From mboxrd@z Thu Jan 1 00:00:00 1970 From: Patrick McHardy Subject: Re: [DCCP] [RFC] [Patchv2 1/1]: Queuing policies -- reworked version of Tomasz's patch set Date: Tue, 22 Apr 2008 06:56:26 +0200 Message-ID: <480D6FFA.80205@trash.net> References: <200804111224.06958.tomasz@grobelny.oswiecenia.net> <480C7E50.3050508@trash.net> <20080421131253.GB12221@ghostprotocols.net> <200804211817.01189.tomasz@grobelny.oswiecenia.net> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: Arnaldo Carvalho de Melo , Gerrit Renker , dccp@vger.kernel.org, netdev@vger.kernel.org To: Tomasz Grobelny Return-path: Received: from stinky.trash.net ([213.144.137.162]:58985 "EHLO stinky.trash.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751221AbYDVE4c (ORCPT ); Tue, 22 Apr 2008 00:56:32 -0400 In-Reply-To: <200804211817.01189.tomasz@grobelny.oswiecenia.net> Sender: netdev-owner@vger.kernel.org List-ID: Tomasz Grobelny wrote: > Dnia Monday 21 of April 2008, Arnaldo Carvalho de Melo napisa=C5=82: >> Nod, if we don't need the space reserved for the lower layer protoco= ls >> in DCCP it is actually the best solution, as we don't need to zero t= he >> cb again before passing it to IP, it gets zeroed at alloc_skb time a= nd >> that is it. If we need the space, we have to pay the price of >> memset before passing to IP. >> > Ok, so in this case the patch for DCCP could be reverted in test tree= , is that=20 > right? Were these two deleted memsets zeroing all that was necessary = or were=20 > there any other bugs fixed by the patch? No, those two memsets became unnecessary by the addition of the new cb members. If you want to remove them again, you need to add those memsets back and additionally add memsets that zero the first sizeof(inet_skb_parm)/ sizeof(inet6_skb_parm) bytes everywhere else where packets are passed to IP(v6).