std::size_t length, std::size_t offset = 0) = 0;
 
-    // Reads exactly one byte at the offset provided, returning boost::none on error.
-    virtual boost::optional<u8> ReadByte(std::size_t offset = 0) const;
+    // Reads exactly one byte at the offset provided, returning std::nullopt on error.
+    virtual std::optional<u8> ReadByte(std::size_t offset = 0) const;
     // Reads size bytes starting at offset in file into a vector.
     virtual std::vector<u8> ReadBytes(std::size_t size, std::size_t offset = 0) const;
     // Reads all the bytes from the file into a vector. Equivalent to 'file->Read(file->GetSize(),

src/core/file_sys/vfs_offset.cpp

@@ -57,11 +57,11 @@ std::size_t OffsetVfsFile::Write(const u8* data, std::size_t length, std::size_t
     return file->Write(data, TrimToFit(length, r_offset), offset + r_offset);
 }
 
-boost::optional<u8> OffsetVfsFile::ReadByte(std::size_t r_offset) const {
+std::optional<u8> OffsetVfsFile::ReadByte(std::size_t r_offset) const {
     if (r_offset < size)
         return file->ReadByte(offset + r_offset);
 
-    return boost::none;
+    return {};
 }
 
 std::vector<u8> OffsetVfsFile::ReadBytes(std::size_t r_size, std::size_t r_offset) const {

src/core/file_sys/vfs_offset.h

@@ -29,7 +29,7 @@ public:
     bool IsReadable() const override;
     std::size_t Read(u8* data, std::size_t length, std::size_t offset) const override;
     std::size_t Write(const u8* data, std::size_t length, std::size_t offset) override;
-    boost::optional<u8> ReadByte(std::size_t offset) const override;
+    std::optional<u8> ReadByte(std::size_t offset) const override;
     std::vector<u8> ReadBytes(std::size_t size, std::size_t offset) const override;
     std::vector<u8> ReadAllBytes() const override;
     bool WriteByte(u8 data, std::size_t offset) override;

src/core/file_sys/vfs_static.h

@@ -53,10 +53,10 @@ public:
         return 0;
     }
 
-    boost::optional<u8> ReadByte(std::size_t offset) const override {
+    std::optional<u8> ReadByte(std::size_t offset) const override {
         if (offset < size)
             return value;
-        return boost::none;
+        return {};
     }
 
     std::vector<u8> ReadBytes(std::size_t length, std::size_t offset) const override {

src/core/hle/ipc_helpers.h

@@ -117,8 +117,7 @@ public:
 
         AlignWithPadding();
 
-        const bool request_has_domain_header{context.GetDomainMessageHeader() != nullptr};
-        if (context.Session()->IsDomain() && request_has_domain_header) {
+        if (context.Session()->IsDomain() && context.HasDomainMessageHeader()) {
             IPC::DomainMessageHeader domain_header{};
             domain_header.num_objects = num_domain_objects;
             PushRaw(domain_header);

src/core/hle/kernel/hle_ipc.h

@@ -161,8 +161,12 @@ public:
         return buffer_c_desciptors;
     }
 
-    const std::shared_ptr<IPC::DomainMessageHeader>& GetDomainMessageHeader() const {
-        return domain_message_header;
+    const IPC::DomainMessageHeader* GetDomainMessageHeader() const {
+        return domain_message_header.get();
+    }
+
+    bool HasDomainMessageHeader() const {
+        return domain_message_header != nullptr;
     }
 
     /// Helper function to read a buffer using the appropriate buffer descriptor

src/core/hle/kernel/kernel.cpp

@@ -32,7 +32,7 @@ namespace Kernel {
  */
 static void ThreadWakeupCallback(u64 thread_handle, [[maybe_unused]] int cycles_late) {
     const auto proper_handle = static_cast<Handle>(thread_handle);
-    auto& system = Core::System::GetInstance();
+    const auto& system = Core::System::GetInstance();
 
     // Lock the global kernel mutex when we enter the kernel HLE.
     std::lock_guard<std::recursive_mutex> lock(HLE::g_hle_lock);
@@ -90,7 +90,7 @@ static void ThreadWakeupCallback(u64 thread_handle, [[maybe_unused]] int cycles_
 /// The timer callback event, called when a timer is fired
 static void TimerCallback(u64 timer_handle, int cycles_late) {
     const auto proper_handle = static_cast<Handle>(timer_handle);
-    auto& system = Core::System::GetInstance();
+    const auto& system = Core::System::GetInstance();
     SharedPtr<Timer> timer = system.Kernel().RetrieveTimerFromCallbackHandleTable(proper_handle);
 
     if (timer == nullptr) {

src/core/hle/kernel/mutex.cpp

@@ -6,8 +6,6 @@
 #include <utility>
 #include <vector>
 
-#include <boost/range/algorithm_ext/erase.hpp>
-
 #include "common/assert.h"
 #include "core/core.h"
 #include "core/hle/kernel/errors.h"

src/core/hle/kernel/process.h

@@ -202,6 +202,16 @@ public:
         return is_64bit_process;
     }
 
+    /// Gets the total running time of the process instance in ticks.
+    u64 GetCPUTimeTicks() const {
+        return total_process_running_time_ticks;
+    }
+
+    /// Updates the total running time, adding the given ticks to it.
+    void UpdateCPUTimeTicks(u64 ticks) {
+        total_process_running_time_ticks += ticks;
+    }
+
     /**
      * Loads process-specifics configuration info with metadata provided
      * by an executable.
@@ -305,6 +315,9 @@ private:
     /// specified by metadata provided to the process during loading.
     bool is_64bit_process = true;
 
+    /// Total running time for the process in ticks.
+    u64 total_process_running_time_ticks = 0;
+
     /// Per-process handle table for storing created object handles in.
     HandleTable handle_table;
 

src/core/hle/kernel/scheduler.cpp

@@ -9,6 +9,7 @@
 #include "common/logging/log.h"
 #include "core/arm/arm_interface.h"
 #include "core/core.h"
+#include "core/core_timing.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/scheduler.h"
@@ -34,6 +35,10 @@ Thread* Scheduler::GetCurrentThread() const {
     return current_thread.get();
 }
 
+u64 Scheduler::GetLastContextSwitchTicks() const {
+    return last_context_switch_time;
+}
+
 Thread* Scheduler::PopNextReadyThread() {
     Thread* next = nullptr;
     Thread* thread = GetCurrentThread();
@@ -54,7 +59,10 @@ Thread* Scheduler::PopNextReadyThread() {
 }
 
 void Scheduler::SwitchContext(Thread* new_thread) {
-    Thread* previous_thread = GetCurrentThread();
+    Thread* const previous_thread = GetCurrentThread();
+    Process* const previous_process = Core::CurrentProcess();
+
+    UpdateLastContextSwitchTime(previous_thread, previous_process);
 
     // Save context for previous thread
     if (previous_thread) {
@@ -78,8 +86,6 @@ void Scheduler::SwitchContext(Thread* new_thread) {
         // Cancel any outstanding wakeup events for this thread
         new_thread->CancelWakeupTimer();
 
-        auto* const previous_process = Core::CurrentProcess();
-
         current_thread = new_thread;
 
         ready_queue.remove(new_thread->GetPriority(), new_thread);
@@ -102,6 +108,22 @@ void Scheduler::SwitchContext(Thread* new_thread) {
     }
 }
 
+void Scheduler::UpdateLastContextSwitchTime(Thread* thread, Process* process) {
+    const u64 prev_switch_ticks = last_context_switch_time;
+    const u64 most_recent_switch_ticks = CoreTiming::GetTicks();
+    const u64 update_ticks = most_recent_switch_ticks - prev_switch_ticks;
+
+    if (thread != nullptr) {
+        thread->UpdateCPUTimeTicks(update_ticks);
+    }
+
+    if (process != nullptr) {
+        process->UpdateCPUTimeTicks(update_ticks);
+    }
+
+    last_context_switch_time = most_recent_switch_ticks;
+}
+
 void Scheduler::Reschedule() {
     std::lock_guard<std::mutex> lock(scheduler_mutex);
 

src/core/hle/kernel/scheduler.h

@@ -17,6 +17,8 @@ class ARM_Interface;
 
 namespace Kernel {
 
+class Process;
+
 class Scheduler final {
 public:
     explicit Scheduler(Core::ARM_Interface& cpu_core);
@@ -31,6 +33,9 @@ public:
     /// Gets the current running thread
     Thread* GetCurrentThread() const;
 
+    /// Gets the timestamp for the last context switch in ticks.
+    u64 GetLastContextSwitchTicks() const;
+
     /// Adds a new thread to the scheduler
     void AddThread(SharedPtr<Thread> thread, u32 priority);
 
@@ -64,6 +69,19 @@ private:
      */
     void SwitchContext(Thread* new_thread);
 
+    /**
+     * Called on every context switch to update the internal timestamp
+     * This also updates the running time ticks for the given thread and
+     * process using the following difference:
+     *
+     * ticks += most_recent_ticks - last_context_switch_ticks
+     *
+     * The internal tick timestamp for the scheduler is simply the
+     * most recent tick count retrieved. No special arithmetic is
+     * applied to it.
+     */
+    void UpdateLastContextSwitchTime(Thread* thread, Process* process);
+
     /// Lists all thread ids that aren't deleted/etc.
     std::vector<SharedPtr<Thread>> thread_list;
 
@@ -73,6 +91,7 @@ private:
     SharedPtr<Thread> current_thread = nullptr;
 
     Core::ARM_Interface& cpu_core;
+    u64 last_context_switch_time = 0;
 
     static std::mutex scheduler_mutex;
 };

src/core/hle/kernel/server_session.cpp

@@ -63,7 +63,7 @@ void ServerSession::Acquire(Thread* thread) {
 }
 
 ResultCode ServerSession::HandleDomainSyncRequest(Kernel::HLERequestContext& context) {
-    auto& domain_message_header = context.GetDomainMessageHeader();
+    auto* const domain_message_header = context.GetDomainMessageHeader();
     if (domain_message_header) {
         // Set domain handlers in HLE context, used for domain objects (IPC interfaces) as inputs
         context.SetDomainRequestHandlers(domain_request_handlers);
@@ -111,7 +111,7 @@ ResultCode ServerSession::HandleSyncRequest(SharedPtr<Thread> thread) {
 
     ResultCode result = RESULT_SUCCESS;
     // If the session has been converted to a domain, handle the domain request
-    if (IsDomain() && context.GetDomainMessageHeader()) {
+    if (IsDomain() && context.HasDomainMessageHeader()) {
         result = HandleDomainSyncRequest(context);
         // If there is no domain header, the regular session handler is used
     } else if (hle_handler != nullptr) {

src/core/hle/kernel/svc.cpp

@@ -34,6 +34,7 @@
 #include "core/hle/lock.h"
 #include "core/hle/result.h"
 #include "core/hle/service/service.h"
+#include "core/settings.h"
 
 namespace Kernel {
 namespace {
@@ -395,16 +396,42 @@ struct BreakReason {
 /// Break program execution
 static void Break(u32 reason, u64 info1, u64 info2) {
     BreakReason break_reason{reason};
+    bool has_dumped_buffer{};
 
+    const auto handle_debug_buffer = [&](VAddr addr, u64 sz) {
+        if (sz == 0 || addr == 0 || has_dumped_buffer) {
+            return;
+        }
+
+        // This typically is an error code so we're going to assume this is the case
+        if (sz == sizeof(u32)) {
+            LOG_CRITICAL(Debug_Emulated, "debug_buffer_err_code={:X}", Memory::Read32(addr));
+        } else {
+            // We don't know what's in here so we'll hexdump it
+            std::vector<u8> debug_buffer(sz);
+            Memory::ReadBlock(addr, debug_buffer.data(), sz);
+            std::string hexdump;
+            for (std::size_t i = 0; i < debug_buffer.size(); i++) {
+                hexdump += fmt::format("{:02X} ", debug_buffer[i]);
+                if (i != 0 && i % 16 == 0) {
+                    hexdump += '\n';
+                }
+            }
+            LOG_CRITICAL(Debug_Emulated, "debug_buffer=\n{}", hexdump);
+        }
+        has_dumped_buffer = true;
+    };
     switch (break_reason.break_type) {
     case BreakType::Panic:
         LOG_CRITICAL(Debug_Emulated, "Signalling debugger, PANIC! info1=0x{:016X}, info2=0x{:016X}",
                      info1, info2);
+        handle_debug_buffer(info1, info2);
         break;
     case BreakType::AssertionFailed:
         LOG_CRITICAL(Debug_Emulated,
                      "Signalling debugger, Assertion failed! info1=0x{:016X}, info2=0x{:016X}",
                      info1, info2);
+        handle_debug_buffer(info1, info2);
         break;
     case BreakType::PreNROLoad:
         LOG_WARNING(
@@ -433,6 +460,7 @@ static void Break(u32 reason, u64 info1, u64 info2) {
             Debug_Emulated,
             "Signalling debugger, Unknown break reason {}, info1=0x{:016X}, info2=0x{:016X}",
             static_cast<u32>(break_reason.break_type.Value()), info1, info2);
+        handle_debug_buffer(info1, info2);
         break;
     }
 
@@ -441,6 +469,7 @@ static void Break(u32 reason, u64 info1, u64 info2) {
             Debug_Emulated,
             "Emulated program broke execution! reason=0x{:016X}, info1=0x{:016X}, info2=0x{:016X}",
             reason, info1, info2);
+        handle_debug_buffer(info1, info2);
         ASSERT(false);
 
         Core::CurrentProcess()->PrepareForTermination();
@@ -467,6 +496,37 @@ static ResultCode GetInfo(u64* result, u64 info_id, u64 handle, u64 info_sub_id)
     LOG_TRACE(Kernel_SVC, "called info_id=0x{:X}, info_sub_id=0x{:X}, handle=0x{:08X}", info_id,
               info_sub_id, handle);
 
+    enum class GetInfoType : u64 {
+        // 1.0.0+
+        AllowedCpuIdBitmask = 0,
+        AllowedThreadPrioBitmask = 1,
+        MapRegionBaseAddr = 2,
+        MapRegionSize = 3,
+        HeapRegionBaseAddr = 4,
+        HeapRegionSize = 5,
+        TotalMemoryUsage = 6,
+        TotalHeapUsage = 7,
+        IsCurrentProcessBeingDebugged = 8,
+        ResourceHandleLimit = 9,
+        IdleTickCount = 10,
+        RandomEntropy = 11,
+        PerformanceCounter = 0xF0000002,
+        // 2.0.0+
+        ASLRRegionBaseAddr = 12,
+        ASLRRegionSize = 13,
+        NewMapRegionBaseAddr = 14,
+        NewMapRegionSize = 15,
+        // 3.0.0+
+        IsVirtualAddressMemoryEnabled = 16,
+        PersonalMmHeapUsage = 17,
+        TitleId = 18,
+        // 4.0.0+
+        PrivilegedProcessId = 19,
+        // 5.0.0+
+        UserExceptionContextAddr = 20,
+        ThreadTickCount = 0xF0000002,
+    };
+
     const auto* current_process = Core::CurrentProcess();
     const auto& vm_manager = current_process->VMManager();
 
@@ -499,7 +559,7 @@ static ResultCode GetInfo(u64* result, u64 info_id, u64 handle, u64 info_sub_id)
         *result = 0;
         break;
     case GetInfoType::RandomEntropy:
-        *result = 0;
+        *result = Settings::values.rng_seed.value_or(0);
         break;
     case GetInfoType::ASLRRegionBaseAddr:
         *result = vm_manager.GetASLRRegionBaseAddress();
@@ -529,6 +589,36 @@ static ResultCode GetInfo(u64* result, u64 info_id, u64 handle, u64 info_sub_id)
                     "(STUBBED) Attempted to query user exception context address, returned 0");
         *result = 0;
         break;
+    case GetInfoType::ThreadTickCount: {
+        constexpr u64 num_cpus = 4;
+        if (info_sub_id != 0xFFFFFFFFFFFFFFFF && info_sub_id >= num_cpus) {
+            return ERR_INVALID_COMBINATION_KERNEL;
+        }
+
+        const auto thread =
+            current_process->GetHandleTable().Get<Thread>(static_cast<Handle>(handle));
+        if (!thread) {
+            return ERR_INVALID_HANDLE;
+        }
+
+        const auto& system = Core::System::GetInstance();
+        const auto& scheduler = system.CurrentScheduler();
+        const auto* const current_thread = scheduler.GetCurrentThread();
+        const bool same_thread = current_thread == thread;
+
+        const u64 prev_ctx_ticks = scheduler.GetLastContextSwitchTicks();
+        u64 out_ticks = 0;
+        if (same_thread && info_sub_id == 0xFFFFFFFFFFFFFFFF) {
+            const u64 thread_ticks = current_thread->GetTotalCPUTimeTicks();
+
+            out_ticks = thread_ticks + (CoreTiming::GetTicks() - prev_ctx_ticks);
+        } else if (same_thread && info_sub_id == system.CurrentCoreIndex()) {
+            out_ticks = CoreTiming::GetTicks() - prev_ctx_ticks;
+        }
+
+        *result = out_ticks;
+        break;
+    }
     default:
         UNIMPLEMENTED();
     }

src/core/hle/kernel/svc.h

@@ -24,37 +24,6 @@ struct PageInfo {
     u64 flags;
 };
 
-/// Values accepted by svcGetInfo
-enum class GetInfoType : u64 {
-    // 1.0.0+
-    AllowedCpuIdBitmask = 0,
-    AllowedThreadPrioBitmask = 1,
-    MapRegionBaseAddr = 2,
-    MapRegionSize = 3,
-    HeapRegionBaseAddr = 4,
-    HeapRegionSize = 5,
-    TotalMemoryUsage = 6,
-    TotalHeapUsage = 7,
-    IsCurrentProcessBeingDebugged = 8,
-    ResourceHandleLimit = 9,
-    IdleTickCount = 10,
-    RandomEntropy = 11,
-    PerformanceCounter = 0xF0000002,
-    // 2.0.0+
-    ASLRRegionBaseAddr = 12,
-    ASLRRegionSize = 13,
-    NewMapRegionBaseAddr = 14,
-    NewMapRegionSize = 15,
-    // 3.0.0+
-    IsVirtualAddressMemoryEnabled = 16,
-    PersonalMmHeapUsage = 17,
-    TitleId = 18,
-    // 4.0.0+
-    PrivilegedProcessId = 19,
-    // 5.0.0+
-    UserExceptionContextAddr = 20,
-};
-
 void CallSVC(u32 immediate);
 
 } // namespace Kernel

src/core/hle/kernel/thread.cpp

@@ -4,9 +4,9 @@
 
 #include <algorithm>
 #include <cinttypes>
+#include <optional>
 #include <vector>
 
-#include <boost/optional.hpp>
 #include <boost/range/algorithm_ext/erase.hpp>
 
 #include "common/assert.h"
@@ -94,7 +94,7 @@ void Thread::CancelWakeupTimer() {
     CoreTiming::UnscheduleEventThreadsafe(kernel.ThreadWakeupCallbackEventType(), callback_handle);
 }
 
-static boost::optional<s32> GetNextProcessorId(u64 mask) {
+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()) {
@@ -142,7 +142,7 @@ void Thread::ResumeFromWait() {
 
     status = ThreadStatus::Ready;
 
-    boost::optional<s32> new_processor_id = GetNextProcessorId(affinity_mask);
+    std::optional<s32> new_processor_id = GetNextProcessorId(affinity_mask);
     if (!new_processor_id) {
         new_processor_id = processor_id;
     }
@@ -369,7 +369,7 @@ void Thread::ChangeCore(u32 core, u64 mask) {
         return;
     }
 
-    boost::optional<s32> new_processor_id{GetNextProcessorId(affinity_mask)};
+    std::optional<s32> new_processor_id{GetNextProcessorId(affinity_mask)};
 
     if (!new_processor_id) {
         new_processor_id = processor_id;

src/core/hle/kernel/thread.h

@@ -258,6 +258,14 @@ public:
         return last_running_ticks;
     }
 
+    u64 GetTotalCPUTimeTicks() const {
+        return total_cpu_time_ticks;
+    }
+
+    void UpdateCPUTimeTicks(u64 ticks) {
+        total_cpu_time_ticks += ticks;
+    }
+
     s32 GetProcessorID() const {
         return processor_id;
     }
@@ -378,7 +386,8 @@ private:
     u32 nominal_priority = 0; ///< Nominal thread priority, as set by the emulated application
     u32 current_priority = 0; ///< Current thread priority, can be temporarily changed
 
-    u64 last_running_ticks = 0; ///< CPU tick when thread was last running
+    u64 total_cpu_time_ticks = 0; ///< Total CPU running ticks.
+    u64 last_running_ticks = 0;   ///< CPU tick when thread was last running
 
     s32 processor_id = 0;
 

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

@@ -242,6 +242,28 @@ void Module::Interface::GetBaasAccountManagerForApplication(Kernel::HLERequestCo
     LOG_DEBUG(Service_ACC, "called");
 }
 
+void Module::Interface::TrySelectUserWithoutInteraction(Kernel::HLERequestContext& ctx) {
+    LOG_DEBUG(Service_ACC, "called");
+    // A u8 is passed into this function which we can safely ignore. It's to determine if we have
+    // access to use the network or not by the looks of it
+    IPC::ResponseBuilder rb{ctx, 6};
+    if (profile_manager->GetUserCount() != 1) {
+        rb.Push(RESULT_SUCCESS);
+        rb.PushRaw<u128>(INVALID_UUID);
+        return;
+    }
+    auto user_list = profile_manager->GetAllUsers();
+    if (user_list.empty()) {
+        rb.Push(ResultCode(-1)); // TODO(ogniK): Find the correct error code
+        rb.PushRaw<u128>(INVALID_UUID);
+        return;
+    }
+
+    // Select the first user we have
+    rb.Push(RESULT_SUCCESS);
+    rb.PushRaw<u128>(profile_manager->GetUser(0)->uuid);
+}
+
 Module::Interface::Interface(std::shared_ptr<Module> module,
                              std::shared_ptr<ProfileManager> profile_manager, const char* name)
     : ServiceFramework(name), module(std::move(module)),

src/core/hle/service/acc/acc.h

@@ -27,6 +27,7 @@ public:
         void InitializeApplicationInfo(Kernel::HLERequestContext& ctx);
         void GetBaasAccountManagerForApplication(Kernel::HLERequestContext& ctx);
         void IsUserRegistrationRequestPermitted(Kernel::HLERequestContext& ctx);
+        void TrySelectUserWithoutInteraction(Kernel::HLERequestContext& ctx);
 
     protected:
         std::shared_ptr<Module> module;

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

@@ -17,7 +17,7 @@ ACC_SU::ACC_SU(std::shared_ptr<Module> module, std::shared_ptr<ProfileManager> p
         {5, &ACC_SU::GetProfile, "GetProfile"},
         {6, nullptr, "GetProfileDigest"},
         {50, &ACC_SU::IsUserRegistrationRequestPermitted, "IsUserRegistrationRequestPermitted"},
-        {51, nullptr, "TrySelectUserWithoutInteraction"},
+        {51, &ACC_SU::TrySelectUserWithoutInteraction, "TrySelectUserWithoutInteraction"},
         {60, nullptr, "ListOpenContextStoredUsers"},
         {100, nullptr, "GetUserRegistrationNotifier"},
         {101, nullptr, "GetUserStateChangeNotifier"},

src/core/hle/service/acc/acc_u0.cpp

@@ -17,7 +17,7 @@ ACC_U0::ACC_U0(std::shared_ptr<Module> module, std::shared_ptr<ProfileManager> p
         {5, &ACC_U0::GetProfile, "GetProfile"},
         {6, nullptr, "GetProfileDigest"},
         {50, &ACC_U0::IsUserRegistrationRequestPermitted, "IsUserRegistrationRequestPermitted"},
-        {51, nullptr, "TrySelectUserWithoutInteraction"},
+        {51, &ACC_U0::TrySelectUserWithoutInteraction, "TrySelectUserWithoutInteraction"},
         {60, nullptr, "ListOpenContextStoredUsers"},
         {100, &ACC_U0::InitializeApplicationInfo, "InitializeApplicationInfo"},
         {101, &ACC_U0::GetBaasAccountManagerForApplication, "GetBaasAccountManagerForApplication"},

src/core/hle/service/acc/acc_u1.cpp

@@ -17,7 +17,7 @@ ACC_U1::ACC_U1(std::shared_ptr<Module> module, std::shared_ptr<ProfileManager> p
         {5, &ACC_U1::GetProfile, "GetProfile"},
         {6, nullptr, "GetProfileDigest"},
         {50, &ACC_U1::IsUserRegistrationRequestPermitted, "IsUserRegistrationRequestPermitted"},
-        {51, nullptr, "TrySelectUserWithoutInteraction"},
+        {51, &ACC_U1::TrySelectUserWithoutInteraction, "TrySelectUserWithoutInteraction"},
         {60, nullptr, "ListOpenContextStoredUsers"},
         {100, nullptr, "GetUserRegistrationNotifier"},
         {101, nullptr, "GetUserStateChangeNotifier"},

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

@@ -195,7 +195,7 @@ std::size_t ProfileManager::GetOpenUserCount() const {
 
 /// Checks if a user id exists in our profile manager
 bool ProfileManager::UserExists(UUID uuid) const {
-    return GetUserIndex(uuid) != std::nullopt;
+    return GetUserIndex(uuid).has_value();
 }
 
 bool ProfileManager::UserExistsIndex(std::size_t index) const {

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

@@ -57,7 +57,8 @@ struct UUID {
 };
 static_assert(sizeof(UUID) == 16, "UUID is an invalid size!");
 
-using ProfileUsername = std::array<u8, 0x20>;
+constexpr std::size_t profile_username_size = 32;
+using ProfileUsername = std::array<u8, profile_username_size>;
 using ProfileData = std::array<u8, MAX_DATA>;
 using UserIDArray = std::array<UUID, MAX_USERS>;
 

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

@@ -338,7 +338,54 @@ void ISelfController::GetIdleTimeDetectionExtension(Kernel::HLERequestContext& c
     LOG_WARNING(Service_AM, "(STUBBED) called");
 }
 
-ICommonStateGetter::ICommonStateGetter() : ServiceFramework("ICommonStateGetter") {
+AppletMessageQueue::AppletMessageQueue() {
+    auto& kernel = Core::System::GetInstance().Kernel();
+    on_new_message = Kernel::Event::Create(kernel, Kernel::ResetType::Sticky,
+                                           "AMMessageQueue:OnMessageRecieved");
+    on_operation_mode_changed = Kernel::Event::Create(kernel, Kernel::ResetType::OneShot,
+                                                      "AMMessageQueue:OperationModeChanged");
+}
+
+AppletMessageQueue::~AppletMessageQueue() = default;
+
+const Kernel::SharedPtr<Kernel::Event>& AppletMessageQueue::GetMesssageRecieveEvent() const {
+    return on_new_message;
+}
+
+const Kernel::SharedPtr<Kernel::Event>& AppletMessageQueue::GetOperationModeChangedEvent() const {
+    return on_operation_mode_changed;
+}
+
+void AppletMessageQueue::PushMessage(AppletMessage msg) {
+    messages.push(msg);
+    on_new_message->Signal();
+}
+
+AppletMessageQueue::AppletMessage AppletMessageQueue::PopMessage() {
+    if (messages.empty()) {
+        on_new_message->Clear();
+        return AppletMessage::NoMessage;
+    }
+    auto msg = messages.front();
+    messages.pop();
+    if (messages.empty()) {
+        on_new_message->Clear();
+    }
+    return msg;
+}
+
+std::size_t AppletMessageQueue::GetMessageCount() const {
+    return messages.size();
+}
+
+void AppletMessageQueue::OperationModeChanged() {
+    PushMessage(AppletMessage::OperationModeChanged);
+    PushMessage(AppletMessage::PerformanceModeChanged);
+    on_operation_mode_changed->Signal();
+}
+
+ICommonStateGetter::ICommonStateGetter(std::shared_ptr<AppletMessageQueue> msg_queue)
+    : ServiceFramework("ICommonStateGetter"), msg_queue(std::move(msg_queue)) {
     // clang-format off
     static const FunctionInfo functions[] = {
         {0, &ICommonStateGetter::GetEventHandle, "GetEventHandle"},
@@ -388,21 +435,19 @@ void ICommonStateGetter::GetBootMode(Kernel::HLERequestContext& ctx) {
 }
 
 void ICommonStateGetter::GetEventHandle(Kernel::HLERequestContext& ctx) {
-    event->Signal();
-
     IPC::ResponseBuilder rb{ctx, 2, 1};
     rb.Push(RESULT_SUCCESS);
-    rb.PushCopyObjects(event);
+    rb.PushCopyObjects(msg_queue->GetMesssageRecieveEvent());
 
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    LOG_DEBUG(Service_AM, "called");
 }
 
 void ICommonStateGetter::ReceiveMessage(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 3};
     rb.Push(RESULT_SUCCESS);
-    rb.Push<u32>(15);
+    rb.PushEnum<AppletMessageQueue::AppletMessage>(msg_queue->PopMessage());
 
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    LOG_DEBUG(Service_AM, "called");
 }
 
 void ICommonStateGetter::GetCurrentFocusState(Kernel::HLERequestContext& ctx) {
@@ -414,13 +459,11 @@ void ICommonStateGetter::GetCurrentFocusState(Kernel::HLERequestContext& ctx) {
 }
 
 void ICommonStateGetter::GetDefaultDisplayResolutionChangeEvent(Kernel::HLERequestContext& ctx) {
-    event->Signal();
-
     IPC::ResponseBuilder rb{ctx, 2, 1};
     rb.Push(RESULT_SUCCESS);
-    rb.PushCopyObjects(event);
+    rb.PushCopyObjects(msg_queue->GetOperationModeChangedEvent());
 
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    LOG_DEBUG(Service_AM, "called");
 }
 
 void ICommonStateGetter::GetDefaultDisplayResolution(Kernel::HLERequestContext& ctx) {
@@ -444,7 +487,7 @@ void ICommonStateGetter::GetOperationMode(Kernel::HLERequestContext& ctx) {
     rb.Push(RESULT_SUCCESS);
     rb.Push(static_cast<u8>(use_docked_mode ? OperationMode::Docked : OperationMode::Handheld));
 
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    LOG_DEBUG(Service_AM, "called");
 }
 
 void ICommonStateGetter::GetPerformanceMode(Kernel::HLERequestContext& ctx) {
@@ -454,7 +497,7 @@ void ICommonStateGetter::GetPerformanceMode(Kernel::HLERequestContext& ctx) {
     rb.Push(static_cast<u32>(use_docked_mode ? APM::PerformanceMode::Docked
                                              : APM::PerformanceMode::Handheld));
 
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    LOG_DEBUG(Service_AM, "called");
 }
 
 class IStorageAccessor final : public ServiceFramework<IStorageAccessor> {
@@ -743,7 +786,7 @@ void IApplicationFunctions::PopLaunchParameter(Kernel::HLERequestContext& ctx) {
 
     Account::ProfileManager profile_manager{};
     const auto uuid = profile_manager.GetUser(Settings::values.current_user);
-    ASSERT(uuid != std::nullopt);
+    ASSERT(uuid);
     params.current_user = uuid->uuid;
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
@@ -840,8 +883,12 @@ void IApplicationFunctions::GetPseudoDeviceId(Kernel::HLERequestContext& ctx) {
 
 void InstallInterfaces(SM::ServiceManager& service_manager,
                        std::shared_ptr<NVFlinger::NVFlinger> nvflinger) {
-    std::make_shared<AppletAE>(nvflinger)->InstallAsService(service_manager);
-    std::make_shared<AppletOE>(nvflinger)->InstallAsService(service_manager);
+    auto message_queue = std::make_shared<AppletMessageQueue>();
+    message_queue->PushMessage(
+        AppletMessageQueue::AppletMessage::FocusStateChanged); // Needed on game boot
+
+    std::make_shared<AppletAE>(nvflinger, message_queue)->InstallAsService(service_manager);
+    std::make_shared<AppletOE>(nvflinger, message_queue)->InstallAsService(service_manager);
     std::make_shared<IdleSys>()->InstallAsService(service_manager);
     std::make_shared<OMM>()->InstallAsService(service_manager);
     std::make_shared<SPSM>()->InstallAsService(service_manager);

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

@@ -5,6 +5,7 @@
 #pragma once
 
 #include <memory>
+#include <queue>
 #include "core/hle/service/service.h"
 
 namespace Kernel {
@@ -39,6 +40,31 @@ enum SystemLanguage {
     TraditionalChinese = 16,
 };
 
+class AppletMessageQueue {
+public:
+    enum class AppletMessage : u32 {
+        NoMessage = 0,
+        FocusStateChanged = 15,
+        OperationModeChanged = 30,
+        PerformanceModeChanged = 31,
+    };
+
+    AppletMessageQueue();
+    ~AppletMessageQueue();
+
+    const Kernel::SharedPtr<Kernel::Event>& GetMesssageRecieveEvent() const;
+    const Kernel::SharedPtr<Kernel::Event>& GetOperationModeChangedEvent() const;
+    void PushMessage(AppletMessage msg);
+    AppletMessage PopMessage();
+    std::size_t GetMessageCount() const;
+    void OperationModeChanged();
+
+private:
+    std::queue<AppletMessage> messages;
+    Kernel::SharedPtr<Kernel::Event> on_new_message;
+    Kernel::SharedPtr<Kernel::Event> on_operation_mode_changed;
+};
+
 class IWindowController final : public ServiceFramework<IWindowController> {
 public:
     IWindowController();
@@ -102,7 +128,7 @@ private:
 
 class ICommonStateGetter final : public ServiceFramework<ICommonStateGetter> {
 public:
-    ICommonStateGetter();
+    explicit ICommonStateGetter(std::shared_ptr<AppletMessageQueue> msg_queue);
     ~ICommonStateGetter() override;
 
 private:
@@ -126,6 +152,7 @@ private:
     void GetDefaultDisplayResolution(Kernel::HLERequestContext& ctx);
 
     Kernel::SharedPtr<Kernel::Event> event;
+    std::shared_ptr<AppletMessageQueue> msg_queue;
 };
 
 class ILibraryAppletCreator final : public ServiceFramework<ILibraryAppletCreator> {

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

@@ -12,8 +12,10 @@ namespace Service::AM {
 
 class ILibraryAppletProxy final : public ServiceFramework<ILibraryAppletProxy> {
 public:
-    explicit ILibraryAppletProxy(std::shared_ptr<NVFlinger::NVFlinger> nvflinger)
-        : ServiceFramework("ILibraryAppletProxy"), nvflinger(std::move(nvflinger)) {
+    explicit ILibraryAppletProxy(std::shared_ptr<NVFlinger::NVFlinger> nvflinger,
+                                 std::shared_ptr<AppletMessageQueue> msg_queue)
+        : ServiceFramework("ILibraryAppletProxy"), nvflinger(std::move(nvflinger)),
+          msg_queue(std::move(msg_queue)) {
         static const FunctionInfo functions[] = {
             {0, &ILibraryAppletProxy::GetCommonStateGetter, "GetCommonStateGetter"},
             {1, &ILibraryAppletProxy::GetSelfController, "GetSelfController"},
@@ -32,7 +34,7 @@ private:
     void GetCommonStateGetter(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
-        rb.PushIpcInterface<ICommonStateGetter>();
+        rb.PushIpcInterface<ICommonStateGetter>(msg_queue);
         LOG_DEBUG(Service_AM, "called");
     }
 
@@ -93,12 +95,15 @@ private:
     }
 
     std::shared_ptr<NVFlinger::NVFlinger> nvflinger;
+    std::shared_ptr<AppletMessageQueue> msg_queue;
 };
 
 class ISystemAppletProxy final : public ServiceFramework<ISystemAppletProxy> {
 public:
-    explicit ISystemAppletProxy(std::shared_ptr<NVFlinger::NVFlinger> nvflinger)
-        : ServiceFramework("ISystemAppletProxy"), nvflinger(std::move(nvflinger)) {
+    explicit ISystemAppletProxy(std::shared_ptr<NVFlinger::NVFlinger> nvflinger,
+                                std::shared_ptr<AppletMessageQueue> msg_queue)
+        : ServiceFramework("ISystemAppletProxy"), nvflinger(std::move(nvflinger)),
+          msg_queue(std::move(msg_queue)) {
         static const FunctionInfo functions[] = {
             {0, &ISystemAppletProxy::GetCommonStateGetter, "GetCommonStateGetter"},
             {1, &ISystemAppletProxy::GetSelfController, "GetSelfController"},
@@ -119,7 +124,7 @@ private:
     void GetCommonStateGetter(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
-        rb.PushIpcInterface<ICommonStateGetter>();
+        rb.PushIpcInterface<ICommonStateGetter>(msg_queue);
         LOG_DEBUG(Service_AM, "called");
     }
 
@@ -186,31 +191,34 @@ private:
         LOG_DEBUG(Service_AM, "called");
     }
     std::shared_ptr<NVFlinger::NVFlinger> nvflinger;
+    std::shared_ptr<AppletMessageQueue> msg_queue;
 };
 
 void AppletAE::OpenSystemAppletProxy(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
-    rb.PushIpcInterface<ISystemAppletProxy>(nvflinger);
+    rb.PushIpcInterface<ISystemAppletProxy>(nvflinger, msg_queue);
     LOG_DEBUG(Service_AM, "called");
 }
 
 void AppletAE::OpenLibraryAppletProxy(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
-    rb.PushIpcInterface<ILibraryAppletProxy>(nvflinger);
+    rb.PushIpcInterface<ILibraryAppletProxy>(nvflinger, msg_queue);
     LOG_DEBUG(Service_AM, "called");
 }
 
 void AppletAE::OpenLibraryAppletProxyOld(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
-    rb.PushIpcInterface<ILibraryAppletProxy>(nvflinger);
+    rb.PushIpcInterface<ILibraryAppletProxy>(nvflinger, msg_queue);
     LOG_DEBUG(Service_AM, "called");
 }
 
-AppletAE::AppletAE(std::shared_ptr<NVFlinger::NVFlinger> nvflinger)
-    : ServiceFramework("appletAE"), nvflinger(std::move(nvflinger)) {
+AppletAE::AppletAE(std::shared_ptr<NVFlinger::NVFlinger> nvflinger,
+                   std::shared_ptr<AppletMessageQueue> msg_queue)
+    : ServiceFramework("appletAE"), nvflinger(std::move(nvflinger)),
+      msg_queue(std::move(msg_queue)) {
     // clang-format off
     static const FunctionInfo functions[] = {
         {100, &AppletAE::OpenSystemAppletProxy, "OpenSystemAppletProxy"},
@@ -228,4 +236,8 @@ AppletAE::AppletAE(std::shared_ptr<NVFlinger::NVFlinger> nvflinger)
 
 AppletAE::~AppletAE() = default;
 
+const std::shared_ptr<AppletMessageQueue>& AppletAE::GetMessageQueue() const {
+    return msg_queue;
+}
+
 } // namespace Service::AM

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

@@ -17,15 +17,19 @@ namespace AM {
 
 class AppletAE final : public ServiceFramework<AppletAE> {
 public:
-    explicit AppletAE(std::shared_ptr<NVFlinger::NVFlinger> nvflinger);
+    explicit AppletAE(std::shared_ptr<NVFlinger::NVFlinger> nvflinger,
+                      std::shared_ptr<AppletMessageQueue> msg_queue);
     ~AppletAE() override;
 
+    const std::shared_ptr<AppletMessageQueue>& GetMessageQueue() const;
+
 private:
     void OpenSystemAppletProxy(Kernel::HLERequestContext& ctx);
     void OpenLibraryAppletProxy(Kernel::HLERequestContext& ctx);
     void OpenLibraryAppletProxyOld(Kernel::HLERequestContext& ctx);
 
     std::shared_ptr<NVFlinger::NVFlinger> nvflinger;
+    std::shared_ptr<AppletMessageQueue> msg_queue;
 };
 
 } // namespace AM

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

@@ -12,8 +12,10 @@ namespace Service::AM {
 
 class IApplicationProxy final : public ServiceFramework<IApplicationProxy> {
 public:
-    explicit IApplicationProxy(std::shared_ptr<NVFlinger::NVFlinger> nvflinger)
-        : ServiceFramework("IApplicationProxy"), nvflinger(std::move(nvflinger)) {
+    explicit IApplicationProxy(std::shared_ptr<NVFlinger::NVFlinger> nvflinger,
+                               std::shared_ptr<AppletMessageQueue> msg_queue)
+        : ServiceFramework("IApplicationProxy"), nvflinger(std::move(nvflinger)),
+          msg_queue(std::move(msg_queue)) {
         // clang-format off
         static const FunctionInfo functions[] = {
             {0, &IApplicationProxy::GetCommonStateGetter, "GetCommonStateGetter"},
@@ -70,7 +72,7 @@ private:
     void GetCommonStateGetter(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
-        rb.PushIpcInterface<ICommonStateGetter>();
+        rb.PushIpcInterface<ICommonStateGetter>(msg_queue);
         LOG_DEBUG(Service_AM, "called");
     }
 
@@ -89,17 +91,20 @@ private:
     }
 
     std::shared_ptr<NVFlinger::NVFlinger> nvflinger;
+    std::shared_ptr<AppletMessageQueue> msg_queue;
 };
 
 void AppletOE::OpenApplicationProxy(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
-    rb.PushIpcInterface<IApplicationProxy>(nvflinger);
+    rb.PushIpcInterface<IApplicationProxy>(nvflinger, msg_queue);
     LOG_DEBUG(Service_AM, "called");
 }
 
-AppletOE::AppletOE(std::shared_ptr<NVFlinger::NVFlinger> nvflinger)
-    : ServiceFramework("appletOE"), nvflinger(std::move(nvflinger)) {
+AppletOE::AppletOE(std::shared_ptr<NVFlinger::NVFlinger> nvflinger,
+                   std::shared_ptr<AppletMessageQueue> msg_queue)
+    : ServiceFramework("appletOE"), nvflinger(std::move(nvflinger)),
+      msg_queue(std::move(msg_queue)) {
     static const FunctionInfo functions[] = {
         {0, &AppletOE::OpenApplicationProxy, "OpenApplicationProxy"},
     };
@@ -108,4 +113,8 @@ AppletOE::AppletOE(std::shared_ptr<NVFlinger::NVFlinger> nvflinger)
 
 AppletOE::~AppletOE() = default;
 
+const std::shared_ptr<AppletMessageQueue>& AppletOE::GetMessageQueue() const {
+    return msg_queue;
+}
+
 } // namespace Service::AM

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

@@ -17,13 +17,17 @@ namespace AM {
 
 class AppletOE final : public ServiceFramework<AppletOE> {
 public:
-    explicit AppletOE(std::shared_ptr<NVFlinger::NVFlinger> nvflinger);
+    explicit AppletOE(std::shared_ptr<NVFlinger::NVFlinger> nvflinger,
+                      std::shared_ptr<AppletMessageQueue> msg_queue);
     ~AppletOE() override;
 
+    const std::shared_ptr<AppletMessageQueue>& GetMessageQueue() const;
+
 private:
     void OpenApplicationProxy(Kernel::HLERequestContext& ctx);
 
     std::shared_ptr<NVFlinger::NVFlinger> nvflinger;
+    std::shared_ptr<AppletMessageQueue> msg_queue;
 };
 
 } // namespace AM

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

@@ -161,7 +161,7 @@ void HwOpus::OpenOpusDecoder(Kernel::HLERequestContext& ctx) {
     ASSERT_MSG(channel_count == 1 || channel_count == 2, "Invalid channel count");
 
     std::size_t worker_sz = WorkerBufferSize(channel_count);
-    ASSERT_MSG(buffer_sz < worker_sz, "Worker buffer too large");
+    ASSERT_MSG(buffer_sz >= worker_sz, "Worker buffer too large");
     std::unique_ptr<OpusDecoder, OpusDeleter> decoder{
         static_cast<OpusDecoder*>(operator new(worker_sz))};
     if (opus_decoder_init(decoder.get(), sample_rate, channel_count)) {

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

@@ -273,8 +273,8 @@ public:
             {0, &IFileSystem::CreateFile, "CreateFile"},
             {1, &IFileSystem::DeleteFile, "DeleteFile"},
             {2, &IFileSystem::CreateDirectory, "CreateDirectory"},
-            {3, nullptr, "DeleteDirectory"},
-            {4, nullptr, "DeleteDirectoryRecursively"},
+            {3, &IFileSystem::DeleteDirectory, "DeleteDirectory"},
+            {4, &IFileSystem::DeleteDirectoryRecursively, "DeleteDirectoryRecursively"},
             {5, &IFileSystem::RenameFile, "RenameFile"},
             {6, nullptr, "RenameDirectory"},
             {7, &IFileSystem::GetEntryType, "GetEntryType"},
@@ -329,6 +329,30 @@ public:
         rb.Push(backend.CreateDirectory(name));
     }
 
+    void DeleteDirectory(Kernel::HLERequestContext& ctx) {
+        const IPC::RequestParser rp{ctx};
+
+        const auto file_buffer = ctx.ReadBuffer();
+        std::string name = Common::StringFromBuffer(file_buffer);
+
+        LOG_DEBUG(Service_FS, "called directory {}", name);
+
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(backend.DeleteDirectory(name));
+    }
+
+    void DeleteDirectoryRecursively(Kernel::HLERequestContext& ctx) {
+        const IPC::RequestParser rp{ctx};
+
+        const auto file_buffer = ctx.ReadBuffer();
+        std::string name = Common::StringFromBuffer(file_buffer);
+
+        LOG_DEBUG(Service_FS, "called directory {}", name);
+
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(backend.DeleteDirectoryRecursively(name));
+    }
+
     void RenameFile(Kernel::HLERequestContext& ctx) {
         IPC::RequestParser rp{ctx};
 

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

@@ -392,8 +392,10 @@ std::size_t Controller_NPad::GetSupportedNPadIdTypesSize() const {
 }
 
 void Controller_NPad::SetHoldType(NpadHoldType joy_hold_type) {
+    styleset_changed_event->Signal();
     hold_type = joy_hold_type;
 }
+
 Controller_NPad::NpadHoldType Controller_NPad::GetHoldType() const {
     return hold_type;
 }
@@ -427,6 +429,9 @@ void Controller_NPad::VibrateController(const std::vector<u32>& controller_ids,
 }
 
 Kernel::SharedPtr<Kernel::Event> Controller_NPad::GetStyleSetChangedEvent() const {
+    // TODO(ogniK): Figure out the best time to signal this event. This event seems that it should
+    // be signalled at least once, and signaled after a new controller is connected?
+    styleset_changed_event->Signal();
     return styleset_changed_event;
 }
 

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

@@ -96,6 +96,8 @@ public:
         // TODO(shinyquagsire23): Other update callbacks? (accel, gyro?)
 
         CoreTiming::ScheduleEvent(pad_update_ticks, pad_update_event);
+
+        ReloadInputDevices();
     }
 
     void ActivateController(HidController controller) {

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

@@ -212,7 +212,7 @@ private:
         IPC::ResponseBuilder rb{ctx, 2};
         auto amiibo = nfp_interface.GetAmiiboBuffer();
         TagInfo tag_info{};
-        std::memcpy(tag_info.uuid.data(), amiibo.uuid.data(), sizeof(tag_info.uuid.size()));
+        tag_info.uuid = amiibo.uuid;
         tag_info.uuid_length = static_cast<u8>(tag_info.uuid.size());
 
         tag_info.protocol = 1; // TODO(ogniK): Figure out actual values

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

@@ -31,7 +31,7 @@ void BufferQueue::SetPreallocatedBuffer(u32 slot, const IGBPBuffer& igbp_buffer)
     buffer_wait_event->Signal();
 }
 
-boost::optional<u32> BufferQueue::DequeueBuffer(u32 width, u32 height) {
+std::optional<u32> BufferQueue::DequeueBuffer(u32 width, u32 height) {
     auto itr = std::find_if(queue.begin(), queue.end(), [&](const Buffer& buffer) {
         // Only consider free buffers. Buffers become free once again after they've been Acquired
         // and Released by the compositor, see the NVFlinger::Compose method.
@@ -44,7 +44,7 @@ boost::optional<u32> BufferQueue::DequeueBuffer(u32 width, u32 height) {
     });
 
     if (itr == queue.end()) {
-        return boost::none;
+        return {};
     }
 
     itr->status = Buffer::Status::Dequeued;
@@ -70,12 +70,12 @@ void BufferQueue::QueueBuffer(u32 slot, BufferTransformFlags transform,
     itr->crop_rect = crop_rect;
 }
 
-boost::optional<const BufferQueue::Buffer&> BufferQueue::AcquireBuffer() {
+std::optional<std::reference_wrapper<const BufferQueue::Buffer>> BufferQueue::AcquireBuffer() {
     auto itr = std::find_if(queue.begin(), queue.end(), [](const Buffer& buffer) {
         return buffer.status == Buffer::Status::Queued;
     });
     if (itr == queue.end())
-        return boost::none;
+        return {};
     itr->status = Buffer::Status::Acquired;
     return *itr;
 }

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

@@ -4,8 +4,9 @@
 
 #pragma once
 
+#include <optional>
 #include <vector>
-#include <boost/optional.hpp>
+
 #include "common/common_funcs.h"
 #include "common/math_util.h"
 #include "common/swap.h"
@@ -57,9 +58,9 @@ public:
         /// Rotate source image 90 degrees clockwise
         Rotate90 = 0x04,
         /// Rotate source image 180 degrees
-        Roate180 = 0x03,
+        Rotate180 = 0x03,
         /// Rotate source image 270 degrees clockwise
-        Roate270 = 0x07,
+        Rotate270 = 0x07,
     };
 
     struct Buffer {
@@ -73,11 +74,11 @@ public:
     };
 
     void SetPreallocatedBuffer(u32 slot, const IGBPBuffer& igbp_buffer);
-    boost::optional<u32> DequeueBuffer(u32 width, u32 height);
+    std::optional<u32> DequeueBuffer(u32 width, u32 height);
     const IGBPBuffer& RequestBuffer(u32 slot) const;
     void QueueBuffer(u32 slot, BufferTransformFlags transform,
                      const MathUtil::Rectangle<int>& crop_rect);
-    boost::optional<const Buffer&> AcquireBuffer();
+    std::optional<std::reference_wrapper<const Buffer>> AcquireBuffer();
     void ReleaseBuffer(u32 slot);
     u32 Query(QueryType type);
 

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

@@ -3,7 +3,7 @@
 // Refer to the license.txt file included.
 
 #include <algorithm>
-#include <boost/optional.hpp>
+#include <optional>
 
 #include "common/alignment.h"
 #include "common/assert.h"
@@ -134,7 +134,7 @@ void NVFlinger::Compose() {
 
         MicroProfileFlip();
 
-        if (buffer == boost::none) {
+        if (!buffer) {
             auto& system_instance = Core::System::GetInstance();
 
             // There was no queued buffer to draw, render previous frame
@@ -143,7 +143,7 @@ void NVFlinger::Compose() {
             continue;
         }
 
-        auto& igbp_buffer = buffer->igbp_buffer;
+        auto& igbp_buffer = buffer->get().igbp_buffer;
 
         // Now send the buffer to the GPU for drawing.
         // TODO(Subv): Support more than just disp0. The display device selection is probably based
@@ -152,10 +152,10 @@ void NVFlinger::Compose() {
         ASSERT(nvdisp);
 
         nvdisp->flip(igbp_buffer.gpu_buffer_id, igbp_buffer.offset, igbp_buffer.format,
-                     igbp_buffer.width, igbp_buffer.height, igbp_buffer.stride, buffer->transform,
-                     buffer->crop_rect);
+                     igbp_buffer.width, igbp_buffer.height, igbp_buffer.stride,
+                     buffer->get().transform, buffer->get().crop_rect);
 
-        buffer_queue->ReleaseBuffer(buffer->slot);
+        buffer_queue->ReleaseBuffer(buffer->get().slot);
     }
 }
 

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

@@ -3,18 +3,23 @@
 // Refer to the license.txt file included.
 
 #include <algorithm>
+#include <chrono>
 #include <cstdlib>
+#include <ctime>
+#include <functional>
 #include <vector>
 #include "common/logging/log.h"
 #include "core/hle/ipc_helpers.h"
 #include "core/hle/service/spl/csrng.h"
 #include "core/hle/service/spl/module.h"
 #include "core/hle/service/spl/spl.h"
+#include "core/settings.h"
 
 namespace Service::SPL {
 
 Module::Interface::Interface(std::shared_ptr<Module> module, const char* name)
-    : ServiceFramework(name), module(std::move(module)) {}
+    : ServiceFramework(name), module(std::move(module)),
+      rng(Settings::values.rng_seed.value_or(std::time(nullptr))) {}
 
 Module::Interface::~Interface() = default;
 
@@ -24,7 +29,7 @@ void Module::Interface::GetRandomBytes(Kernel::HLERequestContext& ctx) {
     std::size_t size = ctx.GetWriteBufferSize();
 
     std::vector<u8> data(size);
-    std::generate(data.begin(), data.end(), std::rand);
+    std::generate(data.begin(), data.end(), rng);
 
     ctx.WriteBuffer(data);
 

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

@@ -4,6 +4,7 @@
 
 #pragma once
 
+#include <random>
 #include "core/hle/service/service.h"
 
 namespace Service::SPL {
@@ -19,6 +20,9 @@ public:
 
     protected:
         std::shared_ptr<Module> module;
+
+    private:
+        std::mt19937 rng;
     };
 };
 

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

@@ -21,7 +21,7 @@ Time::Time(std::shared_ptr<Module> time, const char* name)
         {102, nullptr, "GetStandardUserSystemClockInitialYear"},
         {200, nullptr, "IsStandardNetworkSystemClockAccuracySufficient"},
         {300, nullptr, "CalculateMonotonicSystemClockBaseTimePoint"},
-        {400, nullptr, "GetClockSnapshot"},
+        {400, &Time::GetClockSnapshot, "GetClockSnapshot"},
         {401, nullptr, "GetClockSnapshotFromSystemClockContext"},
         {500, nullptr, "CalculateStandardUserSystemClockDifferenceByUser"},
         {501, nullptr, "CalculateSpanBetween"},

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

@@ -15,6 +15,44 @@
 
 namespace Service::Time {
 
+static void PosixToCalendar(u64 posix_time, CalendarTime& calendar_time,
+                            CalendarAdditionalInfo& additional_info,
+                            [[maybe_unused]] const TimeZoneRule& /*rule*/) {
+    const std::time_t time(posix_time);
+    const std::tm* tm = std::localtime(&time);
+    if (tm == nullptr) {
+        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;
+
+    additional_info.day_of_week = tm->tm_wday;
+    additional_info.day_of_year = tm->tm_yday;
+    std::memcpy(additional_info.name.data(), "UTC", sizeof("UTC"));
+    additional_info.utc_offset = 0;
+}
+
+static u64 CalendarToPosix(const CalendarTime& calendar_time,
+                           [[maybe_unused]] const TimeZoneRule& /*rule*/) {
+    std::tm time{};
+    time.tm_year = calendar_time.year - 1900;
+    time.tm_mon = calendar_time.month - 1;
+    time.tm_mday = calendar_time.day;
+
+    time.tm_hour = calendar_time.hour;
+    time.tm_min = calendar_time.minute;
+    time.tm_sec = calendar_time.second;
+
+    std::time_t epoch_time = std::mktime(&time);
+    return static_cast<u64>(epoch_time);
+}
+
 class ISystemClock final : public ServiceFramework<ISystemClock> {
 public:
     ISystemClock() : ServiceFramework("ISystemClock") {
@@ -80,8 +118,8 @@ public:
             {5, nullptr, "GetTimeZoneRuleVersion"},
             {100, &ITimeZoneService::ToCalendarTime, "ToCalendarTime"},
             {101, &ITimeZoneService::ToCalendarTimeWithMyRule, "ToCalendarTimeWithMyRule"},
-            {201, nullptr, "ToPosixTime"},
-            {202, nullptr, "ToPosixTimeWithMyRule"},
+            {201, &ITimeZoneService::ToPosixTime, "ToPosixTime"},
+            {202, &ITimeZoneService::ToPosixTimeWithMyRule, "ToPosixTimeWithMyRule"},
         };
         RegisterHandlers(functions);
     }
@@ -151,24 +189,29 @@ private:
         rb.PushRaw(additional_info);
     }
 
-    void PosixToCalendar(u64 posix_time, CalendarTime& calendar_time,
-                         CalendarAdditionalInfo& additional_info, const TimeZoneRule& /*rule*/) {
-        std::time_t t(posix_time);
-        std::tm* tm = std::localtime(&t);
-        if (!tm) {
-            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;
+    void ToPosixTime(Kernel::HLERequestContext& ctx) {
+        // TODO(ogniK): Figure out how to handle multiple times
+        LOG_WARNING(Service_Time, "(STUBBED) called");
+        IPC::RequestParser rp{ctx};
+        auto calendar_time = rp.PopRaw<CalendarTime>();
+        auto posix_time = CalendarToPosix(calendar_time, {});
 
-        additional_info.day_of_week = tm->tm_wday;
-        additional_info.day_of_year = tm->tm_yday;
-        std::memcpy(additional_info.name.data(), "UTC", sizeof("UTC"));
-        additional_info.utc_offset = 0;
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushRaw<u32>(1); // Amount of times we're returning
+        ctx.WriteBuffer(&posix_time, sizeof(u64));
+    }
+
+    void ToPosixTimeWithMyRule(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_Time, "(STUBBED) called");
+        IPC::RequestParser rp{ctx};
+        auto calendar_time = rp.PopRaw<CalendarTime>();
+        auto posix_time = CalendarToPosix(calendar_time, {});
+
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushRaw<u32>(1); // Amount of times we're returning
+        ctx.WriteBuffer(&posix_time, sizeof(u64));
     }
 };
 
@@ -207,6 +250,55 @@ void Module::Interface::GetStandardLocalSystemClock(Kernel::HLERequestContext& c
     LOG_DEBUG(Service_Time, "called");
 }
 
+void Module::Interface::GetClockSnapshot(Kernel::HLERequestContext& ctx) {
+    LOG_DEBUG(Service_Time, "called");
+
+    IPC::RequestParser rp{ctx};
+    auto unknown_u8 = rp.PopRaw<u8>();
+
+    ClockSnapshot clock_snapshot{};
+
+    const s64 time_since_epoch{std::chrono::duration_cast<std::chrono::seconds>(
+                                   std::chrono::system_clock::now().time_since_epoch())
+                                   .count()};
+    CalendarTime calendar_time{};
+    const std::time_t time(time_since_epoch);
+    const std::tm* tm = std::localtime(&time);
+    if (tm == nullptr) {
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(ResultCode(-1)); // TODO(ogniK): Find appropriate error code
+        return;
+    }
+    SteadyClockTimePoint steady_clock_time_point{CoreTiming::cyclesToMs(CoreTiming::GetTicks()) /
+                                                 1000};
+
+    LocationName location_name{"UTC"};
+    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;
+    clock_snapshot.system_posix_time = time_since_epoch;
+    clock_snapshot.network_posix_time = time_since_epoch;
+    clock_snapshot.system_calendar_time = calendar_time;
+    clock_snapshot.network_calendar_time = calendar_time;
+
+    CalendarAdditionalInfo additional_info{};
+    PosixToCalendar(time_since_epoch, calendar_time, additional_info, {});
+
+    clock_snapshot.system_calendar_info = additional_info;
+    clock_snapshot.network_calendar_info = additional_info;
+
+    clock_snapshot.steady_clock_timepoint = steady_clock_time_point;
+    clock_snapshot.location_name = location_name;
+    clock_snapshot.clock_auto_adjustment_enabled = 1;
+    clock_snapshot.ipc_u8 = unknown_u8;
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(RESULT_SUCCESS);
+    ctx.WriteBuffer(&clock_snapshot, sizeof(ClockSnapshot));
+}
+
 Module::Interface::Interface(std::shared_ptr<Module> time, const char* name)
     : ServiceFramework(name), time(std::move(time)) {}
 

src/core/hle/service/time/time.h

@@ -5,6 +5,7 @@
 #pragma once
 
 #include <array>
+#include "common/common_funcs.h"
 #include "core/hle/service/service.h"
 
 namespace Service::Time {
@@ -53,6 +54,23 @@ struct SystemClockContext {
 static_assert(sizeof(SystemClockContext) == 0x20,
               "SystemClockContext structure has incorrect size");
 
+struct ClockSnapshot {
+    SystemClockContext user_clock_context;
+    SystemClockContext network_clock_context;
+    s64_le system_posix_time;
+    s64_le network_posix_time;
+    CalendarTime system_calendar_time;
+    CalendarTime network_calendar_time;
+    CalendarAdditionalInfo system_calendar_info;
+    CalendarAdditionalInfo network_calendar_info;
+    SteadyClockTimePoint steady_clock_timepoint;
+    LocationName location_name;
+    u8 clock_auto_adjustment_enabled;
+    u8 ipc_u8;
+    INSERT_PADDING_BYTES(2);
+};
+static_assert(sizeof(ClockSnapshot) == 0xd0, "ClockSnapshot is an invalid size");
+
 class Module final {
 public:
     class Interface : public ServiceFramework<Interface> {
@@ -65,6 +83,7 @@ public:
         void GetStandardSteadyClock(Kernel::HLERequestContext& ctx);
         void GetTimeZoneService(Kernel::HLERequestContext& ctx);
         void GetStandardLocalSystemClock(Kernel::HLERequestContext& ctx);
+        void GetClockSnapshot(Kernel::HLERequestContext& ctx);
 
     protected:
         std::shared_ptr<Module> time;

src/core/hle/service/usb/usb.cpp

@@ -132,11 +132,11 @@ public:
         // clang-format off
         static const FunctionInfo functions[] = {
             {0, nullptr, "BindNoticeEvent"},
-            {1, nullptr, "Unknown1"},
+            {1, nullptr, "UnbindNoticeEvent"},
             {2, nullptr, "GetStatus"},
             {3, nullptr, "GetNotice"},
-            {4, nullptr, "Unknown2"},
-            {5, nullptr, "Unknown3"},
+            {4, nullptr, "EnablePowerRequestNotice"},
+            {5, nullptr, "DisablePowerRequestNotice"},
             {6, nullptr, "ReplyPowerRequest"},
         };
         // clang-format on

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

@@ -6,9 +6,10 @@
 #include <array>
 #include <cstring>
 #include <memory>
+#include <optional>
 #include <type_traits>
 #include <utility>
-#include <boost/optional.hpp>
+
 #include "common/alignment.h"
 #include "common/assert.h"
 #include "common/common_funcs.h"
@@ -506,9 +507,9 @@ private:
             IGBPDequeueBufferRequestParcel request{ctx.ReadBuffer()};
             const u32 width{request.data.width};
             const u32 height{request.data.height};
-            boost::optional<u32> slot = buffer_queue->DequeueBuffer(width, height);
+            std::optional<u32> slot = buffer_queue->DequeueBuffer(width, height);
 
-            if (slot != boost::none) {
+            if (slot) {
                 // Buffer is available
                 IGBPDequeueBufferResponseParcel response{*slot};
                 ctx.WriteBuffer(response.Serialize());
@@ -520,7 +521,7 @@ private:
                         Kernel::ThreadWakeupReason reason) {
                         // Repeat TransactParcel DequeueBuffer when a buffer is available
                         auto buffer_queue = nv_flinger->GetBufferQueue(id);
-                        boost::optional<u32> slot = buffer_queue->DequeueBuffer(width, height);
+                        std::optional<u32> slot = buffer_queue->DequeueBuffer(width, height);
                         IGBPDequeueBufferResponseParcel response{*slot};
                         ctx.WriteBuffer(response.Serialize());
                         IPC::ResponseBuilder rb{ctx, 2};

src/core/loader/loader.h

@@ -6,10 +6,11 @@
 
 #include <iosfwd>
 #include <memory>
+#include <optional>
 #include <string>
 #include <utility>
 #include <vector>
-#include <boost/optional.hpp>
+
 #include "common/common_types.h"
 #include "core/file_sys/vfs.h"
 
@@ -145,7 +146,7 @@ public:
      * information.
      * @returns A pair with the optional system mode, and and the status.
      */
-    virtual std::pair<boost::optional<u32>, ResultStatus> LoadKernelSystemMode() {
+    virtual std::pair<std::optional<u32>, ResultStatus> LoadKernelSystemMode() {
         // 96MB allocated to the application.
         return std::make_pair(2, ResultStatus::Success);
     }

src/core/loader/nsp.cpp

@@ -36,6 +36,16 @@ AppLoader_NSP::AppLoader_NSP(FileSys::VirtualFile file)
 
     std::tie(nacp_file, icon_file) =
         FileSys::PatchManager(nsp->GetProgramTitleID()).ParseControlNCA(*control_nca);
+
+    if (nsp->IsExtractedType()) {
+        secondary_loader = std::make_unique<AppLoader_DeconstructedRomDirectory>(nsp->GetExeFS());
+    } else {
+        if (title_id == 0)
+            return;
+
+        secondary_loader = std::make_unique<AppLoader_NCA>(
+            nsp->GetNCAFile(title_id, FileSys::ContentRecordType::Program));
+    }
 }
 
 AppLoader_NSP::~AppLoader_NSP() = default;
@@ -67,26 +77,19 @@ ResultStatus AppLoader_NSP::Load(Kernel::Process& process) {
         return ResultStatus::ErrorAlreadyLoaded;
     }
 
-    if (nsp->IsExtractedType()) {
-        secondary_loader = std::make_unique<AppLoader_DeconstructedRomDirectory>(nsp->GetExeFS());
-    } else {
-        if (title_id == 0)
-            return ResultStatus::ErrorNSPMissingProgramNCA;
+    if (title_id == 0)
+        return ResultStatus::ErrorNSPMissingProgramNCA;
 
-        secondary_loader = std::make_unique<AppLoader_NCA>(
-            nsp->GetNCAFile(title_id, FileSys::ContentRecordType::Program));
+    if (nsp->GetStatus() != ResultStatus::Success)
+        return nsp->GetStatus();
 
-        if (nsp->GetStatus() != ResultStatus::Success)
-            return nsp->GetStatus();
+    if (nsp->GetProgramStatus(title_id) != ResultStatus::Success)
+        return nsp->GetProgramStatus(title_id);
 
-        if (nsp->GetProgramStatus(title_id) != ResultStatus::Success)
-            return nsp->GetProgramStatus(title_id);
-
-        if (nsp->GetNCA(title_id, FileSys::ContentRecordType::Program) == nullptr) {
-            if (!Core::Crypto::KeyManager::KeyFileExists(false))
-                return ResultStatus::ErrorMissingProductionKeyFile;
-            return ResultStatus::ErrorNSPMissingProgramNCA;
-        }
+    if (nsp->GetNCA(title_id, FileSys::ContentRecordType::Program) == nullptr) {
+        if (!Core::Crypto::KeyManager::KeyFileExists(false))
+            return ResultStatus::ErrorMissingProductionKeyFile;
+        return ResultStatus::ErrorNSPMissingProgramNCA;
     }
 
     const auto result = secondary_loader->Load(process);

src/core/memory.cpp

@@ -4,9 +4,9 @@
 
 #include <algorithm>
 #include <cstring>
+#include <optional>
 #include <utility>
 
-#include <boost/optional.hpp>
 #include "common/assert.h"
 #include "common/common_types.h"
 #include "common/logging/log.h"

src/core/memory_hook.h

@@ -5,7 +5,8 @@
 #pragma once
 
 #include <memory>
-#include <boost/optional.hpp>
+#include <optional>
+
 #include "common/common_types.h"
 
 namespace Memory {
@@ -18,19 +19,19 @@ namespace Memory {
  *
  * A hook may be mapped to multiple regions of memory.
  *
- * If a boost::none or false is returned from a function, the read/write request is passed through
+ * If a std::nullopt or false is returned from a function, the read/write request is passed through
  * to the underlying memory region.
  */
 class MemoryHook {
 public:
     virtual ~MemoryHook();
 
-    virtual boost::optional<bool> IsValidAddress(VAddr addr) = 0;
+    virtual std::optional<bool> IsValidAddress(VAddr addr) = 0;
 
-    virtual boost::optional<u8> Read8(VAddr addr) = 0;
-    virtual boost::optional<u16> Read16(VAddr addr) = 0;
-    virtual boost::optional<u32> Read32(VAddr addr) = 0;
-    virtual boost::optional<u64> Read64(VAddr addr) = 0;
+    virtual std::optional<u8> Read8(VAddr addr) = 0;
+    virtual std::optional<u16> Read16(VAddr addr) = 0;
+    virtual std::optional<u32> Read32(VAddr addr) = 0;
+    virtual std::optional<u64> Read64(VAddr addr) = 0;
 
     virtual bool ReadBlock(VAddr src_addr, void* dest_buffer, std::size_t size) = 0;
 

src/core/settings.h

@@ -6,6 +6,7 @@
 
 #include <array>
 #include <atomic>
+#include <optional>
 #include <string>
 #include "common/common_types.h"
 
@@ -114,8 +115,9 @@ struct Values {
     // System
     bool use_docked_mode;
     bool enable_nfc;
-    int current_user;
-    int language_index;
+    std::optional<u64> rng_seed;
+    s32 current_user;
+    s32 language_index;
 
     // Controls
     std::array<std::string, NativeButton::NumButtons> buttons;

src/core/telemetry_session.cpp

@@ -184,4 +184,13 @@ TelemetrySession::~TelemetrySession() {
     backend = nullptr;
 }
 
+bool TelemetrySession::SubmitTestcase() {
+#ifdef ENABLE_WEB_SERVICE
+    field_collection.Accept(*backend);
+    return backend->SubmitTestcase();
+#else
+    return false;
+#endif
+}
+
 } // namespace Core

src/core/telemetry_session.h

@@ -31,6 +31,12 @@ public:
         field_collection.AddField(type, name, std::move(value));
     }
 
+    /**
+     * Submits a Testcase.
+     * @returns A bool indicating whether the submission succeeded
+     */
+    bool SubmitTestcase();
+
 private:
     Telemetry::FieldCollection field_collection; ///< Tracks all added fields for the session
     std::unique_ptr<Telemetry::VisitorInterface> backend; ///< Backend interface that logs fields

src/tests/core/arm/arm_test_common.cpp

@@ -64,11 +64,11 @@ void TestEnvironment::ClearWriteRecords() {
 
 TestEnvironment::TestMemory::~TestMemory() {}
 
-boost::optional<bool> TestEnvironment::TestMemory::IsValidAddress(VAddr addr) {
+std::optional<bool> TestEnvironment::TestMemory::IsValidAddress(VAddr addr) {
     return true;
 }
 
-boost::optional<u8> TestEnvironment::TestMemory::Read8(VAddr addr) {
+std::optional<u8> TestEnvironment::TestMemory::Read8(VAddr addr) {
     const auto iter = data.find(addr);
 
     if (iter == data.end()) {
@@ -79,15 +79,15 @@ boost::optional<u8> TestEnvironment::TestMemory::Read8(VAddr addr) {
     return iter->second;
 }
 
-boost::optional<u16> TestEnvironment::TestMemory::Read16(VAddr addr) {
+std::optional<u16> TestEnvironment::TestMemory::Read16(VAddr addr) {
     return *Read8(addr) | static_cast<u16>(*Read8(addr + 1)) << 8;
 }
 
-boost::optional<u32> TestEnvironment::TestMemory::Read32(VAddr addr) {
+std::optional<u32> TestEnvironment::TestMemory::Read32(VAddr addr) {
     return *Read16(addr) | static_cast<u32>(*Read16(addr + 2)) << 16;
 }
 
-boost::optional<u64> TestEnvironment::TestMemory::Read64(VAddr addr) {
+std::optional<u64> TestEnvironment::TestMemory::Read64(VAddr addr) {
     return *Read32(addr) | static_cast<u64>(*Read32(addr + 4)) << 32;
 }
 

src/tests/core/arm/arm_test_common.h

@@ -64,12 +64,12 @@ private:
 
         ~TestMemory() override;
 
-        boost::optional<bool> IsValidAddress(VAddr addr) override;
+        std::optional<bool> IsValidAddress(VAddr addr) override;
 
-        boost::optional<u8> Read8(VAddr addr) override;
-        boost::optional<u16> Read16(VAddr addr) override;
-        boost::optional<u32> Read32(VAddr addr) override;
-        boost::optional<u64> Read64(VAddr addr) override;
+        std::optional<u8> Read8(VAddr addr) override;
+        std::optional<u16> Read16(VAddr addr) override;
+        std::optional<u32> Read32(VAddr addr) override;
+        std::optional<u64> Read64(VAddr addr) override;
 
         bool ReadBlock(VAddr src_addr, void* dest_buffer, std::size_t size) override;
 

src/video_core/CMakeLists.txt

@@ -21,6 +21,7 @@ add_library(video_core STATIC
     macro_interpreter.h
     memory_manager.cpp
     memory_manager.h
+    rasterizer_cache.cpp
     rasterizer_cache.h
     rasterizer_interface.h
     renderer_base.cpp
@@ -33,6 +34,7 @@ add_library(video_core STATIC
     renderer_opengl/gl_rasterizer.h
     renderer_opengl/gl_rasterizer_cache.cpp
     renderer_opengl/gl_rasterizer_cache.h
+    renderer_opengl/gl_resource_manager.cpp
     renderer_opengl/gl_resource_manager.h
     renderer_opengl/gl_shader_cache.cpp
     renderer_opengl/gl_shader_cache.h
@@ -51,6 +53,10 @@ add_library(video_core STATIC
     renderer_opengl/maxwell_to_gl.h
     renderer_opengl/renderer_opengl.cpp
     renderer_opengl/renderer_opengl.h
+    renderer_opengl/utils.cpp
+    renderer_opengl/utils.h
+    surface.cpp
+    surface.h
     textures/astc.cpp
     textures/astc.h
     textures/decoders.cpp

src/video_core/command_processor.cpp

@@ -81,7 +81,7 @@ void GPU::ProcessCommandLists(const std::vector<CommandListHeader>& commands) {
     for (auto entry : commands) {
         Tegra::GPUVAddr address = entry.Address();
         u32 size = entry.sz;
-        const boost::optional<VAddr> head_address = memory_manager->GpuToCpuAddress(address);
+        const std::optional<VAddr> head_address = memory_manager->GpuToCpuAddress(address);
         VAddr current_addr = *head_address;
         while (current_addr < *head_address + size * sizeof(CommandHeader)) {
             const CommandHeader header = {Memory::Read32(current_addr)};

src/video_core/engines/maxwell_3d.cpp

@@ -37,21 +37,52 @@ void Maxwell3D::InitializeRegisterDefaults() {
         regs.viewport[viewport].depth_range_near = 0.0f;
         regs.viewport[viewport].depth_range_far = 1.0f;
     }
+    // Doom and Bomberman seems to use the uninitialized registers and just enable blend
+    // so initialize blend registers with sane values
+    regs.blend.equation_rgb = Regs::Blend::Equation::Add;
+    regs.blend.factor_source_rgb = Regs::Blend::Factor::One;
+    regs.blend.factor_dest_rgb = Regs::Blend::Factor::Zero;
+    regs.blend.equation_a = Regs::Blend::Equation::Add;
+    regs.blend.factor_source_a = Regs::Blend::Factor::One;
+    regs.blend.factor_dest_a = Regs::Blend::Factor::Zero;
+    for (std::size_t blend_index = 0; blend_index < Regs::NumRenderTargets; blend_index++) {
+        regs.independent_blend[blend_index].equation_rgb = Regs::Blend::Equation::Add;
+        regs.independent_blend[blend_index].factor_source_rgb = Regs::Blend::Factor::One;
+        regs.independent_blend[blend_index].factor_dest_rgb = Regs::Blend::Factor::Zero;
+        regs.independent_blend[blend_index].equation_a = Regs::Blend::Equation::Add;
+        regs.independent_blend[blend_index].factor_source_a = Regs::Blend::Factor::One;
+        regs.independent_blend[blend_index].factor_dest_a = Regs::Blend::Factor::Zero;
+    }
+    regs.stencil_front_op_fail = Regs::StencilOp::Keep;
+    regs.stencil_front_op_zfail = Regs::StencilOp::Keep;
+    regs.stencil_front_op_zpass = Regs::StencilOp::Keep;
+    regs.stencil_front_func_func = Regs::ComparisonOp::Always;
+    regs.stencil_front_func_mask = 0xFFFFFFFF;
+    regs.stencil_front_mask = 0xFFFFFFFF;
+    regs.stencil_two_side_enable = 1;
+    regs.stencil_back_op_fail = Regs::StencilOp::Keep;
+    regs.stencil_back_op_zfail = Regs::StencilOp::Keep;
+    regs.stencil_back_op_zpass = Regs::StencilOp::Keep;
+    regs.stencil_back_func_func = Regs::ComparisonOp::Always;
+    regs.stencil_back_func_mask = 0xFFFFFFFF;
+    regs.stencil_back_mask = 0xFFFFFFFF;
 }
 
 void Maxwell3D::CallMacroMethod(u32 method, std::vector<u32> parameters) {
     // Reset the current macro.
     executing_macro = 0;
 
-    // The requested macro must have been uploaded already.
-    auto macro_code = uploaded_macros.find(method);
-    if (macro_code == uploaded_macros.end()) {
-        LOG_ERROR(HW_GPU, "Macro {:04X} was not uploaded", method);
+    // Lookup the macro offset
+    const u32 entry{(method - MacroRegistersStart) >> 1};
+    const auto& search{macro_offsets.find(entry)};
+    if (search == macro_offsets.end()) {
+        LOG_CRITICAL(HW_GPU, "macro not found for method 0x{:X}!", method);
+        UNREACHABLE();
         return;
     }
 
     // Execute the current macro.
-    macro_interpreter.Execute(macro_code->second, std::move(parameters));
+    macro_interpreter.Execute(search->second, std::move(parameters));
 }
 
 void Maxwell3D::WriteReg(u32 method, u32 value, u32 remaining_params) {
@@ -90,13 +121,25 @@ void Maxwell3D::WriteReg(u32 method, u32 value, u32 remaining_params) {
         debug_context->OnEvent(Tegra::DebugContext::Event::MaxwellCommandLoaded, nullptr);
     }
 
+    u32 old = regs.reg_array[method];
     regs.reg_array[method] = value;
 
+    if (value != old) {
+        if (method >= MAXWELL3D_REG_INDEX(vertex_attrib_format) &&
+            method < MAXWELL3D_REG_INDEX(vertex_attrib_format) + regs.vertex_attrib_format.size()) {
+            dirty_flags.vertex_attrib_format = true;
+        }
+    }
+
     switch (method) {
     case MAXWELL3D_REG_INDEX(macros.data): {
         ProcessMacroUpload(value);
         break;
     }
+    case MAXWELL3D_REG_INDEX(macros.bind): {
+        ProcessMacroBind(value);
+        break;
+    }
     case MAXWELL3D_REG_INDEX(const_buffer.cb_data[0]):
     case MAXWELL3D_REG_INDEX(const_buffer.cb_data[1]):
     case MAXWELL3D_REG_INDEX(const_buffer.cb_data[2]):
@@ -158,16 +201,20 @@ void Maxwell3D::WriteReg(u32 method, u32 value, u32 remaining_params) {
 }
 
 void Maxwell3D::ProcessMacroUpload(u32 data) {
-    // Store the uploaded macro code to interpret them when they're called.
-    auto& macro = uploaded_macros[regs.macros.entry * 2 + MacroRegistersStart];
-    macro.push_back(data);
+    ASSERT_MSG(regs.macros.upload_address < macro_memory.size(),
+               "upload_address exceeded macro_memory size!");
+    macro_memory[regs.macros.upload_address++] = data;
+}
+
+void Maxwell3D::ProcessMacroBind(u32 data) {
+    macro_offsets[regs.macros.entry] = data;
 }
 
 void Maxwell3D::ProcessQueryGet() {
     GPUVAddr sequence_address = regs.query.QueryAddress();
     // Since the sequence address is given as a GPU VAddr, we have to convert it to an application
     // VAddr before writing.
-    boost::optional<VAddr> address = memory_manager.GpuToCpuAddress(sequence_address);
+    std::optional<VAddr> address = memory_manager.GpuToCpuAddress(sequence_address);
 
     // TODO(Subv): Support the other query units.
     ASSERT_MSG(regs.query.query_get.unit == Regs::QueryUnit::Crop,
@@ -285,7 +332,7 @@ void Maxwell3D::ProcessCBData(u32 value) {
     // Don't allow writing past the end of the buffer.
     ASSERT(regs.const_buffer.cb_pos + sizeof(u32) <= regs.const_buffer.cb_size);
 
-    boost::optional<VAddr> address =
+    std::optional<VAddr> address =
         memory_manager.GpuToCpuAddress(buffer_address + regs.const_buffer.cb_pos);
 
     Memory::Write32(*address, value);
@@ -298,7 +345,7 @@ Texture::TICEntry Maxwell3D::GetTICEntry(u32 tic_index) const {
     GPUVAddr tic_base_address = regs.tic.TICAddress();
 
     GPUVAddr tic_address_gpu = tic_base_address + tic_index * sizeof(Texture::TICEntry);
-    boost::optional<VAddr> tic_address_cpu = memory_manager.GpuToCpuAddress(tic_address_gpu);
+    std::optional<VAddr> tic_address_cpu = memory_manager.GpuToCpuAddress(tic_address_gpu);
 
     Texture::TICEntry tic_entry;
     Memory::ReadBlock(*tic_address_cpu, &tic_entry, sizeof(Texture::TICEntry));
@@ -322,7 +369,7 @@ Texture::TSCEntry Maxwell3D::GetTSCEntry(u32 tsc_index) const {
     GPUVAddr tsc_base_address = regs.tsc.TSCAddress();
 
     GPUVAddr tsc_address_gpu = tsc_base_address + tsc_index * sizeof(Texture::TSCEntry);
-    boost::optional<VAddr> tsc_address_cpu = memory_manager.GpuToCpuAddress(tsc_address_gpu);
+    std::optional<VAddr> tsc_address_cpu = memory_manager.GpuToCpuAddress(tsc_address_gpu);
 
     Texture::TSCEntry tsc_entry;
     Memory::ReadBlock(*tsc_address_cpu, &tsc_entry, sizeof(Texture::TSCEntry));
@@ -386,7 +433,7 @@ Texture::FullTextureInfo Maxwell3D::GetStageTexture(Regs::ShaderStage stage,
 
     ASSERT(tex_info_address < tex_info_buffer.address + tex_info_buffer.size);
 
-    boost::optional<VAddr> tex_address_cpu = memory_manager.GpuToCpuAddress(tex_info_address);
+    std::optional<VAddr> tex_address_cpu = memory_manager.GpuToCpuAddress(tex_info_address);
     Texture::TextureHandle tex_handle{Memory::Read32(*tex_address_cpu)};
 
     Texture::FullTextureInfo tex_info{};

src/video_core/engines/maxwell_3d.h

@@ -345,6 +345,14 @@ public:
             Invert = 6,
             IncrWrap = 7,
             DecrWrap = 8,
+            KeepOGL = 0x1E00,
+            ZeroOGL = 0,
+            ReplaceOGL = 0x1E01,
+            IncrOGL = 0x1E02,
+            DecrOGL = 0x1E03,
+            InvertOGL = 0x150A,
+            IncrWrapOGL = 0x8507,
+            DecrWrapOGL = 0x8508,
         };
 
         enum class MemoryLayout : u32 {
@@ -462,6 +470,16 @@ public:
             }
         };
 
+        struct ColorMask {
+            union {
+                u32 raw;
+                BitField<0, 4, u32> R;
+                BitField<4, 4, u32> G;
+                BitField<8, 4, u32> B;
+                BitField<12, 4, u32> A;
+            };
+        };
+
         bool IsShaderConfigEnabled(std::size_t index) const {
             // The VertexB is always enabled.
             if (index == static_cast<std::size_t>(Regs::ShaderProgram::VertexB)) {
@@ -475,12 +493,13 @@ public:
                 INSERT_PADDING_WORDS(0x45);
 
                 struct {
-                    INSERT_PADDING_WORDS(1);
+                    u32 upload_address;
                     u32 data;
                     u32 entry;
+                    u32 bind;
                 } macros;
 
-                INSERT_PADDING_WORDS(0x189);
+                INSERT_PADDING_WORDS(0x188);
 
                 u32 tfb_enabled;
 
@@ -570,7 +589,11 @@ public:
                 u32 stencil_back_mask;
                 u32 stencil_back_func_mask;
 
-                INSERT_PADDING_WORDS(0x13);
+                INSERT_PADDING_WORDS(0xC);
+
+                u32 color_mask_common;
+
+                INSERT_PADDING_WORDS(0x6);
 
                 u32 rt_separate_frag_data;
 
@@ -645,8 +668,14 @@ public:
                 ComparisonOp depth_test_func;
                 float alpha_test_ref;
                 ComparisonOp alpha_test_func;
-
-                INSERT_PADDING_WORDS(0x9);
+                u32 draw_tfb_stride;
+                struct {
+                    float r;
+                    float g;
+                    float b;
+                    float a;
+                } blend_color;
+                INSERT_PADDING_WORDS(0x4);
 
                 struct {
                     u32 separate_alpha;
@@ -723,7 +752,11 @@ public:
                 StencilOp stencil_back_op_zpass;
                 ComparisonOp stencil_back_func_func;
 
-                INSERT_PADDING_WORDS(0x17);
+                INSERT_PADDING_WORDS(0x4);
+
+                u32 framebuffer_srgb;
+
+                INSERT_PADDING_WORDS(0x12);
 
                 union {
                     BitField<2, 1, u32> coord_origin;
@@ -751,7 +784,14 @@ public:
                     };
                 } draw;
 
-                INSERT_PADDING_WORDS(0x6B);
+                INSERT_PADDING_WORDS(0xA);
+
+                struct {
+                    u32 enabled;
+                    u32 index;
+                } primitive_restart;
+
+                INSERT_PADDING_WORDS(0x5F);
 
                 struct {
                     u32 start_addr_high;
@@ -829,8 +869,9 @@ public:
                     BitField<6, 4, u32> RT;
                     BitField<10, 11, u32> layer;
                 } clear_buffers;
-
-                INSERT_PADDING_WORDS(0x4B);
+                INSERT_PADDING_WORDS(0xB);
+                std::array<ColorMask, NumRenderTargets> color_mask;
+                INSERT_PADDING_WORDS(0x38);
 
                 struct {
                     u32 query_address_high;
@@ -971,6 +1012,12 @@ public:
     State state{};
     MemoryManager& memory_manager;
 
+    struct DirtyFlags {
+        bool vertex_attrib_format = true;
+    };
+
+    DirtyFlags dirty_flags;
+
     /// Reads a register value located at the input method address
     u32 GetRegisterValue(u32 method) const;
 
@@ -983,12 +1030,25 @@ public:
     /// Returns the texture information for a specific texture in a specific shader stage.
     Texture::FullTextureInfo GetStageTexture(Regs::ShaderStage stage, std::size_t offset) const;
 
+    /// Memory for macro code - it's undetermined how big this is, however 1MB is much larger than
+    /// we've seen used.
+    using MacroMemory = std::array<u32, 0x40000>;
+
+    /// Gets a reference to macro memory.
+    const MacroMemory& GetMacroMemory() const {
+        return macro_memory;
+    }
+
 private:
     void InitializeRegisterDefaults();
 
     VideoCore::RasterizerInterface& rasterizer;
 
-    std::unordered_map<u32, std::vector<u32>> uploaded_macros;
+    /// Start offsets of each macro in macro_memory
+    std::unordered_map<u32, u32> macro_offsets;
+
+    /// Memory for macro code
+    MacroMemory macro_memory;
 
     /// Macro method that is currently being executed / being fed parameters.
     u32 executing_macro = 0;
@@ -1011,9 +1071,12 @@ private:
      */
     void CallMacroMethod(u32 method, std::vector<u32> parameters);
 
-    /// Handles writes to the macro uploading registers.
+    /// Handles writes to the macro uploading register.
     void ProcessMacroUpload(u32 data);
 
+    /// Handles writes to the macro bind register.
+    void ProcessMacroBind(u32 data);
+
     /// Handles a write to the CLEAR_BUFFERS register.
     void ProcessClearBuffers();
 
@@ -1047,6 +1110,7 @@ ASSERT_REG_POSITION(scissor_test, 0x380);
 ASSERT_REG_POSITION(stencil_back_func_ref, 0x3D5);
 ASSERT_REG_POSITION(stencil_back_mask, 0x3D6);
 ASSERT_REG_POSITION(stencil_back_func_mask, 0x3D7);
+ASSERT_REG_POSITION(color_mask_common, 0x3E4);
 ASSERT_REG_POSITION(rt_separate_frag_data, 0x3EB);
 ASSERT_REG_POSITION(zeta, 0x3F8);
 ASSERT_REG_POSITION(vertex_attrib_format, 0x458);
@@ -1059,6 +1123,10 @@ ASSERT_REG_POSITION(depth_write_enabled, 0x4BA);
 ASSERT_REG_POSITION(alpha_test_enabled, 0x4BB);
 ASSERT_REG_POSITION(d3d_cull_mode, 0x4C2);
 ASSERT_REG_POSITION(depth_test_func, 0x4C3);
+ASSERT_REG_POSITION(alpha_test_ref, 0x4C4);
+ASSERT_REG_POSITION(alpha_test_func, 0x4C5);
+ASSERT_REG_POSITION(draw_tfb_stride, 0x4C6);
+ASSERT_REG_POSITION(blend_color, 0x4C7);
 ASSERT_REG_POSITION(blend, 0x4CF);
 ASSERT_REG_POSITION(stencil_enable, 0x4E0);
 ASSERT_REG_POSITION(stencil_front_op_fail, 0x4E1);
@@ -1079,14 +1147,17 @@ ASSERT_REG_POSITION(stencil_back_op_fail, 0x566);
 ASSERT_REG_POSITION(stencil_back_op_zfail, 0x567);
 ASSERT_REG_POSITION(stencil_back_op_zpass, 0x568);
 ASSERT_REG_POSITION(stencil_back_func_func, 0x569);
+ASSERT_REG_POSITION(framebuffer_srgb, 0x56E);
 ASSERT_REG_POSITION(point_coord_replace, 0x581);
 ASSERT_REG_POSITION(code_address, 0x582);
 ASSERT_REG_POSITION(draw, 0x585);
+ASSERT_REG_POSITION(primitive_restart, 0x591);
 ASSERT_REG_POSITION(index_array, 0x5F2);
 ASSERT_REG_POSITION(instanced_arrays, 0x620);
 ASSERT_REG_POSITION(cull, 0x646);
 ASSERT_REG_POSITION(logic_op, 0x671);
 ASSERT_REG_POSITION(clear_buffers, 0x674);
+ASSERT_REG_POSITION(color_mask, 0x680);
 ASSERT_REG_POSITION(query, 0x6C0);
 ASSERT_REG_POSITION(vertex_array[0], 0x700);
 ASSERT_REG_POSITION(independent_blend, 0x780);

src/video_core/engines/shader_bytecode.h

@@ -5,12 +5,11 @@
 #pragma once
 
 #include <bitset>
+#include <optional>
 #include <string>
 #include <tuple>
 #include <vector>
 
-#include <boost/optional.hpp>
-
 #include "common/assert.h"
 #include "common/bit_field.h"
 #include "common/common_types.h"
@@ -578,6 +577,10 @@ union Instruction {
         BitField<55, 1, u64> saturate;
     } fmul32;
 
+    union {
+        BitField<52, 1, u64> generates_cc;
+    } op_32;
+
     union {
         BitField<48, 1, u64> is_signed;
     } shift;
@@ -1232,6 +1235,7 @@ union Instruction {
     BitField<60, 1, u64> is_b_gpr;
     BitField<59, 1, u64> is_c_gpr;
     BitField<20, 24, s64> smem_imm;
+    BitField<0, 5, ControlCode> flow_control_code;
 
     Attribute attribute;
     Sampler sampler;
@@ -1456,7 +1460,7 @@ public:
         Type type;
     };
 
-    static boost::optional<const Matcher&> Decode(Instruction instr) {
+    static std::optional<std::reference_wrapper<const Matcher>> Decode(Instruction instr) {
         static const auto table{GetDecodeTable()};
 
         const auto matches_instruction = [instr](const auto& matcher) {
@@ -1464,7 +1468,8 @@ public:
         };
 
         auto iter = std::find_if(table.begin(), table.end(), matches_instruction);
-        return iter != table.end() ? boost::optional<const Matcher&>(*iter) : boost::none;
+        return iter != table.end() ? std::optional<std::reference_wrapper<const Matcher>>(*iter)
+                                   : std::nullopt;
     }
 
 private:
@@ -1658,4 +1663,4 @@ private:
     }
 };
 
-} // namespace Tegra::Shader
+} // namespace Tegra::Shader

src/video_core/macro_interpreter.cpp

@@ -11,7 +11,7 @@ namespace Tegra {
 
 MacroInterpreter::MacroInterpreter(Engines::Maxwell3D& maxwell3d) : maxwell3d(maxwell3d) {}
 
-void MacroInterpreter::Execute(const std::vector<u32>& code, std::vector<u32> parameters) {
+void MacroInterpreter::Execute(u32 offset, std::vector<u32> parameters) {
     Reset();
     registers[1] = parameters[0];
     this->parameters = std::move(parameters);
@@ -19,7 +19,7 @@ void MacroInterpreter::Execute(const std::vector<u32>& code, std::vector<u32> pa
     // Execute the code until we hit an exit condition.
     bool keep_executing = true;
     while (keep_executing) {
-        keep_executing = Step(code, false);
+        keep_executing = Step(offset, false);
     }
 
     // Assert the the macro used all the input parameters
@@ -29,7 +29,7 @@ void MacroInterpreter::Execute(const std::vector<u32>& code, std::vector<u32> pa
 void MacroInterpreter::Reset() {
     registers = {};
     pc = 0;
-    delayed_pc = boost::none;
+    delayed_pc = {};
     method_address.raw = 0;
     parameters.clear();
     // The next parameter index starts at 1, because $r1 already has the value of the first
@@ -37,17 +37,17 @@ void MacroInterpreter::Reset() {
     next_parameter_index = 1;
 }
 
-bool MacroInterpreter::Step(const std::vector<u32>& code, bool is_delay_slot) {
+bool MacroInterpreter::Step(u32 offset, bool is_delay_slot) {
     u32 base_address = pc;
 
-    Opcode opcode = GetOpcode(code);
+    Opcode opcode = GetOpcode(offset);
     pc += 4;
 
     // Update the program counter if we were delayed
-    if (delayed_pc != boost::none) {
+    if (delayed_pc) {
         ASSERT(is_delay_slot);
         pc = *delayed_pc;
-        delayed_pc = boost::none;
+        delayed_pc = {};
     }
 
     switch (opcode.operation) {
@@ -108,7 +108,7 @@ bool MacroInterpreter::Step(const std::vector<u32>& code, bool is_delay_slot) {
 
             delayed_pc = base_address + opcode.GetBranchTarget();
             // Execute one more instruction due to the delay slot.
-            return Step(code, true);
+            return Step(offset, true);
         }
         break;
     }
@@ -121,17 +121,18 @@ bool MacroInterpreter::Step(const std::vector<u32>& code, bool is_delay_slot) {
         // Exit has a delay slot, execute the next instruction
         // Note: Executing an exit during a branch delay slot will cause the instruction at the
         // branch target to be executed before exiting.
-        Step(code, true);
+        Step(offset, true);
         return false;
     }
 
     return true;
 }
 
-MacroInterpreter::Opcode MacroInterpreter::GetOpcode(const std::vector<u32>& code) const {
+MacroInterpreter::Opcode MacroInterpreter::GetOpcode(u32 offset) const {
+    const auto& macro_memory{maxwell3d.GetMacroMemory()};
     ASSERT((pc % sizeof(u32)) == 0);
-    ASSERT(pc < code.size() * sizeof(u32));
-    return {code[pc / sizeof(u32)]};
+    ASSERT((pc + offset) < macro_memory.size() * sizeof(u32));
+    return {macro_memory[offset + pc / sizeof(u32)]};
 }
 
 u32 MacroInterpreter::GetALUResult(ALUOperation operation, u32 src_a, u32 src_b) const {

src/video_core/macro_interpreter.h

@@ -5,8 +5,9 @@
 #pragma once
 
 #include <array>
+#include <optional>
 #include <vector>
-#include <boost/optional.hpp>
+
 #include "common/bit_field.h"
 #include "common/common_types.h"
 
@@ -21,10 +22,10 @@ public:
 
     /**
      * Executes the macro code with the specified input parameters.
-     * @param code The macro byte code to execute
-     * @param parameters The parameters of the macro
+     * @param offset Offset to start execution at.
+     * @param parameters The parameters of the macro.
      */
-    void Execute(const std::vector<u32>& code, std::vector<u32> parameters);
+    void Execute(u32 offset, std::vector<u32> parameters);
 
 private:
     enum class Operation : u32 {
@@ -109,11 +110,11 @@ private:
     /**
      * Executes a single macro instruction located at the current program counter. Returns whether
      * the interpreter should keep running.
-     * @param code The macro code to execute.
+     * @param offset Offset to start execution at.
      * @param is_delay_slot Whether the current step is being executed due to a delay slot in a
      * previous instruction.
      */
-    bool Step(const std::vector<u32>& code, bool is_delay_slot);
+    bool Step(u32 offset, bool is_delay_slot);
 
     /// Calculates the result of an ALU operation. src_a OP src_b;
     u32 GetALUResult(ALUOperation operation, u32 src_a, u32 src_b) const;
@@ -126,7 +127,7 @@ private:
     bool EvaluateBranchCondition(BranchCondition cond, u32 value) const;
 
     /// Reads an opcode at the current program counter location.
-    Opcode GetOpcode(const std::vector<u32>& code) const;
+    Opcode GetOpcode(u32 offset) const;
 
     /// Returns the specified register's value. Register 0 is hardcoded to always return 0.
     u32 GetRegister(u32 register_id) const;
@@ -149,7 +150,7 @@ private:
     Engines::Maxwell3D& maxwell3d;
 
     u32 pc; ///< Current program counter
-    boost::optional<u32>
+    std::optional<u32>
         delayed_pc; ///< Program counter to execute at after the delay slot is executed.
 
     static constexpr std::size_t NumMacroRegisters = 8;

src/video_core/memory_manager.cpp

@@ -4,18 +4,21 @@
 
 #include "common/alignment.h"
 #include "common/assert.h"
+#include "common/logging/log.h"
 #include "video_core/memory_manager.h"
 
 namespace Tegra {
 
 GPUVAddr MemoryManager::AllocateSpace(u64 size, u64 align) {
-    boost::optional<GPUVAddr> gpu_addr = FindFreeBlock(size, align);
-    ASSERT(gpu_addr);
+    const std::optional<GPUVAddr> gpu_addr{FindFreeBlock(0, size, align, PageStatus::Unmapped)};
 
-    for (u64 offset = 0; offset < size; offset += PAGE_SIZE) {
-        VAddr& slot = PageSlot(*gpu_addr + offset);
+    ASSERT_MSG(gpu_addr, "unable to find available GPU memory");
+
+    for (u64 offset{}; offset < size; offset += PAGE_SIZE) {
+        VAddr& slot{PageSlot(*gpu_addr + offset)};
 
         ASSERT(slot == static_cast<u64>(PageStatus::Unmapped));
+
         slot = static_cast<u64>(PageStatus::Allocated);
     }
 
@@ -23,10 +26,11 @@ GPUVAddr MemoryManager::AllocateSpace(u64 size, u64 align) {
 }
 
 GPUVAddr MemoryManager::AllocateSpace(GPUVAddr gpu_addr, u64 size, u64 align) {
-    for (u64 offset = 0; offset < size; offset += PAGE_SIZE) {
-        VAddr& slot = PageSlot(gpu_addr + offset);
+    for (u64 offset{}; offset < size; offset += PAGE_SIZE) {
+        VAddr& slot{PageSlot(gpu_addr + offset)};
 
         ASSERT(slot == static_cast<u64>(PageStatus::Unmapped));
+
         slot = static_cast<u64>(PageStatus::Allocated);
     }
 
@@ -34,17 +38,19 @@ GPUVAddr MemoryManager::AllocateSpace(GPUVAddr gpu_addr, u64 size, u64 align) {
 }
 
 GPUVAddr MemoryManager::MapBufferEx(VAddr cpu_addr, u64 size) {
-    boost::optional<GPUVAddr> gpu_addr = FindFreeBlock(size, PAGE_SIZE);
-    ASSERT(gpu_addr);
+    const std::optional<GPUVAddr> gpu_addr{FindFreeBlock(0, size, PAGE_SIZE, PageStatus::Unmapped)};
 
-    for (u64 offset = 0; offset < size; offset += PAGE_SIZE) {
-        VAddr& slot = PageSlot(*gpu_addr + offset);
+    ASSERT_MSG(gpu_addr, "unable to find available GPU memory");
+
+    for (u64 offset{}; offset < size; offset += PAGE_SIZE) {
+        VAddr& slot{PageSlot(*gpu_addr + offset)};
 
         ASSERT(slot == static_cast<u64>(PageStatus::Unmapped));
+
         slot = cpu_addr + offset;
     }
 
-    MappedRegion region{cpu_addr, *gpu_addr, size};
+    const MappedRegion region{cpu_addr, *gpu_addr, size};
     mapped_regions.push_back(region);
 
     return *gpu_addr;
@@ -53,14 +59,31 @@ GPUVAddr MemoryManager::MapBufferEx(VAddr cpu_addr, u64 size) {
 GPUVAddr MemoryManager::MapBufferEx(VAddr cpu_addr, GPUVAddr gpu_addr, u64 size) {
     ASSERT((gpu_addr & PAGE_MASK) == 0);
 
-    for (u64 offset = 0; offset < size; offset += PAGE_SIZE) {
-        VAddr& slot = PageSlot(gpu_addr + offset);
+    if (PageSlot(gpu_addr) != static_cast<u64>(PageStatus::Allocated)) {
+        // Page has been already mapped. In this case, we must find a new area of memory to use that
+        // is different than the specified one. Super Mario Odyssey hits this scenario when changing
+        // areas, but we do not want to overwrite the old pages.
+        // TODO(bunnei): We need to write a hardware test to confirm this behavior.
+
+        LOG_ERROR(HW_GPU, "attempting to map addr 0x{:016X}, which is not available!", gpu_addr);
+
+        const std::optional<GPUVAddr> new_gpu_addr{
+            FindFreeBlock(gpu_addr, size, PAGE_SIZE, PageStatus::Allocated)};
+
+        ASSERT_MSG(new_gpu_addr, "unable to find available GPU memory");
+
+        gpu_addr = *new_gpu_addr;
+    }
+
+    for (u64 offset{}; offset < size; offset += PAGE_SIZE) {
+        VAddr& slot{PageSlot(gpu_addr + offset)};
 
         ASSERT(slot == static_cast<u64>(PageStatus::Allocated));
+
         slot = cpu_addr + offset;
     }
 
-    MappedRegion region{cpu_addr, gpu_addr, size};
+    const MappedRegion region{cpu_addr, gpu_addr, size};
     mapped_regions.push_back(region);
 
     return gpu_addr;
@@ -69,11 +92,12 @@ GPUVAddr MemoryManager::MapBufferEx(VAddr cpu_addr, GPUVAddr gpu_addr, u64 size)
 GPUVAddr MemoryManager::UnmapBuffer(GPUVAddr gpu_addr, u64 size) {
     ASSERT((gpu_addr & PAGE_MASK) == 0);
 
-    for (u64 offset = 0; offset < size; offset += PAGE_SIZE) {
-        VAddr& slot = PageSlot(gpu_addr + offset);
+    for (u64 offset{}; offset < size; offset += PAGE_SIZE) {
+        VAddr& slot{PageSlot(gpu_addr + offset)};
 
         ASSERT(slot != static_cast<u64>(PageStatus::Allocated) &&
                slot != static_cast<u64>(PageStatus::Unmapped));
+
         slot = static_cast<u64>(PageStatus::Unmapped);
     }
 
@@ -97,13 +121,14 @@ GPUVAddr MemoryManager::GetRegionEnd(GPUVAddr region_start) const {
     return {};
 }
 
-boost::optional<GPUVAddr> MemoryManager::FindFreeBlock(u64 size, u64 align) {
-    GPUVAddr gpu_addr = 0;
-    u64 free_space = 0;
+std::optional<GPUVAddr> MemoryManager::FindFreeBlock(GPUVAddr region_start, u64 size, u64 align,
+                                                     PageStatus status) {
+    GPUVAddr gpu_addr{region_start};
+    u64 free_space{};
     align = (align + PAGE_MASK) & ~PAGE_MASK;
 
     while (gpu_addr + free_space < MAX_ADDRESS) {
-        if (!IsPageMapped(gpu_addr + free_space)) {
+        if (PageSlot(gpu_addr + free_space) == static_cast<u64>(status)) {
             free_space += PAGE_SIZE;
             if (free_space >= size) {
                 return gpu_addr;
@@ -118,8 +143,8 @@ boost::optional<GPUVAddr> MemoryManager::FindFreeBlock(u64 size, u64 align) {
     return {};
 }
 
-boost::optional<VAddr> MemoryManager::GpuToCpuAddress(GPUVAddr gpu_addr) {
-    VAddr base_addr = PageSlot(gpu_addr);
+std::optional<VAddr> MemoryManager::GpuToCpuAddress(GPUVAddr gpu_addr) {
+    const VAddr base_addr{PageSlot(gpu_addr)};
 
     if (base_addr == static_cast<u64>(PageStatus::Allocated) ||
         base_addr == static_cast<u64>(PageStatus::Unmapped)) {
@@ -133,19 +158,15 @@ std::vector<GPUVAddr> MemoryManager::CpuToGpuAddress(VAddr cpu_addr) const {
     std::vector<GPUVAddr> results;
     for (const auto& region : mapped_regions) {
         if (cpu_addr >= region.cpu_addr && cpu_addr < (region.cpu_addr + region.size)) {
-            u64 offset = cpu_addr - region.cpu_addr;
+            const u64 offset{cpu_addr - region.cpu_addr};
             results.push_back(region.gpu_addr + offset);
         }
     }
     return results;
 }
 
-bool MemoryManager::IsPageMapped(GPUVAddr gpu_addr) {
-    return PageSlot(gpu_addr) != static_cast<u64>(PageStatus::Unmapped);
-}
-
 VAddr& MemoryManager::PageSlot(GPUVAddr gpu_addr) {
-    auto& block = page_table[(gpu_addr >> (PAGE_BITS + PAGE_TABLE_BITS)) & PAGE_TABLE_MASK];
+    auto& block{page_table[(gpu_addr >> (PAGE_BITS + PAGE_TABLE_BITS)) & PAGE_TABLE_MASK]};
     if (!block) {
         block = std::make_unique<PageBlock>();
         block->fill(static_cast<VAddr>(PageStatus::Unmapped));

src/video_core/memory_manager.h

@@ -6,10 +6,9 @@
 
 #include <array>
 #include <memory>
+#include <optional>
 #include <vector>
 
-#include <boost/optional.hpp>
-
 #include "common/common_types.h"
 
 namespace Tegra {
@@ -27,7 +26,7 @@ public:
     GPUVAddr MapBufferEx(VAddr cpu_addr, GPUVAddr gpu_addr, u64 size);
     GPUVAddr UnmapBuffer(GPUVAddr gpu_addr, u64 size);
     GPUVAddr GetRegionEnd(GPUVAddr region_start) const;
-    boost::optional<VAddr> GpuToCpuAddress(GPUVAddr gpu_addr);
+    std::optional<VAddr> GpuToCpuAddress(GPUVAddr gpu_addr);
     std::vector<GPUVAddr> CpuToGpuAddress(VAddr cpu_addr) const;
 
     static constexpr u64 PAGE_BITS = 16;
@@ -35,15 +34,15 @@ public:
     static constexpr u64 PAGE_MASK = PAGE_SIZE - 1;
 
 private:
-    boost::optional<GPUVAddr> FindFreeBlock(u64 size, u64 align = 1);
-    bool IsPageMapped(GPUVAddr gpu_addr);
-    VAddr& PageSlot(GPUVAddr gpu_addr);
-
     enum class PageStatus : u64 {
         Unmapped = 0xFFFFFFFFFFFFFFFFULL,
         Allocated = 0xFFFFFFFFFFFFFFFEULL,
     };
 
+    std::optional<GPUVAddr> FindFreeBlock(GPUVAddr region_start, u64 size, u64 align,
+                                          PageStatus status);
+    VAddr& PageSlot(GPUVAddr gpu_addr);
+
     static constexpr u64 MAX_ADDRESS{0x10000000000ULL};
     static constexpr u64 PAGE_TABLE_BITS{10};
     static constexpr u64 PAGE_TABLE_SIZE{1 << PAGE_TABLE_BITS};

src/video_core/rasterizer_cache.cpp

@@ -0,0 +1,7 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "video_core/rasterizer_cache.h"
+
+RasterizerCacheObject::~RasterizerCacheObject() = default;

src/video_core/rasterizer_cache.h

@@ -10,13 +10,13 @@
 #include <boost/range/iterator_range_core.hpp>
 
 #include "common/common_types.h"
-#include "core/core.h"
 #include "core/settings.h"
 #include "video_core/rasterizer_interface.h"
-#include "video_core/renderer_base.h"
 
 class RasterizerCacheObject {
 public:
+    virtual ~RasterizerCacheObject();
+
     /// Gets the address of the shader in guest memory, required for cache management
     virtual VAddr GetAddr() const = 0;
 
@@ -64,6 +64,8 @@ class RasterizerCache : NonCopyable {
     friend class RasterizerCacheObject;
 
 public:
+    explicit RasterizerCache(VideoCore::RasterizerInterface& rasterizer) : rasterizer{rasterizer} {}
+
     /// Write any cached resources overlapping the specified region back to memory
     void FlushRegion(Tegra::GPUVAddr addr, size_t size) {
         const auto& objects{GetSortedObjectsFromRegion(addr, size)};
@@ -109,14 +111,12 @@ protected:
     void Register(const T& object) {
         object->SetIsRegistered(true);
         object_cache.add({GetInterval(object), ObjectSet{object}});
-        auto& rasterizer = Core::System::GetInstance().Renderer().Rasterizer();
         rasterizer.UpdatePagesCachedCount(object->GetAddr(), object->GetSizeInBytes(), 1);
     }
 
     /// Unregisters an object from the cache
     void Unregister(const T& object) {
         object->SetIsRegistered(false);
-        auto& rasterizer = Core::System::GetInstance().Renderer().Rasterizer();
         rasterizer.UpdatePagesCachedCount(object->GetAddr(), object->GetSizeInBytes(), -1);
 
         // Only flush if use_accurate_gpu_emulation is enabled, as it incurs a performance hit
@@ -177,4 +177,5 @@ private:
 
     ObjectCache object_cache; ///< Cache of objects
     u64 modified_ticks{};     ///< Counter of cache state ticks, used for in-order flushing
+    VideoCore::RasterizerInterface& rasterizer;
 };

src/video_core/renderer_base.h

@@ -6,7 +6,8 @@
 
 #include <atomic>
 #include <memory>
-#include <boost/optional.hpp>
+#include <optional>
+
 #include "common/common_types.h"
 #include "video_core/gpu.h"
 #include "video_core/rasterizer_interface.h"
@@ -28,7 +29,8 @@ public:
     virtual ~RendererBase();
 
     /// Swap buffers (render frame)
-    virtual void SwapBuffers(boost::optional<const Tegra::FramebufferConfig&> framebuffer) = 0;
+    virtual void SwapBuffers(
+        std::optional<std::reference_wrapper<const Tegra::FramebufferConfig>> framebuffer) = 0;
 
     /// Initialize the renderer
     virtual bool Init() = 0;

src/video_core/renderer_opengl/gl_buffer_cache.cpp

@@ -9,15 +9,17 @@
 #include "core/core.h"
 #include "core/memory.h"
 #include "video_core/renderer_opengl/gl_buffer_cache.h"
+#include "video_core/renderer_opengl/gl_rasterizer.h"
 
 namespace OpenGL {
 
-OGLBufferCache::OGLBufferCache(std::size_t size) : stream_buffer(GL_ARRAY_BUFFER, size) {}
+OGLBufferCache::OGLBufferCache(RasterizerOpenGL& rasterizer, std::size_t size)
+    : RasterizerCache{rasterizer}, stream_buffer(GL_ARRAY_BUFFER, size) {}
 
 GLintptr OGLBufferCache::UploadMemory(Tegra::GPUVAddr gpu_addr, std::size_t size,
                                       std::size_t alignment, bool cache) {
     auto& memory_manager = Core::System::GetInstance().GPU().MemoryManager();
-    const boost::optional<VAddr> cpu_addr{memory_manager.GpuToCpuAddress(gpu_addr)};
+    const std::optional<VAddr> cpu_addr{memory_manager.GpuToCpuAddress(gpu_addr)};
 
     // Cache management is a big overhead, so only cache entries with a given size.
     // TODO: Figure out which size is the best for given games.

src/video_core/renderer_opengl/gl_buffer_cache.h

@@ -15,6 +15,8 @@
 
 namespace OpenGL {
 
+class RasterizerOpenGL;
+
 struct CachedBufferEntry final : public RasterizerCacheObject {
     VAddr GetAddr() const override {
         return addr;
@@ -35,7 +37,7 @@ struct CachedBufferEntry final : public RasterizerCacheObject {
 
 class OGLBufferCache final : public RasterizerCache<std::shared_ptr<CachedBufferEntry>> {
 public:
-    explicit OGLBufferCache(std::size_t size);
+    explicit OGLBufferCache(RasterizerOpenGL& rasterizer, std::size_t size);
 
     /// Uploads data from a guest GPU address. Returns host's buffer offset where it's been
     /// allocated.

src/video_core/renderer_opengl/gl_primitive_assembler.cpp

@@ -6,6 +6,7 @@
 #include <array>
 #include "common/assert.h"
 #include "common/common_types.h"
+#include "core/core.h"
 #include "core/memory.h"
 #include "video_core/renderer_opengl/gl_buffer_cache.h"
 #include "video_core/renderer_opengl/gl_primitive_assembler.h"
@@ -45,7 +46,7 @@ GLintptr PrimitiveAssembler::MakeQuadIndexed(Tegra::GPUVAddr gpu_addr, std::size
     auto [dst_pointer, index_offset] = buffer_cache.ReserveMemory(map_size);
 
     auto& memory_manager = Core::System::GetInstance().GPU().MemoryManager();
-    const boost::optional<VAddr> cpu_addr{memory_manager.GpuToCpuAddress(gpu_addr)};
+    const std::optional<VAddr> cpu_addr{memory_manager.GpuToCpuAddress(gpu_addr)};
     const u8* source{Memory::GetPointer(*cpu_addr)};
 
     for (u32 primitive = 0; primitive < count / 4; ++primitive) {

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -30,10 +30,11 @@
 namespace OpenGL {
 
 using Maxwell = Tegra::Engines::Maxwell3D::Regs;
-using PixelFormat = SurfaceParams::PixelFormat;
-using SurfaceType = SurfaceParams::SurfaceType;
+using PixelFormat = VideoCore::Surface::PixelFormat;
+using SurfaceType = VideoCore::Surface::SurfaceType;
 
-MICROPROFILE_DEFINE(OpenGL_VAO, "OpenGL", "Vertex Array Setup", MP_RGB(128, 128, 192));
+MICROPROFILE_DEFINE(OpenGL_VAO, "OpenGL", "Vertex Format Setup", MP_RGB(128, 128, 192));
+MICROPROFILE_DEFINE(OpenGL_VB, "OpenGL", "Vertex Buffer Setup", MP_RGB(128, 128, 192));
 MICROPROFILE_DEFINE(OpenGL_Shader, "OpenGL", "Shader Setup", MP_RGB(128, 128, 192));
 MICROPROFILE_DEFINE(OpenGL_UBO, "OpenGL", "Const Buffer Setup", MP_RGB(128, 128, 192));
 MICROPROFILE_DEFINE(OpenGL_Index, "OpenGL", "Index Buffer Setup", MP_RGB(128, 128, 192));
@@ -79,7 +80,8 @@ struct DrawParameters {
 };
 
 RasterizerOpenGL::RasterizerOpenGL(Core::Frontend::EmuWindow& window, ScreenInfo& info)
-    : emu_window{window}, screen_info{info}, buffer_cache(STREAM_BUFFER_SIZE) {
+    : res_cache{*this}, shader_cache{*this}, emu_window{window}, screen_info{info},
+      buffer_cache(*this, STREAM_BUFFER_SIZE) {
     // Create sampler objects
     for (std::size_t i = 0; i < texture_samplers.size(); ++i) {
         texture_samplers[i].Create();
@@ -104,7 +106,7 @@ RasterizerOpenGL::RasterizerOpenGL(Core::Frontend::EmuWindow& window, ScreenInfo
     }
 
     ASSERT_MSG(has_ARB_separate_shader_objects, "has_ARB_separate_shader_objects is unsupported");
-
+    OpenGLState::ApplyDefaultState();
     // Clipping plane 0 is always enabled for PICA fixed clip plane z <= 0
     state.clip_distance[0] = true;
 
@@ -115,8 +117,6 @@ RasterizerOpenGL::RasterizerOpenGL(Core::Frontend::EmuWindow& window, ScreenInfo
     state.draw.shader_program = 0;
     state.Apply();
 
-    glEnable(GL_BLEND);
-
     glGetIntegerv(GL_UNIFORM_BUFFER_OFFSET_ALIGNMENT, &uniform_buffer_alignment);
 
     LOG_CRITICAL(Render_OpenGL, "Sync fixed function OpenGL state here!");
@@ -124,18 +124,23 @@ RasterizerOpenGL::RasterizerOpenGL(Core::Frontend::EmuWindow& window, ScreenInfo
 
 RasterizerOpenGL::~RasterizerOpenGL() {}
 
-void RasterizerOpenGL::SetupVertexArrays() {
-    MICROPROFILE_SCOPE(OpenGL_VAO);
-    const auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
+void RasterizerOpenGL::SetupVertexFormat() {
+    auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
     const auto& regs = gpu.regs;
 
+    if (!gpu.dirty_flags.vertex_attrib_format)
+        return;
+    gpu.dirty_flags.vertex_attrib_format = false;
+
+    MICROPROFILE_SCOPE(OpenGL_VAO);
+
     auto [iter, is_cache_miss] = vertex_array_cache.try_emplace(regs.vertex_attrib_format);
     auto& VAO = iter->second;
 
     if (is_cache_miss) {
         VAO.Create();
         state.draw.vertex_array = VAO.handle;
-        state.Apply();
+        state.ApplyVertexBufferState();
 
         // The index buffer binding is stored within the VAO. Stupid OpenGL, but easy to work
         // around.
@@ -177,8 +182,13 @@ void RasterizerOpenGL::SetupVertexArrays() {
         }
     }
     state.draw.vertex_array = VAO.handle;
-    state.draw.vertex_buffer = buffer_cache.GetHandle();
-    state.Apply();
+    state.ApplyVertexBufferState();
+}
+
+void RasterizerOpenGL::SetupVertexBuffer() {
+    MICROPROFILE_SCOPE(OpenGL_VB);
+    const auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
+    const auto& regs = gpu.regs;
 
     // Upload all guest vertex arrays sequentially to our buffer
     for (u32 index = 0; index < Maxwell::NumVertexArrays; ++index) {
@@ -205,6 +215,9 @@ void RasterizerOpenGL::SetupVertexArrays() {
             glVertexBindingDivisor(index, 0);
         }
     }
+
+    // Implicit set by glBindVertexBuffer. Stupid glstate handling...
+    state.draw.vertex_buffer = buffer_cache.GetHandle();
 }
 
 DrawParameters RasterizerOpenGL::SetupDraw() {
@@ -329,8 +342,6 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
             index++;
         }
     }
-
-    state.Apply();
 }
 
 std::size_t RasterizerOpenGL::CalculateVertexArraysSize() const {
@@ -399,9 +410,9 @@ void RasterizerOpenGL::UpdatePagesCachedCount(VAddr addr, u64 size, int delta) {
         cached_pages.add({pages_interval, delta});
 }
 
-void RasterizerOpenGL::ConfigureFramebuffers(bool using_color_fb, bool using_depth_fb,
-                                             bool preserve_contents,
-                                             boost::optional<std::size_t> single_color_target) {
+void RasterizerOpenGL::ConfigureFramebuffers(OpenGLState& current_state, bool using_color_fb,
+                                             bool using_depth_fb, bool preserve_contents,
+                                             std::optional<std::size_t> single_color_target) {
     MICROPROFILE_SCOPE(OpenGL_Framebuffer);
     const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
 
@@ -416,8 +427,9 @@ void RasterizerOpenGL::ConfigureFramebuffers(bool using_color_fb, bool using_dep
     ASSERT_MSG(regs.rt_separate_frag_data == 0, "Unimplemented");
 
     // Bind the framebuffer surfaces
-    state.draw.draw_framebuffer = framebuffer.handle;
-    state.Apply();
+    current_state.draw.draw_framebuffer = framebuffer.handle;
+    current_state.ApplyFramebufferState();
+    current_state.framebuffer_srgb.enabled = regs.framebuffer_srgb != 0;
 
     if (using_color_fb) {
         if (single_color_target) {
@@ -429,6 +441,9 @@ void RasterizerOpenGL::ConfigureFramebuffers(bool using_color_fb, bool using_dep
                 // Assume that a surface will be written to if it is used as a framebuffer, even if
                 // the shader doesn't actually write to it.
                 color_surface->MarkAsModified(true, res_cache);
+                // Workaround for and issue in nvidia drivers
+                // https://devtalk.nvidia.com/default/topic/776591/opengl/gl_framebuffer_srgb-functions-incorrectly/
+                state.framebuffer_srgb.enabled |= color_surface->GetSurfaceParams().srgb_conversion;
             }
 
             glFramebufferTexture2D(
@@ -446,6 +461,11 @@ void RasterizerOpenGL::ConfigureFramebuffers(bool using_color_fb, bool using_dep
                     // Assume that a surface will be written to if it is used as a framebuffer, even
                     // if the shader doesn't actually write to it.
                     color_surface->MarkAsModified(true, res_cache);
+                    // Enable sRGB only for supported formats
+                    // Workaround for and issue in nvidia drivers
+                    // https://devtalk.nvidia.com/default/topic/776591/opengl/gl_framebuffer_srgb-functions-incorrectly/
+                    state.framebuffer_srgb.enabled |=
+                        color_surface->GetSurfaceParams().srgb_conversion;
                 }
 
                 buffers[index] = GL_COLOR_ATTACHMENT0 + regs.rt_control.GetMap(index);
@@ -487,10 +507,7 @@ void RasterizerOpenGL::ConfigureFramebuffers(bool using_color_fb, bool using_dep
         glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0,
                                0);
     }
-
-    SyncViewport();
-
-    state.Apply();
+    SyncViewport(current_state);
 }
 
 void RasterizerOpenGL::Clear() {
@@ -503,22 +520,23 @@ void RasterizerOpenGL::Clear() {
     bool use_stencil{};
 
     OpenGLState clear_state;
-    clear_state.draw.draw_framebuffer = framebuffer.handle;
-    clear_state.color_mask.red_enabled = regs.clear_buffers.R ? GL_TRUE : GL_FALSE;
-    clear_state.color_mask.green_enabled = regs.clear_buffers.G ? GL_TRUE : GL_FALSE;
-    clear_state.color_mask.blue_enabled = regs.clear_buffers.B ? GL_TRUE : GL_FALSE;
-    clear_state.color_mask.alpha_enabled = regs.clear_buffers.A ? GL_TRUE : GL_FALSE;
-
     if (regs.clear_buffers.R || regs.clear_buffers.G || regs.clear_buffers.B ||
         regs.clear_buffers.A) {
         use_color = true;
     }
+    if (use_color) {
+        clear_state.color_mask[0].red_enabled = regs.clear_buffers.R ? GL_TRUE : GL_FALSE;
+        clear_state.color_mask[0].green_enabled = regs.clear_buffers.G ? GL_TRUE : GL_FALSE;
+        clear_state.color_mask[0].blue_enabled = regs.clear_buffers.B ? GL_TRUE : GL_FALSE;
+        clear_state.color_mask[0].alpha_enabled = regs.clear_buffers.A ? GL_TRUE : GL_FALSE;
+    }
     if (regs.clear_buffers.Z) {
         ASSERT_MSG(regs.zeta_enable != 0, "Tried to clear Z but buffer is not enabled!");
         use_depth = true;
 
         // Always enable the depth write when clearing the depth buffer. The depth write mask is
-        // ignored when clearing the buffer in the Switch, but OpenGL obeys it so we set it to true.
+        // ignored when clearing the buffer in the Switch, but OpenGL obeys it so we set it to
+        // true.
         clear_state.depth.test_enabled = true;
         clear_state.depth.test_func = GL_ALWAYS;
     }
@@ -535,9 +553,8 @@ void RasterizerOpenGL::Clear() {
 
     ScopeAcquireGLContext acquire_context{emu_window};
 
-    ConfigureFramebuffers(use_color, use_depth || use_stencil, false,
+    ConfigureFramebuffers(clear_state, use_color, use_depth || use_stencil, false,
                           regs.clear_buffers.RT.Value());
-
     clear_state.Apply();
 
     if (use_color) {
@@ -563,14 +580,14 @@ void RasterizerOpenGL::DrawArrays() {
 
     ScopeAcquireGLContext acquire_context{emu_window};
 
-    ConfigureFramebuffers();
-
+    ConfigureFramebuffers(state);
+    SyncColorMask();
     SyncDepthTestState();
     SyncStencilTestState();
     SyncBlendState();
     SyncLogicOpState();
     SyncCullMode();
-    SyncDepthRange();
+    SyncPrimitiveRestart();
     SyncScissorTest();
     // Alpha Testing is synced on shaders.
     SyncTransformFeedback();
@@ -584,7 +601,7 @@ void RasterizerOpenGL::DrawArrays() {
     const bool is_indexed = accelerate_draw == AccelDraw::Indexed;
 
     state.draw.vertex_buffer = buffer_cache.GetHandle();
-    state.Apply();
+    state.ApplyVertexBufferState();
 
     std::size_t buffer_size = CalculateVertexArraysSize();
 
@@ -611,7 +628,8 @@ void RasterizerOpenGL::DrawArrays() {
 
     buffer_cache.Map(buffer_size);
 
-    SetupVertexArrays();
+    SetupVertexFormat();
+    SetupVertexBuffer();
     DrawParameters params = SetupDraw();
     SetupShaders(params.primitive_mode);
 
@@ -691,7 +709,8 @@ bool RasterizerOpenGL::AccelerateDisplay(const Tegra::FramebufferConfig& config,
 
     // Verify that the cached surface is the same size and format as the requested framebuffer
     const auto& params{surface->GetSurfaceParams()};
-    const auto& pixel_format{SurfaceParams::PixelFormatFromGPUPixelFormat(config.pixel_format)};
+    const auto& pixel_format{
+        VideoCore::Surface::PixelFormatFromGPUPixelFormat(config.pixel_format)};
     ASSERT_MSG(params.width == config.width, "Framebuffer width is different");
     ASSERT_MSG(params.height == config.height, "Framebuffer height is different");
     ASSERT_MSG(params.pixel_format == pixel_format, "Framebuffer pixel_format is different");
@@ -714,16 +733,20 @@ void RasterizerOpenGL::SamplerInfo::Create() {
     glSamplerParameteri(sampler.handle, GL_TEXTURE_COMPARE_FUNC, GL_NEVER);
 }
 
-void RasterizerOpenGL::SamplerInfo::SyncWithConfig(const Tegra::Texture::TSCEntry& config) {
+void RasterizerOpenGL::SamplerInfo::SyncWithConfig(const Tegra::Texture::FullTextureInfo& info) {
     const GLuint s = sampler.handle;
-
+    const Tegra::Texture::TSCEntry& config = info.tsc;
     if (mag_filter != config.mag_filter) {
         mag_filter = config.mag_filter;
-        glSamplerParameteri(s, GL_TEXTURE_MAG_FILTER, MaxwellToGL::TextureFilterMode(mag_filter));
+        glSamplerParameteri(
+            s, GL_TEXTURE_MAG_FILTER,
+            MaxwellToGL::TextureFilterMode(mag_filter, Tegra::Texture::TextureMipmapFilter::None));
     }
-    if (min_filter != config.min_filter) {
+    if (min_filter != config.min_filter || mip_filter != config.mip_filter) {
         min_filter = config.min_filter;
-        glSamplerParameteri(s, GL_TEXTURE_MIN_FILTER, MaxwellToGL::TextureFilterMode(min_filter));
+        mip_filter = config.mip_filter;
+        glSamplerParameteri(s, GL_TEXTURE_MIN_FILTER,
+                            MaxwellToGL::TextureFilterMode(min_filter, mip_filter));
     }
 
     if (wrap_u != config.wrap_u) {
@@ -763,6 +786,22 @@ void RasterizerOpenGL::SamplerInfo::SyncWithConfig(const Tegra::Texture::TSCEntr
             glSamplerParameterfv(s, GL_TEXTURE_BORDER_COLOR, border_color.data());
         }
     }
+    if (info.tic.use_header_opt_control == 0) {
+        if (GLAD_GL_ARB_texture_filter_anisotropic) {
+            glSamplerParameterf(s, GL_TEXTURE_MAX_ANISOTROPY,
+                                static_cast<float>(1 << info.tic.max_anisotropy.Value()));
+        } else if (GLAD_GL_EXT_texture_filter_anisotropic) {
+            glSamplerParameterf(s, GL_TEXTURE_MAX_ANISOTROPY_EXT,
+                                static_cast<float>(1 << info.tic.max_anisotropy.Value()));
+        }
+        glSamplerParameterf(s, GL_TEXTURE_MIN_LOD,
+                            static_cast<float>(info.tic.res_min_mip_level.Value()));
+        glSamplerParameterf(s, GL_TEXTURE_MAX_LOD,
+                            static_cast<float>(info.tic.res_max_mip_level.Value() == 0
+                                                   ? 16
+                                                   : info.tic.res_max_mip_level.Value()));
+        glSamplerParameterf(s, GL_TEXTURE_LOD_BIAS, info.tic.mip_lod_bias.Value() / 256.f);
+    }
 }
 
 u32 RasterizerOpenGL::SetupConstBuffers(Maxwell::ShaderStage stage, Shader& shader,
@@ -860,7 +899,7 @@ u32 RasterizerOpenGL::SetupTextures(Maxwell::ShaderStage stage, Shader& shader,
             continue;
         }
 
-        texture_samplers[current_bindpoint].SyncWithConfig(texture.tsc);
+        texture_samplers[current_bindpoint].SyncWithConfig(texture);
         Surface surface = res_cache.GetTextureSurface(texture, entry);
         if (surface != nullptr) {
             state.texture_units[current_bindpoint].texture = surface->Texture().handle;
@@ -882,14 +921,18 @@ u32 RasterizerOpenGL::SetupTextures(Maxwell::ShaderStage stage, Shader& shader,
     return current_unit + static_cast<u32>(entries.size());
 }
 
-void RasterizerOpenGL::SyncViewport() {
+void RasterizerOpenGL::SyncViewport(OpenGLState& current_state) {
     const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
-    const MathUtil::Rectangle<s32> viewport_rect{regs.viewport_transform[0].GetRect()};
-
-    state.viewport.x = viewport_rect.left;
-    state.viewport.y = viewport_rect.bottom;
-    state.viewport.width = static_cast<GLsizei>(viewport_rect.GetWidth());
-    state.viewport.height = static_cast<GLsizei>(viewport_rect.GetHeight());
+    for (size_t i = 0; i < Tegra::Engines::Maxwell3D::Regs::NumRenderTargets; i++) {
+        const MathUtil::Rectangle<s32> viewport_rect{regs.viewport_transform[i].GetRect()};
+        auto& viewport = current_state.viewports[i];
+        viewport.x = viewport_rect.left;
+        viewport.y = viewport_rect.bottom;
+        viewport.width = static_cast<GLfloat>(viewport_rect.GetWidth());
+        viewport.height = static_cast<GLfloat>(viewport_rect.GetHeight());
+        viewport.depth_range_far = regs.viewport[i].depth_range_far;
+        viewport.depth_range_near = regs.viewport[i].depth_range_near;
+    }
 }
 
 void RasterizerOpenGL::SyncClipEnabled() {
@@ -924,11 +967,11 @@ void RasterizerOpenGL::SyncCullMode() {
     }
 }
 
-void RasterizerOpenGL::SyncDepthRange() {
+void RasterizerOpenGL::SyncPrimitiveRestart() {
     const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
 
-    state.depth.depth_range_near = regs.viewport->depth_range_near;
-    state.depth.depth_range_far = regs.viewport->depth_range_far;
+    state.primitive_restart.enabled = regs.primitive_restart.enabled;
+    state.primitive_restart.index = regs.primitive_restart.index;
 }
 
 void RasterizerOpenGL::SyncDepthTestState() {
@@ -951,9 +994,6 @@ void RasterizerOpenGL::SyncStencilTestState() {
         return;
     }
 
-    // TODO(bunnei): Verify behavior when this is not set
-    ASSERT(regs.stencil_two_side_enable);
-
     state.stencil.front.test_func = MaxwellToGL::ComparisonOp(regs.stencil_front_func_func);
     state.stencil.front.test_ref = regs.stencil_front_func_ref;
     state.stencil.front.test_mask = regs.stencil_front_func_mask;
@@ -961,36 +1001,79 @@ void RasterizerOpenGL::SyncStencilTestState() {
     state.stencil.front.action_depth_fail = MaxwellToGL::StencilOp(regs.stencil_front_op_zfail);
     state.stencil.front.action_depth_pass = MaxwellToGL::StencilOp(regs.stencil_front_op_zpass);
     state.stencil.front.write_mask = regs.stencil_front_mask;
+    if (regs.stencil_two_side_enable) {
+        state.stencil.back.test_func = MaxwellToGL::ComparisonOp(regs.stencil_back_func_func);
+        state.stencil.back.test_ref = regs.stencil_back_func_ref;
+        state.stencil.back.test_mask = regs.stencil_back_func_mask;
+        state.stencil.back.action_stencil_fail = MaxwellToGL::StencilOp(regs.stencil_back_op_fail);
+        state.stencil.back.action_depth_fail = MaxwellToGL::StencilOp(regs.stencil_back_op_zfail);
+        state.stencil.back.action_depth_pass = MaxwellToGL::StencilOp(regs.stencil_back_op_zpass);
+        state.stencil.back.write_mask = regs.stencil_back_mask;
+    } else {
+        state.stencil.back.test_func = GL_ALWAYS;
+        state.stencil.back.test_ref = 0;
+        state.stencil.back.test_mask = 0xFFFFFFFF;
+        state.stencil.back.write_mask = 0xFFFFFFFF;
+        state.stencil.back.action_stencil_fail = GL_KEEP;
+        state.stencil.back.action_depth_fail = GL_KEEP;
+        state.stencil.back.action_depth_pass = GL_KEEP;
+    }
+}
 
-    state.stencil.back.test_func = MaxwellToGL::ComparisonOp(regs.stencil_back_func_func);
-    state.stencil.back.test_ref = regs.stencil_back_func_ref;
-    state.stencil.back.test_mask = regs.stencil_back_func_mask;
-    state.stencil.back.action_stencil_fail = MaxwellToGL::StencilOp(regs.stencil_back_op_fail);
-    state.stencil.back.action_depth_fail = MaxwellToGL::StencilOp(regs.stencil_back_op_zfail);
-    state.stencil.back.action_depth_pass = MaxwellToGL::StencilOp(regs.stencil_back_op_zpass);
-    state.stencil.back.write_mask = regs.stencil_back_mask;
+void RasterizerOpenGL::SyncColorMask() {
+    const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+    for (size_t i = 0; i < Tegra::Engines::Maxwell3D::Regs::NumRenderTargets; i++) {
+        const auto& source = regs.color_mask[regs.color_mask_common ? 0 : i];
+        auto& dest = state.color_mask[i];
+        dest.red_enabled = (source.R == 0) ? GL_FALSE : GL_TRUE;
+        dest.green_enabled = (source.G == 0) ? GL_FALSE : GL_TRUE;
+        dest.blue_enabled = (source.B == 0) ? GL_FALSE : GL_TRUE;
+        dest.alpha_enabled = (source.A == 0) ? GL_FALSE : GL_TRUE;
+    }
 }
 
 void RasterizerOpenGL::SyncBlendState() {
     const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
 
-    // TODO(Subv): Support more than just render target 0.
-    state.blend.enabled = regs.blend.enable[0] != 0;
+    state.blend_color.red = regs.blend_color.r;
+    state.blend_color.green = regs.blend_color.g;
+    state.blend_color.blue = regs.blend_color.b;
+    state.blend_color.alpha = regs.blend_color.a;
 
-    if (!state.blend.enabled)
+    state.independant_blend.enabled = regs.independent_blend_enable;
+    if (!state.independant_blend.enabled) {
+        auto& blend = state.blend[0];
+        blend.enabled = regs.blend.enable[0] != 0;
+        blend.separate_alpha = regs.blend.separate_alpha;
+        blend.rgb_equation = MaxwellToGL::BlendEquation(regs.blend.equation_rgb);
+        blend.src_rgb_func = MaxwellToGL::BlendFunc(regs.blend.factor_source_rgb);
+        blend.dst_rgb_func = MaxwellToGL::BlendFunc(regs.blend.factor_dest_rgb);
+        if (blend.separate_alpha) {
+            blend.a_equation = MaxwellToGL::BlendEquation(regs.blend.equation_a);
+            blend.src_a_func = MaxwellToGL::BlendFunc(regs.blend.factor_source_a);
+            blend.dst_a_func = MaxwellToGL::BlendFunc(regs.blend.factor_dest_a);
+        }
+        for (size_t i = 1; i < Tegra::Engines::Maxwell3D::Regs::NumRenderTargets; i++) {
+            state.blend[i].enabled = false;
+        }
         return;
+    }
 
-    ASSERT_MSG(regs.logic_op.enable == 0,
-               "Blending and logic op can't be enabled at the same time.");
-
-    ASSERT_MSG(regs.independent_blend_enable == 1, "Only independent blending is implemented");
-    ASSERT_MSG(!regs.independent_blend[0].separate_alpha, "Unimplemented");
-    state.blend.rgb_equation = MaxwellToGL::BlendEquation(regs.independent_blend[0].equation_rgb);
-    state.blend.src_rgb_func = MaxwellToGL::BlendFunc(regs.independent_blend[0].factor_source_rgb);
-    state.blend.dst_rgb_func = MaxwellToGL::BlendFunc(regs.independent_blend[0].factor_dest_rgb);
-    state.blend.a_equation = MaxwellToGL::BlendEquation(regs.independent_blend[0].equation_a);
-    state.blend.src_a_func = MaxwellToGL::BlendFunc(regs.independent_blend[0].factor_source_a);
-    state.blend.dst_a_func = MaxwellToGL::BlendFunc(regs.independent_blend[0].factor_dest_a);
+    for (size_t i = 0; i < Tegra::Engines::Maxwell3D::Regs::NumRenderTargets; i++) {
+        auto& blend = state.blend[i];
+        blend.enabled = regs.blend.enable[i] != 0;
+        if (!blend.enabled)
+            continue;
+        blend.separate_alpha = regs.independent_blend[i].separate_alpha;
+        blend.rgb_equation = MaxwellToGL::BlendEquation(regs.independent_blend[i].equation_rgb);
+        blend.src_rgb_func = MaxwellToGL::BlendFunc(regs.independent_blend[i].factor_source_rgb);
+        blend.dst_rgb_func = MaxwellToGL::BlendFunc(regs.independent_blend[i].factor_dest_rgb);
+        if (blend.separate_alpha) {
+            blend.a_equation = MaxwellToGL::BlendEquation(regs.independent_blend[i].equation_a);
+            blend.src_a_func = MaxwellToGL::BlendFunc(regs.independent_blend[i].factor_source_a);
+            blend.dst_a_func = MaxwellToGL::BlendFunc(regs.independent_blend[i].factor_dest_a);
+        }
+    }
 }
 
 void RasterizerOpenGL::SyncLogicOpState() {
@@ -1009,19 +1092,19 @@ void RasterizerOpenGL::SyncLogicOpState() {
 }
 
 void RasterizerOpenGL::SyncScissorTest() {
+    // TODO: what is the correct behavior here, a single scissor for all targets
+    // or scissor disabled for the rest of the targets?
     const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
-
     state.scissor.enabled = (regs.scissor_test.enable != 0);
-    // TODO(Blinkhawk): Figure if the hardware supports scissor testing per viewport and how it's
-    // implemented.
-    if (regs.scissor_test.enable != 0) {
-        const u32 width = regs.scissor_test.max_x - regs.scissor_test.min_x;
-        const u32 height = regs.scissor_test.max_y - regs.scissor_test.min_y;
-        state.scissor.x = regs.scissor_test.min_x;
-        state.scissor.y = regs.scissor_test.min_y;
-        state.scissor.width = width;
-        state.scissor.height = height;
+    if (regs.scissor_test.enable == 0) {
+        return;
     }
+    const u32 width = regs.scissor_test.max_x - regs.scissor_test.min_x;
+    const u32 height = regs.scissor_test.max_y - regs.scissor_test.min_y;
+    state.scissor.x = regs.scissor_test.min_x;
+    state.scissor.y = regs.scissor_test.min_y;
+    state.scissor.width = width;
+    state.scissor.height = height;
 }
 
 void RasterizerOpenGL::SyncTransformFeedback() {
@@ -1046,9 +1129,8 @@ void RasterizerOpenGL::CheckAlphaTests() {
     const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
 
     if (regs.alpha_test_enabled != 0 && regs.rt_control.count > 1) {
-        LOG_CRITICAL(
-            Render_OpenGL,
-            "Alpha Testing is enabled with Multiple Render Targets, this behavior is undefined.");
+        LOG_CRITICAL(Render_OpenGL, "Alpha Testing is enabled with Multiple Render Targets, "
+                                    "this behavior is undefined.");
         UNREACHABLE();
     }
 }

src/video_core/renderer_opengl/gl_rasterizer.h

@@ -8,12 +8,12 @@
 #include <cstddef>
 #include <map>
 #include <memory>
+#include <optional>
 #include <tuple>
 #include <utility>
 #include <vector>
 
 #include <boost/icl/interval_map.hpp>
-#include <boost/optional.hpp>
 #include <boost/range/iterator_range.hpp>
 #include <glad/glad.h>
 
@@ -88,11 +88,12 @@ private:
         /// SamplerInfo struct.
         void Create();
         /// Syncs the sampler object with the config, updating any necessary state.
-        void SyncWithConfig(const Tegra::Texture::TSCEntry& config);
+        void SyncWithConfig(const Tegra::Texture::FullTextureInfo& info);
 
     private:
         Tegra::Texture::TextureFilter mag_filter;
         Tegra::Texture::TextureFilter min_filter;
+        Tegra::Texture::TextureMipmapFilter mip_filter;
         Tegra::Texture::WrapMode wrap_u;
         Tegra::Texture::WrapMode wrap_v;
         Tegra::Texture::WrapMode wrap_p;
@@ -108,9 +109,9 @@ private:
      * @param preserve_contents If true, tries to preserve data from a previously used framebuffer.
      * @param single_color_target Specifies if a single color buffer target should be used.
      */
-    void ConfigureFramebuffers(bool use_color_fb = true, bool using_depth_fb = true,
-                               bool preserve_contents = true,
-                               boost::optional<std::size_t> single_color_target = {});
+    void ConfigureFramebuffers(OpenGLState& current_state, bool use_color_fb = true,
+                               bool using_depth_fb = true, bool preserve_contents = true,
+                               std::optional<std::size_t> single_color_target = {});
 
     /*
      * Configures the current constbuffers to use for the draw command.
@@ -132,8 +133,8 @@ private:
     u32 SetupTextures(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage, Shader& shader,
                       GLenum primitive_mode, u32 current_unit);
 
-    /// Syncs the viewport to match the guest state
-    void SyncViewport();
+    /// Syncs the viewport and depth range to match the guest state
+    void SyncViewport(OpenGLState& current_state);
 
     /// Syncs the clip enabled status to match the guest state
     void SyncClipEnabled();
@@ -144,8 +145,8 @@ private:
     /// Syncs the cull mode to match the guest state
     void SyncCullMode();
 
-    /// Syncs the depth range to match the guest state
-    void SyncDepthRange();
+    /// Syncs the primitve restart to match the guest state
+    void SyncPrimitiveRestart();
 
     /// Syncs the depth test state to match the guest state
     void SyncDepthTestState();
@@ -168,6 +169,9 @@ private:
     /// Syncs the point state to match the guest state
     void SyncPointState();
 
+    /// Syncs Color Mask
+    void SyncColorMask();
+
     /// Check asserts for alpha testing.
     void CheckAlphaTests();
 
@@ -203,7 +207,8 @@ private:
 
     std::size_t CalculateIndexBufferSize() const;
 
-    void SetupVertexArrays();
+    void SetupVertexFormat();
+    void SetupVertexBuffer();
 
     DrawParameters SetupDraw();
 

src/video_core/renderer_opengl/gl_rasterizer_cache.cpp

@@ -15,16 +15,24 @@
 #include "core/memory.h"
 #include "core/settings.h"
 #include "video_core/engines/maxwell_3d.h"
+#include "video_core/renderer_opengl/gl_rasterizer.h"
 #include "video_core/renderer_opengl/gl_rasterizer_cache.h"
+#include "video_core/renderer_opengl/gl_state.h"
+#include "video_core/renderer_opengl/utils.h"
+#include "video_core/surface.h"
 #include "video_core/textures/astc.h"
 #include "video_core/textures/decoders.h"
 #include "video_core/utils.h"
 
 namespace OpenGL {
 
-using SurfaceType = SurfaceParams::SurfaceType;
-using PixelFormat = SurfaceParams::PixelFormat;
-using ComponentType = SurfaceParams::ComponentType;
+using VideoCore::Surface::ComponentTypeFromDepthFormat;
+using VideoCore::Surface::ComponentTypeFromRenderTarget;
+using VideoCore::Surface::ComponentTypeFromTexture;
+using VideoCore::Surface::PixelFormatFromDepthFormat;
+using VideoCore::Surface::PixelFormatFromRenderTargetFormat;
+using VideoCore::Surface::PixelFormatFromTextureFormat;
+using VideoCore::Surface::SurfaceTargetFromTextureType;
 
 struct FormatTuple {
     GLint internal_format;
@@ -34,34 +42,6 @@ struct FormatTuple {
     bool compressed;
 };
 
-static bool IsPixelFormatASTC(PixelFormat format) {
-    switch (format) {
-    case PixelFormat::ASTC_2D_4X4:
-    case PixelFormat::ASTC_2D_5X4:
-    case PixelFormat::ASTC_2D_8X8:
-    case PixelFormat::ASTC_2D_8X5:
-        return true;
-    default:
-        return false;
-    }
-}
-
-static std::pair<u32, u32> GetASTCBlockSize(PixelFormat format) {
-    switch (format) {
-    case PixelFormat::ASTC_2D_4X4:
-        return {4, 4};
-    case PixelFormat::ASTC_2D_5X4:
-        return {5, 4};
-    case PixelFormat::ASTC_2D_8X8:
-        return {8, 8};
-    case PixelFormat::ASTC_2D_8X5:
-        return {8, 5};
-    default:
-        LOG_CRITICAL(HW_GPU, "Unhandled format: {}", static_cast<u32>(format));
-        UNREACHABLE();
-    }
-}
-
 void SurfaceParams::InitCacheParameters(Tegra::GPUVAddr gpu_addr_) {
     auto& memory_manager{Core::System::GetInstance().GPU().MemoryManager()};
     const auto cpu_addr{memory_manager.GpuToCpuAddress(gpu_addr_)};
@@ -78,27 +58,34 @@ void SurfaceParams::InitCacheParameters(Tegra::GPUVAddr gpu_addr_) {
     }
 }
 
-std::size_t SurfaceParams::InnerMemorySize(bool layer_only) const {
-    const u32 compression_factor{GetCompressionFactor(pixel_format)};
+std::size_t SurfaceParams::InnerMipmapMemorySize(u32 mip_level, bool force_gl, bool layer_only,
+                                                 bool uncompressed) const {
+    const u32 tile_x{GetDefaultBlockWidth(pixel_format)};
+    const u32 tile_y{GetDefaultBlockHeight(pixel_format)};
     const u32 bytes_per_pixel{GetBytesPerPixel(pixel_format)};
     u32 m_depth = (layer_only ? 1U : depth);
-    u32 m_width = std::max(1U, width / compression_factor);
-    u32 m_height = std::max(1U, height / compression_factor);
-    std::size_t size = Tegra::Texture::CalculateSize(is_tiled, bytes_per_pixel, m_width, m_height,
-                                                     m_depth, block_height, block_depth);
-    u32 m_block_height = block_height;
-    u32 m_block_depth = block_depth;
-    std::size_t block_size_bytes = 512 * block_height * block_depth; // 512 is GOB size
-    for (u32 i = 1; i < max_mip_level; i++) {
-        m_width = std::max(1U, m_width / 2);
-        m_height = std::max(1U, m_height / 2);
-        m_depth = std::max(1U, m_depth / 2);
-        m_block_height = std::max(1U, m_block_height / 2);
-        m_block_depth = std::max(1U, m_block_depth / 2);
-        size += Tegra::Texture::CalculateSize(is_tiled, bytes_per_pixel, m_width, m_height, m_depth,
-                                              m_block_height, m_block_depth);
+    u32 m_width = MipWidth(mip_level);
+    u32 m_height = MipHeight(mip_level);
+    m_width = uncompressed ? m_width : std::max(1U, (m_width + tile_x - 1) / tile_x);
+    m_height = uncompressed ? m_height : std::max(1U, (m_height + tile_y - 1) / tile_y);
+    m_depth = std::max(1U, m_depth >> mip_level);
+    u32 m_block_height = MipBlockHeight(mip_level);
+    u32 m_block_depth = MipBlockDepth(mip_level);
+    return Tegra::Texture::CalculateSize(force_gl ? false : is_tiled, bytes_per_pixel, m_width,
+                                         m_height, m_depth, m_block_height, m_block_depth);
+}
+
+std::size_t SurfaceParams::InnerMemorySize(bool force_gl, bool layer_only,
+                                           bool uncompressed) const {
+    std::size_t block_size_bytes = Tegra::Texture::GetGOBSize() * block_height * block_depth;
+    std::size_t size = 0;
+    for (u32 i = 0; i < max_mip_level; i++) {
+        size += InnerMipmapMemorySize(i, force_gl, layer_only, uncompressed);
     }
-    return is_tiled ? Common::AlignUp(size, block_size_bytes) : size;
+    if (!force_gl && is_tiled) {
+        size = Common::AlignUp(size, block_size_bytes);
+    }
+    return size;
 }
 
 /*static*/ SurfaceParams SurfaceParams::CreateForTexture(
@@ -108,8 +95,9 @@ std::size_t SurfaceParams::InnerMemorySize(bool layer_only) const {
     params.block_width = params.is_tiled ? config.tic.BlockWidth() : 0,
     params.block_height = params.is_tiled ? config.tic.BlockHeight() : 0,
     params.block_depth = params.is_tiled ? config.tic.BlockDepth() : 0,
-    params.pixel_format =
-        PixelFormatFromTextureFormat(config.tic.format, config.tic.r_type.Value());
+    params.srgb_conversion = config.tic.IsSrgbConversionEnabled();
+    params.pixel_format = PixelFormatFromTextureFormat(config.tic.format, config.tic.r_type.Value(),
+                                                       params.srgb_conversion);
     params.component_type = ComponentTypeFromTexture(config.tic.r_type.Value());
     params.type = GetFormatType(params.pixel_format);
     params.width = Common::AlignUp(config.tic.Width(), GetCompressionFactor(params.pixel_format));
@@ -140,6 +128,13 @@ std::size_t SurfaceParams::InnerMemorySize(bool layer_only) const {
             params.target = SurfaceTarget::Texture2D;
         }
         break;
+    case SurfaceTarget::TextureCubeArray:
+        params.depth = config.tic.Depth() * 6;
+        if (!entry.IsArray()) {
+            ASSERT(params.depth == 6);
+            params.target = SurfaceTarget::TextureCubemap;
+        }
+        break;
     default:
         LOG_CRITICAL(HW_GPU, "Unknown depth for target={}", static_cast<u32>(params.target));
         UNREACHABLE();
@@ -166,6 +161,8 @@ std::size_t SurfaceParams::InnerMemorySize(bool layer_only) const {
     params.block_height = 1 << config.memory_layout.block_height;
     params.block_depth = 1 << config.memory_layout.block_depth;
     params.pixel_format = PixelFormatFromRenderTargetFormat(config.format);
+    params.srgb_conversion = config.format == Tegra::RenderTargetFormat::BGRA8_SRGB ||
+                             config.format == Tegra::RenderTargetFormat::RGBA8_SRGB;
     params.component_type = ComponentTypeFromRenderTarget(config.format);
     params.type = GetFormatType(params.pixel_format);
     params.width = config.width;
@@ -173,7 +170,7 @@ std::size_t SurfaceParams::InnerMemorySize(bool layer_only) const {
     params.unaligned_height = config.height;
     params.target = SurfaceTarget::Texture2D;
     params.depth = 1;
-    params.max_mip_level = 0;
+    params.max_mip_level = 1;
     params.is_layered = false;
 
     // Render target specific parameters, not used for caching
@@ -201,12 +198,13 @@ std::size_t SurfaceParams::InnerMemorySize(bool layer_only) const {
     params.pixel_format = PixelFormatFromDepthFormat(format);
     params.component_type = ComponentTypeFromDepthFormat(format);
     params.type = GetFormatType(params.pixel_format);
+    params.srgb_conversion = false;
     params.width = zeta_width;
     params.height = zeta_height;
     params.unaligned_height = zeta_height;
     params.target = SurfaceTarget::Texture2D;
     params.depth = 1;
-    params.max_mip_level = 0;
+    params.max_mip_level = 1;
     params.is_layered = false;
     params.rt = {};
 
@@ -224,6 +222,8 @@ std::size_t SurfaceParams::InnerMemorySize(bool layer_only) const {
     params.block_height = params.is_tiled ? std::min(config.BlockHeight(), 32U) : 0,
     params.block_depth = params.is_tiled ? std::min(config.BlockDepth(), 32U) : 0,
     params.pixel_format = PixelFormatFromRenderTargetFormat(config.format);
+    params.srgb_conversion = config.format == Tegra::RenderTargetFormat::BGRA8_SRGB ||
+                             config.format == Tegra::RenderTargetFormat::RGBA8_SRGB;
     params.component_type = ComponentTypeFromRenderTarget(config.format);
     params.type = GetFormatType(params.pixel_format);
     params.width = config.width;
@@ -231,7 +231,7 @@ std::size_t SurfaceParams::InnerMemorySize(bool layer_only) const {
     params.unaligned_height = config.height;
     params.target = SurfaceTarget::Texture2D;
     params.depth = 1;
-    params.max_mip_level = 0;
+    params.max_mip_level = 1;
     params.rt = {};
 
     params.InitCacheParameters(config.Address());
@@ -239,7 +239,7 @@ std::size_t SurfaceParams::InnerMemorySize(bool layer_only) const {
     return params;
 }
 
-static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_format_tuples = {{
+static 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
@@ -255,7 +255,7 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form
     {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_RGB_S3TC_DXT1_EXT, GL_RGB, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
+    {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
@@ -289,14 +289,33 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form
     {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}, // SRGBA8
-    {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_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_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_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
+    // 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_ARB, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8,
+     ComponentType::UNorm, true},                                              // BC7U_SRGB
+    {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
 
     // Depth formats
     {GL_DEPTH_COMPONENT32F, GL_DEPTH_COMPONENT, GL_FLOAT, ComponentType::Float, false}, // Z32F
@@ -312,20 +331,22 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form
      ComponentType::Float, false}, // Z32FS8
 }};
 
-static GLenum SurfaceTargetToGL(SurfaceParams::SurfaceTarget target) {
+static GLenum SurfaceTargetToGL(SurfaceTarget target) {
     switch (target) {
-    case SurfaceParams::SurfaceTarget::Texture1D:
+    case SurfaceTarget::Texture1D:
         return GL_TEXTURE_1D;
-    case SurfaceParams::SurfaceTarget::Texture2D:
+    case SurfaceTarget::Texture2D:
         return GL_TEXTURE_2D;
-    case SurfaceParams::SurfaceTarget::Texture3D:
+    case SurfaceTarget::Texture3D:
         return GL_TEXTURE_3D;
-    case SurfaceParams::SurfaceTarget::Texture1DArray:
+    case SurfaceTarget::Texture1DArray:
         return GL_TEXTURE_1D_ARRAY;
-    case SurfaceParams::SurfaceTarget::Texture2DArray:
+    case SurfaceTarget::Texture2DArray:
         return GL_TEXTURE_2D_ARRAY;
-    case SurfaceParams::SurfaceTarget::TextureCubemap:
+    case SurfaceTarget::TextureCubemap:
         return GL_TEXTURE_CUBE_MAP;
+    case SurfaceTarget::TextureCubeArray:
+        return GL_TEXTURE_CUBE_MAP_ARRAY_ARB;
     }
     LOG_CRITICAL(Render_OpenGL, "Unimplemented texture target={}", static_cast<u32>(target));
     UNREACHABLE();
@@ -340,55 +361,41 @@ static const FormatTuple& GetFormatTuple(PixelFormat pixel_format, ComponentType
     return format;
 }
 
-MathUtil::Rectangle<u32> SurfaceParams::GetRect() const {
-    u32 actual_height{unaligned_height};
+MathUtil::Rectangle<u32> SurfaceParams::GetRect(u32 mip_level) const {
+    u32 actual_height{std::max(1U, unaligned_height >> mip_level)};
     if (IsPixelFormatASTC(pixel_format)) {
         // ASTC formats must stop at the ATSC block size boundary
         actual_height = Common::AlignDown(actual_height, GetASTCBlockSize(pixel_format).second);
     }
-    return {0, actual_height, width, 0};
-}
-
-/// Returns true if the specified PixelFormat is a BCn format, e.g. DXT or DXN
-static bool IsFormatBCn(PixelFormat format) {
-    switch (format) {
-    case PixelFormat::DXT1:
-    case PixelFormat::DXT23:
-    case PixelFormat::DXT45:
-    case PixelFormat::DXN1:
-    case PixelFormat::DXN2SNORM:
-    case PixelFormat::DXN2UNORM:
-    case PixelFormat::BC7U:
-    case PixelFormat::BC6H_UF16:
-    case PixelFormat::BC6H_SF16:
-        return true;
-    }
-    return false;
+    return {0, actual_height, MipWidth(mip_level), 0};
 }
 
 template <bool morton_to_gl, PixelFormat format>
 void MortonCopy(u32 stride, u32 block_height, u32 height, u32 block_depth, u32 depth, u8* gl_buffer,
                 std::size_t gl_buffer_size, VAddr addr) {
-    constexpr u32 bytes_per_pixel = SurfaceParams::GetBytesPerPixel(format);
+    constexpr u32 bytes_per_pixel = GetBytesPerPixel(format);
 
     // With the BCn formats (DXT and DXN), each 4x4 tile is swizzled instead of just individual
     // pixel values.
-    const u32 tile_size{IsFormatBCn(format) ? 4U : 1U};
+    const u32 tile_size_x{GetDefaultBlockWidth(format)};
+    const u32 tile_size_y{GetDefaultBlockHeight(format)};
 
     if (morton_to_gl) {
-        const std::vector<u8> data = Tegra::Texture::UnswizzleTexture(
-            addr, tile_size, bytes_per_pixel, stride, height, depth, block_height, block_depth);
+        const std::vector<u8> data =
+            Tegra::Texture::UnswizzleTexture(addr, tile_size_x, tile_size_y, bytes_per_pixel,
+                                             stride, height, depth, block_height, block_depth);
         const std::size_t size_to_copy{std::min(gl_buffer_size, data.size())};
         memcpy(gl_buffer, data.data(), size_to_copy);
     } else {
-        Tegra::Texture::CopySwizzledData(stride / tile_size, height / tile_size, depth,
+        Tegra::Texture::CopySwizzledData((stride + tile_size_x - 1) / tile_size_x,
+                                         (height + tile_size_y - 1) / tile_size_y, depth,
                                          bytes_per_pixel, bytes_per_pixel, Memory::GetPointer(addr),
                                          gl_buffer, false, block_height, block_depth);
     }
 }
 
 using GLConversionArray = std::array<void (*)(u32, u32, u32, u32, u32, u8*, std::size_t, VAddr),
-                                     SurfaceParams::MaxPixelFormat>;
+                                     VideoCore::Surface::MaxPixelFormat>;
 
 static constexpr GLConversionArray morton_to_gl_fns = {
     // clang-format off
@@ -432,7 +439,7 @@ static constexpr GLConversionArray morton_to_gl_fns = {
         MortonCopy<true, PixelFormat::RG16I>,
         MortonCopy<true, PixelFormat::RG16S>,
         MortonCopy<true, PixelFormat::RGB32F>,
-        MortonCopy<true, PixelFormat::SRGBA8>,
+        MortonCopy<true, PixelFormat::RGBA8_SRGB>,
         MortonCopy<true, PixelFormat::RG8U>,
         MortonCopy<true, PixelFormat::RG8S>,
         MortonCopy<true, PixelFormat::RG32UI>,
@@ -440,6 +447,19 @@ static constexpr GLConversionArray morton_to_gl_fns = {
         MortonCopy<true, PixelFormat::ASTC_2D_8X8>,
         MortonCopy<true, PixelFormat::ASTC_2D_8X5>,
         MortonCopy<true, PixelFormat::ASTC_2D_5X4>,
+        MortonCopy<true, PixelFormat::BGRA8_SRGB>,
+        MortonCopy<true, PixelFormat::DXT1_SRGB>,
+        MortonCopy<true, PixelFormat::DXT23_SRGB>,
+        MortonCopy<true, PixelFormat::DXT45_SRGB>,
+        MortonCopy<true, PixelFormat::BC7U_SRGB>,
+        MortonCopy<true, PixelFormat::ASTC_2D_4X4_SRGB>,
+        MortonCopy<true, PixelFormat::ASTC_2D_8X8_SRGB>,
+        MortonCopy<true, PixelFormat::ASTC_2D_8X5_SRGB>,
+        MortonCopy<true, PixelFormat::ASTC_2D_5X4_SRGB>,
+        MortonCopy<true, PixelFormat::ASTC_2D_5X5>,
+        MortonCopy<true, PixelFormat::ASTC_2D_5X5_SRGB>,
+        MortonCopy<true, PixelFormat::ASTC_2D_10X8>,
+        MortonCopy<true, PixelFormat::ASTC_2D_10X8_SRGB>,
         MortonCopy<true, PixelFormat::Z32F>,
         MortonCopy<true, PixelFormat::Z16>,
         MortonCopy<true, PixelFormat::Z24S8>,
@@ -491,7 +511,7 @@ static constexpr GLConversionArray gl_to_morton_fns = {
         MortonCopy<false, PixelFormat::RG16I>,
         MortonCopy<false, PixelFormat::RG16S>,
         MortonCopy<false, PixelFormat::RGB32F>,
-        MortonCopy<false, PixelFormat::SRGBA8>,
+        MortonCopy<false, PixelFormat::RGBA8_SRGB>,
         MortonCopy<false, PixelFormat::RG8U>,
         MortonCopy<false, PixelFormat::RG8S>,
         MortonCopy<false, PixelFormat::RG32UI>,
@@ -499,6 +519,19 @@ static constexpr GLConversionArray gl_to_morton_fns = {
         nullptr,
         nullptr,
         nullptr,
+        MortonCopy<false, PixelFormat::BGRA8_SRGB>,
+        MortonCopy<false, PixelFormat::DXT1_SRGB>,
+        MortonCopy<false, PixelFormat::DXT23_SRGB>,
+        MortonCopy<false, PixelFormat::DXT45_SRGB>,
+        MortonCopy<false, PixelFormat::BC7U_SRGB>,
+        nullptr,
+        nullptr,
+        nullptr,
+        nullptr,
+        nullptr,
+        nullptr,
+        nullptr,
+        nullptr,
         MortonCopy<false, PixelFormat::Z32F>,
         MortonCopy<false, PixelFormat::Z16>,
         MortonCopy<false, PixelFormat::Z24S8>,
@@ -508,34 +541,39 @@ static constexpr GLConversionArray gl_to_morton_fns = {
 };
 
 void SwizzleFunc(const GLConversionArray& functions, const SurfaceParams& params,
-                 std::vector<u8>& gl_buffer) {
-    u32 depth = params.depth;
-    if (params.target == SurfaceParams::SurfaceTarget::Texture2D) {
+                 std::vector<u8>& gl_buffer, u32 mip_level) {
+    u32 depth = params.MipDepth(mip_level);
+    if (params.target == SurfaceTarget::Texture2D) {
         // TODO(Blinkhawk): Eliminate this condition once all texture types are implemented.
         depth = 1U;
     }
     if (params.is_layered) {
-        u64 offset = 0;
+        u64 offset = params.GetMipmapLevelOffset(mip_level);
         u64 offset_gl = 0;
         u64 layer_size = params.LayerMemorySize();
-        u64 gl_size = params.LayerSizeGL();
-        for (u32 i = 0; i < depth; i++) {
+        u64 gl_size = params.LayerSizeGL(mip_level);
+        for (u32 i = 0; i < params.depth; i++) {
             functions[static_cast<std::size_t>(params.pixel_format)](
-                params.width, params.block_height, params.height, params.block_depth, 1,
+                params.MipWidth(mip_level), params.MipBlockHeight(mip_level),
+                params.MipHeight(mip_level), params.MipBlockDepth(mip_level), 1,
                 gl_buffer.data() + offset_gl, gl_size, params.addr + offset);
             offset += layer_size;
             offset_gl += gl_size;
         }
     } else {
+        u64 offset = params.GetMipmapLevelOffset(mip_level);
         functions[static_cast<std::size_t>(params.pixel_format)](
-            params.width, params.block_height, params.height, params.block_depth, depth,
-            gl_buffer.data(), gl_buffer.size(), params.addr);
+            params.MipWidth(mip_level), params.MipBlockHeight(mip_level),
+            params.MipHeight(mip_level), params.MipBlockDepth(mip_level), depth, gl_buffer.data(),
+            gl_buffer.size(), params.addr + offset);
     }
 }
 
+MICROPROFILE_DEFINE(OpenGL_BlitSurface, "OpenGL", "BlitSurface", MP_RGB(128, 192, 64));
 static bool BlitSurface(const Surface& src_surface, const Surface& dst_surface,
                         GLuint read_fb_handle, GLuint draw_fb_handle, GLenum src_attachment = 0,
                         GLenum dst_attachment = 0, std::size_t cubemap_face = 0) {
+    MICROPROFILE_SCOPE(OpenGL_BlitSurface);
 
     const auto& src_params{src_surface->GetSurfaceParams()};
     const auto& dst_params{dst_surface->GetSurfaceParams()};
@@ -546,19 +584,21 @@ static bool BlitSurface(const Surface& src_surface, const Surface& dst_surface,
     OpenGLState state;
     state.draw.read_framebuffer = read_fb_handle;
     state.draw.draw_framebuffer = draw_fb_handle;
-    state.Apply();
+    // Set sRGB enabled if the destination surfaces need it
+    state.framebuffer_srgb.enabled = dst_params.srgb_conversion;
+    state.ApplyFramebufferState();
 
     u32 buffers{};
 
     if (src_params.type == SurfaceType::ColorTexture) {
         switch (src_params.target) {
-        case SurfaceParams::SurfaceTarget::Texture2D:
+        case SurfaceTarget::Texture2D:
             glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0 + src_attachment,
                                    GL_TEXTURE_2D, src_surface->Texture().handle, 0);
             glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D,
                                    0, 0);
             break;
-        case SurfaceParams::SurfaceTarget::TextureCubemap:
+        case SurfaceTarget::TextureCubemap:
             glFramebufferTexture2D(
                 GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0 + src_attachment,
                 static_cast<GLenum>(GL_TEXTURE_CUBE_MAP_POSITIVE_X + cubemap_face),
@@ -567,12 +607,12 @@ static bool BlitSurface(const Surface& src_surface, const Surface& dst_surface,
                 GL_READ_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT,
                 static_cast<GLenum>(GL_TEXTURE_CUBE_MAP_POSITIVE_X + cubemap_face), 0, 0);
             break;
-        case SurfaceParams::SurfaceTarget::Texture2DArray:
+        case SurfaceTarget::Texture2DArray:
             glFramebufferTextureLayer(GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0 + src_attachment,
                                       src_surface->Texture().handle, 0, 0);
             glFramebufferTextureLayer(GL_READ_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, 0, 0, 0);
             break;
-        case SurfaceParams::SurfaceTarget::Texture3D:
+        case SurfaceTarget::Texture3D:
             glFramebufferTexture3D(GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0 + src_attachment,
                                    SurfaceTargetToGL(src_params.target),
                                    src_surface->Texture().handle, 0, 0);
@@ -588,13 +628,13 @@ static bool BlitSurface(const Surface& src_surface, const Surface& dst_surface,
         }
 
         switch (dst_params.target) {
-        case SurfaceParams::SurfaceTarget::Texture2D:
+        case SurfaceTarget::Texture2D:
             glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0 + dst_attachment,
                                    GL_TEXTURE_2D, dst_surface->Texture().handle, 0);
             glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D,
                                    0, 0);
             break;
-        case SurfaceParams::SurfaceTarget::TextureCubemap:
+        case SurfaceTarget::TextureCubemap:
             glFramebufferTexture2D(
                 GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0 + dst_attachment,
                 static_cast<GLenum>(GL_TEXTURE_CUBE_MAP_POSITIVE_X + cubemap_face),
@@ -603,13 +643,13 @@ static bool BlitSurface(const Surface& src_surface, const Surface& dst_surface,
                 GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT,
                 static_cast<GLenum>(GL_TEXTURE_CUBE_MAP_POSITIVE_X + cubemap_face), 0, 0);
             break;
-        case SurfaceParams::SurfaceTarget::Texture2DArray:
+        case SurfaceTarget::Texture2DArray:
             glFramebufferTextureLayer(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0 + dst_attachment,
                                       dst_surface->Texture().handle, 0, 0);
             glFramebufferTextureLayer(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, 0, 0, 0);
             break;
 
-        case SurfaceParams::SurfaceTarget::Texture3D:
+        case SurfaceTarget::Texture3D:
             glFramebufferTexture3D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0 + dst_attachment,
                                    SurfaceTargetToGL(dst_params.target),
                                    dst_surface->Texture().handle, 0, 0);
@@ -673,9 +713,11 @@ static void FastCopySurface(const Surface& src_surface, const Surface& dst_surfa
                        0, 0, width, height, 1);
 }
 
+MICROPROFILE_DEFINE(OpenGL_CopySurface, "OpenGL", "CopySurface", MP_RGB(128, 192, 64));
 static void CopySurface(const Surface& src_surface, const Surface& dst_surface,
                         GLuint copy_pbo_handle, GLenum src_attachment = 0,
                         GLenum dst_attachment = 0, std::size_t cubemap_face = 0) {
+    MICROPROFILE_SCOPE(OpenGL_CopySurface);
     ASSERT_MSG(dst_attachment == 0, "Unimplemented");
 
     const auto& src_params{src_surface->GetSurfaceParams()};
@@ -730,21 +772,22 @@ static void CopySurface(const Surface& src_surface, const Surface& dst_surface,
         UNREACHABLE();
     } else {
         switch (dst_params.target) {
-        case SurfaceParams::SurfaceTarget::Texture1D:
+        case SurfaceTarget::Texture1D:
             glTextureSubImage1D(dst_surface->Texture().handle, 0, 0, width, dest_format.format,
                                 dest_format.type, nullptr);
             break;
-        case SurfaceParams::SurfaceTarget::Texture2D:
+        case SurfaceTarget::Texture2D:
             glTextureSubImage2D(dst_surface->Texture().handle, 0, 0, 0, width, height,
                                 dest_format.format, dest_format.type, nullptr);
             break;
-        case SurfaceParams::SurfaceTarget::Texture3D:
-        case SurfaceParams::SurfaceTarget::Texture2DArray:
+        case SurfaceTarget::Texture3D:
+        case SurfaceTarget::Texture2DArray:
+        case SurfaceTarget::TextureCubeArray:
             glTextureSubImage3D(dst_surface->Texture().handle, 0, 0, 0, 0, width, height,
                                 static_cast<GLsizei>(dst_params.depth), dest_format.format,
                                 dest_format.type, nullptr);
             break;
-        case SurfaceParams::SurfaceTarget::TextureCubemap:
+        case SurfaceTarget::TextureCubemap:
             glTextureSubImage3D(dst_surface->Texture().handle, 0, 0, 0,
                                 static_cast<GLint>(cubemap_face), width, height, 1,
                                 dest_format.format, dest_format.type, nullptr);
@@ -781,35 +824,43 @@ CachedSurface::CachedSurface(const SurfaceParams& params)
     if (!format_tuple.compressed) {
         // Only pre-create the texture for non-compressed textures.
         switch (params.target) {
-        case SurfaceParams::SurfaceTarget::Texture1D:
-            glTexStorage1D(SurfaceTargetToGL(params.target), 1, format_tuple.internal_format,
-                           rect.GetWidth());
+        case SurfaceTarget::Texture1D:
+            glTexStorage1D(SurfaceTargetToGL(params.target), params.max_mip_level,
+                           format_tuple.internal_format, rect.GetWidth());
             break;
-        case SurfaceParams::SurfaceTarget::Texture2D:
-        case SurfaceParams::SurfaceTarget::TextureCubemap:
-            glTexStorage2D(SurfaceTargetToGL(params.target), 1, format_tuple.internal_format,
-                           rect.GetWidth(), rect.GetHeight());
+        case SurfaceTarget::Texture2D:
+        case SurfaceTarget::TextureCubemap:
+            glTexStorage2D(SurfaceTargetToGL(params.target), params.max_mip_level,
+                           format_tuple.internal_format, rect.GetWidth(), rect.GetHeight());
             break;
-        case SurfaceParams::SurfaceTarget::Texture3D:
-        case SurfaceParams::SurfaceTarget::Texture2DArray:
-            glTexStorage3D(SurfaceTargetToGL(params.target), 1, format_tuple.internal_format,
-                           rect.GetWidth(), rect.GetHeight(), params.depth);
+        case SurfaceTarget::Texture3D:
+        case SurfaceTarget::Texture2DArray:
+        case SurfaceTarget::TextureCubeArray:
+            glTexStorage3D(SurfaceTargetToGL(params.target), params.max_mip_level,
+                           format_tuple.internal_format, rect.GetWidth(), rect.GetHeight(),
+                           params.depth);
             break;
         default:
             LOG_CRITICAL(Render_OpenGL, "Unimplemented surface target={}",
                          static_cast<u32>(params.target));
             UNREACHABLE();
-            glTexStorage2D(GL_TEXTURE_2D, 1, format_tuple.internal_format, rect.GetWidth(),
-                           rect.GetHeight());
+            glTexStorage2D(GL_TEXTURE_2D, params.max_mip_level, format_tuple.internal_format,
+                           rect.GetWidth(), rect.GetHeight());
         }
     }
 
     glTexParameteri(SurfaceTargetToGL(params.target), GL_TEXTURE_MIN_FILTER, GL_LINEAR);
+    glTexParameteri(SurfaceTargetToGL(params.target), GL_TEXTURE_MAG_FILTER, GL_LINEAR);
     glTexParameteri(SurfaceTargetToGL(params.target), GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
     glTexParameteri(SurfaceTargetToGL(params.target), GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
+    glTexParameteri(SurfaceTargetToGL(params.target), GL_TEXTURE_MAX_LEVEL,
+                    params.max_mip_level - 1);
+    if (params.max_mip_level == 1) {
+        glTexParameterf(SurfaceTargetToGL(params.target), GL_TEXTURE_LOD_BIAS, 1000.0);
+    }
 
-    VideoCore::LabelGLObject(GL_TEXTURE, texture.handle, params.addr,
-                             SurfaceParams::SurfaceTargetName(params.target));
+    LabelGLObject(GL_TEXTURE, texture.handle, params.addr,
+                  SurfaceParams::SurfaceTargetName(params.target));
 
     // Clamp size to mapped GPU memory region
     // TODO(bunnei): Super Mario Odyssey maps a 0x40000 byte region and then uses it for a 0x80000
@@ -839,7 +890,7 @@ static void ConvertS8Z24ToZ24S8(std::vector<u8>& data, u32 width, u32 height, bo
 
     S8Z24 s8z24_pixel{};
     Z24S8 z24s8_pixel{};
-    constexpr auto bpp{SurfaceParams::GetBytesPerPixel(PixelFormat::S8Z24)};
+    constexpr auto bpp{GetBytesPerPixel(PixelFormat::S8Z24)};
     for (std::size_t y = 0; y < height; ++y) {
         for (std::size_t x = 0; x < width; ++x) {
             const std::size_t offset{bpp * (y * width + x)};
@@ -859,7 +910,7 @@ static void ConvertS8Z24ToZ24S8(std::vector<u8>& data, u32 width, u32 height, bo
 }
 
 static void ConvertG8R8ToR8G8(std::vector<u8>& data, u32 width, u32 height) {
-    constexpr auto bpp{SurfaceParams::GetBytesPerPixel(PixelFormat::G8R8U)};
+    constexpr auto bpp{GetBytesPerPixel(PixelFormat::G8R8U)};
     for (std::size_t y = 0; y < height; ++y) {
         for (std::size_t x = 0; x < width; ++x) {
             const std::size_t offset{bpp * (y * width + x)};
@@ -876,17 +927,26 @@ static void ConvertG8R8ToR8G8(std::vector<u8>& data, u32 width, u32 height) {
  * typical desktop GPUs.
  */
 static void ConvertFormatAsNeeded_LoadGLBuffer(std::vector<u8>& data, PixelFormat pixel_format,
-                                               u32 width, u32 height) {
+                                               u32 width, u32 height, u32 depth) {
     switch (pixel_format) {
     case PixelFormat::ASTC_2D_4X4:
     case PixelFormat::ASTC_2D_8X8:
     case PixelFormat::ASTC_2D_8X5:
-    case PixelFormat::ASTC_2D_5X4: {
+    case PixelFormat::ASTC_2D_5X4:
+    case PixelFormat::ASTC_2D_5X5:
+    case PixelFormat::ASTC_2D_4X4_SRGB:
+    case PixelFormat::ASTC_2D_8X8_SRGB:
+    case PixelFormat::ASTC_2D_8X5_SRGB:
+    case PixelFormat::ASTC_2D_5X4_SRGB:
+    case PixelFormat::ASTC_2D_5X5_SRGB:
+    case PixelFormat::ASTC_2D_10X8:
+    case PixelFormat::ASTC_2D_10X8_SRGB: {
         // Convert ASTC pixel formats to RGBA8, as most desktop GPUs do not support ASTC.
         u32 block_width{};
         u32 block_height{};
         std::tie(block_width, block_height) = GetASTCBlockSize(pixel_format);
-        data = Tegra::Texture::ASTC::Decompress(data, width, height, block_width, block_height);
+        data =
+            Tegra::Texture::ASTC::Decompress(data, width, height, depth, block_width, block_height);
         break;
     }
     case PixelFormat::S8Z24:
@@ -913,7 +973,13 @@ static void ConvertFormatAsNeeded_FlushGLBuffer(std::vector<u8>& data, PixelForm
     case PixelFormat::G8R8U:
     case PixelFormat::G8R8S:
     case PixelFormat::ASTC_2D_4X4:
-    case PixelFormat::ASTC_2D_8X8: {
+    case PixelFormat::ASTC_2D_8X8:
+    case PixelFormat::ASTC_2D_4X4_SRGB:
+    case PixelFormat::ASTC_2D_8X8_SRGB:
+    case PixelFormat::ASTC_2D_5X5:
+    case PixelFormat::ASTC_2D_5X5_SRGB:
+    case PixelFormat::ASTC_2D_10X8:
+    case PixelFormat::ASTC_2D_10X8_SRGB: {
         LOG_CRITICAL(HW_GPU, "Conversion of format {} after texture flushing is not implemented",
                      static_cast<u32>(pixel_format));
         UNREACHABLE();
@@ -926,23 +992,25 @@ static void ConvertFormatAsNeeded_FlushGLBuffer(std::vector<u8>& data, PixelForm
     }
 }
 
-MICROPROFILE_DEFINE(OpenGL_SurfaceLoad, "OpenGL", "Surface Load", MP_RGB(128, 64, 192));
+MICROPROFILE_DEFINE(OpenGL_SurfaceLoad, "OpenGL", "Surface Load", MP_RGB(128, 192, 64));
 void CachedSurface::LoadGLBuffer() {
     MICROPROFILE_SCOPE(OpenGL_SurfaceLoad);
-
-    gl_buffer.resize(params.size_in_bytes_gl);
+    gl_buffer.resize(params.max_mip_level);
+    for (u32 i = 0; i < params.max_mip_level; i++)
+        gl_buffer[i].resize(params.GetMipmapSizeGL(i));
     if (params.is_tiled) {
         ASSERT_MSG(params.block_width == 1, "Block width is defined as {} on texture type {}",
                    params.block_width, static_cast<u32>(params.target));
-
-        SwizzleFunc(morton_to_gl_fns, params, gl_buffer);
+        for (u32 i = 0; i < params.max_mip_level; i++)
+            SwizzleFunc(morton_to_gl_fns, params, gl_buffer[i], i);
     } else {
         const auto texture_src_data{Memory::GetPointer(params.addr)};
         const auto texture_src_data_end{texture_src_data + params.size_in_bytes_gl};
-        gl_buffer.assign(texture_src_data, texture_src_data_end);
+        gl_buffer[0].assign(texture_src_data, texture_src_data_end);
     }
-
-    ConvertFormatAsNeeded_LoadGLBuffer(gl_buffer, params.pixel_format, params.width, params.height);
+    for (u32 i = 0; i < params.max_mip_level; i++)
+        ConvertFormatAsNeeded_LoadGLBuffer(gl_buffer[i], params.pixel_format, params.MipWidth(i),
+                                           params.MipHeight(i), params.MipDepth(i));
 }
 
 MICROPROFILE_DEFINE(OpenGL_SurfaceFlush, "OpenGL", "Surface Flush", MP_RGB(128, 192, 64));
@@ -952,18 +1020,19 @@ void CachedSurface::FlushGLBuffer() {
     ASSERT_MSG(!IsPixelFormatASTC(params.pixel_format), "Unimplemented");
 
     // OpenGL temporary buffer needs to be big enough to store raw texture size
-    gl_buffer.resize(GetSizeInBytes());
+    gl_buffer.resize(1);
+    gl_buffer[0].resize(GetSizeInBytes());
 
     const FormatTuple& tuple = GetFormatTuple(params.pixel_format, params.component_type);
     // Ensure no bad interactions with GL_UNPACK_ALIGNMENT
-    ASSERT(params.width * SurfaceParams::GetBytesPerPixel(params.pixel_format) % 4 == 0);
+    ASSERT(params.width * GetBytesPerPixel(params.pixel_format) % 4 == 0);
     glPixelStorei(GL_PACK_ROW_LENGTH, static_cast<GLint>(params.width));
     ASSERT(!tuple.compressed);
     glBindBuffer(GL_PIXEL_PACK_BUFFER, 0);
-    glGetTextureImage(texture.handle, 0, tuple.format, tuple.type, gl_buffer.size(),
-                      gl_buffer.data());
+    glGetTextureImage(texture.handle, 0, tuple.format, tuple.type,
+                      static_cast<GLsizei>(gl_buffer[0].size()), gl_buffer[0].data());
     glPixelStorei(GL_PACK_ROW_LENGTH, 0);
-    ConvertFormatAsNeeded_FlushGLBuffer(gl_buffer, params.pixel_format, params.width,
+    ConvertFormatAsNeeded_FlushGLBuffer(gl_buffer[0], params.pixel_format, params.width,
                                         params.height);
     ASSERT(params.type != SurfaceType::Fill);
     const u8* const texture_src_data = Memory::GetPointer(params.addr);
@@ -972,28 +1041,23 @@ void CachedSurface::FlushGLBuffer() {
         ASSERT_MSG(params.block_width == 1, "Block width is defined as {} on texture type {}",
                    params.block_width, static_cast<u32>(params.target));
 
-        SwizzleFunc(gl_to_morton_fns, params, gl_buffer);
+        SwizzleFunc(gl_to_morton_fns, params, gl_buffer[0], 0);
     } else {
-        std::memcpy(Memory::GetPointer(GetAddr()), gl_buffer.data(), GetSizeInBytes());
+        std::memcpy(Memory::GetPointer(GetAddr()), gl_buffer[0].data(), GetSizeInBytes());
     }
 }
 
-MICROPROFILE_DEFINE(OpenGL_TextureUL, "OpenGL", "Texture Upload", MP_RGB(128, 64, 192));
-void CachedSurface::UploadGLTexture(GLuint read_fb_handle, GLuint draw_fb_handle) {
-    if (params.type == SurfaceType::Fill)
-        return;
-
-    MICROPROFILE_SCOPE(OpenGL_TextureUL);
-
-    const auto& rect{params.GetRect()};
+void CachedSurface::UploadGLMipmapTexture(u32 mip_map, GLuint read_fb_handle,
+                                          GLuint draw_fb_handle) {
+    const auto& rect{params.GetRect(mip_map)};
 
     // Load data from memory to the surface
     const GLint x0 = static_cast<GLint>(rect.left);
     const GLint y0 = static_cast<GLint>(rect.bottom);
     std::size_t buffer_offset =
-        static_cast<std::size_t>(static_cast<std::size_t>(y0) * params.width +
+        static_cast<std::size_t>(static_cast<std::size_t>(y0) * params.MipWidth(mip_map) +
                                  static_cast<std::size_t>(x0)) *
-        SurfaceParams::GetBytesPerPixel(params.pixel_format);
+        GetBytesPerPixel(params.pixel_format);
 
     const FormatTuple& tuple = GetFormatTuple(params.pixel_format, params.component_type);
     const GLuint target_tex = texture.handle;
@@ -1009,89 +1073,120 @@ void CachedSurface::UploadGLTexture(GLuint read_fb_handle, GLuint draw_fb_handle
     cur_state.Apply();
 
     // Ensure no bad interactions with GL_UNPACK_ALIGNMENT
-    ASSERT(params.width * SurfaceParams::GetBytesPerPixel(params.pixel_format) % 4 == 0);
-    glPixelStorei(GL_UNPACK_ROW_LENGTH, static_cast<GLint>(params.width));
+    ASSERT(params.MipWidth(mip_map) * GetBytesPerPixel(params.pixel_format) % 4 == 0);
+    glPixelStorei(GL_UNPACK_ROW_LENGTH, static_cast<GLint>(params.MipWidth(mip_map)));
 
+    GLsizei image_size = static_cast<GLsizei>(params.GetMipmapSizeGL(mip_map, false));
     glActiveTexture(GL_TEXTURE0);
     if (tuple.compressed) {
         switch (params.target) {
-        case SurfaceParams::SurfaceTarget::Texture2D:
-            glCompressedTexImage2D(
-                SurfaceTargetToGL(params.target), 0, tuple.internal_format,
-                static_cast<GLsizei>(params.width), static_cast<GLsizei>(params.height), 0,
-                static_cast<GLsizei>(params.size_in_bytes_gl), &gl_buffer[buffer_offset]);
+        case SurfaceTarget::Texture2D:
+            glCompressedTexImage2D(SurfaceTargetToGL(params.target), mip_map, tuple.internal_format,
+                                   static_cast<GLsizei>(params.MipWidth(mip_map)),
+                                   static_cast<GLsizei>(params.MipHeight(mip_map)), 0, image_size,
+                                   &gl_buffer[mip_map][buffer_offset]);
             break;
-        case SurfaceParams::SurfaceTarget::Texture3D:
-        case SurfaceParams::SurfaceTarget::Texture2DArray:
-            glCompressedTexImage3D(
-                SurfaceTargetToGL(params.target), 0, tuple.internal_format,
-                static_cast<GLsizei>(params.width), static_cast<GLsizei>(params.height),
-                static_cast<GLsizei>(params.depth), 0,
-                static_cast<GLsizei>(params.size_in_bytes_gl), &gl_buffer[buffer_offset]);
+        case SurfaceTarget::Texture3D:
+            glCompressedTexImage3D(SurfaceTargetToGL(params.target), mip_map, tuple.internal_format,
+                                   static_cast<GLsizei>(params.MipWidth(mip_map)),
+                                   static_cast<GLsizei>(params.MipHeight(mip_map)),
+                                   static_cast<GLsizei>(params.MipDepth(mip_map)), 0, image_size,
+                                   &gl_buffer[mip_map][buffer_offset]);
             break;
-        case SurfaceParams::SurfaceTarget::TextureCubemap:
+        case SurfaceTarget::Texture2DArray:
+        case SurfaceTarget::TextureCubeArray:
+            glCompressedTexImage3D(SurfaceTargetToGL(params.target), mip_map, tuple.internal_format,
+                                   static_cast<GLsizei>(params.MipWidth(mip_map)),
+                                   static_cast<GLsizei>(params.MipHeight(mip_map)),
+                                   static_cast<GLsizei>(params.depth), 0, image_size,
+                                   &gl_buffer[mip_map][buffer_offset]);
+            break;
+        case SurfaceTarget::TextureCubemap: {
+            GLsizei layer_size = static_cast<GLsizei>(params.LayerSizeGL(mip_map));
             for (std::size_t face = 0; face < params.depth; ++face) {
                 glCompressedTexImage2D(static_cast<GLenum>(GL_TEXTURE_CUBE_MAP_POSITIVE_X + face),
-                                       0, tuple.internal_format, static_cast<GLsizei>(params.width),
-                                       static_cast<GLsizei>(params.height), 0,
-                                       static_cast<GLsizei>(params.SizeInBytesCubeFaceGL()),
-                                       &gl_buffer[buffer_offset]);
-                buffer_offset += params.SizeInBytesCubeFace();
+                                       mip_map, tuple.internal_format,
+                                       static_cast<GLsizei>(params.MipWidth(mip_map)),
+                                       static_cast<GLsizei>(params.MipHeight(mip_map)), 0,
+                                       layer_size, &gl_buffer[mip_map][buffer_offset]);
+                buffer_offset += layer_size;
             }
             break;
+        }
         default:
             LOG_CRITICAL(Render_OpenGL, "Unimplemented surface target={}",
                          static_cast<u32>(params.target));
             UNREACHABLE();
-            glCompressedTexImage2D(
-                GL_TEXTURE_2D, 0, tuple.internal_format, static_cast<GLsizei>(params.width),
-                static_cast<GLsizei>(params.height), 0,
-                static_cast<GLsizei>(params.size_in_bytes_gl), &gl_buffer[buffer_offset]);
+            glCompressedTexImage2D(GL_TEXTURE_2D, mip_map, tuple.internal_format,
+                                   static_cast<GLsizei>(params.MipWidth(mip_map)),
+                                   static_cast<GLsizei>(params.MipHeight(mip_map)), 0,
+                                   static_cast<GLsizei>(params.size_in_bytes_gl),
+                                   &gl_buffer[mip_map][buffer_offset]);
         }
     } else {
 
         switch (params.target) {
-        case SurfaceParams::SurfaceTarget::Texture1D:
-            glTexSubImage1D(SurfaceTargetToGL(params.target), 0, x0,
+        case SurfaceTarget::Texture1D:
+            glTexSubImage1D(SurfaceTargetToGL(params.target), mip_map, x0,
                             static_cast<GLsizei>(rect.GetWidth()), tuple.format, tuple.type,
-                            &gl_buffer[buffer_offset]);
+                            &gl_buffer[mip_map][buffer_offset]);
             break;
-        case SurfaceParams::SurfaceTarget::Texture2D:
-            glTexSubImage2D(SurfaceTargetToGL(params.target), 0, x0, y0,
+        case SurfaceTarget::Texture2D:
+            glTexSubImage2D(SurfaceTargetToGL(params.target), mip_map, x0, y0,
                             static_cast<GLsizei>(rect.GetWidth()),
                             static_cast<GLsizei>(rect.GetHeight()), tuple.format, tuple.type,
-                            &gl_buffer[buffer_offset]);
+                            &gl_buffer[mip_map][buffer_offset]);
             break;
-        case SurfaceParams::SurfaceTarget::Texture3D:
-        case SurfaceParams::SurfaceTarget::Texture2DArray:
-            glTexSubImage3D(SurfaceTargetToGL(params.target), 0, x0, y0, 0,
+        case SurfaceTarget::Texture3D:
+            glTexSubImage3D(SurfaceTargetToGL(params.target), mip_map, x0, y0, 0,
+                            static_cast<GLsizei>(rect.GetWidth()),
+                            static_cast<GLsizei>(rect.GetHeight()), params.MipDepth(mip_map),
+                            tuple.format, tuple.type, &gl_buffer[mip_map][buffer_offset]);
+            break;
+        case SurfaceTarget::Texture2DArray:
+        case SurfaceTarget::TextureCubeArray:
+            glTexSubImage3D(SurfaceTargetToGL(params.target), mip_map, x0, y0, 0,
                             static_cast<GLsizei>(rect.GetWidth()),
                             static_cast<GLsizei>(rect.GetHeight()), params.depth, tuple.format,
-                            tuple.type, &gl_buffer[buffer_offset]);
+                            tuple.type, &gl_buffer[mip_map][buffer_offset]);
             break;
-        case SurfaceParams::SurfaceTarget::TextureCubemap:
+        case SurfaceTarget::TextureCubemap: {
+            std::size_t start = buffer_offset;
             for (std::size_t face = 0; face < params.depth; ++face) {
-                glTexSubImage2D(static_cast<GLenum>(GL_TEXTURE_CUBE_MAP_POSITIVE_X + face), 0, x0,
-                                y0, static_cast<GLsizei>(rect.GetWidth()),
+                glTexSubImage2D(static_cast<GLenum>(GL_TEXTURE_CUBE_MAP_POSITIVE_X + face), mip_map,
+                                x0, y0, static_cast<GLsizei>(rect.GetWidth()),
                                 static_cast<GLsizei>(rect.GetHeight()), tuple.format, tuple.type,
-                                &gl_buffer[buffer_offset]);
-                buffer_offset += params.SizeInBytesCubeFace();
+                                &gl_buffer[mip_map][buffer_offset]);
+                buffer_offset += params.LayerSizeGL(mip_map);
             }
             break;
+        }
         default:
             LOG_CRITICAL(Render_OpenGL, "Unimplemented surface target={}",
                          static_cast<u32>(params.target));
             UNREACHABLE();
-            glTexSubImage2D(GL_TEXTURE_2D, 0, x0, y0, static_cast<GLsizei>(rect.GetWidth()),
+            glTexSubImage2D(GL_TEXTURE_2D, mip_map, x0, y0, static_cast<GLsizei>(rect.GetWidth()),
                             static_cast<GLsizei>(rect.GetHeight()), tuple.format, tuple.type,
-                            &gl_buffer[buffer_offset]);
+                            &gl_buffer[mip_map][buffer_offset]);
         }
     }
 
     glPixelStorei(GL_UNPACK_ROW_LENGTH, 0);
 }
 
-RasterizerCacheOpenGL::RasterizerCacheOpenGL() {
+MICROPROFILE_DEFINE(OpenGL_TextureUL, "OpenGL", "Texture Upload", MP_RGB(128, 192, 64));
+void CachedSurface::UploadGLTexture(GLuint read_fb_handle, GLuint draw_fb_handle) {
+    if (params.type == SurfaceType::Fill)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_TextureUL);
+
+    for (u32 i = 0; i < params.max_mip_level; i++)
+        UploadGLMipmapTexture(i, read_fb_handle, draw_fb_handle);
+}
+
+RasterizerCacheOpenGL::RasterizerCacheOpenGL(RasterizerOpenGL& rasterizer)
+    : RasterizerCache{rasterizer} {
     read_framebuffer.Create();
     draw_framebuffer.Create();
     copy_pbo.Create();
@@ -1231,8 +1326,7 @@ Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& old_surface,
     // For compatible surfaces, we can just do fast glCopyImageSubData based copy
     if (old_params.target == new_params.target && old_params.type == new_params.type &&
         old_params.depth == new_params.depth && old_params.depth == 1 &&
-        SurfaceParams::GetFormatBpp(old_params.pixel_format) ==
-            SurfaceParams::GetFormatBpp(new_params.pixel_format)) {
+        GetFormatBpp(old_params.pixel_format) == GetFormatBpp(new_params.pixel_format)) {
         FastCopySurface(old_surface, new_surface);
         return new_surface;
     }
@@ -1245,15 +1339,17 @@ Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& old_surface,
     const bool is_blit{old_params.pixel_format == new_params.pixel_format};
 
     switch (new_params.target) {
-    case SurfaceParams::SurfaceTarget::Texture2D:
+    case SurfaceTarget::Texture2D:
         if (is_blit) {
             BlitSurface(old_surface, new_surface, read_framebuffer.handle, draw_framebuffer.handle);
         } else {
             CopySurface(old_surface, new_surface, copy_pbo.handle);
         }
         break;
-    case SurfaceParams::SurfaceTarget::TextureCubemap:
-    case SurfaceParams::SurfaceTarget::Texture3D:
+    case SurfaceTarget::TextureCubemap:
+    case SurfaceTarget::Texture3D:
+    case SurfaceTarget::Texture2DArray:
+    case SurfaceTarget::TextureCubeArray:
         AccurateCopySurface(old_surface, new_surface);
         break;
     default:
@@ -1263,7 +1359,7 @@ Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& old_surface,
     }
 
     return new_surface;
-} // namespace OpenGL
+}
 
 Surface RasterizerCacheOpenGL::TryFindFramebufferSurface(VAddr addr) const {
     return TryGet(addr);

src/video_core/renderer_opengl/gl_rasterizer_cache.h

@@ -7,6 +7,7 @@
 #include <array>
 #include <map>
 #include <memory>
+#include <string>
 #include <vector>
 
 #include "common/alignment.h"
@@ -18,6 +19,7 @@
 #include "video_core/rasterizer_cache.h"
 #include "video_core/renderer_opengl/gl_resource_manager.h"
 #include "video_core/renderer_opengl/gl_shader_gen.h"
+#include "video_core/surface.h"
 #include "video_core/textures/decoders.h"
 #include "video_core/textures/texture.h"
 
@@ -27,126 +29,12 @@ class CachedSurface;
 using Surface = std::shared_ptr<CachedSurface>;
 using SurfaceSurfaceRect_Tuple = std::tuple<Surface, Surface, MathUtil::Rectangle<u32>>;
 
+using SurfaceTarget = VideoCore::Surface::SurfaceTarget;
+using SurfaceType = VideoCore::Surface::SurfaceType;
+using PixelFormat = VideoCore::Surface::PixelFormat;
+using ComponentType = VideoCore::Surface::ComponentType;
+
 struct SurfaceParams {
-    enum class PixelFormat {
-        ABGR8U = 0,
-        ABGR8S = 1,
-        ABGR8UI = 2,
-        B5G6R5U = 3,
-        A2B10G10R10U = 4,
-        A1B5G5R5U = 5,
-        R8U = 6,
-        R8UI = 7,
-        RGBA16F = 8,
-        RGBA16U = 9,
-        RGBA16UI = 10,
-        R11FG11FB10F = 11,
-        RGBA32UI = 12,
-        DXT1 = 13,
-        DXT23 = 14,
-        DXT45 = 15,
-        DXN1 = 16, // This is also known as BC4
-        DXN2UNORM = 17,
-        DXN2SNORM = 18,
-        BC7U = 19,
-        BC6H_UF16 = 20,
-        BC6H_SF16 = 21,
-        ASTC_2D_4X4 = 22,
-        G8R8U = 23,
-        G8R8S = 24,
-        BGRA8 = 25,
-        RGBA32F = 26,
-        RG32F = 27,
-        R32F = 28,
-        R16F = 29,
-        R16U = 30,
-        R16S = 31,
-        R16UI = 32,
-        R16I = 33,
-        RG16 = 34,
-        RG16F = 35,
-        RG16UI = 36,
-        RG16I = 37,
-        RG16S = 38,
-        RGB32F = 39,
-        SRGBA8 = 40,
-        RG8U = 41,
-        RG8S = 42,
-        RG32UI = 43,
-        R32UI = 44,
-        ASTC_2D_8X8 = 45,
-        ASTC_2D_8X5 = 46,
-        ASTC_2D_5X4 = 47,
-
-        MaxColorFormat,
-
-        // Depth formats
-        Z32F = 48,
-        Z16 = 49,
-
-        MaxDepthFormat,
-
-        // DepthStencil formats
-        Z24S8 = 50,
-        S8Z24 = 51,
-        Z32FS8 = 52,
-
-        MaxDepthStencilFormat,
-
-        Max = MaxDepthStencilFormat,
-        Invalid = 255,
-    };
-
-    static constexpr std::size_t MaxPixelFormat = static_cast<std::size_t>(PixelFormat::Max);
-
-    enum class ComponentType {
-        Invalid = 0,
-        SNorm = 1,
-        UNorm = 2,
-        SInt = 3,
-        UInt = 4,
-        Float = 5,
-    };
-
-    enum class SurfaceType {
-        ColorTexture = 0,
-        Depth = 1,
-        DepthStencil = 2,
-        Fill = 3,
-        Invalid = 4,
-    };
-
-    enum class SurfaceTarget {
-        Texture1D,
-        Texture2D,
-        Texture3D,
-        Texture1DArray,
-        Texture2DArray,
-        TextureCubemap,
-    };
-
-    static SurfaceTarget SurfaceTargetFromTextureType(Tegra::Texture::TextureType texture_type) {
-        switch (texture_type) {
-        case Tegra::Texture::TextureType::Texture1D:
-            return SurfaceTarget::Texture1D;
-        case Tegra::Texture::TextureType::Texture2D:
-        case Tegra::Texture::TextureType::Texture2DNoMipmap:
-            return SurfaceTarget::Texture2D;
-        case Tegra::Texture::TextureType::Texture3D:
-            return SurfaceTarget::Texture3D;
-        case Tegra::Texture::TextureType::TextureCubemap:
-            return SurfaceTarget::TextureCubemap;
-        case Tegra::Texture::TextureType::Texture1DArray:
-            return SurfaceTarget::Texture1DArray;
-        case Tegra::Texture::TextureType::Texture2DArray:
-            return SurfaceTarget::Texture2DArray;
-        default:
-            LOG_CRITICAL(HW_GPU, "Unimplemented texture_type={}", static_cast<u32>(texture_type));
-            UNREACHABLE();
-            return SurfaceTarget::Texture2D;
-        }
-    }
-
     static std::string SurfaceTargetName(SurfaceTarget target) {
         switch (target) {
         case SurfaceTarget::Texture1D:
@@ -161,6 +49,8 @@ struct SurfaceParams {
             return "Texture2DArray";
         case SurfaceTarget::TextureCubemap:
             return "TextureCubemap";
+        case SurfaceTarget::TextureCubeArray:
+            return "TextureCubeArray";
         default:
             LOG_CRITICAL(HW_GPU, "Unimplemented surface_target={}", static_cast<u32>(target));
             UNREACHABLE();
@@ -168,569 +58,12 @@ struct SurfaceParams {
         }
     }
 
-    static bool SurfaceTargetIsLayered(SurfaceTarget target) {
-        switch (target) {
-        case SurfaceTarget::Texture1D:
-        case SurfaceTarget::Texture2D:
-        case SurfaceTarget::Texture3D:
-            return false;
-        case SurfaceTarget::Texture1DArray:
-        case SurfaceTarget::Texture2DArray:
-        case SurfaceTarget::TextureCubemap:
-            return true;
-        default:
-            LOG_CRITICAL(HW_GPU, "Unimplemented surface_target={}", static_cast<u32>(target));
-            UNREACHABLE();
-            return false;
-        }
-    }
-
-    /**
-     * Gets the compression factor for the specified PixelFormat. This applies to just the
-     * "compressed width" and "compressed height", not the overall compression factor of a
-     * compressed image. This is used for maintaining proper surface sizes for compressed
-     * texture formats.
-     */
-    static constexpr u32 GetCompressionFactor(PixelFormat format) {
-        if (format == PixelFormat::Invalid)
-            return 0;
-
-        constexpr std::array<u32, MaxPixelFormat> compression_factor_table = {{
-            1, // ABGR8U
-            1, // ABGR8S
-            1, // ABGR8UI
-            1, // B5G6R5U
-            1, // A2B10G10R10U
-            1, // A1B5G5R5U
-            1, // R8U
-            1, // R8UI
-            1, // RGBA16F
-            1, // RGBA16U
-            1, // RGBA16UI
-            1, // R11FG11FB10F
-            1, // RGBA32UI
-            4, // DXT1
-            4, // DXT23
-            4, // DXT45
-            4, // DXN1
-            4, // DXN2UNORM
-            4, // DXN2SNORM
-            4, // BC7U
-            4, // BC6H_UF16
-            4, // BC6H_SF16
-            4, // ASTC_2D_4X4
-            1, // G8R8U
-            1, // G8R8S
-            1, // BGRA8
-            1, // RGBA32F
-            1, // RG32F
-            1, // R32F
-            1, // R16F
-            1, // R16U
-            1, // R16S
-            1, // R16UI
-            1, // R16I
-            1, // RG16
-            1, // RG16F
-            1, // RG16UI
-            1, // RG16I
-            1, // RG16S
-            1, // RGB32F
-            1, // SRGBA8
-            1, // RG8U
-            1, // RG8S
-            1, // RG32UI
-            1, // R32UI
-            4, // ASTC_2D_8X8
-            4, // ASTC_2D_8X5
-            4, // ASTC_2D_5X4
-            1, // Z32F
-            1, // Z16
-            1, // Z24S8
-            1, // S8Z24
-            1, // Z32FS8
-        }};
-
-        ASSERT(static_cast<std::size_t>(format) < compression_factor_table.size());
-        return compression_factor_table[static_cast<std::size_t>(format)];
-    }
-
-    static constexpr u32 GetFormatBpp(PixelFormat format) {
-        if (format == PixelFormat::Invalid)
-            return 0;
-
-        constexpr std::array<u32, MaxPixelFormat> bpp_table = {{
-            32,  // ABGR8U
-            32,  // ABGR8S
-            32,  // ABGR8UI
-            16,  // B5G6R5U
-            32,  // A2B10G10R10U
-            16,  // A1B5G5R5U
-            8,   // R8U
-            8,   // R8UI
-            64,  // RGBA16F
-            64,  // RGBA16U
-            64,  // RGBA16UI
-            32,  // R11FG11FB10F
-            128, // RGBA32UI
-            64,  // DXT1
-            128, // DXT23
-            128, // DXT45
-            64,  // DXN1
-            128, // DXN2UNORM
-            128, // DXN2SNORM
-            128, // BC7U
-            128, // BC6H_UF16
-            128, // BC6H_SF16
-            32,  // ASTC_2D_4X4
-            16,  // G8R8U
-            16,  // G8R8S
-            32,  // BGRA8
-            128, // RGBA32F
-            64,  // RG32F
-            32,  // R32F
-            16,  // R16F
-            16,  // R16U
-            16,  // R16S
-            16,  // R16UI
-            16,  // R16I
-            32,  // RG16
-            32,  // RG16F
-            32,  // RG16UI
-            32,  // RG16I
-            32,  // RG16S
-            96,  // RGB32F
-            32,  // SRGBA8
-            16,  // RG8U
-            16,  // RG8S
-            64,  // RG32UI
-            32,  // R32UI
-            16,  // ASTC_2D_8X8
-            32,  // ASTC_2D_8X5
-            32,  // ASTC_2D_5X4
-            32,  // Z32F
-            16,  // Z16
-            32,  // Z24S8
-            32,  // S8Z24
-            64,  // Z32FS8
-        }};
-
-        ASSERT(static_cast<std::size_t>(format) < bpp_table.size());
-        return bpp_table[static_cast<std::size_t>(format)];
-    }
-
     u32 GetFormatBpp() const {
-        return GetFormatBpp(pixel_format);
-    }
-
-    static PixelFormat PixelFormatFromDepthFormat(Tegra::DepthFormat format) {
-        switch (format) {
-        case Tegra::DepthFormat::S8_Z24_UNORM:
-            return PixelFormat::S8Z24;
-        case Tegra::DepthFormat::Z24_S8_UNORM:
-            return PixelFormat::Z24S8;
-        case Tegra::DepthFormat::Z32_FLOAT:
-            return PixelFormat::Z32F;
-        case Tegra::DepthFormat::Z16_UNORM:
-            return PixelFormat::Z16;
-        case Tegra::DepthFormat::Z32_S8_X24_FLOAT:
-            return PixelFormat::Z32FS8;
-        default:
-            LOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
-            UNREACHABLE();
-        }
-    }
-
-    static PixelFormat PixelFormatFromRenderTargetFormat(Tegra::RenderTargetFormat format) {
-        switch (format) {
-        // TODO (Hexagon12): Converting SRGBA to RGBA is a hack and doesn't completely correct the
-        // gamma.
-        case Tegra::RenderTargetFormat::RGBA8_SRGB:
-        case Tegra::RenderTargetFormat::RGBA8_UNORM:
-            return PixelFormat::ABGR8U;
-        case Tegra::RenderTargetFormat::RGBA8_SNORM:
-            return PixelFormat::ABGR8S;
-        case Tegra::RenderTargetFormat::RGBA8_UINT:
-            return PixelFormat::ABGR8UI;
-        case Tegra::RenderTargetFormat::BGRA8_SRGB:
-        case Tegra::RenderTargetFormat::BGRA8_UNORM:
-            return PixelFormat::BGRA8;
-        case Tegra::RenderTargetFormat::RGB10_A2_UNORM:
-            return PixelFormat::A2B10G10R10U;
-        case Tegra::RenderTargetFormat::RGBA16_FLOAT:
-            return PixelFormat::RGBA16F;
-        case Tegra::RenderTargetFormat::RGBA16_UNORM:
-            return PixelFormat::RGBA16U;
-        case Tegra::RenderTargetFormat::RGBA16_UINT:
-            return PixelFormat::RGBA16UI;
-        case Tegra::RenderTargetFormat::RGBA32_FLOAT:
-            return PixelFormat::RGBA32F;
-        case Tegra::RenderTargetFormat::RG32_FLOAT:
-            return PixelFormat::RG32F;
-        case Tegra::RenderTargetFormat::R11G11B10_FLOAT:
-            return PixelFormat::R11FG11FB10F;
-        case Tegra::RenderTargetFormat::B5G6R5_UNORM:
-            return PixelFormat::B5G6R5U;
-        case Tegra::RenderTargetFormat::BGR5A1_UNORM:
-            return PixelFormat::A1B5G5R5U;
-        case Tegra::RenderTargetFormat::RGBA32_UINT:
-            return PixelFormat::RGBA32UI;
-        case Tegra::RenderTargetFormat::R8_UNORM:
-            return PixelFormat::R8U;
-        case Tegra::RenderTargetFormat::R8_UINT:
-            return PixelFormat::R8UI;
-        case Tegra::RenderTargetFormat::RG16_FLOAT:
-            return PixelFormat::RG16F;
-        case Tegra::RenderTargetFormat::RG16_UINT:
-            return PixelFormat::RG16UI;
-        case Tegra::RenderTargetFormat::RG16_SINT:
-            return PixelFormat::RG16I;
-        case Tegra::RenderTargetFormat::RG16_UNORM:
-            return PixelFormat::RG16;
-        case Tegra::RenderTargetFormat::RG16_SNORM:
-            return PixelFormat::RG16S;
-        case Tegra::RenderTargetFormat::RG8_UNORM:
-            return PixelFormat::RG8U;
-        case Tegra::RenderTargetFormat::RG8_SNORM:
-            return PixelFormat::RG8S;
-        case Tegra::RenderTargetFormat::R16_FLOAT:
-            return PixelFormat::R16F;
-        case Tegra::RenderTargetFormat::R16_UNORM:
-            return PixelFormat::R16U;
-        case Tegra::RenderTargetFormat::R16_SNORM:
-            return PixelFormat::R16S;
-        case Tegra::RenderTargetFormat::R16_UINT:
-            return PixelFormat::R16UI;
-        case Tegra::RenderTargetFormat::R16_SINT:
-            return PixelFormat::R16I;
-        case Tegra::RenderTargetFormat::R32_FLOAT:
-            return PixelFormat::R32F;
-        case Tegra::RenderTargetFormat::R32_UINT:
-            return PixelFormat::R32UI;
-        case Tegra::RenderTargetFormat::RG32_UINT:
-            return PixelFormat::RG32UI;
-        default:
-            LOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
-            UNREACHABLE();
-        }
-    }
-
-    static PixelFormat PixelFormatFromTextureFormat(Tegra::Texture::TextureFormat format,
-                                                    Tegra::Texture::ComponentType component_type) {
-        // TODO(Subv): Properly implement this
-        switch (format) {
-        case Tegra::Texture::TextureFormat::A8R8G8B8:
-            switch (component_type) {
-            case Tegra::Texture::ComponentType::UNORM:
-                return PixelFormat::ABGR8U;
-            case Tegra::Texture::ComponentType::SNORM:
-                return PixelFormat::ABGR8S;
-            case Tegra::Texture::ComponentType::UINT:
-                return PixelFormat::ABGR8UI;
-            }
-            LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
-                         static_cast<u32>(component_type));
-            UNREACHABLE();
-        case Tegra::Texture::TextureFormat::B5G6R5:
-            switch (component_type) {
-            case Tegra::Texture::ComponentType::UNORM:
-                return PixelFormat::B5G6R5U;
-            }
-            LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
-                         static_cast<u32>(component_type));
-            UNREACHABLE();
-        case Tegra::Texture::TextureFormat::A2B10G10R10:
-            switch (component_type) {
-            case Tegra::Texture::ComponentType::UNORM:
-                return PixelFormat::A2B10G10R10U;
-            }
-            LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
-                         static_cast<u32>(component_type));
-            UNREACHABLE();
-        case Tegra::Texture::TextureFormat::A1B5G5R5:
-            switch (component_type) {
-            case Tegra::Texture::ComponentType::UNORM:
-                return PixelFormat::A1B5G5R5U;
-            }
-            LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
-                         static_cast<u32>(component_type));
-            UNREACHABLE();
-        case Tegra::Texture::TextureFormat::R8:
-            switch (component_type) {
-            case Tegra::Texture::ComponentType::UNORM:
-                return PixelFormat::R8U;
-            case Tegra::Texture::ComponentType::UINT:
-                return PixelFormat::R8UI;
-            }
-            LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
-                         static_cast<u32>(component_type));
-            UNREACHABLE();
-        case Tegra::Texture::TextureFormat::G8R8:
-            switch (component_type) {
-            case Tegra::Texture::ComponentType::UNORM:
-                return PixelFormat::G8R8U;
-            case Tegra::Texture::ComponentType::SNORM:
-                return PixelFormat::G8R8S;
-            }
-            LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
-                         static_cast<u32>(component_type));
-            UNREACHABLE();
-        case Tegra::Texture::TextureFormat::R16_G16_B16_A16:
-            switch (component_type) {
-            case Tegra::Texture::ComponentType::UNORM:
-                return PixelFormat::RGBA16U;
-            case Tegra::Texture::ComponentType::FLOAT:
-                return PixelFormat::RGBA16F;
-            }
-            LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
-                         static_cast<u32>(component_type));
-            UNREACHABLE();
-        case Tegra::Texture::TextureFormat::BF10GF11RF11:
-            switch (component_type) {
-            case Tegra::Texture::ComponentType::FLOAT:
-                return PixelFormat::R11FG11FB10F;
-            }
-            LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
-                         static_cast<u32>(component_type));
-            UNREACHABLE();
-        case Tegra::Texture::TextureFormat::R32_G32_B32_A32:
-            switch (component_type) {
-            case Tegra::Texture::ComponentType::FLOAT:
-                return PixelFormat::RGBA32F;
-            case Tegra::Texture::ComponentType::UINT:
-                return PixelFormat::RGBA32UI;
-            }
-            LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
-                         static_cast<u32>(component_type));
-            UNREACHABLE();
-        case Tegra::Texture::TextureFormat::R32_G32:
-            switch (component_type) {
-            case Tegra::Texture::ComponentType::FLOAT:
-                return PixelFormat::RG32F;
-            case Tegra::Texture::ComponentType::UINT:
-                return PixelFormat::RG32UI;
-            }
-            LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
-                         static_cast<u32>(component_type));
-            UNREACHABLE();
-        case Tegra::Texture::TextureFormat::R32_G32_B32:
-            switch (component_type) {
-            case Tegra::Texture::ComponentType::FLOAT:
-                return PixelFormat::RGB32F;
-            }
-            LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
-                         static_cast<u32>(component_type));
-            UNREACHABLE();
-        case Tegra::Texture::TextureFormat::R16:
-            switch (component_type) {
-            case Tegra::Texture::ComponentType::FLOAT:
-                return PixelFormat::R16F;
-            case Tegra::Texture::ComponentType::UNORM:
-                return PixelFormat::R16U;
-            case Tegra::Texture::ComponentType::SNORM:
-                return PixelFormat::R16S;
-            case Tegra::Texture::ComponentType::UINT:
-                return PixelFormat::R16UI;
-            case Tegra::Texture::ComponentType::SINT:
-                return PixelFormat::R16I;
-            }
-            LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
-                         static_cast<u32>(component_type));
-            UNREACHABLE();
-        case Tegra::Texture::TextureFormat::R32:
-            switch (component_type) {
-            case Tegra::Texture::ComponentType::FLOAT:
-                return PixelFormat::R32F;
-            case Tegra::Texture::ComponentType::UINT:
-                return PixelFormat::R32UI;
-            }
-            LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
-                         static_cast<u32>(component_type));
-            UNREACHABLE();
-        case Tegra::Texture::TextureFormat::ZF32:
-            return PixelFormat::Z32F;
-        case Tegra::Texture::TextureFormat::Z16:
-            return PixelFormat::Z16;
-        case Tegra::Texture::TextureFormat::Z24S8:
-            return PixelFormat::Z24S8;
-        case Tegra::Texture::TextureFormat::DXT1:
-            return PixelFormat::DXT1;
-        case Tegra::Texture::TextureFormat::DXT23:
-            return PixelFormat::DXT23;
-        case Tegra::Texture::TextureFormat::DXT45:
-            return PixelFormat::DXT45;
-        case Tegra::Texture::TextureFormat::DXN1:
-            return PixelFormat::DXN1;
-        case Tegra::Texture::TextureFormat::DXN2:
-            switch (component_type) {
-            case Tegra::Texture::ComponentType::UNORM:
-                return PixelFormat::DXN2UNORM;
-            case Tegra::Texture::ComponentType::SNORM:
-                return PixelFormat::DXN2SNORM;
-            }
-            LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
-                         static_cast<u32>(component_type));
-            UNREACHABLE();
-        case Tegra::Texture::TextureFormat::BC7U:
-            return PixelFormat::BC7U;
-        case Tegra::Texture::TextureFormat::BC6H_UF16:
-            return PixelFormat::BC6H_UF16;
-        case Tegra::Texture::TextureFormat::BC6H_SF16:
-            return PixelFormat::BC6H_SF16;
-        case Tegra::Texture::TextureFormat::ASTC_2D_4X4:
-            return PixelFormat::ASTC_2D_4X4;
-        case Tegra::Texture::TextureFormat::ASTC_2D_5X4:
-            return PixelFormat::ASTC_2D_5X4;
-        case Tegra::Texture::TextureFormat::ASTC_2D_8X8:
-            return PixelFormat::ASTC_2D_8X8;
-        case Tegra::Texture::TextureFormat::ASTC_2D_8X5:
-            return PixelFormat::ASTC_2D_8X5;
-        case Tegra::Texture::TextureFormat::R16_G16:
-            switch (component_type) {
-            case Tegra::Texture::ComponentType::FLOAT:
-                return PixelFormat::RG16F;
-            case Tegra::Texture::ComponentType::UNORM:
-                return PixelFormat::RG16;
-            case Tegra::Texture::ComponentType::SNORM:
-                return PixelFormat::RG16S;
-            case Tegra::Texture::ComponentType::UINT:
-                return PixelFormat::RG16UI;
-            case Tegra::Texture::ComponentType::SINT:
-                return PixelFormat::RG16I;
-            }
-            LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
-                         static_cast<u32>(component_type));
-            UNREACHABLE();
-        default:
-            LOG_CRITICAL(HW_GPU, "Unimplemented format={}, component_type={}",
-                         static_cast<u32>(format), static_cast<u32>(component_type));
-            UNREACHABLE();
-        }
-    }
-
-    static ComponentType ComponentTypeFromTexture(Tegra::Texture::ComponentType type) {
-        // TODO(Subv): Implement more component types
-        switch (type) {
-        case Tegra::Texture::ComponentType::UNORM:
-            return ComponentType::UNorm;
-        case Tegra::Texture::ComponentType::FLOAT:
-            return ComponentType::Float;
-        case Tegra::Texture::ComponentType::SNORM:
-            return ComponentType::SNorm;
-        case Tegra::Texture::ComponentType::UINT:
-            return ComponentType::UInt;
-        case Tegra::Texture::ComponentType::SINT:
-            return ComponentType::SInt;
-        default:
-            LOG_CRITICAL(HW_GPU, "Unimplemented component type={}", static_cast<u32>(type));
-            UNREACHABLE();
-        }
-    }
-
-    static ComponentType ComponentTypeFromRenderTarget(Tegra::RenderTargetFormat format) {
-        // TODO(Subv): Implement more render targets
-        switch (format) {
-        case Tegra::RenderTargetFormat::RGBA8_UNORM:
-        case Tegra::RenderTargetFormat::RGBA8_SRGB:
-        case Tegra::RenderTargetFormat::BGRA8_UNORM:
-        case Tegra::RenderTargetFormat::BGRA8_SRGB:
-        case Tegra::RenderTargetFormat::RGB10_A2_UNORM:
-        case Tegra::RenderTargetFormat::R8_UNORM:
-        case Tegra::RenderTargetFormat::RG16_UNORM:
-        case Tegra::RenderTargetFormat::R16_UNORM:
-        case Tegra::RenderTargetFormat::B5G6R5_UNORM:
-        case Tegra::RenderTargetFormat::BGR5A1_UNORM:
-        case Tegra::RenderTargetFormat::RG8_UNORM:
-        case Tegra::RenderTargetFormat::RGBA16_UNORM:
-            return ComponentType::UNorm;
-        case Tegra::RenderTargetFormat::RGBA8_SNORM:
-        case Tegra::RenderTargetFormat::RG16_SNORM:
-        case Tegra::RenderTargetFormat::R16_SNORM:
-        case Tegra::RenderTargetFormat::RG8_SNORM:
-            return ComponentType::SNorm;
-        case Tegra::RenderTargetFormat::RGBA16_FLOAT:
-        case Tegra::RenderTargetFormat::R11G11B10_FLOAT:
-        case Tegra::RenderTargetFormat::RGBA32_FLOAT:
-        case Tegra::RenderTargetFormat::RG32_FLOAT:
-        case Tegra::RenderTargetFormat::RG16_FLOAT:
-        case Tegra::RenderTargetFormat::R16_FLOAT:
-        case Tegra::RenderTargetFormat::R32_FLOAT:
-            return ComponentType::Float;
-        case Tegra::RenderTargetFormat::RGBA32_UINT:
-        case Tegra::RenderTargetFormat::RGBA16_UINT:
-        case Tegra::RenderTargetFormat::RG16_UINT:
-        case Tegra::RenderTargetFormat::R8_UINT:
-        case Tegra::RenderTargetFormat::R16_UINT:
-        case Tegra::RenderTargetFormat::RG32_UINT:
-        case Tegra::RenderTargetFormat::R32_UINT:
-        case Tegra::RenderTargetFormat::RGBA8_UINT:
-            return ComponentType::UInt;
-        case Tegra::RenderTargetFormat::RG16_SINT:
-        case Tegra::RenderTargetFormat::R16_SINT:
-            return ComponentType::SInt;
-        default:
-            LOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
-            UNREACHABLE();
-        }
-    }
-
-    static PixelFormat PixelFormatFromGPUPixelFormat(Tegra::FramebufferConfig::PixelFormat format) {
-        switch (format) {
-        case Tegra::FramebufferConfig::PixelFormat::ABGR8:
-            return PixelFormat::ABGR8U;
-        default:
-            LOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
-            UNREACHABLE();
-        }
-    }
-
-    static ComponentType ComponentTypeFromDepthFormat(Tegra::DepthFormat format) {
-        switch (format) {
-        case Tegra::DepthFormat::Z16_UNORM:
-        case Tegra::DepthFormat::S8_Z24_UNORM:
-        case Tegra::DepthFormat::Z24_S8_UNORM:
-            return ComponentType::UNorm;
-        case Tegra::DepthFormat::Z32_FLOAT:
-        case Tegra::DepthFormat::Z32_S8_X24_FLOAT:
-            return ComponentType::Float;
-        default:
-            LOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
-            UNREACHABLE();
-        }
-    }
-
-    static SurfaceType GetFormatType(PixelFormat pixel_format) {
-        if (static_cast<std::size_t>(pixel_format) <
-            static_cast<std::size_t>(PixelFormat::MaxColorFormat)) {
-            return SurfaceType::ColorTexture;
-        }
-
-        if (static_cast<std::size_t>(pixel_format) <
-            static_cast<std::size_t>(PixelFormat::MaxDepthFormat)) {
-            return SurfaceType::Depth;
-        }
-
-        if (static_cast<std::size_t>(pixel_format) <
-            static_cast<std::size_t>(PixelFormat::MaxDepthStencilFormat)) {
-            return SurfaceType::DepthStencil;
-        }
-
-        // TODO(Subv): Implement the other formats
-        ASSERT(false);
-
-        return SurfaceType::Invalid;
-    }
-
-    /// Returns the sizer in bytes of the specified pixel format
-    static constexpr u32 GetBytesPerPixel(PixelFormat pixel_format) {
-        if (pixel_format == SurfaceParams::PixelFormat::Invalid) {
-            return 0;
-        }
-        return GetFormatBpp(pixel_format) / CHAR_BIT;
+        return VideoCore::Surface::GetFormatBpp(pixel_format);
     }
 
     /// Returns the rectangle corresponding to this surface
-    MathUtil::Rectangle<u32> GetRect() const;
+    MathUtil::Rectangle<u32> GetRect(u32 mip_level = 0) const;
 
     /// Returns the total size of this surface in bytes, adjusted for compression
     std::size_t SizeInBytesRaw(bool ignore_tiled = false) const {
@@ -761,7 +94,7 @@ struct SurfaceParams {
 
     /// Returns the exact size of memory occupied by the texture in VRAM, including mipmaps.
     std::size_t MemorySize() const {
-        std::size_t size = InnerMemorySize(is_layered);
+        std::size_t size = InnerMemorySize(false, is_layered);
         if (is_layered)
             return size * depth;
         return size;
@@ -770,12 +103,78 @@ struct SurfaceParams {
     /// Returns the exact size of the memory occupied by a layer in a texture in VRAM, including
     /// mipmaps.
     std::size_t LayerMemorySize() const {
-        return InnerMemorySize(true);
+        return InnerMemorySize(false, true);
     }
 
     /// Returns the size of a layer of this surface in OpenGL.
-    std::size_t LayerSizeGL() const {
-        return SizeInBytesRaw(true) / depth;
+    std::size_t LayerSizeGL(u32 mip_level) const {
+        return InnerMipmapMemorySize(mip_level, true, is_layered, false);
+    }
+
+    std::size_t GetMipmapSizeGL(u32 mip_level, bool ignore_compressed = true) const {
+        std::size_t size = InnerMipmapMemorySize(mip_level, true, is_layered, ignore_compressed);
+        if (is_layered)
+            return size * depth;
+        return size;
+    }
+
+    std::size_t GetMipmapLevelOffset(u32 mip_level) const {
+        std::size_t offset = 0;
+        for (u32 i = 0; i < mip_level; i++)
+            offset += InnerMipmapMemorySize(i, false, is_layered);
+        return offset;
+    }
+
+    std::size_t GetMipmapLevelOffsetGL(u32 mip_level) const {
+        std::size_t offset = 0;
+        for (u32 i = 0; i < mip_level; i++)
+            offset += InnerMipmapMemorySize(i, true, is_layered);
+        return offset;
+    }
+
+    u32 MipWidth(u32 mip_level) const {
+        return std::max(1U, width >> mip_level);
+    }
+
+    u32 MipHeight(u32 mip_level) const {
+        return std::max(1U, height >> mip_level);
+    }
+
+    u32 MipDepth(u32 mip_level) const {
+        return is_layered ? depth : std::max(1U, depth >> mip_level);
+    }
+
+    // Auto block resizing algorithm from:
+    // https://cgit.freedesktop.org/mesa/mesa/tree/src/gallium/drivers/nouveau/nv50/nv50_miptree.c
+    u32 MipBlockHeight(u32 mip_level) const {
+        if (mip_level == 0)
+            return block_height;
+        u32 alt_height = MipHeight(mip_level);
+        u32 h = GetDefaultBlockHeight(pixel_format);
+        u32 blocks_in_y = (alt_height + h - 1) / h;
+        u32 bh = 16;
+        while (bh > 1 && blocks_in_y <= bh * 4) {
+            bh >>= 1;
+        }
+        return bh;
+    }
+
+    u32 MipBlockDepth(u32 mip_level) const {
+        if (mip_level == 0)
+            return block_depth;
+        if (is_layered)
+            return 1;
+        u32 depth = MipDepth(mip_level);
+        u32 bd = 32;
+        while (bd > 1 && depth * 2 <= bd) {
+            bd >>= 1;
+        }
+        if (bd == 32) {
+            u32 bh = MipBlockHeight(mip_level);
+            if (bh >= 4)
+                return 16;
+        }
+        return bd;
     }
 
     /// Creates SurfaceParams from a texture configuration
@@ -819,7 +218,7 @@ struct SurfaceParams {
     SurfaceTarget target;
     u32 max_mip_level;
     bool is_layered;
-
+    bool srgb_conversion;
     // Parameters used for caching
     VAddr addr;
     Tegra::GPUVAddr gpu_addr;
@@ -836,7 +235,10 @@ struct SurfaceParams {
     } rt;
 
 private:
-    std::size_t InnerMemorySize(bool layer_only = false) const;
+    std::size_t InnerMipmapMemorySize(u32 mip_level, bool force_gl = false, bool layer_only = false,
+                                      bool uncompressed = false) const;
+    std::size_t InnerMemorySize(bool force_gl = false, bool layer_only = false,
+                                bool uncompressed = false) const;
 };
 
 }; // namespace OpenGL
@@ -862,6 +264,8 @@ struct hash<SurfaceReserveKey> {
 
 namespace OpenGL {
 
+class RasterizerOpenGL;
+
 class CachedSurface final : public RasterizerCacheObject {
 public:
     CachedSurface(const SurfaceParams& params);
@@ -898,8 +302,10 @@ public:
     void UploadGLTexture(GLuint read_fb_handle, GLuint draw_fb_handle);
 
 private:
+    void UploadGLMipmapTexture(u32 mip_map, GLuint read_fb_handle, GLuint draw_fb_handle);
+
     OGLTexture texture;
-    std::vector<u8> gl_buffer;
+    std::vector<std::vector<u8>> gl_buffer;
     SurfaceParams params;
     GLenum gl_target;
     std::size_t cached_size_in_bytes;
@@ -907,7 +313,7 @@ private:
 
 class RasterizerCacheOpenGL final : public RasterizerCache<Surface> {
 public:
-    RasterizerCacheOpenGL();
+    explicit RasterizerCacheOpenGL(RasterizerOpenGL& rasterizer);
 
     /// Get a surface based on the texture configuration
     Surface GetTextureSurface(const Tegra::Texture::FullTextureInfo& config,

src/video_core/renderer_opengl/gl_resource_manager.cpp

@@ -0,0 +1,186 @@
+// Copyright 2015 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <utility>
+#include <glad/glad.h>
+#include "common/common_types.h"
+#include "common/microprofile.h"
+#include "video_core/renderer_opengl/gl_resource_manager.h"
+#include "video_core/renderer_opengl/gl_shader_util.h"
+#include "video_core/renderer_opengl/gl_state.h"
+
+MICROPROFILE_DEFINE(OpenGL_ResourceCreation, "OpenGL", "Resource Creation", MP_RGB(128, 128, 192));
+MICROPROFILE_DEFINE(OpenGL_ResourceDeletion, "OpenGL", "Resource Deletion", MP_RGB(128, 128, 192));
+
+namespace OpenGL {
+
+void OGLTexture::Create() {
+    if (handle != 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceCreation);
+    glGenTextures(1, &handle);
+}
+
+void OGLTexture::Release() {
+    if (handle == 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceDeletion);
+    glDeleteTextures(1, &handle);
+    OpenGLState::GetCurState().UnbindTexture(handle).Apply();
+    handle = 0;
+}
+
+void OGLSampler::Create() {
+    if (handle != 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceCreation);
+    glGenSamplers(1, &handle);
+}
+
+void OGLSampler::Release() {
+    if (handle == 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceDeletion);
+    glDeleteSamplers(1, &handle);
+    OpenGLState::GetCurState().ResetSampler(handle).Apply();
+    handle = 0;
+}
+
+void OGLShader::Create(const char* source, GLenum type) {
+    if (handle != 0)
+        return;
+    if (source == nullptr)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceCreation);
+    handle = GLShader::LoadShader(source, type);
+}
+
+void OGLShader::Release() {
+    if (handle == 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceDeletion);
+    glDeleteShader(handle);
+    handle = 0;
+}
+
+void OGLProgram::CreateFromSource(const char* vert_shader, const char* geo_shader,
+                                  const char* frag_shader, bool separable_program) {
+    OGLShader vert, geo, frag;
+    if (vert_shader)
+        vert.Create(vert_shader, GL_VERTEX_SHADER);
+    if (geo_shader)
+        geo.Create(geo_shader, GL_GEOMETRY_SHADER);
+    if (frag_shader)
+        frag.Create(frag_shader, GL_FRAGMENT_SHADER);
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceCreation);
+    Create(separable_program, vert.handle, geo.handle, frag.handle);
+}
+
+void OGLProgram::Release() {
+    if (handle == 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceDeletion);
+    glDeleteProgram(handle);
+    OpenGLState::GetCurState().ResetProgram(handle).Apply();
+    handle = 0;
+}
+
+void OGLPipeline::Create() {
+    if (handle != 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceCreation);
+    glGenProgramPipelines(1, &handle);
+}
+
+void OGLPipeline::Release() {
+    if (handle == 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceDeletion);
+    glDeleteProgramPipelines(1, &handle);
+    OpenGLState::GetCurState().ResetPipeline(handle).Apply();
+    handle = 0;
+}
+
+void OGLBuffer::Create() {
+    if (handle != 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceCreation);
+    glGenBuffers(1, &handle);
+}
+
+void OGLBuffer::Release() {
+    if (handle == 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceDeletion);
+    glDeleteBuffers(1, &handle);
+    OpenGLState::GetCurState().ResetBuffer(handle).Apply();
+    handle = 0;
+}
+
+void OGLSync::Create() {
+    if (handle != 0)
+        return;
+
+    // Don't profile here, this one is expected to happen ingame.
+    handle = glFenceSync(GL_SYNC_GPU_COMMANDS_COMPLETE, 0);
+}
+
+void OGLSync::Release() {
+    if (handle == 0)
+        return;
+
+    // Don't profile here, this one is expected to happen ingame.
+    glDeleteSync(handle);
+    handle = 0;
+}
+
+void OGLVertexArray::Create() {
+    if (handle != 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceCreation);
+    glGenVertexArrays(1, &handle);
+}
+
+void OGLVertexArray::Release() {
+    if (handle == 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceDeletion);
+    glDeleteVertexArrays(1, &handle);
+    OpenGLState::GetCurState().ResetVertexArray(handle).Apply();
+    handle = 0;
+}
+
+void OGLFramebuffer::Create() {
+    if (handle != 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceCreation);
+    glGenFramebuffers(1, &handle);
+}
+
+void OGLFramebuffer::Release() {
+    if (handle == 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceDeletion);
+    glDeleteFramebuffers(1, &handle);
+    OpenGLState::GetCurState().ResetFramebuffer(handle).Apply();
+    handle = 0;
+}
+
+} // namespace OpenGL

src/video_core/renderer_opengl/gl_resource_manager.h

@@ -8,7 +8,6 @@
 #include <glad/glad.h>
 #include "common/common_types.h"
 #include "video_core/renderer_opengl/gl_shader_util.h"
-#include "video_core/renderer_opengl/gl_state.h"
 
 namespace OpenGL {
 
@@ -29,20 +28,10 @@ public:
     }
 
     /// Creates a new internal OpenGL resource and stores the handle
-    void Create() {
-        if (handle != 0)
-            return;
-        glGenTextures(1, &handle);
-    }
+    void Create();
 
     /// Deletes the internal OpenGL resource
-    void Release() {
-        if (handle == 0)
-            return;
-        glDeleteTextures(1, &handle);
-        OpenGLState::GetCurState().UnbindTexture(handle).Apply();
-        handle = 0;
-    }
+    void Release();
 
     GLuint handle = 0;
 };
@@ -64,20 +53,10 @@ public:
     }
 
     /// Creates a new internal OpenGL resource and stores the handle
-    void Create() {
-        if (handle != 0)
-            return;
-        glGenSamplers(1, &handle);
-    }
+    void Create();
 
     /// Deletes the internal OpenGL resource
-    void Release() {
-        if (handle == 0)
-            return;
-        glDeleteSamplers(1, &handle);
-        OpenGLState::GetCurState().ResetSampler(handle).Apply();
-        handle = 0;
-    }
+    void Release();
 
     GLuint handle = 0;
 };
@@ -98,20 +77,9 @@ public:
         return *this;
     }
 
-    void Create(const char* source, GLenum type) {
-        if (handle != 0)
-            return;
-        if (source == nullptr)
-            return;
-        handle = GLShader::LoadShader(source, type);
-    }
+    void Create(const char* source, GLenum type);
 
-    void Release() {
-        if (handle == 0)
-            return;
-        glDeleteShader(handle);
-        handle = 0;
-    }
+    void Release();
 
     GLuint handle = 0;
 };
@@ -141,25 +109,10 @@ public:
 
     /// Creates a new internal OpenGL resource and stores the handle
     void CreateFromSource(const char* vert_shader, const char* geo_shader, const char* frag_shader,
-                          bool separable_program = false) {
-        OGLShader vert, geo, frag;
-        if (vert_shader)
-            vert.Create(vert_shader, GL_VERTEX_SHADER);
-        if (geo_shader)
-            geo.Create(geo_shader, GL_GEOMETRY_SHADER);
-        if (frag_shader)
-            frag.Create(frag_shader, GL_FRAGMENT_SHADER);
-        Create(separable_program, vert.handle, geo.handle, frag.handle);
-    }
+                          bool separable_program = false);
 
     /// Deletes the internal OpenGL resource
-    void Release() {
-        if (handle == 0)
-            return;
-        glDeleteProgram(handle);
-        OpenGLState::GetCurState().ResetProgram(handle).Apply();
-        handle = 0;
-    }
+    void Release();
 
     GLuint handle = 0;
 };
@@ -178,20 +131,10 @@ public:
     }
 
     /// Creates a new internal OpenGL resource and stores the handle
-    void Create() {
-        if (handle != 0)
-            return;
-        glGenProgramPipelines(1, &handle);
-    }
+    void Create();
 
     /// Deletes the internal OpenGL resource
-    void Release() {
-        if (handle == 0)
-            return;
-        glDeleteProgramPipelines(1, &handle);
-        OpenGLState::GetCurState().ResetPipeline(handle).Apply();
-        handle = 0;
-    }
+    void Release();
 
     GLuint handle = 0;
 };
@@ -213,20 +156,10 @@ public:
     }
 
     /// Creates a new internal OpenGL resource and stores the handle
-    void Create() {
-        if (handle != 0)
-            return;
-        glGenBuffers(1, &handle);
-    }
+    void Create();
 
     /// Deletes the internal OpenGL resource
-    void Release() {
-        if (handle == 0)
-            return;
-        glDeleteBuffers(1, &handle);
-        OpenGLState::GetCurState().ResetBuffer(handle).Apply();
-        handle = 0;
-    }
+    void Release();
 
     GLuint handle = 0;
 };
@@ -247,19 +180,10 @@ public:
     }
 
     /// Creates a new internal OpenGL resource and stores the handle
-    void Create() {
-        if (handle != 0)
-            return;
-        handle = glFenceSync(GL_SYNC_GPU_COMMANDS_COMPLETE, 0);
-    }
+    void Create();
 
     /// Deletes the internal OpenGL resource
-    void Release() {
-        if (handle == 0)
-            return;
-        glDeleteSync(handle);
-        handle = 0;
-    }
+    void Release();
 
     GLsync handle = 0;
 };
@@ -281,20 +205,10 @@ public:
     }
 
     /// Creates a new internal OpenGL resource and stores the handle
-    void Create() {
-        if (handle != 0)
-            return;
-        glGenVertexArrays(1, &handle);
-    }
+    void Create();
 
     /// Deletes the internal OpenGL resource
-    void Release() {
-        if (handle == 0)
-            return;
-        glDeleteVertexArrays(1, &handle);
-        OpenGLState::GetCurState().ResetVertexArray(handle).Apply();
-        handle = 0;
-    }
+    void Release();
 
     GLuint handle = 0;
 };
@@ -316,20 +230,10 @@ public:
     }
 
     /// Creates a new internal OpenGL resource and stores the handle
-    void Create() {
-        if (handle != 0)
-            return;
-        glGenFramebuffers(1, &handle);
-    }
+    void Create();
 
     /// Deletes the internal OpenGL resource
-    void Release() {
-        if (handle == 0)
-            return;
-        glDeleteFramebuffers(1, &handle);
-        OpenGLState::GetCurState().ResetFramebuffer(handle).Apply();
-        handle = 0;
-    }
+    void Release();
 
     GLuint handle = 0;
 };

src/video_core/renderer_opengl/gl_shader_cache.cpp

@@ -6,9 +6,10 @@
 #include "core/core.h"
 #include "core/memory.h"
 #include "video_core/engines/maxwell_3d.h"
+#include "video_core/renderer_opengl/gl_rasterizer.h"
 #include "video_core/renderer_opengl/gl_shader_cache.h"
 #include "video_core/renderer_opengl/gl_shader_manager.h"
-#include "video_core/utils.h"
+#include "video_core/renderer_opengl/utils.h"
 
 namespace OpenGL {
 
@@ -89,7 +90,7 @@ CachedShader::CachedShader(VAddr addr, Maxwell::ShaderProgram program_type)
         shader.Create(program_result.first.c_str(), gl_type);
         program.Create(true, shader.handle);
         SetShaderUniformBlockBindings(program.handle);
-        VideoCore::LabelGLObject(GL_PROGRAM, program.handle, addr);
+        LabelGLObject(GL_PROGRAM, program.handle, addr);
     } else {
         // Store shader's code to lazily build it on draw
         geometry_programs.code = program_result.first;
@@ -120,20 +121,26 @@ GLint CachedShader::GetUniformLocation(const GLShader::SamplerEntry& sampler) {
 }
 
 GLuint CachedShader::LazyGeometryProgram(OGLProgram& target_program,
-                                         const std::string& glsl_topology,
+                                         const std::string& glsl_topology, u32 max_vertices,
                                          const std::string& debug_name) {
     if (target_program.handle != 0) {
         return target_program.handle;
     }
-    const std::string source{geometry_programs.code + "layout (" + glsl_topology + ") in;\n"};
+    std::string source = "#version 430 core\n";
+    source += "layout (" + glsl_topology + ") in;\n";
+    source += "#define MAX_VERTEX_INPUT " + std::to_string(max_vertices) + '\n';
+    source += geometry_programs.code;
+
     OGLShader shader;
     shader.Create(source.c_str(), GL_GEOMETRY_SHADER);
     target_program.Create(true, shader.handle);
     SetShaderUniformBlockBindings(target_program.handle);
-    VideoCore::LabelGLObject(GL_PROGRAM, target_program.handle, addr, debug_name);
+    LabelGLObject(GL_PROGRAM, target_program.handle, addr, debug_name);
     return target_program.handle;
 };
 
+ShaderCacheOpenGL::ShaderCacheOpenGL(RasterizerOpenGL& rasterizer) : RasterizerCache{rasterizer} {}
+
 Shader ShaderCacheOpenGL::GetStageProgram(Maxwell::ShaderProgram program) {
     const VAddr program_addr{GetShaderAddress(program)};
 

src/video_core/renderer_opengl/gl_shader_cache.h

@@ -16,6 +16,8 @@
 namespace OpenGL {
 
 class CachedShader;
+class RasterizerOpenGL;
+
 using Shader = std::shared_ptr<CachedShader>;
 using Maxwell = Tegra::Engines::Maxwell3D::Regs;
 
@@ -46,22 +48,23 @@ public:
         }
         switch (primitive_mode) {
         case GL_POINTS:
-            return LazyGeometryProgram(geometry_programs.points, "points", "ShaderPoints");
+            return LazyGeometryProgram(geometry_programs.points, "points", 1, "ShaderPoints");
         case GL_LINES:
         case GL_LINE_STRIP:
-            return LazyGeometryProgram(geometry_programs.lines, "lines", "ShaderLines");
+            return LazyGeometryProgram(geometry_programs.lines, "lines", 2, "ShaderLines");
         case GL_LINES_ADJACENCY:
         case GL_LINE_STRIP_ADJACENCY:
-            return LazyGeometryProgram(geometry_programs.lines_adjacency, "lines_adjacency",
+            return LazyGeometryProgram(geometry_programs.lines_adjacency, "lines_adjacency", 4,
                                        "ShaderLinesAdjacency");
         case GL_TRIANGLES:
         case GL_TRIANGLE_STRIP:
         case GL_TRIANGLE_FAN:
-            return LazyGeometryProgram(geometry_programs.triangles, "triangles", "ShaderTriangles");
+            return LazyGeometryProgram(geometry_programs.triangles, "triangles", 3,
+                                       "ShaderTriangles");
         case GL_TRIANGLES_ADJACENCY:
         case GL_TRIANGLE_STRIP_ADJACENCY:
             return LazyGeometryProgram(geometry_programs.triangles_adjacency, "triangles_adjacency",
-                                       "ShaderLines");
+                                       6, "ShaderTrianglesAdjacency");
         default:
             UNREACHABLE_MSG("Unknown primitive mode.");
         }
@@ -76,7 +79,7 @@ public:
 private:
     /// Generates a geometry shader or returns one that already exists.
     GLuint LazyGeometryProgram(OGLProgram& target_program, const std::string& glsl_topology,
-                               const std::string& debug_name);
+                               u32 max_vertices, const std::string& debug_name);
 
     VAddr addr;
     Maxwell::ShaderProgram program_type;
@@ -104,6 +107,8 @@ private:
 
 class ShaderCacheOpenGL final : public RasterizerCache<Shader> {
 public:
+    explicit ShaderCacheOpenGL(RasterizerOpenGL& rasterizer);
+
     /// Gets the current specified shader stage program
     Shader GetStageProgram(Maxwell::ShaderProgram program);
 };

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -3,12 +3,12 @@
 // Refer to the license.txt file included.
 
 #include <map>
+#include <optional>
 #include <set>
 #include <string>
 #include <string_view>
 #include <unordered_set>
 
-#include <boost/optional.hpp>
 #include <fmt/format.h>
 
 #include "common/assert.h"
@@ -144,7 +144,7 @@ private:
         for (u32 offset = begin; offset != end && offset != PROGRAM_END; ++offset) {
             const Instruction instr = {program_code[offset]};
             if (const auto opcode = OpCode::Decode(instr)) {
-                switch (opcode->GetId()) {
+                switch (opcode->get().GetId()) {
                 case OpCode::Id::EXIT: {
                     // The EXIT instruction can be predicated, which means that the shader can
                     // conditionally end on this instruction. We have to consider the case where the
@@ -341,10 +341,10 @@ public:
      */
     void SetRegisterToFloat(const Register& reg, u64 elem, const std::string& value,
                             u64 dest_num_components, u64 value_num_components,
-                            bool is_saturated = false, u64 dest_elem = 0) {
+                            bool is_saturated = false, u64 dest_elem = 0, bool precise = false) {
 
         SetRegister(reg, elem, is_saturated ? "clamp(" + value + ", 0.0, 1.0)" : value,
-                    dest_num_components, value_num_components, dest_elem);
+                    dest_num_components, value_num_components, dest_elem, precise);
     }
 
     /**
@@ -368,11 +368,12 @@ public:
         const std::string func{is_signed ? "intBitsToFloat" : "uintBitsToFloat"};
 
         SetRegister(reg, elem, func + '(' + ConvertIntegerSize(value, size) + ')',
-                    dest_num_components, value_num_components, dest_elem);
+                    dest_num_components, value_num_components, dest_elem, false);
 
         if (sets_cc) {
             const std::string zero_condition = "( " + ConvertIntegerSize(value, size) + " == 0 )";
             SetInternalFlag(InternalFlag::ZeroFlag, zero_condition);
+            LOG_WARNING(HW_GPU, "Control Codes Imcomplete.");
         }
     }
 
@@ -416,7 +417,7 @@ public:
             }
         }();
 
-        SetRegister(reg, elem, result, dest_num_components, value_num_components, dest_elem);
+        SetRegister(reg, elem, result, dest_num_components, value_num_components, dest_elem, false);
     }
 
     /**
@@ -430,7 +431,7 @@ public:
      */
     void SetRegisterToInputAttibute(const Register& reg, u64 elem, Attribute::Index attribute,
                                     const Tegra::Shader::IpaMode& input_mode,
-                                    boost::optional<Register> vertex = {}) {
+                                    std::optional<Register> vertex = {}) {
         const std::string dest = GetRegisterAsFloat(reg);
         const std::string src = GetInputAttribute(attribute, input_mode, vertex) + GetSwizzle(elem);
         shader.AddLine(dest + " = " + src + ';');
@@ -493,10 +494,10 @@ public:
             // instruction for now.
             if (stage == Maxwell3D::Regs::ShaderStage::Geometry) {
                 // TODO(Rodrigo): nouveau sets some attributes after setting emitting a geometry
-                // shader. These instructions use a dirty register as buffer index. To avoid some
-                // drivers from complaining for the out of boundary writes, guard them.
-                const std::string buf_index{"min(" + GetRegisterAsInteger(buf_reg) + ", " +
-                                            std::to_string(MAX_GEOMETRY_BUFFERS - 1) + ')'};
+                // shader. These instructions use a dirty register as buffer index, to avoid some
+                // drivers from complaining about out of boundary writes, guard them.
+                const std::string buf_index{"((" + GetRegisterAsInteger(buf_reg) + ") % " +
+                                            std::to_string(MAX_GEOMETRY_BUFFERS) + ')'};
                 shader.AddLine("amem[" + buf_index + "][" +
                                std::to_string(static_cast<u32>(attribute)) + ']' +
                                GetSwizzle(elem) + " = " + src + ';');
@@ -757,7 +758,8 @@ private:
      * @param dest_elem Optional, the destination element to use for the operation.
      */
     void SetRegister(const Register& reg, u64 elem, const std::string& value,
-                     u64 dest_num_components, u64 value_num_components, u64 dest_elem) {
+                     u64 dest_num_components, u64 value_num_components, u64 dest_elem,
+                     bool precise) {
         if (reg == Register::ZeroIndex) {
             LOG_CRITICAL(HW_GPU, "Cannot set Register::ZeroIndex");
             UNREACHABLE();
@@ -774,7 +776,18 @@ private:
             src += GetSwizzle(elem);
         }
 
-        shader.AddLine(dest + " = " + src + ';');
+        if (precise && stage != Maxwell3D::Regs::ShaderStage::Fragment) {
+            shader.AddLine('{');
+            ++shader.scope;
+            // This avoids optimizations of constant propagation and keeps the code as the original
+            // Sadly using the precise keyword causes "linking" errors on fragment shaders.
+            shader.AddLine("precise float tmp = " + src + ';');
+            shader.AddLine(dest + " = tmp;");
+            --shader.scope;
+            shader.AddLine('}');
+        } else {
+            shader.AddLine(dest + " = " + src + ';');
+        }
     }
 
     /// Build the GLSL register list.
@@ -795,10 +808,14 @@ private:
     /// Generates code representing an input attribute register.
     std::string GetInputAttribute(Attribute::Index attribute,
                                   const Tegra::Shader::IpaMode& input_mode,
-                                  boost::optional<Register> vertex = {}) {
+                                  std::optional<Register> vertex = {}) {
         auto GeometryPass = [&](const std::string& name) {
             if (stage == Maxwell3D::Regs::ShaderStage::Geometry && vertex) {
-                return "gs_" + name + '[' + GetRegisterAsInteger(vertex.value(), 0, false) + ']';
+                // 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.
+                return "gs_" + name + '[' + GetRegisterAsInteger(*vertex, 0, false) +
+                       " % MAX_VERTEX_INPUT]";
             }
             return name;
         };
@@ -1453,7 +1470,7 @@ private:
         }
 
         shader.AddLine(
-            fmt::format("// {}: {} (0x{:016x})", offset, opcode->GetName(), instr.value));
+            fmt::format("// {}: {} (0x{:016x})", offset, opcode->get().GetName(), instr.value));
 
         using Tegra::Shader::Pred;
         ASSERT_MSG(instr.pred.full_pred != Pred::NeverExecute,
@@ -1461,7 +1478,7 @@ private:
 
         // Some instructions (like SSY) don't have a predicate field, they are always
         // unconditionally executed.
-        bool can_be_predicated = OpCode::IsPredicatedInstruction(opcode->GetId());
+        bool can_be_predicated = OpCode::IsPredicatedInstruction(opcode->get().GetId());
 
         if (can_be_predicated && instr.pred.pred_index != static_cast<u64>(Pred::UnusedIndex)) {
             shader.AddLine("if (" +
@@ -1471,7 +1488,7 @@ private:
             ++shader.scope;
         }
 
-        switch (opcode->GetType()) {
+        switch (opcode->get().GetType()) {
         case OpCode::Type::Arithmetic: {
             std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
 
@@ -1488,7 +1505,7 @@ private:
                 }
             }
 
-            switch (opcode->GetId()) {
+            switch (opcode->get().GetId()) {
             case OpCode::Id::MOV_C:
             case OpCode::Id::MOV_R: {
                 // MOV does not have neither 'abs' nor 'neg' bits.
@@ -1510,8 +1527,13 @@ private:
                 ASSERT_MSG(instr.fmul.cc == 0, "FMUL cc is not implemented");
 
                 op_b = GetOperandAbsNeg(op_b, false, instr.fmul.negate_b);
+
                 regs.SetRegisterToFloat(instr.gpr0, 0, op_a + " * " + op_b, 1, 1,
-                                        instr.alu.saturate_d);
+                                        instr.alu.saturate_d, 0, true);
+                if (instr.generates_cc) {
+                    LOG_CRITICAL(HW_GPU, "FMUL Generates an unhandled Control Code");
+                    UNREACHABLE();
+                }
                 break;
             }
             case OpCode::Id::FADD_C:
@@ -1519,8 +1541,13 @@ private:
             case OpCode::Id::FADD_IMM: {
                 op_a = GetOperandAbsNeg(op_a, instr.alu.abs_a, instr.alu.negate_a);
                 op_b = GetOperandAbsNeg(op_b, instr.alu.abs_b, instr.alu.negate_b);
+
                 regs.SetRegisterToFloat(instr.gpr0, 0, op_a + " + " + op_b, 1, 1,
-                                        instr.alu.saturate_d);
+                                        instr.alu.saturate_d, 0, true);
+                if (instr.generates_cc) {
+                    LOG_CRITICAL(HW_GPU, "FADD Generates an unhandled Control Code");
+                    UNREACHABLE();
+                }
                 break;
             }
             case OpCode::Id::MUFU: {
@@ -1528,31 +1555,31 @@ private:
                 switch (instr.sub_op) {
                 case SubOp::Cos:
                     regs.SetRegisterToFloat(instr.gpr0, 0, "cos(" + op_a + ')', 1, 1,
-                                            instr.alu.saturate_d);
+                                            instr.alu.saturate_d, 0, true);
                     break;
                 case SubOp::Sin:
                     regs.SetRegisterToFloat(instr.gpr0, 0, "sin(" + op_a + ')', 1, 1,
-                                            instr.alu.saturate_d);
+                                            instr.alu.saturate_d, 0, true);
                     break;
                 case SubOp::Ex2:
                     regs.SetRegisterToFloat(instr.gpr0, 0, "exp2(" + op_a + ')', 1, 1,
-                                            instr.alu.saturate_d);
+                                            instr.alu.saturate_d, 0, true);
                     break;
                 case SubOp::Lg2:
                     regs.SetRegisterToFloat(instr.gpr0, 0, "log2(" + op_a + ')', 1, 1,
-                                            instr.alu.saturate_d);
+                                            instr.alu.saturate_d, 0, true);
                     break;
                 case SubOp::Rcp:
                     regs.SetRegisterToFloat(instr.gpr0, 0, "1.0 / " + op_a, 1, 1,
-                                            instr.alu.saturate_d);
+                                            instr.alu.saturate_d, 0, true);
                     break;
                 case SubOp::Rsq:
                     regs.SetRegisterToFloat(instr.gpr0, 0, "inversesqrt(" + op_a + ')', 1, 1,
-                                            instr.alu.saturate_d);
+                                            instr.alu.saturate_d, 0, true);
                     break;
                 case SubOp::Sqrt:
                     regs.SetRegisterToFloat(instr.gpr0, 0, "sqrt(" + op_a + ')', 1, 1,
-                                            instr.alu.saturate_d);
+                                            instr.alu.saturate_d, 0, true);
                     break;
                 default:
                     LOG_CRITICAL(HW_GPU, "Unhandled MUFU sub op: {0:x}",
@@ -1573,7 +1600,11 @@ private:
                 regs.SetRegisterToFloat(instr.gpr0, 0,
                                         '(' + condition + ") ? min(" + parameters + ") : max(" +
                                             parameters + ')',
-                                        1, 1);
+                                        1, 1, false, 0, true);
+                if (instr.generates_cc) {
+                    LOG_CRITICAL(HW_GPU, "FMNMX Generates an unhandled Control Code");
+                    UNREACHABLE();
+                }
                 break;
             }
             case OpCode::Id::RRO_C:
@@ -1586,14 +1617,15 @@ private:
                 break;
             }
             default: {
-                LOG_CRITICAL(HW_GPU, "Unhandled arithmetic instruction: {}", opcode->GetName());
+                LOG_CRITICAL(HW_GPU, "Unhandled arithmetic instruction: {}",
+                             opcode->get().GetName());
                 UNREACHABLE();
             }
             }
             break;
         }
         case OpCode::Type::ArithmeticImmediate: {
-            switch (opcode->GetId()) {
+            switch (opcode->get().GetId()) {
             case OpCode::Id::MOV32_IMM: {
                 regs.SetRegisterToFloat(instr.gpr0, 0, GetImmediate32(instr), 1, 1);
                 break;
@@ -1602,7 +1634,11 @@ private:
                 regs.SetRegisterToFloat(instr.gpr0, 0,
                                         regs.GetRegisterAsFloat(instr.gpr8) + " * " +
                                             GetImmediate32(instr),
-                                        1, 1, instr.fmul32.saturate);
+                                        1, 1, instr.fmul32.saturate, 0, true);
+                if (instr.op_32.generates_cc) {
+                    LOG_CRITICAL(HW_GPU, "FMUL32 Generates an unhandled Control Code");
+                    UNREACHABLE();
+                }
                 break;
             }
             case OpCode::Id::FADD32I: {
@@ -1625,7 +1661,11 @@ private:
                     op_b = "-(" + op_b + ')';
                 }
 
-                regs.SetRegisterToFloat(instr.gpr0, 0, op_a + " + " + op_b, 1, 1);
+                regs.SetRegisterToFloat(instr.gpr0, 0, op_a + " + " + op_b, 1, 1, false, 0, true);
+                if (instr.op_32.generates_cc) {
+                    LOG_CRITICAL(HW_GPU, "FADD32 Generates an unhandled Control Code");
+                    UNREACHABLE();
+                }
                 break;
             }
             }
@@ -1637,7 +1677,7 @@ private:
             std::string op_a = instr.bfe.negate_a ? "-" : "";
             op_a += regs.GetRegisterAsInteger(instr.gpr8);
 
-            switch (opcode->GetId()) {
+            switch (opcode->get().GetId()) {
             case OpCode::Id::BFE_IMM: {
                 std::string inner_shift =
                     '(' + op_a + " << " + std::to_string(instr.bfe.GetLeftShiftValue()) + ')';
@@ -1646,10 +1686,14 @@ private:
                     std::to_string(instr.bfe.GetLeftShiftValue() + instr.bfe.shift_position) + ')';
 
                 regs.SetRegisterToInteger(instr.gpr0, true, 0, outer_shift, 1, 1);
+                if (instr.generates_cc) {
+                    LOG_CRITICAL(HW_GPU, "BFE Generates an unhandled Control Code");
+                    UNREACHABLE();
+                }
                 break;
             }
             default: {
-                LOG_CRITICAL(HW_GPU, "Unhandled BFE instruction: {}", opcode->GetName());
+                LOG_CRITICAL(HW_GPU, "Unhandled BFE instruction: {}", opcode->get().GetName());
                 UNREACHABLE();
             }
             }
@@ -1671,7 +1715,7 @@ private:
                 }
             }
 
-            switch (opcode->GetId()) {
+            switch (opcode->get().GetId()) {
             case OpCode::Id::SHR_C:
             case OpCode::Id::SHR_R:
             case OpCode::Id::SHR_IMM: {
@@ -1683,15 +1727,23 @@ private:
                 // Cast to int is superfluous for arithmetic shift, it's only for a logical shift
                 regs.SetRegisterToInteger(instr.gpr0, true, 0, "int(" + op_a + " >> " + op_b + ')',
                                           1, 1);
+                if (instr.generates_cc) {
+                    LOG_CRITICAL(HW_GPU, "SHR Generates an unhandled Control Code");
+                    UNREACHABLE();
+                }
                 break;
             }
             case OpCode::Id::SHL_C:
             case OpCode::Id::SHL_R:
             case OpCode::Id::SHL_IMM:
                 regs.SetRegisterToInteger(instr.gpr0, true, 0, op_a + " << " + op_b, 1, 1);
+                if (instr.generates_cc) {
+                    LOG_CRITICAL(HW_GPU, "SHL Generates an unhandled Control Code");
+                    UNREACHABLE();
+                }
                 break;
             default: {
-                LOG_CRITICAL(HW_GPU, "Unhandled shift instruction: {}", opcode->GetName());
+                LOG_CRITICAL(HW_GPU, "Unhandled shift instruction: {}", opcode->get().GetName());
                 UNREACHABLE();
             }
             }
@@ -1701,13 +1753,17 @@ private:
             std::string op_a = regs.GetRegisterAsInteger(instr.gpr8);
             std::string op_b = std::to_string(instr.alu.imm20_32.Value());
 
-            switch (opcode->GetId()) {
+            switch (opcode->get().GetId()) {
             case OpCode::Id::IADD32I:
                 if (instr.iadd32i.negate_a)
                     op_a = "-(" + op_a + ')';
 
                 regs.SetRegisterToInteger(instr.gpr0, true, 0, op_a + " + " + op_b, 1, 1,
                                           instr.iadd32i.saturate != 0);
+                if (instr.op_32.generates_cc) {
+                    LOG_CRITICAL(HW_GPU, "IADD32 Generates an unhandled Control Code");
+                    UNREACHABLE();
+                }
                 break;
             case OpCode::Id::LOP32I: {
                 if (instr.alu.lop32i.invert_a)
@@ -1719,11 +1775,15 @@ private:
                 WriteLogicOperation(instr.gpr0, instr.alu.lop32i.operation, op_a, op_b,
                                     Tegra::Shader::PredicateResultMode::None,
                                     Tegra::Shader::Pred::UnusedIndex);
+                if (instr.op_32.generates_cc) {
+                    LOG_CRITICAL(HW_GPU, "LOP32I Generates an unhandled Control Code");
+                    UNREACHABLE();
+                }
                 break;
             }
             default: {
                 LOG_CRITICAL(HW_GPU, "Unhandled ArithmeticIntegerImmediate instruction: {}",
-                             opcode->GetName());
+                             opcode->get().GetName());
                 UNREACHABLE();
             }
             }
@@ -1743,7 +1803,7 @@ private:
                 }
             }
 
-            switch (opcode->GetId()) {
+            switch (opcode->get().GetId()) {
             case OpCode::Id::IADD_C:
             case OpCode::Id::IADD_R:
             case OpCode::Id::IADD_IMM: {
@@ -1755,6 +1815,10 @@ private:
 
                 regs.SetRegisterToInteger(instr.gpr0, true, 0, op_a + " + " + op_b, 1, 1,
                                           instr.alu.saturate_d);
+                if (instr.generates_cc) {
+                    LOG_CRITICAL(HW_GPU, "IADD Generates an unhandled Control Code");
+                    UNREACHABLE();
+                }
                 break;
             }
             case OpCode::Id::IADD3_C:
@@ -1779,7 +1843,7 @@ private:
                     }
                 };
 
-                if (opcode->GetId() == OpCode::Id::IADD3_R) {
+                if (opcode->get().GetId() == OpCode::Id::IADD3_R) {
                     apply_height(instr.iadd3.height_a, op_a);
                     apply_height(instr.iadd3.height_b, op_b);
                     apply_height(instr.iadd3.height_c, op_c);
@@ -1795,7 +1859,7 @@ private:
                     op_c = "-(" + op_c + ')';
 
                 std::string result;
-                if (opcode->GetId() == OpCode::Id::IADD3_R) {
+                if (opcode->get().GetId() == OpCode::Id::IADD3_R) {
                     switch (instr.iadd3.mode) {
                     case Tegra::Shader::IAdd3Mode::RightShift:
                         // TODO(tech4me): According to
@@ -1816,6 +1880,11 @@ private:
                 }
 
                 regs.SetRegisterToInteger(instr.gpr0, true, 0, result, 1, 1);
+
+                if (instr.generates_cc) {
+                    LOG_CRITICAL(HW_GPU, "IADD3 Generates an unhandled Control Code");
+                    UNREACHABLE();
+                }
                 break;
             }
             case OpCode::Id::ISCADD_C:
@@ -1831,6 +1900,10 @@ private:
 
                 regs.SetRegisterToInteger(instr.gpr0, true, 0,
                                           "((" + op_a + " << " + shift + ") + " + op_b + ')', 1, 1);
+                if (instr.generates_cc) {
+                    LOG_CRITICAL(HW_GPU, "ISCADD Generates an unhandled Control Code");
+                    UNREACHABLE();
+                }
                 break;
             }
             case OpCode::Id::POPC_C:
@@ -1862,6 +1935,10 @@ private:
 
                 WriteLogicOperation(instr.gpr0, instr.alu.lop.operation, op_a, op_b,
                                     instr.alu.lop.pred_result_mode, instr.alu.lop.pred48);
+                if (instr.generates_cc) {
+                    LOG_CRITICAL(HW_GPU, "LOP Generates an unhandled Control Code");
+                    UNREACHABLE();
+                }
                 break;
             }
             case OpCode::Id::LOP3_C:
@@ -1870,13 +1947,17 @@ private:
                 const std::string op_c = regs.GetRegisterAsInteger(instr.gpr39);
                 std::string lut;
 
-                if (opcode->GetId() == OpCode::Id::LOP3_R) {
+                if (opcode->get().GetId() == OpCode::Id::LOP3_R) {
                     lut = '(' + std::to_string(instr.alu.lop3.GetImmLut28()) + ')';
                 } else {
                     lut = '(' + std::to_string(instr.alu.lop3.GetImmLut48()) + ')';
                 }
 
                 WriteLop3Instruction(instr.gpr0, op_a, op_b, op_c, lut);
+                if (instr.generates_cc) {
+                    LOG_CRITICAL(HW_GPU, "LOP3 Generates an unhandled Control Code");
+                    UNREACHABLE();
+                }
                 break;
             }
             case OpCode::Id::IMNMX_C:
@@ -1891,6 +1972,10 @@ private:
                                           '(' + condition + ") ? min(" + parameters + ") : max(" +
                                               parameters + ')',
                                           1, 1);
+                if (instr.generates_cc) {
+                    LOG_CRITICAL(HW_GPU, "IMNMX Generates an unhandled Control Code");
+                    UNREACHABLE();
+                }
                 break;
             }
             case OpCode::Id::LEA_R2:
@@ -1900,7 +1985,7 @@ private:
             case OpCode::Id::LEA_HI: {
                 std::string op_c;
 
-                switch (opcode->GetId()) {
+                switch (opcode->get().GetId()) {
                 case OpCode::Id::LEA_R2: {
                     op_a = regs.GetRegisterAsInteger(instr.gpr20);
                     op_b = regs.GetRegisterAsInteger(instr.gpr39);
@@ -1945,7 +2030,8 @@ private:
                     op_b = regs.GetRegisterAsInteger(instr.gpr8);
                     op_a = std::to_string(instr.lea.imm.entry_a);
                     op_c = std::to_string(instr.lea.imm.entry_b);
-                    LOG_CRITICAL(HW_GPU, "Unhandled LEA subinstruction: {}", opcode->GetName());
+                    LOG_CRITICAL(HW_GPU, "Unhandled LEA subinstruction: {}",
+                                 opcode->get().GetName());
                     UNREACHABLE();
                 }
                 }
@@ -1960,7 +2046,7 @@ private:
             }
             default: {
                 LOG_CRITICAL(HW_GPU, "Unhandled ArithmeticInteger instruction: {}",
-                             opcode->GetName());
+                             opcode->get().GetName());
                 UNREACHABLE();
             }
             }
@@ -1968,20 +2054,21 @@ private:
             break;
         }
         case OpCode::Type::ArithmeticHalf: {
-            if (opcode->GetId() == OpCode::Id::HADD2_C || opcode->GetId() == OpCode::Id::HADD2_R) {
+            if (opcode->get().GetId() == OpCode::Id::HADD2_C ||
+                opcode->get().GetId() == OpCode::Id::HADD2_R) {
                 ASSERT_MSG(instr.alu_half.ftz == 0, "Unimplemented");
             }
             const bool negate_a =
-                opcode->GetId() != OpCode::Id::HMUL2_R && instr.alu_half.negate_a != 0;
+                opcode->get().GetId() != OpCode::Id::HMUL2_R && instr.alu_half.negate_a != 0;
             const bool negate_b =
-                opcode->GetId() != OpCode::Id::HMUL2_C && instr.alu_half.negate_b != 0;
+                opcode->get().GetId() != OpCode::Id::HMUL2_C && instr.alu_half.negate_b != 0;
 
             const std::string op_a =
                 GetHalfFloat(regs.GetRegisterAsInteger(instr.gpr8, 0, false), instr.alu_half.type_a,
                              instr.alu_half.abs_a != 0, negate_a);
 
             std::string op_b;
-            switch (opcode->GetId()) {
+            switch (opcode->get().GetId()) {
             case OpCode::Id::HADD2_C:
             case OpCode::Id::HMUL2_C:
                 op_b = regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
@@ -1999,7 +2086,7 @@ private:
             op_b = GetHalfFloat(op_b, instr.alu_half.type_b, instr.alu_half.abs_b != 0, negate_b);
 
             const std::string result = [&]() {
-                switch (opcode->GetId()) {
+                switch (opcode->get().GetId()) {
                 case OpCode::Id::HADD2_C:
                 case OpCode::Id::HADD2_R:
                     return '(' + op_a + " + " + op_b + ')';
@@ -2007,7 +2094,8 @@ private:
                 case OpCode::Id::HMUL2_R:
                     return '(' + op_a + " * " + op_b + ')';
                 default:
-                    LOG_CRITICAL(HW_GPU, "Unhandled half float instruction: {}", opcode->GetName());
+                    LOG_CRITICAL(HW_GPU, "Unhandled half float instruction: {}",
+                                 opcode->get().GetName());
                     UNREACHABLE();
                     return std::string("0");
                 }
@@ -2018,7 +2106,7 @@ private:
             break;
         }
         case OpCode::Type::ArithmeticHalfImmediate: {
-            if (opcode->GetId() == OpCode::Id::HADD2_IMM) {
+            if (opcode->get().GetId() == OpCode::Id::HADD2_IMM) {
                 ASSERT_MSG(instr.alu_half_imm.ftz == 0, "Unimplemented");
             } else {
                 ASSERT_MSG(instr.alu_half_imm.precision == Tegra::Shader::HalfPrecision::None,
@@ -2032,7 +2120,7 @@ private:
             const std::string op_b = UnpackHalfImmediate(instr, true);
 
             const std::string result = [&]() {
-                switch (opcode->GetId()) {
+                switch (opcode->get().GetId()) {
                 case OpCode::Id::HADD2_IMM:
                     return op_a + " + " + op_b;
                 case OpCode::Id::HMUL2_IMM:
@@ -2058,7 +2146,7 @@ private:
             ASSERT_MSG(instr.ffma.tab5980_1 == 0, "FFMA tab5980_1({}) not implemented",
                        instr.ffma.tab5980_1.Value());
 
-            switch (opcode->GetId()) {
+            switch (opcode->get().GetId()) {
             case OpCode::Id::FFMA_CR: {
                 op_b += regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
                                         GLSLRegister::Type::Float);
@@ -2082,24 +2170,29 @@ private:
                 break;
             }
             default: {
-                LOG_CRITICAL(HW_GPU, "Unhandled FFMA instruction: {}", opcode->GetName());
+                LOG_CRITICAL(HW_GPU, "Unhandled FFMA instruction: {}", opcode->get().GetName());
                 UNREACHABLE();
             }
             }
 
-            regs.SetRegisterToFloat(instr.gpr0, 0, op_a + " * " + op_b + " + " + op_c, 1, 1,
-                                    instr.alu.saturate_d);
+            regs.SetRegisterToFloat(instr.gpr0, 0, "fma(" + op_a + ", " + op_b + ", " + op_c + ')',
+                                    1, 1, instr.alu.saturate_d, 0, true);
+            if (instr.generates_cc) {
+                LOG_CRITICAL(HW_GPU, "FFMA Generates an unhandled Control Code");
+                UNREACHABLE();
+            }
+
             break;
         }
         case OpCode::Type::Hfma2: {
-            if (opcode->GetId() == OpCode::Id::HFMA2_RR) {
+            if (opcode->get().GetId() == OpCode::Id::HFMA2_RR) {
                 ASSERT_MSG(instr.hfma2.rr.precision == Tegra::Shader::HalfPrecision::None,
                            "Unimplemented");
             } else {
                 ASSERT_MSG(instr.hfma2.precision == Tegra::Shader::HalfPrecision::None,
                            "Unimplemented");
             }
-            const bool saturate = opcode->GetId() == OpCode::Id::HFMA2_RR
+            const bool saturate = opcode->get().GetId() == OpCode::Id::HFMA2_RR
                                       ? instr.hfma2.rr.saturate != 0
                                       : instr.hfma2.saturate != 0;
 
@@ -2107,7 +2200,7 @@ private:
                 GetHalfFloat(regs.GetRegisterAsInteger(instr.gpr8, 0, false), instr.hfma2.type_a);
             std::string op_b, op_c;
 
-            switch (opcode->GetId()) {
+            switch (opcode->get().GetId()) {
             case OpCode::Id::HFMA2_CR:
                 op_b = GetHalfFloat(regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
                                                     GLSLRegister::Type::UnsignedInteger),
@@ -2145,7 +2238,7 @@ private:
             break;
         }
         case OpCode::Type::Conversion: {
-            switch (opcode->GetId()) {
+            switch (opcode->get().GetId()) {
             case OpCode::Id::I2I_R: {
                 ASSERT_MSG(!instr.conversion.selector, "Unimplemented");
 
@@ -2193,6 +2286,11 @@ private:
                 }
 
                 regs.SetRegisterToFloat(instr.gpr0, 0, op_a, 1, 1);
+
+                if (instr.generates_cc) {
+                    LOG_CRITICAL(HW_GPU, "I2F Generates an unhandled Control Code");
+                    UNREACHABLE();
+                }
                 break;
             }
             case OpCode::Id::F2F_R: {
@@ -2231,6 +2329,11 @@ private:
                 }
 
                 regs.SetRegisterToFloat(instr.gpr0, 0, op_a, 1, 1, instr.alu.saturate_d);
+
+                if (instr.generates_cc) {
+                    LOG_CRITICAL(HW_GPU, "F2F Generates an unhandled Control Code");
+                    UNREACHABLE();
+                }
                 break;
             }
             case OpCode::Id::F2I_R:
@@ -2280,17 +2383,22 @@ private:
 
                 regs.SetRegisterToInteger(instr.gpr0, instr.conversion.is_output_signed, 0, op_a, 1,
                                           1, false, 0, instr.conversion.dest_size);
+                if (instr.generates_cc) {
+                    LOG_CRITICAL(HW_GPU, "F2I Generates an unhandled Control Code");
+                    UNREACHABLE();
+                }
                 break;
             }
             default: {
-                LOG_CRITICAL(HW_GPU, "Unhandled conversion instruction: {}", opcode->GetName());
+                LOG_CRITICAL(HW_GPU, "Unhandled conversion instruction: {}",
+                             opcode->get().GetName());
                 UNREACHABLE();
             }
             }
             break;
         }
         case OpCode::Type::Memory: {
-            switch (opcode->GetId()) {
+            switch (opcode->get().GetId()) {
             case OpCode::Id::LD_A: {
                 // Note: Shouldn't this be interp mode flat? As in no interpolation made.
                 ASSERT_MSG(instr.gpr8.Value() == Register::ZeroIndex,
@@ -2638,12 +2746,12 @@ private:
                 }
                 case 3: {
                     if (is_array) {
-                        UNIMPLEMENTED_MSG("3-coordinate arrays not fully implemented");
-                        const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
-                        const std::string y = regs.GetRegisterAsFloat(instr.gpr20);
-                        coord = "vec2 coords = vec2(" + x + ", " + y + ");";
-                        texture_type = Tegra::Shader::TextureType::Texture2D;
-                        is_array = false;
+                        const std::string index = regs.GetRegisterAsInteger(instr.gpr8);
+                        const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                        const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 2);
+                        const std::string z = regs.GetRegisterAsFloat(instr.gpr20);
+                        coord =
+                            "vec4 coords = vec4(" + x + ", " + y + ", " + z + ", " + index + ");";
                     } else {
                         const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
                         const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
@@ -2673,7 +2781,11 @@ private:
                     break;
                 }
                 case Tegra::Shader::TextureProcessMode::LZ: {
-                    texture = "textureLod(" + sampler + ", coords, 0.0)";
+                    if (depth_compare && is_array) {
+                        texture = "texture(" + sampler + ", coords)";
+                    } else {
+                        texture = "textureLod(" + sampler + ", coords, 0.0)";
+                    }
                     break;
                 }
                 case Tegra::Shader::TextureProcessMode::LL: {
@@ -2934,7 +3046,7 @@ private:
                 break;
             }
             default: {
-                LOG_CRITICAL(HW_GPU, "Unhandled memory instruction: {}", opcode->GetName());
+                LOG_CRITICAL(HW_GPU, "Unhandled memory instruction: {}", opcode->get().GetName());
                 UNREACHABLE();
             }
             }
@@ -3028,7 +3140,7 @@ private:
                              instr.hsetp2.abs_a, instr.hsetp2.negate_a);
 
             const std::string op_b = [&]() {
-                switch (opcode->GetId()) {
+                switch (opcode->get().GetId()) {
                 case OpCode::Id::HSETP2_R:
                     return GetHalfFloat(regs.GetRegisterAsInteger(instr.gpr20, 0, false),
                                         instr.hsetp2.type_b, instr.hsetp2.abs_a,
@@ -3087,10 +3199,15 @@ private:
                 regs.SetRegisterToFloat(instr.gpr0, 0, value, 1, 1);
             }
 
+            if (instr.generates_cc) {
+                LOG_CRITICAL(HW_GPU, "PSET Generates an unhandled Control Code");
+                UNREACHABLE();
+            }
+
             break;
         }
         case OpCode::Type::PredicateSetPredicate: {
-            switch (opcode->GetId()) {
+            switch (opcode->get().GetId()) {
             case OpCode::Id::PSETP: {
                 const std::string op_a =
                     GetPredicateCondition(instr.psetp.pred12, instr.psetp.neg_pred12 != 0);
@@ -3136,7 +3253,8 @@ private:
                 break;
             }
             default: {
-                LOG_CRITICAL(HW_GPU, "Unhandled predicate instruction: {}", opcode->GetName());
+                LOG_CRITICAL(HW_GPU, "Unhandled predicate instruction: {}",
+                             opcode->get().GetName());
                 UNREACHABLE();
             }
             }
@@ -3224,7 +3342,7 @@ private:
                              instr.hset2.abs_a != 0, instr.hset2.negate_a != 0);
 
             const std::string op_b = [&]() {
-                switch (opcode->GetId()) {
+                switch (opcode->get().GetId()) {
                 case OpCode::Id::HSET2_R:
                     return GetHalfFloat(regs.GetRegisterAsInteger(instr.gpr20, 0, false),
                                         instr.hset2.type_b, instr.hset2.abs_b != 0,
@@ -3273,7 +3391,7 @@ private:
             const bool is_signed{instr.xmad.sign_a == 1};
 
             bool is_merge{};
-            switch (opcode->GetId()) {
+            switch (opcode->get().GetId()) {
             case OpCode::Id::XMAD_CR: {
                 is_merge = instr.xmad.merge_56;
                 op_b += regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
@@ -3302,7 +3420,7 @@ private:
                 break;
             }
             default: {
-                LOG_CRITICAL(HW_GPU, "Unhandled XMAD instruction: {}", opcode->GetName());
+                LOG_CRITICAL(HW_GPU, "Unhandled XMAD instruction: {}", opcode->get().GetName());
                 UNREACHABLE();
             }
             }
@@ -3351,15 +3469,25 @@ private:
             }
 
             regs.SetRegisterToInteger(instr.gpr0, is_signed, 0, sum, 1, 1);
+            if (instr.generates_cc) {
+                LOG_CRITICAL(HW_GPU, "XMAD Generates an unhandled Control Code");
+                UNREACHABLE();
+            }
             break;
         }
         default: {
-            switch (opcode->GetId()) {
+            switch (opcode->get().GetId()) {
             case OpCode::Id::EXIT: {
                 if (stage == Maxwell3D::Regs::ShaderStage::Fragment) {
                     EmitFragmentOutputsWrite();
                 }
 
+                const Tegra::Shader::ControlCode cc = instr.flow_control_code;
+                if (cc != Tegra::Shader::ControlCode::T) {
+                    LOG_CRITICAL(HW_GPU, "EXIT Control Code used: {}", static_cast<u32>(cc));
+                    UNREACHABLE();
+                }
+
                 switch (instr.flow.cond) {
                 case Tegra::Shader::FlowCondition::Always:
                     shader.AddLine("return true;");
@@ -3389,6 +3517,11 @@ private:
 
                 // Enclose "discard" in a conditional, so that GLSL compilation does not complain
                 // about unexecuted instructions that may follow this.
+                const Tegra::Shader::ControlCode cc = instr.flow_control_code;
+                if (cc != Tegra::Shader::ControlCode::T) {
+                    LOG_CRITICAL(HW_GPU, "KIL Control Code used: {}", static_cast<u32>(cc));
+                    UNREACHABLE();
+                }
                 shader.AddLine("if (true) {");
                 ++shader.scope;
                 shader.AddLine("discard;");
@@ -3446,6 +3579,11 @@ private:
             case OpCode::Id::BRA: {
                 ASSERT_MSG(instr.bra.constant_buffer == 0,
                            "BRA with constant buffers are not implemented");
+                const Tegra::Shader::ControlCode cc = instr.flow_control_code;
+                if (cc != Tegra::Shader::ControlCode::T) {
+                    LOG_CRITICAL(HW_GPU, "BRA Control Code used: {}", static_cast<u32>(cc));
+                    UNREACHABLE();
+                }
                 const u32 target = offset + instr.bra.GetBranchTarget();
                 shader.AddLine("{ jmp_to = " + std::to_string(target) + "u; break; }");
                 break;
@@ -3486,13 +3624,21 @@ private:
             }
             case OpCode::Id::SYNC: {
                 // The SYNC opcode jumps to the address previously set by the SSY opcode
-                ASSERT(instr.flow.cond == Tegra::Shader::FlowCondition::Always);
+                const Tegra::Shader::ControlCode cc = instr.flow_control_code;
+                if (cc != Tegra::Shader::ControlCode::T) {
+                    LOG_CRITICAL(HW_GPU, "SYNC Control Code used: {}", static_cast<u32>(cc));
+                    UNREACHABLE();
+                }
                 EmitPopFromFlowStack();
                 break;
             }
             case OpCode::Id::BRK: {
                 // The BRK opcode jumps to the address previously set by the PBK opcode
-                ASSERT(instr.flow.cond == Tegra::Shader::FlowCondition::Always);
+                const Tegra::Shader::ControlCode cc = instr.flow_control_code;
+                if (cc != Tegra::Shader::ControlCode::T) {
+                    LOG_CRITICAL(HW_GPU, "BRK Control Code used: {}", static_cast<u32>(cc));
+                    UNREACHABLE();
+                }
                 EmitPopFromFlowStack();
                 break;
             }
@@ -3522,6 +3668,11 @@ private:
                 regs.SetRegisterToInteger(instr.gpr0, result_signed, 1, result, 1, 1,
                                           instr.vmad.saturate == 1, 0, Register::Size::Word,
                                           instr.vmad.cc);
+                if (instr.generates_cc) {
+                    LOG_CRITICAL(HW_GPU, "VMAD Generates an unhandled Control Code");
+                    UNREACHABLE();
+                }
+
                 break;
             }
             case OpCode::Id::VSETP: {
@@ -3549,7 +3700,7 @@ private:
                 break;
             }
             default: {
-                LOG_CRITICAL(HW_GPU, "Unhandled instruction: {}", opcode->GetName());
+                LOG_CRITICAL(HW_GPU, "Unhandled instruction: {}", opcode->get().GetName());
                 UNREACHABLE();
             }
             }
@@ -3690,9 +3841,9 @@ std::string GetCommonDeclarations() {
                        RasterizerOpenGL::MaxConstbufferSize / sizeof(GLvec4));
 }
 
-boost::optional<ProgramResult> DecompileProgram(const ProgramCode& program_code, u32 main_offset,
-                                                Maxwell3D::Regs::ShaderStage stage,
-                                                const std::string& suffix) {
+std::optional<ProgramResult> DecompileProgram(const ProgramCode& program_code, u32 main_offset,
+                                              Maxwell3D::Regs::ShaderStage stage,
+                                              const std::string& suffix) {
     try {
         const auto subroutines =
             ControlFlowAnalyzer(program_code, main_offset, suffix).GetSubroutines();
@@ -3701,7 +3852,7 @@ boost::optional<ProgramResult> DecompileProgram(const ProgramCode& program_code,
     } catch (const DecompileFail& exception) {
         LOG_ERROR(HW_GPU, "Shader decompilation failed: {}", exception.what());
     }
-    return boost::none;
+    return {};
 }
 
 } // namespace OpenGL::GLShader::Decompiler

src/video_core/renderer_opengl/gl_shader_decompiler.h

@@ -6,8 +6,8 @@
 
 #include <array>
 #include <functional>
+#include <optional>
 #include <string>
-#include <boost/optional.hpp>
 #include "common/common_types.h"
 #include "video_core/engines/maxwell_3d.h"
 #include "video_core/renderer_opengl/gl_shader_gen.h"
@@ -18,8 +18,8 @@ using Tegra::Engines::Maxwell3D;
 
 std::string GetCommonDeclarations();
 
-boost::optional<ProgramResult> DecompileProgram(const ProgramCode& program_code, u32 main_offset,
-                                                Maxwell3D::Regs::ShaderStage stage,
-                                                const std::string& suffix);
+std::optional<ProgramResult> DecompileProgram(const ProgramCode& program_code, u32 main_offset,
+                                              Maxwell3D::Regs::ShaderStage stage,
+                                              const std::string& suffix);
 
 } // namespace OpenGL::GLShader::Decompiler

src/video_core/renderer_opengl/gl_shader_gen.cpp

@@ -37,7 +37,7 @@ layout(std140) uniform vs_config {
     ProgramResult program =
         Decompiler::DecompileProgram(setup.program.code, PROGRAM_OFFSET,
                                      Maxwell3D::Regs::ShaderStage::Vertex, "vertex")
-            .get_value_or({});
+            .value_or(ProgramResult());
 
     out += program.first;
 
@@ -45,7 +45,7 @@ layout(std140) uniform vs_config {
         ProgramResult program_b =
             Decompiler::DecompileProgram(setup.program.code_b, PROGRAM_OFFSET,
                                          Maxwell3D::Regs::ShaderStage::Vertex, "vertex_b")
-                .get_value_or({});
+                .value_or(ProgramResult());
         out += program_b.first;
     }
 
@@ -82,15 +82,15 @@ void main() {
 }
 
 ProgramResult GenerateGeometryShader(const ShaderSetup& setup) {
-    std::string out = "#version 430 core\n";
-    out += "#extension GL_ARB_separate_shader_objects : enable\n\n";
+    // Version is intentionally skipped in shader generation, it's added by the lazy compilation.
+    std::string out = "#extension GL_ARB_separate_shader_objects : enable\n\n";
     out += Decompiler::GetCommonDeclarations();
     out += "bool exec_geometry();\n";
 
     ProgramResult program =
         Decompiler::DecompileProgram(setup.program.code, PROGRAM_OFFSET,
                                      Maxwell3D::Regs::ShaderStage::Geometry, "geometry")
-            .get_value_or({});
+            .value_or(ProgramResult());
     out += R"(
 out gl_PerVertex {
     vec4 gl_Position;
@@ -124,7 +124,7 @@ ProgramResult GenerateFragmentShader(const ShaderSetup& setup) {
     ProgramResult program =
         Decompiler::DecompileProgram(setup.program.code, PROGRAM_OFFSET,
                                      Maxwell3D::Regs::ShaderStage::Fragment, "fragment")
-            .get_value_or({});
+            .value_or(ProgramResult());
     out += R"(
 layout(location = 0) out vec4 FragColor0;
 layout(location = 1) out vec4 FragColor1;

src/video_core/renderer_opengl/gl_shader_manager.h

@@ -7,6 +7,7 @@
 #include <glad/glad.h>
 
 #include "video_core/renderer_opengl/gl_resource_manager.h"
+#include "video_core/renderer_opengl/gl_state.h"
 #include "video_core/renderer_opengl/maxwell_to_gl.h"
 
 namespace OpenGL::GLShader {

src/video_core/renderer_opengl/gl_state.cpp

@@ -11,9 +11,10 @@
 namespace OpenGL {
 
 OpenGLState OpenGLState::cur_state;
-
+bool OpenGLState::s_rgb_used;
 OpenGLState::OpenGLState() {
     // These all match default OpenGL values
+    framebuffer_srgb.enabled = false;
     cull.enabled = false;
     cull.mode = GL_BACK;
     cull.front_face = GL_CCW;
@@ -21,14 +22,15 @@ OpenGLState::OpenGLState() {
     depth.test_enabled = false;
     depth.test_func = GL_LESS;
     depth.write_mask = GL_TRUE;
-    depth.depth_range_near = 0.0f;
-    depth.depth_range_far = 1.0f;
-
-    color_mask.red_enabled = GL_TRUE;
-    color_mask.green_enabled = GL_TRUE;
-    color_mask.blue_enabled = GL_TRUE;
-    color_mask.alpha_enabled = GL_TRUE;
 
+    primitive_restart.enabled = false;
+    primitive_restart.index = 0;
+    for (auto& item : color_mask) {
+        item.red_enabled = GL_TRUE;
+        item.green_enabled = GL_TRUE;
+        item.blue_enabled = GL_TRUE;
+        item.alpha_enabled = GL_TRUE;
+    }
     stencil.test_enabled = false;
     auto reset_stencil = [](auto& config) {
         config.test_func = GL_ALWAYS;
@@ -41,19 +43,33 @@ OpenGLState::OpenGLState() {
     };
     reset_stencil(stencil.front);
     reset_stencil(stencil.back);
-
-    blend.enabled = true;
-    blend.rgb_equation = GL_FUNC_ADD;
-    blend.a_equation = GL_FUNC_ADD;
-    blend.src_rgb_func = GL_ONE;
-    blend.dst_rgb_func = GL_ZERO;
-    blend.src_a_func = GL_ONE;
-    blend.dst_a_func = GL_ZERO;
-    blend.color.red = 0.0f;
-    blend.color.green = 0.0f;
-    blend.color.blue = 0.0f;
-    blend.color.alpha = 0.0f;
-
+    for (auto& item : viewports) {
+        item.x = 0;
+        item.y = 0;
+        item.width = 0;
+        item.height = 0;
+        item.depth_range_near = 0.0f;
+        item.depth_range_far = 1.0f;
+    }
+    scissor.enabled = false;
+    scissor.x = 0;
+    scissor.y = 0;
+    scissor.width = 0;
+    scissor.height = 0;
+    for (auto& item : blend) {
+        item.enabled = true;
+        item.rgb_equation = GL_FUNC_ADD;
+        item.a_equation = GL_FUNC_ADD;
+        item.src_rgb_func = GL_ONE;
+        item.dst_rgb_func = GL_ZERO;
+        item.src_a_func = GL_ONE;
+        item.dst_a_func = GL_ZERO;
+    }
+    independant_blend.enabled = false;
+    blend_color.red = 0.0f;
+    blend_color.green = 0.0f;
+    blend_color.blue = 0.0f;
+    blend_color.alpha = 0.0f;
     logic_op.enabled = false;
     logic_op.operation = GL_COPY;
 
@@ -69,144 +85,308 @@ OpenGLState::OpenGLState() {
     draw.shader_program = 0;
     draw.program_pipeline = 0;
 
-    scissor.enabled = false;
-    scissor.x = 0;
-    scissor.y = 0;
-    scissor.width = 0;
-    scissor.height = 0;
-
-    viewport.x = 0;
-    viewport.y = 0;
-    viewport.width = 0;
-    viewport.height = 0;
-
     clip_distance = {};
 
     point.size = 1;
 }
 
-void OpenGLState::Apply() const {
+void OpenGLState::ApplyDefaultState() {
+    glDisable(GL_FRAMEBUFFER_SRGB);
+    glDisable(GL_CULL_FACE);
+    glDisable(GL_DEPTH_TEST);
+    glDisable(GL_PRIMITIVE_RESTART);
+    glDisable(GL_STENCIL_TEST);
+    glEnable(GL_BLEND);
+    glDisable(GL_COLOR_LOGIC_OP);
+    glDisable(GL_SCISSOR_TEST);
+}
+
+void OpenGLState::ApplySRgb() const {
+    // sRGB
+    if (framebuffer_srgb.enabled != cur_state.framebuffer_srgb.enabled) {
+        if (framebuffer_srgb.enabled) {
+            // Track if sRGB is used
+            s_rgb_used = true;
+            glEnable(GL_FRAMEBUFFER_SRGB);
+        } else {
+            glDisable(GL_FRAMEBUFFER_SRGB);
+        }
+    }
+}
+
+void OpenGLState::ApplyCulling() const {
     // Culling
-    if (cull.enabled != cur_state.cull.enabled) {
+    const bool cull_changed = cull.enabled != cur_state.cull.enabled;
+    if (cull_changed) {
         if (cull.enabled) {
             glEnable(GL_CULL_FACE);
         } else {
             glDisable(GL_CULL_FACE);
         }
     }
+    if (cull.enabled) {
+        if (cull_changed || cull.mode != cur_state.cull.mode) {
+            glCullFace(cull.mode);
+        }
 
-    if (cull.mode != cur_state.cull.mode) {
-        glCullFace(cull.mode);
+        if (cull_changed || cull.front_face != cur_state.cull.front_face) {
+            glFrontFace(cull.front_face);
+        }
     }
+}
 
-    if (cull.front_face != cur_state.cull.front_face) {
-        glFrontFace(cull.front_face);
+void OpenGLState::ApplyColorMask() const {
+    if (GLAD_GL_ARB_viewport_array) {
+        for (size_t i = 0; i < Tegra::Engines::Maxwell3D::Regs::NumRenderTargets; i++) {
+            const auto& updated = color_mask[i];
+            const auto& current = cur_state.color_mask[i];
+            if (updated.red_enabled != current.red_enabled ||
+                updated.green_enabled != current.green_enabled ||
+                updated.blue_enabled != current.blue_enabled ||
+                updated.alpha_enabled != current.alpha_enabled) {
+                glColorMaski(static_cast<GLuint>(i), updated.red_enabled, updated.green_enabled,
+                             updated.blue_enabled, updated.alpha_enabled);
+            }
+        }
+    } else {
+        const auto& updated = color_mask[0];
+        const auto& current = cur_state.color_mask[0];
+        if (updated.red_enabled != current.red_enabled ||
+            updated.green_enabled != current.green_enabled ||
+            updated.blue_enabled != current.blue_enabled ||
+            updated.alpha_enabled != current.alpha_enabled) {
+            glColorMask(updated.red_enabled, updated.green_enabled, updated.blue_enabled,
+                        updated.alpha_enabled);
+        }
     }
+}
 
+void OpenGLState::ApplyDepth() const {
     // Depth test
-    if (depth.test_enabled != cur_state.depth.test_enabled) {
+    const bool depth_test_changed = depth.test_enabled != cur_state.depth.test_enabled;
+    if (depth_test_changed) {
         if (depth.test_enabled) {
             glEnable(GL_DEPTH_TEST);
         } else {
             glDisable(GL_DEPTH_TEST);
         }
     }
-
-    if (depth.test_func != cur_state.depth.test_func) {
+    if (depth.test_enabled &&
+        (depth_test_changed || depth.test_func != cur_state.depth.test_func)) {
         glDepthFunc(depth.test_func);
     }
-
     // Depth mask
     if (depth.write_mask != cur_state.depth.write_mask) {
         glDepthMask(depth.write_mask);
     }
+}
 
-    // Depth range
-    if (depth.depth_range_near != cur_state.depth.depth_range_near ||
-        depth.depth_range_far != cur_state.depth.depth_range_far) {
-        glDepthRange(depth.depth_range_near, depth.depth_range_far);
+void OpenGLState::ApplyPrimitiveRestart() const {
+    const bool primitive_restart_changed =
+        primitive_restart.enabled != cur_state.primitive_restart.enabled;
+    if (primitive_restart_changed) {
+        if (primitive_restart.enabled) {
+            glEnable(GL_PRIMITIVE_RESTART);
+        } else {
+            glDisable(GL_PRIMITIVE_RESTART);
+        }
     }
-
-    // Color mask
-    if (color_mask.red_enabled != cur_state.color_mask.red_enabled ||
-        color_mask.green_enabled != cur_state.color_mask.green_enabled ||
-        color_mask.blue_enabled != cur_state.color_mask.blue_enabled ||
-        color_mask.alpha_enabled != cur_state.color_mask.alpha_enabled) {
-        glColorMask(color_mask.red_enabled, color_mask.green_enabled, color_mask.blue_enabled,
-                    color_mask.alpha_enabled);
+    if (primitive_restart_changed ||
+        (primitive_restart.enabled &&
+         primitive_restart.index != cur_state.primitive_restart.index)) {
+        glPrimitiveRestartIndex(primitive_restart.index);
     }
+}
 
-    // Stencil test
-    if (stencil.test_enabled != cur_state.stencil.test_enabled) {
+void OpenGLState::ApplyStencilTest() const {
+    const bool stencil_test_changed = stencil.test_enabled != cur_state.stencil.test_enabled;
+    if (stencil_test_changed) {
         if (stencil.test_enabled) {
             glEnable(GL_STENCIL_TEST);
         } else {
             glDisable(GL_STENCIL_TEST);
         }
     }
-    auto config_stencil = [](GLenum face, const auto& config, const auto& prev_config) {
-        if (config.test_func != prev_config.test_func || config.test_ref != prev_config.test_ref ||
-            config.test_mask != prev_config.test_mask) {
-            glStencilFuncSeparate(face, config.test_func, config.test_ref, config.test_mask);
-        }
-        if (config.action_depth_fail != prev_config.action_depth_fail ||
-            config.action_depth_pass != prev_config.action_depth_pass ||
-            config.action_stencil_fail != prev_config.action_stencil_fail) {
-            glStencilOpSeparate(face, config.action_stencil_fail, config.action_depth_fail,
-                                config.action_depth_pass);
-        }
-        if (config.write_mask != prev_config.write_mask) {
-            glStencilMaskSeparate(face, config.write_mask);
-        }
-    };
-    config_stencil(GL_FRONT, stencil.front, cur_state.stencil.front);
-    config_stencil(GL_BACK, stencil.back, cur_state.stencil.back);
+    if (stencil.test_enabled) {
+        auto config_stencil = [stencil_test_changed](GLenum face, const auto& config,
+                                                     const auto& prev_config) {
+            if (stencil_test_changed || config.test_func != prev_config.test_func ||
+                config.test_ref != prev_config.test_ref ||
+                config.test_mask != prev_config.test_mask) {
+                glStencilFuncSeparate(face, config.test_func, config.test_ref, config.test_mask);
+            }
+            if (stencil_test_changed || config.action_depth_fail != prev_config.action_depth_fail ||
+                config.action_depth_pass != prev_config.action_depth_pass ||
+                config.action_stencil_fail != prev_config.action_stencil_fail) {
+                glStencilOpSeparate(face, config.action_stencil_fail, config.action_depth_fail,
+                                    config.action_depth_pass);
+            }
+            if (config.write_mask != prev_config.write_mask) {
+                glStencilMaskSeparate(face, config.write_mask);
+            }
+        };
+        config_stencil(GL_FRONT, stencil.front, cur_state.stencil.front);
+        config_stencil(GL_BACK, stencil.back, cur_state.stencil.back);
+    }
+}
 
-    // Blending
-    if (blend.enabled != cur_state.blend.enabled) {
-        if (blend.enabled) {
-            ASSERT(!logic_op.enabled);
+void OpenGLState::ApplyScissor() const {
+    const bool scissor_changed = scissor.enabled != cur_state.scissor.enabled;
+    if (scissor_changed) {
+        if (scissor.enabled) {
+            glEnable(GL_SCISSOR_TEST);
+        } else {
+            glDisable(GL_SCISSOR_TEST);
+        }
+    }
+    if (scissor.enabled &&
+        (scissor_changed || scissor.x != cur_state.scissor.x || scissor.y != cur_state.scissor.y ||
+         scissor.width != cur_state.scissor.width || scissor.height != cur_state.scissor.height)) {
+        glScissor(scissor.x, scissor.y, scissor.width, scissor.height);
+    }
+}
+
+void OpenGLState::ApplyViewport() const {
+    if (GLAD_GL_ARB_viewport_array) {
+        for (GLuint i = 0;
+             i < static_cast<GLuint>(Tegra::Engines::Maxwell3D::Regs::NumRenderTargets); i++) {
+            const auto& current = cur_state.viewports[i];
+            const auto& updated = viewports[i];
+            if (updated.x != current.x || updated.y != current.y ||
+                updated.width != current.width || updated.height != current.height) {
+                glViewportIndexedf(i, updated.x, updated.y, updated.width, updated.height);
+            }
+            if (updated.depth_range_near != current.depth_range_near ||
+                updated.depth_range_far != current.depth_range_far) {
+                glDepthRangeIndexed(i, updated.depth_range_near, updated.depth_range_far);
+            }
+        }
+    } else {
+        const auto& current = cur_state.viewports[0];
+        const auto& updated = viewports[0];
+        if (updated.x != current.x || updated.y != current.y || updated.width != current.width ||
+            updated.height != current.height) {
+            glViewport(updated.x, updated.y, updated.width, updated.height);
+        }
+        if (updated.depth_range_near != current.depth_range_near ||
+            updated.depth_range_far != current.depth_range_far) {
+            glDepthRange(updated.depth_range_near, updated.depth_range_far);
+        }
+    }
+}
+
+void OpenGLState::ApplyGlobalBlending() const {
+    const Blend& current = cur_state.blend[0];
+    const Blend& updated = blend[0];
+    const bool blend_changed = updated.enabled != current.enabled;
+    if (blend_changed) {
+        if (updated.enabled) {
             glEnable(GL_BLEND);
         } else {
             glDisable(GL_BLEND);
         }
     }
-
-    if (blend.color.red != cur_state.blend.color.red ||
-        blend.color.green != cur_state.blend.color.green ||
-        blend.color.blue != cur_state.blend.color.blue ||
-        blend.color.alpha != cur_state.blend.color.alpha) {
-        glBlendColor(blend.color.red, blend.color.green, blend.color.blue, blend.color.alpha);
+    if (!updated.enabled) {
+        return;
     }
+    if (updated.separate_alpha) {
+        if (blend_changed || updated.src_rgb_func != current.src_rgb_func ||
+            updated.dst_rgb_func != current.dst_rgb_func ||
+            updated.src_a_func != current.src_a_func || updated.dst_a_func != current.dst_a_func) {
+            glBlendFuncSeparate(updated.src_rgb_func, updated.dst_rgb_func, updated.src_a_func,
+                                updated.dst_a_func);
+        }
 
-    if (blend.src_rgb_func != cur_state.blend.src_rgb_func ||
-        blend.dst_rgb_func != cur_state.blend.dst_rgb_func ||
-        blend.src_a_func != cur_state.blend.src_a_func ||
-        blend.dst_a_func != cur_state.blend.dst_a_func) {
-        glBlendFuncSeparate(blend.src_rgb_func, blend.dst_rgb_func, blend.src_a_func,
-                            blend.dst_a_func);
+        if (blend_changed || updated.rgb_equation != current.rgb_equation ||
+            updated.a_equation != current.a_equation) {
+            glBlendEquationSeparate(updated.rgb_equation, updated.a_equation);
+        }
+    } else {
+        if (blend_changed || updated.src_rgb_func != current.src_rgb_func ||
+            updated.dst_rgb_func != current.dst_rgb_func) {
+            glBlendFunc(updated.src_rgb_func, updated.dst_rgb_func);
+        }
+
+        if (blend_changed || updated.rgb_equation != current.rgb_equation) {
+            glBlendEquation(updated.rgb_equation);
+        }
     }
+}
 
-    if (blend.rgb_equation != cur_state.blend.rgb_equation ||
-        blend.a_equation != cur_state.blend.a_equation) {
-        glBlendEquationSeparate(blend.rgb_equation, blend.a_equation);
+void OpenGLState::ApplyTargetBlending(int target, bool force) const {
+    const Blend& updated = blend[target];
+    const Blend& current = cur_state.blend[target];
+    const bool blend_changed = updated.enabled != current.enabled || force;
+    if (blend_changed) {
+        if (updated.enabled) {
+            glEnablei(GL_BLEND, static_cast<GLuint>(target));
+        } else {
+            glDisablei(GL_BLEND, static_cast<GLuint>(target));
+        }
     }
+    if (!updated.enabled) {
+        return;
+    }
+    if (updated.separate_alpha) {
+        if (blend_changed || updated.src_rgb_func != current.src_rgb_func ||
+            updated.dst_rgb_func != current.dst_rgb_func ||
+            updated.src_a_func != current.src_a_func || updated.dst_a_func != current.dst_a_func) {
+            glBlendFuncSeparateiARB(static_cast<GLuint>(target), updated.src_rgb_func,
+                                    updated.dst_rgb_func, updated.src_a_func, updated.dst_a_func);
+        }
 
-    // Logic Operation
-    if (logic_op.enabled != cur_state.logic_op.enabled) {
+        if (blend_changed || updated.rgb_equation != current.rgb_equation ||
+            updated.a_equation != current.a_equation) {
+            glBlendEquationSeparateiARB(static_cast<GLuint>(target), updated.rgb_equation,
+                                        updated.a_equation);
+        }
+    } else {
+        if (blend_changed || updated.src_rgb_func != current.src_rgb_func ||
+            updated.dst_rgb_func != current.dst_rgb_func) {
+            glBlendFunciARB(static_cast<GLuint>(target), updated.src_rgb_func,
+                            updated.dst_rgb_func);
+        }
+
+        if (blend_changed || updated.rgb_equation != current.rgb_equation) {
+            glBlendEquationiARB(static_cast<GLuint>(target), updated.rgb_equation);
+        }
+    }
+}
+
+void OpenGLState::ApplyBlending() const {
+    if (independant_blend.enabled) {
+        for (size_t i = 0; i < Tegra::Engines::Maxwell3D::Regs::NumRenderTargets; i++) {
+            ApplyTargetBlending(i,
+                                independant_blend.enabled != cur_state.independant_blend.enabled);
+        }
+    } else {
+        ApplyGlobalBlending();
+    }
+    if (blend_color.red != cur_state.blend_color.red ||
+        blend_color.green != cur_state.blend_color.green ||
+        blend_color.blue != cur_state.blend_color.blue ||
+        blend_color.alpha != cur_state.blend_color.alpha) {
+        glBlendColor(blend_color.red, blend_color.green, blend_color.blue, blend_color.alpha);
+    }
+}
+
+void OpenGLState::ApplyLogicOp() const {
+    const bool logic_op_changed = logic_op.enabled != cur_state.logic_op.enabled;
+    if (logic_op_changed) {
         if (logic_op.enabled) {
-            ASSERT(!blend.enabled);
             glEnable(GL_COLOR_LOGIC_OP);
         } else {
             glDisable(GL_COLOR_LOGIC_OP);
         }
     }
 
-    if (logic_op.operation != cur_state.logic_op.operation) {
+    if (logic_op.enabled &&
+        (logic_op_changed || logic_op.operation != cur_state.logic_op.operation)) {
         glLogicOp(logic_op.operation);
     }
+}
 
-    // Textures
+void OpenGLState::ApplyTextures() const {
     for (std::size_t i = 0; i < std::size(texture_units); ++i) {
         const auto& texture_unit = texture_units[i];
         const auto& cur_state_texture_unit = cur_state.texture_units[i];
@@ -225,28 +405,29 @@ void OpenGLState::Apply() const {
             glTexParameteriv(texture_unit.target, GL_TEXTURE_SWIZZLE_RGBA, mask.data());
         }
     }
+}
 
-    // Samplers
-    {
-        bool has_delta{};
-        std::size_t first{}, last{};
-        std::array<GLuint, Tegra::Engines::Maxwell3D::Regs::NumTextureSamplers> samplers;
-        for (std::size_t i = 0; i < std::size(samplers); ++i) {
-            samplers[i] = texture_units[i].sampler;
-            if (samplers[i] != cur_state.texture_units[i].sampler) {
-                if (!has_delta) {
-                    first = i;
-                    has_delta = true;
-                }
-                last = i;
+void OpenGLState::ApplySamplers() const {
+    bool has_delta{};
+    std::size_t first{}, last{};
+    std::array<GLuint, Tegra::Engines::Maxwell3D::Regs::NumTextureSamplers> samplers;
+    for (std::size_t i = 0; i < std::size(samplers); ++i) {
+        samplers[i] = texture_units[i].sampler;
+        if (samplers[i] != cur_state.texture_units[i].sampler) {
+            if (!has_delta) {
+                first = i;
+                has_delta = true;
             }
-        }
-        if (has_delta) {
-            glBindSamplers(static_cast<GLuint>(first), static_cast<GLsizei>(last - first + 1),
-                           samplers.data());
+            last = i;
         }
     }
+    if (has_delta) {
+        glBindSamplers(static_cast<GLuint>(first), static_cast<GLsizei>(last - first + 1),
+                       samplers.data());
+    }
+}
 
+void OpenGLState::ApplyFramebufferState() const {
     // Framebuffer
     if (draw.read_framebuffer != cur_state.draw.read_framebuffer) {
         glBindFramebuffer(GL_READ_FRAMEBUFFER, draw.read_framebuffer);
@@ -254,7 +435,9 @@ void OpenGLState::Apply() const {
     if (draw.draw_framebuffer != cur_state.draw.draw_framebuffer) {
         glBindFramebuffer(GL_DRAW_FRAMEBUFFER, draw.draw_framebuffer);
     }
+}
 
+void OpenGLState::ApplyVertexBufferState() const {
     // Vertex array
     if (draw.vertex_array != cur_state.draw.vertex_array) {
         glBindVertexArray(draw.vertex_array);
@@ -264,7 +447,11 @@ void OpenGLState::Apply() const {
     if (draw.vertex_buffer != cur_state.draw.vertex_buffer) {
         glBindBuffer(GL_ARRAY_BUFFER, draw.vertex_buffer);
     }
+}
 
+void OpenGLState::Apply() const {
+    ApplyFramebufferState();
+    ApplyVertexBufferState();
     // Uniform buffer
     if (draw.uniform_buffer != cur_state.draw.uniform_buffer) {
         glBindBuffer(GL_UNIFORM_BUFFER, draw.uniform_buffer);
@@ -279,27 +466,6 @@ void OpenGLState::Apply() const {
     if (draw.program_pipeline != cur_state.draw.program_pipeline) {
         glBindProgramPipeline(draw.program_pipeline);
     }
-
-    // Scissor test
-    if (scissor.enabled != cur_state.scissor.enabled) {
-        if (scissor.enabled) {
-            glEnable(GL_SCISSOR_TEST);
-        } else {
-            glDisable(GL_SCISSOR_TEST);
-        }
-    }
-
-    if (scissor.x != cur_state.scissor.x || scissor.y != cur_state.scissor.y ||
-        scissor.width != cur_state.scissor.width || scissor.height != cur_state.scissor.height) {
-        glScissor(scissor.x, scissor.y, scissor.width, scissor.height);
-    }
-
-    if (viewport.x != cur_state.viewport.x || viewport.y != cur_state.viewport.y ||
-        viewport.width != cur_state.viewport.width ||
-        viewport.height != cur_state.viewport.height) {
-        glViewport(viewport.x, viewport.y, viewport.width, viewport.height);
-    }
-
     // Clip distance
     for (std::size_t i = 0; i < clip_distance.size(); ++i) {
         if (clip_distance[i] != cur_state.clip_distance[i]) {
@@ -310,12 +476,22 @@ void OpenGLState::Apply() const {
             }
         }
     }
-
     // Point
     if (point.size != cur_state.point.size) {
         glPointSize(point.size);
     }
-
+    ApplyColorMask();
+    ApplyViewport();
+    ApplyScissor();
+    ApplyStencilTest();
+    ApplySRgb();
+    ApplyCulling();
+    ApplyDepth();
+    ApplyPrimitiveRestart();
+    ApplyBlending();
+    ApplyLogicOp();
+    ApplyTextures();
+    ApplySamplers();
     cur_state = *this;
 }
 

src/video_core/renderer_opengl/gl_state.h

@@ -35,6 +35,10 @@ constexpr TextureUnit ProcTexDiffLUT{9};
 
 class OpenGLState {
 public:
+    struct {
+        bool enabled; // GL_FRAMEBUFFER_SRGB
+    } framebuffer_srgb;
+
     struct {
         bool enabled;      // GL_CULL_FACE
         GLenum mode;       // GL_CULL_FACE_MODE
@@ -42,20 +46,24 @@ public:
     } cull;
 
     struct {
-        bool test_enabled;        // GL_DEPTH_TEST
-        GLenum test_func;         // GL_DEPTH_FUNC
-        GLboolean write_mask;     // GL_DEPTH_WRITEMASK
-        GLfloat depth_range_near; // GL_DEPTH_RANGE
-        GLfloat depth_range_far;  // GL_DEPTH_RANGE
+        bool test_enabled;    // GL_DEPTH_TEST
+        GLenum test_func;     // GL_DEPTH_FUNC
+        GLboolean write_mask; // GL_DEPTH_WRITEMASK
     } depth;
 
     struct {
+        bool enabled;
+        GLuint index;
+    } primitive_restart; // GL_PRIMITIVE_RESTART
+
+    struct ColorMask {
         GLboolean red_enabled;
         GLboolean green_enabled;
         GLboolean blue_enabled;
         GLboolean alpha_enabled;
-    } color_mask; // GL_COLOR_WRITEMASK
-
+    };
+    std::array<ColorMask, Tegra::Engines::Maxwell3D::Regs::NumRenderTargets>
+        color_mask; // GL_COLOR_WRITEMASK
     struct {
         bool test_enabled; // GL_STENCIL_TEST
         struct {
@@ -69,22 +77,28 @@ public:
         } front, back;
     } stencil;
 
-    struct {
+    struct Blend {
         bool enabled;        // GL_BLEND
+        bool separate_alpha; // Independent blend enabled
         GLenum rgb_equation; // GL_BLEND_EQUATION_RGB
         GLenum a_equation;   // GL_BLEND_EQUATION_ALPHA
         GLenum src_rgb_func; // GL_BLEND_SRC_RGB
         GLenum dst_rgb_func; // GL_BLEND_DST_RGB
         GLenum src_a_func;   // GL_BLEND_SRC_ALPHA
         GLenum dst_a_func;   // GL_BLEND_DST_ALPHA
+    };
+    std::array<Blend, Tegra::Engines::Maxwell3D::Regs::NumRenderTargets> blend;
 
-        struct {
-            GLclampf red;
-            GLclampf green;
-            GLclampf blue;
-            GLclampf alpha;
-        } color; // GL_BLEND_COLOR
-    } blend;
+    struct {
+        bool enabled;
+    } independant_blend;
+
+    struct {
+        GLclampf red;
+        GLclampf green;
+        GLclampf blue;
+        GLclampf alpha;
+    } blend_color; // GL_BLEND_COLOR
 
     struct {
         bool enabled; // GL_LOGIC_OP_MODE
@@ -129,6 +143,16 @@ public:
         GLuint program_pipeline; // GL_PROGRAM_PIPELINE_BINDING
     } draw;
 
+    struct viewport {
+        GLfloat x;
+        GLfloat y;
+        GLfloat width;
+        GLfloat height;
+        GLfloat depth_range_near; // GL_DEPTH_RANGE
+        GLfloat depth_range_far;  // GL_DEPTH_RANGE
+    };
+    std::array<viewport, Tegra::Engines::Maxwell3D::Regs::NumRenderTargets> viewports;
+
     struct {
         bool enabled; // GL_SCISSOR_TEST
         GLint x;
@@ -137,13 +161,6 @@ public:
         GLsizei height;
     } scissor;
 
-    struct {
-        GLint x;
-        GLint y;
-        GLsizei width;
-        GLsizei height;
-    } viewport;
-
     struct {
         float size; // GL_POINT_SIZE
     } point;
@@ -156,10 +173,20 @@ public:
     static OpenGLState GetCurState() {
         return cur_state;
     }
-
+    static bool GetsRGBUsed() {
+        return s_rgb_used;
+    }
+    static void ClearsRGBUsed() {
+        s_rgb_used = false;
+    }
     /// Apply this state as the current OpenGL state
     void Apply() const;
-
+    /// Apply only the state afecting the framebuffer
+    void ApplyFramebufferState() const;
+    /// Apply only the state afecting the vertex buffer
+    void ApplyVertexBufferState() const;
+    /// Set the initial OpenGL state
+    static void ApplyDefaultState();
     /// Resets any references to the given resource
     OpenGLState& UnbindTexture(GLuint handle);
     OpenGLState& ResetSampler(GLuint handle);
@@ -171,6 +198,23 @@ public:
 
 private:
     static OpenGLState cur_state;
+    // Workaround for sRGB problems caused by
+    // QT not supporting srgb output
+    static bool s_rgb_used;
+    void ApplySRgb() const;
+    void ApplyCulling() const;
+    void ApplyColorMask() const;
+    void ApplyDepth() const;
+    void ApplyPrimitiveRestart() const;
+    void ApplyStencilTest() const;
+    void ApplyViewport() const;
+    void ApplyTargetBlending(int target, bool force) const;
+    void ApplyGlobalBlending() const;
+    void ApplyBlending() const;
+    void ApplyLogicOp() const;
+    void ApplyTextures() const;
+    void ApplySamplers() const;
+    void ApplyScissor() const;
 };
 
 } // namespace OpenGL