Re: vfs: INFO: possible circular locking dependency detected

From: Al Viro
Date: Wed May 09 2012 - 12:37:30 EST

On Wed, May 09, 2012 at 05:28:54PM +0100, Al Viro wrote:
> On Wed, May 09, 2012 at 06:23:30PM +0200, Sasha Levin wrote:
> > On Wed, May 9, 2012 at 6:12 PM, Al Viro <viro@xxxxxxxxxxxxxxxxxx> wrote:
> > > On Wed, May 09, 2012 at 05:25:14PM +0200, Sasha Levin wrote:
> > >> Hi all,
> > >>
> > >> I've started seeing the following warning while fuzzing inside a KVM guest with the latest -next:
> > > ? ? ? ?It's not a realistic attack, fortunately, since you need root
> > > to get past open_exec() on any of those... ?Wait. ?How _did_ you get
> > > past open_exec(), anyway? ?MAY_EXEC is not supposed to be granted on
> > > anything that has no exec bits at all and AFAICS none of those files
> > > have them.
> >
> > You could chmod +x and run them, no?
> Can't. proc_setattr() will give you -EPERM and refuse to do anything
> if you call it with ATTR_MODE in ->ia_valid.

OTOH, you probably can do that on unrelated seq_file outside of per-process
part of procfs. So, yes, one could get a warning like that if they, as root,
would do e.g.
chmod +x /proc/swaps
attempt to execve() /proc/swaps
cat /proc/self/environ
and enjoy the hard-earned false positive (it's a different seq_file, so
we have no deadlock). If that's _all_ that happened, I'm not particulary
concerned; it's not pretty, but saying "thou shalt not grab ->cred_guard_mutex
anywhere in ->read() on anything that has exec bits or might get one" is
not too terrible. If that's something else, though, we might have a real
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at
Please read the FAQ at