From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Hook, Gary" Subject: Re: [PATCH] iommu/amd - Use dev_err to send events to the system log Date: Wed, 14 Feb 2018 09:58:31 -0500 Message-ID: <0c024682-2621-44a3-3988-787c00aee1c1@amd.com> References: <151736066756.21649.4739957712713863861.stgit@sosam4> <20180213132216.gyhndywyu23cdzbv@8bytes.org> <325d7406-e01b-f7a1-3da6-390b0cf9bc94@amd.com> <20180213160018.fg7baclgu5o5rgxc@8bytes.org> <20180214114957.GI16484@8bytes.org> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; Format="flowed" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <20180214114957.GI16484-zLv9SwRftAIdnm+yROfE0A@public.gmane.org> Content-Language: en-US List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: iommu-bounces-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org Errors-To: iommu-bounces-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org To: Joerg Roedel Cc: iommu-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org, linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org List-Id: iommu@lists.linux-foundation.org On 2/14/2018 6:49 AM, Joerg Roedel wrote: > On Tue, Feb 13, 2018 at 02:24:56PM -0500, Hook, Gary wrote: >> Without actually running a driver and getting some debug info, I'll just >> say that my example compiled, the amd_iommu structure points to a pci_dev >> which contains a device, and the two possibilities are likely equivalent. >> >> I'll vet that, of course, but working under the assumption that they are >> equivalent, the question becomes, do you have a preference? > > They are not equivalent, your version points to the pci_dev containing > the IOMMU, my version points to the (virtual) iommu-device for that > particular iommu in the kernel. > > So the difference in the prefix is: > > your version: '0000:00:00.2:' > > vs. my version: 'ivhd0:' > > The latter is the better prefix, because it points to the iommu. A > pci_dev can contain more than one iommu, so that prefix is not specific > enough. I know that there is currently no hardware implementing multiple > iommus on one pci_dev, but it is allowed according to the spec. Excellent; specificity is good. I'll crank a new version using that device. Thank you!