From mboxrd@z Thu Jan 1 00:00:00 1970 From: John Blackwood Subject: Re: ANTS Report Date: Fri, 26 Feb 2010 08:59:21 -0600 Message-ID: <4B87E1C9.8030209@ccur.com> Mime-Version: 1.0 Content-Type: text/plain; charset="ISO-8859-1"; format=flowed Content-Transfer-Encoding: 7bit To: , Jeff Hollensen Return-path: Received: from flmx07.ccur.com ([12.192.68.12]:17605 "EHLO flmx07.ccur.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S964837Ab0BZPEd (ORCPT ); Fri, 26 Feb 2010 10:04:33 -0500 Sender: linux-rt-users-owner@vger.kernel.org List-ID: > > Friday, February 26, 2010 > > HAWKER FAILURES > 2.6.31.12-RedHawk-5.4.2-debug (32-bit): > ihawk1 shield_all For what ever it's worth, this failure was the 'invalidate_ktlb' kernel variety, not the user space 'invalidate_tlb_x' type of tlbflush. Just an idea (maybe a bad one)... I'm wondering if this shield_all test script should be changed to not fail if it sees just one invalidate_ktlb occur during a 10 second shield trace on a given cpu.... maybe this would be only for 32bit systems with HIGHMEM. Feel free to ignore this suggestion. It's just that we'll (I'll) have to look at manufacturing systems everytime this test fails, and say whether or not this is a user or kernel failure and whether to ignore it.