diff options
Diffstat (limited to '')
107 files changed, 1139 insertions, 1138 deletions
diff --git a/src/audio_core/audio_renderer.cpp b/src/audio_core/audio_renderer.cpp index e6f38d600..6b0167acd 100644 --- a/src/audio_core/audio_renderer.cpp +++ b/src/audio_core/audio_renderer.cpp @@ -73,7 +73,7 @@ private: EffectInStatus info{}; }; AudioRenderer::AudioRenderer(Core::Timing::CoreTiming& core_timing, AudioRendererParameter params, - Kernel::SharedPtr<Kernel::WritableEvent> buffer_event, + std::shared_ptr<Kernel::WritableEvent> buffer_event, std::size_t instance_number) : worker_params{params}, buffer_event{buffer_event}, voices(params.voice_count), effects(params.effect_count) { diff --git a/src/audio_core/audio_renderer.h b/src/audio_core/audio_renderer.h index 4f14b91cd..abed224bb 100644 --- a/src/audio_core/audio_renderer.h +++ b/src/audio_core/audio_renderer.h @@ -218,8 +218,7 @@ static_assert(sizeof(UpdateDataHeader) == 0x40, "UpdateDataHeader has wrong size class AudioRenderer { public: AudioRenderer(Core::Timing::CoreTiming& core_timing, AudioRendererParameter params, - Kernel::SharedPtr<Kernel::WritableEvent> buffer_event, - std::size_t instance_number); + std::shared_ptr<Kernel::WritableEvent> buffer_event, std::size_t instance_number); ~AudioRenderer(); std::vector<u8> UpdateAudioRenderer(const std::vector<u8>& input_params); @@ -235,7 +234,7 @@ private: class VoiceState; AudioRendererParameter worker_params; - Kernel::SharedPtr<Kernel::WritableEvent> buffer_event; + std::shared_ptr<Kernel::WritableEvent> buffer_event; std::vector<VoiceState> voices; std::vector<EffectState> effects; std::unique_ptr<AudioOut> audio_out; diff --git a/src/common/assert.h b/src/common/assert.h index 4b0e3f64e..5b67c5c52 100644 --- a/src/common/assert.h +++ b/src/common/assert.h @@ -41,8 +41,9 @@ __declspec(noinline, noreturn) } \ while (0) -#define UNREACHABLE() ASSERT_MSG(false, "Unreachable code!") -#define UNREACHABLE_MSG(...) ASSERT_MSG(false, __VA_ARGS__) +#define UNREACHABLE() assert_noinline_call([] { LOG_CRITICAL(Debug, "Unreachable code!"); }) +#define UNREACHABLE_MSG(...) \ + assert_noinline_call([&] { LOG_CRITICAL(Debug, "Unreachable code!\n" __VA_ARGS__); }) #ifdef _DEBUG #define DEBUG_ASSERT(_a_) ASSERT(_a_) diff --git a/src/core/core.h b/src/core/core.h index 984074ce3..f9b1a2866 100644 --- a/src/core/core.h +++ b/src/core/core.h @@ -7,6 +7,7 @@ #include <cstddef> #include <memory> #include <string> +#include <vector> #include "common/common_types.h" #include "core/file_sys/vfs_types.h" diff --git a/src/core/hle/ipc_helpers.h b/src/core/hle/ipc_helpers.h index 5bb139483..b9f9ae1fa 100644 --- a/src/core/hle/ipc_helpers.h +++ b/src/core/hle/ipc_helpers.h @@ -203,10 +203,10 @@ public: void PushRaw(const T& value); template <typename... O> - void PushMoveObjects(Kernel::SharedPtr<O>... pointers); + void PushMoveObjects(std::shared_ptr<O>... pointers); template <typename... O> - void PushCopyObjects(Kernel::SharedPtr<O>... pointers); + void PushCopyObjects(std::shared_ptr<O>... pointers); private: u32 normal_params_size{}; @@ -298,7 +298,7 @@ void ResponseBuilder::Push(const First& first_value, const Other&... other_value } template <typename... O> -inline void ResponseBuilder::PushCopyObjects(Kernel::SharedPtr<O>... pointers) { +inline void ResponseBuilder::PushCopyObjects(std::shared_ptr<O>... pointers) { auto objects = {pointers...}; for (auto& object : objects) { context->AddCopyObject(std::move(object)); @@ -306,7 +306,7 @@ inline void ResponseBuilder::PushCopyObjects(Kernel::SharedPtr<O>... pointers) { } template <typename... O> -inline void ResponseBuilder::PushMoveObjects(Kernel::SharedPtr<O>... pointers) { +inline void ResponseBuilder::PushMoveObjects(std::shared_ptr<O>... pointers) { auto objects = {pointers...}; for (auto& object : objects) { context->AddMoveObject(std::move(object)); @@ -357,10 +357,10 @@ public: T PopRaw(); template <typename T> - Kernel::SharedPtr<T> GetMoveObject(std::size_t index); + std::shared_ptr<T> GetMoveObject(std::size_t index); template <typename T> - Kernel::SharedPtr<T> GetCopyObject(std::size_t index); + std::shared_ptr<T> GetCopyObject(std::size_t index); template <class T> std::shared_ptr<T> PopIpcInterface() { @@ -465,12 +465,12 @@ void RequestParser::Pop(First& first_value, Other&... other_values) { } template <typename T> -Kernel::SharedPtr<T> RequestParser::GetMoveObject(std::size_t index) { +std::shared_ptr<T> RequestParser::GetMoveObject(std::size_t index) { return context->GetMoveObject<T>(index); } template <typename T> -Kernel::SharedPtr<T> RequestParser::GetCopyObject(std::size_t index) { +std::shared_ptr<T> RequestParser::GetCopyObject(std::size_t index) { return context->GetCopyObject<T>(index); } diff --git a/src/core/hle/kernel/address_arbiter.cpp b/src/core/hle/kernel/address_arbiter.cpp index de0a9064e..4859954cb 100644 --- a/src/core/hle/kernel/address_arbiter.cpp +++ b/src/core/hle/kernel/address_arbiter.cpp @@ -21,7 +21,7 @@ namespace Kernel { namespace { // Wake up num_to_wake (or all) threads in a vector. -void WakeThreads(const std::vector<SharedPtr<Thread>>& waiting_threads, s32 num_to_wake) { +void WakeThreads(const std::vector<std::shared_ptr<Thread>>& waiting_threads, s32 num_to_wake) { auto& system = Core::System::GetInstance(); // Only process up to 'target' threads, unless 'target' is <= 0, in which case process // them all. @@ -59,7 +59,8 @@ ResultCode AddressArbiter::SignalToAddress(VAddr address, SignalType type, s32 v } ResultCode AddressArbiter::SignalToAddressOnly(VAddr address, s32 num_to_wake) { - const std::vector<SharedPtr<Thread>> waiting_threads = GetThreadsWaitingOnAddress(address); + const std::vector<std::shared_ptr<Thread>> waiting_threads = + GetThreadsWaitingOnAddress(address); WakeThreads(waiting_threads, num_to_wake); return RESULT_SUCCESS; } @@ -87,7 +88,8 @@ ResultCode AddressArbiter::ModifyByWaitingCountAndSignalToAddressIfEqual(VAddr a } // Get threads waiting on the address. - const std::vector<SharedPtr<Thread>> waiting_threads = GetThreadsWaitingOnAddress(address); + const std::vector<std::shared_ptr<Thread>> waiting_threads = + GetThreadsWaitingOnAddress(address); // Determine the modified value depending on the waiting count. s32 updated_value; @@ -172,21 +174,21 @@ ResultCode AddressArbiter::WaitForAddressIfEqual(VAddr address, s32 value, s64 t } ResultCode AddressArbiter::WaitForAddressImpl(VAddr address, s64 timeout) { - SharedPtr<Thread> current_thread = system.CurrentScheduler().GetCurrentThread(); + Thread* current_thread = system.CurrentScheduler().GetCurrentThread(); current_thread->SetArbiterWaitAddress(address); current_thread->SetStatus(ThreadStatus::WaitArb); current_thread->InvalidateWakeupCallback(); - current_thread->WakeAfterDelay(timeout); system.PrepareReschedule(current_thread->GetProcessorID()); return RESULT_TIMEOUT; } -std::vector<SharedPtr<Thread>> AddressArbiter::GetThreadsWaitingOnAddress(VAddr address) const { +std::vector<std::shared_ptr<Thread>> AddressArbiter::GetThreadsWaitingOnAddress( + VAddr address) const { // Retrieve all threads that are waiting for this address. - std::vector<SharedPtr<Thread>> threads; + std::vector<std::shared_ptr<Thread>> threads; const auto& scheduler = system.GlobalScheduler(); const auto& thread_list = scheduler.GetThreadList(); @@ -198,7 +200,7 @@ std::vector<SharedPtr<Thread>> AddressArbiter::GetThreadsWaitingOnAddress(VAddr // Sort them by priority, such that the highest priority ones come first. std::sort(threads.begin(), threads.end(), - [](const SharedPtr<Thread>& lhs, const SharedPtr<Thread>& rhs) { + [](const std::shared_ptr<Thread>& lhs, const std::shared_ptr<Thread>& rhs) { return lhs->GetPriority() < rhs->GetPriority(); }); diff --git a/src/core/hle/kernel/address_arbiter.h b/src/core/hle/kernel/address_arbiter.h index ed0d0e69f..608918de5 100644 --- a/src/core/hle/kernel/address_arbiter.h +++ b/src/core/hle/kernel/address_arbiter.h @@ -72,7 +72,7 @@ private: ResultCode WaitForAddressImpl(VAddr address, s64 timeout); // Gets the threads waiting on an address. - std::vector<SharedPtr<Thread>> GetThreadsWaitingOnAddress(VAddr address) const; + std::vector<std::shared_ptr<Thread>> GetThreadsWaitingOnAddress(VAddr address) const; Core::System& system; }; diff --git a/src/core/hle/kernel/client_port.cpp b/src/core/hle/kernel/client_port.cpp index 744b1697d..4637b6017 100644 --- a/src/core/hle/kernel/client_port.cpp +++ b/src/core/hle/kernel/client_port.cpp @@ -15,11 +15,11 @@ namespace Kernel { ClientPort::ClientPort(KernelCore& kernel) : Object{kernel} {} ClientPort::~ClientPort() = default; -SharedPtr<ServerPort> ClientPort::GetServerPort() const { +std::shared_ptr<ServerPort> ClientPort::GetServerPort() const { return server_port; } -ResultVal<SharedPtr<ClientSession>> ClientPort::Connect() { +ResultVal<std::shared_ptr<ClientSession>> ClientPort::Connect() { // Note: Threads do not wait for the server endpoint to call // AcceptSession before returning from this call. @@ -29,7 +29,8 @@ ResultVal<SharedPtr<ClientSession>> ClientPort::Connect() { active_sessions++; // Create a new session pair, let the created sessions inherit the parent port's HLE handler. - auto [server, client] = ServerSession::CreateSessionPair(kernel, server_port->GetName(), this); + auto [server, client] = + ServerSession::CreateSessionPair(kernel, server_port->GetName(), SharedFrom(this)); if (server_port->HasHLEHandler()) { server_port->GetHLEHandler()->ClientConnected(server); diff --git a/src/core/hle/kernel/client_port.h b/src/core/hle/kernel/client_port.h index 4921ad4f0..715edd18c 100644 --- a/src/core/hle/kernel/client_port.h +++ b/src/core/hle/kernel/client_port.h @@ -17,6 +17,9 @@ class ServerPort; class ClientPort final : public Object { public: + explicit ClientPort(KernelCore& kernel); + ~ClientPort() override; + friend class ServerPort; std::string GetTypeName() const override { return "ClientPort"; @@ -30,7 +33,7 @@ public: return HANDLE_TYPE; } - SharedPtr<ServerPort> GetServerPort() const; + std::shared_ptr<ServerPort> GetServerPort() const; /** * Creates a new Session pair, adds the created ServerSession to the associated ServerPort's @@ -38,7 +41,7 @@ public: * waiting on it to awake. * @returns ClientSession The client endpoint of the created Session pair, or error code. */ - ResultVal<SharedPtr<ClientSession>> Connect(); + ResultVal<std::shared_ptr<ClientSession>> Connect(); /** * Signifies that a previously active connection has been closed, @@ -47,10 +50,7 @@ public: void ConnectionClosed(); private: - explicit ClientPort(KernelCore& kernel); - ~ClientPort() override; - - SharedPtr<ServerPort> server_port; ///< ServerPort associated with this client port. + std::shared_ptr<ServerPort> server_port; ///< ServerPort associated with this client port. u32 max_sessions = 0; ///< Maximum number of simultaneous sessions the port can have u32 active_sessions = 0; ///< Number of currently open sessions to this port std::string name; ///< Name of client port (optional) diff --git a/src/core/hle/kernel/client_session.cpp b/src/core/hle/kernel/client_session.cpp index c17baa50a..bc59d3306 100644 --- a/src/core/hle/kernel/client_session.cpp +++ b/src/core/hle/kernel/client_session.cpp @@ -16,25 +16,20 @@ ClientSession::ClientSession(KernelCore& kernel) : Object{kernel} {} ClientSession::~ClientSession() { // This destructor will be called automatically when the last ClientSession handle is closed by // the emulated application. - - // A local reference to the ServerSession is necessary to guarantee it - // will be kept alive until after ClientDisconnected() returns. - SharedPtr<ServerSession> server = parent->server; - if (server) { - server->ClientDisconnected(); + if (parent->server) { + parent->server->ClientDisconnected(); } parent->client = nullptr; } -ResultCode ClientSession::SendSyncRequest(SharedPtr<Thread> thread) { +ResultCode ClientSession::SendSyncRequest(Thread* thread) { // Keep ServerSession alive until we're done working with it. - SharedPtr<ServerSession> server = parent->server; - if (server == nullptr) + if (parent->server == nullptr) return ERR_SESSION_CLOSED_BY_REMOTE; // Signal the server session that new data is available - return server->HandleSyncRequest(std::move(thread)); + return parent->server->HandleSyncRequest(SharedFrom(thread)); } } // namespace Kernel diff --git a/src/core/hle/kernel/client_session.h b/src/core/hle/kernel/client_session.h index 09cdff588..5ae41db29 100644 --- a/src/core/hle/kernel/client_session.h +++ b/src/core/hle/kernel/client_session.h @@ -19,6 +19,9 @@ class Thread; class ClientSession final : public Object { public: + explicit ClientSession(KernelCore& kernel); + ~ClientSession() override; + friend class ServerSession; std::string GetTypeName() const override { @@ -34,12 +37,9 @@ public: return HANDLE_TYPE; } - ResultCode SendSyncRequest(SharedPtr<Thread> thread); + ResultCode SendSyncRequest(Thread* thread); private: - explicit ClientSession(KernelCore& kernel); - ~ClientSession() override; - /// The parent session, which links to the server endpoint. std::shared_ptr<Session> parent; diff --git a/src/core/hle/kernel/handle_table.cpp b/src/core/hle/kernel/handle_table.cpp index 2cc5d536b..e441a27fc 100644 --- a/src/core/hle/kernel/handle_table.cpp +++ b/src/core/hle/kernel/handle_table.cpp @@ -44,7 +44,7 @@ ResultCode HandleTable::SetSize(s32 handle_table_size) { return RESULT_SUCCESS; } -ResultVal<Handle> HandleTable::Create(SharedPtr<Object> obj) { +ResultVal<Handle> HandleTable::Create(std::shared_ptr<Object> obj) { DEBUG_ASSERT(obj != nullptr); const u16 slot = next_free_slot; @@ -70,7 +70,7 @@ ResultVal<Handle> HandleTable::Create(SharedPtr<Object> obj) { } ResultVal<Handle> HandleTable::Duplicate(Handle handle) { - SharedPtr<Object> object = GetGeneric(handle); + std::shared_ptr<Object> object = GetGeneric(handle); if (object == nullptr) { LOG_ERROR(Kernel, "Tried to duplicate invalid handle: {:08X}", handle); return ERR_INVALID_HANDLE; @@ -99,11 +99,11 @@ bool HandleTable::IsValid(Handle handle) const { return slot < table_size && objects[slot] != nullptr && generations[slot] == generation; } -SharedPtr<Object> HandleTable::GetGeneric(Handle handle) const { +std::shared_ptr<Object> HandleTable::GetGeneric(Handle handle) const { if (handle == CurrentThread) { - return GetCurrentThread(); + return SharedFrom(GetCurrentThread()); } else if (handle == CurrentProcess) { - return Core::System::GetInstance().CurrentProcess(); + return SharedFrom(Core::System::GetInstance().CurrentProcess()); } if (!IsValid(handle)) { diff --git a/src/core/hle/kernel/handle_table.h b/src/core/hle/kernel/handle_table.h index 44901391b..9fcb4cc15 100644 --- a/src/core/hle/kernel/handle_table.h +++ b/src/core/hle/kernel/handle_table.h @@ -68,7 +68,7 @@ public: * @return The created Handle or one of the following errors: * - `ERR_HANDLE_TABLE_FULL`: the maximum number of handles has been exceeded. */ - ResultVal<Handle> Create(SharedPtr<Object> obj); + ResultVal<Handle> Create(std::shared_ptr<Object> obj); /** * Returns a new handle that points to the same object as the passed in handle. @@ -92,7 +92,7 @@ public: * Looks up a handle. * @return Pointer to the looked-up object, or `nullptr` if the handle is not valid. */ - SharedPtr<Object> GetGeneric(Handle handle) const; + std::shared_ptr<Object> GetGeneric(Handle handle) const; /** * Looks up a handle while verifying its type. @@ -100,7 +100,7 @@ public: * type differs from the requested one. */ template <class T> - SharedPtr<T> Get(Handle handle) const { + std::shared_ptr<T> Get(Handle handle) const { return DynamicObjectCast<T>(GetGeneric(handle)); } @@ -109,7 +109,7 @@ public: private: /// Stores the Object referenced by the handle or null if the slot is empty. - std::array<SharedPtr<Object>, MAX_COUNT> objects; + std::array<std::shared_ptr<Object>, MAX_COUNT> objects; /** * The value of `next_generation` when the handle was created, used to check for validity. For diff --git a/src/core/hle/kernel/hle_ipc.cpp b/src/core/hle/kernel/hle_ipc.cpp index a7b5849b0..be24cef06 100644 --- a/src/core/hle/kernel/hle_ipc.cpp +++ b/src/core/hle/kernel/hle_ipc.cpp @@ -32,23 +32,25 @@ SessionRequestHandler::SessionRequestHandler() = default; SessionRequestHandler::~SessionRequestHandler() = default; -void SessionRequestHandler::ClientConnected(SharedPtr<ServerSession> server_session) { +void SessionRequestHandler::ClientConnected(std::shared_ptr<ServerSession> server_session) { server_session->SetHleHandler(shared_from_this()); connected_sessions.push_back(std::move(server_session)); } -void SessionRequestHandler::ClientDisconnected(const SharedPtr<ServerSession>& server_session) { +void SessionRequestHandler::ClientDisconnected( + const std::shared_ptr<ServerSession>& server_session) { server_session->SetHleHandler(nullptr); boost::range::remove_erase(connected_sessions, server_session); } -SharedPtr<WritableEvent> HLERequestContext::SleepClientThread( +std::shared_ptr<WritableEvent> HLERequestContext::SleepClientThread( const std::string& reason, u64 timeout, WakeupCallback&& callback, - SharedPtr<WritableEvent> writable_event) { + std::shared_ptr<WritableEvent> writable_event) { // Put the client thread to sleep until the wait event is signaled or the timeout expires. - thread->SetWakeupCallback([context = *this, callback]( - ThreadWakeupReason reason, SharedPtr<Thread> thread, - SharedPtr<WaitObject> object, std::size_t index) mutable -> bool { + thread->SetWakeupCallback([context = *this, callback](ThreadWakeupReason reason, + std::shared_ptr<Thread> thread, + std::shared_ptr<WaitObject> object, + std::size_t index) mutable -> bool { ASSERT(thread->GetStatus() == ThreadStatus::WaitHLEEvent); callback(thread, context, reason); context.WriteToOutgoingCommandBuffer(*thread); @@ -75,8 +77,8 @@ SharedPtr<WritableEvent> HLERequestContext::SleepClientThread( return writable_event; } -HLERequestContext::HLERequestContext(SharedPtr<Kernel::ServerSession> server_session, - SharedPtr<Thread> thread) +HLERequestContext::HLERequestContext(std::shared_ptr<Kernel::ServerSession> server_session, + std::shared_ptr<Thread> thread) : server_session(std::move(server_session)), thread(std::move(thread)) { cmd_buf[0] = 0; } diff --git a/src/core/hle/kernel/hle_ipc.h b/src/core/hle/kernel/hle_ipc.h index ccf5e56aa..dab37ba0d 100644 --- a/src/core/hle/kernel/hle_ipc.h +++ b/src/core/hle/kernel/hle_ipc.h @@ -5,6 +5,7 @@ #pragma once #include <array> +#include <functional> #include <memory> #include <optional> #include <string> @@ -60,20 +61,20 @@ public: * associated ServerSession alive for the duration of the connection. * @param server_session Owning pointer to the ServerSession associated with the connection. */ - void ClientConnected(SharedPtr<ServerSession> server_session); + void ClientConnected(std::shared_ptr<ServerSession> server_session); /** * Signals that a client has just disconnected from this HLE handler and releases the * associated ServerSession. * @param server_session ServerSession associated with the connection. */ - void ClientDisconnected(const SharedPtr<ServerSession>& server_session); + void ClientDisconnected(const std::shared_ptr<ServerSession>& server_session); protected: /// List of sessions that are connected to this handler. /// A ServerSession whose server endpoint is an HLE implementation is kept alive by this list /// for the duration of the connection. - std::vector<SharedPtr<ServerSession>> connected_sessions; + std::vector<std::shared_ptr<ServerSession>> connected_sessions; }; /** @@ -97,7 +98,8 @@ protected: */ class HLERequestContext { public: - explicit HLERequestContext(SharedPtr<ServerSession> session, SharedPtr<Thread> thread); + explicit HLERequestContext(std::shared_ptr<ServerSession> session, + std::shared_ptr<Thread> thread); ~HLERequestContext(); /// Returns a pointer to the IPC command buffer for this request. @@ -109,12 +111,12 @@ public: * Returns the session through which this request was made. This can be used as a map key to * access per-client data on services. */ - const SharedPtr<Kernel::ServerSession>& Session() const { + const std::shared_ptr<Kernel::ServerSession>& Session() const { return server_session; } - using WakeupCallback = std::function<void(SharedPtr<Thread> thread, HLERequestContext& context, - ThreadWakeupReason reason)>; + using WakeupCallback = std::function<void( + std::shared_ptr<Thread> thread, HLERequestContext& context, ThreadWakeupReason reason)>; /** * Puts the specified guest thread to sleep until the returned event is signaled or until the @@ -129,9 +131,9 @@ public: * created. * @returns Event that when signaled will resume the thread and call the callback function. */ - SharedPtr<WritableEvent> SleepClientThread(const std::string& reason, u64 timeout, - WakeupCallback&& callback, - SharedPtr<WritableEvent> writable_event = nullptr); + std::shared_ptr<WritableEvent> SleepClientThread( + const std::string& reason, u64 timeout, WakeupCallback&& callback, + std::shared_ptr<WritableEvent> writable_event = nullptr); /// Populates this context with data from the requesting process/thread. ResultCode PopulateFromIncomingCommandBuffer(const HandleTable& handle_table, @@ -209,20 +211,20 @@ public: std::size_t GetWriteBufferSize(int buffer_index = 0) const; template <typename T> - SharedPtr<T> GetCopyObject(std::size_t index) { + std::shared_ptr<T> GetCopyObject(std::size_t index) { return DynamicObjectCast<T>(copy_objects.at(index)); } template <typename T> - SharedPtr<T> GetMoveObject(std::size_t index) { + std::shared_ptr<T> GetMoveObject(std::size_t index) { return DynamicObjectCast<T>(move_objects.at(index)); } - void AddMoveObject(SharedPtr<Object> object) { + void AddMoveObject(std::shared_ptr<Object> object) { move_objects.emplace_back(std::move(object)); } - void AddCopyObject(SharedPtr<Object> object) { + void AddCopyObject(std::shared_ptr<Object> object) { copy_objects.emplace_back(std::move(object)); } @@ -266,11 +268,11 @@ private: void ParseCommandBuffer(const HandleTable& handle_table, u32_le* src_cmdbuf, bool incoming); std::array<u32, IPC::COMMAND_BUFFER_LENGTH> cmd_buf; - SharedPtr<Kernel::ServerSession> server_session; - SharedPtr<Thread> thread; + std::shared_ptr<Kernel::ServerSession> server_session; + std::shared_ptr<Thread> thread; // TODO(yuriks): Check common usage of this and optimize size accordingly - boost::container::small_vector<SharedPtr<Object>, 8> move_objects; - boost::container::small_vector<SharedPtr<Object>, 8> copy_objects; + boost::container::small_vector<std::shared_ptr<Object>, 8> move_objects; + boost::container::small_vector<std::shared_ptr<Object>, 8> copy_objects; boost::container::small_vector<std::shared_ptr<SessionRequestHandler>, 8> domain_objects; std::optional<IPC::CommandHeader> command_header; diff --git a/src/core/hle/kernel/kernel.cpp b/src/core/hle/kernel/kernel.cpp index 9d3b309b3..63ad07950 100644 --- a/src/core/hle/kernel/kernel.cpp +++ b/src/core/hle/kernel/kernel.cpp @@ -40,7 +40,7 @@ static void ThreadWakeupCallback(u64 thread_handle, [[maybe_unused]] s64 cycles_ // Lock the global kernel mutex when we enter the kernel HLE. std::lock_guard lock{HLE::g_hle_lock}; - SharedPtr<Thread> thread = + std::shared_ptr<Thread> thread = system.Kernel().RetrieveThreadFromWakeupCallbackHandleTable(proper_handle); if (thread == nullptr) { LOG_CRITICAL(Kernel, "Callback fired for invalid thread {:08X}", proper_handle); @@ -53,7 +53,7 @@ static void ThreadWakeupCallback(u64 thread_handle, [[maybe_unused]] s64 cycles_ thread->GetStatus() == ThreadStatus::WaitHLEEvent) { // Remove the thread from each of its waiting objects' waitlists for (const auto& object : thread->GetWaitObjects()) { - object->RemoveWaitingThread(thread.get()); + object->RemoveWaitingThread(thread); } thread->ClearWaitObjects(); @@ -160,11 +160,11 @@ struct KernelCore::Impl { std::atomic<u64> next_thread_id{1}; // Lists all processes that exist in the current session. - std::vector<SharedPtr<Process>> process_list; + std::vector<std::shared_ptr<Process>> process_list; Process* current_process = nullptr; Kernel::GlobalScheduler global_scheduler; - SharedPtr<ResourceLimit> system_resource_limit; + std::shared_ptr<ResourceLimit> system_resource_limit; Core::Timing::EventType* thread_wakeup_event_type = nullptr; Core::Timing::EventType* preemption_event = nullptr; @@ -193,15 +193,16 @@ void KernelCore::Shutdown() { impl->Shutdown(); } -SharedPtr<ResourceLimit> KernelCore::GetSystemResourceLimit() const { +std::shared_ptr<ResourceLimit> KernelCore::GetSystemResourceLimit() const { return impl->system_resource_limit; } -SharedPtr<Thread> KernelCore::RetrieveThreadFromWakeupCallbackHandleTable(Handle handle) const { +std::shared_ptr<Thread> KernelCore::RetrieveThreadFromWakeupCallbackHandleTable( + Handle handle) const { return impl->thread_wakeup_callback_handle_table.Get<Thread>(handle); } -void KernelCore::AppendNewProcess(SharedPtr<Process> process) { +void KernelCore::AppendNewProcess(std::shared_ptr<Process> process) { impl->process_list.push_back(std::move(process)); } @@ -223,7 +224,7 @@ const Process* KernelCore::CurrentProcess() const { return impl->current_process; } -const std::vector<SharedPtr<Process>>& KernelCore::GetProcessList() const { +const std::vector<std::shared_ptr<Process>>& KernelCore::GetProcessList() const { return impl->process_list; } @@ -235,7 +236,7 @@ const Kernel::GlobalScheduler& KernelCore::GlobalScheduler() const { return impl->global_scheduler; } -void KernelCore::AddNamedPort(std::string name, SharedPtr<ClientPort> port) { +void KernelCore::AddNamedPort(std::string name, std::shared_ptr<ClientPort> port) { impl->named_ports.emplace(std::move(name), std::move(port)); } diff --git a/src/core/hle/kernel/kernel.h b/src/core/hle/kernel/kernel.h index c4397fc77..c74b9078f 100644 --- a/src/core/hle/kernel/kernel.h +++ b/src/core/hle/kernel/kernel.h @@ -6,6 +6,7 @@ #include <string> #include <unordered_map> +#include <vector> #include "core/hle/kernel/object.h" namespace Core { @@ -30,7 +31,7 @@ class Thread; /// Represents a single instance of the kernel. class KernelCore { private: - using NamedPortTable = std::unordered_map<std::string, SharedPtr<ClientPort>>; + using NamedPortTable = std::unordered_map<std::string, std::shared_ptr<ClientPort>>; public: /// Constructs an instance of the kernel using the given System @@ -56,13 +57,13 @@ public: void Shutdown(); /// Retrieves a shared pointer to the system resource limit instance. - SharedPtr<ResourceLimit> GetSystemResourceLimit() const; + std::shared_ptr<ResourceLimit> GetSystemResourceLimit() const; /// Retrieves a shared pointer to a Thread instance within the thread wakeup handle table. - SharedPtr<Thread> RetrieveThreadFromWakeupCallbackHandleTable(Handle handle) const; + std::shared_ptr<Thread> RetrieveThreadFromWakeupCallbackHandleTable(Handle handle) const; /// Adds the given shared pointer to an internal list of active processes. - void AppendNewProcess(SharedPtr<Process> process); + void AppendNewProcess(std::shared_ptr<Process> process); /// Makes the given process the new current process. void MakeCurrentProcess(Process* process); @@ -74,7 +75,7 @@ public: const Process* CurrentProcess() const; /// Retrieves the list of processes. - const std::vector<SharedPtr<Process>>& GetProcessList() const; + const std::vector<std::shared_ptr<Process>>& GetProcessList() const; /// Gets the sole instance of the global scheduler Kernel::GlobalScheduler& GlobalScheduler(); @@ -83,7 +84,7 @@ public: const Kernel::GlobalScheduler& GlobalScheduler() const; /// Adds a port to the named port table - void AddNamedPort(std::string name, SharedPtr<ClientPort> port); + void AddNamedPort(std::string name, std::shared_ptr<ClientPort> port); /// Finds a port within the named port table with the given name. NamedPortTable::iterator FindNamedPort(const std::string& name); diff --git a/src/core/hle/kernel/mutex.cpp b/src/core/hle/kernel/mutex.cpp index 663d0f4b6..8493d0f78 100644 --- a/src/core/hle/kernel/mutex.cpp +++ b/src/core/hle/kernel/mutex.cpp @@ -22,10 +22,10 @@ namespace Kernel { /// Returns the number of threads that are waiting for a mutex, and the highest priority one among /// those. -static std::pair<SharedPtr<Thread>, u32> GetHighestPriorityMutexWaitingThread( - const SharedPtr<Thread>& current_thread, VAddr mutex_addr) { +static std::pair<std::shared_ptr<Thread>, u32> GetHighestPriorityMutexWaitingThread( + const std::shared_ptr<Thread>& current_thread, VAddr mutex_addr) { - SharedPtr<Thread> highest_priority_thread; + std::shared_ptr<Thread> highest_priority_thread; u32 num_waiters = 0; for (const auto& thread : current_thread->GetMutexWaitingThreads()) { @@ -45,14 +45,14 @@ static std::pair<SharedPtr<Thread>, u32> GetHighestPriorityMutexWaitingThread( } /// Update the mutex owner field of all threads waiting on the mutex to point to the new owner. -static void TransferMutexOwnership(VAddr mutex_addr, SharedPtr<Thread> current_thread, - SharedPtr<Thread> new_owner) { +static void TransferMutexOwnership(VAddr mutex_addr, std::shared_ptr<Thread> current_thread, + std::shared_ptr<Thread> new_owner) { const auto threads = current_thread->GetMutexWaitingThreads(); for (const auto& thread : threads) { if (thread->GetMutexWaitAddress() != mutex_addr) continue; - ASSERT(thread->GetLockOwner() == current_thread); + ASSERT(thread->GetLockOwner() == current_thread.get()); current_thread->RemoveMutexWaiter(thread); if (new_owner != thread) new_owner->AddMutexWaiter(thread); @@ -70,9 +70,10 @@ ResultCode Mutex::TryAcquire(VAddr address, Handle holding_thread_handle, } const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable(); - Thread* const current_thread = system.CurrentScheduler().GetCurrentThread(); - SharedPtr<Thread> holding_thread = handle_table.Get<Thread>(holding_thread_handle); - SharedPtr<Thread> requesting_thread = handle_table.Get<Thread>(requesting_thread_handle); + std::shared_ptr<Thread> current_thread = + SharedFrom(system.CurrentScheduler().GetCurrentThread()); + std::shared_ptr<Thread> holding_thread = handle_table.Get<Thread>(holding_thread_handle); + std::shared_ptr<Thread> requesting_thread = handle_table.Get<Thread>(requesting_thread_handle); // TODO(Subv): It is currently unknown if it is possible to lock a mutex in behalf of another // thread. @@ -110,7 +111,8 @@ ResultCode Mutex::Release(VAddr address) { return ERR_INVALID_ADDRESS; } - auto* const current_thread = system.CurrentScheduler().GetCurrentThread(); + std::shared_ptr<Thread> current_thread = + SharedFrom(system.CurrentScheduler().GetCurrentThread()); auto [thread, num_waiters] = GetHighestPriorityMutexWaitingThread(current_thread, address); // There are no more threads waiting for the mutex, release it completely. diff --git a/src/core/hle/kernel/object.h b/src/core/hle/kernel/object.h index a6faeb83b..bbbb4e7cc 100644 --- a/src/core/hle/kernel/object.h +++ b/src/core/hle/kernel/object.h @@ -5,10 +5,9 @@ #pragma once #include <atomic> +#include <memory> #include <string> -#include <boost/smart_ptr/intrusive_ptr.hpp> - #include "common/common_types.h" namespace Kernel { @@ -32,7 +31,7 @@ enum class HandleType : u32 { ServerSession, }; -class Object : NonCopyable { +class Object : NonCopyable, public std::enable_shared_from_this<Object> { public: explicit Object(KernelCore& kernel); virtual ~Object(); @@ -61,35 +60,24 @@ protected: KernelCore& kernel; private: - friend void intrusive_ptr_add_ref(Object*); - friend void intrusive_ptr_release(Object*); - - std::atomic<u32> ref_count{0}; std::atomic<u32> object_id{0}; }; -// Special functions used by boost::instrusive_ptr to do automatic ref-counting -inline void intrusive_ptr_add_ref(Object* object) { - object->ref_count.fetch_add(1, std::memory_order_relaxed); -} - -inline void intrusive_ptr_release(Object* object) { - if (object->ref_count.fetch_sub(1, std::memory_order_acq_rel) == 1) { - delete object; - } -} - template <typename T> -using SharedPtr = boost::intrusive_ptr<T>; +std::shared_ptr<T> SharedFrom(T* raw) { + if (raw == nullptr) + return nullptr; + return std::static_pointer_cast<T>(raw->shared_from_this()); +} /** * Attempts to downcast the given Object pointer to a pointer to T. * @return Derived pointer to the object, or `nullptr` if `object` isn't of type T. */ template <typename T> -inline SharedPtr<T> DynamicObjectCast(SharedPtr<Object> object) { +inline std::shared_ptr<T> DynamicObjectCast(std::shared_ptr<Object> object) { if (object != nullptr && object->GetHandleType() == T::HANDLE_TYPE) { - return boost::static_pointer_cast<T>(object); + return std::static_pointer_cast<T>(object); } return nullptr; } diff --git a/src/core/hle/kernel/process.cpp b/src/core/hle/kernel/process.cpp index a4e0dd385..12ea4ebe3 100644 --- a/src/core/hle/kernel/process.cpp +++ b/src/core/hle/kernel/process.cpp @@ -38,7 +38,7 @@ void SetupMainThread(Process& owner_process, KernelCore& kernel, u32 priority) { auto thread_res = Thread::Create(kernel, "main", entry_point, priority, 0, owner_process.GetIdealCore(), stack_top, owner_process); - SharedPtr<Thread> thread = std::move(thread_res).Unwrap(); + std::shared_ptr<Thread> thread = std::move(thread_res).Unwrap(); // Register 1 must be a handle to the main thread const Handle thread_handle = owner_process.GetHandleTable().Create(thread).Unwrap(); @@ -100,10 +100,10 @@ private: std::bitset<num_slot_entries> is_slot_used; }; -SharedPtr<Process> Process::Create(Core::System& system, std::string name, ProcessType type) { +std::shared_ptr<Process> Process::Create(Core::System& system, std::string name, ProcessType type) { auto& kernel = system.Kernel(); - SharedPtr<Process> process(new Process(system)); + std::shared_ptr<Process> process = std::make_shared<Process>(system); process->name = std::move(name); process->resource_limit = kernel.GetSystemResourceLimit(); process->status = ProcessStatus::Created; @@ -121,7 +121,7 @@ SharedPtr<Process> Process::Create(Core::System& system, std::string name, Proce return process; } -SharedPtr<ResourceLimit> Process::GetResourceLimit() const { +std::shared_ptr<ResourceLimit> Process::GetResourceLimit() const { return resource_limit; } @@ -142,12 +142,12 @@ u64 Process::GetTotalPhysicalMemoryUsedWithoutSystemResource() const { return GetTotalPhysicalMemoryUsed() - GetSystemResourceUsage(); } -void Process::InsertConditionVariableThread(SharedPtr<Thread> thread) { +void Process::InsertConditionVariableThread(std::shared_ptr<Thread> thread) { VAddr cond_var_addr = thread->GetCondVarWaitAddress(); - std::list<SharedPtr<Thread>>& thread_list = cond_var_threads[cond_var_addr]; + std::list<std::shared_ptr<Thread>>& thread_list = cond_var_threads[cond_var_addr]; auto it = thread_list.begin(); while (it != thread_list.end()) { - const SharedPtr<Thread> current_thread = *it; + const std::shared_ptr<Thread> current_thread = *it; if (current_thread->GetPriority() > thread->GetPriority()) { thread_list.insert(it, thread); return; @@ -157,12 +157,12 @@ void Process::InsertConditionVariableThread(SharedPtr<Thread> thread) { thread_list.push_back(thread); } -void Process::RemoveConditionVariableThread(SharedPtr<Thread> thread) { +void Process::RemoveConditionVariableThread(std::shared_ptr<Thread> thread) { VAddr cond_var_addr = thread->GetCondVarWaitAddress(); - std::list<SharedPtr<Thread>>& thread_list = cond_var_threads[cond_var_addr]; + std::list<std::shared_ptr<Thread>>& thread_list = cond_var_threads[cond_var_addr]; auto it = thread_list.begin(); while (it != thread_list.end()) { - const SharedPtr<Thread> current_thread = *it; + const std::shared_ptr<Thread> current_thread = *it; if (current_thread.get() == thread.get()) { thread_list.erase(it); return; @@ -172,12 +172,13 @@ void Process::RemoveConditionVariableThread(SharedPtr<Thread> thread) { UNREACHABLE(); } -std::vector<SharedPtr<Thread>> Process::GetConditionVariableThreads(const VAddr cond_var_addr) { - std::vector<SharedPtr<Thread>> result{}; - std::list<SharedPtr<Thread>>& thread_list = cond_var_threads[cond_var_addr]; +std::vector<std::shared_ptr<Thread>> Process::GetConditionVariableThreads( + const VAddr cond_var_addr) { + std::vector<std::shared_ptr<Thread>> result{}; + std::list<std::shared_ptr<Thread>>& thread_list = cond_var_threads[cond_var_addr]; auto it = thread_list.begin(); while (it != thread_list.end()) { - SharedPtr<Thread> current_thread = *it; + std::shared_ptr<Thread> current_thread = *it; result.push_back(current_thread); ++it; } @@ -239,12 +240,12 @@ void Process::Run(s32 main_thread_priority, u64 stack_size) { void Process::PrepareForTermination() { ChangeStatus(ProcessStatus::Exiting); - const auto stop_threads = [this](const std::vector<SharedPtr<Thread>>& thread_list) { + const auto stop_threads = [this](const std::vector<std::shared_ptr<Thread>>& thread_list) { for (auto& thread : thread_list) { if (thread->GetOwnerProcess() != this) continue; - if (thread == system.CurrentScheduler().GetCurrentThread()) + if (thread.get() == system.CurrentScheduler().GetCurrentThread()) continue; // TODO(Subv): When are the other running/ready threads terminated? diff --git a/src/core/hle/kernel/process.h b/src/core/hle/kernel/process.h index e2eda26b9..3483fa19d 100644 --- a/src/core/hle/kernel/process.h +++ b/src/core/hle/kernel/process.h @@ -62,6 +62,9 @@ enum class ProcessStatus { class Process final : public WaitObject { public: + explicit Process(Core::System& system); + ~Process() override; + enum : u64 { /// Lowest allowed process ID for a kernel initial process. InitialKIPIDMin = 1, @@ -82,7 +85,8 @@ public: static constexpr std::size_t RANDOM_ENTROPY_SIZE = 4; - static SharedPtr<Process> Create(Core::System& system, std::string name, ProcessType type); + static std::shared_ptr<Process> Create(Core::System& system, std::string name, + ProcessType type); std::string GetTypeName() const override { return "Process"; @@ -157,7 +161,7 @@ public: } /// Gets the resource limit descriptor for this process - SharedPtr<ResourceLimit> GetResourceLimit() const; + std::shared_ptr<ResourceLimit> GetResourceLimit() const; /// Gets the ideal CPU core ID for this process u8 GetIdealCore() const { @@ -234,13 +238,13 @@ public: } /// Insert a thread into the condition variable wait container - void InsertConditionVariableThread(SharedPtr<Thread> thread); + void InsertConditionVariableThread(std::shared_ptr<Thread> thread); /// Remove a thread from the condition variable wait container - void RemoveConditionVariableThread(SharedPtr<Thread> thread); + void RemoveConditionVariableThread(std::shared_ptr<Thread> thread); /// Obtain all condition variable threads waiting for some address - std::vector<SharedPtr<Thread>> GetConditionVariableThreads(VAddr cond_var_addr); + std::vector<std::shared_ptr<Thread>> GetConditionVariableThreads(VAddr cond_var_addr); /// Registers a thread as being created under this process, /// adding it to this process' thread list. @@ -297,9 +301,6 @@ public: void FreeTLSRegion(VAddr tls_address); private: - explicit Process(Core::System& system); - ~Process() override; - /// Checks if the specified thread should wait until this process is available. bool ShouldWait(const Thread* thread) const override; @@ -338,7 +339,7 @@ private: u32 system_resource_size = 0; /// Resource limit descriptor for this process - SharedPtr<ResourceLimit> resource_limit; + std::shared_ptr<ResourceLimit> resource_limit; /// The ideal CPU core for this process, threads are scheduled on this core by default. u8 ideal_core = 0; @@ -386,7 +387,7 @@ private: std::list<const Thread*> thread_list; /// List of threads waiting for a condition variable - std::unordered_map<VAddr, std::list<SharedPtr<Thread>>> cond_var_threads; + std::unordered_map<VAddr, std::list<std::shared_ptr<Thread>>> cond_var_threads; /// System context Core::System& system; diff --git a/src/core/hle/kernel/resource_limit.cpp b/src/core/hle/kernel/resource_limit.cpp index 173f69915..b53423462 100644 --- a/src/core/hle/kernel/resource_limit.cpp +++ b/src/core/hle/kernel/resource_limit.cpp @@ -16,8 +16,8 @@ constexpr std::size_t ResourceTypeToIndex(ResourceType type) { ResourceLimit::ResourceLimit(KernelCore& kernel) : Object{kernel} {} ResourceLimit::~ResourceLimit() = default; -SharedPtr<ResourceLimit> ResourceLimit::Create(KernelCore& kernel) { - return new ResourceLimit(kernel); +std::shared_ptr<ResourceLimit> ResourceLimit::Create(KernelCore& kernel) { + return std::make_shared<ResourceLimit>(kernel); } s64 ResourceLimit::GetCurrentResourceValue(ResourceType resource) const { diff --git a/src/core/hle/kernel/resource_limit.h b/src/core/hle/kernel/resource_limit.h index 2613a6bb5..b5534620d 100644 --- a/src/core/hle/kernel/resource_limit.h +++ b/src/core/hle/kernel/resource_limit.h @@ -31,8 +31,11 @@ constexpr bool IsValidResourceType(ResourceType type) { class ResourceLimit final : public Object { public: + explicit ResourceLimit(KernelCore& kernel); + ~ResourceLimit() override; + /// Creates a resource limit object. - static SharedPtr<ResourceLimit> Create(KernelCore& kernel); + static std::shared_ptr<ResourceLimit> Create(KernelCore& kernel); std::string GetTypeName() const override { return "ResourceLimit"; @@ -76,9 +79,6 @@ public: ResultCode SetLimitValue(ResourceType resource, s64 value); private: - explicit ResourceLimit(KernelCore& kernel); - ~ResourceLimit() override; - // TODO(Subv): Increment resource limit current values in their respective Kernel::T::Create // functions // diff --git a/src/core/hle/kernel/scheduler.cpp b/src/core/hle/kernel/scheduler.cpp index 16e95381b..3f5192087 100644 --- a/src/core/hle/kernel/scheduler.cpp +++ b/src/core/hle/kernel/scheduler.cpp @@ -26,11 +26,11 @@ GlobalScheduler::GlobalScheduler(Core::System& system) : system{system} {} GlobalScheduler::~GlobalScheduler() = default; -void GlobalScheduler::AddThread(SharedPtr<Thread> thread) { +void GlobalScheduler::AddThread(std::shared_ptr<Thread> thread) { thread_list.push_back(std::move(thread)); } -void GlobalScheduler::RemoveThread(const Thread* thread) { +void GlobalScheduler::RemoveThread(std::shared_ptr<Thread> thread) { thread_list.erase(std::remove(thread_list.begin(), thread_list.end(), thread), thread_list.end()); } @@ -42,11 +42,11 @@ void GlobalScheduler::UnloadThread(std::size_t core) { void GlobalScheduler::SelectThread(std::size_t core) { const auto update_thread = [](Thread* thread, Scheduler& sched) { - if (thread != sched.selected_thread) { + if (thread != sched.selected_thread.get()) { if (thread == nullptr) { ++sched.idle_selection_count; } - sched.selected_thread = thread; + sched.selected_thread = SharedFrom(thread); } sched.is_context_switch_pending = sched.selected_thread != sched.current_thread; std::atomic_thread_fence(std::memory_order_seq_cst); @@ -446,7 +446,7 @@ void Scheduler::SwitchContext() { // Cancel any outstanding wakeup events for this thread new_thread->CancelWakeupTimer(); - current_thread = new_thread; + current_thread = SharedFrom(new_thread); new_thread->SetStatus(ThreadStatus::Running); new_thread->SetIsRunning(true); diff --git a/src/core/hle/kernel/scheduler.h b/src/core/hle/kernel/scheduler.h index 311849dfb..3c5c21346 100644 --- a/src/core/hle/kernel/scheduler.h +++ b/src/core/hle/kernel/scheduler.h @@ -28,13 +28,13 @@ public: ~GlobalScheduler(); /// Adds a new thread to the scheduler - void AddThread(SharedPtr<Thread> thread); + void AddThread(std::shared_ptr<Thread> thread); /// Removes a thread from the scheduler - void RemoveThread(const Thread* thread); + void RemoveThread(std::shared_ptr<Thread> thread); /// Returns a list of all threads managed by the scheduler - const std::vector<SharedPtr<Thread>>& GetThreadList() const { + const std::vector<std::shared_ptr<Thread>>& GetThreadList() const { return thread_list; } @@ -157,7 +157,7 @@ private: std::array<u32, NUM_CPU_CORES> preemption_priorities = {59, 59, 59, 62}; /// Lists all thread ids that aren't deleted/etc. - std::vector<SharedPtr<Thread>> thread_list; + std::vector<std::shared_ptr<Thread>> thread_list; Core::System& system; }; @@ -213,8 +213,8 @@ private: */ void UpdateLastContextSwitchTime(Thread* thread, Process* process); - SharedPtr<Thread> current_thread = nullptr; - SharedPtr<Thread> selected_thread = nullptr; + std::shared_ptr<Thread> current_thread = nullptr; + std::shared_ptr<Thread> selected_thread = nullptr; Core::System& system; Core::ARM_Interface& cpu_core; diff --git a/src/core/hle/kernel/server_port.cpp b/src/core/hle/kernel/server_port.cpp index 02e7c60e6..a4ccfa35e 100644 --- a/src/core/hle/kernel/server_port.cpp +++ b/src/core/hle/kernel/server_port.cpp @@ -16,7 +16,7 @@ namespace Kernel { ServerPort::ServerPort(KernelCore& kernel) : WaitObject{kernel} {} ServerPort::~ServerPort() = default; -ResultVal<SharedPtr<ServerSession>> ServerPort::Accept() { +ResultVal<std::shared_ptr<ServerSession>> ServerPort::Accept() { if (pending_sessions.empty()) { return ERR_NOT_FOUND; } @@ -26,7 +26,7 @@ ResultVal<SharedPtr<ServerSession>> ServerPort::Accept() { return MakeResult(std::move(session)); } -void ServerPort::AppendPendingSession(SharedPtr<ServerSession> pending_session) { +void ServerPort::AppendPendingSession(std::shared_ptr<ServerSession> pending_session) { pending_sessions.push_back(std::move(pending_session)); } @@ -41,8 +41,8 @@ void ServerPort::Acquire(Thread* thread) { ServerPort::PortPair ServerPort::CreatePortPair(KernelCore& kernel, u32 max_sessions, std::string name) { - SharedPtr<ServerPort> server_port(new ServerPort(kernel)); - SharedPtr<ClientPort> client_port(new ClientPort(kernel)); + std::shared_ptr<ServerPort> server_port = std::make_shared<ServerPort>(kernel); + std::shared_ptr<ClientPort> client_port = std::make_shared<ClientPort>(kernel); server_port->name = name + "_Server"; client_port->name = name + "_Client"; diff --git a/src/core/hle/kernel/server_port.h b/src/core/hle/kernel/server_port.h index dc88a1ebd..8be8a75ea 100644 --- a/src/core/hle/kernel/server_port.h +++ b/src/core/hle/kernel/server_port.h @@ -22,8 +22,11 @@ class SessionRequestHandler; class ServerPort final : public WaitObject { public: + explicit ServerPort(KernelCore& kernel); + ~ServerPort() override; + using HLEHandler = std::shared_ptr<SessionRequestHandler>; - using PortPair = std::pair<SharedPtr<ServerPort>, SharedPtr<ClientPort>>; + using PortPair = std::pair<std::shared_ptr<ServerPort>, std::shared_ptr<ClientPort>>; /** * Creates a pair of ServerPort and an associated ClientPort. @@ -52,7 +55,7 @@ public: * Accepts a pending incoming connection on this port. If there are no pending sessions, will * return ERR_NO_PENDING_SESSIONS. */ - ResultVal<SharedPtr<ServerSession>> Accept(); + ResultVal<std::shared_ptr<ServerSession>> Accept(); /// Whether or not this server port has an HLE handler available. bool HasHLEHandler() const { @@ -74,17 +77,14 @@ public: /// Appends a ServerSession to the collection of ServerSessions /// waiting to be accepted by this port. - void AppendPendingSession(SharedPtr<ServerSession> pending_session); + void AppendPendingSession(std::shared_ptr<ServerSession> pending_session); bool ShouldWait(const Thread* thread) const override; void Acquire(Thread* thread) override; private: - explicit ServerPort(KernelCore& kernel); - ~ServerPort() override; - /// ServerSessions waiting to be accepted by the port - std::vector<SharedPtr<ServerSession>> pending_sessions; + std::vector<std::shared_ptr<ServerSession>> pending_sessions; /// This session's HLE request handler template (optional) /// ServerSessions created from this port inherit a reference to this handler. diff --git a/src/core/hle/kernel/server_session.cpp b/src/core/hle/kernel/server_session.cpp index 30b2bfb5a..2994fa0ac 100644 --- a/src/core/hle/kernel/server_session.cpp +++ b/src/core/hle/kernel/server_session.cpp @@ -35,8 +35,9 @@ ServerSession::~ServerSession() { parent->server = nullptr; } -ResultVal<SharedPtr<ServerSession>> ServerSession::Create(KernelCore& kernel, std::string name) { - SharedPtr<ServerSession> server_session(new ServerSession(kernel)); +ResultVal<std::shared_ptr<ServerSession>> ServerSession::Create(KernelCore& kernel, + std::string name) { + std::shared_ptr<ServerSession> server_session = std::make_shared<ServerSession>(kernel); server_session->name = std::move(name); server_session->parent = nullptr; @@ -69,7 +70,7 @@ void ServerSession::ClientDisconnected() { if (handler) { // Note that after this returns, this server session's hle_handler is // invalidated (set to null). - handler->ClientDisconnected(this); + handler->ClientDisconnected(SharedFrom(this)); } // Clean up the list of client threads with pending requests, they are unneeded now that the @@ -126,11 +127,11 @@ ResultCode ServerSession::HandleDomainSyncRequest(Kernel::HLERequestContext& con return RESULT_SUCCESS; } -ResultCode ServerSession::HandleSyncRequest(SharedPtr<Thread> thread) { +ResultCode ServerSession::HandleSyncRequest(std::shared_ptr<Thread> thread) { // The ServerSession received a sync request, this means that there's new data available // from its ClientSession, so wake up any threads that may be waiting on a svcReplyAndReceive or // similar. - Kernel::HLERequestContext context(this, thread); + Kernel::HLERequestContext context(SharedFrom(this), thread); u32* cmd_buf = (u32*)Memory::GetPointer(thread->GetTLSAddress()); context.PopulateFromIncomingCommandBuffer(kernel.CurrentProcess()->GetHandleTable(), cmd_buf); @@ -186,9 +187,9 @@ ResultCode ServerSession::HandleSyncRequest(SharedPtr<Thread> thread) { ServerSession::SessionPair ServerSession::CreateSessionPair(KernelCore& kernel, const std::string& name, - SharedPtr<ClientPort> port) { + std::shared_ptr<ClientPort> port) { auto server_session = ServerSession::Create(kernel, name + "_Server").Unwrap(); - SharedPtr<ClientSession> client_session(new ClientSession(kernel)); + std::shared_ptr<ClientSession> client_session = std::make_shared<ClientSession>(kernel); client_session->name = name + "_Client"; std::shared_ptr<Session> parent(new Session); diff --git a/src/core/hle/kernel/server_session.h b/src/core/hle/kernel/server_session.h index 738df30f8..8a65647b6 100644 --- a/src/core/hle/kernel/server_session.h +++ b/src/core/hle/kernel/server_session.h @@ -38,6 +38,9 @@ class Thread; */ class ServerSession final : public WaitObject { public: + explicit ServerSession(KernelCore& kernel); + ~ServerSession() override; + std::string GetTypeName() const override { return "ServerSession"; } @@ -59,7 +62,7 @@ public: return parent.get(); } - using SessionPair = std::pair<SharedPtr<ServerSession>, SharedPtr<ClientSession>>; + using SessionPair = std::pair<std::shared_ptr<ServerSession>, std::shared_ptr<ClientSession>>; /** * Creates a pair of ServerSession and an associated ClientSession. @@ -69,7 +72,7 @@ public: * @return The created session tuple */ static SessionPair CreateSessionPair(KernelCore& kernel, const std::string& name = "Unknown", - SharedPtr<ClientPort> client_port = nullptr); + std::shared_ptr<ClientPort> client_port = nullptr); /** * Sets the HLE handler for the session. This handler will be called to service IPC requests @@ -85,7 +88,7 @@ public: * @param thread Thread that initiated the request. * @returns ResultCode from the operation. */ - ResultCode HandleSyncRequest(SharedPtr<Thread> thread); + ResultCode HandleSyncRequest(std::shared_ptr<Thread> thread); bool ShouldWait(const Thread* thread) const override; @@ -118,9 +121,6 @@ public: } private: - explicit ServerSession(KernelCore& kernel); - ~ServerSession() override; - /** * Creates a server session. The server session can have an optional HLE handler, * which will be invoked to handle the IPC requests that this session receives. @@ -128,8 +128,8 @@ private: * @param name Optional name of the server session. * @return The created server session */ - static ResultVal<SharedPtr<ServerSession>> Create(KernelCore& kernel, - std::string name = "Unknown"); + static ResultVal<std::shared_ptr<ServerSession>> Create(KernelCore& kernel, + std::string name = "Unknown"); /// Handles a SyncRequest to a domain, forwarding the request to the proper object or closing an /// object handle. @@ -147,12 +147,12 @@ private: /// List of threads that are pending a response after a sync request. This list is processed in /// a LIFO manner, thus, the last request will be dispatched first. /// TODO(Subv): Verify if this is indeed processed in LIFO using a hardware test. - std::vector<SharedPtr<Thread>> pending_requesting_threads; + std::vector<std::shared_ptr<Thread>> pending_requesting_threads; /// Thread whose request is currently being handled. A request is considered "handled" when a /// response is sent via svcReplyAndReceive. /// TODO(Subv): Find a better name for this. - SharedPtr<Thread> currently_handling; + std::shared_ptr<Thread> currently_handling; /// When set to True, converts the session to a domain at the end of the command bool convert_to_domain{}; diff --git a/src/core/hle/kernel/session.h b/src/core/hle/kernel/session.h index 7a551f5e4..ea956813b 100644 --- a/src/core/hle/kernel/session.h +++ b/src/core/hle/kernel/session.h @@ -20,8 +20,8 @@ class ServerSession; */ class Session final { public: - ClientSession* client = nullptr; ///< The client endpoint of the session. - ServerSession* server = nullptr; ///< The server endpoint of the session. - SharedPtr<ClientPort> port; ///< The port that this session is associated with (optional). + ClientSession* client = nullptr; ///< The client endpoint of the session. + ServerSession* server = nullptr; ///< The server endpoint of the session. + std::shared_ptr<ClientPort> port; ///< The port that this session is associated with (optional). }; } // namespace Kernel diff --git a/src/core/hle/kernel/shared_memory.cpp b/src/core/hle/kernel/shared_memory.cpp index a815c4eea..afb2e3fc2 100644 --- a/src/core/hle/kernel/shared_memory.cpp +++ b/src/core/hle/kernel/shared_memory.cpp @@ -15,11 +15,12 @@ namespace Kernel { SharedMemory::SharedMemory(KernelCore& kernel) : Object{kernel} {} SharedMemory::~SharedMemory() = default; -SharedPtr<SharedMemory> SharedMemory::Create(KernelCore& kernel, Process* owner_process, u64 size, - MemoryPermission permissions, - MemoryPermission other_permissions, VAddr address, - MemoryRegion region, std::string name) { - SharedPtr<SharedMemory> shared_memory(new SharedMemory(kernel)); +std::shared_ptr<SharedMemory> SharedMemory::Create(KernelCore& kernel, Process* owner_process, + u64 size, MemoryPermission permissions, + MemoryPermission other_permissions, + VAddr address, MemoryRegion region, + std::string name) { + std::shared_ptr<SharedMemory> shared_memory = std::make_shared<SharedMemory>(kernel); shared_memory->owner_process = owner_process; shared_memory->name = std::move(name); @@ -58,10 +59,10 @@ SharedPtr<SharedMemory> SharedMemory::Create(KernelCore& kernel, Process* owner_ return shared_memory; } -SharedPtr<SharedMemory> SharedMemory::CreateForApplet( +std::shared_ptr<SharedMemory> SharedMemory::CreateForApplet( KernelCore& kernel, std::shared_ptr<Kernel::PhysicalMemory> heap_block, std::size_t offset, u64 size, MemoryPermission permissions, MemoryPermission other_permissions, std::string name) { - SharedPtr<SharedMemory> shared_memory(new SharedMemory(kernel)); + std::shared_ptr<SharedMemory> shared_memory = std::make_shared<SharedMemory>(kernel); shared_memory->owner_process = nullptr; shared_memory->name = std::move(name); diff --git a/src/core/hle/kernel/shared_memory.h b/src/core/hle/kernel/shared_memory.h index 01ca6dcd2..18400a5ad 100644 --- a/src/core/hle/kernel/shared_memory.h +++ b/src/core/hle/kernel/shared_memory.h @@ -33,6 +33,9 @@ enum class MemoryPermission : u32 { class SharedMemory final : public Object { public: + explicit SharedMemory(KernelCore& kernel); + ~SharedMemory() override; + /** * Creates a shared memory object. * @param kernel The kernel instance to create a shared memory instance under. @@ -46,11 +49,12 @@ public: * linear heap. * @param name Optional object name, used for debugging purposes. */ - static SharedPtr<SharedMemory> Create(KernelCore& kernel, Process* owner_process, u64 size, - MemoryPermission permissions, - MemoryPermission other_permissions, VAddr address = 0, - MemoryRegion region = MemoryRegion::BASE, - std::string name = "Unknown"); + static std::shared_ptr<SharedMemory> Create(KernelCore& kernel, Process* owner_process, + u64 size, MemoryPermission permissions, + MemoryPermission other_permissions, + VAddr address = 0, + MemoryRegion region = MemoryRegion::BASE, + std::string name = "Unknown"); /** * Creates a shared memory object from a block of memory managed by an HLE applet. @@ -63,7 +67,7 @@ public: * block. * @param name Optional object name, used for debugging purposes. */ - static SharedPtr<SharedMemory> CreateForApplet( + static std::shared_ptr<SharedMemory> CreateForApplet( KernelCore& kernel, std::shared_ptr<Kernel::PhysicalMemory> heap_block, std::size_t offset, u64 size, MemoryPermission permissions, MemoryPermission other_permissions, std::string name = "Unknown Applet"); @@ -130,9 +134,6 @@ public: const u8* GetPointer(std::size_t offset = 0) const; private: - explicit SharedMemory(KernelCore& kernel); - ~SharedMemory() override; - /// Backing memory for this shared memory block. std::shared_ptr<PhysicalMemory> backing_block; /// Offset into the backing block for this shared memory. diff --git a/src/core/hle/kernel/svc.cpp b/src/core/hle/kernel/svc.cpp index 8c67ada43..9928b3a26 100644 --- a/src/core/hle/kernel/svc.cpp +++ b/src/core/hle/kernel/svc.cpp @@ -17,6 +17,7 @@ #include "core/core.h" #include "core/core_cpu.h" #include "core/core_timing.h" +#include "core/core_timing_util.h" #include "core/hle/kernel/address_arbiter.h" #include "core/hle/kernel/client_port.h" #include "core/hle/kernel/client_session.h" @@ -358,7 +359,7 @@ static ResultCode ConnectToNamedPort(Core::System& system, Handle* out_handle, auto client_port = it->second; - SharedPtr<ClientSession> client_session; + std::shared_ptr<ClientSession> client_session; CASCADE_RESULT(client_session, client_port->Connect()); // Return the client session @@ -370,7 +371,7 @@ static ResultCode ConnectToNamedPort(Core::System& system, Handle* out_handle, /// Makes a blocking IPC call to an OS service. static ResultCode SendSyncRequest(Core::System& system, Handle handle) { const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable(); - SharedPtr<ClientSession> session = handle_table.Get<ClientSession>(handle); + std::shared_ptr<ClientSession> session = handle_table.Get<ClientSession>(handle); if (!session) { LOG_ERROR(Kernel_SVC, "called with invalid handle=0x{:08X}", handle); return ERR_INVALID_HANDLE; @@ -390,7 +391,7 @@ static ResultCode GetThreadId(Core::System& system, u64* thread_id, Handle threa LOG_TRACE(Kernel_SVC, "called thread=0x{:08X}", thread_handle); const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable(); - const SharedPtr<Thread> thread = handle_table.Get<Thread>(thread_handle); + const std::shared_ptr<Thread> thread = handle_table.Get<Thread>(thread_handle); if (!thread) { LOG_ERROR(Kernel_SVC, "Thread handle does not exist, handle=0x{:08X}", thread_handle); return ERR_INVALID_HANDLE; @@ -405,13 +406,13 @@ static ResultCode GetProcessId(Core::System& system, u64* process_id, Handle han LOG_DEBUG(Kernel_SVC, "called handle=0x{:08X}", handle); const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable(); - const SharedPtr<Process> process = handle_table.Get<Process>(handle); + const std::shared_ptr<Process> process = handle_table.Get<Process>(handle); if (process) { *process_id = process->GetProcessID(); return RESULT_SUCCESS; } - const SharedPtr<Thread> thread = handle_table.Get<Thread>(handle); + const std::shared_ptr<Thread> thread = handle_table.Get<Thread>(handle); if (thread) { const Process* const owner_process = thread->GetOwnerProcess(); if (!owner_process) { @@ -430,8 +431,8 @@ static ResultCode GetProcessId(Core::System& system, u64* process_id, Handle han } /// Default thread wakeup callback for WaitSynchronization -static bool DefaultThreadWakeupCallback(ThreadWakeupReason reason, SharedPtr<Thread> thread, - SharedPtr<WaitObject> object, std::size_t index) { +static bool DefaultThreadWakeupCallback(ThreadWakeupReason reason, std::shared_ptr<Thread> thread, + std::shared_ptr<WaitObject> object, std::size_t index) { ASSERT(thread->GetStatus() == ThreadStatus::WaitSynch); if (reason == ThreadWakeupReason::Timeout) { @@ -511,7 +512,7 @@ static ResultCode WaitSynchronization(Core::System& system, Handle* index, VAddr } for (auto& object : objects) { - object->AddWaitingThread(thread); + object->AddWaitingThread(SharedFrom(thread)); } thread->SetWaitObjects(std::move(objects)); @@ -531,7 +532,7 @@ static ResultCode CancelSynchronization(Core::System& system, Handle thread_hand LOG_TRACE(Kernel_SVC, "called thread=0x{:X}", thread_handle); const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable(); - SharedPtr<Thread> thread = handle_table.Get<Thread>(thread_handle); + std::shared_ptr<Thread> thread = handle_table.Get<Thread>(thread_handle); if (!thread) { LOG_ERROR(Kernel_SVC, "Thread handle does not exist, thread_handle=0x{:08X}", thread_handle); @@ -940,7 +941,7 @@ static ResultCode GetInfo(Core::System& system, u64* result, u64 info_id, u64 ha const auto& core_timing = system.CoreTiming(); const auto& scheduler = system.CurrentScheduler(); const auto* const current_thread = scheduler.GetCurrentThread(); - const bool same_thread = current_thread == thread; + const bool same_thread = current_thread == thread.get(); const u64 prev_ctx_ticks = scheduler.GetLastContextSwitchTicks(); u64 out_ticks = 0; @@ -1050,7 +1051,7 @@ static ResultCode SetThreadActivity(Core::System& system, Handle handle, u32 act } const auto* current_process = system.Kernel().CurrentProcess(); - const SharedPtr<Thread> thread = current_process->GetHandleTable().Get<Thread>(handle); + const std::shared_ptr<Thread> thread = current_process->GetHandleTable().Get<Thread>(handle); if (!thread) { LOG_ERROR(Kernel_SVC, "Thread handle does not exist, handle=0x{:08X}", handle); return ERR_INVALID_HANDLE; @@ -1066,7 +1067,7 @@ static ResultCode SetThreadActivity(Core::System& system, Handle handle, u32 act return ERR_INVALID_HANDLE; } - if (thread == system.CurrentScheduler().GetCurrentThread()) { + if (thread.get() == system.CurrentScheduler().GetCurrentThread()) { LOG_ERROR(Kernel_SVC, "The thread handle specified is the current running thread"); return ERR_BUSY; } @@ -1082,7 +1083,7 @@ static ResultCode GetThreadContext(Core::System& system, VAddr thread_context, H LOG_DEBUG(Kernel_SVC, "called, context=0x{:08X}, thread=0x{:X}", thread_context, handle); const auto* current_process = system.Kernel().CurrentProcess(); - const SharedPtr<Thread> thread = current_process->GetHandleTable().Get<Thread>(handle); + const std::shared_ptr<Thread> thread = current_process->GetHandleTable().Get<Thread>(handle); if (!thread) { LOG_ERROR(Kernel_SVC, "Thread handle does not exist, handle=0x{:08X}", handle); return ERR_INVALID_HANDLE; @@ -1098,7 +1099,7 @@ static ResultCode GetThreadContext(Core::System& system, VAddr thread_context, H return ERR_INVALID_HANDLE; } - if (thread == system.CurrentScheduler().GetCurrentThread()) { + if (thread.get() == system.CurrentScheduler().GetCurrentThread()) { LOG_ERROR(Kernel_SVC, "The thread handle specified is the current running thread"); return ERR_BUSY; } @@ -1123,7 +1124,7 @@ static ResultCode GetThreadPriority(Core::System& system, u32* priority, Handle LOG_TRACE(Kernel_SVC, "called"); const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable(); - const SharedPtr<Thread> thread = handle_table.Get<Thread>(handle); + const std::shared_ptr<Thread> thread = handle_table.Get<Thread>(handle); if (!thread) { LOG_ERROR(Kernel_SVC, "Thread handle does not exist, handle=0x{:08X}", handle); return ERR_INVALID_HANDLE; @@ -1147,7 +1148,7 @@ static ResultCode SetThreadPriority(Core::System& system, Handle handle, u32 pri const auto* const current_process = system.Kernel().CurrentProcess(); - SharedPtr<Thread> thread = current_process->GetHandleTable().Get<Thread>(handle); + std::shared_ptr<Thread> thread = current_process->GetHandleTable().Get<Thread>(handle); if (!thread) { LOG_ERROR(Kernel_SVC, "Thread handle does not exist, handle=0x{:08X}", handle); return ERR_INVALID_HANDLE; @@ -1267,7 +1268,7 @@ static ResultCode QueryProcessMemory(Core::System& system, VAddr memory_info_add VAddr address) { LOG_TRACE(Kernel_SVC, "called process=0x{:08X} address={:X}", process_handle, address); const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable(); - SharedPtr<Process> process = handle_table.Get<Process>(process_handle); + std::shared_ptr<Process> process = handle_table.Get<Process>(process_handle); if (!process) { LOG_ERROR(Kernel_SVC, "Process handle does not exist, process_handle=0x{:08X}", process_handle); @@ -1495,7 +1496,7 @@ static ResultCode CreateThread(Core::System& system, Handle* out_handle, VAddr e } auto& kernel = system.Kernel(); - CASCADE_RESULT(SharedPtr<Thread> thread, + CASCADE_RESULT(std::shared_ptr<Thread> thread, Thread::Create(kernel, "", entry_point, priority, arg, processor_id, stack_top, *current_process)); @@ -1521,7 +1522,7 @@ static ResultCode StartThread(Core::System& system, Handle thread_handle) { LOG_DEBUG(Kernel_SVC, "called thread=0x{:08X}", thread_handle); const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable(); - const SharedPtr<Thread> thread = handle_table.Get<Thread>(thread_handle); + const std::shared_ptr<Thread> thread = handle_table.Get<Thread>(thread_handle); if (!thread) { LOG_ERROR(Kernel_SVC, "Thread handle does not exist, thread_handle=0x{:08X}", thread_handle); @@ -1545,7 +1546,7 @@ static void ExitThread(Core::System& system) { auto* const current_thread = system.CurrentScheduler().GetCurrentThread(); current_thread->Stop(); - system.GlobalScheduler().RemoveThread(current_thread); + system.GlobalScheduler().RemoveThread(SharedFrom(current_thread)); system.PrepareReschedule(); } @@ -1617,7 +1618,7 @@ static ResultCode WaitProcessWideKeyAtomic(Core::System& system, VAddr mutex_add auto* const current_process = system.Kernel().CurrentProcess(); const auto& handle_table = current_process->GetHandleTable(); - SharedPtr<Thread> thread = handle_table.Get<Thread>(thread_handle); + std::shared_ptr<Thread> thread = handle_table.Get<Thread>(thread_handle); ASSERT(thread); const auto release_result = current_process->GetMutex().Release(mutex_addr); @@ -1625,13 +1626,13 @@ static ResultCode WaitProcessWideKeyAtomic(Core::System& system, VAddr mutex_add return release_result; } - SharedPtr<Thread> current_thread = system.CurrentScheduler().GetCurrentThread(); + Thread* current_thread = system.CurrentScheduler().GetCurrentThread(); current_thread->SetCondVarWaitAddress(condition_variable_addr); current_thread->SetMutexWaitAddress(mutex_addr); current_thread->SetWaitHandle(thread_handle); current_thread->SetStatus(ThreadStatus::WaitCondVar); current_thread->InvalidateWakeupCallback(); - current_process->InsertConditionVariableThread(current_thread); + current_process->InsertConditionVariableThread(SharedFrom(current_thread)); current_thread->WakeAfterDelay(nano_seconds); @@ -1651,7 +1652,7 @@ static ResultCode SignalProcessWideKey(Core::System& system, VAddr condition_var // Retrieve a list of all threads that are waiting for this condition variable. auto* const current_process = system.Kernel().CurrentProcess(); - std::vector<SharedPtr<Thread>> waiting_threads = + std::vector<std::shared_ptr<Thread>> waiting_threads = current_process->GetConditionVariableThreads(condition_variable_addr); // Only process up to 'target' threads, unless 'target' is less equal 0, in which case process @@ -1777,7 +1778,9 @@ static u64 GetSystemTick(Core::System& system) { LOG_TRACE(Kernel_SVC, "called"); auto& core_timing = system.CoreTiming(); - const u64 result{core_timing.GetTicks()}; + + // Returns the value of cntpct_el0 (https://switchbrew.org/wiki/SVC#svcGetSystemTick) + const u64 result{Core::Timing::CpuCyclesToClockCycles(system.CoreTiming().GetTicks())}; // Advance time to defeat dumb games that busy-wait for the frame to end. core_timing.AddTicks(400); @@ -1966,7 +1969,7 @@ static ResultCode GetThreadCoreMask(Core::System& system, Handle thread_handle, LOG_TRACE(Kernel_SVC, "called, handle=0x{:08X}", thread_handle); const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable(); - const SharedPtr<Thread> thread = handle_table.Get<Thread>(thread_handle); + const std::shared_ptr<Thread> thread = handle_table.Get<Thread>(thread_handle); if (!thread) { LOG_ERROR(Kernel_SVC, "Thread handle does not exist, thread_handle=0x{:08X}", thread_handle); @@ -2025,7 +2028,7 @@ static ResultCode SetThreadCoreMask(Core::System& system, Handle thread_handle, } const auto& handle_table = current_process->GetHandleTable(); - const SharedPtr<Thread> thread = handle_table.Get<Thread>(thread_handle); + const std::shared_ptr<Thread> thread = handle_table.Get<Thread>(thread_handle); if (!thread) { LOG_ERROR(Kernel_SVC, "Thread handle does not exist, thread_handle=0x{:08X}", thread_handle); diff --git a/src/core/hle/kernel/thread.cpp b/src/core/hle/kernel/thread.cpp index 7166e9b07..735019d96 100644 --- a/src/core/hle/kernel/thread.cpp +++ b/src/core/hle/kernel/thread.cpp @@ -50,7 +50,7 @@ void Thread::Stop() { // Clean up any dangling references in objects that this thread was waiting for for (auto& wait_object : wait_objects) { - wait_object->RemoveWaitingThread(this); + wait_object->RemoveWaitingThread(SharedFrom(this)); } wait_objects.clear(); @@ -147,9 +147,10 @@ static void ResetThreadContext(Core::ARM_Interface::ThreadContext& context, VAdd context.fpcr = 0x03C00000; } -ResultVal<SharedPtr<Thread>> Thread::Create(KernelCore& kernel, std::string name, VAddr entry_point, - u32 priority, u64 arg, s32 processor_id, - VAddr stack_top, Process& owner_process) { +ResultVal<std::shared_ptr<Thread>> Thread::Create(KernelCore& kernel, std::string name, + VAddr entry_point, u32 priority, u64 arg, + s32 processor_id, VAddr stack_top, + Process& owner_process) { // Check if priority is in ranged. Lowest priority -> highest priority id. if (priority > THREADPRIO_LOWEST) { LOG_ERROR(Kernel_SVC, "Invalid thread priority: {}", priority); @@ -168,7 +169,7 @@ ResultVal<SharedPtr<Thread>> Thread::Create(KernelCore& kernel, std::string name } auto& system = Core::System::GetInstance(); - SharedPtr<Thread> thread(new Thread(kernel)); + std::shared_ptr<Thread> thread = std::make_shared<Thread>(kernel); thread->thread_id = kernel.CreateNewThreadID(); thread->status = ThreadStatus::Dormant; @@ -197,7 +198,7 @@ ResultVal<SharedPtr<Thread>> Thread::Create(KernelCore& kernel, std::string name // to initialize the context ResetThreadContext(thread->context, stack_top, entry_point, arg); - return MakeResult<SharedPtr<Thread>>(std::move(thread)); + return MakeResult<std::shared_ptr<Thread>>(std::move(thread)); } void Thread::SetPriority(u32 priority) { @@ -215,7 +216,7 @@ void Thread::SetWaitSynchronizationOutput(s32 output) { context.cpu_registers[1] = output; } -s32 Thread::GetWaitObjectIndex(const WaitObject* object) const { +s32 Thread::GetWaitObjectIndex(std::shared_ptr<WaitObject> object) const { ASSERT_MSG(!wait_objects.empty(), "Thread is not waiting for anything"); const auto match = std::find(wait_objects.rbegin(), wait_objects.rend(), object); return static_cast<s32>(std::distance(match, wait_objects.rend()) - 1); @@ -255,8 +256,8 @@ void Thread::SetStatus(ThreadStatus new_status) { status = new_status; } -void Thread::AddMutexWaiter(SharedPtr<Thread> thread) { - if (thread->lock_owner == this) { +void Thread::AddMutexWaiter(std::shared_ptr<Thread> thread) { + if (thread->lock_owner.get() == this) { // If the thread is already waiting for this thread to release the mutex, ensure that the // waiters list is consistent and return without doing anything. const auto iter = std::find(wait_mutex_threads.begin(), wait_mutex_threads.end(), thread); @@ -276,13 +277,13 @@ void Thread::AddMutexWaiter(SharedPtr<Thread> thread) { wait_mutex_threads.begin(), wait_mutex_threads.end(), [&thread](const auto& entry) { return entry->GetPriority() > thread->GetPriority(); }); wait_mutex_threads.insert(insertion_point, thread); - thread->lock_owner = this; + thread->lock_owner = SharedFrom(this); UpdatePriority(); } -void Thread::RemoveMutexWaiter(SharedPtr<Thread> thread) { - ASSERT(thread->lock_owner == this); +void Thread::RemoveMutexWaiter(std::shared_ptr<Thread> thread) { + ASSERT(thread->lock_owner.get() == this); // Ensure that the thread is in the list of mutex waiters const auto iter = std::find(wait_mutex_threads.begin(), wait_mutex_threads.end(), thread); @@ -310,13 +311,13 @@ void Thread::UpdatePriority() { } if (GetStatus() == ThreadStatus::WaitCondVar) { - owner_process->RemoveConditionVariableThread(this); + owner_process->RemoveConditionVariableThread(SharedFrom(this)); } SetCurrentPriority(new_priority); if (GetStatus() == ThreadStatus::WaitCondVar) { - owner_process->InsertConditionVariableThread(this); + owner_process->InsertConditionVariableThread(SharedFrom(this)); } if (!lock_owner) { @@ -325,8 +326,8 @@ void Thread::UpdatePriority() { // Ensure that the thread is within the correct location in the waiting list. auto old_owner = lock_owner; - lock_owner->RemoveMutexWaiter(this); - old_owner->AddMutexWaiter(this); + lock_owner->RemoveMutexWaiter(SharedFrom(this)); + old_owner->AddMutexWaiter(SharedFrom(this)); // Recursively update the priority of the thread that depends on the priority of this one. lock_owner->UpdatePriority(); @@ -339,11 +340,11 @@ void Thread::ChangeCore(u32 core, u64 mask) { bool Thread::AllWaitObjectsReady() const { return std::none_of( wait_objects.begin(), wait_objects.end(), - [this](const SharedPtr<WaitObject>& object) { return object->ShouldWait(this); }); + [this](const std::shared_ptr<WaitObject>& object) { return object->ShouldWait(this); }); } -bool Thread::InvokeWakeupCallback(ThreadWakeupReason reason, SharedPtr<Thread> thread, - SharedPtr<WaitObject> object, std::size_t index) { +bool Thread::InvokeWakeupCallback(ThreadWakeupReason reason, std::shared_ptr<Thread> thread, + std::shared_ptr<WaitObject> object, std::size_t index) { ASSERT(wakeup_callback); return wakeup_callback(reason, std::move(thread), std::move(object), index); } diff --git a/src/core/hle/kernel/thread.h b/src/core/hle/kernel/thread.h index 25a6ed234..3bcf9e137 100644 --- a/src/core/hle/kernel/thread.h +++ b/src/core/hle/kernel/thread.h @@ -97,14 +97,18 @@ enum class ThreadSchedMasks : u32 { class Thread final : public WaitObject { public: - using MutexWaitingThreads = std::vector<SharedPtr<Thread>>; + explicit Thread(KernelCore& kernel); + ~Thread() override; + + using MutexWaitingThreads = std::vector<std::shared_ptr<Thread>>; using ThreadContext = Core::ARM_Interface::ThreadContext; - using ThreadWaitObjects = std::vector<SharedPtr<WaitObject>>; + using ThreadWaitObjects = std::vector<std::shared_ptr<WaitObject>>; - using WakeupCallback = std::function<bool(ThreadWakeupReason reason, SharedPtr<Thread> thread, - SharedPtr<WaitObject> object, std::size_t index)>; + using WakeupCallback = + std::function<bool(ThreadWakeupReason reason, std::shared_ptr<Thread> thread, + std::shared_ptr<WaitObject> object, std::size_t index)>; /** * Creates and returns a new thread. The new thread is immediately scheduled @@ -118,10 +122,10 @@ public: * @param owner_process The parent process for the thread * @return A shared pointer to the newly created thread */ - static ResultVal<SharedPtr<Thread>> Create(KernelCore& kernel, std::string name, - VAddr entry_point, u32 priority, u64 arg, - s32 processor_id, VAddr stack_top, - Process& owner_process); + static ResultVal<std::shared_ptr<Thread>> Create(KernelCore& kernel, std::string name, + VAddr entry_point, u32 priority, u64 arg, + s32 processor_id, VAddr stack_top, + Process& owner_process); std::string GetName() const override { return name; @@ -166,10 +170,10 @@ public: void SetPriority(u32 priority); /// Adds a thread to the list of threads that are waiting for a lock held by this thread. - void AddMutexWaiter(SharedPtr<Thread> thread); + void AddMutexWaiter(std::shared_ptr<Thread> thread); /// Removes a thread from the list of threads that are waiting for a lock held by this thread. - void RemoveMutexWaiter(SharedPtr<Thread> thread); + void RemoveMutexWaiter(std::shared_ptr<Thread> thread); /// Recalculates the current priority taking into account priority inheritance. void UpdatePriority(); @@ -229,7 +233,7 @@ public: * * @param object Object to query the index of. */ - s32 GetWaitObjectIndex(const WaitObject* object) const; + s32 GetWaitObjectIndex(std::shared_ptr<WaitObject> object) const; /** * Stops a thread, invalidating it from further use @@ -320,7 +324,7 @@ public: void ClearWaitObjects() { for (const auto& waiting_object : wait_objects) { - waiting_object->RemoveWaitingThread(this); + waiting_object->RemoveWaitingThread(SharedFrom(this)); } wait_objects.clear(); } @@ -336,7 +340,7 @@ public: return lock_owner.get(); } - void SetLockOwner(SharedPtr<Thread> owner) { + void SetLockOwner(std::shared_ptr<Thread> owner) { lock_owner = std::move(owner); } @@ -390,8 +394,8 @@ public: * @pre A valid wakeup callback has been set. Violating this precondition * will cause an assertion to trigger. */ - bool InvokeWakeupCallback(ThreadWakeupReason reason, SharedPtr<Thread> thread, - SharedPtr<WaitObject> object, std::size_t index); + bool InvokeWakeupCallback(ThreadWakeupReason reason, std::shared_ptr<Thread> thread, + std::shared_ptr<WaitObject> object, std::size_t index); u32 GetIdealCore() const { return ideal_core; @@ -449,9 +453,6 @@ public: } private: - explicit Thread(KernelCore& kernel); - ~Thread() override; - void SetSchedulingStatus(ThreadSchedStatus new_status); void SetCurrentPriority(u32 new_priority); ResultCode SetCoreAndAffinityMask(s32 new_core, u64 new_affinity_mask); @@ -499,7 +500,7 @@ private: MutexWaitingThreads wait_mutex_threads; /// Thread that owns the lock that this thread is waiting for. - SharedPtr<Thread> lock_owner; + std::shared_ptr<Thread> lock_owner; /// If waiting on a ConditionVariable, this is the ConditionVariable address VAddr condvar_wait_address = 0; diff --git a/src/core/hle/kernel/transfer_memory.cpp b/src/core/hle/kernel/transfer_memory.cpp index 1113c815e..f0e73f57b 100644 --- a/src/core/hle/kernel/transfer_memory.cpp +++ b/src/core/hle/kernel/transfer_memory.cpp @@ -14,9 +14,9 @@ namespace Kernel { TransferMemory::TransferMemory(KernelCore& kernel) : Object{kernel} {} TransferMemory::~TransferMemory() = default; -SharedPtr<TransferMemory> TransferMemory::Create(KernelCore& kernel, VAddr base_address, u64 size, - MemoryPermission permissions) { - SharedPtr<TransferMemory> transfer_memory{new TransferMemory(kernel)}; +std::shared_ptr<TransferMemory> TransferMemory::Create(KernelCore& kernel, VAddr base_address, + u64 size, MemoryPermission permissions) { + std::shared_ptr<TransferMemory> transfer_memory{std::make_shared<TransferMemory>(kernel)}; transfer_memory->base_address = base_address; transfer_memory->memory_size = size; diff --git a/src/core/hle/kernel/transfer_memory.h b/src/core/hle/kernel/transfer_memory.h index 6be9dc094..556e6c62b 100644 --- a/src/core/hle/kernel/transfer_memory.h +++ b/src/core/hle/kernel/transfer_memory.h @@ -27,10 +27,13 @@ enum class MemoryPermission : u32; /// class TransferMemory final : public Object { public: + explicit TransferMemory(KernelCore& kernel); + ~TransferMemory() override; + static constexpr HandleType HANDLE_TYPE = HandleType::TransferMemory; - static SharedPtr<TransferMemory> Create(KernelCore& kernel, VAddr base_address, u64 size, - MemoryPermission permissions); + static std::shared_ptr<TransferMemory> Create(KernelCore& kernel, VAddr base_address, u64 size, + MemoryPermission permissions); TransferMemory(const TransferMemory&) = delete; TransferMemory& operator=(const TransferMemory&) = delete; @@ -79,9 +82,6 @@ public: ResultCode UnmapMemory(VAddr address, u64 size); private: - explicit TransferMemory(KernelCore& kernel); - ~TransferMemory() override; - /// Memory block backing this instance. std::shared_ptr<PhysicalMemory> backing_block; diff --git a/src/core/hle/kernel/wait_object.cpp b/src/core/hle/kernel/wait_object.cpp index c00cef062..745f2c4e8 100644 --- a/src/core/hle/kernel/wait_object.cpp +++ b/src/core/hle/kernel/wait_object.cpp @@ -18,13 +18,13 @@ namespace Kernel { WaitObject::WaitObject(KernelCore& kernel) : Object{kernel} {} WaitObject::~WaitObject() = default; -void WaitObject::AddWaitingThread(SharedPtr<Thread> thread) { +void WaitObject::AddWaitingThread(std::shared_ptr<Thread> thread) { auto itr = std::find(waiting_threads.begin(), waiting_threads.end(), thread); if (itr == waiting_threads.end()) waiting_threads.push_back(std::move(thread)); } -void WaitObject::RemoveWaitingThread(Thread* thread) { +void WaitObject::RemoveWaitingThread(std::shared_ptr<Thread> thread) { auto itr = std::find(waiting_threads.begin(), waiting_threads.end(), thread); // If a thread passed multiple handles to the same object, // the kernel might attempt to remove the thread from the object's @@ -33,7 +33,7 @@ void WaitObject::RemoveWaitingThread(Thread* thread) { waiting_threads.erase(itr); } -SharedPtr<Thread> WaitObject::GetHighestPriorityReadyThread() const { +std::shared_ptr<Thread> WaitObject::GetHighestPriorityReadyThread() const { Thread* candidate = nullptr; u32 candidate_priority = THREADPRIO_LOWEST + 1; @@ -64,10 +64,10 @@ SharedPtr<Thread> WaitObject::GetHighestPriorityReadyThread() const { } } - return candidate; + return SharedFrom(candidate); } -void WaitObject::WakeupWaitingThread(SharedPtr<Thread> thread) { +void WaitObject::WakeupWaitingThread(std::shared_ptr<Thread> thread) { ASSERT(!ShouldWait(thread.get())); if (!thread) { @@ -83,7 +83,7 @@ void WaitObject::WakeupWaitingThread(SharedPtr<Thread> thread) { Acquire(thread.get()); } - const std::size_t index = thread->GetWaitObjectIndex(this); + const std::size_t index = thread->GetWaitObjectIndex(SharedFrom(this)); thread->ClearWaitObjects(); @@ -91,7 +91,8 @@ void WaitObject::WakeupWaitingThread(SharedPtr<Thread> thread) { bool resume = true; if (thread->HasWakeupCallback()) { - resume = thread->InvokeWakeupCallback(ThreadWakeupReason::Signal, thread, this, index); + resume = thread->InvokeWakeupCallback(ThreadWakeupReason::Signal, thread, SharedFrom(this), + index); } if (resume) { thread->ResumeFromWait(); @@ -105,7 +106,7 @@ void WaitObject::WakeupAllWaitingThreads() { } } -const std::vector<SharedPtr<Thread>>& WaitObject::GetWaitingThreads() const { +const std::vector<std::shared_ptr<Thread>>& WaitObject::GetWaitingThreads() const { return waiting_threads; } diff --git a/src/core/hle/kernel/wait_object.h b/src/core/hle/kernel/wait_object.h index 3271a30a7..f9d596db9 100644 --- a/src/core/hle/kernel/wait_object.h +++ b/src/core/hle/kernel/wait_object.h @@ -33,13 +33,13 @@ public: * Add a thread to wait on this object * @param thread Pointer to thread to add */ - void AddWaitingThread(SharedPtr<Thread> thread); + void AddWaitingThread(std::shared_ptr<Thread> thread); /** * Removes a thread from waiting on this object (e.g. if it was resumed already) * @param thread Pointer to thread to remove */ - void RemoveWaitingThread(Thread* thread); + void RemoveWaitingThread(std::shared_ptr<Thread> thread); /** * Wake up all threads waiting on this object that can be awoken, in priority order, @@ -51,24 +51,24 @@ public: * Wakes up a single thread waiting on this object. * @param thread Thread that is waiting on this object to wakeup. */ - void WakeupWaitingThread(SharedPtr<Thread> thread); + void WakeupWaitingThread(std::shared_ptr<Thread> thread); /// Obtains the highest priority thread that is ready to run from this object's waiting list. - SharedPtr<Thread> GetHighestPriorityReadyThread() const; + std::shared_ptr<Thread> GetHighestPriorityReadyThread() const; /// Get a const reference to the waiting threads list for debug use - const std::vector<SharedPtr<Thread>>& GetWaitingThreads() const; + const std::vector<std::shared_ptr<Thread>>& GetWaitingThreads() const; private: /// Threads waiting for this object to become available - std::vector<SharedPtr<Thread>> waiting_threads; + std::vector<std::shared_ptr<Thread>> waiting_threads; }; // Specialization of DynamicObjectCast for WaitObjects template <> -inline SharedPtr<WaitObject> DynamicObjectCast<WaitObject>(SharedPtr<Object> object) { +inline std::shared_ptr<WaitObject> DynamicObjectCast<WaitObject>(std::shared_ptr<Object> object) { if (object != nullptr && object->IsWaitable()) { - return boost::static_pointer_cast<WaitObject>(object); + return std::static_pointer_cast<WaitObject>(object); } return nullptr; } diff --git a/src/core/hle/kernel/writable_event.cpp b/src/core/hle/kernel/writable_event.cpp index c783a34ee..c9332e3e1 100644 --- a/src/core/hle/kernel/writable_event.cpp +++ b/src/core/hle/kernel/writable_event.cpp @@ -16,8 +16,8 @@ WritableEvent::WritableEvent(KernelCore& kernel) : Object{kernel} {} WritableEvent::~WritableEvent() = default; EventPair WritableEvent::CreateEventPair(KernelCore& kernel, std::string name) { - SharedPtr<WritableEvent> writable_event(new WritableEvent(kernel)); - SharedPtr<ReadableEvent> readable_event(new ReadableEvent(kernel)); + std::shared_ptr<WritableEvent> writable_event(new WritableEvent(kernel)); + std::shared_ptr<ReadableEvent> readable_event(new ReadableEvent(kernel)); writable_event->name = name + ":Writable"; writable_event->readable = readable_event; @@ -27,7 +27,7 @@ EventPair WritableEvent::CreateEventPair(KernelCore& kernel, std::string name) { return {std::move(readable_event), std::move(writable_event)}; } -SharedPtr<ReadableEvent> WritableEvent::GetReadableEvent() const { +std::shared_ptr<ReadableEvent> WritableEvent::GetReadableEvent() const { return readable; } diff --git a/src/core/hle/kernel/writable_event.h b/src/core/hle/kernel/writable_event.h index f46cf1dd8..afe97f3a9 100644 --- a/src/core/hle/kernel/writable_event.h +++ b/src/core/hle/kernel/writable_event.h @@ -13,8 +13,8 @@ class ReadableEvent; class WritableEvent; struct EventPair { - SharedPtr<ReadableEvent> readable; - SharedPtr<WritableEvent> writable; + std::shared_ptr<ReadableEvent> readable; + std::shared_ptr<WritableEvent> writable; }; class WritableEvent final : public Object { @@ -40,7 +40,7 @@ public: return HANDLE_TYPE; } - SharedPtr<ReadableEvent> GetReadableEvent() const; + std::shared_ptr<ReadableEvent> GetReadableEvent() const; void Signal(); void Clear(); @@ -49,7 +49,7 @@ public: private: explicit WritableEvent(KernelCore& kernel); - SharedPtr<ReadableEvent> readable; + std::shared_ptr<ReadableEvent> readable; std::string name; ///< Name of event (optional) }; diff --git a/src/core/hle/service/am/am.cpp b/src/core/hle/service/am/am.cpp index 701f05019..3ebb92994 100644 --- a/src/core/hle/service/am/am.cpp +++ b/src/core/hle/service/am/am.cpp @@ -531,12 +531,11 @@ AppletMessageQueue::AppletMessageQueue(Kernel::KernelCore& kernel) { AppletMessageQueue::~AppletMessageQueue() = default; -const Kernel::SharedPtr<Kernel::ReadableEvent>& AppletMessageQueue::GetMesssageRecieveEvent() - const { +const std::shared_ptr<Kernel::ReadableEvent>& AppletMessageQueue::GetMesssageRecieveEvent() const { return on_new_message.readable; } -const Kernel::SharedPtr<Kernel::ReadableEvent>& AppletMessageQueue::GetOperationModeChangedEvent() +const std::shared_ptr<Kernel::ReadableEvent>& AppletMessageQueue::GetOperationModeChangedEvent() const { return on_operation_mode_changed.readable; } diff --git a/src/core/hle/service/am/am.h b/src/core/hle/service/am/am.h index 06a65b5ed..448817be9 100644 --- a/src/core/hle/service/am/am.h +++ b/src/core/hle/service/am/am.h @@ -54,8 +54,8 @@ public: explicit AppletMessageQueue(Kernel::KernelCore& kernel); ~AppletMessageQueue(); - const Kernel::SharedPtr<Kernel::ReadableEvent>& GetMesssageRecieveEvent() const; - const Kernel::SharedPtr<Kernel::ReadableEvent>& GetOperationModeChangedEvent() const; + const std::shared_ptr<Kernel::ReadableEvent>& GetMesssageRecieveEvent() const; + const std::shared_ptr<Kernel::ReadableEvent>& GetOperationModeChangedEvent() const; void PushMessage(AppletMessage msg); AppletMessage PopMessage(); std::size_t GetMessageCount() const; diff --git a/src/core/hle/service/am/applets/applets.cpp b/src/core/hle/service/am/applets/applets.cpp index 673ad1f7f..92f995f8f 100644 --- a/src/core/hle/service/am/applets/applets.cpp +++ b/src/core/hle/service/am/applets/applets.cpp @@ -108,15 +108,15 @@ void AppletDataBroker::SignalStateChanged() const { state_changed_event.writable->Signal(); } -Kernel::SharedPtr<Kernel::ReadableEvent> AppletDataBroker::GetNormalDataEvent() const { +std::shared_ptr<Kernel::ReadableEvent> AppletDataBroker::GetNormalDataEvent() const { return pop_out_data_event.readable; } -Kernel::SharedPtr<Kernel::ReadableEvent> AppletDataBroker::GetInteractiveDataEvent() const { +std::shared_ptr<Kernel::ReadableEvent> AppletDataBroker::GetInteractiveDataEvent() const { return pop_interactive_out_data_event.readable; } -Kernel::SharedPtr<Kernel::ReadableEvent> AppletDataBroker::GetStateChangedEvent() const { +std::shared_ptr<Kernel::ReadableEvent> AppletDataBroker::GetStateChangedEvent() const { return state_changed_event.readable; } diff --git a/src/core/hle/service/am/applets/applets.h b/src/core/hle/service/am/applets/applets.h index 226be88b1..16e61fc6f 100644 --- a/src/core/hle/service/am/applets/applets.h +++ b/src/core/hle/service/am/applets/applets.h @@ -86,9 +86,9 @@ public: void SignalStateChanged() const; - Kernel::SharedPtr<Kernel::ReadableEvent> GetNormalDataEvent() const; - Kernel::SharedPtr<Kernel::ReadableEvent> GetInteractiveDataEvent() const; - Kernel::SharedPtr<Kernel::ReadableEvent> GetStateChangedEvent() const; + std::shared_ptr<Kernel::ReadableEvent> GetNormalDataEvent() const; + std::shared_ptr<Kernel::ReadableEvent> GetInteractiveDataEvent() const; + std::shared_ptr<Kernel::ReadableEvent> GetStateChangedEvent() const; private: // Queues are named from applet's perspective diff --git a/src/core/hle/service/bcat/backend/backend.cpp b/src/core/hle/service/bcat/backend/backend.cpp index dec0849b8..6f5ea095a 100644 --- a/src/core/hle/service/bcat/backend/backend.cpp +++ b/src/core/hle/service/bcat/backend/backend.cpp @@ -16,7 +16,7 @@ ProgressServiceBackend::ProgressServiceBackend(Kernel::KernelCore& kernel, kernel, std::string("ProgressServiceBackend:UpdateEvent:").append(event_name)); } -Kernel::SharedPtr<Kernel::ReadableEvent> ProgressServiceBackend::GetEvent() const { +std::shared_ptr<Kernel::ReadableEvent> ProgressServiceBackend::GetEvent() const { return event.readable; } diff --git a/src/core/hle/service/bcat/backend/backend.h b/src/core/hle/service/bcat/backend/backend.h index ea4b16ad0..48bbbe66f 100644 --- a/src/core/hle/service/bcat/backend/backend.h +++ b/src/core/hle/service/bcat/backend/backend.h @@ -98,7 +98,7 @@ public: private: explicit ProgressServiceBackend(Kernel::KernelCore& kernel, std::string_view event_name); - Kernel::SharedPtr<Kernel::ReadableEvent> GetEvent() const; + std::shared_ptr<Kernel::ReadableEvent> GetEvent() const; DeliveryCacheProgressImpl& GetImpl(); void SignalUpdate() const; diff --git a/src/core/hle/service/bcat/module.cpp b/src/core/hle/service/bcat/module.cpp index 8a7304f86..920d7269b 100644 --- a/src/core/hle/service/bcat/module.cpp +++ b/src/core/hle/service/bcat/module.cpp @@ -87,7 +87,7 @@ struct DeliveryCacheDirectoryEntry { class IDeliveryCacheProgressService final : public ServiceFramework<IDeliveryCacheProgressService> { public: - IDeliveryCacheProgressService(Kernel::SharedPtr<Kernel::ReadableEvent> event, + IDeliveryCacheProgressService(std::shared_ptr<Kernel::ReadableEvent> event, const DeliveryCacheProgressImpl& impl) : ServiceFramework{"IDeliveryCacheProgressService"}, event(std::move(event)), impl(impl) { // clang-format off @@ -118,7 +118,7 @@ private: rb.Push(RESULT_SUCCESS); } - Kernel::SharedPtr<Kernel::ReadableEvent> event; + std::shared_ptr<Kernel::ReadableEvent> event; const DeliveryCacheProgressImpl& impl; }; diff --git a/src/core/hle/service/hid/controllers/npad.cpp b/src/core/hle/service/hid/controllers/npad.cpp index 79fff517e..4d952adc0 100644 --- a/src/core/hle/service/hid/controllers/npad.cpp +++ b/src/core/hle/service/hid/controllers/npad.cpp @@ -501,8 +501,7 @@ void Controller_NPad::VibrateController(const std::vector<u32>& controller_ids, last_processed_vibration = vibrations.back(); } -Kernel::SharedPtr<Kernel::ReadableEvent> Controller_NPad::GetStyleSetChangedEvent( - u32 npad_id) const { +std::shared_ptr<Kernel::ReadableEvent> Controller_NPad::GetStyleSetChangedEvent(u32 npad_id) const { // TODO(ogniK): Figure out the best time to signal this event. This event seems that it should // be signalled at least once, and signaled after a new controller is connected? const auto& styleset_event = styleset_changed_events[NPadIdToIndex(npad_id)]; diff --git a/src/core/hle/service/hid/controllers/npad.h b/src/core/hle/service/hid/controllers/npad.h index 16c4caa1f..931f03430 100644 --- a/src/core/hle/service/hid/controllers/npad.h +++ b/src/core/hle/service/hid/controllers/npad.h @@ -109,7 +109,7 @@ public: void VibrateController(const std::vector<u32>& controller_ids, const std::vector<Vibration>& vibrations); - Kernel::SharedPtr<Kernel::ReadableEvent> GetStyleSetChangedEvent(u32 npad_id) const; + std::shared_ptr<Kernel::ReadableEvent> GetStyleSetChangedEvent(u32 npad_id) const; Vibration GetLastVibration() const; void AddNewController(NPadControllerType controller); diff --git a/src/core/hle/service/hid/hid.h b/src/core/hle/service/hid/hid.h index f08e036a3..923762fff 100644 --- a/src/core/hle/service/hid/hid.h +++ b/src/core/hle/service/hid/hid.h @@ -67,7 +67,7 @@ private: void GetSharedMemoryHandle(Kernel::HLERequestContext& ctx); void UpdateControllers(u64 userdata, s64 cycles_late); - Kernel::SharedPtr<Kernel::SharedMemory> shared_mem; + std::shared_ptr<Kernel::SharedMemory> shared_mem; Core::Timing::EventType* pad_update_event; Core::System& system; diff --git a/src/core/hle/service/hid/irs.h b/src/core/hle/service/hid/irs.h index eb4e898dd..8918ad6ca 100644 --- a/src/core/hle/service/hid/irs.h +++ b/src/core/hle/service/hid/irs.h @@ -37,7 +37,7 @@ private: void RunIrLedProcessor(Kernel::HLERequestContext& ctx); void StopImageProcessorAsync(Kernel::HLERequestContext& ctx); void ActivateIrsensorWithFunctionLevel(Kernel::HLERequestContext& ctx); - Kernel::SharedPtr<Kernel::SharedMemory> shared_mem; + std::shared_ptr<Kernel::SharedMemory> shared_mem; const u32 device_handle{0xABCD}; Core::System& system; }; diff --git a/src/core/hle/service/nfp/nfp.cpp b/src/core/hle/service/nfp/nfp.cpp index 3bf753dee..ec0367978 100644 --- a/src/core/hle/service/nfp/nfp.cpp +++ b/src/core/hle/service/nfp/nfp.cpp @@ -342,7 +342,7 @@ bool Module::Interface::LoadAmiibo(const std::vector<u8>& buffer) { return true; } -const Kernel::SharedPtr<Kernel::ReadableEvent>& Module::Interface::GetNFCEvent() const { +const std::shared_ptr<Kernel::ReadableEvent>& Module::Interface::GetNFCEvent() const { return nfc_tag_load.readable; } diff --git a/src/core/hle/service/nfp/nfp.h b/src/core/hle/service/nfp/nfp.h index 9718ef745..200013795 100644 --- a/src/core/hle/service/nfp/nfp.h +++ b/src/core/hle/service/nfp/nfp.h @@ -34,7 +34,7 @@ public: void CreateUserInterface(Kernel::HLERequestContext& ctx); bool LoadAmiibo(const std::vector<u8>& buffer); - const Kernel::SharedPtr<Kernel::ReadableEvent>& GetNFCEvent() const; + const std::shared_ptr<Kernel::ReadableEvent>& GetNFCEvent() const; const AmiiboFile& GetAmiiboBuffer() const; private: diff --git a/src/core/hle/service/ns/pl_u.cpp b/src/core/hle/service/ns/pl_u.cpp index db433305f..8dc103bf0 100644 --- a/src/core/hle/service/ns/pl_u.cpp +++ b/src/core/hle/service/ns/pl_u.cpp @@ -141,7 +141,7 @@ struct PL_U::Impl { } /// Handle to shared memory region designated for a shared font - Kernel::SharedPtr<Kernel::SharedMemory> shared_font_mem; + std::shared_ptr<Kernel::SharedMemory> shared_font_mem; /// Backing memory for the shared font data std::shared_ptr<Kernel::PhysicalMemory> shared_font; diff --git a/src/core/hle/service/nvdrv/interface.cpp b/src/core/hle/service/nvdrv/interface.cpp index 68d139cfb..c8ea6c661 100644 --- a/src/core/hle/service/nvdrv/interface.cpp +++ b/src/core/hle/service/nvdrv/interface.cpp @@ -61,7 +61,7 @@ void NVDRV::IoctlBase(Kernel::HLERequestContext& ctx, IoctlVersion version) { if (ctrl.must_delay) { ctrl.fresh_call = false; ctx.SleepClientThread("NVServices::DelayedResponse", ctrl.timeout, - [=](Kernel::SharedPtr<Kernel::Thread> thread, + [=](std::shared_ptr<Kernel::Thread> thread, Kernel::HLERequestContext& ctx, Kernel::ThreadWakeupReason reason) { IoctlCtrl ctrl2{ctrl}; diff --git a/src/core/hle/service/nvdrv/nvdrv.cpp b/src/core/hle/service/nvdrv/nvdrv.cpp index cc9cd3fd1..197c77db0 100644 --- a/src/core/hle/service/nvdrv/nvdrv.cpp +++ b/src/core/hle/service/nvdrv/nvdrv.cpp @@ -100,11 +100,11 @@ void Module::SignalSyncpt(const u32 syncpoint_id, const u32 value) { } } -Kernel::SharedPtr<Kernel::ReadableEvent> Module::GetEvent(const u32 event_id) const { +std::shared_ptr<Kernel::ReadableEvent> Module::GetEvent(const u32 event_id) const { return events_interface.events[event_id].readable; } -Kernel::SharedPtr<Kernel::WritableEvent> Module::GetEventWriteable(const u32 event_id) const { +std::shared_ptr<Kernel::WritableEvent> Module::GetEventWriteable(const u32 event_id) const { return events_interface.events[event_id].writable; } diff --git a/src/core/hle/service/nvdrv/nvdrv.h b/src/core/hle/service/nvdrv/nvdrv.h index f8bb28969..d7a1bef91 100644 --- a/src/core/hle/service/nvdrv/nvdrv.h +++ b/src/core/hle/service/nvdrv/nvdrv.h @@ -114,9 +114,9 @@ public: void SignalSyncpt(const u32 syncpoint_id, const u32 value); - Kernel::SharedPtr<Kernel::ReadableEvent> GetEvent(u32 event_id) const; + std::shared_ptr<Kernel::ReadableEvent> GetEvent(u32 event_id) const; - Kernel::SharedPtr<Kernel::WritableEvent> GetEventWriteable(u32 event_id) const; + std::shared_ptr<Kernel::WritableEvent> GetEventWriteable(u32 event_id) const; private: /// Id to use for the next open file descriptor. diff --git a/src/core/hle/service/nvflinger/buffer_queue.cpp b/src/core/hle/service/nvflinger/buffer_queue.cpp index 1af11e80c..32b6f4b27 100644 --- a/src/core/hle/service/nvflinger/buffer_queue.cpp +++ b/src/core/hle/service/nvflinger/buffer_queue.cpp @@ -117,11 +117,11 @@ u32 BufferQueue::Query(QueryType type) { return 0; } -Kernel::SharedPtr<Kernel::WritableEvent> BufferQueue::GetWritableBufferWaitEvent() const { +std::shared_ptr<Kernel::WritableEvent> BufferQueue::GetWritableBufferWaitEvent() const { return buffer_wait_event.writable; } -Kernel::SharedPtr<Kernel::ReadableEvent> BufferQueue::GetBufferWaitEvent() const { +std::shared_ptr<Kernel::ReadableEvent> BufferQueue::GetBufferWaitEvent() const { return buffer_wait_event.readable; } diff --git a/src/core/hle/service/nvflinger/buffer_queue.h b/src/core/hle/service/nvflinger/buffer_queue.h index 8f9b18547..f4bbfd945 100644 --- a/src/core/hle/service/nvflinger/buffer_queue.h +++ b/src/core/hle/service/nvflinger/buffer_queue.h @@ -93,9 +93,9 @@ public: return id; } - Kernel::SharedPtr<Kernel::WritableEvent> GetWritableBufferWaitEvent() const; + std::shared_ptr<Kernel::WritableEvent> GetWritableBufferWaitEvent() const; - Kernel::SharedPtr<Kernel::ReadableEvent> GetBufferWaitEvent() const; + std::shared_ptr<Kernel::ReadableEvent> GetBufferWaitEvent() const; private: u32 id; diff --git a/src/core/hle/service/nvflinger/nvflinger.cpp b/src/core/hle/service/nvflinger/nvflinger.cpp index cc9522aad..cd07ab362 100644 --- a/src/core/hle/service/nvflinger/nvflinger.cpp +++ b/src/core/hle/service/nvflinger/nvflinger.cpp @@ -98,7 +98,7 @@ std::optional<u32> NVFlinger::FindBufferQueueId(u64 display_id, u64 layer_id) co return layer->GetBufferQueue().GetId(); } -Kernel::SharedPtr<Kernel::ReadableEvent> NVFlinger::FindVsyncEvent(u64 display_id) const { +std::shared_ptr<Kernel::ReadableEvent> NVFlinger::FindVsyncEvent(u64 display_id) const { auto* const display = FindDisplay(display_id); if (display == nullptr) { diff --git a/src/core/hle/service/nvflinger/nvflinger.h b/src/core/hle/service/nvflinger/nvflinger.h index 5d7e3bfb8..9cc41f2e6 100644 --- a/src/core/hle/service/nvflinger/nvflinger.h +++ b/src/core/hle/service/nvflinger/nvflinger.h @@ -62,7 +62,7 @@ public: /// Gets the vsync event for the specified display. /// /// If an invalid display ID is provided, then nullptr is returned. - Kernel::SharedPtr<Kernel::ReadableEvent> FindVsyncEvent(u64 display_id) const; + std::shared_ptr<Kernel::ReadableEvent> FindVsyncEvent(u64 display_id) const; /// Obtains a buffer queue identified by the ID. BufferQueue& FindBufferQueue(u32 id); diff --git a/src/core/hle/service/pm/pm.cpp b/src/core/hle/service/pm/pm.cpp index fe6b5f798..38b569da2 100644 --- a/src/core/hle/service/pm/pm.cpp +++ b/src/core/hle/service/pm/pm.cpp @@ -16,9 +16,9 @@ constexpr ResultCode ERROR_PROCESS_NOT_FOUND{ErrorModule::PM, 1}; constexpr u64 NO_PROCESS_FOUND_PID{0}; -std::optional<Kernel::SharedPtr<Kernel::Process>> SearchProcessList( - const std::vector<Kernel::SharedPtr<Kernel::Process>>& process_list, - std::function<bool(const Kernel::SharedPtr<Kernel::Process>&)> predicate) { +std::optional<std::shared_ptr<Kernel::Process>> SearchProcessList( + const std::vector<std::shared_ptr<Kernel::Process>>& process_list, + std::function<bool(const std::shared_ptr<Kernel::Process>&)> predicate) { const auto iter = std::find_if(process_list.begin(), process_list.end(), predicate); if (iter == process_list.end()) { @@ -29,7 +29,7 @@ std::optional<Kernel::SharedPtr<Kernel::Process>> SearchProcessList( } void GetApplicationPidGeneric(Kernel::HLERequestContext& ctx, - const std::vector<Kernel::SharedPtr<Kernel::Process>>& process_list) { + const std::vector<std::shared_ptr<Kernel::Process>>& process_list) { const auto process = SearchProcessList(process_list, [](const auto& process) { return process->GetProcessID() == Kernel::Process::ProcessIDMin; }); @@ -124,7 +124,7 @@ private: class Info final : public ServiceFramework<Info> { public: - explicit Info(const std::vector<Kernel::SharedPtr<Kernel::Process>>& process_list) + explicit Info(const std::vector<std::shared_ptr<Kernel::Process>>& process_list) : ServiceFramework{"pm:info"}, process_list(process_list) { static const FunctionInfo functions[] = { {0, &Info::GetTitleId, "GetTitleId"}, @@ -154,7 +154,7 @@ private: rb.Push((*process)->GetTitleID()); } - const std::vector<Kernel::SharedPtr<Kernel::Process>>& process_list; + const std::vector<std::shared_ptr<Kernel::Process>>& process_list; }; class Shell final : public ServiceFramework<Shell> { diff --git a/src/core/hle/service/service.cpp b/src/core/hle/service/service.cpp index 7c5302017..5698f429f 100644 --- a/src/core/hle/service/service.cpp +++ b/src/core/hle/service/service.cpp @@ -116,7 +116,7 @@ void ServiceFrameworkBase::InstallAsNamedPort() { port_installed = true; } -Kernel::SharedPtr<Kernel::ClientPort> ServiceFrameworkBase::CreatePort() { +std::shared_ptr<Kernel::ClientPort> ServiceFrameworkBase::CreatePort() { ASSERT(!port_installed); auto& kernel = Core::System::GetInstance().Kernel(); diff --git a/src/core/hle/service/service.h b/src/core/hle/service/service.h index aef964861..022d885b6 100644 --- a/src/core/hle/service/service.h +++ b/src/core/hle/service/service.h @@ -65,7 +65,7 @@ public: /// Creates a port pair and registers it on the kernel's global port registry. void InstallAsNamedPort(); /// Creates and returns an unregistered port for the service. - Kernel::SharedPtr<Kernel::ClientPort> CreatePort(); + std::shared_ptr<Kernel::ClientPort> CreatePort(); void InvokeRequest(Kernel::HLERequestContext& ctx); diff --git a/src/core/hle/service/sm/controller.cpp b/src/core/hle/service/sm/controller.cpp index e9ee73710..af2fadcef 100644 --- a/src/core/hle/service/sm/controller.cpp +++ b/src/core/hle/service/sm/controller.cpp @@ -30,7 +30,7 @@ void Controller::DuplicateSession(Kernel::HLERequestContext& ctx) { IPC::ResponseBuilder rb{ctx, 2, 0, 1, IPC::ResponseBuilder::Flags::AlwaysMoveHandles}; rb.Push(RESULT_SUCCESS); - Kernel::SharedPtr<Kernel::ClientSession> session{ctx.Session()->GetParent()->client}; + std::shared_ptr<Kernel::ClientSession> session{ctx.Session()->GetParent()->client}; rb.PushMoveObjects(session); LOG_DEBUG(Service, "session={}", session->GetObjectId()); diff --git a/src/core/hle/service/sm/sm.cpp b/src/core/hle/service/sm/sm.cpp index 142929124..a0a7206bb 100644 --- a/src/core/hle/service/sm/sm.cpp +++ b/src/core/hle/service/sm/sm.cpp @@ -45,7 +45,7 @@ void ServiceManager::InstallInterfaces(std::shared_ptr<ServiceManager> self) { self->controller_interface = std::make_unique<Controller>(); } -ResultVal<Kernel::SharedPtr<Kernel::ServerPort>> ServiceManager::RegisterService( +ResultVal<std::shared_ptr<Kernel::ServerPort>> ServiceManager::RegisterService( std::string name, unsigned int max_sessions) { CASCADE_CODE(ValidateServiceName(name)); @@ -72,7 +72,7 @@ ResultCode ServiceManager::UnregisterService(const std::string& name) { return RESULT_SUCCESS; } -ResultVal<Kernel::SharedPtr<Kernel::ClientPort>> ServiceManager::GetServicePort( +ResultVal<std::shared_ptr<Kernel::ClientPort>> ServiceManager::GetServicePort( const std::string& name) { CASCADE_CODE(ValidateServiceName(name)); @@ -84,7 +84,7 @@ ResultVal<Kernel::SharedPtr<Kernel::ClientPort>> ServiceManager::GetServicePort( return MakeResult(it->second); } -ResultVal<Kernel::SharedPtr<Kernel::ClientSession>> ServiceManager::ConnectToService( +ResultVal<std::shared_ptr<Kernel::ClientSession>> ServiceManager::ConnectToService( const std::string& name) { CASCADE_RESULT(auto client_port, GetServicePort(name)); diff --git a/src/core/hle/service/sm/sm.h b/src/core/hle/service/sm/sm.h index b9d6381b4..3de22268b 100644 --- a/src/core/hle/service/sm/sm.h +++ b/src/core/hle/service/sm/sm.h @@ -48,11 +48,11 @@ public: ServiceManager(); ~ServiceManager(); - ResultVal<Kernel::SharedPtr<Kernel::ServerPort>> RegisterService(std::string name, - unsigned int max_sessions); + ResultVal<std::shared_ptr<Kernel::ServerPort>> RegisterService(std::string name, + unsigned int max_sessions); ResultCode UnregisterService(const std::string& name); - ResultVal<Kernel::SharedPtr<Kernel::ClientPort>> GetServicePort(const std::string& name); - ResultVal<Kernel::SharedPtr<Kernel::ClientSession>> ConnectToService(const std::string& name); + ResultVal<std::shared_ptr<Kernel::ClientPort>> GetServicePort(const std::string& name); + ResultVal<std::shared_ptr<Kernel::ClientSession>> ConnectToService(const std::string& name); template <typename T> std::shared_ptr<T> GetService(const std::string& service_name) const { @@ -77,7 +77,7 @@ private: std::unique_ptr<Controller> controller_interface; /// Map of registered services, retrieved using GetServicePort or ConnectToService. - std::unordered_map<std::string, Kernel::SharedPtr<Kernel::ClientPort>> registered_services; + std::unordered_map<std::string, std::shared_ptr<Kernel::ClientPort>> registered_services; }; } // namespace Service::SM diff --git a/src/core/hle/service/time/time_sharedmemory.cpp b/src/core/hle/service/time/time_sharedmemory.cpp index bfc81b83c..4035f5072 100644 --- a/src/core/hle/service/time/time_sharedmemory.cpp +++ b/src/core/hle/service/time/time_sharedmemory.cpp @@ -21,7 +21,7 @@ SharedMemory::SharedMemory(Core::System& system) : system(system) { SharedMemory::~SharedMemory() = default; -Kernel::SharedPtr<Kernel::SharedMemory> SharedMemory::GetSharedMemoryHolder() const { +std::shared_ptr<Kernel::SharedMemory> SharedMemory::GetSharedMemoryHolder() const { return shared_memory_holder; } diff --git a/src/core/hle/service/time/time_sharedmemory.h b/src/core/hle/service/time/time_sharedmemory.h index cb8253541..904a96430 100644 --- a/src/core/hle/service/time/time_sharedmemory.h +++ b/src/core/hle/service/time/time_sharedmemory.h @@ -15,7 +15,7 @@ public: ~SharedMemory(); // Return the shared memory handle - Kernel::SharedPtr<Kernel::SharedMemory> GetSharedMemoryHolder() const; + std::shared_ptr<Kernel::SharedMemory> GetSharedMemoryHolder() const; // Set memory barriers in shared memory and update them void SetStandardSteadyClockTimepoint(const SteadyClockTimePoint& timepoint); @@ -66,7 +66,7 @@ public: static_assert(sizeof(Format) == 0xd8, "Format is an invalid size"); private: - Kernel::SharedPtr<Kernel::SharedMemory> shared_memory_holder{}; + std::shared_ptr<Kernel::SharedMemory> shared_memory_holder{}; Core::System& system; Format shared_memory_format{}; }; diff --git a/src/core/hle/service/vi/display/vi_display.cpp b/src/core/hle/service/vi/display/vi_display.cpp index 07033fb98..cd18c1610 100644 --- a/src/core/hle/service/vi/display/vi_display.cpp +++ b/src/core/hle/service/vi/display/vi_display.cpp @@ -31,7 +31,7 @@ const Layer& Display::GetLayer(std::size_t index) const { return layers.at(index); } -Kernel::SharedPtr<Kernel::ReadableEvent> Display::GetVSyncEvent() const { +std::shared_ptr<Kernel::ReadableEvent> Display::GetVSyncEvent() const { return vsync_event.readable; } diff --git a/src/core/hle/service/vi/display/vi_display.h b/src/core/hle/service/vi/display/vi_display.h index f56b5badc..8bb966a85 100644 --- a/src/core/hle/service/vi/display/vi_display.h +++ b/src/core/hle/service/vi/display/vi_display.h @@ -57,7 +57,7 @@ public: const Layer& GetLayer(std::size_t index) const; /// Gets the readable vsync event. - Kernel::SharedPtr<Kernel::ReadableEvent> GetVSyncEvent() const; + std::shared_ptr<Kernel::ReadableEvent> GetVSyncEvent() const; /// Signals the internal vsync event. void SignalVSyncEvent(); diff --git a/src/core/hle/service/vi/vi.cpp b/src/core/hle/service/vi/vi.cpp index abfc3a801..4cab49071 100644 --- a/src/core/hle/service/vi/vi.cpp +++ b/src/core/hle/service/vi/vi.cpp @@ -542,7 +542,7 @@ private: // Wait the current thread until a buffer becomes available ctx.SleepClientThread( "IHOSBinderDriver::DequeueBuffer", UINT64_MAX, - [=](Kernel::SharedPtr<Kernel::Thread> thread, Kernel::HLERequestContext& ctx, + [=](std::shared_ptr<Kernel::Thread> thread, Kernel::HLERequestContext& ctx, Kernel::ThreadWakeupReason reason) { // Repeat TransactParcel DequeueBuffer when a buffer is available auto& buffer_queue = nv_flinger->FindBufferQueue(id); diff --git a/src/video_core/CMakeLists.txt b/src/video_core/CMakeLists.txt index 6f3f2aa9f..3b20c7d34 100644 --- a/src/video_core/CMakeLists.txt +++ b/src/video_core/CMakeLists.txt @@ -22,6 +22,7 @@ add_library(video_core STATIC engines/maxwell_dma.h engines/shader_bytecode.h engines/shader_header.h + engines/shader_type.h gpu.cpp gpu.h gpu_asynch.cpp diff --git a/src/video_core/buffer_cache/buffer_cache.h b/src/video_core/buffer_cache/buffer_cache.h index 4408b5001..0510ed777 100644 --- a/src/video_core/buffer_cache/buffer_cache.h +++ b/src/video_core/buffer_cache/buffer_cache.h @@ -427,8 +427,8 @@ private: VideoCore::RasterizerInterface& rasterizer; Core::System& system; - std::unique_ptr<StreamBuffer> stream_buffer; + std::unique_ptr<StreamBuffer> stream_buffer; TBufferType stream_buffer_handle{}; bool invalidated = false; @@ -440,18 +440,18 @@ private: using IntervalSet = boost::icl::interval_set<CacheAddr>; using IntervalCache = boost::icl::interval_map<CacheAddr, MapInterval>; using IntervalType = typename IntervalCache::interval_type; - IntervalCache mapped_addresses{}; + IntervalCache mapped_addresses; - static constexpr u64 write_page_bit{11}; - std::unordered_map<u64, u32> written_pages{}; + static constexpr u64 write_page_bit = 11; + std::unordered_map<u64, u32> written_pages; - static constexpr u64 block_page_bits{21}; - static constexpr u64 block_page_size{1 << block_page_bits}; - std::unordered_map<u64, TBuffer> blocks{}; + static constexpr u64 block_page_bits = 21; + static constexpr u64 block_page_size = 1ULL << block_page_bits; + std::unordered_map<u64, TBuffer> blocks; - std::list<TBuffer> pending_destruction{}; - u64 epoch{}; - u64 modified_ticks{}; + std::list<TBuffer> pending_destruction; + u64 epoch = 0; + u64 modified_ticks = 0; std::recursive_mutex mutex; }; diff --git a/src/video_core/engines/const_buffer_engine_interface.h b/src/video_core/engines/const_buffer_engine_interface.h index ac27b6cbe..44b8b8d22 100644 --- a/src/video_core/engines/const_buffer_engine_interface.h +++ b/src/video_core/engines/const_buffer_engine_interface.h @@ -8,19 +8,11 @@ #include "common/bit_field.h" #include "common/common_types.h" #include "video_core/engines/shader_bytecode.h" +#include "video_core/engines/shader_type.h" #include "video_core/textures/texture.h" namespace Tegra::Engines { -enum class ShaderType : u32 { - Vertex = 0, - TesselationControl = 1, - TesselationEval = 2, - Geometry = 3, - Fragment = 4, - Compute = 5, -}; - struct SamplerDescriptor { union { BitField<0, 20, Tegra::Shader::TextureType> texture_type; diff --git a/src/video_core/engines/kepler_compute.cpp b/src/video_core/engines/kepler_compute.cpp index 3a39aeabe..110406f2f 100644 --- a/src/video_core/engines/kepler_compute.cpp +++ b/src/video_core/engines/kepler_compute.cpp @@ -8,6 +8,7 @@ #include "core/core.h" #include "video_core/engines/kepler_compute.h" #include "video_core/engines/maxwell_3d.h" +#include "video_core/engines/shader_type.h" #include "video_core/memory_manager.h" #include "video_core/rasterizer_interface.h" #include "video_core/renderer_base.h" diff --git a/src/video_core/engines/kepler_compute.h b/src/video_core/engines/kepler_compute.h index 5259d92bd..4ef3e0613 100644 --- a/src/video_core/engines/kepler_compute.h +++ b/src/video_core/engines/kepler_compute.h @@ -12,6 +12,7 @@ #include "common/common_types.h" #include "video_core/engines/const_buffer_engine_interface.h" #include "video_core/engines/engine_upload.h" +#include "video_core/engines/shader_type.h" #include "video_core/gpu.h" #include "video_core/textures/texture.h" @@ -140,7 +141,7 @@ public: INSERT_PADDING_WORDS(0x3); - BitField<0, 16, u32> shared_alloc; + BitField<0, 18, u32> shared_alloc; BitField<16, 16, u32> block_dim_x; union { @@ -178,7 +179,12 @@ public: BitField<24, 5, u32> gpr_alloc; }; - INSERT_PADDING_WORDS(0x11); + union { + BitField<0, 20, u32> local_crs_alloc; + BitField<24, 5, u32> sass_version; + }; + + INSERT_PADDING_WORDS(0x10); } launch_description{}; struct { diff --git a/src/video_core/engines/maxwell_3d.cpp b/src/video_core/engines/maxwell_3d.cpp index a44c09003..15a7a9d6a 100644 --- a/src/video_core/engines/maxwell_3d.cpp +++ b/src/video_core/engines/maxwell_3d.cpp @@ -9,6 +9,7 @@ #include "core/core_timing.h" #include "video_core/debug_utils/debug_utils.h" #include "video_core/engines/maxwell_3d.h" +#include "video_core/engines/shader_type.h" #include "video_core/memory_manager.h" #include "video_core/rasterizer_interface.h" #include "video_core/textures/texture.h" @@ -368,24 +369,24 @@ void Maxwell3D::CallMethod(const GPU::MethodCall& method_call) { StartCBData(method); break; } - case MAXWELL3D_REG_INDEX(cb_bind[0].raw_config): { - ProcessCBBind(Regs::ShaderStage::Vertex); + case MAXWELL3D_REG_INDEX(cb_bind[0]): { + ProcessCBBind(0); break; } - case MAXWELL3D_REG_INDEX(cb_bind[1].raw_config): { - ProcessCBBind(Regs::ShaderStage::TesselationControl); + case MAXWELL3D_REG_INDEX(cb_bind[1]): { + ProcessCBBind(1); break; } - case MAXWELL3D_REG_INDEX(cb_bind[2].raw_config): { - ProcessCBBind(Regs::ShaderStage::TesselationEval); + case MAXWELL3D_REG_INDEX(cb_bind[2]): { + ProcessCBBind(2); break; } - case MAXWELL3D_REG_INDEX(cb_bind[3].raw_config): { - ProcessCBBind(Regs::ShaderStage::Geometry); + case MAXWELL3D_REG_INDEX(cb_bind[3]): { + ProcessCBBind(3); break; } - case MAXWELL3D_REG_INDEX(cb_bind[4].raw_config): { - ProcessCBBind(Regs::ShaderStage::Fragment); + case MAXWELL3D_REG_INDEX(cb_bind[4]): { + ProcessCBBind(4); break; } case MAXWELL3D_REG_INDEX(draw.vertex_end_gl): { @@ -687,10 +688,10 @@ void Maxwell3D::DrawArrays() { } } -void Maxwell3D::ProcessCBBind(Regs::ShaderStage stage) { +void Maxwell3D::ProcessCBBind(std::size_t stage_index) { // Bind the buffer currently in CB_ADDRESS to the specified index in the desired shader stage. - auto& shader = state.shader_stages[static_cast<std::size_t>(stage)]; - auto& bind_data = regs.cb_bind[static_cast<std::size_t>(stage)]; + auto& shader = state.shader_stages[stage_index]; + auto& bind_data = regs.cb_bind[stage_index]; ASSERT(bind_data.index < Regs::MaxConstBuffers); auto& buffer = shader.const_buffers[bind_data.index]; @@ -757,9 +758,9 @@ Texture::FullTextureInfo Maxwell3D::GetTextureInfo(Texture::TextureHandle tex_ha return Texture::FullTextureInfo{GetTICEntry(tex_handle.tic_id), GetTSCEntry(tex_handle.tsc_id)}; } -Texture::FullTextureInfo Maxwell3D::GetStageTexture(Regs::ShaderStage stage, - std::size_t offset) const { - const auto& shader = state.shader_stages[static_cast<std::size_t>(stage)]; +Texture::FullTextureInfo Maxwell3D::GetStageTexture(ShaderType stage, std::size_t offset) const { + const auto stage_index = static_cast<std::size_t>(stage); + const auto& shader = state.shader_stages[stage_index]; const auto& tex_info_buffer = shader.const_buffers[regs.tex_cb_index]; ASSERT(tex_info_buffer.enabled && tex_info_buffer.address != 0); diff --git a/src/video_core/engines/maxwell_3d.h b/src/video_core/engines/maxwell_3d.h index 1aa7c274f..4cb7339b5 100644 --- a/src/video_core/engines/maxwell_3d.h +++ b/src/video_core/engines/maxwell_3d.h @@ -18,6 +18,7 @@ #include "video_core/engines/const_buffer_engine_interface.h" #include "video_core/engines/const_buffer_info.h" #include "video_core/engines/engine_upload.h" +#include "video_core/engines/shader_type.h" #include "video_core/gpu.h" #include "video_core/macro_interpreter.h" #include "video_core/textures/texture.h" @@ -62,7 +63,6 @@ public: static constexpr std::size_t NumVertexArrays = 32; static constexpr std::size_t NumVertexAttributes = 32; static constexpr std::size_t NumVaryings = 31; - static constexpr std::size_t NumTextureSamplers = 32; static constexpr std::size_t NumImages = 8; // TODO(Rodrigo): Investigate this number static constexpr std::size_t NumClipDistances = 8; static constexpr std::size_t MaxShaderProgram = 6; @@ -130,14 +130,6 @@ public: Fragment = 5, }; - enum class ShaderStage : u32 { - Vertex = 0, - TesselationControl = 1, - TesselationEval = 2, - Geometry = 3, - Fragment = 4, - }; - struct VertexAttribute { enum class Size : u32 { Invalid = 0x0, @@ -677,8 +669,8 @@ public: INSERT_UNION_PADDING_WORDS(0x15); s32 stencil_back_func_ref; - u32 stencil_back_mask; u32 stencil_back_func_mask; + u32 stencil_back_mask; INSERT_UNION_PADDING_WORDS(0xC); @@ -1254,7 +1246,7 @@ public: Texture::FullTextureInfo GetTextureInfo(Texture::TextureHandle tex_handle) const; /// Returns the texture information for a specific texture in a specific shader stage. - Texture::FullTextureInfo GetStageTexture(Regs::ShaderStage stage, std::size_t offset) const; + Texture::FullTextureInfo GetStageTexture(ShaderType stage, std::size_t offset) const; u32 AccessConstBuffer32(ShaderType stage, u64 const_buffer, u64 offset) const override; @@ -1376,7 +1368,7 @@ private: void FinishCBData(); /// Handles a write to the CB_BIND register. - void ProcessCBBind(Regs::ShaderStage stage); + void ProcessCBBind(std::size_t stage_index); /// Handles a write to the VERTEX_END_GL register, triggering a draw. void DrawArrays(); @@ -1407,8 +1399,8 @@ ASSERT_REG_POSITION(polygon_offset_line_enable, 0x371); ASSERT_REG_POSITION(polygon_offset_fill_enable, 0x372); ASSERT_REG_POSITION(scissor_test, 0x380); ASSERT_REG_POSITION(stencil_back_func_ref, 0x3D5); -ASSERT_REG_POSITION(stencil_back_mask, 0x3D6); -ASSERT_REG_POSITION(stencil_back_func_mask, 0x3D7); +ASSERT_REG_POSITION(stencil_back_func_mask, 0x3D6); +ASSERT_REG_POSITION(stencil_back_mask, 0x3D7); ASSERT_REG_POSITION(color_mask_common, 0x3E4); ASSERT_REG_POSITION(rt_separate_frag_data, 0x3EB); ASSERT_REG_POSITION(depth_bounds, 0x3EC); diff --git a/src/video_core/engines/shader_type.h b/src/video_core/engines/shader_type.h new file mode 100644 index 000000000..49ce5cde5 --- /dev/null +++ b/src/video_core/engines/shader_type.h @@ -0,0 +1,21 @@ +// Copyright 2019 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include "common/common_types.h" + +namespace Tegra::Engines { + +enum class ShaderType : u32 { + Vertex = 0, + TesselationControl = 1, + TesselationEval = 2, + Geometry = 3, + Fragment = 4, + Compute = 5, +}; +static constexpr std::size_t MaxShaderTypes = 6; + +} // namespace Tegra::Engines diff --git a/src/video_core/renderer_opengl/gl_device.cpp b/src/video_core/renderer_opengl/gl_device.cpp index b30d5be74..a95bd4b2c 100644 --- a/src/video_core/renderer_opengl/gl_device.cpp +++ b/src/video_core/renderer_opengl/gl_device.cpp @@ -5,7 +5,9 @@ #include <algorithm> #include <array> #include <cstddef> +#include <optional> #include <vector> + #include <glad/glad.h> #include "common/logging/log.h" @@ -17,6 +19,30 @@ namespace OpenGL { namespace { +// One uniform block is reserved for emulation purposes +constexpr u32 ReservedUniformBlocks = 1; + +constexpr u32 NumStages = 5; + +constexpr std::array LimitUBOs = {GL_MAX_VERTEX_UNIFORM_BLOCKS, GL_MAX_TESS_CONTROL_UNIFORM_BLOCKS, + GL_MAX_TESS_EVALUATION_UNIFORM_BLOCKS, + GL_MAX_GEOMETRY_UNIFORM_BLOCKS, GL_MAX_FRAGMENT_UNIFORM_BLOCKS}; + +constexpr std::array LimitSSBOs = { + GL_MAX_VERTEX_SHADER_STORAGE_BLOCKS, GL_MAX_TESS_CONTROL_SHADER_STORAGE_BLOCKS, + GL_MAX_TESS_EVALUATION_SHADER_STORAGE_BLOCKS, GL_MAX_GEOMETRY_SHADER_STORAGE_BLOCKS, + GL_MAX_FRAGMENT_SHADER_STORAGE_BLOCKS}; + +constexpr std::array LimitSamplers = { + GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS, GL_MAX_TESS_CONTROL_TEXTURE_IMAGE_UNITS, + GL_MAX_TESS_EVALUATION_TEXTURE_IMAGE_UNITS, GL_MAX_GEOMETRY_TEXTURE_IMAGE_UNITS, + GL_MAX_TEXTURE_IMAGE_UNITS}; + +constexpr std::array LimitImages = {GL_MAX_VERTEX_IMAGE_UNIFORMS, + GL_MAX_TESS_CONTROL_IMAGE_UNIFORMS, + GL_MAX_TESS_EVALUATION_IMAGE_UNIFORMS, + GL_MAX_GEOMETRY_IMAGE_UNIFORMS, GL_MAX_FRAGMENT_IMAGE_UNIFORMS}; + template <typename T> T GetInteger(GLenum pname) { GLint temporary; @@ -48,13 +74,70 @@ bool HasExtension(const std::vector<std::string_view>& images, std::string_view return std::find(images.begin(), images.end(), extension) != images.end(); } +u32 Extract(u32& base, u32& num, u32 amount, std::optional<GLenum> limit = {}) { + ASSERT(num >= amount); + if (limit) { + amount = std::min(amount, GetInteger<u32>(*limit)); + } + num -= amount; + return std::exchange(base, base + amount); +} + +std::array<Device::BaseBindings, Tegra::Engines::MaxShaderTypes> BuildBaseBindings() noexcept { + std::array<Device::BaseBindings, Tegra::Engines::MaxShaderTypes> bindings; + + static std::array<std::size_t, 5> stage_swizzle = {0, 1, 2, 3, 4}; + const u32 total_ubos = GetInteger<u32>(GL_MAX_UNIFORM_BUFFER_BINDINGS); + const u32 total_ssbos = GetInteger<u32>(GL_MAX_SHADER_STORAGE_BUFFER_BINDINGS); + const u32 total_samplers = GetInteger<u32>(GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS); + + u32 num_ubos = total_ubos - ReservedUniformBlocks; + u32 num_ssbos = total_ssbos; + u32 num_samplers = total_samplers; + + u32 base_ubo = ReservedUniformBlocks; + u32 base_ssbo = 0; + u32 base_samplers = 0; + + for (std::size_t i = 0; i < NumStages; ++i) { + const std::size_t stage = stage_swizzle[i]; + bindings[stage] = { + Extract(base_ubo, num_ubos, total_ubos / NumStages, LimitUBOs[stage]), + Extract(base_ssbo, num_ssbos, total_ssbos / NumStages, LimitSSBOs[stage]), + Extract(base_samplers, num_samplers, total_samplers / NumStages, LimitSamplers[stage])}; + } + + u32 num_images = GetInteger<u32>(GL_MAX_IMAGE_UNITS); + u32 base_images = 0; + + // Reserve more image bindings on fragment and vertex stages. + bindings[4].image = + Extract(base_images, num_images, num_images / NumStages + 2, LimitImages[4]); + bindings[0].image = + Extract(base_images, num_images, num_images / NumStages + 1, LimitImages[0]); + + // Reserve the other image bindings. + const u32 total_extracted_images = num_images / (NumStages - 2); + for (std::size_t i = 2; i < NumStages; ++i) { + const std::size_t stage = stage_swizzle[i]; + bindings[stage].image = + Extract(base_images, num_images, total_extracted_images, LimitImages[stage]); + } + + // Compute doesn't care about any of this. + bindings[5] = {0, 0, 0, 0}; + + return bindings; +} + } // Anonymous namespace -Device::Device() { +Device::Device() : base_bindings{BuildBaseBindings()} { const std::string_view vendor = reinterpret_cast<const char*>(glGetString(GL_VENDOR)); const std::vector extensions = GetExtensions(); const bool is_nvidia = vendor == "NVIDIA Corporation"; + const bool is_intel = vendor == "Intel"; uniform_buffer_alignment = GetInteger<std::size_t>(GL_UNIFORM_BUFFER_OFFSET_ALIGNMENT); shader_storage_alignment = GetInteger<std::size_t>(GL_SHADER_STORAGE_BUFFER_OFFSET_ALIGNMENT); @@ -68,6 +151,7 @@ Device::Device() { has_variable_aoffi = TestVariableAoffi(); has_component_indexing_bug = TestComponentIndexingBug(); has_precise_bug = TestPreciseBug(); + has_broken_compute = is_intel; has_fast_buffer_sub_data = is_nvidia; LOG_INFO(Render_OpenGL, "Renderer_VariableAOFFI: {}", has_variable_aoffi); @@ -85,6 +169,7 @@ Device::Device(std::nullptr_t) { has_image_load_formatted = true; has_variable_aoffi = true; has_component_indexing_bug = false; + has_broken_compute = false; has_precise_bug = false; } diff --git a/src/video_core/renderer_opengl/gl_device.h b/src/video_core/renderer_opengl/gl_device.h index 6c86fe207..5433815b9 100644 --- a/src/video_core/renderer_opengl/gl_device.h +++ b/src/video_core/renderer_opengl/gl_device.h @@ -6,14 +6,32 @@ #include <cstddef> #include "common/common_types.h" +#include "video_core/engines/shader_type.h" namespace OpenGL { -class Device { +static constexpr u32 EmulationUniformBlockBinding = 0; + +class Device final { public: + struct BaseBindings final { + u32 uniform_buffer{}; + u32 shader_storage_buffer{}; + u32 sampler{}; + u32 image{}; + }; + explicit Device(); explicit Device(std::nullptr_t); + const BaseBindings& GetBaseBindings(std::size_t stage_index) const noexcept { + return base_bindings[stage_index]; + } + + const BaseBindings& GetBaseBindings(Tegra::Engines::ShaderType shader_type) const noexcept { + return GetBaseBindings(static_cast<std::size_t>(shader_type)); + } + std::size_t GetUniformBufferAlignment() const { return uniform_buffer_alignment; } @@ -58,6 +76,10 @@ public: return has_precise_bug; } + bool HasBrokenCompute() const { + return has_broken_compute; + } + bool HasFastBufferSubData() const { return has_fast_buffer_sub_data; } @@ -67,6 +89,7 @@ private: static bool TestComponentIndexingBug(); static bool TestPreciseBug(); + std::array<BaseBindings, Tegra::Engines::MaxShaderTypes> base_bindings; std::size_t uniform_buffer_alignment{}; std::size_t shader_storage_alignment{}; u32 max_vertex_attributes{}; @@ -78,6 +101,7 @@ private: bool has_variable_aoffi{}; bool has_component_indexing_bug{}; bool has_precise_bug{}; + bool has_broken_compute{}; bool has_fast_buffer_sub_data{}; }; diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp index 05f8e511b..f97ec06f0 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer.cpp +++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp @@ -22,6 +22,7 @@ #include "core/settings.h" #include "video_core/engines/kepler_compute.h" #include "video_core/engines/maxwell_3d.h" +#include "video_core/engines/shader_type.h" #include "video_core/memory_manager.h" #include "video_core/renderer_opengl/gl_rasterizer.h" #include "video_core/renderer_opengl/gl_shader_cache.h" @@ -49,8 +50,25 @@ MICROPROFILE_DEFINE(OpenGL_Blits, "OpenGL", "Blits", MP_RGB(128, 128, 192)); MICROPROFILE_DEFINE(OpenGL_CacheManagement, "OpenGL", "Cache Mgmt", MP_RGB(100, 255, 100)); MICROPROFILE_DEFINE(OpenGL_PrimitiveAssembly, "OpenGL", "Prim Asmbl", MP_RGB(255, 100, 100)); -static std::size_t GetConstBufferSize(const Tegra::Engines::ConstBufferInfo& buffer, - const GLShader::ConstBufferEntry& entry) { +namespace { + +template <typename Engine, typename Entry> +Tegra::Texture::FullTextureInfo GetTextureInfo(const Engine& engine, const Entry& entry, + Tegra::Engines::ShaderType shader_type) { + if (entry.IsBindless()) { + const Tegra::Texture::TextureHandle tex_handle = + engine.AccessConstBuffer32(shader_type, entry.GetBuffer(), entry.GetOffset()); + return engine.GetTextureInfo(tex_handle); + } + if constexpr (std::is_same_v<Engine, Tegra::Engines::Maxwell3D>) { + return engine.GetStageTexture(shader_type, entry.GetOffset()); + } else { + return engine.GetTexture(entry.GetOffset()); + } +} + +std::size_t GetConstBufferSize(const Tegra::Engines::ConstBufferInfo& buffer, + const GLShader::ConstBufferEntry& entry) { if (!entry.IsIndirect()) { return entry.GetSize(); } @@ -64,6 +82,8 @@ static std::size_t GetConstBufferSize(const Tegra::Engines::ConstBufferInfo& buf return buffer.size; } +} // Anonymous namespace + RasterizerOpenGL::RasterizerOpenGL(Core::System& system, Core::Frontend::EmuWindow& emu_window, ScreenInfo& info) : texture_cache{system, *this, device}, shader_cache{*this, system, emu_window, device}, @@ -238,12 +258,11 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) { MICROPROFILE_SCOPE(OpenGL_Shader); auto& gpu = system.GPU().Maxwell3D(); - BaseBindings base_bindings; std::array<bool, Maxwell::NumClipDistances> clip_distances{}; for (std::size_t index = 0; index < Maxwell::MaxShaderProgram; ++index) { const auto& shader_config = gpu.regs.shader_config[index]; - const Maxwell::ShaderProgram program{static_cast<Maxwell::ShaderProgram>(index)}; + const auto program{static_cast<Maxwell::ShaderProgram>(index)}; // Skip stages that are not enabled if (!gpu.regs.IsShaderConfigEnabled(index)) { @@ -257,24 +276,17 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) { continue; } - GLShader::MaxwellUniformData ubo{}; - ubo.SetFromRegs(gpu); - const auto [buffer, offset] = - buffer_cache.UploadHostMemory(&ubo, sizeof(ubo), device.GetUniformBufferAlignment()); - - // Bind the emulation info buffer - bind_ubo_pushbuffer.Push(buffer, offset, static_cast<GLsizeiptr>(sizeof(ubo))); - Shader shader{shader_cache.GetStageProgram(program)}; // Stage indices are 0 - 5 - const auto stage = static_cast<Maxwell::ShaderStage>(index == 0 ? 0 : index - 1); + const std::size_t stage = index == 0 ? 0 : index - 1; SetupDrawConstBuffers(stage, shader); SetupDrawGlobalMemory(stage, shader); - const auto texture_buffer_usage{SetupDrawTextures(stage, shader, base_bindings)}; + SetupDrawTextures(stage, shader); + SetupDrawImages(stage, shader); - const ProgramVariant variant{base_bindings, primitive_mode, texture_buffer_usage}; - const auto [program_handle, next_bindings] = shader->GetProgramHandle(variant); + const ProgramVariant variant(primitive_mode); + const auto program_handle = shader->GetHandle(variant); switch (program) { case Maxwell::ShaderProgram::VertexA: @@ -303,10 +315,8 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) { // When VertexA is enabled, we have dual vertex shaders if (program == Maxwell::ShaderProgram::VertexA) { // VertexB was combined with VertexA, so we skip the VertexB iteration - index++; + ++index; } - - base_bindings = next_bindings; } SyncClipEnabled(clip_distances); @@ -591,8 +601,16 @@ void RasterizerOpenGL::DrawPrelude() { index_buffer_offset = SetupIndexBuffer(); // Prepare packed bindings. - bind_ubo_pushbuffer.Setup(0); - bind_ssbo_pushbuffer.Setup(0); + bind_ubo_pushbuffer.Setup(); + bind_ssbo_pushbuffer.Setup(); + + // Setup emulation uniform buffer. + GLShader::MaxwellUniformData ubo; + ubo.SetFromRegs(gpu); + const auto [buffer, offset] = + buffer_cache.UploadHostMemory(&ubo, sizeof(ubo), device.GetUniformBufferAlignment()); + bind_ubo_pushbuffer.Push(EmulationUniformBlockBinding, buffer, offset, + static_cast<GLsizeiptr>(sizeof(ubo))); // Setup shaders and their used resources. texture_cache.GuardSamplers(true); @@ -725,19 +743,21 @@ bool RasterizerOpenGL::DrawMultiBatch(bool is_indexed) { } void RasterizerOpenGL::DispatchCompute(GPUVAddr code_addr) { - if (!GLAD_GL_ARB_compute_variable_group_size) { - LOG_ERROR(Render_OpenGL, "Compute is currently not supported on this device due to the " - "lack of GL_ARB_compute_variable_group_size"); + if (device.HasBrokenCompute()) { return; } + buffer_cache.Acquire(); + auto kernel = shader_cache.GetComputeKernel(code_addr); - ProgramVariant variant; - variant.texture_buffer_usage = SetupComputeTextures(kernel); + SetupComputeTextures(kernel); SetupComputeImages(kernel); - const auto [program, next_bindings] = kernel->GetProgramHandle(variant); - state.draw.shader_program = program; + const auto& launch_desc = system.GPU().KeplerCompute().launch_description; + const ProgramVariant variant(launch_desc.block_dim_x, launch_desc.block_dim_y, + launch_desc.block_dim_z, launch_desc.shared_alloc, + launch_desc.local_pos_alloc); + state.draw.shader_program = kernel->GetHandle(variant); state.draw.program_pipeline = 0; const std::size_t buffer_size = @@ -745,8 +765,8 @@ void RasterizerOpenGL::DispatchCompute(GPUVAddr code_addr) { (Maxwell::MaxConstBufferSize + device.GetUniformBufferAlignment()); buffer_cache.Map(buffer_size); - bind_ubo_pushbuffer.Setup(0); - bind_ssbo_pushbuffer.Setup(0); + bind_ubo_pushbuffer.Setup(); + bind_ssbo_pushbuffer.Setup(); SetupComputeConstBuffers(kernel); SetupComputeGlobalMemory(kernel); @@ -761,10 +781,7 @@ void RasterizerOpenGL::DispatchCompute(GPUVAddr code_addr) { state.ApplyShaderProgram(); state.ApplyProgramPipeline(); - const auto& launch_desc = system.GPU().KeplerCompute().launch_description; - glDispatchComputeGroupSizeARB(launch_desc.grid_dim_x, launch_desc.grid_dim_y, - launch_desc.grid_dim_z, launch_desc.block_dim_x, - launch_desc.block_dim_y, launch_desc.block_dim_z); + glDispatchCompute(launch_desc.grid_dim_x, launch_desc.grid_dim_y, launch_desc.grid_dim_z); } void RasterizerOpenGL::FlushAll() {} @@ -833,7 +850,7 @@ bool RasterizerOpenGL::AccelerateDisplay(const Tegra::FramebufferConfig& config, ASSERT_MSG(params.height == config.height, "Framebuffer height is different"); if (params.pixel_format != pixel_format) { - LOG_WARNING(Render_OpenGL, "Framebuffer pixel_format is different"); + LOG_DEBUG(Render_OpenGL, "Framebuffer pixel_format is different"); } screen_info.display_texture = surface->GetTexture(); @@ -842,20 +859,23 @@ bool RasterizerOpenGL::AccelerateDisplay(const Tegra::FramebufferConfig& config, return true; } -void RasterizerOpenGL::SetupDrawConstBuffers(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage, - const Shader& shader) { +void RasterizerOpenGL::SetupDrawConstBuffers(std::size_t stage_index, const Shader& shader) { MICROPROFILE_SCOPE(OpenGL_UBO); const auto& stages = system.GPU().Maxwell3D().state.shader_stages; - const auto& shader_stage = stages[static_cast<std::size_t>(stage)]; + const auto& shader_stage = stages[stage_index]; + + u32 binding = device.GetBaseBindings(stage_index).uniform_buffer; for (const auto& entry : shader->GetShaderEntries().const_buffers) { const auto& buffer = shader_stage.const_buffers[entry.GetIndex()]; - SetupConstBuffer(buffer, entry); + SetupConstBuffer(binding++, buffer, entry); } } void RasterizerOpenGL::SetupComputeConstBuffers(const Shader& kernel) { MICROPROFILE_SCOPE(OpenGL_UBO); const auto& launch_desc = system.GPU().KeplerCompute().launch_description; + + u32 binding = 0; for (const auto& entry : kernel->GetShaderEntries().const_buffers) { const auto& config = launch_desc.const_buffer_config[entry.GetIndex()]; const std::bitset<8> mask = launch_desc.const_buffer_enable_mask.Value(); @@ -863,15 +883,16 @@ void RasterizerOpenGL::SetupComputeConstBuffers(const Shader& kernel) { buffer.address = config.Address(); buffer.size = config.size; buffer.enabled = mask[entry.GetIndex()]; - SetupConstBuffer(buffer, entry); + SetupConstBuffer(binding++, buffer, entry); } } -void RasterizerOpenGL::SetupConstBuffer(const Tegra::Engines::ConstBufferInfo& buffer, +void RasterizerOpenGL::SetupConstBuffer(u32 binding, const Tegra::Engines::ConstBufferInfo& buffer, const GLShader::ConstBufferEntry& entry) { if (!buffer.enabled) { // Set values to zero to unbind buffers - bind_ubo_pushbuffer.Push(buffer_cache.GetEmptyBuffer(sizeof(float)), 0, sizeof(float)); + bind_ubo_pushbuffer.Push(binding, buffer_cache.GetEmptyBuffer(sizeof(float)), 0, + sizeof(float)); return; } @@ -882,19 +903,20 @@ void RasterizerOpenGL::SetupConstBuffer(const Tegra::Engines::ConstBufferInfo& b const auto alignment = device.GetUniformBufferAlignment(); const auto [cbuf, offset] = buffer_cache.UploadMemory(buffer.address, size, alignment, false, device.HasFastBufferSubData()); - bind_ubo_pushbuffer.Push(cbuf, offset, size); + bind_ubo_pushbuffer.Push(binding, cbuf, offset, size); } -void RasterizerOpenGL::SetupDrawGlobalMemory(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage, - const Shader& shader) { +void RasterizerOpenGL::SetupDrawGlobalMemory(std::size_t stage_index, const Shader& shader) { auto& gpu{system.GPU()}; auto& memory_manager{gpu.MemoryManager()}; - const auto cbufs{gpu.Maxwell3D().state.shader_stages[static_cast<std::size_t>(stage)]}; + const auto cbufs{gpu.Maxwell3D().state.shader_stages[stage_index]}; + + u32 binding = device.GetBaseBindings(stage_index).shader_storage_buffer; for (const auto& entry : shader->GetShaderEntries().global_memory_entries) { const auto addr{cbufs.const_buffers[entry.GetCbufIndex()].address + entry.GetCbufOffset()}; const auto gpu_addr{memory_manager.Read<u64>(addr)}; const auto size{memory_manager.Read<u32>(addr + 8)}; - SetupGlobalMemory(entry, gpu_addr, size); + SetupGlobalMemory(binding++, entry, gpu_addr, size); } } @@ -902,120 +924,82 @@ void RasterizerOpenGL::SetupComputeGlobalMemory(const Shader& kernel) { auto& gpu{system.GPU()}; auto& memory_manager{gpu.MemoryManager()}; const auto cbufs{gpu.KeplerCompute().launch_description.const_buffer_config}; + + u32 binding = 0; for (const auto& entry : kernel->GetShaderEntries().global_memory_entries) { const auto addr{cbufs[entry.GetCbufIndex()].Address() + entry.GetCbufOffset()}; const auto gpu_addr{memory_manager.Read<u64>(addr)}; const auto size{memory_manager.Read<u32>(addr + 8)}; - SetupGlobalMemory(entry, gpu_addr, size); + SetupGlobalMemory(binding++, entry, gpu_addr, size); } } -void RasterizerOpenGL::SetupGlobalMemory(const GLShader::GlobalMemoryEntry& entry, +void RasterizerOpenGL::SetupGlobalMemory(u32 binding, const GLShader::GlobalMemoryEntry& entry, GPUVAddr gpu_addr, std::size_t size) { const auto alignment{device.GetShaderStorageBufferAlignment()}; const auto [ssbo, buffer_offset] = buffer_cache.UploadMemory(gpu_addr, size, alignment, entry.IsWritten()); - bind_ssbo_pushbuffer.Push(ssbo, buffer_offset, static_cast<GLsizeiptr>(size)); + bind_ssbo_pushbuffer.Push(binding, ssbo, buffer_offset, static_cast<GLsizeiptr>(size)); } -TextureBufferUsage RasterizerOpenGL::SetupDrawTextures(Maxwell::ShaderStage stage, - const Shader& shader, - BaseBindings base_bindings) { +void RasterizerOpenGL::SetupDrawTextures(std::size_t stage_index, const Shader& shader) { MICROPROFILE_SCOPE(OpenGL_Texture); - const auto& gpu = system.GPU(); - const auto& maxwell3d = gpu.Maxwell3D(); - const auto& entries = shader->GetShaderEntries().samplers; - - ASSERT_MSG(base_bindings.sampler + entries.size() <= std::size(state.textures), - "Exceeded the number of active textures."); - - TextureBufferUsage texture_buffer_usage{0}; - - for (u32 bindpoint = 0; bindpoint < entries.size(); ++bindpoint) { - const auto& entry = entries[bindpoint]; - const auto texture = [&] { - if (!entry.IsBindless()) { - return maxwell3d.GetStageTexture(stage, entry.GetOffset()); - } - const auto shader_type = static_cast<Tegra::Engines::ShaderType>(stage); - const Tegra::Texture::TextureHandle tex_handle = - maxwell3d.AccessConstBuffer32(shader_type, entry.GetBuffer(), entry.GetOffset()); - return maxwell3d.GetTextureInfo(tex_handle); - }(); - - if (SetupTexture(base_bindings.sampler + bindpoint, texture, entry)) { - texture_buffer_usage.set(bindpoint); - } + const auto& maxwell3d = system.GPU().Maxwell3D(); + u32 binding = device.GetBaseBindings(stage_index).sampler; + for (const auto& entry : shader->GetShaderEntries().samplers) { + const auto shader_type = static_cast<Tegra::Engines::ShaderType>(stage_index); + const auto texture = GetTextureInfo(maxwell3d, entry, shader_type); + SetupTexture(binding++, texture, entry); } - - return texture_buffer_usage; } -TextureBufferUsage RasterizerOpenGL::SetupComputeTextures(const Shader& kernel) { +void RasterizerOpenGL::SetupComputeTextures(const Shader& kernel) { MICROPROFILE_SCOPE(OpenGL_Texture); const auto& compute = system.GPU().KeplerCompute(); - const auto& entries = kernel->GetShaderEntries().samplers; - - ASSERT_MSG(entries.size() <= std::size(state.textures), - "Exceeded the number of active textures."); - - TextureBufferUsage texture_buffer_usage{0}; - - for (u32 bindpoint = 0; bindpoint < entries.size(); ++bindpoint) { - const auto& entry = entries[bindpoint]; - const auto texture = [&] { - if (!entry.IsBindless()) { - return compute.GetTexture(entry.GetOffset()); - } - const Tegra::Texture::TextureHandle tex_handle = compute.AccessConstBuffer32( - Tegra::Engines::ShaderType::Compute, entry.GetBuffer(), entry.GetOffset()); - return compute.GetTextureInfo(tex_handle); - }(); - - if (SetupTexture(bindpoint, texture, entry)) { - texture_buffer_usage.set(bindpoint); - } + u32 binding = 0; + for (const auto& entry : kernel->GetShaderEntries().samplers) { + const auto texture = GetTextureInfo(compute, entry, Tegra::Engines::ShaderType::Compute); + SetupTexture(binding++, texture, entry); } - - return texture_buffer_usage; } -bool RasterizerOpenGL::SetupTexture(u32 binding, const Tegra::Texture::FullTextureInfo& texture, +void RasterizerOpenGL::SetupTexture(u32 binding, const Tegra::Texture::FullTextureInfo& texture, const GLShader::SamplerEntry& entry) { - state.samplers[binding] = sampler_cache.GetSampler(texture.tsc); - const auto view = texture_cache.GetTextureSurface(texture.tic, entry); if (!view) { // Can occur when texture addr is null or its memory is unmapped/invalid + state.samplers[binding] = 0; state.textures[binding] = 0; - return false; + return; } state.textures[binding] = view->GetTexture(); if (view->GetSurfaceParams().IsBuffer()) { - return true; + return; } + state.samplers[binding] = sampler_cache.GetSampler(texture.tsc); // Apply swizzle to textures that are not buffers. view->ApplySwizzle(texture.tic.x_source, texture.tic.y_source, texture.tic.z_source, texture.tic.w_source); - return false; +} + +void RasterizerOpenGL::SetupDrawImages(std::size_t stage_index, const Shader& shader) { + const auto& maxwell3d = system.GPU().Maxwell3D(); + u32 binding = device.GetBaseBindings(stage_index).image; + for (const auto& entry : shader->GetShaderEntries().images) { + const auto shader_type = static_cast<Tegra::Engines::ShaderType>(stage_index); + const auto tic = GetTextureInfo(maxwell3d, entry, shader_type).tic; + SetupImage(binding++, tic, entry); + } } void RasterizerOpenGL::SetupComputeImages(const Shader& shader) { const auto& compute = system.GPU().KeplerCompute(); - const auto& entries = shader->GetShaderEntries().images; - for (u32 bindpoint = 0; bindpoint < entries.size(); ++bindpoint) { - const auto& entry = entries[bindpoint]; - const auto tic = [&] { - if (!entry.IsBindless()) { - return compute.GetTexture(entry.GetOffset()).tic; - } - const Tegra::Texture::TextureHandle tex_handle = compute.AccessConstBuffer32( - Tegra::Engines::ShaderType::Compute, entry.GetBuffer(), entry.GetOffset()); - return compute.GetTextureInfo(tex_handle).tic; - }(); - SetupImage(bindpoint, tic, entry); + u32 binding = 0; + for (const auto& entry : shader->GetShaderEntries().images) { + const auto tic = GetTextureInfo(compute, entry, Tegra::Engines::ShaderType::Compute).tic; + SetupImage(binding++, tic, entry); } } diff --git a/src/video_core/renderer_opengl/gl_rasterizer.h b/src/video_core/renderer_opengl/gl_rasterizer.h index bd6fe5c3a..0e47d71df 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer.h +++ b/src/video_core/renderer_opengl/gl_rasterizer.h @@ -83,42 +83,41 @@ private: bool using_depth_fb, bool using_stencil_fb); /// Configures the current constbuffers to use for the draw command. - void SetupDrawConstBuffers(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage, - const Shader& shader); + void SetupDrawConstBuffers(std::size_t stage_index, const Shader& shader); /// Configures the current constbuffers to use for the kernel invocation. void SetupComputeConstBuffers(const Shader& kernel); /// Configures a constant buffer. - void SetupConstBuffer(const Tegra::Engines::ConstBufferInfo& buffer, + void SetupConstBuffer(u32 binding, const Tegra::Engines::ConstBufferInfo& buffer, const GLShader::ConstBufferEntry& entry); /// Configures the current global memory entries to use for the draw command. - void SetupDrawGlobalMemory(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage, - const Shader& shader); + void SetupDrawGlobalMemory(std::size_t stage_index, const Shader& shader); /// Configures the current global memory entries to use for the kernel invocation. void SetupComputeGlobalMemory(const Shader& kernel); /// Configures a constant buffer. - void SetupGlobalMemory(const GLShader::GlobalMemoryEntry& entry, GPUVAddr gpu_addr, + void SetupGlobalMemory(u32 binding, const GLShader::GlobalMemoryEntry& entry, GPUVAddr gpu_addr, std::size_t size); /// Syncs all the state, shaders, render targets and textures setting before a draw call. void DrawPrelude(); - /// Configures the current textures to use for the draw command. Returns shaders texture buffer - /// usage. - TextureBufferUsage SetupDrawTextures(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage, - const Shader& shader, BaseBindings base_bindings); + /// Configures the current textures to use for the draw command. + void SetupDrawTextures(std::size_t stage_index, const Shader& shader); - /// Configures the textures used in a compute shader. Returns texture buffer usage. - TextureBufferUsage SetupComputeTextures(const Shader& kernel); + /// Configures the textures used in a compute shader. + void SetupComputeTextures(const Shader& kernel); - /// Configures a texture. Returns true when the texture is a texture buffer. - bool SetupTexture(u32 binding, const Tegra::Texture::FullTextureInfo& texture, + /// Configures a texture. + void SetupTexture(u32 binding, const Tegra::Texture::FullTextureInfo& texture, const GLShader::SamplerEntry& entry); + /// Configures images in a graphics shader. + void SetupDrawImages(std::size_t stage_index, const Shader& shader); + /// Configures images in a compute shader. void SetupComputeImages(const Shader& shader); diff --git a/src/video_core/renderer_opengl/gl_shader_cache.cpp b/src/video_core/renderer_opengl/gl_shader_cache.cpp index 04a239a39..370bdf052 100644 --- a/src/video_core/renderer_opengl/gl_shader_cache.cpp +++ b/src/video_core/renderer_opengl/gl_shader_cache.cpp @@ -8,12 +8,15 @@ #include <thread> #include <unordered_set> #include <boost/functional/hash.hpp> +#include "common/alignment.h" #include "common/assert.h" +#include "common/logging/log.h" #include "common/scope_exit.h" #include "core/core.h" #include "core/frontend/emu_window.h" #include "video_core/engines/kepler_compute.h" #include "video_core/engines/maxwell_3d.h" +#include "video_core/engines/shader_type.h" #include "video_core/memory_manager.h" #include "video_core/renderer_opengl/gl_rasterizer.h" #include "video_core/renderer_opengl/gl_shader_cache.h" @@ -82,28 +85,26 @@ std::size_t CalculateProgramSize(const GLShader::ProgramCode& program) { /// Gets the shader program code from memory for the specified address ProgramCode GetShaderCode(Tegra::MemoryManager& memory_manager, const GPUVAddr gpu_addr, const u8* host_ptr) { - ProgramCode program_code(VideoCommon::Shader::MAX_PROGRAM_LENGTH); + ProgramCode code(VideoCommon::Shader::MAX_PROGRAM_LENGTH); ASSERT_OR_EXECUTE(host_ptr != nullptr, { - std::fill(program_code.begin(), program_code.end(), 0); - return program_code; + std::fill(code.begin(), code.end(), 0); + return code; }); - memory_manager.ReadBlockUnsafe(gpu_addr, program_code.data(), - program_code.size() * sizeof(u64)); - program_code.resize(CalculateProgramSize(program_code)); - return program_code; + memory_manager.ReadBlockUnsafe(gpu_addr, code.data(), code.size() * sizeof(u64)); + code.resize(CalculateProgramSize(code)); + return code; } /// Gets the shader type from a Maxwell program type -constexpr GLenum GetShaderType(ProgramType program_type) { - switch (program_type) { - case ProgramType::VertexA: - case ProgramType::VertexB: +constexpr GLenum GetGLShaderType(ShaderType shader_type) { + switch (shader_type) { + case ShaderType::Vertex: return GL_VERTEX_SHADER; - case ProgramType::Geometry: + case ShaderType::Geometry: return GL_GEOMETRY_SHADER; - case ProgramType::Fragment: + case ShaderType::Fragment: return GL_FRAGMENT_SHADER; - case ProgramType::Compute: + case ShaderType::Compute: return GL_COMPUTE_SHADER; default: return GL_NONE; @@ -133,30 +134,11 @@ constexpr std::tuple<const char*, const char*, u32> GetPrimitiveDescription(GLen } } -ProgramType GetProgramType(Maxwell::ShaderProgram program) { - switch (program) { - case Maxwell::ShaderProgram::VertexA: - return ProgramType::VertexA; - case Maxwell::ShaderProgram::VertexB: - return ProgramType::VertexB; - case Maxwell::ShaderProgram::TesselationControl: - return ProgramType::TessellationControl; - case Maxwell::ShaderProgram::TesselationEval: - return ProgramType::TessellationEval; - case Maxwell::ShaderProgram::Geometry: - return ProgramType::Geometry; - case Maxwell::ShaderProgram::Fragment: - return ProgramType::Fragment; - } - UNREACHABLE(); - return {}; -} - /// Hashes one (or two) program streams -u64 GetUniqueIdentifier(ProgramType program_type, const ProgramCode& code, +u64 GetUniqueIdentifier(ShaderType shader_type, bool is_a, const ProgramCode& code, const ProgramCode& code_b) { u64 unique_identifier = boost::hash_value(code); - if (program_type == ProgramType::VertexA) { + if (is_a) { // VertexA programs include two programs boost::hash_combine(unique_identifier, boost::hash_value(code_b)); } @@ -164,79 +146,74 @@ u64 GetUniqueIdentifier(ProgramType program_type, const ProgramCode& code, } /// Creates an unspecialized program from code streams -std::string GenerateGLSL(const Device& device, ProgramType program_type, const ShaderIR& ir, +std::string GenerateGLSL(const Device& device, ShaderType shader_type, const ShaderIR& ir, const std::optional<ShaderIR>& ir_b) { - switch (program_type) { - case ProgramType::VertexA: - case ProgramType::VertexB: + switch (shader_type) { + case ShaderType::Vertex: return GLShader::GenerateVertexShader(device, ir, ir_b ? &*ir_b : nullptr); - case ProgramType::Geometry: + case ShaderType::Geometry: return GLShader::GenerateGeometryShader(device, ir); - case ProgramType::Fragment: + case ShaderType::Fragment: return GLShader::GenerateFragmentShader(device, ir); - case ProgramType::Compute: + case ShaderType::Compute: return GLShader::GenerateComputeShader(device, ir); default: - UNIMPLEMENTED_MSG("Unimplemented program_type={}", static_cast<u32>(program_type)); + UNIMPLEMENTED_MSG("Unimplemented shader_type={}", static_cast<u32>(shader_type)); return {}; } } -constexpr const char* GetProgramTypeName(ProgramType program_type) { - switch (program_type) { - case ProgramType::VertexA: - case ProgramType::VertexB: +constexpr const char* GetShaderTypeName(ShaderType shader_type) { + switch (shader_type) { + case ShaderType::Vertex: return "VS"; - case ProgramType::TessellationControl: - return "TCS"; - case ProgramType::TessellationEval: - return "TES"; - case ProgramType::Geometry: + case ShaderType::TesselationControl: + return "HS"; + case ShaderType::TesselationEval: + return "DS"; + case ShaderType::Geometry: return "GS"; - case ProgramType::Fragment: + case ShaderType::Fragment: return "FS"; - case ProgramType::Compute: + case ShaderType::Compute: return "CS"; } return "UNK"; } -Tegra::Engines::ShaderType GetEnginesShaderType(ProgramType program_type) { +constexpr ShaderType GetShaderType(Maxwell::ShaderProgram program_type) { switch (program_type) { - case ProgramType::VertexA: - case ProgramType::VertexB: - return Tegra::Engines::ShaderType::Vertex; - case ProgramType::TessellationControl: - return Tegra::Engines::ShaderType::TesselationControl; - case ProgramType::TessellationEval: - return Tegra::Engines::ShaderType::TesselationEval; - case ProgramType::Geometry: - return Tegra::Engines::ShaderType::Geometry; - case ProgramType::Fragment: - return Tegra::Engines::ShaderType::Fragment; - case ProgramType::Compute: - return Tegra::Engines::ShaderType::Compute; - } - UNREACHABLE(); + case Maxwell::ShaderProgram::VertexA: + case Maxwell::ShaderProgram::VertexB: + return ShaderType::Vertex; + case Maxwell::ShaderProgram::TesselationControl: + return ShaderType::TesselationControl; + case Maxwell::ShaderProgram::TesselationEval: + return ShaderType::TesselationEval; + case Maxwell::ShaderProgram::Geometry: + return ShaderType::Geometry; + case Maxwell::ShaderProgram::Fragment: + return ShaderType::Fragment; + } return {}; } -std::string GetShaderId(u64 unique_identifier, ProgramType program_type) { - return fmt::format("{}{:016X}", GetProgramTypeName(program_type), unique_identifier); +std::string GetShaderId(u64 unique_identifier, ShaderType shader_type) { + return fmt::format("{}{:016X}", GetShaderTypeName(shader_type), unique_identifier); } -Tegra::Engines::ConstBufferEngineInterface& GetConstBufferEngineInterface( - Core::System& system, ProgramType program_type) { - if (program_type == ProgramType::Compute) { +Tegra::Engines::ConstBufferEngineInterface& GetConstBufferEngineInterface(Core::System& system, + ShaderType shader_type) { + if (shader_type == ShaderType::Compute) { return system.GPU().KeplerCompute(); } else { return system.GPU().Maxwell3D(); } } -std::unique_ptr<ConstBufferLocker> MakeLocker(Core::System& system, ProgramType program_type) { - return std::make_unique<ConstBufferLocker>(GetEnginesShaderType(program_type), - GetConstBufferEngineInterface(system, program_type)); +std::unique_ptr<ConstBufferLocker> MakeLocker(Core::System& system, ShaderType shader_type) { + return std::make_unique<ConstBufferLocker>(shader_type, + GetConstBufferEngineInterface(system, shader_type)); } void FillLocker(ConstBufferLocker& locker, const ShaderDiskCacheUsage& usage) { @@ -253,33 +230,26 @@ void FillLocker(ConstBufferLocker& locker, const ShaderDiskCacheUsage& usage) { } } -CachedProgram BuildShader(const Device& device, u64 unique_identifier, ProgramType program_type, - const ProgramCode& program_code, const ProgramCode& program_code_b, - const ProgramVariant& variant, ConstBufferLocker& locker, +CachedProgram BuildShader(const Device& device, u64 unique_identifier, ShaderType shader_type, + const ProgramCode& code, const ProgramCode& code_b, + ConstBufferLocker& locker, const ProgramVariant& variant, bool hint_retrievable = false) { - LOG_INFO(Render_OpenGL, "called. {}", GetShaderId(unique_identifier, program_type)); + LOG_INFO(Render_OpenGL, "called. {}", GetShaderId(unique_identifier, shader_type)); - const bool is_compute = program_type == ProgramType::Compute; + const bool is_compute = shader_type == ShaderType::Compute; const u32 main_offset = is_compute ? KERNEL_MAIN_OFFSET : STAGE_MAIN_OFFSET; - const ShaderIR ir(program_code, main_offset, COMPILER_SETTINGS, locker); + const ShaderIR ir(code, main_offset, COMPILER_SETTINGS, locker); std::optional<ShaderIR> ir_b; - if (!program_code_b.empty()) { - ir_b.emplace(program_code_b, main_offset, COMPILER_SETTINGS, locker); + if (!code_b.empty()) { + ir_b.emplace(code_b, main_offset, COMPILER_SETTINGS, locker); } const auto entries = GLShader::GetEntries(ir); - auto base_bindings{variant.base_bindings}; - const auto primitive_mode{variant.primitive_mode}; - const auto texture_buffer_usage{variant.texture_buffer_usage}; - std::string source = fmt::format(R"(// {} #version 430 core #extension GL_ARB_separate_shader_objects : enable )", - GetShaderId(unique_identifier, program_type)); - if (is_compute) { - source += "#extension GL_ARB_compute_variable_group_size : require\n"; - } + GetShaderId(unique_identifier, shader_type)); if (device.HasShaderBallot()) { source += "#extension GL_ARB_shader_ballot : require\n"; } @@ -296,54 +266,35 @@ CachedProgram BuildShader(const Device& device, u64 unique_identifier, ProgramTy } source += '\n'; - if (!is_compute) { - source += fmt::format("#define EMULATION_UBO_BINDING {}\n", base_bindings.cbuf++); - } + if (shader_type == ShaderType::Geometry) { + const auto [glsl_topology, debug_name, max_vertices] = + GetPrimitiveDescription(variant.primitive_mode); - for (const auto& cbuf : entries.const_buffers) { - source += - fmt::format("#define CBUF_BINDING_{} {}\n", cbuf.GetIndex(), base_bindings.cbuf++); + source += fmt::format("layout ({}) in;\n\n", glsl_topology); + source += fmt::format("#define MAX_VERTEX_INPUT {}\n", max_vertices); } - for (const auto& gmem : entries.global_memory_entries) { - source += fmt::format("#define GMEM_BINDING_{}_{} {}\n", gmem.GetCbufIndex(), - gmem.GetCbufOffset(), base_bindings.gmem++); - } - for (const auto& sampler : entries.samplers) { - source += fmt::format("#define SAMPLER_BINDING_{} {}\n", sampler.GetIndex(), - base_bindings.sampler++); - } - for (const auto& image : entries.images) { + if (shader_type == ShaderType::Compute) { source += - fmt::format("#define IMAGE_BINDING_{} {}\n", image.GetIndex(), base_bindings.image++); - } + fmt::format("layout (local_size_x = {}, local_size_y = {}, local_size_z = {}) in;\n", + variant.block_x, variant.block_y, variant.block_z); - // Transform 1D textures to texture samplers by declaring its preprocessor macros. - for (std::size_t i = 0; i < texture_buffer_usage.size(); ++i) { - if (!texture_buffer_usage.test(i)) { - continue; + if (variant.shared_memory_size > 0) { + // TODO(Rodrigo): We should divide by four here, but having a larger shared memory pool + // avoids out of bound stores. Find out why shared memory size is being invalid. + source += fmt::format("shared uint smem[{}];", variant.shared_memory_size); } - source += fmt::format("#define SAMPLER_{}_IS_BUFFER\n", i); - } - if (texture_buffer_usage.any()) { - source += '\n'; - } - if (program_type == ProgramType::Geometry) { - const auto [glsl_topology, debug_name, max_vertices] = - GetPrimitiveDescription(primitive_mode); - - source += "layout (" + std::string(glsl_topology) + ") in;\n\n"; - source += "#define MAX_VERTEX_INPUT " + std::to_string(max_vertices) + '\n'; - } - if (program_type == ProgramType::Compute) { - source += "layout (local_size_variable) in;\n"; + if (variant.local_memory_size > 0) { + source += fmt::format("#define LOCAL_MEMORY_SIZE {}", + Common::AlignUp(variant.local_memory_size, 4) / 4); + } } source += '\n'; - source += GenerateGLSL(device, program_type, ir, ir_b); + source += GenerateGLSL(device, shader_type, ir, ir_b); OGLShader shader; - shader.Create(source.c_str(), GetShaderType(program_type)); + shader.Create(source.c_str(), GetGLShaderType(shader_type)); auto program = std::make_shared<OGLProgram>(); program->Create(true, hint_retrievable, shader.handle); @@ -366,18 +317,16 @@ std::unordered_set<GLenum> GetSupportedFormats() { } // Anonymous namespace -CachedShader::CachedShader(const ShaderParameters& params, ProgramType program_type, - GLShader::ShaderEntries entries, ProgramCode program_code, - ProgramCode program_code_b) - : RasterizerCacheObject{params.host_ptr}, system{params.system}, - disk_cache{params.disk_cache}, device{params.device}, cpu_addr{params.cpu_addr}, - unique_identifier{params.unique_identifier}, program_type{program_type}, entries{entries}, - program_code{std::move(program_code)}, program_code_b{std::move(program_code_b)} { +CachedShader::CachedShader(const ShaderParameters& params, ShaderType shader_type, + GLShader::ShaderEntries entries, ProgramCode code, ProgramCode code_b) + : RasterizerCacheObject{params.host_ptr}, system{params.system}, disk_cache{params.disk_cache}, + device{params.device}, cpu_addr{params.cpu_addr}, unique_identifier{params.unique_identifier}, + shader_type{shader_type}, entries{entries}, code{std::move(code)}, code_b{std::move(code_b)} { if (!params.precompiled_variants) { return; } for (const auto& pair : *params.precompiled_variants) { - auto locker = MakeLocker(system, program_type); + auto locker = MakeLocker(system, shader_type); const auto& usage = pair->first; FillLocker(*locker, usage); @@ -398,94 +347,83 @@ CachedShader::CachedShader(const ShaderParameters& params, ProgramType program_t } Shader CachedShader::CreateStageFromMemory(const ShaderParameters& params, - Maxwell::ShaderProgram program_type, - ProgramCode program_code, ProgramCode program_code_b) { - params.disk_cache.SaveRaw(ShaderDiskCacheRaw( - params.unique_identifier, GetProgramType(program_type), program_code, program_code_b)); - - ConstBufferLocker locker(GetEnginesShaderType(GetProgramType(program_type)), - params.system.GPU().Maxwell3D()); - const ShaderIR ir(program_code, STAGE_MAIN_OFFSET, COMPILER_SETTINGS, locker); + Maxwell::ShaderProgram program_type, ProgramCode code, + ProgramCode code_b) { + const auto shader_type = GetShaderType(program_type); + params.disk_cache.SaveRaw( + ShaderDiskCacheRaw(params.unique_identifier, shader_type, code, code_b)); + + ConstBufferLocker locker(shader_type, params.system.GPU().Maxwell3D()); + const ShaderIR ir(code, STAGE_MAIN_OFFSET, COMPILER_SETTINGS, locker); // TODO(Rodrigo): Handle VertexA shaders // std::optional<ShaderIR> ir_b; - // if (!program_code_b.empty()) { - // ir_b.emplace(program_code_b, STAGE_MAIN_OFFSET); + // if (!code_b.empty()) { + // ir_b.emplace(code_b, STAGE_MAIN_OFFSET); // } - return std::shared_ptr<CachedShader>( - new CachedShader(params, GetProgramType(program_type), GLShader::GetEntries(ir), - std::move(program_code), std::move(program_code_b))); + return std::shared_ptr<CachedShader>(new CachedShader( + params, shader_type, GLShader::GetEntries(ir), std::move(code), std::move(code_b))); } Shader CachedShader::CreateKernelFromMemory(const ShaderParameters& params, ProgramCode code) { params.disk_cache.SaveRaw( - ShaderDiskCacheRaw(params.unique_identifier, ProgramType::Compute, code)); + ShaderDiskCacheRaw(params.unique_identifier, ShaderType::Compute, code)); ConstBufferLocker locker(Tegra::Engines::ShaderType::Compute, params.system.GPU().KeplerCompute()); const ShaderIR ir(code, KERNEL_MAIN_OFFSET, COMPILER_SETTINGS, locker); return std::shared_ptr<CachedShader>(new CachedShader( - params, ProgramType::Compute, GLShader::GetEntries(ir), std::move(code), {})); + params, ShaderType::Compute, GLShader::GetEntries(ir), std::move(code), {})); } Shader CachedShader::CreateFromCache(const ShaderParameters& params, const UnspecializedShader& unspecialized) { - return std::shared_ptr<CachedShader>(new CachedShader(params, unspecialized.program_type, + return std::shared_ptr<CachedShader>(new CachedShader(params, unspecialized.type, unspecialized.entries, unspecialized.code, unspecialized.code_b)); } -std::tuple<GLuint, BaseBindings> CachedShader::GetProgramHandle(const ProgramVariant& variant) { - UpdateVariant(); +GLuint CachedShader::GetHandle(const ProgramVariant& variant) { + EnsureValidLockerVariant(); - const auto [entry, is_cache_miss] = curr_variant->programs.try_emplace(variant); + const auto [entry, is_cache_miss] = curr_locker_variant->programs.try_emplace(variant); auto& program = entry->second; - if (is_cache_miss) { - program = BuildShader(device, unique_identifier, program_type, program_code, program_code_b, - variant, *curr_variant->locker); - disk_cache.SaveUsage(GetUsage(variant, *curr_variant->locker)); - - LabelGLObject(GL_PROGRAM, program->handle, cpu_addr); + if (!is_cache_miss) { + return program->handle; } - auto base_bindings = variant.base_bindings; - base_bindings.cbuf += static_cast<u32>(entries.const_buffers.size()); - if (program_type != ProgramType::Compute) { - base_bindings.cbuf += STAGE_RESERVED_UBOS; - } - base_bindings.gmem += static_cast<u32>(entries.global_memory_entries.size()); - base_bindings.sampler += static_cast<u32>(entries.samplers.size()); + program = BuildShader(device, unique_identifier, shader_type, code, code_b, + *curr_locker_variant->locker, variant); + disk_cache.SaveUsage(GetUsage(variant, *curr_locker_variant->locker)); - return {program->handle, base_bindings}; + LabelGLObject(GL_PROGRAM, program->handle, cpu_addr); + return program->handle; } -void CachedShader::UpdateVariant() { - if (curr_variant && !curr_variant->locker->IsConsistent()) { - curr_variant = nullptr; +bool CachedShader::EnsureValidLockerVariant() { + const auto previous_variant = curr_locker_variant; + if (curr_locker_variant && !curr_locker_variant->locker->IsConsistent()) { + curr_locker_variant = nullptr; } - if (!curr_variant) { + if (!curr_locker_variant) { for (auto& variant : locker_variants) { if (variant->locker->IsConsistent()) { - curr_variant = variant.get(); + curr_locker_variant = variant.get(); } } } - if (!curr_variant) { + if (!curr_locker_variant) { auto& new_variant = locker_variants.emplace_back(); new_variant = std::make_unique<LockerVariant>(); - new_variant->locker = MakeLocker(system, program_type); - curr_variant = new_variant.get(); + new_variant->locker = MakeLocker(system, shader_type); + curr_locker_variant = new_variant.get(); } + return previous_variant == curr_locker_variant; } ShaderDiskCacheUsage CachedShader::GetUsage(const ProgramVariant& variant, const ConstBufferLocker& locker) const { - ShaderDiskCacheUsage usage; - usage.unique_identifier = unique_identifier; - usage.variant = variant; - usage.keys = locker.GetKeys(); - usage.bound_samplers = locker.GetBoundSamplers(); - usage.bindless_samplers = locker.GetBindlessSamplers(); - return usage; + return ShaderDiskCacheUsage{unique_identifier, variant, locker.GetKeys(), + locker.GetBoundSamplers(), locker.GetBindlessSamplers()}; } ShaderCacheOpenGL::ShaderCacheOpenGL(RasterizerOpenGL& rasterizer, Core::System& system, @@ -544,11 +482,12 @@ void ShaderCacheOpenGL::LoadDiskCache(const std::atomic_bool& stop_loading, } } if (!shader) { - auto locker{MakeLocker(system, unspecialized.program_type)}; + auto locker{MakeLocker(system, unspecialized.type)}; FillLocker(*locker, usage); - shader = BuildShader(device, usage.unique_identifier, unspecialized.program_type, - unspecialized.code, unspecialized.code_b, usage.variant, - *locker, true); + + shader = BuildShader(device, usage.unique_identifier, unspecialized.type, + unspecialized.code, unspecialized.code_b, *locker, + usage.variant, true); } std::scoped_lock lock{mutex}; @@ -651,7 +590,7 @@ bool ShaderCacheOpenGL::GenerateUnspecializedShaders( const auto& raw{raws[i]}; const u64 unique_identifier{raw.GetUniqueIdentifier()}; const u64 calculated_hash{ - GetUniqueIdentifier(raw.GetProgramType(), raw.GetProgramCode(), raw.GetProgramCodeB())}; + GetUniqueIdentifier(raw.GetType(), raw.HasProgramA(), raw.GetCode(), raw.GetCodeB())}; if (unique_identifier != calculated_hash) { LOG_ERROR(Render_OpenGL, "Invalid hash in entry={:016x} (obtained hash={:016x}) - " @@ -662,9 +601,9 @@ bool ShaderCacheOpenGL::GenerateUnspecializedShaders( } const u32 main_offset = - raw.GetProgramType() == ProgramType::Compute ? KERNEL_MAIN_OFFSET : STAGE_MAIN_OFFSET; - ConstBufferLocker locker(GetEnginesShaderType(raw.GetProgramType())); - const ShaderIR ir(raw.GetProgramCode(), main_offset, COMPILER_SETTINGS, locker); + raw.GetType() == ShaderType::Compute ? KERNEL_MAIN_OFFSET : STAGE_MAIN_OFFSET; + ConstBufferLocker locker(raw.GetType()); + const ShaderIR ir(raw.GetCode(), main_offset, COMPILER_SETTINGS, locker); // TODO(Rodrigo): Handle VertexA shaders // std::optional<ShaderIR> ir_b; // if (raw.HasProgramA()) { @@ -673,9 +612,9 @@ bool ShaderCacheOpenGL::GenerateUnspecializedShaders( UnspecializedShader unspecialized; unspecialized.entries = GLShader::GetEntries(ir); - unspecialized.program_type = raw.GetProgramType(); - unspecialized.code = raw.GetProgramCode(); - unspecialized.code_b = raw.GetProgramCodeB(); + unspecialized.type = raw.GetType(); + unspecialized.code = raw.GetCode(); + unspecialized.code_b = raw.GetCodeB(); unspecialized_shaders.emplace(raw.GetUniqueIdentifier(), unspecialized); if (callback) { @@ -708,7 +647,8 @@ Shader ShaderCacheOpenGL::GetStageProgram(Maxwell::ShaderProgram program) { code_b = GetShaderCode(memory_manager, address_b, memory_manager.GetPointer(address_b)); } - const auto unique_identifier = GetUniqueIdentifier(GetProgramType(program), code, code_b); + const auto unique_identifier = GetUniqueIdentifier( + GetShaderType(program), program == Maxwell::ShaderProgram::VertexA, code, code_b); const auto precompiled_variants = GetPrecompiledVariants(unique_identifier); const auto cpu_addr{*memory_manager.GpuToCpuAddress(address)}; const ShaderParameters params{system, disk_cache, precompiled_variants, device, @@ -736,7 +676,7 @@ Shader ShaderCacheOpenGL::GetComputeKernel(GPUVAddr code_addr) { // No kernel found - create a new one auto code{GetShaderCode(memory_manager, code_addr, host_ptr)}; - const auto unique_identifier{GetUniqueIdentifier(ProgramType::Compute, code, {})}; + const auto unique_identifier{GetUniqueIdentifier(ShaderType::Compute, false, code, {})}; const auto precompiled_variants = GetPrecompiledVariants(unique_identifier); const auto cpu_addr{*memory_manager.GpuToCpuAddress(code_addr)}; const ShaderParameters params{system, disk_cache, precompiled_variants, device, diff --git a/src/video_core/renderer_opengl/gl_shader_cache.h b/src/video_core/renderer_opengl/gl_shader_cache.h index 6bd7c9cf1..7b1470db3 100644 --- a/src/video_core/renderer_opengl/gl_shader_cache.h +++ b/src/video_core/renderer_opengl/gl_shader_cache.h @@ -17,6 +17,7 @@ #include <glad/glad.h> #include "common/common_types.h" +#include "video_core/engines/shader_type.h" #include "video_core/rasterizer_cache.h" #include "video_core/renderer_opengl/gl_resource_manager.h" #include "video_core/renderer_opengl/gl_shader_decompiler.h" @@ -47,7 +48,7 @@ using PrecompiledVariants = std::vector<PrecompiledPrograms::iterator>; struct UnspecializedShader { GLShader::ShaderEntries entries; - ProgramType program_type; + Tegra::Engines::ShaderType type; ProgramCode code; ProgramCode code_b; }; @@ -77,7 +78,7 @@ public: } std::size_t GetSizeInBytes() const override { - return program_code.size() * sizeof(u64); + return code.size() * sizeof(u64); } /// Gets the shader entries for the shader @@ -86,7 +87,7 @@ public: } /// Gets the GL program handle for the shader - std::tuple<GLuint, BaseBindings> GetProgramHandle(const ProgramVariant& variant); + GLuint GetHandle(const ProgramVariant& variant); private: struct LockerVariant { @@ -94,11 +95,11 @@ private: std::unordered_map<ProgramVariant, CachedProgram> programs; }; - explicit CachedShader(const ShaderParameters& params, ProgramType program_type, + explicit CachedShader(const ShaderParameters& params, Tegra::Engines::ShaderType shader_type, GLShader::ShaderEntries entries, ProgramCode program_code, ProgramCode program_code_b); - void UpdateVariant(); + bool EnsureValidLockerVariant(); ShaderDiskCacheUsage GetUsage(const ProgramVariant& variant, const VideoCommon::Shader::ConstBufferLocker& locker) const; @@ -110,14 +111,14 @@ private: VAddr cpu_addr{}; u64 unique_identifier{}; - ProgramType program_type{}; + Tegra::Engines::ShaderType shader_type{}; GLShader::ShaderEntries entries; - ProgramCode program_code; - ProgramCode program_code_b; + ProgramCode code; + ProgramCode code_b; - LockerVariant* curr_variant = nullptr; + LockerVariant* curr_locker_variant = nullptr; std::vector<std::unique_ptr<LockerVariant>> locker_variants; }; diff --git a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp index 4f2b49170..b17c4e703 100644 --- a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp +++ b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp @@ -16,6 +16,7 @@ #include "common/common_types.h" #include "common/logging/log.h" #include "video_core/engines/maxwell_3d.h" +#include "video_core/engines/shader_type.h" #include "video_core/renderer_opengl/gl_device.h" #include "video_core/renderer_opengl/gl_rasterizer.h" #include "video_core/renderer_opengl/gl_shader_decompiler.h" @@ -27,6 +28,7 @@ namespace OpenGL::GLShader { namespace { +using Tegra::Engines::ShaderType; using Tegra::Shader::Attribute; using Tegra::Shader::AttributeUse; using Tegra::Shader::Header; @@ -41,6 +43,9 @@ using namespace VideoCommon::Shader; using Maxwell = Tegra::Engines::Maxwell3D::Regs; using Operation = const OperationNode&; +class ASTDecompiler; +class ExprDecompiler; + enum class Type { Void, Bool, Bool2, Float, Int, Uint, HalfFloat }; struct TextureAoffi {}; @@ -223,7 +228,7 @@ private: Type type{}; }; -constexpr const char* GetTypeString(Type type) { +const char* GetTypeString(Type type) { switch (type) { case Type::Bool: return "bool"; @@ -243,7 +248,7 @@ constexpr const char* GetTypeString(Type type) { } } -constexpr const char* GetImageTypeDeclaration(Tegra::Shader::ImageType image_type) { +const char* GetImageTypeDeclaration(Tegra::Shader::ImageType image_type) { switch (image_type) { case Tegra::Shader::ImageType::Texture1D: return "1D"; @@ -331,16 +336,13 @@ std::string FlowStackTopName(MetaStackClass stack) { return fmt::format("{}_flow_stack_top", GetFlowStackPrefix(stack)); } -constexpr bool IsVertexShader(ProgramType stage) { - return stage == ProgramType::VertexA || stage == ProgramType::VertexB; +[[deprecated]] constexpr bool IsVertexShader(ShaderType stage) { + return stage == ShaderType::Vertex; } -class ASTDecompiler; -class ExprDecompiler; - class GLSLDecompiler final { public: - explicit GLSLDecompiler(const Device& device, const ShaderIR& ir, ProgramType stage, + explicit GLSLDecompiler(const Device& device, const ShaderIR& ir, ShaderType stage, std::string suffix) : device{device}, ir{ir}, stage{stage}, suffix{suffix}, header{ir.GetHeader()} {} @@ -427,7 +429,7 @@ private: } void DeclareGeometry() { - if (stage != ProgramType::Geometry) { + if (stage != ShaderType::Geometry) { return; } @@ -510,10 +512,14 @@ private: } void DeclareLocalMemory() { - // TODO(Rodrigo): Unstub kernel local memory size and pass it from a register at - // specialization time. - const u64 local_memory_size = - stage == ProgramType::Compute ? 0x400 : header.GetLocalMemorySize(); + if (stage == ShaderType::Compute) { + code.AddLine("#ifdef LOCAL_MEMORY_SIZE"); + code.AddLine("uint {}[LOCAL_MEMORY_SIZE];", GetLocalMemory()); + code.AddLine("#endif"); + return; + } + + const u64 local_memory_size = header.GetLocalMemorySize(); if (local_memory_size == 0) { return; } @@ -522,13 +528,6 @@ private: code.AddNewLine(); } - void DeclareSharedMemory() { - if (stage != ProgramType::Compute) { - return; - } - code.AddLine("shared uint {}[];", GetSharedMemory()); - } - void DeclareInternalFlags() { for (u32 flag = 0; flag < static_cast<u32>(InternalFlag::Amount); flag++) { const auto flag_code = static_cast<InternalFlag>(flag); @@ -578,12 +577,12 @@ private: const u32 location{GetGenericAttributeIndex(index)}; std::string name{GetInputAttribute(index)}; - if (stage == ProgramType::Geometry) { + if (stage == ShaderType::Geometry) { name = "gs_" + name + "[]"; } std::string suffix; - if (stage == ProgramType::Fragment) { + if (stage == ShaderType::Fragment) { const auto input_mode{header.ps.GetAttributeUse(location)}; if (skip_unused && input_mode == AttributeUse::Unused) { return; @@ -595,7 +594,7 @@ private: } void DeclareOutputAttributes() { - if (ir.HasPhysicalAttributes() && stage != ProgramType::Fragment) { + if (ir.HasPhysicalAttributes() && stage != ShaderType::Fragment) { for (u32 i = 0; i < GetNumPhysicalVaryings(); ++i) { DeclareOutputAttribute(ToGenericAttribute(i)); } @@ -620,9 +619,9 @@ private: } void DeclareConstantBuffers() { - for (const auto& entry : ir.GetConstantBuffers()) { - const auto [index, size] = entry; - code.AddLine("layout (std140, binding = CBUF_BINDING_{}) uniform {} {{", index, + u32 binding = device.GetBaseBindings(stage).uniform_buffer; + for (const auto& [index, cbuf] : ir.GetConstantBuffers()) { + code.AddLine("layout (std140, binding = {}) uniform {} {{", binding++, GetConstBufferBlock(index)); code.AddLine(" uvec4 {}[{}];", GetConstBuffer(index), MAX_CONSTBUFFER_ELEMENTS); code.AddLine("}};"); @@ -631,9 +630,8 @@ private: } void DeclareGlobalMemory() { - for (const auto& gmem : ir.GetGlobalMemory()) { - const auto& [base, usage] = gmem; - + u32 binding = device.GetBaseBindings(stage).shader_storage_buffer; + for (const auto& [base, usage] : ir.GetGlobalMemory()) { // Since we don't know how the shader will use the shader, hint the driver to disable as // much optimizations as possible std::string qualifier = "coherent volatile"; @@ -643,8 +641,8 @@ private: qualifier += " writeonly"; } - code.AddLine("layout (std430, binding = GMEM_BINDING_{}_{}) {} buffer {} {{", - base.cbuf_index, base.cbuf_offset, qualifier, GetGlobalMemoryBlock(base)); + code.AddLine("layout (std430, binding = {}) {} buffer {} {{", binding++, qualifier, + GetGlobalMemoryBlock(base)); code.AddLine(" uint {}[];", GetGlobalMemory(base)); code.AddLine("}};"); code.AddNewLine(); @@ -652,15 +650,17 @@ private: } void DeclareSamplers() { - const auto& samplers = ir.GetSamplers(); - for (const auto& sampler : samplers) { - const std::string name{GetSampler(sampler)}; - const std::string description{"layout (binding = SAMPLER_BINDING_" + - std::to_string(sampler.GetIndex()) + ") uniform"}; + u32 binding = device.GetBaseBindings(stage).sampler; + for (const auto& sampler : ir.GetSamplers()) { + const std::string name = GetSampler(sampler); + const std::string description = fmt::format("layout (binding = {}) uniform", binding++); + std::string sampler_type = [&]() { + if (sampler.IsBuffer()) { + return "samplerBuffer"; + } switch (sampler.GetType()) { case Tegra::Shader::TextureType::Texture1D: - // Special cased, read below. return "sampler1D"; case Tegra::Shader::TextureType::Texture2D: return "sampler2D"; @@ -680,21 +680,9 @@ private: sampler_type += "Shadow"; } - if (sampler.GetType() == Tegra::Shader::TextureType::Texture1D) { - // 1D textures can be aliased to texture buffers, hide the declarations behind a - // preprocessor flag and use one or the other from the GPU state. This has to be - // done because shaders don't have enough information to determine the texture type. - EmitIfdefIsBuffer(sampler); - code.AddLine("{} samplerBuffer {};", description, name); - code.AddLine("#else"); - code.AddLine("{} {} {};", description, sampler_type, name); - code.AddLine("#endif"); - } else { - // The other texture types (2D, 3D and cubes) don't have this issue. - code.AddLine("{} {} {};", description, sampler_type, name); - } + code.AddLine("{} {} {};", description, sampler_type, name); } - if (!samplers.empty()) { + if (!ir.GetSamplers().empty()) { code.AddNewLine(); } } @@ -717,7 +705,7 @@ private: constexpr u32 element_stride = 4; const u32 address{generic_base + index * generic_stride + element * element_stride}; - const bool declared = stage != ProgramType::Fragment || + const bool declared = stage != ShaderType::Fragment || header.ps.GetAttributeUse(index) != AttributeUse::Unused; const std::string value = declared ? ReadAttribute(attribute, element).AsFloat() : "0.0f"; @@ -734,8 +722,8 @@ private: } void DeclareImages() { - const auto& images{ir.GetImages()}; - for (const auto& image : images) { + u32 binding = device.GetBaseBindings(stage).image; + for (const auto& image : ir.GetImages()) { std::string qualifier = "coherent volatile"; if (image.IsRead() && !image.IsWritten()) { qualifier += " readonly"; @@ -745,10 +733,10 @@ private: const char* format = image.IsAtomic() ? "r32ui, " : ""; const char* type_declaration = GetImageTypeDeclaration(image.GetType()); - code.AddLine("layout ({}binding = IMAGE_BINDING_{}) {} uniform uimage{} {};", format, - image.GetIndex(), qualifier, type_declaration, GetImage(image)); + code.AddLine("layout ({}binding = {}) {} uniform uimage{} {};", format, binding++, + qualifier, type_declaration, GetImage(image)); } - if (!images.empty()) { + if (!ir.GetImages().empty()) { code.AddNewLine(); } } @@ -809,7 +797,7 @@ private: } if (const auto abuf = std::get_if<AbufNode>(&*node)) { - UNIMPLEMENTED_IF_MSG(abuf->IsPhysicalBuffer() && stage == ProgramType::Geometry, + UNIMPLEMENTED_IF_MSG(abuf->IsPhysicalBuffer() && stage == ShaderType::Geometry, "Physical attributes in geometry shaders are not implemented"); if (abuf->IsPhysicalBuffer()) { return {fmt::format("ReadPhysicalAttribute({})", @@ -868,18 +856,13 @@ private: } if (const auto lmem = std::get_if<LmemNode>(&*node)) { - if (stage == ProgramType::Compute) { - LOG_WARNING(Render_OpenGL, "Local memory is stubbed on compute shaders"); - } return { fmt::format("{}[{} >> 2]", GetLocalMemory(), Visit(lmem->GetAddress()).AsUint()), Type::Uint}; } if (const auto smem = std::get_if<SmemNode>(&*node)) { - return { - fmt::format("{}[{} >> 2]", GetSharedMemory(), Visit(smem->GetAddress()).AsUint()), - Type::Uint}; + return {fmt::format("smem[{} >> 2]", Visit(smem->GetAddress()).AsUint()), Type::Uint}; } if (const auto internal_flag = std::get_if<InternalFlagNode>(&*node)) { @@ -909,7 +892,7 @@ private: Expression ReadAttribute(Attribute::Index attribute, u32 element, const Node& buffer = {}) { const auto GeometryPass = [&](std::string_view name) { - if (stage == ProgramType::Geometry && buffer) { + if (stage == ShaderType::Geometry && buffer) { // TODO(Rodrigo): Guard geometry inputs against out of bound reads. Some games // set an 0x80000000 index for those and the shader fails to build. Find out why // this happens and what's its intent. @@ -921,11 +904,11 @@ private: switch (attribute) { case Attribute::Index::Position: switch (stage) { - case ProgramType::Geometry: + case ShaderType::Geometry: return {fmt::format("gl_in[{}].gl_Position{}", Visit(buffer).AsUint(), GetSwizzle(element)), Type::Float}; - case ProgramType::Fragment: + case ShaderType::Fragment: return {element == 3 ? "1.0f" : ("gl_FragCoord"s + GetSwizzle(element)), Type::Float}; default: @@ -959,7 +942,7 @@ private: return {"0", Type::Int}; case Attribute::Index::FrontFacing: // TODO(Subv): Find out what the values are for the other elements. - ASSERT(stage == ProgramType::Fragment); + ASSERT(stage == ShaderType::Fragment); switch (element) { case 3: return {"(gl_FrontFacing ? -1 : 0)", Type::Int}; @@ -985,7 +968,7 @@ private: // be found in fragment shaders, so we disable precise there. There are vertex shaders that // also fail to build but nobody seems to care about those. // Note: Only bugged drivers will skip precise. - const bool disable_precise = device.HasPreciseBug() && stage == ProgramType::Fragment; + const bool disable_precise = device.HasPreciseBug() && stage == ShaderType::Fragment; std::string temporary = code.GenerateTemporary(); code.AddLine("{}{} {} = {};", disable_precise ? "" : "precise ", GetTypeString(type), @@ -1247,17 +1230,12 @@ private: } target = std::move(*output); } else if (const auto lmem = std::get_if<LmemNode>(&*dest)) { - if (stage == ProgramType::Compute) { - LOG_WARNING(Render_OpenGL, "Local memory is stubbed on compute shaders"); - } target = { fmt::format("{}[{} >> 2]", GetLocalMemory(), Visit(lmem->GetAddress()).AsUint()), Type::Uint}; } else if (const auto smem = std::get_if<SmemNode>(&*dest)) { - ASSERT(stage == ProgramType::Compute); - target = { - fmt::format("{}[{} >> 2]", GetSharedMemory(), Visit(smem->GetAddress()).AsUint()), - Type::Uint}; + ASSERT(stage == ShaderType::Compute); + target = {fmt::format("smem[{} >> 2]", Visit(smem->GetAddress()).AsUint()), Type::Uint}; } else if (const auto gmem = std::get_if<GmemNode>(&*dest)) { const std::string real = Visit(gmem->GetRealAddress()).AsUint(); const std::string base = Visit(gmem->GetBaseAddress()).AsUint(); @@ -1749,27 +1727,14 @@ private: expr += ", "; } - // Store a copy of the expression without the lod to be used with texture buffers - std::string expr_buffer = expr; - - if (meta->lod) { + if (meta->lod && !meta->sampler.IsBuffer()) { expr += ", "; expr += Visit(meta->lod).AsInt(); } expr += ')'; expr += GetSwizzle(meta->element); - expr_buffer += ')'; - expr_buffer += GetSwizzle(meta->element); - - const std::string tmp{code.GenerateTemporary()}; - EmitIfdefIsBuffer(meta->sampler); - code.AddLine("float {} = {};", tmp, expr_buffer); - code.AddLine("#else"); - code.AddLine("float {} = {};", tmp, expr); - code.AddLine("#endif"); - - return {tmp, Type::Float}; + return {std::move(expr), Type::Float}; } Expression ImageLoad(Operation operation) { @@ -1837,7 +1802,7 @@ private: } void PreExit() { - if (stage != ProgramType::Fragment) { + if (stage != ShaderType::Fragment) { return; } const auto& used_registers = ir.GetRegisters(); @@ -1890,14 +1855,14 @@ private: } Expression EmitVertex(Operation operation) { - ASSERT_MSG(stage == ProgramType::Geometry, + ASSERT_MSG(stage == ShaderType::Geometry, "EmitVertex is expected to be used in a geometry shader."); code.AddLine("EmitVertex();"); return {}; } Expression EndPrimitive(Operation operation) { - ASSERT_MSG(stage == ProgramType::Geometry, + ASSERT_MSG(stage == ShaderType::Geometry, "EndPrimitive is expected to be used in a geometry shader."); code.AddLine("EndPrimitive();"); return {}; @@ -2193,10 +2158,6 @@ private: return "lmem_" + suffix; } - std::string GetSharedMemory() const { - return fmt::format("smem_{}", suffix); - } - std::string GetInternalFlag(InternalFlag flag) const { constexpr std::array InternalFlagNames = {"zero_flag", "sign_flag", "carry_flag", "overflow_flag"}; @@ -2214,10 +2175,6 @@ private: return GetDeclarationWithSuffix(static_cast<u32>(image.GetIndex()), "image"); } - void EmitIfdefIsBuffer(const Sampler& sampler) { - code.AddLine("#ifdef SAMPLER_{}_IS_BUFFER", sampler.GetIndex()); - } - std::string GetDeclarationWithSuffix(u32 index, std::string_view name) const { return fmt::format("{}_{}_{}", name, index, suffix); } @@ -2236,7 +2193,7 @@ private: const Device& device; const ShaderIR& ir; - const ProgramType stage; + const ShaderType stage; const std::string suffix; const Header header; @@ -2491,7 +2448,7 @@ const float fswzadd_modifiers_b[] = float[4](-1.0f, -1.0f, 1.0f, -1.0f ); )"; } -std::string Decompile(const Device& device, const ShaderIR& ir, ProgramType stage, +std::string Decompile(const Device& device, const ShaderIR& ir, ShaderType stage, const std::string& suffix) { GLSLDecompiler decompiler(device, ir, stage, suffix); decompiler.Decompile(); diff --git a/src/video_core/renderer_opengl/gl_shader_decompiler.h b/src/video_core/renderer_opengl/gl_shader_decompiler.h index b1e75e6cc..7876f48d6 100644 --- a/src/video_core/renderer_opengl/gl_shader_decompiler.h +++ b/src/video_core/renderer_opengl/gl_shader_decompiler.h @@ -10,6 +10,7 @@ #include <vector> #include "common/common_types.h" #include "video_core/engines/maxwell_3d.h" +#include "video_core/engines/shader_type.h" #include "video_core/shader/shader_ir.h" namespace VideoCommon::Shader { @@ -17,20 +18,8 @@ class ShaderIR; } namespace OpenGL { - class Device; - -enum class ProgramType : u32 { - VertexA = 0, - VertexB = 1, - TessellationControl = 2, - TessellationEval = 3, - Geometry = 4, - Fragment = 5, - Compute = 6 -}; - -} // namespace OpenGL +} namespace OpenGL::GLShader { @@ -94,6 +83,6 @@ ShaderEntries GetEntries(const VideoCommon::Shader::ShaderIR& ir); std::string GetCommonDeclarations(); std::string Decompile(const Device& device, const VideoCommon::Shader::ShaderIR& ir, - ProgramType stage, const std::string& suffix); + Tegra::Engines::ShaderType stage, const std::string& suffix); } // namespace OpenGL::GLShader diff --git a/src/video_core/renderer_opengl/gl_shader_disk_cache.cpp b/src/video_core/renderer_opengl/gl_shader_disk_cache.cpp index 184a565e6..cf874a09a 100644 --- a/src/video_core/renderer_opengl/gl_shader_disk_cache.cpp +++ b/src/video_core/renderer_opengl/gl_shader_disk_cache.cpp @@ -3,6 +3,7 @@ // Refer to the license.txt file included. #include <cstring> + #include <fmt/format.h> #include "common/assert.h" @@ -12,50 +13,50 @@ #include "common/logging/log.h" #include "common/scm_rev.h" #include "common/zstd_compression.h" - #include "core/core.h" #include "core/hle/kernel/process.h" #include "core/settings.h" - +#include "video_core/engines/shader_type.h" #include "video_core/renderer_opengl/gl_shader_cache.h" #include "video_core/renderer_opengl/gl_shader_disk_cache.h" namespace OpenGL { +using Tegra::Engines::ShaderType; using VideoCommon::Shader::BindlessSamplerMap; using VideoCommon::Shader::BoundSamplerMap; using VideoCommon::Shader::KeyMap; namespace { +using ShaderCacheVersionHash = std::array<u8, 64>; + +enum class TransferableEntryKind : u32 { + Raw, + Usage, +}; + struct ConstBufferKey { - u32 cbuf; - u32 offset; - u32 value; + u32 cbuf{}; + u32 offset{}; + u32 value{}; }; struct BoundSamplerKey { - u32 offset; - Tegra::Engines::SamplerDescriptor sampler; + u32 offset{}; + Tegra::Engines::SamplerDescriptor sampler{}; }; struct BindlessSamplerKey { - u32 cbuf; - u32 offset; - Tegra::Engines::SamplerDescriptor sampler; -}; - -using ShaderCacheVersionHash = std::array<u8, 64>; - -enum class TransferableEntryKind : u32 { - Raw, - Usage, + u32 cbuf{}; + u32 offset{}; + Tegra::Engines::SamplerDescriptor sampler{}; }; -constexpr u32 NativeVersion = 5; +constexpr u32 NativeVersion = 11; // Making sure sizes doesn't change by accident -static_assert(sizeof(BaseBindings) == 16); +static_assert(sizeof(ProgramVariant) == 20); ShaderCacheVersionHash GetShaderCacheVersionHash() { ShaderCacheVersionHash hash{}; @@ -66,10 +67,10 @@ ShaderCacheVersionHash GetShaderCacheVersionHash() { } // Anonymous namespace -ShaderDiskCacheRaw::ShaderDiskCacheRaw(u64 unique_identifier, ProgramType program_type, - ProgramCode program_code, ProgramCode program_code_b) - : unique_identifier{unique_identifier}, program_type{program_type}, - program_code{std::move(program_code)}, program_code_b{std::move(program_code_b)} {} +ShaderDiskCacheRaw::ShaderDiskCacheRaw(u64 unique_identifier, ShaderType type, ProgramCode code, + ProgramCode code_b) + : unique_identifier{unique_identifier}, type{type}, code{std::move(code)}, code_b{std::move( + code_b)} {} ShaderDiskCacheRaw::ShaderDiskCacheRaw() = default; @@ -77,42 +78,39 @@ ShaderDiskCacheRaw::~ShaderDiskCacheRaw() = default; bool ShaderDiskCacheRaw::Load(FileUtil::IOFile& file) { if (file.ReadBytes(&unique_identifier, sizeof(u64)) != sizeof(u64) || - file.ReadBytes(&program_type, sizeof(u32)) != sizeof(u32)) { + file.ReadBytes(&type, sizeof(u32)) != sizeof(u32)) { return false; } - u32 program_code_size{}; - u32 program_code_size_b{}; - if (file.ReadBytes(&program_code_size, sizeof(u32)) != sizeof(u32) || - file.ReadBytes(&program_code_size_b, sizeof(u32)) != sizeof(u32)) { + u32 code_size{}; + u32 code_size_b{}; + if (file.ReadBytes(&code_size, sizeof(u32)) != sizeof(u32) || + file.ReadBytes(&code_size_b, sizeof(u32)) != sizeof(u32)) { return false; } - program_code.resize(program_code_size); - program_code_b.resize(program_code_size_b); + code.resize(code_size); + code_b.resize(code_size_b); - if (file.ReadArray(program_code.data(), program_code_size) != program_code_size) + if (file.ReadArray(code.data(), code_size) != code_size) return false; - if (HasProgramA() && - file.ReadArray(program_code_b.data(), program_code_size_b) != program_code_size_b) { + if (HasProgramA() && file.ReadArray(code_b.data(), code_size_b) != code_size_b) { return false; } return true; } bool ShaderDiskCacheRaw::Save(FileUtil::IOFile& file) const { - if (file.WriteObject(unique_identifier) != 1 || - file.WriteObject(static_cast<u32>(program_type)) != 1 || - file.WriteObject(static_cast<u32>(program_code.size())) != 1 || - file.WriteObject(static_cast<u32>(program_code_b.size())) != 1) { + if (file.WriteObject(unique_identifier) != 1 || file.WriteObject(static_cast<u32>(type)) != 1 || + file.WriteObject(static_cast<u32>(code.size())) != 1 || + file.WriteObject(static_cast<u32>(code_b.size())) != 1) { return false; } - if (file.WriteArray(program_code.data(), program_code.size()) != program_code.size()) + if (file.WriteArray(code.data(), code.size()) != code.size()) return false; - if (HasProgramA() && - file.WriteArray(program_code_b.data(), program_code_b.size()) != program_code_b.size()) { + if (HasProgramA() && file.WriteArray(code_b.data(), code_b.size()) != code_b.size()) { return false; } return true; diff --git a/src/video_core/renderer_opengl/gl_shader_disk_cache.h b/src/video_core/renderer_opengl/gl_shader_disk_cache.h index db23ada93..69a2fbdda 100644 --- a/src/video_core/renderer_opengl/gl_shader_disk_cache.h +++ b/src/video_core/renderer_opengl/gl_shader_disk_cache.h @@ -4,7 +4,6 @@ #pragma once -#include <bitset> #include <optional> #include <string> #include <tuple> @@ -19,6 +18,7 @@ #include "common/assert.h" #include "common/common_types.h" #include "core/file_sys/vfs_vector.h" +#include "video_core/engines/shader_type.h" #include "video_core/renderer_opengl/gl_shader_gen.h" #include "video_core/shader/const_buffer_locker.h" @@ -37,42 +37,42 @@ struct ShaderDiskCacheDump; using ProgramCode = std::vector<u64>; using ShaderDumpsMap = std::unordered_map<ShaderDiskCacheUsage, ShaderDiskCacheDump>; -using TextureBufferUsage = std::bitset<64>; - -/// Allocated bindings used by an OpenGL shader program -struct BaseBindings { - u32 cbuf{}; - u32 gmem{}; - u32 sampler{}; - u32 image{}; - - bool operator==(const BaseBindings& rhs) const { - return std::tie(cbuf, gmem, sampler, image) == - std::tie(rhs.cbuf, rhs.gmem, rhs.sampler, rhs.image); - } - bool operator!=(const BaseBindings& rhs) const { - return !operator==(rhs); - } -}; -static_assert(std::is_trivially_copyable_v<BaseBindings>); +/// Describes the different variants a program can be compiled with. +struct ProgramVariant final { + ProgramVariant() = default; + + /// Graphics constructor. + explicit constexpr ProgramVariant(GLenum primitive_mode) noexcept + : primitive_mode{primitive_mode} {} + + /// Compute constructor. + explicit constexpr ProgramVariant(u32 block_x, u32 block_y, u32 block_z, u32 shared_memory_size, + u32 local_memory_size) noexcept + : block_x{block_x}, block_y{static_cast<u16>(block_y)}, block_z{static_cast<u16>(block_z)}, + shared_memory_size{shared_memory_size}, local_memory_size{local_memory_size} {} -/// Describes the different variants a single program can be compiled. -struct ProgramVariant { - BaseBindings base_bindings; + // Graphics specific parameters. GLenum primitive_mode{}; - TextureBufferUsage texture_buffer_usage{}; - bool operator==(const ProgramVariant& rhs) const { - return std::tie(base_bindings, primitive_mode, texture_buffer_usage) == - std::tie(rhs.base_bindings, rhs.primitive_mode, rhs.texture_buffer_usage); + // Compute specific parameters. + u32 block_x{}; + u16 block_y{}; + u16 block_z{}; + u32 shared_memory_size{}; + u32 local_memory_size{}; + + bool operator==(const ProgramVariant& rhs) const noexcept { + return std::tie(primitive_mode, block_x, block_y, block_z, shared_memory_size, + local_memory_size) == std::tie(rhs.primitive_mode, rhs.block_x, rhs.block_y, + rhs.block_z, rhs.shared_memory_size, + rhs.local_memory_size); } - bool operator!=(const ProgramVariant& rhs) const { + bool operator!=(const ProgramVariant& rhs) const noexcept { return !operator==(rhs); } }; - static_assert(std::is_trivially_copyable_v<ProgramVariant>); /// Describes how a shader is used. @@ -99,21 +99,14 @@ struct ShaderDiskCacheUsage { namespace std { template <> -struct hash<OpenGL::BaseBindings> { - std::size_t operator()(const OpenGL::BaseBindings& bindings) const noexcept { - return static_cast<std::size_t>(bindings.cbuf) ^ - (static_cast<std::size_t>(bindings.gmem) << 8) ^ - (static_cast<std::size_t>(bindings.sampler) << 16) ^ - (static_cast<std::size_t>(bindings.image) << 24); - } -}; - -template <> struct hash<OpenGL::ProgramVariant> { std::size_t operator()(const OpenGL::ProgramVariant& variant) const noexcept { - return std::hash<OpenGL::BaseBindings>()(variant.base_bindings) ^ - std::hash<OpenGL::TextureBufferUsage>()(variant.texture_buffer_usage) ^ - (static_cast<std::size_t>(variant.primitive_mode) << 6); + return (static_cast<std::size_t>(variant.primitive_mode) << 6) ^ + static_cast<std::size_t>(variant.block_x) ^ + (static_cast<std::size_t>(variant.block_y) << 32) ^ + (static_cast<std::size_t>(variant.block_z) << 48) ^ + (static_cast<std::size_t>(variant.shared_memory_size) << 16) ^ + (static_cast<std::size_t>(variant.local_memory_size) << 36); } }; @@ -121,7 +114,7 @@ template <> struct hash<OpenGL::ShaderDiskCacheUsage> { std::size_t operator()(const OpenGL::ShaderDiskCacheUsage& usage) const noexcept { return static_cast<std::size_t>(usage.unique_identifier) ^ - std::hash<OpenGL::ProgramVariant>()(usage.variant); + std::hash<OpenGL::ProgramVariant>{}(usage.variant); } }; @@ -132,8 +125,8 @@ namespace OpenGL { /// Describes a shader how it's used by the guest GPU class ShaderDiskCacheRaw { public: - explicit ShaderDiskCacheRaw(u64 unique_identifier, ProgramType program_type, - ProgramCode program_code, ProgramCode program_code_b = {}); + explicit ShaderDiskCacheRaw(u64 unique_identifier, Tegra::Engines::ShaderType type, + ProgramCode code, ProgramCode code_b = {}); ShaderDiskCacheRaw(); ~ShaderDiskCacheRaw(); @@ -146,27 +139,26 @@ public: } bool HasProgramA() const { - return program_type == ProgramType::VertexA; + return !code.empty() && !code_b.empty(); } - ProgramType GetProgramType() const { - return program_type; + Tegra::Engines::ShaderType GetType() const { + return type; } - const ProgramCode& GetProgramCode() const { - return program_code; + const ProgramCode& GetCode() const { + return code; } - const ProgramCode& GetProgramCodeB() const { - return program_code_b; + const ProgramCode& GetCodeB() const { + return code_b; } private: u64 unique_identifier{}; - ProgramType program_type{}; - - ProgramCode program_code; - ProgramCode program_code_b; + Tegra::Engines::ShaderType type{}; + ProgramCode code; + ProgramCode code_b; }; /// Contains an OpenGL dumped binary program diff --git a/src/video_core/renderer_opengl/gl_shader_gen.cpp b/src/video_core/renderer_opengl/gl_shader_gen.cpp index a63c1a6b8..34946fb47 100644 --- a/src/video_core/renderer_opengl/gl_shader_gen.cpp +++ b/src/video_core/renderer_opengl/gl_shader_gen.cpp @@ -2,8 +2,13 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. +#include <string> + #include <fmt/format.h> + #include "video_core/engines/maxwell_3d.h" +#include "video_core/engines/shader_type.h" +#include "video_core/renderer_opengl/gl_device.h" #include "video_core/renderer_opengl/gl_shader_decompiler.h" #include "video_core/renderer_opengl/gl_shader_gen.h" #include "video_core/shader/shader_ir.h" @@ -11,6 +16,7 @@ namespace OpenGL::GLShader { using Tegra::Engines::Maxwell3D; +using Tegra::Engines::ShaderType; using VideoCommon::Shader::CompileDepth; using VideoCommon::Shader::CompilerSettings; using VideoCommon::Shader::ProgramCode; @@ -18,16 +24,16 @@ using VideoCommon::Shader::ShaderIR; std::string GenerateVertexShader(const Device& device, const ShaderIR& ir, const ShaderIR* ir_b) { std::string out = GetCommonDeclarations(); - out += R"( -layout (std140, binding = EMULATION_UBO_BINDING) uniform vs_config { + out += fmt::format(R"( +layout (std140, binding = {}) uniform vs_config {{ float y_direction; -}; +}}; -)"; - const auto stage = ir_b ? ProgramType::VertexA : ProgramType::VertexB; - out += Decompile(device, ir, stage, "vertex"); +)", + EmulationUniformBlockBinding); + out += Decompile(device, ir, ShaderType::Vertex, "vertex"); if (ir_b) { - out += Decompile(device, *ir_b, ProgramType::VertexB, "vertex_b"); + out += Decompile(device, *ir_b, ShaderType::Vertex, "vertex_b"); } out += R"( @@ -44,13 +50,14 @@ void main() { std::string GenerateGeometryShader(const Device& device, const ShaderIR& ir) { std::string out = GetCommonDeclarations(); - out += R"( -layout (std140, binding = EMULATION_UBO_BINDING) uniform gs_config { + out += fmt::format(R"( +layout (std140, binding = {}) uniform gs_config {{ float y_direction; -}; +}}; -)"; - out += Decompile(device, ir, ProgramType::Geometry, "geometry"); +)", + EmulationUniformBlockBinding); + out += Decompile(device, ir, ShaderType::Geometry, "geometry"); out += R"( void main() { @@ -62,7 +69,7 @@ void main() { std::string GenerateFragmentShader(const Device& device, const ShaderIR& ir) { std::string out = GetCommonDeclarations(); - out += R"( + out += fmt::format(R"( layout (location = 0) out vec4 FragColor0; layout (location = 1) out vec4 FragColor1; layout (location = 2) out vec4 FragColor2; @@ -72,12 +79,13 @@ layout (location = 5) out vec4 FragColor5; layout (location = 6) out vec4 FragColor6; layout (location = 7) out vec4 FragColor7; -layout (std140, binding = EMULATION_UBO_BINDING) uniform fs_config { +layout (std140, binding = {}) uniform fs_config {{ float y_direction; -}; +}}; -)"; - out += Decompile(device, ir, ProgramType::Fragment, "fragment"); +)", + EmulationUniformBlockBinding); + out += Decompile(device, ir, ShaderType::Fragment, "fragment"); out += R"( void main() { @@ -89,7 +97,7 @@ void main() { std::string GenerateComputeShader(const Device& device, const ShaderIR& ir) { std::string out = GetCommonDeclarations(); - out += Decompile(device, ir, ProgramType::Compute, "compute"); + out += Decompile(device, ir, ShaderType::Compute, "compute"); out += R"( void main() { execute_compute(); diff --git a/src/video_core/renderer_opengl/gl_state.cpp b/src/video_core/renderer_opengl/gl_state.cpp index ccbe5912e..39b3986d3 100644 --- a/src/video_core/renderer_opengl/gl_state.cpp +++ b/src/video_core/renderer_opengl/gl_state.cpp @@ -417,14 +417,24 @@ void OpenGLState::ApplyClipControl() { } void OpenGLState::ApplyTextures() { - if (const auto update = UpdateArray(cur_state.textures, textures)) { - glBindTextures(update->first, update->second, textures.data() + update->first); + const std::size_t size = std::size(textures); + for (std::size_t i = 0; i < size; ++i) { + if (UpdateValue(cur_state.textures[i], textures[i])) { + // BindTextureUnit doesn't support binding null textures, skip those binds. + // TODO(Rodrigo): Stop using null textures + if (textures[i] != 0) { + glBindTextureUnit(static_cast<GLuint>(i), textures[i]); + } + } } } void OpenGLState::ApplySamplers() { - if (const auto update = UpdateArray(cur_state.samplers, samplers)) { - glBindSamplers(update->first, update->second, samplers.data() + update->first); + const std::size_t size = std::size(samplers); + for (std::size_t i = 0; i < size; ++i) { + if (UpdateValue(cur_state.samplers[i], samplers[i])) { + glBindSampler(static_cast<GLuint>(i), samplers[i]); + } } } diff --git a/src/video_core/renderer_opengl/gl_state.h b/src/video_core/renderer_opengl/gl_state.h index eaff22bda..e53c2c5f2 100644 --- a/src/video_core/renderer_opengl/gl_state.h +++ b/src/video_core/renderer_opengl/gl_state.h @@ -96,9 +96,11 @@ public: GLenum operation = GL_COPY; } logic_op; - std::array<GLuint, Tegra::Engines::Maxwell3D::Regs::NumTextureSamplers> textures = {}; - std::array<GLuint, Tegra::Engines::Maxwell3D::Regs::NumTextureSamplers> samplers = {}; - std::array<GLuint, Tegra::Engines::Maxwell3D::Regs::NumImages> images = {}; + static constexpr std::size_t NumSamplers = 32 * 5; + static constexpr std::size_t NumImages = 8 * 5; + std::array<GLuint, NumSamplers> textures = {}; + std::array<GLuint, NumSamplers> samplers = {}; + std::array<GLuint, NumImages> images = {}; struct { GLuint read_framebuffer = 0; // GL_READ_FRAMEBUFFER_BINDING diff --git a/src/video_core/renderer_opengl/utils.cpp b/src/video_core/renderer_opengl/utils.cpp index c504a2c1a..9770dda1c 100644 --- a/src/video_core/renderer_opengl/utils.cpp +++ b/src/video_core/renderer_opengl/utils.cpp @@ -3,7 +3,10 @@ // Refer to the license.txt file included. #include <string> +#include <vector> + #include <fmt/format.h> + #include <glad/glad.h> #include "common/assert.h" @@ -48,34 +51,19 @@ BindBuffersRangePushBuffer::BindBuffersRangePushBuffer(GLenum target) : target{t BindBuffersRangePushBuffer::~BindBuffersRangePushBuffer() = default; -void BindBuffersRangePushBuffer::Setup(GLuint first_) { - first = first_; - buffer_pointers.clear(); - offsets.clear(); - sizes.clear(); +void BindBuffersRangePushBuffer::Setup() { + entries.clear(); } -void BindBuffersRangePushBuffer::Push(const GLuint* buffer, GLintptr offset, GLsizeiptr size) { - buffer_pointers.push_back(buffer); - offsets.push_back(offset); - sizes.push_back(size); +void BindBuffersRangePushBuffer::Push(GLuint binding, const GLuint* buffer, GLintptr offset, + GLsizeiptr size) { + entries.push_back(Entry{binding, buffer, offset, size}); } void BindBuffersRangePushBuffer::Bind() { - // Ensure sizes are valid. - const std::size_t count{buffer_pointers.size()}; - DEBUG_ASSERT(count == offsets.size() && count == sizes.size()); - if (count == 0) { - return; + for (const Entry& entry : entries) { + glBindBufferRange(target, entry.binding, *entry.buffer, entry.offset, entry.size); } - - // Dereference buffers. - buffers.resize(count); - std::transform(buffer_pointers.begin(), buffer_pointers.end(), buffers.begin(), - [](const GLuint* pointer) { return *pointer; }); - - glBindBuffersRange(target, first, static_cast<GLsizei>(count), buffers.data(), offsets.data(), - sizes.data()); } void LabelGLObject(GLenum identifier, GLuint handle, VAddr addr, std::string_view extra_info) { diff --git a/src/video_core/renderer_opengl/utils.h b/src/video_core/renderer_opengl/utils.h index 6c2b45546..d56153fe7 100644 --- a/src/video_core/renderer_opengl/utils.h +++ b/src/video_core/renderer_opengl/utils.h @@ -43,20 +43,22 @@ public: explicit BindBuffersRangePushBuffer(GLenum target); ~BindBuffersRangePushBuffer(); - void Setup(GLuint first_); + void Setup(); - void Push(const GLuint* buffer, GLintptr offset, GLsizeiptr size); + void Push(GLuint binding, const GLuint* buffer, GLintptr offset, GLsizeiptr size); void Bind(); private: - GLenum target{}; - GLuint first{}; - std::vector<const GLuint*> buffer_pointers; + struct Entry { + GLuint binding; + const GLuint* buffer; + GLintptr offset; + GLsizeiptr size; + }; - std::vector<GLuint> buffers; - std::vector<GLintptr> offsets; - std::vector<GLsizeiptr> sizes; + GLenum target; + std::vector<Entry> entries; }; void LabelGLObject(GLenum identifier, GLuint handle, VAddr addr, std::string_view extra_info = {}); diff --git a/src/video_core/renderer_vulkan/maxwell_to_vk.cpp b/src/video_core/renderer_vulkan/maxwell_to_vk.cpp index 463ed43ae..7f0eb6b74 100644 --- a/src/video_core/renderer_vulkan/maxwell_to_vk.cpp +++ b/src/video_core/renderer_vulkan/maxwell_to_vk.cpp @@ -13,6 +13,8 @@ namespace Vulkan::MaxwellToVK { +using Maxwell = Tegra::Engines::Maxwell3D::Regs; + namespace Sampler { vk::Filter Filter(Tegra::Texture::TextureFilter filter) { @@ -196,17 +198,17 @@ std::pair<vk::Format, bool> SurfaceFormat(const VKDevice& device, FormatType for return {device.GetSupportedFormat(tuple.format, usage, format_type), tuple.attachable}; } -vk::ShaderStageFlagBits ShaderStage(Maxwell::ShaderStage stage) { +vk::ShaderStageFlagBits ShaderStage(Tegra::Engines::ShaderType stage) { switch (stage) { - case Maxwell::ShaderStage::Vertex: + case Tegra::Engines::ShaderType::Vertex: return vk::ShaderStageFlagBits::eVertex; - case Maxwell::ShaderStage::TesselationControl: + case Tegra::Engines::ShaderType::TesselationControl: return vk::ShaderStageFlagBits::eTessellationControl; - case Maxwell::ShaderStage::TesselationEval: + case Tegra::Engines::ShaderType::TesselationEval: return vk::ShaderStageFlagBits::eTessellationEvaluation; - case Maxwell::ShaderStage::Geometry: + case Tegra::Engines::ShaderType::Geometry: return vk::ShaderStageFlagBits::eGeometry; - case Maxwell::ShaderStage::Fragment: + case Tegra::Engines::ShaderType::Fragment: return vk::ShaderStageFlagBits::eFragment; } UNIMPLEMENTED_MSG("Unimplemented shader stage={}", static_cast<u32>(stage)); diff --git a/src/video_core/renderer_vulkan/maxwell_to_vk.h b/src/video_core/renderer_vulkan/maxwell_to_vk.h index 5b0ffd87a..904a32e01 100644 --- a/src/video_core/renderer_vulkan/maxwell_to_vk.h +++ b/src/video_core/renderer_vulkan/maxwell_to_vk.h @@ -32,7 +32,7 @@ vk::CompareOp DepthCompareFunction(Tegra::Texture::DepthCompareFunc depth_compar std::pair<vk::Format, bool> SurfaceFormat(const VKDevice& device, FormatType format_type, PixelFormat pixel_format); -vk::ShaderStageFlagBits ShaderStage(Maxwell::ShaderStage stage); +vk::ShaderStageFlagBits ShaderStage(Tegra::Engines::ShaderType stage); vk::PrimitiveTopology PrimitiveTopology(Maxwell::PrimitiveTopology topology); diff --git a/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp b/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp index 2850d5b59..80738d3d0 100644 --- a/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp +++ b/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp @@ -17,6 +17,7 @@ #include "video_core/engines/maxwell_3d.h" #include "video_core/engines/shader_bytecode.h" #include "video_core/engines/shader_header.h" +#include "video_core/engines/shader_type.h" #include "video_core/renderer_vulkan/vk_device.h" #include "video_core/renderer_vulkan/vk_shader_decompiler.h" #include "video_core/shader/node.h" @@ -25,13 +26,13 @@ namespace Vulkan::VKShader { using Sirit::Id; +using Tegra::Engines::ShaderType; using Tegra::Shader::Attribute; using Tegra::Shader::AttributeUse; using Tegra::Shader::Register; using namespace VideoCommon::Shader; using Maxwell = Tegra::Engines::Maxwell3D::Regs; -using ShaderStage = Tegra::Engines::Maxwell3D::Regs::ShaderStage; using Operation = const OperationNode&; // TODO(Rodrigo): Use rasterizer's value @@ -93,7 +94,7 @@ class ExprDecompiler; class SPIRVDecompiler : public Sirit::Module { public: - explicit SPIRVDecompiler(const VKDevice& device, const ShaderIR& ir, ShaderStage stage) + explicit SPIRVDecompiler(const VKDevice& device, const ShaderIR& ir, ShaderType stage) : Module(0x00010300), device{device}, ir{ir}, stage{stage}, header{ir.GetHeader()} { AddCapability(spv::Capability::Shader); AddExtension("SPV_KHR_storage_buffer_storage_class"); @@ -256,21 +257,21 @@ private: } void DeclareVertex() { - if (stage != ShaderStage::Vertex) + if (stage != ShaderType::Vertex) return; DeclareVertexRedeclarations(); } void DeclareGeometry() { - if (stage != ShaderStage::Geometry) + if (stage != ShaderType::Geometry) return; UNIMPLEMENTED(); } void DeclareFragment() { - if (stage != ShaderStage::Fragment) + if (stage != ShaderType::Fragment) return; for (u32 rt = 0; rt < static_cast<u32>(frag_colors.size()); ++rt) { @@ -354,7 +355,7 @@ private: continue; } - UNIMPLEMENTED_IF(stage == ShaderStage::Geometry); + UNIMPLEMENTED_IF(stage == ShaderType::Geometry); const u32 location = GetGenericAttributeLocation(index); const Id id = OpVariable(t_in_float4, spv::StorageClass::Input); @@ -364,7 +365,7 @@ private: Decorate(id, spv::Decoration::Location, location); - if (stage != ShaderStage::Fragment) { + if (stage != ShaderType::Fragment) { continue; } switch (header.ps.GetAttributeUse(location)) { @@ -548,7 +549,7 @@ private: switch (attribute) { case Attribute::Index::Position: - if (stage != ShaderStage::Fragment) { + if (stage != ShaderType::Fragment) { UNIMPLEMENTED(); break; } else { @@ -561,7 +562,7 @@ private: // TODO(Subv): Find out what the values are for the first two elements when inside a // vertex shader, and what's the value of the fourth element when inside a Tess Eval // shader. - ASSERT(stage == ShaderStage::Vertex); + ASSERT(stage == ShaderType::Vertex); switch (element) { case 2: return BitcastFrom<Type::Uint>(Emit(OpLoad(t_uint, instance_index))); @@ -572,7 +573,7 @@ private: return Constant(t_float, 0); case Attribute::Index::FrontFacing: // TODO(Subv): Find out what the values are for the other elements. - ASSERT(stage == ShaderStage::Fragment); + ASSERT(stage == ShaderType::Fragment); if (element == 3) { const Id is_front_facing = Emit(OpLoad(t_bool, front_facing)); const Id true_value = @@ -1075,7 +1076,7 @@ private: Id PreExit() { switch (stage) { - case ShaderStage::Vertex: { + case ShaderType::Vertex: { // TODO(Rodrigo): We should use VK_EXT_depth_range_unrestricted instead, but it doesn't // seem to be working on Nvidia's drivers and Intel (mesa and blob) doesn't support it. const Id z_pointer = AccessElement(t_out_float, per_vertex, position_index, 2u); @@ -1085,7 +1086,7 @@ private: Emit(OpStore(z_pointer, depth)); break; } - case ShaderStage::Fragment: { + case ShaderType::Fragment: { const auto SafeGetRegister = [&](u32 reg) { // TODO(Rodrigo): Replace with contains once C++20 releases if (const auto it = registers.find(reg); it != registers.end()) { @@ -1511,7 +1512,7 @@ private: const VKDevice& device; const ShaderIR& ir; - const ShaderStage stage; + const ShaderType stage; const Tegra::Shader::Header header; u64 conditional_nest_count{}; u64 inside_branch{}; @@ -1843,7 +1844,7 @@ void SPIRVDecompiler::DecompileAST() { } DecompilerResult Decompile(const VKDevice& device, const VideoCommon::Shader::ShaderIR& ir, - Maxwell::ShaderStage stage) { + ShaderType stage) { auto decompiler = std::make_unique<SPIRVDecompiler>(device, ir, stage); decompiler->Decompile(); return {std::move(decompiler), decompiler->GetShaderEntries()}; diff --git a/src/video_core/renderer_vulkan/vk_shader_decompiler.h b/src/video_core/renderer_vulkan/vk_shader_decompiler.h index f90541cc1..203fc00d0 100644 --- a/src/video_core/renderer_vulkan/vk_shader_decompiler.h +++ b/src/video_core/renderer_vulkan/vk_shader_decompiler.h @@ -79,6 +79,6 @@ struct ShaderEntries { using DecompilerResult = std::pair<std::unique_ptr<Sirit::Module>, ShaderEntries>; DecompilerResult Decompile(const VKDevice& device, const VideoCommon::Shader::ShaderIR& ir, - Maxwell::ShaderStage stage); + Tegra::Engines::ShaderType stage); } // namespace Vulkan::VKShader diff --git a/src/video_core/shader/const_buffer_locker.cpp b/src/video_core/shader/const_buffer_locker.cpp index fe467608e..b65399f91 100644 --- a/src/video_core/shader/const_buffer_locker.cpp +++ b/src/video_core/shader/const_buffer_locker.cpp @@ -9,6 +9,7 @@ #include "common/assert.h" #include "common/common_types.h" #include "video_core/engines/maxwell_3d.h" +#include "video_core/engines/shader_type.h" #include "video_core/shader/const_buffer_locker.h" namespace VideoCommon::Shader { diff --git a/src/video_core/shader/const_buffer_locker.h b/src/video_core/shader/const_buffer_locker.h index 600e2f3c3..50a8ce42a 100644 --- a/src/video_core/shader/const_buffer_locker.h +++ b/src/video_core/shader/const_buffer_locker.h @@ -8,6 +8,7 @@ #include "common/common_types.h" #include "common/hash.h" #include "video_core/engines/const_buffer_engine_interface.h" +#include "video_core/engines/shader_type.h" namespace VideoCommon::Shader { @@ -20,7 +21,7 @@ using BindlessSamplerMap = * The ConstBufferLocker is a class use to interface the 3D and compute engines with the shader * compiler. with it, the shader can obtain required data from GPU state and store it for disk * shader compilation. - **/ + */ class ConstBufferLocker { public: explicit ConstBufferLocker(Tegra::Engines::ShaderType shader_stage); diff --git a/src/video_core/shader/decode/texture.cpp b/src/video_core/shader/decode/texture.cpp index bb926a132..b094e5a06 100644 --- a/src/video_core/shader/decode/texture.cpp +++ b/src/video_core/shader/decode/texture.cpp @@ -128,8 +128,8 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) { } const Node component = Immediate(static_cast<u32>(instr.tld4s.component)); - const auto& sampler = - GetSampler(instr.sampler, {{TextureType::Texture2D, false, depth_compare}}); + const SamplerInfo info{TextureType::Texture2D, false, depth_compare}; + const auto& sampler = GetSampler(instr.sampler, info); Node4 values; for (u32 element = 0; element < values.size(); ++element) { @@ -149,7 +149,7 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) { // Sadly, not all texture instructions specify the type of texture their sampler // uses. This must be fixed at a later instance. const auto& sampler = - is_bindless ? GetBindlessSampler(instr.gpr8, {}) : GetSampler(instr.sampler, {}); + is_bindless ? GetBindlessSampler(instr.gpr8) : GetSampler(instr.sampler); u32 indexer = 0; switch (instr.txq.query_type) { @@ -185,8 +185,7 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) { auto texture_type = instr.tmml.texture_type.Value(); const bool is_array = instr.tmml.array != 0; const auto& sampler = - is_bindless ? GetBindlessSampler(instr.gpr20, {{texture_type, is_array, false}}) - : GetSampler(instr.sampler, {{texture_type, is_array, false}}); + is_bindless ? GetBindlessSampler(instr.gpr20) : GetSampler(instr.sampler); std::vector<Node> coords; @@ -254,67 +253,50 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) { return pc; } -const Sampler& ShaderIR::GetSampler(const Tegra::Shader::Sampler& sampler, - std::optional<SamplerInfo> sampler_info) { - const auto offset = static_cast<u32>(sampler.index.Value()); - - TextureType type; - bool is_array; - bool is_shadow; +ShaderIR::SamplerInfo ShaderIR::GetSamplerInfo(std::optional<SamplerInfo> sampler_info, u32 offset, + std::optional<u32> buffer) { if (sampler_info) { - type = sampler_info->type; - is_array = sampler_info->is_array; - is_shadow = sampler_info->is_shadow; - } else if (const auto sampler = locker.ObtainBoundSampler(offset)) { - type = sampler->texture_type.Value(); - is_array = sampler->is_array.Value() != 0; - is_shadow = sampler->is_shadow.Value() != 0; - } else { + return *sampler_info; + } + const auto sampler = + buffer ? locker.ObtainBindlessSampler(*buffer, offset) : locker.ObtainBoundSampler(offset); + if (!sampler) { LOG_WARNING(HW_GPU, "Unknown sampler info"); - type = TextureType::Texture2D; - is_array = false; - is_shadow = false; + return SamplerInfo{TextureType::Texture2D, false, false, false}; } + return SamplerInfo{sampler->texture_type, sampler->is_array != 0, sampler->is_shadow != 0, + sampler->is_buffer != 0}; +} + +const Sampler& ShaderIR::GetSampler(const Tegra::Shader::Sampler& sampler, + std::optional<SamplerInfo> sampler_info) { + const auto offset = static_cast<u32>(sampler.index.Value()); + const auto info = GetSamplerInfo(sampler_info, offset); // If this sampler has already been used, return the existing mapping. const auto it = std::find_if(used_samplers.begin(), used_samplers.end(), [offset](const Sampler& entry) { return entry.GetOffset() == offset; }); if (it != used_samplers.end()) { - ASSERT(!it->IsBindless() && it->GetType() == type && it->IsArray() == is_array && - it->IsShadow() == is_shadow); + ASSERT(!it->IsBindless() && it->GetType() == info.type && it->IsArray() == info.is_array && + it->IsShadow() == info.is_shadow && it->IsBuffer() == info.is_buffer); return *it; } // Otherwise create a new mapping for this sampler const auto next_index = static_cast<u32>(used_samplers.size()); - return used_samplers.emplace_back(Sampler(next_index, offset, type, is_array, is_shadow)); + return used_samplers.emplace_back(next_index, offset, info.type, info.is_array, info.is_shadow, + info.is_buffer); } -const Sampler& ShaderIR::GetBindlessSampler(const Tegra::Shader::Register& reg, +const Sampler& ShaderIR::GetBindlessSampler(Tegra::Shader::Register reg, std::optional<SamplerInfo> sampler_info) { const Node sampler_register = GetRegister(reg); const auto [base_sampler, buffer, offset] = TrackCbuf(sampler_register, global_code, static_cast<s64>(global_code.size())); ASSERT(base_sampler != nullptr); - TextureType type; - bool is_array; - bool is_shadow; - if (sampler_info) { - type = sampler_info->type; - is_array = sampler_info->is_array; - is_shadow = sampler_info->is_shadow; - } else if (const auto sampler = locker.ObtainBindlessSampler(buffer, offset)) { - type = sampler->texture_type.Value(); - is_array = sampler->is_array.Value() != 0; - is_shadow = sampler->is_shadow.Value() != 0; - } else { - LOG_WARNING(HW_GPU, "Unknown sampler info"); - type = TextureType::Texture2D; - is_array = false; - is_shadow = false; - } + const auto info = GetSamplerInfo(sampler_info, offset, buffer); // If this sampler has already been used, return the existing mapping. const auto it = @@ -323,15 +305,15 @@ const Sampler& ShaderIR::GetBindlessSampler(const Tegra::Shader::Register& reg, return entry.GetBuffer() == buffer && entry.GetOffset() == offset; }); if (it != used_samplers.end()) { - ASSERT(it->IsBindless() && it->GetType() == type && it->IsArray() == is_array && - it->IsShadow() == is_shadow); + ASSERT(it->IsBindless() && it->GetType() == info.type && it->IsArray() == info.is_array && + it->IsShadow() == info.is_shadow); return *it; } // Otherwise create a new mapping for this sampler const auto next_index = static_cast<u32>(used_samplers.size()); - return used_samplers.emplace_back( - Sampler(next_index, offset, buffer, type, is_array, is_shadow)); + return used_samplers.emplace_back(next_index, offset, buffer, info.type, info.is_array, + info.is_shadow, info.is_buffer); } void ShaderIR::WriteTexInstructionFloat(NodeBlock& bb, Instruction instr, const Node4& components) { @@ -416,17 +398,16 @@ Node4 ShaderIR::GetTextureCode(Instruction instr, TextureType texture_type, (texture_type == TextureType::TextureCube && is_array && is_shadow), "This method is not supported."); + const SamplerInfo info{texture_type, is_array, is_shadow, false}; const auto& sampler = - is_bindless ? GetBindlessSampler(*bindless_reg, {{texture_type, is_array, is_shadow}}) - : GetSampler(instr.sampler, {{texture_type, is_array, is_shadow}}); + is_bindless ? GetBindlessSampler(*bindless_reg, info) : GetSampler(instr.sampler, info); const bool lod_needed = process_mode == TextureProcessMode::LZ || process_mode == TextureProcessMode::LL || process_mode == TextureProcessMode::LLA; - // LOD selection (either via bias or explicit textureLod) not - // supported in GL for sampler2DArrayShadow and - // samplerCubeArrayShadow. + // LOD selection (either via bias or explicit textureLod) not supported in GL for + // sampler2DArrayShadow and samplerCubeArrayShadow. const bool gl_lod_supported = !((texture_type == Tegra::Shader::TextureType::Texture2D && is_array && is_shadow) || (texture_type == Tegra::Shader::TextureType::TextureCube && is_array && is_shadow)); @@ -436,8 +417,8 @@ Node4 ShaderIR::GetTextureCode(Instruction instr, TextureType texture_type, UNIMPLEMENTED_IF(process_mode != TextureProcessMode::None && !gl_lod_supported); - Node bias = {}; - Node lod = {}; + Node bias; + Node lod; if (process_mode != TextureProcessMode::None && gl_lod_supported) { switch (process_mode) { case TextureProcessMode::LZ: @@ -573,10 +554,9 @@ Node4 ShaderIR::GetTld4Code(Instruction instr, TextureType texture_type, bool de u64 parameter_register = instr.gpr20.Value(); - const auto& sampler = - is_bindless - ? GetBindlessSampler(parameter_register++, {{texture_type, is_array, depth_compare}}) - : GetSampler(instr.sampler, {{texture_type, is_array, depth_compare}}); + const SamplerInfo info{texture_type, is_array, depth_compare, false}; + const auto& sampler = is_bindless ? GetBindlessSampler(parameter_register++, info) + : GetSampler(instr.sampler, info); std::vector<Node> aoffi; if (is_aoffi) { @@ -623,7 +603,7 @@ Node4 ShaderIR::GetTldCode(Tegra::Shader::Instruction instr) { // const Node aoffi_register{is_aoffi ? GetRegister(gpr20_cursor++) : nullptr}; // const Node multisample{is_multisample ? GetRegister(gpr20_cursor++) : nullptr}; - const auto& sampler = GetSampler(instr.sampler, {{texture_type, is_array, false}}); + const auto& sampler = GetSampler(instr.sampler); Node4 values; for (u32 element = 0; element < values.size(); ++element) { @@ -636,6 +616,8 @@ Node4 ShaderIR::GetTldCode(Tegra::Shader::Instruction instr) { } Node4 ShaderIR::GetTldsCode(Instruction instr, TextureType texture_type, bool is_array) { + const auto& sampler = GetSampler(instr.sampler); + const std::size_t type_coord_count = GetCoordCount(texture_type); const bool lod_enabled = instr.tlds.GetTextureProcessMode() == TextureProcessMode::LL; @@ -659,7 +641,14 @@ Node4 ShaderIR::GetTldsCode(Instruction instr, TextureType texture_type, bool is // When lod is used always is in gpr20 const Node lod = lod_enabled ? GetRegister(instr.gpr20) : Immediate(0); - const auto& sampler = GetSampler(instr.sampler, {{texture_type, is_array, false}}); + // Fill empty entries from the guest sampler. + const std::size_t entry_coord_count = GetCoordCount(sampler.GetType()); + if (type_coord_count != entry_coord_count) { + LOG_WARNING(HW_GPU, "Bound and built texture types mismatch"); + } + for (std::size_t i = type_coord_count; i < entry_coord_count; ++i) { + coords.push_back(GetRegister(Register::ZeroIndex)); + } Node4 values; for (u32 element = 0; element < values.size(); ++element) { diff --git a/src/video_core/shader/node.h b/src/video_core/shader/node.h index 54217e6a4..44d85d434 100644 --- a/src/video_core/shader/node.h +++ b/src/video_core/shader/node.h @@ -225,14 +225,15 @@ class Sampler { public: /// This constructor is for bound samplers constexpr explicit Sampler(u32 index, u32 offset, Tegra::Shader::TextureType type, - bool is_array, bool is_shadow) - : index{index}, offset{offset}, type{type}, is_array{is_array}, is_shadow{is_shadow} {} + bool is_array, bool is_shadow, bool is_buffer) + : index{index}, offset{offset}, type{type}, is_array{is_array}, is_shadow{is_shadow}, + is_buffer{is_buffer} {} /// This constructor is for bindless samplers constexpr explicit Sampler(u32 index, u32 offset, u32 buffer, Tegra::Shader::TextureType type, - bool is_array, bool is_shadow) + bool is_array, bool is_shadow, bool is_buffer) : index{index}, offset{offset}, buffer{buffer}, type{type}, is_array{is_array}, - is_shadow{is_shadow}, is_bindless{true} {} + is_shadow{is_shadow}, is_buffer{is_buffer}, is_bindless{true} {} constexpr u32 GetIndex() const { return index; @@ -258,6 +259,10 @@ public: return is_shadow; } + constexpr bool IsBuffer() const { + return is_buffer; + } + constexpr bool IsBindless() const { return is_bindless; } @@ -270,6 +275,7 @@ private: Tegra::Shader::TextureType type{}; ///< The type used to sample this texture (Texture2D, etc) bool is_array{}; ///< Whether the texture is being sampled as an array texture or not. bool is_shadow{}; ///< Whether the texture is being sampled as a depth texture or not. + bool is_buffer{}; ///< Whether the texture is a texture buffer without sampler. bool is_bindless{}; ///< Whether this sampler belongs to a bindless texture or not. }; diff --git a/src/video_core/shader/shader_ir.h b/src/video_core/shader/shader_ir.h index 76a849818..2f71a50d2 100644 --- a/src/video_core/shader/shader_ir.h +++ b/src/video_core/shader/shader_ir.h @@ -179,6 +179,7 @@ private: Tegra::Shader::TextureType type; bool is_array; bool is_shadow; + bool is_buffer; }; void Decode(); @@ -303,13 +304,17 @@ private: /// Returns a predicate combiner operation OperationCode GetPredicateCombiner(Tegra::Shader::PredOperation operation); + /// Queries the missing sampler info from the execution context. + SamplerInfo GetSamplerInfo(std::optional<SamplerInfo> sampler_info, u32 offset, + std::optional<u32> buffer = std::nullopt); + /// Accesses a texture sampler const Sampler& GetSampler(const Tegra::Shader::Sampler& sampler, - std::optional<SamplerInfo> sampler_info); + std::optional<SamplerInfo> sampler_info = std::nullopt); - // Accesses a texture sampler for a bindless texture. - const Sampler& GetBindlessSampler(const Tegra::Shader::Register& reg, - std::optional<SamplerInfo> sampler_info); + /// Accesses a texture sampler for a bindless texture. + const Sampler& GetBindlessSampler(Tegra::Shader::Register reg, + std::optional<SamplerInfo> sampler_info = std::nullopt); /// Accesses an image. Image& GetImage(Tegra::Shader::Image image, Tegra::Shader::ImageType type); diff --git a/src/yuzu/debugger/wait_tree.cpp b/src/yuzu/debugger/wait_tree.cpp index 188f798c0..535b3ce90 100644 --- a/src/yuzu/debugger/wait_tree.cpp +++ b/src/yuzu/debugger/wait_tree.cpp @@ -57,7 +57,7 @@ std::size_t WaitTreeItem::Row() const { std::vector<std::unique_ptr<WaitTreeThread>> WaitTreeItem::MakeThreadItemList() { std::vector<std::unique_ptr<WaitTreeThread>> item_list; std::size_t row = 0; - auto add_threads = [&](const std::vector<Kernel::SharedPtr<Kernel::Thread>>& threads) { + auto add_threads = [&](const std::vector<std::shared_ptr<Kernel::Thread>>& threads) { for (std::size_t i = 0; i < threads.size(); ++i) { item_list.push_back(std::make_unique<WaitTreeThread>(*threads[i])); item_list.back()->row = row; @@ -172,8 +172,8 @@ std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeWaitObject::GetChildren() con return list; } -WaitTreeObjectList::WaitTreeObjectList( - const std::vector<Kernel::SharedPtr<Kernel::WaitObject>>& list, bool w_all) +WaitTreeObjectList::WaitTreeObjectList(const std::vector<std::shared_ptr<Kernel::WaitObject>>& list, + bool w_all) : object_list(list), wait_all(w_all) {} WaitTreeObjectList::~WaitTreeObjectList() = default; @@ -325,7 +325,7 @@ std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeThread::GetChildren() const { WaitTreeEvent::WaitTreeEvent(const Kernel::ReadableEvent& object) : WaitTreeWaitObject(object) {} WaitTreeEvent::~WaitTreeEvent() = default; -WaitTreeThreadList::WaitTreeThreadList(const std::vector<Kernel::SharedPtr<Kernel::Thread>>& list) +WaitTreeThreadList::WaitTreeThreadList(const std::vector<std::shared_ptr<Kernel::Thread>>& list) : thread_list(list) {} WaitTreeThreadList::~WaitTreeThreadList() = default; diff --git a/src/yuzu/debugger/wait_tree.h b/src/yuzu/debugger/wait_tree.h index f2b13be24..631274a5f 100644 --- a/src/yuzu/debugger/wait_tree.h +++ b/src/yuzu/debugger/wait_tree.h @@ -83,7 +83,7 @@ private: VAddr mutex_address; u32 mutex_value; Kernel::Handle owner_handle; - Kernel::SharedPtr<Kernel::Thread> owner; + std::shared_ptr<Kernel::Thread> owner; }; class WaitTreeCallstack : public WaitTreeExpandableItem { @@ -116,15 +116,14 @@ protected: class WaitTreeObjectList : public WaitTreeExpandableItem { Q_OBJECT public: - WaitTreeObjectList(const std::vector<Kernel::SharedPtr<Kernel::WaitObject>>& list, - bool wait_all); + WaitTreeObjectList(const std::vector<std::shared_ptr<Kernel::WaitObject>>& list, bool wait_all); ~WaitTreeObjectList() override; QString GetText() const override; std::vector<std::unique_ptr<WaitTreeItem>> GetChildren() const override; private: - const std::vector<Kernel::SharedPtr<Kernel::WaitObject>>& object_list; + const std::vector<std::shared_ptr<Kernel::WaitObject>>& object_list; bool wait_all; }; @@ -149,14 +148,14 @@ public: class WaitTreeThreadList : public WaitTreeExpandableItem { Q_OBJECT public: - explicit WaitTreeThreadList(const std::vector<Kernel::SharedPtr<Kernel::Thread>>& list); + explicit WaitTreeThreadList(const std::vector<std::shared_ptr<Kernel::Thread>>& list); ~WaitTreeThreadList() override; QString GetText() const override; std::vector<std::unique_ptr<WaitTreeItem>> GetChildren() const override; private: - const std::vector<Kernel::SharedPtr<Kernel::Thread>>& thread_list; + const std::vector<std::shared_ptr<Kernel::Thread>>& thread_list; }; class WaitTreeModel : public QAbstractItemModel { |