From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754821AbYISQqF (ORCPT ); Fri, 19 Sep 2008 12:46:05 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1751389AbYISQpz (ORCPT ); Fri, 19 Sep 2008 12:45:55 -0400 Received: from outbound-dub.frontbridge.com ([213.199.154.16]:26888 "EHLO IE1EHSOBE003.bigfish.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751116AbYISQpy (ORCPT ); Fri, 19 Sep 2008 12:45:54 -0400 X-BigFish: VPS-22(z1039oz1432R98dR1805M936fQzz10d3izz2f39iz32i6bh43j61h) X-Spam-TCS-SCL: 0:0 X-WSS-ID: 0K7GBW2-03-TE5-01 Date: Fri, 19 Sep 2008 18:45:41 +0200 From: Joerg Roedel To: FUJITA Tomonori CC: mingo@elte.hu, linux-kernel@vger.kernel.org Subject: Re: [PATCH] remove fullflush and nofullflush in IOMMU generic option Message-ID: <20080919164541.GS24392@amd.com> References: <20080920012322R.fujita.tomonori@lab.ntt.co.jp> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Disposition: inline In-Reply-To: <20080920012322R.fujita.tomonori@lab.ntt.co.jp> User-Agent: mutt-ng/devel-r804 (Linux) X-OriginalArrivalTime: 19 Sep 2008 16:45:41.0707 (UTC) FILETIME=[272709B0:01C91A77] Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Sat, Sep 20, 2008 at 01:23:30AM +0900, FUJITA Tomonori wrote: > This patch against tip/x86/iommu virtually reverts > 2842e5bf3115193f05dc9dac20f940e7abf44c1a. But just reverting the > commit breaks AMD IOMMU so this patch also includes some fixes. NACK. > The above commit adds new two options to x86 IOMMU generic kernel boot > options, fullflush and nofullflush. But such change that affects all > the IOMMUs needs more discussion (all IOMMU parties need the chance to > discuss it): It affects only IOMMUs which use the iommu_fullflush variable. This are GART, which used it since ages, and AMD IOMMU which was introduced by the above commit. It absolutly makes sense to have command line parameters which make sense for more than one or most of the IOMMUs in x86 into 'iommu='. Ingo agreed with this, see http://lkml.org/lkml/2008/6/30/155 I agree with that too. The commit you are trying to revert here was a step into this direction. > http://lkml.org/lkml/2008/9/19/106 > > For me, adding these boot parameters doesn't make sense. May be true for nofullflush. For fullflush it makes sense to have it in the generic code since all x86 hardware IOMMUs except Calgary can make use of it. > All the hardware IOMMUs could use 'fullflush' for lazy IOTLB flushing > but Calgary doesn't support it. Intel VT-d has the different option > for it (and we can't rename it). So it might be useful for only GART > and AMD IOMMU. We can keep the current Intel option and have iommu=fullflush in parallel. It could also affect Intel IOMMU. > 'nofullflush' definitely is pointless. 'nofullflush' option doesn't > change any kernel behavior and it was added just for GART > compatibility. We should not have such generic meaningless option just > for GART. So why do you keep it in this patch then? > This patch removes the fullflush and nofullflush from the generic > kernel boot options and adds 'fullflush' support for AMD IOMMU. I disagree. Keep it in the generic code. Joerg -- | AMD Saxony Limited Liability Company & Co. KG Operating | Wilschdorfer Landstr. 101, 01109 Dresden, Germany System | Register Court Dresden: HRA 4896 Research | General Partner authorized to represent: Center | AMD Saxony LLC (Wilmington, Delaware, US) | General Manager of AMD Saxony LLC: Dr. Hans-R. Deppe, Thomas McCoy