From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Ahern Subject: Re: [RFC] --dump-sym-trace Date: Fri, 04 Feb 2011 19:38:50 -0700 Message-ID: <4D4CB83A.2030400@cisco.com> References: <4D3F871B.3060906@cisco.com> <4D433811.7020904@cisco.com> <20110129205050.GH6345@ghostprotocols.net> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit Return-path: Received: from sj-iport-1.cisco.com ([171.71.176.70]:7742 "EHLO sj-iport-1.cisco.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753047Ab1BECiw (ORCPT ); Fri, 4 Feb 2011 21:38:52 -0500 In-Reply-To: Sender: linux-perf-users-owner@vger.kernel.org List-ID: To: Arun Sharma Cc: Arnaldo Carvalho de Melo , linux-perf-users@vger.kernel.org On 02/04/11 18:48, Arun Sharma wrote: > On Sat, Jan 29, 2011 at 12:50 PM, Arnaldo Carvalho de Melo > wrote: >> /me too >> >> I have to get back to that, it is definetely a feature we need to have. > > There are a couple of more minor nits I found in David's patch: > > * perf record --timehist doesn't seem to be doing anything. I used > perf record -T -a. For record the the --timehist flag adds the timestamp and cpu to the IP samples -- hidden attributes from a 'perf report' perspective unless you add the --timehist flag on the report as well in which case the output is completely different. I take it your mileage varied from the above. David > * Some of the printfs are missing checks along the lines of: > > --- a/tools/perf/builtin-report.c > +++ b/tools/perf/builtin-report.c > @@ -228,6 +228,7 @@ static int perf_session__print_sample(struct > perf_session *session, > free(syms); > > } else { > + if (!data || !al || !al->thread || !al->sym || > !al->map) return 0; > printf(TIMEHIST_FMT, > tstr, data->cpu, > al->thread->comm, al->thread->pid, al->addr, > > to avoid SIGSEGVs when the information in the sample is incomplete. > > -Arun