From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Ahern Subject: Re: timing information with perf Date: Thu, 12 Nov 2015 16:22:35 -0700 Message-ID: <56451F3B.8020509@gmail.com> References: <94D0C06D-C3A7-4BD1-9898-124EE0AA923D@netapp.com> <20151112215728.GD7160@kernel.org> <21ECAC65-E51A-4430-A499-FFE216BF856B@netapp.com> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: Received: from mail-io0-f175.google.com ([209.85.223.175]:33136 "EHLO mail-io0-f175.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754386AbbKLXWg (ORCPT ); Thu, 12 Nov 2015 18:22:36 -0500 Received: by iouu10 with SMTP id u10so73966803iou.0 for ; Thu, 12 Nov 2015 15:22:36 -0800 (PST) In-Reply-To: <21ECAC65-E51A-4430-A499-FFE216BF856B@netapp.com> Sender: linux-perf-users-owner@vger.kernel.org List-ID: To: "Kornievskaia, Olga" , Arnaldo Carvalho de Melo Cc: "linux-perf-users@vger.kernel.org" On 11/12/15 4:09 PM, Kornievskaia, Olga wrote: > Just to elaborate further, what I=E2=80=99m interested in timings of = NFS operations, so I=E2=80=99m not using any new tracepoints, but just = the ones that are present in the code. However since NFS doesn=E2=80=99= t various things outside of the process calling into NFS then will =E2=80= =9Cperf trace=E2=80=9D still work? > > I=E2=80=99m using =E2=80=9Cperf record -e =E2=80=98nfs4:*=E2=80=99,=E2= =80=99nfs:*=E2=80=99 -ag=E2=80=9D command to generate tracing info. Tha= t captures tracepoints executed by the application as well as the kerne= l threads that do asynchronous IO. Do you really need callchains? (-g option) After the perf-record run perf-script to dump the samples. Pipe that=20 into an awk script (or python or bash or ...) and you can compute what=20 you want.