From mboxrd@z Thu Jan 1 00:00:00 1970 From: Neil Horman Subject: Re: [PATCH] net: add tracepoints to socket api Date: Wed, 28 Jan 2009 16:28:09 -0500 Message-ID: <20090128212809.GA19228@hmsreliant.think-freely.org> References: <20090127171857.GB14429@infradead.org> <20090127.092017.40699485.davem@davemloft.net> <20090127172323.GA7635@infradead.org> <20090127.102141.99756103.davem@davemloft.net> <20090127184704.GA26403@infradead.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: David Miller , netdev@vger.kernel.org To: Christoph Hellwig Return-path: Received: from charlotte.tuxdriver.com ([70.61.120.58]:56176 "EHLO smtp.tuxdriver.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751318AbZA1V2V (ORCPT ); Wed, 28 Jan 2009 16:28:21 -0500 Content-Disposition: inline In-Reply-To: <20090127184704.GA26403@infradead.org> Sender: netdev-owner@vger.kernel.org List-ID: On Tue, Jan 27, 2009 at 01:47:04PM -0500, Christoph Hellwig wrote: > On Tue, Jan 27, 2009 at 10:21:41AM -0800, David Miller wrote: > > The block layer already merged trace annotations. I therefore see no > > harm in merging the net bits. > > The block layer also has a consumer for them (blktrace) and soon a > second one (the ftrace plugin) > > FWIW, ftrace will also work well on these syscalls (which I think is rather usefull). Another use case came across my path the other day too. I've had several requests from various people to find a way to provide unifying frame drop indicators. Basically people hate to have to dig through various stat files and tools to look at all the places a frame can be dropped within our network stack. I think a nice solution to this problem might be some tracepoints that people could use to gather comprehensive tx/rx frame drop information using one tool. Of course, these provided tracepoints don't do that, but I think it would be good to get the existing tracepoint set in before I go poking at an implementation of that, just to ensure I avoid duplicate work Neil