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;
Organization: Red Hat UK Ltd. Registered Address: Red Hat UK Ltd, Amberley
        Place, 107-111 Peascod Street, Windsor, Berkshire, SI4 1TE, United
        Kingdom.
        Registered in England and Wales under Company Registration No. 3798903
From:   David Howells <dhowells@redhat.com>
In-Reply-To: <Yl7d++G25sNXIR+p@rabbit.intern.cm-ag>
References: <Yl7d++G25sNXIR+p@rabbit.intern.cm-ag> <YlWWbpW5Foynjllo@rabbit.intern.cm-ag> <507518.1650383808@warthog.procyon.org.uk>
To:     Max Kellermann <mk@cm4all.com>
Cc:     dhowells@redhat.com, linux-cachefs@redhat.com,
        linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: fscache corruption in Linux 5.17?
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <509960.1650386569.1@warthog.procyon.org.uk>
Content-Transfer-Encoding: quoted-printable
Date:   Tue, 19 Apr 2022 17:42:49 +0100
Message-ID: <509961.1650386569@warthog.procyon.org.uk>
Precedence: bulk

Max Kellermann <mk@cm4all.com> wrote:

> Did you read this part of my email?:

Sorry, I'm trying to deal with several things at once.

> My theory was that fscache shows a mix of old and new pages after the
> file was modified.  Does this make sense?

Okay - that makes a bit more sense.

Could the file have been modified by a third party?  If you're using NFS3
there's a problem if two clients can modify a file at the same time.  The
second write can mask the first write and the client has no way to detect =
it.
The problem is inherent to the protocol design.  The NFS2 and NFS3 protoco=
ls
don't support anything better than {ctime,mtime,filesize} - the change
attribute only becomes available with NFS4.

If an NFS file is opened for writing locally, the cache for it supposed to=
 be
invalidated and remain unused until there are no open file descriptors lef=
t
referring to it.  This is intended for handling DIO writes, but it should
serve for this also.

The following might be of use in checking if the invalidation happens loca=
lly:

echo 1 >/sys/kernel/debug/tracing/events/fscache/fscache_invalidate/enable

And then this can be used to check if it correctly identifies that it has =
an
obsolete version of the file in the cache when it binds to it:

echo 1 >/sys/kernel/debug/tracing/events/cachefiles/cachefiles_coherency/e=
nable


David