Re: autofs bug report

H. Peter Anvin (hpa@transmeta.com)
19 Sep 1998 19:45:21 GMT

Followup to: <E0zKPwy-0007Ba-00@danube.inka.de>
By author: Olaf Titz <olaf@bigred.inka.de>
In newsgroup: linux.dev.kernel
>
> > This is really a case of "don't do that, then"; this means someone
> > with root privileges has done something fundamentally stupid. And
> > yes, it does cause deadlock. I don't really consider this a bug; a
>
> But it shouldn't be too difficult (although I don't know the code) to
> put in sanity checks to avoid the error. Like we do have sanity checks
> in syscalls only available to root which prevent us from Oopsing. To
> take an example of mine, I put in a number of such checks into the
> kernel part of CIPE although that is not expected to be called by
> anything but its own accompanying driver program, _and_ everything in
> there requires UID 0 (or in future versions CAP_NET_ADMIN).
>

autofs won't even see the mount request, so it is nothing it can do to
stop it. /bin/mount doesn't know that it isn't being called by the
autofs daemon, so there is nothing *it* can do about it. All in all,
this is a hard one to fix, requires root to mess up in a fairly
obvious manner, and even if the sysadmin messes up doesn't cause
anything like an oops -- that automount point will be out of
commission until you kill the mount process; depending on the version
of /bin/mount you may not be able to mount other filesystem until you
remove /etc/mtab~.

I really don't consider this one a problem.

-hpa

-- 
    PGP: 2047/2A960705 BA 03 D3 2C 14 A8 A8 BD  1E DF FE 69 EE 35 BD 74
    See http://www.zytor.com/~hpa/ for web page and full PGP public key
        I am Bahá'í -- ask me about it or see http://www.bahai.org/
   "To love another person is to see the face of God." -- Les Misérables


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/