Check the status of IOMMU usage on Chrome OS |
||||||
Issue descriptionWiFi chips can use DMA for transfers, we should see if we can enable IOMMUs for them. If so, file bugs to actually enable them.
,
Aug 2 2017
,
Aug 2 2017
,
Aug 3 2017
We are going to have to enable this carefully, one chipset at a time. I would suggest creating a list of chip sets and prioritize. My primary concern is by turning on the IOMMU, we are very likely to expose bugs (a good thing) in the existing device drivers that do DMA. "use after free" type bugs are present in many drivers and I expect we will expose them with modprobe/rmmod or suspend/resume cyclic testing. In "a previous life", I've exposed quite a list of such bugs in tulip driver and ISTR it still has several more races between DMA and "shutdown"-like activities.
,
Sep 26 2017
,
Sep 27
This issue has been Available for over a year. If it's no longer important or seems unlikely to be fixed, please consider closing it out. If it is important, please re-triage the issue. Sorry for the inconvenience if the bug really should have been left as Available. For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot
,
Oct 4
Assigning this to Mattias since he started the conversation around IOMMUs in AVL. |
||||||
►
Sign in to add a comment |
||||||
Comment 1 by jorgelo@chromium.org
, Aug 2 2017Kees says: """ AFAICT, the main problem on Chromebooks was that most SKUs lacked IOMMU support at all. The Pixels, I think, were the first to have it, but suffered from graphics problems with the IOMMU enabled. I'd be curious to see if we could enable it but blacklist the GPU ("intel_iommu=on,igfx_off") if there are still problems. """