From mboxrd@z Thu Jan 1 00:00:00 1970 From: Darren Hart Subject: Re: RFC: methods of sched event tracing and my assumptions Date: Wed, 02 Sep 2009 14:09:10 -0700 Message-ID: <4A9EDEF6.2000002@us.ibm.com> References: <4A9C1371.6050101@us.ibm.com> <4A9C5600.8060007@us.ibm.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: Peter Zijlstra , Steven Rostedt , Thomas Gleixner , Dinakar Guniguntala , "lkml, " To: linux-rt-users Return-path: Received: from e8.ny.us.ibm.com ([32.97.182.138]:47013 "EHLO e8.ny.us.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751130AbZIBVJL (ORCPT ); Wed, 2 Sep 2009 17:09:11 -0400 In-Reply-To: <4A9C5600.8060007@us.ibm.com> Sender: linux-rt-users-owner@vger.kernel.org List-ID: Darren Hart wrote: > > After some gentle resistance to CVS and non-single-file testcases, I > prepared a standalone prio-wake-min.c. It ran to completion on linus > with slightly fewer wakeups than the LTP version (I removed a number of > things that weren't strictly necessary for this specific exercise), png > attached. It's hung part way through the scripts on rt though. I'll be > looking into that, but in case you prefer to look at this version, it is > available here: > > http://dvhart.com/darren/linux/prio-wake-min.c I have just now updated this test to fix a possible race when using a requeue PI enabled glibc (likely a bug in glibc) which also reduces the tests runtime from 1s to about 15ms (yeah, the sleep loops were rather stupid in the original version :/). Thanks, -- Darren Hart IBM Linux Technology Center Real-Time Linux Team