env_posix.cc 14.2 KB
Newer Older
J
jorlow@chromium.org 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
// 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 <deque>
#include <dirent.h>
#include <errno.h>
#include <fcntl.h>
#include <pthread.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/mman.h>
#include <sys/stat.h>
#include <sys/time.h>
#include <sys/types.h>
#include <time.h>
#include <unistd.h>
#if defined(LEVELDB_PLATFORM_ANDROID)
#include <sys/stat.h>
#endif
22 23
#include "leveldb/env.h"
#include "leveldb/slice.h"
J
jorlow@chromium.org 已提交
24 25
#include "port/port.h"
#include "util/logging.h"
26
#include "util/posix_logger.h"
J
jorlow@chromium.org 已提交
27 28 29 30 31

namespace leveldb {

namespace {

32 33 34 35
static Status IOError(const std::string& context, int err_number) {
  return Status::IOError(context, strerror(err_number));
}

J
jorlow@chromium.org 已提交
36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54
class PosixSequentialFile: public SequentialFile {
 private:
  std::string filename_;
  FILE* file_;

 public:
  PosixSequentialFile(const std::string& fname, FILE* f)
      : filename_(fname), file_(f) { }
  virtual ~PosixSequentialFile() { fclose(file_); }

  virtual Status Read(size_t n, Slice* result, char* scratch) {
    Status s;
    size_t r = fread_unlocked(scratch, 1, n, file_);
    *result = Slice(scratch, r);
    if (r < n) {
      if (feof(file_)) {
        // We leave status as ok if we hit the end of the file
      } else {
        // A partial read with an error: return a non-ok status
55
        s = IOError(filename_, errno);
J
jorlow@chromium.org 已提交
56 57 58 59
      }
    }
    return s;
  }
60 61 62

  virtual Status Skip(uint64_t n) {
    if (fseek(file_, n, SEEK_CUR)) {
63
      return IOError(filename_, errno);
64 65 66
    }
    return Status::OK();
  }
J
jorlow@chromium.org 已提交
67 68 69 70 71 72 73 74
};

class PosixRandomAccessFile: public RandomAccessFile {
 private:
  std::string filename_;
  int fd_;

 public:
J
jorlow@chromium.org 已提交
75 76
  PosixRandomAccessFile(const std::string& fname, int fd)
      : filename_(fname), fd_(fd) { }
J
jorlow@chromium.org 已提交
77 78 79 80 81 82 83 84 85
  virtual ~PosixRandomAccessFile() { close(fd_); }

  virtual Status Read(uint64_t offset, size_t n, Slice* result,
                      char* scratch) const {
    Status s;
    ssize_t r = pread(fd_, scratch, n, static_cast<off_t>(offset));
    *result = Slice(scratch, (r < 0) ? 0 : r);
    if (r < 0) {
      // An error: return a non-ok status
86
      s = IOError(filename_, errno);
J
jorlow@chromium.org 已提交
87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121
    }
    return s;
  }
};

// We preallocate up to an extra megabyte and use memcpy to append new
// data to the file.  This is safe since we either properly close the
// file before reading from it, or for log files, the reading code
// knows enough to skip zero suffixes.
class PosixMmapFile : public WritableFile {
 private:
  std::string filename_;
  int fd_;
  size_t page_size_;
  size_t map_size_;       // How much extra memory to map at a time
  char* base_;            // The mapped region
  char* limit_;           // Limit of the mapped region
  char* dst_;             // Where to write next  (in range [base_,limit_])
  char* last_sync_;       // Where have we synced up to
  uint64_t file_offset_;  // Offset of base_ in file

  // Have we done an munmap of unsynced data?
  bool pending_sync_;

  // Roundup x to a multiple of y
  static size_t Roundup(size_t x, size_t y) {
    return ((x + y - 1) / y) * y;
  }

  size_t TruncateToPageBoundary(size_t s) {
    s -= (s & (page_size_ - 1));
    assert((s % page_size_) == 0);
    return s;
  }

122 123
  bool UnmapCurrentRegion() {
    bool result = true;
J
jorlow@chromium.org 已提交
124 125 126 127 128
    if (base_ != NULL) {
      if (last_sync_ < limit_) {
        // Defer syncing this data until next Sync() call, if any
        pending_sync_ = true;
      }
129 130 131
      if (munmap(base_, limit_ - base_) != 0) {
        result = false;
      }
J
jorlow@chromium.org 已提交
132 133 134 135 136 137 138 139 140 141 142
      file_offset_ += limit_ - base_;
      base_ = NULL;
      limit_ = NULL;
      last_sync_ = NULL;
      dst_ = NULL;

      // Increase the amount we map the next time, but capped at 1MB
      if (map_size_ < (1<<20)) {
        map_size_ *= 2;
      }
    }
143
    return result;
J
jorlow@chromium.org 已提交
144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192
  }

  bool MapNewRegion() {
    assert(base_ == NULL);
    if (ftruncate(fd_, file_offset_ + map_size_) < 0) {
      return false;
    }
    void* ptr = mmap(NULL, map_size_, PROT_READ | PROT_WRITE, MAP_SHARED,
                     fd_, file_offset_);
    if (ptr == MAP_FAILED) {
      return false;
    }
    base_ = reinterpret_cast<char*>(ptr);
    limit_ = base_ + map_size_;
    dst_ = base_;
    last_sync_ = base_;
    return true;
  }

 public:
  PosixMmapFile(const std::string& fname, int fd, size_t page_size)
      : filename_(fname),
        fd_(fd),
        page_size_(page_size),
        map_size_(Roundup(65536, page_size)),
        base_(NULL),
        limit_(NULL),
        dst_(NULL),
        last_sync_(NULL),
        file_offset_(0),
        pending_sync_(false) {
    assert((page_size & (page_size - 1)) == 0);
  }


  ~PosixMmapFile() {
    if (fd_ >= 0) {
      PosixMmapFile::Close();
    }
  }

  virtual Status Append(const Slice& data) {
    const char* src = data.data();
    size_t left = data.size();
    while (left > 0) {
      assert(base_ <= dst_);
      assert(dst_ <= limit_);
      size_t avail = limit_ - dst_;
      if (avail == 0) {
193 194 195 196
        if (!UnmapCurrentRegion() ||
            !MapNewRegion()) {
          return IOError(filename_, errno);
        }
J
jorlow@chromium.org 已提交
197 198 199 200 201 202 203 204 205 206 207 208 209 210
      }

      size_t n = (left <= avail) ? left : avail;
      memcpy(dst_, src, n);
      dst_ += n;
      src += n;
      left -= n;
    }
    return Status::OK();
  }

  virtual Status Close() {
    Status s;
    size_t unused = limit_ - dst_;
211 212 213
    if (!UnmapCurrentRegion()) {
      s = IOError(filename_, errno);
    } else if (unused > 0) {
J
jorlow@chromium.org 已提交
214 215
      // Trim the extra space at the end of the file
      if (ftruncate(fd_, file_offset_ - unused) < 0) {
216
        s = IOError(filename_, errno);
J
jorlow@chromium.org 已提交
217 218 219 220 221
      }
    }

    if (close(fd_) < 0) {
      if (s.ok()) {
222
        s = IOError(filename_, errno);
J
jorlow@chromium.org 已提交
223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242
      }
    }

    fd_ = -1;
    base_ = NULL;
    limit_ = NULL;
    return s;
  }

  virtual Status Flush() {
    return Status::OK();
  }

  virtual Status Sync() {
    Status s;

    if (pending_sync_) {
      // Some unmapped data was not synced
      pending_sync_ = false;
      if (fdatasync(fd_) < 0) {
243
        s = IOError(filename_, errno);
J
jorlow@chromium.org 已提交
244 245 246 247 248 249 250 251 252 253
      }
    }

    if (dst_ > last_sync_) {
      // Find the beginnings of the pages that contain the first and last
      // bytes to be synced.
      size_t p1 = TruncateToPageBoundary(last_sync_ - base_);
      size_t p2 = TruncateToPageBoundary(dst_ - base_ - 1);
      last_sync_ = dst_;
      if (msync(base_ + p1, p2 - p1 + page_size_, MS_SYNC) < 0) {
254
        s = IOError(filename_, errno);
J
jorlow@chromium.org 已提交
255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290
      }
    }

    return s;
  }
};

static int LockOrUnlock(int fd, bool lock) {
  errno = 0;
  struct flock f;
  memset(&f, 0, sizeof(f));
  f.l_type = (lock ? F_WRLCK : F_UNLCK);
  f.l_whence = SEEK_SET;
  f.l_start = 0;
  f.l_len = 0;        // Lock/unlock entire file
  return fcntl(fd, F_SETLK, &f);
}

class PosixFileLock : public FileLock {
 public:
  int fd_;
};

class PosixEnv : public Env {
 public:
  PosixEnv();
  virtual ~PosixEnv() {
    fprintf(stderr, "Destroying Env::Default()\n");
    exit(1);
  }

  virtual Status NewSequentialFile(const std::string& fname,
                                   SequentialFile** result) {
    FILE* f = fopen(fname.c_str(), "r");
    if (f == NULL) {
      *result = NULL;
291
      return IOError(fname, errno);
J
jorlow@chromium.org 已提交
292 293 294 295 296 297 298 299 300 301 302
    } else {
      *result = new PosixSequentialFile(fname, f);
      return Status::OK();
    }
  }

  virtual Status NewRandomAccessFile(const std::string& fname,
                                     RandomAccessFile** result) {
    int fd = open(fname.c_str(), O_RDONLY);
    if (fd < 0) {
      *result = NULL;
303
      return IOError(fname, errno);
J
jorlow@chromium.org 已提交
304
    }
J
jorlow@chromium.org 已提交
305
    *result = new PosixRandomAccessFile(fname, fd);
J
jorlow@chromium.org 已提交
306 307 308 309 310 311 312 313 314
    return Status::OK();
  }

  virtual Status NewWritableFile(const std::string& fname,
                                 WritableFile** result) {
    Status s;
    const int fd = open(fname.c_str(), O_CREAT | O_RDWR | O_TRUNC, 0644);
    if (fd < 0) {
      *result = NULL;
315
      s = IOError(fname, errno);
J
jorlow@chromium.org 已提交
316 317 318 319 320 321 322 323 324 325 326 327 328 329 330
    } else {
      *result = new PosixMmapFile(fname, fd, page_size_);
    }
    return s;
  }

  virtual bool FileExists(const std::string& fname) {
    return access(fname.c_str(), F_OK) == 0;
  }

  virtual Status GetChildren(const std::string& dir,
                             std::vector<std::string>* result) {
    result->clear();
    DIR* d = opendir(dir.c_str());
    if (d == NULL) {
331
      return IOError(dir, errno);
J
jorlow@chromium.org 已提交
332 333 334 335 336 337 338 339 340 341 342 343
    }
    struct dirent* entry;
    while ((entry = readdir(d)) != NULL) {
      result->push_back(entry->d_name);
    }
    closedir(d);
    return Status::OK();
  }

  virtual Status DeleteFile(const std::string& fname) {
    Status result;
    if (unlink(fname.c_str()) != 0) {
344
      result = IOError(fname, errno);
J
jorlow@chromium.org 已提交
345 346 347 348 349 350 351
    }
    return result;
  };

  virtual Status CreateDir(const std::string& name) {
    Status result;
    if (mkdir(name.c_str(), 0755) != 0) {
352
      result = IOError(name, errno);
J
jorlow@chromium.org 已提交
353 354 355 356 357 358 359
    }
    return result;
  };

  virtual Status DeleteDir(const std::string& name) {
    Status result;
    if (rmdir(name.c_str()) != 0) {
360
      result = IOError(name, errno);
J
jorlow@chromium.org 已提交
361 362 363 364 365 366 367 368 369
    }
    return result;
  };

  virtual Status GetFileSize(const std::string& fname, uint64_t* size) {
    Status s;
    struct stat sbuf;
    if (stat(fname.c_str(), &sbuf) != 0) {
      *size = 0;
370
      s = IOError(fname, errno);
J
jorlow@chromium.org 已提交
371 372 373 374 375 376 377 378 379
    } else {
      *size = sbuf.st_size;
    }
    return s;
  }

  virtual Status RenameFile(const std::string& src, const std::string& target) {
    Status result;
    if (rename(src.c_str(), target.c_str()) != 0) {
380
      result = IOError(src, errno);
J
jorlow@chromium.org 已提交
381 382 383 384 385 386 387 388 389
    }
    return result;
  }

  virtual Status LockFile(const std::string& fname, FileLock** lock) {
    *lock = NULL;
    Status result;
    int fd = open(fname.c_str(), O_RDWR | O_CREAT, 0644);
    if (fd < 0) {
390
      result = IOError(fname, errno);
J
jorlow@chromium.org 已提交
391
    } else if (LockOrUnlock(fd, true) == -1) {
392
      result = IOError("lock " + fname, errno);
J
jorlow@chromium.org 已提交
393 394 395 396 397 398 399 400 401 402 403 404 405
      close(fd);
    } else {
      PosixFileLock* my_lock = new PosixFileLock;
      my_lock->fd_ = fd;
      *lock = my_lock;
    }
    return result;
  }

  virtual Status UnlockFile(FileLock* lock) {
    PosixFileLock* my_lock = reinterpret_cast<PosixFileLock*>(lock);
    Status result;
    if (LockOrUnlock(my_lock->fd_, false) == -1) {
406
      result = IOError("unlock", errno);
J
jorlow@chromium.org 已提交
407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430
    }
    close(my_lock->fd_);
    delete my_lock;
    return result;
  }

  virtual void Schedule(void (*function)(void*), void* arg);

  virtual void StartThread(void (*function)(void* arg), void* arg);

  virtual Status GetTestDirectory(std::string* result) {
    const char* env = getenv("TEST_TMPDIR");
    if (env && env[0] != '\0') {
      *result = env;
    } else {
      char buf[100];
      snprintf(buf, sizeof(buf), "/tmp/leveldbtest-%d", int(geteuid()));
      *result = buf;
    }
    // Directory may already exist
    CreateDir(*result);
    return Status::OK();
  }

431
  static uint64_t gettid() {
J
jorlow@chromium.org 已提交
432 433
    pthread_t tid = pthread_self();
    uint64_t thread_id = 0;
J
jorlow@chromium.org 已提交
434
    memcpy(&thread_id, &tid, std::min(sizeof(thread_id), sizeof(tid)));
435 436
    return thread_id;
  }
J
jorlow@chromium.org 已提交
437

438 439 440 441 442 443 444 445
  virtual Status NewLogger(const std::string& fname, Logger** result) {
    FILE* f = fopen(fname.c_str(), "w");
    if (f == NULL) {
      *result = NULL;
      return IOError(fname, errno);
    } else {
      *result = new PosixLogger(f, &PosixEnv::gettid);
      return Status::OK();
J
jorlow@chromium.org 已提交
446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555
    }
  }

  virtual uint64_t NowMicros() {
    struct timeval tv;
    gettimeofday(&tv, NULL);
    return static_cast<uint64_t>(tv.tv_sec) * 1000000 + tv.tv_usec;
  }

  virtual void SleepForMicroseconds(int micros) {
    usleep(micros);
  }

 private:
  void PthreadCall(const char* label, int result) {
    if (result != 0) {
      fprintf(stderr, "pthread %s: %s\n", label, strerror(result));
      exit(1);
    }
  }

  // BGThread() is the body of the background thread
  void BGThread();
  static void* BGThreadWrapper(void* arg) {
    reinterpret_cast<PosixEnv*>(arg)->BGThread();
    return NULL;
  }

  size_t page_size_;
  pthread_mutex_t mu_;
  pthread_cond_t bgsignal_;
  pthread_t bgthread_;
  bool started_bgthread_;

  // Entry per Schedule() call
  struct BGItem { void* arg; void (*function)(void*); };
  typedef std::deque<BGItem> BGQueue;
  BGQueue queue_;
};

PosixEnv::PosixEnv() : page_size_(getpagesize()),
                       started_bgthread_(false) {
  PthreadCall("mutex_init", pthread_mutex_init(&mu_, NULL));
  PthreadCall("cvar_init", pthread_cond_init(&bgsignal_, NULL));
}

void PosixEnv::Schedule(void (*function)(void*), void* arg) {
  PthreadCall("lock", pthread_mutex_lock(&mu_));

  // Start background thread if necessary
  if (!started_bgthread_) {
    started_bgthread_ = true;
    PthreadCall(
        "create thread",
        pthread_create(&bgthread_, NULL,  &PosixEnv::BGThreadWrapper, this));
  }

  // If the queue is currently empty, the background thread may currently be
  // waiting.
  if (queue_.empty()) {
    PthreadCall("signal", pthread_cond_signal(&bgsignal_));
  }

  // Add to priority queue
  queue_.push_back(BGItem());
  queue_.back().function = function;
  queue_.back().arg = arg;

  PthreadCall("unlock", pthread_mutex_unlock(&mu_));
}

void PosixEnv::BGThread() {
  while (true) {
    // Wait until there is an item that is ready to run
    PthreadCall("lock", pthread_mutex_lock(&mu_));
    while (queue_.empty()) {
      PthreadCall("wait", pthread_cond_wait(&bgsignal_, &mu_));
    }

    void (*function)(void*) = queue_.front().function;
    void* arg = queue_.front().arg;
    queue_.pop_front();

    PthreadCall("unlock", pthread_mutex_unlock(&mu_));
    (*function)(arg);
  }
}

namespace {
struct StartThreadState {
  void (*user_function)(void*);
  void* arg;
};
}
static void* StartThreadWrapper(void* arg) {
  StartThreadState* state = reinterpret_cast<StartThreadState*>(arg);
  state->user_function(state->arg);
  delete state;
  return NULL;
}

void PosixEnv::StartThread(void (*function)(void* arg), void* arg) {
  pthread_t t;
  StartThreadState* state = new StartThreadState;
  state->user_function = function;
  state->arg = arg;
  PthreadCall("start thread",
              pthread_create(&t, NULL,  &StartThreadWrapper, state));
}

H
Hans Wennborg 已提交
556
}  // namespace
J
jorlow@chromium.org 已提交
557 558 559 560 561 562 563 564 565 566

static pthread_once_t once = PTHREAD_ONCE_INIT;
static Env* default_env;
static void InitDefaultEnv() { default_env = new PosixEnv; }

Env* Env::Default() {
  pthread_once(&once, InitDefaultEnv);
  return default_env;
}

H
Hans Wennborg 已提交
567
}  // namespace leveldb