From mboxrd@z Thu Jan 1 00:00:00 1970 From: Ian Campbell Subject: Re: [RFC PATCH V4 02/13] netback: add module unload function. Date: Thu, 2 Feb 2012 19:59:21 +0000 Message-ID: <1328212761.28964.77.camel@dagon.hellion.org.uk> References: <1328201363-13915-1-git-send-email-wei.liu2@citrix.com> <1328201363-13915-3-git-send-email-wei.liu2@citrix.com> <1328202524.11534.3.camel@edumazet-HP-Compaq-6005-Pro-SFF-PC> <1328203710.5553.94.camel@leeds.uk.xensource.com> <1328204936.13262.4.camel@edumazet-HP-Compaq-6005-Pro-SFF-PC> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: "Wei Liu (Intern)" , "netdev@vger.kernel.org" , "xen-devel@lists.xensource.com" , "konrad.wilk@oracle.com" To: Eric Dumazet Return-path: Received: from smtp.eu.citrix.com ([62.200.22.115]:62927 "EHLO SMTP.EU.CITRIX.COM" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932506Ab2BBT7X (ORCPT ); Thu, 2 Feb 2012 14:59:23 -0500 In-Reply-To: <1328204936.13262.4.camel@edumazet-HP-Compaq-6005-Pro-SFF-PC> Sender: netdev-owner@vger.kernel.org List-ID: On Thu, 2012-02-02 at 17:48 +0000, Eric Dumazet wrote: > Le jeudi 02 f=E9vrier 2012 =E0 17:28 +0000, Wei Liu a =E9crit : >=20 > > You're right about this. > >=20 > > But this part is destined to get wiped out (in the very near future= ?) -- > > see following patches. So I don't think it is worthy to fix this. > >=20 >=20 > Before adding new bugs, you must fix previous ones. I've never heard of this requirement before! It's a wonder anyone ever gets anything done. Anyway, I think it would be reasonable to just remove the kthread_bind call from this loop. We don't actually want/need a thread per online CP= U in any strict sense, we just want there to be some number of worker threads available and ~numcpus at start of day is a good enough approximation for that number. There have been patches floating around in the past which make the number of groups a module parameter which would also be a reasonable thing to dig out if we weren't just about to remove all this code anyway. So removing the kthread_bind is good enough for the short term, and for stable if people feel that is necessary, and we can continue in mainlin= e with the direction Wei's patches are taking us. Ian.