Re: [RFC 0/4] Fix gfs2 readahead deadlocks

From: Andreas Gruenbacher <agruenba@redhat.com>
To: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Matthew Wilcox <willy@infradead.org>,
	Dave Chinner <david@fromorbit.com>,
	linux-fsdevel <linux-fsdevel@vger.kernel.org>,
	Linux-MM <linux-mm@kvack.org>,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>
Subject: Re: [RFC 0/4] Fix gfs2 readahead deadlocks
Date: Thu, 2 Jul 2020 20:23:03 +0200	[thread overview]
Message-ID: <CAHc6FU7ZWJb308yfMaskFeSwNxgxqn89pxT4F7Ud4HthhrC5CA@mail.gmail.com> (raw)
In-Reply-To: <CAHk-=whb4H3ywKcwGxgjFSTEap_WuFj5SW7CYw0J2j=WGUs4nQ@mail.gmail.com>

On Thu, Jul 2, 2020 at 8:11 PM Linus Torvalds
<torvalds@linux-foundation.org> wrote:
> On Thu, Jul 2, 2020 at 9:51 AM Andreas Gruenbacher <agruenba@redhat.com> wrote:
> >
> > Of this patch queue, either only the first patch or all four patches can
> > be applied to fix gfs2's current issues in 5.8.  Please let me know what
> > you think.
>
> I think the IOCB_NOIO flag looks fine (apart from the nit I pointed
> out), and we could do that.

Ok, that's a step forward.

> However, is the "revert and reinstate" looks odd. Is the reinstate so
> different from the original that it makes sense to do that way?
>
> Or was it done that way only to give the choice of just doing the revert?
>
> Because if so, I think I'd rather just see a "fix" rather than
> "revert+reinstate".

I only did the "revert and reinstate" so that the revert alone will
give us a working gfs2 in 5.8. If there's agreement to add the
IOCB_NOIO flag, then we can just fix gfs2 (basically
https://lore.kernel.org/linux-fsdevel/20200619093916.1081129-3-agruenba@redhat.com/
with IOCB_CACHED renamed to IOCB_NOIO).

Thanks,
Andreas