Unable to unmount idle filesystem on 6.2

[Darren Pilgrim]

Kris Kennaway wrote:
> Darren Pilgrim wrote:
>> I'm unable to unmount an idle filesystem (or even drop it to
>> read-only):
>>
>> # mount
>> /dev/da0s1a on / (ufs, local, noatime)
>> devfs on /dev (devfs, local)
>> /dev/da0s1d on /var (ufs, local, noatime, soft-updates)
>> /dev/da0s1e on /usr (ufs, local, noatime, soft-updates)
>> /dev/da0s1fp1 on /usr/obj (ufs, asynchronous, local, noatime)
>> /dev/da0s1fp2 on /usr/ports (ufs, local, soft-updates)
>> /dev/da0s1fp3 on /usr/src (ufs, local, soft-updates)
>> /dev/da0s2d on /data (ufs, local, noatime, soft-updates)
>>
>> # fstat -f /usr/ports
>> USER     CMD          PID   FD MOUNT      INUM MODE         SZ|DV R/W
>>
>> # umount /usr/ports
>> umount: unmount of /usr/ports failed: Device busy
>>
>> # umount -f /usr/ports
>> umount: unmount of /usr/ports failed: Device busy
>>
>> # mount -o ro /usr/ports
>> mount: /dev/da0s1fp2: Operation not permitted
>>
>> # uname -r
>> 6.2-RELEASE-p8
>> _______________________________________________
>> freebsd-questions at freebsd.org mailing list
>> http://lists.freebsd.org/mailman/listinfo/freebsd-questions
>> To unsubscribe, send any mail to 
>> "freebsd-questions-unsubscribe at freebsd.org"
>>
>>
> 
> Strange, can you break to DDB and do 'show lockedvnods'?

I haven't done that yet; however, I did find 12 instances of the 
following in the log:

softdep_waitidle: Failed to flush worklist for 0xc66e5298

A quick check and that message gets spit out whenever I issue any 
of the following commands:

# mount -uo ro /usr/ports
# umount /usr/ports
# umount -f /usr/ports

A bit of searching on that error message tells me I've hit some 
kind of a corner case with soft-updates.  The filesystem was 
mounted read-only, then upgraded to rw so I could update the ports 
tree.  After cvsup was done, I tried to take the filesystem back 
down to read-only.  The common case seems to be that the mount 
change is followed too quickly after the large number of writes and 
it somehow wedges soft-updates.

Unfortunately, I haven't been able to find a fix other than 
rebooting the machine.  The problem is that the search results[1] 
also tell me the filesystem may well be hosed and the reboot won't 
be clean.  Luckily for me, I can just drop the FS from /etc/fstab 
and newfs the partition after the box comes back up.

[1]: 
http://lists.freebsd.org/pipermail/freebsd-current/2007-February/069178.html