From mboxrd@z Thu Jan 1 00:00:00 1970 From: Eric Dumazet Subject: Re: [Bugme-new] [Bug 38102] New: BUG kmalloc-2048: Poison overwritten Date: Tue, 05 Jul 2011 06:14:18 +0200 Message-ID: <1309839258.2720.17.camel@edumazet-laptop> References: <20110629145135.5179fb0c.akpm@linux-foundation.org> <1309707971.2523.20.camel@edumazet-laptop> <20110704130531.37cf876e@Nokia-N900> <1309787822.2247.6.camel@edumazet-HP-Compaq-6005-Pro-SFF-PC> <20110704162726.4072e715@maggie> <20110704164351.338dc12e@maggie> <1309792323.2247.33.camel@edumazet-HP-Compaq-6005-Pro-SFF-PC> <1309837443.2720.8.camel@edumazet-laptop> <1309839068.2720.15.camel@edumazet-laptop> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: Michael =?ISO-8859-1?Q?B=FCsch?= , Andrew Morton , netdev@vger.kernel.org, Gary Zambrano , bugme-daemon@bugzilla.kernel.org, "David S. Miller" , Pekka Pietikainen , Florian Schirmer , Felix Fietkau , Michael Buesch To: Alexey Zaytsev Return-path: Received: from mail-ww0-f44.google.com ([74.125.82.44]:53582 "EHLO mail-ww0-f44.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751002Ab1GEEOX (ORCPT ); Tue, 5 Jul 2011 00:14:23 -0400 Received: by wwe5 with SMTP id 5so5537462wwe.1 for ; Mon, 04 Jul 2011 21:14:22 -0700 (PDT) In-Reply-To: <1309839068.2720.15.camel@edumazet-laptop> Sender: netdev-owner@vger.kernel.org List-ID: Le mardi 05 juillet 2011 =C3=A0 06:11 +0200, Eric Dumazet a =C3=A9crit = : > Le mardi 05 juillet 2011 =C3=A0 07:56 +0400, Alexey Zaytsev a =C3=A9c= rit : > > On Tue, Jul 5, 2011 at 07:44, Eric Dumazet = wrote: > > > > > > I dont care about duplicate acks at this point. > > > > > > Thats a separate issue (TCP layer) > > > > >=20 > > Maybe some tx packets are just sent out more then once? Or a single > > packet is sent out instead of some other packets? > > The delays between two dups is short, and they come in bursts, up t= o a > > few hundreds of duplicate packets at a time. > >=20 >=20 > Thats a completely different problem. SSH is very expensive for your > receiver (your dump1 file has small packets (560 bytes)), and it cann= ot > cope with the stress. >=20 > You're filling the b44 rx ring, and b44 driver has no choice to zap 2= 00 > packets at once. This sure is a problem for tcp, as it stalls the thi= ng. >=20 > You could avoid this by doing this at b44 machine (the receiver) >=20 > echo "4096 32768 87380" >/proc/sys/net/ipv4/tcp_rmem >=20 > So that sender wont be able to push so many packets You can also try using more packets in rx ring : (default is 200 packets, limit ~511) ethtool -G eth0 rx 400