From mboxrd@z Thu Jan 1 00:00:00 1970 From: John Kacur Subject: Re: tracefs .. idea Date: Tue, 16 Oct 2012 13:04:13 +0200 (CEST) Message-ID: References: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Cc: linux-rt-users@vger.kernel.org To: Lars Segerlund Return-path: Received: from mail-bk0-f46.google.com ([209.85.214.46]:54417 "EHLO mail-bk0-f46.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753558Ab2JPLCZ (ORCPT ); Tue, 16 Oct 2012 07:02:25 -0400 Received: by mail-bk0-f46.google.com with SMTP id jk13so2911690bkc.19 for ; Tue, 16 Oct 2012 04:02:23 -0700 (PDT) In-Reply-To: Sender: linux-rt-users-owner@vger.kernel.org List-ID: On Tue, 16 Oct 2012, Lars Segerlund wrote: > I was thinking of tracking locks and above all interrupt disable > through debugfs. > > Something along the lines of a buffer with the top offenders in it, > and thus getting a list of the longest latency introducing sections. > Or would it be 'better' to instrument with something like kprobes and > have a utility for this functionality ? > > Any further idea of stuff to trace ? or has this already been done > and I am unaware of it ? > > I basicly want a list of the sections in the kernel that disables > preemption for the longest time. Hi Lars - Have a look at ftrace, I think it does what you are looking for. In the kernel source read, Documentation/trace/ftrace.txt Look at "preemptoff" to see how to record the amount of time preemption is disabled. John