Re: [cgroup or VFS ?] WARNING: at fs/namespace.c:636 mntput_no_expire+0xac/0xf2()

[Al Viro <viro@ZenIV.linux.org.uk>]

On Fri, Feb 13, 2009 at 02:12:13PM +0800, Li Zefan wrote:
> Al Viro wrote:
> > On Fri, Feb 13, 2009 at 01:09:17PM +0800, Li Zefan wrote:
> > 
> >> I ran following testcase, and triggered the warning in 1 hour:
> >>
> >> thread 1:
> >> for ((; ;))
> >> {
> >>         mount --bind /cgroup /mnt > /dev/null 2>&1
> >>         umount /mnt > /dev/null 2>&1
> >> }
> >>
> >> tread 2:
> >> for ((; ;))
> >> {
> >>         mount -t cgroup -o cpu xxx /cgroup > /dev/null 2>&1
> >>         mkdir /cgroup/0 > /dev/null 2>&1
> >>         rmdir /cgroup/0 > /dev/null 2>&1
> >>         umount -l /cgroup > /dev/null 2>&1
> >> }
> > 
> > Wow.  You know, at that point these redirects could probably be removed.
> 
> Ah, yes.
> 
> > If anything in there ends up producing an output, we very much want to
> > see that.  Actually, I'd even make that
> > 	mount --bind /cgroup/mnt || (echo mount1: ; date)
> > etc., so we'd see when do they fail and which one fails (if any)...
> >  
> > Which umount has failed in the above, BTW?
> > 
> > 
> 
> the first one sometimes failed, and the second one hasn't failed:

> mount: wrong fs type, bad option, bad superblock on /cgroup,
>        missing codepage or helper program, or other error
>        In some cases useful info is found in syslog - try
>        dmesg | tail  or so
> 
> mount1

Hold on.  In your last example the first one was doing mount --bind;
has _that_ failed?  Oh, wait...  It can fail, all right, if lookup on
/cgroup gives you your filesystem with the second thread managing to
detach it before we get the namespace_sem.  Then we'll fail that way -
and clean up properly.

Oh, well...  The original question still stands: with those two
scripts, which umount produces that WARN_ON?  The trivial way
to check would be to have a copy of /sbin/umount under a different
name and use _that_ in one of the threads instead of umount.
Then reproduce the WARN_ON and look at the process name in dmesg...