From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754811AbYIWW44 (ORCPT ); Tue, 23 Sep 2008 18:56:56 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1751415AbYIWW4t (ORCPT ); Tue, 23 Sep 2008 18:56:49 -0400 Received: from smtp-out.google.com ([216.239.33.17]:3478 "EHLO smtp-out.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750884AbYIWW4s (ORCPT ); Tue, 23 Sep 2008 18:56:48 -0400 DomainKey-Signature: a=rsa-sha1; s=beta; d=google.com; c=nofws; q=dns; h=subject:from:to:cc:in-reply-to:references:content-type: organization:date:message-id:mime-version:x-mailer: content-transfer-encoding:x-gmailtapped-by; b=o/TIKzmmxuCfLga/Se3t9enMc0DhsABN4Pv4DmR+LdTJS6N/l3F16jxrZnNT94Y6P 2bsd2MHjiS54fO0kulwBQ== Subject: Re: [PATCH 2.6.27-rc5 re-resubmit] Fix itimer/many thread hang. From: Frank Mayhar To: Oleg Nesterov Cc: Ingo Molnar , linux-kernel , Roland McGrath , Thomas Gleixner , Alexey Dobriyan , Andrew Morton In-Reply-To: <20080923135917.GA285@tv-sign.ru> References: <1221238479.30136.2.camel@bobble.smo.corp.google.com> <20080914150651.GK12522@elte.hu> <20080914150923.GB26984@elte.hu> <1221502142.19012.35.camel@bobble.smo.corp.google.com> <20080916084143.GC17287@elte.hu> <1221678187.13420.17.camel@bobble.smo.corp.google.com> <20080918102353.GD20967@elte.hu> <1222114936.21579.20.camel@bobble.smo.corp.google.com> <20080923114033.GB14840@elte.hu> <20080923135917.GA285@tv-sign.ru> Content-Type: text/plain Organization: Google, Inc. Date: Tue, 23 Sep 2008 15:56:00 -0700 Message-Id: <1222210560.3429.20.camel@bobble.smo.corp.google.com> Mime-Version: 1.0 X-Mailer: Evolution 2.6.1 Content-Transfer-Encoding: 7bit X-GMailtapped-By: 172.25.146.18 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Tue, 2008-09-23 at 17:59 +0400, Oleg Nesterov wrote: > Looks like task_delta_exec() forgets to unlock rq. Actually it shouldn't be locking rq. I failed to delete the lock. > Confused. Unless I misread the patch, tgtimes->totals can be NULL > on UP too? Yeah. I totally messed up on this, too many things going on at once. I've eliminated the SMP/UP split in sched_stats.h and pulled it all into the three basic functions. > (of course, this can't explain the crash) I'll be investigating this in the morning (I'm at home today and without a test box). -- Frank Mayhar Google, Inc.