Re: [PATCH 2/2] fs/lock: show locks taken by processes from another pidns

[Andrey Vagin]

On Thu, Jun 14, 2018 at 07:00:07AM -0400, Jeff Layton wrote:
> On Fri, 2018-06-08 at 17:27 +0300, Konstantin Khorenko wrote:
> > Currently if we face a lock taken by a process invisible in the current
> > pidns we skip the lock completely, but this
> > 
> > 1) makes the output not that nice
> >     (root@vz7)/: cat /proc/${PID_A2}/fdinfo/3
> >     pos:    4
> >     flags:  02100002
> >     mnt_id: 257
> >     lock:   (root@vz7)/:
> > 
> > 2) makes it more difficult to debug issues with leaked flocks
> >    if you get error on lock, but don't see any locks in /proc/$id/fdinfo/$file
> > 
> > Let's show information about such locks again as previously, but
> > show zero in the owner pid field.
> > 
> > After the patch:
> > ===============
> > (root@vz7)/:cat /proc/${PID_A2}/fdinfo/3
> > pos:    4
> > flags:  02100002
> > mnt_id: 295
> > lock:   1: FLOCK  ADVISORY  WRITE 0 b6:f8a61:529946 0 EOF
> > 
> > Fixes: 9d5b86ac13c5 ("fs/locks: Remove fl_nspid and use fs-specific l_pid for remote locks")
> > Signed-off-by: Konstantin Khorenko <khorenko@xxxxxxxxxxxxx>
> > ---
> >  fs/locks.c | 8 +++-----
> >  1 file changed, 3 insertions(+), 5 deletions(-)
> > 
> > diff --git a/fs/locks.c b/fs/locks.c
> > index bfee5b7f2862..e533623e2e99 100644
> > --- a/fs/locks.c
> > +++ b/fs/locks.c
> > @@ -2633,12 +2633,10 @@ static void lock_get_status(struct seq_file *f, struct file_lock *fl,
> >  
> >  	fl_pid = locks_translate_pid(fl, proc_pidns);
> >  	/*
> > -	 * If there isn't a fl_pid don't display who is waiting on
> > -	 * the lock if we are called from locks_show, or if we are
> > -	 * called from __show_fd_info - skip lock entirely
> > +	 * If lock owner is dead (and pid is freed) or not visible in current
> > +	 * pidns, zero is shown as a pid value. Check lock info from
> > +	 * init_pid_ns to get saved lock pid value.
> >  	 */
> > -	if (fl_pid == 0)
> > -		return;
> >  
> >  	if (fl->fl_file != NULL)
> >  		inode = locks_inode(fl->fl_file);
> 
> (cc'ing Nickolay)
> 
> As Andrey points out, this behavior was originally added in commit
> d67fd44f697d to address performance issues when there are a lot of locks
> held by tasks in other namespaces.
> 
> Will allowing this code to show these again cause a problem there?

No, it will not. The content of /proc/locks is still filtered. As for
fdinfo, it shows locks for one file descriptor, there will not be a lot of
locks. And fdinfo was designed to show all locks for a file descriptor,
it doesn't matter in what pidns they were taken.

Thanks,
Andrei

> -- 
> Jeff Layton <jlayton@xxxxxxxxxx>