Address Comments

input_common: First implementation of controller rumble

.ci/yuzu-patreon-step2.yml

@@ -9,6 +9,7 @@ stages:
   displayName: 'build'
   jobs:
   - job: build
+    timeoutInMinutes: 120
     displayName: 'windows-msvc'
     pool:
       vmImage: windows-2019

src/audio_core/CMakeLists.txt

@@ -44,6 +44,16 @@ add_library(audio_core STATIC
 
 create_target_directory_groups(audio_core)
 
+if (NOT MSVC)
+    target_compile_options(audio_core PRIVATE
+        -Werror=ignored-qualifiers
+        -Werror=implicit-fallthrough
+        -Werror=reorder
+        -Werror=sign-compare
+        -Werror=unused-variable
+    )
+endif()
+
 target_link_libraries(audio_core PUBLIC common core)
 target_link_libraries(audio_core PRIVATE SoundTouch)
 

src/audio_core/audio_renderer.cpp

@@ -3,16 +3,13 @@
 // Refer to the license.txt file included.
 
 #include <vector>
-#include "audio_core/algorithm/interpolate.h"
+
 #include "audio_core/audio_out.h"
 #include "audio_core/audio_renderer.h"
-#include "audio_core/codec.h"
 #include "audio_core/common.h"
 #include "audio_core/info_updater.h"
 #include "audio_core/voice_context.h"
-#include "common/assert.h"
 #include "common/logging/log.h"
-#include "core/core.h"
 #include "core/hle/kernel/writable_event.h"
 #include "core/memory.h"
 #include "core/settings.h"

src/audio_core/audio_renderer.h

@@ -21,7 +21,6 @@
 #include "common/common_funcs.h"
 #include "common/common_types.h"
 #include "common/swap.h"
-#include "core/hle/kernel/object.h"
 #include "core/hle/result.h"
 
 namespace Core::Timing {

src/audio_core/behavior_info.cpp

@@ -57,15 +57,15 @@ bool BehaviorInfo::IsLongSizePreDelaySupported() const {
     return AudioCommon::IsRevisionSupported(3, user_revision);
 }
 
-bool BehaviorInfo::IsAudioRenererProcessingTimeLimit80PercentSupported() const {
+bool BehaviorInfo::IsAudioRendererProcessingTimeLimit80PercentSupported() const {
     return AudioCommon::IsRevisionSupported(5, user_revision);
 }
 
-bool BehaviorInfo::IsAudioRenererProcessingTimeLimit75PercentSupported() const {
+bool BehaviorInfo::IsAudioRendererProcessingTimeLimit75PercentSupported() const {
     return AudioCommon::IsRevisionSupported(4, user_revision);
 }
 
-bool BehaviorInfo::IsAudioRenererProcessingTimeLimit70PercentSupported() const {
+bool BehaviorInfo::IsAudioRendererProcessingTimeLimit70PercentSupported() const {
     return AudioCommon::IsRevisionSupported(1, user_revision);
 }
 

src/audio_core/behavior_info.h

@@ -49,9 +49,9 @@ public:
     bool IsAdpcmLoopContextBugFixed() const;
     bool IsSplitterSupported() const;
     bool IsLongSizePreDelaySupported() const;
-    bool IsAudioRenererProcessingTimeLimit80PercentSupported() const;
-    bool IsAudioRenererProcessingTimeLimit75PercentSupported() const;
-    bool IsAudioRenererProcessingTimeLimit70PercentSupported() const;
+    bool IsAudioRendererProcessingTimeLimit80PercentSupported() const;
+    bool IsAudioRendererProcessingTimeLimit75PercentSupported() const;
+    bool IsAudioRendererProcessingTimeLimit70PercentSupported() const;
     bool IsElapsedFrameCountSupported() const;
     bool IsMemoryPoolForceMappingEnabled() const;
     bool IsFlushVoiceWaveBuffersSupported() const;

src/audio_core/codec.cpp

@@ -16,8 +16,9 @@ std::vector<s16> DecodeADPCM(const u8* const data, std::size_t size, const ADPCM
 
     constexpr std::size_t FRAME_LEN = 8;
     constexpr std::size_t SAMPLES_PER_FRAME = 14;
-    constexpr std::array<int, 16> SIGNED_NIBBLES = {
-        {0, 1, 2, 3, 4, 5, 6, 7, -8, -7, -6, -5, -4, -3, -2, -1}};
+    static constexpr std::array<int, 16> SIGNED_NIBBLES{
+        0, 1, 2, 3, 4, 5, 6, 7, -8, -7, -6, -5, -4, -3, -2, -1,
+    };
 
     const std::size_t sample_count = (size / FRAME_LEN) * SAMPLES_PER_FRAME;
     const std::size_t ret_size =

src/audio_core/codec.h

@@ -38,7 +38,7 @@ using ADPCM_Coeff = std::array<s16, 16>;
  * @param state ADPCM state, this is updated with new state
  * @return Decoded stereo signed PCM16 data, sample_count in length
  */
-std::vector<s16> DecodeADPCM(const u8* const data, std::size_t size, const ADPCM_Coeff& coeff,
+std::vector<s16> DecodeADPCM(const u8* data, std::size_t size, const ADPCM_Coeff& coeff,
                              ADPCMState& state);
 
 }; // namespace AudioCore::Codec

src/audio_core/command_generator.cpp

@@ -152,7 +152,7 @@ void CommandGenerator::GenerateVoiceCommand(ServerVoiceInfo& voice_info) {
                     if (!destination_data->IsConfigured()) {
                         continue;
                     }
-                    if (destination_data->GetMixId() >= mix_context.GetCount()) {
+                    if (destination_data->GetMixId() >= static_cast<int>(mix_context.GetCount())) {
                         continue;
                     }
 
@@ -435,7 +435,7 @@ void CommandGenerator::GenerateAuxCommand(s32 mix_buffer_offset, EffectBase* inf
                     GetMixBuffer(output_index), worker_params.sample_count, offset, write_count);
                 memory.WriteBlock(aux->GetRecvInfo(), &recv_info, sizeof(AuxInfoDSP));
 
-                if (samples_read != worker_params.sample_count &&
+                if (samples_read != static_cast<int>(worker_params.sample_count) &&
                     samples_read <= params.sample_count) {
                     std::memset(GetMixBuffer(output_index), 0, params.sample_count - samples_read);
                 }
@@ -611,7 +611,8 @@ void CommandGenerator::GenerateMixCommands(ServerMixInfo& mix_info) {
             const auto& dest_mix = mix_context.GetInfo(destination_data->GetMixId());
             const auto& dest_in_params = dest_mix.GetInParams();
             const auto mix_index = (base - 1) % in_params.buffer_count + in_params.buffer_offset;
-            for (std::size_t i = 0; i < dest_in_params.buffer_count; i++) {
+            for (std::size_t i = 0; i < static_cast<std::size_t>(dest_in_params.buffer_count);
+                 i++) {
                 const auto mixed_volume = in_params.volume * destination_data->GetMixVolume(i);
                 if (mixed_volume != 0.0f) {
                     GenerateMixCommand(dest_in_params.buffer_offset + i, mix_index, mixed_volume,
@@ -704,7 +705,7 @@ s32 CommandGenerator::DecodePcm16(ServerVoiceInfo& voice_info, VoiceState& dsp_s
         std::vector<s16> buffer(samples_processed * channel_count);
         memory.ReadBlock(buffer_pos, buffer.data(), buffer.size() * sizeof(s16));
 
-        for (std::size_t i = 0; i < samples_processed; i++) {
+        for (std::size_t i = 0; i < static_cast<std::size_t>(samples_processed); i++) {
             sample_buffer[mix_offset + i] = buffer[i * channel_count + channel];
         }
     }
@@ -726,8 +727,9 @@ s32 CommandGenerator::DecodeAdpcm(ServerVoiceInfo& voice_info, VoiceState& dsp_s
         return 0;
     }
 
-    constexpr std::array<int, 16> SIGNED_NIBBLES = {
-        {0, 1, 2, 3, 4, 5, 6, 7, -8, -7, -6, -5, -4, -3, -2, -1}};
+    static constexpr std::array<int, 16> SIGNED_NIBBLES{
+        0, 1, 2, 3, 4, 5, 6, 7, -8, -7, -6, -5, -4, -3, -2, -1,
+    };
 
     constexpr std::size_t FRAME_LEN = 8;
     constexpr std::size_t NIBBLES_PER_SAMPLE = 16;
@@ -789,7 +791,7 @@ s32 CommandGenerator::DecodeAdpcm(ServerVoiceInfo& voice_info, VoiceState& dsp_s
             position_in_frame += 2;
 
             // Decode entire frame
-            if (remaining_samples >= SAMPLES_PER_FRAME) {
+            if (remaining_samples >= static_cast<int>(SAMPLES_PER_FRAME)) {
                 for (std::size_t i = 0; i < SAMPLES_PER_FRAME / 2; i++) {
 
                     // Sample 1
@@ -866,7 +868,6 @@ void CommandGenerator::DecodeFromWaveBuffers(ServerVoiceInfo& voice_info, s32* o
     const auto resample_rate = static_cast<s32>(
         static_cast<float>(in_params.sample_rate) / static_cast<float>(target_sample_rate) *
         static_cast<float>(static_cast<s32>(in_params.pitch * 32768.0f)));
-    auto* output_base = output;
     if (dsp_state.fraction + sample_count * resample_rate >
         static_cast<s32>(SCALED_MIX_BUFFER_SIZE - 4ULL)) {
         return;

src/audio_core/command_generator.h

@@ -7,7 +7,6 @@
 #include <array>
 #include "audio_core/common.h"
 #include "audio_core/voice_context.h"
-#include "common/common_funcs.h"
 #include "common/common_types.h"
 
 namespace Core::Memory {

src/audio_core/common.h

@@ -3,6 +3,7 @@
 // Refer to the license.txt file included.
 
 #pragma once
+
 #include "common/common_funcs.h"
 #include "common/common_types.h"
 #include "common/swap.h"

src/audio_core/cubeb_sink.cpp

@@ -192,8 +192,8 @@ SinkStream& CubebSink::AcquireSinkStream(u32 sample_rate, u32 num_channels,
 
 long CubebSinkStream::DataCallback(cubeb_stream* stream, void* user_data, const void* input_buffer,
                                    void* output_buffer, long num_frames) {
-    CubebSinkStream* impl = static_cast<CubebSinkStream*>(user_data);
-    u8* buffer = reinterpret_cast<u8*>(output_buffer);
+    auto* impl = static_cast<CubebSinkStream*>(user_data);
+    auto* buffer = static_cast<u8*>(output_buffer);
 
     if (!impl) {
         return {};

src/audio_core/effect_context.cpp

@@ -184,19 +184,19 @@ void EffectAuxInfo::UpdateForCommandGeneration() {
     }
 }
 
-const VAddr EffectAuxInfo::GetSendInfo() const {
+VAddr EffectAuxInfo::GetSendInfo() const {
     return send_info;
 }
 
-const VAddr EffectAuxInfo::GetSendBuffer() const {
+VAddr EffectAuxInfo::GetSendBuffer() const {
     return send_buffer;
 }
 
-const VAddr EffectAuxInfo::GetRecvInfo() const {
+VAddr EffectAuxInfo::GetRecvInfo() const {
     return recv_info;
 }
 
-const VAddr EffectAuxInfo::GetRecvBuffer() const {
+VAddr EffectAuxInfo::GetRecvBuffer() const {
     return recv_buffer;
 }
 

src/audio_core/effect_context.h

@@ -166,13 +166,13 @@ public:
             std::array<u8, 0xa0> raw;
         };
     };
-    static_assert(sizeof(EffectInfo::InParams) == 0xc0, "InParams is an invalid size");
+    static_assert(sizeof(InParams) == 0xc0, "InParams is an invalid size");
 
     struct OutParams {
         UsageStatus status{};
         INSERT_PADDING_BYTES(15);
     };
-    static_assert(sizeof(EffectInfo::OutParams) == 0x10, "OutParams is an invalid size");
+    static_assert(sizeof(OutParams) == 0x10, "OutParams is an invalid size");
 };
 
 struct AuxAddress {
@@ -184,8 +184,8 @@ struct AuxAddress {
 
 class EffectBase {
 public:
-    EffectBase(EffectType effect_type);
-    ~EffectBase();
+    explicit EffectBase(EffectType effect_type);
+    virtual ~EffectBase();
 
     virtual void Update(EffectInfo::InParams& in_params) = 0;
     virtual void UpdateForCommandGeneration() = 0;
@@ -206,8 +206,7 @@ protected:
 template <typename T>
 class EffectGeneric : public EffectBase {
 public:
-    EffectGeneric(EffectType effect_type) : EffectBase::EffectBase(effect_type) {}
-    ~EffectGeneric() = default;
+    explicit EffectGeneric(EffectType effect_type) : EffectBase(effect_type) {}
 
     T& GetParams() {
         return internal_params;
@@ -224,7 +223,7 @@ private:
 class EffectStubbed : public EffectBase {
 public:
     explicit EffectStubbed();
-    ~EffectStubbed();
+    ~EffectStubbed() override;
 
     void Update(EffectInfo::InParams& in_params) override;
     void UpdateForCommandGeneration() override;
@@ -233,7 +232,7 @@ public:
 class EffectI3dl2Reverb : public EffectGeneric<I3dl2ReverbParams> {
 public:
     explicit EffectI3dl2Reverb();
-    ~EffectI3dl2Reverb();
+    ~EffectI3dl2Reverb() override;
 
     void Update(EffectInfo::InParams& in_params) override;
     void UpdateForCommandGeneration() override;
@@ -245,7 +244,7 @@ private:
 class EffectBiquadFilter : public EffectGeneric<BiquadFilterParams> {
 public:
     explicit EffectBiquadFilter();
-    ~EffectBiquadFilter();
+    ~EffectBiquadFilter() override;
 
     void Update(EffectInfo::InParams& in_params) override;
     void UpdateForCommandGeneration() override;
@@ -254,14 +253,14 @@ public:
 class EffectAuxInfo : public EffectGeneric<AuxInfo> {
 public:
     explicit EffectAuxInfo();
-    ~EffectAuxInfo();
+    ~EffectAuxInfo() override;
 
     void Update(EffectInfo::InParams& in_params) override;
     void UpdateForCommandGeneration() override;
-    const VAddr GetSendInfo() const;
-    const VAddr GetSendBuffer() const;
-    const VAddr GetRecvInfo() const;
-    const VAddr GetRecvBuffer() const;
+    VAddr GetSendInfo() const;
+    VAddr GetSendBuffer() const;
+    VAddr GetRecvInfo() const;
+    VAddr GetRecvBuffer() const;
 
 private:
     VAddr send_info{};
@@ -275,7 +274,7 @@ private:
 class EffectDelay : public EffectGeneric<DelayParams> {
 public:
     explicit EffectDelay();
-    ~EffectDelay();
+    ~EffectDelay() override;
 
     void Update(EffectInfo::InParams& in_params) override;
     void UpdateForCommandGeneration() override;
@@ -287,7 +286,7 @@ private:
 class EffectBufferMixer : public EffectGeneric<BufferMixerParams> {
 public:
     explicit EffectBufferMixer();
-    ~EffectBufferMixer();
+    ~EffectBufferMixer() override;
 
     void Update(EffectInfo::InParams& in_params) override;
     void UpdateForCommandGeneration() override;
@@ -296,7 +295,7 @@ public:
 class EffectReverb : public EffectGeneric<ReverbParams> {
 public:
     explicit EffectReverb();
-    ~EffectReverb();
+    ~EffectReverb() override;
 
     void Update(EffectInfo::InParams& in_params) override;
     void UpdateForCommandGeneration() override;

src/audio_core/info_updater.cpp

@@ -64,7 +64,6 @@ bool InfoUpdater::UpdateBehaviorInfo(BehaviorInfo& in_behavior_info) {
 }
 
 bool InfoUpdater::UpdateMemoryPools(std::vector<ServerMemoryPoolInfo>& memory_pool_info) {
-    const auto force_mapping = behavior_info.IsMemoryPoolForceMappingEnabled();
     const auto memory_pool_count = memory_pool_info.size();
     const auto total_memory_pool_in = sizeof(ServerMemoryPoolInfo::InParams) * memory_pool_count;
     const auto total_memory_pool_out = sizeof(ServerMemoryPoolInfo::OutParams) * memory_pool_count;
@@ -174,7 +173,7 @@ bool InfoUpdater::UpdateVoices(VoiceContext& voice_context,
         }
         // Voice states for each channel
         std::array<VoiceState*, AudioCommon::MAX_CHANNEL_COUNT> voice_states{};
-        ASSERT(in_params.id < voice_count);
+        ASSERT(static_cast<std::size_t>(in_params.id) < voice_count);
 
         // Grab our current voice info
         auto& voice_info = voice_context.GetInfo(static_cast<std::size_t>(in_params.id));
@@ -352,8 +351,8 @@ ResultCode InfoUpdater::UpdateMixes(MixContext& mix_context, std::size_t mix_buf
         for (std::size_t i = 0; i < mix_count; i++) {
             const auto& in = mix_in_params[i];
             total_buffer_count += in.buffer_count;
-            if (in.dest_mix_id > mix_count && in.dest_mix_id != AudioCommon::NO_MIX &&
-                in.mix_id != AudioCommon::FINAL_MIX) {
+            if (static_cast<std::size_t>(in.dest_mix_id) > mix_count &&
+                in.dest_mix_id != AudioCommon::NO_MIX && in.mix_id != AudioCommon::FINAL_MIX) {
                 LOG_ERROR(
                     Audio,
                     "Invalid mix destination, mix_id={:X}, dest_mix_id={:X}, mix_buffer_count={:X}",

src/audio_core/mix_context.cpp

@@ -53,7 +53,7 @@ void MixContext::UpdateDistancesFromFinalMix() {
         auto mix_id = in_params.mix_id;
         // Needs to be referenced out of scope
         s32 distance_to_final_mix{AudioCommon::FINAL_MIX};
-        for (; distance_to_final_mix < info_count; distance_to_final_mix++) {
+        for (; distance_to_final_mix < static_cast<s32>(info_count); distance_to_final_mix++) {
             if (mix_id == AudioCommon::FINAL_MIX) {
                 // If we're at the final mix, we're done
                 break;
@@ -77,7 +77,7 @@ void MixContext::UpdateDistancesFromFinalMix() {
         }
 
         // If we're out of range for our distance, mark it as no final mix
-        if (distance_to_final_mix >= info_count) {
+        if (distance_to_final_mix >= static_cast<s32>(info_count)) {
             distance_to_final_mix = AudioCommon::NO_FINAL_MIX;
         }
 

src/audio_core/splitter_context.cpp

@@ -306,7 +306,7 @@ bool SplitterContext::UpdateInfo(const std::vector<u8>& input, std::size_t& inpu
             break;
         }
 
-        if (header.send_id < 0 || header.send_id > info_count) {
+        if (header.send_id < 0 || static_cast<std::size_t>(header.send_id) > info_count) {
             LOG_ERROR(Audio, "Bad splitter data id");
             break;
         }
@@ -348,7 +348,7 @@ bool SplitterContext::UpdateData(const std::vector<u8>& input, std::size_t& inpu
             break;
         }
 
-        if (header.splitter_id < 0 || header.splitter_id > data_count) {
+        if (header.splitter_id < 0 || static_cast<std::size_t>(header.splitter_id) > data_count) {
             LOG_ERROR(Audio, "Bad splitter data id");
             break;
         }
@@ -434,7 +434,7 @@ const std::vector<s32>& NodeStates::GetIndexList() const {
 }
 
 void NodeStates::PushTsortResult(s32 index) {
-    ASSERT(index < node_count);
+    ASSERT(index < static_cast<s32>(node_count));
     index_list[index_pos++] = index;
 }
 

src/audio_core/stream.cpp

@@ -12,7 +12,6 @@
 #include "common/assert.h"
 #include "common/logging/log.h"
 #include "core/core_timing.h"
-#include "core/core_timing_util.h"
 #include "core/settings.h"
 
 namespace AudioCore {

src/audio_core/voice_context.cpp

@@ -488,11 +488,11 @@ s32 VoiceContext::DecodePcm16(s32* output_buffer, ServerWaveBuffer* wave_buffer,
 
     // Fast path
     if (channel_count == 1) {
-        for (std::size_t i = 0; i < samples_processed; i++) {
+        for (std::ptrdiff_t i = 0; i < samples_processed; i++) {
             output_buffer[i] = buffer_data[i];
         }
     } else {
-        for (std::size_t i = 0; i < samples_processed; i++) {
+        for (std::ptrdiff_t i = 0; i < samples_processed; i++) {
             output_buffer[i] = buffer_data[i * channel_count + channel];
         }
     }

src/common/wall_clock.cpp

@@ -15,7 +15,7 @@ namespace Common {
 using base_timer = std::chrono::steady_clock;
 using base_time_point = std::chrono::time_point<base_timer>;
 
-class StandardWallClock : public WallClock {
+class StandardWallClock final : public WallClock {
 public:
     StandardWallClock(u64 emulated_cpu_frequency, u64 emulated_clock_frequency)
         : WallClock(emulated_cpu_frequency, emulated_clock_frequency, false) {

src/common/wall_clock.h

@@ -13,6 +13,8 @@ namespace Common {
 
 class WallClock {
 public:
+    virtual ~WallClock() = default;
+
     /// Returns current wall time in nanoseconds
     [[nodiscard]] virtual std::chrono::nanoseconds GetTimeNS() = 0;
 

src/common/x64/native_clock.h

@@ -12,7 +12,7 @@
 namespace Common {
 
 namespace X64 {
-class NativeClock : public WallClock {
+class NativeClock final : public WallClock {
 public:
     NativeClock(u64 emulated_cpu_frequency, u64 emulated_clock_frequency, u64 rtsc_frequency);
 

src/core/arm/dynarmic/arm_dynarmic_cp15.cpp

@@ -34,7 +34,7 @@ std::optional<Callback> DynarmicCP15::CompileInternalOperation(bool two, unsigne
                                                                CoprocReg CRm, unsigned opc2) {
     LOG_CRITICAL(Core_ARM, "CP15: cdp{} p15, {}, {}, {}, {}, {}", two ? "2" : "", opc1, CRd, CRn,
                  CRm, opc2);
-    return {};
+    return std::nullopt;
 }
 
 CallbackOrAccessOneWord DynarmicCP15::CompileSendOneWord(bool two, unsigned opc1, CoprocReg CRn,
@@ -115,7 +115,7 @@ std::optional<Callback> DynarmicCP15::CompileLoadWords(bool two, bool long_trans
         LOG_CRITICAL(Core_ARM, "CP15: mrrc{}{} p15, {}, [...]", two ? "2" : "",
                      long_transfer ? "l" : "", CRd);
     }
-    return {};
+    return std::nullopt;
 }
 
 std::optional<Callback> DynarmicCP15::CompileStoreWords(bool two, bool long_transfer, CoprocReg CRd,
@@ -127,7 +127,7 @@ std::optional<Callback> DynarmicCP15::CompileStoreWords(bool two, bool long_tran
         LOG_CRITICAL(Core_ARM, "CP15: mrrc{}{} p15, {}, [...]", two ? "2" : "",
                      long_transfer ? "l" : "", CRd);
     }
-    return {};
+    return std::nullopt;
 }
 
 } // namespace Core

src/core/core.h

@@ -120,7 +120,7 @@ public:
      * Gets the instance of the System singleton class.
      * @returns Reference to the instance of the System singleton class.
      */
-    static System& GetInstance() {
+    [[deprecated("Use of the global system instance is deprecated")]] static System& GetInstance() {
         return s_instance;
     }
 

src/core/file_sys/content_archive.cpp

@@ -323,7 +323,7 @@ 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});
 
-        std::optional<Core::Crypto::Key128> key = {};
+        std::optional<Core::Crypto::Key128> key;
         if (encrypted) {
             if (has_rights_id) {
                 status = Loader::ResultStatus::Success;
@@ -442,18 +442,18 @@ std::optional<Core::Crypto::Key128> NCA::GetTitlekey() {
     memcpy(rights_id.data(), header.rights_id.data(), 16);
     if (rights_id == u128{}) {
         status = Loader::ResultStatus::ErrorInvalidRightsID;
-        return {};
+        return std::nullopt;
     }
 
     auto titlekey = keys.GetKey(Core::Crypto::S128KeyType::Titlekey, rights_id[1], rights_id[0]);
     if (titlekey == Core::Crypto::Key128{}) {
         status = Loader::ResultStatus::ErrorMissingTitlekey;
-        return {};
+        return std::nullopt;
     }
 
     if (!keys.HasKey(Core::Crypto::S128KeyType::Titlekek, master_key_id)) {
         status = Loader::ResultStatus::ErrorMissingTitlekek;
-        return {};
+        return std::nullopt;
     }
 
     Core::Crypto::AESCipher<Core::Crypto::Key128> cipher(
@@ -477,7 +477,7 @@ VirtualFile NCA::Decrypt(const NCASectionHeader& s_header, VirtualFile in, u64 s
     case NCASectionCryptoType::BKTR:
         LOG_TRACE(Crypto, "called with mode=CTR, starting_offset={:016X}", starting_offset);
         {
-            std::optional<Core::Crypto::Key128> key = {};
+            std::optional<Core::Crypto::Key128> key;
             if (has_rights_id) {
                 status = Loader::ResultStatus::Success;
                 key = GetTitlekey();

src/core/file_sys/ips_layer.cpp

@@ -245,9 +245,11 @@ void IPSwitchCompiler::Parse() {
 
             // Read rest of patch
             while (true) {
-                if (i + 1 >= lines.size())
+                if (i + 1 >= lines.size()) {
                     break;
-                const auto patch_line = lines[++i];
+                }
+
+                const auto& patch_line = lines[++i];
 
                 // Start of new patch
                 if (StartsWith(patch_line, "@enabled") || StartsWith(patch_line, "@disabled")) {

src/core/file_sys/submission_package.cpp

@@ -267,9 +267,9 @@ void NSP::ReadNCAs(const std::vector<VirtualFile>& files) {
             }
 
             const CNMT cnmt(inner_file);
-            auto& ncas_title = ncas[cnmt.GetTitleID()];
 
-            ncas_title[{cnmt.GetType(), ContentRecordType::Meta}] = nca;
+            ncas[cnmt.GetTitleID()][{cnmt.GetType(), ContentRecordType::Meta}] = nca;
+
             for (const auto& rec : cnmt.GetContentRecords()) {
                 const auto id_string = Common::HexToString(rec.nca_id, false);
                 auto next_file = pfs->GetFile(fmt::format("{}.nca", id_string));
@@ -287,12 +287,12 @@ void NSP::ReadNCAs(const std::vector<VirtualFile>& files) {
 
                 auto next_nca = std::make_shared<NCA>(std::move(next_file), nullptr, 0);
                 if (next_nca->GetType() == NCAContentType::Program) {
-                    program_status[cnmt.GetTitleID()] = next_nca->GetStatus();
+                    program_status[next_nca->GetTitleId()] = next_nca->GetStatus();
                 }
                 if (next_nca->GetStatus() == Loader::ResultStatus::Success ||
                     (next_nca->GetStatus() == Loader::ResultStatus::ErrorMissingBKTRBaseRomFS &&
-                     (cnmt.GetTitleID() & 0x800) != 0)) {
-                    ncas_title[{cnmt.GetType(), rec.type}] = std::move(next_nca);
+                     (next_nca->GetTitleId() & 0x800) != 0)) {
+                    ncas[next_nca->GetTitleId()][{cnmt.GetType(), rec.type}] = std::move(next_nca);
                 }
             }
 

src/core/file_sys/vfs.cpp

@@ -169,11 +169,12 @@ VfsDirectory::~VfsDirectory() = default;
 
 std::optional<u8> VfsFile::ReadByte(std::size_t offset) const {
     u8 out{};
-    std::size_t size = Read(&out, 1, offset);
-    if (size == 1)
+    const std::size_t size = Read(&out, sizeof(u8), offset);
+    if (size == 1) {
         return out;
+    }
 
-    return {};
+    return std::nullopt;
 }
 
 std::vector<u8> VfsFile::ReadBytes(std::size_t size, std::size_t offset) const {

src/core/file_sys/vfs_offset.cpp

@@ -58,10 +58,11 @@ std::size_t OffsetVfsFile::Write(const u8* data, std::size_t length, std::size_t
 }
 
 std::optional<u8> OffsetVfsFile::ReadByte(std::size_t r_offset) const {
-    if (r_offset < size)
-        return file->ReadByte(offset + r_offset);
+    if (r_offset >= size) {
+        return std::nullopt;
+    }
 
-    return {};
+    return file->ReadByte(offset + r_offset);
 }
 
 std::vector<u8> OffsetVfsFile::ReadBytes(std::size_t r_size, std::size_t r_offset) const {

src/core/file_sys/vfs_static.h

@@ -54,9 +54,11 @@ public:
     }
 
     std::optional<u8> ReadByte(std::size_t offset) const override {
-        if (offset < size)
-            return value;
-        return {};
+        if (offset >= size) {
+            return std::nullopt;
+        }
+
+        return value;
     }
 
     std::vector<u8> ReadBytes(std::size_t length, std::size_t offset) const override {

src/core/frontend/applets/controller.cpp

@@ -4,7 +4,6 @@
 
 #include "common/assert.h"
 #include "common/logging/log.h"
-#include "core/core.h"
 #include "core/frontend/applets/controller.h"
 #include "core/hle/service/hid/controllers/npad.h"
 #include "core/hle/service/hid/hid.h"
@@ -14,6 +13,9 @@ namespace Core::Frontend {
 
 ControllerApplet::~ControllerApplet() = default;
 
+DefaultControllerApplet::DefaultControllerApplet(Service::SM::ServiceManager& service_manager_)
+    : service_manager{service_manager_} {}
+
 DefaultControllerApplet::~DefaultControllerApplet() = default;
 
 void DefaultControllerApplet::ReconfigureControllers(std::function<void()> callback,
@@ -21,9 +23,7 @@ void DefaultControllerApplet::ReconfigureControllers(std::function<void()> callb
     LOG_INFO(Service_HID, "called, deducing the best configuration based on the given parameters!");
 
     auto& npad =
-        Core::System::GetInstance()
-            .ServiceManager()
-            .GetService<Service::HID::Hid>("hid")
+        service_manager.GetService<Service::HID::Hid>("hid")
             ->GetAppletResource()
             ->GetController<Service::HID::Controller_NPad>(Service::HID::HidController::NPad);
 

src/core/frontend/applets/controller.h

@@ -8,6 +8,10 @@
 
 #include "common/common_types.h"
 
+namespace Service::SM {
+class ServiceManager;
+}
+
 namespace Core::Frontend {
 
 using BorderColor = std::array<u8, 4>;
@@ -39,10 +43,14 @@ public:
 
 class DefaultControllerApplet final : public ControllerApplet {
 public:
+    explicit DefaultControllerApplet(Service::SM::ServiceManager& service_manager_);
     ~DefaultControllerApplet() override;
 
     void ReconfigureControllers(std::function<void()> callback,
                                 ControllerParameters parameters) const override;
+
+private:
+    Service::SM::ServiceManager& service_manager;
 };
 
 } // namespace Core::Frontend

src/core/frontend/input.h

@@ -33,6 +33,9 @@ public:
     virtual bool GetAnalogDirectionStatus(AnalogDirection direction) const {
         return {};
     }
+    virtual bool SetRumblePlay(f32 amp_high, f32 amp_low, f32 freq_high, f32 freq_low) const {
+        return {};
+    }
 };
 
 /// An abstract class template for a factory that can create input devices.

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

@@ -774,6 +774,17 @@ void Module::Interface::ListQualifiedUsers(Kernel::HLERequestContext& ctx) {
     rb.Push(RESULT_SUCCESS);
 }
 
+void Module::Interface::LoadOpenContext(Kernel::HLERequestContext& ctx) {
+    LOG_WARNING(Service_ACC, "(STUBBED) called");
+
+    // This is similar to GetBaasAccountManagerForApplication
+    // This command is used concurrently with ListOpenContextStoredUsers
+    // TODO: Find the differences between this and GetBaasAccountManagerForApplication
+    IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+    rb.Push(RESULT_SUCCESS);
+    rb.PushIpcInterface<IManagerForApplication>(profile_manager->GetLastOpenedUser());
+}
+
 void Module::Interface::ListOpenContextStoredUsers(Kernel::HLERequestContext& ctx) {
     LOG_WARNING(Service_ACC, "(STUBBED) called");
 

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

@@ -34,6 +34,7 @@ public:
         void IsUserAccountSwitchLocked(Kernel::HLERequestContext& ctx);
         void GetProfileEditor(Kernel::HLERequestContext& ctx);
         void ListQualifiedUsers(Kernel::HLERequestContext& ctx);
+        void LoadOpenContext(Kernel::HLERequestContext& ctx);
         void ListOpenContextStoredUsers(Kernel::HLERequestContext& ctx);
 
     private:

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

@@ -29,7 +29,7 @@ ACC_U0::ACC_U0(std::shared_ptr<Module> module, std::shared_ptr<ProfileManager> p
         {110, nullptr, "StoreSaveDataThumbnail"},
         {111, nullptr, "ClearSaveDataThumbnail"},
         {120, nullptr, "CreateGuestLoginRequest"},
-        {130, nullptr, "LoadOpenContext"}, // 5.0.0+
+        {130, &ACC_U0::LoadOpenContext, "LoadOpenContext"}, // 5.0.0+
         {131, &ACC_U0::ListOpenContextStoredUsers, "ListOpenContextStoredUsers"}, // 6.0.0+
         {140, &ACC_U0::InitializeApplicationInfoRestricted, "InitializeApplicationInfoRestricted"}, // 6.0.0+
         {141, &ACC_U0::ListQualifiedUsers, "ListQualifiedUsers"}, // 6.0.0+

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

@@ -206,7 +206,8 @@ void AppletManager::SetDefaultAppletFrontendSet() {
 
 void AppletManager::SetDefaultAppletsIfMissing() {
     if (frontend.controller == nullptr) {
-        frontend.controller = std::make_unique<Core::Frontend::DefaultControllerApplet>();
+        frontend.controller =
+            std::make_unique<Core::Frontend::DefaultControllerApplet>(system.ServiceManager());
     }
 
     if (frontend.e_commerce == nullptr) {

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

@@ -31,6 +31,10 @@ public:
     virtual void OnUpdate(const Core::Timing::CoreTiming& core_timing, u8* data,
                           std::size_t size) = 0;
 
+    // When the controller is requesting a motion update for the shared memory
+    virtual void OnMotionUpdate(const Core::Timing::CoreTiming& core_timing, u8* data,
+                                std::size_t size) {}
+
     // Called when input devices should be loaded
     virtual void OnLoadInputDevices() = 0;
 

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

@@ -365,44 +365,6 @@ void Controller_NPad::OnUpdate(const Core::Timing::CoreTiming& core_timing, u8*
         }
         const u32 npad_index = static_cast<u32>(i);
 
-        const std::array<SixAxisGeneric*, 6> controller_sixaxes{
-            &npad.sixaxis_full,       &npad.sixaxis_handheld, &npad.sixaxis_dual_left,
-            &npad.sixaxis_dual_right, &npad.sixaxis_left,     &npad.sixaxis_right,
-        };
-
-        for (auto* sixaxis_sensor : controller_sixaxes) {
-            sixaxis_sensor->common.entry_count = 16;
-            sixaxis_sensor->common.total_entry_count = 17;
-
-            const auto& last_entry =
-                sixaxis_sensor->sixaxis[sixaxis_sensor->common.last_entry_index];
-
-            sixaxis_sensor->common.timestamp = core_timing.GetCPUTicks();
-            sixaxis_sensor->common.last_entry_index =
-                (sixaxis_sensor->common.last_entry_index + 1) % 17;
-
-            auto& cur_entry = sixaxis_sensor->sixaxis[sixaxis_sensor->common.last_entry_index];
-
-            cur_entry.timestamp = last_entry.timestamp + 1;
-            cur_entry.timestamp2 = cur_entry.timestamp;
-        }
-
-        // Try to read sixaxis sensor states
-        std::array<MotionDevice, 2> motion_devices;
-
-        if (sixaxis_sensors_enabled && Settings::values.motion_enabled) {
-            sixaxis_at_rest = true;
-            for (std::size_t e = 0; e < motion_devices.size(); ++e) {
-                const auto& device = motions[i][e];
-                if (device) {
-                    std::tie(motion_devices[e].accel, motion_devices[e].gyro,
-                             motion_devices[e].rotation, motion_devices[e].orientation) =
-                        device->GetStatus();
-                    sixaxis_at_rest = sixaxis_at_rest && motion_devices[e].gyro.Length2() < 0.0001f;
-                }
-            }
-        }
-
         RequestPadStateUpdate(npad_index);
         auto& pad_state = npad_pad_states[npad_index];
 
@@ -446,13 +408,6 @@ void Controller_NPad::OnUpdate(const Core::Timing::CoreTiming& core_timing, u8*
             main_controller.pad.r_stick = pad_state.r_stick;
 
             libnx_entry.connection_status.IsWired.Assign(1);
-
-            if (sixaxis_sensors_enabled && motions[i][0]) {
-                full_sixaxis_entry.accel = motion_devices[0].accel;
-                full_sixaxis_entry.gyro = motion_devices[0].gyro;
-                full_sixaxis_entry.rotation = motion_devices[0].rotation;
-                full_sixaxis_entry.orientation = motion_devices[0].orientation;
-            }
             break;
         case NPadControllerType::Handheld:
             handheld_entry.connection_status.raw = 0;
@@ -471,13 +426,6 @@ void Controller_NPad::OnUpdate(const Core::Timing::CoreTiming& core_timing, u8*
             libnx_entry.connection_status.IsRightJoyConnected.Assign(1);
             libnx_entry.connection_status.IsLeftJoyWired.Assign(1);
             libnx_entry.connection_status.IsRightJoyWired.Assign(1);
-
-            if (sixaxis_sensors_enabled && motions[i][0]) {
-                handheld_sixaxis_entry.accel = motion_devices[0].accel;
-                handheld_sixaxis_entry.gyro = motion_devices[0].gyro;
-                handheld_sixaxis_entry.rotation = motion_devices[0].rotation;
-                handheld_sixaxis_entry.orientation = motion_devices[0].orientation;
-            }
             break;
         case NPadControllerType::JoyDual:
             dual_entry.connection_status.raw = 0;
@@ -490,21 +438,6 @@ void Controller_NPad::OnUpdate(const Core::Timing::CoreTiming& core_timing, u8*
 
             libnx_entry.connection_status.IsLeftJoyConnected.Assign(1);
             libnx_entry.connection_status.IsRightJoyConnected.Assign(1);
-
-            if (sixaxis_sensors_enabled && motions[i][0]) {
-                // Set motion for the left joycon
-                dual_left_sixaxis_entry.accel = motion_devices[0].accel;
-                dual_left_sixaxis_entry.gyro = motion_devices[0].gyro;
-                dual_left_sixaxis_entry.rotation = motion_devices[0].rotation;
-                dual_left_sixaxis_entry.orientation = motion_devices[0].orientation;
-            }
-            if (sixaxis_sensors_enabled && motions[i][1]) {
-                // Set motion for the right joycon
-                dual_right_sixaxis_entry.accel = motion_devices[1].accel;
-                dual_right_sixaxis_entry.gyro = motion_devices[1].gyro;
-                dual_right_sixaxis_entry.rotation = motion_devices[1].rotation;
-                dual_right_sixaxis_entry.orientation = motion_devices[1].orientation;
-            }
             break;
         case NPadControllerType::JoyLeft:
             left_entry.connection_status.raw = 0;
@@ -515,13 +448,6 @@ void Controller_NPad::OnUpdate(const Core::Timing::CoreTiming& core_timing, u8*
             left_entry.pad.r_stick = pad_state.r_stick;
 
             libnx_entry.connection_status.IsLeftJoyConnected.Assign(1);
-
-            if (sixaxis_sensors_enabled && motions[i][0]) {
-                left_sixaxis_entry.accel = motion_devices[0].accel;
-                left_sixaxis_entry.gyro = motion_devices[0].gyro;
-                left_sixaxis_entry.rotation = motion_devices[0].rotation;
-                left_sixaxis_entry.orientation = motion_devices[0].orientation;
-            }
             break;
         case NPadControllerType::JoyRight:
             right_entry.connection_status.raw = 0;
@@ -532,13 +458,6 @@ void Controller_NPad::OnUpdate(const Core::Timing::CoreTiming& core_timing, u8*
             right_entry.pad.r_stick = pad_state.r_stick;
 
             libnx_entry.connection_status.IsRightJoyConnected.Assign(1);
-
-            if (sixaxis_sensors_enabled && motions[i][1]) {
-                right_sixaxis_entry.accel = motion_devices[1].accel;
-                right_sixaxis_entry.gyro = motion_devices[1].gyro;
-                right_sixaxis_entry.rotation = motion_devices[1].rotation;
-                right_sixaxis_entry.orientation = motion_devices[1].orientation;
-            }
             break;
         case NPadControllerType::Pokeball:
             pokeball_entry.connection_status.raw = 0;
@@ -561,6 +480,143 @@ void Controller_NPad::OnUpdate(const Core::Timing::CoreTiming& core_timing, u8*
                 shared_memory_entries.size() * sizeof(NPadEntry));
 }
 
+void Controller_NPad::OnMotionUpdate(const Core::Timing::CoreTiming& core_timing, u8* data,
+                                     std::size_t data_len) {
+    if (!IsControllerActivated()) {
+        return;
+    }
+    for (std::size_t i = 0; i < shared_memory_entries.size(); i++) {
+        auto& npad = shared_memory_entries[i];
+
+        const auto& controller_type = connected_controllers[i].type;
+
+        if (controller_type == NPadControllerType::None || !connected_controllers[i].is_connected) {
+            continue;
+        }
+
+        const std::array<SixAxisGeneric*, 6> controller_sixaxes{
+            &npad.sixaxis_full,       &npad.sixaxis_handheld, &npad.sixaxis_dual_left,
+            &npad.sixaxis_dual_right, &npad.sixaxis_left,     &npad.sixaxis_right,
+        };
+
+        for (auto* sixaxis_sensor : controller_sixaxes) {
+            sixaxis_sensor->common.entry_count = 16;
+            sixaxis_sensor->common.total_entry_count = 17;
+
+            const auto& last_entry =
+                sixaxis_sensor->sixaxis[sixaxis_sensor->common.last_entry_index];
+
+            sixaxis_sensor->common.timestamp = core_timing.GetCPUTicks();
+            sixaxis_sensor->common.last_entry_index =
+                (sixaxis_sensor->common.last_entry_index + 1) % 17;
+
+            auto& cur_entry = sixaxis_sensor->sixaxis[sixaxis_sensor->common.last_entry_index];
+
+            cur_entry.timestamp = last_entry.timestamp + 1;
+            cur_entry.timestamp2 = cur_entry.timestamp;
+        }
+
+        // Try to read sixaxis sensor states
+        std::array<MotionDevice, 2> motion_devices;
+
+        if (sixaxis_sensors_enabled && Settings::values.motion_enabled) {
+            sixaxis_at_rest = true;
+            for (std::size_t e = 0; e < motion_devices.size(); ++e) {
+                const auto& device = motions[i][e];
+                if (device) {
+                    std::tie(motion_devices[e].accel, motion_devices[e].gyro,
+                             motion_devices[e].rotation, motion_devices[e].orientation) =
+                        device->GetStatus();
+                    sixaxis_at_rest = sixaxis_at_rest && motion_devices[e].gyro.Length2() < 0.0001f;
+                }
+            }
+        }
+
+        auto& main_controller =
+            npad.main_controller_states.npad[npad.main_controller_states.common.last_entry_index];
+        auto& handheld_entry =
+            npad.handheld_states.npad[npad.handheld_states.common.last_entry_index];
+        auto& dual_entry = npad.dual_states.npad[npad.dual_states.common.last_entry_index];
+        auto& left_entry = npad.left_joy_states.npad[npad.left_joy_states.common.last_entry_index];
+        auto& right_entry =
+            npad.right_joy_states.npad[npad.right_joy_states.common.last_entry_index];
+        auto& pokeball_entry =
+            npad.pokeball_states.npad[npad.pokeball_states.common.last_entry_index];
+        auto& libnx_entry = npad.libnx.npad[npad.libnx.common.last_entry_index];
+
+        auto& full_sixaxis_entry =
+            npad.sixaxis_full.sixaxis[npad.sixaxis_full.common.last_entry_index];
+        auto& handheld_sixaxis_entry =
+            npad.sixaxis_handheld.sixaxis[npad.sixaxis_handheld.common.last_entry_index];
+        auto& dual_left_sixaxis_entry =
+            npad.sixaxis_dual_left.sixaxis[npad.sixaxis_dual_left.common.last_entry_index];
+        auto& dual_right_sixaxis_entry =
+            npad.sixaxis_dual_right.sixaxis[npad.sixaxis_dual_right.common.last_entry_index];
+        auto& left_sixaxis_entry =
+            npad.sixaxis_left.sixaxis[npad.sixaxis_left.common.last_entry_index];
+        auto& right_sixaxis_entry =
+            npad.sixaxis_right.sixaxis[npad.sixaxis_right.common.last_entry_index];
+
+        switch (controller_type) {
+        case NPadControllerType::None:
+            UNREACHABLE();
+            break;
+        case NPadControllerType::ProController:
+            if (sixaxis_sensors_enabled && motions[i][0]) {
+                full_sixaxis_entry.accel = motion_devices[0].accel;
+                full_sixaxis_entry.gyro = motion_devices[0].gyro;
+                full_sixaxis_entry.rotation = motion_devices[0].rotation;
+                full_sixaxis_entry.orientation = motion_devices[0].orientation;
+            }
+            break;
+        case NPadControllerType::Handheld:
+            if (sixaxis_sensors_enabled && motions[i][0]) {
+                handheld_sixaxis_entry.accel = motion_devices[0].accel;
+                handheld_sixaxis_entry.gyro = motion_devices[0].gyro;
+                handheld_sixaxis_entry.rotation = motion_devices[0].rotation;
+                handheld_sixaxis_entry.orientation = motion_devices[0].orientation;
+            }
+            break;
+        case NPadControllerType::JoyDual:
+            if (sixaxis_sensors_enabled && motions[i][0]) {
+                // Set motion for the left joycon
+                dual_left_sixaxis_entry.accel = motion_devices[0].accel;
+                dual_left_sixaxis_entry.gyro = motion_devices[0].gyro;
+                dual_left_sixaxis_entry.rotation = motion_devices[0].rotation;
+                dual_left_sixaxis_entry.orientation = motion_devices[0].orientation;
+            }
+            if (sixaxis_sensors_enabled && motions[i][1]) {
+                // Set motion for the right joycon
+                dual_right_sixaxis_entry.accel = motion_devices[1].accel;
+                dual_right_sixaxis_entry.gyro = motion_devices[1].gyro;
+                dual_right_sixaxis_entry.rotation = motion_devices[1].rotation;
+                dual_right_sixaxis_entry.orientation = motion_devices[1].orientation;
+            }
+            break;
+        case NPadControllerType::JoyLeft:
+            if (sixaxis_sensors_enabled && motions[i][0]) {
+                left_sixaxis_entry.accel = motion_devices[0].accel;
+                left_sixaxis_entry.gyro = motion_devices[0].gyro;
+                left_sixaxis_entry.rotation = motion_devices[0].rotation;
+                left_sixaxis_entry.orientation = motion_devices[0].orientation;
+            }
+            break;
+        case NPadControllerType::JoyRight:
+            if (sixaxis_sensors_enabled && motions[i][1]) {
+                right_sixaxis_entry.accel = motion_devices[1].accel;
+                right_sixaxis_entry.gyro = motion_devices[1].gyro;
+                right_sixaxis_entry.rotation = motion_devices[1].rotation;
+                right_sixaxis_entry.orientation = motion_devices[1].orientation;
+            }
+            break;
+        case NPadControllerType::Pokeball:
+            break;
+        }
+    }
+    std::memcpy(data + NPAD_OFFSET, shared_memory_entries.data(),
+                shared_memory_entries.size() * sizeof(NPadEntry));
+}
+
 void Controller_NPad::SetSupportedStyleSet(NPadType style_set) {
     style.raw = style_set.raw;
 }
@@ -593,6 +649,14 @@ Controller_NPad::NpadHoldType Controller_NPad::GetHoldType() const {
     return hold_type;
 }
 
+void Controller_NPad::SetNpadHandheldActivationMode(NpadHandheldActivationMode activation_mode) {
+    handheld_activation_mode = activation_mode;
+}
+
+Controller_NPad::NpadHandheldActivationMode Controller_NPad::GetNpadHandheldActivationMode() const {
+    return handheld_activation_mode;
+}
+
 void Controller_NPad::SetNpadMode(u32 npad_id, NPadAssignments assignment_mode) {
     const std::size_t npad_index = NPadIdToIndex(npad_id);
     ASSERT(npad_index < shared_memory_entries.size());
@@ -601,20 +665,31 @@ void Controller_NPad::SetNpadMode(u32 npad_id, NPadAssignments assignment_mode)
     }
 }
 
-void Controller_NPad::VibrateController(const std::vector<u32>& controller_ids,
+void Controller_NPad::VibrateController(const std::vector<u32>& controllers,
                                         const std::vector<Vibration>& vibrations) {
-    LOG_DEBUG(Service_HID, "(STUBBED) called");
+    LOG_TRACE(Service_HID, "called");
 
     if (!Settings::values.vibration_enabled || !can_controllers_vibrate) {
         return;
     }
-    for (std::size_t i = 0; i < controller_ids.size(); i++) {
-        std::size_t controller_pos = NPadIdToIndex(static_cast<u32>(i));
-        if (connected_controllers[controller_pos].is_connected) {
-            // TODO(ogniK): Vibrate the physical controller
+    bool success = true;
+    for (std::size_t i = 0; i < controllers.size(); ++i) {
+        if (!connected_controllers[i].is_connected) {
+            continue;
+        }
+        using namespace Settings::NativeButton;
+        const auto& button_state = buttons[i];
+        if (button_state[A - BUTTON_HID_BEGIN]) {
+            if (button_state[A - BUTTON_HID_BEGIN]->SetRumblePlay(
+                    vibrations[0].amp_high, vibrations[0].amp_low, vibrations[0].freq_high,
+                    vibrations[0].freq_low)) {
+                success = false;
+            }
         }
     }
-    last_processed_vibration = vibrations.back();
+    if (success) {
+        last_processed_vibration = vibrations.back();
+    }
 }
 
 Controller_NPad::Vibration Controller_NPad::GetLastVibration() const {

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

@@ -32,6 +32,10 @@ public:
     // When the controller is requesting an update for the shared memory
     void OnUpdate(const Core::Timing::CoreTiming& core_timing, u8* data, std::size_t size) override;
 
+    // When the controller is requesting a motion update for the shared memory
+    void OnMotionUpdate(const Core::Timing::CoreTiming& core_timing, u8* data,
+                        std::size_t size) override;
+
     // Called when input devices should be loaded
     void OnLoadInputDevices() override;
 
@@ -74,6 +78,12 @@ public:
         Single = 1,
     };
 
+    enum class NpadHandheldActivationMode : u64 {
+        Dual = 0,
+        Single = 1,
+        None = 2,
+    };
+
     enum class NPadControllerType {
         None,
         ProController,
@@ -110,9 +120,12 @@ public:
     void SetHoldType(NpadHoldType joy_hold_type);
     NpadHoldType GetHoldType() const;
 
+    void SetNpadHandheldActivationMode(NpadHandheldActivationMode activation_mode);
+    NpadHandheldActivationMode GetNpadHandheldActivationMode() const;
+
     void SetNpadMode(u32 npad_id, NPadAssignments assignment_mode);
 
-    void VibrateController(const std::vector<u32>& controller_ids,
+    void VibrateController(const std::vector<u32>& controllers,
                            const std::vector<Vibration>& vibrations);
 
     Vibration GetLastVibration() const;
@@ -369,6 +382,7 @@ private:
     MotionArray motions;
     std::vector<u32> supported_npad_id_types{};
     NpadHoldType hold_type{NpadHoldType::Vertical};
+    NpadHandheldActivationMode handheld_activation_mode{NpadHandheldActivationMode::Dual};
     // Each controller should have their own styleset changed event
     std::array<Kernel::EventPair, 10> styleset_changed_events;
     Vibration last_processed_vibration{};

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

@@ -40,7 +40,8 @@ namespace Service::HID {
 // Updating period for each HID device.
 // HID is polled every 15ms, this value was derived from
 // https://github.com/dekuNukem/Nintendo_Switch_Reverse_Engineering#joy-con-status-data-packet
-constexpr auto pad_update_ns = std::chrono::nanoseconds{15 * 1000 * 1000}; // (15ms, 66.6Hz)
+constexpr auto pad_update_ns = std::chrono::nanoseconds{1000 * 1000};         // (1ms, 1000Hz)
+constexpr auto motion_update_ns = std::chrono::nanoseconds{15 * 1000 * 1000}; // (15ms, 66.666Hz)
 constexpr std::size_t SHARED_MEMORY_SIZE = 0x40000;
 
 IAppletResource::IAppletResource(Core::System& system)
@@ -79,10 +80,14 @@ IAppletResource::IAppletResource(Core::System& system)
         [this](std::uintptr_t user_data, std::chrono::nanoseconds ns_late) {
             UpdateControllers(user_data, ns_late);
         });
-
-    // TODO(shinyquagsire23): Other update callbacks? (accel, gyro?)
+    motion_update_event = Core::Timing::CreateEvent(
+        "HID::MotionPadCallback",
+        [this](std::uintptr_t user_data, std::chrono::nanoseconds ns_late) {
+            UpdateMotion(user_data, ns_late);
+        });
 
     system.CoreTiming().ScheduleEvent(pad_update_ns, pad_update_event);
+    system.CoreTiming().ScheduleEvent(motion_update_ns, motion_update_event);
 
     ReloadInputDevices();
 }
@@ -122,6 +127,16 @@ void IAppletResource::UpdateControllers(std::uintptr_t user_data,
     core_timing.ScheduleEvent(pad_update_ns - ns_late, pad_update_event);
 }
 
+void IAppletResource::UpdateMotion(std::uintptr_t user_data, std::chrono::nanoseconds ns_late) {
+    auto& core_timing = system.CoreTiming();
+
+    for (const auto& controller : controllers) {
+        controller->OnMotionUpdate(core_timing, shared_mem->GetPointer(), SHARED_MEMORY_SIZE);
+    }
+
+    core_timing.ScheduleEvent(motion_update_ns - ns_late, motion_update_event);
+}
+
 class IActiveVibrationDeviceList final : public ServiceFramework<IActiveVibrationDeviceList> {
 public:
     IActiveVibrationDeviceList() : ServiceFramework("IActiveVibrationDeviceList") {
@@ -739,8 +754,11 @@ void Hid::SetNpadHandheldActivationMode(Kernel::HLERequestContext& ctx) {
     const auto applet_resource_user_id{rp.Pop<u64>()};
     const auto mode{rp.Pop<u64>()};
 
-    LOG_WARNING(Service_HID, "(STUBBED) called, applet_resource_user_id={}, mode={}",
-                applet_resource_user_id, mode);
+    LOG_DEBUG(Service_HID, "called, applet_resource_user_id={}, mode={}", applet_resource_user_id,
+              mode);
+
+    applet_resource->GetController<Controller_NPad>(HidController::NPad)
+        .SetNpadHandheldActivationMode(Controller_NPad::NpadHandheldActivationMode{mode});
 
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
@@ -750,11 +768,13 @@ void Hid::GetNpadHandheldActivationMode(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
     const auto applet_resource_user_id{rp.Pop<u64>()};
 
-    LOG_WARNING(Service_HID, "(STUBBED) called, applet_resource_user_id={}",
-                applet_resource_user_id);
+    LOG_DEBUG(Service_HID, "called, applet_resource_user_id={}", applet_resource_user_id);
 
-    IPC::ResponseBuilder rb{ctx, 2};
+    IPC::ResponseBuilder rb{ctx, 4};
     rb.Push(RESULT_SUCCESS);
+    rb.Push<u64>(
+        static_cast<u64>(applet_resource->GetController<Controller_NPad>(HidController::NPad)
+                             .GetNpadHandheldActivationMode()));
 }
 
 void Hid::SwapNpadAssignment(Kernel::HLERequestContext& ctx) {
@@ -797,18 +817,18 @@ void Hid::EndPermitVibrationSession(Kernel::HLERequestContext& ctx) {
 
 void Hid::SendVibrationValue(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
-    const auto controller_id{rp.Pop<u32>()};
+    const auto controller{rp.Pop<u32>()};
     const auto vibration_values{rp.PopRaw<Controller_NPad::Vibration>()};
     const auto applet_resource_user_id{rp.Pop<u64>()};
 
-    LOG_DEBUG(Service_HID, "called, controller_id={}, applet_resource_user_id={}", controller_id,
+    LOG_DEBUG(Service_HID, "called, controller={}, applet_resource_user_id={}", controller,
               applet_resource_user_id);
 
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
 
     applet_resource->GetController<Controller_NPad>(HidController::NPad)
-        .VibrateController({controller_id}, {vibration_values});
+        .VibrateController({controller}, {vibration_values});
 }
 
 void Hid::SendVibrationValues(Kernel::HLERequestContext& ctx) {
@@ -826,8 +846,6 @@ void Hid::SendVibrationValues(Kernel::HLERequestContext& ctx) {
 
     std::memcpy(controller_list.data(), controllers.data(), controllers.size());
     std::memcpy(vibration_list.data(), vibrations.data(), vibrations.size());
-    std::transform(controller_list.begin(), controller_list.end(), controller_list.begin(),
-                   [](u32 controller_id) { return controller_id - 3; });
 
     applet_resource->GetController<Controller_NPad>(HidController::NPad)
         .VibrateController(controller_list, vibration_list);

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

@@ -65,10 +65,12 @@ private:
 
     void GetSharedMemoryHandle(Kernel::HLERequestContext& ctx);
     void UpdateControllers(std::uintptr_t user_data, std::chrono::nanoseconds ns_late);
+    void UpdateMotion(std::uintptr_t user_data, std::chrono::nanoseconds ns_late);
 
     std::shared_ptr<Kernel::SharedMemory> shared_mem;
 
     std::shared_ptr<Core::Timing::EventType> pad_update_event;
+    std::shared_ptr<Core::Timing::EventType> motion_update_event;
     Core::System& system;
 
     std::array<std::unique_ptr<ControllerBase>, static_cast<size_t>(HidController::MaxControllers)>

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

@@ -265,7 +265,7 @@ std::optional<nvhost_as_gpu::BufferMap> nvhost_as_gpu::FindBufferMap(GPUVAddr gp
         }
     }
 
-    return {};
+    return std::nullopt;
 }
 
 void nvhost_as_gpu::AddBufferMap(GPUVAddr gpu_addr, std::size_t size, VAddr cpu_addr,
@@ -286,7 +286,7 @@ std::optional<std::size_t> nvhost_as_gpu::RemoveBufferMap(GPUVAddr gpu_addr) {
         return size;
     }
 
-    return {};
+    return std::nullopt;
 }
 
 } // namespace Service::Nvidia::Devices

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

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

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

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

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

@@ -22,6 +22,18 @@ u32 nvhost_vic::ioctl(Ioctl command, const std::vector<u8>& input, const std::ve
     switch (static_cast<IoctlCommand>(command.raw)) {
     case IoctlCommand::IocSetNVMAPfdCommand:
         return SetNVMAPfd(input, output);
+    case IoctlCommand::IocSubmit:
+        return Submit(input, output);
+    case IoctlCommand::IocGetSyncpoint:
+        return GetSyncpoint(input, output);
+    case IoctlCommand::IocGetWaitbase:
+        return GetWaitbase(input, output);
+    case IoctlCommand::IocMapBuffer:
+        return MapBuffer(input, output);
+    case IoctlCommand::IocMapBufferEx:
+        return MapBuffer(input, output);
+    case IoctlCommand::IocUnmapBufferEx:
+        return UnmapBufferEx(input, output);
     }
 
     UNIMPLEMENTED_MSG("Unimplemented ioctl");
@@ -30,11 +42,71 @@ u32 nvhost_vic::ioctl(Ioctl command, const std::vector<u8>& input, const std::ve
 
 u32 nvhost_vic::SetNVMAPfd(const std::vector<u8>& input, std::vector<u8>& output) {
     IoctlSetNvmapFD params{};
-    std::memcpy(&params, input.data(), input.size());
+    std::memcpy(&params, input.data(), sizeof(IoctlSetNvmapFD));
     LOG_DEBUG(Service_NVDRV, "called, fd={}", params.nvmap_fd);
 
     nvmap_fd = params.nvmap_fd;
     return 0;
 }
 
+u32 nvhost_vic::Submit(const std::vector<u8>& input, std::vector<u8>& output) {
+    IoctlSubmit params{};
+    std::memcpy(&params, input.data(), sizeof(IoctlSubmit));
+    LOG_WARNING(Service_NVDRV, "(STUBBED) called");
+
+    // Workaround for Luigi's Mansion 3, as nvhost_vic is not implemented for asynch GPU
+    params.command_buffer = {};
+
+    std::memcpy(output.data(), &params, sizeof(IoctlSubmit));
+    return 0;
+}
+
+u32 nvhost_vic::GetSyncpoint(const std::vector<u8>& input, std::vector<u8>& output) {
+    IoctlGetSyncpoint params{};
+    std::memcpy(&params, input.data(), sizeof(IoctlGetSyncpoint));
+    LOG_INFO(Service_NVDRV, "called, unknown=0x{:X}", params.unknown);
+    params.value = 0; // Seems to be hard coded at 0
+    std::memcpy(output.data(), &params, sizeof(IoctlGetSyncpoint));
+    return 0;
+}
+
+u32 nvhost_vic::GetWaitbase(const std::vector<u8>& input, std::vector<u8>& output) {
+    IoctlGetWaitbase params{};
+    std::memcpy(&params, input.data(), sizeof(IoctlGetWaitbase));
+    LOG_INFO(Service_NVDRV, "called, unknown=0x{:X}", params.unknown);
+    params.value = 0; // Seems to be hard coded at 0
+    std::memcpy(output.data(), &params, sizeof(IoctlGetWaitbase));
+    return 0;
+}
+
+u32 nvhost_vic::MapBuffer(const std::vector<u8>& input, std::vector<u8>& output) {
+    IoctlMapBuffer params{};
+    std::memcpy(&params, input.data(), sizeof(IoctlMapBuffer));
+    LOG_WARNING(Service_NVDRV, "(STUBBED) called with address={:08X}{:08X}", params.address_2,
+                params.address_1);
+    params.address_1 = 0;
+    params.address_2 = 0;
+    std::memcpy(output.data(), &params, sizeof(IoctlMapBuffer));
+    return 0;
+}
+
+u32 nvhost_vic::MapBufferEx(const std::vector<u8>& input, std::vector<u8>& output) {
+    IoctlMapBufferEx params{};
+    std::memcpy(&params, input.data(), sizeof(IoctlMapBufferEx));
+    LOG_WARNING(Service_NVDRV, "(STUBBED) called with address={:08X}{:08X}", params.address_2,
+                params.address_1);
+    params.address_1 = 0;
+    params.address_2 = 0;
+    std::memcpy(output.data(), &params, sizeof(IoctlMapBufferEx));
+    return 0;
+}
+
+u32 nvhost_vic::UnmapBufferEx(const std::vector<u8>& input, std::vector<u8>& output) {
+    IoctlUnmapBufferEx params{};
+    std::memcpy(&params, input.data(), sizeof(IoctlUnmapBufferEx));
+    LOG_WARNING(Service_NVDRV, "(STUBBED) called");
+    std::memcpy(output.data(), &params, sizeof(IoctlUnmapBufferEx));
+    return 0;
+}
+
 } // namespace Service::Nvidia::Devices

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

@@ -4,6 +4,7 @@
 
 #pragma once
 
+#include <array>
 #include <vector>
 #include "common/common_types.h"
 #include "common/swap.h"
@@ -23,6 +24,12 @@ public:
 private:
     enum class IoctlCommand : u32_le {
         IocSetNVMAPfdCommand = 0x40044801,
+        IocSubmit = 0xC0400001,
+        IocGetSyncpoint = 0xC0080002,
+        IocGetWaitbase = 0xC0080003,
+        IocMapBuffer = 0xC01C0009,
+        IocMapBufferEx = 0xC03C0009,
+        IocUnmapBufferEx = 0xC03C000A,
     };
 
     struct IoctlSetNvmapFD {
@@ -30,9 +37,65 @@ private:
     };
     static_assert(sizeof(IoctlSetNvmapFD) == 4, "IoctlSetNvmapFD is incorrect size");
 
+    struct IoctlSubmitCommandBuffer {
+        u32 id;
+        u32 offset;
+        u32 count;
+    };
+    static_assert(sizeof(IoctlSubmitCommandBuffer) == 0xC,
+                  "IoctlSubmitCommandBuffer is incorrect size");
+
+    struct IoctlSubmit {
+        u32 command_buffer_count;
+        u32 relocations_count;
+        u32 syncpt_count;
+        u32 wait_count;
+        std::array<IoctlSubmitCommandBuffer, 4> command_buffer;
+    };
+    static_assert(sizeof(IoctlSubmit) == 0x40, "IoctlSubmit is incorrect size");
+
+    struct IoctlGetSyncpoint {
+        u32 unknown; // seems to be ignored? Nintendo added this
+        u32 value;
+    };
+    static_assert(sizeof(IoctlGetSyncpoint) == 0x8, "IoctlGetSyncpoint is incorrect size");
+
+    struct IoctlGetWaitbase {
+        u32 unknown; // seems to be ignored? Nintendo added this
+        u32 value;
+    };
+    static_assert(sizeof(IoctlGetWaitbase) == 0x8, "IoctlGetWaitbase is incorrect size");
+
+    struct IoctlMapBuffer {
+        u32 unknown;
+        u32 address_1;
+        u32 address_2;
+        INSERT_PADDING_BYTES(0x10); // TODO(DarkLordZach): RE this structure
+    };
+    static_assert(sizeof(IoctlMapBuffer) == 0x1C, "IoctlMapBuffer is incorrect size");
+
+    struct IoctlMapBufferEx {
+        u32 unknown;
+        u32 address_1;
+        u32 address_2;
+        INSERT_PADDING_BYTES(0x30); // TODO(DarkLordZach): RE this structure
+    };
+    static_assert(sizeof(IoctlMapBufferEx) == 0x3C, "IoctlMapBufferEx is incorrect size");
+
+    struct IoctlUnmapBufferEx {
+        INSERT_PADDING_BYTES(0x3C); // TODO(DarkLordZach): RE this structure
+    };
+    static_assert(sizeof(IoctlUnmapBufferEx) == 0x3C, "IoctlUnmapBufferEx is incorrect size");
+
     u32_le nvmap_fd{};
 
     u32 SetNVMAPfd(const std::vector<u8>& input, std::vector<u8>& output);
+    u32 Submit(const std::vector<u8>& input, std::vector<u8>& output);
+    u32 GetSyncpoint(const std::vector<u8>& input, std::vector<u8>& output);
+    u32 GetWaitbase(const std::vector<u8>& input, std::vector<u8>& output);
+    u32 MapBuffer(const std::vector<u8>& input, std::vector<u8>& output);
+    u32 MapBufferEx(const std::vector<u8>& input, std::vector<u8>& output);
+    u32 UnmapBufferEx(const std::vector<u8>& input, std::vector<u8>& output);
 };
 
 } // namespace Service::Nvidia::Devices

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

@@ -54,7 +54,7 @@ struct EventInterface {
             }
             mask = mask >> 1;
         }
-        return {};
+        return std::nullopt;
     }
     void SetEventStatus(const u32 event_id, EventState new_status) {
         EventState old_status = status[event_id];

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

@@ -114,7 +114,7 @@ std::optional<u64> NVFlinger::OpenDisplay(std::string_view name) {
                      [&](const VI::Display& display) { return display.GetName() == name; });
 
     if (itr == displays.end()) {
-        return {};
+        return std::nullopt;
     }
 
     return itr->GetID();
@@ -124,7 +124,7 @@ std::optional<u64> NVFlinger::CreateLayer(u64 display_id) {
     auto* const display = FindDisplay(display_id);
 
     if (display == nullptr) {
-        return {};
+        return std::nullopt;
     }
 
     const u64 layer_id = next_layer_id++;
@@ -144,7 +144,7 @@ std::optional<u32> NVFlinger::FindBufferQueueId(u64 display_id, u64 layer_id) co
     const auto* const layer = FindLayer(display_id, layer_id);
 
     if (layer == nullptr) {
-        return {};
+        return std::nullopt;
     }
 
     return layer->GetBufferQueue().GetId();

src/core/hle/service/service.cpp

@@ -72,6 +72,23 @@
 
 namespace Service {
 
+/**
+ * Creates a function string for logging, complete with the name (or header code, depending
+ * on what's passed in) the port name, and all the cmd_buff arguments.
+ */
+[[maybe_unused]] static std::string MakeFunctionString(std::string_view name,
+                                                       std::string_view port_name,
+                                                       const u32* cmd_buff) {
+    // Number of params == bits 0-5 + bits 6-11
+    int num_params = (cmd_buff[0] & 0x3F) + ((cmd_buff[0] >> 6) & 0x3F);
+
+    std::string function_string = fmt::format("function '{}': port={}", name, port_name);
+    for (int i = 1; i <= num_params; ++i) {
+        function_string += fmt::format(", cmd_buff[{}]=0x{:X}", i, cmd_buff[i]);
+    }
+    return function_string;
+}
+
 ServiceFrameworkBase::ServiceFrameworkBase(const char* service_name, u32 max_sessions,
                                            InvokerFn* handler_invoker)
     : service_name(service_name), max_sessions(max_sessions), handler_invoker(handler_invoker) {}

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

@@ -497,7 +497,7 @@ std::pair<s32, Errno> BSD::PollImpl(std::vector<u8>& write_buffer, std::vector<u
             return {0, Errno::SUCCESS};
         }
 
-        std::optional<FileDescriptor>& descriptor = file_descriptors[pollfd.fd];
+        const std::optional<FileDescriptor>& descriptor = file_descriptors[pollfd.fd];
         if (!descriptor) {
             LOG_ERROR(Service, "File descriptor handle={} is not allocated", pollfd.fd);
             pollfd.revents = POLL_NVAL;

src/core/loader/nso.cpp

@@ -76,16 +76,16 @@ std::optional<VAddr> AppLoader_NSO::LoadModule(Kernel::Process& process, Core::S
                                                bool should_pass_arguments, bool load_into_process,
                                                std::optional<FileSys::PatchManager> pm) {
     if (file.GetSize() < sizeof(NSOHeader)) {
-        return {};
+        return std::nullopt;
     }
 
     NSOHeader nso_header{};
     if (sizeof(NSOHeader) != file.ReadObject(&nso_header)) {
-        return {};
+        return std::nullopt;
     }
 
     if (nso_header.magic != Common::MakeMagic('N', 'S', 'O', '0')) {
-        return {};
+        return std::nullopt;
     }
 
     // Build program image

src/core/memory.cpp

@@ -567,7 +567,7 @@ struct Memory::Impl {
      * @param page_table The page table to use to perform the mapping.
      * @param base       The base address to begin mapping at.
      * @param size       The total size of the range in bytes.
-     * @param memory     The memory to map.
+     * @param target     The target address to begin mapping from.
      * @param type       The page type to map the memory as.
      */
     void MapPages(Common::PageTable& page_table, VAddr base, u64 size, PAddr target,

src/input_common/gcadapter/gc_adapter.cpp

@@ -15,7 +15,10 @@
 #endif
 
 #include "common/logging/log.h"
+#include "common/param_package.h"
+#include "core/settings.h"
 #include "input_common/gcadapter/gc_adapter.h"
+#include "input_common/settings.h"
 
 namespace GCAdapter {
 
@@ -127,6 +130,23 @@ void Adapter::Read() {
             adapter_thread_running = false; // error reading from adapter, stop reading.
             break;
         }
+
+        if (Settings::values.vibration_enabled) {
+            // Modulate vibrations to 8 states
+            vibration_counter = (vibration_counter + 1) % 8;
+            const bool p1 = rumble[0] > vibration_counter;
+            const bool p2 = rumble[1] > vibration_counter;
+            const bool p3 = rumble[2] > vibration_counter;
+            const bool p4 = rumble[3] > vibration_counter;
+            Vibrate(p1, p2, p3, p4);
+        } else {
+            // stop all onging vibrations
+            if (vibration_counter != 8) {
+                vibration_counter = 8;
+                Vibrate(false, false, false, false);
+            }
+        }
+
         for (std::size_t port = 0; port < pads.size(); ++port) {
             pads[port] = GetPadStatus(port, adapter_payload);
             if (DeviceConnected(port) && configuring) {
@@ -158,6 +178,8 @@ void Adapter::Setup() {
     adapter_controllers_status.fill(ControllerTypes::None);
     // Initialize all ports to store axis origin values
     get_origin.fill(true);
+    // Initialize all rumble to no vibrate
+    rumble.fill(0);
 
     // pointer to list of connected usb devices
     libusb_device** devices{};
@@ -292,6 +314,109 @@ void Adapter::Reset() {
     }
 }
 
+std::vector<Common::ParamPackage> Adapter::GetInputDevices() const {
+    std::vector<Common::ParamPackage> devices;
+    for (std::size_t port = 0; port < state.size(); ++port) {
+        if (!DeviceConnected(port)) {
+            continue;
+        }
+        std::string name = fmt::format("Gamecube Controller {}", port);
+        devices.emplace_back(Common::ParamPackage{
+            {"class", "gcpad"},
+            {"display", std::move(name)},
+            {"port", std::to_string(port)},
+        });
+    }
+    return devices;
+}
+
+InputCommon::ButtonMapping Adapter::GetButtonMappingForDevice(
+    const Common::ParamPackage& params) const {
+    // This list is missing ZL/ZR since those are not considered buttons.
+    // We will add those afterwards
+    // This list also excludes any button that can't be really mapped
+    static constexpr std::array<std::pair<Settings::NativeButton::Values, PadButton>, 12>
+        switch_to_gcadapter_button = {
+            std::pair{Settings::NativeButton::A, PadButton::PAD_BUTTON_A},
+            {Settings::NativeButton::B, PadButton::PAD_BUTTON_B},
+            {Settings::NativeButton::X, PadButton::PAD_BUTTON_X},
+            {Settings::NativeButton::Y, PadButton::PAD_BUTTON_Y},
+            {Settings::NativeButton::Plus, PadButton::PAD_BUTTON_START},
+            {Settings::NativeButton::DLeft, PadButton::PAD_BUTTON_LEFT},
+            {Settings::NativeButton::DUp, PadButton::PAD_BUTTON_UP},
+            {Settings::NativeButton::DRight, PadButton::PAD_BUTTON_RIGHT},
+            {Settings::NativeButton::DDown, PadButton::PAD_BUTTON_DOWN},
+            {Settings::NativeButton::SL, PadButton::PAD_TRIGGER_L},
+            {Settings::NativeButton::SR, PadButton::PAD_TRIGGER_R},
+            {Settings::NativeButton::R, PadButton::PAD_TRIGGER_Z},
+        };
+    if (!params.Has("port")) {
+        return {};
+    }
+
+    InputCommon::ButtonMapping mapping{};
+    for (const auto& [switch_button, gcadapter_button] : switch_to_gcadapter_button) {
+        Common::ParamPackage button_params({{"engine", "gcpad"}});
+        button_params.Set("port", params.Get("port", 0));
+        button_params.Set("button", static_cast<int>(gcadapter_button));
+        mapping.insert_or_assign(switch_button, std::move(button_params));
+    }
+
+    // Add the missing bindings for ZL/ZR
+    static constexpr std::array<std::pair<Settings::NativeButton::Values, PadAxes>, 2>
+        switch_to_gcadapter_axis = {
+            std::pair{Settings::NativeButton::ZL, PadAxes::TriggerLeft},
+            {Settings::NativeButton::ZR, PadAxes::TriggerRight},
+        };
+    for (const auto& [switch_button, gcadapter_axis] : switch_to_gcadapter_axis) {
+        Common::ParamPackage button_params({{"engine", "gcpad"}});
+        button_params.Set("port", params.Get("port", 0));
+        button_params.Set("button", static_cast<int>(PadButton::PAD_STICK));
+        button_params.Set("axis", static_cast<int>(gcadapter_axis));
+        mapping.insert_or_assign(switch_button, std::move(button_params));
+    }
+    return mapping;
+}
+
+InputCommon::AnalogMapping Adapter::GetAnalogMappingForDevice(
+    const Common::ParamPackage& params) const {
+    if (!params.Has("port")) {
+        return {};
+    }
+
+    InputCommon::AnalogMapping mapping = {};
+    Common::ParamPackage left_analog_params;
+    left_analog_params.Set("engine", "gcpad");
+    left_analog_params.Set("port", params.Get("port", 0));
+    left_analog_params.Set("axis_x", static_cast<int>(PadAxes::StickX));
+    left_analog_params.Set("axis_y", static_cast<int>(PadAxes::StickY));
+    mapping.insert_or_assign(Settings::NativeAnalog::LStick, std::move(left_analog_params));
+    Common::ParamPackage right_analog_params;
+    right_analog_params.Set("engine", "gcpad");
+    right_analog_params.Set("port", params.Get("port", 0));
+    right_analog_params.Set("axis_x", static_cast<int>(PadAxes::SubstickX));
+    right_analog_params.Set("axis_y", static_cast<int>(PadAxes::SubstickY));
+    mapping.insert_or_assign(Settings::NativeAnalog::RStick, std::move(right_analog_params));
+    return mapping;
+}
+
+void Adapter::Vibrate(bool p1, bool p2, bool p3, bool p4) {
+    int size = 0;
+    std::array<u8, 5> payload = {0x11, p1, p2, p3, p4};
+    const int err = libusb_interrupt_transfer(usb_adapter_handle, output_endpoint, payload.data(),
+                                              sizeof(payload), &size, 16);
+    if (err) {
+        LOG_ERROR(Input, "Adapter libusb write failed: {}", libusb_error_name(err));
+    }
+}
+
+bool Adapter::RumblePlay(int port, f32 amplitude) {
+    const u8 raw_amp = static_cast<u8>(amplitude * 0x8);
+    rumble[port] = raw_amp;
+
+    return false;
+}
+
 bool Adapter::DeviceConnected(std::size_t port) const {
     return adapter_controllers_status[port] != ControllerTypes::None;
 }

src/input_common/gcadapter/gc_adapter.h

@@ -10,6 +10,7 @@
 #include <unordered_map>
 #include "common/common_types.h"
 #include "common/threadsafe_queue.h"
+#include "input_common/main.h"
 
 struct libusb_context;
 struct libusb_device;
@@ -75,6 +76,10 @@ public:
     void BeginConfiguration();
     void EndConfiguration();
 
+    std::vector<Common::ParamPackage> GetInputDevices() const;
+    InputCommon::ButtonMapping GetButtonMappingForDevice(const Common::ParamPackage& params) const;
+    InputCommon::AnalogMapping GetAnalogMappingForDevice(const Common::ParamPackage& params) const;
+
     /// Returns true if there is a device connected to port
     bool DeviceConnected(std::size_t port) const;
 
@@ -86,6 +91,8 @@ public:
 
     int GetOriginValue(int port, int axis) const;
 
+    bool RumblePlay(int port, f32 amplitude);
+
 private:
     GCPadStatus GetPadStatus(std::size_t port, const std::array<u8, 37>& adapter_payload);
 
@@ -108,6 +115,9 @@ private:
     /// For use in initialization, querying devices to find the adapter
     void Setup();
 
+    // Updates vibration state of all controllers
+    void Vibrate(bool p1, bool p2, bool p3, bool p4);
+
     libusb_device_handle* usb_adapter_handle = nullptr;
 
     std::thread adapter_input_thread;
@@ -117,10 +127,12 @@ private:
 
     u8 input_endpoint = 0;
     u8 output_endpoint = 0;
+    u8 vibration_counter = 0;
 
     bool configuring = false;
 
     std::array<GCState, 4> state;
+    std::array<u8, 4> rumble;
     std::array<bool, 4> get_origin;
     std::array<GCPadStatus, 4> origin_status;
     std::array<Common::SPSCQueue<GCPadStatus>, 4> pad_queue;

src/input_common/gcadapter/gc_poller.cpp

@@ -15,7 +15,7 @@ namespace InputCommon {
 
 class GCButton final : public Input::ButtonDevice {
 public:
-    explicit GCButton(int port_, int button_, const GCAdapter::Adapter* adapter)
+    explicit GCButton(int port_, int button_, GCAdapter::Adapter* adapter)
         : port(port_), button(button_), gcadapter(adapter) {}
 
     ~GCButton() override;
@@ -27,10 +27,17 @@ public:
         return false;
     }
 
+    bool SetRumblePlay(f32 amp_high, f32 amp_low, f32 freq_high, f32 freq_low) const override {
+        const f32 amplitude = amp_high + amp_low > 2 ? 1.0f : (amp_high + amp_low) * 0.5f;
+        const f32 new_amp = pow(amplitude, 0.5f) * (3.0f - 2.0f * pow(amplitude, 0.15f));
+
+        return gcadapter->RumblePlay(port, new_amp);
+    }
+
 private:
     const int port;
     const int button;
-    const GCAdapter::Adapter* gcadapter;
+    GCAdapter::Adapter* gcadapter;
 };
 
 class GCAxisButton final : public Input::ButtonDevice {

src/input_common/main.cpp

@@ -22,7 +22,7 @@ namespace InputCommon {
 
 struct InputSubsystem::Impl {
     void Initialize() {
-        auto gcadapter = std::make_shared<GCAdapter::Adapter>();
+        gcadapter = std::make_shared<GCAdapter::Adapter>();
         gcbuttons = std::make_shared<GCButtonFactory>(gcadapter);
         Input::RegisterFactory<Input::ButtonDevice>("gcpad", gcbuttons);
         gcanalog = std::make_shared<GCAnalogFactory>(gcadapter);
@@ -82,6 +82,8 @@ struct InputSubsystem::Impl {
 #endif
         auto udp_devices = udp->GetInputDevices();
         devices.insert(devices.end(), udp_devices.begin(), udp_devices.end());
+        auto gcpad_devices = gcadapter->GetInputDevices();
+        devices.insert(devices.end(), gcpad_devices.begin(), gcpad_devices.end());
         return devices;
     }
 
@@ -94,6 +96,9 @@ struct InputSubsystem::Impl {
             // TODO consider returning the SDL key codes for the default keybindings
             return {};
         }
+        if (params.Get("class", "") == "gcpad") {
+            return gcadapter->GetAnalogMappingForDevice(params);
+        }
 #ifdef HAVE_SDL2
         if (params.Get("class", "") == "sdl") {
             return sdl->GetAnalogMappingForDevice(params);
@@ -111,6 +116,9 @@ struct InputSubsystem::Impl {
             // TODO consider returning the SDL key codes for the default keybindings
             return {};
         }
+        if (params.Get("class", "") == "gcpad") {
+            return gcadapter->GetButtonMappingForDevice(params);
+        }
 #ifdef HAVE_SDL2
         if (params.Get("class", "") == "sdl") {
             return sdl->GetButtonMappingForDevice(params);
@@ -141,6 +149,7 @@ struct InputSubsystem::Impl {
     std::shared_ptr<UDPMotionFactory> udpmotion;
     std::shared_ptr<UDPTouchFactory> udptouch;
     std::shared_ptr<CemuhookUDP::Client> udp;
+    std::shared_ptr<GCAdapter::Adapter> gcadapter;
 };
 
 InputSubsystem::InputSubsystem() : impl{std::make_unique<Impl>()} {}

src/input_common/sdl/sdl_impl.cpp

@@ -5,6 +5,7 @@
 #include <algorithm>
 #include <array>
 #include <atomic>
+#include <chrono>
 #include <cmath>
 #include <functional>
 #include <mutex>
@@ -78,6 +79,33 @@ public:
         return state.axes.at(axis) / (32767.0f * range);
     }
 
+    bool RumblePlay(f32 amp_low, f32 amp_high, int time) {
+        const u16 raw_amp_low = static_cast<u16>(amp_low * 0xFFFF);
+        const u16 raw_amp_high = static_cast<u16>(amp_high * 0xFFFF);
+        // Lower drastically the number of state changes
+        if (raw_amp_low >> 11 == last_state_rumble_low >> 11 &&
+            raw_amp_high >> 11 == last_state_rumble_high >> 11) {
+            if (raw_amp_low + raw_amp_high != 0 ||
+                last_state_rumble_low + last_state_rumble_high == 0) {
+                return false;
+            }
+        }
+        // Don't change state if last vibration was < 20ms
+        const auto now = std::chrono::system_clock::now();
+        if (std::chrono::duration_cast<std::chrono::milliseconds>(now - last_vibration) <
+            std::chrono::milliseconds(20)) {
+            return raw_amp_low + raw_amp_high == 0;
+        }
+
+        last_vibration = now;
+        last_state_rumble_low = raw_amp_low;
+        last_state_rumble_high = raw_amp_high;
+        if (sdl_joystick) {
+            SDL_JoystickRumble(sdl_joystick.get(), raw_amp_low, raw_amp_high, time);
+        }
+        return false;
+    }
+
     std::tuple<float, float> GetAnalog(int axis_x, int axis_y, float range) const {
         float x = GetAxis(axis_x, range);
         float y = GetAxis(axis_y, range);
@@ -139,6 +167,9 @@ private:
     } state;
     std::string guid;
     int port;
+    u16 last_state_rumble_high;
+    u16 last_state_rumble_low;
+    std::chrono::time_point<std::chrono::system_clock> last_vibration;
     std::unique_ptr<SDL_Joystick, decltype(&SDL_JoystickClose)> sdl_joystick;
     std::unique_ptr<SDL_GameController, decltype(&SDL_GameControllerClose)> sdl_controller;
     mutable std::mutex mutex;
@@ -207,7 +238,7 @@ void SDLState::InitJoystick(int joystick_index) {
         sdl_gamecontroller = SDL_GameControllerOpen(joystick_index);
     }
     if (!sdl_joystick) {
-        LOG_ERROR(Input, "failed to open joystick {}", joystick_index);
+        LOG_ERROR(Input, "Failed to open joystick {}", joystick_index);
         return;
     }
     const std::string guid = GetGUID(sdl_joystick);
@@ -303,6 +334,12 @@ public:
         return joystick->GetButton(button);
     }
 
+    bool SetRumblePlay(f32 amp_high, f32 amp_low, f32 freq_high, f32 freq_low) const override {
+        const f32 new_amp_low = pow(amp_low, 0.5f) * (3.0f - 2.0f * pow(amp_low, 0.15f));
+        const f32 new_amp_high = pow(amp_high, 0.5f) * (3.0f - 2.0f * pow(amp_high, 0.15f));
+        return joystick->RumblePlay(new_amp_low, new_amp_high, 250);
+    }
+
 private:
     std::shared_ptr<SDLJoystick> joystick;
     int button;

src/video_core/CMakeLists.txt

@@ -190,6 +190,8 @@ if (ENABLE_VULKAN)
         renderer_vulkan/vk_blit_screen.h
         renderer_vulkan/vk_buffer_cache.cpp
         renderer_vulkan/vk_buffer_cache.h
+        renderer_vulkan/vk_command_pool.cpp
+        renderer_vulkan/vk_command_pool.h
         renderer_vulkan/vk_compute_pass.cpp
         renderer_vulkan/vk_compute_pass.h
         renderer_vulkan/vk_compute_pipeline.cpp
@@ -204,6 +206,8 @@ if (ENABLE_VULKAN)
         renderer_vulkan/vk_graphics_pipeline.h
         renderer_vulkan/vk_image.cpp
         renderer_vulkan/vk_image.h
+        renderer_vulkan/vk_master_semaphore.cpp
+        renderer_vulkan/vk_master_semaphore.h
         renderer_vulkan/vk_memory_manager.cpp
         renderer_vulkan/vk_memory_manager.h
         renderer_vulkan/vk_pipeline_cache.cpp
@@ -214,8 +218,8 @@ if (ENABLE_VULKAN)
         renderer_vulkan/vk_rasterizer.h
         renderer_vulkan/vk_renderpass_cache.cpp
         renderer_vulkan/vk_renderpass_cache.h
-        renderer_vulkan/vk_resource_manager.cpp
-        renderer_vulkan/vk_resource_manager.h
+        renderer_vulkan/vk_resource_pool.cpp
+        renderer_vulkan/vk_resource_pool.h
         renderer_vulkan/vk_sampler_cache.cpp
         renderer_vulkan/vk_sampler_cache.h
         renderer_vulkan/vk_scheduler.cpp

src/video_core/engines/fermi_2d.cpp

@@ -87,12 +87,12 @@ void Fermi2D::HandleSurfaceCopy() {
     const Common::Rectangle<u32> src_rect{src_blit_x1, src_blit_y1, src_blit_x2, src_blit_y2};
     const Common::Rectangle<u32> dst_rect{regs.blit_dst_x, regs.blit_dst_y, dst_blit_x2,
                                           dst_blit_y2};
-    Config copy_config;
-    copy_config.operation = regs.operation;
-    copy_config.filter = regs.blit_control.filter;
-    copy_config.src_rect = src_rect;
-    copy_config.dst_rect = dst_rect;
-
+    const Config copy_config{
+        .operation = regs.operation,
+        .filter = regs.blit_control.filter,
+        .src_rect = src_rect,
+        .dst_rect = dst_rect,
+    };
     if (!rasterizer->AccelerateSurfaceCopy(regs.src, regs.dst, copy_config)) {
         UNIMPLEMENTED();
     }

src/video_core/engines/fermi_2d.h

@@ -145,8 +145,8 @@ public:
     } regs{};
 
     struct Config {
-        Operation operation;
-        Filter filter;
+        Operation operation{};
+        Filter filter{};
         Common::Rectangle<u32> src_rect;
         Common::Rectangle<u32> dst_rect;
     };

src/video_core/engines/maxwell_3d.cpp

@@ -597,7 +597,7 @@ std::optional<u64> Maxwell3D::GetQueryResult() {
         // Deferred.
         rasterizer->Query(regs.query.QueryAddress(), VideoCore::QueryType::SamplesPassed,
                           system.GPU().GetTicks());
-        return {};
+        return std::nullopt;
     default:
         LOG_DEBUG(HW_GPU, "Unimplemented query select type {}",
                   static_cast<u32>(regs.query.query_get.select.Value()));

src/video_core/macro/macro.cpp

@@ -36,7 +36,7 @@ void MacroEngine::Execute(Engines::Maxwell3D& maxwell3d, u32 method,
         }
     } else {
         // Macro not compiled, check if it's uploaded and if so, compile it
-        std::optional<u32> mid_method = std::nullopt;
+        std::optional<u32> mid_method;
         const auto macro_code = uploaded_macro_code.find(method);
         if (macro_code == uploaded_macro_code.end()) {
             for (const auto& [method_base, code] : uploaded_macro_code) {

src/video_core/memory_manager.cpp

@@ -58,7 +58,7 @@ void MemoryManager::Unmap(GPUVAddr gpu_addr, std::size_t size) {
 std::optional<GPUVAddr> MemoryManager::AllocateFixed(GPUVAddr gpu_addr, std::size_t size) {
     for (u64 offset{}; offset < size; offset += page_size) {
         if (!GetPageEntry(gpu_addr + offset).IsUnmapped()) {
-            return {};
+            return std::nullopt;
         }
     }
 
@@ -135,13 +135,13 @@ std::optional<GPUVAddr> MemoryManager::FindFreeRange(std::size_t size, std::size
         }
     }
 
-    return {};
+    return std::nullopt;
 }
 
 std::optional<VAddr> MemoryManager::GpuToCpuAddress(GPUVAddr gpu_addr) const {
     const auto page_entry{GetPageEntry(gpu_addr)};
     if (!page_entry.IsValid()) {
-        return {};
+        return std::nullopt;
     }
 
     return page_entry.ToAddress() + (gpu_addr & page_mask);

src/video_core/query_cache.h

@@ -91,8 +91,7 @@ private:
     std::shared_ptr<HostCounter> last;
 };
 
-template <class QueryCache, class CachedQuery, class CounterStream, class HostCounter,
-          class QueryPool>
+template <class QueryCache, class CachedQuery, class CounterStream, class HostCounter>
 class QueryCacheBase {
 public:
     explicit QueryCacheBase(VideoCore::RasterizerInterface& rasterizer_,
@@ -206,9 +205,6 @@ public:
         committed_flushes.pop_front();
     }
 
-protected:
-    std::array<QueryPool, VideoCore::NumQueryTypes> query_pools;
-
 private:
     /// Flushes a memory range to guest memory and removes it from the cache.
     void FlushAndRemoveRegion(VAddr addr, std::size_t size) {

src/video_core/renderer_base.h

@@ -46,11 +46,6 @@ public:
     /// Finalize rendering the guest frame and draw into the presentation texture
     virtual void SwapBuffers(const Tegra::FramebufferConfig* framebuffer) = 0;
 
-    /// Draws the latest frame to the window waiting timeout_ms for a frame to arrive (Renderer
-    /// specific implementation)
-    /// Returns true if a frame was drawn
-    virtual bool TryPresent(int timeout_ms) = 0;
-
     // Getter/setter functions:
     // ------------------------
 

src/video_core/renderer_opengl/gl_query_cache.cpp

@@ -32,10 +32,8 @@ constexpr GLenum GetTarget(VideoCore::QueryType type) {
 
 QueryCache::QueryCache(RasterizerOpenGL& rasterizer, Tegra::Engines::Maxwell3D& maxwell3d,
                        Tegra::MemoryManager& gpu_memory)
-    : VideoCommon::QueryCacheBase<
-          QueryCache, CachedQuery, CounterStream, HostCounter,
-          std::vector<OGLQuery>>{static_cast<VideoCore::RasterizerInterface&>(rasterizer),
-                                 maxwell3d, gpu_memory},
+    : VideoCommon::QueryCacheBase<QueryCache, CachedQuery, CounterStream, HostCounter>(
+          rasterizer, maxwell3d, gpu_memory),
       gl_rasterizer{rasterizer} {}
 
 QueryCache::~QueryCache() = default;
@@ -91,6 +89,8 @@ u64 HostCounter::BlockingQuery() const {
 CachedQuery::CachedQuery(QueryCache& cache, VideoCore::QueryType type, VAddr cpu_addr, u8* host_ptr)
     : VideoCommon::CachedQueryBase<HostCounter>{cpu_addr, host_ptr}, cache{&cache}, type{type} {}
 
+CachedQuery::~CachedQuery() = default;
+
 CachedQuery::CachedQuery(CachedQuery&& rhs) noexcept
     : VideoCommon::CachedQueryBase<HostCounter>(std::move(rhs)), cache{rhs.cache}, type{rhs.type} {}
 

src/video_core/renderer_opengl/gl_query_cache.h

@@ -26,8 +26,8 @@ class RasterizerOpenGL;
 
 using CounterStream = VideoCommon::CounterStreamBase<QueryCache, HostCounter>;
 
-class QueryCache final : public VideoCommon::QueryCacheBase<QueryCache, CachedQuery, CounterStream,
-                                                            HostCounter, std::vector<OGLQuery>> {
+class QueryCache final
+    : public VideoCommon::QueryCacheBase<QueryCache, CachedQuery, CounterStream, HostCounter> {
 public:
     explicit QueryCache(RasterizerOpenGL& rasterizer, Tegra::Engines::Maxwell3D& maxwell3d,
                         Tegra::MemoryManager& gpu_memory);
@@ -41,6 +41,7 @@ public:
 
 private:
     RasterizerOpenGL& gl_rasterizer;
+    std::array<std::vector<OGLQuery>, VideoCore::NumQueryTypes> query_pools;
 };
 
 class HostCounter final : public VideoCommon::HostCounterBase<QueryCache, HostCounter> {
@@ -63,10 +64,12 @@ class CachedQuery final : public VideoCommon::CachedQueryBase<HostCounter> {
 public:
     explicit CachedQuery(QueryCache& cache, VideoCore::QueryType type, VAddr cpu_addr,
                          u8* host_ptr);
-    CachedQuery(CachedQuery&& rhs) noexcept;
-    CachedQuery(const CachedQuery&) = delete;
+    ~CachedQuery() override;
 
+    CachedQuery(CachedQuery&& rhs) noexcept;
     CachedQuery& operator=(CachedQuery&& rhs) noexcept;
+
+    CachedQuery(const CachedQuery&) = delete;
     CachedQuery& operator=(const CachedQuery&) = delete;
 
     void Flush() override;

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -813,7 +813,7 @@ private:
         const u8 location = static_cast<u8>(static_cast<u32>(index) * 4 + element);
         const auto it = transform_feedback.find(location);
         if (it == transform_feedback.end()) {
-            return {};
+            return std::nullopt;
         }
         return it->second.components;
     }
@@ -1295,21 +1295,21 @@ private:
             switch (element) {
             case 0:
                 UNIMPLEMENTED();
-                return {};
+                return std::nullopt;
             case 1:
                 if (stage == ShaderType::Vertex && !device.HasVertexViewportLayer()) {
-                    return {};
+                    return std::nullopt;
                 }
                 return {{"gl_Layer", Type::Int}};
             case 2:
                 if (stage == ShaderType::Vertex && !device.HasVertexViewportLayer()) {
-                    return {};
+                    return std::nullopt;
                 }
                 return {{"gl_ViewportIndex", Type::Int}};
             case 3:
                 return {{"gl_PointSize", Type::Float}};
             }
-            return {};
+            return std::nullopt;
         case Attribute::Index::FrontColor:
             return {{"gl_FrontColor"s + GetSwizzle(element), Type::Float}};
         case Attribute::Index::FrontSecondaryColor:
@@ -1332,7 +1332,7 @@ private:
                          Type::Float}};
             }
             UNIMPLEMENTED_MSG("Unhandled output attribute: {}", static_cast<u32>(attribute));
-            return {};
+            return std::nullopt;
         }
     }
 

src/video_core/renderer_opengl/renderer_opengl.cpp

@@ -32,20 +32,6 @@ namespace OpenGL {
 
 namespace {
 
-constexpr std::size_t SWAP_CHAIN_SIZE = 3;
-
-struct Frame {
-    u32 width{};                      /// Width of the frame (to detect resize)
-    u32 height{};                     /// Height of the frame
-    bool color_reloaded{};            /// Texture attachment was recreated (ie: resized)
-    OpenGL::OGLRenderbuffer color{};  /// Buffer shared between the render/present FBO
-    OpenGL::OGLFramebuffer render{};  /// FBO created on the render thread
-    OpenGL::OGLFramebuffer present{}; /// FBO created on the present thread
-    GLsync render_fence{};            /// Fence created on the render thread
-    GLsync present_fence{};           /// Fence created on the presentation thread
-    bool is_srgb{};                   /// Framebuffer is sRGB or RGB
-};
-
 constexpr GLint PositionLocation = 0;
 constexpr GLint TexCoordLocation = 1;
 constexpr GLint ModelViewMatrixLocation = 0;
@@ -58,24 +44,6 @@ struct ScreenRectVertex {
     std::array<GLfloat, 2> tex_coord;
 };
 
-/// Returns true if any debug tool is attached
-bool HasDebugTool() {
-    const bool nsight = std::getenv("NVTX_INJECTION64_PATH") || std::getenv("NSIGHT_LAUNCHED");
-    if (nsight) {
-        return true;
-    }
-
-    GLint num_extensions;
-    glGetIntegerv(GL_NUM_EXTENSIONS, &num_extensions);
-    for (GLuint index = 0; index < static_cast<GLuint>(num_extensions); ++index) {
-        const auto name = reinterpret_cast<const char*>(glGetStringi(GL_EXTENSIONS, index));
-        if (!std::strcmp(name, "GL_EXT_debug_tool")) {
-            return true;
-        }
-    }
-    return false;
-}
-
 /**
  * Defines a 1:1 pixel ortographic projection matrix with (0,0) on the top-left
  * corner and (width, height) on the lower-bottom.
@@ -159,135 +127,15 @@ void APIENTRY DebugHandler(GLenum source, GLenum type, GLuint id, GLenum severit
 
 } // Anonymous namespace
 
-/**
- * For smooth Vsync rendering, we want to always present the latest frame that the core generates,
- * but also make sure that rendering happens at the pace that the frontend dictates. This is a
- * helper class that the renderer uses to sync frames between the render thread and the presentation
- * thread
- */
-class FrameMailbox {
-public:
-    std::mutex swap_chain_lock;
-    std::condition_variable present_cv;
-    std::array<Frame, SWAP_CHAIN_SIZE> swap_chain{};
-    std::queue<Frame*> free_queue;
-    std::deque<Frame*> present_queue;
-    Frame* previous_frame{};
-
-    FrameMailbox() {
-        for (auto& frame : swap_chain) {
-            free_queue.push(&frame);
-        }
-    }
-
-    ~FrameMailbox() {
-        // lock the mutex and clear out the present and free_queues and notify any people who are
-        // blocked to prevent deadlock on shutdown
-        std::scoped_lock lock{swap_chain_lock};
-        std::queue<Frame*>().swap(free_queue);
-        present_queue.clear();
-        present_cv.notify_all();
-    }
-
-    void ReloadPresentFrame(Frame* frame, u32 height, u32 width) {
-        frame->present.Release();
-        frame->present.Create();
-        GLint previous_draw_fbo{};
-        glGetIntegerv(GL_DRAW_FRAMEBUFFER_BINDING, &previous_draw_fbo);
-        glBindFramebuffer(GL_FRAMEBUFFER, frame->present.handle);
-        glFramebufferRenderbuffer(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_RENDERBUFFER,
-                                  frame->color.handle);
-        if (glCheckFramebufferStatus(GL_FRAMEBUFFER) != GL_FRAMEBUFFER_COMPLETE) {
-            LOG_CRITICAL(Render_OpenGL, "Failed to recreate present FBO!");
-        }
-        glBindFramebuffer(GL_DRAW_FRAMEBUFFER, previous_draw_fbo);
-        frame->color_reloaded = false;
-    }
-
-    void ReloadRenderFrame(Frame* frame, u32 width, u32 height) {
-        // Recreate the color texture attachment
-        frame->color.Release();
-        frame->color.Create();
-        const GLenum internal_format = frame->is_srgb ? GL_SRGB8 : GL_RGB8;
-        glNamedRenderbufferStorage(frame->color.handle, internal_format, width, height);
-
-        // Recreate the FBO for the render target
-        frame->render.Release();
-        frame->render.Create();
-        glBindFramebuffer(GL_FRAMEBUFFER, frame->render.handle);
-        glFramebufferRenderbuffer(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_RENDERBUFFER,
-                                  frame->color.handle);
-        if (glCheckFramebufferStatus(GL_FRAMEBUFFER) != GL_FRAMEBUFFER_COMPLETE) {
-            LOG_CRITICAL(Render_OpenGL, "Failed to recreate render FBO!");
-        }
-
-        frame->width = width;
-        frame->height = height;
-        frame->color_reloaded = true;
-    }
-
-    Frame* GetRenderFrame() {
-        std::unique_lock lock{swap_chain_lock};
-
-        // If theres no free frames, we will reuse the oldest render frame
-        if (free_queue.empty()) {
-            auto frame = present_queue.back();
-            present_queue.pop_back();
-            return frame;
-        }
-
-        Frame* frame = free_queue.front();
-        free_queue.pop();
-        return frame;
-    }
-
-    void ReleaseRenderFrame(Frame* frame) {
-        std::unique_lock lock{swap_chain_lock};
-        present_queue.push_front(frame);
-        present_cv.notify_one();
-    }
-
-    Frame* TryGetPresentFrame(int timeout_ms) {
-        std::unique_lock lock{swap_chain_lock};
-        // wait for new entries in the present_queue
-        present_cv.wait_for(lock, std::chrono::milliseconds(timeout_ms),
-                            [&] { return !present_queue.empty(); });
-        if (present_queue.empty()) {
-            // timed out waiting for a frame to draw so return the previous frame
-            return previous_frame;
-        }
-
-        // free the previous frame and add it back to the free queue
-        if (previous_frame) {
-            free_queue.push(previous_frame);
-        }
-
-        // the newest entries are pushed to the front of the queue
-        Frame* frame = present_queue.front();
-        present_queue.pop_front();
-        // remove all old entries from the present queue and move them back to the free_queue
-        for (auto f : present_queue) {
-            free_queue.push(f);
-        }
-        present_queue.clear();
-        previous_frame = frame;
-        return frame;
-    }
-};
-
 RendererOpenGL::RendererOpenGL(Core::TelemetrySession& telemetry_session_,
                                Core::Frontend::EmuWindow& emu_window_,
                                Core::Memory::Memory& cpu_memory_, Tegra::GPU& gpu_,
                                std::unique_ptr<Core::Frontend::GraphicsContext> context)
     : RendererBase{emu_window_, std::move(context)}, telemetry_session{telemetry_session_},
-      emu_window{emu_window_}, cpu_memory{cpu_memory_}, gpu{gpu_}, program_manager{device},
-      has_debug_tool{HasDebugTool()} {}
+      emu_window{emu_window_}, cpu_memory{cpu_memory_}, gpu{gpu_}, program_manager{device} {}
 
 RendererOpenGL::~RendererOpenGL() = default;
 
-MICROPROFILE_DEFINE(OpenGL_RenderFrame, "OpenGL", "Render Frame", MP_RGB(128, 128, 64));
-MICROPROFILE_DEFINE(OpenGL_WaitPresent, "OpenGL", "Wait For Present", MP_RGB(128, 128, 128));
-
 void RendererOpenGL::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
     if (!framebuffer) {
         return;
@@ -296,79 +144,34 @@ void RendererOpenGL::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
     PrepareRendertarget(framebuffer);
     RenderScreenshot();
 
-    Frame* frame;
-    {
-        MICROPROFILE_SCOPE(OpenGL_WaitPresent);
+    glBindFramebuffer(GL_DRAW_FRAMEBUFFER, 0);
+    DrawScreen(emu_window.GetFramebufferLayout());
 
-        frame = frame_mailbox->GetRenderFrame();
+    ++m_current_frame;
 
-        // Clean up sync objects before drawing
-
-        // INTEL driver workaround. We can't delete the previous render sync object until we are
-        // sure that the presentation is done
-        if (frame->present_fence) {
-            glClientWaitSync(frame->present_fence, 0, GL_TIMEOUT_IGNORED);
-        }
-
-        // delete the draw fence if the frame wasn't presented
-        if (frame->render_fence) {
-            glDeleteSync(frame->render_fence);
-            frame->render_fence = 0;
-        }
-
-        // wait for the presentation to be done
-        if (frame->present_fence) {
-            glWaitSync(frame->present_fence, 0, GL_TIMEOUT_IGNORED);
-            glDeleteSync(frame->present_fence);
-            frame->present_fence = 0;
-        }
-    }
-
-    {
-        MICROPROFILE_SCOPE(OpenGL_RenderFrame);
-        const auto& layout = render_window.GetFramebufferLayout();
-
-        // Recreate the frame if the size of the window has changed
-        if (layout.width != frame->width || layout.height != frame->height ||
-            screen_info.display_srgb != frame->is_srgb) {
-            LOG_DEBUG(Render_OpenGL, "Reloading render frame");
-            frame->is_srgb = screen_info.display_srgb;
-            frame_mailbox->ReloadRenderFrame(frame, layout.width, layout.height);
-        }
-        glBindFramebuffer(GL_DRAW_FRAMEBUFFER, frame->render.handle);
-        DrawScreen(layout);
-        // Create a fence for the frontend to wait on and swap this frame to OffTex
-        frame->render_fence = glFenceSync(GL_SYNC_GPU_COMMANDS_COMPLETE, 0);
-        glFlush();
-        frame_mailbox->ReleaseRenderFrame(frame);
-        m_current_frame++;
-        rasterizer->TickFrame();
-    }
+    rasterizer->TickFrame();
 
     render_window.PollEvents();
-    if (has_debug_tool) {
-        glBindFramebuffer(GL_DRAW_FRAMEBUFFER, 0);
-        Present(0);
-        context->SwapBuffers();
-    }
+    context->SwapBuffers();
 }
 
 void RendererOpenGL::PrepareRendertarget(const Tegra::FramebufferConfig* framebuffer) {
-    if (framebuffer) {
-        // If framebuffer is provided, reload it from memory to a texture
-        if (screen_info.texture.width != static_cast<GLsizei>(framebuffer->width) ||
-            screen_info.texture.height != static_cast<GLsizei>(framebuffer->height) ||
-            screen_info.texture.pixel_format != framebuffer->pixel_format ||
-            gl_framebuffer_data.empty()) {
-            // Reallocate texture if the framebuffer size has changed.
-            // This is expected to not happen very often and hence should not be a
-            // performance problem.
-            ConfigureFramebufferTexture(screen_info.texture, *framebuffer);
-        }
-
-        // Load the framebuffer from memory, draw it to the screen, and swap buffers
-        LoadFBToScreenInfo(*framebuffer);
+    if (!framebuffer) {
+        return;
     }
+    // If framebuffer is provided, reload it from memory to a texture
+    if (screen_info.texture.width != static_cast<GLsizei>(framebuffer->width) ||
+        screen_info.texture.height != static_cast<GLsizei>(framebuffer->height) ||
+        screen_info.texture.pixel_format != framebuffer->pixel_format ||
+        gl_framebuffer_data.empty()) {
+        // Reallocate texture if the framebuffer size has changed.
+        // This is expected to not happen very often and hence should not be a
+        // performance problem.
+        ConfigureFramebufferTexture(screen_info.texture, *framebuffer);
+    }
+
+    // Load the framebuffer from memory, draw it to the screen, and swap buffers
+    LoadFBToScreenInfo(*framebuffer);
 }
 
 void RendererOpenGL::LoadFBToScreenInfo(const Tegra::FramebufferConfig& framebuffer) {
@@ -418,8 +221,6 @@ void RendererOpenGL::LoadColorToActiveGLTexture(u8 color_r, u8 color_g, u8 color
 }
 
 void RendererOpenGL::InitOpenGLObjects() {
-    frame_mailbox = std::make_unique<FrameMailbox>();
-
     glClearColor(Settings::values.bg_red.GetValue(), Settings::values.bg_green.GetValue(),
                  Settings::values.bg_blue.GetValue(), 0.0f);
 
@@ -647,51 +448,6 @@ void RendererOpenGL::DrawScreen(const Layout::FramebufferLayout& layout) {
     program_manager.RestoreGuestPipeline();
 }
 
-bool RendererOpenGL::TryPresent(int timeout_ms) {
-    if (has_debug_tool) {
-        LOG_DEBUG(Render_OpenGL,
-                  "Skipping presentation because we are presenting on the main context");
-        return false;
-    }
-    return Present(timeout_ms);
-}
-
-bool RendererOpenGL::Present(int timeout_ms) {
-    const auto& layout = render_window.GetFramebufferLayout();
-    auto frame = frame_mailbox->TryGetPresentFrame(timeout_ms);
-    if (!frame) {
-        LOG_DEBUG(Render_OpenGL, "TryGetPresentFrame returned no frame to present");
-        return false;
-    }
-
-    // Clearing before a full overwrite of a fbo can signal to drivers that they can avoid a
-    // readback since we won't be doing any blending
-    glClear(GL_COLOR_BUFFER_BIT);
-
-    // Recreate the presentation FBO if the color attachment was changed
-    if (frame->color_reloaded) {
-        LOG_DEBUG(Render_OpenGL, "Reloading present frame");
-        frame_mailbox->ReloadPresentFrame(frame, layout.width, layout.height);
-    }
-    glWaitSync(frame->render_fence, 0, GL_TIMEOUT_IGNORED);
-    // INTEL workaround.
-    // Normally we could just delete the draw fence here, but due to driver bugs, we can just delete
-    // it on the emulation thread without too much penalty
-    // glDeleteSync(frame.render_sync);
-    // frame.render_sync = 0;
-
-    glBindFramebuffer(GL_READ_FRAMEBUFFER, frame->present.handle);
-    glBlitFramebuffer(0, 0, frame->width, frame->height, 0, 0, layout.width, layout.height,
-                      GL_COLOR_BUFFER_BIT, GL_LINEAR);
-
-    // Insert fence for the main thread to block on
-    frame->present_fence = glFenceSync(GL_SYNC_GPU_COMMANDS_COMPLETE, 0);
-    glFlush();
-
-    glBindFramebuffer(GL_READ_FRAMEBUFFER, 0);
-    return true;
-}
-
 void RendererOpenGL::RenderScreenshot() {
     if (!renderer_settings.screenshot_requested) {
         return;
@@ -706,7 +462,7 @@ void RendererOpenGL::RenderScreenshot() {
     screenshot_framebuffer.Create();
     glBindFramebuffer(GL_FRAMEBUFFER, screenshot_framebuffer.handle);
 
-    Layout::FramebufferLayout layout{renderer_settings.screenshot_framebuffer_layout};
+    const Layout::FramebufferLayout layout{renderer_settings.screenshot_framebuffer_layout};
 
     GLuint renderbuffer;
     glGenRenderbuffers(1, &renderbuffer);

src/video_core/renderer_opengl/renderer_opengl.h

@@ -55,14 +55,6 @@ struct ScreenInfo {
     TextureInfo texture;
 };
 
-struct PresentationTexture {
-    u32 width = 0;
-    u32 height = 0;
-    OGLTexture texture;
-};
-
-class FrameMailbox;
-
 class RendererOpenGL final : public VideoCore::RendererBase {
 public:
     explicit RendererOpenGL(Core::TelemetrySession& telemetry_session,
@@ -74,7 +66,6 @@ public:
     bool Init() override;
     void ShutDown() override;
     void SwapBuffers(const Tegra::FramebufferConfig* framebuffer) override;
-    bool TryPresent(int timeout_ms) override;
 
 private:
     /// Initializes the OpenGL state and creates persistent objects.
@@ -102,8 +93,6 @@ private:
 
     void PrepareRendertarget(const Tegra::FramebufferConfig* framebuffer);
 
-    bool Present(int timeout_ms);
-
     Core::TelemetrySession& telemetry_session;
     Core::Frontend::EmuWindow& emu_window;
     Core::Memory::Memory& cpu_memory;
@@ -134,11 +123,6 @@ private:
     /// Used for transforming the framebuffer orientation
     Tegra::FramebufferConfig::TransformFlags framebuffer_transform_flags{};
     Common::Rectangle<int> framebuffer_crop_rect;
-
-    /// Frame presentation mailbox
-    std::unique_ptr<FrameMailbox> frame_mailbox;
-
-    bool has_debug_tool = false;
 };
 
 } // namespace OpenGL

src/video_core/renderer_vulkan/renderer_vulkan.cpp

@@ -25,9 +25,9 @@
 #include "video_core/renderer_vulkan/renderer_vulkan.h"
 #include "video_core/renderer_vulkan/vk_blit_screen.h"
 #include "video_core/renderer_vulkan/vk_device.h"
+#include "video_core/renderer_vulkan/vk_master_semaphore.h"
 #include "video_core/renderer_vulkan/vk_memory_manager.h"
 #include "video_core/renderer_vulkan/vk_rasterizer.h"
-#include "video_core/renderer_vulkan/vk_resource_manager.h"
 #include "video_core/renderer_vulkan/vk_scheduler.h"
 #include "video_core/renderer_vulkan/vk_state_tracker.h"
 #include "video_core/renderer_vulkan/vk_swapchain.h"
@@ -56,7 +56,7 @@ VkBool32 DebugCallback(VkDebugUtilsMessageSeverityFlagBitsEXT severity,
                        VkDebugUtilsMessageTypeFlagsEXT type,
                        const VkDebugUtilsMessengerCallbackDataEXT* data,
                        [[maybe_unused]] void* user_data) {
-    const char* message{data->pMessage};
+    const char* const message{data->pMessage};
 
     if (severity & VK_DEBUG_UTILS_MESSAGE_SEVERITY_ERROR_BIT_EXT) {
         LOG_CRITICAL(Render_Vulkan, "{}", message);
@@ -269,11 +269,11 @@ void RendererVulkan::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
         scheduler->WaitWorker();
 
         swapchain->AcquireNextImage();
-        const auto [fence, render_semaphore] = blit_screen->Draw(*framebuffer, use_accelerated);
+        const VkSemaphore render_semaphore = blit_screen->Draw(*framebuffer, use_accelerated);
 
-        scheduler->Flush(false, render_semaphore);
+        scheduler->Flush(render_semaphore);
 
-        if (swapchain->Present(render_semaphore, fence)) {
+        if (swapchain->Present(render_semaphore)) {
             blit_screen->Recreate();
         }
 
@@ -283,11 +283,6 @@ void RendererVulkan::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
     render_window.PollEvents();
 }
 
-bool RendererVulkan::TryPresent(int /*timeout_ms*/) {
-    // TODO (bunnei): ImplementMe
-    return true;
-}
-
 bool RendererVulkan::Init() {
     library = OpenVulkanLibrary();
     instance = CreateInstance(library, dld, render_window.GetWindowInfo().type,
@@ -300,23 +295,21 @@ bool RendererVulkan::Init() {
 
     memory_manager = std::make_unique<VKMemoryManager>(*device);
 
-    resource_manager = std::make_unique<VKResourceManager>(*device);
-
-    const auto& framebuffer = render_window.GetFramebufferLayout();
-    swapchain = std::make_unique<VKSwapchain>(*surface, *device);
-    swapchain->Create(framebuffer.width, framebuffer.height, false);
-
     state_tracker = std::make_unique<StateTracker>(gpu);
 
-    scheduler = std::make_unique<VKScheduler>(*device, *resource_manager, *state_tracker);
+    scheduler = std::make_unique<VKScheduler>(*device, *state_tracker);
 
-    rasterizer = std::make_unique<RasterizerVulkan>(
-        render_window, gpu, gpu.MemoryManager(), cpu_memory, screen_info, *device,
-        *resource_manager, *memory_manager, *state_tracker, *scheduler);
+    const auto& framebuffer = render_window.GetFramebufferLayout();
+    swapchain = std::make_unique<VKSwapchain>(*surface, *device, *scheduler);
+    swapchain->Create(framebuffer.width, framebuffer.height, false);
 
-    blit_screen = std::make_unique<VKBlitScreen>(cpu_memory, render_window, *rasterizer, *device,
-                                                 *resource_manager, *memory_manager, *swapchain,
-                                                 *scheduler, screen_info);
+    rasterizer = std::make_unique<RasterizerVulkan>(render_window, gpu, gpu.MemoryManager(),
+                                                    cpu_memory, screen_info, *device,
+                                                    *memory_manager, *state_tracker, *scheduler);
+
+    blit_screen =
+        std::make_unique<VKBlitScreen>(cpu_memory, render_window, *rasterizer, *device,
+                                       *memory_manager, *swapchain, *scheduler, screen_info);
 
     return true;
 }
@@ -334,7 +327,6 @@ void RendererVulkan::ShutDown() {
     scheduler.reset();
     swapchain.reset();
     memory_manager.reset();
-    resource_manager.reset();
     device.reset();
 }
 

src/video_core/renderer_vulkan/renderer_vulkan.h

@@ -30,9 +30,7 @@ namespace Vulkan {
 class StateTracker;
 class VKBlitScreen;
 class VKDevice;
-class VKFence;
 class VKMemoryManager;
-class VKResourceManager;
 class VKSwapchain;
 class VKScheduler;
 class VKImage;
@@ -55,7 +53,6 @@ public:
     bool Init() override;
     void ShutDown() override;
     void SwapBuffers(const Tegra::FramebufferConfig* framebuffer) override;
-    bool TryPresent(int timeout_ms) override;
 
     static std::vector<std::string> EnumerateDevices();
 
@@ -82,11 +79,10 @@ private:
 
     vk::DebugCallback debug_callback;
     std::unique_ptr<VKDevice> device;
-    std::unique_ptr<VKSwapchain> swapchain;
     std::unique_ptr<VKMemoryManager> memory_manager;
-    std::unique_ptr<VKResourceManager> resource_manager;
     std::unique_ptr<StateTracker> state_tracker;
     std::unique_ptr<VKScheduler> scheduler;
+    std::unique_ptr<VKSwapchain> swapchain;
     std::unique_ptr<VKBlitScreen> blit_screen;
 };
 

src/video_core/renderer_vulkan/vk_blit_screen.cpp

@@ -12,11 +12,9 @@
 #include "common/assert.h"
 #include "common/common_types.h"
 #include "common/math_util.h"
-
 #include "core/core.h"
 #include "core/frontend/emu_window.h"
 #include "core/memory.h"
-
 #include "video_core/gpu.h"
 #include "video_core/morton.h"
 #include "video_core/rasterizer_interface.h"
@@ -24,8 +22,8 @@
 #include "video_core/renderer_vulkan/vk_blit_screen.h"
 #include "video_core/renderer_vulkan/vk_device.h"
 #include "video_core/renderer_vulkan/vk_image.h"
+#include "video_core/renderer_vulkan/vk_master_semaphore.h"
 #include "video_core/renderer_vulkan/vk_memory_manager.h"
-#include "video_core/renderer_vulkan/vk_resource_manager.h"
 #include "video_core/renderer_vulkan/vk_scheduler.h"
 #include "video_core/renderer_vulkan/vk_shader_util.h"
 #include "video_core/renderer_vulkan/vk_swapchain.h"
@@ -213,16 +211,12 @@ struct VKBlitScreen::BufferData {
 VKBlitScreen::VKBlitScreen(Core::Memory::Memory& cpu_memory_,
                            Core::Frontend::EmuWindow& render_window_,
                            VideoCore::RasterizerInterface& rasterizer_, const VKDevice& device_,
-                           VKResourceManager& resource_manager_, VKMemoryManager& memory_manager_,
-                           VKSwapchain& swapchain_, VKScheduler& scheduler_,
-                           const VKScreenInfo& screen_info_)
-    : cpu_memory{cpu_memory_}, render_window{render_window_},
-      rasterizer{rasterizer_}, device{device_}, resource_manager{resource_manager_},
-      memory_manager{memory_manager_}, swapchain{swapchain_}, scheduler{scheduler_},
-      image_count{swapchain.GetImageCount()}, screen_info{screen_info_} {
-    watches.resize(image_count);
-    std::generate(watches.begin(), watches.end(),
-                  []() { return std::make_unique<VKFenceWatch>(); });
+                           VKMemoryManager& memory_manager_, VKSwapchain& swapchain_,
+                           VKScheduler& scheduler_, const VKScreenInfo& screen_info_)
+    : cpu_memory{cpu_memory_}, render_window{render_window_}, rasterizer{rasterizer_},
+      device{device_}, memory_manager{memory_manager_}, swapchain{swapchain_},
+      scheduler{scheduler_}, image_count{swapchain.GetImageCount()}, screen_info{screen_info_} {
+    resource_ticks.resize(image_count);
 
     CreateStaticResources();
     CreateDynamicResources();
@@ -234,15 +228,16 @@ void VKBlitScreen::Recreate() {
     CreateDynamicResources();
 }
 
-std::tuple<VKFence&, VkSemaphore> VKBlitScreen::Draw(const Tegra::FramebufferConfig& framebuffer,
-                                                     bool use_accelerated) {
+VkSemaphore VKBlitScreen::Draw(const Tegra::FramebufferConfig& framebuffer, bool use_accelerated) {
     RefreshResources(framebuffer);
 
     // Finish any pending renderpass
     scheduler.RequestOutsideRenderPassOperationContext();
 
     const std::size_t image_index = swapchain.GetImageIndex();
-    watches[image_index]->Watch(scheduler.GetFence());
+
+    scheduler.Wait(resource_ticks[image_index]);
+    resource_ticks[image_index] = scheduler.CurrentTick();
 
     VKImage* blit_image = use_accelerated ? screen_info.image : raw_images[image_index].get();
 
@@ -345,7 +340,7 @@ std::tuple<VKFence&, VkSemaphore> VKBlitScreen::Draw(const Tegra::FramebufferCon
         cmdbuf.EndRenderPass();
     });
 
-    return {scheduler.GetFence(), *semaphores[image_index]};
+    return *semaphores[image_index];
 }
 
 void VKBlitScreen::CreateStaticResources() {
@@ -713,7 +708,7 @@ void VKBlitScreen::CreateFramebuffers() {
 
 void VKBlitScreen::ReleaseRawImages() {
     for (std::size_t i = 0; i < raw_images.size(); ++i) {
-        watches[i]->Wait();
+        scheduler.Wait(resource_ticks.at(i));
     }
     raw_images.clear();
     raw_buffer_commits.clear();

src/video_core/renderer_vulkan/vk_blit_screen.h

@@ -5,10 +5,8 @@
 #pragma once
 
 #include <memory>
-#include <tuple>
 
 #include "video_core/renderer_vulkan/vk_memory_manager.h"
-#include "video_core/renderer_vulkan/vk_resource_manager.h"
 #include "video_core/renderer_vulkan/wrapper.h"
 
 namespace Core {
@@ -34,9 +32,9 @@ class RasterizerInterface;
 namespace Vulkan {
 
 struct ScreenInfo;
+
 class RasterizerVulkan;
 class VKDevice;
-class VKFence;
 class VKImage;
 class VKScheduler;
 class VKSwapchain;
@@ -46,15 +44,14 @@ public:
     explicit VKBlitScreen(Core::Memory::Memory& cpu_memory,
                           Core::Frontend::EmuWindow& render_window,
                           VideoCore::RasterizerInterface& rasterizer, const VKDevice& device,
-                          VKResourceManager& resource_manager, VKMemoryManager& memory_manager,
-                          VKSwapchain& swapchain, VKScheduler& scheduler,
-                          const VKScreenInfo& screen_info);
+                          VKMemoryManager& memory_manager, VKSwapchain& swapchain,
+                          VKScheduler& scheduler, const VKScreenInfo& screen_info);
     ~VKBlitScreen();
 
     void Recreate();
 
-    std::tuple<VKFence&, VkSemaphore> Draw(const Tegra::FramebufferConfig& framebuffer,
-                                           bool use_accelerated);
+    [[nodiscard]] VkSemaphore Draw(const Tegra::FramebufferConfig& framebuffer,
+                                   bool use_accelerated);
 
 private:
     struct BufferData;
@@ -90,7 +87,6 @@ private:
     Core::Frontend::EmuWindow& render_window;
     VideoCore::RasterizerInterface& rasterizer;
     const VKDevice& device;
-    VKResourceManager& resource_manager;
     VKMemoryManager& memory_manager;
     VKSwapchain& swapchain;
     VKScheduler& scheduler;
@@ -111,7 +107,7 @@ private:
     vk::Buffer buffer;
     VKMemoryCommit buffer_commit;
 
-    std::vector<std::unique_ptr<VKFenceWatch>> watches;
+    std::vector<u64> resource_ticks;
 
     std::vector<vk::Semaphore> semaphores;
     std::vector<std::unique_ptr<VKImage>> raw_images;

src/video_core/renderer_vulkan/vk_command_pool.cpp

@@ -0,0 +1,46 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <cstddef>
+
+#include "video_core/renderer_vulkan/vk_command_pool.h"
+#include "video_core/renderer_vulkan/vk_device.h"
+#include "video_core/renderer_vulkan/wrapper.h"
+
+namespace Vulkan {
+
+constexpr size_t COMMAND_BUFFER_POOL_SIZE = 0x1000;
+
+struct CommandPool::Pool {
+    vk::CommandPool handle;
+    vk::CommandBuffers cmdbufs;
+};
+
+CommandPool::CommandPool(MasterSemaphore& master_semaphore, const VKDevice& device)
+    : ResourcePool(master_semaphore, COMMAND_BUFFER_POOL_SIZE), device{device} {}
+
+CommandPool::~CommandPool() = default;
+
+void CommandPool::Allocate(size_t begin, size_t end) {
+    // Command buffers are going to be commited, recorded, executed every single usage cycle.
+    // They are also going to be reseted when commited.
+    Pool& pool = pools.emplace_back();
+    pool.handle = device.GetLogical().CreateCommandPool({
+        .sType = VK_STRUCTURE_TYPE_COMMAND_POOL_CREATE_INFO,
+        .pNext = nullptr,
+        .flags =
+            VK_COMMAND_POOL_CREATE_TRANSIENT_BIT | VK_COMMAND_POOL_CREATE_RESET_COMMAND_BUFFER_BIT,
+        .queueFamilyIndex = device.GetGraphicsFamily(),
+    });
+    pool.cmdbufs = pool.handle.Allocate(COMMAND_BUFFER_POOL_SIZE);
+}
+
+VkCommandBuffer CommandPool::Commit() {
+    const size_t index = CommitResource();
+    const auto pool_index = index / COMMAND_BUFFER_POOL_SIZE;
+    const auto sub_index = index % COMMAND_BUFFER_POOL_SIZE;
+    return pools[pool_index].cmdbufs[sub_index];
+}
+
+} // namespace Vulkan

src/video_core/renderer_vulkan/vk_command_pool.h

@@ -0,0 +1,34 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <cstddef>
+#include <vector>
+
+#include "video_core/renderer_vulkan/vk_resource_pool.h"
+#include "video_core/renderer_vulkan/wrapper.h"
+
+namespace Vulkan {
+
+class MasterSemaphore;
+class VKDevice;
+
+class CommandPool final : public ResourcePool {
+public:
+    explicit CommandPool(MasterSemaphore& master_semaphore, const VKDevice& device);
+    ~CommandPool() override;
+
+    void Allocate(size_t begin, size_t end) override;
+
+    VkCommandBuffer Commit();
+
+private:
+    struct Pool;
+
+    const VKDevice& device;
+    std::vector<Pool> pools;
+};
+
+} // namespace Vulkan

src/video_core/renderer_vulkan/vk_compute_pass.cpp

@@ -112,7 +112,8 @@ constexpr u8 quad_array[] = {
     0xf9, 0x00, 0x02, 0x00, 0x21, 0x00, 0x00, 0x00, 0xf8, 0x00, 0x02, 0x00, 0x23, 0x00, 0x00, 0x00,
     0xf9, 0x00, 0x02, 0x00, 0x4b, 0x00, 0x00, 0x00, 0xf8, 0x00, 0x02, 0x00, 0x4e, 0x00, 0x00, 0x00,
     0xf9, 0x00, 0x02, 0x00, 0x4c, 0x00, 0x00, 0x00, 0xf8, 0x00, 0x02, 0x00, 0x4b, 0x00, 0x00, 0x00,
-    0xfd, 0x00, 0x01, 0x00, 0x38, 0x00, 0x01, 0x00};
+    0xfd, 0x00, 0x01, 0x00, 0x38, 0x00, 0x01, 0x00,
+};
 
 VkDescriptorSetLayoutBinding BuildQuadArrayPassDescriptorSetLayoutBinding() {
     return {
@@ -218,7 +219,8 @@ constexpr u8 uint8_pass[] = {
     0x2a, 0x00, 0x00, 0x00, 0x2b, 0x00, 0x00, 0x00, 0x22, 0x00, 0x00, 0x00, 0x23, 0x00, 0x00, 0x00,
     0x24, 0x00, 0x00, 0x00, 0x3e, 0x00, 0x03, 0x00, 0x2b, 0x00, 0x00, 0x00, 0x29, 0x00, 0x00, 0x00,
     0xf9, 0x00, 0x02, 0x00, 0x1d, 0x00, 0x00, 0x00, 0xf8, 0x00, 0x02, 0x00, 0x1d, 0x00, 0x00, 0x00,
-    0xfd, 0x00, 0x01, 0x00, 0x38, 0x00, 0x01, 0x00};
+    0xfd, 0x00, 0x01, 0x00, 0x38, 0x00, 0x01, 0x00,
+};
 
 // Quad indexed SPIR-V module. Generated from the "shaders/" directory.
 constexpr u8 QUAD_INDEXED_SPV[] = {
@@ -341,7 +343,8 @@ constexpr u8 QUAD_INDEXED_SPV[] = {
     0xf9, 0x00, 0x02, 0x00, 0x35, 0x00, 0x00, 0x00, 0xf8, 0x00, 0x02, 0x00, 0x37, 0x00, 0x00, 0x00,
     0xf9, 0x00, 0x02, 0x00, 0x73, 0x00, 0x00, 0x00, 0xf8, 0x00, 0x02, 0x00, 0x76, 0x00, 0x00, 0x00,
     0xf9, 0x00, 0x02, 0x00, 0x74, 0x00, 0x00, 0x00, 0xf8, 0x00, 0x02, 0x00, 0x73, 0x00, 0x00, 0x00,
-    0xfd, 0x00, 0x01, 0x00, 0x38, 0x00, 0x01, 0x00};
+    0xfd, 0x00, 0x01, 0x00, 0x38, 0x00, 0x01, 0x00,
+};
 
 std::array<VkDescriptorSetLayoutBinding, 2> BuildInputOutputDescriptorSetBindings() {
     return {{
@@ -448,12 +451,12 @@ VKComputePass::VKComputePass(const VKDevice& device, VKDescriptorPool& descripto
 
 VKComputePass::~VKComputePass() = default;
 
-VkDescriptorSet VKComputePass::CommitDescriptorSet(VKUpdateDescriptorQueue& update_descriptor_queue,
-                                                   VKFence& fence) {
+VkDescriptorSet VKComputePass::CommitDescriptorSet(
+    VKUpdateDescriptorQueue& update_descriptor_queue) {
     if (!descriptor_template) {
         return nullptr;
     }
-    const auto set = descriptor_allocator->Commit(fence);
+    const VkDescriptorSet set = descriptor_allocator->Commit();
     update_descriptor_queue.Send(*descriptor_template, set);
     return set;
 }
@@ -477,7 +480,7 @@ std::pair<VkBuffer, VkDeviceSize> QuadArrayPass::Assemble(u32 num_vertices, u32
 
     update_descriptor_queue.Acquire();
     update_descriptor_queue.AddBuffer(*buffer.handle, 0, staging_size);
-    const auto set = CommitDescriptorSet(update_descriptor_queue, scheduler.GetFence());
+    const VkDescriptorSet set = CommitDescriptorSet(update_descriptor_queue);
 
     scheduler.RequestOutsideRenderPassOperationContext();
 
@@ -520,13 +523,13 @@ Uint8Pass::~Uint8Pass() = default;
 
 std::pair<VkBuffer, u64> Uint8Pass::Assemble(u32 num_vertices, VkBuffer src_buffer,
                                              u64 src_offset) {
-    const auto staging_size = static_cast<u32>(num_vertices * sizeof(u16));
+    const u32 staging_size = static_cast<u32>(num_vertices * sizeof(u16));
     auto& buffer = staging_buffer_pool.GetUnusedBuffer(staging_size, false);
 
     update_descriptor_queue.Acquire();
     update_descriptor_queue.AddBuffer(src_buffer, src_offset, num_vertices);
     update_descriptor_queue.AddBuffer(*buffer.handle, 0, staging_size);
-    const auto set = CommitDescriptorSet(update_descriptor_queue, scheduler.GetFence());
+    const VkDescriptorSet set = CommitDescriptorSet(update_descriptor_queue);
 
     scheduler.RequestOutsideRenderPassOperationContext();
     scheduler.Record([layout = *layout, pipeline = *pipeline, buffer = *buffer.handle, set,
@@ -589,7 +592,7 @@ std::pair<VkBuffer, u64> QuadIndexedPass::Assemble(
     update_descriptor_queue.Acquire();
     update_descriptor_queue.AddBuffer(src_buffer, src_offset, input_size);
     update_descriptor_queue.AddBuffer(*buffer.handle, 0, staging_size);
-    const auto set = CommitDescriptorSet(update_descriptor_queue, scheduler.GetFence());
+    const VkDescriptorSet set = CommitDescriptorSet(update_descriptor_queue);
 
     scheduler.RequestOutsideRenderPassOperationContext();
     scheduler.Record([layout = *layout, pipeline = *pipeline, buffer = *buffer.handle, set,

src/video_core/renderer_vulkan/vk_compute_pass.h

@@ -15,7 +15,6 @@
 namespace Vulkan {
 
 class VKDevice;
-class VKFence;
 class VKScheduler;
 class VKStagingBufferPool;
 class VKUpdateDescriptorQueue;
@@ -30,8 +29,7 @@ public:
     ~VKComputePass();
 
 protected:
-    VkDescriptorSet CommitDescriptorSet(VKUpdateDescriptorQueue& update_descriptor_queue,
-                                        VKFence& fence);
+    VkDescriptorSet CommitDescriptorSet(VKUpdateDescriptorQueue& update_descriptor_queue);
 
     vk::DescriptorUpdateTemplateKHR descriptor_template;
     vk::PipelineLayout layout;

src/video_core/renderer_vulkan/vk_compute_pipeline.cpp

@@ -32,7 +32,7 @@ VkDescriptorSet VKComputePipeline::CommitDescriptorSet() {
     if (!descriptor_template) {
         return {};
     }
-    const auto set = descriptor_allocator.Commit(scheduler.GetFence());
+    const VkDescriptorSet set = descriptor_allocator.Commit();
     update_descriptor_queue.Send(*descriptor_template, set);
     return set;
 }

src/video_core/renderer_vulkan/vk_descriptor_pool.cpp

@@ -7,7 +7,8 @@
 #include "common/common_types.h"
 #include "video_core/renderer_vulkan/vk_descriptor_pool.h"
 #include "video_core/renderer_vulkan/vk_device.h"
-#include "video_core/renderer_vulkan/vk_resource_manager.h"
+#include "video_core/renderer_vulkan/vk_resource_pool.h"
+#include "video_core/renderer_vulkan/vk_scheduler.h"
 #include "video_core/renderer_vulkan/wrapper.h"
 
 namespace Vulkan {
@@ -15,14 +16,15 @@ namespace Vulkan {
 // Prefer small grow rates to avoid saturating the descriptor pool with barely used pipelines.
 constexpr std::size_t SETS_GROW_RATE = 0x20;
 
-DescriptorAllocator::DescriptorAllocator(VKDescriptorPool& descriptor_pool,
-                                         VkDescriptorSetLayout layout)
-    : VKFencedPool{SETS_GROW_RATE}, descriptor_pool{descriptor_pool}, layout{layout} {}
+DescriptorAllocator::DescriptorAllocator(VKDescriptorPool& descriptor_pool_,
+                                         VkDescriptorSetLayout layout_)
+    : ResourcePool(descriptor_pool_.master_semaphore, SETS_GROW_RATE),
+      descriptor_pool{descriptor_pool_}, layout{layout_} {}
 
 DescriptorAllocator::~DescriptorAllocator() = default;
 
-VkDescriptorSet DescriptorAllocator::Commit(VKFence& fence) {
-    const std::size_t index = CommitResource(fence);
+VkDescriptorSet DescriptorAllocator::Commit() {
+    const std::size_t index = CommitResource();
     return descriptors_allocations[index / SETS_GROW_RATE][index % SETS_GROW_RATE];
 }
 
@@ -30,8 +32,9 @@ void DescriptorAllocator::Allocate(std::size_t begin, std::size_t end) {
     descriptors_allocations.push_back(descriptor_pool.AllocateDescriptors(layout, end - begin));
 }
 
-VKDescriptorPool::VKDescriptorPool(const VKDevice& device)
-    : device{device}, active_pool{AllocateNewPool()} {}
+VKDescriptorPool::VKDescriptorPool(const VKDevice& device_, VKScheduler& scheduler)
+    : device{device_}, master_semaphore{scheduler.GetMasterSemaphore()}, active_pool{
+                                                                             AllocateNewPool()} {}
 
 VKDescriptorPool::~VKDescriptorPool() = default;
 

src/video_core/renderer_vulkan/vk_descriptor_pool.h

@@ -6,21 +6,24 @@
 
 #include <vector>
 
-#include "video_core/renderer_vulkan/vk_resource_manager.h"
+#include "video_core/renderer_vulkan/vk_resource_pool.h"
 #include "video_core/renderer_vulkan/wrapper.h"
 
 namespace Vulkan {
 
+class VKDevice;
 class VKDescriptorPool;
+class VKScheduler;
 
-class DescriptorAllocator final : public VKFencedPool {
+class DescriptorAllocator final : public ResourcePool {
 public:
     explicit DescriptorAllocator(VKDescriptorPool& descriptor_pool, VkDescriptorSetLayout layout);
     ~DescriptorAllocator() override;
 
+    DescriptorAllocator& operator=(const DescriptorAllocator&) = delete;
     DescriptorAllocator(const DescriptorAllocator&) = delete;
 
-    VkDescriptorSet Commit(VKFence& fence);
+    VkDescriptorSet Commit();
 
 protected:
     void Allocate(std::size_t begin, std::size_t end) override;
@@ -36,15 +39,19 @@ class VKDescriptorPool final {
     friend DescriptorAllocator;
 
 public:
-    explicit VKDescriptorPool(const VKDevice& device);
+    explicit VKDescriptorPool(const VKDevice& device, VKScheduler& scheduler);
     ~VKDescriptorPool();
 
+    VKDescriptorPool(const VKDescriptorPool&) = delete;
+    VKDescriptorPool& operator=(const VKDescriptorPool&) = delete;
+
 private:
     vk::DescriptorPool* AllocateNewPool();
 
     vk::DescriptorSets AllocateDescriptors(VkDescriptorSetLayout layout, std::size_t count);
 
     const VKDevice& device;
+    MasterSemaphore& master_semaphore;
 
     std::vector<vk::DescriptorPool> pools;
     vk::DescriptorPool* active_pool;

src/video_core/renderer_vulkan/vk_device.cpp

@@ -42,6 +42,7 @@ constexpr std::array REQUIRED_EXTENSIONS{
     VK_KHR_8BIT_STORAGE_EXTENSION_NAME,
     VK_KHR_DRIVER_PROPERTIES_EXTENSION_NAME,
     VK_KHR_DESCRIPTOR_UPDATE_TEMPLATE_EXTENSION_NAME,
+    VK_KHR_TIMELINE_SEMAPHORE_EXTENSION_NAME,
     VK_EXT_VERTEX_ATTRIBUTE_DIVISOR_EXTENSION_NAME,
     VK_EXT_SHADER_SUBGROUP_BALLOT_EXTENSION_NAME,
     VK_EXT_SHADER_SUBGROUP_VOTE_EXTENSION_NAME,
@@ -250,6 +251,13 @@ bool VKDevice::Create() {
         .inheritedQueries = false,
     };
 
+    VkPhysicalDeviceTimelineSemaphoreFeaturesKHR timeline_semaphore{
+        .sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_TIMELINE_SEMAPHORE_FEATURES_KHR,
+        .pNext = nullptr,
+        .timelineSemaphore = true,
+    };
+    SetNext(next, timeline_semaphore);
+
     VkPhysicalDevice16BitStorageFeaturesKHR bit16_storage{
         .sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_16BIT_STORAGE_FEATURES_KHR,
         .pNext = nullptr,
@@ -380,14 +388,6 @@ bool VKDevice::Create() {
 
     CollectTelemetryParameters();
 
-    if (ext_extended_dynamic_state && driver_id == VK_DRIVER_ID_AMD_PROPRIETARY_KHR) {
-        // AMD's proprietary driver supports VK_EXT_extended_dynamic_state but the <stride> field
-        // seems to be bugged. Blacklisting it for now.
-        LOG_WARNING(Render_Vulkan,
-                    "Blacklisting AMD proprietary from VK_EXT_extended_dynamic_state");
-        ext_extended_dynamic_state = false;
-    }
-
     graphics_queue = logical.GetQueue(graphics_family);
     present_queue = logical.GetQueue(present_family);
 

src/video_core/renderer_vulkan/vk_fence_manager.cpp

@@ -29,8 +29,8 @@ void InnerFence::Queue() {
     }
     ASSERT(!event);
 
-    event = device.GetLogical().CreateNewEvent();
-    ticks = scheduler.Ticks();
+    event = device.GetLogical().CreateEvent();
+    ticks = scheduler.CurrentTick();
 
     scheduler.RequestOutsideRenderPassOperationContext();
     scheduler.Record([event = *event](vk::CommandBuffer cmdbuf) {
@@ -52,7 +52,7 @@ void InnerFence::Wait() {
     }
     ASSERT(event);
 
-    if (ticks >= scheduler.Ticks()) {
+    if (ticks >= scheduler.CurrentTick()) {
         scheduler.Flush();
     }
     while (!IsEventSignalled()) {

src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp

@@ -93,7 +93,7 @@ VkDescriptorSet VKGraphicsPipeline::CommitDescriptorSet() {
     if (!descriptor_template) {
         return {};
     }
-    const auto set = descriptor_allocator.Commit(scheduler.GetFence());
+    const VkDescriptorSet set = descriptor_allocator.Commit();
     update_descriptor_queue.Send(*descriptor_template, set);
     return set;
 }

src/video_core/renderer_vulkan/vk_master_semaphore.cpp

@@ -0,0 +1,56 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <atomic>
+#include <chrono>
+
+#include "core/settings.h"
+#include "video_core/renderer_vulkan/vk_device.h"
+#include "video_core/renderer_vulkan/vk_master_semaphore.h"
+#include "video_core/renderer_vulkan/wrapper.h"
+
+namespace Vulkan {
+
+using namespace std::chrono_literals;
+
+MasterSemaphore::MasterSemaphore(const VKDevice& device) {
+    static constexpr VkSemaphoreTypeCreateInfoKHR semaphore_type_ci{
+        .sType = VK_STRUCTURE_TYPE_SEMAPHORE_TYPE_CREATE_INFO_KHR,
+        .pNext = nullptr,
+        .semaphoreType = VK_SEMAPHORE_TYPE_TIMELINE_KHR,
+        .initialValue = 0,
+    };
+    static constexpr VkSemaphoreCreateInfo semaphore_ci{
+        .sType = VK_STRUCTURE_TYPE_SEMAPHORE_CREATE_INFO,
+        .pNext = &semaphore_type_ci,
+        .flags = 0,
+    };
+    semaphore = device.GetLogical().CreateSemaphore(semaphore_ci);
+
+    if (!Settings::values.renderer_debug) {
+        return;
+    }
+    // Validation layers have a bug where they fail to track resource usage when using timeline
+    // semaphores and synchronizing with GetSemaphoreCounterValueKHR. To workaround this issue, have
+    // a separate thread waiting for each timeline semaphore value.
+    debug_thread = std::thread([this] {
+        u64 counter = 0;
+        while (!shutdown) {
+            if (semaphore.Wait(counter, 10'000'000)) {
+                ++counter;
+            }
+        }
+    });
+}
+
+MasterSemaphore::~MasterSemaphore() {
+    shutdown = true;
+
+    // This thread might not be started
+    if (debug_thread.joinable()) {
+        debug_thread.join();
+    }
+}
+
+} // namespace Vulkan

src/video_core/renderer_vulkan/vk_master_semaphore.h

@@ -0,0 +1,70 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <atomic>
+#include <thread>
+
+#include "common/common_types.h"
+#include "video_core/renderer_vulkan/wrapper.h"
+
+namespace Vulkan {
+
+class VKDevice;
+
+class MasterSemaphore {
+public:
+    explicit MasterSemaphore(const VKDevice& device);
+    ~MasterSemaphore();
+
+    /// Returns the current logical tick.
+    [[nodiscard]] u64 CurrentTick() const noexcept {
+        return current_tick;
+    }
+
+    /// Returns the timeline semaphore handle.
+    [[nodiscard]] VkSemaphore Handle() const noexcept {
+        return *semaphore;
+    }
+
+    /// Returns true when a tick has been hit by the GPU.
+    [[nodiscard]] bool IsFree(u64 tick) {
+        return gpu_tick >= tick;
+    }
+
+    /// Advance to the logical tick.
+    void NextTick() noexcept {
+        ++current_tick;
+    }
+
+    /// Refresh the known GPU tick
+    void Refresh() {
+        gpu_tick = semaphore.GetCounter();
+    }
+
+    /// Waits for a tick to be hit on the GPU
+    void Wait(u64 tick) {
+        // No need to wait if the GPU is ahead of the tick
+        if (IsFree(tick)) {
+            return;
+        }
+        // Update the GPU tick and try again
+        Refresh();
+        if (IsFree(tick)) {
+            return;
+        }
+        // If none of the above is hit, fallback to a regular wait
+        semaphore.Wait(tick);
+    }
+
+private:
+    vk::Semaphore semaphore;           ///< Timeline semaphore.
+    std::atomic<u64> gpu_tick{0};      ///< Current known GPU tick.
+    std::atomic<u64> current_tick{1};  ///< Current logical tick.
+    std::atomic<bool> shutdown{false}; ///< True when the object is being destroyed.
+    std::thread debug_thread;          ///< Debug thread to workaround validation layer bugs.
+};
+
+} // namespace Vulkan

src/video_core/renderer_vulkan/vk_pipeline_cache.h

@@ -38,7 +38,6 @@ class RasterizerVulkan;
 class VKComputePipeline;
 class VKDescriptorPool;
 class VKDevice;
-class VKFence;
 class VKScheduler;
 class VKUpdateDescriptorQueue;
 

src/video_core/renderer_vulkan/vk_query_cache.cpp

@@ -9,35 +9,33 @@
 
 #include "video_core/renderer_vulkan/vk_device.h"
 #include "video_core/renderer_vulkan/vk_query_cache.h"
-#include "video_core/renderer_vulkan/vk_resource_manager.h"
+#include "video_core/renderer_vulkan/vk_resource_pool.h"
 #include "video_core/renderer_vulkan/vk_scheduler.h"
 #include "video_core/renderer_vulkan/wrapper.h"
 
 namespace Vulkan {
 
+using VideoCore::QueryType;
+
 namespace {
 
 constexpr std::array QUERY_TARGETS = {VK_QUERY_TYPE_OCCLUSION};
 
-constexpr VkQueryType GetTarget(VideoCore::QueryType type) {
+constexpr VkQueryType GetTarget(QueryType type) {
     return QUERY_TARGETS[static_cast<std::size_t>(type)];
 }
 
 } // Anonymous namespace
 
-QueryPool::QueryPool() : VKFencedPool{GROW_STEP} {}
+QueryPool::QueryPool(const VKDevice& device_, VKScheduler& scheduler, QueryType type_)
+    : ResourcePool{scheduler.GetMasterSemaphore(), GROW_STEP}, device{device_}, type{type_} {}
 
 QueryPool::~QueryPool() = default;
 
-void QueryPool::Initialize(const VKDevice& device_, VideoCore::QueryType type_) {
-    device = &device_;
-    type = type_;
-}
-
-std::pair<VkQueryPool, u32> QueryPool::Commit(VKFence& fence) {
+std::pair<VkQueryPool, u32> QueryPool::Commit() {
     std::size_t index;
     do {
-        index = CommitResource(fence);
+        index = CommitResource();
     } while (usage[index]);
     usage[index] = true;
 
@@ -47,7 +45,7 @@ std::pair<VkQueryPool, u32> QueryPool::Commit(VKFence& fence) {
 void QueryPool::Allocate(std::size_t begin, std::size_t end) {
     usage.resize(end);
 
-    pools.push_back(device->GetLogical().CreateQueryPool({
+    pools.push_back(device.GetLogical().CreateQueryPool({
         .sType = VK_STRUCTURE_TYPE_QUERY_POOL_CREATE_INFO,
         .pNext = nullptr,
         .flags = 0,
@@ -71,28 +69,36 @@ void QueryPool::Reserve(std::pair<VkQueryPool, u32> query) {
 VKQueryCache::VKQueryCache(VideoCore::RasterizerInterface& rasterizer,
                            Tegra::Engines::Maxwell3D& maxwell3d, Tegra::MemoryManager& gpu_memory,
                            const VKDevice& device, VKScheduler& scheduler)
-    : VideoCommon::QueryCacheBase<VKQueryCache, CachedQuery, CounterStream, HostCounter,
-                                  QueryPool>{rasterizer, maxwell3d, gpu_memory},
-      device{device}, scheduler{scheduler} {
-    for (std::size_t i = 0; i < static_cast<std::size_t>(VideoCore::NumQueryTypes); ++i) {
-        query_pools[i].Initialize(device, static_cast<VideoCore::QueryType>(i));
+    : VideoCommon::QueryCacheBase<VKQueryCache, CachedQuery, CounterStream,
+                                  HostCounter>{rasterizer, maxwell3d, gpu_memory},
+      device{device}, scheduler{scheduler}, query_pools{
+                                                QueryPool{device, scheduler,
+                                                          QueryType::SamplesPassed},
+                                            } {}
+
+VKQueryCache::~VKQueryCache() {
+    // TODO(Rodrigo): This is a hack to destroy all HostCounter instances before the base class
+    // destructor is called. The query cache should be redesigned to have a proper ownership model
+    // instead of using shared pointers.
+    for (size_t query_type = 0; query_type < VideoCore::NumQueryTypes; ++query_type) {
+        auto& stream = Stream(static_cast<QueryType>(query_type));
+        stream.Update(false);
+        stream.Reset();
     }
 }
 
-VKQueryCache::~VKQueryCache() = default;
-
-std::pair<VkQueryPool, u32> VKQueryCache::AllocateQuery(VideoCore::QueryType type) {
-    return query_pools[static_cast<std::size_t>(type)].Commit(scheduler.GetFence());
+std::pair<VkQueryPool, u32> VKQueryCache::AllocateQuery(QueryType type) {
+    return query_pools[static_cast<std::size_t>(type)].Commit();
 }
 
-void VKQueryCache::Reserve(VideoCore::QueryType type, std::pair<VkQueryPool, u32> query) {
+void VKQueryCache::Reserve(QueryType type, std::pair<VkQueryPool, u32> query) {
     query_pools[static_cast<std::size_t>(type)].Reserve(query);
 }
 
 HostCounter::HostCounter(VKQueryCache& cache, std::shared_ptr<HostCounter> dependency,
-                         VideoCore::QueryType type)
+                         QueryType type)
     : VideoCommon::HostCounterBase<VKQueryCache, HostCounter>{std::move(dependency)}, cache{cache},
-      type{type}, query{cache.AllocateQuery(type)}, ticks{cache.Scheduler().Ticks()} {
+      type{type}, query{cache.AllocateQuery(type)}, tick{cache.Scheduler().CurrentTick()} {
     const vk::Device* logical = &cache.Device().GetLogical();
     cache.Scheduler().Record([logical, query = query](vk::CommandBuffer cmdbuf) {
         logical->ResetQueryPoolEXT(query.first, query.second, 1);
@@ -110,7 +116,7 @@ void HostCounter::EndQuery() {
 }
 
 u64 HostCounter::BlockingQuery() const {
-    if (ticks >= cache.Scheduler().Ticks()) {
+    if (tick >= cache.Scheduler().CurrentTick()) {
         cache.Scheduler().Flush();
     }
     u64 data;

src/video_core/renderer_vulkan/vk_query_cache.h

@@ -11,7 +11,7 @@
 
 #include "common/common_types.h"
 #include "video_core/query_cache.h"
-#include "video_core/renderer_vulkan/vk_resource_manager.h"
+#include "video_core/renderer_vulkan/vk_resource_pool.h"
 #include "video_core/renderer_vulkan/wrapper.h"
 
 namespace VideoCore {
@@ -28,14 +28,12 @@ class VKScheduler;
 
 using CounterStream = VideoCommon::CounterStreamBase<VKQueryCache, HostCounter>;
 
-class QueryPool final : public VKFencedPool {
+class QueryPool final : public ResourcePool {
 public:
-    explicit QueryPool();
+    explicit QueryPool(const VKDevice& device, VKScheduler& scheduler, VideoCore::QueryType type);
     ~QueryPool() override;
 
-    void Initialize(const VKDevice& device, VideoCore::QueryType type);
-
-    std::pair<VkQueryPool, u32> Commit(VKFence& fence);
+    std::pair<VkQueryPool, u32> Commit();
 
     void Reserve(std::pair<VkQueryPool, u32> query);
 
@@ -45,16 +43,15 @@ protected:
 private:
     static constexpr std::size_t GROW_STEP = 512;
 
-    const VKDevice* device = nullptr;
-    VideoCore::QueryType type = {};
+    const VKDevice& device;
+    const VideoCore::QueryType type;
 
     std::vector<vk::QueryPool> pools;
     std::vector<bool> usage;
 };
 
 class VKQueryCache final
-    : public VideoCommon::QueryCacheBase<VKQueryCache, CachedQuery, CounterStream, HostCounter,
-                                         QueryPool> {
+    : public VideoCommon::QueryCacheBase<VKQueryCache, CachedQuery, CounterStream, HostCounter> {
 public:
     explicit VKQueryCache(VideoCore::RasterizerInterface& rasterizer,
                           Tegra::Engines::Maxwell3D& maxwell3d, Tegra::MemoryManager& gpu_memory,
@@ -76,6 +73,7 @@ public:
 private:
     const VKDevice& device;
     VKScheduler& scheduler;
+    std::array<QueryPool, VideoCore::NumQueryTypes> query_pools;
 };
 
 class HostCounter final : public VideoCommon::HostCounterBase<VKQueryCache, HostCounter> {
@@ -92,7 +90,7 @@ private:
     VKQueryCache& cache;
     const VideoCore::QueryType type;
     const std::pair<VkQueryPool, u32> query;
-    const u64 ticks;
+    const u64 tick;
 };
 
 class CachedQuery : public VideoCommon::CachedQueryBase<HostCounter> {

src/video_core/renderer_vulkan/vk_rasterizer.cpp

@@ -31,7 +31,6 @@
 #include "video_core/renderer_vulkan/vk_pipeline_cache.h"
 #include "video_core/renderer_vulkan/vk_rasterizer.h"
 #include "video_core/renderer_vulkan/vk_renderpass_cache.h"
-#include "video_core/renderer_vulkan/vk_resource_manager.h"
 #include "video_core/renderer_vulkan/vk_sampler_cache.h"
 #include "video_core/renderer_vulkan/vk_scheduler.h"
 #include "video_core/renderer_vulkan/vk_staging_buffer_pool.h"
@@ -384,27 +383,25 @@ void RasterizerVulkan::DrawParameters::Draw(vk::CommandBuffer cmdbuf) const {
 RasterizerVulkan::RasterizerVulkan(Core::Frontend::EmuWindow& emu_window, Tegra::GPU& gpu_,
                                    Tegra::MemoryManager& gpu_memory_,
                                    Core::Memory::Memory& cpu_memory, VKScreenInfo& screen_info_,
-                                   const VKDevice& device_, VKResourceManager& resource_manager_,
-                                   VKMemoryManager& memory_manager_, StateTracker& state_tracker_,
-                                   VKScheduler& scheduler_)
+                                   const VKDevice& device_, VKMemoryManager& memory_manager_,
+                                   StateTracker& state_tracker_, VKScheduler& scheduler_)
     : RasterizerAccelerated(cpu_memory), gpu(gpu_), gpu_memory(gpu_memory_),
       maxwell3d(gpu.Maxwell3D()), kepler_compute(gpu.KeplerCompute()), screen_info(screen_info_),
-      device(device_), resource_manager(resource_manager_), memory_manager(memory_manager_),
-      state_tracker(state_tracker_), scheduler(scheduler_),
-      staging_pool(device, memory_manager, scheduler), descriptor_pool(device),
-      update_descriptor_queue(device, scheduler), renderpass_cache(device),
+      device(device_), memory_manager(memory_manager_), state_tracker(state_tracker_),
+      scheduler(scheduler_), staging_pool(device, memory_manager, scheduler),
+      descriptor_pool(device, scheduler_), update_descriptor_queue(device, scheduler),
+      renderpass_cache(device),
       quad_array_pass(device, scheduler, descriptor_pool, staging_pool, update_descriptor_queue),
       quad_indexed_pass(device, scheduler, descriptor_pool, staging_pool, update_descriptor_queue),
       uint8_pass(device, scheduler, descriptor_pool, staging_pool, update_descriptor_queue),
-      texture_cache(*this, maxwell3d, gpu_memory, device, resource_manager, memory_manager,
-                    scheduler, staging_pool),
+      texture_cache(*this, maxwell3d, gpu_memory, device, memory_manager, scheduler, staging_pool),
       pipeline_cache(*this, gpu, maxwell3d, kepler_compute, gpu_memory, device, scheduler,
                      descriptor_pool, update_descriptor_queue, renderpass_cache),
       buffer_cache(*this, gpu_memory, cpu_memory, device, memory_manager, scheduler, staging_pool),
       sampler_cache(device), query_cache(*this, maxwell3d, gpu_memory, device, scheduler),
       fence_manager(*this, gpu, gpu_memory, texture_cache, buffer_cache, query_cache, device,
                     scheduler),
-      wfi_event(device.GetLogical().CreateNewEvent()), async_shaders(emu_window) {
+      wfi_event(device.GetLogical().CreateEvent()), async_shaders(emu_window) {
     scheduler.SetQueryCache(query_cache);
     if (device.UseAsynchronousShaders()) {
         async_shaders.AllocateWorkers();

src/video_core/renderer_vulkan/vk_rasterizer.h

@@ -25,7 +25,6 @@
 #include "video_core/renderer_vulkan/vk_pipeline_cache.h"
 #include "video_core/renderer_vulkan/vk_query_cache.h"
 #include "video_core/renderer_vulkan/vk_renderpass_cache.h"
-#include "video_core/renderer_vulkan/vk_resource_manager.h"
 #include "video_core/renderer_vulkan/vk_sampler_cache.h"
 #include "video_core/renderer_vulkan/vk_scheduler.h"
 #include "video_core/renderer_vulkan/vk_staging_buffer_pool.h"
@@ -109,8 +108,8 @@ public:
     explicit RasterizerVulkan(Core::Frontend::EmuWindow& emu_window, Tegra::GPU& gpu,
                               Tegra::MemoryManager& gpu_memory, Core::Memory::Memory& cpu_memory,
                               VKScreenInfo& screen_info, const VKDevice& device,
-                              VKResourceManager& resource_manager, VKMemoryManager& memory_manager,
-                              StateTracker& state_tracker, VKScheduler& scheduler);
+                              VKMemoryManager& memory_manager, StateTracker& state_tracker,
+                              VKScheduler& scheduler);
     ~RasterizerVulkan() override;
 
     void Draw(bool is_indexed, bool is_instanced) override;
@@ -286,7 +285,6 @@ private:
 
     VKScreenInfo& screen_info;
     const VKDevice& device;
-    VKResourceManager& resource_manager;
     VKMemoryManager& memory_manager;
     StateTracker& state_tracker;
     VKScheduler& scheduler;

src/video_core/renderer_vulkan/vk_resource_manager.cpp

@@ -1,311 +0,0 @@
-// Copyright 2018 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include <algorithm>
-#include <optional>
-#include "common/assert.h"
-#include "common/logging/log.h"
-#include "video_core/renderer_vulkan/vk_device.h"
-#include "video_core/renderer_vulkan/vk_resource_manager.h"
-#include "video_core/renderer_vulkan/wrapper.h"
-
-namespace Vulkan {
-
-namespace {
-
-// TODO(Rodrigo): Fine tune these numbers.
-constexpr std::size_t COMMAND_BUFFER_POOL_SIZE = 0x1000;
-constexpr std::size_t FENCES_GROW_STEP = 0x40;
-
-constexpr VkFenceCreateInfo BuildFenceCreateInfo() {
-    return {
-        .sType = VK_STRUCTURE_TYPE_FENCE_CREATE_INFO,
-        .pNext = nullptr,
-        .flags = 0,
-    };
-}
-
-} // Anonymous namespace
-
-class CommandBufferPool final : public VKFencedPool {
-public:
-    explicit CommandBufferPool(const VKDevice& device)
-        : VKFencedPool(COMMAND_BUFFER_POOL_SIZE), device{device} {}
-
-    void Allocate(std::size_t begin, std::size_t end) override {
-        // Command buffers are going to be commited, recorded, executed every single usage cycle.
-        // They are also going to be reseted when commited.
-        Pool& pool = pools.emplace_back();
-        pool.handle = device.GetLogical().CreateCommandPool({
-            .sType = VK_STRUCTURE_TYPE_COMMAND_POOL_CREATE_INFO,
-            .pNext = nullptr,
-            .flags = VK_COMMAND_POOL_CREATE_TRANSIENT_BIT |
-                     VK_COMMAND_POOL_CREATE_RESET_COMMAND_BUFFER_BIT,
-            .queueFamilyIndex = device.GetGraphicsFamily(),
-        });
-        pool.cmdbufs = pool.handle.Allocate(COMMAND_BUFFER_POOL_SIZE);
-    }
-
-    VkCommandBuffer Commit(VKFence& fence) {
-        const std::size_t index = CommitResource(fence);
-        const auto pool_index = index / COMMAND_BUFFER_POOL_SIZE;
-        const auto sub_index = index % COMMAND_BUFFER_POOL_SIZE;
-        return pools[pool_index].cmdbufs[sub_index];
-    }
-
-private:
-    struct Pool {
-        vk::CommandPool handle;
-        vk::CommandBuffers cmdbufs;
-    };
-
-    const VKDevice& device;
-    std::vector<Pool> pools;
-};
-
-VKResource::VKResource() = default;
-
-VKResource::~VKResource() = default;
-
-VKFence::VKFence(const VKDevice& device)
-    : device{device}, handle{device.GetLogical().CreateFence(BuildFenceCreateInfo())} {}
-
-VKFence::~VKFence() = default;
-
-void VKFence::Wait() {
-    switch (const VkResult result = handle.Wait()) {
-    case VK_SUCCESS:
-        return;
-    case VK_ERROR_DEVICE_LOST:
-        device.ReportLoss();
-        [[fallthrough]];
-    default:
-        throw vk::Exception(result);
-    }
-}
-
-void VKFence::Release() {
-    ASSERT(is_owned);
-    is_owned = false;
-}
-
-void VKFence::Commit() {
-    is_owned = true;
-    is_used = true;
-}
-
-bool VKFence::Tick(bool gpu_wait, bool owner_wait) {
-    if (!is_used) {
-        // If a fence is not used it's always free.
-        return true;
-    }
-    if (is_owned && !owner_wait) {
-        // The fence is still being owned (Release has not been called) and ownership wait has
-        // not been asked.
-        return false;
-    }
-
-    if (gpu_wait) {
-        // Wait for the fence if it has been requested.
-        (void)handle.Wait();
-    } else {
-        if (handle.GetStatus() != VK_SUCCESS) {
-            // Vulkan fence is not ready, not much it can do here
-            return false;
-        }
-    }
-
-    // Broadcast resources their free state.
-    for (auto* resource : protected_resources) {
-        resource->OnFenceRemoval(this);
-    }
-    protected_resources.clear();
-
-    // Prepare fence for reusage.
-    handle.Reset();
-    is_used = false;
-    return true;
-}
-
-void VKFence::Protect(VKResource* resource) {
-    protected_resources.push_back(resource);
-}
-
-void VKFence::Unprotect(VKResource* resource) {
-    const auto it = std::find(protected_resources.begin(), protected_resources.end(), resource);
-    ASSERT(it != protected_resources.end());
-
-    resource->OnFenceRemoval(this);
-    protected_resources.erase(it);
-}
-
-void VKFence::RedirectProtection(VKResource* old_resource, VKResource* new_resource) noexcept {
-    std::replace(std::begin(protected_resources), std::end(protected_resources), old_resource,
-                 new_resource);
-}
-
-VKFenceWatch::VKFenceWatch() = default;
-
-VKFenceWatch::VKFenceWatch(VKFence& initial_fence) {
-    Watch(initial_fence);
-}
-
-VKFenceWatch::VKFenceWatch(VKFenceWatch&& rhs) noexcept {
-    fence = std::exchange(rhs.fence, nullptr);
-    if (fence) {
-        fence->RedirectProtection(&rhs, this);
-    }
-}
-
-VKFenceWatch& VKFenceWatch::operator=(VKFenceWatch&& rhs) noexcept {
-    fence = std::exchange(rhs.fence, nullptr);
-    if (fence) {
-        fence->RedirectProtection(&rhs, this);
-    }
-    return *this;
-}
-
-VKFenceWatch::~VKFenceWatch() {
-    if (fence) {
-        fence->Unprotect(this);
-    }
-}
-
-void VKFenceWatch::Wait() {
-    if (fence == nullptr) {
-        return;
-    }
-    fence->Wait();
-    fence->Unprotect(this);
-}
-
-void VKFenceWatch::Watch(VKFence& new_fence) {
-    Wait();
-    fence = &new_fence;
-    fence->Protect(this);
-}
-
-bool VKFenceWatch::TryWatch(VKFence& new_fence) {
-    if (fence) {
-        return false;
-    }
-    fence = &new_fence;
-    fence->Protect(this);
-    return true;
-}
-
-void VKFenceWatch::OnFenceRemoval(VKFence* signaling_fence) {
-    ASSERT_MSG(signaling_fence == fence, "Removing the wrong fence");
-    fence = nullptr;
-}
-
-VKFencedPool::VKFencedPool(std::size_t grow_step) : grow_step{grow_step} {}
-
-VKFencedPool::~VKFencedPool() = default;
-
-std::size_t VKFencedPool::CommitResource(VKFence& fence) {
-    const auto Search = [&](std::size_t begin, std::size_t end) -> std::optional<std::size_t> {
-        for (std::size_t iterator = begin; iterator < end; ++iterator) {
-            if (watches[iterator]->TryWatch(fence)) {
-                // The resource is now being watched, a free resource was successfully found.
-                return iterator;
-            }
-        }
-        return {};
-    };
-    // Try to find a free resource from the hinted position to the end.
-    auto found = Search(free_iterator, watches.size());
-    if (!found) {
-        // Search from beginning to the hinted position.
-        found = Search(0, free_iterator);
-        if (!found) {
-            // Both searches failed, the pool is full; handle it.
-            const std::size_t free_resource = ManageOverflow();
-
-            // Watch will wait for the resource to be free.
-            watches[free_resource]->Watch(fence);
-            found = free_resource;
-        }
-    }
-    // Free iterator is hinted to the resource after the one that's been commited.
-    free_iterator = (*found + 1) % watches.size();
-    return *found;
-}
-
-std::size_t VKFencedPool::ManageOverflow() {
-    const std::size_t old_capacity = watches.size();
-    Grow();
-
-    // The last entry is guaranted to be free, since it's the first element of the freshly
-    // allocated resources.
-    return old_capacity;
-}
-
-void VKFencedPool::Grow() {
-    const std::size_t old_capacity = watches.size();
-    watches.resize(old_capacity + grow_step);
-    std::generate(watches.begin() + old_capacity, watches.end(),
-                  []() { return std::make_unique<VKFenceWatch>(); });
-    Allocate(old_capacity, old_capacity + grow_step);
-}
-
-VKResourceManager::VKResourceManager(const VKDevice& device) : device{device} {
-    GrowFences(FENCES_GROW_STEP);
-    command_buffer_pool = std::make_unique<CommandBufferPool>(device);
-}
-
-VKResourceManager::~VKResourceManager() = default;
-
-VKFence& VKResourceManager::CommitFence() {
-    const auto StepFences = [&](bool gpu_wait, bool owner_wait) -> VKFence* {
-        const auto Tick = [=](auto& fence) { return fence->Tick(gpu_wait, owner_wait); };
-        const auto hinted = fences.begin() + fences_iterator;
-
-        auto it = std::find_if(hinted, fences.end(), Tick);
-        if (it == fences.end()) {
-            it = std::find_if(fences.begin(), hinted, Tick);
-            if (it == hinted) {
-                return nullptr;
-            }
-        }
-        fences_iterator = std::distance(fences.begin(), it) + 1;
-        if (fences_iterator >= fences.size())
-            fences_iterator = 0;
-
-        auto& fence = *it;
-        fence->Commit();
-        return fence.get();
-    };
-
-    VKFence* found_fence = StepFences(false, false);
-    if (!found_fence) {
-        // Try again, this time waiting.
-        found_fence = StepFences(true, false);
-
-        if (!found_fence) {
-            // Allocate new fences and try again.
-            LOG_INFO(Render_Vulkan, "Allocating new fences {} -> {}", fences.size(),
-                     fences.size() + FENCES_GROW_STEP);
-
-            GrowFences(FENCES_GROW_STEP);
-            found_fence = StepFences(true, false);
-            ASSERT(found_fence != nullptr);
-        }
-    }
-    return *found_fence;
-}
-
-VkCommandBuffer VKResourceManager::CommitCommandBuffer(VKFence& fence) {
-    return command_buffer_pool->Commit(fence);
-}
-
-void VKResourceManager::GrowFences(std::size_t new_fences_count) {
-    const std::size_t previous_size = fences.size();
-    fences.resize(previous_size + new_fences_count);
-
-    std::generate(fences.begin() + previous_size, fences.end(),
-                  [this] { return std::make_unique<VKFence>(device); });
-}
-
-} // namespace Vulkan

src/video_core/renderer_vulkan/vk_resource_manager.h

@@ -1,196 +0,0 @@
-// Copyright 2018 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <cstddef>
-#include <memory>
-#include <vector>
-#include "video_core/renderer_vulkan/wrapper.h"
-
-namespace Vulkan {
-
-class VKDevice;
-class VKFence;
-class VKResourceManager;
-
-class CommandBufferPool;
-
-/// Interface for a Vulkan resource
-class VKResource {
-public:
-    explicit VKResource();
-    virtual ~VKResource();
-
-    /**
-     * Signals the object that an owning fence has been signaled.
-     * @param signaling_fence Fence that signals its usage end.
-     */
-    virtual void OnFenceRemoval(VKFence* signaling_fence) = 0;
-};
-
-/**
- * Fences take ownership of objects, protecting them from GPU-side or driver-side concurrent access.
- * They must be commited from the resource manager. Their usage flow is: commit the fence from the
- * resource manager, protect resources with it and use them, send the fence to an execution queue
- * and Wait for it if needed and then call Release. Used resources will automatically be signaled
- * when they are free to be reused.
- * @brief Protects resources for concurrent usage and signals its release.
- */
-class VKFence {
-    friend class VKResourceManager;
-
-public:
-    explicit VKFence(const VKDevice& device);
-    ~VKFence();
-
-    /**
-     * Waits for the fence to be signaled.
-     * @warning You must have ownership of the fence and it has to be previously sent to a queue to
-     * call this function.
-     */
-    void Wait();
-
-    /**
-     * Releases ownership of the fence. Pass after it has been sent to an execution queue.
-     * Unmanaged usage of the fence after the call will result in undefined behavior because it may
-     * be being used for something else.
-     */
-    void Release();
-
-    /// Protects a resource with this fence.
-    void Protect(VKResource* resource);
-
-    /// Removes protection for a resource.
-    void Unprotect(VKResource* resource);
-
-    /// Redirects one protected resource to a new address.
-    void RedirectProtection(VKResource* old_resource, VKResource* new_resource) noexcept;
-
-    /// Retreives the fence.
-    operator VkFence() const {
-        return *handle;
-    }
-
-private:
-    /// Take ownership of the fence.
-    void Commit();
-
-    /**
-     * Updates the fence status.
-     * @warning Waiting for the owner might soft lock the execution.
-     * @param gpu_wait Wait for the fence to be signaled by the driver.
-     * @param owner_wait Wait for the owner to signal its freedom.
-     * @returns True if the fence is free. Waiting for gpu and owner will always return true.
-     */
-    bool Tick(bool gpu_wait, bool owner_wait);
-
-    const VKDevice& device;                       ///< Device handler
-    vk::Fence handle;                             ///< Vulkan fence
-    std::vector<VKResource*> protected_resources; ///< List of resources protected by this fence
-    bool is_owned = false; ///< The fence has been commited but not released yet.
-    bool is_used = false;  ///< The fence has been commited but it has not been checked to be free.
-};
-
-/**
- * A fence watch is used to keep track of the usage of a fence and protect a resource or set of
- * resources without having to inherit VKResource from their handlers.
- */
-class VKFenceWatch final : public VKResource {
-public:
-    explicit VKFenceWatch();
-    VKFenceWatch(VKFence& initial_fence);
-    VKFenceWatch(VKFenceWatch&&) noexcept;
-    VKFenceWatch(const VKFenceWatch&) = delete;
-    ~VKFenceWatch() override;
-
-    VKFenceWatch& operator=(VKFenceWatch&&) noexcept;
-
-    /// Waits for the fence to be released.
-    void Wait();
-
-    /**
-     * Waits for a previous fence and watches a new one.
-     * @param new_fence New fence to wait to.
-     */
-    void Watch(VKFence& new_fence);
-
-    /**
-     * Checks if it's currently being watched and starts watching it if it's available.
-     * @returns True if a watch has started, false if it's being watched.
-     */
-    bool TryWatch(VKFence& new_fence);
-
-    void OnFenceRemoval(VKFence* signaling_fence) override;
-
-    /**
-     * Do not use it paired with Watch. Use TryWatch instead.
-     * Returns true when the watch is free.
-     */
-    bool IsUsed() const {
-        return fence != nullptr;
-    }
-
-private:
-    VKFence* fence{}; ///< Fence watching this resource. nullptr when the watch is free.
-};
-
-/**
- * Handles a pool of resources protected by fences. Manages resource overflow allocating more
- * resources.
- */
-class VKFencedPool {
-public:
-    explicit VKFencedPool(std::size_t grow_step);
-    virtual ~VKFencedPool();
-
-protected:
-    /**
-     * Commits a free resource and protects it with a fence. It may allocate new resources.
-     * @param fence Fence that protects the commited resource.
-     * @returns Index of the resource commited.
-     */
-    std::size_t CommitResource(VKFence& fence);
-
-    /// Called when a chunk of resources have to be allocated.
-    virtual void Allocate(std::size_t begin, std::size_t end) = 0;
-
-private:
-    /// Manages pool overflow allocating new resources.
-    std::size_t ManageOverflow();
-
-    /// Allocates a new page of resources.
-    void Grow();
-
-    std::size_t grow_step = 0;     ///< Number of new resources created after an overflow
-    std::size_t free_iterator = 0; ///< Hint to where the next free resources is likely to be found
-    std::vector<std::unique_ptr<VKFenceWatch>> watches; ///< Set of watched resources
-};
-
-/**
- * The resource manager handles all resources that can be protected with a fence avoiding
- * driver-side or GPU-side concurrent usage. Usage is documented in VKFence.
- */
-class VKResourceManager final {
-public:
-    explicit VKResourceManager(const VKDevice& device);
-    ~VKResourceManager();
-
-    /// Commits a fence. It has to be sent to a queue and released.
-    VKFence& CommitFence();
-
-    /// Commits an unused command buffer and protects it with a fence.
-    VkCommandBuffer CommitCommandBuffer(VKFence& fence);
-
-private:
-    /// Allocates new fences.
-    void GrowFences(std::size_t new_fences_count);
-
-    const VKDevice& device;          ///< Device handler.
-    std::size_t fences_iterator = 0; ///< Index where a free fence is likely to be found.
-    std::vector<std::unique_ptr<VKFence>> fences;           ///< Pool of fences.
-    std::unique_ptr<CommandBufferPool> command_buffer_pool; ///< Pool of command buffers.
-};
-
-} // namespace Vulkan

src/video_core/renderer_vulkan/vk_resource_pool.cpp

@@ -0,0 +1,63 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <optional>
+
+#include "video_core/renderer_vulkan/vk_master_semaphore.h"
+#include "video_core/renderer_vulkan/vk_resource_pool.h"
+
+namespace Vulkan {
+
+ResourcePool::ResourcePool(MasterSemaphore& master_semaphore_, size_t grow_step_)
+    : master_semaphore{master_semaphore_}, grow_step{grow_step_} {}
+
+ResourcePool::~ResourcePool() = default;
+
+size_t ResourcePool::CommitResource() {
+    // Refresh semaphore to query updated results
+    master_semaphore.Refresh();
+
+    const auto search = [this](size_t begin, size_t end) -> std::optional<size_t> {
+        for (size_t iterator = begin; iterator < end; ++iterator) {
+            if (master_semaphore.IsFree(ticks[iterator])) {
+                ticks[iterator] = master_semaphore.CurrentTick();
+                return iterator;
+            }
+        }
+        return {};
+    };
+    // Try to find a free resource from the hinted position to the end.
+    auto found = search(free_iterator, ticks.size());
+    if (!found) {
+        // Search from beginning to the hinted position.
+        found = search(0, free_iterator);
+        if (!found) {
+            // Both searches failed, the pool is full; handle it.
+            const size_t free_resource = ManageOverflow();
+
+            ticks[free_resource] = master_semaphore.CurrentTick();
+            found = free_resource;
+        }
+    }
+    // Free iterator is hinted to the resource after the one that's been commited.
+    free_iterator = (*found + 1) % ticks.size();
+    return *found;
+}
+
+size_t ResourcePool::ManageOverflow() {
+    const size_t old_capacity = ticks.size();
+    Grow();
+
+    // The last entry is guaranted to be free, since it's the first element of the freshly
+    // allocated resources.
+    return old_capacity;
+}
+
+void ResourcePool::Grow() {
+    const size_t old_capacity = ticks.size();
+    ticks.resize(old_capacity + grow_step);
+    Allocate(old_capacity, old_capacity + grow_step);
+}
+
+} // namespace Vulkan

src/video_core/renderer_vulkan/vk_resource_pool.h

@@ -0,0 +1,43 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <vector>
+
+#include "common/common_types.h"
+
+namespace Vulkan {
+
+class MasterSemaphore;
+
+/**
+ * Handles a pool of resources protected by fences. Manages resource overflow allocating more
+ * resources.
+ */
+class ResourcePool {
+public:
+    explicit ResourcePool(MasterSemaphore& master_semaphore, size_t grow_step);
+    virtual ~ResourcePool();
+
+protected:
+    size_t CommitResource();
+
+    /// Called when a chunk of resources have to be allocated.
+    virtual void Allocate(size_t begin, size_t end) = 0;
+
+private:
+    /// Manages pool overflow allocating new resources.
+    size_t ManageOverflow();
+
+    /// Allocates a new page of resources.
+    void Grow();
+
+    MasterSemaphore& master_semaphore;
+    size_t grow_step = 0;     ///< Number of new resources created after an overflow
+    size_t free_iterator = 0; ///< Hint to where the next free resources is likely to be found
+    std::vector<u64> ticks;   ///< Ticks for each resource
+};
+
+} // namespace Vulkan

src/video_core/renderer_vulkan/vk_scheduler.cpp

@@ -10,9 +10,10 @@
 
 #include "common/microprofile.h"
 #include "common/thread.h"
+#include "video_core/renderer_vulkan/vk_command_pool.h"
 #include "video_core/renderer_vulkan/vk_device.h"
+#include "video_core/renderer_vulkan/vk_master_semaphore.h"
 #include "video_core/renderer_vulkan/vk_query_cache.h"
-#include "video_core/renderer_vulkan/vk_resource_manager.h"
 #include "video_core/renderer_vulkan/vk_scheduler.h"
 #include "video_core/renderer_vulkan/vk_state_tracker.h"
 #include "video_core/renderer_vulkan/wrapper.h"
@@ -35,10 +36,10 @@ void VKScheduler::CommandChunk::ExecuteAll(vk::CommandBuffer cmdbuf) {
     last = nullptr;
 }
 
-VKScheduler::VKScheduler(const VKDevice& device, VKResourceManager& resource_manager,
-                         StateTracker& state_tracker)
-    : device{device}, resource_manager{resource_manager}, state_tracker{state_tracker},
-      next_fence{&resource_manager.CommitFence()} {
+VKScheduler::VKScheduler(const VKDevice& device_, StateTracker& state_tracker_)
+    : device{device_}, state_tracker{state_tracker_},
+      master_semaphore{std::make_unique<MasterSemaphore>(device)},
+      command_pool{std::make_unique<CommandPool>(*master_semaphore, device)} {
     AcquireNewChunk();
     AllocateNewContext();
     worker_thread = std::thread(&VKScheduler::WorkerThread, this);
@@ -50,20 +51,27 @@ VKScheduler::~VKScheduler() {
     worker_thread.join();
 }
 
-void VKScheduler::Flush(bool release_fence, VkSemaphore semaphore) {
+u64 VKScheduler::CurrentTick() const noexcept {
+    return master_semaphore->CurrentTick();
+}
+
+bool VKScheduler::IsFree(u64 tick) const noexcept {
+    return master_semaphore->IsFree(tick);
+}
+
+void VKScheduler::Wait(u64 tick) {
+    master_semaphore->Wait(tick);
+}
+
+void VKScheduler::Flush(VkSemaphore semaphore) {
     SubmitExecution(semaphore);
-    if (release_fence) {
-        current_fence->Release();
-    }
     AllocateNewContext();
 }
 
-void VKScheduler::Finish(bool release_fence, VkSemaphore semaphore) {
+void VKScheduler::Finish(VkSemaphore semaphore) {
+    const u64 presubmit_tick = CurrentTick();
     SubmitExecution(semaphore);
-    current_fence->Wait();
-    if (release_fence) {
-        current_fence->Release();
-    }
+    Wait(presubmit_tick);
     AllocateNewContext();
 }
 
@@ -160,18 +168,38 @@ void VKScheduler::SubmitExecution(VkSemaphore semaphore) {
 
     current_cmdbuf.End();
 
+    const VkSemaphore timeline_semaphore = master_semaphore->Handle();
+    const u32 num_signal_semaphores = semaphore ? 2U : 1U;
+
+    const u64 signal_value = master_semaphore->CurrentTick();
+    const u64 wait_value = signal_value - 1;
+    const VkPipelineStageFlags wait_stage_mask = VK_PIPELINE_STAGE_ALL_COMMANDS_BIT;
+
+    master_semaphore->NextTick();
+
+    const std::array signal_values{signal_value, u64(0)};
+    const std::array signal_semaphores{timeline_semaphore, semaphore};
+
+    const VkTimelineSemaphoreSubmitInfoKHR timeline_si{
+        .sType = VK_STRUCTURE_TYPE_TIMELINE_SEMAPHORE_SUBMIT_INFO_KHR,
+        .pNext = nullptr,
+        .waitSemaphoreValueCount = 1,
+        .pWaitSemaphoreValues = &wait_value,
+        .signalSemaphoreValueCount = num_signal_semaphores,
+        .pSignalSemaphoreValues = signal_values.data(),
+    };
     const VkSubmitInfo submit_info{
         .sType = VK_STRUCTURE_TYPE_SUBMIT_INFO,
-        .pNext = nullptr,
-        .waitSemaphoreCount = 0,
-        .pWaitSemaphores = nullptr,
-        .pWaitDstStageMask = nullptr,
+        .pNext = &timeline_si,
+        .waitSemaphoreCount = 1,
+        .pWaitSemaphores = &timeline_semaphore,
+        .pWaitDstStageMask = &wait_stage_mask,
         .commandBufferCount = 1,
         .pCommandBuffers = current_cmdbuf.address(),
-        .signalSemaphoreCount = semaphore ? 1U : 0U,
-        .pSignalSemaphores = &semaphore,
+        .signalSemaphoreCount = num_signal_semaphores,
+        .pSignalSemaphores = signal_semaphores.data(),
     };
-    switch (const VkResult result = device.GetGraphicsQueue().Submit(submit_info, *current_fence)) {
+    switch (const VkResult result = device.GetGraphicsQueue().Submit(submit_info)) {
     case VK_SUCCESS:
         break;
     case VK_ERROR_DEVICE_LOST:
@@ -183,14 +211,9 @@ void VKScheduler::SubmitExecution(VkSemaphore semaphore) {
 }
 
 void VKScheduler::AllocateNewContext() {
-    ++ticks;
-
     std::unique_lock lock{mutex};
-    current_fence = next_fence;
-    next_fence = &resource_manager.CommitFence();
 
-    current_cmdbuf = vk::CommandBuffer(resource_manager.CommitCommandBuffer(*current_fence),
-                                       device.GetDispatchLoader());
+    current_cmdbuf = vk::CommandBuffer(command_pool->Commit(), device.GetDispatchLoader());
     current_cmdbuf.Begin({
         .sType = VK_STRUCTURE_TYPE_COMMAND_BUFFER_BEGIN_INFO,
         .pNext = nullptr,

src/video_core/renderer_vulkan/vk_scheduler.h

@@ -16,42 +16,33 @@
 
 namespace Vulkan {
 
+class CommandPool;
+class MasterSemaphore;
 class StateTracker;
 class VKDevice;
-class VKFence;
 class VKQueryCache;
-class VKResourceManager;
-
-class VKFenceView {
-public:
-    VKFenceView() = default;
-    VKFenceView(VKFence* const& fence) : fence{fence} {}
-
-    VKFence* operator->() const noexcept {
-        return fence;
-    }
-
-    operator VKFence&() const noexcept {
-        return *fence;
-    }
-
-private:
-    VKFence* const& fence;
-};
 
 /// The scheduler abstracts command buffer and fence management with an interface that's able to do
 /// OpenGL-like operations on Vulkan command buffers.
 class VKScheduler {
 public:
-    explicit VKScheduler(const VKDevice& device, VKResourceManager& resource_manager,
-                         StateTracker& state_tracker);
+    explicit VKScheduler(const VKDevice& device, StateTracker& state_tracker);
     ~VKScheduler();
 
+    /// Returns the current command buffer tick.
+    [[nodiscard]] u64 CurrentTick() const noexcept;
+
+    /// Returns true when a tick has been triggered by the GPU.
+    [[nodiscard]] bool IsFree(u64 tick) const noexcept;
+
+    /// Waits for the given tick to trigger on the GPU.
+    void Wait(u64 tick);
+
     /// Sends the current execution context to the GPU.
-    void Flush(bool release_fence = true, VkSemaphore semaphore = nullptr);
+    void Flush(VkSemaphore semaphore = nullptr);
 
     /// Sends the current execution context to the GPU and waits for it to complete.
-    void Finish(bool release_fence = true, VkSemaphore semaphore = nullptr);
+    void Finish(VkSemaphore semaphore = nullptr);
 
     /// Waits for the worker thread to finish executing everything. After this function returns it's
     /// safe to touch worker resources.
@@ -86,14 +77,9 @@ public:
         (void)chunk->Record(command);
     }
 
-    /// Gets a reference to the current fence.
-    VKFenceView GetFence() const {
-        return current_fence;
-    }
-
-    /// Returns the current command buffer tick.
-    u64 Ticks() const {
-        return ticks;
+    /// Returns the master timeline semaphore.
+    [[nodiscard]] MasterSemaphore& GetMasterSemaphore() const noexcept {
+        return *master_semaphore;
     }
 
 private:
@@ -171,6 +157,13 @@ private:
         std::array<u8, 0x8000> data{};
     };
 
+    struct State {
+        VkRenderPass renderpass = nullptr;
+        VkFramebuffer framebuffer = nullptr;
+        VkExtent2D render_area = {0, 0};
+        VkPipeline graphics_pipeline = nullptr;
+    };
+
     void WorkerThread();
 
     void SubmitExecution(VkSemaphore semaphore);
@@ -186,30 +179,23 @@ private:
     void AcquireNewChunk();
 
     const VKDevice& device;
-    VKResourceManager& resource_manager;
     StateTracker& state_tracker;
 
+    std::unique_ptr<MasterSemaphore> master_semaphore;
+    std::unique_ptr<CommandPool> command_pool;
+
     VKQueryCache* query_cache = nullptr;
 
     vk::CommandBuffer current_cmdbuf;
-    VKFence* current_fence = nullptr;
-    VKFence* next_fence = nullptr;
-
-    struct State {
-        VkRenderPass renderpass = nullptr;
-        VkFramebuffer framebuffer = nullptr;
-        VkExtent2D render_area = {0, 0};
-        VkPipeline graphics_pipeline = nullptr;
-    } state;
 
     std::unique_ptr<CommandChunk> chunk;
     std::thread worker_thread;
 
+    State state;
     Common::SPSCQueue<std::unique_ptr<CommandChunk>> chunk_queue;
     Common::SPSCQueue<std::unique_ptr<CommandChunk>> chunk_reserve;
     std::mutex mutex;
     std::condition_variable cv;
-    std::atomic<u64> ticks = 0;
     bool quit = false;
 };
 

src/video_core/renderer_vulkan/vk_staging_buffer_pool.cpp

@@ -10,36 +10,18 @@
 #include "common/bit_util.h"
 #include "common/common_types.h"
 #include "video_core/renderer_vulkan/vk_device.h"
-#include "video_core/renderer_vulkan/vk_resource_manager.h"
 #include "video_core/renderer_vulkan/vk_scheduler.h"
 #include "video_core/renderer_vulkan/vk_staging_buffer_pool.h"
 #include "video_core/renderer_vulkan/wrapper.h"
 
 namespace Vulkan {
 
-VKStagingBufferPool::StagingBuffer::StagingBuffer(std::unique_ptr<VKBuffer> buffer, VKFence& fence,
-                                                  u64 last_epoch)
-    : buffer{std::move(buffer)}, watch{fence}, last_epoch{last_epoch} {}
+VKStagingBufferPool::StagingBuffer::StagingBuffer(std::unique_ptr<VKBuffer> buffer_)
+    : buffer{std::move(buffer_)} {}
 
-VKStagingBufferPool::StagingBuffer::StagingBuffer(StagingBuffer&& rhs) noexcept {
-    buffer = std::move(rhs.buffer);
-    watch = std::move(rhs.watch);
-    last_epoch = rhs.last_epoch;
-}
-
-VKStagingBufferPool::StagingBuffer::~StagingBuffer() = default;
-
-VKStagingBufferPool::StagingBuffer& VKStagingBufferPool::StagingBuffer::operator=(
-    StagingBuffer&& rhs) noexcept {
-    buffer = std::move(rhs.buffer);
-    watch = std::move(rhs.watch);
-    last_epoch = rhs.last_epoch;
-    return *this;
-}
-
-VKStagingBufferPool::VKStagingBufferPool(const VKDevice& device, VKMemoryManager& memory_manager,
-                                         VKScheduler& scheduler)
-    : device{device}, memory_manager{memory_manager}, scheduler{scheduler} {}
+VKStagingBufferPool::VKStagingBufferPool(const VKDevice& device_, VKMemoryManager& memory_manager_,
+                                         VKScheduler& scheduler_)
+    : device{device_}, memory_manager{memory_manager_}, scheduler{scheduler_} {}
 
 VKStagingBufferPool::~VKStagingBufferPool() = default;
 
@@ -51,7 +33,6 @@ VKBuffer& VKStagingBufferPool::GetUnusedBuffer(std::size_t size, bool host_visib
 }
 
 void VKStagingBufferPool::TickFrame() {
-    ++epoch;
     current_delete_level = (current_delete_level + 1) % NumLevels;
 
     ReleaseCache(true);
@@ -59,11 +40,12 @@ void VKStagingBufferPool::TickFrame() {
 }
 
 VKBuffer* VKStagingBufferPool::TryGetReservedBuffer(std::size_t size, bool host_visible) {
-    for (auto& entry : GetCache(host_visible)[Common::Log2Ceil64(size)].entries) {
-        if (entry.watch.TryWatch(scheduler.GetFence())) {
-            entry.last_epoch = epoch;
-            return &*entry.buffer;
+    for (StagingBuffer& entry : GetCache(host_visible)[Common::Log2Ceil64(size)].entries) {
+        if (!scheduler.IsFree(entry.tick)) {
+            continue;
         }
+        entry.tick = scheduler.CurrentTick();
+        return &*entry.buffer;
     }
     return nullptr;
 }
@@ -86,8 +68,10 @@ VKBuffer& VKStagingBufferPool::CreateStagingBuffer(std::size_t size, bool host_v
     });
     buffer->commit = memory_manager.Commit(buffer->handle, host_visible);
 
-    auto& entries = GetCache(host_visible)[log2].entries;
-    return *entries.emplace_back(std::move(buffer), scheduler.GetFence(), epoch).buffer;
+    std::vector<StagingBuffer>& entries = GetCache(host_visible)[log2].entries;
+    StagingBuffer& entry = entries.emplace_back(std::move(buffer));
+    entry.tick = scheduler.CurrentTick();
+    return *entry.buffer;
 }
 
 VKStagingBufferPool::StagingBuffersCache& VKStagingBufferPool::GetCache(bool host_visible) {
@@ -109,9 +93,8 @@ u64 VKStagingBufferPool::ReleaseLevel(StagingBuffersCache& cache, std::size_t lo
     auto& entries = staging.entries;
     const std::size_t old_size = entries.size();
 
-    const auto is_deleteable = [this](const auto& entry) {
-        static constexpr u64 epochs_to_destroy = 180;
-        return entry.last_epoch + epochs_to_destroy < epoch && !entry.watch.IsUsed();
+    const auto is_deleteable = [this](const StagingBuffer& entry) {
+        return scheduler.IsFree(entry.tick);
     };
     const std::size_t begin_offset = staging.delete_index;
     const std::size_t end_offset = std::min(begin_offset + deletions_per_tick, old_size);

src/video_core/renderer_vulkan/vk_staging_buffer_pool.h

@@ -10,13 +10,11 @@
 #include "common/common_types.h"
 
 #include "video_core/renderer_vulkan/vk_memory_manager.h"
-#include "video_core/renderer_vulkan/vk_resource_manager.h"
 #include "video_core/renderer_vulkan/wrapper.h"
 
 namespace Vulkan {
 
 class VKDevice;
-class VKFenceWatch;
 class VKScheduler;
 
 struct VKBuffer final {
@@ -36,16 +34,10 @@ public:
 
 private:
     struct StagingBuffer final {
-        explicit StagingBuffer(std::unique_ptr<VKBuffer> buffer, VKFence& fence, u64 last_epoch);
-        StagingBuffer(StagingBuffer&& rhs) noexcept;
-        StagingBuffer(const StagingBuffer&) = delete;
-        ~StagingBuffer();
-
-        StagingBuffer& operator=(StagingBuffer&& rhs) noexcept;
+        explicit StagingBuffer(std::unique_ptr<VKBuffer> buffer);
 
         std::unique_ptr<VKBuffer> buffer;
-        VKFenceWatch watch;
-        u64 last_epoch = 0;
+        u64 tick = 0;
     };
 
     struct StagingBuffers final {
@@ -73,8 +65,6 @@ private:
     StagingBuffersCache host_staging_buffers;
     StagingBuffersCache device_staging_buffers;
 
-    u64 epoch = 0;
-
     std::size_t current_delete_level = 0;
 };