diff options
Diffstat (limited to 'src/core/hle')
42 files changed, 1475 insertions, 812 deletions
diff --git a/src/core/hle/ipc.h b/src/core/hle/ipc.h index fae54bcc7..7ce313190 100644 --- a/src/core/hle/ipc.h +++ b/src/core/hle/ipc.h @@ -160,7 +160,7 @@ struct DomainMessageHeader { // Used when responding to an IPC request, Server -> Client. struct { u32_le num_objects; - INSERT_PADDING_WORDS(3); + INSERT_UNION_PADDING_WORDS(3); }; // Used when performing an IPC request, Client -> Server. @@ -171,8 +171,10 @@ struct DomainMessageHeader { BitField<16, 16, u32> size; }; u32_le object_id; - INSERT_PADDING_WORDS(2); + INSERT_UNION_PADDING_WORDS(2); }; + + std::array<u32, 4> raw{}; }; }; static_assert(sizeof(DomainMessageHeader) == 16, "DomainMessageHeader size is incorrect"); diff --git a/src/core/hle/kernel/address_arbiter.cpp b/src/core/hle/kernel/address_arbiter.cpp index c8842410b..de0a9064e 100644 --- a/src/core/hle/kernel/address_arbiter.cpp +++ b/src/core/hle/kernel/address_arbiter.cpp @@ -22,6 +22,7 @@ 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) { + auto& system = Core::System::GetInstance(); // Only process up to 'target' threads, unless 'target' is <= 0, in which case process // them all. std::size_t last = waiting_threads.size(); @@ -35,6 +36,7 @@ void WakeThreads(const std::vector<SharedPtr<Thread>>& waiting_threads, s32 num_ waiting_threads[i]->SetWaitSynchronizationResult(RESULT_SUCCESS); waiting_threads[i]->SetArbiterWaitAddress(0); waiting_threads[i]->ResumeFromWait(); + system.PrepareReschedule(waiting_threads[i]->GetProcessorID()); } } } // Anonymous namespace @@ -89,12 +91,20 @@ ResultCode AddressArbiter::ModifyByWaitingCountAndSignalToAddressIfEqual(VAddr a // Determine the modified value depending on the waiting count. s32 updated_value; - if (waiting_threads.empty()) { - updated_value = value + 1; - } else if (num_to_wake <= 0 || waiting_threads.size() <= static_cast<u32>(num_to_wake)) { - updated_value = value - 1; + if (num_to_wake <= 0) { + if (waiting_threads.empty()) { + updated_value = value + 1; + } else { + updated_value = value - 1; + } } else { - updated_value = value; + if (waiting_threads.empty()) { + updated_value = value + 1; + } else if (waiting_threads.size() <= static_cast<u32>(num_to_wake)) { + updated_value = value - 1; + } else { + updated_value = value; + } } if (static_cast<s32>(Memory::Read32(address)) != value) { @@ -169,30 +179,22 @@ ResultCode AddressArbiter::WaitForAddressImpl(VAddr address, s64 timeout) { current_thread->WakeAfterDelay(timeout); - system.CpuCore(current_thread->GetProcessorID()).PrepareReschedule(); + system.PrepareReschedule(current_thread->GetProcessorID()); return RESULT_TIMEOUT; } std::vector<SharedPtr<Thread>> AddressArbiter::GetThreadsWaitingOnAddress(VAddr address) const { - const auto RetrieveWaitingThreads = [this](std::size_t core_index, - std::vector<SharedPtr<Thread>>& waiting_threads, - VAddr arb_addr) { - const auto& scheduler = system.Scheduler(core_index); - const auto& thread_list = scheduler.GetThreadList(); - - for (const auto& thread : thread_list) { - if (thread->GetArbiterWaitAddress() == arb_addr) { - waiting_threads.push_back(thread); - } - } - }; // Retrieve all threads that are waiting for this address. std::vector<SharedPtr<Thread>> threads; - RetrieveWaitingThreads(0, threads, address); - RetrieveWaitingThreads(1, threads, address); - RetrieveWaitingThreads(2, threads, address); - RetrieveWaitingThreads(3, threads, address); + const auto& scheduler = system.GlobalScheduler(); + const auto& thread_list = scheduler.GetThreadList(); + + for (const auto& thread : thread_list) { + if (thread->GetArbiterWaitAddress() == address) { + threads.push_back(thread); + } + } // Sort them by priority, such that the highest priority ones come first. std::sort(threads.begin(), threads.end(), diff --git a/src/core/hle/kernel/kernel.cpp b/src/core/hle/kernel/kernel.cpp index 799e5e0d8..f94ac150d 100644 --- a/src/core/hle/kernel/kernel.cpp +++ b/src/core/hle/kernel/kernel.cpp @@ -12,12 +12,15 @@ #include "core/core.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/errors.h" #include "core/hle/kernel/handle_table.h" #include "core/hle/kernel/kernel.h" #include "core/hle/kernel/process.h" #include "core/hle/kernel/resource_limit.h" +#include "core/hle/kernel/scheduler.h" #include "core/hle/kernel/thread.h" #include "core/hle/lock.h" #include "core/hle/result.h" @@ -58,12 +61,8 @@ static void ThreadWakeupCallback(u64 thread_handle, [[maybe_unused]] s64 cycles_ if (thread->HasWakeupCallback()) { resume = thread->InvokeWakeupCallback(ThreadWakeupReason::Timeout, thread, nullptr, 0); } - } - - if (thread->GetMutexWaitAddress() != 0 || thread->GetCondVarWaitAddress() != 0 || - thread->GetWaitHandle() != 0) { - ASSERT(thread->GetStatus() == ThreadStatus::WaitMutex || - thread->GetStatus() == ThreadStatus::WaitCondVar); + } else if (thread->GetStatus() == ThreadStatus::WaitMutex || + thread->GetStatus() == ThreadStatus::WaitCondVar) { thread->SetMutexWaitAddress(0); thread->SetCondVarWaitAddress(0); thread->SetWaitHandle(0); @@ -83,18 +82,23 @@ static void ThreadWakeupCallback(u64 thread_handle, [[maybe_unused]] s64 cycles_ } if (resume) { + if (thread->GetStatus() == ThreadStatus::WaitCondVar || + thread->GetStatus() == ThreadStatus::WaitArb) { + thread->SetWaitSynchronizationResult(RESULT_TIMEOUT); + } thread->ResumeFromWait(); } } struct KernelCore::Impl { - explicit Impl(Core::System& system) : system{system} {} + explicit Impl(Core::System& system) : system{system}, global_scheduler{system} {} void Initialize(KernelCore& kernel) { Shutdown(); InitializeSystemResourceLimit(kernel); InitializeThreads(); + InitializePreemption(); } void Shutdown() { @@ -110,6 +114,9 @@ struct KernelCore::Impl { thread_wakeup_callback_handle_table.Clear(); thread_wakeup_event_type = nullptr; + preemption_event = nullptr; + + global_scheduler.Shutdown(); named_ports.clear(); } @@ -132,6 +139,18 @@ struct KernelCore::Impl { system.CoreTiming().RegisterEvent("ThreadWakeupCallback", ThreadWakeupCallback); } + void InitializePreemption() { + preemption_event = system.CoreTiming().RegisterEvent( + "PreemptionCallback", [this](u64 userdata, s64 cycles_late) { + global_scheduler.PreemptThreads(); + s64 time_interval = Core::Timing::msToCycles(std::chrono::milliseconds(10)); + system.CoreTiming().ScheduleEvent(time_interval, preemption_event); + }); + + s64 time_interval = Core::Timing::msToCycles(std::chrono::milliseconds(10)); + system.CoreTiming().ScheduleEvent(time_interval, preemption_event); + } + std::atomic<u32> next_object_id{0}; std::atomic<u64> next_kernel_process_id{Process::InitialKIPIDMin}; std::atomic<u64> next_user_process_id{Process::ProcessIDMin}; @@ -140,10 +159,12 @@ struct KernelCore::Impl { // Lists all processes that exist in the current session. std::vector<SharedPtr<Process>> process_list; Process* current_process = nullptr; + Kernel::GlobalScheduler global_scheduler; SharedPtr<ResourceLimit> system_resource_limit; Core::Timing::EventType* thread_wakeup_event_type = nullptr; + Core::Timing::EventType* preemption_event = nullptr; // TODO(yuriks): This can be removed if Thread objects are explicitly pooled in the future, // allowing us to simply use a pool index or similar. Kernel::HandleTable thread_wakeup_callback_handle_table; @@ -203,6 +224,14 @@ const std::vector<SharedPtr<Process>>& KernelCore::GetProcessList() const { return impl->process_list; } +Kernel::GlobalScheduler& KernelCore::GlobalScheduler() { + return impl->global_scheduler; +} + +const Kernel::GlobalScheduler& KernelCore::GlobalScheduler() const { + return impl->global_scheduler; +} + void KernelCore::AddNamedPort(std::string name, SharedPtr<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 0cc44ee76..c4397fc77 100644 --- a/src/core/hle/kernel/kernel.h +++ b/src/core/hle/kernel/kernel.h @@ -21,6 +21,7 @@ namespace Kernel { class AddressArbiter; class ClientPort; +class GlobalScheduler; class HandleTable; class Process; class ResourceLimit; @@ -75,6 +76,12 @@ public: /// Retrieves the list of processes. const std::vector<SharedPtr<Process>>& GetProcessList() const; + /// Gets the sole instance of the global scheduler + Kernel::GlobalScheduler& GlobalScheduler(); + + /// Gets the sole instance of the global scheduler + const Kernel::GlobalScheduler& GlobalScheduler() const; + /// Adds a port to the named port table void AddNamedPort(std::string name, SharedPtr<ClientPort> port); diff --git a/src/core/hle/kernel/mutex.cpp b/src/core/hle/kernel/mutex.cpp index 98e87313b..663d0f4b6 100644 --- a/src/core/hle/kernel/mutex.cpp +++ b/src/core/hle/kernel/mutex.cpp @@ -139,6 +139,9 @@ ResultCode Mutex::Release(VAddr address) { thread->SetCondVarWaitAddress(0); thread->SetMutexWaitAddress(0); thread->SetWaitHandle(0); + thread->SetWaitSynchronizationResult(RESULT_SUCCESS); + + system.PrepareReschedule(); return RESULT_SUCCESS; } diff --git a/src/core/hle/kernel/process.cpp b/src/core/hle/kernel/process.cpp index e80a12ac3..12a900bcc 100644 --- a/src/core/hle/kernel/process.cpp +++ b/src/core/hle/kernel/process.cpp @@ -213,10 +213,7 @@ void Process::PrepareForTermination() { } }; - stop_threads(system.Scheduler(0).GetThreadList()); - stop_threads(system.Scheduler(1).GetThreadList()); - stop_threads(system.Scheduler(2).GetThreadList()); - stop_threads(system.Scheduler(3).GetThreadList()); + stop_threads(system.GlobalScheduler().GetThreadList()); FreeTLSRegion(tls_region_address); tls_region_address = 0; diff --git a/src/core/hle/kernel/scheduler.cpp b/src/core/hle/kernel/scheduler.cpp index e8447b69a..0e2dbf13e 100644 --- a/src/core/hle/kernel/scheduler.cpp +++ b/src/core/hle/kernel/scheduler.cpp @@ -1,8 +1,13 @@ // Copyright 2018 yuzu emulator team // Licensed under GPLv2 or any later version // Refer to the license.txt file included. +// +// SelectThreads, Yield functions originally by TuxSH. +// licensed under GPLv2 or later under exception provided by the author. #include <algorithm> +#include <set> +#include <unordered_set> #include <utility> #include "common/assert.h" @@ -17,56 +22,403 @@ namespace Kernel { -std::mutex Scheduler::scheduler_mutex; +GlobalScheduler::GlobalScheduler(Core::System& system) : system{system} {} -Scheduler::Scheduler(Core::System& system, Core::ARM_Interface& cpu_core) - : cpu_core{cpu_core}, system{system} {} +GlobalScheduler::~GlobalScheduler() = default; -Scheduler::~Scheduler() { - for (auto& thread : thread_list) { - thread->Stop(); +void GlobalScheduler::AddThread(SharedPtr<Thread> thread) { + thread_list.push_back(std::move(thread)); +} + +void GlobalScheduler::RemoveThread(const Thread* thread) { + thread_list.erase(std::remove(thread_list.begin(), thread_list.end(), thread), + thread_list.end()); +} + +void GlobalScheduler::UnloadThread(s32 core) { + Scheduler& sched = system.Scheduler(core); + sched.UnloadThread(); +} + +void GlobalScheduler::SelectThread(u32 core) { + const auto update_thread = [](Thread* thread, Scheduler& sched) { + if (thread != sched.selected_thread) { + if (thread == nullptr) { + ++sched.idle_selection_count; + } + sched.selected_thread = thread; + } + sched.is_context_switch_pending = sched.selected_thread != sched.current_thread; + std::atomic_thread_fence(std::memory_order_seq_cst); + }; + Scheduler& sched = system.Scheduler(core); + Thread* current_thread = nullptr; + // Step 1: Get top thread in schedule queue. + current_thread = scheduled_queue[core].empty() ? nullptr : scheduled_queue[core].front(); + if (current_thread) { + update_thread(current_thread, sched); + return; } + // Step 2: Try selecting a suggested thread. + Thread* winner = nullptr; + std::set<s32> sug_cores; + for (auto thread : suggested_queue[core]) { + s32 this_core = thread->GetProcessorID(); + Thread* thread_on_core = nullptr; + if (this_core >= 0) { + thread_on_core = scheduled_queue[this_core].front(); + } + if (this_core < 0 || thread != thread_on_core) { + winner = thread; + break; + } + sug_cores.insert(this_core); + } + // if we got a suggested thread, select it, else do a second pass. + if (winner && winner->GetPriority() > 2) { + if (winner->IsRunning()) { + UnloadThread(winner->GetProcessorID()); + } + TransferToCore(winner->GetPriority(), core, winner); + update_thread(winner, sched); + return; + } + // Step 3: Select a suggested thread from another core + for (auto& src_core : sug_cores) { + auto it = scheduled_queue[src_core].begin(); + it++; + if (it != scheduled_queue[src_core].end()) { + Thread* thread_on_core = scheduled_queue[src_core].front(); + Thread* to_change = *it; + if (thread_on_core->IsRunning() || to_change->IsRunning()) { + UnloadThread(src_core); + } + TransferToCore(thread_on_core->GetPriority(), core, thread_on_core); + current_thread = thread_on_core; + break; + } + } + update_thread(current_thread, sched); } +bool GlobalScheduler::YieldThread(Thread* yielding_thread) { + // Note: caller should use critical section, etc. + const u32 core_id = static_cast<u32>(yielding_thread->GetProcessorID()); + const u32 priority = yielding_thread->GetPriority(); + + // Yield the thread + const Thread* const winner = scheduled_queue[core_id].front(priority); + ASSERT_MSG(yielding_thread == winner, "Thread yielding without being in front"); + scheduled_queue[core_id].yield(priority); + + return AskForReselectionOrMarkRedundant(yielding_thread, winner); +} + +bool GlobalScheduler::YieldThreadAndBalanceLoad(Thread* yielding_thread) { + // Note: caller should check if !thread.IsSchedulerOperationRedundant and use critical section, + // etc. + const u32 core_id = static_cast<u32>(yielding_thread->GetProcessorID()); + const u32 priority = yielding_thread->GetPriority(); + + // Yield the thread + ASSERT_MSG(yielding_thread == scheduled_queue[core_id].front(priority), + "Thread yielding without being in front"); + scheduled_queue[core_id].yield(priority); + + std::array<Thread*, NUM_CPU_CORES> current_threads; + for (u32 i = 0; i < NUM_CPU_CORES; i++) { + current_threads[i] = scheduled_queue[i].empty() ? nullptr : scheduled_queue[i].front(); + } + + Thread* next_thread = scheduled_queue[core_id].front(priority); + Thread* winner = nullptr; + for (auto& thread : suggested_queue[core_id]) { + const s32 source_core = thread->GetProcessorID(); + if (source_core >= 0) { + if (current_threads[source_core] != nullptr) { + if (thread == current_threads[source_core] || + current_threads[source_core]->GetPriority() < min_regular_priority) { + continue; + } + } + } + if (next_thread->GetLastRunningTicks() >= thread->GetLastRunningTicks() || + next_thread->GetPriority() < thread->GetPriority()) { + if (thread->GetPriority() <= priority) { + winner = thread; + break; + } + } + } + + if (winner != nullptr) { + if (winner != yielding_thread) { + if (winner->IsRunning()) { + UnloadThread(winner->GetProcessorID()); + } + TransferToCore(winner->GetPriority(), core_id, winner); + } + } else { + winner = next_thread; + } + + return AskForReselectionOrMarkRedundant(yielding_thread, winner); +} + +bool GlobalScheduler::YieldThreadAndWaitForLoadBalancing(Thread* yielding_thread) { + // Note: caller should check if !thread.IsSchedulerOperationRedundant and use critical section, + // etc. + Thread* winner = nullptr; + const u32 core_id = static_cast<u32>(yielding_thread->GetProcessorID()); + + // Remove the thread from its scheduled mlq, put it on the corresponding "suggested" one instead + TransferToCore(yielding_thread->GetPriority(), -1, yielding_thread); + + // If the core is idle, perform load balancing, excluding the threads that have just used this + // function... + if (scheduled_queue[core_id].empty()) { + // Here, "current_threads" is calculated after the ""yield"", unlike yield -1 + std::array<Thread*, NUM_CPU_CORES> current_threads; + for (u32 i = 0; i < NUM_CPU_CORES; i++) { + current_threads[i] = scheduled_queue[i].empty() ? nullptr : scheduled_queue[i].front(); + } + for (auto& thread : suggested_queue[core_id]) { + const s32 source_core = thread->GetProcessorID(); + if (source_core < 0 || thread == current_threads[source_core]) { + continue; + } + if (current_threads[source_core] == nullptr || + current_threads[source_core]->GetPriority() >= min_regular_priority) { + winner = thread; + } + break; + } + if (winner != nullptr) { + if (winner != yielding_thread) { + if (winner->IsRunning()) { + UnloadThread(winner->GetProcessorID()); + } + TransferToCore(winner->GetPriority(), core_id, winner); + } + } else { + winner = yielding_thread; + } + } + + return AskForReselectionOrMarkRedundant(yielding_thread, winner); +} + +void GlobalScheduler::PreemptThreads() { + for (std::size_t core_id = 0; core_id < NUM_CPU_CORES; core_id++) { + const u32 priority = preemption_priorities[core_id]; + + if (scheduled_queue[core_id].size(priority) > 0) { + scheduled_queue[core_id].front(priority)->IncrementYieldCount(); + scheduled_queue[core_id].yield(priority); + if (scheduled_queue[core_id].size(priority) > 1) { + scheduled_queue[core_id].front(priority)->IncrementYieldCount(); + } + } + + Thread* current_thread = + scheduled_queue[core_id].empty() ? nullptr : scheduled_queue[core_id].front(); + Thread* winner = nullptr; + for (auto& thread : suggested_queue[core_id]) { + const s32 source_core = thread->GetProcessorID(); + if (thread->GetPriority() != priority) { + continue; + } + if (source_core >= 0) { + Thread* next_thread = scheduled_queue[source_core].empty() + ? nullptr + : scheduled_queue[source_core].front(); + if (next_thread != nullptr && next_thread->GetPriority() < 2) { + break; + } + if (next_thread == thread) { + continue; + } + } + if (current_thread != nullptr && + current_thread->GetLastRunningTicks() >= thread->GetLastRunningTicks()) { + winner = thread; + break; + } + } + + if (winner != nullptr) { + if (winner->IsRunning()) { + UnloadThread(winner->GetProcessorID()); + } + TransferToCore(winner->GetPriority(), s32(core_id), winner); + current_thread = + winner->GetPriority() <= current_thread->GetPriority() ? winner : current_thread; + } + + if (current_thread != nullptr && current_thread->GetPriority() > priority) { + for (auto& thread : suggested_queue[core_id]) { + const s32 source_core = thread->GetProcessorID(); + if (thread->GetPriority() < priority) { + continue; + } + if (source_core >= 0) { + Thread* next_thread = scheduled_queue[source_core].empty() + ? nullptr + : scheduled_queue[source_core].front(); + if (next_thread != nullptr && next_thread->GetPriority() < 2) { + break; + } + if (next_thread == thread) { + continue; + } + } + if (current_thread != nullptr && + current_thread->GetLastRunningTicks() >= thread->GetLastRunningTicks()) { + winner = thread; + break; + } + } + + if (winner != nullptr) { + if (winner->IsRunning()) { + UnloadThread(winner->GetProcessorID()); + } + TransferToCore(winner->GetPriority(), s32(core_id), winner); + current_thread = winner; + } + } + + is_reselection_pending.store(true, std::memory_order_release); + } +} + +void GlobalScheduler::Suggest(u32 priority, u32 core, Thread* thread) { + suggested_queue[core].add(thread, priority); +} + +void GlobalScheduler::Unsuggest(u32 priority, u32 core, Thread* thread) { + suggested_queue[core].remove(thread, priority); +} + +void GlobalScheduler::Schedule(u32 priority, u32 core, Thread* thread) { + ASSERT_MSG(thread->GetProcessorID() == s32(core), "Thread must be assigned to this core."); + scheduled_queue[core].add(thread, priority); +} + +void GlobalScheduler::SchedulePrepend(u32 priority, u32 core, Thread* thread) { + ASSERT_MSG(thread->GetProcessorID() == s32(core), "Thread must be assigned to this core."); + scheduled_queue[core].add(thread, priority, false); +} + +void GlobalScheduler::Reschedule(u32 priority, u32 core, Thread* thread) { + scheduled_queue[core].remove(thread, priority); + scheduled_queue[core].add(thread, priority); +} + +void GlobalScheduler::Unschedule(u32 priority, u32 core, Thread* thread) { + scheduled_queue[core].remove(thread, priority); +} + +void GlobalScheduler::TransferToCore(u32 priority, s32 destination_core, Thread* thread) { + const bool schedulable = thread->GetPriority() < THREADPRIO_COUNT; + const s32 source_core = thread->GetProcessorID(); + if (source_core == destination_core || !schedulable) { + return; + } + thread->SetProcessorID(destination_core); + if (source_core >= 0) { + Unschedule(priority, source_core, thread); + } + if (destination_core >= 0) { + Unsuggest(priority, destination_core, thread); + Schedule(priority, destination_core, thread); + } + if (source_core >= 0) { + Suggest(priority, source_core, thread); + } +} + +bool GlobalScheduler::AskForReselectionOrMarkRedundant(Thread* current_thread, + const Thread* winner) { + if (current_thread == winner) { + current_thread->IncrementYieldCount(); + return true; + } else { + is_reselection_pending.store(true, std::memory_order_release); + return false; + } +} + +void GlobalScheduler::Shutdown() { + for (std::size_t core = 0; core < NUM_CPU_CORES; core++) { + scheduled_queue[core].clear(); + suggested_queue[core].clear(); + } + thread_list.clear(); +} + +Scheduler::Scheduler(Core::System& system, Core::ARM_Interface& cpu_core, u32 core_id) + : system(system), cpu_core(cpu_core), core_id(core_id) {} + +Scheduler::~Scheduler() = default; + bool Scheduler::HaveReadyThreads() const { - std::lock_guard lock{scheduler_mutex}; - return !ready_queue.empty(); + return system.GlobalScheduler().HaveReadyThreads(core_id); } Thread* Scheduler::GetCurrentThread() const { return current_thread.get(); } +Thread* Scheduler::GetSelectedThread() const { + return selected_thread.get(); +} + +void Scheduler::SelectThreads() { + system.GlobalScheduler().SelectThread(core_id); +} + u64 Scheduler::GetLastContextSwitchTicks() const { return last_context_switch_time; } -Thread* Scheduler::PopNextReadyThread() { - Thread* next = nullptr; - Thread* thread = GetCurrentThread(); +void Scheduler::TryDoContextSwitch() { + if (is_context_switch_pending) { + SwitchContext(); + } +} - if (thread && thread->GetStatus() == ThreadStatus::Running) { - if (ready_queue.empty()) { - return thread; - } - // We have to do better than the current thread. - // This call returns null when that's not possible. - next = ready_queue.front(); - if (next == nullptr || next->GetPriority() >= thread->GetPriority()) { - next = thread; - } - } else { - if (ready_queue.empty()) { - return nullptr; +void Scheduler::UnloadThread() { + Thread* const previous_thread = GetCurrentThread(); + Process* const previous_process = system.Kernel().CurrentProcess(); + + UpdateLastContextSwitchTime(previous_thread, previous_process); + + // Save context for previous thread + if (previous_thread) { + cpu_core.SaveContext(previous_thread->GetContext()); + // Save the TPIDR_EL0 system register in case it was modified. + previous_thread->SetTPIDR_EL0(cpu_core.GetTPIDR_EL0()); + + if (previous_thread->GetStatus() == ThreadStatus::Running) { + // This is only the case when a reschedule is triggered without the current thread + // yielding execution (i.e. an event triggered, system core time-sliced, etc) + previous_thread->SetStatus(ThreadStatus::Ready); } - next = ready_queue.front(); + previous_thread->SetIsRunning(false); } - - return next; + current_thread = nullptr; } -void Scheduler::SwitchContext(Thread* new_thread) { - Thread* previous_thread = GetCurrentThread(); +void Scheduler::SwitchContext() { + Thread* const previous_thread = GetCurrentThread(); + Thread* const new_thread = GetSelectedThread(); + + is_context_switch_pending = false; + if (new_thread == previous_thread) { + return; + } + Process* const previous_process = system.Kernel().CurrentProcess(); UpdateLastContextSwitchTime(previous_thread, previous_process); @@ -80,23 +432,23 @@ void Scheduler::SwitchContext(Thread* new_thread) { if (previous_thread->GetStatus() == ThreadStatus::Running) { // This is only the case when a reschedule is triggered without the current thread // yielding execution (i.e. an event triggered, system core time-sliced, etc) - ready_queue.add(previous_thread, previous_thread->GetPriority(), false); previous_thread->SetStatus(ThreadStatus::Ready); } + previous_thread->SetIsRunning(false); } // Load context of new thread if (new_thread) { + ASSERT_MSG(new_thread->GetProcessorID() == s32(this->core_id), + "Thread must be assigned to this core."); ASSERT_MSG(new_thread->GetStatus() == ThreadStatus::Ready, "Thread must be ready to become running."); // Cancel any outstanding wakeup events for this thread new_thread->CancelWakeupTimer(); - current_thread = new_thread; - - ready_queue.remove(new_thread, new_thread->GetPriority()); new_thread->SetStatus(ThreadStatus::Running); + new_thread->SetIsRunning(true); auto* const thread_owner_process = current_thread->GetOwnerProcess(); if (previous_process != thread_owner_process) { @@ -130,124 +482,9 @@ void Scheduler::UpdateLastContextSwitchTime(Thread* thread, Process* process) { last_context_switch_time = most_recent_switch_ticks; } -void Scheduler::Reschedule() { - std::lock_guard lock{scheduler_mutex}; - - Thread* cur = GetCurrentThread(); - Thread* next = PopNextReadyThread(); - - if (cur && next) { - LOG_TRACE(Kernel, "context switch {} -> {}", cur->GetObjectId(), next->GetObjectId()); - } else if (cur) { - LOG_TRACE(Kernel, "context switch {} -> idle", cur->GetObjectId()); - } else if (next) { - LOG_TRACE(Kernel, "context switch idle -> {}", next->GetObjectId()); - } - - SwitchContext(next); -} - -void Scheduler::AddThread(SharedPtr<Thread> thread) { - std::lock_guard lock{scheduler_mutex}; - - thread_list.push_back(std::move(thread)); -} - -void Scheduler::RemoveThread(Thread* thread) { - std::lock_guard lock{scheduler_mutex}; - - thread_list.erase(std::remove(thread_list.begin(), thread_list.end(), thread), - thread_list.end()); -} - -void Scheduler::ScheduleThread(Thread* thread, u32 priority) { - std::lock_guard lock{scheduler_mutex}; - - ASSERT(thread->GetStatus() == ThreadStatus::Ready); - ready_queue.add(thread, priority); -} - -void Scheduler::UnscheduleThread(Thread* thread, u32 priority) { - std::lock_guard lock{scheduler_mutex}; - - ASSERT(thread->GetStatus() == ThreadStatus::Ready); - ready_queue.remove(thread, priority); -} - -void Scheduler::SetThreadPriority(Thread* thread, u32 priority) { - std::lock_guard lock{scheduler_mutex}; - if (thread->GetPriority() == priority) { - return; - } - - // If thread was ready, adjust queues - if (thread->GetStatus() == ThreadStatus::Ready) - ready_queue.adjust(thread, thread->GetPriority(), priority); -} - -Thread* Scheduler::GetNextSuggestedThread(u32 core, u32 maximum_priority) const { - std::lock_guard lock{scheduler_mutex}; - - const u32 mask = 1U << core; - for (auto* thread : ready_queue) { - if ((thread->GetAffinityMask() & mask) != 0 && thread->GetPriority() < maximum_priority) { - return thread; - } - } - return nullptr; -} - -void Scheduler::YieldWithoutLoadBalancing(Thread* thread) { - ASSERT(thread != nullptr); - // Avoid yielding if the thread isn't even running. - ASSERT(thread->GetStatus() == ThreadStatus::Running); - - // Sanity check that the priority is valid - ASSERT(thread->GetPriority() < THREADPRIO_COUNT); - - // Yield this thread -- sleep for zero time and force reschedule to different thread - GetCurrentThread()->Sleep(0); -} - -void Scheduler::YieldWithLoadBalancing(Thread* thread) { - ASSERT(thread != nullptr); - const auto priority = thread->GetPriority(); - const auto core = static_cast<u32>(thread->GetProcessorID()); - - // Avoid yielding if the thread isn't even running. - ASSERT(thread->GetStatus() == ThreadStatus::Running); - - // Sanity check that the priority is valid - ASSERT(priority < THREADPRIO_COUNT); - - // Sleep for zero time to be able to force reschedule to different thread - GetCurrentThread()->Sleep(0); - - Thread* suggested_thread = nullptr; - - // Search through all of the cpu cores (except this one) for a suggested thread. - // Take the first non-nullptr one - for (unsigned cur_core = 0; cur_core < Core::NUM_CPU_CORES; ++cur_core) { - const auto res = - system.CpuCore(cur_core).Scheduler().GetNextSuggestedThread(core, priority); - - // If scheduler provides a suggested thread - if (res != nullptr) { - // And its better than the current suggested thread (or is the first valid one) - if (suggested_thread == nullptr || - suggested_thread->GetPriority() > res->GetPriority()) { - suggested_thread = res; - } - } - } - - // If a suggested thread was found, queue that for this core - if (suggested_thread != nullptr) - suggested_thread->ChangeCore(core, suggested_thread->GetAffinityMask()); -} - -void Scheduler::YieldAndWaitForLoadBalancing(Thread* thread) { - UNIMPLEMENTED_MSG("Wait for load balancing thread yield type is not implemented!"); +void Scheduler::Shutdown() { + current_thread = nullptr; + selected_thread = nullptr; } } // namespace Kernel diff --git a/src/core/hle/kernel/scheduler.h b/src/core/hle/kernel/scheduler.h index b29bf7be8..f2d6311b8 100644 --- a/src/core/hle/kernel/scheduler.h +++ b/src/core/hle/kernel/scheduler.h @@ -20,124 +20,185 @@ namespace Kernel { class Process; -class Scheduler final { +class GlobalScheduler final { public: - explicit Scheduler(Core::System& system, Core::ARM_Interface& cpu_core); - ~Scheduler(); - - /// Returns whether there are any threads that are ready to run. - bool HaveReadyThreads() const; - - /// Reschedules to the next available thread (call after current thread is suspended) - void Reschedule(); - - /// Gets the current running thread - Thread* GetCurrentThread() const; + static constexpr u32 NUM_CPU_CORES = 4; - /// Gets the timestamp for the last context switch in ticks. - u64 GetLastContextSwitchTicks() const; + explicit GlobalScheduler(Core::System& system); + ~GlobalScheduler(); /// Adds a new thread to the scheduler void AddThread(SharedPtr<Thread> thread); /// Removes a thread from the scheduler - void RemoveThread(Thread* thread); + void RemoveThread(const Thread* thread); - /// Schedules a thread that has become "ready" - void ScheduleThread(Thread* thread, u32 priority); + /// Returns a list of all threads managed by the scheduler + const std::vector<SharedPtr<Thread>>& GetThreadList() const { + return thread_list; + } - /// Unschedules a thread that was already scheduled - void UnscheduleThread(Thread* thread, u32 priority); + /** + * Add a thread to the suggested queue of a cpu core. Suggested threads may be + * picked if no thread is scheduled to run on the core. + */ + void Suggest(u32 priority, u32 core, Thread* thread); - /// Sets the priority of a thread in the scheduler - void SetThreadPriority(Thread* thread, u32 priority); + /** + * Remove a thread to the suggested queue of a cpu core. Suggested threads may be + * picked if no thread is scheduled to run on the core. + */ + void Unsuggest(u32 priority, u32 core, Thread* thread); - /// Gets the next suggested thread for load balancing - Thread* GetNextSuggestedThread(u32 core, u32 minimum_priority) const; + /** + * Add a thread to the scheduling queue of a cpu core. The thread is added at the + * back the queue in its priority level. + */ + void Schedule(u32 priority, u32 core, Thread* thread); /** - * YieldWithoutLoadBalancing -- analogous to normal yield on a system - * Moves the thread to the end of the ready queue for its priority, and then reschedules the - * system to the new head of the queue. - * - * Example (Single Core -- but can be extrapolated to multi): - * ready_queue[prio=0]: ThreadA, ThreadB, ThreadC (->exec order->) - * Currently Running: ThreadR - * - * ThreadR calls YieldWithoutLoadBalancing - * - * ThreadR is moved to the end of ready_queue[prio=0]: - * ready_queue[prio=0]: ThreadA, ThreadB, ThreadC, ThreadR (->exec order->) - * Currently Running: Nothing - * - * System is rescheduled (ThreadA is popped off of queue): - * ready_queue[prio=0]: ThreadB, ThreadC, ThreadR (->exec order->) - * Currently Running: ThreadA - * - * If the queue is empty at time of call, no yielding occurs. This does not cross between cores - * or priorities at all. + * Add a thread to the scheduling queue of a cpu core. The thread is added at the + * front the queue in its priority level. + */ + void SchedulePrepend(u32 priority, u32 core, Thread* thread); + + /// Reschedule an already scheduled thread based on a new priority + void Reschedule(u32 priority, u32 core, Thread* thread); + + /// Unschedules a thread. + void Unschedule(u32 priority, u32 core, Thread* thread); + + /** + * Transfers a thread into an specific core. If the destination_core is -1 + * it will be unscheduled from its source code and added into its suggested + * queue. */ - void YieldWithoutLoadBalancing(Thread* thread); + void TransferToCore(u32 priority, s32 destination_core, Thread* thread); + + /// Selects a core and forces it to unload its current thread's context + void UnloadThread(s32 core); /** - * YieldWithLoadBalancing -- yield but with better selection of the new running thread - * Moves the current thread to the end of the ready queue for its priority, then selects a - * 'suggested thread' (a thread on a different core that could run on this core) from the - * scheduler, changes its core, and reschedules the current core to that thread. + * Takes care of selecting the new scheduled thread in three steps: * - * Example (Dual Core -- can be extrapolated to Quad Core, this is just normal yield if it were - * single core): - * ready_queue[core=0][prio=0]: ThreadA, ThreadB (affinities not pictured as irrelevant - * ready_queue[core=1][prio=0]: ThreadC[affinity=both], ThreadD[affinity=core1only] - * Currently Running: ThreadQ on Core 0 || ThreadP on Core 1 + * 1. First a thread is selected from the top of the priority queue. If no thread + * is obtained then we move to step two, else we are done. * - * ThreadQ calls YieldWithLoadBalancing + * 2. Second we try to get a suggested thread that's not assigned to any core or + * that is not the top thread in that core. * - * ThreadQ is moved to the end of ready_queue[core=0][prio=0]: - * ready_queue[core=0][prio=0]: ThreadA, ThreadB - * ready_queue[core=1][prio=0]: ThreadC[affinity=both], ThreadD[affinity=core1only] - * Currently Running: ThreadQ on Core 0 || ThreadP on Core 1 + * 3. Third is no suggested thread is found, we do a second pass and pick a running + * thread in another core and swap it with its current thread. + */ + void SelectThread(u32 core); + + bool HaveReadyThreads(u32 core_id) const { + return !scheduled_queue[core_id].empty(); + } + + /** + * Takes a thread and moves it to the back of the it's priority list. * - * A list of suggested threads for each core is compiled - * Suggested Threads: {ThreadC on Core 1} - * If this were quad core (as the switch is), there could be between 0 and 3 threads in this - * list. If there are more than one, the thread is selected by highest prio. + * @note This operation can be redundant and no scheduling is changed if marked as so. + */ + bool YieldThread(Thread* thread); + + /** + * Takes a thread and moves it to the back of the it's priority list. + * Afterwards, tries to pick a suggested thread from the suggested queue that has worse time or + * a better priority than the next thread in the core. * - * ThreadC is core changed to Core 0: - * ready_queue[core=0][prio=0]: ThreadC, ThreadA, ThreadB, ThreadQ - * ready_queue[core=1][prio=0]: ThreadD - * Currently Running: None on Core 0 || ThreadP on Core 1 + * @note This operation can be redundant and no scheduling is changed if marked as so. + */ + bool YieldThreadAndBalanceLoad(Thread* thread); + + /** + * Takes a thread and moves it out of the scheduling queue. + * and into the suggested queue. If no thread can be scheduled afterwards in that core, + * a suggested thread is obtained instead. * - * System is rescheduled (ThreadC is popped off of queue): - * ready_queue[core=0][prio=0]: ThreadA, ThreadB, ThreadQ - * ready_queue[core=1][prio=0]: ThreadD - * Currently Running: ThreadC on Core 0 || ThreadP on Core 1 + * @note This operation can be redundant and no scheduling is changed if marked as so. + */ + bool YieldThreadAndWaitForLoadBalancing(Thread* thread); + + /** + * Rotates the scheduling queues of threads at a preemption priority and then does + * some core rebalancing. Preemption priorities can be found in the array + * 'preemption_priorities'. * - * If no suggested threads can be found this will behave just as normal yield. If there are - * multiple candidates for the suggested thread on a core, the highest prio is taken. + * @note This operation happens every 10ms. */ - void YieldWithLoadBalancing(Thread* thread); + void PreemptThreads(); - /// Currently unknown -- asserts as unimplemented on call - void YieldAndWaitForLoadBalancing(Thread* thread); + u32 CpuCoresCount() const { + return NUM_CPU_CORES; + } - /// Returns a list of all threads managed by the scheduler - const std::vector<SharedPtr<Thread>>& GetThreadList() const { - return thread_list; + void SetReselectionPending() { + is_reselection_pending.store(true, std::memory_order_release); + } + + bool IsReselectionPending() const { + return is_reselection_pending.load(std::memory_order_acquire); } + void Shutdown(); + private: - /** - * Pops and returns the next thread from the thread queue - * @return A pointer to the next ready thread - */ - Thread* PopNextReadyThread(); + bool AskForReselectionOrMarkRedundant(Thread* current_thread, const Thread* winner); - /** - * Switches the CPU's active thread context to that of the specified thread - * @param new_thread The thread to switch to - */ - void SwitchContext(Thread* new_thread); + static constexpr u32 min_regular_priority = 2; + std::array<Common::MultiLevelQueue<Thread*, THREADPRIO_COUNT>, NUM_CPU_CORES> scheduled_queue; + std::array<Common::MultiLevelQueue<Thread*, THREADPRIO_COUNT>, NUM_CPU_CORES> suggested_queue; + std::atomic<bool> is_reselection_pending{false}; + + // The priority levels at which the global scheduler preempts threads every 10 ms. They are + // ordered from Core 0 to Core 3. + 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; + Core::System& system; +}; + +class Scheduler final { +public: + explicit Scheduler(Core::System& system, Core::ARM_Interface& cpu_core, u32 core_id); + ~Scheduler(); + + /// Returns whether there are any threads that are ready to run. + bool HaveReadyThreads() const; + + /// Reschedules to the next available thread (call after current thread is suspended) + void TryDoContextSwitch(); + + /// Unloads currently running thread + void UnloadThread(); + + /// Select the threads in top of the scheduling multilist. + void SelectThreads(); + + /// Gets the current running thread + Thread* GetCurrentThread() const; + + /// Gets the currently selected thread from the top of the multilevel queue + Thread* GetSelectedThread() const; + + /// Gets the timestamp for the last context switch in ticks. + u64 GetLastContextSwitchTicks() const; + + bool ContextSwitchPending() const { + return is_context_switch_pending; + } + + /// Shutdowns the scheduler. + void Shutdown(); + +private: + friend class GlobalScheduler; + + /// Switches the CPU's active thread context to that of the specified thread + void SwitchContext(); /** * Called on every context switch to update the internal timestamp @@ -152,19 +213,16 @@ private: */ void UpdateLastContextSwitchTime(Thread* thread, Process* process); - /// Lists all thread ids that aren't deleted/etc. - std::vector<SharedPtr<Thread>> thread_list; - - /// Lists only ready thread ids. - Common::MultiLevelQueue<Thread*, THREADPRIO_LOWEST + 1> ready_queue; - SharedPtr<Thread> current_thread = nullptr; + SharedPtr<Thread> selected_thread = nullptr; + Core::System& system; Core::ARM_Interface& cpu_core; u64 last_context_switch_time = 0; + u64 idle_selection_count = 0; + const u32 core_id; - Core::System& system; - static std::mutex scheduler_mutex; + bool is_context_switch_pending = false; }; } // namespace Kernel diff --git a/src/core/hle/kernel/svc.cpp b/src/core/hle/kernel/svc.cpp index 1fd1a732a..f64236be1 100644 --- a/src/core/hle/kernel/svc.cpp +++ b/src/core/hle/kernel/svc.cpp @@ -516,7 +516,7 @@ static ResultCode WaitSynchronization(Core::System& system, Handle* index, VAddr thread->WakeAfterDelay(nano_seconds); thread->SetWakeupCallback(DefaultThreadWakeupCallback); - system.CpuCore(thread->GetProcessorID()).PrepareReschedule(); + system.PrepareReschedule(thread->GetProcessorID()); return RESULT_TIMEOUT; } @@ -534,6 +534,7 @@ static ResultCode CancelSynchronization(Core::System& system, Handle thread_hand } thread->CancelWait(); + system.PrepareReschedule(thread->GetProcessorID()); return RESULT_SUCCESS; } @@ -1066,6 +1067,8 @@ static ResultCode SetThreadActivity(Core::System& system, Handle handle, u32 act } thread->SetActivity(static_cast<ThreadActivity>(activity)); + + system.PrepareReschedule(thread->GetProcessorID()); return RESULT_SUCCESS; } @@ -1147,7 +1150,7 @@ static ResultCode SetThreadPriority(Core::System& system, Handle handle, u32 pri thread->SetPriority(priority); - system.CpuCore(thread->GetProcessorID()).PrepareReschedule(); + system.PrepareReschedule(thread->GetProcessorID()); return RESULT_SUCCESS; } @@ -1503,7 +1506,7 @@ static ResultCode CreateThread(Core::System& system, Handle* out_handle, VAddr e thread->SetName( fmt::format("thread[entry_point={:X}, handle={:X}]", entry_point, *new_thread_handle)); - system.CpuCore(thread->GetProcessorID()).PrepareReschedule(); + system.PrepareReschedule(thread->GetProcessorID()); return RESULT_SUCCESS; } @@ -1525,7 +1528,7 @@ static ResultCode StartThread(Core::System& system, Handle thread_handle) { thread->ResumeFromWait(); if (thread->GetStatus() == ThreadStatus::Ready) { - system.CpuCore(thread->GetProcessorID()).PrepareReschedule(); + system.PrepareReschedule(thread->GetProcessorID()); } return RESULT_SUCCESS; @@ -1537,7 +1540,7 @@ static void ExitThread(Core::System& system) { auto* const current_thread = system.CurrentScheduler().GetCurrentThread(); current_thread->Stop(); - system.CurrentScheduler().RemoveThread(current_thread); + system.GlobalScheduler().RemoveThread(current_thread); system.PrepareReschedule(); } @@ -1553,17 +1556,18 @@ static void SleepThread(Core::System& system, s64 nanoseconds) { auto& scheduler = system.CurrentScheduler(); auto* const current_thread = scheduler.GetCurrentThread(); + bool is_redundant = false; if (nanoseconds <= 0) { switch (static_cast<SleepType>(nanoseconds)) { case SleepType::YieldWithoutLoadBalancing: - scheduler.YieldWithoutLoadBalancing(current_thread); + is_redundant = current_thread->YieldSimple(); break; case SleepType::YieldWithLoadBalancing: - scheduler.YieldWithLoadBalancing(current_thread); + is_redundant = current_thread->YieldAndBalanceLoad(); break; case SleepType::YieldAndWaitForLoadBalancing: - scheduler.YieldAndWaitForLoadBalancing(current_thread); + is_redundant = current_thread->YieldAndWaitForLoadBalancing(); break; default: UNREACHABLE_MSG("Unimplemented sleep yield type '{:016X}'!", nanoseconds); @@ -1572,10 +1576,13 @@ static void SleepThread(Core::System& system, s64 nanoseconds) { current_thread->Sleep(nanoseconds); } - // Reschedule all CPU cores - for (std::size_t i = 0; i < Core::NUM_CPU_CORES; ++i) { - system.CpuCore(i).PrepareReschedule(); + if (is_redundant) { + // If it's redundant, the core is pretty much idle. Some games keep idling + // a core while it's doing nothing, we advance timing to avoid costly continuous + // calls. + system.CoreTiming().AddTicks(2000); } + system.PrepareReschedule(current_thread->GetProcessorID()); } /// Wait process wide key atomic @@ -1601,6 +1608,8 @@ static ResultCode WaitProcessWideKeyAtomic(Core::System& system, VAddr mutex_add return ERR_INVALID_ADDRESS; } + ASSERT(condition_variable_addr == Common::AlignDown(condition_variable_addr, 4)); + auto* const current_process = system.Kernel().CurrentProcess(); const auto& handle_table = current_process->GetHandleTable(); SharedPtr<Thread> thread = handle_table.Get<Thread>(thread_handle); @@ -1622,7 +1631,7 @@ static ResultCode WaitProcessWideKeyAtomic(Core::System& system, VAddr mutex_add // Note: Deliberately don't attempt to inherit the lock owner's priority. - system.CpuCore(current_thread->GetProcessorID()).PrepareReschedule(); + system.PrepareReschedule(current_thread->GetProcessorID()); return RESULT_SUCCESS; } @@ -1632,24 +1641,19 @@ static ResultCode SignalProcessWideKey(Core::System& system, VAddr condition_var LOG_TRACE(Kernel_SVC, "called, condition_variable_addr=0x{:X}, target=0x{:08X}", condition_variable_addr, target); - const auto RetrieveWaitingThreads = [&system](std::size_t core_index, - std::vector<SharedPtr<Thread>>& waiting_threads, - VAddr condvar_addr) { - const auto& scheduler = system.Scheduler(core_index); - const auto& thread_list = scheduler.GetThreadList(); - - for (const auto& thread : thread_list) { - if (thread->GetCondVarWaitAddress() == condvar_addr) - waiting_threads.push_back(thread); - } - }; + ASSERT(condition_variable_addr == Common::AlignDown(condition_variable_addr, 4)); // Retrieve a list of all threads that are waiting for this condition variable. std::vector<SharedPtr<Thread>> waiting_threads; - RetrieveWaitingThreads(0, waiting_threads, condition_variable_addr); - RetrieveWaitingThreads(1, waiting_threads, condition_variable_addr); - RetrieveWaitingThreads(2, waiting_threads, condition_variable_addr); - RetrieveWaitingThreads(3, waiting_threads, condition_variable_addr); + const auto& scheduler = system.GlobalScheduler(); + const auto& thread_list = scheduler.GetThreadList(); + + for (const auto& thread : thread_list) { + if (thread->GetCondVarWaitAddress() == condition_variable_addr) { + waiting_threads.push_back(thread); + } + } + // Sort them by priority, such that the highest priority ones come first. std::sort(waiting_threads.begin(), waiting_threads.end(), [](const SharedPtr<Thread>& lhs, const SharedPtr<Thread>& rhs) { @@ -1679,18 +1683,20 @@ static ResultCode SignalProcessWideKey(Core::System& system, VAddr condition_var // Atomically read the value of the mutex. u32 mutex_val = 0; + u32 update_val = 0; + const VAddr mutex_address = thread->GetMutexWaitAddress(); do { - monitor.SetExclusive(current_core, thread->GetMutexWaitAddress()); + monitor.SetExclusive(current_core, mutex_address); // If the mutex is not yet acquired, acquire it. - mutex_val = Memory::Read32(thread->GetMutexWaitAddress()); + mutex_val = Memory::Read32(mutex_address); if (mutex_val != 0) { - monitor.ClearExclusive(); - break; + update_val = mutex_val | Mutex::MutexHasWaitersFlag; + } else { + update_val = thread->GetWaitHandle(); } - } while (!monitor.ExclusiveWrite32(current_core, thread->GetMutexWaitAddress(), - thread->GetWaitHandle())); + } while (!monitor.ExclusiveWrite32(current_core, mutex_address, update_val)); if (mutex_val == 0) { // We were able to acquire the mutex, resume this thread. ASSERT(thread->GetStatus() == ThreadStatus::WaitCondVar); @@ -1704,20 +1710,9 @@ static ResultCode SignalProcessWideKey(Core::System& system, VAddr condition_var thread->SetLockOwner(nullptr); thread->SetMutexWaitAddress(0); thread->SetWaitHandle(0); - system.CpuCore(thread->GetProcessorID()).PrepareReschedule(); + thread->SetWaitSynchronizationResult(RESULT_SUCCESS); + system.PrepareReschedule(thread->GetProcessorID()); } else { - // Atomically signal that the mutex now has a waiting thread. - do { - monitor.SetExclusive(current_core, thread->GetMutexWaitAddress()); - - // Ensure that the mutex value is still what we expect. - u32 value = Memory::Read32(thread->GetMutexWaitAddress()); - // TODO(Subv): When this happens, the kernel just clears the exclusive state and - // retries the initial read for this thread. - ASSERT_MSG(mutex_val == value, "Unhandled synchronization primitive case"); - } while (!monitor.ExclusiveWrite32(current_core, thread->GetMutexWaitAddress(), - mutex_val | Mutex::MutexHasWaitersFlag)); - // The mutex is already owned by some other thread, make this thread wait on it. const Handle owner_handle = static_cast<Handle>(mutex_val & Mutex::MutexOwnerMask); const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable(); @@ -1728,6 +1723,7 @@ static ResultCode SignalProcessWideKey(Core::System& system, VAddr condition_var thread->SetStatus(ThreadStatus::WaitMutex); owner->AddMutexWaiter(thread); + system.PrepareReschedule(thread->GetProcessorID()); } } @@ -1754,7 +1750,12 @@ static ResultCode WaitForAddress(Core::System& system, VAddr address, u32 type, const auto arbitration_type = static_cast<AddressArbiter::ArbitrationType>(type); auto& address_arbiter = system.Kernel().CurrentProcess()->GetAddressArbiter(); - return address_arbiter.WaitForAddress(address, arbitration_type, value, timeout); + const ResultCode result = + address_arbiter.WaitForAddress(address, arbitration_type, value, timeout); + if (result == RESULT_SUCCESS) { + system.PrepareReschedule(); + } + return result; } // Signals to an address (via Address Arbiter) @@ -2040,7 +2041,10 @@ static ResultCode SetThreadCoreMask(Core::System& system, Handle thread_handle, return ERR_INVALID_HANDLE; } + system.PrepareReschedule(thread->GetProcessorID()); thread->ChangeCore(core, affinity_mask); + system.PrepareReschedule(thread->GetProcessorID()); + return RESULT_SUCCESS; } @@ -2151,6 +2155,7 @@ static ResultCode SignalEvent(Core::System& system, Handle handle) { } writable_event->Signal(); + system.PrepareReschedule(); return RESULT_SUCCESS; } diff --git a/src/core/hle/kernel/thread.cpp b/src/core/hle/kernel/thread.cpp index ec529e7f2..962530d2d 100644 --- a/src/core/hle/kernel/thread.cpp +++ b/src/core/hle/kernel/thread.cpp @@ -45,15 +45,7 @@ void Thread::Stop() { callback_handle); kernel.ThreadWakeupCallbackHandleTable().Close(callback_handle); callback_handle = 0; - - // Clean up thread from ready queue - // This is only needed when the thread is terminated forcefully (SVC TerminateProcess) - if (status == ThreadStatus::Ready || status == ThreadStatus::Paused) { - scheduler->UnscheduleThread(this, current_priority); - } - - status = ThreadStatus::Dead; - + SetStatus(ThreadStatus::Dead); WakeupAllWaitingThreads(); // Clean up any dangling references in objects that this thread was waiting for @@ -132,17 +124,16 @@ void Thread::ResumeFromWait() { wakeup_callback = nullptr; if (activity == ThreadActivity::Paused) { - status = ThreadStatus::Paused; + SetStatus(ThreadStatus::Paused); return; } - status = ThreadStatus::Ready; - - ChangeScheduler(); + SetStatus(ThreadStatus::Ready); } void Thread::CancelWait() { ASSERT(GetStatus() == ThreadStatus::WaitSynch); + ClearWaitObjects(); SetWaitSynchronizationResult(ERR_SYNCHRONIZATION_CANCELED); ResumeFromWait(); } @@ -205,9 +196,9 @@ ResultVal<SharedPtr<Thread>> Thread::Create(KernelCore& kernel, std::string name thread->name = std::move(name); thread->callback_handle = kernel.ThreadWakeupCallbackHandleTable().Create(thread).Unwrap(); thread->owner_process = &owner_process; + auto& scheduler = kernel.GlobalScheduler(); + scheduler.AddThread(thread); thread->tls_address = thread->owner_process->CreateTLSRegion(); - thread->scheduler = &system.Scheduler(processor_id); - thread->scheduler->AddThread(thread); thread->owner_process->RegisterThread(thread.get()); @@ -250,6 +241,22 @@ void Thread::SetStatus(ThreadStatus new_status) { return; } + switch (new_status) { + case ThreadStatus::Ready: + case ThreadStatus::Running: + SetSchedulingStatus(ThreadSchedStatus::Runnable); + break; + case ThreadStatus::Dormant: + SetSchedulingStatus(ThreadSchedStatus::None); + break; + case ThreadStatus::Dead: + SetSchedulingStatus(ThreadSchedStatus::Exited); + break; + default: + SetSchedulingStatus(ThreadSchedStatus::Paused); + break; + } + if (status == ThreadStatus::Running) { last_running_ticks = Core::System::GetInstance().CoreTiming().GetTicks(); } @@ -311,8 +318,7 @@ void Thread::UpdatePriority() { return; } - scheduler->SetThreadPriority(this, new_priority); - current_priority = new_priority; + SetCurrentPriority(new_priority); if (!lock_owner) { return; @@ -328,47 +334,7 @@ void Thread::UpdatePriority() { } void Thread::ChangeCore(u32 core, u64 mask) { - ideal_core = core; - affinity_mask = mask; - ChangeScheduler(); -} - -void Thread::ChangeScheduler() { - if (status != ThreadStatus::Ready) { - return; - } - - auto& system = Core::System::GetInstance(); - std::optional<s32> new_processor_id{GetNextProcessorId(affinity_mask)}; - - if (!new_processor_id) { - new_processor_id = processor_id; - } - if (ideal_core != -1 && system.Scheduler(ideal_core).GetCurrentThread() == nullptr) { - new_processor_id = ideal_core; - } - - ASSERT(*new_processor_id < 4); - - // Add thread to new core's scheduler - auto& next_scheduler = system.Scheduler(*new_processor_id); - - if (*new_processor_id != processor_id) { - // Remove thread from previous core's scheduler - scheduler->RemoveThread(this); - next_scheduler.AddThread(this); - } - - processor_id = *new_processor_id; - - // If the thread was ready, unschedule from the previous core and schedule on the new core - scheduler->UnscheduleThread(this, current_priority); - next_scheduler.ScheduleThread(this, current_priority); - - // Change thread's scheduler - scheduler = &next_scheduler; - - system.CpuCore(processor_id).PrepareReschedule(); + SetCoreAndAffinityMask(core, mask); } bool Thread::AllWaitObjectsReady() const { @@ -388,10 +354,8 @@ void Thread::SetActivity(ThreadActivity value) { if (value == ThreadActivity::Paused) { // Set status if not waiting - if (status == ThreadStatus::Ready) { - status = ThreadStatus::Paused; - } else if (status == ThreadStatus::Running) { - status = ThreadStatus::Paused; + if (status == ThreadStatus::Ready || status == ThreadStatus::Running) { + SetStatus(ThreadStatus::Paused); Core::System::GetInstance().CpuCore(processor_id).PrepareReschedule(); } } else if (status == ThreadStatus::Paused) { @@ -408,6 +372,170 @@ void Thread::Sleep(s64 nanoseconds) { WakeAfterDelay(nanoseconds); } +bool Thread::YieldSimple() { + auto& scheduler = kernel.GlobalScheduler(); + return scheduler.YieldThread(this); +} + +bool Thread::YieldAndBalanceLoad() { + auto& scheduler = kernel.GlobalScheduler(); + return scheduler.YieldThreadAndBalanceLoad(this); +} + +bool Thread::YieldAndWaitForLoadBalancing() { + auto& scheduler = kernel.GlobalScheduler(); + return scheduler.YieldThreadAndWaitForLoadBalancing(this); +} + +void Thread::SetSchedulingStatus(ThreadSchedStatus new_status) { + const u32 old_flags = scheduling_state; + scheduling_state = (scheduling_state & static_cast<u32>(ThreadSchedMasks::HighMask)) | + static_cast<u32>(new_status); + AdjustSchedulingOnStatus(old_flags); +} + +void Thread::SetCurrentPriority(u32 new_priority) { + const u32 old_priority = std::exchange(current_priority, new_priority); + AdjustSchedulingOnPriority(old_priority); +} + +ResultCode Thread::SetCoreAndAffinityMask(s32 new_core, u64 new_affinity_mask) { + const auto HighestSetCore = [](u64 mask, u32 max_cores) { + for (s32 core = max_cores - 1; core >= 0; core--) { + if (((mask >> core) & 1) != 0) { + return core; + } + } + return -1; + }; + + const bool use_override = affinity_override_count != 0; + if (new_core == THREADPROCESSORID_DONT_UPDATE) { + new_core = use_override ? ideal_core_override : ideal_core; + if ((new_affinity_mask & (1ULL << new_core)) == 0) { + return ERR_INVALID_COMBINATION; + } + } + if (use_override) { + ideal_core_override = new_core; + affinity_mask_override = new_affinity_mask; + } else { + const u64 old_affinity_mask = std::exchange(affinity_mask, new_affinity_mask); + ideal_core = new_core; + if (old_affinity_mask != new_affinity_mask) { + const s32 old_core = processor_id; + if (processor_id >= 0 && ((affinity_mask >> processor_id) & 1) == 0) { + if (ideal_core < 0) { + processor_id = HighestSetCore(affinity_mask, GlobalScheduler::NUM_CPU_CORES); + } else { + processor_id = ideal_core; + } + } + AdjustSchedulingOnAffinity(old_affinity_mask, old_core); + } + } + return RESULT_SUCCESS; +} + +void Thread::AdjustSchedulingOnStatus(u32 old_flags) { + if (old_flags == scheduling_state) { + return; + } + + auto& scheduler = kernel.GlobalScheduler(); + if (static_cast<ThreadSchedStatus>(old_flags & static_cast<u32>(ThreadSchedMasks::LowMask)) == + ThreadSchedStatus::Runnable) { + // In this case the thread was running, now it's pausing/exitting + if (processor_id >= 0) { + scheduler.Unschedule(current_priority, processor_id, this); + } + + for (s32 core = 0; core < GlobalScheduler::NUM_CPU_CORES; core++) { + if (core != processor_id && ((affinity_mask >> core) & 1) != 0) { + scheduler.Unsuggest(current_priority, static_cast<u32>(core), this); + } + } + } else if (GetSchedulingStatus() == ThreadSchedStatus::Runnable) { + // The thread is now set to running from being stopped + if (processor_id >= 0) { + scheduler.Schedule(current_priority, processor_id, this); + } + + for (s32 core = 0; core < GlobalScheduler::NUM_CPU_CORES; core++) { + if (core != processor_id && ((affinity_mask >> core) & 1) != 0) { + scheduler.Suggest(current_priority, static_cast<u32>(core), this); + } + } + } + + scheduler.SetReselectionPending(); +} + +void Thread::AdjustSchedulingOnPriority(u32 old_priority) { + if (GetSchedulingStatus() != ThreadSchedStatus::Runnable) { + return; + } + auto& scheduler = Core::System::GetInstance().GlobalScheduler(); + if (processor_id >= 0) { + scheduler.Unschedule(old_priority, processor_id, this); + } + + for (u32 core = 0; core < GlobalScheduler::NUM_CPU_CORES; core++) { + if (core != processor_id && ((affinity_mask >> core) & 1) != 0) { + scheduler.Unsuggest(old_priority, core, this); + } + } + + // Add thread to the new priority queues. + Thread* current_thread = GetCurrentThread(); + + if (processor_id >= 0) { + if (current_thread == this) { + scheduler.SchedulePrepend(current_priority, processor_id, this); + } else { + scheduler.Schedule(current_priority, processor_id, this); + } + } + + for (u32 core = 0; core < GlobalScheduler::NUM_CPU_CORES; core++) { + if (core != processor_id && ((affinity_mask >> core) & 1) != 0) { + scheduler.Suggest(current_priority, core, this); + } + } + + scheduler.SetReselectionPending(); +} + +void Thread::AdjustSchedulingOnAffinity(u64 old_affinity_mask, s32 old_core) { + auto& scheduler = Core::System::GetInstance().GlobalScheduler(); + if (GetSchedulingStatus() != ThreadSchedStatus::Runnable || + current_priority >= THREADPRIO_COUNT) { + return; + } + + for (u32 core = 0; core < GlobalScheduler::NUM_CPU_CORES; core++) { + if (((old_affinity_mask >> core) & 1) != 0) { + if (core == old_core) { + scheduler.Unschedule(current_priority, core, this); + } else { + scheduler.Unsuggest(current_priority, core, this); + } + } + } + + for (u32 core = 0; core < GlobalScheduler::NUM_CPU_CORES; core++) { + if (((affinity_mask >> core) & 1) != 0) { + if (core == processor_id) { + scheduler.Schedule(current_priority, core, this); + } else { + scheduler.Suggest(current_priority, core, this); + } + } + } + + scheduler.SetReselectionPending(); +} + //////////////////////////////////////////////////////////////////////////////////////////////////// /** diff --git a/src/core/hle/kernel/thread.h b/src/core/hle/kernel/thread.h index 07e989637..c9870873d 100644 --- a/src/core/hle/kernel/thread.h +++ b/src/core/hle/kernel/thread.h @@ -75,6 +75,26 @@ enum class ThreadActivity : u32 { Paused = 1, }; +enum class ThreadSchedStatus : u32 { + None = 0, + Paused = 1, + Runnable = 2, + Exited = 3, +}; + +enum class ThreadSchedFlags : u32 { + ProcessPauseFlag = 1 << 4, + ThreadPauseFlag = 1 << 5, + ProcessDebugPauseFlag = 1 << 6, + KernelInitPauseFlag = 1 << 8, +}; + +enum class ThreadSchedMasks : u32 { + LowMask = 0x000f, + HighMask = 0xfff0, + ForcePauseMask = 0x0070, +}; + class Thread final : public WaitObject { public: using MutexWaitingThreads = std::vector<SharedPtr<Thread>>; @@ -278,6 +298,10 @@ public: return processor_id; } + void SetProcessorID(s32 new_core) { + processor_id = new_core; + } + Process* GetOwnerProcess() { return owner_process; } @@ -295,6 +319,9 @@ public: } void ClearWaitObjects() { + for (const auto& waiting_object : wait_objects) { + waiting_object->RemoveWaitingThread(this); + } wait_objects.clear(); } @@ -383,11 +410,47 @@ public: /// Sleeps this thread for the given amount of nanoseconds. void Sleep(s64 nanoseconds); + /// Yields this thread without rebalancing loads. + bool YieldSimple(); + + /// Yields this thread and does a load rebalancing. + bool YieldAndBalanceLoad(); + + /// Yields this thread and if the core is left idle, loads are rebalanced + bool YieldAndWaitForLoadBalancing(); + + void IncrementYieldCount() { + yield_count++; + } + + u64 GetYieldCount() const { + return yield_count; + } + + ThreadSchedStatus GetSchedulingStatus() const { + return static_cast<ThreadSchedStatus>(scheduling_state & + static_cast<u32>(ThreadSchedMasks::LowMask)); + } + + bool IsRunning() const { + return is_running; + } + + void SetIsRunning(bool value) { + is_running = value; + } + private: explicit Thread(KernelCore& kernel); ~Thread() override; - void ChangeScheduler(); + void SetSchedulingStatus(ThreadSchedStatus new_status); + void SetCurrentPriority(u32 new_priority); + ResultCode SetCoreAndAffinityMask(s32 new_core, u64 new_affinity_mask); + + void AdjustSchedulingOnStatus(u32 old_flags); + void AdjustSchedulingOnPriority(u32 old_priority); + void AdjustSchedulingOnAffinity(u64 old_affinity_mask, s32 old_core); Core::ARM_Interface::ThreadContext context{}; @@ -409,6 +472,8 @@ private: u64 total_cpu_time_ticks = 0; ///< Total CPU running ticks. u64 last_running_ticks = 0; ///< CPU tick when thread was last running + u64 yield_count = 0; ///< Number of redundant yields carried by this thread. + ///< a redundant yield is one where no scheduling is changed s32 processor_id = 0; @@ -453,6 +518,13 @@ private: ThreadActivity activity = ThreadActivity::Normal; + s32 ideal_core_override = -1; + u64 affinity_mask_override = 0x1; + u32 affinity_override_count = 0; + + u32 scheduling_state = 0; + bool is_running = false; + std::string name; }; diff --git a/src/core/hle/kernel/wait_object.cpp b/src/core/hle/kernel/wait_object.cpp index 0e96ba872..c00cef062 100644 --- a/src/core/hle/kernel/wait_object.cpp +++ b/src/core/hle/kernel/wait_object.cpp @@ -6,6 +6,9 @@ #include "common/assert.h" #include "common/common_types.h" #include "common/logging/log.h" +#include "core/core.h" +#include "core/core_cpu.h" +#include "core/hle/kernel/kernel.h" #include "core/hle/kernel/object.h" #include "core/hle/kernel/process.h" #include "core/hle/kernel/thread.h" @@ -82,9 +85,6 @@ void WaitObject::WakeupWaitingThread(SharedPtr<Thread> thread) { const std::size_t index = thread->GetWaitObjectIndex(this); - for (const auto& object : thread->GetWaitObjects()) { - object->RemoveWaitingThread(thread.get()); - } thread->ClearWaitObjects(); thread->CancelWakeupTimer(); @@ -95,6 +95,7 @@ void WaitObject::WakeupWaitingThread(SharedPtr<Thread> thread) { } if (resume) { thread->ResumeFromWait(); + Core::System::GetInstance().PrepareReschedule(thread->GetProcessorID()); } } diff --git a/src/core/hle/service/am/am.cpp b/src/core/hle/service/am/am.cpp index 941ebc93a..3d8a91d22 100644 --- a/src/core/hle/service/am/am.cpp +++ b/src/core/hle/service/am/am.cpp @@ -1073,9 +1073,9 @@ IApplicationFunctions::IApplicationFunctions(Core::System& system_) {71, nullptr, "RequestToReboot"}, {80, nullptr, "ExitAndRequestToShowThanksMessage"}, {90, &IApplicationFunctions::EnableApplicationCrashReport, "EnableApplicationCrashReport"}, - {100, nullptr, "InitializeApplicationCopyrightFrameBuffer"}, - {101, nullptr, "SetApplicationCopyrightImage"}, - {102, nullptr, "SetApplicationCopyrightVisibility"}, + {100, &IApplicationFunctions::InitializeApplicationCopyrightFrameBuffer, "InitializeApplicationCopyrightFrameBuffer"}, + {101, &IApplicationFunctions::SetApplicationCopyrightImage, "SetApplicationCopyrightImage"}, + {102, &IApplicationFunctions::SetApplicationCopyrightVisibility, "SetApplicationCopyrightVisibility"}, {110, nullptr, "QueryApplicationPlayStatistics"}, {120, nullptr, "ExecuteProgram"}, {121, nullptr, "ClearUserChannel"}, @@ -1103,6 +1103,31 @@ void IApplicationFunctions::EnableApplicationCrashReport(Kernel::HLERequestConte rb.Push(RESULT_SUCCESS); } +void IApplicationFunctions::InitializeApplicationCopyrightFrameBuffer( + Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_AM, "(STUBBED) called"); + + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); +} + +void IApplicationFunctions::SetApplicationCopyrightImage(Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_AM, "(STUBBED) called"); + + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); +} + +void IApplicationFunctions::SetApplicationCopyrightVisibility(Kernel::HLERequestContext& ctx) { + IPC::RequestParser rp{ctx}; + const auto is_visible = rp.Pop<bool>(); + + LOG_WARNING(Service_AM, "(STUBBED) called, is_visible={}", is_visible); + + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); +} + void IApplicationFunctions::BeginBlockingHomeButtonShortAndLongPressed( Kernel::HLERequestContext& ctx) { LOG_WARNING(Service_AM, "(STUBBED) called"); @@ -1140,8 +1165,9 @@ void IApplicationFunctions::PopLaunchParameter(Kernel::HLERequestContext& ctx) { LOG_DEBUG(Service_AM, "called, kind={:08X}", static_cast<u8>(kind)); if (kind == LaunchParameterKind::ApplicationSpecific && !launch_popped_application_specific) { - const auto backend = BCAT::CreateBackendFromSettings( - [this](u64 tid) { return system.GetFileSystemController().GetBCATDirectory(tid); }); + const auto backend = BCAT::CreateBackendFromSettings(system, [this](u64 tid) { + return system.GetFileSystemController().GetBCATDirectory(tid); + }); const auto build_id_full = system.GetCurrentProcessBuildID(); u64 build_id{}; std::memcpy(&build_id, build_id_full.data(), sizeof(u64)); diff --git a/src/core/hle/service/am/am.h b/src/core/hle/service/am/am.h index ccd053c13..2ae9402a8 100644 --- a/src/core/hle/service/am/am.h +++ b/src/core/hle/service/am/am.h @@ -252,6 +252,9 @@ private: void BeginBlockingHomeButton(Kernel::HLERequestContext& ctx); void EndBlockingHomeButton(Kernel::HLERequestContext& ctx); void EnableApplicationCrashReport(Kernel::HLERequestContext& ctx); + void InitializeApplicationCopyrightFrameBuffer(Kernel::HLERequestContext& ctx); + void SetApplicationCopyrightImage(Kernel::HLERequestContext& ctx); + void SetApplicationCopyrightVisibility(Kernel::HLERequestContext& ctx); void GetGpuErrorDetectedSystemEvent(Kernel::HLERequestContext& ctx); bool launch_popped_application_specific = false; diff --git a/src/core/hle/service/am/applets/error.cpp b/src/core/hle/service/am/applets/error.cpp index a7db26725..eab0d42c9 100644 --- a/src/core/hle/service/am/applets/error.cpp +++ b/src/core/hle/service/am/applets/error.cpp @@ -20,9 +20,9 @@ namespace Service::AM::Applets { struct ShowError { u8 mode; bool jump; - INSERT_PADDING_BYTES(4); + INSERT_UNION_PADDING_BYTES(4); bool use_64bit_error_code; - INSERT_PADDING_BYTES(1); + INSERT_UNION_PADDING_BYTES(1); u64 error_code_64; u32 error_code_32; }; @@ -32,7 +32,7 @@ static_assert(sizeof(ShowError) == 0x14, "ShowError has incorrect size."); struct ShowErrorRecord { u8 mode; bool jump; - INSERT_PADDING_BYTES(6); + INSERT_UNION_PADDING_BYTES(6); u64 error_code_64; u64 posix_time; }; @@ -41,7 +41,7 @@ static_assert(sizeof(ShowErrorRecord) == 0x18, "ShowErrorRecord has incorrect si struct SystemErrorArg { u8 mode; bool jump; - INSERT_PADDING_BYTES(6); + INSERT_UNION_PADDING_BYTES(6); u64 error_code_64; std::array<char, 8> language_code; std::array<char, 0x800> main_text; @@ -52,7 +52,7 @@ static_assert(sizeof(SystemErrorArg) == 0x1018, "SystemErrorArg has incorrect si struct ApplicationErrorArg { u8 mode; bool jump; - INSERT_PADDING_BYTES(6); + INSERT_UNION_PADDING_BYTES(6); u32 error_code; std::array<char, 8> language_code; std::array<char, 0x800> main_text; @@ -65,6 +65,7 @@ union Error::ErrorArguments { ShowErrorRecord error_record; SystemErrorArg system_error; ApplicationErrorArg application_error; + std::array<u8, 0x1018> raw{}; }; namespace { diff --git a/src/core/hle/service/apm/controller.cpp b/src/core/hle/service/apm/controller.cpp index 073d0f6fa..25a886238 100644 --- a/src/core/hle/service/apm/controller.cpp +++ b/src/core/hle/service/apm/controller.cpp @@ -2,6 +2,10 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. +#include <algorithm> +#include <array> +#include <utility> + #include "common/logging/log.h" #include "core/core_timing.h" #include "core/hle/service/apm/controller.h" @@ -9,8 +13,7 @@ namespace Service::APM { -constexpr PerformanceConfiguration DEFAULT_PERFORMANCE_CONFIGURATION = - PerformanceConfiguration::Config7; +constexpr auto DEFAULT_PERFORMANCE_CONFIGURATION = PerformanceConfiguration::Config7; Controller::Controller(Core::Timing::CoreTiming& core_timing) : core_timing{core_timing}, configs{ @@ -22,18 +25,35 @@ Controller::~Controller() = default; void Controller::SetPerformanceConfiguration(PerformanceMode mode, PerformanceConfiguration config) { - static const std::map<PerformanceConfiguration, u32> PCONFIG_TO_SPEED_MAP{ - {PerformanceConfiguration::Config1, 1020}, {PerformanceConfiguration::Config2, 1020}, - {PerformanceConfiguration::Config3, 1224}, {PerformanceConfiguration::Config4, 1020}, - {PerformanceConfiguration::Config5, 1020}, {PerformanceConfiguration::Config6, 1224}, - {PerformanceConfiguration::Config7, 1020}, {PerformanceConfiguration::Config8, 1020}, - {PerformanceConfiguration::Config9, 1020}, {PerformanceConfiguration::Config10, 1020}, - {PerformanceConfiguration::Config11, 1020}, {PerformanceConfiguration::Config12, 1020}, - {PerformanceConfiguration::Config13, 1785}, {PerformanceConfiguration::Config14, 1785}, - {PerformanceConfiguration::Config15, 1020}, {PerformanceConfiguration::Config16, 1020}, - }; - - SetClockSpeed(PCONFIG_TO_SPEED_MAP.find(config)->second); + static constexpr std::array<std::pair<PerformanceConfiguration, u32>, 16> config_to_speed{{ + {PerformanceConfiguration::Config1, 1020}, + {PerformanceConfiguration::Config2, 1020}, + {PerformanceConfiguration::Config3, 1224}, + {PerformanceConfiguration::Config4, 1020}, + {PerformanceConfiguration::Config5, 1020}, + {PerformanceConfiguration::Config6, 1224}, + {PerformanceConfiguration::Config7, 1020}, + {PerformanceConfiguration::Config8, 1020}, + {PerformanceConfiguration::Config9, 1020}, + {PerformanceConfiguration::Config10, 1020}, + {PerformanceConfiguration::Config11, 1020}, + {PerformanceConfiguration::Config12, 1020}, + {PerformanceConfiguration::Config13, 1785}, + {PerformanceConfiguration::Config14, 1785}, + {PerformanceConfiguration::Config15, 1020}, + {PerformanceConfiguration::Config16, 1020}, + }}; + + const auto iter = std::find_if(config_to_speed.cbegin(), config_to_speed.cend(), + [config](const auto& entry) { return entry.first == config; }); + + if (iter == config_to_speed.cend()) { + LOG_ERROR(Service_APM, "Invalid performance configuration value provided: {}", + static_cast<u32>(config)); + return; + } + + SetClockSpeed(iter->second); configs.insert_or_assign(mode, config); } @@ -48,7 +68,7 @@ void Controller::SetFromCpuBoostMode(CpuBoostMode mode) { BOOST_MODE_TO_CONFIG_MAP.at(static_cast<u32>(mode))); } -PerformanceMode Controller::GetCurrentPerformanceMode() { +PerformanceMode Controller::GetCurrentPerformanceMode() const { return Settings::values.use_docked_mode ? PerformanceMode::Docked : PerformanceMode::Handheld; } diff --git a/src/core/hle/service/apm/controller.h b/src/core/hle/service/apm/controller.h index 454caa6eb..af0c4cd34 100644 --- a/src/core/hle/service/apm/controller.h +++ b/src/core/hle/service/apm/controller.h @@ -56,7 +56,7 @@ public: void SetPerformanceConfiguration(PerformanceMode mode, PerformanceConfiguration config); void SetFromCpuBoostMode(CpuBoostMode mode); - PerformanceMode GetCurrentPerformanceMode(); + PerformanceMode GetCurrentPerformanceMode() const; PerformanceConfiguration GetCurrentPerformanceConfiguration(PerformanceMode mode); private: diff --git a/src/core/hle/service/bcat/backend/backend.cpp b/src/core/hle/service/bcat/backend/backend.cpp index 9d6946bc5..b86fda29a 100644 --- a/src/core/hle/service/bcat/backend/backend.cpp +++ b/src/core/hle/service/bcat/backend/backend.cpp @@ -10,8 +10,8 @@ namespace Service::BCAT { -ProgressServiceBackend::ProgressServiceBackend(std::string_view event_name) { - auto& kernel{Core::System::GetInstance().Kernel()}; +ProgressServiceBackend::ProgressServiceBackend(Kernel::KernelCore& kernel, + std::string_view event_name) { event = Kernel::WritableEvent::CreateEventPair( kernel, Kernel::ResetType::Automatic, std::string("ProgressServiceBackend:UpdateEvent:").append(event_name)); diff --git a/src/core/hle/service/bcat/backend/backend.h b/src/core/hle/service/bcat/backend/backend.h index 51dbd3316..ea4b16ad0 100644 --- a/src/core/hle/service/bcat/backend/backend.h +++ b/src/core/hle/service/bcat/backend/backend.h @@ -15,6 +15,14 @@ #include "core/hle/kernel/writable_event.h" #include "core/hle/result.h" +namespace Core { +class System; +} + +namespace Kernel { +class KernelCore; +} + namespace Service::BCAT { struct DeliveryCacheProgressImpl; @@ -88,7 +96,7 @@ public: void FinishDownload(ResultCode result); private: - explicit ProgressServiceBackend(std::string_view event_name); + explicit ProgressServiceBackend(Kernel::KernelCore& kernel, std::string_view event_name); Kernel::SharedPtr<Kernel::ReadableEvent> GetEvent() const; DeliveryCacheProgressImpl& GetImpl(); @@ -145,6 +153,6 @@ public: std::optional<std::vector<u8>> GetLaunchParameter(TitleIDVersion title) override; }; -std::unique_ptr<Backend> CreateBackendFromSettings(DirectoryGetter getter); +std::unique_ptr<Backend> CreateBackendFromSettings(Core::System& system, DirectoryGetter getter); } // namespace Service::BCAT diff --git a/src/core/hle/service/bcat/backend/boxcat.cpp b/src/core/hle/service/bcat/backend/boxcat.cpp index 64022982b..918159e11 100644 --- a/src/core/hle/service/bcat/backend/boxcat.cpp +++ b/src/core/hle/service/bcat/backend/boxcat.cpp @@ -104,14 +104,15 @@ std::string GetZIPFilePath(u64 title_id) { // If the error is something the user should know about (build ID mismatch, bad client version), // display an error. -void HandleDownloadDisplayResult(DownloadResult res) { +void HandleDownloadDisplayResult(const AM::Applets::AppletManager& applet_manager, + DownloadResult res) { if (res == DownloadResult::Success || res == DownloadResult::NoResponse || res == DownloadResult::GeneralWebError || res == DownloadResult::GeneralFSError || res == DownloadResult::NoMatchTitleId || res == DownloadResult::InvalidContentType) { return; } - const auto& frontend{Core::System::GetInstance().GetAppletManager().GetAppletFrontendSet()}; + const auto& frontend{applet_manager.GetAppletFrontendSet()}; frontend.error->ShowCustomErrorText( ResultCode(-1), "There was an error while attempting to use Boxcat.", DOWNLOAD_RESULT_LOG_MESSAGES[static_cast<std::size_t>(res)], [] {}); @@ -264,12 +265,13 @@ private: u64 build_id; }; -Boxcat::Boxcat(DirectoryGetter getter) : Backend(std::move(getter)) {} +Boxcat::Boxcat(AM::Applets::AppletManager& applet_manager_, DirectoryGetter getter) + : Backend(std::move(getter)), applet_manager{applet_manager_} {} Boxcat::~Boxcat() = default; -void SynchronizeInternal(DirectoryGetter dir_getter, TitleIDVersion title, - ProgressServiceBackend& progress, +void SynchronizeInternal(AM::Applets::AppletManager& applet_manager, DirectoryGetter dir_getter, + TitleIDVersion title, ProgressServiceBackend& progress, std::optional<std::string> dir_name = {}) { progress.SetNeedHLELock(true); @@ -295,7 +297,7 @@ void SynchronizeInternal(DirectoryGetter dir_getter, TitleIDVersion title, FileUtil::Delete(zip_path); } - HandleDownloadDisplayResult(res); + HandleDownloadDisplayResult(applet_manager, res); progress.FinishDownload(ERROR_GENERAL_BCAT_FAILURE); return; } @@ -364,17 +366,24 @@ void SynchronizeInternal(DirectoryGetter dir_getter, TitleIDVersion title, bool Boxcat::Synchronize(TitleIDVersion title, ProgressServiceBackend& progress) { is_syncing.exchange(true); - std::thread([this, title, &progress] { SynchronizeInternal(dir_getter, title, progress); }) + + std::thread([this, title, &progress] { + SynchronizeInternal(applet_manager, dir_getter, title, progress); + }) .detach(); + return true; } bool Boxcat::SynchronizeDirectory(TitleIDVersion title, std::string name, ProgressServiceBackend& progress) { is_syncing.exchange(true); - std::thread( - [this, title, name, &progress] { SynchronizeInternal(dir_getter, title, progress, name); }) + + std::thread([this, title, name, &progress] { + SynchronizeInternal(applet_manager, dir_getter, title, progress, name); + }) .detach(); + return true; } @@ -420,7 +429,7 @@ std::optional<std::vector<u8>> Boxcat::GetLaunchParameter(TitleIDVersion title) FileUtil::Delete(path); } - HandleDownloadDisplayResult(res); + HandleDownloadDisplayResult(applet_manager, res); return std::nullopt; } } diff --git a/src/core/hle/service/bcat/backend/boxcat.h b/src/core/hle/service/bcat/backend/boxcat.h index 601151189..d65b42e58 100644 --- a/src/core/hle/service/bcat/backend/boxcat.h +++ b/src/core/hle/service/bcat/backend/boxcat.h @@ -9,6 +9,10 @@ #include <optional> #include "core/hle/service/bcat/backend/backend.h" +namespace Service::AM::Applets { +class AppletManager; +} + namespace Service::BCAT { struct EventStatus { @@ -20,12 +24,13 @@ struct EventStatus { /// Boxcat is yuzu's custom backend implementation of Nintendo's BCAT service. It is free to use and /// doesn't require a switch or nintendo account. The content is controlled by the yuzu team. class Boxcat final : public Backend { - friend void SynchronizeInternal(DirectoryGetter dir_getter, TitleIDVersion title, + friend void SynchronizeInternal(AM::Applets::AppletManager& applet_manager, + DirectoryGetter dir_getter, TitleIDVersion title, ProgressServiceBackend& progress, std::optional<std::string> dir_name); public: - explicit Boxcat(DirectoryGetter getter); + explicit Boxcat(AM::Applets::AppletManager& applet_manager_, DirectoryGetter getter); ~Boxcat() override; bool Synchronize(TitleIDVersion title, ProgressServiceBackend& progress) override; @@ -53,6 +58,7 @@ private: class Client; std::unique_ptr<Client> client; + AM::Applets::AppletManager& applet_manager; }; } // namespace Service::BCAT diff --git a/src/core/hle/service/bcat/module.cpp b/src/core/hle/service/bcat/module.cpp index 4e4aa758b..6d9d1527d 100644 --- a/src/core/hle/service/bcat/module.cpp +++ b/src/core/hle/service/bcat/module.cpp @@ -125,7 +125,11 @@ private: class IBcatService final : public ServiceFramework<IBcatService> { public: explicit IBcatService(Core::System& system_, Backend& backend_) - : ServiceFramework("IBcatService"), system{system_}, backend{backend_} { + : ServiceFramework("IBcatService"), system{system_}, backend{backend_}, + progress{{ + ProgressServiceBackend{system_.Kernel(), "Normal"}, + ProgressServiceBackend{system_.Kernel(), "Directory"}, + }} { // clang-format off static const FunctionInfo functions[] = { {10100, &IBcatService::RequestSyncDeliveryCache, "RequestSyncDeliveryCache"}, @@ -249,10 +253,7 @@ private: Core::System& system; Backend& backend; - std::array<ProgressServiceBackend, static_cast<std::size_t>(SyncType::Count)> progress{ - ProgressServiceBackend{"Normal"}, - ProgressServiceBackend{"Directory"}, - }; + std::array<ProgressServiceBackend, static_cast<std::size_t>(SyncType::Count)> progress; }; void Module::Interface::CreateBcatService(Kernel::HLERequestContext& ctx) { @@ -557,12 +558,12 @@ void Module::Interface::CreateDeliveryCacheStorageServiceWithApplicationId( rb.PushIpcInterface<IDeliveryCacheStorageService>(fsc.GetBCATDirectory(title_id)); } -std::unique_ptr<Backend> CreateBackendFromSettings(DirectoryGetter getter) { - const auto backend = Settings::values.bcat_backend; - +std::unique_ptr<Backend> CreateBackendFromSettings([[maybe_unused]] Core::System& system, + DirectoryGetter getter) { #ifdef YUZU_ENABLE_BOXCAT - if (backend == "boxcat") - return std::make_unique<Boxcat>(std::move(getter)); + if (Settings::values.bcat_backend == "boxcat") { + return std::make_unique<Boxcat>(system.GetAppletManager(), std::move(getter)); + } #endif return std::make_unique<NullBackend>(std::move(getter)); @@ -571,7 +572,8 @@ std::unique_ptr<Backend> CreateBackendFromSettings(DirectoryGetter getter) { Module::Interface::Interface(Core::System& system_, std::shared_ptr<Module> module_, FileSystem::FileSystemController& fsc_, const char* name) : ServiceFramework(name), fsc{fsc_}, module{std::move(module_)}, - backend{CreateBackendFromSettings([&fsc_](u64 tid) { return fsc_.GetBCATDirectory(tid); })}, + backend{CreateBackendFromSettings(system_, + [&fsc_](u64 tid) { return fsc_.GetBCATDirectory(tid); })}, system{system_} {} Module::Interface::~Interface() = default; diff --git a/src/core/hle/service/es/es.cpp b/src/core/hle/service/es/es.cpp index af70d174d..f77ddd739 100644 --- a/src/core/hle/service/es/es.cpp +++ b/src/core/hle/service/es/es.cpp @@ -128,7 +128,7 @@ private: void CountCommonTicket(Kernel::HLERequestContext& ctx) { LOG_DEBUG(Service_ETicket, "called"); - const auto count = keys.GetCommonTickets().size(); + const u32 count = static_cast<u32>(keys.GetCommonTickets().size()); IPC::ResponseBuilder rb{ctx, 3}; rb.Push(RESULT_SUCCESS); @@ -138,7 +138,7 @@ private: void CountPersonalizedTicket(Kernel::HLERequestContext& ctx) { LOG_DEBUG(Service_ETicket, "called"); - const auto count = keys.GetPersonalizedTickets().size(); + const u32 count = static_cast<u32>(keys.GetPersonalizedTickets().size()); IPC::ResponseBuilder rb{ctx, 3}; rb.Push(RESULT_SUCCESS); @@ -150,7 +150,7 @@ private: if (keys.GetCommonTickets().empty()) out_entries = 0; else - out_entries = ctx.GetWriteBufferSize() / sizeof(u128); + out_entries = static_cast<u32>(ctx.GetWriteBufferSize() / sizeof(u128)); LOG_DEBUG(Service_ETicket, "called, entries={:016X}", out_entries); @@ -160,7 +160,7 @@ private: std::transform(tickets.begin(), tickets.end(), std::back_inserter(ids), [](const auto& pair) { return pair.first; }); - out_entries = std::min<u32>(ids.size(), out_entries); + out_entries = static_cast<u32>(std::min<std::size_t>(ids.size(), out_entries)); ctx.WriteBuffer(ids.data(), out_entries * sizeof(u128)); IPC::ResponseBuilder rb{ctx, 3}; @@ -173,7 +173,7 @@ private: if (keys.GetPersonalizedTickets().empty()) out_entries = 0; else - out_entries = ctx.GetWriteBufferSize() / sizeof(u128); + out_entries = static_cast<u32>(ctx.GetWriteBufferSize() / sizeof(u128)); LOG_DEBUG(Service_ETicket, "called, entries={:016X}", out_entries); @@ -183,7 +183,7 @@ private: std::transform(tickets.begin(), tickets.end(), std::back_inserter(ids), [](const auto& pair) { return pair.first; }); - out_entries = std::min<u32>(ids.size(), out_entries); + out_entries = static_cast<u32>(std::min<std::size_t>(ids.size(), out_entries)); ctx.WriteBuffer(ids.data(), out_entries * sizeof(u128)); IPC::ResponseBuilder rb{ctx, 3}; diff --git a/src/core/hle/service/hid/controllers/npad.cpp b/src/core/hle/service/hid/controllers/npad.cpp index a2b25a796..81bd2f3cb 100644 --- a/src/core/hle/service/hid/controllers/npad.cpp +++ b/src/core/hle/service/hid/controllers/npad.cpp @@ -583,36 +583,6 @@ bool Controller_NPad::SwapNpadAssignment(u32 npad_id_1, u32 npad_id_2) { return true; } -bool Controller_NPad::IsControllerSupported(NPadControllerType controller) { - if (controller == NPadControllerType::Handheld) { - // Handheld is not even a supported type, lets stop here - if (std::find(supported_npad_id_types.begin(), supported_npad_id_types.end(), - NPAD_HANDHELD) == supported_npad_id_types.end()) { - return false; - } - // Handheld should not be supported in docked mode - if (Settings::values.use_docked_mode) { - return false; - } - } - switch (controller) { - case NPadControllerType::ProController: - return style.pro_controller; - case NPadControllerType::Handheld: - return style.handheld; - case NPadControllerType::JoyDual: - return style.joycon_dual; - case NPadControllerType::JoyLeft: - return style.joycon_left; - case NPadControllerType::JoyRight: - return style.joycon_right; - case NPadControllerType::Pokeball: - return style.pokeball; - default: - return false; - } -} - Controller_NPad::LedPattern Controller_NPad::GetLedPattern(u32 npad_id) { if (npad_id == npad_id_list.back() || npad_id == npad_id_list[npad_id_list.size() - 2]) { // These are controllers without led patterns @@ -659,25 +629,24 @@ void Controller_NPad::ClearAllConnectedControllers() { } void Controller_NPad::DisconnectAllConnectedControllers() { - std::for_each(connected_controllers.begin(), connected_controllers.end(), - [](ControllerHolder& controller) { controller.is_connected = false; }); + for (ControllerHolder& controller : connected_controllers) { + controller.is_connected = false; + } } void Controller_NPad::ConnectAllDisconnectedControllers() { - std::for_each(connected_controllers.begin(), connected_controllers.end(), - [](ControllerHolder& controller) { - if (controller.type != NPadControllerType::None && !controller.is_connected) { - controller.is_connected = false; - } - }); + for (ControllerHolder& controller : connected_controllers) { + if (controller.type != NPadControllerType::None && !controller.is_connected) { + controller.is_connected = true; + } + } } void Controller_NPad::ClearAllControllers() { - std::for_each(connected_controllers.begin(), connected_controllers.end(), - [](ControllerHolder& controller) { - controller.type = NPadControllerType::None; - controller.is_connected = false; - }); + for (ControllerHolder& controller : connected_controllers) { + controller.type = NPadControllerType::None; + controller.is_connected = false; + } } u32 Controller_NPad::GetAndResetPressState() { @@ -685,10 +654,10 @@ u32 Controller_NPad::GetAndResetPressState() { } bool Controller_NPad::IsControllerSupported(NPadControllerType controller) const { - const bool support_handheld = - std::find(supported_npad_id_types.begin(), supported_npad_id_types.end(), NPAD_HANDHELD) != - supported_npad_id_types.end(); if (controller == NPadControllerType::Handheld) { + const bool support_handheld = + std::find(supported_npad_id_types.begin(), supported_npad_id_types.end(), + NPAD_HANDHELD) != supported_npad_id_types.end(); // Handheld is not even a supported type, lets stop here if (!support_handheld) { return false; @@ -700,6 +669,7 @@ bool Controller_NPad::IsControllerSupported(NPadControllerType controller) const return true; } + if (std::any_of(supported_npad_id_types.begin(), supported_npad_id_types.end(), [](u32 npad_id) { return npad_id <= MAX_NPAD_ID; })) { switch (controller) { @@ -717,6 +687,7 @@ bool Controller_NPad::IsControllerSupported(NPadControllerType controller) const return false; } } + return false; } @@ -795,6 +766,7 @@ Controller_NPad::NPadControllerType Controller_NPad::DecideBestController( priority_list.push_back(NPadControllerType::JoyLeft); priority_list.push_back(NPadControllerType::JoyRight); priority_list.push_back(NPadControllerType::JoyDual); + break; } const auto iter = std::find_if(priority_list.begin(), priority_list.end(), diff --git a/src/core/hle/service/hid/controllers/npad.h b/src/core/hle/service/hid/controllers/npad.h index 1bc3d55d6..16c4caa1f 100644 --- a/src/core/hle/service/hid/controllers/npad.h +++ b/src/core/hle/service/hid/controllers/npad.h @@ -301,6 +301,11 @@ private: bool is_connected; }; + void InitNewlyAddedControler(std::size_t controller_idx); + bool IsControllerSupported(NPadControllerType controller) const; + NPadControllerType DecideBestController(NPadControllerType priority) const; + void RequestPadStateUpdate(u32 npad_id); + u32 press_state{}; NPadType style{}; @@ -321,12 +326,7 @@ private: std::array<ControllerHolder, 10> connected_controllers{}; bool can_controllers_vibrate{true}; - void InitNewlyAddedControler(std::size_t controller_idx); - bool IsControllerSupported(NPadControllerType controller) const; - NPadControllerType DecideBestController(NPadControllerType priority) const; - void RequestPadStateUpdate(u32 npad_id); std::array<ControllerPad, 10> npad_pad_states{}; - bool IsControllerSupported(NPadControllerType controller); bool is_in_lr_assignment_mode{false}; Core::System& system; }; diff --git a/src/core/hle/service/hid/hid.cpp b/src/core/hle/service/hid/hid.cpp index 9ab8b930e..ecc130f6c 100644 --- a/src/core/hle/service/hid/hid.cpp +++ b/src/core/hle/service/hid/hid.cpp @@ -195,7 +195,7 @@ Hid::Hid(Core::System& system) : ServiceFramework("hid"), system(system) { {101, &Hid::GetSupportedNpadStyleSet, "GetSupportedNpadStyleSet"}, {102, &Hid::SetSupportedNpadIdType, "SetSupportedNpadIdType"}, {103, &Hid::ActivateNpad, "ActivateNpad"}, - {104, nullptr, "DeactivateNpad"}, + {104, &Hid::DeactivateNpad, "DeactivateNpad"}, {106, &Hid::AcquireNpadStyleSetUpdateEventHandle, "AcquireNpadStyleSetUpdateEventHandle"}, {107, &Hid::DisconnectNpad, "DisconnectNpad"}, {108, &Hid::GetPlayerLedPattern, "GetPlayerLedPattern"}, @@ -470,6 +470,17 @@ void Hid::ActivateNpad(Kernel::HLERequestContext& ctx) { applet_resource->ActivateController(HidController::NPad); } +void Hid::DeactivateNpad(Kernel::HLERequestContext& ctx) { + IPC::RequestParser rp{ctx}; + const auto applet_resource_user_id{rp.Pop<u64>()}; + + LOG_DEBUG(Service_HID, "called, applet_resource_user_id={}", applet_resource_user_id); + + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); + applet_resource->DeactivateController(HidController::NPad); +} + void Hid::AcquireNpadStyleSetUpdateEventHandle(Kernel::HLERequestContext& ctx) { IPC::RequestParser rp{ctx}; const auto npad_id{rp.Pop<u32>()}; diff --git a/src/core/hle/service/hid/hid.h b/src/core/hle/service/hid/hid.h index 10abd10ed..f08e036a3 100644 --- a/src/core/hle/service/hid/hid.h +++ b/src/core/hle/service/hid/hid.h @@ -99,6 +99,7 @@ private: void GetSupportedNpadStyleSet(Kernel::HLERequestContext& ctx); void SetSupportedNpadIdType(Kernel::HLERequestContext& ctx); void ActivateNpad(Kernel::HLERequestContext& ctx); + void DeactivateNpad(Kernel::HLERequestContext& ctx); void AcquireNpadStyleSetUpdateEventHandle(Kernel::HLERequestContext& ctx); void DisconnectNpad(Kernel::HLERequestContext& ctx); void GetPlayerLedPattern(Kernel::HLERequestContext& ctx); diff --git a/src/core/hle/service/lm/lm.cpp b/src/core/hle/service/lm/lm.cpp index 2a61593e2..435f2d286 100644 --- a/src/core/hle/service/lm/lm.cpp +++ b/src/core/hle/service/lm/lm.cpp @@ -6,8 +6,10 @@ #include <string> #include "common/logging/log.h" +#include "common/scope_exit.h" #include "core/hle/ipc_helpers.h" #include "core/hle/service/lm/lm.h" +#include "core/hle/service/lm/manager.h" #include "core/hle/service/service.h" #include "core/memory.h" @@ -15,65 +17,16 @@ namespace Service::LM { class ILogger final : public ServiceFramework<ILogger> { public: - ILogger() : ServiceFramework("ILogger") { + ILogger(Manager& manager) : ServiceFramework("ILogger"), manager(manager) { static const FunctionInfo functions[] = { - {0x00000000, &ILogger::Initialize, "Initialize"}, - {0x00000001, &ILogger::SetDestination, "SetDestination"}, + {0, &ILogger::Log, "Log"}, + {1, &ILogger::SetDestination, "SetDestination"}, }; RegisterHandlers(functions); } private: - struct MessageHeader { - enum Flags : u32_le { - IsHead = 1, - IsTail = 2, - }; - enum Severity : u32_le { - Trace, - Info, - Warning, - Error, - Critical, - }; - - u64_le pid; - u64_le threadContext; - union { - BitField<0, 16, Flags> flags; - BitField<16, 8, Severity> severity; - BitField<24, 8, u32> verbosity; - }; - u32_le payload_size; - - bool IsHeadLog() const { - return flags & Flags::IsHead; - } - bool IsTailLog() const { - return flags & Flags::IsTail; - } - }; - static_assert(sizeof(MessageHeader) == 0x18, "MessageHeader is incorrect size"); - - /// Log field type - enum class Field : u8 { - Skip = 1, - Message = 2, - Line = 3, - Filename = 4, - Function = 5, - Module = 6, - Thread = 7, - }; - - /** - * ILogger::Initialize service function - * Inputs: - * 0: 0x00000000 - * Outputs: - * 0: ResultCode - */ - void Initialize(Kernel::HLERequestContext& ctx) { + void Log(Kernel::HLERequestContext& ctx) { // This function only succeeds - Get that out of the way IPC::ResponseBuilder rb{ctx, 2}; rb.Push(RESULT_SUCCESS); @@ -85,140 +38,70 @@ private: Memory::ReadBlock(addr, &header, sizeof(MessageHeader)); addr += sizeof(MessageHeader); - if (header.IsHeadLog()) { - log_stream.str(""); - log_stream.clear(); - } - - // Parse out log metadata - u32 line{}; - std::string module; - std::string message; - std::string filename; - std::string function; - std::string thread; + FieldMap fields; while (addr < end_addr) { - const Field field{static_cast<Field>(Memory::Read8(addr++))}; - const std::size_t length{Memory::Read8(addr++)}; + const auto field = static_cast<Field>(Memory::Read8(addr++)); + const auto length = Memory::Read8(addr++); if (static_cast<Field>(Memory::Read8(addr)) == Field::Skip) { ++addr; } - switch (field) { - case Field::Skip: - break; - case Field::Message: - message = Memory::ReadCString(addr, length); - break; - case Field::Line: - line = Memory::Read32(addr); - break; - case Field::Filename: - filename = Memory::ReadCString(addr, length); - break; - case Field::Function: - function = Memory::ReadCString(addr, length); - break; - case Field::Module: - module = Memory::ReadCString(addr, length); - break; - case Field::Thread: - thread = Memory::ReadCString(addr, length); - break; - } + SCOPE_EXIT({ addr += length; }); - addr += length; - } + if (field == Field::Skip) { + continue; + } - // Empty log - nothing to do here - if (log_stream.str().empty() && message.empty()) { - return; + std::vector<u8> data(length); + Memory::ReadBlock(addr, data.data(), length); + fields.emplace(field, std::move(data)); } - // Format a nicely printable string out of the log metadata - if (!filename.empty()) { - log_stream << filename << ':'; - } - if (!module.empty()) { - log_stream << module << ':'; - } - if (!function.empty()) { - log_stream << function << ':'; - } - if (line) { - log_stream << std::to_string(line) << ':'; - } - if (!thread.empty()) { - log_stream << thread << ':'; - } - if (log_stream.str().length() > 0 && log_stream.str().back() == ':') { - log_stream << ' '; - } - log_stream << message; - - if (header.IsTailLog()) { - switch (header.severity) { - case MessageHeader::Severity::Trace: - LOG_DEBUG(Debug_Emulated, "{}", log_stream.str()); - break; - case MessageHeader::Severity::Info: - LOG_INFO(Debug_Emulated, "{}", log_stream.str()); - break; - case MessageHeader::Severity::Warning: - LOG_WARNING(Debug_Emulated, "{}", log_stream.str()); - break; - case MessageHeader::Severity::Error: - LOG_ERROR(Debug_Emulated, "{}", log_stream.str()); - break; - case MessageHeader::Severity::Critical: - LOG_CRITICAL(Debug_Emulated, "{}", log_stream.str()); - break; - } - } + manager.Log({header, std::move(fields)}); } - // This service function is intended to be used as a way to - // redirect logging output to different destinations, however, - // given we always want to see the logging output, it's sufficient - // to do nothing and return success here. void SetDestination(Kernel::HLERequestContext& ctx) { - LOG_DEBUG(Service_LM, "called"); + IPC::RequestParser rp{ctx}; + const auto destination = rp.PopEnum<DestinationFlag>(); + + LOG_DEBUG(Service_LM, "called, destination={:08X}", static_cast<u32>(destination)); + + manager.SetDestination(destination); IPC::ResponseBuilder rb{ctx, 2}; rb.Push(RESULT_SUCCESS); } - std::ostringstream log_stream; + Manager& manager; }; class LM final : public ServiceFramework<LM> { public: - explicit LM() : ServiceFramework{"lm"} { + explicit LM(Manager& manager) : ServiceFramework{"lm"}, manager(manager) { + // clang-format off static const FunctionInfo functions[] = { - {0x00000000, &LM::OpenLogger, "OpenLogger"}, + {0, &LM::OpenLogger, "OpenLogger"}, }; + // clang-format on + RegisterHandlers(functions); } - /** - * LM::OpenLogger service function - * Inputs: - * 0: 0x00000000 - * Outputs: - * 0: ResultCode - */ +private: void OpenLogger(Kernel::HLERequestContext& ctx) { LOG_DEBUG(Service_LM, "called"); IPC::ResponseBuilder rb{ctx, 2, 0, 1}; rb.Push(RESULT_SUCCESS); - rb.PushIpcInterface<ILogger>(); + rb.PushIpcInterface<ILogger>(manager); } + + Manager& manager; }; -void InstallInterfaces(SM::ServiceManager& service_manager) { - std::make_shared<LM>()->InstallAsService(service_manager); +void InstallInterfaces(Core::System& system) { + std::make_shared<LM>(system.GetLogManager())->InstallAsService(system.ServiceManager()); } } // namespace Service::LM diff --git a/src/core/hle/service/lm/lm.h b/src/core/hle/service/lm/lm.h index 7806ae27b..d40410b5c 100644 --- a/src/core/hle/service/lm/lm.h +++ b/src/core/hle/service/lm/lm.h @@ -4,13 +4,13 @@ #pragma once -namespace Service::SM { -class ServiceManager; +namespace Core { +class System; } namespace Service::LM { /// Registers all LM services with the specified service manager. -void InstallInterfaces(SM::ServiceManager& service_manager); +void InstallInterfaces(Core::System& system); } // namespace Service::LM diff --git a/src/core/hle/service/lm/manager.cpp b/src/core/hle/service/lm/manager.cpp new file mode 100644 index 000000000..b67081b86 --- /dev/null +++ b/src/core/hle/service/lm/manager.cpp @@ -0,0 +1,133 @@ +// Copyright 2019 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include "common/assert.h" +#include "common/logging/log.h" +#include "common/string_util.h" +#include "core/hle/service/lm/manager.h" +#include "core/reporter.h" + +namespace Service::LM { + +std::ostream& operator<<(std::ostream& os, DestinationFlag dest) { + std::vector<std::string> array; + const auto check_single_flag = [dest, &array](DestinationFlag check, std::string name) { + if ((static_cast<u32>(check) & static_cast<u32>(dest)) != 0) { + array.emplace_back(std::move(name)); + } + }; + + check_single_flag(DestinationFlag::Default, "Default"); + check_single_flag(DestinationFlag::UART, "UART"); + check_single_flag(DestinationFlag::UARTSleeping, "UART (Sleeping)"); + + os << "["; + for (const auto& entry : array) { + os << entry << ", "; + } + return os << "]"; +} + +std::ostream& operator<<(std::ostream& os, MessageHeader::Severity severity) { + switch (severity) { + case MessageHeader::Severity::Trace: + return os << "Trace"; + case MessageHeader::Severity::Info: + return os << "Info"; + case MessageHeader::Severity::Warning: + return os << "Warning"; + case MessageHeader::Severity::Error: + return os << "Error"; + case MessageHeader::Severity::Critical: + return os << "Critical"; + default: + return os << fmt::format("{:08X}", static_cast<u32>(severity)); + } +} + +std::ostream& operator<<(std::ostream& os, Field field) { + switch (field) { + case Field::Skip: + return os << "Skip"; + case Field::Message: + return os << "Message"; + case Field::Line: + return os << "Line"; + case Field::Filename: + return os << "Filename"; + case Field::Function: + return os << "Function"; + case Field::Module: + return os << "Module"; + case Field::Thread: + return os << "Thread"; + default: + return os << fmt::format("{:08X}", static_cast<u32>(field)); + } +} + +std::string FormatField(Field type, const std::vector<u8>& data) { + switch (type) { + case Field::Skip: + return ""; + case Field::Line: + if (data.size() >= sizeof(u32)) { + u32 line; + std::memcpy(&line, data.data(), sizeof(u32)); + return fmt::format("{}", line); + } + return "[ERROR DECODING LINE NUMBER]"; + case Field::Message: + case Field::Filename: + case Field::Function: + case Field::Module: + case Field::Thread: + return Common::StringFromFixedZeroTerminatedBuffer( + reinterpret_cast<const char*>(data.data()), data.size()); + default: + UNIMPLEMENTED(); + } +} + +Manager::Manager(Core::Reporter& reporter) : reporter(reporter) {} + +Manager::~Manager() = default; + +void Manager::SetEnabled(bool enabled) { + this->enabled = enabled; +} + +void Manager::SetDestination(DestinationFlag destination) { + this->destination = destination; +} + +void Manager::Log(LogMessage message) { + if (message.header.IsHeadLog()) { + InitializeLog(); + } + + current_log.emplace_back(std::move(message)); + + if (current_log.back().header.IsTailLog()) { + FinalizeLog(); + } +} + +void Manager::Flush() { + FinalizeLog(); +} + +void Manager::InitializeLog() { + current_log.clear(); + + LOG_INFO(Service_LM, "Initialized new log session"); +} + +void Manager::FinalizeLog() { + reporter.SaveLogReport(static_cast<u32>(destination), std::move(current_log)); + + LOG_INFO(Service_LM, "Finalized current log session"); +} + +} // namespace Service::LM diff --git a/src/core/hle/service/lm/manager.h b/src/core/hle/service/lm/manager.h new file mode 100644 index 000000000..544e636ba --- /dev/null +++ b/src/core/hle/service/lm/manager.h @@ -0,0 +1,106 @@ +// Copyright 2019 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <map> +#include <ostream> +#include <vector> +#include "common/bit_field.h" +#include "common/common_types.h" +#include "common/swap.h" + +namespace Core { +class Reporter; +} + +namespace Service::LM { + +enum class DestinationFlag : u32 { + Default = 1, + UART = 2, + UARTSleeping = 4, + + All = 0xFFFF, +}; + +struct MessageHeader { + enum Flags : u32_le { + IsHead = 1, + IsTail = 2, + }; + enum Severity : u32_le { + Trace, + Info, + Warning, + Error, + Critical, + }; + + u64_le pid; + u64_le thread_context; + union { + BitField<0, 16, Flags> flags; + BitField<16, 8, Severity> severity; + BitField<24, 8, u32> verbosity; + }; + u32_le payload_size; + + bool IsHeadLog() const { + return flags & IsHead; + } + bool IsTailLog() const { + return flags & IsTail; + } +}; +static_assert(sizeof(MessageHeader) == 0x18, "MessageHeader is incorrect size"); + +enum class Field : u8 { + Skip = 1, + Message = 2, + Line = 3, + Filename = 4, + Function = 5, + Module = 6, + Thread = 7, +}; + +std::ostream& operator<<(std::ostream& os, DestinationFlag dest); +std::ostream& operator<<(std::ostream& os, MessageHeader::Severity severity); +std::ostream& operator<<(std::ostream& os, Field field); + +using FieldMap = std::map<Field, std::vector<u8>>; + +struct LogMessage { + MessageHeader header; + FieldMap fields; +}; + +std::string FormatField(Field type, const std::vector<u8>& data); + +class Manager { +public: + explicit Manager(Core::Reporter& reporter); + ~Manager(); + + void SetEnabled(bool enabled); + void SetDestination(DestinationFlag destination); + + void Log(LogMessage message); + + void Flush(); + +private: + void InitializeLog(); + void FinalizeLog(); + + bool enabled = true; + DestinationFlag destination = DestinationFlag::All; + + std::vector<LogMessage> current_log; + + Core::Reporter& reporter; +}; + +} // namespace Service::LM diff --git a/src/core/hle/service/ns/pl_u.cpp b/src/core/hle/service/ns/pl_u.cpp index f64535237..23477315f 100644 --- a/src/core/hle/service/ns/pl_u.cpp +++ b/src/core/hle/service/ns/pl_u.cpp @@ -6,13 +6,6 @@ #include <cstring> #include <vector> -#include <FontChineseSimplified.h> -#include <FontChineseTraditional.h> -#include <FontExtendedChineseSimplified.h> -#include <FontKorean.h> -#include <FontNintendoExtended.h> -#include <FontStandard.h> - #include "common/assert.h" #include "common/common_paths.h" #include "common/common_types.h" @@ -24,7 +17,9 @@ #include "core/file_sys/nca_metadata.h" #include "core/file_sys/registered_cache.h" #include "core/file_sys/romfs.h" +#include "core/file_sys/system_archive/system_archive.h" #include "core/hle/ipc_helpers.h" +#include "core/hle/kernel/physical_memory.h" #include "core/hle/kernel/shared_memory.h" #include "core/hle/service/filesystem/filesystem.h" #include "core/hle/service/ns/pl_u.h" @@ -94,15 +89,20 @@ static void DecryptSharedFont(const std::vector<u32>& input, Kernel::PhysicalMem offset += transformed_font.size() * sizeof(u32); } -static void EncryptSharedFont(const std::vector<u8>& input, Kernel::PhysicalMemory& output, - std::size_t& offset) { - ASSERT_MSG(offset + input.size() + 8 < SHARED_FONT_MEM_SIZE, "Shared fonts exceeds 17mb!"); - const u32 KEY = EXPECTED_MAGIC ^ EXPECTED_RESULT; - std::memcpy(output.data() + offset, &EXPECTED_RESULT, sizeof(u32)); // Magic header - const u32 ENC_SIZE = static_cast<u32>(input.size()) ^ KEY; - std::memcpy(output.data() + offset + sizeof(u32), &ENC_SIZE, sizeof(u32)); - std::memcpy(output.data() + offset + (sizeof(u32) * 2), input.data(), input.size()); - offset += input.size() + (sizeof(u32) * 2); +void EncryptSharedFont(const std::vector<u32>& input, std::vector<u8>& output, + std::size_t& offset) { + ASSERT_MSG(offset + (input.size() * sizeof(u32)) < SHARED_FONT_MEM_SIZE, + "Shared fonts exceeds 17mb!"); + + const auto key = Common::swap32(EXPECTED_RESULT ^ EXPECTED_MAGIC); + std::vector<u32> transformed_font(input.size() + 2); + transformed_font[0] = Common::swap32(EXPECTED_MAGIC); + transformed_font[1] = Common::swap32(input.size() * sizeof(u32)) ^ key; + std::transform(input.begin(), input.end(), transformed_font.begin() + 2, + [key](u32 in) { return in ^ key; }); + std::memcpy(output.data() + offset, transformed_font.data(), + transformed_font.size() * sizeof(u32)); + offset += transformed_font.size() * sizeof(u32); } // Helper function to make BuildSharedFontsRawRegions a bit nicer @@ -168,114 +168,49 @@ PL_U::PL_U(Core::System& system) // Attempt to load shared font data from disk const auto* nand = fsc.GetSystemNANDContents(); std::size_t offset = 0; - // Rebuild shared fonts from data ncas - if (nand->HasEntry(static_cast<u64>(FontArchives::Standard), - FileSys::ContentRecordType::Data)) { - impl->shared_font = std::make_shared<Kernel::PhysicalMemory>(SHARED_FONT_MEM_SIZE); - for (auto font : SHARED_FONTS) { - const auto nca = - nand->GetEntry(static_cast<u64>(font.first), FileSys::ContentRecordType::Data); - if (!nca) { - LOG_ERROR(Service_NS, "Failed to find {:016X}! Skipping", - static_cast<u64>(font.first)); - continue; - } - const auto romfs = nca->GetRomFS(); - if (!romfs) { - LOG_ERROR(Service_NS, "{:016X} has no RomFS! Skipping", - static_cast<u64>(font.first)); - continue; - } - const auto extracted_romfs = FileSys::ExtractRomFS(romfs); - if (!extracted_romfs) { - LOG_ERROR(Service_NS, "Failed to extract RomFS for {:016X}! Skipping", - static_cast<u64>(font.first)); - continue; - } - const auto font_fp = extracted_romfs->GetFile(font.second); - if (!font_fp) { - LOG_ERROR(Service_NS, "{:016X} has no file \"{}\"! Skipping", - static_cast<u64>(font.first), font.second); - continue; - } - std::vector<u32> font_data_u32(font_fp->GetSize() / sizeof(u32)); - font_fp->ReadBytes<u32>(font_data_u32.data(), font_fp->GetSize()); - // We need to be BigEndian as u32s for the xor encryption - std::transform(font_data_u32.begin(), font_data_u32.end(), font_data_u32.begin(), - Common::swap32); - FontRegion region{ - static_cast<u32>(offset + 8), - static_cast<u32>((font_data_u32.size() * sizeof(u32)) - - 8)}; // Font offset and size do not account for the header - DecryptSharedFont(font_data_u32, *impl->shared_font, offset); - impl->shared_font_regions.push_back(region); + // Rebuild shared fonts from data ncas or synthesize + + impl->shared_font = std::make_shared<Kernel::PhysicalMemory>(SHARED_FONT_MEM_SIZE); + for (auto font : SHARED_FONTS) { + FileSys::VirtualFile romfs; + const auto nca = + nand->GetEntry(static_cast<u64>(font.first), FileSys::ContentRecordType::Data); + if (nca) { + romfs = nca->GetRomFS(); } - } else { - impl->shared_font = std::make_shared<Kernel::PhysicalMemory>( - SHARED_FONT_MEM_SIZE); // Shared memory needs to always be allocated and a fixed size - - const std::string user_path = FileUtil::GetUserPath(FileUtil::UserPath::SysDataDir); - const std::string filepath{user_path + SHARED_FONT}; + if (!romfs) { + romfs = FileSys::SystemArchive::SynthesizeSystemArchive(static_cast<u64>(font.first)); + } - // Create path if not already created - if (!FileUtil::CreateFullPath(filepath)) { - LOG_ERROR(Service_NS, "Failed to create sharedfonts path \"{}\"!", filepath); - return; + if (!romfs) { + LOG_ERROR(Service_NS, "Failed to find or synthesize {:016X}! Skipping", + static_cast<u64>(font.first)); + continue; } - bool using_ttf = false; - for (const char* font_ttf : SHARED_FONTS_TTF) { - if (FileUtil::Exists(user_path + font_ttf)) { - using_ttf = true; - FileUtil::IOFile file(user_path + font_ttf, "rb"); - if (file.IsOpen()) { - std::vector<u8> ttf_bytes(file.GetSize()); - file.ReadBytes<u8>(ttf_bytes.data(), ttf_bytes.size()); - FontRegion region{ - static_cast<u32>(offset + 8), - static_cast<u32>(ttf_bytes.size())}; // Font offset and size do not account - // for the header - EncryptSharedFont(ttf_bytes, *impl->shared_font, offset); - impl->shared_font_regions.push_back(region); - } else { - LOG_WARNING(Service_NS, "Unable to load font: {}", font_ttf); - } - } else if (using_ttf) { - LOG_WARNING(Service_NS, "Unable to find font: {}", font_ttf); - } + const auto extracted_romfs = FileSys::ExtractRomFS(romfs); + if (!extracted_romfs) { + LOG_ERROR(Service_NS, "Failed to extract RomFS for {:016X}! Skipping", + static_cast<u64>(font.first)); + continue; } - if (using_ttf) - return; - FileUtil::IOFile file(filepath, "rb"); - - if (file.IsOpen()) { - // Read shared font data - ASSERT(file.GetSize() == SHARED_FONT_MEM_SIZE); - file.ReadBytes(impl->shared_font->data(), impl->shared_font->size()); - impl->BuildSharedFontsRawRegions(*impl->shared_font); - } else { - LOG_WARNING(Service_NS, - "Shared Font file missing. Loading open source replacement from memory"); - - // clang-format off - const std::vector<std::vector<u8>> open_source_shared_fonts_ttf = { - {std::begin(FontChineseSimplified), std::end(FontChineseSimplified)}, - {std::begin(FontChineseTraditional), std::end(FontChineseTraditional)}, - {std::begin(FontExtendedChineseSimplified), std::end(FontExtendedChineseSimplified)}, - {std::begin(FontKorean), std::end(FontKorean)}, - {std::begin(FontNintendoExtended), std::end(FontNintendoExtended)}, - {std::begin(FontStandard), std::end(FontStandard)}, - }; - // clang-format on - - for (const std::vector<u8>& font_ttf : open_source_shared_fonts_ttf) { - const FontRegion region{static_cast<u32>(offset + 8), - static_cast<u32>(font_ttf.size())}; - EncryptSharedFont(font_ttf, *impl->shared_font, offset); - impl->shared_font_regions.push_back(region); - } + const auto font_fp = extracted_romfs->GetFile(font.second); + if (!font_fp) { + LOG_ERROR(Service_NS, "{:016X} has no file \"{}\"! Skipping", + static_cast<u64>(font.first), font.second); + continue; } + std::vector<u32> font_data_u32(font_fp->GetSize() / sizeof(u32)); + font_fp->ReadBytes<u32>(font_data_u32.data(), font_fp->GetSize()); + // We need to be BigEndian as u32s for the xor encryption + std::transform(font_data_u32.begin(), font_data_u32.end(), font_data_u32.begin(), + Common::swap32); + // Font offset and size do not account for the header + const FontRegion region{static_cast<u32>(offset + 8), + static_cast<u32>((font_data_u32.size() * sizeof(u32)) - 8)}; + DecryptSharedFont(font_data_u32, *impl->shared_font, offset); + impl->shared_font_regions.push_back(region); } } diff --git a/src/core/hle/service/ns/pl_u.h b/src/core/hle/service/ns/pl_u.h index 1063f4204..27161bd7a 100644 --- a/src/core/hle/service/ns/pl_u.h +++ b/src/core/hle/service/ns/pl_u.h @@ -5,6 +5,7 @@ #pragma once #include <memory> +#include <vector> #include "core/hle/service/service.h" namespace Service { @@ -15,6 +16,8 @@ class FileSystemController; namespace NS { +void EncryptSharedFont(const std::vector<u32>& input, std::vector<u8>& output, std::size_t& offset); + class PL_U final : public ServiceFramework<PL_U> { public: explicit PL_U(Core::System& system); diff --git a/src/core/hle/service/nvdrv/devices/nvdisp_disp0.cpp b/src/core/hle/service/nvdrv/devices/nvdisp_disp0.cpp index f764388bc..3f7b8e670 100644 --- a/src/core/hle/service/nvdrv/devices/nvdisp_disp0.cpp +++ b/src/core/hle/service/nvdrv/devices/nvdisp_disp0.cpp @@ -5,6 +5,7 @@ #include "common/assert.h" #include "common/logging/log.h" #include "core/core.h" +#include "core/core_timing.h" #include "core/hle/service/nvdrv/devices/nvdisp_disp0.h" #include "core/hle/service/nvdrv/devices/nvmap.h" #include "core/perf_stats.h" @@ -38,7 +39,10 @@ void nvdisp_disp0::flip(u32 buffer_handle, u32 offset, u32 format, u32 width, u3 transform, crop_rect}; system.GetPerfStats().EndGameFrame(); + system.GetPerfStats().EndSystemFrame(); system.GPU().SwapBuffers(&framebuffer); + system.FrameLimiter().DoFrameLimiting(system.CoreTiming().GetGlobalTimeUs()); + system.GetPerfStats().BeginSystemFrame(); } } // namespace Service::Nvidia::Devices diff --git a/src/core/hle/service/nvdrv/devices/nvhost_ctrl.cpp b/src/core/hle/service/nvdrv/devices/nvhost_ctrl.cpp index eb88fee1b..b27ee0502 100644 --- a/src/core/hle/service/nvdrv/devices/nvhost_ctrl.cpp +++ b/src/core/hle/service/nvdrv/devices/nvhost_ctrl.cpp @@ -63,16 +63,26 @@ u32 nvhost_ctrl::IocCtrlEventWait(const std::vector<u8>& input, std::vector<u8>& return NvResult::BadParameter; } + u32 event_id = params.value & 0x00FF; + + if (event_id >= MaxNvEvents) { + std::memcpy(output.data(), ¶ms, sizeof(params)); + return NvResult::BadParameter; + } + + auto event = events_interface.events[event_id]; auto& gpu = system.GPU(); // This is mostly to take into account unimplemented features. As synced // gpu is always synced. if (!gpu.IsAsync()) { + event.writable->Signal(); return NvResult::Success; } auto lock = gpu.LockSync(); const u32 current_syncpoint_value = gpu.GetSyncpointValue(params.syncpt_id); const s32 diff = current_syncpoint_value - params.threshold; if (diff >= 0) { + event.writable->Signal(); params.value = current_syncpoint_value; std::memcpy(output.data(), ¶ms, sizeof(params)); return NvResult::Success; @@ -88,27 +98,6 @@ u32 nvhost_ctrl::IocCtrlEventWait(const std::vector<u8>& input, std::vector<u8>& return NvResult::Timeout; } - u32 event_id; - if (is_async) { - event_id = params.value & 0x00FF; - if (event_id >= MaxNvEvents) { - std::memcpy(output.data(), ¶ms, sizeof(params)); - return NvResult::BadParameter; - } - } else { - if (ctrl.fresh_call) { - const auto result = events_interface.GetFreeEvent(); - if (result) { - event_id = *result; - } else { - LOG_CRITICAL(Service_NVDRV, "No Free Events available!"); - event_id = params.value & 0x00FF; - } - } else { - event_id = ctrl.event_id; - } - } - EventState status = events_interface.status[event_id]; if (event_id < MaxNvEvents || status == EventState::Free || status == EventState::Registered) { events_interface.SetEventStatus(event_id, EventState::Waiting); @@ -120,7 +109,7 @@ u32 nvhost_ctrl::IocCtrlEventWait(const std::vector<u8>& input, std::vector<u8>& params.value = ((params.syncpt_id & 0xfff) << 16) | 0x10000000; } params.value |= event_id; - events_interface.events[event_id].writable->Clear(); + event.writable->Clear(); gpu.RegisterSyncptInterrupt(params.syncpt_id, target_value); if (!is_async && ctrl.fresh_call) { ctrl.must_delay = true; diff --git a/src/core/hle/service/nvdrv/interface.cpp b/src/core/hle/service/nvdrv/interface.cpp index 5e0c23602..68d139cfb 100644 --- a/src/core/hle/service/nvdrv/interface.cpp +++ b/src/core/hle/service/nvdrv/interface.cpp @@ -134,7 +134,9 @@ void NVDRV::QueryEvent(Kernel::HLERequestContext& ctx) { IPC::ResponseBuilder rb{ctx, 3, 1}; rb.Push(RESULT_SUCCESS); if (event_id < MaxNvEvents) { - rb.PushCopyObjects(nvdrv->GetEvent(event_id)); + auto event = nvdrv->GetEvent(event_id); + event->Clear(); + rb.PushCopyObjects(event); rb.Push<u32>(NvResult::Success); } else { rb.Push<u32>(0); diff --git a/src/core/hle/service/nvdrv/nvdrv.cpp b/src/core/hle/service/nvdrv/nvdrv.cpp index 307a7e928..7bfb99e34 100644 --- a/src/core/hle/service/nvdrv/nvdrv.cpp +++ b/src/core/hle/service/nvdrv/nvdrv.cpp @@ -40,8 +40,8 @@ Module::Module(Core::System& system) { auto& kernel = system.Kernel(); for (u32 i = 0; i < MaxNvEvents; i++) { std::string event_label = fmt::format("NVDRV::NvEvent_{}", i); - events_interface.events[i] = Kernel::WritableEvent::CreateEventPair( - kernel, Kernel::ResetType::Automatic, event_label); + events_interface.events[i] = + Kernel::WritableEvent::CreateEventPair(kernel, Kernel::ResetType::Manual, event_label); events_interface.status[i] = EventState::Free; events_interface.registered[i] = false; } diff --git a/src/core/hle/service/nvflinger/buffer_queue.cpp b/src/core/hle/service/nvflinger/buffer_queue.cpp index e1a07d3ee..55b68eb0c 100644 --- a/src/core/hle/service/nvflinger/buffer_queue.cpp +++ b/src/core/hle/service/nvflinger/buffer_queue.cpp @@ -14,8 +14,8 @@ namespace Service::NVFlinger { -BufferQueue::BufferQueue(u32 id, u64 layer_id) : id(id), layer_id(layer_id) { - auto& kernel = Core::System::GetInstance().Kernel(); +BufferQueue::BufferQueue(Kernel::KernelCore& kernel, u32 id, u64 layer_id) + : id(id), layer_id(layer_id) { buffer_wait_event = Kernel::WritableEvent::CreateEventPair(kernel, Kernel::ResetType::Manual, "BufferQueue NativeHandle"); } diff --git a/src/core/hle/service/nvflinger/buffer_queue.h b/src/core/hle/service/nvflinger/buffer_queue.h index 356bedb81..8f9b18547 100644 --- a/src/core/hle/service/nvflinger/buffer_queue.h +++ b/src/core/hle/service/nvflinger/buffer_queue.h @@ -15,6 +15,10 @@ #include "core/hle/kernel/writable_event.h" #include "core/hle/service/nvdrv/nvdata.h" +namespace Kernel { +class KernelCore; +} + namespace Service::NVFlinger { struct IGBPBuffer { @@ -44,7 +48,7 @@ public: NativeWindowFormat = 2, }; - BufferQueue(u32 id, u64 layer_id); + explicit BufferQueue(Kernel::KernelCore& kernel, u32 id, u64 layer_id); ~BufferQueue(); enum class BufferTransformFlags : u32 { diff --git a/src/core/hle/service/nvflinger/nvflinger.cpp b/src/core/hle/service/nvflinger/nvflinger.cpp index 2e4d707b9..cc9522aad 100644 --- a/src/core/hle/service/nvflinger/nvflinger.cpp +++ b/src/core/hle/service/nvflinger/nvflinger.cpp @@ -83,7 +83,7 @@ std::optional<u64> NVFlinger::CreateLayer(u64 display_id) { const u64 layer_id = next_layer_id++; const u32 buffer_queue_id = next_buffer_queue_id++; - buffer_queues.emplace_back(buffer_queue_id, layer_id); + buffer_queues.emplace_back(system.Kernel(), buffer_queue_id, layer_id); display->CreateLayer(layer_id, buffer_queues.back()); return layer_id; } @@ -187,14 +187,18 @@ void NVFlinger::Compose() { MicroProfileFlip(); if (!buffer) { - // There was no queued buffer to draw, render previous frame - system.GetPerfStats().EndGameFrame(); - system.GPU().SwapBuffers({}); continue; } const auto& igbp_buffer = buffer->get().igbp_buffer; + const auto& gpu = system.GPU(); + const auto& multi_fence = buffer->get().multi_fence; + for (u32 fence_id = 0; fence_id < multi_fence.num_fences; fence_id++) { + const auto& fence = multi_fence.fences[fence_id]; + gpu.WaitFence(fence.id, fence.value); + } + // Now send the buffer to the GPU for drawing. // TODO(Subv): Support more than just disp0. The display device selection is probably based // on which display we're drawing (Default, Internal, External, etc) diff --git a/src/core/hle/service/service.cpp b/src/core/hle/service/service.cpp index f2c6fe9dc..7c5302017 100644 --- a/src/core/hle/service/service.cpp +++ b/src/core/hle/service/service.cpp @@ -226,7 +226,7 @@ void Init(std::shared_ptr<SM::ServiceManager>& sm, Core::System& system) { LBL::InstallInterfaces(*sm); LDN::InstallInterfaces(*sm); LDR::InstallInterfaces(*sm, system); - LM::InstallInterfaces(*sm); + LM::InstallInterfaces(system); Migration::InstallInterfaces(*sm); Mii::InstallInterfaces(*sm); MM::InstallInterfaces(*sm); diff --git a/src/core/hle/service/vi/vi.cpp b/src/core/hle/service/vi/vi.cpp index 199b30635..611cecc20 100644 --- a/src/core/hle/service/vi/vi.cpp +++ b/src/core/hle/service/vi/vi.cpp @@ -45,7 +45,7 @@ struct DisplayInfo { /// Whether or not the display has a limited number of layers. u8 has_limited_layers{1}; - INSERT_PADDING_BYTES(7){}; + INSERT_PADDING_BYTES(7); /// Indicates the total amount of layers supported by the display. /// @note This is only valid if has_limited_layers is set. |