Date: Thu, 23 Apr 2009 20:17:08 +0100
From: Jamie Lokier <jamie@shareable.org>
To: Tim Bird <tim.bird@am.sony.com>
Cc: linux kernel <linux-kernel@vger.kernel.org>,
       linux-arm-kernel <linux-arm-kernel@lists.arm.linux.org.uk>,
       Steven Rostedt <rostedt@goodmis.org>,
       Uwe =?iso-8859-1?Q?Kleine-K=F6nig?= 
	<u.kleine-koenig@pengutronix.de>,
       Ingo Molnar <mingo@redhat.com>,
       Frederic Weisbecker <fweisbec@gmail.com>,
       Russell King <rmk@arm.linux.org.uk>
Subject: Re: [PATCH] Add function graph tracer support for ARM (resend)
Message-ID: <20090423191708.GC13326@shareable.org>
References: <49F0B4FE.8070107@am.sony.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <49F0B4FE.8070107@am.sony.com>
User-Agent: Mutt/1.5.13 (2006-08-11)
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 972
Lines: 24

Tim Bird wrote:
> The use of "already_here" works around trace entry nesting
> (I believe due to instrumentation of either cpu_clock or
> raw_smp_processor_id).  Advice on the best way to deal
> with this would be appreciated.  (that is, whether
> to use the guard variable, or just un-instrument the
> offending routines).

I think it's better to keep functions instrumented where possible, if
checking for context isn't too expensive.  Sometimes it's interesting
to see how many times they are called.

But it's not much use instrumenting those functions if activity on a
different CPU suppresses it.  Is there no way to make "already_here"
task/CPU specific?

What do other architectures do?  Consistency is good.

-- Jamie
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/