Re: [PATCHv2 2/2] tracing,mm - add kernel pagefault tracepoint for x86 & x86_64

From: Steven Rostedt <rostedt@goodmis.org>
To: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Andi Kleen <andi@firstfloor.org>, Jiri Olsa <jolsa@redhat.com>,
	mingo@elte.hu, lwoodman@redhat.com, hch@infradead.org,
	linux-kernel@vger.kernel.org,
	Thomas Gleixner <tglx@linutronix.de>
Subject: Re: [PATCHv2 2/2] tracing,mm - add kernel pagefault tracepoint for x86 & x86_64
Date: Mon, 15 Nov 2010 10:04:18 -0500	[thread overview]
Message-ID: <1289833458.12418.601.camel@gandalf.stny.rr.com> (raw)
In-Reply-To: <20101115145359.GC5410@nowhere>

On Mon, 2010-11-15 at 15:54 +0100, Frederic Weisbecker wrote:
> On Mon, Nov 15, 2010 at 03:06:33PM +0100, Andi Kleen wrote:

> > For tracing the whole page fault me think it's better to have 
> > a generalized exception tracer with a filter on page fault.
> 
> 
> You're right. A tracepoint in handle_mm_fault() would be perhaps
> better. It should catch most tracepoints the users are interested
> in. On the other hand we may miss part of the page fault
> latency, like the mmap_sem contention. This can be measured using
> lock events though.
> 
> But I'm probably missing other important things.

I would have the general exception handler as a tracepoint that would
have a "stable tracepoint" hook to it. That is, when enabling it as a
infield debugging tracepoint, we would get the "raw tracepoint", but the
stable tracepoint would massage it into different general types of
exceptions, and the code there would do the filtering.

-- Steve