rn-leveldb 3.11.0

package/cpp/leveldb/db/log_writer.h

@@ -0,0 +1,54 @@
+// Copyright (c) 2011 The LevelDB Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file. See the AUTHORS file for names of contributors.
+
+#ifndef STORAGE_LEVELDB_DB_LOG_WRITER_H_
+#define STORAGE_LEVELDB_DB_LOG_WRITER_H_
+
+#include <cstdint>
+
+#include "db/log_format.h"
+#include "leveldb/slice.h"
+#include "leveldb/status.h"
+
+namespace leveldb {
+
+class WritableFile;
+
+namespace log {
+
+class Writer {
+ public:
+  // Create a writer that will append data to "*dest".
+  // "*dest" must be initially empty.
+  // "*dest" must remain live while this Writer is in use.
+  explicit Writer(WritableFile* dest);
+
+  // Create a writer that will append data to "*dest".
+  // "*dest" must have initial length "dest_length".
+  // "*dest" must remain live while this Writer is in use.
+  Writer(WritableFile* dest, uint64_t dest_length);
+
+  Writer(const Writer&) = delete;
+  Writer& operator=(const Writer&) = delete;
+
+  ~Writer();
+
+  Status AddRecord(const Slice& slice);
+
+ private:
+  Status EmitPhysicalRecord(RecordType type, const char* ptr, size_t length);
+
+  WritableFile* dest_;
+  int block_offset_;  // Current offset in block
+
+  // crc32c values for all supported record types.  These are
+  // pre-computed to reduce the overhead of computing the crc of the
+  // record type stored in the header.
+  uint32_t type_crc_[kMaxRecordType + 1];
+};
+
+}  // namespace log
+}  // namespace leveldb
+
+#endif  // STORAGE_LEVELDB_DB_LOG_WRITER_H_

package/cpp/leveldb/db/memtable.cc

@@ -0,0 +1,137 @@
+// Copyright (c) 2011 The LevelDB Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file. See the AUTHORS file for names of contributors.
+
+#include "db/memtable.h"
+#include "db/dbformat.h"
+#include "leveldb/comparator.h"
+#include "leveldb/env.h"
+#include "leveldb/iterator.h"
+#include "util/coding.h"
+
+namespace leveldb {
+
+static Slice GetLengthPrefixedSlice(const char* data) {
+  uint32_t len;
+  const char* p = data;
+  p = GetVarint32Ptr(p, p + 5, &len);  // +5: we assume "p" is not corrupted
+  return Slice(p, len);
+}
+
+MemTable::MemTable(const InternalKeyComparator& comparator)
+    : comparator_(comparator), refs_(0), table_(comparator_, &arena_) {}
+
+MemTable::~MemTable() { assert(refs_ == 0); }
+
+size_t MemTable::ApproximateMemoryUsage() { return arena_.MemoryUsage(); }
+
+int MemTable::KeyComparator::operator()(const char* aptr,
+                                        const char* bptr) const {
+  // Internal keys are encoded as length-prefixed strings.
+  Slice a = GetLengthPrefixedSlice(aptr);
+  Slice b = GetLengthPrefixedSlice(bptr);
+  return comparator.Compare(a, b);
+}
+
+// Encode a suitable internal key target for "target" and return it.
+// Uses *scratch as scratch space, and the returned pointer will point
+// into this scratch space.
+static const char* EncodeKey(std::string* scratch, const Slice& target) {
+  scratch->clear();
+  PutVarint32(scratch, target.size());
+  scratch->append(target.data(), target.size());
+  return scratch->data();
+}
+
+class MemTableIterator : public Iterator {
+ public:
+  explicit MemTableIterator(MemTable::Table* table) : iter_(table) {}
+
+  MemTableIterator(const MemTableIterator&) = delete;
+  MemTableIterator& operator=(const MemTableIterator&) = delete;
+
+  ~MemTableIterator() override = default;
+
+  bool Valid() const override { return iter_.Valid(); }
+  void Seek(const Slice& k) override { iter_.Seek(EncodeKey(&tmp_, k)); }
+  void SeekToFirst() override { iter_.SeekToFirst(); }
+  void SeekToLast() override { iter_.SeekToLast(); }
+  void Next() override { iter_.Next(); }
+  void Prev() override { iter_.Prev(); }
+  Slice key() const override { return GetLengthPrefixedSlice(iter_.key()); }
+  Slice value() const override {
+    Slice key_slice = GetLengthPrefixedSlice(iter_.key());
+    return GetLengthPrefixedSlice(key_slice.data() + key_slice.size());
+  }
+
+  Status status() const override { return Status::OK(); }
+
+ private:
+  MemTable::Table::Iterator iter_;
+  std::string tmp_;  // For passing to EncodeKey
+};
+
+Iterator* MemTable::NewIterator() { return new MemTableIterator(&table_); }
+
+void MemTable::Add(SequenceNumber s, ValueType type, const Slice& key,
+                   const Slice& value) {
+  // Format of an entry is concatenation of:
+  //  key_size     : varint32 of internal_key.size()
+  //  key bytes    : char[internal_key.size()]
+  //  value_size   : varint32 of value.size()
+  //  value bytes  : char[value.size()]
+  size_t key_size = key.size();
+  size_t val_size = value.size();
+  size_t internal_key_size = key_size + 8;
+  const size_t encoded_len = VarintLength(internal_key_size) +
+                             internal_key_size + VarintLength(val_size) +
+                             val_size;
+  char* buf = arena_.Allocate(encoded_len);
+  char* p = EncodeVarint32(buf, internal_key_size);
+  std::memcpy(p, key.data(), key_size);
+  p += key_size;
+  EncodeFixed64(p, (s << 8) | type);
+  p += 8;
+  p = EncodeVarint32(p, val_size);
+  std::memcpy(p, value.data(), val_size);
+  assert(p + val_size == buf + encoded_len);
+  table_.Insert(buf);
+}
+
+bool MemTable::Get(const LookupKey& key, std::string* value, Status* s) {
+  Slice memkey = key.memtable_key();
+  Table::Iterator iter(&table_);
+  iter.Seek(memkey.data());
+  if (iter.Valid()) {
+    // entry format is:
+    //    klength  varint32
+    //    userkey  char[klength]
+    //    tag      uint64
+    //    vlength  varint32
+    //    value    char[vlength]
+    // Check that it belongs to same user key.  We do not check the
+    // sequence number since the Seek() call above should have skipped
+    // all entries with overly large sequence numbers.
+    const char* entry = iter.key();
+    uint32_t key_length;
+    const char* key_ptr = GetVarint32Ptr(entry, entry + 5, &key_length);
+    if (comparator_.comparator.user_comparator()->Compare(
+            Slice(key_ptr, key_length - 8), key.user_key()) == 0) {
+      // Correct user key
+      const uint64_t tag = DecodeFixed64(key_ptr + key_length - 8);
+      switch (static_cast<ValueType>(tag & 0xff)) {
+        case kTypeValue: {
+          Slice v = GetLengthPrefixedSlice(key_ptr + key_length);
+          value->assign(v.data(), v.size());
+          return true;
+        }
+        case kTypeDeletion:
+          *s = Status::NotFound(Slice());
+          return true;
+      }
+    }
+  }
+  return false;
+}
+
+}  // namespace leveldb

package/cpp/leveldb/db/memtable.h

@@ -0,0 +1,87 @@
+// Copyright (c) 2011 The LevelDB Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file. See the AUTHORS file for names of contributors.
+
+#ifndef STORAGE_LEVELDB_DB_MEMTABLE_H_
+#define STORAGE_LEVELDB_DB_MEMTABLE_H_
+
+#include <string>
+
+#include "db/dbformat.h"
+#include "db/skiplist.h"
+#include "leveldb/db.h"
+#include "util/arena.h"
+
+namespace leveldb {
+
+class InternalKeyComparator;
+class MemTableIterator;
+
+class MemTable {
+ public:
+  // MemTables are reference counted.  The initial reference count
+  // is zero and the caller must call Ref() at least once.
+  explicit MemTable(const InternalKeyComparator& comparator);
+
+  MemTable(const MemTable&) = delete;
+  MemTable& operator=(const MemTable&) = delete;
+
+  // Increase reference count.
+  void Ref() { ++refs_; }
+
+  // Drop reference count.  Delete if no more references exist.
+  void Unref() {
+    --refs_;
+    assert(refs_ >= 0);
+    if (refs_ <= 0) {
+      delete this;
+    }
+  }
+
+  // Returns an estimate of the number of bytes of data in use by this
+  // data structure. It is safe to call when MemTable is being modified.
+  size_t ApproximateMemoryUsage();
+
+  // Return an iterator that yields the contents of the memtable.
+  //
+  // The caller must ensure that the underlying MemTable remains live
+  // while the returned iterator is live.  The keys returned by this
+  // iterator are internal keys encoded by AppendInternalKey in the
+  // db/format.{h,cc} module.
+  Iterator* NewIterator();
+
+  // Add an entry into memtable that maps key to value at the
+  // specified sequence number and with the specified type.
+  // Typically value will be empty if type==kTypeDeletion.
+  void Add(SequenceNumber seq, ValueType type, const Slice& key,
+           const Slice& value);
+
+  // If memtable contains a value for key, store it in *value and return true.
+  // If memtable contains a deletion for key, store a NotFound() error
+  // in *status and return true.
+  // Else, return false.
+  bool Get(const LookupKey& key, std::string* value, Status* s);
+
+ private:
+  friend class MemTableIterator;
+  friend class MemTableBackwardIterator;
+
+  struct KeyComparator {
+    const InternalKeyComparator comparator;
+    explicit KeyComparator(const InternalKeyComparator& c) : comparator(c) {}
+    int operator()(const char* a, const char* b) const;
+  };
+
+  typedef SkipList<const char*, KeyComparator> Table;
+
+  ~MemTable();  // Private since only Unref() should be used to delete it
+
+  KeyComparator comparator_;
+  int refs_;
+  Arena arena_;
+  Table table_;
+};
+
+}  // namespace leveldb
+
+#endif  // STORAGE_LEVELDB_DB_MEMTABLE_H_

package/cpp/leveldb/db/recovery_test.cc

@@ -0,0 +1,339 @@
+// Copyright (c) 2014 The LevelDB Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file. See the AUTHORS file for names of contributors.
+
+#include "gtest/gtest.h"
+#include "db/db_impl.h"
+#include "db/filename.h"
+#include "db/version_set.h"
+#include "db/write_batch_internal.h"
+#include "leveldb/db.h"
+#include "leveldb/env.h"
+#include "leveldb/write_batch.h"
+#include "util/logging.h"
+#include "util/testutil.h"
+
+namespace leveldb {
+
+class RecoveryTest : public testing::Test {
+ public:
+  RecoveryTest() : env_(Env::Default()), db_(nullptr) {
+    dbname_ = testing::TempDir() + "/recovery_test";
+    DestroyDB(dbname_, Options());
+    Open();
+  }
+
+  ~RecoveryTest() {
+    Close();
+    DestroyDB(dbname_, Options());
+  }
+
+  DBImpl* dbfull() const { return reinterpret_cast<DBImpl*>(db_); }
+  Env* env() const { return env_; }
+
+  bool CanAppend() {
+    WritableFile* tmp;
+    Status s = env_->NewAppendableFile(CurrentFileName(dbname_), &tmp);
+    delete tmp;
+    if (s.IsNotSupportedError()) {
+      return false;
+    } else {
+      return true;
+    }
+  }
+
+  void Close() {
+    delete db_;
+    db_ = nullptr;
+  }
+
+  Status OpenWithStatus(Options* options = nullptr) {
+    Close();
+    Options opts;
+    if (options != nullptr) {
+      opts = *options;
+    } else {
+      opts.reuse_logs = true;  // TODO(sanjay): test both ways
+      opts.create_if_missing = true;
+    }
+    if (opts.env == nullptr) {
+      opts.env = env_;
+    }
+    return DB::Open(opts, dbname_, &db_);
+  }
+
+  void Open(Options* options = nullptr) {
+    ASSERT_LEVELDB_OK(OpenWithStatus(options));
+    ASSERT_EQ(1, NumLogs());
+  }
+
+  Status Put(const std::string& k, const std::string& v) {
+    return db_->Put(WriteOptions(), k, v);
+  }
+
+  std::string Get(const std::string& k, const Snapshot* snapshot = nullptr) {
+    std::string result;
+    Status s = db_->Get(ReadOptions(), k, &result);
+    if (s.IsNotFound()) {
+      result = "NOT_FOUND";
+    } else if (!s.ok()) {
+      result = s.ToString();
+    }
+    return result;
+  }
+
+  std::string ManifestFileName() {
+    std::string current;
+    EXPECT_LEVELDB_OK(
+        ReadFileToString(env_, CurrentFileName(dbname_), &current));
+    size_t len = current.size();
+    if (len > 0 && current[len - 1] == '\n') {
+      current.resize(len - 1);
+    }
+    return dbname_ + "/" + current;
+  }
+
+  std::string LogName(uint64_t number) { return LogFileName(dbname_, number); }
+
+  size_t RemoveLogFiles() {
+    // Linux allows unlinking open files, but Windows does not.
+    // Closing the db allows for file deletion.
+    Close();
+    std::vector<uint64_t> logs = GetFiles(kLogFile);
+    for (size_t i = 0; i < logs.size(); i++) {
+      EXPECT_LEVELDB_OK(env_->RemoveFile(LogName(logs[i]))) << LogName(logs[i]);
+    }
+    return logs.size();
+  }
+
+  void RemoveManifestFile() {
+    ASSERT_LEVELDB_OK(env_->RemoveFile(ManifestFileName()));
+  }
+
+  uint64_t FirstLogFile() { return GetFiles(kLogFile)[0]; }
+
+  std::vector<uint64_t> GetFiles(FileType t) {
+    std::vector<std::string> filenames;
+    EXPECT_LEVELDB_OK(env_->GetChildren(dbname_, &filenames));
+    std::vector<uint64_t> result;
+    for (size_t i = 0; i < filenames.size(); i++) {
+      uint64_t number;
+      FileType type;
+      if (ParseFileName(filenames[i], &number, &type) && type == t) {
+        result.push_back(number);
+      }
+    }
+    return result;
+  }
+
+  int NumLogs() { return GetFiles(kLogFile).size(); }
+
+  int NumTables() { return GetFiles(kTableFile).size(); }
+
+  uint64_t FileSize(const std::string& fname) {
+    uint64_t result;
+    EXPECT_LEVELDB_OK(env_->GetFileSize(fname, &result)) << fname;
+    return result;
+  }
+
+  void CompactMemTable() { dbfull()->TEST_CompactMemTable(); }
+
+  // Directly construct a log file that sets key to val.
+  void MakeLogFile(uint64_t lognum, SequenceNumber seq, Slice key, Slice val) {
+    std::string fname = LogFileName(dbname_, lognum);
+    WritableFile* file;
+    ASSERT_LEVELDB_OK(env_->NewWritableFile(fname, &file));
+    log::Writer writer(file);
+    WriteBatch batch;
+    batch.Put(key, val);
+    WriteBatchInternal::SetSequence(&batch, seq);
+    ASSERT_LEVELDB_OK(writer.AddRecord(WriteBatchInternal::Contents(&batch)));
+    ASSERT_LEVELDB_OK(file->Flush());
+    delete file;
+  }
+
+ private:
+  std::string dbname_;
+  Env* env_;
+  DB* db_;
+};
+
+TEST_F(RecoveryTest, ManifestReused) {
+  if (!CanAppend()) {
+    std::fprintf(stderr,
+                 "skipping test because env does not support appending\n");
+    return;
+  }
+  ASSERT_LEVELDB_OK(Put("foo", "bar"));
+  Close();
+  std::string old_manifest = ManifestFileName();
+  Open();
+  ASSERT_EQ(old_manifest, ManifestFileName());
+  ASSERT_EQ("bar", Get("foo"));
+  Open();
+  ASSERT_EQ(old_manifest, ManifestFileName());
+  ASSERT_EQ("bar", Get("foo"));
+}
+
+TEST_F(RecoveryTest, LargeManifestCompacted) {
+  if (!CanAppend()) {
+    std::fprintf(stderr,
+                 "skipping test because env does not support appending\n");
+    return;
+  }
+  ASSERT_LEVELDB_OK(Put("foo", "bar"));
+  Close();
+  std::string old_manifest = ManifestFileName();
+
+  // Pad with zeroes to make manifest file very big.
+  {
+    uint64_t len = FileSize(old_manifest);
+    WritableFile* file;
+    ASSERT_LEVELDB_OK(env()->NewAppendableFile(old_manifest, &file));
+    std::string zeroes(3 * 1048576 - static_cast<size_t>(len), 0);
+    ASSERT_LEVELDB_OK(file->Append(zeroes));
+    ASSERT_LEVELDB_OK(file->Flush());
+    delete file;
+  }
+
+  Open();
+  std::string new_manifest = ManifestFileName();
+  ASSERT_NE(old_manifest, new_manifest);
+  ASSERT_GT(10000, FileSize(new_manifest));
+  ASSERT_EQ("bar", Get("foo"));
+
+  Open();
+  ASSERT_EQ(new_manifest, ManifestFileName());
+  ASSERT_EQ("bar", Get("foo"));
+}
+
+TEST_F(RecoveryTest, NoLogFiles) {
+  ASSERT_LEVELDB_OK(Put("foo", "bar"));
+  ASSERT_EQ(1, RemoveLogFiles());
+  Open();
+  ASSERT_EQ("NOT_FOUND", Get("foo"));
+  Open();
+  ASSERT_EQ("NOT_FOUND", Get("foo"));
+}
+
+TEST_F(RecoveryTest, LogFileReuse) {
+  if (!CanAppend()) {
+    std::fprintf(stderr,
+                 "skipping test because env does not support appending\n");
+    return;
+  }
+  for (int i = 0; i < 2; i++) {
+    ASSERT_LEVELDB_OK(Put("foo", "bar"));
+    if (i == 0) {
+      // Compact to ensure current log is empty
+      CompactMemTable();
+    }
+    Close();
+    ASSERT_EQ(1, NumLogs());
+    uint64_t number = FirstLogFile();
+    if (i == 0) {
+      ASSERT_EQ(0, FileSize(LogName(number)));
+    } else {
+      ASSERT_LT(0, FileSize(LogName(number)));
+    }
+    Open();
+    ASSERT_EQ(1, NumLogs());
+    ASSERT_EQ(number, FirstLogFile()) << "did not reuse log file";
+    ASSERT_EQ("bar", Get("foo"));
+    Open();
+    ASSERT_EQ(1, NumLogs());
+    ASSERT_EQ(number, FirstLogFile()) << "did not reuse log file";
+    ASSERT_EQ("bar", Get("foo"));
+  }
+}
+
+TEST_F(RecoveryTest, MultipleMemTables) {
+  // Make a large log.
+  const int kNum = 1000;
+  for (int i = 0; i < kNum; i++) {
+    char buf[100];
+    std::snprintf(buf, sizeof(buf), "%050d", i);
+    ASSERT_LEVELDB_OK(Put(buf, buf));
+  }
+  ASSERT_EQ(0, NumTables());
+  Close();
+  ASSERT_EQ(0, NumTables());
+  ASSERT_EQ(1, NumLogs());
+  uint64_t old_log_file = FirstLogFile();
+
+  // Force creation of multiple memtables by reducing the write buffer size.
+  Options opt;
+  opt.reuse_logs = true;
+  opt.write_buffer_size = (kNum * 100) / 2;
+  Open(&opt);
+  ASSERT_LE(2, NumTables());
+  ASSERT_EQ(1, NumLogs());
+  ASSERT_NE(old_log_file, FirstLogFile()) << "must not reuse log";
+  for (int i = 0; i < kNum; i++) {
+    char buf[100];
+    std::snprintf(buf, sizeof(buf), "%050d", i);
+    ASSERT_EQ(buf, Get(buf));
+  }
+}
+
+TEST_F(RecoveryTest, MultipleLogFiles) {
+  ASSERT_LEVELDB_OK(Put("foo", "bar"));
+  Close();
+  ASSERT_EQ(1, NumLogs());
+
+  // Make a bunch of uncompacted log files.
+  uint64_t old_log = FirstLogFile();
+  MakeLogFile(old_log + 1, 1000, "hello", "world");
+  MakeLogFile(old_log + 2, 1001, "hi", "there");
+  MakeLogFile(old_log + 3, 1002, "foo", "bar2");
+
+  // Recover and check that all log files were processed.
+  Open();
+  ASSERT_LE(1, NumTables());
+  ASSERT_EQ(1, NumLogs());
+  uint64_t new_log = FirstLogFile();
+  ASSERT_LE(old_log + 3, new_log);
+  ASSERT_EQ("bar2", Get("foo"));
+  ASSERT_EQ("world", Get("hello"));
+  ASSERT_EQ("there", Get("hi"));
+
+  // Test that previous recovery produced recoverable state.
+  Open();
+  ASSERT_LE(1, NumTables());
+  ASSERT_EQ(1, NumLogs());
+  if (CanAppend()) {
+    ASSERT_EQ(new_log, FirstLogFile());
+  }
+  ASSERT_EQ("bar2", Get("foo"));
+  ASSERT_EQ("world", Get("hello"));
+  ASSERT_EQ("there", Get("hi"));
+
+  // Check that introducing an older log file does not cause it to be re-read.
+  Close();
+  MakeLogFile(old_log + 1, 2000, "hello", "stale write");
+  Open();
+  ASSERT_LE(1, NumTables());
+  ASSERT_EQ(1, NumLogs());
+  if (CanAppend()) {
+    ASSERT_EQ(new_log, FirstLogFile());
+  }
+  ASSERT_EQ("bar2", Get("foo"));
+  ASSERT_EQ("world", Get("hello"));
+  ASSERT_EQ("there", Get("hi"));
+}
+
+TEST_F(RecoveryTest, ManifestMissing) {
+  ASSERT_LEVELDB_OK(Put("foo", "bar"));
+  Close();
+  RemoveManifestFile();
+
+  Status status = OpenWithStatus();
+  ASSERT_TRUE(status.IsCorruption());
+}
+
+}  // namespace leveldb
+
+int main(int argc, char** argv) {
+  testing::InitGoogleTest(&argc, argv);
+  return RUN_ALL_TESTS();
+}