[jira] [Commented] (MESOS-1432) Atomically set close-on-exec where possible.

["Till Toenshoff (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/MESOS-1432?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14011886#comment-14011886 ] 

Till Toenshoff commented on MESOS-1432:
---------------------------------------

Some pretty helpful debug code on the general subject of FD setup and possible leakage due to missing cloexec's:

{noformat}
void showFDInfo(int fd)
{
  using std::cerr;
  using std::endl;

  char buf[256];

  int fd_flags = fcntl(fd, F_GETFD);
  if (fd_flags == -1) return;

  int fl_flags = fcntl(fd, F_GETFL);
  if (fl_flags == -1) return;

  char path[256];
  sprintf(path, "/proc/self/fd/%d", fd);

  memset(&buf[0], 0, 256);
  ssize_t s = readlink( path, &buf[0], 256 );
  if (s == -1)
  {
    cerr << " (" << path << "): " << "not available" << endl;
    return;
  }
  cerr << fd << " (" << buf << "): ";

  if (fd_flags & FD_CLOEXEC)  cerr << "cloexec ";

  // file status
  if (fl_flags & O_APPEND  )  cerr << "append ";
  if (fl_flags & O_NONBLOCK)  cerr << "nonblock ";

  // acc mode
  if (fl_flags & O_RDONLY  )  cerr << "read-only ";
  if (fl_flags & O_RDWR    )  cerr << "read-write ";
  if (fl_flags & O_WRONLY  )  cerr << "write-only ";

  if (fl_flags & O_DSYNC   )  cerr << "dsync ";
  if (fl_flags & O_RSYNC   )  cerr << "rsync ";
  if (fl_flags & O_SYNC    )  cerr << "sync ";

  struct flock fl;
  fl.l_type = F_WRLCK;
  fl.l_whence = 0;
  fl.l_start = 0;
  fl.l_len = 0;
  fcntl( fd, F_GETLK, &fl );
  if (fl.l_type != F_UNLCK)
  {
    if (fl.l_type == F_WRLCK)
      cerr << "write-locked";
    else
      cerr << "read-locked";
    cerr << "(pid:" << fl.l_pid << ") ";
  }
  cerr << endl;
}


void showFDInfo()
{
   int numHandles = getdtablesize();

   for (int i = 0; i < numHandles; i++)
   {
      int fd_flags = fcntl(i, F_GETFD);
      if (fd_flags == -1)
        continue;

      showFDInfo(i);
   }
}
{noformat}

Works on linux only and is taken from http://oroboro.com/file-handle-leaks-server/


> Atomically set close-on-exec where possible.
> --------------------------------------------
>
>                 Key: MESOS-1432
>                 URL: https://issues.apache.org/jira/browse/MESOS-1432
>             Project: Mesos
>          Issue Type: Bug
>            Reporter: Benjamin Mahler
>
> In our multi-threaded code, there is a race between opening a file descriptor and setting close-on-exec on it: if a fork occurs between these operations then we have leaked the file descriptor to the child process.
> Some required reading:
> http://udrepper.livejournal.com/20407.html
> We should do an audit of all of our code to ensure we're doing avoiding this race where possible.



--
This message was sent by Atlassian JIRA
(v6.2#6252)