From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932910Ab3IDUfh (ORCPT ); Wed, 4 Sep 2013 16:35:37 -0400 Received: from aserp1040.oracle.com ([141.146.126.69]:51039 "EHLO aserp1040.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1758229Ab3IDUff (ORCPT ); Wed, 4 Sep 2013 16:35:35 -0400 Date: Wed, 4 Sep 2013 16:35:03 -0400 From: Konrad Rzeszutek Wilk To: Zoltan Kiss Cc: Jeremy Fitzhardinge , Steven Rostedt , Frederic Weisbecker , Ingo Molnar , linux-kernel@vger.kernel.org, xen-devel@lists.xensource.com, virtualization@lists.linux-foundation.org Subject: Re: [PATCH] tracing/events: Add bounce tracing to swiotbl-xen Message-ID: <20130904203502.GA8726@phenom.dumpdata.com> References: <1377208048-8963-1-git-send-email-zoltan.kiss@citrix.com> <20130823125533.GC14306@konrad-lan.dumpdata.com> <5224D0D6.40908@citrix.com> <20130903124237.GB9870@konrad-lan.dumpdata.com> <522793CA.3060002@citrix.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <522793CA.3060002@citrix.com> User-Agent: Mutt/1.5.21 (2010-09-15) X-Source-IP: acsinet21.oracle.com [141.146.126.237] Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Wed, Sep 04, 2013 at 09:10:50PM +0100, Zoltan Kiss wrote: > On 03/09/13 13:42, Konrad Rzeszutek Wilk wrote: > >Correct. The double buffering code is being run in lib/swiotlb.c not the > >xen-swiotlb.c. Hence the question of why not move the tracing in there. > I've put the trace to both locations before swiotlb_tbl_map_single > is called, so the same tracer will be hit both with classic and Xen > SWIOTLB. I used 2 different place instead of calling from > swiotlb_tbl_map_single because I want to print out dev_addr, and > it's calculated differently. > > >Yes. And please (if it adds a benefit) also for unmap/sync which can trigger > >the bounce buffer. > > > >Or if it makes sense just for the bounce buffer copying - then just > >leave it at that. Thanks. > For me the relevant event was to see when we start to do bounce > buffering. When does it end, or when syncing happens is a different > thing. If someone is interested, they can easily extend this patch > with that. > > However one question bothers me: why ftrace doesn't trace these > functions? They are not inlined, static, or marked as notrace. They are usually part of the dma_ops structure. That might be the reason. > > Regards, > > Zoli