From mboxrd@z Thu Jan 1 00:00:00 1970 From: Zoltan Kiss Subject: Re: [PATCH net RFC] xen-netback: Fix grant ref resolution in RX path Date: Thu, 15 May 2014 11:14:25 +0100 Message-ID: <53749381.7010301@citrix.com> References: <1399991500-4432-1-git-send-email-zoltan.kiss@citrix.com> <1399997635.7973.56.camel@edumazet-glaptop2.roam.corp.google.com> <53734F9F.2010101@citrix.com> <1400142682.30687.3.camel@kazak.uk.xensource.com> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8"; format=flowed Content-Transfer-Encoding: 7bit Cc: Eric Dumazet , , , , , , , To: Ian Campbell Return-path: Received: from smtp.citrix.com ([66.165.176.89]:20786 "EHLO SMTP.CITRIX.COM" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751951AbaEOKO3 (ORCPT ); Thu, 15 May 2014 06:14:29 -0400 In-Reply-To: <1400142682.30687.3.camel@kazak.uk.xensource.com> Sender: netdev-owner@vger.kernel.org List-ID: On 15/05/14 09:31, Ian Campbell wrote: > On Wed, 2014-05-14 at 12:12 +0100, Zoltan Kiss wrote: >>> The 'cleanup' of stale ubufs should be right after __pskb_pull_tail(). >> We can't fix every place in the kernel where frags might be changed, >> especially with a netback specific stuff, so unfortunately that won't work > > Is it worth fixing up the ones in netback though so that the things > injected into the stack are consistent when we hand them over? It would > avoid some search overhead on the rx path at the other end I guess? > Perhaps not significant though. There are plans to remove that unconditional pull, as it damages performance. It is better to use during checksum setup maybe_pull_tail, and pull up whatever is needed for checksum setup (this is already done, partially). A sensible netfront would send the header in the first slot anyway, so netback won't pull, and it definitely won't pull the whole first frag. Regards, Zoli