linux - fs io 的用户级反弹缓冲区？为什么写文件缓冲区地址空间要对齐？

Question

我看到一段代码像这样执行磁盘io:

static bool is_aligned(unsigned char *buffer) {
  return ( ((unsigned long)buffer) & (DISK_PAGE_SIZE -1)) == 0;
}


void do_write_IO(int fd, unsignced char *buffer, unsigned long buffer_bytes) {
  ...

  if (is_aligned(buffer)) {
    write_to_file(fd, buffer, io_size);
  } else {
    // bounce buffer is an aligned memory space.
    // if buffer not aligned, copy to an aligned address
    // and do fs write. WHY ?
    // What are the benefits ? 
    memcpy(bounce_buffer, buffer, io_size);
    write_to_file(fd, bounce_buffer, io_size);
  }

  ...
}

// Just call posix write, write output to fd, with bytes_to_write size.
static void write_to_file(int fd, 
          unsigned char *output,
          unsigned long bytes_to_write) 
{
  __sync_fetch_and_add(&stat_bytes_written, bytes_to_write);
  while(bytes_to_write) {
    unsigned long bytes_written = write(fd, output, bytes_to_write);
    if(bytes_written == -1UL) {
      if(errno != EAGAIN) {
        BOOST_LOG_TRIVIAL(fatal) << 
          "Stream writeout unsuccessful:" << strerror(errno);
        exit(-1);
      }
    }
    else {
      output += bytes_written;
      bytes_to_write -= bytes_written;
    }
  }
}

我注意到它使用地址空间对齐的缓冲区写入文件。 那么，写入具有对齐缓冲区的文件有什么好处吗？

Answer 1

我认为“正常工作”是一种好处。为此，您需要满足调用的每个函数的先决条件。

如果 write 需要对齐缓冲区，则

write_to_file 需要对齐缓冲区，并且这是与设备相关的。可能您的fd是用O_DIRECT打开的。 open 手册页显示:

The O_DIRECT flag may impose alignment restrictions on the length and address of user-space buffers and the file offset of I/Os.

和

Under Linux 2.4, transfer sizes, and the alignment of the user buffer and the file offset must all be multiples of the logical block size of the file system. Under Linux 2.6, alignment to 512-byte boundaries suffices.

如果您不遵守此前提条件，write 手册页会记录错误:

EINVAL fd is attached to an object which is unsuitable for writing; or the file was opened with the O_DIRECT flag, and either the address specified in buf, the value specified in count, or the current file offset is not suitably aligned.