From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752403Ab3FYJfq (ORCPT ); Tue, 25 Jun 2013 05:35:46 -0400 Received: from 8bytes.org ([85.214.48.195]:40678 "EHLO mail.8bytes.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752123Ab3FYJfp (ORCPT ); Tue, 25 Jun 2013 05:35:45 -0400 Date: Tue, 25 Jun 2013 11:35:42 +0200 From: Joerg Roedel To: James Georgas Cc: linux-kernel@vger.kernel.org Subject: Re: Integrated MegaRAID controller and AMD IOMMU issue Message-ID: <20130625093542.GR11309@8bytes.org> References: MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.21 (2010-09-15) X-DSPAM-Result: Whitelisted X-DSPAM-Processed: Tue Jun 25 11:35:43 2013 X-DSPAM-Confidence: 0.9955 X-DSPAM-Probability: 0.0000 X-DSPAM-Signature: 51c9646f20864878855447 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Hi James, On Tue, Jun 25, 2013 at 12:12:53AM -0600, James Georgas wrote: > I discussed the issue with an LSI engineer, and he said that since the > integrated controller uses system memory, which it gets from the BIOS > during initialization, there will be issues with the IOMMU. I assume > he is referring to DMA address problems due to IOMMU translation. Looks like the problem is in the LSI driver. The driver needs to dma-map the system memory it gets from the BIOS (for example with dma_map_single) and use the dma_handle returned for transactions. Otherwise the IOMMU will block the accesses from the device and make it unusable. Joerg