Add new simple disk cache backend.

net/disk_cache/simple/simple_synchronous_entry.cc

+// Copyright (c) 2013 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/simple/simple_synchronous_entry.h"
+
+#include <algorithm>
+#include <cstring>
+
+#include "base/basictypes.h"
+#include "base/file_util.h"
+#include "base/hash.h"
+#include "base/location.h"
+#include "base/message_loop_proxy.h"
+#include "base/sha1.h"
+#include "base/stringprintf.h"
+#include "base/task_runner.h"
+#include "net/base/io_buffer.h"
+#include "net/base/net_errors.h"
+#include "net/disk_cache/simple/simple_disk_format.h"
+
+using base::ClosePlatformFile;
+using base::GetPlatformFileInfo;
+using base::PlatformFileError;
+using base::PlatformFileInfo;
+using base::PLATFORM_FILE_CREATE_ALWAYS;
+using base::PLATFORM_FILE_OK;
+using base::PLATFORM_FILE_OPEN;
+using base::PLATFORM_FILE_READ;
+using base::PLATFORM_FILE_WRITE;
+using base::ReadPlatformFile;
+using base::TaskRunner;
+using base::Time;
+using base::TruncatePlatformFile;
+using base::WritePlatformFile;
+
+namespace {
+
+std::string GetFilenameForKeyAndIndex(const std::string& key, int index) {
+  const std::string sha_hash = base::SHA1HashString(key);
+  return StringPrintf("%02x%02x%02x%02x%02x_%1d",
+                      implicit_cast<unsigned char>(sha_hash[0]),
+                      implicit_cast<unsigned char>(sha_hash[1]),
+                      implicit_cast<unsigned char>(sha_hash[2]),
+                      implicit_cast<unsigned char>(sha_hash[3]),
+                      implicit_cast<unsigned char>(sha_hash[4]), index);
+}
+
+int32 DataSizeFromKeyAndFileSize(size_t key_size, int64 file_size) {
+  return file_size - key_size - sizeof(disk_cache::SimpleFileHeader);

[rvargas (doing something else), 2013/02/13 01:48:46]

There's a size mismatch here that should be handled (and signed vs unsigned). I
would suggest going for 32 bit ints.

[gavinp, 2013/02/14 15:29:55]

Done. I used std::numeric_limits<> to guard the cast. It's not the most common
Chrome style, but it's in the standard and I very much like the consistant
interface. I'm open to any other suggestions.

+}
+
+int64 FileOffsetFromDataOffset(size_t key_size, int offset) {

[rvargas (doing something else), 2013/02/13 01:48:46]

Do you expect to have large files?

[gavinp, 2013/02/14 15:29:55]

No.

[rvargas (doing something else), 2013/02/14 21:09:30]

My point is that then you can return the size that you expect from this code,
and the caller code would still work as is. This makes it explicit that there is
no support / expectation for huge files.

+  const int64 headers_size = sizeof(disk_cache::SimpleFileHeader) +
+                             key_size;
+  return headers_size + offset;
+}
+
+}  // namespace
+
+namespace disk_cache {
+
+// static
+void SimpleSynchronousEntry::OpenEntry(
+    const FilePath& path,
+    const std::string& key,
+    const scoped_refptr<TaskRunner>& callback_runner,
+    const SynchronousEntryCallback& callback) {
+  SimpleSynchronousEntry* sync_entry =
+      new SimpleSynchronousEntry(callback_runner, path, key);
+
+  if (!sync_entry->InitializeForOpen()) {
+    delete sync_entry;
+    sync_entry = NULL;
+  }
+  callback_runner->PostTask(FROM_HERE,
+                            base::Bind(callback, sync_entry,
+                                       sync_entry ? net::OK : net::ERR_FAILED));

[rvargas (doing something else), 2013/02/13 01:48:46]

Sounds simpler (on both sides) to remove the last argument (error code).

[gavinp, 2013/02/14 15:29:55]

Done.

+}
+
+// static
+void SimpleSynchronousEntry::CreateEntry(
+    const FilePath& path,
+    const std::string& key,
+    const scoped_refptr<TaskRunner>& callback_runner,
+    const SynchronousEntryCallback& callback) {
+  SimpleSynchronousEntry* sync_entry =
+      new SimpleSynchronousEntry(callback_runner, path, key);
+
+  if (!sync_entry->InitializeForCreate()) {
+    delete sync_entry;
+    sync_entry = NULL;
+  }
+  callback_runner->PostTask(FROM_HERE,
+                            base::Bind(callback, sync_entry,
+                                       sync_entry ? net::OK : net::ERR_FAILED));
+}
+
+// static
+void SimpleSynchronousEntry::DoomEntry(
+    const FilePath& path,
+    const std::string& key,
+    scoped_refptr<TaskRunner> callback_runner,
+    const net::CompletionCallback& callback) {
+  for (int i = 0; i < kIndexCount; ++i) {
+    bool delete_result =
+        file_util::Delete(path.AppendASCII(GetFilenameForKeyAndIndex(key, i)),
+                          false);
+    DCHECK(delete_result);

[rvargas (doing something else), 2013/02/13 01:48:46]

should not dcheck something coming from the OS

[gavinp, 2013/02/14 15:29:55]

Done.

+  }
+  if (!callback.is_null())
+    callback_runner->PostTask(FROM_HERE, base::Bind(callback, net::OK));
+}
+
+void SimpleSynchronousEntry::DoomAndClose() {
+  scoped_refptr<TaskRunner> callback_runner = callback_runner_;
+  FilePath path = path_;
+  std::string key = key_;
+
+  Close();
+  // |this| is now deleted.
+
+  DoomEntry(path, key, callback_runner, net::CompletionCallback());
+}
+
+void SimpleSynchronousEntry::Close() {
+  for (int i = 0; i < kIndexCount; ++i) {
+    bool result = ClosePlatformFile(files_[i]);
+    DCHECK(result);
+  }
+  delete this;
+}
+
+void SimpleSynchronousEntry::ReadData(
+    int index,
+    int offset,
+    net::IOBuffer* buf,
+    int buf_len,
+    const SynchronousEntryCallback& callback) {
+  DCHECK(initialized_);
+  if (status_[index].mode != EntryStatus::ENTRY_READER)

[rvargas (doing something else), 2013/02/13 01:48:46]

What's the purpose of this?

[gavinp, 2013/02/14 15:29:55]

Removed.

+    status_[index].data_offset = 0;
+  DCHECK_EQ(status_[index].data_offset, offset);
+  status_[index].mode = EntryStatus::ENTRY_READER;
+
+  int64 file_offset = FileOffsetFromDataOffset(key_.size(), offset);
+  int bytes_read = ReadPlatformFile(files_[index], file_offset,
+                                    buf->data(), buf_len);
+  if (bytes_read > 0) {
+    last_used_ = Time::Now();
+    status_[index].data_offset += bytes_read;
+  }
+  int result = (bytes_read >= 0) ? bytes_read : net::ERR_FAILED;
+  callback_runner_->PostTask(FROM_HERE, base::Bind(callback, this, result));
+}
+
+void SimpleSynchronousEntry::WriteData(
+    int index,
+    int offset,
+    net::IOBuffer* buf,
+    int buf_len,
+    const SynchronousEntryCallback& callback,
+    bool truncate) {
+  DCHECK(initialized_);
+  if (status_[index].mode != EntryStatus::ENTRY_WRITER ||
+      (truncate && offset == 0))
+    status_[index].data_offset = 0;
+  DCHECK_EQ(status_[index].data_offset, offset);
+  status_[index].mode = EntryStatus::ENTRY_WRITER;
+
+  int64 file_offset = FileOffsetFromDataOffset(key_.size(), offset);
+  if (buf_len > 0) {
+    if (WritePlatformFile(files_[index], file_offset, buf->data(), buf_len) !=
+        buf_len) {
+      callback_runner_->PostTask(FROM_HERE,
+                                 base::Bind(callback, this, net::ERR_FAILED));
+      return;
+    }
+    data_size_[index] = std::max(data_size_[index], offset + buf_len);
+    status_[index].data_offset += buf_len;
+  }
+  if (truncate) {
+    data_size_[index] = offset + buf_len;
+    if (!TruncatePlatformFile(files_[index], file_offset + buf_len)) {
+      callback_runner_->PostTask(FROM_HERE,
+                                 base::Bind(callback, this, net::ERR_FAILED));
+      return;
+    }
+  }
+  last_modified_ = Time::Now();
+  callback_runner_->PostTask(FROM_HERE, base::Bind(callback, this, buf_len));
+}
+
+SimpleSynchronousEntry::EntryStatus::EntryStatus()
+    : mode(ENTRY_UNINITIALIZED),
+      data_offset(0) {
+}
+
+SimpleSynchronousEntry::SimpleSynchronousEntry(
+    const scoped_refptr<TaskRunner>& callback_runner,
+    const FilePath& path,
+    const std::string& key) : callback_runner_(callback_runner),

[rvargas (doing something else), 2013/02/13 01:48:46]

nit: initialization list on the next line

[gavinp, 2013/02/14 15:29:55]

Done.

+                              path_(path),
+                              key_(key),
+                              initialized_(false) {
+}
+
+SimpleSynchronousEntry::~SimpleSynchronousEntry() {
+}
+
+bool SimpleSynchronousEntry::OpenOrCreateFiles(bool create) {
+  for (int i = 0; i < kIndexCount; ++i) {
+    FilePath filename = path_.AppendASCII(GetFilenameForKeyAndIndex(key_, i));
+    int flags = PLATFORM_FILE_READ | PLATFORM_FILE_WRITE;
+    if (create)
+      flags |= PLATFORM_FILE_CREATE_ALWAYS;

[rvargas (doing something else), 2013/02/13 01:48:46]

overwrite would be an error

[gavinp, 2013/02/14 15:29:55]

Done.

+    else
+      flags |= PLATFORM_FILE_OPEN;
+    PlatformFileError error;
+    files_[i] = CreatePlatformFile(filename, flags, NULL, &error);
+    if (error != PLATFORM_FILE_OK) {

[pasko-google - do not use, 2013/02/13 11:36:24]

please log the error here

[gavinp, 2013/02/14 15:29:55]

Done.

+      while (--i >= 0) {
+        bool did_close = ClosePlatformFile(files_[i]);
+        DCHECK(did_close);
+      }
+      return false;
+    }
+  }
+
+  for (int i = 0; i < kIndexCount; ++i) {
+    PlatformFileInfo file_info;
+    CHECK(GetPlatformFileInfo(files_[i], &file_info));
+    last_used_ = std::max(last_used_, file_info.last_accessed);
+    last_modified_ = std::max(last_modified_, file_info.last_modified);
+    data_size_[i] = DataSizeFromKeyAndFileSize(key_.size(), file_info.size);
+  }
+
+  return true;
+}
+
+bool SimpleSynchronousEntry::InitializeForOpen() {
+  DCHECK(!initialized_);
+  if (!OpenOrCreateFiles(false))
+    return false;
+
+  for (int i = 0; i < kIndexCount; ++i) {
+    SimpleFileHeader header;
+    if (ReadPlatformFile(files_[i], 0, reinterpret_cast<char*>(&header),
+                         sizeof(header)) != sizeof(header)) {
+      return false;
+    }
+
+    if (header.initial_magic_number != kSimpleInitialMagicNumber) {
+      // TODO(gavinp): This seems very bad; for now we log at WARNING, but we
+      // should give consideration to not saturating the log with these if that
+      // becomes a problem.
+      LOG(WARNING) << "Magic number did not match, saw "
+                   << header.initial_magic_number << " expecting "

[rvargas (doing something else), 2013/02/13 01:48:46]

nit: consider removing the saw vs expected parts. (one is a compile time
constant, the other... well, it doesn't match)

[gavinp, 2013/02/14 15:29:55]

Done.

+                   << kSimpleInitialMagicNumber;
+    }
+
+    if (header.version != kSimpleVersion) {
+      LOG(INFO) << "Unreadable version. Got " << header.version

[rvargas (doing something else), 2013/02/13 01:48:46]

nit: remove the variable parts?.
Shouldn't this be at the same level of the previous log?

[gavinp, 2013/02/14 15:29:55]

Done.

+                << " but can only read " << kSimpleVersion;
+      return false;
+    }
+
+    char key[4096];
+    DCHECK_LE(header.key_length, sizeof(key));

[rvargas (doing something else), 2013/02/13 01:48:46]

this doesn't work (in general).

[gavinp, 2013/02/14 15:29:55]

Now replaced with the thing that works in general.

+    if (ReadPlatformFile(files_[i], sizeof(header), key, header.key_length) !=
+        implicit_cast<int>(header.key_length)) {

[rvargas (doing something else), 2013/02/13 01:48:46]

why is key_length not an int?

[gavinp, 2013/02/14 15:29:55]

Because there are no negative length keys.

[rvargas (doing something else), 2013/02/14 21:09:30]

That's not the principle that determines signed vs unsigned on the net code (or
in Chrome/Google). Casts and signed/unsigned mismatches seem worse than gaining
one extra bit of resolution.

[gavinp, 2013/02/15 16:04:01]

Good point; I"d been assuming file formats were "different", and presuming the
size_t exception covered this case. That's stretching it. I'll look very
carefully at this in the next rev of any of the disk formats.

+      LOG(ERROR) << "Cannot read key from entry.";

[rvargas (doing something else), 2013/02/13 01:48:46]

nit: I would expect all failures from this method to be at the same level,
because the consequence of the failure is the same.

[gavinp, 2013/02/14 15:29:55]

Done.

+      return false;
+    }
+    if (header.key_length != key_.size() ||
+        std::memcmp(static_cast<const void*>(key_.data()),
+                    static_cast<const void*>(key),

[rvargas (doing something else), 2013/02/13 01:48:46]

shouldn't need the casts here

[gavinp, 2013/02/14 15:29:55]

Done.

+                    key_.size()) != 0) {
+      // TODO(gavinp): Since the way we use Entry SHA to name entries means this
+      // is expected to occur at some frequency, add unit_tests that this does
+      // is handled gracefully at higher levels.
+      LOG(INFO) << "Key mismatch on open, expecting " << key_
+                << " but file stored " << std::string(key, header.key_length);
+      return false;
+    }
+
+    if (base::Hash(key, header.key_length) != header.key_hash) {
+      LOG(ERROR) << "Hash mismatch on key.";
+      return false;
+    }
+  }
+
+  initialized_ = true;
+  return true;
+}
+
+bool SimpleSynchronousEntry::InitializeForCreate() {
+  DCHECK(!initialized_);
+  if (!OpenOrCreateFiles(true)) {
+    LOG(ERROR) << "Could not create platform files for key " << key_;
+    return false;
+  }
+
+  for (int i = 0; i < kIndexCount; ++i) {
+    SimpleFileHeader header;
+    header.initial_magic_number = kSimpleInitialMagicNumber;
+    header.version = kSimpleVersion;
+
+    header.key_length = key_.size();
+    header.key_hash = base::Hash(key_);
+
+    if (WritePlatformFile(files_[i], 0, reinterpret_cast<char*>(&header),
+                          sizeof(header)) != sizeof(header)) {
+      // TODO(gavinp): Clean up created files.
+      LOG(WARNING) << "Could not write headers to new cache entry.";
+      return false;
+    }
+
+    if (WritePlatformFile(files_[i], sizeof(header), key_.data(),
+                          key_.size()) != implicit_cast<int>(key_.size())) {
+      // TODO(gavinp): Clean up created files.
+      LOG(WARNING) << "Could not write keys to new cache entry.";
+      return false;
+    }
+  }
+
+  initialized_ = true;
+  return true;
+}
+
+}  // namespace disk_cache