Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20230606124800.1151665-1-agruenba@redhat.com> <CAHk-=wi_P4tOWCbZYo11u5vbB+tTxCSGwKC9m71y-QdbZP8zKA@mail.gmail.com>
In-Reply-To: <CAHk-=wi_P4tOWCbZYo11u5vbB+tTxCSGwKC9m71y-QdbZP8zKA@mail.gmail.com>
From:   Andreas Gruenbacher <agruenba@redhat.com>
Date:   Tue, 6 Jun 2023 15:32:34 +0200
Message-ID: <CAHc6FU7e-ZwFg71S4a6UqdW5aagFPJAqGRhwwov+OMUHcY4UPA@mail.gmail.com>
Subject: Re: [GIT PULL] gfs2 fix
To:     Linus Torvalds <torvalds@linux-foundation.org>
Cc:     cluster-devel@redhat.com, linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Tue, Jun 6, 2023 at 2:55=E2=80=AFPM Linus Torvalds
<torvalds@linux-foundation.org> wrote:
> On Tue, Jun 6, 2023 at 5:48=E2=80=AFAM Andreas Gruenbacher <agruenba@redh=
at.com> wrote:
> >
> > - Don't get stuck writing page onto itself under direct I/O.
>
> Btw, is there a test for this DIO case?

The previous test case I wrote for these kinds of page faults is:

  "generic: Test page faults during read and write"
  https://git.kernel.org/pub/scm/fs/xfs/xfstests-dev.git/commit/?id=3Dd3cbd=
abff

I've added a check for this specific case, but this change hasn't been
posted/merged yet:

  "generic/728: Add mmap + DIO write test"
  https://gitlab.com/agruenba/xfstests/-/commit/8c37de03

> We've had the deadlock issue on t page lock (or for inode locks or
> whatever) for normal IO when faulting in the same page that is written
> to, and we have as pattern for solving that and I think there are
> filesystem tests that trigger this.
>
> But the DIO pattern is a bit different, with the whole "invalidate
> page cache: issue, and the fact that you send this patch now (rather
> than years ago) makes me wonder about test coverage for this all?

Yes, this case wasn't covered so far. The other page fault issues are
covered since 2021, and were fixed in gfs2 back then.

Thanks,
Andreas