Re: CVE Request: cronie fd leak

[Vincent Danen <vdanen () redhat com>]

* [2013-01-09 09:24:23 +0100] Sebastian Krahmer wrote:

> Possible that you have got a different cron implementation.

No, we're using cronie as well.

> There is no hidden info in our bugzilla; the reproducer
> is using lvm commands, but I remember it worked with any
> command. Actually its not about the warnings, a "cat" will probably
> also do, if you check its /proc/$pid/fd when its invoked.
> But cat doesnt emit warnings about open fd's.
>
> Not sure about upstream, probably not.

Ok, so did some more digging based on some info from one of our
developers that we had patched this in Fedora.

Looks like this patch introduced the leak on 2011-04-28:

http://git.fedorahosted.org/cgit/cronie.git/commit/src/cron.c?id=acdf4ae8456888ed78201906ef528f4c28f54582

And this patch reverted it on 2011-06-29:

http://git.fedorahosted.org/cgit/cronie.git/commit/src/cron.c?id=b19007ca9fddd62ecef3af4a7d2d252f1d5e0419

So it looks like only 1.4.8 was affected by this (which, judging by the
patch in your bugzilla is the same version you're seeing as affected).

That might be a better patch to use than what you're using.  Anyways,
this only affects 1.4.8 (for any others using cronie and concerned as to
whether or not they might be affected).

This was also reported to our bugzilla here:

https://bugzilla.redhat.com/show_bug.cgi?id=717505

> On Tue, Jan 08, 2013 at 09:01:19PM -0700, Vincent Danen wrote:
> > * [2013-01-08 13:56:40 +0100] Sebastian Krahmer wrote:
> >
> > > "Hello Kurt, Steve, vendors,"
> > >
> > > cronie leaks read-only fd's, please check here:
> > >
> > > https://bugzilla.novell.com/show_bug.cgi?id=786096
> > >
> > > can someone assign a CVE?
> >
> > Sebastian, do you have a specific command that you're using?  I'm trying
> > to reproduce this in Fedora and RHEL using lvdisplay (maybe a bad
> > choice?) and also using "lvm vgck -v vg_thor && lvm pvs" in
> > /etc/crontab.
> >
> > The output is mailed to me fine with no warnings?  Can you share what
> > command was being used to reproduce this?  It's possible that something
> > you added (or we added) makes this a non-issue on other platforms.
> >
> > Has upstream been informed of this yet?
> >
> > --
> > Vincent Danen / Red Hat Security Response Team
>
> --
>
> ~ perl self.pl
> ~ $_='print"\$_=\47$_\47;eval"';eval
> ~ krahmer () suse de - SuSE Security Team

--
Vincent Danen / Red Hat Security Response Team