Received-SPF: pass (google.com: domain of linux-nfs+bounces-1198-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.80.249 as permitted sender) client-ip=147.75.80.249;
Message-ID: <3b4b548b5d7ede4632a113304cab38002f4aa2e1.camel@pengutronix.de>
Subject: Re: [PATCH] SUNRPC: use request size to initialize bio_vec in
 svc_udp_sendto()
From: Lucas Stach <l.stach@pengutronix.de>
To: Chuck Lever <chuck.lever@oracle.com>
Cc: Jeff Layton <jlayton@kernel.org>, Neil Brown <neilb@suse.de>, Olga
 Kornievskaia <kolga@netapp.com>, Dai Ngo <Dai.Ngo@oracle.com>, Tom Talpey
 <tom@talpey.com>, Trond Myklebust <trond.myklebust@hammerspace.com>, Anna
 Schumaker <anna@kernel.org>,  linux-nfs@vger.kernel.org,
 netdev@vger.kernel.org, kernel@pengutronix.de,  patchwork-lst@pengutronix.de
Date: Thu, 18 Jan 2024 09:42:42 +0100
In-Reply-To: <ZahOmmMZxHcK8Amn@tissot.1015granger.net>
References: <20240117210628.1534046-1-l.stach@pengutronix.de>
	 <ZahOmmMZxHcK8Amn@tissot.1015granger.net>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.48.4 (3.48.4-1.fc38) 
Precedence: bulk
MIME-Version: 1.0

Am Mittwoch, dem 17.01.2024 um 17:03 -0500 schrieb Chuck Lever:
> On Wed, Jan 17, 2024 at 10:06:28PM +0100, Lucas Stach wrote:
> > Use the proper size when setting up the bio_vec, as otherwise only
> > zero-length UDP packets will be sent.
> >=20
> > Fixes: baabf59c2414 ("SUNRPC: Convert svc_udp_sendto() to use the per-s=
ocket bio_vec array")
> > Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
> > ---
> >  net/sunrpc/svcsock.c | 4 ++--
> >  1 file changed, 2 insertions(+), 2 deletions(-)
> >=20
> > diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
> > index 998687421fa6..e0ce4276274b 100644
> > --- a/net/sunrpc/svcsock.c
> > +++ b/net/sunrpc/svcsock.c
> > @@ -717,12 +717,12 @@ static int svc_udp_sendto(struct svc_rqst *rqstp)
> >  				ARRAY_SIZE(rqstp->rq_bvec), xdr);
> > =20
> >  	iov_iter_bvec(&msg.msg_iter, ITER_SOURCE, rqstp->rq_bvec,
> > -		      count, 0);
> > +		      count, rqstp->rq_res.len);
> >  	err =3D sock_sendmsg(svsk->sk_sock, &msg);
> >  	if (err =3D=3D -ECONNREFUSED) {
> >  		/* ICMP error on earlier request. */
> >  		iov_iter_bvec(&msg.msg_iter, ITER_SOURCE, rqstp->rq_bvec,
> > -			      count, 0);
> > +			      count, rqstp->rq_res.len);
> >  		err =3D sock_sendmsg(svsk->sk_sock, &msg);
> >  	}
> > =20
> > --=20
> > 2.43.0
> >=20
>=20
> I can't fathom why I would have chosen zero for the @count argument.
>=20
> We currently have zero test coverage for UDP. I'll look into that.
>=20
> I've applied this to the nfsd-fixes branch. It should appear in
> v6.8-rc if I can get it tested.

Thanks. For what it is worth, this fix didn't come out of pure code
inspection, but fixes a real world setup for me. While I can't claim
that I have any kind of comprehensive testing, this fix has at least
shown to fix the issue introduced in the referenced commit in my setup.

Regards,
Lucas