Received-SPF: pass (google.com: domain of linux-nfs-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: <165660978413.2453.15153844664543877314.stgit@klimt.1015granger.net>
 <faae3bdaf5fc4b1c2ff481977bd1bf091bb0c8b6.camel@hammerspace.com>
In-Reply-To: <faae3bdaf5fc4b1c2ff481977bd1bf091bb0c8b6.camel@hammerspace.com>
From:   Anna Schumaker <schumaker.anna@gmail.com>
Date:   Thu, 30 Jun 2022 14:09:02 -0400
Message-ID: <CAFX2Jfmxi2_M=Ptt=2M7dbks62Gkdia2913yod5zaZ8bV9vGwA@mail.gmail.com>
Subject: Re: [PATCH v1] SUNRPC: Fix READ_PLUS crasher
To:     Trond Myklebust <trondmy@hammerspace.com>
Cc:     "linux-nfs@vger.kernel.org" <linux-nfs@vger.kernel.org>,
        "chuck.lever@oracle.com" <chuck.lever@oracle.com>,
        "bfields@fieldses.org" <bfields@fieldses.org>,
        "anna.schumaker@netapp.com" <anna.schumaker@netapp.com>,
        "zlang@redhat.com" <zlang@redhat.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Thu, Jun 30, 2022 at 2:03 PM Trond Myklebust <trondmy@hammerspace.com> wrote:
>
> On Thu, 2022-06-30 at 13:24 -0400, Chuck Lever wrote:
> > Looks like there are still cases when "space_left - frag1bytes" can
> > legitimately exceed PAGE_SIZE. Ensure that xdr->end always remains
> > within the current encode buffer.
> >
> > Reported-by: Bruce Fields <bfields@fieldses.org>
> > Reported-by: Zorro Lang <zlang@redhat.com>
> > Link: https://bugzilla.kernel.org/show_bug.cgi?id=216151
> > Fixes: 6c254bf3b637 ("SUNRPC: Fix the calculation of xdr->end in
> > xdr_get_next_encode_buffer()")
> > Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
> > ---
> >  net/sunrpc/xdr.c |    2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > Hi-
> >
> > I had a few minutes yesterday afternoon to look into this one. The
> > following one-liner seems to address the issue and passes my smoke
> > tests with NFSv4.1/TCP and NFSv3/RDMA. Any thoughts?
> >
> >
> > diff --git a/net/sunrpc/xdr.c b/net/sunrpc/xdr.c
> > index f87a2d8f23a7..916659be2774 100644
> > --- a/net/sunrpc/xdr.c
> > +++ b/net/sunrpc/xdr.c
> > @@ -987,7 +987,7 @@ static noinline __be32
> > *xdr_get_next_encode_buffer(struct xdr_stream *xdr,
> >         if (space_left - nbytes >= PAGE_SIZE)
> >                 xdr->end = p + PAGE_SIZE;
> >         else
> > -               xdr->end = p + space_left - frag1bytes;
> > +               xdr->end = p + min_t(int, space_left - frag1bytes,
> > PAGE_SIZE);
>
> Since we know that frag1bytes <= nbytes (that is determined in
> xdr_reserve_space()), isn't this effectively the same thing as changing
> the condition to
>
>         if (space_left - frag1bytes >= PAGE_SIZE)
>                 xdr->end = p + PAGE_SIZE;
>         else
>                 xdr->end = p + space_left - frag1bytes;

I added some printk's without this patch, and I'm seeing space_left
larger than PAGE_SIZE and frag1bytes set to 0 in that branch right
before the crash. So the min_t() will choose the PAGE_SIZE option
sometimes.

Anna

>
> ?
> >
> >         xdr->buf->page_len += frag2bytes;
> >         xdr->buf->len += nbytes;
> >
> >
>
> --
> Trond Myklebust
> Linux NFS client maintainer, Hammerspace
> trond.myklebust@hammerspace.com
>
>