1
0
Fork 0
mirror of https://github.com/Atmosphere-NX/Atmosphere.git synced 2024-11-23 12:22:08 +00:00
Atmosphere/libraries/libmesosphere/source/kern_k_thread.cpp
2021-04-11 03:42:16 -07:00

1357 lines
51 KiB
C++

/*
* Copyright (c) 2018-2020 Atmosphère-NX
*
* This program is free software; you can redistribute it and/or modify it
* under the terms and conditions of the GNU General Public License,
* version 2, as published by the Free Software Foundation.
*
* This program is distributed in the hope it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
* more details.
*
* You should have received a copy of the GNU General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include <mesosphere.hpp>
namespace ams::kern {
namespace {
constexpr inline s32 TerminatingThreadPriority = ams::svc::SystemThreadPriorityHighest - 1;
constexpr bool IsKernelAddressKey(KProcessAddress key) {
const uintptr_t key_uptr = GetInteger(key);
return KernelVirtualAddressSpaceBase <= key_uptr && key_uptr <= KernelVirtualAddressSpaceLast && (key_uptr & 1) == 0;
}
void InitializeKernelStack(uintptr_t stack_top) {
#if defined(MESOSPHERE_ENABLE_KERNEL_STACK_USAGE)
const uintptr_t stack_bottom = stack_top - PageSize;
std::memset(reinterpret_cast<void *>(stack_bottom), 0xCC, PageSize - sizeof(KThread::StackParameters));
#else
MESOSPHERE_UNUSED(stack_top);
#endif
}
void CleanupKernelStack(uintptr_t stack_top) {
const uintptr_t stack_bottom = stack_top - PageSize;
KPhysicalAddress stack_paddr = Null<KPhysicalAddress>;
MESOSPHERE_ABORT_UNLESS(Kernel::GetKernelPageTable().GetPhysicalAddress(&stack_paddr, stack_bottom));
MESOSPHERE_R_ABORT_UNLESS(Kernel::GetKernelPageTable().UnmapPages(stack_bottom, 1, KMemoryState_Kernel));
/* Free the stack page. */
KPageBuffer::Free(KPageBuffer::FromPhysicalAddress(stack_paddr));
}
}
Result KThread::Initialize(KThreadFunction func, uintptr_t arg, void *kern_stack_top, KProcessAddress user_stack_top, s32 prio, s32 virt_core, KProcess *owner, ThreadType type) {
/* Assert parameters are valid. */
MESOSPHERE_ASSERT_THIS();
MESOSPHERE_ASSERT(kern_stack_top != nullptr);
MESOSPHERE_ASSERT((type == ThreadType_Main) || (ams::svc::HighestThreadPriority <= prio && prio <= ams::svc::LowestThreadPriority));
MESOSPHERE_ASSERT((owner != nullptr) || (type != ThreadType_User));
MESOSPHERE_ASSERT(0 <= virt_core && virt_core < static_cast<s32>(BITSIZEOF(u64)));
/* Convert the virtual core to a physical core. */
const s32 phys_core = cpu::VirtualToPhysicalCoreMap[virt_core];
MESOSPHERE_ASSERT(0 <= phys_core && phys_core < static_cast<s32>(cpu::NumCores));
/* First, clear the TLS address. */
m_tls_address = Null<KProcessAddress>;
const uintptr_t kern_stack_top_address = reinterpret_cast<uintptr_t>(kern_stack_top);
MESOSPHERE_UNUSED(kern_stack_top_address);
/* Next, assert things based on the type. */
switch (type) {
case ThreadType_Main:
{
MESOSPHERE_ASSERT(arg == 0);
}
[[fallthrough]];
case ThreadType_HighPriority:
if (type != ThreadType_Main) {
MESOSPHERE_ASSERT(phys_core == GetCurrentCoreId());
}
[[fallthrough]];
case ThreadType_Kernel:
{
MESOSPHERE_ASSERT(user_stack_top == 0);
MESOSPHERE_ASSERT(util::IsAligned(kern_stack_top_address, PageSize));
}
[[fallthrough]];
case ThreadType_User:
{
MESOSPHERE_ASSERT(((owner == nullptr) || (owner->GetCoreMask() | (1ul << virt_core)) == owner->GetCoreMask()));
MESOSPHERE_ASSERT(((owner == nullptr) || (owner->GetPriorityMask() | (1ul << prio)) == owner->GetPriorityMask()));
}
break;
default:
MESOSPHERE_PANIC("KThread::Initialize: Unknown ThreadType %u", static_cast<u32>(type));
break;
}
/* Set the ideal core ID and affinity mask. */
m_virtual_ideal_core_id = virt_core;
m_physical_ideal_core_id = phys_core;
m_virtual_affinity_mask = (static_cast<u64>(1) << virt_core);
m_physical_affinity_mask.SetAffinity(phys_core, true);
/* Set the thread state. */
m_thread_state = (type == ThreadType_Main) ? ThreadState_Runnable : ThreadState_Initialized;
/* Set TLS address and TLS heap address. */
/* NOTE: Nintendo wrote TLS address above already, but official code really does write tls address twice. */
m_tls_address = 0;
m_tls_heap_address = 0;
/* Set parent and condvar tree. */
m_parent = nullptr;
m_condvar_tree = nullptr;
/* Set sync booleans. */
m_signaled = false;
m_termination_requested = false;
m_wait_cancelled = false;
m_cancellable = false;
/* Set core ID and wait result. */
m_core_id = phys_core;
m_wait_result = svc::ResultNoSynchronizationObject();
/* Set the stack top. */
m_kernel_stack_top = kern_stack_top;
/* Set priorities. */
m_priority = prio;
m_base_priority = prio;
/* Set sync object and waiting lock to null. */
m_synced_object = nullptr;
m_waiting_lock = nullptr;
/* Initialize sleeping queue. */
m_sleeping_queue = nullptr;
/* Set suspend flags. */
m_suspend_request_flags = 0;
m_suspend_allowed_flags = ThreadState_SuspendFlagMask;
/* We're neither debug attached, nor are we nesting our priority inheritance. */
m_debug_attached = false;
m_priority_inheritance_count = 0;
/* We haven't been scheduled, and we have done no light IPC. */
m_schedule_count = -1;
m_last_scheduled_tick = 0;
m_light_ipc_data = nullptr;
/* We're not waiting for a lock, and we haven't disabled migration. */
m_lock_owner = nullptr;
m_num_core_migration_disables = 0;
/* We have no waiters, but we do have an entrypoint. */
m_num_kernel_waiters = 0;
/* Set our current core id. */
m_current_core_id = phys_core;
/* We haven't released our resource limit hint, and we've spent no time on the cpu. */
m_resource_limit_release_hint = 0;
m_cpu_time = 0;
/* Setup our kernel stack. */
if (type != ThreadType_Main) {
InitializeKernelStack(reinterpret_cast<uintptr_t>(kern_stack_top));
}
/* Clear our stack parameters. */
std::memset(static_cast<void *>(std::addressof(this->GetStackParameters())), 0, sizeof(StackParameters));
/* Setup the TLS, if needed. */
if (type == ThreadType_User) {
R_TRY(owner->CreateThreadLocalRegion(std::addressof(m_tls_address)));
m_tls_heap_address = owner->GetThreadLocalRegionPointer(m_tls_address);
std::memset(m_tls_heap_address, 0, ams::svc::ThreadLocalRegionSize);
}
/* Set parent, if relevant. */
if (owner != nullptr) {
m_parent = owner;
m_parent->Open();
m_parent->IncrementThreadCount();
}
/* Initialize thread context. */
constexpr bool IsDefault64Bit = sizeof(uintptr_t) == sizeof(u64);
const bool is_64_bit = m_parent ? m_parent->Is64Bit() : IsDefault64Bit;
const bool is_user = (type == ThreadType_User);
const bool is_main = (type == ThreadType_Main);
m_thread_context.Initialize(reinterpret_cast<uintptr_t>(func), reinterpret_cast<uintptr_t>(this->GetStackTop()), GetInteger(user_stack_top), arg, is_user, is_64_bit, is_main);
/* Setup the stack parameters. */
StackParameters &sp = this->GetStackParameters();
if (m_parent != nullptr) {
m_parent->CopySvcPermissionsTo(sp);
}
sp.context = std::addressof(m_thread_context);
sp.cur_thread = this;
sp.disable_count = 1;
this->SetInExceptionHandler();
/* Set thread ID. */
m_thread_id = s_next_thread_id++;
/* We initialized! */
m_initialized = true;
/* Register ourselves with our parent process. */
if (m_parent != nullptr) {
m_parent->RegisterThread(this);
if (m_parent->IsSuspended()) {
this->RequestSuspend(SuspendType_Process);
}
}
return ResultSuccess();
}
Result KThread::InitializeThread(KThread *thread, KThreadFunction func, uintptr_t arg, KProcessAddress user_stack_top, s32 prio, s32 core, KProcess *owner, ThreadType type) {
/* Get stack region for the thread. */
const auto &stack_region = KMemoryLayout::GetKernelStackRegion();
MESOSPHERE_ABORT_UNLESS(stack_region.GetEndAddress() != 0);
/* Allocate a page to use as the thread. */
KPageBuffer *page = KPageBuffer::Allocate();
R_UNLESS(page != nullptr, svc::ResultOutOfResource());
/* Map the stack page. */
KProcessAddress stack_top = Null<KProcessAddress>;
{
KProcessAddress stack_bottom = Null<KProcessAddress>;
auto page_guard = SCOPE_GUARD { KPageBuffer::Free(page); };
R_TRY(Kernel::GetKernelPageTable().MapPages(std::addressof(stack_bottom), 1, PageSize, page->GetPhysicalAddress(), stack_region.GetAddress(),
stack_region.GetSize() / PageSize, KMemoryState_Kernel, KMemoryPermission_KernelReadWrite));
page_guard.Cancel();
/* Calculate top of the stack. */
stack_top = stack_bottom + PageSize;
}
/* Initialize the thread. */
auto map_guard = SCOPE_GUARD { CleanupKernelStack(GetInteger(stack_top)); };
R_TRY(thread->Initialize(func, arg, GetVoidPointer(stack_top), user_stack_top, prio, core, owner, type));
map_guard.Cancel();
return ResultSuccess();
}
void KThread::PostDestroy(uintptr_t arg) {
KProcess *owner = reinterpret_cast<KProcess *>(arg & ~1ul);
const bool resource_limit_release_hint = (arg & 1);
const s64 hint_value = (resource_limit_release_hint ? 0 : 1);
if (owner != nullptr) {
owner->ReleaseResource(ams::svc::LimitableResource_ThreadCountMax, 1, hint_value);
owner->Close();
} else {
Kernel::GetSystemResourceLimit().Release(ams::svc::LimitableResource_ThreadCountMax, 1, hint_value);
}
}
void KThread::ResumeThreadsSuspendedForInit() {
KThread::ListAccessor list_accessor;
{
KScopedSchedulerLock sl;
for (auto &thread : list_accessor) {
static_cast<KThread &>(thread).Resume(SuspendType_Init);
}
}
}
void KThread::Finalize() {
MESOSPHERE_ASSERT_THIS();
/* If the thread has an owner process, unregister it. */
if (m_parent != nullptr) {
m_parent->UnregisterThread(this);
}
/* If the thread has a local region, delete it. */
if (m_tls_address != Null<KProcessAddress>) {
MESOSPHERE_R_ABORT_UNLESS(m_parent->DeleteThreadLocalRegion(m_tls_address));
}
/* Release any waiters. */
{
MESOSPHERE_ASSERT(m_lock_owner == nullptr);
KScopedSchedulerLock sl;
auto it = m_waiter_list.begin();
while (it != m_waiter_list.end()) {
/* The thread shouldn't be a kernel waiter. */
MESOSPHERE_ASSERT(!IsKernelAddressKey(it->GetAddressKey()));
it->SetLockOwner(nullptr);
it->SetSyncedObject(nullptr, svc::ResultInvalidState());
it->Wakeup();
it = m_waiter_list.erase(it);
}
}
/* Finalize the thread context. */
m_thread_context.Finalize();
/* Cleanup the kernel stack. */
if (m_kernel_stack_top != nullptr) {
CleanupKernelStack(reinterpret_cast<uintptr_t>(m_kernel_stack_top));
}
/* Decrement the parent process's thread count. */
if (m_parent != nullptr) {
m_parent->DecrementThreadCount();
}
/* Perform inherited finalization. */
KAutoObjectWithSlabHeapAndContainer<KThread, KSynchronizationObject>::Finalize();
}
bool KThread::IsSignaled() const {
return m_signaled;
}
void KThread::Wakeup() {
MESOSPHERE_ASSERT_THIS();
KScopedSchedulerLock sl;
if (this->GetState() == ThreadState_Waiting) {
if (m_sleeping_queue != nullptr) {
m_sleeping_queue->WakeupThread(this);
} else {
this->SetState(ThreadState_Runnable);
}
}
}
void KThread::OnTimer() {
MESOSPHERE_ASSERT_THIS();
MESOSPHERE_ASSERT(KScheduler::IsSchedulerLockedByCurrentThread());
this->Wakeup();
}
void KThread::StartTermination() {
MESOSPHERE_ASSERT_THIS();
MESOSPHERE_ASSERT(KScheduler::IsSchedulerLockedByCurrentThread());
/* Release user exception and unpin, if relevant. */
if (m_parent != nullptr) {
m_parent->ReleaseUserException(this);
if (m_parent->GetPinnedThread(GetCurrentCoreId()) == this) {
m_parent->UnpinCurrentThread();
}
}
/* Set state to terminated. */
this->SetState(KThread::ThreadState_Terminated);
/* Clear the thread's status as running in parent. */
if (m_parent != nullptr) {
m_parent->ClearRunningThread(this);
}
/* Signal. */
m_signaled = true;
this->NotifyAvailable();
/* Call the on thread termination handler. */
KThreadContext::OnThreadTerminating(this);
/* Clear previous thread in KScheduler. */
KScheduler::ClearPreviousThread(this);
/* Register terminated dpc flag. */
this->RegisterDpc(DpcFlag_Terminated);
}
void KThread::FinishTermination() {
MESOSPHERE_ASSERT_THIS();
/* Ensure that the thread is not executing on any core. */
if (m_parent != nullptr) {
for (size_t i = 0; i < cpu::NumCores; ++i) {
KThread *core_thread;
do {
core_thread = Kernel::GetScheduler(i).GetSchedulerCurrentThread();
} while (core_thread == this);
}
}
/* Close the thread. */
this->Close();
}
void KThread::DoWorkerTask() {
/* Finish the termination that was begun by Exit(). */
this->FinishTermination();
}
void KThread::Pin() {
MESOSPHERE_ASSERT_THIS();
MESOSPHERE_ASSERT(KScheduler::IsSchedulerLockedByCurrentThread());
/* Set ourselves as pinned. */
this->GetStackParameters().is_pinned = true;
/* Disable core migration. */
MESOSPHERE_ASSERT(m_num_core_migration_disables == 0);
{
++m_num_core_migration_disables;
/* Save our ideal state to restore when we're unpinned. */
m_original_physical_ideal_core_id = m_physical_ideal_core_id;
m_original_physical_affinity_mask = m_physical_affinity_mask;
/* Bind ourselves to this core. */
const s32 active_core = this->GetActiveCore();
const s32 current_core = GetCurrentCoreId();
this->SetActiveCore(current_core);
m_physical_ideal_core_id = current_core;
m_physical_affinity_mask.SetAffinityMask(1ul << current_core);
if (active_core != current_core || m_physical_affinity_mask.GetAffinityMask() != m_original_physical_affinity_mask.GetAffinityMask()) {
KScheduler::OnThreadAffinityMaskChanged(this, m_original_physical_affinity_mask, active_core);
}
/* Set base priority-on-unpin. */
const s32 old_base_priority = m_base_priority;
m_base_priority_on_unpin = old_base_priority;
/* Set base priority to higher than any possible process priority. */
m_base_priority = std::min<s32>(old_base_priority, __builtin_ctzll(this->GetOwnerProcess()->GetPriorityMask()));
RestorePriority(this);
}
/* Disallow performing thread suspension. */
{
/* Update our allow flags. */
m_suspend_allowed_flags &= ~(1 << (SuspendType_Thread + ThreadState_SuspendShift));
/* Update our state. */
const ThreadState old_state = m_thread_state;
m_thread_state = static_cast<ThreadState>(this->GetSuspendFlags() | (old_state & ThreadState_Mask));
if (m_thread_state != old_state) {
KScheduler::OnThreadStateChanged(this, old_state);
}
}
/* Update our SVC access permissions. */
MESOSPHERE_ASSERT(m_parent != nullptr);
m_parent->CopyPinnedSvcPermissionsTo(this->GetStackParameters());
}
void KThread::Unpin() {
MESOSPHERE_ASSERT_THIS();
MESOSPHERE_ASSERT(KScheduler::IsSchedulerLockedByCurrentThread());
/* Set ourselves as unpinned. */
this->GetStackParameters().is_pinned = false;
/* Enable core migration. */
MESOSPHERE_ASSERT(m_num_core_migration_disables == 1);
{
--m_num_core_migration_disables;
/* Restore our original state. */
const KAffinityMask old_mask = m_physical_affinity_mask;
m_physical_ideal_core_id = m_original_physical_ideal_core_id;
m_physical_affinity_mask = m_original_physical_affinity_mask;
if (m_physical_affinity_mask.GetAffinityMask() != old_mask.GetAffinityMask()) {
const s32 active_core = this->GetActiveCore();
if (!m_physical_affinity_mask.GetAffinity(active_core)) {
if (m_physical_ideal_core_id >= 0) {
this->SetActiveCore(m_physical_ideal_core_id);
} else {
this->SetActiveCore(BITSIZEOF(unsigned long long) - 1 - __builtin_clzll(m_physical_affinity_mask.GetAffinityMask()));
}
}
KScheduler::OnThreadAffinityMaskChanged(this, old_mask, active_core);
}
m_base_priority = m_base_priority_on_unpin;
RestorePriority(this);
}
/* Allow performing thread suspension (if termination hasn't been requested). */
{
/* Update our allow flags. */
if (!this->IsTerminationRequested()) {
m_suspend_allowed_flags |= (1 << (SuspendType_Thread + ThreadState_SuspendShift));
}
/* Update our state. */
const ThreadState old_state = m_thread_state;
m_thread_state = static_cast<ThreadState>(this->GetSuspendFlags() | (old_state & ThreadState_Mask));
if (m_thread_state != old_state) {
KScheduler::OnThreadStateChanged(this, old_state);
}
}
/* Update our SVC access permissions. */
MESOSPHERE_ASSERT(m_parent != nullptr);
m_parent->CopyUnpinnedSvcPermissionsTo(this->GetStackParameters());
/* Resume any threads that began waiting on us while we were pinned. */
for (auto it = m_pinned_waiter_list.begin(); it != m_pinned_waiter_list.end(); ++it) {
if (it->GetState() == ThreadState_Waiting) {
it->SetState(ThreadState_Runnable);
}
}
}
void KThread::DisableCoreMigration() {
MESOSPHERE_ASSERT_THIS();
MESOSPHERE_ASSERT(this == GetCurrentThreadPointer());
KScopedSchedulerLock sl;
MESOSPHERE_ASSERT(m_num_core_migration_disables >= 0);
if ((m_num_core_migration_disables++) == 0) {
/* Save our ideal state to restore when we can migrate again. */
m_original_physical_ideal_core_id = m_physical_ideal_core_id;
m_original_physical_affinity_mask = m_physical_affinity_mask;
/* Bind ourselves to this core. */
const s32 active_core = this->GetActiveCore();
m_physical_ideal_core_id = active_core;
m_physical_affinity_mask.SetAffinityMask(1ul << active_core);
if (m_physical_affinity_mask.GetAffinityMask() != m_original_physical_affinity_mask.GetAffinityMask()) {
KScheduler::OnThreadAffinityMaskChanged(this, m_original_physical_affinity_mask, active_core);
}
}
}
void KThread::EnableCoreMigration() {
MESOSPHERE_ASSERT_THIS();
MESOSPHERE_ASSERT(this == GetCurrentThreadPointer());
KScopedSchedulerLock sl;
MESOSPHERE_ASSERT(m_num_core_migration_disables > 0);
if ((--m_num_core_migration_disables) == 0) {
const KAffinityMask old_mask = m_physical_affinity_mask;
/* Restore our ideals. */
m_physical_ideal_core_id = m_original_physical_ideal_core_id;
m_physical_affinity_mask = m_original_physical_affinity_mask;
if (m_physical_affinity_mask.GetAffinityMask() != old_mask.GetAffinityMask()) {
const s32 active_core = this->GetActiveCore();
if (!m_physical_affinity_mask.GetAffinity(active_core)) {
if (m_physical_ideal_core_id >= 0) {
this->SetActiveCore(m_physical_ideal_core_id);
} else {
this->SetActiveCore(BITSIZEOF(unsigned long long) - 1 - __builtin_clzll(m_physical_affinity_mask.GetAffinityMask()));
}
}
KScheduler::OnThreadAffinityMaskChanged(this, old_mask, active_core);
}
}
}
Result KThread::GetCoreMask(int32_t *out_ideal_core, u64 *out_affinity_mask) {
MESOSPHERE_ASSERT_THIS();
{
KScopedSchedulerLock sl;
/* Get the virtual mask. */
*out_ideal_core = m_virtual_ideal_core_id;
*out_affinity_mask = m_virtual_affinity_mask;
}
return ResultSuccess();
}
Result KThread::GetPhysicalCoreMask(int32_t *out_ideal_core, u64 *out_affinity_mask) {
MESOSPHERE_ASSERT_THIS();
{
KScopedSchedulerLock sl;
MESOSPHERE_ASSERT(m_num_core_migration_disables >= 0);
/* Select between core mask and original core mask. */
if (m_num_core_migration_disables == 0) {
*out_ideal_core = m_physical_ideal_core_id;
*out_affinity_mask = m_physical_affinity_mask.GetAffinityMask();
} else {
*out_ideal_core = m_original_physical_ideal_core_id;
*out_affinity_mask = m_original_physical_affinity_mask.GetAffinityMask();
}
}
return ResultSuccess();
}
Result KThread::SetCoreMask(int32_t core_id, u64 v_affinity_mask) {
MESOSPHERE_ASSERT_THIS();
MESOSPHERE_ASSERT(m_parent != nullptr);
MESOSPHERE_ASSERT(v_affinity_mask != 0);
KScopedLightLock lk(m_activity_pause_lock);
/* Set the core mask. */
u64 p_affinity_mask = 0;
{
KScopedSchedulerLock sl;
MESOSPHERE_ASSERT(m_num_core_migration_disables >= 0);
/* If we're updating, set our ideal virtual core. */
if (core_id != ams::svc::IdealCoreNoUpdate) {
m_virtual_ideal_core_id = core_id;
} else {
/* Preserve our ideal core id. */
core_id = m_virtual_ideal_core_id;
R_UNLESS(((1ul << core_id) & v_affinity_mask) != 0, svc::ResultInvalidCombination());
}
/* Set our affinity mask. */
m_virtual_affinity_mask = v_affinity_mask;
/* Translate the virtual core to a physical core. */
if (core_id >= 0) {
core_id = cpu::VirtualToPhysicalCoreMap[core_id];
}
/* Translate the virtual affinity mask to a physical one. */
while (v_affinity_mask != 0) {
const u64 next = __builtin_ctzll(v_affinity_mask);
v_affinity_mask &= ~(1ul << next);
p_affinity_mask |= (1ul << cpu::VirtualToPhysicalCoreMap[next]);
}
/* If we haven't disabled migration, perform an affinity change. */
if (m_num_core_migration_disables == 0) {
const KAffinityMask old_mask = m_physical_affinity_mask;
/* Set our new ideals. */
m_physical_ideal_core_id = core_id;
m_physical_affinity_mask.SetAffinityMask(p_affinity_mask);
if (m_physical_affinity_mask.GetAffinityMask() != old_mask.GetAffinityMask()) {
const s32 active_core = this->GetActiveCore();
if (active_core >= 0 && !m_physical_affinity_mask.GetAffinity(active_core)) {
const s32 new_core = m_physical_ideal_core_id >= 0 ? m_physical_ideal_core_id : BITSIZEOF(unsigned long long) - 1 - __builtin_clzll(m_physical_affinity_mask.GetAffinityMask());
this->SetActiveCore(new_core);
}
KScheduler::OnThreadAffinityMaskChanged(this, old_mask, active_core);
}
} else {
/* Otherwise, we edit the original affinity for restoration later. */
m_original_physical_ideal_core_id = core_id;
m_original_physical_affinity_mask.SetAffinityMask(p_affinity_mask);
}
}
/* Update the pinned waiter list. */
{
bool retry_update;
bool thread_is_pinned = false;
do {
/* Lock the scheduler. */
KScopedSchedulerLock sl;
/* Don't do any further management if our termination has been requested. */
R_SUCCEED_IF(this->IsTerminationRequested());
/* By default, we won't need to retry. */
retry_update = false;
/* Check if the thread is currently running. */
bool thread_is_current = false;
s32 thread_core;
for (thread_core = 0; thread_core < static_cast<s32>(cpu::NumCores); ++thread_core) {
if (Kernel::GetScheduler(thread_core).GetSchedulerCurrentThread() == this) {
thread_is_current = true;
break;
}
}
/* If the thread is currently running, check whether it's no longer allowed under the new mask. */
if (thread_is_current && ((1ul << thread_core) & p_affinity_mask) == 0) {
/* If the thread is pinned, we want to wait until it's not pinned. */
if (this->GetStackParameters().is_pinned) {
/* Verify that the current thread isn't terminating. */
R_UNLESS(!GetCurrentThread().IsTerminationRequested(), svc::ResultTerminationRequested());
/* Note that the thread was pinned. */
thread_is_pinned = true;
/* Wait until the thread isn't pinned any more. */
m_pinned_waiter_list.push_back(GetCurrentThread());
GetCurrentThread().SetState(ThreadState_Waiting);
} else {
/* If the thread isn't pinned, release the scheduler lock and retry until it's not current. */
retry_update = true;
}
}
} while (retry_update);
/* If the thread was pinned, it no longer is, and we should remove the current thread from our waiter list. */
if (thread_is_pinned) {
/* Lock the scheduler. */
KScopedSchedulerLock sl;
/* Remove from the list. */
m_pinned_waiter_list.erase(m_pinned_waiter_list.iterator_to(GetCurrentThread()));
}
}
return ResultSuccess();
}
void KThread::SetBasePriority(s32 priority) {
MESOSPHERE_ASSERT_THIS();
MESOSPHERE_ASSERT(ams::svc::HighestThreadPriority <= priority && priority <= ams::svc::LowestThreadPriority);
KScopedSchedulerLock sl;
/* Determine the priority value to use. */
const s32 target_priority = m_termination_requested.load() && priority >= TerminatingThreadPriority ? TerminatingThreadPriority : priority;
/* Change our base priority. */
if (this->GetStackParameters().is_pinned) {
m_base_priority_on_unpin = target_priority;
} else {
m_base_priority = target_priority;
}
/* Perform a priority restoration. */
RestorePriority(this);
}
void KThread::IncreaseBasePriority(s32 priority) {
MESOSPHERE_ASSERT_THIS();
MESOSPHERE_ASSERT(ams::svc::HighestThreadPriority <= priority && priority <= ams::svc::LowestThreadPriority);
/* Set our unpin base priority, if we're pinned. */
if (this->GetStackParameters().is_pinned && m_base_priority_on_unpin > priority) {
m_base_priority_on_unpin = priority;
}
/* Set our base priority. */
if (m_base_priority > priority) {
m_base_priority = priority;
/* Perform a priority restoration. */
RestorePriority(this);
}
}
Result KThread::SetPriorityToIdle() {
MESOSPHERE_ASSERT_THIS();
KScopedSchedulerLock sl;
/* Change both our priorities to the idle thread priority. */
const s32 old_priority = m_priority;
m_priority = IdleThreadPriority;
m_base_priority = IdleThreadPriority;
KScheduler::OnThreadPriorityChanged(this, old_priority);
return ResultSuccess();
}
void KThread::RequestSuspend(SuspendType type) {
MESOSPHERE_ASSERT_THIS();
KScopedSchedulerLock lk;
/* Note the request in our flags. */
m_suspend_request_flags |= (1u << (ThreadState_SuspendShift + type));
/* Try to perform the suspend. */
this->TrySuspend();
}
void KThread::Resume(SuspendType type) {
MESOSPHERE_ASSERT_THIS();
KScopedSchedulerLock sl;
/* Clear the request in our flags. */
m_suspend_request_flags &= ~(1u << (ThreadState_SuspendShift + type));
/* Update our state. */
const ThreadState old_state = m_thread_state;
m_thread_state = static_cast<ThreadState>(this->GetSuspendFlags() | (old_state & ThreadState_Mask));
if (m_thread_state != old_state) {
KScheduler::OnThreadStateChanged(this, old_state);
}
}
void KThread::WaitCancel() {
MESOSPHERE_ASSERT_THIS();
KScopedSchedulerLock sl;
/* Check if we're waiting and cancellable. */
if (this->GetState() == ThreadState_Waiting && m_cancellable) {
if (m_sleeping_queue != nullptr) {
m_sleeping_queue->WakeupThread(this);
m_wait_cancelled = true;
} else {
this->SetSyncedObject(nullptr, svc::ResultCancelled());
this->SetState(ThreadState_Runnable);
m_wait_cancelled = false;
}
} else {
/* Otherwise, note that we cancelled a wait. */
m_wait_cancelled = true;
}
}
void KThread::TrySuspend() {
MESOSPHERE_ASSERT_THIS();
MESOSPHERE_ASSERT(KScheduler::IsSchedulerLockedByCurrentThread());
MESOSPHERE_ASSERT(this->IsSuspendRequested());
/* Ensure that we have no waiters. */
if (this->GetNumKernelWaiters() > 0) {
return;
}
MESOSPHERE_ABORT_UNLESS(this->GetNumKernelWaiters() == 0);
/* Perform the suspend. */
this->Suspend();
}
void KThread::Suspend() {
MESOSPHERE_ASSERT_THIS();
MESOSPHERE_ASSERT(KScheduler::IsSchedulerLockedByCurrentThread());
MESOSPHERE_ASSERT(this->IsSuspendRequested());
/* Set our suspend flags in state. */
const auto old_state = m_thread_state;
m_thread_state = static_cast<ThreadState>(this->GetSuspendFlags() | (old_state & ThreadState_Mask));
/* Note the state change in scheduler. */
KScheduler::OnThreadStateChanged(this, old_state);
}
void KThread::Continue() {
MESOSPHERE_ASSERT_THIS();
MESOSPHERE_ASSERT(KScheduler::IsSchedulerLockedByCurrentThread());
/* Clear our suspend flags in state. */
const auto old_state = m_thread_state;
m_thread_state = static_cast<ThreadState>(old_state & ThreadState_Mask);
/* Note the state change in scheduler. */
KScheduler::OnThreadStateChanged(this, old_state);
}
size_t KThread::GetKernelStackUsage() const {
MESOSPHERE_ASSERT_THIS();
MESOSPHERE_ASSERT(m_kernel_stack_top != nullptr);
#if defined(MESOSPHERE_ENABLE_KERNEL_STACK_USAGE)
const u8 *stack = static_cast<const u8 *>(m_kernel_stack_top) - PageSize;
size_t i;
for (i = 0; i < PageSize; ++i) {
if (stack[i] != 0xCC) {
break;
}
}
return PageSize - i;
#else
return 0;
#endif
}
Result KThread::SetActivity(ams::svc::ThreadActivity activity) {
/* Lock ourselves. */
KScopedLightLock lk(m_activity_pause_lock);
/* Set the activity. */
{
/* Lock the scheduler. */
KScopedSchedulerLock sl;
/* Verify our state. */
const auto cur_state = this->GetState();
R_UNLESS((cur_state == ThreadState_Waiting || cur_state == ThreadState_Runnable), svc::ResultInvalidState());
/* Either pause or resume. */
if (activity == ams::svc::ThreadActivity_Paused) {
/* Verify that we're not suspended. */
R_UNLESS(!this->IsSuspendRequested(SuspendType_Thread), svc::ResultInvalidState());
/* Suspend. */
this->RequestSuspend(SuspendType_Thread);
} else {
MESOSPHERE_ASSERT(activity == ams::svc::ThreadActivity_Runnable);
/* Verify that we're suspended. */
R_UNLESS(this->IsSuspendRequested(SuspendType_Thread), svc::ResultInvalidState());
/* Resume. */
this->Resume(SuspendType_Thread);
}
}
/* If the thread is now paused, update the pinned waiter list. */
if (activity == ams::svc::ThreadActivity_Paused) {
bool thread_is_pinned = false;
bool thread_is_current;
do {
/* Lock the scheduler. */
KScopedSchedulerLock sl;
/* Don't do any further management if our termination has been requested. */
R_SUCCEED_IF(this->IsTerminationRequested());
/* Check whether the thread is pinned. */
if (this->GetStackParameters().is_pinned) {
/* Verify that the current thread isn't terminating. */
R_UNLESS(!GetCurrentThread().IsTerminationRequested(), svc::ResultTerminationRequested());
/* Note that the thread was pinned and not current. */
thread_is_pinned = true;
thread_is_current = false;
/* Wait until the thread isn't pinned any more. */
m_pinned_waiter_list.push_back(GetCurrentThread());
GetCurrentThread().SetState(ThreadState_Waiting);
} else {
/* Check if the thread is currently running. */
/* If it is, we'll need to retry. */
thread_is_current = false;
for (auto i = 0; i < static_cast<s32>(cpu::NumCores); ++i) {
if (Kernel::GetScheduler(i).GetSchedulerCurrentThread() == this) {
thread_is_current = true;
break;
}
}
}
} while (thread_is_current);
/* If the thread was pinned, it no longer is, and we should remove the current thread from our waiter list. */
if (thread_is_pinned) {
/* Lock the scheduler. */
KScopedSchedulerLock sl;
/* Remove from the list. */
m_pinned_waiter_list.erase(m_pinned_waiter_list.iterator_to(GetCurrentThread()));
}
}
return ResultSuccess();
}
Result KThread::GetThreadContext3(ams::svc::ThreadContext *out) {
/* Lock ourselves. */
KScopedLightLock lk(m_activity_pause_lock);
/* Get the context. */
{
/* Lock the scheduler. */
KScopedSchedulerLock sl;
/* Verify that we're suspended. */
R_UNLESS(this->IsSuspendRequested(SuspendType_Thread), svc::ResultInvalidState());
/* If we're not terminating, get the thread's user context. */
if (!this->IsTerminationRequested()) {
GetUserContext(out, this);
}
}
return ResultSuccess();
}
void KThread::AddWaiterImpl(KThread *thread) {
MESOSPHERE_ASSERT_THIS();
MESOSPHERE_ASSERT(KScheduler::IsSchedulerLockedByCurrentThread());
/* Find the right spot to insert the waiter. */
auto it = m_waiter_list.begin();
while (it != m_waiter_list.end()) {
if (it->GetPriority() > thread->GetPriority()) {
break;
}
it++;
}
/* Keep track of how many kernel waiters we have. */
if (IsKernelAddressKey(thread->GetAddressKey())) {
MESOSPHERE_ABORT_UNLESS((m_num_kernel_waiters++) >= 0);
KScheduler::SetSchedulerUpdateNeeded();
}
/* Insert the waiter. */
m_waiter_list.insert(it, *thread);
thread->SetLockOwner(this);
}
void KThread::RemoveWaiterImpl(KThread *thread) {
MESOSPHERE_ASSERT_THIS();
MESOSPHERE_ASSERT(KScheduler::IsSchedulerLockedByCurrentThread());
/* Keep track of how many kernel waiters we have. */
if (IsKernelAddressKey(thread->GetAddressKey())) {
MESOSPHERE_ABORT_UNLESS((m_num_kernel_waiters--) > 0);
KScheduler::SetSchedulerUpdateNeeded();
}
/* Remove the waiter. */
m_waiter_list.erase(m_waiter_list.iterator_to(*thread));
thread->SetLockOwner(nullptr);
}
void KThread::RestorePriority(KThread *thread) {
MESOSPHERE_ASSERT(KScheduler::IsSchedulerLockedByCurrentThread());
while (true) {
/* We want to inherit priority where possible. */
s32 new_priority = thread->GetBasePriority();
if (thread->HasWaiters()) {
new_priority = std::min(new_priority, thread->m_waiter_list.front().GetPriority());
}
/* If the priority we would inherit is not different from ours, don't do anything. */
if (new_priority == thread->GetPriority()) {
return;
}
/* Ensure we don't violate condition variable red black tree invariants. */
if (auto *cv_tree = thread->GetConditionVariableTree(); cv_tree != nullptr) {
BeforeUpdatePriority(cv_tree, thread);
}
/* Change the priority. */
const s32 old_priority = thread->GetPriority();
thread->SetPriority(new_priority);
/* Restore the condition variable, if relevant. */
if (auto *cv_tree = thread->GetConditionVariableTree(); cv_tree != nullptr) {
AfterUpdatePriority(cv_tree, thread);
}
/* Update the scheduler. */
KScheduler::OnThreadPriorityChanged(thread, old_priority);
/* Keep the lock owner up to date. */
KThread *lock_owner = thread->GetLockOwner();
if (lock_owner == nullptr) {
return;
}
/* Update the thread in the lock owner's sorted list, and continue inheriting. */
lock_owner->RemoveWaiterImpl(thread);
lock_owner->AddWaiterImpl(thread);
thread = lock_owner;
}
}
void KThread::AddWaiter(KThread *thread) {
MESOSPHERE_ASSERT_THIS();
this->AddWaiterImpl(thread);
RestorePriority(this);
}
void KThread::RemoveWaiter(KThread *thread) {
MESOSPHERE_ASSERT_THIS();
this->RemoveWaiterImpl(thread);
RestorePriority(this);
}
KThread *KThread::RemoveWaiterByKey(s32 *out_num_waiters, KProcessAddress key) {
MESOSPHERE_ASSERT_THIS();
MESOSPHERE_ASSERT(KScheduler::IsSchedulerLockedByCurrentThread());
s32 num_waiters = 0;
KThread *next_lock_owner = nullptr;
auto it = m_waiter_list.begin();
while (it != m_waiter_list.end()) {
if (it->GetAddressKey() == key) {
KThread *thread = std::addressof(*it);
/* Keep track of how many kernel waiters we have. */
if (IsKernelAddressKey(thread->GetAddressKey())) {
MESOSPHERE_ABORT_UNLESS((m_num_kernel_waiters--) > 0);
KScheduler::SetSchedulerUpdateNeeded();
}
it = m_waiter_list.erase(it);
/* Update the next lock owner. */
if (next_lock_owner == nullptr) {
next_lock_owner = thread;
next_lock_owner->SetLockOwner(nullptr);
} else {
next_lock_owner->AddWaiterImpl(thread);
}
num_waiters++;
} else {
it++;
}
}
/* Do priority updates, if we have a next owner. */
if (next_lock_owner) {
RestorePriority(this);
RestorePriority(next_lock_owner);
}
/* Return output. */
*out_num_waiters = num_waiters;
return next_lock_owner;
}
Result KThread::Run() {
MESOSPHERE_ASSERT_THIS();
/* If the kernel hasn't finished initializing, then we should suspend. */
if (Kernel::GetState() != Kernel::State::Initialized) {
this->RequestSuspend(SuspendType_Init);
}
while (true) {
KScopedSchedulerLock lk;
/* If either this thread or the current thread are requesting termination, note it. */
R_UNLESS(!this->IsTerminationRequested(), svc::ResultTerminationRequested());
R_UNLESS(!GetCurrentThread().IsTerminationRequested(), svc::ResultTerminationRequested());
/* Ensure our thread state is correct. */
R_UNLESS(this->GetState() == ThreadState_Initialized, svc::ResultInvalidState());
/* If the current thread has been asked to suspend, suspend it and retry. */
if (GetCurrentThread().IsSuspended()) {
GetCurrentThread().Suspend();
continue;
}
/* If we're not a kernel thread and we've been asked to suspend, suspend ourselves. */
if (this->IsUserThread() && this->IsSuspended()) {
this->Suspend();
}
/* Set our state and finish. */
this->SetState(KThread::ThreadState_Runnable);
return ResultSuccess();
}
}
void KThread::Exit() {
MESOSPHERE_ASSERT_THIS();
MESOSPHERE_ASSERT(this == GetCurrentThreadPointer());
/* Call the debug callback. */
KDebug::OnExitThread(this);
/* Release the thread resource hint from parent. */
if (m_parent != nullptr) {
m_parent->ReleaseResource(ams::svc::LimitableResource_ThreadCountMax, 0, 1);
m_resource_limit_release_hint = true;
}
/* Perform termination. */
{
KScopedSchedulerLock sl;
/* Disallow all suspension. */
m_suspend_allowed_flags = 0;
/* Start termination. */
this->StartTermination();
/* Register the thread as a work task. */
KWorkerTaskManager::AddTask(KWorkerTaskManager::WorkerType_Exit, this);
}
MESOSPHERE_PANIC("KThread::Exit() would return");
}
void KThread::Terminate() {
MESOSPHERE_ASSERT_THIS();
MESOSPHERE_ASSERT(this != GetCurrentThreadPointer());
/* Request the thread terminate. */
if (const auto new_state = this->RequestTerminate(); new_state != ThreadState_Terminated) {
/* If the thread isn't terminated, wait for it to terminate. */
s32 index;
KSynchronizationObject *objects[] = { this };
KSynchronizationObject::Wait(std::addressof(index), objects, 1, ams::svc::WaitInfinite);
}
}
KThread::ThreadState KThread::RequestTerminate() {
MESOSPHERE_ASSERT_THIS();
MESOSPHERE_ASSERT(this != GetCurrentThreadPointer());
KScopedSchedulerLock sl;
/* Determine if this is the first termination request. */
const bool first_request = [&] ALWAYS_INLINE_LAMBDA () -> bool {
/* Perform an atomic compare-and-swap from false to true. */
u8 expected = false;
return m_termination_requested.compare_exchange_strong(expected, true);
}();
/* If this is the first request, start termination procedure. */
if (first_request) {
/* If the thread is in initialized state, just change state to terminated. */
if (this->GetState() == ThreadState_Initialized) {
m_thread_state = ThreadState_Terminated;
return ThreadState_Terminated;
}
/* Register the terminating dpc. */
this->RegisterDpc(DpcFlag_Terminating);
/* If the thread is suspended, continue it. */
if (this->IsSuspended()) {
m_suspend_allowed_flags = 0;
this->Continue();
}
/* Change the thread's priority to be higher than any system thread's. */
if (this->GetBasePriority() >= ams::svc::SystemThreadPriorityHighest) {
this->SetBasePriority(TerminatingThreadPriority);
}
/* If the thread is runnable, send a termination interrupt to other cores. */
if (this->GetState() == ThreadState_Runnable) {
if (const u64 core_mask = m_physical_affinity_mask.GetAffinityMask() & ~(1ul << GetCurrentCoreId()); core_mask != 0) {
cpu::DataSynchronizationBarrier();
Kernel::GetInterruptManager().SendInterProcessorInterrupt(KInterruptName_ThreadTerminate, core_mask);
}
}
/* Wake up the thread. */
this->SetSyncedObject(nullptr, svc::ResultTerminationRequested());
this->Wakeup();
}
return this->GetState();
}
Result KThread::Sleep(s64 timeout) {
MESOSPHERE_ASSERT_THIS();
MESOSPHERE_ASSERT(!KScheduler::IsSchedulerLockedByCurrentThread());
MESOSPHERE_ASSERT(this == GetCurrentThreadPointer());
MESOSPHERE_ASSERT(timeout > 0);
KHardwareTimer *timer;
{
/* Setup the scheduling lock and sleep. */
KScopedSchedulerLockAndSleep slp(std::addressof(timer), this, timeout);
/* Check if the thread should terminate. */
if (this->IsTerminationRequested()) {
slp.CancelSleep();
return svc::ResultTerminationRequested();
}
/* Mark the thread as waiting. */
this->SetState(KThread::ThreadState_Waiting);
}
/* The lock/sleep is done. */
/* Cancel the timer. */
timer->CancelTask(this);
return ResultSuccess();
}
void KThread::SetState(ThreadState state) {
MESOSPHERE_ASSERT_THIS();
KScopedSchedulerLock sl;
const ThreadState old_state = m_thread_state;
m_thread_state = static_cast<ThreadState>((old_state & ~ThreadState_Mask) | (state & ThreadState_Mask));
if (m_thread_state != old_state) {
KScheduler::OnThreadStateChanged(this, old_state);
}
}
KThreadContext *KThread::GetContextForSchedulerLoop() {
return std::addressof(this->GetContext());
}
KThread *KThread::GetThreadFromId(u64 thread_id) {
/* Lock the list. */
KThread::ListAccessor accessor;
const auto end = accessor.end();
/* Define helper object to find the thread. */
class IdObjectHelper : public KAutoObjectWithListContainer::ListType::value_type {
private:
u64 m_id;
public:
constexpr explicit IdObjectHelper(u64 id) : m_id(id) { /* ... */ }
virtual u64 GetId() const override { return m_id; }
};
/* Find the object with the right id. */
const auto it = accessor.find(IdObjectHelper(thread_id));
/* Check to make sure we found the thread. */
if (it == end) {
return nullptr;
}
/* Get the thread. */
KThread *thread = static_cast<KThread *>(std::addressof(*it));
/* Open the thread. */
if (AMS_LIKELY(thread->Open())) {
MESOSPHERE_ASSERT(thread->GetId() == thread_id);
return thread;
}
/* We failed to find the thread. */
return nullptr;
}
Result KThread::GetThreadList(s32 *out_num_threads, ams::kern::svc::KUserPointer<u64 *> out_thread_ids, s32 max_out_count) {
/* Lock the list. */
KThread::ListAccessor accessor;
const auto end = accessor.end();
/* Iterate over the list. */
s32 count = 0;
for (auto it = accessor.begin(); it != end; ++it) {
/* If we're within array bounds, write the id. */
if (count < max_out_count) {
/* Get the thread id. */
KThread *thread = static_cast<KThread *>(std::addressof(*it));
const u64 id = thread->GetId();
/* Copy the id to userland. */
R_TRY(out_thread_ids.CopyArrayElementFrom(std::addressof(id), count));
}
/* Increment the count. */
++count;
}
/* We successfully iterated the list. */
*out_num_threads = count;
return ResultSuccess();
}
}