Re: [LKP] d50d82faa0 [ 33.671845] WARNING: possible circular locking dependency detected

[Mikulas Patocka <mpatocka@redhat.com>]

On Wed, 7 Nov 2018, Andrew Morton wrote:

> On Wed, 7 Nov 2018 15:43:36 -0800 Andrew Morton <akpm@linux-foundation.org> wrote:
> 
> > On Tue, 23 Oct 2018 08:30:04 +0800 kernel test robot <rong.a.chen@intel.com> wrote:
> > 
> > > Greetings,
> > > 
> > > 0day kernel testing robot got the below dmesg and the first bad commit is
> > > 
> > > https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master
> > > 
> > > commit d50d82faa0c964e31f7a946ba8aba7c715ca7ab0
> > > Author:     Mikulas Patocka <mpatocka@redhat.com>
> > > AuthorDate: Wed Jun 27 23:26:09 2018 -0700
> > > Commit:     Linus Torvalds <torvalds@linux-foundation.org>
> > > CommitDate: Thu Jun 28 11:16:44 2018 -0700
> > > 
> > >     slub: fix failure when we delete and create a slab cache
> > 
> > This is ugly.  Is there an alternative way of fixing the race which
> > Mikulas attempted to address?  Possibly cancel the work and reuse the
> > existing sysfs file, or is that too stupid to live?
> > 
> > 3b7b314053d021 ("slub: make sysfs file removal asynchronous") was
> > pretty lame, really.  As mentioned,
> > 
> > : It'd be the cleanest to deal with the issue by removing sysfs files
> > : without holding slab_mutex before the rest of shutdown; however, given
> > : the current code structure, it is pretty difficult to do so.
> > 
> > Would be a preferable approach.
> > 
> > >     
> > >     This uncovered a bug in the slub subsystem - if we delete a cache and
> > >     immediatelly create another cache with the same attributes, it fails
> > >     because of duplicate filename in /sys/kernel/slab/.  The slub subsystem
> > >     offloads freeing the cache to a workqueue - and if we create the new
> > >     cache before the workqueue runs, it complains because of duplicate
> > >     filename in sysfs.
> 
> Alternatively, could we flush the workqueue before attempting to
> (re)create the sysfs file?

What if someone creates the slab cache from the workqueue?

> Extra points for only doing this if the
> first (re)creation attempt returned -EEXIST?

If it returns -EEXIST, it has already written the warning to the log.

Mikulas