From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 32551 invoked by alias); 7 Aug 2002 07:51:50 -0000 Mailing-List: contact ecos-discuss-help@sources.redhat.com; run by ezmlm Precedence: bulk List-Subscribe: List-Archive: List-Post: List-Help: , Sender: ecos-discuss-owner@sources.redhat.com Received: (qmail 32543 invoked from network); 7 Aug 2002 07:51:48 -0000 Received: from unknown (HELO blackstar.vidisys.com) (194.25.115.114) by sources.redhat.com with SMTP; 7 Aug 2002 07:51:48 -0000 Received: from doms.vidisys.com (doms.vidisys.com [192.168.3.4]) by blackstar.vidisys.com (8.11.3/8.11.3/SuSE Linux 8.11.1-0.5) with ESMTP id g777qOL03439; Wed, 7 Aug 2002 09:52:24 +0200 Received: (from larice@localhost) by doms.vidisys.com (8.11.3/8.11.3/SuSE Linux 8.11.1-0.5) id g777qNF03831; Wed, 7 Aug 2002 09:52:23 +0200 From: Larice Robert Message-Id: <200208070752.g777qNF03831@doms.vidisys.com> In-Reply-To: from Thomas BINDER at "Aug 6, 2002 05:05:39 pm" To: Thomas BINDER Date: Wed, 07 Aug 2002 00:51:00 -0000 CC: ecos-discuss@sources.redhat.com MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Subject: Re: [ECOS] Possible Bug in Cyg_Alarm ... X-SW-Source: 2002-08/txt/msg00085.txt.bz2 > Hi folks! > I think I might have discovered a bug in the alarm handling of ECOS. If two (or more) alarms are set up to fire at the very same counter value, only one (the first that is initialized) istriggered , even if an interval is specified. > The problem is due to the fact that an alarm is only added to a Cyg_Counter if its trigger value is > (instead of >=) than the counter value. hello Thomas, a tiny improvement of your problem description is to say: Cyg_Counter::add_alarm() fails to add a given alarm to alarm_list (CYGIMP_KERNEL_COUNTERS_SORT_LIST implementation) if there happens to be an alarm for the same time at the tail of alarm_list (note the tail aspect, if you just look for > versus >= one would not understand why this should make any difference, unless one checks for the end of the do .. while condition) i vote for your patch. but it is necessairy to check whether there is somewhere a specification which gurantees the execution order of alarm callbacks. your patch reverses this order. Robert Larice -- Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos and search the list archive: http://sources.redhat.com/ml/ecos-discuss