From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751263AbWFPKhD (ORCPT ); Fri, 16 Jun 2006 06:37:03 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1751316AbWFPKhC (ORCPT ); Fri, 16 Jun 2006 06:37:02 -0400 Received: from cantor.suse.de ([195.135.220.2]:48040 "EHLO mx1.suse.de") by vger.kernel.org with ESMTP id S1751263AbWFPKhA (ORCPT ); Fri, 16 Jun 2006 06:37:00 -0400 From: Andi Kleen To: KAMEZAWA Hiroyuki Subject: Re: [RFC][PATCH] avoid cpu hot remove of cpus which have special RT tasks. Date: Fri, 16 Jun 2006 12:36:50 +0200 User-Agent: KMail/1.9.3 Cc: ashok.raj@intel.com, linux-kernel@vger.kernel.org References: <20060616162343.02c3ce62.kamezawa.hiroyu@jp.fujitsu.com> <20060616192654.50f4f6b7.kamezawa.hiroyu@jp.fujitsu.com> In-Reply-To: <20060616192654.50f4f6b7.kamezawa.hiroyu@jp.fujitsu.com> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200606161236.50302.ak@suse.de> Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Friday 16 June 2006 12:26, KAMEZAWA Hiroyuki wrote: > On 16 Jun 2006 11:14:57 +0200 > Andi Kleen wrote: > > > KAMEZAWA Hiroyuki writes: > > > > > > This is a bit pessimistic. But forecd migration of RT task which is bounded > > > to the special cpu will cause unpredictable trouble, I think. > > > > More trouble than running it on a CPU that is about to fail? > > Doubtful. > > > With my patch, RT tasks will continute to run. That's the problem - if the CPU is failing and you have to remove it the task will likely corrupt its data or fail in other ways if it doesn't allow it. Better to let RT tasks run a little slower on another CPU. > Assume there are some multi-threaded tasks with SCHED_FIFO. > If they uses some kind of synchronization in user land and task is migrated to > other cpus, it will cause dead-lock. If its CPU fails much worse things than that will happen. One way might be to break affinity of all processes in the system on hot unplug - then your deadlock would be avoided - but it might be a bit radical. -Andi