Henrik Grindal Bakken <
hgb@ifi.uio.no> writes:
> Hi, Charles. I've been testing this nfs patch a bit, and I have one
> problem in my setup. This might be related to other parts than
> yaffs2, but then again, it could be a yaffs bug.
>
> When I export a yaffs2 filesystem with the kernel-space nfs daemon, I
> cannot seem to umount the filesystem, even though rpc.nfsd and friends
> have been terminated.
>
> I can see an [nfsd] process (i.e. kernel thread), which I believe
> might be the culprit. No other processes have any files on my file
> system open, no other processes have /proc/pid/{cwd,root} pointing to
> the yaffs mount point.
>
> Have you tested this bit? It seems like some kernel-side resources
> might not be properly released when we unexport the file system.
An update...
Yesterday, I fiddled with this, and among other things tried to kill
the [lockd] and [nfsd] processes that were still running. I managed
to kill [lockd], but not [nfsd] (with kill -TERM or kill -KILL).
However, when I arrived today (after the box just sitting there for
the night), the [nfsd] process was gone too, and the file system could
be properlt umounted.
Seems there might be some kind of timeout or grace period somewhere.
--
Henrik Grindal Bakken <
hgb@ifi.uio.no>
PGP ID: 8D436E52
Fingerprint: 131D 9590 F0CF 47EF 7963 02AF 9236 D25A 8D43 6E52
(text/plain)