From mboxrd@z Thu Jan 1 00:00:00 1970 From: Stephen Hemminger Subject: Re: TCP event tracking via netlink... Date: Wed, 5 Dec 2007 21:06:49 -0700 Message-ID: <20071205210649.5a3d1a52@shemminger-laptop> References: <20071205.053031.87154402.davem@davemloft.net> <1196873587.10819.30.camel@localhost> <20071205163338.50aff45d@shemminger-laptop> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: Joe Perches , David Miller , Netdev To: "Ilpo =?UTF-8?B?SsOkcnZpbmVu?=" Return-path: Received: from smtp2.linux-foundation.org ([207.189.120.14]:59138 "EHLO smtp2.linux-foundation.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751963AbXLFEIA convert rfc822-to-8bit (ORCPT ); Wed, 5 Dec 2007 23:08:00 -0500 In-Reply-To: Sender: netdev-owner@vger.kernel.org List-ID: On Thu, 6 Dec 2007 00:15:49 +0200 (EET) "Ilpo J=C3=A4rvinen" wrote: > On Wed, 5 Dec 2007, Stephen Hemminger wrote: >=20 > > On Wed, 05 Dec 2007 08:53:07 -0800 > > Joe Perches wrote: > >=20 > > > > it occurred to me that we might want to do something > > > > like a state change event generator. > > >=20 > > > This could be a basis for an interesting TCP > > > performance tester. > >=20 > > That is what tcpprobe does but it isn't detailed enough to address = SACK > > issues. >=20 > ...It would be nice if that could be generalized so that the probe co= uld=20 > be attached to some other functions than tcp_rcv_established instead. >=20 > If we convert remaining functions that don't have sk or tp as first=20 > argument so that sk is listed first (should be many with wrong orderi= ng=20 > if any), then maybe a generic handler could be of type: >=20 > jtcp_entry(struct sock *sk, ...) >=20 > or when available: >=20 > jtcp_entry(struct sock *sk, struct sk_buff *ack, ...) >=20 >=20 > --=20 > i. An earlier version had hooks in send as well, it is trivial to extend. = as long as the prototypes match, any function arg ordering is okay.