Add database recovery for FileSystemOriginDatabase

webkit/fileapi/file_system_directory_database.cc

 // 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 "webkit/fileapi/file_system_directory_database.h"
 
 #include <math.h>
 
 #include "base/location.h"
 #include "base/pickle.h"
 #include "base/string_number_conversions.h"
 #include "base/string_util.h"
 #include "base/sys_string_conversions.h"
 #include "third_party/leveldatabase/src/include/leveldb/iterator.h"
+#include "third_party/leveldatabase/src/include/leveldb/status.h"
 #include "third_party/leveldatabase/src/include/leveldb/write_batch.h"
 #include "webkit/fileapi/file_system_util.h"
 
 namespace {
 
 bool PickleFromFileInfo(
     const fileapi::FileSystemDirectoryDatabase::FileInfo& info,
     Pickle* pickle) {
   DCHECK(pickle);
   std::string data_path;
@@ skipping 98 matching lines @@
 #elif defined(OS_WIN)
   path_ = base::SysWideToUTF8(path.value());
 #endif
 }
 
 FileSystemDirectoryDatabase::~FileSystemDirectoryDatabase() {
 }
 
 bool FileSystemDirectoryDatabase::GetChildWithName(
     FileId parent_id, const FilePath::StringType& name, FileId* child_id) {
-  if (!Init())
+  if (!Init(true))

[jsbell, 2012/03/09 23:10:11]

The Boolean argument isn't very readable. Use an enum, or const bool recover =
true;

[tzik, 2012/03/10 00:16:55]

Done.

     return false;
   DCHECK(child_id);
   std::string child_key = GetChildLookupKey(parent_id, name);
   std::string child_id_string;
   leveldb::Status status =
       db_->Get(leveldb::ReadOptions(), child_key, &child_id_string);
   if (status.IsNotFound())
     return false;
   if (status.ok()) {
     if (!base::StringToInt64(child_id_string, child_id)) {
@@ skipping 20 matching lines @@
     if (!GetChildWithName(local_id, name, &local_id))
       return false;
   }
   *file_id = local_id;
   return true;
 }
 
 bool FileSystemDirectoryDatabase::ListChildren(
     FileId parent_id, std::vector<FileId>* children) {
   // Check to add later: fail if parent is a file, at least in debug builds.
-  if (!Init())
+  if (!Init(true))
     return false;
   DCHECK(children);
   std::string child_key_prefix = GetChildListingKeyPrefix(parent_id);
 
   scoped_ptr<leveldb::Iterator> iter(db_->NewIterator(leveldb::ReadOptions()));
   iter->Seek(child_key_prefix);
   children->clear();
   while (iter->Valid() &&
       StartsWithASCII(iter->key().ToString(), child_key_prefix, true)) {
     std::string child_id_string = iter->value().ToString();
     FileId child_id;
     if (!base::StringToInt64(child_id_string, &child_id)) {
       LOG(ERROR) << "Hit database corruption!";
       return false;
     }
     children->push_back(child_id);
     iter->Next();
   }
   return true;
 }
 
 bool FileSystemDirectoryDatabase::GetFileInfo(FileId file_id, FileInfo* info) {
-  if (!Init())
+  if (!Init(true))
     return false;
   DCHECK(info);
   std::string file_key = GetFileLookupKey(file_id);
   std::string file_data_string;
   leveldb::Status status =
       db_->Get(leveldb::ReadOptions(), file_key, &file_data_string);
   if (status.ok()) {
     return FileInfoFromPickle(
         Pickle(file_data_string.data(), file_data_string.length()), info);
   }
   // Special-case the root, for databases that haven't been initialized yet.
   // Without this, a query for the root's file info, made before creating the
   // first file in the database, will fail and confuse callers.
   if (status.IsNotFound() && !file_id) {
     info->name = FilePath::StringType();
     info->data_path = FilePath();
     info->modification_time = base::Time::Now();
     info->parent_id = 0;
     return true;
   }
   HandleError(FROM_HERE, status);
   return false;
 }
 
 bool FileSystemDirectoryDatabase::AddFileInfo(
     const FileInfo& info, FileId* file_id) {
-  if (!Init())
+  if (!Init(true))
     return false;
   DCHECK(file_id);
   std::string child_key = GetChildLookupKey(info.parent_id, info.name);
   std::string child_id_string;
   leveldb::Status status =
       db_->Get(leveldb::ReadOptions(), child_key, &child_id_string);
   if (status.ok()) {
     LOG(ERROR) << "File exists already!";
     return false;
   }
@@ skipping 21 matching lines @@
   status = db_->Write(leveldb::WriteOptions(), &batch);
   if (!status.ok()) {
     HandleError(FROM_HERE, status);
     return false;
   }
   *file_id = temp_id;
   return true;
 }
 
 bool FileSystemDirectoryDatabase::RemoveFileInfo(FileId file_id) {
-  if (!Init())
+  if (!Init(true))
     return false;
   leveldb::WriteBatch batch;
   if (!RemoveFileInfoHelper(file_id, &batch))
     return false;
   leveldb::Status status = db_->Write(leveldb::WriteOptions(), &batch);
   if (!status.ok()) {
     HandleError(FROM_HERE, status);
     return false;
   }
   return true;
 }
 
 bool FileSystemDirectoryDatabase::UpdateFileInfo(
     FileId file_id, const FileInfo& new_info) {
   // TODO: We should also check to see that this doesn't create a loop, but
   // perhaps only in a debug build.
-  if (!Init())
+  if (!Init(true))
     return false;
   DCHECK(file_id);  // You can't remove the root, ever.  Just delete the DB.
   FileInfo old_info;
   if (!GetFileInfo(file_id, &old_info))
     return false;
   if (old_info.parent_id != new_info.parent_id &&
       !VerifyIsDirectory(new_info.parent_id))
     return false;
   if (old_info.parent_id != new_info.parent_id ||
       old_info.name != new_info.name) {
@@ skipping 63 matching lines @@
                      pickle.size()));
   leveldb::Status status = db_->Write(leveldb::WriteOptions(), &batch);
   if (!status.ok()) {
     HandleError(FROM_HERE, status);
     return false;
   }
   return true;
 }
 
 bool FileSystemDirectoryDatabase::GetNextInteger(int64* next) {
-  if (!Init())
+  if (!Init(true))
     return false;
   DCHECK(next);
   std::string int_string;
   leveldb::Status status =
       db_->Get(leveldb::ReadOptions(), LastIntegerKey(), &int_string);
   if (status.ok()) {
     int64 temp;
     if (!base::StringToInt64(int_string, &temp)) {
       LOG(ERROR) << "Hit database corruption!";
       return false;
@@ skipping 28 matching lines @@
   name = base::SysWideToUTF8(path.value());
 #endif
   leveldb::Status status = leveldb::DestroyDB(name, leveldb::Options());
   if (status.ok())
     return true;
   LOG(WARNING) << "Failed to destroy a database with status " <<
       status.ToString();
   return false;
 }
 
-bool FileSystemDirectoryDatabase::Init() {
- if (db_.get())
-   return true;
+bool FileSystemDirectoryDatabase::Init(bool cleanup_if_corrupted) {
+  if (db_.get())
+    return true;
 
- leveldb::Options options;
- options.create_if_missing = true;
- leveldb::DB* db;
- leveldb::Status status = leveldb::DB::Open(options, path_, &db);
- if (status.ok()) {
-   db_.reset(db);
-   return true;
- }
- HandleError(FROM_HERE, status);
- return false;
+  leveldb::Options options;
+  options.create_if_missing = true;
+  leveldb::DB* db;
+  leveldb::Status status = leveldb::DB::Open(options, path_, &db);
+  // TODO(tzik): Collect status metrics here.
+  if (status.ok()) {
+    db_.reset(db);
+    return true;
+  }
+  HandleError(FROM_HERE, status);
+
+  if (cleanup_if_corrupted && CheckIfDatabaseCorrupted(status)) {
+    LOG(WARNING) << "FileSystem API directory database is corrupted."
+                 << " Attempting cleanup.";
+    if (leveldb::DestroyDB(path_, leveldb::Options()).ok()) {
+      LOG(WARNING) << "FileSystem API directory database cleanup completed."
+                   << " Reopening.";
+      return Init(false);
+    }
+    LOG(WARNING) << "Failed to cleanup FileSystem API directory database.";
+  }
+
+  return false;
+}
+
+bool FileSystemDirectoryDatabase::CheckIfDatabaseCorrupted(

[jsbell, 2012/03/09 23:10:11]

What is the benefit in factoring this method out?

[tzik, 2012/03/10 00:16:55]

I'd initially thought, we have many thing to do here to detect the corruption.
Omitted.

+    const leveldb::Status& status) const {
+  // TODO(tzik): Return status.code() == leveldb::Status::kCorruption, after
+  // leveldb roll.
+  return !status.ok();
 }
 
 bool FileSystemDirectoryDatabase::StoreDefaultValues() {
   // Verify that this is a totally new database, and initialize it.
   scoped_ptr<leveldb::Iterator> iter(db_->NewIterator(leveldb::ReadOptions()));
   iter->SeekToFirst();
   if (iter->Valid()) {  // DB was not empty--we shouldn't have been called.
     LOG(ERROR) << "File system origin database is corrupt!";
     return false;
   }
   // This is always the first write into the database.  If we ever add a
   // version number, it should go in this transaction too.
   FileInfo root;
   root.parent_id = 0;
   root.modification_time = base::Time::Now();
   leveldb::WriteBatch batch;
   if (!AddFileInfoHelper(root, 0, &batch))
     return false;
   batch.Put(LastFileIdKey(), base::Int64ToString(0));
   batch.Put(LastIntegerKey(), base::Int64ToString(-1));
   leveldb::Status status = db_->Write(leveldb::WriteOptions(), &batch);
   if (!status.ok()) {
     HandleError(FROM_HERE, status);
     return false;
   }
   return true;
 }
 
 bool FileSystemDirectoryDatabase::GetLastFileId(FileId* file_id) {
-  if (!Init())
+  if (!Init(true))
     return false;
   DCHECK(file_id);
   std::string id_string;
   leveldb::Status status =
       db_->Get(leveldb::ReadOptions(), LastFileIdKey(), &id_string);
   if (status.ok()) {
     if (!base::StringToInt64(id_string, file_id)) {
       LOG(ERROR) << "Hit database corruption!";
       return false;
     }
@@ skipping 69 matching lines @@
 
 void FileSystemDirectoryDatabase::HandleError(
     const tracked_objects::Location& from_here,
     leveldb::Status status) {
   LOG(ERROR) << "FileSystemDirectoryDatabase failed at: "
              << from_here.ToString() << " with error: " << status.ToString();
   db_.reset();
 }
 
 }  // namespace fileapi