// Copyright (c) 2012 The Chromium Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. #include "net/disk_cache/block_files.h" #include "base/atomicops.h" #include "base/file_util.h" #include "base/metrics/histogram.h" #include "base/string_util.h" #include "base/stringprintf.h" #include "base/threading/thread_checker.h" #include "base/time.h" #include "net/disk_cache/cache_util.h" #include "net/disk_cache/file_lock.h" #include "net/disk_cache/trace.h" using base::TimeTicks; namespace { const char* kBlockName = "data_"; // This array is used to perform a fast lookup of the nibble bit pattern to the // type of entry that can be stored there (number of consecutive blocks). const char s_types[16] = {4, 3, 2, 2, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0}; // Returns the type of block (number of consecutive blocks that can be stored) // for a given nibble of the bitmap. inline int GetMapBlockType(uint8 value) { value &= 0xf; return s_types[value]; } void FixAllocationCounters(disk_cache::BlockFileHeader* header); // Creates a new entry on the allocation map, updating the apropriate counters. // target is the type of block to use (number of empty blocks), and size is the // actual number of blocks to use. bool CreateMapBlock(int target, int size, disk_cache::BlockFileHeader* header, int* index) { if (target <= 0 || target > disk_cache::kMaxNumBlocks || size <= 0 || size > disk_cache::kMaxNumBlocks) { NOTREACHED(); return false; } TimeTicks start = TimeTicks::Now(); // We are going to process the map on 32-block chunks (32 bits), and on every // chunk, iterate through the 8 nibbles where the new block can be located. int current = header->hints[target - 1]; for (int i = 0; i < header->max_entries / 32; i++, current++) { if (current == header->max_entries / 32) current = 0; uint32 map_block = header->allocation_map[current]; for (int j = 0; j < 8; j++, map_block >>= 4) { if (GetMapBlockType(map_block) != target) continue; disk_cache::FileLock lock(header); int index_offset = j * 4 + 4 - target; *index = current * 32 + index_offset; DCHECK_EQ(*index / 4, (*index + size - 1) / 4); uint32 to_add = ((1 << size) - 1) << index_offset; header->num_entries++; // Note that there is no race in the normal sense here, but if we enforce // the order of memory accesses between num_entries and allocation_map, we // can assert that even if we crash here, num_entries will never be less // than the actual number of used blocks. base::subtle::MemoryBarrier(); header->allocation_map[current] |= to_add; header->hints[target - 1] = current; header->empty[target - 1]--; DCHECK_GE(header->empty[target - 1], 0); if (target != size) { header->empty[target - size - 1]++; } HISTOGRAM_TIMES("DiskCache.CreateBlock", TimeTicks::Now() - start); return true; } } // It is possible to have an undetected corruption (for example when the OS // crashes), fix it here. LOG(ERROR) << "Failing CreateMapBlock"; FixAllocationCounters(header); return false; } // Deletes the block pointed by index from allocation_map, and updates the // relevant counters on the header. void DeleteMapBlock(int index, int size, disk_cache::BlockFileHeader* header) { if (size < 0 || size > disk_cache::kMaxNumBlocks) { NOTREACHED(); return; } TimeTicks start = TimeTicks::Now(); int byte_index = index / 8; uint8* byte_map = reinterpret_cast(header->allocation_map); uint8 map_block = byte_map[byte_index]; if (index % 8 >= 4) map_block >>= 4; // See what type of block will be availabe after we delete this one. int bits_at_end = 4 - size - index % 4; uint8 end_mask = (0xf << (4 - bits_at_end)) & 0xf; bool update_counters = (map_block & end_mask) == 0; uint8 new_value = map_block & ~(((1 << size) - 1) << (index % 4)); int new_type = GetMapBlockType(new_value); disk_cache::FileLock lock(header); DCHECK((((1 << size) - 1) << (index % 8)) < 0x100); uint8 to_clear = ((1 << size) - 1) << (index % 8); DCHECK((byte_map[byte_index] & to_clear) == to_clear); byte_map[byte_index] &= ~to_clear; if (update_counters) { if (bits_at_end) header->empty[bits_at_end - 1]--; header->empty[new_type - 1]++; DCHECK_GE(header->empty[bits_at_end - 1], 0); } base::subtle::MemoryBarrier(); header->num_entries--; DCHECK_GE(header->num_entries, 0); HISTOGRAM_TIMES("DiskCache.DeleteBlock", TimeTicks::Now() - start); } #ifndef NDEBUG // Returns true if the specified block is used. Note that this is a simplified // version of DeleteMapBlock(). bool UsedMapBlock(int index, int size, disk_cache::BlockFileHeader* header) { if (size < 0 || size > disk_cache::kMaxNumBlocks) { NOTREACHED(); return false; } int byte_index = index / 8; uint8* byte_map = reinterpret_cast(header->allocation_map); uint8 map_block = byte_map[byte_index]; if (index % 8 >= 4) map_block >>= 4; DCHECK((((1 << size) - 1) << (index % 8)) < 0x100); uint8 to_clear = ((1 << size) - 1) << (index % 8); return ((byte_map[byte_index] & to_clear) == to_clear); } #endif // NDEBUG // Restores the "empty counters" and allocation hints. void FixAllocationCounters(disk_cache::BlockFileHeader* header) { for (int i = 0; i < disk_cache::kMaxNumBlocks; i++) { header->hints[i] = 0; header->empty[i] = 0; } for (int i = 0; i < header->max_entries / 32; i++) { uint32 map_block = header->allocation_map[i]; for (int j = 0; j < 8; j++, map_block >>= 4) { int type = GetMapBlockType(map_block); if (type) header->empty[type -1]++; } } } // Returns true if the current block file should not be used as-is to store more // records. |block_count| is the number of blocks to allocate. bool NeedToGrowBlockFile(const disk_cache::BlockFileHeader* header, int block_count) { bool have_space = false; int empty_blocks = 0; for (int i = 0; i < disk_cache::kMaxNumBlocks; i++) { empty_blocks += header->empty[i] * (i + 1); if (i >= block_count - 1 && header->empty[i]) have_space = true; } if (header->next_file && (empty_blocks < disk_cache::kMaxBlocks / 10)) { // This file is almost full but we already created another one, don't use // this file yet so that it is easier to find empty blocks when we start // using this file again. return true; } return !have_space; } // Returns the number of empty blocks for this file. int EmptyBlocks(const disk_cache::BlockFileHeader* header) { int empty_blocks = 0; for (int i = 0; i < disk_cache::kMaxNumBlocks; i++) { empty_blocks += header->empty[i] * (i + 1); if (header->empty[i] < 0) return false; } return empty_blocks; } // Returns true if the counters look OK. bool ValidateCounters(const disk_cache::BlockFileHeader* header) { if (header->max_entries < 0 || header->max_entries > disk_cache::kMaxBlocks || header->num_entries < 0) return false; int empty_blocks = EmptyBlocks(header); if (empty_blocks + header->num_entries > header->max_entries) return false; return true; } } // namespace namespace disk_cache { BlockFiles::BlockFiles(const FilePath& path) : init_(false), zero_buffer_(NULL), path_(path) { } BlockFiles::~BlockFiles() { if (zero_buffer_) delete[] zero_buffer_; CloseFiles(); } bool BlockFiles::Init(bool create_files) { DCHECK(!init_); if (init_) return false; thread_checker_.reset(new base::ThreadChecker); block_files_.resize(kFirstAdditionalBlockFile); for (int i = 0; i < kFirstAdditionalBlockFile; i++) { if (create_files) if (!CreateBlockFile(i, static_cast(i + 1), true)) return false; if (!OpenBlockFile(i)) return false; // Walk this chain of files removing empty ones. if (!RemoveEmptyFile(static_cast(i + 1))) return false; } init_ = true; return true; } MappedFile* BlockFiles::GetFile(Addr address) { DCHECK(thread_checker_->CalledOnValidThread()); DCHECK(block_files_.size() >= 4); DCHECK(address.is_block_file() || !address.is_initialized()); if (!address.is_initialized()) return NULL; int file_index = address.FileNumber(); if (static_cast(file_index) >= block_files_.size() || !block_files_[file_index]) { // We need to open the file if (!OpenBlockFile(file_index)) return NULL; } DCHECK(block_files_.size() >= static_cast(file_index)); return block_files_[file_index]; } bool BlockFiles::CreateBlock(FileType block_type, int block_count, Addr* block_address) { DCHECK(thread_checker_->CalledOnValidThread()); if (block_type < RANKINGS || block_type > BLOCK_4K || block_count < 1 || block_count > 4) return false; if (!init_) return false; MappedFile* file = FileForNewBlock(block_type, block_count); if (!file) return false; ScopedFlush flush(file); BlockFileHeader* header = reinterpret_cast(file->buffer()); int target_size = 0; for (int i = block_count; i <= 4; i++) { if (header->empty[i - 1]) { target_size = i; break; } } DCHECK(target_size); int index; if (!CreateMapBlock(target_size, block_count, header, &index)) return false; Addr address(block_type, block_count, header->this_file, index); block_address->set_value(address.value()); Trace("CreateBlock 0x%x", address.value()); return true; } void BlockFiles::DeleteBlock(Addr address, bool deep) { DCHECK(thread_checker_->CalledOnValidThread()); if (!address.is_initialized() || address.is_separate_file()) return; if (!zero_buffer_) { zero_buffer_ = new char[Addr::BlockSizeForFileType(BLOCK_4K) * 4]; memset(zero_buffer_, 0, Addr::BlockSizeForFileType(BLOCK_4K) * 4); } MappedFile* file = GetFile(address); if (!file) return; Trace("DeleteBlock 0x%x", address.value()); size_t size = address.BlockSize() * address.num_blocks(); size_t offset = address.start_block() * address.BlockSize() + kBlockHeaderSize; if (deep) file->Write(zero_buffer_, size, offset); BlockFileHeader* header = reinterpret_cast(file->buffer()); DeleteMapBlock(address.start_block(), address.num_blocks(), header); file->Flush(); if (!header->num_entries) { // This file is now empty. Let's try to delete it. FileType type = Addr::RequiredFileType(header->entry_size); if (Addr::BlockSizeForFileType(RANKINGS) == header->entry_size) type = RANKINGS; RemoveEmptyFile(type); // Ignore failures. } } void BlockFiles::CloseFiles() { if (init_) { DCHECK(thread_checker_->CalledOnValidThread()); } init_ = false; for (unsigned int i = 0; i < block_files_.size(); i++) { if (block_files_[i]) { block_files_[i]->Release(); block_files_[i] = NULL; } } block_files_.clear(); } void BlockFiles::ReportStats() { DCHECK(thread_checker_->CalledOnValidThread()); int used_blocks[kFirstAdditionalBlockFile]; int load[kFirstAdditionalBlockFile]; for (int i = 0; i < kFirstAdditionalBlockFile; i++) { GetFileStats(i, &used_blocks[i], &load[i]); } UMA_HISTOGRAM_COUNTS("DiskCache.Blocks_0", used_blocks[0]); UMA_HISTOGRAM_COUNTS("DiskCache.Blocks_1", used_blocks[1]); UMA_HISTOGRAM_COUNTS("DiskCache.Blocks_2", used_blocks[2]); UMA_HISTOGRAM_COUNTS("DiskCache.Blocks_3", used_blocks[3]); UMA_HISTOGRAM_ENUMERATION("DiskCache.BlockLoad_0", load[0], 101); UMA_HISTOGRAM_ENUMERATION("DiskCache.BlockLoad_1", load[1], 101); UMA_HISTOGRAM_ENUMERATION("DiskCache.BlockLoad_2", load[2], 101); UMA_HISTOGRAM_ENUMERATION("DiskCache.BlockLoad_3", load[3], 101); } bool BlockFiles::IsValid(Addr address) { #ifdef NDEBUG return true; #else if (!address.is_initialized() || address.is_separate_file()) return false; MappedFile* file = GetFile(address); if (!file) return false; BlockFileHeader* header = reinterpret_cast(file->buffer()); bool rv = UsedMapBlock(address.start_block(), address.num_blocks(), header); DCHECK(rv); static bool read_contents = false; if (read_contents) { scoped_array buffer; buffer.reset(new char[Addr::BlockSizeForFileType(BLOCK_4K) * 4]); size_t size = address.BlockSize() * address.num_blocks(); size_t offset = address.start_block() * address.BlockSize() + kBlockHeaderSize; bool ok = file->Read(buffer.get(), size, offset); DCHECK(ok); } return rv; #endif } bool BlockFiles::CreateBlockFile(int index, FileType file_type, bool force) { FilePath name = Name(index); int flags = force ? base::PLATFORM_FILE_CREATE_ALWAYS : base::PLATFORM_FILE_CREATE; flags |= base::PLATFORM_FILE_WRITE | base::PLATFORM_FILE_EXCLUSIVE_WRITE; scoped_refptr file(new File( base::CreatePlatformFile(name, flags, NULL, NULL))); if (!file->IsValid()) return false; BlockFileHeader header; header.entry_size = Addr::BlockSizeForFileType(file_type); header.this_file = static_cast(index); DCHECK(index <= kint16max && index >= 0); return file->Write(&header, sizeof(header), 0); } bool BlockFiles::OpenBlockFile(int index) { if (block_files_.size() - 1 < static_cast(index)) { DCHECK(index > 0); int to_add = index - static_cast(block_files_.size()) + 1; block_files_.resize(block_files_.size() + to_add); } FilePath name = Name(index); scoped_refptr file(new MappedFile()); if (!file->Init(name, kBlockHeaderSize)) { LOG(ERROR) << "Failed to open " << name.value(); return false; } size_t file_len = file->GetLength(); if (file_len < static_cast(kBlockHeaderSize)) { LOG(ERROR) << "File too small " << name.value(); return false; } BlockFileHeader* header = reinterpret_cast(file->buffer()); if (kBlockMagic != header->magic || kCurrentVersion != header->version) { LOG(ERROR) << "Invalid file version or magic " << name.value(); return false; } if (header->updating || !ValidateCounters(header)) { // Last instance was not properly shutdown, or counters are out of sync. if (!FixBlockFileHeader(file)) { LOG(ERROR) << "Unable to fix block file " << name.value(); return false; } } if (static_cast(file_len) < header->max_entries * header->entry_size + kBlockHeaderSize) { LOG(ERROR) << "File too small " << name.value(); return false; } if (index == 0) { // Load the links file into memory with a single read. scoped_array buf(new char[file_len]); if (!file->Read(buf.get(), file_len, 0)) return false; } ScopedFlush flush(file); DCHECK(!block_files_[index]); file.swap(&block_files_[index]); return true; } bool BlockFiles::GrowBlockFile(MappedFile* file, BlockFileHeader* header) { if (kMaxBlocks == header->max_entries) return false; ScopedFlush flush(file); DCHECK(!header->empty[3]); int new_size = header->max_entries + 1024; if (new_size > kMaxBlocks) new_size = kMaxBlocks; int new_size_bytes = new_size * header->entry_size + sizeof(*header); if (!file->SetLength(new_size_bytes)) { // Most likely we are trying to truncate the file, so the header is wrong. if (header->updating < 10 && !FixBlockFileHeader(file)) { // If we can't fix the file increase the lock guard so we'll pick it on // the next start and replace it. header->updating = 100; return false; } return (header->max_entries >= new_size); } FileLock lock(header); header->empty[3] = (new_size - header->max_entries) / 4; // 4 blocks entries header->max_entries = new_size; return true; } MappedFile* BlockFiles::FileForNewBlock(FileType block_type, int block_count) { COMPILE_ASSERT(RANKINGS == 1, invalid_file_type); MappedFile* file = block_files_[block_type - 1]; BlockFileHeader* header = reinterpret_cast(file->buffer()); TimeTicks start = TimeTicks::Now(); while (NeedToGrowBlockFile(header, block_count)) { if (kMaxBlocks == header->max_entries) { file = NextFile(file); if (!file) return NULL; header = reinterpret_cast(file->buffer()); continue; } if (!GrowBlockFile(file, header)) return NULL; break; } HISTOGRAM_TIMES("DiskCache.GetFileForNewBlock", TimeTicks::Now() - start); return file; } MappedFile* BlockFiles::NextFile(MappedFile* file) { ScopedFlush flush(file); BlockFileHeader* header = reinterpret_cast(file->buffer()); int new_file = header->next_file; if (!new_file) { // RANKINGS is not reported as a type for small entries, but we may be // extending the rankings block file. FileType type = Addr::RequiredFileType(header->entry_size); if (header->entry_size == Addr::BlockSizeForFileType(RANKINGS)) type = RANKINGS; new_file = CreateNextBlockFile(type); if (!new_file) return NULL; FileLock lock(header); header->next_file = new_file; } // Only the block_file argument is relevant for what we want. Addr address(BLOCK_256, 1, new_file, 0); return GetFile(address); } int BlockFiles::CreateNextBlockFile(FileType block_type) { for (int i = kFirstAdditionalBlockFile; i <= kMaxBlockFile; i++) { if (CreateBlockFile(i, block_type, false)) return i; } return 0; } // We walk the list of files for this particular block type, deleting the ones // that are empty. bool BlockFiles::RemoveEmptyFile(FileType block_type) { MappedFile* file = block_files_[block_type - 1]; BlockFileHeader* header = reinterpret_cast(file->buffer()); while (header->next_file) { // Only the block_file argument is relevant for what we want. Addr address(BLOCK_256, 1, header->next_file, 0); MappedFile* next_file = GetFile(address); if (!next_file) return false; BlockFileHeader* next_header = reinterpret_cast(next_file->buffer()); if (!next_header->num_entries) { DCHECK_EQ(next_header->entry_size, header->entry_size); // Delete next_file and remove it from the chain. int file_index = header->next_file; header->next_file = next_header->next_file; DCHECK(block_files_.size() >= static_cast(file_index)); file->Flush(); // We get a new handle to the file and release the old one so that the // file gets unmmaped... so we can delete it. FilePath name = Name(file_index); scoped_refptr this_file(new File(false)); this_file->Init(name); block_files_[file_index]->Release(); block_files_[file_index] = NULL; int failure = DeleteCacheFile(name) ? 0 : 1; UMA_HISTOGRAM_COUNTS("DiskCache.DeleteFailed2", failure); if (failure) LOG(ERROR) << "Failed to delete " << name.value() << " from the cache."; continue; } header = next_header; file = next_file; } return true; } // Note that we expect to be called outside of a FileLock... however, we cannot // DCHECK on header->updating because we may be fixing a crash. bool BlockFiles::FixBlockFileHeader(MappedFile* file) { ScopedFlush flush(file); BlockFileHeader* header = reinterpret_cast(file->buffer()); int file_size = static_cast(file->GetLength()); if (file_size < static_cast(sizeof(*header))) return false; // file_size > 2GB is also an error. const int kMinBlockSize = 36; const int kMaxBlockSize = 4096; if (header->entry_size < kMinBlockSize || header->entry_size > kMaxBlockSize || header->num_entries < 0) return false; // Make sure that we survive crashes. header->updating = 1; int expected = header->entry_size * header->max_entries + sizeof(*header); if (file_size != expected) { int max_expected = header->entry_size * kMaxBlocks + sizeof(*header); if (file_size < expected || header->empty[3] || file_size > max_expected) { NOTREACHED(); LOG(ERROR) << "Unexpected file size"; return false; } // We were in the middle of growing the file. int num_entries = (file_size - sizeof(*header)) / header->entry_size; header->max_entries = num_entries; } FixAllocationCounters(header); int empty_blocks = EmptyBlocks(header); if (empty_blocks + header->num_entries > header->max_entries) header->num_entries = header->max_entries - empty_blocks; if (!ValidateCounters(header)) return false; header->updating = 0; return true; } // We are interested in the total number of blocks used by this file type, and // the max number of blocks that we can store (reported as the percentage of // used blocks). In order to find out the number of used blocks, we have to // substract the empty blocks from the total blocks for each file in the chain. void BlockFiles::GetFileStats(int index, int* used_count, int* load) { int max_blocks = 0; *used_count = 0; *load = 0; for (;;) { if (!block_files_[index] && !OpenBlockFile(index)) return; BlockFileHeader* header = reinterpret_cast(block_files_[index]->buffer()); max_blocks += header->max_entries; int used = header->max_entries; for (int i = 0; i < 4; i++) { used -= header->empty[i] * (i + 1); DCHECK_GE(used, 0); } *used_count += used; if (!header->next_file) break; index = header->next_file; } if (max_blocks) *load = *used_count * 100 / max_blocks; } FilePath BlockFiles::Name(int index) { // The file format allows for 256 files. DCHECK(index < 256 || index >= 0); std::string tmp = base::StringPrintf("%s%d", kBlockName, index); return path_.AppendASCII(tmp); } } // namespace disk_cache