bert hubert wrote:
>
> Yesterday, I shut down my computer manually by doing alt-sysreq-s a few
> times (sync), and then emergency remount read/only. This used to work fine
> on my workstation.
'shutdown -nr now' has a nicer user interface :-)
> This morning when I booted it appeared that the sync had not been performed,
> there were lots of 'deleted inode X has unset dtime' notices, and even 2
> unattached inodes.
Yes, there is a problem with umount. From linux-fsdevel:
> I can demonstrate this 100% with NFS. "foo" is an NFS server:
>
>
> mnm:/# mount foo:/ /mnt/foo
> mnm:/# cd /mnt/foo/usr/src/linux
> mnm:/mnt/foo/usr/src/linux# make dep
> [ elided ]
> mnm:/mnt/foo/usr/src/linux# cd /
> mnm:/# umount -f /mnt/foo
> umount2: Device or resource busy
> umount: /mnt/foo: device is busy
> mnm:/# lsof|grep foo
> mnm:/# fuser /mnt/foo
> mnm:/# fuser /mnt/foo/usr/src/linux
>
> It seems you need to do quite a lot of I/O on the mounted fs to trigger
> this. 'make dep' does the job.
>
> The umount is failing because root->d_count == 2 in d_umount().
The same happens with an ext2 fs. I should point out that in this
example, "make" is aliased to "make -j3".
I assume this is on one of the fs guys' to-do list.
-- -akpm-- 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/
This archive was generated by hypermail 2b29 : Sat Apr 15 2000 - 21:00:24 EST