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: <00000000000058d01705ecddccb0@google.com> <CAG_fn=WAyOc+1GEC+P3PpTM2zLcLcepAX1pPXkj5C6aPyrDVUA@mail.gmail.com>
 <Y2lGu/QTIWNpzFI3@sol.localdomain>
In-Reply-To: <Y2lGu/QTIWNpzFI3@sol.localdomain>
From:   Alexander Potapenko <glider@google.com>
Date:   Mon, 7 Nov 2022 19:14:55 +0100
Message-ID: <CAG_fn=VQBv-sgPhT0gLVChAtMNx0F3RcQYDKdvhBL4mBpiDkFA@mail.gmail.com>
Subject: Re: [syzbot] KMSAN: uninit-value in pagecache_write
To:     Eric Biggers <ebiggers@kernel.org>
Cc:     syzbot <syzbot+9767be679ef5016b6082@syzkaller.appspotmail.com>,
        adilger.kernel@dilger.ca, linux-ext4@vger.kernel.org,
        linux-kernel@vger.kernel.org, syzkaller-bugs@googlegroups.com,
        tytso@mit.edu
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Mon, Nov 7, 2022 at 6:56 PM Eric Biggers <ebiggers@kernel.org> wrote:
>
> On Mon, Nov 07, 2022 at 10:46:13AM +0100, 'Alexander Potapenko' via syzka=
ller-bugs wrote:
> >    ext4: initialize fsdata in pagecache_write()
> >
> >     When aops->write_begin() does not initialize fsdata, KMSAN reports
> >     an error passing the latter to aops->write_end().
> >
> >     Fix this by unconditionally initializing fsdata.
> >
> >     Fixes: c93d8f885809 ("ext4: add basic fs-verity support")
> >     Reported-by: syzbot+9767be679ef5016b6082@syzkaller.appspotmail.com
> >     Signed-off-by: Alexander Potapenko <glider@google.com>
> >
> > diff --git a/fs/ext4/verity.c b/fs/ext4/verity.c
> > index 3c640bd7ecaeb..30e3b65798b50 100644
> > --- a/fs/ext4/verity.c
> > +++ b/fs/ext4/verity.c
> > @@ -79,7 +79,7 @@ static int pagecache_write(struct inode *inode,
> > const void *buf, size_t count,
> >                 size_t n =3D min_t(size_t, count,
> >                                  PAGE_SIZE - offset_in_page(pos));
> >                 struct page *page;
> > -               void *fsdata;
> > +               void *fsdata =3D NULL;
> >                 int res;
> >
> >                 res =3D aops->write_begin(NULL, mapping, pos, n, &page,=
 &fsdata);
>
> Are you sure that KMSAN should be reporting this?  The uninitialized valu=
e is
> passed as a function parameter, but it's never actually used.

To summarize what's written here:
https://lore.kernel.org/lkml/20220701142310.2188015-44-glider@google.com/
:
 - this is UB from the C11 standpoint;
 - Linus despises UB in general, but expecting function arguments to
be initialized is reasonable;
 - if a function ends up being a no-op, tools should not warn about
uninits passed to it;
 - KMSAN only warns about functions that survive inlining.

In addition, one cannot really rely on write_end() being consistent
with write_begin(), in particular if fsdata was ignored initially and
only appeared at a later point.

>
> Anyway, this patch doesn't hurt, I suppose.  Can please you send it out a=
s a
> formal patch to linux-ext4?  It would be easy for people to miss this pat=
ch
> buried in this thread.  Also, can you please send a patch to linux-f2fs-d=
evel
> for the same code in fs/f2fs/verity.c?

Will do!

> Thanks!
>
> - Eric


--=20
Alexander Potapenko
Software Engineer

Google Germany GmbH
Erika-Mann-Stra=C3=9Fe, 33
80636 M=C3=BCnchen

Gesch=C3=A4ftsf=C3=BChrer: Paul Manicle, Liana Sebastian
Registergericht und -nummer: Hamburg, HRB 86891
Sitz der Gesellschaft: Hamburg