From mboxrd@z Thu Jan 1 00:00:00 1970 From: Arnaldo Carvalho de Melo Subject: Re: [PATCH] msi: skip calling pci_find_capability from msi_set_mask_bits Date: Mon, 19 May 2008 17:26:54 -0300 Message-ID: <20080519202654.GE17967@ghostprotocols.net> References: <20080515160426.GD14846@ghostprotocols.net> <200805161044.34938.jbarnes@virtuousgeek.org> <48310691.9050706@jp.fujitsu.com> <200805191311.46162.jbarnes@virtuousgeek.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Hidetoshi Seto , Matthew Wilcox , linux-kernel@vger.kernel.org, linux-pci@vger.kernel.org, linux-rt-users@vger.kernel.org To: Jesse Barnes Return-path: Received: from mx1.redhat.com ([66.187.233.31]:48056 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1763284AbYESU1A (ORCPT ); Mon, 19 May 2008 16:27:00 -0400 Content-Disposition: inline In-Reply-To: <200805191311.46162.jbarnes@virtuousgeek.org> Sender: linux-rt-users-owner@vger.kernel.org List-ID: Em Mon, May 19, 2008 at 01:11:45PM -0700, Jesse Barnes escreveu: > On Sunday, May 18, 2008 9:48 pm Hidetoshi Seto wrote: > > Jesse Barnes wrote: > > > Since we know MSI is a problem, let's just go with fixing that for now. > > > If we find that other caps are also causing problems we can revisit > > > caching all of them; the patch is simple enough. > > > > Humm... > > I suppose it can be more simple. How about this patch? > > > > > everytime handle_edge_irq is called it needs to mask and unmask MSI, and > > > that leads to a series of very expensive calls to pci_find_capability > > > > The position of MSI capability is already cached in the msi_desc when > > we enter the msi_set_mask_bits(). Use it instead. > > > > Signed-off-by: Hidetoshi Seto > > Yeah, this looks really nice. It should also fix Arnaldo's latency problem, > and really looks like a bug fix for the MSI code more than anything. > > Arnaldo, can you take a look & test and ack/nack? I'll do that now - Arnaldo