diff options
Diffstat (limited to 'src/leveldb/db')
45 files changed, 14616 insertions, 0 deletions
diff --git a/src/leveldb/db/autocompact_test.cc b/src/leveldb/db/autocompact_test.cc new file mode 100644 index 0000000000..d20a2362c3 --- /dev/null +++ b/src/leveldb/db/autocompact_test.cc @@ -0,0 +1,118 @@ +// Copyright (c) 2013 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 "leveldb/db.h" +#include "db/db_impl.h" +#include "leveldb/cache.h" +#include "util/testharness.h" +#include "util/testutil.h" + +namespace leveldb { + +class AutoCompactTest { + public: + std::string dbname_; + Cache* tiny_cache_; + Options options_; + DB* db_; + + AutoCompactTest() { + dbname_ = test::TmpDir() + "/autocompact_test"; + tiny_cache_ = NewLRUCache(100); + options_.block_cache = tiny_cache_; + DestroyDB(dbname_, options_); + options_.create_if_missing = true; + options_.compression = kNoCompression; + ASSERT_OK(DB::Open(options_, dbname_, &db_)); + } + + ~AutoCompactTest() { + delete db_; + DestroyDB(dbname_, Options()); + delete tiny_cache_; + } + + std::string Key(int i) { + char buf[100]; + snprintf(buf, sizeof(buf), "key%06d", i); + return std::string(buf); + } + + uint64_t Size(const Slice& start, const Slice& limit) { + Range r(start, limit); + uint64_t size; + db_->GetApproximateSizes(&r, 1, &size); + return size; + } + + void DoReads(int n); +}; + +static const int kValueSize = 200 * 1024; +static const int kTotalSize = 100 * 1024 * 1024; +static const int kCount = kTotalSize / kValueSize; + +// Read through the first n keys repeatedly and check that they get +// compacted (verified by checking the size of the key space). +void AutoCompactTest::DoReads(int n) { + std::string value(kValueSize, 'x'); + DBImpl* dbi = reinterpret_cast<DBImpl*>(db_); + + // Fill database + for (int i = 0; i < kCount; i++) { + ASSERT_OK(db_->Put(WriteOptions(), Key(i), value)); + } + ASSERT_OK(dbi->TEST_CompactMemTable()); + + // Delete everything + for (int i = 0; i < kCount; i++) { + ASSERT_OK(db_->Delete(WriteOptions(), Key(i))); + } + ASSERT_OK(dbi->TEST_CompactMemTable()); + + // Get initial measurement of the space we will be reading. + const int64_t initial_size = Size(Key(0), Key(n)); + const int64_t initial_other_size = Size(Key(n), Key(kCount)); + + // Read until size drops significantly. + std::string limit_key = Key(n); + for (int read = 0; true; read++) { + ASSERT_LT(read, 100) << "Taking too long to compact"; + Iterator* iter = db_->NewIterator(ReadOptions()); + for (iter->SeekToFirst(); + iter->Valid() && iter->key().ToString() < limit_key; + iter->Next()) { + // Drop data + } + delete iter; + // Wait a little bit to allow any triggered compactions to complete. + Env::Default()->SleepForMicroseconds(1000000); + uint64_t size = Size(Key(0), Key(n)); + fprintf(stderr, "iter %3d => %7.3f MB [other %7.3f MB]\n", + read+1, size/1048576.0, Size(Key(n), Key(kCount))/1048576.0); + if (size <= initial_size/10) { + break; + } + } + + // Verify that the size of the key space not touched by the reads + // is pretty much unchanged. + const int64_t final_other_size = Size(Key(n), Key(kCount)); + ASSERT_LE(final_other_size, initial_other_size + 1048576); + ASSERT_GE(final_other_size, initial_other_size/5 - 1048576); +} + +TEST(AutoCompactTest, ReadAll) { + DoReads(kCount); +} + +TEST(AutoCompactTest, ReadHalf) { + DoReads(kCount/2); +} + +} // namespace leveldb + +int main(int argc, char** argv) { + return leveldb::test::RunAllTests(); +} diff --git a/src/leveldb/db/builder.cc b/src/leveldb/db/builder.cc new file mode 100644 index 0000000000..f419882197 --- /dev/null +++ b/src/leveldb/db/builder.cc @@ -0,0 +1,88 @@ +// 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/builder.h" + +#include "db/filename.h" +#include "db/dbformat.h" +#include "db/table_cache.h" +#include "db/version_edit.h" +#include "leveldb/db.h" +#include "leveldb/env.h" +#include "leveldb/iterator.h" + +namespace leveldb { + +Status BuildTable(const std::string& dbname, + Env* env, + const Options& options, + TableCache* table_cache, + Iterator* iter, + FileMetaData* meta) { + Status s; + meta->file_size = 0; + iter->SeekToFirst(); + + std::string fname = TableFileName(dbname, meta->number); + if (iter->Valid()) { + WritableFile* file; + s = env->NewWritableFile(fname, &file); + if (!s.ok()) { + return s; + } + + TableBuilder* builder = new TableBuilder(options, file); + meta->smallest.DecodeFrom(iter->key()); + for (; iter->Valid(); iter->Next()) { + Slice key = iter->key(); + meta->largest.DecodeFrom(key); + builder->Add(key, iter->value()); + } + + // Finish and check for builder errors + if (s.ok()) { + s = builder->Finish(); + if (s.ok()) { + meta->file_size = builder->FileSize(); + assert(meta->file_size > 0); + } + } else { + builder->Abandon(); + } + delete builder; + + // Finish and check for file errors + if (s.ok()) { + s = file->Sync(); + } + if (s.ok()) { + s = file->Close(); + } + delete file; + file = NULL; + + if (s.ok()) { + // Verify that the table is usable + Iterator* it = table_cache->NewIterator(ReadOptions(), + meta->number, + meta->file_size); + s = it->status(); + delete it; + } + } + + // Check for input iterator errors + if (!iter->status().ok()) { + s = iter->status(); + } + + if (s.ok() && meta->file_size > 0) { + // Keep it + } else { + env->DeleteFile(fname); + } + return s; +} + +} // namespace leveldb diff --git a/src/leveldb/db/builder.h b/src/leveldb/db/builder.h new file mode 100644 index 0000000000..62431fcf44 --- /dev/null +++ b/src/leveldb/db/builder.h @@ -0,0 +1,34 @@ +// 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_BUILDER_H_ +#define STORAGE_LEVELDB_DB_BUILDER_H_ + +#include "leveldb/status.h" + +namespace leveldb { + +struct Options; +struct FileMetaData; + +class Env; +class Iterator; +class TableCache; +class VersionEdit; + +// Build a Table file from the contents of *iter. The generated file +// will be named according to meta->number. On success, the rest of +// *meta will be filled with metadata about the generated table. +// If no data is present in *iter, meta->file_size will be set to +// zero, and no Table file will be produced. +extern Status BuildTable(const std::string& dbname, + Env* env, + const Options& options, + TableCache* table_cache, + Iterator* iter, + FileMetaData* meta); + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_BUILDER_H_ diff --git a/src/leveldb/db/c.cc b/src/leveldb/db/c.cc new file mode 100644 index 0000000000..08ff0ad90a --- /dev/null +++ b/src/leveldb/db/c.cc @@ -0,0 +1,595 @@ +// 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 "leveldb/c.h" + +#include <stdlib.h> +#include <unistd.h> +#include "leveldb/cache.h" +#include "leveldb/comparator.h" +#include "leveldb/db.h" +#include "leveldb/env.h" +#include "leveldb/filter_policy.h" +#include "leveldb/iterator.h" +#include "leveldb/options.h" +#include "leveldb/status.h" +#include "leveldb/write_batch.h" + +using leveldb::Cache; +using leveldb::Comparator; +using leveldb::CompressionType; +using leveldb::DB; +using leveldb::Env; +using leveldb::FileLock; +using leveldb::FilterPolicy; +using leveldb::Iterator; +using leveldb::kMajorVersion; +using leveldb::kMinorVersion; +using leveldb::Logger; +using leveldb::NewBloomFilterPolicy; +using leveldb::NewLRUCache; +using leveldb::Options; +using leveldb::RandomAccessFile; +using leveldb::Range; +using leveldb::ReadOptions; +using leveldb::SequentialFile; +using leveldb::Slice; +using leveldb::Snapshot; +using leveldb::Status; +using leveldb::WritableFile; +using leveldb::WriteBatch; +using leveldb::WriteOptions; + +extern "C" { + +struct leveldb_t { DB* rep; }; +struct leveldb_iterator_t { Iterator* rep; }; +struct leveldb_writebatch_t { WriteBatch rep; }; +struct leveldb_snapshot_t { const Snapshot* rep; }; +struct leveldb_readoptions_t { ReadOptions rep; }; +struct leveldb_writeoptions_t { WriteOptions rep; }; +struct leveldb_options_t { Options rep; }; +struct leveldb_cache_t { Cache* rep; }; +struct leveldb_seqfile_t { SequentialFile* rep; }; +struct leveldb_randomfile_t { RandomAccessFile* rep; }; +struct leveldb_writablefile_t { WritableFile* rep; }; +struct leveldb_logger_t { Logger* rep; }; +struct leveldb_filelock_t { FileLock* rep; }; + +struct leveldb_comparator_t : public Comparator { + void* state_; + void (*destructor_)(void*); + int (*compare_)( + void*, + const char* a, size_t alen, + const char* b, size_t blen); + const char* (*name_)(void*); + + virtual ~leveldb_comparator_t() { + (*destructor_)(state_); + } + + virtual int Compare(const Slice& a, const Slice& b) const { + return (*compare_)(state_, a.data(), a.size(), b.data(), b.size()); + } + + virtual const char* Name() const { + return (*name_)(state_); + } + + // No-ops since the C binding does not support key shortening methods. + virtual void FindShortestSeparator(std::string*, const Slice&) const { } + virtual void FindShortSuccessor(std::string* key) const { } +}; + +struct leveldb_filterpolicy_t : public FilterPolicy { + void* state_; + void (*destructor_)(void*); + const char* (*name_)(void*); + char* (*create_)( + void*, + const char* const* key_array, const size_t* key_length_array, + int num_keys, + size_t* filter_length); + unsigned char (*key_match_)( + void*, + const char* key, size_t length, + const char* filter, size_t filter_length); + + virtual ~leveldb_filterpolicy_t() { + (*destructor_)(state_); + } + + virtual const char* Name() const { + return (*name_)(state_); + } + + virtual void CreateFilter(const Slice* keys, int n, std::string* dst) const { + std::vector<const char*> key_pointers(n); + std::vector<size_t> key_sizes(n); + for (int i = 0; i < n; i++) { + key_pointers[i] = keys[i].data(); + key_sizes[i] = keys[i].size(); + } + size_t len; + char* filter = (*create_)(state_, &key_pointers[0], &key_sizes[0], n, &len); + dst->append(filter, len); + free(filter); + } + + virtual bool KeyMayMatch(const Slice& key, const Slice& filter) const { + return (*key_match_)(state_, key.data(), key.size(), + filter.data(), filter.size()); + } +}; + +struct leveldb_env_t { + Env* rep; + bool is_default; +}; + +static bool SaveError(char** errptr, const Status& s) { + assert(errptr != NULL); + if (s.ok()) { + return false; + } else if (*errptr == NULL) { + *errptr = strdup(s.ToString().c_str()); + } else { + // TODO(sanjay): Merge with existing error? + free(*errptr); + *errptr = strdup(s.ToString().c_str()); + } + return true; +} + +static char* CopyString(const std::string& str) { + char* result = reinterpret_cast<char*>(malloc(sizeof(char) * str.size())); + memcpy(result, str.data(), sizeof(char) * str.size()); + return result; +} + +leveldb_t* leveldb_open( + const leveldb_options_t* options, + const char* name, + char** errptr) { + DB* db; + if (SaveError(errptr, DB::Open(options->rep, std::string(name), &db))) { + return NULL; + } + leveldb_t* result = new leveldb_t; + result->rep = db; + return result; +} + +void leveldb_close(leveldb_t* db) { + delete db->rep; + delete db; +} + +void leveldb_put( + leveldb_t* db, + const leveldb_writeoptions_t* options, + const char* key, size_t keylen, + const char* val, size_t vallen, + char** errptr) { + SaveError(errptr, + db->rep->Put(options->rep, Slice(key, keylen), Slice(val, vallen))); +} + +void leveldb_delete( + leveldb_t* db, + const leveldb_writeoptions_t* options, + const char* key, size_t keylen, + char** errptr) { + SaveError(errptr, db->rep->Delete(options->rep, Slice(key, keylen))); +} + + +void leveldb_write( + leveldb_t* db, + const leveldb_writeoptions_t* options, + leveldb_writebatch_t* batch, + char** errptr) { + SaveError(errptr, db->rep->Write(options->rep, &batch->rep)); +} + +char* leveldb_get( + leveldb_t* db, + const leveldb_readoptions_t* options, + const char* key, size_t keylen, + size_t* vallen, + char** errptr) { + char* result = NULL; + std::string tmp; + Status s = db->rep->Get(options->rep, Slice(key, keylen), &tmp); + if (s.ok()) { + *vallen = tmp.size(); + result = CopyString(tmp); + } else { + *vallen = 0; + if (!s.IsNotFound()) { + SaveError(errptr, s); + } + } + return result; +} + +leveldb_iterator_t* leveldb_create_iterator( + leveldb_t* db, + const leveldb_readoptions_t* options) { + leveldb_iterator_t* result = new leveldb_iterator_t; + result->rep = db->rep->NewIterator(options->rep); + return result; +} + +const leveldb_snapshot_t* leveldb_create_snapshot( + leveldb_t* db) { + leveldb_snapshot_t* result = new leveldb_snapshot_t; + result->rep = db->rep->GetSnapshot(); + return result; +} + +void leveldb_release_snapshot( + leveldb_t* db, + const leveldb_snapshot_t* snapshot) { + db->rep->ReleaseSnapshot(snapshot->rep); + delete snapshot; +} + +char* leveldb_property_value( + leveldb_t* db, + const char* propname) { + std::string tmp; + if (db->rep->GetProperty(Slice(propname), &tmp)) { + // We use strdup() since we expect human readable output. + return strdup(tmp.c_str()); + } else { + return NULL; + } +} + +void leveldb_approximate_sizes( + leveldb_t* db, + int num_ranges, + const char* const* range_start_key, const size_t* range_start_key_len, + const char* const* range_limit_key, const size_t* range_limit_key_len, + uint64_t* sizes) { + Range* ranges = new Range[num_ranges]; + for (int i = 0; i < num_ranges; i++) { + ranges[i].start = Slice(range_start_key[i], range_start_key_len[i]); + ranges[i].limit = Slice(range_limit_key[i], range_limit_key_len[i]); + } + db->rep->GetApproximateSizes(ranges, num_ranges, sizes); + delete[] ranges; +} + +void leveldb_compact_range( + leveldb_t* db, + const char* start_key, size_t start_key_len, + const char* limit_key, size_t limit_key_len) { + Slice a, b; + db->rep->CompactRange( + // Pass NULL Slice if corresponding "const char*" is NULL + (start_key ? (a = Slice(start_key, start_key_len), &a) : NULL), + (limit_key ? (b = Slice(limit_key, limit_key_len), &b) : NULL)); +} + +void leveldb_destroy_db( + const leveldb_options_t* options, + const char* name, + char** errptr) { + SaveError(errptr, DestroyDB(name, options->rep)); +} + +void leveldb_repair_db( + const leveldb_options_t* options, + const char* name, + char** errptr) { + SaveError(errptr, RepairDB(name, options->rep)); +} + +void leveldb_iter_destroy(leveldb_iterator_t* iter) { + delete iter->rep; + delete iter; +} + +unsigned char leveldb_iter_valid(const leveldb_iterator_t* iter) { + return iter->rep->Valid(); +} + +void leveldb_iter_seek_to_first(leveldb_iterator_t* iter) { + iter->rep->SeekToFirst(); +} + +void leveldb_iter_seek_to_last(leveldb_iterator_t* iter) { + iter->rep->SeekToLast(); +} + +void leveldb_iter_seek(leveldb_iterator_t* iter, const char* k, size_t klen) { + iter->rep->Seek(Slice(k, klen)); +} + +void leveldb_iter_next(leveldb_iterator_t* iter) { + iter->rep->Next(); +} + +void leveldb_iter_prev(leveldb_iterator_t* iter) { + iter->rep->Prev(); +} + +const char* leveldb_iter_key(const leveldb_iterator_t* iter, size_t* klen) { + Slice s = iter->rep->key(); + *klen = s.size(); + return s.data(); +} + +const char* leveldb_iter_value(const leveldb_iterator_t* iter, size_t* vlen) { + Slice s = iter->rep->value(); + *vlen = s.size(); + return s.data(); +} + +void leveldb_iter_get_error(const leveldb_iterator_t* iter, char** errptr) { + SaveError(errptr, iter->rep->status()); +} + +leveldb_writebatch_t* leveldb_writebatch_create() { + return new leveldb_writebatch_t; +} + +void leveldb_writebatch_destroy(leveldb_writebatch_t* b) { + delete b; +} + +void leveldb_writebatch_clear(leveldb_writebatch_t* b) { + b->rep.Clear(); +} + +void leveldb_writebatch_put( + leveldb_writebatch_t* b, + const char* key, size_t klen, + const char* val, size_t vlen) { + b->rep.Put(Slice(key, klen), Slice(val, vlen)); +} + +void leveldb_writebatch_delete( + leveldb_writebatch_t* b, + const char* key, size_t klen) { + b->rep.Delete(Slice(key, klen)); +} + +void leveldb_writebatch_iterate( + leveldb_writebatch_t* b, + void* state, + void (*put)(void*, const char* k, size_t klen, const char* v, size_t vlen), + void (*deleted)(void*, const char* k, size_t klen)) { + class H : public WriteBatch::Handler { + public: + void* state_; + void (*put_)(void*, const char* k, size_t klen, const char* v, size_t vlen); + void (*deleted_)(void*, const char* k, size_t klen); + virtual void Put(const Slice& key, const Slice& value) { + (*put_)(state_, key.data(), key.size(), value.data(), value.size()); + } + virtual void Delete(const Slice& key) { + (*deleted_)(state_, key.data(), key.size()); + } + }; + H handler; + handler.state_ = state; + handler.put_ = put; + handler.deleted_ = deleted; + b->rep.Iterate(&handler); +} + +leveldb_options_t* leveldb_options_create() { + return new leveldb_options_t; +} + +void leveldb_options_destroy(leveldb_options_t* options) { + delete options; +} + +void leveldb_options_set_comparator( + leveldb_options_t* opt, + leveldb_comparator_t* cmp) { + opt->rep.comparator = cmp; +} + +void leveldb_options_set_filter_policy( + leveldb_options_t* opt, + leveldb_filterpolicy_t* policy) { + opt->rep.filter_policy = policy; +} + +void leveldb_options_set_create_if_missing( + leveldb_options_t* opt, unsigned char v) { + opt->rep.create_if_missing = v; +} + +void leveldb_options_set_error_if_exists( + leveldb_options_t* opt, unsigned char v) { + opt->rep.error_if_exists = v; +} + +void leveldb_options_set_paranoid_checks( + leveldb_options_t* opt, unsigned char v) { + opt->rep.paranoid_checks = v; +} + +void leveldb_options_set_env(leveldb_options_t* opt, leveldb_env_t* env) { + opt->rep.env = (env ? env->rep : NULL); +} + +void leveldb_options_set_info_log(leveldb_options_t* opt, leveldb_logger_t* l) { + opt->rep.info_log = (l ? l->rep : NULL); +} + +void leveldb_options_set_write_buffer_size(leveldb_options_t* opt, size_t s) { + opt->rep.write_buffer_size = s; +} + +void leveldb_options_set_max_open_files(leveldb_options_t* opt, int n) { + opt->rep.max_open_files = n; +} + +void leveldb_options_set_cache(leveldb_options_t* opt, leveldb_cache_t* c) { + opt->rep.block_cache = c->rep; +} + +void leveldb_options_set_block_size(leveldb_options_t* opt, size_t s) { + opt->rep.block_size = s; +} + +void leveldb_options_set_block_restart_interval(leveldb_options_t* opt, int n) { + opt->rep.block_restart_interval = n; +} + +void leveldb_options_set_compression(leveldb_options_t* opt, int t) { + opt->rep.compression = static_cast<CompressionType>(t); +} + +leveldb_comparator_t* leveldb_comparator_create( + void* state, + void (*destructor)(void*), + int (*compare)( + void*, + const char* a, size_t alen, + const char* b, size_t blen), + const char* (*name)(void*)) { + leveldb_comparator_t* result = new leveldb_comparator_t; + result->state_ = state; + result->destructor_ = destructor; + result->compare_ = compare; + result->name_ = name; + return result; +} + +void leveldb_comparator_destroy(leveldb_comparator_t* cmp) { + delete cmp; +} + +leveldb_filterpolicy_t* leveldb_filterpolicy_create( + void* state, + void (*destructor)(void*), + char* (*create_filter)( + void*, + const char* const* key_array, const size_t* key_length_array, + int num_keys, + size_t* filter_length), + unsigned char (*key_may_match)( + void*, + const char* key, size_t length, + const char* filter, size_t filter_length), + const char* (*name)(void*)) { + leveldb_filterpolicy_t* result = new leveldb_filterpolicy_t; + result->state_ = state; + result->destructor_ = destructor; + result->create_ = create_filter; + result->key_match_ = key_may_match; + result->name_ = name; + return result; +} + +void leveldb_filterpolicy_destroy(leveldb_filterpolicy_t* filter) { + delete filter; +} + +leveldb_filterpolicy_t* leveldb_filterpolicy_create_bloom(int bits_per_key) { + // Make a leveldb_filterpolicy_t, but override all of its methods so + // they delegate to a NewBloomFilterPolicy() instead of user + // supplied C functions. + struct Wrapper : public leveldb_filterpolicy_t { + const FilterPolicy* rep_; + ~Wrapper() { delete rep_; } + const char* Name() const { return rep_->Name(); } + void CreateFilter(const Slice* keys, int n, std::string* dst) const { + return rep_->CreateFilter(keys, n, dst); + } + bool KeyMayMatch(const Slice& key, const Slice& filter) const { + return rep_->KeyMayMatch(key, filter); + } + static void DoNothing(void*) { } + }; + Wrapper* wrapper = new Wrapper; + wrapper->rep_ = NewBloomFilterPolicy(bits_per_key); + wrapper->state_ = NULL; + wrapper->destructor_ = &Wrapper::DoNothing; + return wrapper; +} + +leveldb_readoptions_t* leveldb_readoptions_create() { + return new leveldb_readoptions_t; +} + +void leveldb_readoptions_destroy(leveldb_readoptions_t* opt) { + delete opt; +} + +void leveldb_readoptions_set_verify_checksums( + leveldb_readoptions_t* opt, + unsigned char v) { + opt->rep.verify_checksums = v; +} + +void leveldb_readoptions_set_fill_cache( + leveldb_readoptions_t* opt, unsigned char v) { + opt->rep.fill_cache = v; +} + +void leveldb_readoptions_set_snapshot( + leveldb_readoptions_t* opt, + const leveldb_snapshot_t* snap) { + opt->rep.snapshot = (snap ? snap->rep : NULL); +} + +leveldb_writeoptions_t* leveldb_writeoptions_create() { + return new leveldb_writeoptions_t; +} + +void leveldb_writeoptions_destroy(leveldb_writeoptions_t* opt) { + delete opt; +} + +void leveldb_writeoptions_set_sync( + leveldb_writeoptions_t* opt, unsigned char v) { + opt->rep.sync = v; +} + +leveldb_cache_t* leveldb_cache_create_lru(size_t capacity) { + leveldb_cache_t* c = new leveldb_cache_t; + c->rep = NewLRUCache(capacity); + return c; +} + +void leveldb_cache_destroy(leveldb_cache_t* cache) { + delete cache->rep; + delete cache; +} + +leveldb_env_t* leveldb_create_default_env() { + leveldb_env_t* result = new leveldb_env_t; + result->rep = Env::Default(); + result->is_default = true; + return result; +} + +void leveldb_env_destroy(leveldb_env_t* env) { + if (!env->is_default) delete env->rep; + delete env; +} + +void leveldb_free(void* ptr) { + free(ptr); +} + +int leveldb_major_version() { + return kMajorVersion; +} + +int leveldb_minor_version() { + return kMinorVersion; +} + +} // end extern "C" diff --git a/src/leveldb/db/c_test.c b/src/leveldb/db/c_test.c new file mode 100644 index 0000000000..7cd5ee0207 --- /dev/null +++ b/src/leveldb/db/c_test.c @@ -0,0 +1,390 @@ +/* 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 "leveldb/c.h" + +#include <stddef.h> +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <sys/types.h> +#include <unistd.h> + +const char* phase = ""; +static char dbname[200]; + +static void StartPhase(const char* name) { + fprintf(stderr, "=== Test %s\n", name); + phase = name; +} + +static const char* GetTempDir(void) { + const char* ret = getenv("TEST_TMPDIR"); + if (ret == NULL || ret[0] == '\0') + ret = "/tmp"; + return ret; +} + +#define CheckNoError(err) \ + if ((err) != NULL) { \ + fprintf(stderr, "%s:%d: %s: %s\n", __FILE__, __LINE__, phase, (err)); \ + abort(); \ + } + +#define CheckCondition(cond) \ + if (!(cond)) { \ + fprintf(stderr, "%s:%d: %s: %s\n", __FILE__, __LINE__, phase, #cond); \ + abort(); \ + } + +static void CheckEqual(const char* expected, const char* v, size_t n) { + if (expected == NULL && v == NULL) { + // ok + } else if (expected != NULL && v != NULL && n == strlen(expected) && + memcmp(expected, v, n) == 0) { + // ok + return; + } else { + fprintf(stderr, "%s: expected '%s', got '%s'\n", + phase, + (expected ? expected : "(null)"), + (v ? v : "(null")); + abort(); + } +} + +static void Free(char** ptr) { + if (*ptr) { + free(*ptr); + *ptr = NULL; + } +} + +static void CheckGet( + leveldb_t* db, + const leveldb_readoptions_t* options, + const char* key, + const char* expected) { + char* err = NULL; + size_t val_len; + char* val; + val = leveldb_get(db, options, key, strlen(key), &val_len, &err); + CheckNoError(err); + CheckEqual(expected, val, val_len); + Free(&val); +} + +static void CheckIter(leveldb_iterator_t* iter, + const char* key, const char* val) { + size_t len; + const char* str; + str = leveldb_iter_key(iter, &len); + CheckEqual(key, str, len); + str = leveldb_iter_value(iter, &len); + CheckEqual(val, str, len); +} + +// Callback from leveldb_writebatch_iterate() +static void CheckPut(void* ptr, + const char* k, size_t klen, + const char* v, size_t vlen) { + int* state = (int*) ptr; + CheckCondition(*state < 2); + switch (*state) { + case 0: + CheckEqual("bar", k, klen); + CheckEqual("b", v, vlen); + break; + case 1: + CheckEqual("box", k, klen); + CheckEqual("c", v, vlen); + break; + } + (*state)++; +} + +// Callback from leveldb_writebatch_iterate() +static void CheckDel(void* ptr, const char* k, size_t klen) { + int* state = (int*) ptr; + CheckCondition(*state == 2); + CheckEqual("bar", k, klen); + (*state)++; +} + +static void CmpDestroy(void* arg) { } + +static int CmpCompare(void* arg, const char* a, size_t alen, + const char* b, size_t blen) { + int n = (alen < blen) ? alen : blen; + int r = memcmp(a, b, n); + if (r == 0) { + if (alen < blen) r = -1; + else if (alen > blen) r = +1; + } + return r; +} + +static const char* CmpName(void* arg) { + return "foo"; +} + +// Custom filter policy +static unsigned char fake_filter_result = 1; +static void FilterDestroy(void* arg) { } +static const char* FilterName(void* arg) { + return "TestFilter"; +} +static char* FilterCreate( + void* arg, + const char* const* key_array, const size_t* key_length_array, + int num_keys, + size_t* filter_length) { + *filter_length = 4; + char* result = malloc(4); + memcpy(result, "fake", 4); + return result; +} +unsigned char FilterKeyMatch( + void* arg, + const char* key, size_t length, + const char* filter, size_t filter_length) { + CheckCondition(filter_length == 4); + CheckCondition(memcmp(filter, "fake", 4) == 0); + return fake_filter_result; +} + +int main(int argc, char** argv) { + leveldb_t* db; + leveldb_comparator_t* cmp; + leveldb_cache_t* cache; + leveldb_env_t* env; + leveldb_options_t* options; + leveldb_readoptions_t* roptions; + leveldb_writeoptions_t* woptions; + char* err = NULL; + int run = -1; + + CheckCondition(leveldb_major_version() >= 1); + CheckCondition(leveldb_minor_version() >= 1); + + snprintf(dbname, sizeof(dbname), + "%s/leveldb_c_test-%d", + GetTempDir(), + ((int) geteuid())); + + StartPhase("create_objects"); + cmp = leveldb_comparator_create(NULL, CmpDestroy, CmpCompare, CmpName); + env = leveldb_create_default_env(); + cache = leveldb_cache_create_lru(100000); + + options = leveldb_options_create(); + leveldb_options_set_comparator(options, cmp); + leveldb_options_set_error_if_exists(options, 1); + leveldb_options_set_cache(options, cache); + leveldb_options_set_env(options, env); + leveldb_options_set_info_log(options, NULL); + leveldb_options_set_write_buffer_size(options, 100000); + leveldb_options_set_paranoid_checks(options, 1); + leveldb_options_set_max_open_files(options, 10); + leveldb_options_set_block_size(options, 1024); + leveldb_options_set_block_restart_interval(options, 8); + leveldb_options_set_compression(options, leveldb_no_compression); + + roptions = leveldb_readoptions_create(); + leveldb_readoptions_set_verify_checksums(roptions, 1); + leveldb_readoptions_set_fill_cache(roptions, 0); + + woptions = leveldb_writeoptions_create(); + leveldb_writeoptions_set_sync(woptions, 1); + + StartPhase("destroy"); + leveldb_destroy_db(options, dbname, &err); + Free(&err); + + StartPhase("open_error"); + db = leveldb_open(options, dbname, &err); + CheckCondition(err != NULL); + Free(&err); + + StartPhase("leveldb_free"); + db = leveldb_open(options, dbname, &err); + CheckCondition(err != NULL); + leveldb_free(err); + err = NULL; + + StartPhase("open"); + leveldb_options_set_create_if_missing(options, 1); + db = leveldb_open(options, dbname, &err); + CheckNoError(err); + CheckGet(db, roptions, "foo", NULL); + + StartPhase("put"); + leveldb_put(db, woptions, "foo", 3, "hello", 5, &err); + CheckNoError(err); + CheckGet(db, roptions, "foo", "hello"); + + StartPhase("compactall"); + leveldb_compact_range(db, NULL, 0, NULL, 0); + CheckGet(db, roptions, "foo", "hello"); + + StartPhase("compactrange"); + leveldb_compact_range(db, "a", 1, "z", 1); + CheckGet(db, roptions, "foo", "hello"); + + StartPhase("writebatch"); + { + leveldb_writebatch_t* wb = leveldb_writebatch_create(); + leveldb_writebatch_put(wb, "foo", 3, "a", 1); + leveldb_writebatch_clear(wb); + leveldb_writebatch_put(wb, "bar", 3, "b", 1); + leveldb_writebatch_put(wb, "box", 3, "c", 1); + leveldb_writebatch_delete(wb, "bar", 3); + leveldb_write(db, woptions, wb, &err); + CheckNoError(err); + CheckGet(db, roptions, "foo", "hello"); + CheckGet(db, roptions, "bar", NULL); + CheckGet(db, roptions, "box", "c"); + int pos = 0; + leveldb_writebatch_iterate(wb, &pos, CheckPut, CheckDel); + CheckCondition(pos == 3); + leveldb_writebatch_destroy(wb); + } + + StartPhase("iter"); + { + leveldb_iterator_t* iter = leveldb_create_iterator(db, roptions); + CheckCondition(!leveldb_iter_valid(iter)); + leveldb_iter_seek_to_first(iter); + CheckCondition(leveldb_iter_valid(iter)); + CheckIter(iter, "box", "c"); + leveldb_iter_next(iter); + CheckIter(iter, "foo", "hello"); + leveldb_iter_prev(iter); + CheckIter(iter, "box", "c"); + leveldb_iter_prev(iter); + CheckCondition(!leveldb_iter_valid(iter)); + leveldb_iter_seek_to_last(iter); + CheckIter(iter, "foo", "hello"); + leveldb_iter_seek(iter, "b", 1); + CheckIter(iter, "box", "c"); + leveldb_iter_get_error(iter, &err); + CheckNoError(err); + leveldb_iter_destroy(iter); + } + + StartPhase("approximate_sizes"); + { + int i; + int n = 20000; + char keybuf[100]; + char valbuf[100]; + uint64_t sizes[2]; + const char* start[2] = { "a", "k00000000000000010000" }; + size_t start_len[2] = { 1, 21 }; + const char* limit[2] = { "k00000000000000010000", "z" }; + size_t limit_len[2] = { 21, 1 }; + leveldb_writeoptions_set_sync(woptions, 0); + for (i = 0; i < n; i++) { + snprintf(keybuf, sizeof(keybuf), "k%020d", i); + snprintf(valbuf, sizeof(valbuf), "v%020d", i); + leveldb_put(db, woptions, keybuf, strlen(keybuf), valbuf, strlen(valbuf), + &err); + CheckNoError(err); + } + leveldb_approximate_sizes(db, 2, start, start_len, limit, limit_len, sizes); + CheckCondition(sizes[0] > 0); + CheckCondition(sizes[1] > 0); + } + + StartPhase("property"); + { + char* prop = leveldb_property_value(db, "nosuchprop"); + CheckCondition(prop == NULL); + prop = leveldb_property_value(db, "leveldb.stats"); + CheckCondition(prop != NULL); + Free(&prop); + } + + StartPhase("snapshot"); + { + const leveldb_snapshot_t* snap; + snap = leveldb_create_snapshot(db); + leveldb_delete(db, woptions, "foo", 3, &err); + CheckNoError(err); + leveldb_readoptions_set_snapshot(roptions, snap); + CheckGet(db, roptions, "foo", "hello"); + leveldb_readoptions_set_snapshot(roptions, NULL); + CheckGet(db, roptions, "foo", NULL); + leveldb_release_snapshot(db, snap); + } + + StartPhase("repair"); + { + leveldb_close(db); + leveldb_options_set_create_if_missing(options, 0); + leveldb_options_set_error_if_exists(options, 0); + leveldb_repair_db(options, dbname, &err); + CheckNoError(err); + db = leveldb_open(options, dbname, &err); + CheckNoError(err); + CheckGet(db, roptions, "foo", NULL); + CheckGet(db, roptions, "bar", NULL); + CheckGet(db, roptions, "box", "c"); + leveldb_options_set_create_if_missing(options, 1); + leveldb_options_set_error_if_exists(options, 1); + } + + StartPhase("filter"); + for (run = 0; run < 2; run++) { + // First run uses custom filter, second run uses bloom filter + CheckNoError(err); + leveldb_filterpolicy_t* policy; + if (run == 0) { + policy = leveldb_filterpolicy_create( + NULL, FilterDestroy, FilterCreate, FilterKeyMatch, FilterName); + } else { + policy = leveldb_filterpolicy_create_bloom(10); + } + + // Create new database + leveldb_close(db); + leveldb_destroy_db(options, dbname, &err); + leveldb_options_set_filter_policy(options, policy); + db = leveldb_open(options, dbname, &err); + CheckNoError(err); + leveldb_put(db, woptions, "foo", 3, "foovalue", 8, &err); + CheckNoError(err); + leveldb_put(db, woptions, "bar", 3, "barvalue", 8, &err); + CheckNoError(err); + leveldb_compact_range(db, NULL, 0, NULL, 0); + + fake_filter_result = 1; + CheckGet(db, roptions, "foo", "foovalue"); + CheckGet(db, roptions, "bar", "barvalue"); + if (phase == 0) { + // Must not find value when custom filter returns false + fake_filter_result = 0; + CheckGet(db, roptions, "foo", NULL); + CheckGet(db, roptions, "bar", NULL); + fake_filter_result = 1; + + CheckGet(db, roptions, "foo", "foovalue"); + CheckGet(db, roptions, "bar", "barvalue"); + } + leveldb_options_set_filter_policy(options, NULL); + leveldb_filterpolicy_destroy(policy); + } + + StartPhase("cleanup"); + leveldb_close(db); + leveldb_options_destroy(options); + leveldb_readoptions_destroy(roptions); + leveldb_writeoptions_destroy(woptions); + leveldb_cache_destroy(cache); + leveldb_comparator_destroy(cmp); + leveldb_env_destroy(env); + + fprintf(stderr, "PASS\n"); + return 0; +} diff --git a/src/leveldb/db/corruption_test.cc b/src/leveldb/db/corruption_test.cc new file mode 100644 index 0000000000..37a484d25f --- /dev/null +++ b/src/leveldb/db/corruption_test.cc @@ -0,0 +1,374 @@ +// 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 "leveldb/db.h" + +#include <errno.h> +#include <fcntl.h> +#include <sys/stat.h> +#include <sys/types.h> +#include "leveldb/cache.h" +#include "leveldb/env.h" +#include "leveldb/table.h" +#include "leveldb/write_batch.h" +#include "db/db_impl.h" +#include "db/filename.h" +#include "db/log_format.h" +#include "db/version_set.h" +#include "util/logging.h" +#include "util/testharness.h" +#include "util/testutil.h" + +namespace leveldb { + +static const int kValueSize = 1000; + +class CorruptionTest { + public: + test::ErrorEnv env_; + std::string dbname_; + Cache* tiny_cache_; + Options options_; + DB* db_; + + CorruptionTest() { + tiny_cache_ = NewLRUCache(100); + options_.env = &env_; + options_.block_cache = tiny_cache_; + dbname_ = test::TmpDir() + "/corruption_test"; + DestroyDB(dbname_, options_); + + db_ = NULL; + options_.create_if_missing = true; + Reopen(); + options_.create_if_missing = false; + } + + ~CorruptionTest() { + delete db_; + DestroyDB(dbname_, Options()); + delete tiny_cache_; + } + + Status TryReopen() { + delete db_; + db_ = NULL; + return DB::Open(options_, dbname_, &db_); + } + + void Reopen() { + ASSERT_OK(TryReopen()); + } + + void RepairDB() { + delete db_; + db_ = NULL; + ASSERT_OK(::leveldb::RepairDB(dbname_, options_)); + } + + void Build(int n) { + std::string key_space, value_space; + WriteBatch batch; + for (int i = 0; i < n; i++) { + //if ((i % 100) == 0) fprintf(stderr, "@ %d of %d\n", i, n); + Slice key = Key(i, &key_space); + batch.Clear(); + batch.Put(key, Value(i, &value_space)); + WriteOptions options; + // Corrupt() doesn't work without this sync on windows; stat reports 0 for + // the file size. + if (i == n - 1) { + options.sync = true; + } + ASSERT_OK(db_->Write(options, &batch)); + } + } + + void Check(int min_expected, int max_expected) { + int next_expected = 0; + int missed = 0; + int bad_keys = 0; + int bad_values = 0; + int correct = 0; + std::string value_space; + Iterator* iter = db_->NewIterator(ReadOptions()); + for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { + uint64_t key; + Slice in(iter->key()); + if (in == "" || in == "~") { + // Ignore boundary keys. + continue; + } + if (!ConsumeDecimalNumber(&in, &key) || + !in.empty() || + key < next_expected) { + bad_keys++; + continue; + } + missed += (key - next_expected); + next_expected = key + 1; + if (iter->value() != Value(key, &value_space)) { + bad_values++; + } else { + correct++; + } + } + delete iter; + + fprintf(stderr, + "expected=%d..%d; got=%d; bad_keys=%d; bad_values=%d; missed=%d\n", + min_expected, max_expected, correct, bad_keys, bad_values, missed); + ASSERT_LE(min_expected, correct); + ASSERT_GE(max_expected, correct); + } + + void Corrupt(FileType filetype, int offset, int bytes_to_corrupt) { + // Pick file to corrupt + std::vector<std::string> filenames; + ASSERT_OK(env_.GetChildren(dbname_, &filenames)); + uint64_t number; + FileType type; + std::string fname; + int picked_number = -1; + for (size_t i = 0; i < filenames.size(); i++) { + if (ParseFileName(filenames[i], &number, &type) && + type == filetype && + int(number) > picked_number) { // Pick latest file + fname = dbname_ + "/" + filenames[i]; + picked_number = number; + } + } + ASSERT_TRUE(!fname.empty()) << filetype; + + struct stat sbuf; + if (stat(fname.c_str(), &sbuf) != 0) { + const char* msg = strerror(errno); + ASSERT_TRUE(false) << fname << ": " << msg; + } + + if (offset < 0) { + // Relative to end of file; make it absolute + if (-offset > sbuf.st_size) { + offset = 0; + } else { + offset = sbuf.st_size + offset; + } + } + if (offset > sbuf.st_size) { + offset = sbuf.st_size; + } + if (offset + bytes_to_corrupt > sbuf.st_size) { + bytes_to_corrupt = sbuf.st_size - offset; + } + + // Do it + std::string contents; + Status s = ReadFileToString(Env::Default(), fname, &contents); + ASSERT_TRUE(s.ok()) << s.ToString(); + for (int i = 0; i < bytes_to_corrupt; i++) { + contents[i + offset] ^= 0x80; + } + s = WriteStringToFile(Env::Default(), contents, fname); + ASSERT_TRUE(s.ok()) << s.ToString(); + } + + int Property(const std::string& name) { + std::string property; + int result; + if (db_->GetProperty(name, &property) && + sscanf(property.c_str(), "%d", &result) == 1) { + return result; + } else { + return -1; + } + } + + // Return the ith key + Slice Key(int i, std::string* storage) { + char buf[100]; + snprintf(buf, sizeof(buf), "%016d", i); + storage->assign(buf, strlen(buf)); + return Slice(*storage); + } + + // Return the value to associate with the specified key + Slice Value(int k, std::string* storage) { + Random r(k); + return test::RandomString(&r, kValueSize, storage); + } +}; + +TEST(CorruptionTest, Recovery) { + Build(100); + Check(100, 100); + Corrupt(kLogFile, 19, 1); // WriteBatch tag for first record + Corrupt(kLogFile, log::kBlockSize + 1000, 1); // Somewhere in second block + Reopen(); + + // The 64 records in the first two log blocks are completely lost. + Check(36, 36); +} + +TEST(CorruptionTest, RecoverWriteError) { + env_.writable_file_error_ = true; + Status s = TryReopen(); + ASSERT_TRUE(!s.ok()); +} + +TEST(CorruptionTest, NewFileErrorDuringWrite) { + // Do enough writing to force minor compaction + env_.writable_file_error_ = true; + const int num = 3 + (Options().write_buffer_size / kValueSize); + std::string value_storage; + Status s; + for (int i = 0; s.ok() && i < num; i++) { + WriteBatch batch; + batch.Put("a", Value(100, &value_storage)); + s = db_->Write(WriteOptions(), &batch); + } + ASSERT_TRUE(!s.ok()); + ASSERT_GE(env_.num_writable_file_errors_, 1); + env_.writable_file_error_ = false; + Reopen(); +} + +TEST(CorruptionTest, TableFile) { + Build(100); + DBImpl* dbi = reinterpret_cast<DBImpl*>(db_); + dbi->TEST_CompactMemTable(); + dbi->TEST_CompactRange(0, NULL, NULL); + dbi->TEST_CompactRange(1, NULL, NULL); + + Corrupt(kTableFile, 100, 1); + Check(90, 99); +} + +TEST(CorruptionTest, TableFileRepair) { + options_.block_size = 2 * kValueSize; // Limit scope of corruption + options_.paranoid_checks = true; + Reopen(); + Build(100); + DBImpl* dbi = reinterpret_cast<DBImpl*>(db_); + dbi->TEST_CompactMemTable(); + dbi->TEST_CompactRange(0, NULL, NULL); + dbi->TEST_CompactRange(1, NULL, NULL); + + Corrupt(kTableFile, 100, 1); + RepairDB(); + Reopen(); + Check(95, 99); +} + +TEST(CorruptionTest, TableFileIndexData) { + Build(10000); // Enough to build multiple Tables + DBImpl* dbi = reinterpret_cast<DBImpl*>(db_); + dbi->TEST_CompactMemTable(); + + Corrupt(kTableFile, -2000, 500); + Reopen(); + Check(5000, 9999); +} + +TEST(CorruptionTest, MissingDescriptor) { + Build(1000); + RepairDB(); + Reopen(); + Check(1000, 1000); +} + +TEST(CorruptionTest, SequenceNumberRecovery) { + ASSERT_OK(db_->Put(WriteOptions(), "foo", "v1")); + ASSERT_OK(db_->Put(WriteOptions(), "foo", "v2")); + ASSERT_OK(db_->Put(WriteOptions(), "foo", "v3")); + ASSERT_OK(db_->Put(WriteOptions(), "foo", "v4")); + ASSERT_OK(db_->Put(WriteOptions(), "foo", "v5")); + RepairDB(); + Reopen(); + std::string v; + ASSERT_OK(db_->Get(ReadOptions(), "foo", &v)); + ASSERT_EQ("v5", v); + // Write something. If sequence number was not recovered properly, + // it will be hidden by an earlier write. + ASSERT_OK(db_->Put(WriteOptions(), "foo", "v6")); + ASSERT_OK(db_->Get(ReadOptions(), "foo", &v)); + ASSERT_EQ("v6", v); + Reopen(); + ASSERT_OK(db_->Get(ReadOptions(), "foo", &v)); + ASSERT_EQ("v6", v); +} + +TEST(CorruptionTest, CorruptedDescriptor) { + ASSERT_OK(db_->Put(WriteOptions(), "foo", "hello")); + DBImpl* dbi = reinterpret_cast<DBImpl*>(db_); + dbi->TEST_CompactMemTable(); + dbi->TEST_CompactRange(0, NULL, NULL); + + Corrupt(kDescriptorFile, 0, 1000); + Status s = TryReopen(); + ASSERT_TRUE(!s.ok()); + + RepairDB(); + Reopen(); + std::string v; + ASSERT_OK(db_->Get(ReadOptions(), "foo", &v)); + ASSERT_EQ("hello", v); +} + +TEST(CorruptionTest, CompactionInputError) { + Build(10); + DBImpl* dbi = reinterpret_cast<DBImpl*>(db_); + dbi->TEST_CompactMemTable(); + const int last = config::kMaxMemCompactLevel; + ASSERT_EQ(1, Property("leveldb.num-files-at-level" + NumberToString(last))); + + Corrupt(kTableFile, 100, 1); + Check(5, 9); + + // Force compactions by writing lots of values + Build(10000); + Check(10000, 10000); +} + +TEST(CorruptionTest, CompactionInputErrorParanoid) { + options_.paranoid_checks = true; + options_.write_buffer_size = 512 << 10; + Reopen(); + DBImpl* dbi = reinterpret_cast<DBImpl*>(db_); + + // Make multiple inputs so we need to compact. + for (int i = 0; i < 2; i++) { + Build(10); + dbi->TEST_CompactMemTable(); + Corrupt(kTableFile, 100, 1); + env_.SleepForMicroseconds(100000); + } + dbi->CompactRange(NULL, NULL); + + // Write must fail because of corrupted table + std::string tmp1, tmp2; + Status s = db_->Put(WriteOptions(), Key(5, &tmp1), Value(5, &tmp2)); + ASSERT_TRUE(!s.ok()) << "write did not fail in corrupted paranoid db"; +} + +TEST(CorruptionTest, UnrelatedKeys) { + Build(10); + DBImpl* dbi = reinterpret_cast<DBImpl*>(db_); + dbi->TEST_CompactMemTable(); + Corrupt(kTableFile, 100, 1); + + std::string tmp1, tmp2; + ASSERT_OK(db_->Put(WriteOptions(), Key(1000, &tmp1), Value(1000, &tmp2))); + std::string v; + ASSERT_OK(db_->Get(ReadOptions(), Key(1000, &tmp1), &v)); + ASSERT_EQ(Value(1000, &tmp2).ToString(), v); + dbi->TEST_CompactMemTable(); + ASSERT_OK(db_->Get(ReadOptions(), Key(1000, &tmp1), &v)); + ASSERT_EQ(Value(1000, &tmp2).ToString(), v); +} + +} // namespace leveldb + +int main(int argc, char** argv) { + return leveldb::test::RunAllTests(); +} diff --git a/src/leveldb/db/db_bench.cc b/src/leveldb/db/db_bench.cc new file mode 100644 index 0000000000..7a0f5e08cd --- /dev/null +++ b/src/leveldb/db/db_bench.cc @@ -0,0 +1,1000 @@ +// 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 <sys/types.h> +#include <stdio.h> +#include <stdlib.h> +#include "db/db_impl.h" +#include "db/version_set.h" +#include "leveldb/cache.h" +#include "leveldb/db.h" +#include "leveldb/env.h" +#include "leveldb/write_batch.h" +#include "port/port.h" +#include "util/crc32c.h" +#include "util/histogram.h" +#include "util/mutexlock.h" +#include "util/random.h" +#include "util/testutil.h" + +// Comma-separated list of operations to run in the specified order +// Actual benchmarks: +// fillseq -- write N values in sequential key order in async mode +// fillrandom -- write N values in random key order in async mode +// overwrite -- overwrite N values in random key order in async mode +// fillsync -- write N/100 values in random key order in sync mode +// fill100K -- write N/1000 100K values in random order in async mode +// deleteseq -- delete N keys in sequential order +// deleterandom -- delete N keys in random order +// readseq -- read N times sequentially +// readreverse -- read N times in reverse order +// readrandom -- read N times in random order +// readmissing -- read N missing keys in random order +// readhot -- read N times in random order from 1% section of DB +// seekrandom -- N random seeks +// open -- cost of opening a DB +// crc32c -- repeated crc32c of 4K of data +// acquireload -- load N*1000 times +// Meta operations: +// compact -- Compact the entire DB +// stats -- Print DB stats +// sstables -- Print sstable info +// heapprofile -- Dump a heap profile (if supported by this port) +static const char* FLAGS_benchmarks = + "fillseq," + "fillsync," + "fillrandom," + "overwrite," + "readrandom," + "readrandom," // Extra run to allow previous compactions to quiesce + "readseq," + "readreverse," + "compact," + "readrandom," + "readseq," + "readreverse," + "fill100K," + "crc32c," + "snappycomp," + "snappyuncomp," + "acquireload," + ; + +// Number of key/values to place in database +static int FLAGS_num = 1000000; + +// Number of read operations to do. If negative, do FLAGS_num reads. +static int FLAGS_reads = -1; + +// Number of concurrent threads to run. +static int FLAGS_threads = 1; + +// Size of each value +static int FLAGS_value_size = 100; + +// Arrange to generate values that shrink to this fraction of +// their original size after compression +static double FLAGS_compression_ratio = 0.5; + +// Print histogram of operation timings +static bool FLAGS_histogram = false; + +// Number of bytes to buffer in memtable before compacting +// (initialized to default value by "main") +static int FLAGS_write_buffer_size = 0; + +// Number of bytes to use as a cache of uncompressed data. +// Negative means use default settings. +static int FLAGS_cache_size = -1; + +// Maximum number of files to keep open at the same time (use default if == 0) +static int FLAGS_open_files = 0; + +// Bloom filter bits per key. +// Negative means use default settings. +static int FLAGS_bloom_bits = -1; + +// If true, do not destroy the existing database. If you set this +// flag and also specify a benchmark that wants a fresh database, that +// benchmark will fail. +static bool FLAGS_use_existing_db = false; + +// If true, reuse existing log/MANIFEST files when re-opening a database. +static bool FLAGS_reuse_logs = false; + +// Use the db with the following name. +static const char* FLAGS_db = NULL; + +namespace leveldb { + +namespace { + +// Helper for quickly generating random data. +class RandomGenerator { + private: + std::string data_; + int pos_; + + public: + RandomGenerator() { + // We use a limited amount of data over and over again and ensure + // that it is larger than the compression window (32KB), and also + // large enough to serve all typical value sizes we want to write. + Random rnd(301); + std::string piece; + while (data_.size() < 1048576) { + // Add a short fragment that is as compressible as specified + // by FLAGS_compression_ratio. + test::CompressibleString(&rnd, FLAGS_compression_ratio, 100, &piece); + data_.append(piece); + } + pos_ = 0; + } + + Slice Generate(size_t len) { + if (pos_ + len > data_.size()) { + pos_ = 0; + assert(len < data_.size()); + } + pos_ += len; + return Slice(data_.data() + pos_ - len, len); + } +}; + +#if defined(__linux) +static Slice TrimSpace(Slice s) { + size_t start = 0; + while (start < s.size() && isspace(s[start])) { + start++; + } + size_t limit = s.size(); + while (limit > start && isspace(s[limit-1])) { + limit--; + } + return Slice(s.data() + start, limit - start); +} +#endif + +static void AppendWithSpace(std::string* str, Slice msg) { + if (msg.empty()) return; + if (!str->empty()) { + str->push_back(' '); + } + str->append(msg.data(), msg.size()); +} + +class Stats { + private: + double start_; + double finish_; + double seconds_; + int done_; + int next_report_; + int64_t bytes_; + double last_op_finish_; + Histogram hist_; + std::string message_; + + public: + Stats() { Start(); } + + void Start() { + next_report_ = 100; + last_op_finish_ = start_; + hist_.Clear(); + done_ = 0; + bytes_ = 0; + seconds_ = 0; + start_ = Env::Default()->NowMicros(); + finish_ = start_; + message_.clear(); + } + + void Merge(const Stats& other) { + hist_.Merge(other.hist_); + done_ += other.done_; + bytes_ += other.bytes_; + seconds_ += other.seconds_; + if (other.start_ < start_) start_ = other.start_; + if (other.finish_ > finish_) finish_ = other.finish_; + + // Just keep the messages from one thread + if (message_.empty()) message_ = other.message_; + } + + void Stop() { + finish_ = Env::Default()->NowMicros(); + seconds_ = (finish_ - start_) * 1e-6; + } + + void AddMessage(Slice msg) { + AppendWithSpace(&message_, msg); + } + + void FinishedSingleOp() { + if (FLAGS_histogram) { + double now = Env::Default()->NowMicros(); + double micros = now - last_op_finish_; + hist_.Add(micros); + if (micros > 20000) { + fprintf(stderr, "long op: %.1f micros%30s\r", micros, ""); + fflush(stderr); + } + last_op_finish_ = now; + } + + done_++; + if (done_ >= next_report_) { + if (next_report_ < 1000) next_report_ += 100; + else if (next_report_ < 5000) next_report_ += 500; + else if (next_report_ < 10000) next_report_ += 1000; + else if (next_report_ < 50000) next_report_ += 5000; + else if (next_report_ < 100000) next_report_ += 10000; + else if (next_report_ < 500000) next_report_ += 50000; + else next_report_ += 100000; + fprintf(stderr, "... finished %d ops%30s\r", done_, ""); + fflush(stderr); + } + } + + void AddBytes(int64_t n) { + bytes_ += n; + } + + void Report(const Slice& name) { + // Pretend at least one op was done in case we are running a benchmark + // that does not call FinishedSingleOp(). + if (done_ < 1) done_ = 1; + + std::string extra; + if (bytes_ > 0) { + // Rate is computed on actual elapsed time, not the sum of per-thread + // elapsed times. + double elapsed = (finish_ - start_) * 1e-6; + char rate[100]; + snprintf(rate, sizeof(rate), "%6.1f MB/s", + (bytes_ / 1048576.0) / elapsed); + extra = rate; + } + AppendWithSpace(&extra, message_); + + fprintf(stdout, "%-12s : %11.3f micros/op;%s%s\n", + name.ToString().c_str(), + seconds_ * 1e6 / done_, + (extra.empty() ? "" : " "), + extra.c_str()); + if (FLAGS_histogram) { + fprintf(stdout, "Microseconds per op:\n%s\n", hist_.ToString().c_str()); + } + fflush(stdout); + } +}; + +// State shared by all concurrent executions of the same benchmark. +struct SharedState { + port::Mutex mu; + port::CondVar cv; + int total; + + // Each thread goes through the following states: + // (1) initializing + // (2) waiting for others to be initialized + // (3) running + // (4) done + + int num_initialized; + int num_done; + bool start; + + SharedState() : cv(&mu) { } +}; + +// Per-thread state for concurrent executions of the same benchmark. +struct ThreadState { + int tid; // 0..n-1 when running in n threads + Random rand; // Has different seeds for different threads + Stats stats; + SharedState* shared; + + ThreadState(int index) + : tid(index), + rand(1000 + index) { + } +}; + +} // namespace + +class Benchmark { + private: + Cache* cache_; + const FilterPolicy* filter_policy_; + DB* db_; + int num_; + int value_size_; + int entries_per_batch_; + WriteOptions write_options_; + int reads_; + int heap_counter_; + + void PrintHeader() { + const int kKeySize = 16; + PrintEnvironment(); + fprintf(stdout, "Keys: %d bytes each\n", kKeySize); + fprintf(stdout, "Values: %d bytes each (%d bytes after compression)\n", + FLAGS_value_size, + static_cast<int>(FLAGS_value_size * FLAGS_compression_ratio + 0.5)); + fprintf(stdout, "Entries: %d\n", num_); + fprintf(stdout, "RawSize: %.1f MB (estimated)\n", + ((static_cast<int64_t>(kKeySize + FLAGS_value_size) * num_) + / 1048576.0)); + fprintf(stdout, "FileSize: %.1f MB (estimated)\n", + (((kKeySize + FLAGS_value_size * FLAGS_compression_ratio) * num_) + / 1048576.0)); + PrintWarnings(); + fprintf(stdout, "------------------------------------------------\n"); + } + + void PrintWarnings() { +#if defined(__GNUC__) && !defined(__OPTIMIZE__) + fprintf(stdout, + "WARNING: Optimization is disabled: benchmarks unnecessarily slow\n" + ); +#endif +#ifndef NDEBUG + fprintf(stdout, + "WARNING: Assertions are enabled; benchmarks unnecessarily slow\n"); +#endif + + // See if snappy is working by attempting to compress a compressible string + const char text[] = "yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy"; + std::string compressed; + if (!port::Snappy_Compress(text, sizeof(text), &compressed)) { + fprintf(stdout, "WARNING: Snappy compression is not enabled\n"); + } else if (compressed.size() >= sizeof(text)) { + fprintf(stdout, "WARNING: Snappy compression is not effective\n"); + } + } + + void PrintEnvironment() { + fprintf(stderr, "LevelDB: version %d.%d\n", + kMajorVersion, kMinorVersion); + +#if defined(__linux) + time_t now = time(NULL); + fprintf(stderr, "Date: %s", ctime(&now)); // ctime() adds newline + + FILE* cpuinfo = fopen("/proc/cpuinfo", "r"); + if (cpuinfo != NULL) { + char line[1000]; + int num_cpus = 0; + std::string cpu_type; + std::string cache_size; + while (fgets(line, sizeof(line), cpuinfo) != NULL) { + const char* sep = strchr(line, ':'); + if (sep == NULL) { + continue; + } + Slice key = TrimSpace(Slice(line, sep - 1 - line)); + Slice val = TrimSpace(Slice(sep + 1)); + if (key == "model name") { + ++num_cpus; + cpu_type = val.ToString(); + } else if (key == "cache size") { + cache_size = val.ToString(); + } + } + fclose(cpuinfo); + fprintf(stderr, "CPU: %d * %s\n", num_cpus, cpu_type.c_str()); + fprintf(stderr, "CPUCache: %s\n", cache_size.c_str()); + } +#endif + } + + public: + Benchmark() + : cache_(FLAGS_cache_size >= 0 ? NewLRUCache(FLAGS_cache_size) : NULL), + filter_policy_(FLAGS_bloom_bits >= 0 + ? NewBloomFilterPolicy(FLAGS_bloom_bits) + : NULL), + db_(NULL), + num_(FLAGS_num), + value_size_(FLAGS_value_size), + entries_per_batch_(1), + reads_(FLAGS_reads < 0 ? FLAGS_num : FLAGS_reads), + heap_counter_(0) { + std::vector<std::string> files; + Env::Default()->GetChildren(FLAGS_db, &files); + for (size_t i = 0; i < files.size(); i++) { + if (Slice(files[i]).starts_with("heap-")) { + Env::Default()->DeleteFile(std::string(FLAGS_db) + "/" + files[i]); + } + } + if (!FLAGS_use_existing_db) { + DestroyDB(FLAGS_db, Options()); + } + } + + ~Benchmark() { + delete db_; + delete cache_; + delete filter_policy_; + } + + void Run() { + PrintHeader(); + Open(); + + const char* benchmarks = FLAGS_benchmarks; + while (benchmarks != NULL) { + const char* sep = strchr(benchmarks, ','); + Slice name; + if (sep == NULL) { + name = benchmarks; + benchmarks = NULL; + } else { + name = Slice(benchmarks, sep - benchmarks); + benchmarks = sep + 1; + } + + // Reset parameters that may be overridden below + num_ = FLAGS_num; + reads_ = (FLAGS_reads < 0 ? FLAGS_num : FLAGS_reads); + value_size_ = FLAGS_value_size; + entries_per_batch_ = 1; + write_options_ = WriteOptions(); + + void (Benchmark::*method)(ThreadState*) = NULL; + bool fresh_db = false; + int num_threads = FLAGS_threads; + + if (name == Slice("open")) { + method = &Benchmark::OpenBench; + num_ /= 10000; + if (num_ < 1) num_ = 1; + } else if (name == Slice("fillseq")) { + fresh_db = true; + method = &Benchmark::WriteSeq; + } else if (name == Slice("fillbatch")) { + fresh_db = true; + entries_per_batch_ = 1000; + method = &Benchmark::WriteSeq; + } else if (name == Slice("fillrandom")) { + fresh_db = true; + method = &Benchmark::WriteRandom; + } else if (name == Slice("overwrite")) { + fresh_db = false; + method = &Benchmark::WriteRandom; + } else if (name == Slice("fillsync")) { + fresh_db = true; + num_ /= 1000; + write_options_.sync = true; + method = &Benchmark::WriteRandom; + } else if (name == Slice("fill100K")) { + fresh_db = true; + num_ /= 1000; + value_size_ = 100 * 1000; + method = &Benchmark::WriteRandom; + } else if (name == Slice("readseq")) { + method = &Benchmark::ReadSequential; + } else if (name == Slice("readreverse")) { + method = &Benchmark::ReadReverse; + } else if (name == Slice("readrandom")) { + method = &Benchmark::ReadRandom; + } else if (name == Slice("readmissing")) { + method = &Benchmark::ReadMissing; + } else if (name == Slice("seekrandom")) { + method = &Benchmark::SeekRandom; + } else if (name == Slice("readhot")) { + method = &Benchmark::ReadHot; + } else if (name == Slice("readrandomsmall")) { + reads_ /= 1000; + method = &Benchmark::ReadRandom; + } else if (name == Slice("deleteseq")) { + method = &Benchmark::DeleteSeq; + } else if (name == Slice("deleterandom")) { + method = &Benchmark::DeleteRandom; + } else if (name == Slice("readwhilewriting")) { + num_threads++; // Add extra thread for writing + method = &Benchmark::ReadWhileWriting; + } else if (name == Slice("compact")) { + method = &Benchmark::Compact; + } else if (name == Slice("crc32c")) { + method = &Benchmark::Crc32c; + } else if (name == Slice("acquireload")) { + method = &Benchmark::AcquireLoad; + } else if (name == Slice("snappycomp")) { + method = &Benchmark::SnappyCompress; + } else if (name == Slice("snappyuncomp")) { + method = &Benchmark::SnappyUncompress; + } else if (name == Slice("heapprofile")) { + HeapProfile(); + } else if (name == Slice("stats")) { + PrintStats("leveldb.stats"); + } else if (name == Slice("sstables")) { + PrintStats("leveldb.sstables"); + } else { + if (name != Slice()) { // No error message for empty name + fprintf(stderr, "unknown benchmark '%s'\n", name.ToString().c_str()); + } + } + + if (fresh_db) { + if (FLAGS_use_existing_db) { + fprintf(stdout, "%-12s : skipped (--use_existing_db is true)\n", + name.ToString().c_str()); + method = NULL; + } else { + delete db_; + db_ = NULL; + DestroyDB(FLAGS_db, Options()); + Open(); + } + } + + if (method != NULL) { + RunBenchmark(num_threads, name, method); + } + } + } + + private: + struct ThreadArg { + Benchmark* bm; + SharedState* shared; + ThreadState* thread; + void (Benchmark::*method)(ThreadState*); + }; + + static void ThreadBody(void* v) { + ThreadArg* arg = reinterpret_cast<ThreadArg*>(v); + SharedState* shared = arg->shared; + ThreadState* thread = arg->thread; + { + MutexLock l(&shared->mu); + shared->num_initialized++; + if (shared->num_initialized >= shared->total) { + shared->cv.SignalAll(); + } + while (!shared->start) { + shared->cv.Wait(); + } + } + + thread->stats.Start(); + (arg->bm->*(arg->method))(thread); + thread->stats.Stop(); + + { + MutexLock l(&shared->mu); + shared->num_done++; + if (shared->num_done >= shared->total) { + shared->cv.SignalAll(); + } + } + } + + void RunBenchmark(int n, Slice name, + void (Benchmark::*method)(ThreadState*)) { + SharedState shared; + shared.total = n; + shared.num_initialized = 0; + shared.num_done = 0; + shared.start = false; + + ThreadArg* arg = new ThreadArg[n]; + for (int i = 0; i < n; i++) { + arg[i].bm = this; + arg[i].method = method; + arg[i].shared = &shared; + arg[i].thread = new ThreadState(i); + arg[i].thread->shared = &shared; + Env::Default()->StartThread(ThreadBody, &arg[i]); + } + + shared.mu.Lock(); + while (shared.num_initialized < n) { + shared.cv.Wait(); + } + + shared.start = true; + shared.cv.SignalAll(); + while (shared.num_done < n) { + shared.cv.Wait(); + } + shared.mu.Unlock(); + + for (int i = 1; i < n; i++) { + arg[0].thread->stats.Merge(arg[i].thread->stats); + } + arg[0].thread->stats.Report(name); + + for (int i = 0; i < n; i++) { + delete arg[i].thread; + } + delete[] arg; + } + + void Crc32c(ThreadState* thread) { + // Checksum about 500MB of data total + const int size = 4096; + const char* label = "(4K per op)"; + std::string data(size, 'x'); + int64_t bytes = 0; + uint32_t crc = 0; + while (bytes < 500 * 1048576) { + crc = crc32c::Value(data.data(), size); + thread->stats.FinishedSingleOp(); + bytes += size; + } + // Print so result is not dead + fprintf(stderr, "... crc=0x%x\r", static_cast<unsigned int>(crc)); + + thread->stats.AddBytes(bytes); + thread->stats.AddMessage(label); + } + + void AcquireLoad(ThreadState* thread) { + int dummy; + port::AtomicPointer ap(&dummy); + int count = 0; + void *ptr = NULL; + thread->stats.AddMessage("(each op is 1000 loads)"); + while (count < 100000) { + for (int i = 0; i < 1000; i++) { + ptr = ap.Acquire_Load(); + } + count++; + thread->stats.FinishedSingleOp(); + } + if (ptr == NULL) exit(1); // Disable unused variable warning. + } + + void SnappyCompress(ThreadState* thread) { + RandomGenerator gen; + Slice input = gen.Generate(Options().block_size); + int64_t bytes = 0; + int64_t produced = 0; + bool ok = true; + std::string compressed; + while (ok && bytes < 1024 * 1048576) { // Compress 1G + ok = port::Snappy_Compress(input.data(), input.size(), &compressed); + produced += compressed.size(); + bytes += input.size(); + thread->stats.FinishedSingleOp(); + } + + if (!ok) { + thread->stats.AddMessage("(snappy failure)"); + } else { + char buf[100]; + snprintf(buf, sizeof(buf), "(output: %.1f%%)", + (produced * 100.0) / bytes); + thread->stats.AddMessage(buf); + thread->stats.AddBytes(bytes); + } + } + + void SnappyUncompress(ThreadState* thread) { + RandomGenerator gen; + Slice input = gen.Generate(Options().block_size); + std::string compressed; + bool ok = port::Snappy_Compress(input.data(), input.size(), &compressed); + int64_t bytes = 0; + char* uncompressed = new char[input.size()]; + while (ok && bytes < 1024 * 1048576) { // Compress 1G + ok = port::Snappy_Uncompress(compressed.data(), compressed.size(), + uncompressed); + bytes += input.size(); + thread->stats.FinishedSingleOp(); + } + delete[] uncompressed; + + if (!ok) { + thread->stats.AddMessage("(snappy failure)"); + } else { + thread->stats.AddBytes(bytes); + } + } + + void Open() { + assert(db_ == NULL); + Options options; + options.create_if_missing = !FLAGS_use_existing_db; + options.block_cache = cache_; + options.write_buffer_size = FLAGS_write_buffer_size; + options.max_open_files = FLAGS_open_files; + options.filter_policy = filter_policy_; + options.reuse_logs = FLAGS_reuse_logs; + Status s = DB::Open(options, FLAGS_db, &db_); + if (!s.ok()) { + fprintf(stderr, "open error: %s\n", s.ToString().c_str()); + exit(1); + } + } + + void OpenBench(ThreadState* thread) { + for (int i = 0; i < num_; i++) { + delete db_; + Open(); + thread->stats.FinishedSingleOp(); + } + } + + void WriteSeq(ThreadState* thread) { + DoWrite(thread, true); + } + + void WriteRandom(ThreadState* thread) { + DoWrite(thread, false); + } + + void DoWrite(ThreadState* thread, bool seq) { + if (num_ != FLAGS_num) { + char msg[100]; + snprintf(msg, sizeof(msg), "(%d ops)", num_); + thread->stats.AddMessage(msg); + } + + RandomGenerator gen; + WriteBatch batch; + Status s; + int64_t bytes = 0; + for (int i = 0; i < num_; i += entries_per_batch_) { + batch.Clear(); + for (int j = 0; j < entries_per_batch_; j++) { + const int k = seq ? i+j : (thread->rand.Next() % FLAGS_num); + char key[100]; + snprintf(key, sizeof(key), "%016d", k); + batch.Put(key, gen.Generate(value_size_)); + bytes += value_size_ + strlen(key); + thread->stats.FinishedSingleOp(); + } + s = db_->Write(write_options_, &batch); + if (!s.ok()) { + fprintf(stderr, "put error: %s\n", s.ToString().c_str()); + exit(1); + } + } + thread->stats.AddBytes(bytes); + } + + void ReadSequential(ThreadState* thread) { + Iterator* iter = db_->NewIterator(ReadOptions()); + int i = 0; + int64_t bytes = 0; + for (iter->SeekToFirst(); i < reads_ && iter->Valid(); iter->Next()) { + bytes += iter->key().size() + iter->value().size(); + thread->stats.FinishedSingleOp(); + ++i; + } + delete iter; + thread->stats.AddBytes(bytes); + } + + void ReadReverse(ThreadState* thread) { + Iterator* iter = db_->NewIterator(ReadOptions()); + int i = 0; + int64_t bytes = 0; + for (iter->SeekToLast(); i < reads_ && iter->Valid(); iter->Prev()) { + bytes += iter->key().size() + iter->value().size(); + thread->stats.FinishedSingleOp(); + ++i; + } + delete iter; + thread->stats.AddBytes(bytes); + } + + void ReadRandom(ThreadState* thread) { + ReadOptions options; + std::string value; + int found = 0; + for (int i = 0; i < reads_; i++) { + char key[100]; + const int k = thread->rand.Next() % FLAGS_num; + snprintf(key, sizeof(key), "%016d", k); + if (db_->Get(options, key, &value).ok()) { + found++; + } + thread->stats.FinishedSingleOp(); + } + char msg[100]; + snprintf(msg, sizeof(msg), "(%d of %d found)", found, num_); + thread->stats.AddMessage(msg); + } + + void ReadMissing(ThreadState* thread) { + ReadOptions options; + std::string value; + for (int i = 0; i < reads_; i++) { + char key[100]; + const int k = thread->rand.Next() % FLAGS_num; + snprintf(key, sizeof(key), "%016d.", k); + db_->Get(options, key, &value); + thread->stats.FinishedSingleOp(); + } + } + + void ReadHot(ThreadState* thread) { + ReadOptions options; + std::string value; + const int range = (FLAGS_num + 99) / 100; + for (int i = 0; i < reads_; i++) { + char key[100]; + const int k = thread->rand.Next() % range; + snprintf(key, sizeof(key), "%016d", k); + db_->Get(options, key, &value); + thread->stats.FinishedSingleOp(); + } + } + + void SeekRandom(ThreadState* thread) { + ReadOptions options; + int found = 0; + for (int i = 0; i < reads_; i++) { + Iterator* iter = db_->NewIterator(options); + char key[100]; + const int k = thread->rand.Next() % FLAGS_num; + snprintf(key, sizeof(key), "%016d", k); + iter->Seek(key); + if (iter->Valid() && iter->key() == key) found++; + delete iter; + thread->stats.FinishedSingleOp(); + } + char msg[100]; + snprintf(msg, sizeof(msg), "(%d of %d found)", found, num_); + thread->stats.AddMessage(msg); + } + + void DoDelete(ThreadState* thread, bool seq) { + RandomGenerator gen; + WriteBatch batch; + Status s; + for (int i = 0; i < num_; i += entries_per_batch_) { + batch.Clear(); + for (int j = 0; j < entries_per_batch_; j++) { + const int k = seq ? i+j : (thread->rand.Next() % FLAGS_num); + char key[100]; + snprintf(key, sizeof(key), "%016d", k); + batch.Delete(key); + thread->stats.FinishedSingleOp(); + } + s = db_->Write(write_options_, &batch); + if (!s.ok()) { + fprintf(stderr, "del error: %s\n", s.ToString().c_str()); + exit(1); + } + } + } + + void DeleteSeq(ThreadState* thread) { + DoDelete(thread, true); + } + + void DeleteRandom(ThreadState* thread) { + DoDelete(thread, false); + } + + void ReadWhileWriting(ThreadState* thread) { + if (thread->tid > 0) { + ReadRandom(thread); + } else { + // Special thread that keeps writing until other threads are done. + RandomGenerator gen; + while (true) { + { + MutexLock l(&thread->shared->mu); + if (thread->shared->num_done + 1 >= thread->shared->num_initialized) { + // Other threads have finished + break; + } + } + + const int k = thread->rand.Next() % FLAGS_num; + char key[100]; + snprintf(key, sizeof(key), "%016d", k); + Status s = db_->Put(write_options_, key, gen.Generate(value_size_)); + if (!s.ok()) { + fprintf(stderr, "put error: %s\n", s.ToString().c_str()); + exit(1); + } + } + + // Do not count any of the preceding work/delay in stats. + thread->stats.Start(); + } + } + + void Compact(ThreadState* thread) { + db_->CompactRange(NULL, NULL); + } + + void PrintStats(const char* key) { + std::string stats; + if (!db_->GetProperty(key, &stats)) { + stats = "(failed)"; + } + fprintf(stdout, "\n%s\n", stats.c_str()); + } + + static void WriteToFile(void* arg, const char* buf, int n) { + reinterpret_cast<WritableFile*>(arg)->Append(Slice(buf, n)); + } + + void HeapProfile() { + char fname[100]; + snprintf(fname, sizeof(fname), "%s/heap-%04d", FLAGS_db, ++heap_counter_); + WritableFile* file; + Status s = Env::Default()->NewWritableFile(fname, &file); + if (!s.ok()) { + fprintf(stderr, "%s\n", s.ToString().c_str()); + return; + } + bool ok = port::GetHeapProfile(WriteToFile, file); + delete file; + if (!ok) { + fprintf(stderr, "heap profiling not supported\n"); + Env::Default()->DeleteFile(fname); + } + } +}; + +} // namespace leveldb + +int main(int argc, char** argv) { + FLAGS_write_buffer_size = leveldb::Options().write_buffer_size; + FLAGS_open_files = leveldb::Options().max_open_files; + std::string default_db_path; + + for (int i = 1; i < argc; i++) { + double d; + int n; + char junk; + if (leveldb::Slice(argv[i]).starts_with("--benchmarks=")) { + FLAGS_benchmarks = argv[i] + strlen("--benchmarks="); + } else if (sscanf(argv[i], "--compression_ratio=%lf%c", &d, &junk) == 1) { + FLAGS_compression_ratio = d; + } else if (sscanf(argv[i], "--histogram=%d%c", &n, &junk) == 1 && + (n == 0 || n == 1)) { + FLAGS_histogram = n; + } else if (sscanf(argv[i], "--use_existing_db=%d%c", &n, &junk) == 1 && + (n == 0 || n == 1)) { + FLAGS_use_existing_db = n; + } else if (sscanf(argv[i], "--reuse_logs=%d%c", &n, &junk) == 1 && + (n == 0 || n == 1)) { + FLAGS_reuse_logs = n; + } else if (sscanf(argv[i], "--num=%d%c", &n, &junk) == 1) { + FLAGS_num = n; + } else if (sscanf(argv[i], "--reads=%d%c", &n, &junk) == 1) { + FLAGS_reads = n; + } else if (sscanf(argv[i], "--threads=%d%c", &n, &junk) == 1) { + FLAGS_threads = n; + } else if (sscanf(argv[i], "--value_size=%d%c", &n, &junk) == 1) { + FLAGS_value_size = n; + } else if (sscanf(argv[i], "--write_buffer_size=%d%c", &n, &junk) == 1) { + FLAGS_write_buffer_size = n; + } else if (sscanf(argv[i], "--cache_size=%d%c", &n, &junk) == 1) { + FLAGS_cache_size = n; + } else if (sscanf(argv[i], "--bloom_bits=%d%c", &n, &junk) == 1) { + FLAGS_bloom_bits = n; + } else if (sscanf(argv[i], "--open_files=%d%c", &n, &junk) == 1) { + FLAGS_open_files = n; + } else if (strncmp(argv[i], "--db=", 5) == 0) { + FLAGS_db = argv[i] + 5; + } else { + fprintf(stderr, "Invalid flag '%s'\n", argv[i]); + exit(1); + } + } + + // Choose a location for the test database if none given with --db=<path> + if (FLAGS_db == NULL) { + leveldb::Env::Default()->GetTestDirectory(&default_db_path); + default_db_path += "/dbbench"; + FLAGS_db = default_db_path.c_str(); + } + + leveldb::Benchmark benchmark; + benchmark.Run(); + return 0; +} diff --git a/src/leveldb/db/db_impl.cc b/src/leveldb/db/db_impl.cc new file mode 100644 index 0000000000..60f4e66e55 --- /dev/null +++ b/src/leveldb/db/db_impl.cc @@ -0,0 +1,1567 @@ +// 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/db_impl.h" + +#include <algorithm> +#include <set> +#include <string> +#include <stdint.h> +#include <stdio.h> +#include <vector> +#include "db/builder.h" +#include "db/db_iter.h" +#include "db/dbformat.h" +#include "db/filename.h" +#include "db/log_reader.h" +#include "db/log_writer.h" +#include "db/memtable.h" +#include "db/table_cache.h" +#include "db/version_set.h" +#include "db/write_batch_internal.h" +#include "leveldb/db.h" +#include "leveldb/env.h" +#include "leveldb/status.h" +#include "leveldb/table.h" +#include "leveldb/table_builder.h" +#include "port/port.h" +#include "table/block.h" +#include "table/merger.h" +#include "table/two_level_iterator.h" +#include "util/coding.h" +#include "util/logging.h" +#include "util/mutexlock.h" + +namespace leveldb { + +const int kNumNonTableCacheFiles = 10; + +// Information kept for every waiting writer +struct DBImpl::Writer { + Status status; + WriteBatch* batch; + bool sync; + bool done; + port::CondVar cv; + + explicit Writer(port::Mutex* mu) : cv(mu) { } +}; + +struct DBImpl::CompactionState { + Compaction* const compaction; + + // Sequence numbers < smallest_snapshot are not significant since we + // will never have to service a snapshot below smallest_snapshot. + // Therefore if we have seen a sequence number S <= smallest_snapshot, + // we can drop all entries for the same key with sequence numbers < S. + SequenceNumber smallest_snapshot; + + // Files produced by compaction + struct Output { + uint64_t number; + uint64_t file_size; + InternalKey smallest, largest; + }; + std::vector<Output> outputs; + + // State kept for output being generated + WritableFile* outfile; + TableBuilder* builder; + + uint64_t total_bytes; + + Output* current_output() { return &outputs[outputs.size()-1]; } + + explicit CompactionState(Compaction* c) + : compaction(c), + outfile(NULL), + builder(NULL), + total_bytes(0) { + } +}; + +// Fix user-supplied options to be reasonable +template <class T,class V> +static void ClipToRange(T* ptr, V minvalue, V maxvalue) { + if (static_cast<V>(*ptr) > maxvalue) *ptr = maxvalue; + if (static_cast<V>(*ptr) < minvalue) *ptr = minvalue; +} +Options SanitizeOptions(const std::string& dbname, + const InternalKeyComparator* icmp, + const InternalFilterPolicy* ipolicy, + const Options& src) { + Options result = src; + result.comparator = icmp; + result.filter_policy = (src.filter_policy != NULL) ? ipolicy : NULL; + ClipToRange(&result.max_open_files, 64 + kNumNonTableCacheFiles, 50000); + ClipToRange(&result.write_buffer_size, 64<<10, 1<<30); + ClipToRange(&result.block_size, 1<<10, 4<<20); + if (result.info_log == NULL) { + // Open a log file in the same directory as the db + src.env->CreateDir(dbname); // In case it does not exist + src.env->RenameFile(InfoLogFileName(dbname), OldInfoLogFileName(dbname)); + Status s = src.env->NewLogger(InfoLogFileName(dbname), &result.info_log); + if (!s.ok()) { + // No place suitable for logging + result.info_log = NULL; + } + } + if (result.block_cache == NULL) { + result.block_cache = NewLRUCache(8 << 20); + } + return result; +} + +DBImpl::DBImpl(const Options& raw_options, const std::string& dbname) + : env_(raw_options.env), + internal_comparator_(raw_options.comparator), + internal_filter_policy_(raw_options.filter_policy), + options_(SanitizeOptions(dbname, &internal_comparator_, + &internal_filter_policy_, raw_options)), + owns_info_log_(options_.info_log != raw_options.info_log), + owns_cache_(options_.block_cache != raw_options.block_cache), + dbname_(dbname), + db_lock_(NULL), + shutting_down_(NULL), + bg_cv_(&mutex_), + mem_(NULL), + imm_(NULL), + logfile_(NULL), + logfile_number_(0), + log_(NULL), + seed_(0), + tmp_batch_(new WriteBatch), + bg_compaction_scheduled_(false), + manual_compaction_(NULL) { + has_imm_.Release_Store(NULL); + + // Reserve ten files or so for other uses and give the rest to TableCache. + const int table_cache_size = options_.max_open_files - kNumNonTableCacheFiles; + table_cache_ = new TableCache(dbname_, &options_, table_cache_size); + + versions_ = new VersionSet(dbname_, &options_, table_cache_, + &internal_comparator_); +} + +DBImpl::~DBImpl() { + // Wait for background work to finish + mutex_.Lock(); + shutting_down_.Release_Store(this); // Any non-NULL value is ok + while (bg_compaction_scheduled_) { + bg_cv_.Wait(); + } + mutex_.Unlock(); + + if (db_lock_ != NULL) { + env_->UnlockFile(db_lock_); + } + + delete versions_; + if (mem_ != NULL) mem_->Unref(); + if (imm_ != NULL) imm_->Unref(); + delete tmp_batch_; + delete log_; + delete logfile_; + delete table_cache_; + + if (owns_info_log_) { + delete options_.info_log; + } + if (owns_cache_) { + delete options_.block_cache; + } +} + +Status DBImpl::NewDB() { + VersionEdit new_db; + new_db.SetComparatorName(user_comparator()->Name()); + new_db.SetLogNumber(0); + new_db.SetNextFile(2); + new_db.SetLastSequence(0); + + const std::string manifest = DescriptorFileName(dbname_, 1); + WritableFile* file; + Status s = env_->NewWritableFile(manifest, &file); + if (!s.ok()) { + return s; + } + { + log::Writer log(file); + std::string record; + new_db.EncodeTo(&record); + s = log.AddRecord(record); + if (s.ok()) { + s = file->Close(); + } + } + delete file; + if (s.ok()) { + // Make "CURRENT" file that points to the new manifest file. + s = SetCurrentFile(env_, dbname_, 1); + } else { + env_->DeleteFile(manifest); + } + return s; +} + +void DBImpl::MaybeIgnoreError(Status* s) const { + if (s->ok() || options_.paranoid_checks) { + // No change needed + } else { + Log(options_.info_log, "Ignoring error %s", s->ToString().c_str()); + *s = Status::OK(); + } +} + +void DBImpl::DeleteObsoleteFiles() { + if (!bg_error_.ok()) { + // After a background error, we don't know whether a new version may + // or may not have been committed, so we cannot safely garbage collect. + return; + } + + // Make a set of all of the live files + std::set<uint64_t> live = pending_outputs_; + versions_->AddLiveFiles(&live); + + std::vector<std::string> filenames; + env_->GetChildren(dbname_, &filenames); // Ignoring errors on purpose + uint64_t number; + FileType type; + for (size_t i = 0; i < filenames.size(); i++) { + if (ParseFileName(filenames[i], &number, &type)) { + bool keep = true; + switch (type) { + case kLogFile: + keep = ((number >= versions_->LogNumber()) || + (number == versions_->PrevLogNumber())); + break; + case kDescriptorFile: + // Keep my manifest file, and any newer incarnations' + // (in case there is a race that allows other incarnations) + keep = (number >= versions_->ManifestFileNumber()); + break; + case kTableFile: + keep = (live.find(number) != live.end()); + break; + case kTempFile: + // Any temp files that are currently being written to must + // be recorded in pending_outputs_, which is inserted into "live" + keep = (live.find(number) != live.end()); + break; + case kCurrentFile: + case kDBLockFile: + case kInfoLogFile: + keep = true; + break; + } + + if (!keep) { + if (type == kTableFile) { + table_cache_->Evict(number); + } + Log(options_.info_log, "Delete type=%d #%lld\n", + int(type), + static_cast<unsigned long long>(number)); + env_->DeleteFile(dbname_ + "/" + filenames[i]); + } + } + } +} + +Status DBImpl::Recover(VersionEdit* edit, bool *save_manifest) { + mutex_.AssertHeld(); + + // Ignore error from CreateDir since the creation of the DB is + // committed only when the descriptor is created, and this directory + // may already exist from a previous failed creation attempt. + env_->CreateDir(dbname_); + assert(db_lock_ == NULL); + Status s = env_->LockFile(LockFileName(dbname_), &db_lock_); + if (!s.ok()) { + return s; + } + + if (!env_->FileExists(CurrentFileName(dbname_))) { + if (options_.create_if_missing) { + s = NewDB(); + if (!s.ok()) { + return s; + } + } else { + return Status::InvalidArgument( + dbname_, "does not exist (create_if_missing is false)"); + } + } else { + if (options_.error_if_exists) { + return Status::InvalidArgument( + dbname_, "exists (error_if_exists is true)"); + } + } + + s = versions_->Recover(save_manifest); + if (!s.ok()) { + return s; + } + SequenceNumber max_sequence(0); + + // Recover from all newer log files than the ones named in the + // descriptor (new log files may have been added by the previous + // incarnation without registering them in the descriptor). + // + // Note that PrevLogNumber() is no longer used, but we pay + // attention to it in case we are recovering a database + // produced by an older version of leveldb. + const uint64_t min_log = versions_->LogNumber(); + const uint64_t prev_log = versions_->PrevLogNumber(); + std::vector<std::string> filenames; + s = env_->GetChildren(dbname_, &filenames); + if (!s.ok()) { + return s; + } + std::set<uint64_t> expected; + versions_->AddLiveFiles(&expected); + uint64_t number; + FileType type; + std::vector<uint64_t> logs; + for (size_t i = 0; i < filenames.size(); i++) { + if (ParseFileName(filenames[i], &number, &type)) { + expected.erase(number); + if (type == kLogFile && ((number >= min_log) || (number == prev_log))) + logs.push_back(number); + } + } + if (!expected.empty()) { + char buf[50]; + snprintf(buf, sizeof(buf), "%d missing files; e.g.", + static_cast<int>(expected.size())); + return Status::Corruption(buf, TableFileName(dbname_, *(expected.begin()))); + } + + // Recover in the order in which the logs were generated + std::sort(logs.begin(), logs.end()); + for (size_t i = 0; i < logs.size(); i++) { + s = RecoverLogFile(logs[i], (i == logs.size() - 1), save_manifest, edit, + &max_sequence); + if (!s.ok()) { + return s; + } + + // The previous incarnation may not have written any MANIFEST + // records after allocating this log number. So we manually + // update the file number allocation counter in VersionSet. + versions_->MarkFileNumberUsed(logs[i]); + } + + if (versions_->LastSequence() < max_sequence) { + versions_->SetLastSequence(max_sequence); + } + + return Status::OK(); +} + +Status DBImpl::RecoverLogFile(uint64_t log_number, bool last_log, + bool* save_manifest, VersionEdit* edit, + SequenceNumber* max_sequence) { + struct LogReporter : public log::Reader::Reporter { + Env* env; + Logger* info_log; + const char* fname; + Status* status; // NULL if options_.paranoid_checks==false + virtual void Corruption(size_t bytes, const Status& s) { + Log(info_log, "%s%s: dropping %d bytes; %s", + (this->status == NULL ? "(ignoring error) " : ""), + fname, static_cast<int>(bytes), s.ToString().c_str()); + if (this->status != NULL && this->status->ok()) *this->status = s; + } + }; + + mutex_.AssertHeld(); + + // Open the log file + std::string fname = LogFileName(dbname_, log_number); + SequentialFile* file; + Status status = env_->NewSequentialFile(fname, &file); + if (!status.ok()) { + MaybeIgnoreError(&status); + return status; + } + + // Create the log reader. + LogReporter reporter; + reporter.env = env_; + reporter.info_log = options_.info_log; + reporter.fname = fname.c_str(); + reporter.status = (options_.paranoid_checks ? &status : NULL); + // We intentionally make log::Reader do checksumming even if + // paranoid_checks==false so that corruptions cause entire commits + // to be skipped instead of propagating bad information (like overly + // large sequence numbers). + log::Reader reader(file, &reporter, true/*checksum*/, + 0/*initial_offset*/); + Log(options_.info_log, "Recovering log #%llu", + (unsigned long long) log_number); + + // Read all the records and add to a memtable + std::string scratch; + Slice record; + WriteBatch batch; + int compactions = 0; + MemTable* mem = NULL; + while (reader.ReadRecord(&record, &scratch) && + status.ok()) { + if (record.size() < 12) { + reporter.Corruption( + record.size(), Status::Corruption("log record too small")); + continue; + } + WriteBatchInternal::SetContents(&batch, record); + + if (mem == NULL) { + mem = new MemTable(internal_comparator_); + mem->Ref(); + } + status = WriteBatchInternal::InsertInto(&batch, mem); + MaybeIgnoreError(&status); + if (!status.ok()) { + break; + } + const SequenceNumber last_seq = + WriteBatchInternal::Sequence(&batch) + + WriteBatchInternal::Count(&batch) - 1; + if (last_seq > *max_sequence) { + *max_sequence = last_seq; + } + + if (mem->ApproximateMemoryUsage() > options_.write_buffer_size) { + compactions++; + *save_manifest = true; + status = WriteLevel0Table(mem, edit, NULL); + mem->Unref(); + mem = NULL; + if (!status.ok()) { + // Reflect errors immediately so that conditions like full + // file-systems cause the DB::Open() to fail. + break; + } + } + } + + delete file; + + // See if we should keep reusing the last log file. + if (status.ok() && options_.reuse_logs && last_log && compactions == 0) { + assert(logfile_ == NULL); + assert(log_ == NULL); + assert(mem_ == NULL); + uint64_t lfile_size; + if (env_->GetFileSize(fname, &lfile_size).ok() && + env_->NewAppendableFile(fname, &logfile_).ok()) { + Log(options_.info_log, "Reusing old log %s \n", fname.c_str()); + log_ = new log::Writer(logfile_, lfile_size); + logfile_number_ = log_number; + if (mem != NULL) { + mem_ = mem; + mem = NULL; + } else { + // mem can be NULL if lognum exists but was empty. + mem_ = new MemTable(internal_comparator_); + mem_->Ref(); + } + } + } + + if (mem != NULL) { + // mem did not get reused; compact it. + if (status.ok()) { + *save_manifest = true; + status = WriteLevel0Table(mem, edit, NULL); + } + mem->Unref(); + } + + return status; +} + +Status DBImpl::WriteLevel0Table(MemTable* mem, VersionEdit* edit, + Version* base) { + mutex_.AssertHeld(); + const uint64_t start_micros = env_->NowMicros(); + FileMetaData meta; + meta.number = versions_->NewFileNumber(); + pending_outputs_.insert(meta.number); + Iterator* iter = mem->NewIterator(); + Log(options_.info_log, "Level-0 table #%llu: started", + (unsigned long long) meta.number); + + Status s; + { + mutex_.Unlock(); + s = BuildTable(dbname_, env_, options_, table_cache_, iter, &meta); + mutex_.Lock(); + } + + Log(options_.info_log, "Level-0 table #%llu: %lld bytes %s", + (unsigned long long) meta.number, + (unsigned long long) meta.file_size, + s.ToString().c_str()); + delete iter; + pending_outputs_.erase(meta.number); + + + // Note that if file_size is zero, the file has been deleted and + // should not be added to the manifest. + int level = 0; + if (s.ok() && meta.file_size > 0) { + const Slice min_user_key = meta.smallest.user_key(); + const Slice max_user_key = meta.largest.user_key(); + if (base != NULL) { + level = base->PickLevelForMemTableOutput(min_user_key, max_user_key); + } + edit->AddFile(level, meta.number, meta.file_size, + meta.smallest, meta.largest); + } + + CompactionStats stats; + stats.micros = env_->NowMicros() - start_micros; + stats.bytes_written = meta.file_size; + stats_[level].Add(stats); + return s; +} + +void DBImpl::CompactMemTable() { + mutex_.AssertHeld(); + assert(imm_ != NULL); + + // Save the contents of the memtable as a new Table + VersionEdit edit; + Version* base = versions_->current(); + base->Ref(); + Status s = WriteLevel0Table(imm_, &edit, base); + base->Unref(); + + if (s.ok() && shutting_down_.Acquire_Load()) { + s = Status::IOError("Deleting DB during memtable compaction"); + } + + // Replace immutable memtable with the generated Table + if (s.ok()) { + edit.SetPrevLogNumber(0); + edit.SetLogNumber(logfile_number_); // Earlier logs no longer needed + s = versions_->LogAndApply(&edit, &mutex_); + } + + if (s.ok()) { + // Commit to the new state + imm_->Unref(); + imm_ = NULL; + has_imm_.Release_Store(NULL); + DeleteObsoleteFiles(); + } else { + RecordBackgroundError(s); + } +} + +void DBImpl::CompactRange(const Slice* begin, const Slice* end) { + int max_level_with_files = 1; + { + MutexLock l(&mutex_); + Version* base = versions_->current(); + for (int level = 1; level < config::kNumLevels; level++) { + if (base->OverlapInLevel(level, begin, end)) { + max_level_with_files = level; + } + } + } + TEST_CompactMemTable(); // TODO(sanjay): Skip if memtable does not overlap + for (int level = 0; level < max_level_with_files; level++) { + TEST_CompactRange(level, begin, end); + } +} + +void DBImpl::TEST_CompactRange(int level, const Slice* begin,const Slice* end) { + assert(level >= 0); + assert(level + 1 < config::kNumLevels); + + InternalKey begin_storage, end_storage; + + ManualCompaction manual; + manual.level = level; + manual.done = false; + if (begin == NULL) { + manual.begin = NULL; + } else { + begin_storage = InternalKey(*begin, kMaxSequenceNumber, kValueTypeForSeek); + manual.begin = &begin_storage; + } + if (end == NULL) { + manual.end = NULL; + } else { + end_storage = InternalKey(*end, 0, static_cast<ValueType>(0)); + manual.end = &end_storage; + } + + MutexLock l(&mutex_); + while (!manual.done && !shutting_down_.Acquire_Load() && bg_error_.ok()) { + if (manual_compaction_ == NULL) { // Idle + manual_compaction_ = &manual; + MaybeScheduleCompaction(); + } else { // Running either my compaction or another compaction. + bg_cv_.Wait(); + } + } + if (manual_compaction_ == &manual) { + // Cancel my manual compaction since we aborted early for some reason. + manual_compaction_ = NULL; + } +} + +Status DBImpl::TEST_CompactMemTable() { + // NULL batch means just wait for earlier writes to be done + Status s = Write(WriteOptions(), NULL); + if (s.ok()) { + // Wait until the compaction completes + MutexLock l(&mutex_); + while (imm_ != NULL && bg_error_.ok()) { + bg_cv_.Wait(); + } + if (imm_ != NULL) { + s = bg_error_; + } + } + return s; +} + +void DBImpl::RecordBackgroundError(const Status& s) { + mutex_.AssertHeld(); + if (bg_error_.ok()) { + bg_error_ = s; + bg_cv_.SignalAll(); + } +} + +void DBImpl::MaybeScheduleCompaction() { + mutex_.AssertHeld(); + if (bg_compaction_scheduled_) { + // Already scheduled + } else if (shutting_down_.Acquire_Load()) { + // DB is being deleted; no more background compactions + } else if (!bg_error_.ok()) { + // Already got an error; no more changes + } else if (imm_ == NULL && + manual_compaction_ == NULL && + !versions_->NeedsCompaction()) { + // No work to be done + } else { + bg_compaction_scheduled_ = true; + env_->Schedule(&DBImpl::BGWork, this); + } +} + +void DBImpl::BGWork(void* db) { + reinterpret_cast<DBImpl*>(db)->BackgroundCall(); +} + +void DBImpl::BackgroundCall() { + MutexLock l(&mutex_); + assert(bg_compaction_scheduled_); + if (shutting_down_.Acquire_Load()) { + // No more background work when shutting down. + } else if (!bg_error_.ok()) { + // No more background work after a background error. + } else { + BackgroundCompaction(); + } + + bg_compaction_scheduled_ = false; + + // Previous compaction may have produced too many files in a level, + // so reschedule another compaction if needed. + MaybeScheduleCompaction(); + bg_cv_.SignalAll(); +} + +void DBImpl::BackgroundCompaction() { + mutex_.AssertHeld(); + + if (imm_ != NULL) { + CompactMemTable(); + return; + } + + Compaction* c; + bool is_manual = (manual_compaction_ != NULL); + InternalKey manual_end; + if (is_manual) { + ManualCompaction* m = manual_compaction_; + c = versions_->CompactRange(m->level, m->begin, m->end); + m->done = (c == NULL); + if (c != NULL) { + manual_end = c->input(0, c->num_input_files(0) - 1)->largest; + } + Log(options_.info_log, + "Manual compaction at level-%d from %s .. %s; will stop at %s\n", + m->level, + (m->begin ? m->begin->DebugString().c_str() : "(begin)"), + (m->end ? m->end->DebugString().c_str() : "(end)"), + (m->done ? "(end)" : manual_end.DebugString().c_str())); + } else { + c = versions_->PickCompaction(); + } + + Status status; + if (c == NULL) { + // Nothing to do + } else if (!is_manual && c->IsTrivialMove()) { + // Move file to next level + assert(c->num_input_files(0) == 1); + FileMetaData* f = c->input(0, 0); + c->edit()->DeleteFile(c->level(), f->number); + c->edit()->AddFile(c->level() + 1, f->number, f->file_size, + f->smallest, f->largest); + status = versions_->LogAndApply(c->edit(), &mutex_); + if (!status.ok()) { + RecordBackgroundError(status); + } + VersionSet::LevelSummaryStorage tmp; + Log(options_.info_log, "Moved #%lld to level-%d %lld bytes %s: %s\n", + static_cast<unsigned long long>(f->number), + c->level() + 1, + static_cast<unsigned long long>(f->file_size), + status.ToString().c_str(), + versions_->LevelSummary(&tmp)); + } else { + CompactionState* compact = new CompactionState(c); + status = DoCompactionWork(compact); + if (!status.ok()) { + RecordBackgroundError(status); + } + CleanupCompaction(compact); + c->ReleaseInputs(); + DeleteObsoleteFiles(); + } + delete c; + + if (status.ok()) { + // Done + } else if (shutting_down_.Acquire_Load()) { + // Ignore compaction errors found during shutting down + } else { + Log(options_.info_log, + "Compaction error: %s", status.ToString().c_str()); + } + + if (is_manual) { + ManualCompaction* m = manual_compaction_; + if (!status.ok()) { + m->done = true; + } + if (!m->done) { + // We only compacted part of the requested range. Update *m + // to the range that is left to be compacted. + m->tmp_storage = manual_end; + m->begin = &m->tmp_storage; + } + manual_compaction_ = NULL; + } +} + +void DBImpl::CleanupCompaction(CompactionState* compact) { + mutex_.AssertHeld(); + if (compact->builder != NULL) { + // May happen if we get a shutdown call in the middle of compaction + compact->builder->Abandon(); + delete compact->builder; + } else { + assert(compact->outfile == NULL); + } + delete compact->outfile; + for (size_t i = 0; i < compact->outputs.size(); i++) { + const CompactionState::Output& out = compact->outputs[i]; + pending_outputs_.erase(out.number); + } + delete compact; +} + +Status DBImpl::OpenCompactionOutputFile(CompactionState* compact) { + assert(compact != NULL); + assert(compact->builder == NULL); + uint64_t file_number; + { + mutex_.Lock(); + file_number = versions_->NewFileNumber(); + pending_outputs_.insert(file_number); + CompactionState::Output out; + out.number = file_number; + out.smallest.Clear(); + out.largest.Clear(); + compact->outputs.push_back(out); + mutex_.Unlock(); + } + + // Make the output file + std::string fname = TableFileName(dbname_, file_number); + Status s = env_->NewWritableFile(fname, &compact->outfile); + if (s.ok()) { + compact->builder = new TableBuilder(options_, compact->outfile); + } + return s; +} + +Status DBImpl::FinishCompactionOutputFile(CompactionState* compact, + Iterator* input) { + assert(compact != NULL); + assert(compact->outfile != NULL); + assert(compact->builder != NULL); + + const uint64_t output_number = compact->current_output()->number; + assert(output_number != 0); + + // Check for iterator errors + Status s = input->status(); + const uint64_t current_entries = compact->builder->NumEntries(); + if (s.ok()) { + s = compact->builder->Finish(); + } else { + compact->builder->Abandon(); + } + const uint64_t current_bytes = compact->builder->FileSize(); + compact->current_output()->file_size = current_bytes; + compact->total_bytes += current_bytes; + delete compact->builder; + compact->builder = NULL; + + // Finish and check for file errors + if (s.ok()) { + s = compact->outfile->Sync(); + } + if (s.ok()) { + s = compact->outfile->Close(); + } + delete compact->outfile; + compact->outfile = NULL; + + if (s.ok() && current_entries > 0) { + // Verify that the table is usable + Iterator* iter = table_cache_->NewIterator(ReadOptions(), + output_number, + current_bytes); + s = iter->status(); + delete iter; + if (s.ok()) { + Log(options_.info_log, + "Generated table #%llu@%d: %lld keys, %lld bytes", + (unsigned long long) output_number, + compact->compaction->level(), + (unsigned long long) current_entries, + (unsigned long long) current_bytes); + } + } + return s; +} + + +Status DBImpl::InstallCompactionResults(CompactionState* compact) { + mutex_.AssertHeld(); + Log(options_.info_log, "Compacted %d@%d + %d@%d files => %lld bytes", + compact->compaction->num_input_files(0), + compact->compaction->level(), + compact->compaction->num_input_files(1), + compact->compaction->level() + 1, + static_cast<long long>(compact->total_bytes)); + + // Add compaction outputs + compact->compaction->AddInputDeletions(compact->compaction->edit()); + const int level = compact->compaction->level(); + for (size_t i = 0; i < compact->outputs.size(); i++) { + const CompactionState::Output& out = compact->outputs[i]; + compact->compaction->edit()->AddFile( + level + 1, + out.number, out.file_size, out.smallest, out.largest); + } + return versions_->LogAndApply(compact->compaction->edit(), &mutex_); +} + +Status DBImpl::DoCompactionWork(CompactionState* compact) { + const uint64_t start_micros = env_->NowMicros(); + int64_t imm_micros = 0; // Micros spent doing imm_ compactions + + Log(options_.info_log, "Compacting %d@%d + %d@%d files", + compact->compaction->num_input_files(0), + compact->compaction->level(), + compact->compaction->num_input_files(1), + compact->compaction->level() + 1); + + assert(versions_->NumLevelFiles(compact->compaction->level()) > 0); + assert(compact->builder == NULL); + assert(compact->outfile == NULL); + if (snapshots_.empty()) { + compact->smallest_snapshot = versions_->LastSequence(); + } else { + compact->smallest_snapshot = snapshots_.oldest()->number_; + } + + // Release mutex while we're actually doing the compaction work + mutex_.Unlock(); + + Iterator* input = versions_->MakeInputIterator(compact->compaction); + input->SeekToFirst(); + Status status; + ParsedInternalKey ikey; + std::string current_user_key; + bool has_current_user_key = false; + SequenceNumber last_sequence_for_key = kMaxSequenceNumber; + for (; input->Valid() && !shutting_down_.Acquire_Load(); ) { + // Prioritize immutable compaction work + if (has_imm_.NoBarrier_Load() != NULL) { + const uint64_t imm_start = env_->NowMicros(); + mutex_.Lock(); + if (imm_ != NULL) { + CompactMemTable(); + bg_cv_.SignalAll(); // Wakeup MakeRoomForWrite() if necessary + } + mutex_.Unlock(); + imm_micros += (env_->NowMicros() - imm_start); + } + + Slice key = input->key(); + if (compact->compaction->ShouldStopBefore(key) && + compact->builder != NULL) { + status = FinishCompactionOutputFile(compact, input); + if (!status.ok()) { + break; + } + } + + // Handle key/value, add to state, etc. + bool drop = false; + if (!ParseInternalKey(key, &ikey)) { + // Do not hide error keys + current_user_key.clear(); + has_current_user_key = false; + last_sequence_for_key = kMaxSequenceNumber; + } else { + if (!has_current_user_key || + user_comparator()->Compare(ikey.user_key, + Slice(current_user_key)) != 0) { + // First occurrence of this user key + current_user_key.assign(ikey.user_key.data(), ikey.user_key.size()); + has_current_user_key = true; + last_sequence_for_key = kMaxSequenceNumber; + } + + if (last_sequence_for_key <= compact->smallest_snapshot) { + // Hidden by an newer entry for same user key + drop = true; // (A) + } else if (ikey.type == kTypeDeletion && + ikey.sequence <= compact->smallest_snapshot && + compact->compaction->IsBaseLevelForKey(ikey.user_key)) { + // For this user key: + // (1) there is no data in higher levels + // (2) data in lower levels will have larger sequence numbers + // (3) data in layers that are being compacted here and have + // smaller sequence numbers will be dropped in the next + // few iterations of this loop (by rule (A) above). + // Therefore this deletion marker is obsolete and can be dropped. + drop = true; + } + + last_sequence_for_key = ikey.sequence; + } +#if 0 + Log(options_.info_log, + " Compact: %s, seq %d, type: %d %d, drop: %d, is_base: %d, " + "%d smallest_snapshot: %d", + ikey.user_key.ToString().c_str(), + (int)ikey.sequence, ikey.type, kTypeValue, drop, + compact->compaction->IsBaseLevelForKey(ikey.user_key), + (int)last_sequence_for_key, (int)compact->smallest_snapshot); +#endif + + if (!drop) { + // Open output file if necessary + if (compact->builder == NULL) { + status = OpenCompactionOutputFile(compact); + if (!status.ok()) { + break; + } + } + if (compact->builder->NumEntries() == 0) { + compact->current_output()->smallest.DecodeFrom(key); + } + compact->current_output()->largest.DecodeFrom(key); + compact->builder->Add(key, input->value()); + + // Close output file if it is big enough + if (compact->builder->FileSize() >= + compact->compaction->MaxOutputFileSize()) { + status = FinishCompactionOutputFile(compact, input); + if (!status.ok()) { + break; + } + } + } + + input->Next(); + } + + if (status.ok() && shutting_down_.Acquire_Load()) { + status = Status::IOError("Deleting DB during compaction"); + } + if (status.ok() && compact->builder != NULL) { + status = FinishCompactionOutputFile(compact, input); + } + if (status.ok()) { + status = input->status(); + } + delete input; + input = NULL; + + CompactionStats stats; + stats.micros = env_->NowMicros() - start_micros - imm_micros; + for (int which = 0; which < 2; which++) { + for (int i = 0; i < compact->compaction->num_input_files(which); i++) { + stats.bytes_read += compact->compaction->input(which, i)->file_size; + } + } + for (size_t i = 0; i < compact->outputs.size(); i++) { + stats.bytes_written += compact->outputs[i].file_size; + } + + mutex_.Lock(); + stats_[compact->compaction->level() + 1].Add(stats); + + if (status.ok()) { + status = InstallCompactionResults(compact); + } + if (!status.ok()) { + RecordBackgroundError(status); + } + VersionSet::LevelSummaryStorage tmp; + Log(options_.info_log, + "compacted to: %s", versions_->LevelSummary(&tmp)); + return status; +} + +namespace { +struct IterState { + port::Mutex* mu; + Version* version; + MemTable* mem; + MemTable* imm; +}; + +static void CleanupIteratorState(void* arg1, void* arg2) { + IterState* state = reinterpret_cast<IterState*>(arg1); + state->mu->Lock(); + state->mem->Unref(); + if (state->imm != NULL) state->imm->Unref(); + state->version->Unref(); + state->mu->Unlock(); + delete state; +} +} // namespace + +Iterator* DBImpl::NewInternalIterator(const ReadOptions& options, + SequenceNumber* latest_snapshot, + uint32_t* seed) { + IterState* cleanup = new IterState; + mutex_.Lock(); + *latest_snapshot = versions_->LastSequence(); + + // Collect together all needed child iterators + std::vector<Iterator*> list; + list.push_back(mem_->NewIterator()); + mem_->Ref(); + if (imm_ != NULL) { + list.push_back(imm_->NewIterator()); + imm_->Ref(); + } + versions_->current()->AddIterators(options, &list); + Iterator* internal_iter = + NewMergingIterator(&internal_comparator_, &list[0], list.size()); + versions_->current()->Ref(); + + cleanup->mu = &mutex_; + cleanup->mem = mem_; + cleanup->imm = imm_; + cleanup->version = versions_->current(); + internal_iter->RegisterCleanup(CleanupIteratorState, cleanup, NULL); + + *seed = ++seed_; + mutex_.Unlock(); + return internal_iter; +} + +Iterator* DBImpl::TEST_NewInternalIterator() { + SequenceNumber ignored; + uint32_t ignored_seed; + return NewInternalIterator(ReadOptions(), &ignored, &ignored_seed); +} + +int64_t DBImpl::TEST_MaxNextLevelOverlappingBytes() { + MutexLock l(&mutex_); + return versions_->MaxNextLevelOverlappingBytes(); +} + +Status DBImpl::Get(const ReadOptions& options, + const Slice& key, + std::string* value) { + Status s; + MutexLock l(&mutex_); + SequenceNumber snapshot; + if (options.snapshot != NULL) { + snapshot = reinterpret_cast<const SnapshotImpl*>(options.snapshot)->number_; + } else { + snapshot = versions_->LastSequence(); + } + + MemTable* mem = mem_; + MemTable* imm = imm_; + Version* current = versions_->current(); + mem->Ref(); + if (imm != NULL) imm->Ref(); + current->Ref(); + + bool have_stat_update = false; + Version::GetStats stats; + + // Unlock while reading from files and memtables + { + mutex_.Unlock(); + // First look in the memtable, then in the immutable memtable (if any). + LookupKey lkey(key, snapshot); + if (mem->Get(lkey, value, &s)) { + // Done + } else if (imm != NULL && imm->Get(lkey, value, &s)) { + // Done + } else { + s = current->Get(options, lkey, value, &stats); + have_stat_update = true; + } + mutex_.Lock(); + } + + if (have_stat_update && current->UpdateStats(stats)) { + MaybeScheduleCompaction(); + } + mem->Unref(); + if (imm != NULL) imm->Unref(); + current->Unref(); + return s; +} + +Iterator* DBImpl::NewIterator(const ReadOptions& options) { + SequenceNumber latest_snapshot; + uint32_t seed; + Iterator* iter = NewInternalIterator(options, &latest_snapshot, &seed); + return NewDBIterator( + this, user_comparator(), iter, + (options.snapshot != NULL + ? reinterpret_cast<const SnapshotImpl*>(options.snapshot)->number_ + : latest_snapshot), + seed); +} + +void DBImpl::RecordReadSample(Slice key) { + MutexLock l(&mutex_); + if (versions_->current()->RecordReadSample(key)) { + MaybeScheduleCompaction(); + } +} + +const Snapshot* DBImpl::GetSnapshot() { + MutexLock l(&mutex_); + return snapshots_.New(versions_->LastSequence()); +} + +void DBImpl::ReleaseSnapshot(const Snapshot* s) { + MutexLock l(&mutex_); + snapshots_.Delete(reinterpret_cast<const SnapshotImpl*>(s)); +} + +// Convenience methods +Status DBImpl::Put(const WriteOptions& o, const Slice& key, const Slice& val) { + return DB::Put(o, key, val); +} + +Status DBImpl::Delete(const WriteOptions& options, const Slice& key) { + return DB::Delete(options, key); +} + +Status DBImpl::Write(const WriteOptions& options, WriteBatch* my_batch) { + Writer w(&mutex_); + w.batch = my_batch; + w.sync = options.sync; + w.done = false; + + MutexLock l(&mutex_); + writers_.push_back(&w); + while (!w.done && &w != writers_.front()) { + w.cv.Wait(); + } + if (w.done) { + return w.status; + } + + // May temporarily unlock and wait. + Status status = MakeRoomForWrite(my_batch == NULL); + uint64_t last_sequence = versions_->LastSequence(); + Writer* last_writer = &w; + if (status.ok() && my_batch != NULL) { // NULL batch is for compactions + WriteBatch* updates = BuildBatchGroup(&last_writer); + WriteBatchInternal::SetSequence(updates, last_sequence + 1); + last_sequence += WriteBatchInternal::Count(updates); + + // Add to log and apply to memtable. We can release the lock + // during this phase since &w is currently responsible for logging + // and protects against concurrent loggers and concurrent writes + // into mem_. + { + mutex_.Unlock(); + status = log_->AddRecord(WriteBatchInternal::Contents(updates)); + bool sync_error = false; + if (status.ok() && options.sync) { + status = logfile_->Sync(); + if (!status.ok()) { + sync_error = true; + } + } + if (status.ok()) { + status = WriteBatchInternal::InsertInto(updates, mem_); + } + mutex_.Lock(); + if (sync_error) { + // The state of the log file is indeterminate: the log record we + // just added may or may not show up when the DB is re-opened. + // So we force the DB into a mode where all future writes fail. + RecordBackgroundError(status); + } + } + if (updates == tmp_batch_) tmp_batch_->Clear(); + + versions_->SetLastSequence(last_sequence); + } + + while (true) { + Writer* ready = writers_.front(); + writers_.pop_front(); + if (ready != &w) { + ready->status = status; + ready->done = true; + ready->cv.Signal(); + } + if (ready == last_writer) break; + } + + // Notify new head of write queue + if (!writers_.empty()) { + writers_.front()->cv.Signal(); + } + + return status; +} + +// REQUIRES: Writer list must be non-empty +// REQUIRES: First writer must have a non-NULL batch +WriteBatch* DBImpl::BuildBatchGroup(Writer** last_writer) { + assert(!writers_.empty()); + Writer* first = writers_.front(); + WriteBatch* result = first->batch; + assert(result != NULL); + + size_t size = WriteBatchInternal::ByteSize(first->batch); + + // Allow the group to grow up to a maximum size, but if the + // original write is small, limit the growth so we do not slow + // down the small write too much. + size_t max_size = 1 << 20; + if (size <= (128<<10)) { + max_size = size + (128<<10); + } + + *last_writer = first; + std::deque<Writer*>::iterator iter = writers_.begin(); + ++iter; // Advance past "first" + for (; iter != writers_.end(); ++iter) { + Writer* w = *iter; + if (w->sync && !first->sync) { + // Do not include a sync write into a batch handled by a non-sync write. + break; + } + + if (w->batch != NULL) { + size += WriteBatchInternal::ByteSize(w->batch); + if (size > max_size) { + // Do not make batch too big + break; + } + + // Append to *result + if (result == first->batch) { + // Switch to temporary batch instead of disturbing caller's batch + result = tmp_batch_; + assert(WriteBatchInternal::Count(result) == 0); + WriteBatchInternal::Append(result, first->batch); + } + WriteBatchInternal::Append(result, w->batch); + } + *last_writer = w; + } + return result; +} + +// REQUIRES: mutex_ is held +// REQUIRES: this thread is currently at the front of the writer queue +Status DBImpl::MakeRoomForWrite(bool force) { + mutex_.AssertHeld(); + assert(!writers_.empty()); + bool allow_delay = !force; + Status s; + while (true) { + if (!bg_error_.ok()) { + // Yield previous error + s = bg_error_; + break; + } else if ( + allow_delay && + versions_->NumLevelFiles(0) >= config::kL0_SlowdownWritesTrigger) { + // We are getting close to hitting a hard limit on the number of + // L0 files. Rather than delaying a single write by several + // seconds when we hit the hard limit, start delaying each + // individual write by 1ms to reduce latency variance. Also, + // this delay hands over some CPU to the compaction thread in + // case it is sharing the same core as the writer. + mutex_.Unlock(); + env_->SleepForMicroseconds(1000); + allow_delay = false; // Do not delay a single write more than once + mutex_.Lock(); + } else if (!force && + (mem_->ApproximateMemoryUsage() <= options_.write_buffer_size)) { + // There is room in current memtable + break; + } else if (imm_ != NULL) { + // We have filled up the current memtable, but the previous + // one is still being compacted, so we wait. + Log(options_.info_log, "Current memtable full; waiting...\n"); + bg_cv_.Wait(); + } else if (versions_->NumLevelFiles(0) >= config::kL0_StopWritesTrigger) { + // There are too many level-0 files. + Log(options_.info_log, "Too many L0 files; waiting...\n"); + bg_cv_.Wait(); + } else { + // Attempt to switch to a new memtable and trigger compaction of old + assert(versions_->PrevLogNumber() == 0); + uint64_t new_log_number = versions_->NewFileNumber(); + WritableFile* lfile = NULL; + s = env_->NewWritableFile(LogFileName(dbname_, new_log_number), &lfile); + if (!s.ok()) { + // Avoid chewing through file number space in a tight loop. + versions_->ReuseFileNumber(new_log_number); + break; + } + delete log_; + delete logfile_; + logfile_ = lfile; + logfile_number_ = new_log_number; + log_ = new log::Writer(lfile); + imm_ = mem_; + has_imm_.Release_Store(imm_); + mem_ = new MemTable(internal_comparator_); + mem_->Ref(); + force = false; // Do not force another compaction if have room + MaybeScheduleCompaction(); + } + } + return s; +} + +bool DBImpl::GetProperty(const Slice& property, std::string* value) { + value->clear(); + + MutexLock l(&mutex_); + Slice in = property; + Slice prefix("leveldb."); + if (!in.starts_with(prefix)) return false; + in.remove_prefix(prefix.size()); + + if (in.starts_with("num-files-at-level")) { + in.remove_prefix(strlen("num-files-at-level")); + uint64_t level; + bool ok = ConsumeDecimalNumber(&in, &level) && in.empty(); + if (!ok || level >= config::kNumLevels) { + return false; + } else { + char buf[100]; + snprintf(buf, sizeof(buf), "%d", + versions_->NumLevelFiles(static_cast<int>(level))); + *value = buf; + return true; + } + } else if (in == "stats") { + char buf[200]; + snprintf(buf, sizeof(buf), + " Compactions\n" + "Level Files Size(MB) Time(sec) Read(MB) Write(MB)\n" + "--------------------------------------------------\n" + ); + value->append(buf); + for (int level = 0; level < config::kNumLevels; level++) { + int files = versions_->NumLevelFiles(level); + if (stats_[level].micros > 0 || files > 0) { + snprintf( + buf, sizeof(buf), + "%3d %8d %8.0f %9.0f %8.0f %9.0f\n", + level, + files, + versions_->NumLevelBytes(level) / 1048576.0, + stats_[level].micros / 1e6, + stats_[level].bytes_read / 1048576.0, + stats_[level].bytes_written / 1048576.0); + value->append(buf); + } + } + return true; + } else if (in == "sstables") { + *value = versions_->current()->DebugString(); + return true; + } else if (in == "approximate-memory-usage") { + size_t total_usage = options_.block_cache->TotalCharge(); + if (mem_) { + total_usage += mem_->ApproximateMemoryUsage(); + } + if (imm_) { + total_usage += imm_->ApproximateMemoryUsage(); + } + char buf[50]; + snprintf(buf, sizeof(buf), "%llu", + static_cast<unsigned long long>(total_usage)); + value->append(buf); + return true; + } + + return false; +} + +void DBImpl::GetApproximateSizes( + const Range* range, int n, + uint64_t* sizes) { + // TODO(opt): better implementation + Version* v; + { + MutexLock l(&mutex_); + versions_->current()->Ref(); + v = versions_->current(); + } + + for (int i = 0; i < n; i++) { + // Convert user_key into a corresponding internal key. + InternalKey k1(range[i].start, kMaxSequenceNumber, kValueTypeForSeek); + InternalKey k2(range[i].limit, kMaxSequenceNumber, kValueTypeForSeek); + uint64_t start = versions_->ApproximateOffsetOf(v, k1); + uint64_t limit = versions_->ApproximateOffsetOf(v, k2); + sizes[i] = (limit >= start ? limit - start : 0); + } + + { + MutexLock l(&mutex_); + v->Unref(); + } +} + +// Default implementations of convenience methods that subclasses of DB +// can call if they wish +Status DB::Put(const WriteOptions& opt, const Slice& key, const Slice& value) { + WriteBatch batch; + batch.Put(key, value); + return Write(opt, &batch); +} + +Status DB::Delete(const WriteOptions& opt, const Slice& key) { + WriteBatch batch; + batch.Delete(key); + return Write(opt, &batch); +} + +DB::~DB() { } + +Status DB::Open(const Options& options, const std::string& dbname, + DB** dbptr) { + *dbptr = NULL; + + DBImpl* impl = new DBImpl(options, dbname); + impl->mutex_.Lock(); + VersionEdit edit; + // Recover handles create_if_missing, error_if_exists + bool save_manifest = false; + Status s = impl->Recover(&edit, &save_manifest); + if (s.ok() && impl->mem_ == NULL) { + // Create new log and a corresponding memtable. + uint64_t new_log_number = impl->versions_->NewFileNumber(); + WritableFile* lfile; + s = options.env->NewWritableFile(LogFileName(dbname, new_log_number), + &lfile); + if (s.ok()) { + edit.SetLogNumber(new_log_number); + impl->logfile_ = lfile; + impl->logfile_number_ = new_log_number; + impl->log_ = new log::Writer(lfile); + impl->mem_ = new MemTable(impl->internal_comparator_); + impl->mem_->Ref(); + } + } + if (s.ok() && save_manifest) { + edit.SetPrevLogNumber(0); // No older logs needed after recovery. + edit.SetLogNumber(impl->logfile_number_); + s = impl->versions_->LogAndApply(&edit, &impl->mutex_); + } + if (s.ok()) { + impl->DeleteObsoleteFiles(); + impl->MaybeScheduleCompaction(); + } + impl->mutex_.Unlock(); + if (s.ok()) { + assert(impl->mem_ != NULL); + *dbptr = impl; + } else { + delete impl; + } + return s; +} + +Snapshot::~Snapshot() { +} + +Status DestroyDB(const std::string& dbname, const Options& options) { + Env* env = options.env; + std::vector<std::string> filenames; + // Ignore error in case directory does not exist + env->GetChildren(dbname, &filenames); + if (filenames.empty()) { + return Status::OK(); + } + + FileLock* lock; + const std::string lockname = LockFileName(dbname); + Status result = env->LockFile(lockname, &lock); + if (result.ok()) { + uint64_t number; + FileType type; + for (size_t i = 0; i < filenames.size(); i++) { + if (ParseFileName(filenames[i], &number, &type) && + type != kDBLockFile) { // Lock file will be deleted at end + Status del = env->DeleteFile(dbname + "/" + filenames[i]); + if (result.ok() && !del.ok()) { + result = del; + } + } + } + env->UnlockFile(lock); // Ignore error since state is already gone + env->DeleteFile(lockname); + env->DeleteDir(dbname); // Ignore error in case dir contains other files + } + return result; +} + +} // namespace leveldb diff --git a/src/leveldb/db/db_impl.h b/src/leveldb/db/db_impl.h new file mode 100644 index 0000000000..8ff323e728 --- /dev/null +++ b/src/leveldb/db/db_impl.h @@ -0,0 +1,211 @@ +// 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_DB_IMPL_H_ +#define STORAGE_LEVELDB_DB_DB_IMPL_H_ + +#include <deque> +#include <set> +#include "db/dbformat.h" +#include "db/log_writer.h" +#include "db/snapshot.h" +#include "leveldb/db.h" +#include "leveldb/env.h" +#include "port/port.h" +#include "port/thread_annotations.h" + +namespace leveldb { + +class MemTable; +class TableCache; +class Version; +class VersionEdit; +class VersionSet; + +class DBImpl : public DB { + public: + DBImpl(const Options& options, const std::string& dbname); + virtual ~DBImpl(); + + // Implementations of the DB interface + virtual Status Put(const WriteOptions&, const Slice& key, const Slice& value); + virtual Status Delete(const WriteOptions&, const Slice& key); + virtual Status Write(const WriteOptions& options, WriteBatch* updates); + virtual Status Get(const ReadOptions& options, + const Slice& key, + std::string* value); + virtual Iterator* NewIterator(const ReadOptions&); + virtual const Snapshot* GetSnapshot(); + virtual void ReleaseSnapshot(const Snapshot* snapshot); + virtual bool GetProperty(const Slice& property, std::string* value); + virtual void GetApproximateSizes(const Range* range, int n, uint64_t* sizes); + virtual void CompactRange(const Slice* begin, const Slice* end); + + // Extra methods (for testing) that are not in the public DB interface + + // Compact any files in the named level that overlap [*begin,*end] + void TEST_CompactRange(int level, const Slice* begin, const Slice* end); + + // Force current memtable contents to be compacted. + Status TEST_CompactMemTable(); + + // Return an internal iterator over the current state of the database. + // The keys of this iterator are internal keys (see format.h). + // The returned iterator should be deleted when no longer needed. + Iterator* TEST_NewInternalIterator(); + + // Return the maximum overlapping data (in bytes) at next level for any + // file at a level >= 1. + int64_t TEST_MaxNextLevelOverlappingBytes(); + + // Record a sample of bytes read at the specified internal key. + // Samples are taken approximately once every config::kReadBytesPeriod + // bytes. + void RecordReadSample(Slice key); + + private: + friend class DB; + struct CompactionState; + struct Writer; + + Iterator* NewInternalIterator(const ReadOptions&, + SequenceNumber* latest_snapshot, + uint32_t* seed); + + Status NewDB(); + + // Recover the descriptor from persistent storage. May do a significant + // amount of work to recover recently logged updates. Any changes to + // be made to the descriptor are added to *edit. + Status Recover(VersionEdit* edit, bool* save_manifest) + EXCLUSIVE_LOCKS_REQUIRED(mutex_); + + void MaybeIgnoreError(Status* s) const; + + // Delete any unneeded files and stale in-memory entries. + void DeleteObsoleteFiles(); + + // Compact the in-memory write buffer to disk. Switches to a new + // log-file/memtable and writes a new descriptor iff successful. + // Errors are recorded in bg_error_. + void CompactMemTable() EXCLUSIVE_LOCKS_REQUIRED(mutex_); + + Status RecoverLogFile(uint64_t log_number, bool last_log, bool* save_manifest, + VersionEdit* edit, SequenceNumber* max_sequence) + EXCLUSIVE_LOCKS_REQUIRED(mutex_); + + Status WriteLevel0Table(MemTable* mem, VersionEdit* edit, Version* base) + EXCLUSIVE_LOCKS_REQUIRED(mutex_); + + Status MakeRoomForWrite(bool force /* compact even if there is room? */) + EXCLUSIVE_LOCKS_REQUIRED(mutex_); + WriteBatch* BuildBatchGroup(Writer** last_writer); + + void RecordBackgroundError(const Status& s); + + void MaybeScheduleCompaction() EXCLUSIVE_LOCKS_REQUIRED(mutex_); + static void BGWork(void* db); + void BackgroundCall(); + void BackgroundCompaction() EXCLUSIVE_LOCKS_REQUIRED(mutex_); + void CleanupCompaction(CompactionState* compact) + EXCLUSIVE_LOCKS_REQUIRED(mutex_); + Status DoCompactionWork(CompactionState* compact) + EXCLUSIVE_LOCKS_REQUIRED(mutex_); + + Status OpenCompactionOutputFile(CompactionState* compact); + Status FinishCompactionOutputFile(CompactionState* compact, Iterator* input); + Status InstallCompactionResults(CompactionState* compact) + EXCLUSIVE_LOCKS_REQUIRED(mutex_); + + // Constant after construction + Env* const env_; + const InternalKeyComparator internal_comparator_; + const InternalFilterPolicy internal_filter_policy_; + const Options options_; // options_.comparator == &internal_comparator_ + bool owns_info_log_; + bool owns_cache_; + const std::string dbname_; + + // table_cache_ provides its own synchronization + TableCache* table_cache_; + + // Lock over the persistent DB state. Non-NULL iff successfully acquired. + FileLock* db_lock_; + + // State below is protected by mutex_ + port::Mutex mutex_; + port::AtomicPointer shutting_down_; + port::CondVar bg_cv_; // Signalled when background work finishes + MemTable* mem_; + MemTable* imm_; // Memtable being compacted + port::AtomicPointer has_imm_; // So bg thread can detect non-NULL imm_ + WritableFile* logfile_; + uint64_t logfile_number_; + log::Writer* log_; + uint32_t seed_; // For sampling. + + // Queue of writers. + std::deque<Writer*> writers_; + WriteBatch* tmp_batch_; + + SnapshotList snapshots_; + + // Set of table files to protect from deletion because they are + // part of ongoing compactions. + std::set<uint64_t> pending_outputs_; + + // Has a background compaction been scheduled or is running? + bool bg_compaction_scheduled_; + + // Information for a manual compaction + struct ManualCompaction { + int level; + bool done; + const InternalKey* begin; // NULL means beginning of key range + const InternalKey* end; // NULL means end of key range + InternalKey tmp_storage; // Used to keep track of compaction progress + }; + ManualCompaction* manual_compaction_; + + VersionSet* versions_; + + // Have we encountered a background error in paranoid mode? + Status bg_error_; + + // Per level compaction stats. stats_[level] stores the stats for + // compactions that produced data for the specified "level". + struct CompactionStats { + int64_t micros; + int64_t bytes_read; + int64_t bytes_written; + + CompactionStats() : micros(0), bytes_read(0), bytes_written(0) { } + + void Add(const CompactionStats& c) { + this->micros += c.micros; + this->bytes_read += c.bytes_read; + this->bytes_written += c.bytes_written; + } + }; + CompactionStats stats_[config::kNumLevels]; + + // No copying allowed + DBImpl(const DBImpl&); + void operator=(const DBImpl&); + + const Comparator* user_comparator() const { + return internal_comparator_.user_comparator(); + } +}; + +// Sanitize db options. The caller should delete result.info_log if +// it is not equal to src.info_log. +extern Options SanitizeOptions(const std::string& db, + const InternalKeyComparator* icmp, + const InternalFilterPolicy* ipolicy, + const Options& src); + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_DB_IMPL_H_ diff --git a/src/leveldb/db/db_iter.cc b/src/leveldb/db/db_iter.cc new file mode 100644 index 0000000000..3b2035e9e3 --- /dev/null +++ b/src/leveldb/db/db_iter.cc @@ -0,0 +1,317 @@ +// 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/db_iter.h" + +#include "db/filename.h" +#include "db/db_impl.h" +#include "db/dbformat.h" +#include "leveldb/env.h" +#include "leveldb/iterator.h" +#include "port/port.h" +#include "util/logging.h" +#include "util/mutexlock.h" +#include "util/random.h" + +namespace leveldb { + +#if 0 +static void DumpInternalIter(Iterator* iter) { + for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { + ParsedInternalKey k; + if (!ParseInternalKey(iter->key(), &k)) { + fprintf(stderr, "Corrupt '%s'\n", EscapeString(iter->key()).c_str()); + } else { + fprintf(stderr, "@ '%s'\n", k.DebugString().c_str()); + } + } +} +#endif + +namespace { + +// Memtables and sstables that make the DB representation contain +// (userkey,seq,type) => uservalue entries. DBIter +// combines multiple entries for the same userkey found in the DB +// representation into a single entry while accounting for sequence +// numbers, deletion markers, overwrites, etc. +class DBIter: public Iterator { + public: + // Which direction is the iterator currently moving? + // (1) When moving forward, the internal iterator is positioned at + // the exact entry that yields this->key(), this->value() + // (2) When moving backwards, the internal iterator is positioned + // just before all entries whose user key == this->key(). + enum Direction { + kForward, + kReverse + }; + + DBIter(DBImpl* db, const Comparator* cmp, Iterator* iter, SequenceNumber s, + uint32_t seed) + : db_(db), + user_comparator_(cmp), + iter_(iter), + sequence_(s), + direction_(kForward), + valid_(false), + rnd_(seed), + bytes_counter_(RandomPeriod()) { + } + virtual ~DBIter() { + delete iter_; + } + virtual bool Valid() const { return valid_; } + virtual Slice key() const { + assert(valid_); + return (direction_ == kForward) ? ExtractUserKey(iter_->key()) : saved_key_; + } + virtual Slice value() const { + assert(valid_); + return (direction_ == kForward) ? iter_->value() : saved_value_; + } + virtual Status status() const { + if (status_.ok()) { + return iter_->status(); + } else { + return status_; + } + } + + virtual void Next(); + virtual void Prev(); + virtual void Seek(const Slice& target); + virtual void SeekToFirst(); + virtual void SeekToLast(); + + private: + void FindNextUserEntry(bool skipping, std::string* skip); + void FindPrevUserEntry(); + bool ParseKey(ParsedInternalKey* key); + + inline void SaveKey(const Slice& k, std::string* dst) { + dst->assign(k.data(), k.size()); + } + + inline void ClearSavedValue() { + if (saved_value_.capacity() > 1048576) { + std::string empty; + swap(empty, saved_value_); + } else { + saved_value_.clear(); + } + } + + // Pick next gap with average value of config::kReadBytesPeriod. + ssize_t RandomPeriod() { + return rnd_.Uniform(2*config::kReadBytesPeriod); + } + + DBImpl* db_; + const Comparator* const user_comparator_; + Iterator* const iter_; + SequenceNumber const sequence_; + + Status status_; + std::string saved_key_; // == current key when direction_==kReverse + std::string saved_value_; // == current raw value when direction_==kReverse + Direction direction_; + bool valid_; + + Random rnd_; + ssize_t bytes_counter_; + + // No copying allowed + DBIter(const DBIter&); + void operator=(const DBIter&); +}; + +inline bool DBIter::ParseKey(ParsedInternalKey* ikey) { + Slice k = iter_->key(); + ssize_t n = k.size() + iter_->value().size(); + bytes_counter_ -= n; + while (bytes_counter_ < 0) { + bytes_counter_ += RandomPeriod(); + db_->RecordReadSample(k); + } + if (!ParseInternalKey(k, ikey)) { + status_ = Status::Corruption("corrupted internal key in DBIter"); + return false; + } else { + return true; + } +} + +void DBIter::Next() { + assert(valid_); + + if (direction_ == kReverse) { // Switch directions? + direction_ = kForward; + // iter_ is pointing just before the entries for this->key(), + // so advance into the range of entries for this->key() and then + // use the normal skipping code below. + if (!iter_->Valid()) { + iter_->SeekToFirst(); + } else { + iter_->Next(); + } + if (!iter_->Valid()) { + valid_ = false; + saved_key_.clear(); + return; + } + // saved_key_ already contains the key to skip past. + } else { + // Store in saved_key_ the current key so we skip it below. + SaveKey(ExtractUserKey(iter_->key()), &saved_key_); + } + + FindNextUserEntry(true, &saved_key_); +} + +void DBIter::FindNextUserEntry(bool skipping, std::string* skip) { + // Loop until we hit an acceptable entry to yield + assert(iter_->Valid()); + assert(direction_ == kForward); + do { + ParsedInternalKey ikey; + if (ParseKey(&ikey) && ikey.sequence <= sequence_) { + switch (ikey.type) { + case kTypeDeletion: + // Arrange to skip all upcoming entries for this key since + // they are hidden by this deletion. + SaveKey(ikey.user_key, skip); + skipping = true; + break; + case kTypeValue: + if (skipping && + user_comparator_->Compare(ikey.user_key, *skip) <= 0) { + // Entry hidden + } else { + valid_ = true; + saved_key_.clear(); + return; + } + break; + } + } + iter_->Next(); + } while (iter_->Valid()); + saved_key_.clear(); + valid_ = false; +} + +void DBIter::Prev() { + assert(valid_); + + if (direction_ == kForward) { // Switch directions? + // iter_ is pointing at the current entry. Scan backwards until + // the key changes so we can use the normal reverse scanning code. + assert(iter_->Valid()); // Otherwise valid_ would have been false + SaveKey(ExtractUserKey(iter_->key()), &saved_key_); + while (true) { + iter_->Prev(); + if (!iter_->Valid()) { + valid_ = false; + saved_key_.clear(); + ClearSavedValue(); + return; + } + if (user_comparator_->Compare(ExtractUserKey(iter_->key()), + saved_key_) < 0) { + break; + } + } + direction_ = kReverse; + } + + FindPrevUserEntry(); +} + +void DBIter::FindPrevUserEntry() { + assert(direction_ == kReverse); + + ValueType value_type = kTypeDeletion; + if (iter_->Valid()) { + do { + ParsedInternalKey ikey; + if (ParseKey(&ikey) && ikey.sequence <= sequence_) { + if ((value_type != kTypeDeletion) && + user_comparator_->Compare(ikey.user_key, saved_key_) < 0) { + // We encountered a non-deleted value in entries for previous keys, + break; + } + value_type = ikey.type; + if (value_type == kTypeDeletion) { + saved_key_.clear(); + ClearSavedValue(); + } else { + Slice raw_value = iter_->value(); + if (saved_value_.capacity() > raw_value.size() + 1048576) { + std::string empty; + swap(empty, saved_value_); + } + SaveKey(ExtractUserKey(iter_->key()), &saved_key_); + saved_value_.assign(raw_value.data(), raw_value.size()); + } + } + iter_->Prev(); + } while (iter_->Valid()); + } + + if (value_type == kTypeDeletion) { + // End + valid_ = false; + saved_key_.clear(); + ClearSavedValue(); + direction_ = kForward; + } else { + valid_ = true; + } +} + +void DBIter::Seek(const Slice& target) { + direction_ = kForward; + ClearSavedValue(); + saved_key_.clear(); + AppendInternalKey( + &saved_key_, ParsedInternalKey(target, sequence_, kValueTypeForSeek)); + iter_->Seek(saved_key_); + if (iter_->Valid()) { + FindNextUserEntry(false, &saved_key_ /* temporary storage */); + } else { + valid_ = false; + } +} + +void DBIter::SeekToFirst() { + direction_ = kForward; + ClearSavedValue(); + iter_->SeekToFirst(); + if (iter_->Valid()) { + FindNextUserEntry(false, &saved_key_ /* temporary storage */); + } else { + valid_ = false; + } +} + +void DBIter::SeekToLast() { + direction_ = kReverse; + ClearSavedValue(); + iter_->SeekToLast(); + FindPrevUserEntry(); +} + +} // anonymous namespace + +Iterator* NewDBIterator( + DBImpl* db, + const Comparator* user_key_comparator, + Iterator* internal_iter, + SequenceNumber sequence, + uint32_t seed) { + return new DBIter(db, user_key_comparator, internal_iter, sequence, seed); +} + +} // namespace leveldb diff --git a/src/leveldb/db/db_iter.h b/src/leveldb/db/db_iter.h new file mode 100644 index 0000000000..04927e937b --- /dev/null +++ b/src/leveldb/db/db_iter.h @@ -0,0 +1,28 @@ +// 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_DB_ITER_H_ +#define STORAGE_LEVELDB_DB_DB_ITER_H_ + +#include <stdint.h> +#include "leveldb/db.h" +#include "db/dbformat.h" + +namespace leveldb { + +class DBImpl; + +// Return a new iterator that converts internal keys (yielded by +// "*internal_iter") that were live at the specified "sequence" number +// into appropriate user keys. +extern Iterator* NewDBIterator( + DBImpl* db, + const Comparator* user_key_comparator, + Iterator* internal_iter, + SequenceNumber sequence, + uint32_t seed); + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_DB_ITER_H_ diff --git a/src/leveldb/db/db_test.cc b/src/leveldb/db/db_test.cc new file mode 100644 index 0000000000..a0b08bc19c --- /dev/null +++ b/src/leveldb/db/db_test.cc @@ -0,0 +1,2158 @@ +// 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 "leveldb/db.h" +#include "leveldb/filter_policy.h" +#include "db/db_impl.h" +#include "db/filename.h" +#include "db/version_set.h" +#include "db/write_batch_internal.h" +#include "leveldb/cache.h" +#include "leveldb/env.h" +#include "leveldb/table.h" +#include "util/hash.h" +#include "util/logging.h" +#include "util/mutexlock.h" +#include "util/testharness.h" +#include "util/testutil.h" + +namespace leveldb { + +static std::string RandomString(Random* rnd, int len) { + std::string r; + test::RandomString(rnd, len, &r); + return r; +} + +namespace { +class AtomicCounter { + private: + port::Mutex mu_; + int count_; + public: + AtomicCounter() : count_(0) { } + void Increment() { + IncrementBy(1); + } + void IncrementBy(int count) { + MutexLock l(&mu_); + count_ += count; + } + int Read() { + MutexLock l(&mu_); + return count_; + } + void Reset() { + MutexLock l(&mu_); + count_ = 0; + } +}; + +void DelayMilliseconds(int millis) { + Env::Default()->SleepForMicroseconds(millis * 1000); +} +} + +// Special Env used to delay background operations +class SpecialEnv : public EnvWrapper { + public: + // sstable/log Sync() calls are blocked while this pointer is non-NULL. + port::AtomicPointer delay_data_sync_; + + // sstable/log Sync() calls return an error. + port::AtomicPointer data_sync_error_; + + // Simulate no-space errors while this pointer is non-NULL. + port::AtomicPointer no_space_; + + // Simulate non-writable file system while this pointer is non-NULL + port::AtomicPointer non_writable_; + + // Force sync of manifest files to fail while this pointer is non-NULL + port::AtomicPointer manifest_sync_error_; + + // Force write to manifest files to fail while this pointer is non-NULL + port::AtomicPointer manifest_write_error_; + + bool count_random_reads_; + AtomicCounter random_read_counter_; + + explicit SpecialEnv(Env* base) : EnvWrapper(base) { + delay_data_sync_.Release_Store(NULL); + data_sync_error_.Release_Store(NULL); + no_space_.Release_Store(NULL); + non_writable_.Release_Store(NULL); + count_random_reads_ = false; + manifest_sync_error_.Release_Store(NULL); + manifest_write_error_.Release_Store(NULL); + } + + Status NewWritableFile(const std::string& f, WritableFile** r) { + class DataFile : public WritableFile { + private: + SpecialEnv* env_; + WritableFile* base_; + + public: + DataFile(SpecialEnv* env, WritableFile* base) + : env_(env), + base_(base) { + } + ~DataFile() { delete base_; } + Status Append(const Slice& data) { + if (env_->no_space_.Acquire_Load() != NULL) { + // Drop writes on the floor + return Status::OK(); + } else { + return base_->Append(data); + } + } + Status Close() { return base_->Close(); } + Status Flush() { return base_->Flush(); } + Status Sync() { + if (env_->data_sync_error_.Acquire_Load() != NULL) { + return Status::IOError("simulated data sync error"); + } + while (env_->delay_data_sync_.Acquire_Load() != NULL) { + DelayMilliseconds(100); + } + return base_->Sync(); + } + }; + class ManifestFile : public WritableFile { + private: + SpecialEnv* env_; + WritableFile* base_; + public: + ManifestFile(SpecialEnv* env, WritableFile* b) : env_(env), base_(b) { } + ~ManifestFile() { delete base_; } + Status Append(const Slice& data) { + if (env_->manifest_write_error_.Acquire_Load() != NULL) { + return Status::IOError("simulated writer error"); + } else { + return base_->Append(data); + } + } + Status Close() { return base_->Close(); } + Status Flush() { return base_->Flush(); } + Status Sync() { + if (env_->manifest_sync_error_.Acquire_Load() != NULL) { + return Status::IOError("simulated sync error"); + } else { + return base_->Sync(); + } + } + }; + + if (non_writable_.Acquire_Load() != NULL) { + return Status::IOError("simulated write error"); + } + + Status s = target()->NewWritableFile(f, r); + if (s.ok()) { + if (strstr(f.c_str(), ".ldb") != NULL || + strstr(f.c_str(), ".log") != NULL) { + *r = new DataFile(this, *r); + } else if (strstr(f.c_str(), "MANIFEST") != NULL) { + *r = new ManifestFile(this, *r); + } + } + return s; + } + + Status NewRandomAccessFile(const std::string& f, RandomAccessFile** r) { + class CountingFile : public RandomAccessFile { + private: + RandomAccessFile* target_; + AtomicCounter* counter_; + public: + CountingFile(RandomAccessFile* target, AtomicCounter* counter) + : target_(target), counter_(counter) { + } + virtual ~CountingFile() { delete target_; } + virtual Status Read(uint64_t offset, size_t n, Slice* result, + char* scratch) const { + counter_->Increment(); + return target_->Read(offset, n, result, scratch); + } + }; + + Status s = target()->NewRandomAccessFile(f, r); + if (s.ok() && count_random_reads_) { + *r = new CountingFile(*r, &random_read_counter_); + } + return s; + } +}; + +class DBTest { + private: + const FilterPolicy* filter_policy_; + + // Sequence of option configurations to try + enum OptionConfig { + kDefault, + kReuse, + kFilter, + kUncompressed, + kEnd + }; + int option_config_; + + public: + std::string dbname_; + SpecialEnv* env_; + DB* db_; + + Options last_options_; + + DBTest() : option_config_(kDefault), + env_(new SpecialEnv(Env::Default())) { + filter_policy_ = NewBloomFilterPolicy(10); + dbname_ = test::TmpDir() + "/db_test"; + DestroyDB(dbname_, Options()); + db_ = NULL; + Reopen(); + } + + ~DBTest() { + delete db_; + DestroyDB(dbname_, Options()); + delete env_; + delete filter_policy_; + } + + // Switch to a fresh database with the next option configuration to + // test. Return false if there are no more configurations to test. + bool ChangeOptions() { + option_config_++; + if (option_config_ >= kEnd) { + return false; + } else { + DestroyAndReopen(); + return true; + } + } + + // Return the current option configuration. + Options CurrentOptions() { + Options options; + options.reuse_logs = false; + switch (option_config_) { + case kReuse: + options.reuse_logs = true; + break; + case kFilter: + options.filter_policy = filter_policy_; + break; + case kUncompressed: + options.compression = kNoCompression; + break; + default: + break; + } + return options; + } + + DBImpl* dbfull() { + return reinterpret_cast<DBImpl*>(db_); + } + + void Reopen(Options* options = NULL) { + ASSERT_OK(TryReopen(options)); + } + + void Close() { + delete db_; + db_ = NULL; + } + + void DestroyAndReopen(Options* options = NULL) { + delete db_; + db_ = NULL; + DestroyDB(dbname_, Options()); + ASSERT_OK(TryReopen(options)); + } + + Status TryReopen(Options* options) { + delete db_; + db_ = NULL; + Options opts; + if (options != NULL) { + opts = *options; + } else { + opts = CurrentOptions(); + opts.create_if_missing = true; + } + last_options_ = opts; + + return DB::Open(opts, dbname_, &db_); + } + + Status Put(const std::string& k, const std::string& v) { + return db_->Put(WriteOptions(), k, v); + } + + Status Delete(const std::string& k) { + return db_->Delete(WriteOptions(), k); + } + + std::string Get(const std::string& k, const Snapshot* snapshot = NULL) { + ReadOptions options; + options.snapshot = snapshot; + std::string result; + Status s = db_->Get(options, k, &result); + if (s.IsNotFound()) { + result = "NOT_FOUND"; + } else if (!s.ok()) { + result = s.ToString(); + } + return result; + } + + // Return a string that contains all key,value pairs in order, + // formatted like "(k1->v1)(k2->v2)". + std::string Contents() { + std::vector<std::string> forward; + std::string result; + Iterator* iter = db_->NewIterator(ReadOptions()); + for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { + std::string s = IterStatus(iter); + result.push_back('('); + result.append(s); + result.push_back(')'); + forward.push_back(s); + } + + // Check reverse iteration results are the reverse of forward results + size_t matched = 0; + for (iter->SeekToLast(); iter->Valid(); iter->Prev()) { + ASSERT_LT(matched, forward.size()); + ASSERT_EQ(IterStatus(iter), forward[forward.size() - matched - 1]); + matched++; + } + ASSERT_EQ(matched, forward.size()); + + delete iter; + return result; + } + + std::string AllEntriesFor(const Slice& user_key) { + Iterator* iter = dbfull()->TEST_NewInternalIterator(); + InternalKey target(user_key, kMaxSequenceNumber, kTypeValue); + iter->Seek(target.Encode()); + std::string result; + if (!iter->status().ok()) { + result = iter->status().ToString(); + } else { + result = "[ "; + bool first = true; + while (iter->Valid()) { + ParsedInternalKey ikey; + if (!ParseInternalKey(iter->key(), &ikey)) { + result += "CORRUPTED"; + } else { + if (last_options_.comparator->Compare(ikey.user_key, user_key) != 0) { + break; + } + if (!first) { + result += ", "; + } + first = false; + switch (ikey.type) { + case kTypeValue: + result += iter->value().ToString(); + break; + case kTypeDeletion: + result += "DEL"; + break; + } + } + iter->Next(); + } + if (!first) { + result += " "; + } + result += "]"; + } + delete iter; + return result; + } + + int NumTableFilesAtLevel(int level) { + std::string property; + ASSERT_TRUE( + db_->GetProperty("leveldb.num-files-at-level" + NumberToString(level), + &property)); + return atoi(property.c_str()); + } + + int TotalTableFiles() { + int result = 0; + for (int level = 0; level < config::kNumLevels; level++) { + result += NumTableFilesAtLevel(level); + } + return result; + } + + // Return spread of files per level + std::string FilesPerLevel() { + std::string result; + int last_non_zero_offset = 0; + for (int level = 0; level < config::kNumLevels; level++) { + int f = NumTableFilesAtLevel(level); + char buf[100]; + snprintf(buf, sizeof(buf), "%s%d", (level ? "," : ""), f); + result += buf; + if (f > 0) { + last_non_zero_offset = result.size(); + } + } + result.resize(last_non_zero_offset); + return result; + } + + int CountFiles() { + std::vector<std::string> files; + env_->GetChildren(dbname_, &files); + return static_cast<int>(files.size()); + } + + uint64_t Size(const Slice& start, const Slice& limit) { + Range r(start, limit); + uint64_t size; + db_->GetApproximateSizes(&r, 1, &size); + return size; + } + + void Compact(const Slice& start, const Slice& limit) { + db_->CompactRange(&start, &limit); + } + + // Do n memtable compactions, each of which produces an sstable + // covering the range [small,large]. + void MakeTables(int n, const std::string& small, const std::string& large) { + for (int i = 0; i < n; i++) { + Put(small, "begin"); + Put(large, "end"); + dbfull()->TEST_CompactMemTable(); + } + } + + // Prevent pushing of new sstables into deeper levels by adding + // tables that cover a specified range to all levels. + void FillLevels(const std::string& smallest, const std::string& largest) { + MakeTables(config::kNumLevels, smallest, largest); + } + + void DumpFileCounts(const char* label) { + fprintf(stderr, "---\n%s:\n", label); + fprintf(stderr, "maxoverlap: %lld\n", + static_cast<long long>( + dbfull()->TEST_MaxNextLevelOverlappingBytes())); + for (int level = 0; level < config::kNumLevels; level++) { + int num = NumTableFilesAtLevel(level); + if (num > 0) { + fprintf(stderr, " level %3d : %d files\n", level, num); + } + } + } + + std::string DumpSSTableList() { + std::string property; + db_->GetProperty("leveldb.sstables", &property); + return property; + } + + std::string IterStatus(Iterator* iter) { + std::string result; + if (iter->Valid()) { + result = iter->key().ToString() + "->" + iter->value().ToString(); + } else { + result = "(invalid)"; + } + return result; + } + + bool DeleteAnSSTFile() { + std::vector<std::string> filenames; + ASSERT_OK(env_->GetChildren(dbname_, &filenames)); + uint64_t number; + FileType type; + for (size_t i = 0; i < filenames.size(); i++) { + if (ParseFileName(filenames[i], &number, &type) && type == kTableFile) { + ASSERT_OK(env_->DeleteFile(TableFileName(dbname_, number))); + return true; + } + } + return false; + } + + // Returns number of files renamed. + int RenameLDBToSST() { + std::vector<std::string> filenames; + ASSERT_OK(env_->GetChildren(dbname_, &filenames)); + uint64_t number; + FileType type; + int files_renamed = 0; + for (size_t i = 0; i < filenames.size(); i++) { + if (ParseFileName(filenames[i], &number, &type) && type == kTableFile) { + const std::string from = TableFileName(dbname_, number); + const std::string to = SSTTableFileName(dbname_, number); + ASSERT_OK(env_->RenameFile(from, to)); + files_renamed++; + } + } + return files_renamed; + } +}; + +TEST(DBTest, Empty) { + do { + ASSERT_TRUE(db_ != NULL); + ASSERT_EQ("NOT_FOUND", Get("foo")); + } while (ChangeOptions()); +} + +TEST(DBTest, ReadWrite) { + do { + ASSERT_OK(Put("foo", "v1")); + ASSERT_EQ("v1", Get("foo")); + ASSERT_OK(Put("bar", "v2")); + ASSERT_OK(Put("foo", "v3")); + ASSERT_EQ("v3", Get("foo")); + ASSERT_EQ("v2", Get("bar")); + } while (ChangeOptions()); +} + +TEST(DBTest, PutDeleteGet) { + do { + ASSERT_OK(db_->Put(WriteOptions(), "foo", "v1")); + ASSERT_EQ("v1", Get("foo")); + ASSERT_OK(db_->Put(WriteOptions(), "foo", "v2")); + ASSERT_EQ("v2", Get("foo")); + ASSERT_OK(db_->Delete(WriteOptions(), "foo")); + ASSERT_EQ("NOT_FOUND", Get("foo")); + } while (ChangeOptions()); +} + +TEST(DBTest, GetFromImmutableLayer) { + do { + Options options = CurrentOptions(); + options.env = env_; + options.write_buffer_size = 100000; // Small write buffer + Reopen(&options); + + ASSERT_OK(Put("foo", "v1")); + ASSERT_EQ("v1", Get("foo")); + + env_->delay_data_sync_.Release_Store(env_); // Block sync calls + Put("k1", std::string(100000, 'x')); // Fill memtable + Put("k2", std::string(100000, 'y')); // Trigger compaction + ASSERT_EQ("v1", Get("foo")); + env_->delay_data_sync_.Release_Store(NULL); // Release sync calls + } while (ChangeOptions()); +} + +TEST(DBTest, GetFromVersions) { + do { + ASSERT_OK(Put("foo", "v1")); + dbfull()->TEST_CompactMemTable(); + ASSERT_EQ("v1", Get("foo")); + } while (ChangeOptions()); +} + +TEST(DBTest, GetMemUsage) { + do { + ASSERT_OK(Put("foo", "v1")); + std::string val; + ASSERT_TRUE(db_->GetProperty("leveldb.approximate-memory-usage", &val)); + int mem_usage = atoi(val.c_str()); + ASSERT_GT(mem_usage, 0); + ASSERT_LT(mem_usage, 5*1024*1024); + } while (ChangeOptions()); +} + +TEST(DBTest, GetSnapshot) { + do { + // Try with both a short key and a long key + for (int i = 0; i < 2; i++) { + std::string key = (i == 0) ? std::string("foo") : std::string(200, 'x'); + ASSERT_OK(Put(key, "v1")); + const Snapshot* s1 = db_->GetSnapshot(); + ASSERT_OK(Put(key, "v2")); + ASSERT_EQ("v2", Get(key)); + ASSERT_EQ("v1", Get(key, s1)); + dbfull()->TEST_CompactMemTable(); + ASSERT_EQ("v2", Get(key)); + ASSERT_EQ("v1", Get(key, s1)); + db_->ReleaseSnapshot(s1); + } + } while (ChangeOptions()); +} + +TEST(DBTest, GetLevel0Ordering) { + do { + // Check that we process level-0 files in correct order. The code + // below generates two level-0 files where the earlier one comes + // before the later one in the level-0 file list since the earlier + // one has a smaller "smallest" key. + ASSERT_OK(Put("bar", "b")); + ASSERT_OK(Put("foo", "v1")); + dbfull()->TEST_CompactMemTable(); + ASSERT_OK(Put("foo", "v2")); + dbfull()->TEST_CompactMemTable(); + ASSERT_EQ("v2", Get("foo")); + } while (ChangeOptions()); +} + +TEST(DBTest, GetOrderedByLevels) { + do { + ASSERT_OK(Put("foo", "v1")); + Compact("a", "z"); + ASSERT_EQ("v1", Get("foo")); + ASSERT_OK(Put("foo", "v2")); + ASSERT_EQ("v2", Get("foo")); + dbfull()->TEST_CompactMemTable(); + ASSERT_EQ("v2", Get("foo")); + } while (ChangeOptions()); +} + +TEST(DBTest, GetPicksCorrectFile) { + do { + // Arrange to have multiple files in a non-level-0 level. + ASSERT_OK(Put("a", "va")); + Compact("a", "b"); + ASSERT_OK(Put("x", "vx")); + Compact("x", "y"); + ASSERT_OK(Put("f", "vf")); + Compact("f", "g"); + ASSERT_EQ("va", Get("a")); + ASSERT_EQ("vf", Get("f")); + ASSERT_EQ("vx", Get("x")); + } while (ChangeOptions()); +} + +TEST(DBTest, GetEncountersEmptyLevel) { + do { + // Arrange for the following to happen: + // * sstable A in level 0 + // * nothing in level 1 + // * sstable B in level 2 + // Then do enough Get() calls to arrange for an automatic compaction + // of sstable A. A bug would cause the compaction to be marked as + // occurring at level 1 (instead of the correct level 0). + + // Step 1: First place sstables in levels 0 and 2 + int compaction_count = 0; + while (NumTableFilesAtLevel(0) == 0 || + NumTableFilesAtLevel(2) == 0) { + ASSERT_LE(compaction_count, 100) << "could not fill levels 0 and 2"; + compaction_count++; + Put("a", "begin"); + Put("z", "end"); + dbfull()->TEST_CompactMemTable(); + } + + // Step 2: clear level 1 if necessary. + dbfull()->TEST_CompactRange(1, NULL, NULL); + ASSERT_EQ(NumTableFilesAtLevel(0), 1); + ASSERT_EQ(NumTableFilesAtLevel(1), 0); + ASSERT_EQ(NumTableFilesAtLevel(2), 1); + + // Step 3: read a bunch of times + for (int i = 0; i < 1000; i++) { + ASSERT_EQ("NOT_FOUND", Get("missing")); + } + + // Step 4: Wait for compaction to finish + DelayMilliseconds(1000); + + ASSERT_EQ(NumTableFilesAtLevel(0), 0); + } while (ChangeOptions()); +} + +TEST(DBTest, IterEmpty) { + Iterator* iter = db_->NewIterator(ReadOptions()); + + iter->SeekToFirst(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + + iter->SeekToLast(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + + iter->Seek("foo"); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + + delete iter; +} + +TEST(DBTest, IterSingle) { + ASSERT_OK(Put("a", "va")); + Iterator* iter = db_->NewIterator(ReadOptions()); + + iter->SeekToFirst(); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + iter->SeekToFirst(); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + + iter->SeekToLast(); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + iter->SeekToLast(); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + + iter->Seek(""); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + + iter->Seek("a"); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + + iter->Seek("b"); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + + delete iter; +} + +TEST(DBTest, IterMulti) { + ASSERT_OK(Put("a", "va")); + ASSERT_OK(Put("b", "vb")); + ASSERT_OK(Put("c", "vc")); + Iterator* iter = db_->NewIterator(ReadOptions()); + + iter->SeekToFirst(); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "b->vb"); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "c->vc"); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + iter->SeekToFirst(); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + + iter->SeekToLast(); + ASSERT_EQ(IterStatus(iter), "c->vc"); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "b->vb"); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + iter->SeekToLast(); + ASSERT_EQ(IterStatus(iter), "c->vc"); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + + iter->Seek(""); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Seek("a"); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Seek("ax"); + ASSERT_EQ(IterStatus(iter), "b->vb"); + iter->Seek("b"); + ASSERT_EQ(IterStatus(iter), "b->vb"); + iter->Seek("z"); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + + // Switch from reverse to forward + iter->SeekToLast(); + iter->Prev(); + iter->Prev(); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "b->vb"); + + // Switch from forward to reverse + iter->SeekToFirst(); + iter->Next(); + iter->Next(); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "b->vb"); + + // Make sure iter stays at snapshot + ASSERT_OK(Put("a", "va2")); + ASSERT_OK(Put("a2", "va3")); + ASSERT_OK(Put("b", "vb2")); + ASSERT_OK(Put("c", "vc2")); + ASSERT_OK(Delete("b")); + iter->SeekToFirst(); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "b->vb"); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "c->vc"); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + iter->SeekToLast(); + ASSERT_EQ(IterStatus(iter), "c->vc"); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "b->vb"); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + + delete iter; +} + +TEST(DBTest, IterSmallAndLargeMix) { + ASSERT_OK(Put("a", "va")); + ASSERT_OK(Put("b", std::string(100000, 'b'))); + ASSERT_OK(Put("c", "vc")); + ASSERT_OK(Put("d", std::string(100000, 'd'))); + ASSERT_OK(Put("e", std::string(100000, 'e'))); + + Iterator* iter = db_->NewIterator(ReadOptions()); + + iter->SeekToFirst(); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "b->" + std::string(100000, 'b')); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "c->vc"); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "d->" + std::string(100000, 'd')); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "e->" + std::string(100000, 'e')); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + + iter->SeekToLast(); + ASSERT_EQ(IterStatus(iter), "e->" + std::string(100000, 'e')); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "d->" + std::string(100000, 'd')); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "c->vc"); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "b->" + std::string(100000, 'b')); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + + delete iter; +} + +TEST(DBTest, IterMultiWithDelete) { + do { + ASSERT_OK(Put("a", "va")); + ASSERT_OK(Put("b", "vb")); + ASSERT_OK(Put("c", "vc")); + ASSERT_OK(Delete("b")); + ASSERT_EQ("NOT_FOUND", Get("b")); + + Iterator* iter = db_->NewIterator(ReadOptions()); + iter->Seek("c"); + ASSERT_EQ(IterStatus(iter), "c->vc"); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "a->va"); + delete iter; + } while (ChangeOptions()); +} + +TEST(DBTest, Recover) { + do { + ASSERT_OK(Put("foo", "v1")); + ASSERT_OK(Put("baz", "v5")); + + Reopen(); + ASSERT_EQ("v1", Get("foo")); + + ASSERT_EQ("v1", Get("foo")); + ASSERT_EQ("v5", Get("baz")); + ASSERT_OK(Put("bar", "v2")); + ASSERT_OK(Put("foo", "v3")); + + Reopen(); + ASSERT_EQ("v3", Get("foo")); + ASSERT_OK(Put("foo", "v4")); + ASSERT_EQ("v4", Get("foo")); + ASSERT_EQ("v2", Get("bar")); + ASSERT_EQ("v5", Get("baz")); + } while (ChangeOptions()); +} + +TEST(DBTest, RecoveryWithEmptyLog) { + do { + ASSERT_OK(Put("foo", "v1")); + ASSERT_OK(Put("foo", "v2")); + Reopen(); + Reopen(); + ASSERT_OK(Put("foo", "v3")); + Reopen(); + ASSERT_EQ("v3", Get("foo")); + } while (ChangeOptions()); +} + +// Check that writes done during a memtable compaction are recovered +// if the database is shutdown during the memtable compaction. +TEST(DBTest, RecoverDuringMemtableCompaction) { + do { + Options options = CurrentOptions(); + options.env = env_; + options.write_buffer_size = 1000000; + Reopen(&options); + + // Trigger a long memtable compaction and reopen the database during it + ASSERT_OK(Put("foo", "v1")); // Goes to 1st log file + ASSERT_OK(Put("big1", std::string(10000000, 'x'))); // Fills memtable + ASSERT_OK(Put("big2", std::string(1000, 'y'))); // Triggers compaction + ASSERT_OK(Put("bar", "v2")); // Goes to new log file + + Reopen(&options); + ASSERT_EQ("v1", Get("foo")); + ASSERT_EQ("v2", Get("bar")); + ASSERT_EQ(std::string(10000000, 'x'), Get("big1")); + ASSERT_EQ(std::string(1000, 'y'), Get("big2")); + } while (ChangeOptions()); +} + +static std::string Key(int i) { + char buf[100]; + snprintf(buf, sizeof(buf), "key%06d", i); + return std::string(buf); +} + +TEST(DBTest, MinorCompactionsHappen) { + Options options = CurrentOptions(); + options.write_buffer_size = 10000; + Reopen(&options); + + const int N = 500; + + int starting_num_tables = TotalTableFiles(); + for (int i = 0; i < N; i++) { + ASSERT_OK(Put(Key(i), Key(i) + std::string(1000, 'v'))); + } + int ending_num_tables = TotalTableFiles(); + ASSERT_GT(ending_num_tables, starting_num_tables); + + for (int i = 0; i < N; i++) { + ASSERT_EQ(Key(i) + std::string(1000, 'v'), Get(Key(i))); + } + + Reopen(); + + for (int i = 0; i < N; i++) { + ASSERT_EQ(Key(i) + std::string(1000, 'v'), Get(Key(i))); + } +} + +TEST(DBTest, RecoverWithLargeLog) { + { + Options options = CurrentOptions(); + Reopen(&options); + ASSERT_OK(Put("big1", std::string(200000, '1'))); + ASSERT_OK(Put("big2", std::string(200000, '2'))); + ASSERT_OK(Put("small3", std::string(10, '3'))); + ASSERT_OK(Put("small4", std::string(10, '4'))); + ASSERT_EQ(NumTableFilesAtLevel(0), 0); + } + + // Make sure that if we re-open with a small write buffer size that + // we flush table files in the middle of a large log file. + Options options = CurrentOptions(); + options.write_buffer_size = 100000; + Reopen(&options); + ASSERT_EQ(NumTableFilesAtLevel(0), 3); + ASSERT_EQ(std::string(200000, '1'), Get("big1")); + ASSERT_EQ(std::string(200000, '2'), Get("big2")); + ASSERT_EQ(std::string(10, '3'), Get("small3")); + ASSERT_EQ(std::string(10, '4'), Get("small4")); + ASSERT_GT(NumTableFilesAtLevel(0), 1); +} + +TEST(DBTest, CompactionsGenerateMultipleFiles) { + Options options = CurrentOptions(); + options.write_buffer_size = 100000000; // Large write buffer + Reopen(&options); + + Random rnd(301); + + // Write 8MB (80 values, each 100K) + ASSERT_EQ(NumTableFilesAtLevel(0), 0); + std::vector<std::string> values; + for (int i = 0; i < 80; i++) { + values.push_back(RandomString(&rnd, 100000)); + ASSERT_OK(Put(Key(i), values[i])); + } + + // Reopening moves updates to level-0 + Reopen(&options); + dbfull()->TEST_CompactRange(0, NULL, NULL); + + ASSERT_EQ(NumTableFilesAtLevel(0), 0); + ASSERT_GT(NumTableFilesAtLevel(1), 1); + for (int i = 0; i < 80; i++) { + ASSERT_EQ(Get(Key(i)), values[i]); + } +} + +TEST(DBTest, RepeatedWritesToSameKey) { + Options options = CurrentOptions(); + options.env = env_; + options.write_buffer_size = 100000; // Small write buffer + Reopen(&options); + + // We must have at most one file per level except for level-0, + // which may have up to kL0_StopWritesTrigger files. + const int kMaxFiles = config::kNumLevels + config::kL0_StopWritesTrigger; + + Random rnd(301); + std::string value = RandomString(&rnd, 2 * options.write_buffer_size); + for (int i = 0; i < 5 * kMaxFiles; i++) { + Put("key", value); + ASSERT_LE(TotalTableFiles(), kMaxFiles); + fprintf(stderr, "after %d: %d files\n", int(i+1), TotalTableFiles()); + } +} + +TEST(DBTest, SparseMerge) { + Options options = CurrentOptions(); + options.compression = kNoCompression; + Reopen(&options); + + FillLevels("A", "Z"); + + // Suppose there is: + // small amount of data with prefix A + // large amount of data with prefix B + // small amount of data with prefix C + // and that recent updates have made small changes to all three prefixes. + // Check that we do not do a compaction that merges all of B in one shot. + const std::string value(1000, 'x'); + Put("A", "va"); + // Write approximately 100MB of "B" values + for (int i = 0; i < 100000; i++) { + char key[100]; + snprintf(key, sizeof(key), "B%010d", i); + Put(key, value); + } + Put("C", "vc"); + dbfull()->TEST_CompactMemTable(); + dbfull()->TEST_CompactRange(0, NULL, NULL); + + // Make sparse update + Put("A", "va2"); + Put("B100", "bvalue2"); + Put("C", "vc2"); + dbfull()->TEST_CompactMemTable(); + + // Compactions should not cause us to create a situation where + // a file overlaps too much data at the next level. + ASSERT_LE(dbfull()->TEST_MaxNextLevelOverlappingBytes(), 20*1048576); + dbfull()->TEST_CompactRange(0, NULL, NULL); + ASSERT_LE(dbfull()->TEST_MaxNextLevelOverlappingBytes(), 20*1048576); + dbfull()->TEST_CompactRange(1, NULL, NULL); + ASSERT_LE(dbfull()->TEST_MaxNextLevelOverlappingBytes(), 20*1048576); +} + +static bool Between(uint64_t val, uint64_t low, uint64_t high) { + bool result = (val >= low) && (val <= high); + if (!result) { + fprintf(stderr, "Value %llu is not in range [%llu, %llu]\n", + (unsigned long long)(val), + (unsigned long long)(low), + (unsigned long long)(high)); + } + return result; +} + +TEST(DBTest, ApproximateSizes) { + do { + Options options = CurrentOptions(); + options.write_buffer_size = 100000000; // Large write buffer + options.compression = kNoCompression; + DestroyAndReopen(); + + ASSERT_TRUE(Between(Size("", "xyz"), 0, 0)); + Reopen(&options); + ASSERT_TRUE(Between(Size("", "xyz"), 0, 0)); + + // Write 8MB (80 values, each 100K) + ASSERT_EQ(NumTableFilesAtLevel(0), 0); + const int N = 80; + static const int S1 = 100000; + static const int S2 = 105000; // Allow some expansion from metadata + Random rnd(301); + for (int i = 0; i < N; i++) { + ASSERT_OK(Put(Key(i), RandomString(&rnd, S1))); + } + + // 0 because GetApproximateSizes() does not account for memtable space + ASSERT_TRUE(Between(Size("", Key(50)), 0, 0)); + + if (options.reuse_logs) { + // Recovery will reuse memtable, and GetApproximateSizes() does not + // account for memtable usage; + Reopen(&options); + ASSERT_TRUE(Between(Size("", Key(50)), 0, 0)); + continue; + } + + // Check sizes across recovery by reopening a few times + for (int run = 0; run < 3; run++) { + Reopen(&options); + + for (int compact_start = 0; compact_start < N; compact_start += 10) { + for (int i = 0; i < N; i += 10) { + ASSERT_TRUE(Between(Size("", Key(i)), S1*i, S2*i)); + ASSERT_TRUE(Between(Size("", Key(i)+".suffix"), S1*(i+1), S2*(i+1))); + ASSERT_TRUE(Between(Size(Key(i), Key(i+10)), S1*10, S2*10)); + } + ASSERT_TRUE(Between(Size("", Key(50)), S1*50, S2*50)); + ASSERT_TRUE(Between(Size("", Key(50)+".suffix"), S1*50, S2*50)); + + std::string cstart_str = Key(compact_start); + std::string cend_str = Key(compact_start + 9); + Slice cstart = cstart_str; + Slice cend = cend_str; + dbfull()->TEST_CompactRange(0, &cstart, &cend); + } + + ASSERT_EQ(NumTableFilesAtLevel(0), 0); + ASSERT_GT(NumTableFilesAtLevel(1), 0); + } + } while (ChangeOptions()); +} + +TEST(DBTest, ApproximateSizes_MixOfSmallAndLarge) { + do { + Options options = CurrentOptions(); + options.compression = kNoCompression; + Reopen(); + + Random rnd(301); + std::string big1 = RandomString(&rnd, 100000); + ASSERT_OK(Put(Key(0), RandomString(&rnd, 10000))); + ASSERT_OK(Put(Key(1), RandomString(&rnd, 10000))); + ASSERT_OK(Put(Key(2), big1)); + ASSERT_OK(Put(Key(3), RandomString(&rnd, 10000))); + ASSERT_OK(Put(Key(4), big1)); + ASSERT_OK(Put(Key(5), RandomString(&rnd, 10000))); + ASSERT_OK(Put(Key(6), RandomString(&rnd, 300000))); + ASSERT_OK(Put(Key(7), RandomString(&rnd, 10000))); + + if (options.reuse_logs) { + // Need to force a memtable compaction since recovery does not do so. + ASSERT_OK(dbfull()->TEST_CompactMemTable()); + } + + // Check sizes across recovery by reopening a few times + for (int run = 0; run < 3; run++) { + Reopen(&options); + + ASSERT_TRUE(Between(Size("", Key(0)), 0, 0)); + ASSERT_TRUE(Between(Size("", Key(1)), 10000, 11000)); + ASSERT_TRUE(Between(Size("", Key(2)), 20000, 21000)); + ASSERT_TRUE(Between(Size("", Key(3)), 120000, 121000)); + ASSERT_TRUE(Between(Size("", Key(4)), 130000, 131000)); + ASSERT_TRUE(Between(Size("", Key(5)), 230000, 231000)); + ASSERT_TRUE(Between(Size("", Key(6)), 240000, 241000)); + ASSERT_TRUE(Between(Size("", Key(7)), 540000, 541000)); + ASSERT_TRUE(Between(Size("", Key(8)), 550000, 560000)); + + ASSERT_TRUE(Between(Size(Key(3), Key(5)), 110000, 111000)); + + dbfull()->TEST_CompactRange(0, NULL, NULL); + } + } while (ChangeOptions()); +} + +TEST(DBTest, IteratorPinsRef) { + Put("foo", "hello"); + + // Get iterator that will yield the current contents of the DB. + Iterator* iter = db_->NewIterator(ReadOptions()); + + // Write to force compactions + Put("foo", "newvalue1"); + for (int i = 0; i < 100; i++) { + ASSERT_OK(Put(Key(i), Key(i) + std::string(100000, 'v'))); // 100K values + } + Put("foo", "newvalue2"); + + iter->SeekToFirst(); + ASSERT_TRUE(iter->Valid()); + ASSERT_EQ("foo", iter->key().ToString()); + ASSERT_EQ("hello", iter->value().ToString()); + iter->Next(); + ASSERT_TRUE(!iter->Valid()); + delete iter; +} + +TEST(DBTest, Snapshot) { + do { + Put("foo", "v1"); + const Snapshot* s1 = db_->GetSnapshot(); + Put("foo", "v2"); + const Snapshot* s2 = db_->GetSnapshot(); + Put("foo", "v3"); + const Snapshot* s3 = db_->GetSnapshot(); + + Put("foo", "v4"); + ASSERT_EQ("v1", Get("foo", s1)); + ASSERT_EQ("v2", Get("foo", s2)); + ASSERT_EQ("v3", Get("foo", s3)); + ASSERT_EQ("v4", Get("foo")); + + db_->ReleaseSnapshot(s3); + ASSERT_EQ("v1", Get("foo", s1)); + ASSERT_EQ("v2", Get("foo", s2)); + ASSERT_EQ("v4", Get("foo")); + + db_->ReleaseSnapshot(s1); + ASSERT_EQ("v2", Get("foo", s2)); + ASSERT_EQ("v4", Get("foo")); + + db_->ReleaseSnapshot(s2); + ASSERT_EQ("v4", Get("foo")); + } while (ChangeOptions()); +} + +TEST(DBTest, HiddenValuesAreRemoved) { + do { + Random rnd(301); + FillLevels("a", "z"); + + std::string big = RandomString(&rnd, 50000); + Put("foo", big); + Put("pastfoo", "v"); + const Snapshot* snapshot = db_->GetSnapshot(); + Put("foo", "tiny"); + Put("pastfoo2", "v2"); // Advance sequence number one more + + ASSERT_OK(dbfull()->TEST_CompactMemTable()); + ASSERT_GT(NumTableFilesAtLevel(0), 0); + + ASSERT_EQ(big, Get("foo", snapshot)); + ASSERT_TRUE(Between(Size("", "pastfoo"), 50000, 60000)); + db_->ReleaseSnapshot(snapshot); + ASSERT_EQ(AllEntriesFor("foo"), "[ tiny, " + big + " ]"); + Slice x("x"); + dbfull()->TEST_CompactRange(0, NULL, &x); + ASSERT_EQ(AllEntriesFor("foo"), "[ tiny ]"); + ASSERT_EQ(NumTableFilesAtLevel(0), 0); + ASSERT_GE(NumTableFilesAtLevel(1), 1); + dbfull()->TEST_CompactRange(1, NULL, &x); + ASSERT_EQ(AllEntriesFor("foo"), "[ tiny ]"); + + ASSERT_TRUE(Between(Size("", "pastfoo"), 0, 1000)); + } while (ChangeOptions()); +} + +TEST(DBTest, DeletionMarkers1) { + Put("foo", "v1"); + ASSERT_OK(dbfull()->TEST_CompactMemTable()); + const int last = config::kMaxMemCompactLevel; + ASSERT_EQ(NumTableFilesAtLevel(last), 1); // foo => v1 is now in last level + + // Place a table at level last-1 to prevent merging with preceding mutation + Put("a", "begin"); + Put("z", "end"); + dbfull()->TEST_CompactMemTable(); + ASSERT_EQ(NumTableFilesAtLevel(last), 1); + ASSERT_EQ(NumTableFilesAtLevel(last-1), 1); + + Delete("foo"); + Put("foo", "v2"); + ASSERT_EQ(AllEntriesFor("foo"), "[ v2, DEL, v1 ]"); + ASSERT_OK(dbfull()->TEST_CompactMemTable()); // Moves to level last-2 + ASSERT_EQ(AllEntriesFor("foo"), "[ v2, DEL, v1 ]"); + Slice z("z"); + dbfull()->TEST_CompactRange(last-2, NULL, &z); + // DEL eliminated, but v1 remains because we aren't compacting that level + // (DEL can be eliminated because v2 hides v1). + ASSERT_EQ(AllEntriesFor("foo"), "[ v2, v1 ]"); + dbfull()->TEST_CompactRange(last-1, NULL, NULL); + // Merging last-1 w/ last, so we are the base level for "foo", so + // DEL is removed. (as is v1). + ASSERT_EQ(AllEntriesFor("foo"), "[ v2 ]"); +} + +TEST(DBTest, DeletionMarkers2) { + Put("foo", "v1"); + ASSERT_OK(dbfull()->TEST_CompactMemTable()); + const int last = config::kMaxMemCompactLevel; + ASSERT_EQ(NumTableFilesAtLevel(last), 1); // foo => v1 is now in last level + + // Place a table at level last-1 to prevent merging with preceding mutation + Put("a", "begin"); + Put("z", "end"); + dbfull()->TEST_CompactMemTable(); + ASSERT_EQ(NumTableFilesAtLevel(last), 1); + ASSERT_EQ(NumTableFilesAtLevel(last-1), 1); + + Delete("foo"); + ASSERT_EQ(AllEntriesFor("foo"), "[ DEL, v1 ]"); + ASSERT_OK(dbfull()->TEST_CompactMemTable()); // Moves to level last-2 + ASSERT_EQ(AllEntriesFor("foo"), "[ DEL, v1 ]"); + dbfull()->TEST_CompactRange(last-2, NULL, NULL); + // DEL kept: "last" file overlaps + ASSERT_EQ(AllEntriesFor("foo"), "[ DEL, v1 ]"); + dbfull()->TEST_CompactRange(last-1, NULL, NULL); + // Merging last-1 w/ last, so we are the base level for "foo", so + // DEL is removed. (as is v1). + ASSERT_EQ(AllEntriesFor("foo"), "[ ]"); +} + +TEST(DBTest, OverlapInLevel0) { + do { + ASSERT_EQ(config::kMaxMemCompactLevel, 2) << "Fix test to match config"; + + // Fill levels 1 and 2 to disable the pushing of new memtables to levels > 0. + ASSERT_OK(Put("100", "v100")); + ASSERT_OK(Put("999", "v999")); + dbfull()->TEST_CompactMemTable(); + ASSERT_OK(Delete("100")); + ASSERT_OK(Delete("999")); + dbfull()->TEST_CompactMemTable(); + ASSERT_EQ("0,1,1", FilesPerLevel()); + + // Make files spanning the following ranges in level-0: + // files[0] 200 .. 900 + // files[1] 300 .. 500 + // Note that files are sorted by smallest key. + ASSERT_OK(Put("300", "v300")); + ASSERT_OK(Put("500", "v500")); + dbfull()->TEST_CompactMemTable(); + ASSERT_OK(Put("200", "v200")); + ASSERT_OK(Put("600", "v600")); + ASSERT_OK(Put("900", "v900")); + dbfull()->TEST_CompactMemTable(); + ASSERT_EQ("2,1,1", FilesPerLevel()); + + // Compact away the placeholder files we created initially + dbfull()->TEST_CompactRange(1, NULL, NULL); + dbfull()->TEST_CompactRange(2, NULL, NULL); + ASSERT_EQ("2", FilesPerLevel()); + + // Do a memtable compaction. Before bug-fix, the compaction would + // not detect the overlap with level-0 files and would incorrectly place + // the deletion in a deeper level. + ASSERT_OK(Delete("600")); + dbfull()->TEST_CompactMemTable(); + ASSERT_EQ("3", FilesPerLevel()); + ASSERT_EQ("NOT_FOUND", Get("600")); + } while (ChangeOptions()); +} + +TEST(DBTest, L0_CompactionBug_Issue44_a) { + Reopen(); + ASSERT_OK(Put("b", "v")); + Reopen(); + ASSERT_OK(Delete("b")); + ASSERT_OK(Delete("a")); + Reopen(); + ASSERT_OK(Delete("a")); + Reopen(); + ASSERT_OK(Put("a", "v")); + Reopen(); + Reopen(); + ASSERT_EQ("(a->v)", Contents()); + DelayMilliseconds(1000); // Wait for compaction to finish + ASSERT_EQ("(a->v)", Contents()); +} + +TEST(DBTest, L0_CompactionBug_Issue44_b) { + Reopen(); + Put("",""); + Reopen(); + Delete("e"); + Put("",""); + Reopen(); + Put("c", "cv"); + Reopen(); + Put("",""); + Reopen(); + Put("",""); + DelayMilliseconds(1000); // Wait for compaction to finish + Reopen(); + Put("d","dv"); + Reopen(); + Put("",""); + Reopen(); + Delete("d"); + Delete("b"); + Reopen(); + ASSERT_EQ("(->)(c->cv)", Contents()); + DelayMilliseconds(1000); // Wait for compaction to finish + ASSERT_EQ("(->)(c->cv)", Contents()); +} + +TEST(DBTest, ComparatorCheck) { + class NewComparator : public Comparator { + public: + virtual const char* Name() const { return "leveldb.NewComparator"; } + virtual int Compare(const Slice& a, const Slice& b) const { + return BytewiseComparator()->Compare(a, b); + } + virtual void FindShortestSeparator(std::string* s, const Slice& l) const { + BytewiseComparator()->FindShortestSeparator(s, l); + } + virtual void FindShortSuccessor(std::string* key) const { + BytewiseComparator()->FindShortSuccessor(key); + } + }; + NewComparator cmp; + Options new_options = CurrentOptions(); + new_options.comparator = &cmp; + Status s = TryReopen(&new_options); + ASSERT_TRUE(!s.ok()); + ASSERT_TRUE(s.ToString().find("comparator") != std::string::npos) + << s.ToString(); +} + +TEST(DBTest, CustomComparator) { + class NumberComparator : public Comparator { + public: + virtual const char* Name() const { return "test.NumberComparator"; } + virtual int Compare(const Slice& a, const Slice& b) const { + return ToNumber(a) - ToNumber(b); + } + virtual void FindShortestSeparator(std::string* s, const Slice& l) const { + ToNumber(*s); // Check format + ToNumber(l); // Check format + } + virtual void FindShortSuccessor(std::string* key) const { + ToNumber(*key); // Check format + } + private: + static int ToNumber(const Slice& x) { + // Check that there are no extra characters. + ASSERT_TRUE(x.size() >= 2 && x[0] == '[' && x[x.size()-1] == ']') + << EscapeString(x); + int val; + char ignored; + ASSERT_TRUE(sscanf(x.ToString().c_str(), "[%i]%c", &val, &ignored) == 1) + << EscapeString(x); + return val; + } + }; + NumberComparator cmp; + Options new_options = CurrentOptions(); + new_options.create_if_missing = true; + new_options.comparator = &cmp; + new_options.filter_policy = NULL; // Cannot use bloom filters + new_options.write_buffer_size = 1000; // Compact more often + DestroyAndReopen(&new_options); + ASSERT_OK(Put("[10]", "ten")); + ASSERT_OK(Put("[0x14]", "twenty")); + for (int i = 0; i < 2; i++) { + ASSERT_EQ("ten", Get("[10]")); + ASSERT_EQ("ten", Get("[0xa]")); + ASSERT_EQ("twenty", Get("[20]")); + ASSERT_EQ("twenty", Get("[0x14]")); + ASSERT_EQ("NOT_FOUND", Get("[15]")); + ASSERT_EQ("NOT_FOUND", Get("[0xf]")); + Compact("[0]", "[9999]"); + } + + for (int run = 0; run < 2; run++) { + for (int i = 0; i < 1000; i++) { + char buf[100]; + snprintf(buf, sizeof(buf), "[%d]", i*10); + ASSERT_OK(Put(buf, buf)); + } + Compact("[0]", "[1000000]"); + } +} + +TEST(DBTest, ManualCompaction) { + ASSERT_EQ(config::kMaxMemCompactLevel, 2) + << "Need to update this test to match kMaxMemCompactLevel"; + + MakeTables(3, "p", "q"); + ASSERT_EQ("1,1,1", FilesPerLevel()); + + // Compaction range falls before files + Compact("", "c"); + ASSERT_EQ("1,1,1", FilesPerLevel()); + + // Compaction range falls after files + Compact("r", "z"); + ASSERT_EQ("1,1,1", FilesPerLevel()); + + // Compaction range overlaps files + Compact("p1", "p9"); + ASSERT_EQ("0,0,1", FilesPerLevel()); + + // Populate a different range + MakeTables(3, "c", "e"); + ASSERT_EQ("1,1,2", FilesPerLevel()); + + // Compact just the new range + Compact("b", "f"); + ASSERT_EQ("0,0,2", FilesPerLevel()); + + // Compact all + MakeTables(1, "a", "z"); + ASSERT_EQ("0,1,2", FilesPerLevel()); + db_->CompactRange(NULL, NULL); + ASSERT_EQ("0,0,1", FilesPerLevel()); +} + +TEST(DBTest, DBOpen_Options) { + std::string dbname = test::TmpDir() + "/db_options_test"; + DestroyDB(dbname, Options()); + + // Does not exist, and create_if_missing == false: error + DB* db = NULL; + Options opts; + opts.create_if_missing = false; + Status s = DB::Open(opts, dbname, &db); + ASSERT_TRUE(strstr(s.ToString().c_str(), "does not exist") != NULL); + ASSERT_TRUE(db == NULL); + + // Does not exist, and create_if_missing == true: OK + opts.create_if_missing = true; + s = DB::Open(opts, dbname, &db); + ASSERT_OK(s); + ASSERT_TRUE(db != NULL); + + delete db; + db = NULL; + + // Does exist, and error_if_exists == true: error + opts.create_if_missing = false; + opts.error_if_exists = true; + s = DB::Open(opts, dbname, &db); + ASSERT_TRUE(strstr(s.ToString().c_str(), "exists") != NULL); + ASSERT_TRUE(db == NULL); + + // Does exist, and error_if_exists == false: OK + opts.create_if_missing = true; + opts.error_if_exists = false; + s = DB::Open(opts, dbname, &db); + ASSERT_OK(s); + ASSERT_TRUE(db != NULL); + + delete db; + db = NULL; +} + +TEST(DBTest, Locking) { + DB* db2 = NULL; + Status s = DB::Open(CurrentOptions(), dbname_, &db2); + ASSERT_TRUE(!s.ok()) << "Locking did not prevent re-opening db"; +} + +// Check that number of files does not grow when we are out of space +TEST(DBTest, NoSpace) { + Options options = CurrentOptions(); + options.env = env_; + Reopen(&options); + + ASSERT_OK(Put("foo", "v1")); + ASSERT_EQ("v1", Get("foo")); + Compact("a", "z"); + const int num_files = CountFiles(); + env_->no_space_.Release_Store(env_); // Force out-of-space errors + for (int i = 0; i < 10; i++) { + for (int level = 0; level < config::kNumLevels-1; level++) { + dbfull()->TEST_CompactRange(level, NULL, NULL); + } + } + env_->no_space_.Release_Store(NULL); + ASSERT_LT(CountFiles(), num_files + 3); +} + +TEST(DBTest, NonWritableFileSystem) { + Options options = CurrentOptions(); + options.write_buffer_size = 1000; + options.env = env_; + Reopen(&options); + ASSERT_OK(Put("foo", "v1")); + env_->non_writable_.Release_Store(env_); // Force errors for new files + std::string big(100000, 'x'); + int errors = 0; + for (int i = 0; i < 20; i++) { + fprintf(stderr, "iter %d; errors %d\n", i, errors); + if (!Put("foo", big).ok()) { + errors++; + DelayMilliseconds(100); + } + } + ASSERT_GT(errors, 0); + env_->non_writable_.Release_Store(NULL); +} + +TEST(DBTest, WriteSyncError) { + // Check that log sync errors cause the DB to disallow future writes. + + // (a) Cause log sync calls to fail + Options options = CurrentOptions(); + options.env = env_; + Reopen(&options); + env_->data_sync_error_.Release_Store(env_); + + // (b) Normal write should succeed + WriteOptions w; + ASSERT_OK(db_->Put(w, "k1", "v1")); + ASSERT_EQ("v1", Get("k1")); + + // (c) Do a sync write; should fail + w.sync = true; + ASSERT_TRUE(!db_->Put(w, "k2", "v2").ok()); + ASSERT_EQ("v1", Get("k1")); + ASSERT_EQ("NOT_FOUND", Get("k2")); + + // (d) make sync behave normally + env_->data_sync_error_.Release_Store(NULL); + + // (e) Do a non-sync write; should fail + w.sync = false; + ASSERT_TRUE(!db_->Put(w, "k3", "v3").ok()); + ASSERT_EQ("v1", Get("k1")); + ASSERT_EQ("NOT_FOUND", Get("k2")); + ASSERT_EQ("NOT_FOUND", Get("k3")); +} + +TEST(DBTest, ManifestWriteError) { + // Test for the following problem: + // (a) Compaction produces file F + // (b) Log record containing F is written to MANIFEST file, but Sync() fails + // (c) GC deletes F + // (d) After reopening DB, reads fail since deleted F is named in log record + + // We iterate twice. In the second iteration, everything is the + // same except the log record never makes it to the MANIFEST file. + for (int iter = 0; iter < 2; iter++) { + port::AtomicPointer* error_type = (iter == 0) + ? &env_->manifest_sync_error_ + : &env_->manifest_write_error_; + + // Insert foo=>bar mapping + Options options = CurrentOptions(); + options.env = env_; + options.create_if_missing = true; + options.error_if_exists = false; + DestroyAndReopen(&options); + ASSERT_OK(Put("foo", "bar")); + ASSERT_EQ("bar", Get("foo")); + + // Memtable compaction (will succeed) + dbfull()->TEST_CompactMemTable(); + ASSERT_EQ("bar", Get("foo")); + const int last = config::kMaxMemCompactLevel; + ASSERT_EQ(NumTableFilesAtLevel(last), 1); // foo=>bar is now in last level + + // Merging compaction (will fail) + error_type->Release_Store(env_); + dbfull()->TEST_CompactRange(last, NULL, NULL); // Should fail + ASSERT_EQ("bar", Get("foo")); + + // Recovery: should not lose data + error_type->Release_Store(NULL); + Reopen(&options); + ASSERT_EQ("bar", Get("foo")); + } +} + +TEST(DBTest, MissingSSTFile) { + ASSERT_OK(Put("foo", "bar")); + ASSERT_EQ("bar", Get("foo")); + + // Dump the memtable to disk. + dbfull()->TEST_CompactMemTable(); + ASSERT_EQ("bar", Get("foo")); + + Close(); + ASSERT_TRUE(DeleteAnSSTFile()); + Options options = CurrentOptions(); + options.paranoid_checks = true; + Status s = TryReopen(&options); + ASSERT_TRUE(!s.ok()); + ASSERT_TRUE(s.ToString().find("issing") != std::string::npos) + << s.ToString(); +} + +TEST(DBTest, StillReadSST) { + ASSERT_OK(Put("foo", "bar")); + ASSERT_EQ("bar", Get("foo")); + + // Dump the memtable to disk. + dbfull()->TEST_CompactMemTable(); + ASSERT_EQ("bar", Get("foo")); + Close(); + ASSERT_GT(RenameLDBToSST(), 0); + Options options = CurrentOptions(); + options.paranoid_checks = true; + Status s = TryReopen(&options); + ASSERT_TRUE(s.ok()); + ASSERT_EQ("bar", Get("foo")); +} + +TEST(DBTest, FilesDeletedAfterCompaction) { + ASSERT_OK(Put("foo", "v2")); + Compact("a", "z"); + const int num_files = CountFiles(); + for (int i = 0; i < 10; i++) { + ASSERT_OK(Put("foo", "v2")); + Compact("a", "z"); + } + ASSERT_EQ(CountFiles(), num_files); +} + +TEST(DBTest, BloomFilter) { + env_->count_random_reads_ = true; + Options options = CurrentOptions(); + options.env = env_; + options.block_cache = NewLRUCache(0); // Prevent cache hits + options.filter_policy = NewBloomFilterPolicy(10); + Reopen(&options); + + // Populate multiple layers + const int N = 10000; + for (int i = 0; i < N; i++) { + ASSERT_OK(Put(Key(i), Key(i))); + } + Compact("a", "z"); + for (int i = 0; i < N; i += 100) { + ASSERT_OK(Put(Key(i), Key(i))); + } + dbfull()->TEST_CompactMemTable(); + + // Prevent auto compactions triggered by seeks + env_->delay_data_sync_.Release_Store(env_); + + // Lookup present keys. Should rarely read from small sstable. + env_->random_read_counter_.Reset(); + for (int i = 0; i < N; i++) { + ASSERT_EQ(Key(i), Get(Key(i))); + } + int reads = env_->random_read_counter_.Read(); + fprintf(stderr, "%d present => %d reads\n", N, reads); + ASSERT_GE(reads, N); + ASSERT_LE(reads, N + 2*N/100); + + // Lookup present keys. Should rarely read from either sstable. + env_->random_read_counter_.Reset(); + for (int i = 0; i < N; i++) { + ASSERT_EQ("NOT_FOUND", Get(Key(i) + ".missing")); + } + reads = env_->random_read_counter_.Read(); + fprintf(stderr, "%d missing => %d reads\n", N, reads); + ASSERT_LE(reads, 3*N/100); + + env_->delay_data_sync_.Release_Store(NULL); + Close(); + delete options.block_cache; + delete options.filter_policy; +} + +// Multi-threaded test: +namespace { + +static const int kNumThreads = 4; +static const int kTestSeconds = 10; +static const int kNumKeys = 1000; + +struct MTState { + DBTest* test; + port::AtomicPointer stop; + port::AtomicPointer counter[kNumThreads]; + port::AtomicPointer thread_done[kNumThreads]; +}; + +struct MTThread { + MTState* state; + int id; +}; + +static void MTThreadBody(void* arg) { + MTThread* t = reinterpret_cast<MTThread*>(arg); + int id = t->id; + DB* db = t->state->test->db_; + uintptr_t counter = 0; + fprintf(stderr, "... starting thread %d\n", id); + Random rnd(1000 + id); + std::string value; + char valbuf[1500]; + while (t->state->stop.Acquire_Load() == NULL) { + t->state->counter[id].Release_Store(reinterpret_cast<void*>(counter)); + + int key = rnd.Uniform(kNumKeys); + char keybuf[20]; + snprintf(keybuf, sizeof(keybuf), "%016d", key); + + if (rnd.OneIn(2)) { + // Write values of the form <key, my id, counter>. + // We add some padding for force compactions. + snprintf(valbuf, sizeof(valbuf), "%d.%d.%-1000d", + key, id, static_cast<int>(counter)); + ASSERT_OK(db->Put(WriteOptions(), Slice(keybuf), Slice(valbuf))); + } else { + // Read a value and verify that it matches the pattern written above. + Status s = db->Get(ReadOptions(), Slice(keybuf), &value); + if (s.IsNotFound()) { + // Key has not yet been written + } else { + // Check that the writer thread counter is >= the counter in the value + ASSERT_OK(s); + int k, w, c; + ASSERT_EQ(3, sscanf(value.c_str(), "%d.%d.%d", &k, &w, &c)) << value; + ASSERT_EQ(k, key); + ASSERT_GE(w, 0); + ASSERT_LT(w, kNumThreads); + ASSERT_LE(static_cast<uintptr_t>(c), reinterpret_cast<uintptr_t>( + t->state->counter[w].Acquire_Load())); + } + } + counter++; + } + t->state->thread_done[id].Release_Store(t); + fprintf(stderr, "... stopping thread %d after %d ops\n", id, int(counter)); +} + +} // namespace + +TEST(DBTest, MultiThreaded) { + do { + // Initialize state + MTState mt; + mt.test = this; + mt.stop.Release_Store(0); + for (int id = 0; id < kNumThreads; id++) { + mt.counter[id].Release_Store(0); + mt.thread_done[id].Release_Store(0); + } + + // Start threads + MTThread thread[kNumThreads]; + for (int id = 0; id < kNumThreads; id++) { + thread[id].state = &mt; + thread[id].id = id; + env_->StartThread(MTThreadBody, &thread[id]); + } + + // Let them run for a while + DelayMilliseconds(kTestSeconds * 1000); + + // Stop the threads and wait for them to finish + mt.stop.Release_Store(&mt); + for (int id = 0; id < kNumThreads; id++) { + while (mt.thread_done[id].Acquire_Load() == NULL) { + DelayMilliseconds(100); + } + } + } while (ChangeOptions()); +} + +namespace { +typedef std::map<std::string, std::string> KVMap; +} + +class ModelDB: public DB { + public: + class ModelSnapshot : public Snapshot { + public: + KVMap map_; + }; + + explicit ModelDB(const Options& options): options_(options) { } + ~ModelDB() { } + virtual Status Put(const WriteOptions& o, const Slice& k, const Slice& v) { + return DB::Put(o, k, v); + } + virtual Status Delete(const WriteOptions& o, const Slice& key) { + return DB::Delete(o, key); + } + virtual Status Get(const ReadOptions& options, + const Slice& key, std::string* value) { + assert(false); // Not implemented + return Status::NotFound(key); + } + virtual Iterator* NewIterator(const ReadOptions& options) { + if (options.snapshot == NULL) { + KVMap* saved = new KVMap; + *saved = map_; + return new ModelIter(saved, true); + } else { + const KVMap* snapshot_state = + &(reinterpret_cast<const ModelSnapshot*>(options.snapshot)->map_); + return new ModelIter(snapshot_state, false); + } + } + virtual const Snapshot* GetSnapshot() { + ModelSnapshot* snapshot = new ModelSnapshot; + snapshot->map_ = map_; + return snapshot; + } + + virtual void ReleaseSnapshot(const Snapshot* snapshot) { + delete reinterpret_cast<const ModelSnapshot*>(snapshot); + } + virtual Status Write(const WriteOptions& options, WriteBatch* batch) { + class Handler : public WriteBatch::Handler { + public: + KVMap* map_; + virtual void Put(const Slice& key, const Slice& value) { + (*map_)[key.ToString()] = value.ToString(); + } + virtual void Delete(const Slice& key) { + map_->erase(key.ToString()); + } + }; + Handler handler; + handler.map_ = &map_; + return batch->Iterate(&handler); + } + + virtual bool GetProperty(const Slice& property, std::string* value) { + return false; + } + virtual void GetApproximateSizes(const Range* r, int n, uint64_t* sizes) { + for (int i = 0; i < n; i++) { + sizes[i] = 0; + } + } + virtual void CompactRange(const Slice* start, const Slice* end) { + } + + private: + class ModelIter: public Iterator { + public: + ModelIter(const KVMap* map, bool owned) + : map_(map), owned_(owned), iter_(map_->end()) { + } + ~ModelIter() { + if (owned_) delete map_; + } + virtual bool Valid() const { return iter_ != map_->end(); } + virtual void SeekToFirst() { iter_ = map_->begin(); } + virtual void SeekToLast() { + if (map_->empty()) { + iter_ = map_->end(); + } else { + iter_ = map_->find(map_->rbegin()->first); + } + } + virtual void Seek(const Slice& k) { + iter_ = map_->lower_bound(k.ToString()); + } + virtual void Next() { ++iter_; } + virtual void Prev() { --iter_; } + virtual Slice key() const { return iter_->first; } + virtual Slice value() const { return iter_->second; } + virtual Status status() const { return Status::OK(); } + private: + const KVMap* const map_; + const bool owned_; // Do we own map_ + KVMap::const_iterator iter_; + }; + const Options options_; + KVMap map_; +}; + +static std::string RandomKey(Random* rnd) { + int len = (rnd->OneIn(3) + ? 1 // Short sometimes to encourage collisions + : (rnd->OneIn(100) ? rnd->Skewed(10) : rnd->Uniform(10))); + return test::RandomKey(rnd, len); +} + +static bool CompareIterators(int step, + DB* model, + DB* db, + const Snapshot* model_snap, + const Snapshot* db_snap) { + ReadOptions options; + options.snapshot = model_snap; + Iterator* miter = model->NewIterator(options); + options.snapshot = db_snap; + Iterator* dbiter = db->NewIterator(options); + bool ok = true; + int count = 0; + for (miter->SeekToFirst(), dbiter->SeekToFirst(); + ok && miter->Valid() && dbiter->Valid(); + miter->Next(), dbiter->Next()) { + count++; + if (miter->key().compare(dbiter->key()) != 0) { + fprintf(stderr, "step %d: Key mismatch: '%s' vs. '%s'\n", + step, + EscapeString(miter->key()).c_str(), + EscapeString(dbiter->key()).c_str()); + ok = false; + break; + } + + if (miter->value().compare(dbiter->value()) != 0) { + fprintf(stderr, "step %d: Value mismatch for key '%s': '%s' vs. '%s'\n", + step, + EscapeString(miter->key()).c_str(), + EscapeString(miter->value()).c_str(), + EscapeString(miter->value()).c_str()); + ok = false; + } + } + + if (ok) { + if (miter->Valid() != dbiter->Valid()) { + fprintf(stderr, "step %d: Mismatch at end of iterators: %d vs. %d\n", + step, miter->Valid(), dbiter->Valid()); + ok = false; + } + } + fprintf(stderr, "%d entries compared: ok=%d\n", count, ok); + delete miter; + delete dbiter; + return ok; +} + +TEST(DBTest, Randomized) { + Random rnd(test::RandomSeed()); + do { + ModelDB model(CurrentOptions()); + const int N = 10000; + const Snapshot* model_snap = NULL; + const Snapshot* db_snap = NULL; + std::string k, v; + for (int step = 0; step < N; step++) { + if (step % 100 == 0) { + fprintf(stderr, "Step %d of %d\n", step, N); + } + // TODO(sanjay): Test Get() works + int p = rnd.Uniform(100); + if (p < 45) { // Put + k = RandomKey(&rnd); + v = RandomString(&rnd, + rnd.OneIn(20) + ? 100 + rnd.Uniform(100) + : rnd.Uniform(8)); + ASSERT_OK(model.Put(WriteOptions(), k, v)); + ASSERT_OK(db_->Put(WriteOptions(), k, v)); + + } else if (p < 90) { // Delete + k = RandomKey(&rnd); + ASSERT_OK(model.Delete(WriteOptions(), k)); + ASSERT_OK(db_->Delete(WriteOptions(), k)); + + + } else { // Multi-element batch + WriteBatch b; + const int num = rnd.Uniform(8); + for (int i = 0; i < num; i++) { + if (i == 0 || !rnd.OneIn(10)) { + k = RandomKey(&rnd); + } else { + // Periodically re-use the same key from the previous iter, so + // we have multiple entries in the write batch for the same key + } + if (rnd.OneIn(2)) { + v = RandomString(&rnd, rnd.Uniform(10)); + b.Put(k, v); + } else { + b.Delete(k); + } + } + ASSERT_OK(model.Write(WriteOptions(), &b)); + ASSERT_OK(db_->Write(WriteOptions(), &b)); + } + + if ((step % 100) == 0) { + ASSERT_TRUE(CompareIterators(step, &model, db_, NULL, NULL)); + ASSERT_TRUE(CompareIterators(step, &model, db_, model_snap, db_snap)); + // Save a snapshot from each DB this time that we'll use next + // time we compare things, to make sure the current state is + // preserved with the snapshot + if (model_snap != NULL) model.ReleaseSnapshot(model_snap); + if (db_snap != NULL) db_->ReleaseSnapshot(db_snap); + + Reopen(); + ASSERT_TRUE(CompareIterators(step, &model, db_, NULL, NULL)); + + model_snap = model.GetSnapshot(); + db_snap = db_->GetSnapshot(); + } + } + if (model_snap != NULL) model.ReleaseSnapshot(model_snap); + if (db_snap != NULL) db_->ReleaseSnapshot(db_snap); + } while (ChangeOptions()); +} + +std::string MakeKey(unsigned int num) { + char buf[30]; + snprintf(buf, sizeof(buf), "%016u", num); + return std::string(buf); +} + +void BM_LogAndApply(int iters, int num_base_files) { + std::string dbname = test::TmpDir() + "/leveldb_test_benchmark"; + DestroyDB(dbname, Options()); + + DB* db = NULL; + Options opts; + opts.create_if_missing = true; + Status s = DB::Open(opts, dbname, &db); + ASSERT_OK(s); + ASSERT_TRUE(db != NULL); + + delete db; + db = NULL; + + Env* env = Env::Default(); + + port::Mutex mu; + MutexLock l(&mu); + + InternalKeyComparator cmp(BytewiseComparator()); + Options options; + VersionSet vset(dbname, &options, NULL, &cmp); + bool save_manifest; + ASSERT_OK(vset.Recover(&save_manifest)); + VersionEdit vbase; + uint64_t fnum = 1; + for (int i = 0; i < num_base_files; i++) { + InternalKey start(MakeKey(2*fnum), 1, kTypeValue); + InternalKey limit(MakeKey(2*fnum+1), 1, kTypeDeletion); + vbase.AddFile(2, fnum++, 1 /* file size */, start, limit); + } + ASSERT_OK(vset.LogAndApply(&vbase, &mu)); + + uint64_t start_micros = env->NowMicros(); + + for (int i = 0; i < iters; i++) { + VersionEdit vedit; + vedit.DeleteFile(2, fnum); + InternalKey start(MakeKey(2*fnum), 1, kTypeValue); + InternalKey limit(MakeKey(2*fnum+1), 1, kTypeDeletion); + vedit.AddFile(2, fnum++, 1 /* file size */, start, limit); + vset.LogAndApply(&vedit, &mu); + } + uint64_t stop_micros = env->NowMicros(); + unsigned int us = stop_micros - start_micros; + char buf[16]; + snprintf(buf, sizeof(buf), "%d", num_base_files); + fprintf(stderr, + "BM_LogAndApply/%-6s %8d iters : %9u us (%7.0f us / iter)\n", + buf, iters, us, ((float)us) / iters); +} + +} // namespace leveldb + +int main(int argc, char** argv) { + if (argc > 1 && std::string(argv[1]) == "--benchmark") { + leveldb::BM_LogAndApply(1000, 1); + leveldb::BM_LogAndApply(1000, 100); + leveldb::BM_LogAndApply(1000, 10000); + leveldb::BM_LogAndApply(100, 100000); + return 0; + } + + return leveldb::test::RunAllTests(); +} diff --git a/src/leveldb/db/dbformat.cc b/src/leveldb/db/dbformat.cc new file mode 100644 index 0000000000..20a7ca4462 --- /dev/null +++ b/src/leveldb/db/dbformat.cc @@ -0,0 +1,140 @@ +// 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 <stdio.h> +#include "db/dbformat.h" +#include "port/port.h" +#include "util/coding.h" + +namespace leveldb { + +static uint64_t PackSequenceAndType(uint64_t seq, ValueType t) { + assert(seq <= kMaxSequenceNumber); + assert(t <= kValueTypeForSeek); + return (seq << 8) | t; +} + +void AppendInternalKey(std::string* result, const ParsedInternalKey& key) { + result->append(key.user_key.data(), key.user_key.size()); + PutFixed64(result, PackSequenceAndType(key.sequence, key.type)); +} + +std::string ParsedInternalKey::DebugString() const { + char buf[50]; + snprintf(buf, sizeof(buf), "' @ %llu : %d", + (unsigned long long) sequence, + int(type)); + std::string result = "'"; + result += EscapeString(user_key.ToString()); + result += buf; + return result; +} + +std::string InternalKey::DebugString() const { + std::string result; + ParsedInternalKey parsed; + if (ParseInternalKey(rep_, &parsed)) { + result = parsed.DebugString(); + } else { + result = "(bad)"; + result.append(EscapeString(rep_)); + } + return result; +} + +const char* InternalKeyComparator::Name() const { + return "leveldb.InternalKeyComparator"; +} + +int InternalKeyComparator::Compare(const Slice& akey, const Slice& bkey) const { + // Order by: + // increasing user key (according to user-supplied comparator) + // decreasing sequence number + // decreasing type (though sequence# should be enough to disambiguate) + int r = user_comparator_->Compare(ExtractUserKey(akey), ExtractUserKey(bkey)); + if (r == 0) { + const uint64_t anum = DecodeFixed64(akey.data() + akey.size() - 8); + const uint64_t bnum = DecodeFixed64(bkey.data() + bkey.size() - 8); + if (anum > bnum) { + r = -1; + } else if (anum < bnum) { + r = +1; + } + } + return r; +} + +void InternalKeyComparator::FindShortestSeparator( + std::string* start, + const Slice& limit) const { + // Attempt to shorten the user portion of the key + Slice user_start = ExtractUserKey(*start); + Slice user_limit = ExtractUserKey(limit); + std::string tmp(user_start.data(), user_start.size()); + user_comparator_->FindShortestSeparator(&tmp, user_limit); + if (tmp.size() < user_start.size() && + user_comparator_->Compare(user_start, tmp) < 0) { + // User key has become shorter physically, but larger logically. + // Tack on the earliest possible number to the shortened user key. + PutFixed64(&tmp, PackSequenceAndType(kMaxSequenceNumber,kValueTypeForSeek)); + assert(this->Compare(*start, tmp) < 0); + assert(this->Compare(tmp, limit) < 0); + start->swap(tmp); + } +} + +void InternalKeyComparator::FindShortSuccessor(std::string* key) const { + Slice user_key = ExtractUserKey(*key); + std::string tmp(user_key.data(), user_key.size()); + user_comparator_->FindShortSuccessor(&tmp); + if (tmp.size() < user_key.size() && + user_comparator_->Compare(user_key, tmp) < 0) { + // User key has become shorter physically, but larger logically. + // Tack on the earliest possible number to the shortened user key. + PutFixed64(&tmp, PackSequenceAndType(kMaxSequenceNumber,kValueTypeForSeek)); + assert(this->Compare(*key, tmp) < 0); + key->swap(tmp); + } +} + +const char* InternalFilterPolicy::Name() const { + return user_policy_->Name(); +} + +void InternalFilterPolicy::CreateFilter(const Slice* keys, int n, + std::string* dst) const { + // We rely on the fact that the code in table.cc does not mind us + // adjusting keys[]. + Slice* mkey = const_cast<Slice*>(keys); + for (int i = 0; i < n; i++) { + mkey[i] = ExtractUserKey(keys[i]); + // TODO(sanjay): Suppress dups? + } + user_policy_->CreateFilter(keys, n, dst); +} + +bool InternalFilterPolicy::KeyMayMatch(const Slice& key, const Slice& f) const { + return user_policy_->KeyMayMatch(ExtractUserKey(key), f); +} + +LookupKey::LookupKey(const Slice& user_key, SequenceNumber s) { + size_t usize = user_key.size(); + size_t needed = usize + 13; // A conservative estimate + char* dst; + if (needed <= sizeof(space_)) { + dst = space_; + } else { + dst = new char[needed]; + } + start_ = dst; + dst = EncodeVarint32(dst, usize + 8); + kstart_ = dst; + memcpy(dst, user_key.data(), usize); + dst += usize; + EncodeFixed64(dst, PackSequenceAndType(s, kValueTypeForSeek)); + dst += 8; + end_ = dst; +} + +} // namespace leveldb diff --git a/src/leveldb/db/dbformat.h b/src/leveldb/db/dbformat.h new file mode 100644 index 0000000000..ea897b13c0 --- /dev/null +++ b/src/leveldb/db/dbformat.h @@ -0,0 +1,230 @@ +// 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_DBFORMAT_H_ +#define STORAGE_LEVELDB_DB_DBFORMAT_H_ + +#include <stdio.h> +#include "leveldb/comparator.h" +#include "leveldb/db.h" +#include "leveldb/filter_policy.h" +#include "leveldb/slice.h" +#include "leveldb/table_builder.h" +#include "util/coding.h" +#include "util/logging.h" + +namespace leveldb { + +// Grouping of constants. We may want to make some of these +// parameters set via options. +namespace config { +static const int kNumLevels = 7; + +// Level-0 compaction is started when we hit this many files. +static const int kL0_CompactionTrigger = 4; + +// Soft limit on number of level-0 files. We slow down writes at this point. +static const int kL0_SlowdownWritesTrigger = 8; + +// Maximum number of level-0 files. We stop writes at this point. +static const int kL0_StopWritesTrigger = 12; + +// Maximum level to which a new compacted memtable is pushed if it +// does not create overlap. We try to push to level 2 to avoid the +// relatively expensive level 0=>1 compactions and to avoid some +// expensive manifest file operations. We do not push all the way to +// the largest level since that can generate a lot of wasted disk +// space if the same key space is being repeatedly overwritten. +static const int kMaxMemCompactLevel = 2; + +// Approximate gap in bytes between samples of data read during iteration. +static const int kReadBytesPeriod = 1048576; + +} // namespace config + +class InternalKey; + +// Value types encoded as the last component of internal keys. +// DO NOT CHANGE THESE ENUM VALUES: they are embedded in the on-disk +// data structures. +enum ValueType { + kTypeDeletion = 0x0, + kTypeValue = 0x1 +}; +// kValueTypeForSeek defines the ValueType that should be passed when +// constructing a ParsedInternalKey object for seeking to a particular +// sequence number (since we sort sequence numbers in decreasing order +// and the value type is embedded as the low 8 bits in the sequence +// number in internal keys, we need to use the highest-numbered +// ValueType, not the lowest). +static const ValueType kValueTypeForSeek = kTypeValue; + +typedef uint64_t SequenceNumber; + +// We leave eight bits empty at the bottom so a type and sequence# +// can be packed together into 64-bits. +static const SequenceNumber kMaxSequenceNumber = + ((0x1ull << 56) - 1); + +struct ParsedInternalKey { + Slice user_key; + SequenceNumber sequence; + ValueType type; + + ParsedInternalKey() { } // Intentionally left uninitialized (for speed) + ParsedInternalKey(const Slice& u, const SequenceNumber& seq, ValueType t) + : user_key(u), sequence(seq), type(t) { } + std::string DebugString() const; +}; + +// Return the length of the encoding of "key". +inline size_t InternalKeyEncodingLength(const ParsedInternalKey& key) { + return key.user_key.size() + 8; +} + +// Append the serialization of "key" to *result. +extern void AppendInternalKey(std::string* result, + const ParsedInternalKey& key); + +// Attempt to parse an internal key from "internal_key". On success, +// stores the parsed data in "*result", and returns true. +// +// On error, returns false, leaves "*result" in an undefined state. +extern bool ParseInternalKey(const Slice& internal_key, + ParsedInternalKey* result); + +// Returns the user key portion of an internal key. +inline Slice ExtractUserKey(const Slice& internal_key) { + assert(internal_key.size() >= 8); + return Slice(internal_key.data(), internal_key.size() - 8); +} + +inline ValueType ExtractValueType(const Slice& internal_key) { + assert(internal_key.size() >= 8); + const size_t n = internal_key.size(); + uint64_t num = DecodeFixed64(internal_key.data() + n - 8); + unsigned char c = num & 0xff; + return static_cast<ValueType>(c); +} + +// A comparator for internal keys that uses a specified comparator for +// the user key portion and breaks ties by decreasing sequence number. +class InternalKeyComparator : public Comparator { + private: + const Comparator* user_comparator_; + public: + explicit InternalKeyComparator(const Comparator* c) : user_comparator_(c) { } + virtual const char* Name() const; + virtual int Compare(const Slice& a, const Slice& b) const; + virtual void FindShortestSeparator( + std::string* start, + const Slice& limit) const; + virtual void FindShortSuccessor(std::string* key) const; + + const Comparator* user_comparator() const { return user_comparator_; } + + int Compare(const InternalKey& a, const InternalKey& b) const; +}; + +// Filter policy wrapper that converts from internal keys to user keys +class InternalFilterPolicy : public FilterPolicy { + private: + const FilterPolicy* const user_policy_; + public: + explicit InternalFilterPolicy(const FilterPolicy* p) : user_policy_(p) { } + virtual const char* Name() const; + virtual void CreateFilter(const Slice* keys, int n, std::string* dst) const; + virtual bool KeyMayMatch(const Slice& key, const Slice& filter) const; +}; + +// Modules in this directory should keep internal keys wrapped inside +// the following class instead of plain strings so that we do not +// incorrectly use string comparisons instead of an InternalKeyComparator. +class InternalKey { + private: + std::string rep_; + public: + InternalKey() { } // Leave rep_ as empty to indicate it is invalid + InternalKey(const Slice& user_key, SequenceNumber s, ValueType t) { + AppendInternalKey(&rep_, ParsedInternalKey(user_key, s, t)); + } + + void DecodeFrom(const Slice& s) { rep_.assign(s.data(), s.size()); } + Slice Encode() const { + assert(!rep_.empty()); + return rep_; + } + + Slice user_key() const { return ExtractUserKey(rep_); } + + void SetFrom(const ParsedInternalKey& p) { + rep_.clear(); + AppendInternalKey(&rep_, p); + } + + void Clear() { rep_.clear(); } + + std::string DebugString() const; +}; + +inline int InternalKeyComparator::Compare( + const InternalKey& a, const InternalKey& b) const { + return Compare(a.Encode(), b.Encode()); +} + +inline bool ParseInternalKey(const Slice& internal_key, + ParsedInternalKey* result) { + const size_t n = internal_key.size(); + if (n < 8) return false; + uint64_t num = DecodeFixed64(internal_key.data() + n - 8); + unsigned char c = num & 0xff; + result->sequence = num >> 8; + result->type = static_cast<ValueType>(c); + result->user_key = Slice(internal_key.data(), n - 8); + return (c <= static_cast<unsigned char>(kTypeValue)); +} + +// A helper class useful for DBImpl::Get() +class LookupKey { + public: + // Initialize *this for looking up user_key at a snapshot with + // the specified sequence number. + LookupKey(const Slice& user_key, SequenceNumber sequence); + + ~LookupKey(); + + // Return a key suitable for lookup in a MemTable. + Slice memtable_key() const { return Slice(start_, end_ - start_); } + + // Return an internal key (suitable for passing to an internal iterator) + Slice internal_key() const { return Slice(kstart_, end_ - kstart_); } + + // Return the user key + Slice user_key() const { return Slice(kstart_, end_ - kstart_ - 8); } + + private: + // We construct a char array of the form: + // klength varint32 <-- start_ + // userkey char[klength] <-- kstart_ + // tag uint64 + // <-- end_ + // The array is a suitable MemTable key. + // The suffix starting with "userkey" can be used as an InternalKey. + const char* start_; + const char* kstart_; + const char* end_; + char space_[200]; // Avoid allocation for short keys + + // No copying allowed + LookupKey(const LookupKey&); + void operator=(const LookupKey&); +}; + +inline LookupKey::~LookupKey() { + if (start_ != space_) delete[] start_; +} + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_DBFORMAT_H_ diff --git a/src/leveldb/db/dbformat_test.cc b/src/leveldb/db/dbformat_test.cc new file mode 100644 index 0000000000..5d82f5d313 --- /dev/null +++ b/src/leveldb/db/dbformat_test.cc @@ -0,0 +1,112 @@ +// 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/dbformat.h" +#include "util/logging.h" +#include "util/testharness.h" + +namespace leveldb { + +static std::string IKey(const std::string& user_key, + uint64_t seq, + ValueType vt) { + std::string encoded; + AppendInternalKey(&encoded, ParsedInternalKey(user_key, seq, vt)); + return encoded; +} + +static std::string Shorten(const std::string& s, const std::string& l) { + std::string result = s; + InternalKeyComparator(BytewiseComparator()).FindShortestSeparator(&result, l); + return result; +} + +static std::string ShortSuccessor(const std::string& s) { + std::string result = s; + InternalKeyComparator(BytewiseComparator()).FindShortSuccessor(&result); + return result; +} + +static void TestKey(const std::string& key, + uint64_t seq, + ValueType vt) { + std::string encoded = IKey(key, seq, vt); + + Slice in(encoded); + ParsedInternalKey decoded("", 0, kTypeValue); + + ASSERT_TRUE(ParseInternalKey(in, &decoded)); + ASSERT_EQ(key, decoded.user_key.ToString()); + ASSERT_EQ(seq, decoded.sequence); + ASSERT_EQ(vt, decoded.type); + + ASSERT_TRUE(!ParseInternalKey(Slice("bar"), &decoded)); +} + +class FormatTest { }; + +TEST(FormatTest, InternalKey_EncodeDecode) { + const char* keys[] = { "", "k", "hello", "longggggggggggggggggggggg" }; + const uint64_t seq[] = { + 1, 2, 3, + (1ull << 8) - 1, 1ull << 8, (1ull << 8) + 1, + (1ull << 16) - 1, 1ull << 16, (1ull << 16) + 1, + (1ull << 32) - 1, 1ull << 32, (1ull << 32) + 1 + }; + for (int k = 0; k < sizeof(keys) / sizeof(keys[0]); k++) { + for (int s = 0; s < sizeof(seq) / sizeof(seq[0]); s++) { + TestKey(keys[k], seq[s], kTypeValue); + TestKey("hello", 1, kTypeDeletion); + } + } +} + +TEST(FormatTest, InternalKeyShortSeparator) { + // When user keys are same + ASSERT_EQ(IKey("foo", 100, kTypeValue), + Shorten(IKey("foo", 100, kTypeValue), + IKey("foo", 99, kTypeValue))); + ASSERT_EQ(IKey("foo", 100, kTypeValue), + Shorten(IKey("foo", 100, kTypeValue), + IKey("foo", 101, kTypeValue))); + ASSERT_EQ(IKey("foo", 100, kTypeValue), + Shorten(IKey("foo", 100, kTypeValue), + IKey("foo", 100, kTypeValue))); + ASSERT_EQ(IKey("foo", 100, kTypeValue), + Shorten(IKey("foo", 100, kTypeValue), + IKey("foo", 100, kTypeDeletion))); + + // When user keys are misordered + ASSERT_EQ(IKey("foo", 100, kTypeValue), + Shorten(IKey("foo", 100, kTypeValue), + IKey("bar", 99, kTypeValue))); + + // When user keys are different, but correctly ordered + ASSERT_EQ(IKey("g", kMaxSequenceNumber, kValueTypeForSeek), + Shorten(IKey("foo", 100, kTypeValue), + IKey("hello", 200, kTypeValue))); + + // When start user key is prefix of limit user key + ASSERT_EQ(IKey("foo", 100, kTypeValue), + Shorten(IKey("foo", 100, kTypeValue), + IKey("foobar", 200, kTypeValue))); + + // When limit user key is prefix of start user key + ASSERT_EQ(IKey("foobar", 100, kTypeValue), + Shorten(IKey("foobar", 100, kTypeValue), + IKey("foo", 200, kTypeValue))); +} + +TEST(FormatTest, InternalKeyShortestSuccessor) { + ASSERT_EQ(IKey("g", kMaxSequenceNumber, kValueTypeForSeek), + ShortSuccessor(IKey("foo", 100, kTypeValue))); + ASSERT_EQ(IKey("\xff\xff", 100, kTypeValue), + ShortSuccessor(IKey("\xff\xff", 100, kTypeValue))); +} + +} // namespace leveldb + +int main(int argc, char** argv) { + return leveldb::test::RunAllTests(); +} diff --git a/src/leveldb/db/dumpfile.cc b/src/leveldb/db/dumpfile.cc new file mode 100644 index 0000000000..61c47c2ff9 --- /dev/null +++ b/src/leveldb/db/dumpfile.cc @@ -0,0 +1,225 @@ +// Copyright (c) 2012 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 <stdio.h> +#include "db/dbformat.h" +#include "db/filename.h" +#include "db/log_reader.h" +#include "db/version_edit.h" +#include "db/write_batch_internal.h" +#include "leveldb/env.h" +#include "leveldb/iterator.h" +#include "leveldb/options.h" +#include "leveldb/status.h" +#include "leveldb/table.h" +#include "leveldb/write_batch.h" +#include "util/logging.h" + +namespace leveldb { + +namespace { + +bool GuessType(const std::string& fname, FileType* type) { + size_t pos = fname.rfind('/'); + std::string basename; + if (pos == std::string::npos) { + basename = fname; + } else { + basename = std::string(fname.data() + pos + 1, fname.size() - pos - 1); + } + uint64_t ignored; + return ParseFileName(basename, &ignored, type); +} + +// Notified when log reader encounters corruption. +class CorruptionReporter : public log::Reader::Reporter { + public: + WritableFile* dst_; + virtual void Corruption(size_t bytes, const Status& status) { + std::string r = "corruption: "; + AppendNumberTo(&r, bytes); + r += " bytes; "; + r += status.ToString(); + r.push_back('\n'); + dst_->Append(r); + } +}; + +// Print contents of a log file. (*func)() is called on every record. +Status PrintLogContents(Env* env, const std::string& fname, + void (*func)(uint64_t, Slice, WritableFile*), + WritableFile* dst) { + SequentialFile* file; + Status s = env->NewSequentialFile(fname, &file); + if (!s.ok()) { + return s; + } + CorruptionReporter reporter; + reporter.dst_ = dst; + log::Reader reader(file, &reporter, true, 0); + Slice record; + std::string scratch; + while (reader.ReadRecord(&record, &scratch)) { + (*func)(reader.LastRecordOffset(), record, dst); + } + delete file; + return Status::OK(); +} + +// Called on every item found in a WriteBatch. +class WriteBatchItemPrinter : public WriteBatch::Handler { + public: + WritableFile* dst_; + virtual void Put(const Slice& key, const Slice& value) { + std::string r = " put '"; + AppendEscapedStringTo(&r, key); + r += "' '"; + AppendEscapedStringTo(&r, value); + r += "'\n"; + dst_->Append(r); + } + virtual void Delete(const Slice& key) { + std::string r = " del '"; + AppendEscapedStringTo(&r, key); + r += "'\n"; + dst_->Append(r); + } +}; + + +// Called on every log record (each one of which is a WriteBatch) +// found in a kLogFile. +static void WriteBatchPrinter(uint64_t pos, Slice record, WritableFile* dst) { + std::string r = "--- offset "; + AppendNumberTo(&r, pos); + r += "; "; + if (record.size() < 12) { + r += "log record length "; + AppendNumberTo(&r, record.size()); + r += " is too small\n"; + dst->Append(r); + return; + } + WriteBatch batch; + WriteBatchInternal::SetContents(&batch, record); + r += "sequence "; + AppendNumberTo(&r, WriteBatchInternal::Sequence(&batch)); + r.push_back('\n'); + dst->Append(r); + WriteBatchItemPrinter batch_item_printer; + batch_item_printer.dst_ = dst; + Status s = batch.Iterate(&batch_item_printer); + if (!s.ok()) { + dst->Append(" error: " + s.ToString() + "\n"); + } +} + +Status DumpLog(Env* env, const std::string& fname, WritableFile* dst) { + return PrintLogContents(env, fname, WriteBatchPrinter, dst); +} + +// Called on every log record (each one of which is a WriteBatch) +// found in a kDescriptorFile. +static void VersionEditPrinter(uint64_t pos, Slice record, WritableFile* dst) { + std::string r = "--- offset "; + AppendNumberTo(&r, pos); + r += "; "; + VersionEdit edit; + Status s = edit.DecodeFrom(record); + if (!s.ok()) { + r += s.ToString(); + r.push_back('\n'); + } else { + r += edit.DebugString(); + } + dst->Append(r); +} + +Status DumpDescriptor(Env* env, const std::string& fname, WritableFile* dst) { + return PrintLogContents(env, fname, VersionEditPrinter, dst); +} + +Status DumpTable(Env* env, const std::string& fname, WritableFile* dst) { + uint64_t file_size; + RandomAccessFile* file = NULL; + Table* table = NULL; + Status s = env->GetFileSize(fname, &file_size); + if (s.ok()) { + s = env->NewRandomAccessFile(fname, &file); + } + if (s.ok()) { + // We use the default comparator, which may or may not match the + // comparator used in this database. However this should not cause + // problems since we only use Table operations that do not require + // any comparisons. In particular, we do not call Seek or Prev. + s = Table::Open(Options(), file, file_size, &table); + } + if (!s.ok()) { + delete table; + delete file; + return s; + } + + ReadOptions ro; + ro.fill_cache = false; + Iterator* iter = table->NewIterator(ro); + std::string r; + for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { + r.clear(); + ParsedInternalKey key; + if (!ParseInternalKey(iter->key(), &key)) { + r = "badkey '"; + AppendEscapedStringTo(&r, iter->key()); + r += "' => '"; + AppendEscapedStringTo(&r, iter->value()); + r += "'\n"; + dst->Append(r); + } else { + r = "'"; + AppendEscapedStringTo(&r, key.user_key); + r += "' @ "; + AppendNumberTo(&r, key.sequence); + r += " : "; + if (key.type == kTypeDeletion) { + r += "del"; + } else if (key.type == kTypeValue) { + r += "val"; + } else { + AppendNumberTo(&r, key.type); + } + r += " => '"; + AppendEscapedStringTo(&r, iter->value()); + r += "'\n"; + dst->Append(r); + } + } + s = iter->status(); + if (!s.ok()) { + dst->Append("iterator error: " + s.ToString() + "\n"); + } + + delete iter; + delete table; + delete file; + return Status::OK(); +} + +} // namespace + +Status DumpFile(Env* env, const std::string& fname, WritableFile* dst) { + FileType ftype; + if (!GuessType(fname, &ftype)) { + return Status::InvalidArgument(fname + ": unknown file type"); + } + switch (ftype) { + case kLogFile: return DumpLog(env, fname, dst); + case kDescriptorFile: return DumpDescriptor(env, fname, dst); + case kTableFile: return DumpTable(env, fname, dst); + default: + break; + } + return Status::InvalidArgument(fname + ": not a dump-able file type"); +} + +} // namespace leveldb diff --git a/src/leveldb/db/fault_injection_test.cc b/src/leveldb/db/fault_injection_test.cc new file mode 100644 index 0000000000..875dfe81ee --- /dev/null +++ b/src/leveldb/db/fault_injection_test.cc @@ -0,0 +1,554 @@ +// Copyright 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. + +// This test uses a custom Env to keep track of the state of a filesystem as of +// the last "sync". It then checks for data loss errors by purposely dropping +// file data (or entire files) not protected by a "sync". + +#include "leveldb/db.h" + +#include <map> +#include <set> +#include "db/db_impl.h" +#include "db/filename.h" +#include "db/log_format.h" +#include "db/version_set.h" +#include "leveldb/cache.h" +#include "leveldb/env.h" +#include "leveldb/table.h" +#include "leveldb/write_batch.h" +#include "util/logging.h" +#include "util/mutexlock.h" +#include "util/testharness.h" +#include "util/testutil.h" + +namespace leveldb { + +static const int kValueSize = 1000; +static const int kMaxNumValues = 2000; +static const size_t kNumIterations = 3; + +class FaultInjectionTestEnv; + +namespace { + +// Assume a filename, and not a directory name like "/foo/bar/" +static std::string GetDirName(const std::string filename) { + size_t found = filename.find_last_of("/\\"); + if (found == std::string::npos) { + return ""; + } else { + return filename.substr(0, found); + } +} + +Status SyncDir(const std::string& dir) { + // As this is a test it isn't required to *actually* sync this directory. + return Status::OK(); +} + +// A basic file truncation function suitable for this test. +Status Truncate(const std::string& filename, uint64_t length) { + leveldb::Env* env = leveldb::Env::Default(); + + SequentialFile* orig_file; + Status s = env->NewSequentialFile(filename, &orig_file); + if (!s.ok()) + return s; + + char* scratch = new char[length]; + leveldb::Slice result; + s = orig_file->Read(length, &result, scratch); + delete orig_file; + if (s.ok()) { + std::string tmp_name = GetDirName(filename) + "/truncate.tmp"; + WritableFile* tmp_file; + s = env->NewWritableFile(tmp_name, &tmp_file); + if (s.ok()) { + s = tmp_file->Append(result); + delete tmp_file; + if (s.ok()) { + s = env->RenameFile(tmp_name, filename); + } else { + env->DeleteFile(tmp_name); + } + } + } + + delete[] scratch; + + return s; +} + +struct FileState { + std::string filename_; + ssize_t pos_; + ssize_t pos_at_last_sync_; + ssize_t pos_at_last_flush_; + + FileState(const std::string& filename) + : filename_(filename), + pos_(-1), + pos_at_last_sync_(-1), + pos_at_last_flush_(-1) { } + + FileState() : pos_(-1), pos_at_last_sync_(-1), pos_at_last_flush_(-1) {} + + bool IsFullySynced() const { return pos_ <= 0 || pos_ == pos_at_last_sync_; } + + Status DropUnsyncedData() const; +}; + +} // anonymous namespace + +// A wrapper around WritableFile which informs another Env whenever this file +// is written to or sync'ed. +class TestWritableFile : public WritableFile { + public: + TestWritableFile(const FileState& state, + WritableFile* f, + FaultInjectionTestEnv* env); + virtual ~TestWritableFile(); + virtual Status Append(const Slice& data); + virtual Status Close(); + virtual Status Flush(); + virtual Status Sync(); + + private: + FileState state_; + WritableFile* target_; + bool writable_file_opened_; + FaultInjectionTestEnv* env_; + + Status SyncParent(); +}; + +class FaultInjectionTestEnv : public EnvWrapper { + public: + FaultInjectionTestEnv() : EnvWrapper(Env::Default()), filesystem_active_(true) {} + virtual ~FaultInjectionTestEnv() { } + virtual Status NewWritableFile(const std::string& fname, + WritableFile** result); + virtual Status NewAppendableFile(const std::string& fname, + WritableFile** result); + virtual Status DeleteFile(const std::string& f); + virtual Status RenameFile(const std::string& s, const std::string& t); + + void WritableFileClosed(const FileState& state); + Status DropUnsyncedFileData(); + Status DeleteFilesCreatedAfterLastDirSync(); + void DirWasSynced(); + bool IsFileCreatedSinceLastDirSync(const std::string& filename); + void ResetState(); + void UntrackFile(const std::string& f); + // Setting the filesystem to inactive is the test equivalent to simulating a + // system reset. Setting to inactive will freeze our saved filesystem state so + // that it will stop being recorded. It can then be reset back to the state at + // the time of the reset. + bool IsFilesystemActive() const { return filesystem_active_; } + void SetFilesystemActive(bool active) { filesystem_active_ = active; } + + private: + port::Mutex mutex_; + std::map<std::string, FileState> db_file_state_; + std::set<std::string> new_files_since_last_dir_sync_; + bool filesystem_active_; // Record flushes, syncs, writes +}; + +TestWritableFile::TestWritableFile(const FileState& state, + WritableFile* f, + FaultInjectionTestEnv* env) + : state_(state), + target_(f), + writable_file_opened_(true), + env_(env) { + assert(f != NULL); +} + +TestWritableFile::~TestWritableFile() { + if (writable_file_opened_) { + Close(); + } + delete target_; +} + +Status TestWritableFile::Append(const Slice& data) { + Status s = target_->Append(data); + if (s.ok() && env_->IsFilesystemActive()) { + state_.pos_ += data.size(); + } + return s; +} + +Status TestWritableFile::Close() { + writable_file_opened_ = false; + Status s = target_->Close(); + if (s.ok()) { + env_->WritableFileClosed(state_); + } + return s; +} + +Status TestWritableFile::Flush() { + Status s = target_->Flush(); + if (s.ok() && env_->IsFilesystemActive()) { + state_.pos_at_last_flush_ = state_.pos_; + } + return s; +} + +Status TestWritableFile::SyncParent() { + Status s = SyncDir(GetDirName(state_.filename_)); + if (s.ok()) { + env_->DirWasSynced(); + } + return s; +} + +Status TestWritableFile::Sync() { + if (!env_->IsFilesystemActive()) { + return Status::OK(); + } + // Ensure new files referred to by the manifest are in the filesystem. + Status s = target_->Sync(); + if (s.ok()) { + state_.pos_at_last_sync_ = state_.pos_; + } + if (env_->IsFileCreatedSinceLastDirSync(state_.filename_)) { + Status ps = SyncParent(); + if (s.ok() && !ps.ok()) { + s = ps; + } + } + return s; +} + +Status FaultInjectionTestEnv::NewWritableFile(const std::string& fname, + WritableFile** result) { + WritableFile* actual_writable_file; + Status s = target()->NewWritableFile(fname, &actual_writable_file); + if (s.ok()) { + FileState state(fname); + state.pos_ = 0; + *result = new TestWritableFile(state, actual_writable_file, this); + // NewWritableFile doesn't append to files, so if the same file is + // opened again then it will be truncated - so forget our saved + // state. + UntrackFile(fname); + MutexLock l(&mutex_); + new_files_since_last_dir_sync_.insert(fname); + } + return s; +} + +Status FaultInjectionTestEnv::NewAppendableFile(const std::string& fname, + WritableFile** result) { + WritableFile* actual_writable_file; + Status s = target()->NewAppendableFile(fname, &actual_writable_file); + if (s.ok()) { + FileState state(fname); + state.pos_ = 0; + { + MutexLock l(&mutex_); + if (db_file_state_.count(fname) == 0) { + new_files_since_last_dir_sync_.insert(fname); + } else { + state = db_file_state_[fname]; + } + } + *result = new TestWritableFile(state, actual_writable_file, this); + } + return s; +} + +Status FaultInjectionTestEnv::DropUnsyncedFileData() { + Status s; + MutexLock l(&mutex_); + for (std::map<std::string, FileState>::const_iterator it = + db_file_state_.begin(); + s.ok() && it != db_file_state_.end(); ++it) { + const FileState& state = it->second; + if (!state.IsFullySynced()) { + s = state.DropUnsyncedData(); + } + } + return s; +} + +void FaultInjectionTestEnv::DirWasSynced() { + MutexLock l(&mutex_); + new_files_since_last_dir_sync_.clear(); +} + +bool FaultInjectionTestEnv::IsFileCreatedSinceLastDirSync( + const std::string& filename) { + MutexLock l(&mutex_); + return new_files_since_last_dir_sync_.find(filename) != + new_files_since_last_dir_sync_.end(); +} + +void FaultInjectionTestEnv::UntrackFile(const std::string& f) { + MutexLock l(&mutex_); + db_file_state_.erase(f); + new_files_since_last_dir_sync_.erase(f); +} + +Status FaultInjectionTestEnv::DeleteFile(const std::string& f) { + Status s = EnvWrapper::DeleteFile(f); + ASSERT_OK(s); + if (s.ok()) { + UntrackFile(f); + } + return s; +} + +Status FaultInjectionTestEnv::RenameFile(const std::string& s, + const std::string& t) { + Status ret = EnvWrapper::RenameFile(s, t); + + if (ret.ok()) { + MutexLock l(&mutex_); + if (db_file_state_.find(s) != db_file_state_.end()) { + db_file_state_[t] = db_file_state_[s]; + db_file_state_.erase(s); + } + + if (new_files_since_last_dir_sync_.erase(s) != 0) { + assert(new_files_since_last_dir_sync_.find(t) == + new_files_since_last_dir_sync_.end()); + new_files_since_last_dir_sync_.insert(t); + } + } + + return ret; +} + +void FaultInjectionTestEnv::ResetState() { + // Since we are not destroying the database, the existing files + // should keep their recorded synced/flushed state. Therefore + // we do not reset db_file_state_ and new_files_since_last_dir_sync_. + MutexLock l(&mutex_); + SetFilesystemActive(true); +} + +Status FaultInjectionTestEnv::DeleteFilesCreatedAfterLastDirSync() { + // Because DeleteFile access this container make a copy to avoid deadlock + mutex_.Lock(); + std::set<std::string> new_files(new_files_since_last_dir_sync_.begin(), + new_files_since_last_dir_sync_.end()); + mutex_.Unlock(); + Status s; + std::set<std::string>::const_iterator it; + for (it = new_files.begin(); s.ok() && it != new_files.end(); ++it) { + s = DeleteFile(*it); + } + return s; +} + +void FaultInjectionTestEnv::WritableFileClosed(const FileState& state) { + MutexLock l(&mutex_); + db_file_state_[state.filename_] = state; +} + +Status FileState::DropUnsyncedData() const { + ssize_t sync_pos = pos_at_last_sync_ == -1 ? 0 : pos_at_last_sync_; + return Truncate(filename_, sync_pos); +} + +class FaultInjectionTest { + public: + enum ExpectedVerifResult { VAL_EXPECT_NO_ERROR, VAL_EXPECT_ERROR }; + enum ResetMethod { RESET_DROP_UNSYNCED_DATA, RESET_DELETE_UNSYNCED_FILES }; + + FaultInjectionTestEnv* env_; + std::string dbname_; + Cache* tiny_cache_; + Options options_; + DB* db_; + + FaultInjectionTest() + : env_(new FaultInjectionTestEnv), + tiny_cache_(NewLRUCache(100)), + db_(NULL) { + dbname_ = test::TmpDir() + "/fault_test"; + DestroyDB(dbname_, Options()); // Destroy any db from earlier run + options_.reuse_logs = true; + options_.env = env_; + options_.paranoid_checks = true; + options_.block_cache = tiny_cache_; + options_.create_if_missing = true; + } + + ~FaultInjectionTest() { + CloseDB(); + DestroyDB(dbname_, Options()); + delete tiny_cache_; + delete env_; + } + + void ReuseLogs(bool reuse) { + options_.reuse_logs = reuse; + } + + void Build(int start_idx, int num_vals) { + std::string key_space, value_space; + WriteBatch batch; + for (int i = start_idx; i < start_idx + num_vals; i++) { + Slice key = Key(i, &key_space); + batch.Clear(); + batch.Put(key, Value(i, &value_space)); + WriteOptions options; + ASSERT_OK(db_->Write(options, &batch)); + } + } + + Status ReadValue(int i, std::string* val) const { + std::string key_space, value_space; + Slice key = Key(i, &key_space); + Value(i, &value_space); + ReadOptions options; + return db_->Get(options, key, val); + } + + Status Verify(int start_idx, int num_vals, + ExpectedVerifResult expected) const { + std::string val; + std::string value_space; + Status s; + for (int i = start_idx; i < start_idx + num_vals && s.ok(); i++) { + Value(i, &value_space); + s = ReadValue(i, &val); + if (expected == VAL_EXPECT_NO_ERROR) { + if (s.ok()) { + ASSERT_EQ(value_space, val); + } + } else if (s.ok()) { + fprintf(stderr, "Expected an error at %d, but was OK\n", i); + s = Status::IOError(dbname_, "Expected value error:"); + } else { + s = Status::OK(); // An expected error + } + } + return s; + } + + // Return the ith key + Slice Key(int i, std::string* storage) const { + char buf[100]; + snprintf(buf, sizeof(buf), "%016d", i); + storage->assign(buf, strlen(buf)); + return Slice(*storage); + } + + // Return the value to associate with the specified key + Slice Value(int k, std::string* storage) const { + Random r(k); + return test::RandomString(&r, kValueSize, storage); + } + + Status OpenDB() { + delete db_; + db_ = NULL; + env_->ResetState(); + return DB::Open(options_, dbname_, &db_); + } + + void CloseDB() { + delete db_; + db_ = NULL; + } + + void DeleteAllData() { + Iterator* iter = db_->NewIterator(ReadOptions()); + WriteOptions options; + for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { + ASSERT_OK(db_->Delete(WriteOptions(), iter->key())); + } + + delete iter; + } + + void ResetDBState(ResetMethod reset_method) { + switch (reset_method) { + case RESET_DROP_UNSYNCED_DATA: + ASSERT_OK(env_->DropUnsyncedFileData()); + break; + case RESET_DELETE_UNSYNCED_FILES: + ASSERT_OK(env_->DeleteFilesCreatedAfterLastDirSync()); + break; + default: + assert(false); + } + } + + void PartialCompactTestPreFault(int num_pre_sync, int num_post_sync) { + DeleteAllData(); + Build(0, num_pre_sync); + db_->CompactRange(NULL, NULL); + Build(num_pre_sync, num_post_sync); + } + + void PartialCompactTestReopenWithFault(ResetMethod reset_method, + int num_pre_sync, + int num_post_sync) { + env_->SetFilesystemActive(false); + CloseDB(); + ResetDBState(reset_method); + ASSERT_OK(OpenDB()); + ASSERT_OK(Verify(0, num_pre_sync, FaultInjectionTest::VAL_EXPECT_NO_ERROR)); + ASSERT_OK(Verify(num_pre_sync, num_post_sync, FaultInjectionTest::VAL_EXPECT_ERROR)); + } + + void NoWriteTestPreFault() { + } + + void NoWriteTestReopenWithFault(ResetMethod reset_method) { + CloseDB(); + ResetDBState(reset_method); + ASSERT_OK(OpenDB()); + } + + void DoTest() { + Random rnd(0); + ASSERT_OK(OpenDB()); + for (size_t idx = 0; idx < kNumIterations; idx++) { + int num_pre_sync = rnd.Uniform(kMaxNumValues); + int num_post_sync = rnd.Uniform(kMaxNumValues); + + PartialCompactTestPreFault(num_pre_sync, num_post_sync); + PartialCompactTestReopenWithFault(RESET_DROP_UNSYNCED_DATA, + num_pre_sync, + num_post_sync); + + NoWriteTestPreFault(); + NoWriteTestReopenWithFault(RESET_DROP_UNSYNCED_DATA); + + PartialCompactTestPreFault(num_pre_sync, num_post_sync); + // No new files created so we expect all values since no files will be + // dropped. + PartialCompactTestReopenWithFault(RESET_DELETE_UNSYNCED_FILES, + num_pre_sync + num_post_sync, + 0); + + NoWriteTestPreFault(); + NoWriteTestReopenWithFault(RESET_DELETE_UNSYNCED_FILES); + } + } +}; + +TEST(FaultInjectionTest, FaultTestNoLogReuse) { + ReuseLogs(false); + DoTest(); +} + +TEST(FaultInjectionTest, FaultTestWithLogReuse) { + ReuseLogs(true); + DoTest(); +} + +} // namespace leveldb + +int main(int argc, char** argv) { + return leveldb::test::RunAllTests(); +} diff --git a/src/leveldb/db/filename.cc b/src/leveldb/db/filename.cc new file mode 100644 index 0000000000..da32946d99 --- /dev/null +++ b/src/leveldb/db/filename.cc @@ -0,0 +1,144 @@ +// 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 <ctype.h> +#include <stdio.h> +#include "db/filename.h" +#include "db/dbformat.h" +#include "leveldb/env.h" +#include "util/logging.h" + +namespace leveldb { + +// A utility routine: write "data" to the named file and Sync() it. +extern Status WriteStringToFileSync(Env* env, const Slice& data, + const std::string& fname); + +static std::string MakeFileName(const std::string& name, uint64_t number, + const char* suffix) { + char buf[100]; + snprintf(buf, sizeof(buf), "/%06llu.%s", + static_cast<unsigned long long>(number), + suffix); + return name + buf; +} + +std::string LogFileName(const std::string& name, uint64_t number) { + assert(number > 0); + return MakeFileName(name, number, "log"); +} + +std::string TableFileName(const std::string& name, uint64_t number) { + assert(number > 0); + return MakeFileName(name, number, "ldb"); +} + +std::string SSTTableFileName(const std::string& name, uint64_t number) { + assert(number > 0); + return MakeFileName(name, number, "sst"); +} + +std::string DescriptorFileName(const std::string& dbname, uint64_t number) { + assert(number > 0); + char buf[100]; + snprintf(buf, sizeof(buf), "/MANIFEST-%06llu", + static_cast<unsigned long long>(number)); + return dbname + buf; +} + +std::string CurrentFileName(const std::string& dbname) { + return dbname + "/CURRENT"; +} + +std::string LockFileName(const std::string& dbname) { + return dbname + "/LOCK"; +} + +std::string TempFileName(const std::string& dbname, uint64_t number) { + assert(number > 0); + return MakeFileName(dbname, number, "dbtmp"); +} + +std::string InfoLogFileName(const std::string& dbname) { + return dbname + "/LOG"; +} + +// Return the name of the old info log file for "dbname". +std::string OldInfoLogFileName(const std::string& dbname) { + return dbname + "/LOG.old"; +} + + +// Owned filenames have the form: +// dbname/CURRENT +// dbname/LOCK +// dbname/LOG +// dbname/LOG.old +// dbname/MANIFEST-[0-9]+ +// dbname/[0-9]+.(log|sst|ldb) +bool ParseFileName(const std::string& fname, + uint64_t* number, + FileType* type) { + Slice rest(fname); + if (rest == "CURRENT") { + *number = 0; + *type = kCurrentFile; + } else if (rest == "LOCK") { + *number = 0; + *type = kDBLockFile; + } else if (rest == "LOG" || rest == "LOG.old") { + *number = 0; + *type = kInfoLogFile; + } else if (rest.starts_with("MANIFEST-")) { + rest.remove_prefix(strlen("MANIFEST-")); + uint64_t num; + if (!ConsumeDecimalNumber(&rest, &num)) { + return false; + } + if (!rest.empty()) { + return false; + } + *type = kDescriptorFile; + *number = num; + } else { + // Avoid strtoull() to keep filename format independent of the + // current locale + uint64_t num; + if (!ConsumeDecimalNumber(&rest, &num)) { + return false; + } + Slice suffix = rest; + if (suffix == Slice(".log")) { + *type = kLogFile; + } else if (suffix == Slice(".sst") || suffix == Slice(".ldb")) { + *type = kTableFile; + } else if (suffix == Slice(".dbtmp")) { + *type = kTempFile; + } else { + return false; + } + *number = num; + } + return true; +} + +Status SetCurrentFile(Env* env, const std::string& dbname, + uint64_t descriptor_number) { + // Remove leading "dbname/" and add newline to manifest file name + std::string manifest = DescriptorFileName(dbname, descriptor_number); + Slice contents = manifest; + assert(contents.starts_with(dbname + "/")); + contents.remove_prefix(dbname.size() + 1); + std::string tmp = TempFileName(dbname, descriptor_number); + Status s = WriteStringToFileSync(env, contents.ToString() + "\n", tmp); + if (s.ok()) { + s = env->RenameFile(tmp, CurrentFileName(dbname)); + } + if (!s.ok()) { + env->DeleteFile(tmp); + } + return s; +} + +} // namespace leveldb diff --git a/src/leveldb/db/filename.h b/src/leveldb/db/filename.h new file mode 100644 index 0000000000..87a752605d --- /dev/null +++ b/src/leveldb/db/filename.h @@ -0,0 +1,85 @@ +// 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. +// +// File names used by DB code + +#ifndef STORAGE_LEVELDB_DB_FILENAME_H_ +#define STORAGE_LEVELDB_DB_FILENAME_H_ + +#include <stdint.h> +#include <string> +#include "leveldb/slice.h" +#include "leveldb/status.h" +#include "port/port.h" + +namespace leveldb { + +class Env; + +enum FileType { + kLogFile, + kDBLockFile, + kTableFile, + kDescriptorFile, + kCurrentFile, + kTempFile, + kInfoLogFile // Either the current one, or an old one +}; + +// Return the name of the log file with the specified number +// in the db named by "dbname". The result will be prefixed with +// "dbname". +extern std::string LogFileName(const std::string& dbname, uint64_t number); + +// Return the name of the sstable with the specified number +// in the db named by "dbname". The result will be prefixed with +// "dbname". +extern std::string TableFileName(const std::string& dbname, uint64_t number); + +// Return the legacy file name for an sstable with the specified number +// in the db named by "dbname". The result will be prefixed with +// "dbname". +extern std::string SSTTableFileName(const std::string& dbname, uint64_t number); + +// Return the name of the descriptor file for the db named by +// "dbname" and the specified incarnation number. The result will be +// prefixed with "dbname". +extern std::string DescriptorFileName(const std::string& dbname, + uint64_t number); + +// Return the name of the current file. This file contains the name +// of the current manifest file. The result will be prefixed with +// "dbname". +extern std::string CurrentFileName(const std::string& dbname); + +// Return the name of the lock file for the db named by +// "dbname". The result will be prefixed with "dbname". +extern std::string LockFileName(const std::string& dbname); + +// Return the name of a temporary file owned by the db named "dbname". +// The result will be prefixed with "dbname". +extern std::string TempFileName(const std::string& dbname, uint64_t number); + +// Return the name of the info log file for "dbname". +extern std::string InfoLogFileName(const std::string& dbname); + +// Return the name of the old info log file for "dbname". +extern std::string OldInfoLogFileName(const std::string& dbname); + +// If filename is a leveldb file, store the type of the file in *type. +// The number encoded in the filename is stored in *number. If the +// filename was successfully parsed, returns true. Else return false. +extern bool ParseFileName(const std::string& filename, + uint64_t* number, + FileType* type); + +// Make the CURRENT file point to the descriptor file with the +// specified number. +extern Status SetCurrentFile(Env* env, const std::string& dbname, + uint64_t descriptor_number); + + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_FILENAME_H_ diff --git a/src/leveldb/db/filename_test.cc b/src/leveldb/db/filename_test.cc new file mode 100644 index 0000000000..a32556deaf --- /dev/null +++ b/src/leveldb/db/filename_test.cc @@ -0,0 +1,123 @@ +// 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/filename.h" + +#include "db/dbformat.h" +#include "port/port.h" +#include "util/logging.h" +#include "util/testharness.h" + +namespace leveldb { + +class FileNameTest { }; + +TEST(FileNameTest, Parse) { + Slice db; + FileType type; + uint64_t number; + + // Successful parses + static struct { + const char* fname; + uint64_t number; + FileType type; + } cases[] = { + { "100.log", 100, kLogFile }, + { "0.log", 0, kLogFile }, + { "0.sst", 0, kTableFile }, + { "0.ldb", 0, kTableFile }, + { "CURRENT", 0, kCurrentFile }, + { "LOCK", 0, kDBLockFile }, + { "MANIFEST-2", 2, kDescriptorFile }, + { "MANIFEST-7", 7, kDescriptorFile }, + { "LOG", 0, kInfoLogFile }, + { "LOG.old", 0, kInfoLogFile }, + { "18446744073709551615.log", 18446744073709551615ull, kLogFile }, + }; + for (int i = 0; i < sizeof(cases) / sizeof(cases[0]); i++) { + std::string f = cases[i].fname; + ASSERT_TRUE(ParseFileName(f, &number, &type)) << f; + ASSERT_EQ(cases[i].type, type) << f; + ASSERT_EQ(cases[i].number, number) << f; + } + + // Errors + static const char* errors[] = { + "", + "foo", + "foo-dx-100.log", + ".log", + "", + "manifest", + "CURREN", + "CURRENTX", + "MANIFES", + "MANIFEST", + "MANIFEST-", + "XMANIFEST-3", + "MANIFEST-3x", + "LOC", + "LOCKx", + "LO", + "LOGx", + "18446744073709551616.log", + "184467440737095516150.log", + "100", + "100.", + "100.lop" + }; + for (int i = 0; i < sizeof(errors) / sizeof(errors[0]); i++) { + std::string f = errors[i]; + ASSERT_TRUE(!ParseFileName(f, &number, &type)) << f; + } +} + +TEST(FileNameTest, Construction) { + uint64_t number; + FileType type; + std::string fname; + + fname = CurrentFileName("foo"); + ASSERT_EQ("foo/", std::string(fname.data(), 4)); + ASSERT_TRUE(ParseFileName(fname.c_str() + 4, &number, &type)); + ASSERT_EQ(0, number); + ASSERT_EQ(kCurrentFile, type); + + fname = LockFileName("foo"); + ASSERT_EQ("foo/", std::string(fname.data(), 4)); + ASSERT_TRUE(ParseFileName(fname.c_str() + 4, &number, &type)); + ASSERT_EQ(0, number); + ASSERT_EQ(kDBLockFile, type); + + fname = LogFileName("foo", 192); + ASSERT_EQ("foo/", std::string(fname.data(), 4)); + ASSERT_TRUE(ParseFileName(fname.c_str() + 4, &number, &type)); + ASSERT_EQ(192, number); + ASSERT_EQ(kLogFile, type); + + fname = TableFileName("bar", 200); + ASSERT_EQ("bar/", std::string(fname.data(), 4)); + ASSERT_TRUE(ParseFileName(fname.c_str() + 4, &number, &type)); + ASSERT_EQ(200, number); + ASSERT_EQ(kTableFile, type); + + fname = DescriptorFileName("bar", 100); + ASSERT_EQ("bar/", std::string(fname.data(), 4)); + ASSERT_TRUE(ParseFileName(fname.c_str() + 4, &number, &type)); + ASSERT_EQ(100, number); + ASSERT_EQ(kDescriptorFile, type); + + fname = TempFileName("tmp", 999); + ASSERT_EQ("tmp/", std::string(fname.data(), 4)); + ASSERT_TRUE(ParseFileName(fname.c_str() + 4, &number, &type)); + ASSERT_EQ(999, number); + ASSERT_EQ(kTempFile, type); +} + +} // namespace leveldb + +int main(int argc, char** argv) { + return leveldb::test::RunAllTests(); +} diff --git a/src/leveldb/db/leveldbutil.cc b/src/leveldb/db/leveldbutil.cc new file mode 100644 index 0000000000..9f4b7dd70c --- /dev/null +++ b/src/leveldb/db/leveldbutil.cc @@ -0,0 +1,64 @@ +// Copyright (c) 2012 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 <stdio.h> +#include "leveldb/dumpfile.h" +#include "leveldb/env.h" +#include "leveldb/status.h" + +namespace leveldb { +namespace { + +class StdoutPrinter : public WritableFile { + public: + virtual Status Append(const Slice& data) { + fwrite(data.data(), 1, data.size(), stdout); + return Status::OK(); + } + virtual Status Close() { return Status::OK(); } + virtual Status Flush() { return Status::OK(); } + virtual Status Sync() { return Status::OK(); } +}; + +bool HandleDumpCommand(Env* env, char** files, int num) { + StdoutPrinter printer; + bool ok = true; + for (int i = 0; i < num; i++) { + Status s = DumpFile(env, files[i], &printer); + if (!s.ok()) { + fprintf(stderr, "%s\n", s.ToString().c_str()); + ok = false; + } + } + return ok; +} + +} // namespace +} // namespace leveldb + +static void Usage() { + fprintf( + stderr, + "Usage: leveldbutil command...\n" + " dump files... -- dump contents of specified files\n" + ); +} + +int main(int argc, char** argv) { + leveldb::Env* env = leveldb::Env::Default(); + bool ok = true; + if (argc < 2) { + Usage(); + ok = false; + } else { + std::string command = argv[1]; + if (command == "dump") { + ok = leveldb::HandleDumpCommand(env, argv+2, argc-2); + } else { + Usage(); + ok = false; + } + } + return (ok ? 0 : 1); +} diff --git a/src/leveldb/db/log_format.h b/src/leveldb/db/log_format.h new file mode 100644 index 0000000000..a8c06efe18 --- /dev/null +++ b/src/leveldb/db/log_format.h @@ -0,0 +1,35 @@ +// 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. +// +// Log format information shared by reader and writer. +// See ../doc/log_format.txt for more detail. + +#ifndef STORAGE_LEVELDB_DB_LOG_FORMAT_H_ +#define STORAGE_LEVELDB_DB_LOG_FORMAT_H_ + +namespace leveldb { +namespace log { + +enum RecordType { + // Zero is reserved for preallocated files + kZeroType = 0, + + kFullType = 1, + + // For fragments + kFirstType = 2, + kMiddleType = 3, + kLastType = 4 +}; +static const int kMaxRecordType = kLastType; + +static const int kBlockSize = 32768; + +// Header is checksum (4 bytes), length (2 bytes), type (1 byte). +static const int kHeaderSize = 4 + 2 + 1; + +} // namespace log +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_LOG_FORMAT_H_ diff --git a/src/leveldb/db/log_reader.cc b/src/leveldb/db/log_reader.cc new file mode 100644 index 0000000000..a6d304545d --- /dev/null +++ b/src/leveldb/db/log_reader.cc @@ -0,0 +1,284 @@ +// 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/log_reader.h" + +#include <stdio.h> +#include "leveldb/env.h" +#include "util/coding.h" +#include "util/crc32c.h" + +namespace leveldb { +namespace log { + +Reader::Reporter::~Reporter() { +} + +Reader::Reader(SequentialFile* file, Reporter* reporter, bool checksum, + uint64_t initial_offset) + : file_(file), + reporter_(reporter), + checksum_(checksum), + backing_store_(new char[kBlockSize]), + buffer_(), + eof_(false), + last_record_offset_(0), + end_of_buffer_offset_(0), + initial_offset_(initial_offset), + resyncing_(initial_offset > 0) { +} + +Reader::~Reader() { + delete[] backing_store_; +} + +bool Reader::SkipToInitialBlock() { + size_t offset_in_block = initial_offset_ % kBlockSize; + uint64_t block_start_location = initial_offset_ - offset_in_block; + + // Don't search a block if we'd be in the trailer + if (offset_in_block > kBlockSize - 6) { + offset_in_block = 0; + block_start_location += kBlockSize; + } + + end_of_buffer_offset_ = block_start_location; + + // Skip to start of first block that can contain the initial record + if (block_start_location > 0) { + Status skip_status = file_->Skip(block_start_location); + if (!skip_status.ok()) { + ReportDrop(block_start_location, skip_status); + return false; + } + } + + return true; +} + +bool Reader::ReadRecord(Slice* record, std::string* scratch) { + if (last_record_offset_ < initial_offset_) { + if (!SkipToInitialBlock()) { + return false; + } + } + + scratch->clear(); + record->clear(); + bool in_fragmented_record = false; + // Record offset of the logical record that we're reading + // 0 is a dummy value to make compilers happy + uint64_t prospective_record_offset = 0; + + Slice fragment; + while (true) { + const unsigned int record_type = ReadPhysicalRecord(&fragment); + + // ReadPhysicalRecord may have only had an empty trailer remaining in its + // internal buffer. Calculate the offset of the next physical record now + // that it has returned, properly accounting for its header size. + uint64_t physical_record_offset = + end_of_buffer_offset_ - buffer_.size() - kHeaderSize - fragment.size(); + + if (resyncing_) { + if (record_type == kMiddleType) { + continue; + } else if (record_type == kLastType) { + resyncing_ = false; + continue; + } else { + resyncing_ = false; + } + } + + switch (record_type) { + case kFullType: + if (in_fragmented_record) { + // Handle bug in earlier versions of log::Writer where + // it could emit an empty kFirstType record at the tail end + // of a block followed by a kFullType or kFirstType record + // at the beginning of the next block. + if (scratch->empty()) { + in_fragmented_record = false; + } else { + ReportCorruption(scratch->size(), "partial record without end(1)"); + } + } + prospective_record_offset = physical_record_offset; + scratch->clear(); + *record = fragment; + last_record_offset_ = prospective_record_offset; + return true; + + case kFirstType: + if (in_fragmented_record) { + // Handle bug in earlier versions of log::Writer where + // it could emit an empty kFirstType record at the tail end + // of a block followed by a kFullType or kFirstType record + // at the beginning of the next block. + if (scratch->empty()) { + in_fragmented_record = false; + } else { + ReportCorruption(scratch->size(), "partial record without end(2)"); + } + } + prospective_record_offset = physical_record_offset; + scratch->assign(fragment.data(), fragment.size()); + in_fragmented_record = true; + break; + + case kMiddleType: + if (!in_fragmented_record) { + ReportCorruption(fragment.size(), + "missing start of fragmented record(1)"); + } else { + scratch->append(fragment.data(), fragment.size()); + } + break; + + case kLastType: + if (!in_fragmented_record) { + ReportCorruption(fragment.size(), + "missing start of fragmented record(2)"); + } else { + scratch->append(fragment.data(), fragment.size()); + *record = Slice(*scratch); + last_record_offset_ = prospective_record_offset; + return true; + } + break; + + case kEof: + if (in_fragmented_record) { + // This can be caused by the writer dying immediately after + // writing a physical record but before completing the next; don't + // treat it as a corruption, just ignore the entire logical record. + scratch->clear(); + } + return false; + + case kBadRecord: + if (in_fragmented_record) { + ReportCorruption(scratch->size(), "error in middle of record"); + in_fragmented_record = false; + scratch->clear(); + } + break; + + default: { + char buf[40]; + snprintf(buf, sizeof(buf), "unknown record type %u", record_type); + ReportCorruption( + (fragment.size() + (in_fragmented_record ? scratch->size() : 0)), + buf); + in_fragmented_record = false; + scratch->clear(); + break; + } + } + } + return false; +} + +uint64_t Reader::LastRecordOffset() { + return last_record_offset_; +} + +void Reader::ReportCorruption(uint64_t bytes, const char* reason) { + ReportDrop(bytes, Status::Corruption(reason)); +} + +void Reader::ReportDrop(uint64_t bytes, const Status& reason) { + if (reporter_ != NULL && + end_of_buffer_offset_ - buffer_.size() - bytes >= initial_offset_) { + reporter_->Corruption(static_cast<size_t>(bytes), reason); + } +} + +unsigned int Reader::ReadPhysicalRecord(Slice* result) { + while (true) { + if (buffer_.size() < kHeaderSize) { + if (!eof_) { + // Last read was a full read, so this is a trailer to skip + buffer_.clear(); + Status status = file_->Read(kBlockSize, &buffer_, backing_store_); + end_of_buffer_offset_ += buffer_.size(); + if (!status.ok()) { + buffer_.clear(); + ReportDrop(kBlockSize, status); + eof_ = true; + return kEof; + } else if (buffer_.size() < kBlockSize) { + eof_ = true; + } + continue; + } else { + // Note that if buffer_ is non-empty, we have a truncated header at the + // end of the file, which can be caused by the writer crashing in the + // middle of writing the header. Instead of considering this an error, + // just report EOF. + buffer_.clear(); + return kEof; + } + } + + // Parse the header + const char* header = buffer_.data(); + const uint32_t a = static_cast<uint32_t>(header[4]) & 0xff; + const uint32_t b = static_cast<uint32_t>(header[5]) & 0xff; + const unsigned int type = header[6]; + const uint32_t length = a | (b << 8); + if (kHeaderSize + length > buffer_.size()) { + size_t drop_size = buffer_.size(); + buffer_.clear(); + if (!eof_) { + ReportCorruption(drop_size, "bad record length"); + return kBadRecord; + } + // If the end of the file has been reached without reading |length| bytes + // of payload, assume the writer died in the middle of writing the record. + // Don't report a corruption. + return kEof; + } + + if (type == kZeroType && length == 0) { + // Skip zero length record without reporting any drops since + // such records are produced by the mmap based writing code in + // env_posix.cc that preallocates file regions. + buffer_.clear(); + return kBadRecord; + } + + // Check crc + if (checksum_) { + uint32_t expected_crc = crc32c::Unmask(DecodeFixed32(header)); + uint32_t actual_crc = crc32c::Value(header + 6, 1 + length); + if (actual_crc != expected_crc) { + // Drop the rest of the buffer since "length" itself may have + // been corrupted and if we trust it, we could find some + // fragment of a real log record that just happens to look + // like a valid log record. + size_t drop_size = buffer_.size(); + buffer_.clear(); + ReportCorruption(drop_size, "checksum mismatch"); + return kBadRecord; + } + } + + buffer_.remove_prefix(kHeaderSize + length); + + // Skip physical record that started before initial_offset_ + if (end_of_buffer_offset_ - buffer_.size() - kHeaderSize - length < + initial_offset_) { + result->clear(); + return kBadRecord; + } + + *result = Slice(header + kHeaderSize, length); + return type; + } +} + +} // namespace log +} // namespace leveldb diff --git a/src/leveldb/db/log_reader.h b/src/leveldb/db/log_reader.h new file mode 100644 index 0000000000..8389d61f8f --- /dev/null +++ b/src/leveldb/db/log_reader.h @@ -0,0 +1,113 @@ +// 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_READER_H_ +#define STORAGE_LEVELDB_DB_LOG_READER_H_ + +#include <stdint.h> + +#include "db/log_format.h" +#include "leveldb/slice.h" +#include "leveldb/status.h" + +namespace leveldb { + +class SequentialFile; + +namespace log { + +class Reader { + public: + // Interface for reporting errors. + class Reporter { + public: + virtual ~Reporter(); + + // Some corruption was detected. "size" is the approximate number + // of bytes dropped due to the corruption. + virtual void Corruption(size_t bytes, const Status& status) = 0; + }; + + // Create a reader that will return log records from "*file". + // "*file" must remain live while this Reader is in use. + // + // If "reporter" is non-NULL, it is notified whenever some data is + // dropped due to a detected corruption. "*reporter" must remain + // live while this Reader is in use. + // + // If "checksum" is true, verify checksums if available. + // + // The Reader will start reading at the first record located at physical + // position >= initial_offset within the file. + Reader(SequentialFile* file, Reporter* reporter, bool checksum, + uint64_t initial_offset); + + ~Reader(); + + // Read the next record into *record. Returns true if read + // successfully, false if we hit end of the input. May use + // "*scratch" as temporary storage. The contents filled in *record + // will only be valid until the next mutating operation on this + // reader or the next mutation to *scratch. + bool ReadRecord(Slice* record, std::string* scratch); + + // Returns the physical offset of the last record returned by ReadRecord. + // + // Undefined before the first call to ReadRecord. + uint64_t LastRecordOffset(); + + private: + SequentialFile* const file_; + Reporter* const reporter_; + bool const checksum_; + char* const backing_store_; + Slice buffer_; + bool eof_; // Last Read() indicated EOF by returning < kBlockSize + + // Offset of the last record returned by ReadRecord. + uint64_t last_record_offset_; + // Offset of the first location past the end of buffer_. + uint64_t end_of_buffer_offset_; + + // Offset at which to start looking for the first record to return + uint64_t const initial_offset_; + + // True if we are resynchronizing after a seek (initial_offset_ > 0). In + // particular, a run of kMiddleType and kLastType records can be silently + // skipped in this mode + bool resyncing_; + + // Extend record types with the following special values + enum { + kEof = kMaxRecordType + 1, + // Returned whenever we find an invalid physical record. + // Currently there are three situations in which this happens: + // * The record has an invalid CRC (ReadPhysicalRecord reports a drop) + // * The record is a 0-length record (No drop is reported) + // * The record is below constructor's initial_offset (No drop is reported) + kBadRecord = kMaxRecordType + 2 + }; + + // Skips all blocks that are completely before "initial_offset_". + // + // Returns true on success. Handles reporting. + bool SkipToInitialBlock(); + + // Return type, or one of the preceding special values + unsigned int ReadPhysicalRecord(Slice* result); + + // Reports dropped bytes to the reporter. + // buffer_ must be updated to remove the dropped bytes prior to invocation. + void ReportCorruption(uint64_t bytes, const char* reason); + void ReportDrop(uint64_t bytes, const Status& reason); + + // No copying allowed + Reader(const Reader&); + void operator=(const Reader&); +}; + +} // namespace log +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_LOG_READER_H_ diff --git a/src/leveldb/db/log_test.cc b/src/leveldb/db/log_test.cc new file mode 100644 index 0000000000..48a5928657 --- /dev/null +++ b/src/leveldb/db/log_test.cc @@ -0,0 +1,591 @@ +// 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/log_reader.h" +#include "db/log_writer.h" +#include "leveldb/env.h" +#include "util/coding.h" +#include "util/crc32c.h" +#include "util/random.h" +#include "util/testharness.h" + +namespace leveldb { +namespace log { + +// Construct a string of the specified length made out of the supplied +// partial string. +static std::string BigString(const std::string& partial_string, size_t n) { + std::string result; + while (result.size() < n) { + result.append(partial_string); + } + result.resize(n); + return result; +} + +// Construct a string from a number +static std::string NumberString(int n) { + char buf[50]; + snprintf(buf, sizeof(buf), "%d.", n); + return std::string(buf); +} + +// Return a skewed potentially long string +static std::string RandomSkewedString(int i, Random* rnd) { + return BigString(NumberString(i), rnd->Skewed(17)); +} + +class LogTest { + private: + class StringDest : public WritableFile { + public: + std::string contents_; + + virtual Status Close() { return Status::OK(); } + virtual Status Flush() { return Status::OK(); } + virtual Status Sync() { return Status::OK(); } + virtual Status Append(const Slice& slice) { + contents_.append(slice.data(), slice.size()); + return Status::OK(); + } + }; + + class StringSource : public SequentialFile { + public: + Slice contents_; + bool force_error_; + bool returned_partial_; + StringSource() : force_error_(false), returned_partial_(false) { } + + virtual Status Read(size_t n, Slice* result, char* scratch) { + ASSERT_TRUE(!returned_partial_) << "must not Read() after eof/error"; + + if (force_error_) { + force_error_ = false; + returned_partial_ = true; + return Status::Corruption("read error"); + } + + if (contents_.size() < n) { + n = contents_.size(); + returned_partial_ = true; + } + *result = Slice(contents_.data(), n); + contents_.remove_prefix(n); + return Status::OK(); + } + + virtual Status Skip(uint64_t n) { + if (n > contents_.size()) { + contents_.clear(); + return Status::NotFound("in-memory file skipped past end"); + } + + contents_.remove_prefix(n); + + return Status::OK(); + } + }; + + class ReportCollector : public Reader::Reporter { + public: + size_t dropped_bytes_; + std::string message_; + + ReportCollector() : dropped_bytes_(0) { } + virtual void Corruption(size_t bytes, const Status& status) { + dropped_bytes_ += bytes; + message_.append(status.ToString()); + } + }; + + StringDest dest_; + StringSource source_; + ReportCollector report_; + bool reading_; + Writer* writer_; + Reader* reader_; + + // Record metadata for testing initial offset functionality + static size_t initial_offset_record_sizes_[]; + static uint64_t initial_offset_last_record_offsets_[]; + static int num_initial_offset_records_; + + public: + LogTest() : reading_(false), + writer_(new Writer(&dest_)), + reader_(new Reader(&source_, &report_, true/*checksum*/, + 0/*initial_offset*/)) { + } + + ~LogTest() { + delete writer_; + delete reader_; + } + + void ReopenForAppend() { + delete writer_; + writer_ = new Writer(&dest_, dest_.contents_.size()); + } + + void Write(const std::string& msg) { + ASSERT_TRUE(!reading_) << "Write() after starting to read"; + writer_->AddRecord(Slice(msg)); + } + + size_t WrittenBytes() const { + return dest_.contents_.size(); + } + + std::string Read() { + if (!reading_) { + reading_ = true; + source_.contents_ = Slice(dest_.contents_); + } + std::string scratch; + Slice record; + if (reader_->ReadRecord(&record, &scratch)) { + return record.ToString(); + } else { + return "EOF"; + } + } + + void IncrementByte(int offset, int delta) { + dest_.contents_[offset] += delta; + } + + void SetByte(int offset, char new_byte) { + dest_.contents_[offset] = new_byte; + } + + void ShrinkSize(int bytes) { + dest_.contents_.resize(dest_.contents_.size() - bytes); + } + + void FixChecksum(int header_offset, int len) { + // Compute crc of type/len/data + uint32_t crc = crc32c::Value(&dest_.contents_[header_offset+6], 1 + len); + crc = crc32c::Mask(crc); + EncodeFixed32(&dest_.contents_[header_offset], crc); + } + + void ForceError() { + source_.force_error_ = true; + } + + size_t DroppedBytes() const { + return report_.dropped_bytes_; + } + + std::string ReportMessage() const { + return report_.message_; + } + + // Returns OK iff recorded error message contains "msg" + std::string MatchError(const std::string& msg) const { + if (report_.message_.find(msg) == std::string::npos) { + return report_.message_; + } else { + return "OK"; + } + } + + void WriteInitialOffsetLog() { + for (int i = 0; i < num_initial_offset_records_; i++) { + std::string record(initial_offset_record_sizes_[i], + static_cast<char>('a' + i)); + Write(record); + } + } + + void StartReadingAt(uint64_t initial_offset) { + delete reader_; + reader_ = new Reader(&source_, &report_, true/*checksum*/, initial_offset); + } + + void CheckOffsetPastEndReturnsNoRecords(uint64_t offset_past_end) { + WriteInitialOffsetLog(); + reading_ = true; + source_.contents_ = Slice(dest_.contents_); + Reader* offset_reader = new Reader(&source_, &report_, true/*checksum*/, + WrittenBytes() + offset_past_end); + Slice record; + std::string scratch; + ASSERT_TRUE(!offset_reader->ReadRecord(&record, &scratch)); + delete offset_reader; + } + + void CheckInitialOffsetRecord(uint64_t initial_offset, + int expected_record_offset) { + WriteInitialOffsetLog(); + reading_ = true; + source_.contents_ = Slice(dest_.contents_); + Reader* offset_reader = new Reader(&source_, &report_, true/*checksum*/, + initial_offset); + + // Read all records from expected_record_offset through the last one. + ASSERT_LT(expected_record_offset, num_initial_offset_records_); + for (; expected_record_offset < num_initial_offset_records_; + ++expected_record_offset) { + Slice record; + std::string scratch; + ASSERT_TRUE(offset_reader->ReadRecord(&record, &scratch)); + ASSERT_EQ(initial_offset_record_sizes_[expected_record_offset], + record.size()); + ASSERT_EQ(initial_offset_last_record_offsets_[expected_record_offset], + offset_reader->LastRecordOffset()); + ASSERT_EQ((char)('a' + expected_record_offset), record.data()[0]); + } + delete offset_reader; + } +}; + +size_t LogTest::initial_offset_record_sizes_[] = + {10000, // Two sizable records in first block + 10000, + 2 * log::kBlockSize - 1000, // Span three blocks + 1, + 13716, // Consume all but two bytes of block 3. + log::kBlockSize - kHeaderSize, // Consume the entirety of block 4. + }; + +uint64_t LogTest::initial_offset_last_record_offsets_[] = + {0, + kHeaderSize + 10000, + 2 * (kHeaderSize + 10000), + 2 * (kHeaderSize + 10000) + + (2 * log::kBlockSize - 1000) + 3 * kHeaderSize, + 2 * (kHeaderSize + 10000) + + (2 * log::kBlockSize - 1000) + 3 * kHeaderSize + + kHeaderSize + 1, + 3 * log::kBlockSize, + }; + +// LogTest::initial_offset_last_record_offsets_ must be defined before this. +int LogTest::num_initial_offset_records_ = + sizeof(LogTest::initial_offset_last_record_offsets_)/sizeof(uint64_t); + +TEST(LogTest, Empty) { + ASSERT_EQ("EOF", Read()); +} + +TEST(LogTest, ReadWrite) { + Write("foo"); + Write("bar"); + Write(""); + Write("xxxx"); + ASSERT_EQ("foo", Read()); + ASSERT_EQ("bar", Read()); + ASSERT_EQ("", Read()); + ASSERT_EQ("xxxx", Read()); + ASSERT_EQ("EOF", Read()); + ASSERT_EQ("EOF", Read()); // Make sure reads at eof work +} + +TEST(LogTest, ManyBlocks) { + for (int i = 0; i < 100000; i++) { + Write(NumberString(i)); + } + for (int i = 0; i < 100000; i++) { + ASSERT_EQ(NumberString(i), Read()); + } + ASSERT_EQ("EOF", Read()); +} + +TEST(LogTest, Fragmentation) { + Write("small"); + Write(BigString("medium", 50000)); + Write(BigString("large", 100000)); + ASSERT_EQ("small", Read()); + ASSERT_EQ(BigString("medium", 50000), Read()); + ASSERT_EQ(BigString("large", 100000), Read()); + ASSERT_EQ("EOF", Read()); +} + +TEST(LogTest, MarginalTrailer) { + // Make a trailer that is exactly the same length as an empty record. + const int n = kBlockSize - 2*kHeaderSize; + Write(BigString("foo", n)); + ASSERT_EQ(kBlockSize - kHeaderSize, WrittenBytes()); + Write(""); + Write("bar"); + ASSERT_EQ(BigString("foo", n), Read()); + ASSERT_EQ("", Read()); + ASSERT_EQ("bar", Read()); + ASSERT_EQ("EOF", Read()); +} + +TEST(LogTest, MarginalTrailer2) { + // Make a trailer that is exactly the same length as an empty record. + const int n = kBlockSize - 2*kHeaderSize; + Write(BigString("foo", n)); + ASSERT_EQ(kBlockSize - kHeaderSize, WrittenBytes()); + Write("bar"); + ASSERT_EQ(BigString("foo", n), Read()); + ASSERT_EQ("bar", Read()); + ASSERT_EQ("EOF", Read()); + ASSERT_EQ(0, DroppedBytes()); + ASSERT_EQ("", ReportMessage()); +} + +TEST(LogTest, ShortTrailer) { + const int n = kBlockSize - 2*kHeaderSize + 4; + Write(BigString("foo", n)); + ASSERT_EQ(kBlockSize - kHeaderSize + 4, WrittenBytes()); + Write(""); + Write("bar"); + ASSERT_EQ(BigString("foo", n), Read()); + ASSERT_EQ("", Read()); + ASSERT_EQ("bar", Read()); + ASSERT_EQ("EOF", Read()); +} + +TEST(LogTest, AlignedEof) { + const int n = kBlockSize - 2*kHeaderSize + 4; + Write(BigString("foo", n)); + ASSERT_EQ(kBlockSize - kHeaderSize + 4, WrittenBytes()); + ASSERT_EQ(BigString("foo", n), Read()); + ASSERT_EQ("EOF", Read()); +} + +TEST(LogTest, OpenForAppend) { + Write("hello"); + ReopenForAppend(); + Write("world"); + ASSERT_EQ("hello", Read()); + ASSERT_EQ("world", Read()); + ASSERT_EQ("EOF", Read()); +} + +TEST(LogTest, RandomRead) { + const int N = 500; + Random write_rnd(301); + for (int i = 0; i < N; i++) { + Write(RandomSkewedString(i, &write_rnd)); + } + Random read_rnd(301); + for (int i = 0; i < N; i++) { + ASSERT_EQ(RandomSkewedString(i, &read_rnd), Read()); + } + ASSERT_EQ("EOF", Read()); +} + +// Tests of all the error paths in log_reader.cc follow: + +TEST(LogTest, ReadError) { + Write("foo"); + ForceError(); + ASSERT_EQ("EOF", Read()); + ASSERT_EQ(kBlockSize, DroppedBytes()); + ASSERT_EQ("OK", MatchError("read error")); +} + +TEST(LogTest, BadRecordType) { + Write("foo"); + // Type is stored in header[6] + IncrementByte(6, 100); + FixChecksum(0, 3); + ASSERT_EQ("EOF", Read()); + ASSERT_EQ(3, DroppedBytes()); + ASSERT_EQ("OK", MatchError("unknown record type")); +} + +TEST(LogTest, TruncatedTrailingRecordIsIgnored) { + Write("foo"); + ShrinkSize(4); // Drop all payload as well as a header byte + ASSERT_EQ("EOF", Read()); + // Truncated last record is ignored, not treated as an error. + ASSERT_EQ(0, DroppedBytes()); + ASSERT_EQ("", ReportMessage()); +} + +TEST(LogTest, BadLength) { + const int kPayloadSize = kBlockSize - kHeaderSize; + Write(BigString("bar", kPayloadSize)); + Write("foo"); + // Least significant size byte is stored in header[4]. + IncrementByte(4, 1); + ASSERT_EQ("foo", Read()); + ASSERT_EQ(kBlockSize, DroppedBytes()); + ASSERT_EQ("OK", MatchError("bad record length")); +} + +TEST(LogTest, BadLengthAtEndIsIgnored) { + Write("foo"); + ShrinkSize(1); + ASSERT_EQ("EOF", Read()); + ASSERT_EQ(0, DroppedBytes()); + ASSERT_EQ("", ReportMessage()); +} + +TEST(LogTest, ChecksumMismatch) { + Write("foo"); + IncrementByte(0, 10); + ASSERT_EQ("EOF", Read()); + ASSERT_EQ(10, DroppedBytes()); + ASSERT_EQ("OK", MatchError("checksum mismatch")); +} + +TEST(LogTest, UnexpectedMiddleType) { + Write("foo"); + SetByte(6, kMiddleType); + FixChecksum(0, 3); + ASSERT_EQ("EOF", Read()); + ASSERT_EQ(3, DroppedBytes()); + ASSERT_EQ("OK", MatchError("missing start")); +} + +TEST(LogTest, UnexpectedLastType) { + Write("foo"); + SetByte(6, kLastType); + FixChecksum(0, 3); + ASSERT_EQ("EOF", Read()); + ASSERT_EQ(3, DroppedBytes()); + ASSERT_EQ("OK", MatchError("missing start")); +} + +TEST(LogTest, UnexpectedFullType) { + Write("foo"); + Write("bar"); + SetByte(6, kFirstType); + FixChecksum(0, 3); + ASSERT_EQ("bar", Read()); + ASSERT_EQ("EOF", Read()); + ASSERT_EQ(3, DroppedBytes()); + ASSERT_EQ("OK", MatchError("partial record without end")); +} + +TEST(LogTest, UnexpectedFirstType) { + Write("foo"); + Write(BigString("bar", 100000)); + SetByte(6, kFirstType); + FixChecksum(0, 3); + ASSERT_EQ(BigString("bar", 100000), Read()); + ASSERT_EQ("EOF", Read()); + ASSERT_EQ(3, DroppedBytes()); + ASSERT_EQ("OK", MatchError("partial record without end")); +} + +TEST(LogTest, MissingLastIsIgnored) { + Write(BigString("bar", kBlockSize)); + // Remove the LAST block, including header. + ShrinkSize(14); + ASSERT_EQ("EOF", Read()); + ASSERT_EQ("", ReportMessage()); + ASSERT_EQ(0, DroppedBytes()); +} + +TEST(LogTest, PartialLastIsIgnored) { + Write(BigString("bar", kBlockSize)); + // Cause a bad record length in the LAST block. + ShrinkSize(1); + ASSERT_EQ("EOF", Read()); + ASSERT_EQ("", ReportMessage()); + ASSERT_EQ(0, DroppedBytes()); +} + +TEST(LogTest, SkipIntoMultiRecord) { + // Consider a fragmented record: + // first(R1), middle(R1), last(R1), first(R2) + // If initial_offset points to a record after first(R1) but before first(R2) + // incomplete fragment errors are not actual errors, and must be suppressed + // until a new first or full record is encountered. + Write(BigString("foo", 3*kBlockSize)); + Write("correct"); + StartReadingAt(kBlockSize); + + ASSERT_EQ("correct", Read()); + ASSERT_EQ("", ReportMessage()); + ASSERT_EQ(0, DroppedBytes()); + ASSERT_EQ("EOF", Read()); +} + +TEST(LogTest, ErrorJoinsRecords) { + // Consider two fragmented records: + // first(R1) last(R1) first(R2) last(R2) + // where the middle two fragments disappear. We do not want + // first(R1),last(R2) to get joined and returned as a valid record. + + // Write records that span two blocks + Write(BigString("foo", kBlockSize)); + Write(BigString("bar", kBlockSize)); + Write("correct"); + + // Wipe the middle block + for (int offset = kBlockSize; offset < 2*kBlockSize; offset++) { + SetByte(offset, 'x'); + } + + ASSERT_EQ("correct", Read()); + ASSERT_EQ("EOF", Read()); + const size_t dropped = DroppedBytes(); + ASSERT_LE(dropped, 2*kBlockSize + 100); + ASSERT_GE(dropped, 2*kBlockSize); +} + +TEST(LogTest, ReadStart) { + CheckInitialOffsetRecord(0, 0); +} + +TEST(LogTest, ReadSecondOneOff) { + CheckInitialOffsetRecord(1, 1); +} + +TEST(LogTest, ReadSecondTenThousand) { + CheckInitialOffsetRecord(10000, 1); +} + +TEST(LogTest, ReadSecondStart) { + CheckInitialOffsetRecord(10007, 1); +} + +TEST(LogTest, ReadThirdOneOff) { + CheckInitialOffsetRecord(10008, 2); +} + +TEST(LogTest, ReadThirdStart) { + CheckInitialOffsetRecord(20014, 2); +} + +TEST(LogTest, ReadFourthOneOff) { + CheckInitialOffsetRecord(20015, 3); +} + +TEST(LogTest, ReadFourthFirstBlockTrailer) { + CheckInitialOffsetRecord(log::kBlockSize - 4, 3); +} + +TEST(LogTest, ReadFourthMiddleBlock) { + CheckInitialOffsetRecord(log::kBlockSize + 1, 3); +} + +TEST(LogTest, ReadFourthLastBlock) { + CheckInitialOffsetRecord(2 * log::kBlockSize + 1, 3); +} + +TEST(LogTest, ReadFourthStart) { + CheckInitialOffsetRecord( + 2 * (kHeaderSize + 1000) + (2 * log::kBlockSize - 1000) + 3 * kHeaderSize, + 3); +} + +TEST(LogTest, ReadInitialOffsetIntoBlockPadding) { + CheckInitialOffsetRecord(3 * log::kBlockSize - 3, 5); +} + +TEST(LogTest, ReadEnd) { + CheckOffsetPastEndReturnsNoRecords(0); +} + +TEST(LogTest, ReadPastEnd) { + CheckOffsetPastEndReturnsNoRecords(5); +} + +} // namespace log +} // namespace leveldb + +int main(int argc, char** argv) { + return leveldb::test::RunAllTests(); +} diff --git a/src/leveldb/db/log_writer.cc b/src/leveldb/db/log_writer.cc new file mode 100644 index 0000000000..74a03270da --- /dev/null +++ b/src/leveldb/db/log_writer.cc @@ -0,0 +1,112 @@ +// 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/log_writer.h" + +#include <stdint.h> +#include "leveldb/env.h" +#include "util/coding.h" +#include "util/crc32c.h" + +namespace leveldb { +namespace log { + +static void InitTypeCrc(uint32_t* type_crc) { + for (int i = 0; i <= kMaxRecordType; i++) { + char t = static_cast<char>(i); + type_crc[i] = crc32c::Value(&t, 1); + } +} + +Writer::Writer(WritableFile* dest) + : dest_(dest), + block_offset_(0) { + InitTypeCrc(type_crc_); +} + +Writer::Writer(WritableFile* dest, uint64_t dest_length) + : dest_(dest), block_offset_(dest_length % kBlockSize) { + InitTypeCrc(type_crc_); +} + +Writer::~Writer() { +} + +Status Writer::AddRecord(const Slice& slice) { + const char* ptr = slice.data(); + size_t left = slice.size(); + + // Fragment the record if necessary and emit it. Note that if slice + // is empty, we still want to iterate once to emit a single + // zero-length record + Status s; + bool begin = true; + do { + const int leftover = kBlockSize - block_offset_; + assert(leftover >= 0); + if (leftover < kHeaderSize) { + // Switch to a new block + if (leftover > 0) { + // Fill the trailer (literal below relies on kHeaderSize being 7) + assert(kHeaderSize == 7); + dest_->Append(Slice("\x00\x00\x00\x00\x00\x00", leftover)); + } + block_offset_ = 0; + } + + // Invariant: we never leave < kHeaderSize bytes in a block. + assert(kBlockSize - block_offset_ - kHeaderSize >= 0); + + const size_t avail = kBlockSize - block_offset_ - kHeaderSize; + const size_t fragment_length = (left < avail) ? left : avail; + + RecordType type; + const bool end = (left == fragment_length); + if (begin && end) { + type = kFullType; + } else if (begin) { + type = kFirstType; + } else if (end) { + type = kLastType; + } else { + type = kMiddleType; + } + + s = EmitPhysicalRecord(type, ptr, fragment_length); + ptr += fragment_length; + left -= fragment_length; + begin = false; + } while (s.ok() && left > 0); + return s; +} + +Status Writer::EmitPhysicalRecord(RecordType t, const char* ptr, size_t n) { + assert(n <= 0xffff); // Must fit in two bytes + assert(block_offset_ + kHeaderSize + n <= kBlockSize); + + // Format the header + char buf[kHeaderSize]; + buf[4] = static_cast<char>(n & 0xff); + buf[5] = static_cast<char>(n >> 8); + buf[6] = static_cast<char>(t); + + // Compute the crc of the record type and the payload. + uint32_t crc = crc32c::Extend(type_crc_[t], ptr, n); + crc = crc32c::Mask(crc); // Adjust for storage + EncodeFixed32(buf, crc); + + // Write the header and the payload + Status s = dest_->Append(Slice(buf, kHeaderSize)); + if (s.ok()) { + s = dest_->Append(Slice(ptr, n)); + if (s.ok()) { + s = dest_->Flush(); + } + } + block_offset_ += kHeaderSize + n; + return s; +} + +} // namespace log +} // namespace leveldb diff --git a/src/leveldb/db/log_writer.h b/src/leveldb/db/log_writer.h new file mode 100644 index 0000000000..9e7cc4705b --- /dev/null +++ b/src/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 <stdint.h> +#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(); + + Status AddRecord(const Slice& slice); + + private: + 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]; + + Status EmitPhysicalRecord(RecordType type, const char* ptr, size_t length); + + // No copying allowed + Writer(const Writer&); + void operator=(const Writer&); +}; + +} // namespace log +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_LOG_WRITER_H_ diff --git a/src/leveldb/db/memtable.cc b/src/leveldb/db/memtable.cc new file mode 100644 index 0000000000..bfec0a7e7a --- /dev/null +++ b/src/leveldb/db/memtable.cc @@ -0,0 +1,145 @@ +// 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& cmp) + : comparator_(cmp), + 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) { } + + virtual bool Valid() const { return iter_.Valid(); } + virtual void Seek(const Slice& k) { iter_.Seek(EncodeKey(&tmp_, k)); } + virtual void SeekToFirst() { iter_.SeekToFirst(); } + virtual void SeekToLast() { iter_.SeekToLast(); } + virtual void Next() { iter_.Next(); } + virtual void Prev() { iter_.Prev(); } + virtual Slice key() const { return GetLengthPrefixedSlice(iter_.key()); } + virtual Slice value() const { + Slice key_slice = GetLengthPrefixedSlice(iter_.key()); + return GetLengthPrefixedSlice(key_slice.data() + key_slice.size()); + } + + virtual Status status() const { return Status::OK(); } + + private: + MemTable::Table::Iterator iter_; + std::string tmp_; // For passing to EncodeKey + + // No copying allowed + MemTableIterator(const MemTableIterator&); + void operator=(const MemTableIterator&); +}; + +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); + memcpy(p, key.data(), key_size); + p += key_size; + EncodeFixed64(p, (s << 8) | type); + p += 8; + p = EncodeVarint32(p, val_size); + 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 diff --git a/src/leveldb/db/memtable.h b/src/leveldb/db/memtable.h new file mode 100644 index 0000000000..9f41567cde --- /dev/null +++ b/src/leveldb/db/memtable.h @@ -0,0 +1,88 @@ +// 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 "leveldb/db.h" +#include "db/dbformat.h" +#include "db/skiplist.h" +#include "util/arena.h" + +namespace leveldb { + +class InternalKeyComparator; +class Mutex; +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); + + // 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: + ~MemTable(); // Private since only Unref() should be used to delete it + + struct KeyComparator { + const InternalKeyComparator comparator; + explicit KeyComparator(const InternalKeyComparator& c) : comparator(c) { } + int operator()(const char* a, const char* b) const; + }; + friend class MemTableIterator; + friend class MemTableBackwardIterator; + + typedef SkipList<const char*, KeyComparator> Table; + + KeyComparator comparator_; + int refs_; + Arena arena_; + Table table_; + + // No copying allowed + MemTable(const MemTable&); + void operator=(const MemTable&); +}; + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_MEMTABLE_H_ diff --git a/src/leveldb/db/recovery_test.cc b/src/leveldb/db/recovery_test.cc new file mode 100644 index 0000000000..9596f4288a --- /dev/null +++ b/src/leveldb/db/recovery_test.cc @@ -0,0 +1,324 @@ +// 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 "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/testharness.h" +#include "util/testutil.h" + +namespace leveldb { + +class RecoveryTest { + public: + RecoveryTest() : env_(Env::Default()), db_(NULL) { + dbname_ = test::TmpDir() + "/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_ = NULL; + } + + void Open(Options* options = NULL) { + Close(); + Options opts; + if (options != NULL) { + opts = *options; + } else { + opts.reuse_logs = true; // TODO(sanjay): test both ways + opts.create_if_missing = true; + } + if (opts.env == NULL) { + opts.env = env_; + } + ASSERT_OK(DB::Open(opts, dbname_, &db_)); + 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 = NULL) { + 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; + ASSERT_OK(ReadFileToString(env_, CurrentFileName(dbname_), ¤t)); + 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 DeleteLogFiles() { + std::vector<uint64_t> logs = GetFiles(kLogFile); + for (size_t i = 0; i < logs.size(); i++) { + ASSERT_OK(env_->DeleteFile(LogName(logs[i]))) << LogName(logs[i]); + } + return logs.size(); + } + + uint64_t FirstLogFile() { + return GetFiles(kLogFile)[0]; + } + + std::vector<uint64_t> GetFiles(FileType t) { + std::vector<std::string> filenames; + ASSERT_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; + ASSERT_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_OK(env_->NewWritableFile(fname, &file)); + log::Writer writer(file); + WriteBatch batch; + batch.Put(key, val); + WriteBatchInternal::SetSequence(&batch, seq); + ASSERT_OK(writer.AddRecord(WriteBatchInternal::Contents(&batch))); + ASSERT_OK(file->Flush()); + delete file; + } + + private: + std::string dbname_; + Env* env_; + DB* db_; +}; + +TEST(RecoveryTest, ManifestReused) { + if (!CanAppend()) { + fprintf(stderr, "skipping test because env does not support appending\n"); + return; + } + ASSERT_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(RecoveryTest, LargeManifestCompacted) { + if (!CanAppend()) { + fprintf(stderr, "skipping test because env does not support appending\n"); + return; + } + ASSERT_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_OK(env()->NewAppendableFile(old_manifest, &file)); + std::string zeroes(3*1048576 - static_cast<size_t>(len), 0); + ASSERT_OK(file->Append(zeroes)); + ASSERT_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(RecoveryTest, NoLogFiles) { + ASSERT_OK(Put("foo", "bar")); + ASSERT_EQ(1, DeleteLogFiles()); + Open(); + ASSERT_EQ("NOT_FOUND", Get("foo")); + Open(); + ASSERT_EQ("NOT_FOUND", Get("foo")); +} + +TEST(RecoveryTest, LogFileReuse) { + if (!CanAppend()) { + fprintf(stderr, "skipping test because env does not support appending\n"); + return; + } + for (int i = 0; i < 2; i++) { + ASSERT_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(RecoveryTest, MultipleMemTables) { + // Make a large log. + const int kNum = 1000; + for (int i = 0; i < kNum; i++) { + char buf[100]; + snprintf(buf, sizeof(buf), "%050d", i); + ASSERT_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]; + snprintf(buf, sizeof(buf), "%050d", i); + ASSERT_EQ(buf, Get(buf)); + } +} + +TEST(RecoveryTest, MultipleLogFiles) { + ASSERT_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")); +} + +} // namespace leveldb + +int main(int argc, char** argv) { + return leveldb::test::RunAllTests(); +} diff --git a/src/leveldb/db/repair.cc b/src/leveldb/db/repair.cc new file mode 100644 index 0000000000..4cd4bb047f --- /dev/null +++ b/src/leveldb/db/repair.cc @@ -0,0 +1,461 @@ +// 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. +// +// We recover the contents of the descriptor from the other files we find. +// (1) Any log files are first converted to tables +// (2) We scan every table to compute +// (a) smallest/largest for the table +// (b) largest sequence number in the table +// (3) We generate descriptor contents: +// - log number is set to zero +// - next-file-number is set to 1 + largest file number we found +// - last-sequence-number is set to largest sequence# found across +// all tables (see 2c) +// - compaction pointers are cleared +// - every table file is added at level 0 +// +// Possible optimization 1: +// (a) Compute total size and use to pick appropriate max-level M +// (b) Sort tables by largest sequence# in the table +// (c) For each table: if it overlaps earlier table, place in level-0, +// else place in level-M. +// Possible optimization 2: +// Store per-table metadata (smallest, largest, largest-seq#, ...) +// in the table's meta section to speed up ScanTable. + +#include "db/builder.h" +#include "db/db_impl.h" +#include "db/dbformat.h" +#include "db/filename.h" +#include "db/log_reader.h" +#include "db/log_writer.h" +#include "db/memtable.h" +#include "db/table_cache.h" +#include "db/version_edit.h" +#include "db/write_batch_internal.h" +#include "leveldb/comparator.h" +#include "leveldb/db.h" +#include "leveldb/env.h" + +namespace leveldb { + +namespace { + +class Repairer { + public: + Repairer(const std::string& dbname, const Options& options) + : dbname_(dbname), + env_(options.env), + icmp_(options.comparator), + ipolicy_(options.filter_policy), + options_(SanitizeOptions(dbname, &icmp_, &ipolicy_, options)), + owns_info_log_(options_.info_log != options.info_log), + owns_cache_(options_.block_cache != options.block_cache), + next_file_number_(1) { + // TableCache can be small since we expect each table to be opened once. + table_cache_ = new TableCache(dbname_, &options_, 10); + } + + ~Repairer() { + delete table_cache_; + if (owns_info_log_) { + delete options_.info_log; + } + if (owns_cache_) { + delete options_.block_cache; + } + } + + Status Run() { + Status status = FindFiles(); + if (status.ok()) { + ConvertLogFilesToTables(); + ExtractMetaData(); + status = WriteDescriptor(); + } + if (status.ok()) { + unsigned long long bytes = 0; + for (size_t i = 0; i < tables_.size(); i++) { + bytes += tables_[i].meta.file_size; + } + Log(options_.info_log, + "**** Repaired leveldb %s; " + "recovered %d files; %llu bytes. " + "Some data may have been lost. " + "****", + dbname_.c_str(), + static_cast<int>(tables_.size()), + bytes); + } + return status; + } + + private: + struct TableInfo { + FileMetaData meta; + SequenceNumber max_sequence; + }; + + std::string const dbname_; + Env* const env_; + InternalKeyComparator const icmp_; + InternalFilterPolicy const ipolicy_; + Options const options_; + bool owns_info_log_; + bool owns_cache_; + TableCache* table_cache_; + VersionEdit edit_; + + std::vector<std::string> manifests_; + std::vector<uint64_t> table_numbers_; + std::vector<uint64_t> logs_; + std::vector<TableInfo> tables_; + uint64_t next_file_number_; + + Status FindFiles() { + std::vector<std::string> filenames; + Status status = env_->GetChildren(dbname_, &filenames); + if (!status.ok()) { + return status; + } + if (filenames.empty()) { + return Status::IOError(dbname_, "repair found no files"); + } + + uint64_t number; + FileType type; + for (size_t i = 0; i < filenames.size(); i++) { + if (ParseFileName(filenames[i], &number, &type)) { + if (type == kDescriptorFile) { + manifests_.push_back(filenames[i]); + } else { + if (number + 1 > next_file_number_) { + next_file_number_ = number + 1; + } + if (type == kLogFile) { + logs_.push_back(number); + } else if (type == kTableFile) { + table_numbers_.push_back(number); + } else { + // Ignore other files + } + } + } + } + return status; + } + + void ConvertLogFilesToTables() { + for (size_t i = 0; i < logs_.size(); i++) { + std::string logname = LogFileName(dbname_, logs_[i]); + Status status = ConvertLogToTable(logs_[i]); + if (!status.ok()) { + Log(options_.info_log, "Log #%llu: ignoring conversion error: %s", + (unsigned long long) logs_[i], + status.ToString().c_str()); + } + ArchiveFile(logname); + } + } + + Status ConvertLogToTable(uint64_t log) { + struct LogReporter : public log::Reader::Reporter { + Env* env; + Logger* info_log; + uint64_t lognum; + virtual void Corruption(size_t bytes, const Status& s) { + // We print error messages for corruption, but continue repairing. + Log(info_log, "Log #%llu: dropping %d bytes; %s", + (unsigned long long) lognum, + static_cast<int>(bytes), + s.ToString().c_str()); + } + }; + + // Open the log file + std::string logname = LogFileName(dbname_, log); + SequentialFile* lfile; + Status status = env_->NewSequentialFile(logname, &lfile); + if (!status.ok()) { + return status; + } + + // Create the log reader. + LogReporter reporter; + reporter.env = env_; + reporter.info_log = options_.info_log; + reporter.lognum = log; + // We intentionally make log::Reader do checksumming so that + // corruptions cause entire commits to be skipped instead of + // propagating bad information (like overly large sequence + // numbers). + log::Reader reader(lfile, &reporter, false/*do not checksum*/, + 0/*initial_offset*/); + + // Read all the records and add to a memtable + std::string scratch; + Slice record; + WriteBatch batch; + MemTable* mem = new MemTable(icmp_); + mem->Ref(); + int counter = 0; + while (reader.ReadRecord(&record, &scratch)) { + if (record.size() < 12) { + reporter.Corruption( + record.size(), Status::Corruption("log record too small")); + continue; + } + WriteBatchInternal::SetContents(&batch, record); + status = WriteBatchInternal::InsertInto(&batch, mem); + if (status.ok()) { + counter += WriteBatchInternal::Count(&batch); + } else { + Log(options_.info_log, "Log #%llu: ignoring %s", + (unsigned long long) log, + status.ToString().c_str()); + status = Status::OK(); // Keep going with rest of file + } + } + delete lfile; + + // Do not record a version edit for this conversion to a Table + // since ExtractMetaData() will also generate edits. + FileMetaData meta; + meta.number = next_file_number_++; + Iterator* iter = mem->NewIterator(); + status = BuildTable(dbname_, env_, options_, table_cache_, iter, &meta); + delete iter; + mem->Unref(); + mem = NULL; + if (status.ok()) { + if (meta.file_size > 0) { + table_numbers_.push_back(meta.number); + } + } + Log(options_.info_log, "Log #%llu: %d ops saved to Table #%llu %s", + (unsigned long long) log, + counter, + (unsigned long long) meta.number, + status.ToString().c_str()); + return status; + } + + void ExtractMetaData() { + for (size_t i = 0; i < table_numbers_.size(); i++) { + ScanTable(table_numbers_[i]); + } + } + + Iterator* NewTableIterator(const FileMetaData& meta) { + // Same as compaction iterators: if paranoid_checks are on, turn + // on checksum verification. + ReadOptions r; + r.verify_checksums = options_.paranoid_checks; + return table_cache_->NewIterator(r, meta.number, meta.file_size); + } + + void ScanTable(uint64_t number) { + TableInfo t; + t.meta.number = number; + std::string fname = TableFileName(dbname_, number); + Status status = env_->GetFileSize(fname, &t.meta.file_size); + if (!status.ok()) { + // Try alternate file name. + fname = SSTTableFileName(dbname_, number); + Status s2 = env_->GetFileSize(fname, &t.meta.file_size); + if (s2.ok()) { + status = Status::OK(); + } + } + if (!status.ok()) { + ArchiveFile(TableFileName(dbname_, number)); + ArchiveFile(SSTTableFileName(dbname_, number)); + Log(options_.info_log, "Table #%llu: dropped: %s", + (unsigned long long) t.meta.number, + status.ToString().c_str()); + return; + } + + // Extract metadata by scanning through table. + int counter = 0; + Iterator* iter = NewTableIterator(t.meta); + bool empty = true; + ParsedInternalKey parsed; + t.max_sequence = 0; + for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { + Slice key = iter->key(); + if (!ParseInternalKey(key, &parsed)) { + Log(options_.info_log, "Table #%llu: unparsable key %s", + (unsigned long long) t.meta.number, + EscapeString(key).c_str()); + continue; + } + + counter++; + if (empty) { + empty = false; + t.meta.smallest.DecodeFrom(key); + } + t.meta.largest.DecodeFrom(key); + if (parsed.sequence > t.max_sequence) { + t.max_sequence = parsed.sequence; + } + } + if (!iter->status().ok()) { + status = iter->status(); + } + delete iter; + Log(options_.info_log, "Table #%llu: %d entries %s", + (unsigned long long) t.meta.number, + counter, + status.ToString().c_str()); + + if (status.ok()) { + tables_.push_back(t); + } else { + RepairTable(fname, t); // RepairTable archives input file. + } + } + + void RepairTable(const std::string& src, TableInfo t) { + // We will copy src contents to a new table and then rename the + // new table over the source. + + // Create builder. + std::string copy = TableFileName(dbname_, next_file_number_++); + WritableFile* file; + Status s = env_->NewWritableFile(copy, &file); + if (!s.ok()) { + return; + } + TableBuilder* builder = new TableBuilder(options_, file); + + // Copy data. + Iterator* iter = NewTableIterator(t.meta); + int counter = 0; + for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { + builder->Add(iter->key(), iter->value()); + counter++; + } + delete iter; + + ArchiveFile(src); + if (counter == 0) { + builder->Abandon(); // Nothing to save + } else { + s = builder->Finish(); + if (s.ok()) { + t.meta.file_size = builder->FileSize(); + } + } + delete builder; + builder = NULL; + + if (s.ok()) { + s = file->Close(); + } + delete file; + file = NULL; + + if (counter > 0 && s.ok()) { + std::string orig = TableFileName(dbname_, t.meta.number); + s = env_->RenameFile(copy, orig); + if (s.ok()) { + Log(options_.info_log, "Table #%llu: %d entries repaired", + (unsigned long long) t.meta.number, counter); + tables_.push_back(t); + } + } + if (!s.ok()) { + env_->DeleteFile(copy); + } + } + + Status WriteDescriptor() { + std::string tmp = TempFileName(dbname_, 1); + WritableFile* file; + Status status = env_->NewWritableFile(tmp, &file); + if (!status.ok()) { + return status; + } + + SequenceNumber max_sequence = 0; + for (size_t i = 0; i < tables_.size(); i++) { + if (max_sequence < tables_[i].max_sequence) { + max_sequence = tables_[i].max_sequence; + } + } + + edit_.SetComparatorName(icmp_.user_comparator()->Name()); + edit_.SetLogNumber(0); + edit_.SetNextFile(next_file_number_); + edit_.SetLastSequence(max_sequence); + + for (size_t i = 0; i < tables_.size(); i++) { + // TODO(opt): separate out into multiple levels + const TableInfo& t = tables_[i]; + edit_.AddFile(0, t.meta.number, t.meta.file_size, + t.meta.smallest, t.meta.largest); + } + + //fprintf(stderr, "NewDescriptor:\n%s\n", edit_.DebugString().c_str()); + { + log::Writer log(file); + std::string record; + edit_.EncodeTo(&record); + status = log.AddRecord(record); + } + if (status.ok()) { + status = file->Close(); + } + delete file; + file = NULL; + + if (!status.ok()) { + env_->DeleteFile(tmp); + } else { + // Discard older manifests + for (size_t i = 0; i < manifests_.size(); i++) { + ArchiveFile(dbname_ + "/" + manifests_[i]); + } + + // Install new manifest + status = env_->RenameFile(tmp, DescriptorFileName(dbname_, 1)); + if (status.ok()) { + status = SetCurrentFile(env_, dbname_, 1); + } else { + env_->DeleteFile(tmp); + } + } + return status; + } + + void ArchiveFile(const std::string& fname) { + // Move into another directory. E.g., for + // dir/foo + // rename to + // dir/lost/foo + const char* slash = strrchr(fname.c_str(), '/'); + std::string new_dir; + if (slash != NULL) { + new_dir.assign(fname.data(), slash - fname.data()); + } + new_dir.append("/lost"); + env_->CreateDir(new_dir); // Ignore error + std::string new_file = new_dir; + new_file.append("/"); + new_file.append((slash == NULL) ? fname.c_str() : slash + 1); + Status s = env_->RenameFile(fname, new_file); + Log(options_.info_log, "Archiving %s: %s\n", + fname.c_str(), s.ToString().c_str()); + } +}; +} // namespace + +Status RepairDB(const std::string& dbname, const Options& options) { + Repairer repairer(dbname, options); + return repairer.Run(); +} + +} // namespace leveldb diff --git a/src/leveldb/db/skiplist.h b/src/leveldb/db/skiplist.h new file mode 100644 index 0000000000..8bd77764d8 --- /dev/null +++ b/src/leveldb/db/skiplist.h @@ -0,0 +1,384 @@ +// 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_SKIPLIST_H_ +#define STORAGE_LEVELDB_DB_SKIPLIST_H_ + +// Thread safety +// ------------- +// +// Writes require external synchronization, most likely a mutex. +// Reads require a guarantee that the SkipList will not be destroyed +// while the read is in progress. Apart from that, reads progress +// without any internal locking or synchronization. +// +// Invariants: +// +// (1) Allocated nodes are never deleted until the SkipList is +// destroyed. This is trivially guaranteed by the code since we +// never delete any skip list nodes. +// +// (2) The contents of a Node except for the next/prev pointers are +// immutable after the Node has been linked into the SkipList. +// Only Insert() modifies the list, and it is careful to initialize +// a node and use release-stores to publish the nodes in one or +// more lists. +// +// ... prev vs. next pointer ordering ... + +#include <assert.h> +#include <stdlib.h> +#include "port/port.h" +#include "util/arena.h" +#include "util/random.h" + +namespace leveldb { + +class Arena; + +template<typename Key, class Comparator> +class SkipList { + private: + struct Node; + + public: + // Create a new SkipList object that will use "cmp" for comparing keys, + // and will allocate memory using "*arena". Objects allocated in the arena + // must remain allocated for the lifetime of the skiplist object. + explicit SkipList(Comparator cmp, Arena* arena); + + // Insert key into the list. + // REQUIRES: nothing that compares equal to key is currently in the list. + void Insert(const Key& key); + + // Returns true iff an entry that compares equal to key is in the list. + bool Contains(const Key& key) const; + + // Iteration over the contents of a skip list + class Iterator { + public: + // Initialize an iterator over the specified list. + // The returned iterator is not valid. + explicit Iterator(const SkipList* list); + + // Returns true iff the iterator is positioned at a valid node. + bool Valid() const; + + // Returns the key at the current position. + // REQUIRES: Valid() + const Key& key() const; + + // Advances to the next position. + // REQUIRES: Valid() + void Next(); + + // Advances to the previous position. + // REQUIRES: Valid() + void Prev(); + + // Advance to the first entry with a key >= target + void Seek(const Key& target); + + // Position at the first entry in list. + // Final state of iterator is Valid() iff list is not empty. + void SeekToFirst(); + + // Position at the last entry in list. + // Final state of iterator is Valid() iff list is not empty. + void SeekToLast(); + + private: + const SkipList* list_; + Node* node_; + // Intentionally copyable + }; + + private: + enum { kMaxHeight = 12 }; + + // Immutable after construction + Comparator const compare_; + Arena* const arena_; // Arena used for allocations of nodes + + Node* const head_; + + // Modified only by Insert(). Read racily by readers, but stale + // values are ok. + port::AtomicPointer max_height_; // Height of the entire list + + inline int GetMaxHeight() const { + return static_cast<int>( + reinterpret_cast<intptr_t>(max_height_.NoBarrier_Load())); + } + + // Read/written only by Insert(). + Random rnd_; + + Node* NewNode(const Key& key, int height); + int RandomHeight(); + bool Equal(const Key& a, const Key& b) const { return (compare_(a, b) == 0); } + + // Return true if key is greater than the data stored in "n" + bool KeyIsAfterNode(const Key& key, Node* n) const; + + // Return the earliest node that comes at or after key. + // Return NULL if there is no such node. + // + // If prev is non-NULL, fills prev[level] with pointer to previous + // node at "level" for every level in [0..max_height_-1]. + Node* FindGreaterOrEqual(const Key& key, Node** prev) const; + + // Return the latest node with a key < key. + // Return head_ if there is no such node. + Node* FindLessThan(const Key& key) const; + + // Return the last node in the list. + // Return head_ if list is empty. + Node* FindLast() const; + + // No copying allowed + SkipList(const SkipList&); + void operator=(const SkipList&); +}; + +// Implementation details follow +template<typename Key, class Comparator> +struct SkipList<Key,Comparator>::Node { + explicit Node(const Key& k) : key(k) { } + + Key const key; + + // Accessors/mutators for links. Wrapped in methods so we can + // add the appropriate barriers as necessary. + Node* Next(int n) { + assert(n >= 0); + // Use an 'acquire load' so that we observe a fully initialized + // version of the returned Node. + return reinterpret_cast<Node*>(next_[n].Acquire_Load()); + } + void SetNext(int n, Node* x) { + assert(n >= 0); + // Use a 'release store' so that anybody who reads through this + // pointer observes a fully initialized version of the inserted node. + next_[n].Release_Store(x); + } + + // No-barrier variants that can be safely used in a few locations. + Node* NoBarrier_Next(int n) { + assert(n >= 0); + return reinterpret_cast<Node*>(next_[n].NoBarrier_Load()); + } + void NoBarrier_SetNext(int n, Node* x) { + assert(n >= 0); + next_[n].NoBarrier_Store(x); + } + + private: + // Array of length equal to the node height. next_[0] is lowest level link. + port::AtomicPointer next_[1]; +}; + +template<typename Key, class Comparator> +typename SkipList<Key,Comparator>::Node* +SkipList<Key,Comparator>::NewNode(const Key& key, int height) { + char* mem = arena_->AllocateAligned( + sizeof(Node) + sizeof(port::AtomicPointer) * (height - 1)); + return new (mem) Node(key); +} + +template<typename Key, class Comparator> +inline SkipList<Key,Comparator>::Iterator::Iterator(const SkipList* list) { + list_ = list; + node_ = NULL; +} + +template<typename Key, class Comparator> +inline bool SkipList<Key,Comparator>::Iterator::Valid() const { + return node_ != NULL; +} + +template<typename Key, class Comparator> +inline const Key& SkipList<Key,Comparator>::Iterator::key() const { + assert(Valid()); + return node_->key; +} + +template<typename Key, class Comparator> +inline void SkipList<Key,Comparator>::Iterator::Next() { + assert(Valid()); + node_ = node_->Next(0); +} + +template<typename Key, class Comparator> +inline void SkipList<Key,Comparator>::Iterator::Prev() { + // Instead of using explicit "prev" links, we just search for the + // last node that falls before key. + assert(Valid()); + node_ = list_->FindLessThan(node_->key); + if (node_ == list_->head_) { + node_ = NULL; + } +} + +template<typename Key, class Comparator> +inline void SkipList<Key,Comparator>::Iterator::Seek(const Key& target) { + node_ = list_->FindGreaterOrEqual(target, NULL); +} + +template<typename Key, class Comparator> +inline void SkipList<Key,Comparator>::Iterator::SeekToFirst() { + node_ = list_->head_->Next(0); +} + +template<typename Key, class Comparator> +inline void SkipList<Key,Comparator>::Iterator::SeekToLast() { + node_ = list_->FindLast(); + if (node_ == list_->head_) { + node_ = NULL; + } +} + +template<typename Key, class Comparator> +int SkipList<Key,Comparator>::RandomHeight() { + // Increase height with probability 1 in kBranching + static const unsigned int kBranching = 4; + int height = 1; + while (height < kMaxHeight && ((rnd_.Next() % kBranching) == 0)) { + height++; + } + assert(height > 0); + assert(height <= kMaxHeight); + return height; +} + +template<typename Key, class Comparator> +bool SkipList<Key,Comparator>::KeyIsAfterNode(const Key& key, Node* n) const { + // NULL n is considered infinite + return (n != NULL) && (compare_(n->key, key) < 0); +} + +template<typename Key, class Comparator> +typename SkipList<Key,Comparator>::Node* SkipList<Key,Comparator>::FindGreaterOrEqual(const Key& key, Node** prev) + const { + Node* x = head_; + int level = GetMaxHeight() - 1; + while (true) { + Node* next = x->Next(level); + if (KeyIsAfterNode(key, next)) { + // Keep searching in this list + x = next; + } else { + if (prev != NULL) prev[level] = x; + if (level == 0) { + return next; + } else { + // Switch to next list + level--; + } + } + } +} + +template<typename Key, class Comparator> +typename SkipList<Key,Comparator>::Node* +SkipList<Key,Comparator>::FindLessThan(const Key& key) const { + Node* x = head_; + int level = GetMaxHeight() - 1; + while (true) { + assert(x == head_ || compare_(x->key, key) < 0); + Node* next = x->Next(level); + if (next == NULL || compare_(next->key, key) >= 0) { + if (level == 0) { + return x; + } else { + // Switch to next list + level--; + } + } else { + x = next; + } + } +} + +template<typename Key, class Comparator> +typename SkipList<Key,Comparator>::Node* SkipList<Key,Comparator>::FindLast() + const { + Node* x = head_; + int level = GetMaxHeight() - 1; + while (true) { + Node* next = x->Next(level); + if (next == NULL) { + if (level == 0) { + return x; + } else { + // Switch to next list + level--; + } + } else { + x = next; + } + } +} + +template<typename Key, class Comparator> +SkipList<Key,Comparator>::SkipList(Comparator cmp, Arena* arena) + : compare_(cmp), + arena_(arena), + head_(NewNode(0 /* any key will do */, kMaxHeight)), + max_height_(reinterpret_cast<void*>(1)), + rnd_(0xdeadbeef) { + for (int i = 0; i < kMaxHeight; i++) { + head_->SetNext(i, NULL); + } +} + +template<typename Key, class Comparator> +void SkipList<Key,Comparator>::Insert(const Key& key) { + // TODO(opt): We can use a barrier-free variant of FindGreaterOrEqual() + // here since Insert() is externally synchronized. + Node* prev[kMaxHeight]; + Node* x = FindGreaterOrEqual(key, prev); + + // Our data structure does not allow duplicate insertion + assert(x == NULL || !Equal(key, x->key)); + + int height = RandomHeight(); + if (height > GetMaxHeight()) { + for (int i = GetMaxHeight(); i < height; i++) { + prev[i] = head_; + } + //fprintf(stderr, "Change height from %d to %d\n", max_height_, height); + + // It is ok to mutate max_height_ without any synchronization + // with concurrent readers. A concurrent reader that observes + // the new value of max_height_ will see either the old value of + // new level pointers from head_ (NULL), or a new value set in + // the loop below. In the former case the reader will + // immediately drop to the next level since NULL sorts after all + // keys. In the latter case the reader will use the new node. + max_height_.NoBarrier_Store(reinterpret_cast<void*>(height)); + } + + x = NewNode(key, height); + for (int i = 0; i < height; i++) { + // NoBarrier_SetNext() suffices since we will add a barrier when + // we publish a pointer to "x" in prev[i]. + x->NoBarrier_SetNext(i, prev[i]->NoBarrier_Next(i)); + prev[i]->SetNext(i, x); + } +} + +template<typename Key, class Comparator> +bool SkipList<Key,Comparator>::Contains(const Key& key) const { + Node* x = FindGreaterOrEqual(key, NULL); + if (x != NULL && Equal(key, x->key)) { + return true; + } else { + return false; + } +} + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_SKIPLIST_H_ diff --git a/src/leveldb/db/skiplist_test.cc b/src/leveldb/db/skiplist_test.cc new file mode 100644 index 0000000000..aee1461e1b --- /dev/null +++ b/src/leveldb/db/skiplist_test.cc @@ -0,0 +1,378 @@ +// 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/skiplist.h" +#include <set> +#include "leveldb/env.h" +#include "util/arena.h" +#include "util/hash.h" +#include "util/random.h" +#include "util/testharness.h" + +namespace leveldb { + +typedef uint64_t Key; + +struct Comparator { + int operator()(const Key& a, const Key& b) const { + if (a < b) { + return -1; + } else if (a > b) { + return +1; + } else { + return 0; + } + } +}; + +class SkipTest { }; + +TEST(SkipTest, Empty) { + Arena arena; + Comparator cmp; + SkipList<Key, Comparator> list(cmp, &arena); + ASSERT_TRUE(!list.Contains(10)); + + SkipList<Key, Comparator>::Iterator iter(&list); + ASSERT_TRUE(!iter.Valid()); + iter.SeekToFirst(); + ASSERT_TRUE(!iter.Valid()); + iter.Seek(100); + ASSERT_TRUE(!iter.Valid()); + iter.SeekToLast(); + ASSERT_TRUE(!iter.Valid()); +} + +TEST(SkipTest, InsertAndLookup) { + const int N = 2000; + const int R = 5000; + Random rnd(1000); + std::set<Key> keys; + Arena arena; + Comparator cmp; + SkipList<Key, Comparator> list(cmp, &arena); + for (int i = 0; i < N; i++) { + Key key = rnd.Next() % R; + if (keys.insert(key).second) { + list.Insert(key); + } + } + + for (int i = 0; i < R; i++) { + if (list.Contains(i)) { + ASSERT_EQ(keys.count(i), 1); + } else { + ASSERT_EQ(keys.count(i), 0); + } + } + + // Simple iterator tests + { + SkipList<Key, Comparator>::Iterator iter(&list); + ASSERT_TRUE(!iter.Valid()); + + iter.Seek(0); + ASSERT_TRUE(iter.Valid()); + ASSERT_EQ(*(keys.begin()), iter.key()); + + iter.SeekToFirst(); + ASSERT_TRUE(iter.Valid()); + ASSERT_EQ(*(keys.begin()), iter.key()); + + iter.SeekToLast(); + ASSERT_TRUE(iter.Valid()); + ASSERT_EQ(*(keys.rbegin()), iter.key()); + } + + // Forward iteration test + for (int i = 0; i < R; i++) { + SkipList<Key, Comparator>::Iterator iter(&list); + iter.Seek(i); + + // Compare against model iterator + std::set<Key>::iterator model_iter = keys.lower_bound(i); + for (int j = 0; j < 3; j++) { + if (model_iter == keys.end()) { + ASSERT_TRUE(!iter.Valid()); + break; + } else { + ASSERT_TRUE(iter.Valid()); + ASSERT_EQ(*model_iter, iter.key()); + ++model_iter; + iter.Next(); + } + } + } + + // Backward iteration test + { + SkipList<Key, Comparator>::Iterator iter(&list); + iter.SeekToLast(); + + // Compare against model iterator + for (std::set<Key>::reverse_iterator model_iter = keys.rbegin(); + model_iter != keys.rend(); + ++model_iter) { + ASSERT_TRUE(iter.Valid()); + ASSERT_EQ(*model_iter, iter.key()); + iter.Prev(); + } + ASSERT_TRUE(!iter.Valid()); + } +} + +// We want to make sure that with a single writer and multiple +// concurrent readers (with no synchronization other than when a +// reader's iterator is created), the reader always observes all the +// data that was present in the skip list when the iterator was +// constructor. Because insertions are happening concurrently, we may +// also observe new values that were inserted since the iterator was +// constructed, but we should never miss any values that were present +// at iterator construction time. +// +// We generate multi-part keys: +// <key,gen,hash> +// where: +// key is in range [0..K-1] +// gen is a generation number for key +// hash is hash(key,gen) +// +// The insertion code picks a random key, sets gen to be 1 + the last +// generation number inserted for that key, and sets hash to Hash(key,gen). +// +// At the beginning of a read, we snapshot the last inserted +// generation number for each key. We then iterate, including random +// calls to Next() and Seek(). For every key we encounter, we +// check that it is either expected given the initial snapshot or has +// been concurrently added since the iterator started. +class ConcurrentTest { + private: + static const uint32_t K = 4; + + static uint64_t key(Key key) { return (key >> 40); } + static uint64_t gen(Key key) { return (key >> 8) & 0xffffffffu; } + static uint64_t hash(Key key) { return key & 0xff; } + + static uint64_t HashNumbers(uint64_t k, uint64_t g) { + uint64_t data[2] = { k, g }; + return Hash(reinterpret_cast<char*>(data), sizeof(data), 0); + } + + static Key MakeKey(uint64_t k, uint64_t g) { + assert(sizeof(Key) == sizeof(uint64_t)); + assert(k <= K); // We sometimes pass K to seek to the end of the skiplist + assert(g <= 0xffffffffu); + return ((k << 40) | (g << 8) | (HashNumbers(k, g) & 0xff)); + } + + static bool IsValidKey(Key k) { + return hash(k) == (HashNumbers(key(k), gen(k)) & 0xff); + } + + static Key RandomTarget(Random* rnd) { + switch (rnd->Next() % 10) { + case 0: + // Seek to beginning + return MakeKey(0, 0); + case 1: + // Seek to end + return MakeKey(K, 0); + default: + // Seek to middle + return MakeKey(rnd->Next() % K, 0); + } + } + + // Per-key generation + struct State { + port::AtomicPointer generation[K]; + void Set(int k, intptr_t v) { + generation[k].Release_Store(reinterpret_cast<void*>(v)); + } + intptr_t Get(int k) { + return reinterpret_cast<intptr_t>(generation[k].Acquire_Load()); + } + + State() { + for (int k = 0; k < K; k++) { + Set(k, 0); + } + } + }; + + // Current state of the test + State current_; + + Arena arena_; + + // SkipList is not protected by mu_. We just use a single writer + // thread to modify it. + SkipList<Key, Comparator> list_; + + public: + ConcurrentTest() : list_(Comparator(), &arena_) { } + + // REQUIRES: External synchronization + void WriteStep(Random* rnd) { + const uint32_t k = rnd->Next() % K; + const intptr_t g = current_.Get(k) + 1; + const Key key = MakeKey(k, g); + list_.Insert(key); + current_.Set(k, g); + } + + void ReadStep(Random* rnd) { + // Remember the initial committed state of the skiplist. + State initial_state; + for (int k = 0; k < K; k++) { + initial_state.Set(k, current_.Get(k)); + } + + Key pos = RandomTarget(rnd); + SkipList<Key, Comparator>::Iterator iter(&list_); + iter.Seek(pos); + while (true) { + Key current; + if (!iter.Valid()) { + current = MakeKey(K, 0); + } else { + current = iter.key(); + ASSERT_TRUE(IsValidKey(current)) << current; + } + ASSERT_LE(pos, current) << "should not go backwards"; + + // Verify that everything in [pos,current) was not present in + // initial_state. + while (pos < current) { + ASSERT_LT(key(pos), K) << pos; + + // Note that generation 0 is never inserted, so it is ok if + // <*,0,*> is missing. + ASSERT_TRUE((gen(pos) == 0) || + (gen(pos) > static_cast<Key>(initial_state.Get(key(pos)))) + ) << "key: " << key(pos) + << "; gen: " << gen(pos) + << "; initgen: " + << initial_state.Get(key(pos)); + + // Advance to next key in the valid key space + if (key(pos) < key(current)) { + pos = MakeKey(key(pos) + 1, 0); + } else { + pos = MakeKey(key(pos), gen(pos) + 1); + } + } + + if (!iter.Valid()) { + break; + } + + if (rnd->Next() % 2) { + iter.Next(); + pos = MakeKey(key(pos), gen(pos) + 1); + } else { + Key new_target = RandomTarget(rnd); + if (new_target > pos) { + pos = new_target; + iter.Seek(new_target); + } + } + } + } +}; +const uint32_t ConcurrentTest::K; + +// Simple test that does single-threaded testing of the ConcurrentTest +// scaffolding. +TEST(SkipTest, ConcurrentWithoutThreads) { + ConcurrentTest test; + Random rnd(test::RandomSeed()); + for (int i = 0; i < 10000; i++) { + test.ReadStep(&rnd); + test.WriteStep(&rnd); + } +} + +class TestState { + public: + ConcurrentTest t_; + int seed_; + port::AtomicPointer quit_flag_; + + enum ReaderState { + STARTING, + RUNNING, + DONE + }; + + explicit TestState(int s) + : seed_(s), + quit_flag_(NULL), + state_(STARTING), + state_cv_(&mu_) {} + + void Wait(ReaderState s) { + mu_.Lock(); + while (state_ != s) { + state_cv_.Wait(); + } + mu_.Unlock(); + } + + void Change(ReaderState s) { + mu_.Lock(); + state_ = s; + state_cv_.Signal(); + mu_.Unlock(); + } + + private: + port::Mutex mu_; + ReaderState state_; + port::CondVar state_cv_; +}; + +static void ConcurrentReader(void* arg) { + TestState* state = reinterpret_cast<TestState*>(arg); + Random rnd(state->seed_); + int64_t reads = 0; + state->Change(TestState::RUNNING); + while (!state->quit_flag_.Acquire_Load()) { + state->t_.ReadStep(&rnd); + ++reads; + } + state->Change(TestState::DONE); +} + +static void RunConcurrent(int run) { + const int seed = test::RandomSeed() + (run * 100); + Random rnd(seed); + const int N = 1000; + const int kSize = 1000; + for (int i = 0; i < N; i++) { + if ((i % 100) == 0) { + fprintf(stderr, "Run %d of %d\n", i, N); + } + TestState state(seed + 1); + Env::Default()->Schedule(ConcurrentReader, &state); + state.Wait(TestState::RUNNING); + for (int i = 0; i < kSize; i++) { + state.t_.WriteStep(&rnd); + } + state.quit_flag_.Release_Store(&state); // Any non-NULL arg will do + state.Wait(TestState::DONE); + } +} + +TEST(SkipTest, Concurrent1) { RunConcurrent(1); } +TEST(SkipTest, Concurrent2) { RunConcurrent(2); } +TEST(SkipTest, Concurrent3) { RunConcurrent(3); } +TEST(SkipTest, Concurrent4) { RunConcurrent(4); } +TEST(SkipTest, Concurrent5) { RunConcurrent(5); } + +} // namespace leveldb + +int main(int argc, char** argv) { + return leveldb::test::RunAllTests(); +} diff --git a/src/leveldb/db/snapshot.h b/src/leveldb/db/snapshot.h new file mode 100644 index 0000000000..6ed413c42d --- /dev/null +++ b/src/leveldb/db/snapshot.h @@ -0,0 +1,67 @@ +// 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_SNAPSHOT_H_ +#define STORAGE_LEVELDB_DB_SNAPSHOT_H_ + +#include "db/dbformat.h" +#include "leveldb/db.h" + +namespace leveldb { + +class SnapshotList; + +// Snapshots are kept in a doubly-linked list in the DB. +// Each SnapshotImpl corresponds to a particular sequence number. +class SnapshotImpl : public Snapshot { + public: + SequenceNumber number_; // const after creation + + private: + friend class SnapshotList; + + // SnapshotImpl is kept in a doubly-linked circular list + SnapshotImpl* prev_; + SnapshotImpl* next_; + + SnapshotList* list_; // just for sanity checks +}; + +class SnapshotList { + public: + SnapshotList() { + list_.prev_ = &list_; + list_.next_ = &list_; + } + + bool empty() const { return list_.next_ == &list_; } + SnapshotImpl* oldest() const { assert(!empty()); return list_.next_; } + SnapshotImpl* newest() const { assert(!empty()); return list_.prev_; } + + const SnapshotImpl* New(SequenceNumber seq) { + SnapshotImpl* s = new SnapshotImpl; + s->number_ = seq; + s->list_ = this; + s->next_ = &list_; + s->prev_ = list_.prev_; + s->prev_->next_ = s; + s->next_->prev_ = s; + return s; + } + + void Delete(const SnapshotImpl* s) { + assert(s->list_ == this); + s->prev_->next_ = s->next_; + s->next_->prev_ = s->prev_; + delete s; + } + + private: + // Dummy head of doubly-linked list of snapshots + SnapshotImpl list_; +}; + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_SNAPSHOT_H_ diff --git a/src/leveldb/db/table_cache.cc b/src/leveldb/db/table_cache.cc new file mode 100644 index 0000000000..e3d82cd3ea --- /dev/null +++ b/src/leveldb/db/table_cache.cc @@ -0,0 +1,127 @@ +// 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/table_cache.h" + +#include "db/filename.h" +#include "leveldb/env.h" +#include "leveldb/table.h" +#include "util/coding.h" + +namespace leveldb { + +struct TableAndFile { + RandomAccessFile* file; + Table* table; +}; + +static void DeleteEntry(const Slice& key, void* value) { + TableAndFile* tf = reinterpret_cast<TableAndFile*>(value); + delete tf->table; + delete tf->file; + delete tf; +} + +static void UnrefEntry(void* arg1, void* arg2) { + Cache* cache = reinterpret_cast<Cache*>(arg1); + Cache::Handle* h = reinterpret_cast<Cache::Handle*>(arg2); + cache->Release(h); +} + +TableCache::TableCache(const std::string& dbname, + const Options* options, + int entries) + : env_(options->env), + dbname_(dbname), + options_(options), + cache_(NewLRUCache(entries)) { +} + +TableCache::~TableCache() { + delete cache_; +} + +Status TableCache::FindTable(uint64_t file_number, uint64_t file_size, + Cache::Handle** handle) { + Status s; + char buf[sizeof(file_number)]; + EncodeFixed64(buf, file_number); + Slice key(buf, sizeof(buf)); + *handle = cache_->Lookup(key); + if (*handle == NULL) { + std::string fname = TableFileName(dbname_, file_number); + RandomAccessFile* file = NULL; + Table* table = NULL; + s = env_->NewRandomAccessFile(fname, &file); + if (!s.ok()) { + std::string old_fname = SSTTableFileName(dbname_, file_number); + if (env_->NewRandomAccessFile(old_fname, &file).ok()) { + s = Status::OK(); + } + } + if (s.ok()) { + s = Table::Open(*options_, file, file_size, &table); + } + + if (!s.ok()) { + assert(table == NULL); + delete file; + // We do not cache error results so that if the error is transient, + // or somebody repairs the file, we recover automatically. + } else { + TableAndFile* tf = new TableAndFile; + tf->file = file; + tf->table = table; + *handle = cache_->Insert(key, tf, 1, &DeleteEntry); + } + } + return s; +} + +Iterator* TableCache::NewIterator(const ReadOptions& options, + uint64_t file_number, + uint64_t file_size, + Table** tableptr) { + if (tableptr != NULL) { + *tableptr = NULL; + } + + Cache::Handle* handle = NULL; + Status s = FindTable(file_number, file_size, &handle); + if (!s.ok()) { + return NewErrorIterator(s); + } + + Table* table = reinterpret_cast<TableAndFile*>(cache_->Value(handle))->table; + Iterator* result = table->NewIterator(options); + result->RegisterCleanup(&UnrefEntry, cache_, handle); + if (tableptr != NULL) { + *tableptr = table; + } + return result; +} + +Status TableCache::Get(const ReadOptions& options, + uint64_t file_number, + uint64_t file_size, + const Slice& k, + void* arg, + void (*saver)(void*, const Slice&, const Slice&)) { + Cache::Handle* handle = NULL; + Status s = FindTable(file_number, file_size, &handle); + if (s.ok()) { + Table* t = reinterpret_cast<TableAndFile*>(cache_->Value(handle))->table; + s = t->InternalGet(options, k, arg, saver); + cache_->Release(handle); + } + return s; +} + +void TableCache::Evict(uint64_t file_number) { + char buf[sizeof(file_number)]; + EncodeFixed64(buf, file_number); + cache_->Erase(Slice(buf, sizeof(buf))); +} + +} // namespace leveldb diff --git a/src/leveldb/db/table_cache.h b/src/leveldb/db/table_cache.h new file mode 100644 index 0000000000..8cf4aaf12d --- /dev/null +++ b/src/leveldb/db/table_cache.h @@ -0,0 +1,61 @@ +// 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. +// +// Thread-safe (provides internal synchronization) + +#ifndef STORAGE_LEVELDB_DB_TABLE_CACHE_H_ +#define STORAGE_LEVELDB_DB_TABLE_CACHE_H_ + +#include <string> +#include <stdint.h> +#include "db/dbformat.h" +#include "leveldb/cache.h" +#include "leveldb/table.h" +#include "port/port.h" + +namespace leveldb { + +class Env; + +class TableCache { + public: + TableCache(const std::string& dbname, const Options* options, int entries); + ~TableCache(); + + // Return an iterator for the specified file number (the corresponding + // file length must be exactly "file_size" bytes). If "tableptr" is + // non-NULL, also sets "*tableptr" to point to the Table object + // underlying the returned iterator, or NULL if no Table object underlies + // the returned iterator. The returned "*tableptr" object is owned by + // the cache and should not be deleted, and is valid for as long as the + // returned iterator is live. + Iterator* NewIterator(const ReadOptions& options, + uint64_t file_number, + uint64_t file_size, + Table** tableptr = NULL); + + // If a seek to internal key "k" in specified file finds an entry, + // call (*handle_result)(arg, found_key, found_value). + Status Get(const ReadOptions& options, + uint64_t file_number, + uint64_t file_size, + const Slice& k, + void* arg, + void (*handle_result)(void*, const Slice&, const Slice&)); + + // Evict any entry for the specified file number + void Evict(uint64_t file_number); + + private: + Env* const env_; + const std::string dbname_; + const Options* options_; + Cache* cache_; + + Status FindTable(uint64_t file_number, uint64_t file_size, Cache::Handle**); +}; + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_TABLE_CACHE_H_ diff --git a/src/leveldb/db/version_edit.cc b/src/leveldb/db/version_edit.cc new file mode 100644 index 0000000000..f10a2d58b2 --- /dev/null +++ b/src/leveldb/db/version_edit.cc @@ -0,0 +1,266 @@ +// 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/version_edit.h" + +#include "db/version_set.h" +#include "util/coding.h" + +namespace leveldb { + +// Tag numbers for serialized VersionEdit. These numbers are written to +// disk and should not be changed. +enum Tag { + kComparator = 1, + kLogNumber = 2, + kNextFileNumber = 3, + kLastSequence = 4, + kCompactPointer = 5, + kDeletedFile = 6, + kNewFile = 7, + // 8 was used for large value refs + kPrevLogNumber = 9 +}; + +void VersionEdit::Clear() { + comparator_.clear(); + log_number_ = 0; + prev_log_number_ = 0; + last_sequence_ = 0; + next_file_number_ = 0; + has_comparator_ = false; + has_log_number_ = false; + has_prev_log_number_ = false; + has_next_file_number_ = false; + has_last_sequence_ = false; + deleted_files_.clear(); + new_files_.clear(); +} + +void VersionEdit::EncodeTo(std::string* dst) const { + if (has_comparator_) { + PutVarint32(dst, kComparator); + PutLengthPrefixedSlice(dst, comparator_); + } + if (has_log_number_) { + PutVarint32(dst, kLogNumber); + PutVarint64(dst, log_number_); + } + if (has_prev_log_number_) { + PutVarint32(dst, kPrevLogNumber); + PutVarint64(dst, prev_log_number_); + } + if (has_next_file_number_) { + PutVarint32(dst, kNextFileNumber); + PutVarint64(dst, next_file_number_); + } + if (has_last_sequence_) { + PutVarint32(dst, kLastSequence); + PutVarint64(dst, last_sequence_); + } + + for (size_t i = 0; i < compact_pointers_.size(); i++) { + PutVarint32(dst, kCompactPointer); + PutVarint32(dst, compact_pointers_[i].first); // level + PutLengthPrefixedSlice(dst, compact_pointers_[i].second.Encode()); + } + + for (DeletedFileSet::const_iterator iter = deleted_files_.begin(); + iter != deleted_files_.end(); + ++iter) { + PutVarint32(dst, kDeletedFile); + PutVarint32(dst, iter->first); // level + PutVarint64(dst, iter->second); // file number + } + + for (size_t i = 0; i < new_files_.size(); i++) { + const FileMetaData& f = new_files_[i].second; + PutVarint32(dst, kNewFile); + PutVarint32(dst, new_files_[i].first); // level + PutVarint64(dst, f.number); + PutVarint64(dst, f.file_size); + PutLengthPrefixedSlice(dst, f.smallest.Encode()); + PutLengthPrefixedSlice(dst, f.largest.Encode()); + } +} + +static bool GetInternalKey(Slice* input, InternalKey* dst) { + Slice str; + if (GetLengthPrefixedSlice(input, &str)) { + dst->DecodeFrom(str); + return true; + } else { + return false; + } +} + +static bool GetLevel(Slice* input, int* level) { + uint32_t v; + if (GetVarint32(input, &v) && + v < config::kNumLevels) { + *level = v; + return true; + } else { + return false; + } +} + +Status VersionEdit::DecodeFrom(const Slice& src) { + Clear(); + Slice input = src; + const char* msg = NULL; + uint32_t tag; + + // Temporary storage for parsing + int level; + uint64_t number; + FileMetaData f; + Slice str; + InternalKey key; + + while (msg == NULL && GetVarint32(&input, &tag)) { + switch (tag) { + case kComparator: + if (GetLengthPrefixedSlice(&input, &str)) { + comparator_ = str.ToString(); + has_comparator_ = true; + } else { + msg = "comparator name"; + } + break; + + case kLogNumber: + if (GetVarint64(&input, &log_number_)) { + has_log_number_ = true; + } else { + msg = "log number"; + } + break; + + case kPrevLogNumber: + if (GetVarint64(&input, &prev_log_number_)) { + has_prev_log_number_ = true; + } else { + msg = "previous log number"; + } + break; + + case kNextFileNumber: + if (GetVarint64(&input, &next_file_number_)) { + has_next_file_number_ = true; + } else { + msg = "next file number"; + } + break; + + case kLastSequence: + if (GetVarint64(&input, &last_sequence_)) { + has_last_sequence_ = true; + } else { + msg = "last sequence number"; + } + break; + + case kCompactPointer: + if (GetLevel(&input, &level) && + GetInternalKey(&input, &key)) { + compact_pointers_.push_back(std::make_pair(level, key)); + } else { + msg = "compaction pointer"; + } + break; + + case kDeletedFile: + if (GetLevel(&input, &level) && + GetVarint64(&input, &number)) { + deleted_files_.insert(std::make_pair(level, number)); + } else { + msg = "deleted file"; + } + break; + + case kNewFile: + if (GetLevel(&input, &level) && + GetVarint64(&input, &f.number) && + GetVarint64(&input, &f.file_size) && + GetInternalKey(&input, &f.smallest) && + GetInternalKey(&input, &f.largest)) { + new_files_.push_back(std::make_pair(level, f)); + } else { + msg = "new-file entry"; + } + break; + + default: + msg = "unknown tag"; + break; + } + } + + if (msg == NULL && !input.empty()) { + msg = "invalid tag"; + } + + Status result; + if (msg != NULL) { + result = Status::Corruption("VersionEdit", msg); + } + return result; +} + +std::string VersionEdit::DebugString() const { + std::string r; + r.append("VersionEdit {"); + if (has_comparator_) { + r.append("\n Comparator: "); + r.append(comparator_); + } + if (has_log_number_) { + r.append("\n LogNumber: "); + AppendNumberTo(&r, log_number_); + } + if (has_prev_log_number_) { + r.append("\n PrevLogNumber: "); + AppendNumberTo(&r, prev_log_number_); + } + if (has_next_file_number_) { + r.append("\n NextFile: "); + AppendNumberTo(&r, next_file_number_); + } + if (has_last_sequence_) { + r.append("\n LastSeq: "); + AppendNumberTo(&r, last_sequence_); + } + for (size_t i = 0; i < compact_pointers_.size(); i++) { + r.append("\n CompactPointer: "); + AppendNumberTo(&r, compact_pointers_[i].first); + r.append(" "); + r.append(compact_pointers_[i].second.DebugString()); + } + for (DeletedFileSet::const_iterator iter = deleted_files_.begin(); + iter != deleted_files_.end(); + ++iter) { + r.append("\n DeleteFile: "); + AppendNumberTo(&r, iter->first); + r.append(" "); + AppendNumberTo(&r, iter->second); + } + for (size_t i = 0; i < new_files_.size(); i++) { + const FileMetaData& f = new_files_[i].second; + r.append("\n AddFile: "); + AppendNumberTo(&r, new_files_[i].first); + r.append(" "); + AppendNumberTo(&r, f.number); + r.append(" "); + AppendNumberTo(&r, f.file_size); + r.append(" "); + r.append(f.smallest.DebugString()); + r.append(" .. "); + r.append(f.largest.DebugString()); + } + r.append("\n}\n"); + return r; +} + +} // namespace leveldb diff --git a/src/leveldb/db/version_edit.h b/src/leveldb/db/version_edit.h new file mode 100644 index 0000000000..eaef77b327 --- /dev/null +++ b/src/leveldb/db/version_edit.h @@ -0,0 +1,107 @@ +// 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_VERSION_EDIT_H_ +#define STORAGE_LEVELDB_DB_VERSION_EDIT_H_ + +#include <set> +#include <utility> +#include <vector> +#include "db/dbformat.h" + +namespace leveldb { + +class VersionSet; + +struct FileMetaData { + int refs; + int allowed_seeks; // Seeks allowed until compaction + uint64_t number; + uint64_t file_size; // File size in bytes + InternalKey smallest; // Smallest internal key served by table + InternalKey largest; // Largest internal key served by table + + FileMetaData() : refs(0), allowed_seeks(1 << 30), file_size(0) { } +}; + +class VersionEdit { + public: + VersionEdit() { Clear(); } + ~VersionEdit() { } + + void Clear(); + + void SetComparatorName(const Slice& name) { + has_comparator_ = true; + comparator_ = name.ToString(); + } + void SetLogNumber(uint64_t num) { + has_log_number_ = true; + log_number_ = num; + } + void SetPrevLogNumber(uint64_t num) { + has_prev_log_number_ = true; + prev_log_number_ = num; + } + void SetNextFile(uint64_t num) { + has_next_file_number_ = true; + next_file_number_ = num; + } + void SetLastSequence(SequenceNumber seq) { + has_last_sequence_ = true; + last_sequence_ = seq; + } + void SetCompactPointer(int level, const InternalKey& key) { + compact_pointers_.push_back(std::make_pair(level, key)); + } + + // Add the specified file at the specified number. + // REQUIRES: This version has not been saved (see VersionSet::SaveTo) + // REQUIRES: "smallest" and "largest" are smallest and largest keys in file + void AddFile(int level, uint64_t file, + uint64_t file_size, + const InternalKey& smallest, + const InternalKey& largest) { + FileMetaData f; + f.number = file; + f.file_size = file_size; + f.smallest = smallest; + f.largest = largest; + new_files_.push_back(std::make_pair(level, f)); + } + + // Delete the specified "file" from the specified "level". + void DeleteFile(int level, uint64_t file) { + deleted_files_.insert(std::make_pair(level, file)); + } + + void EncodeTo(std::string* dst) const; + Status DecodeFrom(const Slice& src); + + std::string DebugString() const; + + private: + friend class VersionSet; + + typedef std::set< std::pair<int, uint64_t> > DeletedFileSet; + + std::string comparator_; + uint64_t log_number_; + uint64_t prev_log_number_; + uint64_t next_file_number_; + SequenceNumber last_sequence_; + bool has_comparator_; + bool has_log_number_; + bool has_prev_log_number_; + bool has_next_file_number_; + bool has_last_sequence_; + + std::vector< std::pair<int, InternalKey> > compact_pointers_; + DeletedFileSet deleted_files_; + std::vector< std::pair<int, FileMetaData> > new_files_; +}; + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_VERSION_EDIT_H_ diff --git a/src/leveldb/db/version_edit_test.cc b/src/leveldb/db/version_edit_test.cc new file mode 100644 index 0000000000..280310b49d --- /dev/null +++ b/src/leveldb/db/version_edit_test.cc @@ -0,0 +1,46 @@ +// 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/version_edit.h" +#include "util/testharness.h" + +namespace leveldb { + +static void TestEncodeDecode(const VersionEdit& edit) { + std::string encoded, encoded2; + edit.EncodeTo(&encoded); + VersionEdit parsed; + Status s = parsed.DecodeFrom(encoded); + ASSERT_TRUE(s.ok()) << s.ToString(); + parsed.EncodeTo(&encoded2); + ASSERT_EQ(encoded, encoded2); +} + +class VersionEditTest { }; + +TEST(VersionEditTest, EncodeDecode) { + static const uint64_t kBig = 1ull << 50; + + VersionEdit edit; + for (int i = 0; i < 4; i++) { + TestEncodeDecode(edit); + edit.AddFile(3, kBig + 300 + i, kBig + 400 + i, + InternalKey("foo", kBig + 500 + i, kTypeValue), + InternalKey("zoo", kBig + 600 + i, kTypeDeletion)); + edit.DeleteFile(4, kBig + 700 + i); + edit.SetCompactPointer(i, InternalKey("x", kBig + 900 + i, kTypeValue)); + } + + edit.SetComparatorName("foo"); + edit.SetLogNumber(kBig + 100); + edit.SetNextFile(kBig + 200); + edit.SetLastSequence(kBig + 1000); + TestEncodeDecode(edit); +} + +} // namespace leveldb + +int main(int argc, char** argv) { + return leveldb::test::RunAllTests(); +} diff --git a/src/leveldb/db/version_set.cc b/src/leveldb/db/version_set.cc new file mode 100644 index 0000000000..a5e0f77a6a --- /dev/null +++ b/src/leveldb/db/version_set.cc @@ -0,0 +1,1522 @@ +// 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/version_set.h" + +#include <algorithm> +#include <stdio.h> +#include "db/filename.h" +#include "db/log_reader.h" +#include "db/log_writer.h" +#include "db/memtable.h" +#include "db/table_cache.h" +#include "leveldb/env.h" +#include "leveldb/table_builder.h" +#include "table/merger.h" +#include "table/two_level_iterator.h" +#include "util/coding.h" +#include "util/logging.h" + +namespace leveldb { + +static const int kTargetFileSize = 2 * 1048576; + +// Maximum bytes of overlaps in grandparent (i.e., level+2) before we +// stop building a single file in a level->level+1 compaction. +static const int64_t kMaxGrandParentOverlapBytes = 10 * kTargetFileSize; + +// Maximum number of bytes in all compacted files. We avoid expanding +// the lower level file set of a compaction if it would make the +// total compaction cover more than this many bytes. +static const int64_t kExpandedCompactionByteSizeLimit = 25 * kTargetFileSize; + +static double MaxBytesForLevel(int level) { + // Note: the result for level zero is not really used since we set + // the level-0 compaction threshold based on number of files. + double result = 10 * 1048576.0; // Result for both level-0 and level-1 + while (level > 1) { + result *= 10; + level--; + } + return result; +} + +static uint64_t MaxFileSizeForLevel(int level) { + return kTargetFileSize; // We could vary per level to reduce number of files? +} + +static int64_t TotalFileSize(const std::vector<FileMetaData*>& files) { + int64_t sum = 0; + for (size_t i = 0; i < files.size(); i++) { + sum += files[i]->file_size; + } + return sum; +} + +Version::~Version() { + assert(refs_ == 0); + + // Remove from linked list + prev_->next_ = next_; + next_->prev_ = prev_; + + // Drop references to files + for (int level = 0; level < config::kNumLevels; level++) { + for (size_t i = 0; i < files_[level].size(); i++) { + FileMetaData* f = files_[level][i]; + assert(f->refs > 0); + f->refs--; + if (f->refs <= 0) { + delete f; + } + } + } +} + +int FindFile(const InternalKeyComparator& icmp, + const std::vector<FileMetaData*>& files, + const Slice& key) { + uint32_t left = 0; + uint32_t right = files.size(); + while (left < right) { + uint32_t mid = (left + right) / 2; + const FileMetaData* f = files[mid]; + if (icmp.InternalKeyComparator::Compare(f->largest.Encode(), key) < 0) { + // Key at "mid.largest" is < "target". Therefore all + // files at or before "mid" are uninteresting. + left = mid + 1; + } else { + // Key at "mid.largest" is >= "target". Therefore all files + // after "mid" are uninteresting. + right = mid; + } + } + return right; +} + +static bool AfterFile(const Comparator* ucmp, + const Slice* user_key, const FileMetaData* f) { + // NULL user_key occurs before all keys and is therefore never after *f + return (user_key != NULL && + ucmp->Compare(*user_key, f->largest.user_key()) > 0); +} + +static bool BeforeFile(const Comparator* ucmp, + const Slice* user_key, const FileMetaData* f) { + // NULL user_key occurs after all keys and is therefore never before *f + return (user_key != NULL && + ucmp->Compare(*user_key, f->smallest.user_key()) < 0); +} + +bool SomeFileOverlapsRange( + const InternalKeyComparator& icmp, + bool disjoint_sorted_files, + const std::vector<FileMetaData*>& files, + const Slice* smallest_user_key, + const Slice* largest_user_key) { + const Comparator* ucmp = icmp.user_comparator(); + if (!disjoint_sorted_files) { + // Need to check against all files + for (size_t i = 0; i < files.size(); i++) { + const FileMetaData* f = files[i]; + if (AfterFile(ucmp, smallest_user_key, f) || + BeforeFile(ucmp, largest_user_key, f)) { + // No overlap + } else { + return true; // Overlap + } + } + return false; + } + + // Binary search over file list + uint32_t index = 0; + if (smallest_user_key != NULL) { + // Find the earliest possible internal key for smallest_user_key + InternalKey small(*smallest_user_key, kMaxSequenceNumber,kValueTypeForSeek); + index = FindFile(icmp, files, small.Encode()); + } + + if (index >= files.size()) { + // beginning of range is after all files, so no overlap. + return false; + } + + return !BeforeFile(ucmp, largest_user_key, files[index]); +} + +// An internal iterator. For a given version/level pair, yields +// information about the files in the level. For a given entry, key() +// is the largest key that occurs in the file, and value() is an +// 16-byte value containing the file number and file size, both +// encoded using EncodeFixed64. +class Version::LevelFileNumIterator : public Iterator { + public: + LevelFileNumIterator(const InternalKeyComparator& icmp, + const std::vector<FileMetaData*>* flist) + : icmp_(icmp), + flist_(flist), + index_(flist->size()) { // Marks as invalid + } + virtual bool Valid() const { + return index_ < flist_->size(); + } + virtual void Seek(const Slice& target) { + index_ = FindFile(icmp_, *flist_, target); + } + virtual void SeekToFirst() { index_ = 0; } + virtual void SeekToLast() { + index_ = flist_->empty() ? 0 : flist_->size() - 1; + } + virtual void Next() { + assert(Valid()); + index_++; + } + virtual void Prev() { + assert(Valid()); + if (index_ == 0) { + index_ = flist_->size(); // Marks as invalid + } else { + index_--; + } + } + Slice key() const { + assert(Valid()); + return (*flist_)[index_]->largest.Encode(); + } + Slice value() const { + assert(Valid()); + EncodeFixed64(value_buf_, (*flist_)[index_]->number); + EncodeFixed64(value_buf_+8, (*flist_)[index_]->file_size); + return Slice(value_buf_, sizeof(value_buf_)); + } + virtual Status status() const { return Status::OK(); } + private: + const InternalKeyComparator icmp_; + const std::vector<FileMetaData*>* const flist_; + uint32_t index_; + + // Backing store for value(). Holds the file number and size. + mutable char value_buf_[16]; +}; + +static Iterator* GetFileIterator(void* arg, + const ReadOptions& options, + const Slice& file_value) { + TableCache* cache = reinterpret_cast<TableCache*>(arg); + if (file_value.size() != 16) { + return NewErrorIterator( + Status::Corruption("FileReader invoked with unexpected value")); + } else { + return cache->NewIterator(options, + DecodeFixed64(file_value.data()), + DecodeFixed64(file_value.data() + 8)); + } +} + +Iterator* Version::NewConcatenatingIterator(const ReadOptions& options, + int level) const { + return NewTwoLevelIterator( + new LevelFileNumIterator(vset_->icmp_, &files_[level]), + &GetFileIterator, vset_->table_cache_, options); +} + +void Version::AddIterators(const ReadOptions& options, + std::vector<Iterator*>* iters) { + // Merge all level zero files together since they may overlap + for (size_t i = 0; i < files_[0].size(); i++) { + iters->push_back( + vset_->table_cache_->NewIterator( + options, files_[0][i]->number, files_[0][i]->file_size)); + } + + // For levels > 0, we can use a concatenating iterator that sequentially + // walks through the non-overlapping files in the level, opening them + // lazily. + for (int level = 1; level < config::kNumLevels; level++) { + if (!files_[level].empty()) { + iters->push_back(NewConcatenatingIterator(options, level)); + } + } +} + +// Callback from TableCache::Get() +namespace { +enum SaverState { + kNotFound, + kFound, + kDeleted, + kCorrupt, +}; +struct Saver { + SaverState state; + const Comparator* ucmp; + Slice user_key; + std::string* value; +}; +} +static void SaveValue(void* arg, const Slice& ikey, const Slice& v) { + Saver* s = reinterpret_cast<Saver*>(arg); + ParsedInternalKey parsed_key; + if (!ParseInternalKey(ikey, &parsed_key)) { + s->state = kCorrupt; + } else { + if (s->ucmp->Compare(parsed_key.user_key, s->user_key) == 0) { + s->state = (parsed_key.type == kTypeValue) ? kFound : kDeleted; + if (s->state == kFound) { + s->value->assign(v.data(), v.size()); + } + } + } +} + +static bool NewestFirst(FileMetaData* a, FileMetaData* b) { + return a->number > b->number; +} + +void Version::ForEachOverlapping(Slice user_key, Slice internal_key, + void* arg, + bool (*func)(void*, int, FileMetaData*)) { + // TODO(sanjay): Change Version::Get() to use this function. + const Comparator* ucmp = vset_->icmp_.user_comparator(); + + // Search level-0 in order from newest to oldest. + std::vector<FileMetaData*> tmp; + tmp.reserve(files_[0].size()); + for (uint32_t i = 0; i < files_[0].size(); i++) { + FileMetaData* f = files_[0][i]; + if (ucmp->Compare(user_key, f->smallest.user_key()) >= 0 && + ucmp->Compare(user_key, f->largest.user_key()) <= 0) { + tmp.push_back(f); + } + } + if (!tmp.empty()) { + std::sort(tmp.begin(), tmp.end(), NewestFirst); + for (uint32_t i = 0; i < tmp.size(); i++) { + if (!(*func)(arg, 0, tmp[i])) { + return; + } + } + } + + // Search other levels. + for (int level = 1; level < config::kNumLevels; level++) { + size_t num_files = files_[level].size(); + if (num_files == 0) continue; + + // Binary search to find earliest index whose largest key >= internal_key. + uint32_t index = FindFile(vset_->icmp_, files_[level], internal_key); + if (index < num_files) { + FileMetaData* f = files_[level][index]; + if (ucmp->Compare(user_key, f->smallest.user_key()) < 0) { + // All of "f" is past any data for user_key + } else { + if (!(*func)(arg, level, f)) { + return; + } + } + } + } +} + +Status Version::Get(const ReadOptions& options, + const LookupKey& k, + std::string* value, + GetStats* stats) { + Slice ikey = k.internal_key(); + Slice user_key = k.user_key(); + const Comparator* ucmp = vset_->icmp_.user_comparator(); + Status s; + + stats->seek_file = NULL; + stats->seek_file_level = -1; + FileMetaData* last_file_read = NULL; + int last_file_read_level = -1; + + // We can search level-by-level since entries never hop across + // levels. Therefore we are guaranteed that if we find data + // in an smaller level, later levels are irrelevant. + std::vector<FileMetaData*> tmp; + FileMetaData* tmp2; + for (int level = 0; level < config::kNumLevels; level++) { + size_t num_files = files_[level].size(); + if (num_files == 0) continue; + + // Get the list of files to search in this level + FileMetaData* const* files = &files_[level][0]; + if (level == 0) { + // Level-0 files may overlap each other. Find all files that + // overlap user_key and process them in order from newest to oldest. + tmp.reserve(num_files); + for (uint32_t i = 0; i < num_files; i++) { + FileMetaData* f = files[i]; + if (ucmp->Compare(user_key, f->smallest.user_key()) >= 0 && + ucmp->Compare(user_key, f->largest.user_key()) <= 0) { + tmp.push_back(f); + } + } + if (tmp.empty()) continue; + + std::sort(tmp.begin(), tmp.end(), NewestFirst); + files = &tmp[0]; + num_files = tmp.size(); + } else { + // Binary search to find earliest index whose largest key >= ikey. + uint32_t index = FindFile(vset_->icmp_, files_[level], ikey); + if (index >= num_files) { + files = NULL; + num_files = 0; + } else { + tmp2 = files[index]; + if (ucmp->Compare(user_key, tmp2->smallest.user_key()) < 0) { + // All of "tmp2" is past any data for user_key + files = NULL; + num_files = 0; + } else { + files = &tmp2; + num_files = 1; + } + } + } + + for (uint32_t i = 0; i < num_files; ++i) { + if (last_file_read != NULL && stats->seek_file == NULL) { + // We have had more than one seek for this read. Charge the 1st file. + stats->seek_file = last_file_read; + stats->seek_file_level = last_file_read_level; + } + + FileMetaData* f = files[i]; + last_file_read = f; + last_file_read_level = level; + + Saver saver; + saver.state = kNotFound; + saver.ucmp = ucmp; + saver.user_key = user_key; + saver.value = value; + s = vset_->table_cache_->Get(options, f->number, f->file_size, + ikey, &saver, SaveValue); + if (!s.ok()) { + return s; + } + switch (saver.state) { + case kNotFound: + break; // Keep searching in other files + case kFound: + return s; + case kDeleted: + s = Status::NotFound(Slice()); // Use empty error message for speed + return s; + case kCorrupt: + s = Status::Corruption("corrupted key for ", user_key); + return s; + } + } + } + + return Status::NotFound(Slice()); // Use an empty error message for speed +} + +bool Version::UpdateStats(const GetStats& stats) { + FileMetaData* f = stats.seek_file; + if (f != NULL) { + f->allowed_seeks--; + if (f->allowed_seeks <= 0 && file_to_compact_ == NULL) { + file_to_compact_ = f; + file_to_compact_level_ = stats.seek_file_level; + return true; + } + } + return false; +} + +bool Version::RecordReadSample(Slice internal_key) { + ParsedInternalKey ikey; + if (!ParseInternalKey(internal_key, &ikey)) { + return false; + } + + struct State { + GetStats stats; // Holds first matching file + int matches; + + static bool Match(void* arg, int level, FileMetaData* f) { + State* state = reinterpret_cast<State*>(arg); + state->matches++; + if (state->matches == 1) { + // Remember first match. + state->stats.seek_file = f; + state->stats.seek_file_level = level; + } + // We can stop iterating once we have a second match. + return state->matches < 2; + } + }; + + State state; + state.matches = 0; + ForEachOverlapping(ikey.user_key, internal_key, &state, &State::Match); + + // Must have at least two matches since we want to merge across + // files. But what if we have a single file that contains many + // overwrites and deletions? Should we have another mechanism for + // finding such files? + if (state.matches >= 2) { + // 1MB cost is about 1 seek (see comment in Builder::Apply). + return UpdateStats(state.stats); + } + return false; +} + +void Version::Ref() { + ++refs_; +} + +void Version::Unref() { + assert(this != &vset_->dummy_versions_); + assert(refs_ >= 1); + --refs_; + if (refs_ == 0) { + delete this; + } +} + +bool Version::OverlapInLevel(int level, + const Slice* smallest_user_key, + const Slice* largest_user_key) { + return SomeFileOverlapsRange(vset_->icmp_, (level > 0), files_[level], + smallest_user_key, largest_user_key); +} + +int Version::PickLevelForMemTableOutput( + const Slice& smallest_user_key, + const Slice& largest_user_key) { + int level = 0; + if (!OverlapInLevel(0, &smallest_user_key, &largest_user_key)) { + // Push to next level if there is no overlap in next level, + // and the #bytes overlapping in the level after that are limited. + InternalKey start(smallest_user_key, kMaxSequenceNumber, kValueTypeForSeek); + InternalKey limit(largest_user_key, 0, static_cast<ValueType>(0)); + std::vector<FileMetaData*> overlaps; + while (level < config::kMaxMemCompactLevel) { + if (OverlapInLevel(level + 1, &smallest_user_key, &largest_user_key)) { + break; + } + if (level + 2 < config::kNumLevels) { + // Check that file does not overlap too many grandparent bytes. + GetOverlappingInputs(level + 2, &start, &limit, &overlaps); + const int64_t sum = TotalFileSize(overlaps); + if (sum > kMaxGrandParentOverlapBytes) { + break; + } + } + level++; + } + } + return level; +} + +// Store in "*inputs" all files in "level" that overlap [begin,end] +void Version::GetOverlappingInputs( + int level, + const InternalKey* begin, + const InternalKey* end, + std::vector<FileMetaData*>* inputs) { + assert(level >= 0); + assert(level < config::kNumLevels); + inputs->clear(); + Slice user_begin, user_end; + if (begin != NULL) { + user_begin = begin->user_key(); + } + if (end != NULL) { + user_end = end->user_key(); + } + const Comparator* user_cmp = vset_->icmp_.user_comparator(); + for (size_t i = 0; i < files_[level].size(); ) { + FileMetaData* f = files_[level][i++]; + const Slice file_start = f->smallest.user_key(); + const Slice file_limit = f->largest.user_key(); + if (begin != NULL && user_cmp->Compare(file_limit, user_begin) < 0) { + // "f" is completely before specified range; skip it + } else if (end != NULL && user_cmp->Compare(file_start, user_end) > 0) { + // "f" is completely after specified range; skip it + } else { + inputs->push_back(f); + if (level == 0) { + // Level-0 files may overlap each other. So check if the newly + // added file has expanded the range. If so, restart search. + if (begin != NULL && user_cmp->Compare(file_start, user_begin) < 0) { + user_begin = file_start; + inputs->clear(); + i = 0; + } else if (end != NULL && user_cmp->Compare(file_limit, user_end) > 0) { + user_end = file_limit; + inputs->clear(); + i = 0; + } + } + } + } +} + +std::string Version::DebugString() const { + std::string r; + for (int level = 0; level < config::kNumLevels; level++) { + // E.g., + // --- level 1 --- + // 17:123['a' .. 'd'] + // 20:43['e' .. 'g'] + r.append("--- level "); + AppendNumberTo(&r, level); + r.append(" ---\n"); + const std::vector<FileMetaData*>& files = files_[level]; + for (size_t i = 0; i < files.size(); i++) { + r.push_back(' '); + AppendNumberTo(&r, files[i]->number); + r.push_back(':'); + AppendNumberTo(&r, files[i]->file_size); + r.append("["); + r.append(files[i]->smallest.DebugString()); + r.append(" .. "); + r.append(files[i]->largest.DebugString()); + r.append("]\n"); + } + } + return r; +} + +// A helper class so we can efficiently apply a whole sequence +// of edits to a particular state without creating intermediate +// Versions that contain full copies of the intermediate state. +class VersionSet::Builder { + private: + // Helper to sort by v->files_[file_number].smallest + struct BySmallestKey { + const InternalKeyComparator* internal_comparator; + + bool operator()(FileMetaData* f1, FileMetaData* f2) const { + int r = internal_comparator->Compare(f1->smallest, f2->smallest); + if (r != 0) { + return (r < 0); + } else { + // Break ties by file number + return (f1->number < f2->number); + } + } + }; + + typedef std::set<FileMetaData*, BySmallestKey> FileSet; + struct LevelState { + std::set<uint64_t> deleted_files; + FileSet* added_files; + }; + + VersionSet* vset_; + Version* base_; + LevelState levels_[config::kNumLevels]; + + public: + // Initialize a builder with the files from *base and other info from *vset + Builder(VersionSet* vset, Version* base) + : vset_(vset), + base_(base) { + base_->Ref(); + BySmallestKey cmp; + cmp.internal_comparator = &vset_->icmp_; + for (int level = 0; level < config::kNumLevels; level++) { + levels_[level].added_files = new FileSet(cmp); + } + } + + ~Builder() { + for (int level = 0; level < config::kNumLevels; level++) { + const FileSet* added = levels_[level].added_files; + std::vector<FileMetaData*> to_unref; + to_unref.reserve(added->size()); + for (FileSet::const_iterator it = added->begin(); + it != added->end(); ++it) { + to_unref.push_back(*it); + } + delete added; + for (uint32_t i = 0; i < to_unref.size(); i++) { + FileMetaData* f = to_unref[i]; + f->refs--; + if (f->refs <= 0) { + delete f; + } + } + } + base_->Unref(); + } + + // Apply all of the edits in *edit to the current state. + void Apply(VersionEdit* edit) { + // Update compaction pointers + for (size_t i = 0; i < edit->compact_pointers_.size(); i++) { + const int level = edit->compact_pointers_[i].first; + vset_->compact_pointer_[level] = + edit->compact_pointers_[i].second.Encode().ToString(); + } + + // Delete files + const VersionEdit::DeletedFileSet& del = edit->deleted_files_; + for (VersionEdit::DeletedFileSet::const_iterator iter = del.begin(); + iter != del.end(); + ++iter) { + const int level = iter->first; + const uint64_t number = iter->second; + levels_[level].deleted_files.insert(number); + } + + // Add new files + for (size_t i = 0; i < edit->new_files_.size(); i++) { + const int level = edit->new_files_[i].first; + FileMetaData* f = new FileMetaData(edit->new_files_[i].second); + f->refs = 1; + + // We arrange to automatically compact this file after + // a certain number of seeks. Let's assume: + // (1) One seek costs 10ms + // (2) Writing or reading 1MB costs 10ms (100MB/s) + // (3) A compaction of 1MB does 25MB of IO: + // 1MB read from this level + // 10-12MB read from next level (boundaries may be misaligned) + // 10-12MB written to next level + // This implies that 25 seeks cost the same as the compaction + // of 1MB of data. I.e., one seek costs approximately the + // same as the compaction of 40KB of data. We are a little + // conservative and allow approximately one seek for every 16KB + // of data before triggering a compaction. + f->allowed_seeks = (f->file_size / 16384); + if (f->allowed_seeks < 100) f->allowed_seeks = 100; + + levels_[level].deleted_files.erase(f->number); + levels_[level].added_files->insert(f); + } + } + + // Save the current state in *v. + void SaveTo(Version* v) { + BySmallestKey cmp; + cmp.internal_comparator = &vset_->icmp_; + for (int level = 0; level < config::kNumLevels; level++) { + // Merge the set of added files with the set of pre-existing files. + // Drop any deleted files. Store the result in *v. + const std::vector<FileMetaData*>& base_files = base_->files_[level]; + std::vector<FileMetaData*>::const_iterator base_iter = base_files.begin(); + std::vector<FileMetaData*>::const_iterator base_end = base_files.end(); + const FileSet* added = levels_[level].added_files; + v->files_[level].reserve(base_files.size() + added->size()); + for (FileSet::const_iterator added_iter = added->begin(); + added_iter != added->end(); + ++added_iter) { + // Add all smaller files listed in base_ + for (std::vector<FileMetaData*>::const_iterator bpos + = std::upper_bound(base_iter, base_end, *added_iter, cmp); + base_iter != bpos; + ++base_iter) { + MaybeAddFile(v, level, *base_iter); + } + + MaybeAddFile(v, level, *added_iter); + } + + // Add remaining base files + for (; base_iter != base_end; ++base_iter) { + MaybeAddFile(v, level, *base_iter); + } + +#ifndef NDEBUG + // Make sure there is no overlap in levels > 0 + if (level > 0) { + for (uint32_t i = 1; i < v->files_[level].size(); i++) { + const InternalKey& prev_end = v->files_[level][i-1]->largest; + const InternalKey& this_begin = v->files_[level][i]->smallest; + if (vset_->icmp_.Compare(prev_end, this_begin) >= 0) { + fprintf(stderr, "overlapping ranges in same level %s vs. %s\n", + prev_end.DebugString().c_str(), + this_begin.DebugString().c_str()); + abort(); + } + } + } +#endif + } + } + + void MaybeAddFile(Version* v, int level, FileMetaData* f) { + if (levels_[level].deleted_files.count(f->number) > 0) { + // File is deleted: do nothing + } else { + std::vector<FileMetaData*>* files = &v->files_[level]; + if (level > 0 && !files->empty()) { + // Must not overlap + assert(vset_->icmp_.Compare((*files)[files->size()-1]->largest, + f->smallest) < 0); + } + f->refs++; + files->push_back(f); + } + } +}; + +VersionSet::VersionSet(const std::string& dbname, + const Options* options, + TableCache* table_cache, + const InternalKeyComparator* cmp) + : env_(options->env), + dbname_(dbname), + options_(options), + table_cache_(table_cache), + icmp_(*cmp), + next_file_number_(2), + manifest_file_number_(0), // Filled by Recover() + last_sequence_(0), + log_number_(0), + prev_log_number_(0), + descriptor_file_(NULL), + descriptor_log_(NULL), + dummy_versions_(this), + current_(NULL) { + AppendVersion(new Version(this)); +} + +VersionSet::~VersionSet() { + current_->Unref(); + assert(dummy_versions_.next_ == &dummy_versions_); // List must be empty + delete descriptor_log_; + delete descriptor_file_; +} + +void VersionSet::AppendVersion(Version* v) { + // Make "v" current + assert(v->refs_ == 0); + assert(v != current_); + if (current_ != NULL) { + current_->Unref(); + } + current_ = v; + v->Ref(); + + // Append to linked list + v->prev_ = dummy_versions_.prev_; + v->next_ = &dummy_versions_; + v->prev_->next_ = v; + v->next_->prev_ = v; +} + +Status VersionSet::LogAndApply(VersionEdit* edit, port::Mutex* mu) { + if (edit->has_log_number_) { + assert(edit->log_number_ >= log_number_); + assert(edit->log_number_ < next_file_number_); + } else { + edit->SetLogNumber(log_number_); + } + + if (!edit->has_prev_log_number_) { + edit->SetPrevLogNumber(prev_log_number_); + } + + edit->SetNextFile(next_file_number_); + edit->SetLastSequence(last_sequence_); + + Version* v = new Version(this); + { + Builder builder(this, current_); + builder.Apply(edit); + builder.SaveTo(v); + } + Finalize(v); + + // Initialize new descriptor log file if necessary by creating + // a temporary file that contains a snapshot of the current version. + std::string new_manifest_file; + Status s; + if (descriptor_log_ == NULL) { + // No reason to unlock *mu here since we only hit this path in the + // first call to LogAndApply (when opening the database). + assert(descriptor_file_ == NULL); + new_manifest_file = DescriptorFileName(dbname_, manifest_file_number_); + edit->SetNextFile(next_file_number_); + s = env_->NewWritableFile(new_manifest_file, &descriptor_file_); + if (s.ok()) { + descriptor_log_ = new log::Writer(descriptor_file_); + s = WriteSnapshot(descriptor_log_); + } + } + + // Unlock during expensive MANIFEST log write + { + mu->Unlock(); + + // Write new record to MANIFEST log + if (s.ok()) { + std::string record; + edit->EncodeTo(&record); + s = descriptor_log_->AddRecord(record); + if (s.ok()) { + s = descriptor_file_->Sync(); + } + if (!s.ok()) { + Log(options_->info_log, "MANIFEST write: %s\n", s.ToString().c_str()); + } + } + + // If we just created a new descriptor file, install it by writing a + // new CURRENT file that points to it. + if (s.ok() && !new_manifest_file.empty()) { + s = SetCurrentFile(env_, dbname_, manifest_file_number_); + } + + mu->Lock(); + } + + // Install the new version + if (s.ok()) { + AppendVersion(v); + log_number_ = edit->log_number_; + prev_log_number_ = edit->prev_log_number_; + } else { + delete v; + if (!new_manifest_file.empty()) { + delete descriptor_log_; + delete descriptor_file_; + descriptor_log_ = NULL; + descriptor_file_ = NULL; + env_->DeleteFile(new_manifest_file); + } + } + + return s; +} + +Status VersionSet::Recover(bool *save_manifest) { + struct LogReporter : public log::Reader::Reporter { + Status* status; + virtual void Corruption(size_t bytes, const Status& s) { + if (this->status->ok()) *this->status = s; + } + }; + + // Read "CURRENT" file, which contains a pointer to the current manifest file + std::string current; + Status s = ReadFileToString(env_, CurrentFileName(dbname_), ¤t); + if (!s.ok()) { + return s; + } + if (current.empty() || current[current.size()-1] != '\n') { + return Status::Corruption("CURRENT file does not end with newline"); + } + current.resize(current.size() - 1); + + std::string dscname = dbname_ + "/" + current; + SequentialFile* file; + s = env_->NewSequentialFile(dscname, &file); + if (!s.ok()) { + return s; + } + + bool have_log_number = false; + bool have_prev_log_number = false; + bool have_next_file = false; + bool have_last_sequence = false; + uint64_t next_file = 0; + uint64_t last_sequence = 0; + uint64_t log_number = 0; + uint64_t prev_log_number = 0; + Builder builder(this, current_); + + { + LogReporter reporter; + reporter.status = &s; + log::Reader reader(file, &reporter, true/*checksum*/, 0/*initial_offset*/); + Slice record; + std::string scratch; + while (reader.ReadRecord(&record, &scratch) && s.ok()) { + VersionEdit edit; + s = edit.DecodeFrom(record); + if (s.ok()) { + if (edit.has_comparator_ && + edit.comparator_ != icmp_.user_comparator()->Name()) { + s = Status::InvalidArgument( + edit.comparator_ + " does not match existing comparator ", + icmp_.user_comparator()->Name()); + } + } + + if (s.ok()) { + builder.Apply(&edit); + } + + if (edit.has_log_number_) { + log_number = edit.log_number_; + have_log_number = true; + } + + if (edit.has_prev_log_number_) { + prev_log_number = edit.prev_log_number_; + have_prev_log_number = true; + } + + if (edit.has_next_file_number_) { + next_file = edit.next_file_number_; + have_next_file = true; + } + + if (edit.has_last_sequence_) { + last_sequence = edit.last_sequence_; + have_last_sequence = true; + } + } + } + delete file; + file = NULL; + + if (s.ok()) { + if (!have_next_file) { + s = Status::Corruption("no meta-nextfile entry in descriptor"); + } else if (!have_log_number) { + s = Status::Corruption("no meta-lognumber entry in descriptor"); + } else if (!have_last_sequence) { + s = Status::Corruption("no last-sequence-number entry in descriptor"); + } + + if (!have_prev_log_number) { + prev_log_number = 0; + } + + MarkFileNumberUsed(prev_log_number); + MarkFileNumberUsed(log_number); + } + + if (s.ok()) { + Version* v = new Version(this); + builder.SaveTo(v); + // Install recovered version + Finalize(v); + AppendVersion(v); + manifest_file_number_ = next_file; + next_file_number_ = next_file + 1; + last_sequence_ = last_sequence; + log_number_ = log_number; + prev_log_number_ = prev_log_number; + + // See if we can reuse the existing MANIFEST file. + if (ReuseManifest(dscname, current)) { + // No need to save new manifest + } else { + *save_manifest = true; + } + } + + return s; +} + +bool VersionSet::ReuseManifest(const std::string& dscname, + const std::string& dscbase) { + if (!options_->reuse_logs) { + return false; + } + FileType manifest_type; + uint64_t manifest_number; + uint64_t manifest_size; + if (!ParseFileName(dscbase, &manifest_number, &manifest_type) || + manifest_type != kDescriptorFile || + !env_->GetFileSize(dscname, &manifest_size).ok() || + // Make new compacted MANIFEST if old one is too big + manifest_size >= kTargetFileSize) { + return false; + } + + assert(descriptor_file_ == NULL); + assert(descriptor_log_ == NULL); + Status r = env_->NewAppendableFile(dscname, &descriptor_file_); + if (!r.ok()) { + Log(options_->info_log, "Reuse MANIFEST: %s\n", r.ToString().c_str()); + assert(descriptor_file_ == NULL); + return false; + } + + Log(options_->info_log, "Reusing MANIFEST %s\n", dscname.c_str()); + descriptor_log_ = new log::Writer(descriptor_file_, manifest_size); + manifest_file_number_ = manifest_number; + return true; +} + +void VersionSet::MarkFileNumberUsed(uint64_t number) { + if (next_file_number_ <= number) { + next_file_number_ = number + 1; + } +} + +void VersionSet::Finalize(Version* v) { + // Precomputed best level for next compaction + int best_level = -1; + double best_score = -1; + + for (int level = 0; level < config::kNumLevels-1; level++) { + double score; + if (level == 0) { + // We treat level-0 specially by bounding the number of files + // instead of number of bytes for two reasons: + // + // (1) With larger write-buffer sizes, it is nice not to do too + // many level-0 compactions. + // + // (2) The files in level-0 are merged on every read and + // therefore we wish to avoid too many files when the individual + // file size is small (perhaps because of a small write-buffer + // setting, or very high compression ratios, or lots of + // overwrites/deletions). + score = v->files_[level].size() / + static_cast<double>(config::kL0_CompactionTrigger); + } else { + // Compute the ratio of current size to size limit. + const uint64_t level_bytes = TotalFileSize(v->files_[level]); + score = static_cast<double>(level_bytes) / MaxBytesForLevel(level); + } + + if (score > best_score) { + best_level = level; + best_score = score; + } + } + + v->compaction_level_ = best_level; + v->compaction_score_ = best_score; +} + +Status VersionSet::WriteSnapshot(log::Writer* log) { + // TODO: Break up into multiple records to reduce memory usage on recovery? + + // Save metadata + VersionEdit edit; + edit.SetComparatorName(icmp_.user_comparator()->Name()); + + // Save compaction pointers + for (int level = 0; level < config::kNumLevels; level++) { + if (!compact_pointer_[level].empty()) { + InternalKey key; + key.DecodeFrom(compact_pointer_[level]); + edit.SetCompactPointer(level, key); + } + } + + // Save files + for (int level = 0; level < config::kNumLevels; level++) { + const std::vector<FileMetaData*>& files = current_->files_[level]; + for (size_t i = 0; i < files.size(); i++) { + const FileMetaData* f = files[i]; + edit.AddFile(level, f->number, f->file_size, f->smallest, f->largest); + } + } + + std::string record; + edit.EncodeTo(&record); + return log->AddRecord(record); +} + +int VersionSet::NumLevelFiles(int level) const { + assert(level >= 0); + assert(level < config::kNumLevels); + return current_->files_[level].size(); +} + +const char* VersionSet::LevelSummary(LevelSummaryStorage* scratch) const { + // Update code if kNumLevels changes + assert(config::kNumLevels == 7); + snprintf(scratch->buffer, sizeof(scratch->buffer), + "files[ %d %d %d %d %d %d %d ]", + int(current_->files_[0].size()), + int(current_->files_[1].size()), + int(current_->files_[2].size()), + int(current_->files_[3].size()), + int(current_->files_[4].size()), + int(current_->files_[5].size()), + int(current_->files_[6].size())); + return scratch->buffer; +} + +uint64_t VersionSet::ApproximateOffsetOf(Version* v, const InternalKey& ikey) { + uint64_t result = 0; + for (int level = 0; level < config::kNumLevels; level++) { + const std::vector<FileMetaData*>& files = v->files_[level]; + for (size_t i = 0; i < files.size(); i++) { + if (icmp_.Compare(files[i]->largest, ikey) <= 0) { + // Entire file is before "ikey", so just add the file size + result += files[i]->file_size; + } else if (icmp_.Compare(files[i]->smallest, ikey) > 0) { + // Entire file is after "ikey", so ignore + if (level > 0) { + // Files other than level 0 are sorted by meta->smallest, so + // no further files in this level will contain data for + // "ikey". + break; + } + } else { + // "ikey" falls in the range for this table. Add the + // approximate offset of "ikey" within the table. + Table* tableptr; + Iterator* iter = table_cache_->NewIterator( + ReadOptions(), files[i]->number, files[i]->file_size, &tableptr); + if (tableptr != NULL) { + result += tableptr->ApproximateOffsetOf(ikey.Encode()); + } + delete iter; + } + } + } + return result; +} + +void VersionSet::AddLiveFiles(std::set<uint64_t>* live) { + for (Version* v = dummy_versions_.next_; + v != &dummy_versions_; + v = v->next_) { + for (int level = 0; level < config::kNumLevels; level++) { + const std::vector<FileMetaData*>& files = v->files_[level]; + for (size_t i = 0; i < files.size(); i++) { + live->insert(files[i]->number); + } + } + } +} + +int64_t VersionSet::NumLevelBytes(int level) const { + assert(level >= 0); + assert(level < config::kNumLevels); + return TotalFileSize(current_->files_[level]); +} + +int64_t VersionSet::MaxNextLevelOverlappingBytes() { + int64_t result = 0; + std::vector<FileMetaData*> overlaps; + for (int level = 1; level < config::kNumLevels - 1; level++) { + for (size_t i = 0; i < current_->files_[level].size(); i++) { + const FileMetaData* f = current_->files_[level][i]; + current_->GetOverlappingInputs(level+1, &f->smallest, &f->largest, + &overlaps); + const int64_t sum = TotalFileSize(overlaps); + if (sum > result) { + result = sum; + } + } + } + return result; +} + +// Stores the minimal range that covers all entries in inputs in +// *smallest, *largest. +// REQUIRES: inputs is not empty +void VersionSet::GetRange(const std::vector<FileMetaData*>& inputs, + InternalKey* smallest, + InternalKey* largest) { + assert(!inputs.empty()); + smallest->Clear(); + largest->Clear(); + for (size_t i = 0; i < inputs.size(); i++) { + FileMetaData* f = inputs[i]; + if (i == 0) { + *smallest = f->smallest; + *largest = f->largest; + } else { + if (icmp_.Compare(f->smallest, *smallest) < 0) { + *smallest = f->smallest; + } + if (icmp_.Compare(f->largest, *largest) > 0) { + *largest = f->largest; + } + } + } +} + +// Stores the minimal range that covers all entries in inputs1 and inputs2 +// in *smallest, *largest. +// REQUIRES: inputs is not empty +void VersionSet::GetRange2(const std::vector<FileMetaData*>& inputs1, + const std::vector<FileMetaData*>& inputs2, + InternalKey* smallest, + InternalKey* largest) { + std::vector<FileMetaData*> all = inputs1; + all.insert(all.end(), inputs2.begin(), inputs2.end()); + GetRange(all, smallest, largest); +} + +Iterator* VersionSet::MakeInputIterator(Compaction* c) { + ReadOptions options; + options.verify_checksums = options_->paranoid_checks; + options.fill_cache = false; + + // Level-0 files have to be merged together. For other levels, + // we will make a concatenating iterator per level. + // TODO(opt): use concatenating iterator for level-0 if there is no overlap + const int space = (c->level() == 0 ? c->inputs_[0].size() + 1 : 2); + Iterator** list = new Iterator*[space]; + int num = 0; + for (int which = 0; which < 2; which++) { + if (!c->inputs_[which].empty()) { + if (c->level() + which == 0) { + const std::vector<FileMetaData*>& files = c->inputs_[which]; + for (size_t i = 0; i < files.size(); i++) { + list[num++] = table_cache_->NewIterator( + options, files[i]->number, files[i]->file_size); + } + } else { + // Create concatenating iterator for the files from this level + list[num++] = NewTwoLevelIterator( + new Version::LevelFileNumIterator(icmp_, &c->inputs_[which]), + &GetFileIterator, table_cache_, options); + } + } + } + assert(num <= space); + Iterator* result = NewMergingIterator(&icmp_, list, num); + delete[] list; + return result; +} + +Compaction* VersionSet::PickCompaction() { + Compaction* c; + int level; + + // We prefer compactions triggered by too much data in a level over + // the compactions triggered by seeks. + const bool size_compaction = (current_->compaction_score_ >= 1); + const bool seek_compaction = (current_->file_to_compact_ != NULL); + if (size_compaction) { + level = current_->compaction_level_; + assert(level >= 0); + assert(level+1 < config::kNumLevels); + c = new Compaction(level); + + // Pick the first file that comes after compact_pointer_[level] + for (size_t i = 0; i < current_->files_[level].size(); i++) { + FileMetaData* f = current_->files_[level][i]; + if (compact_pointer_[level].empty() || + icmp_.Compare(f->largest.Encode(), compact_pointer_[level]) > 0) { + c->inputs_[0].push_back(f); + break; + } + } + if (c->inputs_[0].empty()) { + // Wrap-around to the beginning of the key space + c->inputs_[0].push_back(current_->files_[level][0]); + } + } else if (seek_compaction) { + level = current_->file_to_compact_level_; + c = new Compaction(level); + c->inputs_[0].push_back(current_->file_to_compact_); + } else { + return NULL; + } + + c->input_version_ = current_; + c->input_version_->Ref(); + + // Files in level 0 may overlap each other, so pick up all overlapping ones + if (level == 0) { + InternalKey smallest, largest; + GetRange(c->inputs_[0], &smallest, &largest); + // Note that the next call will discard the file we placed in + // c->inputs_[0] earlier and replace it with an overlapping set + // which will include the picked file. + current_->GetOverlappingInputs(0, &smallest, &largest, &c->inputs_[0]); + assert(!c->inputs_[0].empty()); + } + + SetupOtherInputs(c); + + return c; +} + +void VersionSet::SetupOtherInputs(Compaction* c) { + const int level = c->level(); + InternalKey smallest, largest; + GetRange(c->inputs_[0], &smallest, &largest); + + current_->GetOverlappingInputs(level+1, &smallest, &largest, &c->inputs_[1]); + + // Get entire range covered by compaction + InternalKey all_start, all_limit; + GetRange2(c->inputs_[0], c->inputs_[1], &all_start, &all_limit); + + // See if we can grow the number of inputs in "level" without + // changing the number of "level+1" files we pick up. + if (!c->inputs_[1].empty()) { + std::vector<FileMetaData*> expanded0; + current_->GetOverlappingInputs(level, &all_start, &all_limit, &expanded0); + const int64_t inputs0_size = TotalFileSize(c->inputs_[0]); + const int64_t inputs1_size = TotalFileSize(c->inputs_[1]); + const int64_t expanded0_size = TotalFileSize(expanded0); + if (expanded0.size() > c->inputs_[0].size() && + inputs1_size + expanded0_size < kExpandedCompactionByteSizeLimit) { + InternalKey new_start, new_limit; + GetRange(expanded0, &new_start, &new_limit); + std::vector<FileMetaData*> expanded1; + current_->GetOverlappingInputs(level+1, &new_start, &new_limit, + &expanded1); + if (expanded1.size() == c->inputs_[1].size()) { + Log(options_->info_log, + "Expanding@%d %d+%d (%ld+%ld bytes) to %d+%d (%ld+%ld bytes)\n", + level, + int(c->inputs_[0].size()), + int(c->inputs_[1].size()), + long(inputs0_size), long(inputs1_size), + int(expanded0.size()), + int(expanded1.size()), + long(expanded0_size), long(inputs1_size)); + smallest = new_start; + largest = new_limit; + c->inputs_[0] = expanded0; + c->inputs_[1] = expanded1; + GetRange2(c->inputs_[0], c->inputs_[1], &all_start, &all_limit); + } + } + } + + // Compute the set of grandparent files that overlap this compaction + // (parent == level+1; grandparent == level+2) + if (level + 2 < config::kNumLevels) { + current_->GetOverlappingInputs(level + 2, &all_start, &all_limit, + &c->grandparents_); + } + + if (false) { + Log(options_->info_log, "Compacting %d '%s' .. '%s'", + level, + smallest.DebugString().c_str(), + largest.DebugString().c_str()); + } + + // Update the place where we will do the next compaction for this level. + // We update this immediately instead of waiting for the VersionEdit + // to be applied so that if the compaction fails, we will try a different + // key range next time. + compact_pointer_[level] = largest.Encode().ToString(); + c->edit_.SetCompactPointer(level, largest); +} + +Compaction* VersionSet::CompactRange( + int level, + const InternalKey* begin, + const InternalKey* end) { + std::vector<FileMetaData*> inputs; + current_->GetOverlappingInputs(level, begin, end, &inputs); + if (inputs.empty()) { + return NULL; + } + + // Avoid compacting too much in one shot in case the range is large. + // But we cannot do this for level-0 since level-0 files can overlap + // and we must not pick one file and drop another older file if the + // two files overlap. + if (level > 0) { + const uint64_t limit = MaxFileSizeForLevel(level); + uint64_t total = 0; + for (size_t i = 0; i < inputs.size(); i++) { + uint64_t s = inputs[i]->file_size; + total += s; + if (total >= limit) { + inputs.resize(i + 1); + break; + } + } + } + + Compaction* c = new Compaction(level); + c->input_version_ = current_; + c->input_version_->Ref(); + c->inputs_[0] = inputs; + SetupOtherInputs(c); + return c; +} + +Compaction::Compaction(int level) + : level_(level), + max_output_file_size_(MaxFileSizeForLevel(level)), + input_version_(NULL), + grandparent_index_(0), + seen_key_(false), + overlapped_bytes_(0) { + for (int i = 0; i < config::kNumLevels; i++) { + level_ptrs_[i] = 0; + } +} + +Compaction::~Compaction() { + if (input_version_ != NULL) { + input_version_->Unref(); + } +} + +bool Compaction::IsTrivialMove() const { + // Avoid a move if there is lots of overlapping grandparent data. + // Otherwise, the move could create a parent file that will require + // a very expensive merge later on. + return (num_input_files(0) == 1 && + num_input_files(1) == 0 && + TotalFileSize(grandparents_) <= kMaxGrandParentOverlapBytes); +} + +void Compaction::AddInputDeletions(VersionEdit* edit) { + for (int which = 0; which < 2; which++) { + for (size_t i = 0; i < inputs_[which].size(); i++) { + edit->DeleteFile(level_ + which, inputs_[which][i]->number); + } + } +} + +bool Compaction::IsBaseLevelForKey(const Slice& user_key) { + // Maybe use binary search to find right entry instead of linear search? + const Comparator* user_cmp = input_version_->vset_->icmp_.user_comparator(); + for (int lvl = level_ + 2; lvl < config::kNumLevels; lvl++) { + const std::vector<FileMetaData*>& files = input_version_->files_[lvl]; + for (; level_ptrs_[lvl] < files.size(); ) { + FileMetaData* f = files[level_ptrs_[lvl]]; + if (user_cmp->Compare(user_key, f->largest.user_key()) <= 0) { + // We've advanced far enough + if (user_cmp->Compare(user_key, f->smallest.user_key()) >= 0) { + // Key falls in this file's range, so definitely not base level + return false; + } + break; + } + level_ptrs_[lvl]++; + } + } + return true; +} + +bool Compaction::ShouldStopBefore(const Slice& internal_key) { + // Scan to find earliest grandparent file that contains key. + const InternalKeyComparator* icmp = &input_version_->vset_->icmp_; + while (grandparent_index_ < grandparents_.size() && + icmp->Compare(internal_key, + grandparents_[grandparent_index_]->largest.Encode()) > 0) { + if (seen_key_) { + overlapped_bytes_ += grandparents_[grandparent_index_]->file_size; + } + grandparent_index_++; + } + seen_key_ = true; + + if (overlapped_bytes_ > kMaxGrandParentOverlapBytes) { + // Too much overlap for current output; start new output + overlapped_bytes_ = 0; + return true; + } else { + return false; + } +} + +void Compaction::ReleaseInputs() { + if (input_version_ != NULL) { + input_version_->Unref(); + input_version_ = NULL; + } +} + +} // namespace leveldb diff --git a/src/leveldb/db/version_set.h b/src/leveldb/db/version_set.h new file mode 100644 index 0000000000..1dec745673 --- /dev/null +++ b/src/leveldb/db/version_set.h @@ -0,0 +1,398 @@ +// 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. +// +// The representation of a DBImpl consists of a set of Versions. The +// newest version is called "current". Older versions may be kept +// around to provide a consistent view to live iterators. +// +// Each Version keeps track of a set of Table files per level. The +// entire set of versions is maintained in a VersionSet. +// +// Version,VersionSet are thread-compatible, but require external +// synchronization on all accesses. + +#ifndef STORAGE_LEVELDB_DB_VERSION_SET_H_ +#define STORAGE_LEVELDB_DB_VERSION_SET_H_ + +#include <map> +#include <set> +#include <vector> +#include "db/dbformat.h" +#include "db/version_edit.h" +#include "port/port.h" +#include "port/thread_annotations.h" + +namespace leveldb { + +namespace log { class Writer; } + +class Compaction; +class Iterator; +class MemTable; +class TableBuilder; +class TableCache; +class Version; +class VersionSet; +class WritableFile; + +// Return the smallest index i such that files[i]->largest >= key. +// Return files.size() if there is no such file. +// REQUIRES: "files" contains a sorted list of non-overlapping files. +extern int FindFile(const InternalKeyComparator& icmp, + const std::vector<FileMetaData*>& files, + const Slice& key); + +// Returns true iff some file in "files" overlaps the user key range +// [*smallest,*largest]. +// smallest==NULL represents a key smaller than all keys in the DB. +// largest==NULL represents a key largest than all keys in the DB. +// REQUIRES: If disjoint_sorted_files, files[] contains disjoint ranges +// in sorted order. +extern bool SomeFileOverlapsRange( + const InternalKeyComparator& icmp, + bool disjoint_sorted_files, + const std::vector<FileMetaData*>& files, + const Slice* smallest_user_key, + const Slice* largest_user_key); + +class Version { + public: + // Append to *iters a sequence of iterators that will + // yield the contents of this Version when merged together. + // REQUIRES: This version has been saved (see VersionSet::SaveTo) + void AddIterators(const ReadOptions&, std::vector<Iterator*>* iters); + + // Lookup the value for key. If found, store it in *val and + // return OK. Else return a non-OK status. Fills *stats. + // REQUIRES: lock is not held + struct GetStats { + FileMetaData* seek_file; + int seek_file_level; + }; + Status Get(const ReadOptions&, const LookupKey& key, std::string* val, + GetStats* stats); + + // Adds "stats" into the current state. Returns true if a new + // compaction may need to be triggered, false otherwise. + // REQUIRES: lock is held + bool UpdateStats(const GetStats& stats); + + // Record a sample of bytes read at the specified internal key. + // Samples are taken approximately once every config::kReadBytesPeriod + // bytes. Returns true if a new compaction may need to be triggered. + // REQUIRES: lock is held + bool RecordReadSample(Slice key); + + // Reference count management (so Versions do not disappear out from + // under live iterators) + void Ref(); + void Unref(); + + void GetOverlappingInputs( + int level, + const InternalKey* begin, // NULL means before all keys + const InternalKey* end, // NULL means after all keys + std::vector<FileMetaData*>* inputs); + + // Returns true iff some file in the specified level overlaps + // some part of [*smallest_user_key,*largest_user_key]. + // smallest_user_key==NULL represents a key smaller than all keys in the DB. + // largest_user_key==NULL represents a key largest than all keys in the DB. + bool OverlapInLevel(int level, + const Slice* smallest_user_key, + const Slice* largest_user_key); + + // Return the level at which we should place a new memtable compaction + // result that covers the range [smallest_user_key,largest_user_key]. + int PickLevelForMemTableOutput(const Slice& smallest_user_key, + const Slice& largest_user_key); + + int NumFiles(int level) const { return files_[level].size(); } + + // Return a human readable string that describes this version's contents. + std::string DebugString() const; + + private: + friend class Compaction; + friend class VersionSet; + + class LevelFileNumIterator; + Iterator* NewConcatenatingIterator(const ReadOptions&, int level) const; + + // Call func(arg, level, f) for every file that overlaps user_key in + // order from newest to oldest. If an invocation of func returns + // false, makes no more calls. + // + // REQUIRES: user portion of internal_key == user_key. + void ForEachOverlapping(Slice user_key, Slice internal_key, + void* arg, + bool (*func)(void*, int, FileMetaData*)); + + VersionSet* vset_; // VersionSet to which this Version belongs + Version* next_; // Next version in linked list + Version* prev_; // Previous version in linked list + int refs_; // Number of live refs to this version + + // List of files per level + std::vector<FileMetaData*> files_[config::kNumLevels]; + + // Next file to compact based on seek stats. + FileMetaData* file_to_compact_; + int file_to_compact_level_; + + // Level that should be compacted next and its compaction score. + // Score < 1 means compaction is not strictly needed. These fields + // are initialized by Finalize(). + double compaction_score_; + int compaction_level_; + + explicit Version(VersionSet* vset) + : vset_(vset), next_(this), prev_(this), refs_(0), + file_to_compact_(NULL), + file_to_compact_level_(-1), + compaction_score_(-1), + compaction_level_(-1) { + } + + ~Version(); + + // No copying allowed + Version(const Version&); + void operator=(const Version&); +}; + +class VersionSet { + public: + VersionSet(const std::string& dbname, + const Options* options, + TableCache* table_cache, + const InternalKeyComparator*); + ~VersionSet(); + + // Apply *edit to the current version to form a new descriptor that + // is both saved to persistent state and installed as the new + // current version. Will release *mu while actually writing to the file. + // REQUIRES: *mu is held on entry. + // REQUIRES: no other thread concurrently calls LogAndApply() + Status LogAndApply(VersionEdit* edit, port::Mutex* mu) + EXCLUSIVE_LOCKS_REQUIRED(mu); + + // Recover the last saved descriptor from persistent storage. + Status Recover(bool *save_manifest); + + // Return the current version. + Version* current() const { return current_; } + + // Return the current manifest file number + uint64_t ManifestFileNumber() const { return manifest_file_number_; } + + // Allocate and return a new file number + uint64_t NewFileNumber() { return next_file_number_++; } + + // Arrange to reuse "file_number" unless a newer file number has + // already been allocated. + // REQUIRES: "file_number" was returned by a call to NewFileNumber(). + void ReuseFileNumber(uint64_t file_number) { + if (next_file_number_ == file_number + 1) { + next_file_number_ = file_number; + } + } + + // Return the number of Table files at the specified level. + int NumLevelFiles(int level) const; + + // Return the combined file size of all files at the specified level. + int64_t NumLevelBytes(int level) const; + + // Return the last sequence number. + uint64_t LastSequence() const { return last_sequence_; } + + // Set the last sequence number to s. + void SetLastSequence(uint64_t s) { + assert(s >= last_sequence_); + last_sequence_ = s; + } + + // Mark the specified file number as used. + void MarkFileNumberUsed(uint64_t number); + + // Return the current log file number. + uint64_t LogNumber() const { return log_number_; } + + // Return the log file number for the log file that is currently + // being compacted, or zero if there is no such log file. + uint64_t PrevLogNumber() const { return prev_log_number_; } + + // Pick level and inputs for a new compaction. + // Returns NULL if there is no compaction to be done. + // Otherwise returns a pointer to a heap-allocated object that + // describes the compaction. Caller should delete the result. + Compaction* PickCompaction(); + + // Return a compaction object for compacting the range [begin,end] in + // the specified level. Returns NULL if there is nothing in that + // level that overlaps the specified range. Caller should delete + // the result. + Compaction* CompactRange( + int level, + const InternalKey* begin, + const InternalKey* end); + + // Return the maximum overlapping data (in bytes) at next level for any + // file at a level >= 1. + int64_t MaxNextLevelOverlappingBytes(); + + // Create an iterator that reads over the compaction inputs for "*c". + // The caller should delete the iterator when no longer needed. + Iterator* MakeInputIterator(Compaction* c); + + // Returns true iff some level needs a compaction. + bool NeedsCompaction() const { + Version* v = current_; + return (v->compaction_score_ >= 1) || (v->file_to_compact_ != NULL); + } + + // Add all files listed in any live version to *live. + // May also mutate some internal state. + void AddLiveFiles(std::set<uint64_t>* live); + + // Return the approximate offset in the database of the data for + // "key" as of version "v". + uint64_t ApproximateOffsetOf(Version* v, const InternalKey& key); + + // Return a human-readable short (single-line) summary of the number + // of files per level. Uses *scratch as backing store. + struct LevelSummaryStorage { + char buffer[100]; + }; + const char* LevelSummary(LevelSummaryStorage* scratch) const; + + private: + class Builder; + + friend class Compaction; + friend class Version; + + bool ReuseManifest(const std::string& dscname, const std::string& dscbase); + + void Finalize(Version* v); + + void GetRange(const std::vector<FileMetaData*>& inputs, + InternalKey* smallest, + InternalKey* largest); + + void GetRange2(const std::vector<FileMetaData*>& inputs1, + const std::vector<FileMetaData*>& inputs2, + InternalKey* smallest, + InternalKey* largest); + + void SetupOtherInputs(Compaction* c); + + // Save current contents to *log + Status WriteSnapshot(log::Writer* log); + + void AppendVersion(Version* v); + + Env* const env_; + const std::string dbname_; + const Options* const options_; + TableCache* const table_cache_; + const InternalKeyComparator icmp_; + uint64_t next_file_number_; + uint64_t manifest_file_number_; + uint64_t last_sequence_; + uint64_t log_number_; + uint64_t prev_log_number_; // 0 or backing store for memtable being compacted + + // Opened lazily + WritableFile* descriptor_file_; + log::Writer* descriptor_log_; + Version dummy_versions_; // Head of circular doubly-linked list of versions. + Version* current_; // == dummy_versions_.prev_ + + // Per-level key at which the next compaction at that level should start. + // Either an empty string, or a valid InternalKey. + std::string compact_pointer_[config::kNumLevels]; + + // No copying allowed + VersionSet(const VersionSet&); + void operator=(const VersionSet&); +}; + +// A Compaction encapsulates information about a compaction. +class Compaction { + public: + ~Compaction(); + + // Return the level that is being compacted. Inputs from "level" + // and "level+1" will be merged to produce a set of "level+1" files. + int level() const { return level_; } + + // Return the object that holds the edits to the descriptor done + // by this compaction. + VersionEdit* edit() { return &edit_; } + + // "which" must be either 0 or 1 + int num_input_files(int which) const { return inputs_[which].size(); } + + // Return the ith input file at "level()+which" ("which" must be 0 or 1). + FileMetaData* input(int which, int i) const { return inputs_[which][i]; } + + // Maximum size of files to build during this compaction. + uint64_t MaxOutputFileSize() const { return max_output_file_size_; } + + // Is this a trivial compaction that can be implemented by just + // moving a single input file to the next level (no merging or splitting) + bool IsTrivialMove() const; + + // Add all inputs to this compaction as delete operations to *edit. + void AddInputDeletions(VersionEdit* edit); + + // Returns true if the information we have available guarantees that + // the compaction is producing data in "level+1" for which no data exists + // in levels greater than "level+1". + bool IsBaseLevelForKey(const Slice& user_key); + + // Returns true iff we should stop building the current output + // before processing "internal_key". + bool ShouldStopBefore(const Slice& internal_key); + + // Release the input version for the compaction, once the compaction + // is successful. + void ReleaseInputs(); + + private: + friend class Version; + friend class VersionSet; + + explicit Compaction(int level); + + int level_; + uint64_t max_output_file_size_; + Version* input_version_; + VersionEdit edit_; + + // Each compaction reads inputs from "level_" and "level_+1" + std::vector<FileMetaData*> inputs_[2]; // The two sets of inputs + + // State used to check for number of of overlapping grandparent files + // (parent == level_ + 1, grandparent == level_ + 2) + std::vector<FileMetaData*> grandparents_; + size_t grandparent_index_; // Index in grandparent_starts_ + bool seen_key_; // Some output key has been seen + int64_t overlapped_bytes_; // Bytes of overlap between current output + // and grandparent files + + // State for implementing IsBaseLevelForKey + + // level_ptrs_ holds indices into input_version_->levels_: our state + // is that we are positioned at one of the file ranges for each + // higher level than the ones involved in this compaction (i.e. for + // all L >= level_ + 2). + size_t level_ptrs_[config::kNumLevels]; +}; + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_VERSION_SET_H_ diff --git a/src/leveldb/db/version_set_test.cc b/src/leveldb/db/version_set_test.cc new file mode 100644 index 0000000000..501e34d133 --- /dev/null +++ b/src/leveldb/db/version_set_test.cc @@ -0,0 +1,179 @@ +// 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/version_set.h" +#include "util/logging.h" +#include "util/testharness.h" +#include "util/testutil.h" + +namespace leveldb { + +class FindFileTest { + public: + std::vector<FileMetaData*> files_; + bool disjoint_sorted_files_; + + FindFileTest() : disjoint_sorted_files_(true) { } + + ~FindFileTest() { + for (int i = 0; i < files_.size(); i++) { + delete files_[i]; + } + } + + void Add(const char* smallest, const char* largest, + SequenceNumber smallest_seq = 100, + SequenceNumber largest_seq = 100) { + FileMetaData* f = new FileMetaData; + f->number = files_.size() + 1; + f->smallest = InternalKey(smallest, smallest_seq, kTypeValue); + f->largest = InternalKey(largest, largest_seq, kTypeValue); + files_.push_back(f); + } + + int Find(const char* key) { + InternalKey target(key, 100, kTypeValue); + InternalKeyComparator cmp(BytewiseComparator()); + return FindFile(cmp, files_, target.Encode()); + } + + bool Overlaps(const char* smallest, const char* largest) { + InternalKeyComparator cmp(BytewiseComparator()); + Slice s(smallest != NULL ? smallest : ""); + Slice l(largest != NULL ? largest : ""); + return SomeFileOverlapsRange(cmp, disjoint_sorted_files_, files_, + (smallest != NULL ? &s : NULL), + (largest != NULL ? &l : NULL)); + } +}; + +TEST(FindFileTest, Empty) { + ASSERT_EQ(0, Find("foo")); + ASSERT_TRUE(! Overlaps("a", "z")); + ASSERT_TRUE(! Overlaps(NULL, "z")); + ASSERT_TRUE(! Overlaps("a", NULL)); + ASSERT_TRUE(! Overlaps(NULL, NULL)); +} + +TEST(FindFileTest, Single) { + Add("p", "q"); + ASSERT_EQ(0, Find("a")); + ASSERT_EQ(0, Find("p")); + ASSERT_EQ(0, Find("p1")); + ASSERT_EQ(0, Find("q")); + ASSERT_EQ(1, Find("q1")); + ASSERT_EQ(1, Find("z")); + + ASSERT_TRUE(! Overlaps("a", "b")); + ASSERT_TRUE(! Overlaps("z1", "z2")); + ASSERT_TRUE(Overlaps("a", "p")); + ASSERT_TRUE(Overlaps("a", "q")); + ASSERT_TRUE(Overlaps("a", "z")); + ASSERT_TRUE(Overlaps("p", "p1")); + ASSERT_TRUE(Overlaps("p", "q")); + ASSERT_TRUE(Overlaps("p", "z")); + ASSERT_TRUE(Overlaps("p1", "p2")); + ASSERT_TRUE(Overlaps("p1", "z")); + ASSERT_TRUE(Overlaps("q", "q")); + ASSERT_TRUE(Overlaps("q", "q1")); + + ASSERT_TRUE(! Overlaps(NULL, "j")); + ASSERT_TRUE(! Overlaps("r", NULL)); + ASSERT_TRUE(Overlaps(NULL, "p")); + ASSERT_TRUE(Overlaps(NULL, "p1")); + ASSERT_TRUE(Overlaps("q", NULL)); + ASSERT_TRUE(Overlaps(NULL, NULL)); +} + + +TEST(FindFileTest, Multiple) { + Add("150", "200"); + Add("200", "250"); + Add("300", "350"); + Add("400", "450"); + ASSERT_EQ(0, Find("100")); + ASSERT_EQ(0, Find("150")); + ASSERT_EQ(0, Find("151")); + ASSERT_EQ(0, Find("199")); + ASSERT_EQ(0, Find("200")); + ASSERT_EQ(1, Find("201")); + ASSERT_EQ(1, Find("249")); + ASSERT_EQ(1, Find("250")); + ASSERT_EQ(2, Find("251")); + ASSERT_EQ(2, Find("299")); + ASSERT_EQ(2, Find("300")); + ASSERT_EQ(2, Find("349")); + ASSERT_EQ(2, Find("350")); + ASSERT_EQ(3, Find("351")); + ASSERT_EQ(3, Find("400")); + ASSERT_EQ(3, Find("450")); + ASSERT_EQ(4, Find("451")); + + ASSERT_TRUE(! Overlaps("100", "149")); + ASSERT_TRUE(! Overlaps("251", "299")); + ASSERT_TRUE(! Overlaps("451", "500")); + ASSERT_TRUE(! Overlaps("351", "399")); + + ASSERT_TRUE(Overlaps("100", "150")); + ASSERT_TRUE(Overlaps("100", "200")); + ASSERT_TRUE(Overlaps("100", "300")); + ASSERT_TRUE(Overlaps("100", "400")); + ASSERT_TRUE(Overlaps("100", "500")); + ASSERT_TRUE(Overlaps("375", "400")); + ASSERT_TRUE(Overlaps("450", "450")); + ASSERT_TRUE(Overlaps("450", "500")); +} + +TEST(FindFileTest, MultipleNullBoundaries) { + Add("150", "200"); + Add("200", "250"); + Add("300", "350"); + Add("400", "450"); + ASSERT_TRUE(! Overlaps(NULL, "149")); + ASSERT_TRUE(! Overlaps("451", NULL)); + ASSERT_TRUE(Overlaps(NULL, NULL)); + ASSERT_TRUE(Overlaps(NULL, "150")); + ASSERT_TRUE(Overlaps(NULL, "199")); + ASSERT_TRUE(Overlaps(NULL, "200")); + ASSERT_TRUE(Overlaps(NULL, "201")); + ASSERT_TRUE(Overlaps(NULL, "400")); + ASSERT_TRUE(Overlaps(NULL, "800")); + ASSERT_TRUE(Overlaps("100", NULL)); + ASSERT_TRUE(Overlaps("200", NULL)); + ASSERT_TRUE(Overlaps("449", NULL)); + ASSERT_TRUE(Overlaps("450", NULL)); +} + +TEST(FindFileTest, OverlapSequenceChecks) { + Add("200", "200", 5000, 3000); + ASSERT_TRUE(! Overlaps("199", "199")); + ASSERT_TRUE(! Overlaps("201", "300")); + ASSERT_TRUE(Overlaps("200", "200")); + ASSERT_TRUE(Overlaps("190", "200")); + ASSERT_TRUE(Overlaps("200", "210")); +} + +TEST(FindFileTest, OverlappingFiles) { + Add("150", "600"); + Add("400", "500"); + disjoint_sorted_files_ = false; + ASSERT_TRUE(! Overlaps("100", "149")); + ASSERT_TRUE(! Overlaps("601", "700")); + ASSERT_TRUE(Overlaps("100", "150")); + ASSERT_TRUE(Overlaps("100", "200")); + ASSERT_TRUE(Overlaps("100", "300")); + ASSERT_TRUE(Overlaps("100", "400")); + ASSERT_TRUE(Overlaps("100", "500")); + ASSERT_TRUE(Overlaps("375", "400")); + ASSERT_TRUE(Overlaps("450", "450")); + ASSERT_TRUE(Overlaps("450", "500")); + ASSERT_TRUE(Overlaps("450", "700")); + ASSERT_TRUE(Overlaps("600", "700")); +} + +} // namespace leveldb + +int main(int argc, char** argv) { + return leveldb::test::RunAllTests(); +} diff --git a/src/leveldb/db/write_batch.cc b/src/leveldb/db/write_batch.cc new file mode 100644 index 0000000000..33f4a4257e --- /dev/null +++ b/src/leveldb/db/write_batch.cc @@ -0,0 +1,147 @@ +// 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. +// +// WriteBatch::rep_ := +// sequence: fixed64 +// count: fixed32 +// data: record[count] +// record := +// kTypeValue varstring varstring | +// kTypeDeletion varstring +// varstring := +// len: varint32 +// data: uint8[len] + +#include "leveldb/write_batch.h" + +#include "leveldb/db.h" +#include "db/dbformat.h" +#include "db/memtable.h" +#include "db/write_batch_internal.h" +#include "util/coding.h" + +namespace leveldb { + +// WriteBatch header has an 8-byte sequence number followed by a 4-byte count. +static const size_t kHeader = 12; + +WriteBatch::WriteBatch() { + Clear(); +} + +WriteBatch::~WriteBatch() { } + +WriteBatch::Handler::~Handler() { } + +void WriteBatch::Clear() { + rep_.clear(); + rep_.resize(kHeader); +} + +Status WriteBatch::Iterate(Handler* handler) const { + Slice input(rep_); + if (input.size() < kHeader) { + return Status::Corruption("malformed WriteBatch (too small)"); + } + + input.remove_prefix(kHeader); + Slice key, value; + int found = 0; + while (!input.empty()) { + found++; + char tag = input[0]; + input.remove_prefix(1); + switch (tag) { + case kTypeValue: + if (GetLengthPrefixedSlice(&input, &key) && + GetLengthPrefixedSlice(&input, &value)) { + handler->Put(key, value); + } else { + return Status::Corruption("bad WriteBatch Put"); + } + break; + case kTypeDeletion: + if (GetLengthPrefixedSlice(&input, &key)) { + handler->Delete(key); + } else { + return Status::Corruption("bad WriteBatch Delete"); + } + break; + default: + return Status::Corruption("unknown WriteBatch tag"); + } + } + if (found != WriteBatchInternal::Count(this)) { + return Status::Corruption("WriteBatch has wrong count"); + } else { + return Status::OK(); + } +} + +int WriteBatchInternal::Count(const WriteBatch* b) { + return DecodeFixed32(b->rep_.data() + 8); +} + +void WriteBatchInternal::SetCount(WriteBatch* b, int n) { + EncodeFixed32(&b->rep_[8], n); +} + +SequenceNumber WriteBatchInternal::Sequence(const WriteBatch* b) { + return SequenceNumber(DecodeFixed64(b->rep_.data())); +} + +void WriteBatchInternal::SetSequence(WriteBatch* b, SequenceNumber seq) { + EncodeFixed64(&b->rep_[0], seq); +} + +void WriteBatch::Put(const Slice& key, const Slice& value) { + WriteBatchInternal::SetCount(this, WriteBatchInternal::Count(this) + 1); + rep_.push_back(static_cast<char>(kTypeValue)); + PutLengthPrefixedSlice(&rep_, key); + PutLengthPrefixedSlice(&rep_, value); +} + +void WriteBatch::Delete(const Slice& key) { + WriteBatchInternal::SetCount(this, WriteBatchInternal::Count(this) + 1); + rep_.push_back(static_cast<char>(kTypeDeletion)); + PutLengthPrefixedSlice(&rep_, key); +} + +namespace { +class MemTableInserter : public WriteBatch::Handler { + public: + SequenceNumber sequence_; + MemTable* mem_; + + virtual void Put(const Slice& key, const Slice& value) { + mem_->Add(sequence_, kTypeValue, key, value); + sequence_++; + } + virtual void Delete(const Slice& key) { + mem_->Add(sequence_, kTypeDeletion, key, Slice()); + sequence_++; + } +}; +} // namespace + +Status WriteBatchInternal::InsertInto(const WriteBatch* b, + MemTable* memtable) { + MemTableInserter inserter; + inserter.sequence_ = WriteBatchInternal::Sequence(b); + inserter.mem_ = memtable; + return b->Iterate(&inserter); +} + +void WriteBatchInternal::SetContents(WriteBatch* b, const Slice& contents) { + assert(contents.size() >= kHeader); + b->rep_.assign(contents.data(), contents.size()); +} + +void WriteBatchInternal::Append(WriteBatch* dst, const WriteBatch* src) { + SetCount(dst, Count(dst) + Count(src)); + assert(src->rep_.size() >= kHeader); + dst->rep_.append(src->rep_.data() + kHeader, src->rep_.size() - kHeader); +} + +} // namespace leveldb diff --git a/src/leveldb/db/write_batch_internal.h b/src/leveldb/db/write_batch_internal.h new file mode 100644 index 0000000000..9448ef7b21 --- /dev/null +++ b/src/leveldb/db/write_batch_internal.h @@ -0,0 +1,50 @@ +// 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_WRITE_BATCH_INTERNAL_H_ +#define STORAGE_LEVELDB_DB_WRITE_BATCH_INTERNAL_H_ + +#include "db/dbformat.h" +#include "leveldb/write_batch.h" + +namespace leveldb { + +class MemTable; + +// WriteBatchInternal provides static methods for manipulating a +// WriteBatch that we don't want in the public WriteBatch interface. +class WriteBatchInternal { + public: + // Return the number of entries in the batch. + static int Count(const WriteBatch* batch); + + // Set the count for the number of entries in the batch. + static void SetCount(WriteBatch* batch, int n); + + // Return the sequence number for the start of this batch. + static SequenceNumber Sequence(const WriteBatch* batch); + + // Store the specified number as the sequence number for the start of + // this batch. + static void SetSequence(WriteBatch* batch, SequenceNumber seq); + + static Slice Contents(const WriteBatch* batch) { + return Slice(batch->rep_); + } + + static size_t ByteSize(const WriteBatch* batch) { + return batch->rep_.size(); + } + + static void SetContents(WriteBatch* batch, const Slice& contents); + + static Status InsertInto(const WriteBatch* batch, MemTable* memtable); + + static void Append(WriteBatch* dst, const WriteBatch* src); +}; + +} // namespace leveldb + + +#endif // STORAGE_LEVELDB_DB_WRITE_BATCH_INTERNAL_H_ diff --git a/src/leveldb/db/write_batch_test.cc b/src/leveldb/db/write_batch_test.cc new file mode 100644 index 0000000000..9064e3d85e --- /dev/null +++ b/src/leveldb/db/write_batch_test.cc @@ -0,0 +1,120 @@ +// 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 "leveldb/db.h" + +#include "db/memtable.h" +#include "db/write_batch_internal.h" +#include "leveldb/env.h" +#include "util/logging.h" +#include "util/testharness.h" + +namespace leveldb { + +static std::string PrintContents(WriteBatch* b) { + InternalKeyComparator cmp(BytewiseComparator()); + MemTable* mem = new MemTable(cmp); + mem->Ref(); + std::string state; + Status s = WriteBatchInternal::InsertInto(b, mem); + int count = 0; + Iterator* iter = mem->NewIterator(); + for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { + ParsedInternalKey ikey; + ASSERT_TRUE(ParseInternalKey(iter->key(), &ikey)); + switch (ikey.type) { + case kTypeValue: + state.append("Put("); + state.append(ikey.user_key.ToString()); + state.append(", "); + state.append(iter->value().ToString()); + state.append(")"); + count++; + break; + case kTypeDeletion: + state.append("Delete("); + state.append(ikey.user_key.ToString()); + state.append(")"); + count++; + break; + } + state.append("@"); + state.append(NumberToString(ikey.sequence)); + } + delete iter; + if (!s.ok()) { + state.append("ParseError()"); + } else if (count != WriteBatchInternal::Count(b)) { + state.append("CountMismatch()"); + } + mem->Unref(); + return state; +} + +class WriteBatchTest { }; + +TEST(WriteBatchTest, Empty) { + WriteBatch batch; + ASSERT_EQ("", PrintContents(&batch)); + ASSERT_EQ(0, WriteBatchInternal::Count(&batch)); +} + +TEST(WriteBatchTest, Multiple) { + WriteBatch batch; + batch.Put(Slice("foo"), Slice("bar")); + batch.Delete(Slice("box")); + batch.Put(Slice("baz"), Slice("boo")); + WriteBatchInternal::SetSequence(&batch, 100); + ASSERT_EQ(100, WriteBatchInternal::Sequence(&batch)); + ASSERT_EQ(3, WriteBatchInternal::Count(&batch)); + ASSERT_EQ("Put(baz, boo)@102" + "Delete(box)@101" + "Put(foo, bar)@100", + PrintContents(&batch)); +} + +TEST(WriteBatchTest, Corruption) { + WriteBatch batch; + batch.Put(Slice("foo"), Slice("bar")); + batch.Delete(Slice("box")); + WriteBatchInternal::SetSequence(&batch, 200); + Slice contents = WriteBatchInternal::Contents(&batch); + WriteBatchInternal::SetContents(&batch, + Slice(contents.data(),contents.size()-1)); + ASSERT_EQ("Put(foo, bar)@200" + "ParseError()", + PrintContents(&batch)); +} + +TEST(WriteBatchTest, Append) { + WriteBatch b1, b2; + WriteBatchInternal::SetSequence(&b1, 200); + WriteBatchInternal::SetSequence(&b2, 300); + WriteBatchInternal::Append(&b1, &b2); + ASSERT_EQ("", + PrintContents(&b1)); + b2.Put("a", "va"); + WriteBatchInternal::Append(&b1, &b2); + ASSERT_EQ("Put(a, va)@200", + PrintContents(&b1)); + b2.Clear(); + b2.Put("b", "vb"); + WriteBatchInternal::Append(&b1, &b2); + ASSERT_EQ("Put(a, va)@200" + "Put(b, vb)@201", + PrintContents(&b1)); + b2.Delete("foo"); + WriteBatchInternal::Append(&b1, &b2); + ASSERT_EQ("Put(a, va)@200" + "Put(b, vb)@202" + "Put(b, vb)@201" + "Delete(foo)@203", + PrintContents(&b1)); +} + +} // namespace leveldb + +int main(int argc, char** argv) { + return leveldb::test::RunAllTests(); +} |