From mboxrd@z Thu Jan 1 00:00:00 1970 From: Tom St Denis Subject: Re: Feature Request: Ability to decode bus/dma address back into physical address Date: Wed, 2 Aug 2017 15:25:32 -0400 Message-ID: <76c1f638-05a4-1e37-14ad-7f5b21e47028@amd.com> References: <20170801180415.GB3443@gmail.com> <483ecda0-2977-d2ea-794c-320e429d7645@amd.com> <20170801190259.GC3443@gmail.com> <42c5fe2b-f179-cb71-03d3-7ae991543edb@amd.com> <20170801195556.GD3443@gmail.com> <77e557d2-aa75-46c4-88a7-cca5448ea08e@amd.com> <20170801210105.GE3443@gmail.com> <2cd345ee-d5ad-1ad7-508a-86225e65621c@amd.com> <20170802044214.GA6285@gmail.com> <4a2b004b-ebc2-9331-84c4-4e6672dd7b97@amd.com> <20170802164343.GA3105@gmail.com> <5ecbb5c2-fe4e-fd84-43b5-67ae06c5a032@amd.com> <2f9da712-1559-6593-e512-c0508e21d747@arm.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; Format="flowed" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <2f9da712-1559-6593-e512-c0508e21d747-5wv7dgnIgG8@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: Robin Murphy , =?UTF-8?Q?Christian_K=c3=b6nig?= , Jerome Glisse Cc: "Deucher, Alexander" , iommu-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org List-Id: iommu@lists.linux-foundation.org On 02/08/17 02:42 PM, Robin Murphy wrote: > FWIW I'm not aware of anything that does exactly what you want (other > than some trick patches I keep around that only work for arm64), but it > does come to mind that it ought to be fairly straightforward to give > dma-debug the ability to dump out the information it already captures. > It wouldn't be viable for deployment in non-development kernels, but > it's a idea - for production kernels, the simple answer is probably > "boot with iommu=off|pt when GPU debugging" anyway. The genesis of this requirement is that IOMMU is a requirement in a lot of new development (for instance, virtualization) and newer systems cannot operate with it disabled. Previously, ya I'd tell umr (our debugger) users to turn it off but that's becoming increasingly hard to sell :-) Right now our tracepoint seems to work but relatively light workloads but can become somewhat laggy on higher loads and I fear that on data center type loads will become completely unusable. Tom