Received-SPF: pass (google.com: domain of linux-ext4-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: <YjDj3lvlNJK/IPiU@bfoster> <YjJPu/3tYnuKK888@casper.infradead.org>
 <CAHk-=wgPTWoXCa=JembExs8Y7fw7YUi9XR0zn1xaxWLSXBN_vg@mail.gmail.com>
 <YjNN5SzHELGig+U4@casper.infradead.org> <CAHk-=wiZvOpaP0DVyqOnspFqpXRaT6q53=gnA2psxnf5dbt7bw@mail.gmail.com>
 <YjOlJL7xwktKoLFN@casper.infradead.org> <20220318131600.iv7ct2m4o52plkhl@quack3.lan>
In-Reply-To: <20220318131600.iv7ct2m4o52plkhl@quack3.lan>
From:   Linus Torvalds <torvalds@linux-foundation.org>
Date:   Fri, 18 Mar 2022 11:56:04 -0700
Message-ID: <CAHk-=wiky+cT7xF_2S94ToEjm=XNX73CsFHaQJH3tzYQ+Vb1mw@mail.gmail.com>
Subject: Re: writeback completion soft lockup BUG in folio_wake_bit()
To:     Jan Kara <jack@suse.cz>
Cc:     Matthew Wilcox <willy@infradead.org>,
        Brian Foster <bfoster@redhat.com>,
        Linux-MM <linux-mm@kvack.org>,
        linux-fsdevel <linux-fsdevel@vger.kernel.org>,
        linux-xfs <linux-xfs@vger.kernel.org>,
        Hugh Dickins <hughd@google.com>,
        Namjae Jeon <namjae.jeon@samsung.com>,
        Ashish Sangwan <a.sangwan@samsung.com>,
        "Theodore Ts'o" <tytso@mit.edu>,
        Ext4 Developers List <linux-ext4@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Fri, Mar 18, 2022 at 6:16 AM Jan Kara <jack@suse.cz> wrote:
>
> I agree with Dave that 'keep_towrite' thing is kind of self-inflicted
> damage on the ext4 side (we need to write out some blocks underlying the
> page but cannot write all from the transaction commit code, so we need to
> keep xarray tags intact so that data integrity sync cannot miss the page).
> Also it is no longer needed in the current default ext4 setup. But if you
> have blocksize < pagesize and mount the fs with 'dioreadlock,data=ordered'
> mount options, the hack is still needed AFAIK and we don't have a
> reasonable way around it.

I assume you meant 'dioread_lock'.

Which seems to be the default (even if 'data=ordered' is not).

Anyway - if it's not a problem for any current default setting, maybe
the solution is to warn about this case and turn it off?

IOW, we could simply warn about "data=ordered is no longer supported"
and turn it into data=journal.

Obviously *only* do this for the case of "blocksize < PAGE_SIZE".

If this ext4 thing is (a) obsolete and (b) causes VFS-level problems
that nobody else has, I really think we'd be much better off disabling
it than trying to work with it.

                 Linus