Quellcodebibliothek Statistik Leitseite products/Sources/formale Sprachen/C/Android/art/art/runtime/gc/space/   (Android Betriebssystem Version 17©)  Datei vom 26.5.2026 mit Größe 11 kB image not shown  

Quelle  bump_pointer_space.cc

  Sprache: C
 

/*
 * Copyright (C) 2013 The Android Open Source Project
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 *      http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */


#include "bump_pointer_space.h"
#include "bump_pointer_space-inl.h"
#include "mirror/class-inl.h"
#include "mirror/object-inl.h"
#include "thread_list.h"

namespace art HIDDEN {
namespace gc {
namespace space {

BumpPointerSpace* BumpPointerSpace::Create(const std::string& name, size_t capacity) {
  capacity = RoundUp(capacity, gPageSize);
  std::string error_msg;
  MemMap mem_map = MemMap::MapAnonymous(name.c_str(),
                                        capacity,
                                        PROT_READ | PROT_WRITE,
                                        /*low_4gb=*/ true,
                                        &error_msg);
  if (!mem_map.IsValid()) {
    LOG(ERROR) << "Failed to allocate pages for alloc space (" << name << ") of size "
        << PrettySize(capacity) << " with message " << error_msg;
    return nullptr;
  }
  return new BumpPointerSpace(name, std::move(mem_map));
}

BumpPointerSpace* BumpPointerSpace::CreateFromMemMap(const std::string& name, MemMap&& mem_map) {
  return new BumpPointerSpace(name, std::move(mem_map));
}

BumpPointerSpace::BumpPointerSpace(const std::string& name, uint8_t* begin, uint8_t* limit)
    : ContinuousMemMapAllocSpace(
          name, MemMap::Invalid(), begin, begin, limit, kGcRetentionPolicyAlwaysCollect),
      growth_end_(limit),
      objects_allocated_(0),
      bytes_allocated_(0),
      lock_("Bump-pointer space lock"),
      main_block_size_(0) {
  // This constructor gets called only from Heap::PreZygoteFork(), which
  // doesn't require a mark_bitmap.
}

BumpPointerSpace::BumpPointerSpace(const std::string& name, MemMap&& mem_map)
    : ContinuousMemMapAllocSpace(name,
                                 std::move(mem_map),
                                 mem_map.Begin(),
                                 mem_map.Begin(),
                                 mem_map.End(),
                                 kGcRetentionPolicyAlwaysCollect),
      growth_end_(mem_map_.End()),
      objects_allocated_(0),
      bytes_allocated_(0),
      lock_("Bump-pointer space lock", kBumpPointerSpaceBlockLock),
      main_block_size_(0) {
  mark_bitmap_ =
      accounting::ContinuousSpaceBitmap::Create("bump-pointer space live bitmap",
                                                Begin(),
                                                Capacity());
}

void BumpPointerSpace::Clear() {
  // Release the pages back to the operating system.
  if (!kMadviseZeroes) {
    memset(Begin(), 0, Limit() - Begin());
  }
  CHECK_NE(madvise(Begin(), Limit() - Begin(), MADV_DONTNEED), -1) << "madvise failed";
  if (GetMarkBitmap() != nullptr) {
    GetMarkBitmap()->Clear();
  }
  // Reset the end of the space back to the beginning, we move the end forward as we allocate
  // objects.
  SetEnd(Begin());
  objects_allocated_.store(0, std::memory_order_relaxed);
  bytes_allocated_.store(0, std::memory_order_relaxed);
  {
    MutexLock mu(Thread::Current(), lock_);
    growth_end_ = Limit();
    block_sizes_.clear();
    main_block_size_ = 0;
    black_dense_region_size_ = 0;
  }
}

size_t BumpPointerSpace::ClampGrowthLimit(size_t new_capacity) {
  CHECK(gUseUserfaultfd);
  MutexLock mu(Thread::Current(), lock_);
  CHECK_EQ(growth_end_, Limit());
  uint8_t* end = End();
  CHECK_LE(end, growth_end_);
  size_t free_capacity = growth_end_ - end;
  size_t clamp_size = Capacity() - new_capacity;
  if (clamp_size > free_capacity) {
    new_capacity += clamp_size - free_capacity;
  }
  SetLimit(Begin() + new_capacity);
  growth_end_ = Limit();
  GetMemMap()->SetSize(new_capacity);
  if (GetMarkBitmap()->HeapBegin() != 0) {
    GetMarkBitmap()->SetHeapSize(new_capacity);
  }
  return new_capacity;
}

void BumpPointerSpace::Dump(std::ostream& os) const {
  os << GetName() << " "
      << reinterpret_cast<void*>(Begin()) << "-" << reinterpret_cast<void*>(End()) << " - "
      << reinterpret_cast<void*>(Limit());
}

size_t BumpPointerSpace::RevokeThreadLocalBuffers(Thread* thread) {
  MutexLock mu(Thread::Current(), lock_);
  RevokeThreadLocalBuffersLocked(thread);
  return 0U;
}

size_t BumpPointerSpace::RevokeAllThreadLocalBuffers() {
  Thread* self = Thread::Current();
  MutexLock mu(self, *Locks::runtime_shutdown_lock_);
  MutexLock mu2(self, *Locks::thread_list_lock_);
  // TODO: Not do a copy of the thread list?
  std::list<Thread*> thread_list = Runtime::Current()->GetThreadList()->GetList();
  for (Thread* thread : thread_list) {
    RevokeThreadLocalBuffers(thread);
  }
  return 0U;
}

void BumpPointerSpace::AssertThreadLocalBuffersAreRevoked(Thread* thread) {
  if (kIsDebugBuild) {
    MutexLock mu(Thread::Current(), lock_);
    DCHECK(!thread->HasTlab());
  }
}

void BumpPointerSpace::AssertAllThreadLocalBuffersAreRevoked() {
  if (kIsDebugBuild) {
    Thread* self = Thread::Current();
    MutexLock mu(self, *Locks::runtime_shutdown_lock_);
    MutexLock mu2(self, *Locks::thread_list_lock_);
    // TODO: Not do a copy of the thread list?
    std::list<Thread*> thread_list = Runtime::Current()->GetThreadList()->GetList();
    for (Thread* thread : thread_list) {
      AssertThreadLocalBuffersAreRevoked(thread);
    }
  }
}

void BumpPointerSpace::UpdateMainBlock() {
  DCHECK(block_sizes_.empty());
  main_block_size_ = Size();
}

// Returns the start of the storage.
uint8_t* BumpPointerSpace::AllocBlock(size_t bytes) {
  if (block_sizes_.empty()) {
    UpdateMainBlock();
  }
  uint8_t* storage = reinterpret_cast<uint8_t*>(AllocNonvirtualWithoutAccounting(bytes));
  if (LIKELY(storage != nullptr)) {
    block_sizes_.push_back(bytes);
  }
  return storage;
}

accounting::ContinuousSpaceBitmap::SweepCallback* BumpPointerSpace::GetSweepCallback() {
  UNIMPLEMENTED(FATAL);
  UNREACHABLE();
}

uint64_t BumpPointerSpace::GetBytesAllocated() {
  // Start out pre-determined amount (blocks which are not being allocated into).
  uint64_t total = static_cast<uint64_t>(bytes_allocated_.load(std::memory_order_relaxed));
  Thread* self = Thread::Current();
  MutexLock mu(self, *Locks::runtime_shutdown_lock_);
  MutexLock mu2(self, *Locks::thread_list_lock_);
  std::list<Thread*> thread_list = Runtime::Current()->GetThreadList()->GetList();
  MutexLock mu3(Thread::Current(), lock_);
  // If we don't have any blocks, we don't have any thread local buffers. This check is required
  // since there can exist multiple bump pointer spaces which exist at the same time.
  if (!block_sizes_.empty()) {
    for (Thread* thread : thread_list) {
      total += thread->GetThreadLocalBytesAllocated();
    }
  }
  return total;
}

uint64_t BumpPointerSpace::GetObjectsAllocated() {
  // Start out pre-determined amount (blocks which are not being allocated into).
  uint64_t total = static_cast<uint64_t>(objects_allocated_.load(std::memory_order_relaxed));
  Thread* self = Thread::Current();
  MutexLock mu(self, *Locks::runtime_shutdown_lock_);
  MutexLock mu2(self, *Locks::thread_list_lock_);
  std::list<Thread*> thread_list = Runtime::Current()->GetThreadList()->GetList();
  MutexLock mu3(Thread::Current(), lock_);
  // If we don't have any blocks, we don't have any thread local buffers. This check is required
  // since there can exist multiple bump pointer spaces which exist at the same time.
  if (!block_sizes_.empty()) {
    for (Thread* thread : thread_list) {
      total += thread->GetThreadLocalObjectsAllocated();
    }
  }
  return total;
}

void BumpPointerSpace::RevokeThreadLocalBuffersLocked(Thread* thread) {
  objects_allocated_.fetch_add(thread->GetThreadLocalObjectsAllocated(), std::memory_order_relaxed);
  bytes_allocated_.fetch_add(thread->GetThreadLocalBytesAllocated(), std::memory_order_relaxed);
  thread->ResetTlab();
}

bool BumpPointerSpace::AllocNewTlab(Thread* self, size_t bytes, size_t* bytes_tl_bulk_allocated) {
  bytes = RoundUp(bytes, kAlignment);
  MutexLock mu(Thread::Current(), lock_);
  RevokeThreadLocalBuffersLocked(self);
  uint8_t* start = AllocBlock(bytes);
  if (start == nullptr) {
    return false;
  }
  self->SetTlab(start, start + bytes, start + bytes);
  if (bytes_tl_bulk_allocated != nullptr) {
    *bytes_tl_bulk_allocated = bytes;
  }
  return true;
}

bool BumpPointerSpace::LogFragmentationAllocFailure(std::ostream& os,
                                                    size_t failed_alloc_bytes) {
  size_t max_contiguous_allocation = Limit() - End();
  if (failed_alloc_bytes > max_contiguous_allocation) {
    os << "; failed due to fragmentation (largest possible contiguous allocation "
       <<  max_contiguous_allocation << " bytes)";
    return true;
  }
  // Caller's job to print failed_alloc_bytes.
  return false;
}

size_t BumpPointerSpace::AllocationSizeNonvirtual(mirror::Object* obj, size_t* usable_size) {
  size_t num_bytes = obj->SizeOf();
  if (usable_size != nullptr) {
    *usable_size = RoundUp(num_bytes, kAlignment);
  }
  return num_bytes;
}

uint8_t* BumpPointerSpace::AlignEnd(Thread* self, size_t alignment, Heap* heap) {
  Locks::mutator_lock_->AssertExclusiveHeld(self);
  DCHECK(IsAligned<kAlignment>(alignment));
  uint8_t* end = end_.load(std::memory_order_relaxed);
  uint8_t* aligned_end = AlignUp(end, alignment);
  ptrdiff_t diff = aligned_end - end;
  if (diff > 0) {
    end_.store(aligned_end, std::memory_order_relaxed);
    heap->AddBytesAllocated(diff);
    // If we have blocks after the main one. Then just add the diff to the last
    // block.
    MutexLock mu(self, lock_);
    if (!block_sizes_.empty()) {
      block_sizes_.back() += diff;
    }
  }
  return aligned_end;
}

std::vector<size_t>* BumpPointerSpace::GetBlockSizes(Thread* self, size_t* main_block_size) {
  std::vector<size_t>* block_sizes = nullptr;
  MutexLock mu(self, lock_);
  if (!block_sizes_.empty()) {
    block_sizes = new std::vector<size_t>(block_sizes_.begin(), block_sizes_.end());
  } else {
    UpdateMainBlock();
  }
  *main_block_size = main_block_size_;
  return block_sizes;
}

void BumpPointerSpace::SetBlockSizes(Thread* self,
                                     const size_t main_block_size,
                                     const size_t first_valid_idx) {
  MutexLock mu(self, lock_);
  main_block_size_ = main_block_size;
  if (!block_sizes_.empty()) {
    block_sizes_.erase(block_sizes_.begin(), block_sizes_.begin() + first_valid_idx);
  }
  size_t size = main_block_size;
  for (size_t block_size : block_sizes_) {
    size += block_size;
  }
  DCHECK(IsAligned<kAlignment>(size));
  end_.store(Begin() + size, std::memory_order_relaxed);
}

void BumpPointerSpace::SetBlackDenseRegionSize(size_t size) {
  DCHECK_ALIGNED_PARAM(size, gPageSize);
  MutexLock mu(Thread::Current(), lock_);
  black_dense_region_size_ = size;
}

}  // namespace space
}  // namespace gc
}  // namespace art

Messung V0.5 in Prozent
C=90 H=91 G=90

¤ Dauer der Verarbeitung: 0.11 Sekunden  (vorverarbeitet am  2026-06-29) ¤

*© Formatika GbR, Deutschland






Wurzel

Suchen

PVS Prover

Isabelle Prover

NIST Cobol Testsuite

Cephes Mathematical Library

Vienna Development Method

Haftungshinweis

Die Informationen auf dieser Webseite wurden nach bestem Wissen sorgfältig zusammengestellt. Es wird jedoch weder Vollständigkeit, noch Richtigkeit, noch Qualität der bereit gestellten Informationen zugesichert.

Bemerkung:

Die farbliche Syntaxdarstellung und die Messung sind noch experimentell.