From mboxrd@z Thu Jan 1 00:00:00 1970 From: Dong Liu Subject: Re: cpu stall and hyperthread Date: Fri, 06 Jul 2012 02:07:12 -0400 Message-ID: <4FF68090.5050605@gmail.com> References: <4FEBCDDE.60503@gmail.com> <4FF351C6.3000100@gmail.com> <4FF4C94A.6040105@osadl.org> <4FF5DFDC.9010501@gmail.com> <4FF619D5.3090206@osadl.org> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: linux-rt-users@vger.kernel.org To: Carsten Emde Return-path: Received: from mail-qa0-f46.google.com ([209.85.216.46]:44381 "EHLO mail-qa0-f46.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751309Ab2GFGHQ (ORCPT ); Fri, 6 Jul 2012 02:07:16 -0400 Received: by qadb17 with SMTP id b17so241542qad.19 for ; Thu, 05 Jul 2012 23:07:15 -0700 (PDT) In-Reply-To: <4FF619D5.3090206@osadl.org> Sender: linux-rt-users-owner@vger.kernel.org List-ID: Hi Carsten, Thanks! I'll try to take a look at these systems. Cheers! Dong On 7/5/12 6:48 PM, Carsten Emde wrote: > Hi Dong, > >> I tried to disable CONFIG_RT_GROUP_SCHED, but I still got cpu stalls! > Hmm, sorry. I hoped it would work. > >> Is there a recommended config for the RT patch? > Not really. In the x86 distro case, we base everything on the distro > configuration, e.g. using > cp /boot/config-`uname -r` .config > > We then disable CONFIG_RT_GROUP_SCHED and CONFIG_DEBUG_STACKOVERFLOW, > because they are known to cause trouble and long latencies, > respectively. In addition, we enable diagnostics such a latency > histograms. That's all. > > An alternative approach would be to select a QA farm system that is > close to your hardware and distro and use the same settings on your > system. A list of all farm systems sorted by distro is here -> > https://www.osadl.org/?id=1070; clicking on the system name in the first > row of this table will display its entire profile. This includes kernel > parameter, hardware data and, of course, the kernel configuration. > Sometimes, there is also a quilt series, in case we temporarily use > off-tree patches that are on their way to mainline. Would be interesting > to see whether your system also stalls when a configuration is used that > is successful here. > > -Carsten.