Re: Autofs race condition ?
H. Peter Anvin (hpa@transmeta.com)
31 Oct 1999 02:52:38 GMT
Followup to: <19991031023530.A22566@alienor.populi.vox>
By author: Thierry Danis <danis@mail.dotcom.fr>
In newsgroup: linux.dev.kernel
>
> Hello,
>
> I have observed what looks like a race in automount (autofs). It happens
> when an access is reactivating a mount point that is in the process of
> beeing expiring.
>
> Here are the logs in /var/log/messages :
>
> Oct 31 02:14:51 pcodile automount[12272]: expired /alienor/usr.reuse
> Oct 31 02:14:52 pcodile automount[12272]: expired /alienor/usr.reuse.debug
> Oct 31 02:14:52 pcodile automount[693]: attempting to mount entry /alienor/home
> Oct 31 02:14:52 pcodile automount[12272]: expired /alienor/home
> Oct 31 02:14:52 pcodile automount[12277]: >> mount: mount point /alienor/home does not exist
> Oct 31 02:14:52 pcodile automount[12277]: mount(nfs): nfs: mount failure alienor:/home on /alienor/home
> Oct 31 02:14:52 pcodile automount[693]: attempting to mount entry /alienor/home
>
> When that happens, all subsequent accesses to /alienor/home are blocked.
> If I ^C the first offending process, the remote partition can finally
> be mounted.
>
> Kernel version 2.2.12-20 (coming from RH 6.1). Autofs : autofs-3.1.3-9.
>
Please send this to me or to the autofs list in the future. This is
very quite interesting; and definitely a "this shouldn't happen".
What seems to happen is an access in between the mount() and the
rmdir() of the expiring process. I will look into it as soon as
possible.
-hpa
--
<hpa@transmeta.com> at work, <hpa@zytor.com> in private!
-
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/