From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755923Ab2DETTe (ORCPT ); Thu, 5 Apr 2012 15:19:34 -0400 Received: from mail.windriver.com ([147.11.1.11]:58207 "EHLO mail.windriver.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755299Ab2DETTd (ORCPT ); Thu, 5 Apr 2012 15:19:33 -0400 From: Mark Asselstine To: CC: , Subject: [PATCH 0/3] trace-cmd fixes for latency tracing record and report Date: Thu, 5 Apr 2012 15:19:43 -0400 Message-ID: <1333653586-3379-1-git-send-email-mark.asselstine@windriver.com> X-Mailer: git-send-email 1.7.5.4 MIME-Version: 1.0 Content-Type: text/plain X-Originating-IP: [128.224.146.23] Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org The following three patches are for the latency-cmd tree. trace-cmd latency tracing is broken due to segfaults and disappearing 'trace' logs. Recent commits seem to have broken latency tracers by changing when things are being initialized or leaving some things uninitialized since they are unused in certain circumstances. The first two patches address segfaults when running both the record and report commands caused by attempts to access uninitialized pointers. The last patch avoids empty reports due to the premature clearing of the 'trace' log caused by resetting the current_tracer to 'nop'. We still set the 'current_tracer' to 'nop', just after we have grabbed the data we want. These changes have been tested on top of the latest trace-cmd tree as well as on top of the trace-cmd-v1.2 tag point in conjunction with the latest mainline kernel as well as an older 2.6.34 vintage kernel. Tests were done with # ./trace-cmd record -p irqsoff ls /bin # ./trace-cmd report Regards, Mark Asselstine