From mboxrd@z Thu Jan 1 00:00:00 1970 From: Nicholas Mc Guire Subject: Re: BUG - pthread conditional locks and barriers not reliable Date: Mon, 8 Apr 2013 08:20:48 +0200 Message-ID: <20130408062048.GA24078@opentech.at> References: <1364458931.4931.8.camel@debian> <1364520685.8268.2.camel@debian> <1364521778.8491.4.camel@debian> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: linux-rt-users@vger.kernel.org To: Jo?l Kr?hemann Return-path: Received: from hofr.at ([212.69.189.236]:35454 "EHLO mail.hofr.at" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752408Ab3DHG3o (ORCPT ); Mon, 8 Apr 2013 02:29:44 -0400 Content-Disposition: inline In-Reply-To: <1364521778.8491.4.camel@debian> Sender: linux-rt-users-owner@vger.kernel.org List-ID: On Fri, 29 Mar 2013, Jo?l Kr?hemann wrote: > Hi, finished with euphoria. It's an issue with pthread implementation > may be you have to extend kernel API to fix this problem. You can't wrap > pthread_barrier_wait with pthread_mutex_lock and pthread_mutex_unlock > because it has a race condition what ends probably in a deadlock. > are you using a PI mutex ? the POSIX spec actually explicitly states that barriers are not inversion free so for a non-PI mutex that could cause deadlocks in combination with mutexes. If you are not using PI could you give it a shot- if that fixes it ? thx! hofrat