From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from am1ehsobe005.messaging.microsoft.com ([213.199.154.208]:8553 "EHLO AM1EHSOBE005.bigfish.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750753Ab1KRKru (ORCPT ); Fri, 18 Nov 2011 05:47:50 -0500 Date: Fri, 18 Nov 2011 11:46:51 +0100 From: Joerg Roedel To: Alex Williamson CC: , , , Subject: Re: [PATCH] iommu: Include MSI susceptibility to DMA in creating iommu groups Message-ID: <20111118104651.GJ5627@amd.com> References: <20111117170800.3125.84150.stgit@bling.home> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" In-Reply-To: <20111117170800.3125.84150.stgit@bling.home> Sender: linux-pci-owner@vger.kernel.org List-ID: On Thu, Nov 17, 2011 at 10:09:26AM -0700, Alex Williamson wrote: > IOMMU drivers should account for the platform's susceptibility to > DMA triggered MSI interrupts in creating IOMMU groups. Skip > devices when the IOMMU can't isolate MSI from DMA, but allow > an iommu=group_unsafe_msi option for opt-in. This removes the > leap in logic for users that IOMMU_CAP_INTR_REMAP is required for > interrupt security when they may be running on a non-x86 platform > that does not have this dependency. I actually don't see the point in this. The iommu-group thing is to tell user-space what devices the IOMMU can safely distinguish between. The absence of interrupt-remapping changes nothing to that grouping. So why remove it when interrupt remapping is not enabled? Joerg -- AMD Operating System Research Center Advanced Micro Devices GmbH Einsteinring 24 85609 Dornach General Managers: Alberto Bozzo, Andrew Bowd Registration: Dornach, Landkr. Muenchen; Registerger. Muenchen, HRB Nr. 43632