From mboxrd@z Thu Jan 1 00:00:00 1970 From: Alban Crequy Subject: Re: [PATCH] AF_UNIX: Implement SO_TIMESTAMP and SO_TIMETAMPNS on Unix sockets Date: Mon, 4 Oct 2010 19:00:49 +0100 Message-ID: <20101004190049.783f14ab@chocolatine.cbg.collabora.co.uk> References: <1286206706-25733-1-git-send-email-alban.crequy@collabora.co.uk> <1286210504.18293.370.camel@edumazet-laptop> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: Lennart Poettering , "David S. Miller" , Stephen Hemminger , Cyrill Gorcunov , Alexey Dobriyan , linux-kernel@vger.kernel.org, netdev@vger.kernel.org, Alban Crequy To: Eric Dumazet Return-path: In-Reply-To: <1286210504.18293.370.camel@edumazet-laptop> Sender: linux-kernel-owner@vger.kernel.org List-Id: netdev.vger.kernel.org Le Mon, 04 Oct 2010 18:41:44 +0200, Eric Dumazet a =C3=A9crit : > Le lundi 04 octobre 2010 =C3=A0 16:38 +0100, Alban Crequy a =C3=A9cri= t : > > Userspace applications can already request to receive timestamps > > with: setsockopt(sockfd, SOL_SOCKET, SO_TIMESTAMP, ...) > >=20 > > Although setsockopt() returns zero (success), timestamps are not > > added to the ancillary data. This patch fixes that. > >=20 > > Signed-off-by: Alban Crequy > > --- > > net/unix/af_unix.c | 4 ++++ > > 1 files changed, 4 insertions(+), 0 deletions(-) > >=20 > > diff --git a/net/unix/af_unix.c b/net/unix/af_unix.c > > index 617bea4..142ccea 100644 > > --- a/net/unix/af_unix.c > > +++ b/net/unix/af_unix.c > > @@ -1697,6 +1697,8 @@ static int unix_dgram_recvmsg(struct kiocb > > *iocb, struct socket *sock, if (err) > > goto out_free; > > =20 > > + sock_recv_timestamp(msg, sk, skb); > > + > > if (!siocb->scm) { > > siocb->scm =3D &tmp_scm; > > memset(&tmp_scm, 0, sizeof(tmp_scm)); > > @@ -1877,6 +1879,8 @@ static int unix_stream_recvmsg(struct kiocb > > *iocb, struct socket *sock, copied +=3D chunk; > > size -=3D chunk; > > =20 > > + sock_recv_timestamp(msg, sk, skb); > > + > > /* Mark read part of skb as used */ > > if (!(flags & MSG_PEEK)) { > > skb_pull(skb, chunk); >=20 > Are you sure its needed for unix_stream case ? I was about to say it does not cost anything to add it... > We dont do this for TCP for example, only for datagrams. >=20 > As shown in the past, sock_recv_timestamp() is a bit expensive becaus= e > it takes care of many possible options. Is it really expensive? It looks like a few flag checks in an inline function to me. > It would be better to use in AF_UNIX case (only software timestamps) = : >=20 > Solution 1)=20 > if (sock_flag(sk, SOCK_RCVTSTAMP)) > __sock_recv_timestamp(msg, sk, skb); >=20 > Solution 2)=20 > Or something already used elsewhere since 2.6.35 and commit > 767dd03369ac1 (net: speedup sock_recv_ts_and_drops()) : >=20 > sock_recv_ts_and_drops(msg, sk, skb); >=20 >=20 > I would vote for the 1) solution The patch in the next email implements the 1) solution on SOCK_DGRAM an= d SOCK_SEQPACKET (without SOCK_STREAM). --=20 Alban