Re: [PATCH] make autoclear operation synchronous again

[Christoph Hellwig <hch@lst.de>]

On Wed, Jan 05, 2022 at 03:10:23PM +0900, Tetsuo Handa wrote:
> On 2022/01/05 15:02, Jan Stancek wrote:
> > On Thu, Dec 30, 2021 at 07:52:34PM +0900, Tetsuo Handa wrote:
> >> OK. Two patches shown below. Are these look reasonable?
> >>
> >>
> >>
> >>> From 1409a49181efcc474fbae2cf4e60cbc37adf34aa Mon Sep 17 00:00:00 2001
> >> From: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
> >> Date: Thu, 30 Dec 2021 18:41:05 +0900
> >> Subject: [PATCH 1/2] loop: Revert "loop: make autoclear operation asynchronous"
> >>
> > 
> > Thanks, the revert fixes failures we saw recently in LTP tests,
> > which do mount/umount in close succession:
> > 
> > # for i in `seq 1 2`;do mount -t iso9660 -o loop /root/isofs.iso /mnt/isofs; umount /mnt/isofs/; done
> > mount: /mnt/isofs: WARNING: source write-protected, mounted read-only.
> > mount: /mnt/isofs: wrong fs type, bad option, bad superblock on /dev/loop0, missing codepage or helper program, or other error.
> > umount: /mnt/isofs/: not mounted.
> > 
> 
> I'm waiting for Jens to come back to
> https://lkml.kernel.org/r/c205dcd2-db55-a35c-e2ef-20193b5ac0da@i-love.sakura.ne.jp .

So I think the version in this thread is what we should merge.

On top of that we should:

 - remove the probe mechanism (patch already sent)
 - stop taking open_mutex in del_gendisk and bdev_disk_changed (I have
   a series for that now)