Hi Ckristian, Thanks for your root-cause analysis.
Personally, in this situation I would prefer just zero-allocating the array in the first place, so that the zeroed bytes will follow the data wherever it may be used. (Else stale data is interpreted as GEM objects everywhere in the code) The allocated extra slots are still a bit strange to me, although it seems they may be useful after intel_engine_cmd_parser() is invoked. I've also extended the change you made from eb->buffer_count to eb->args->buffer_count, as I believe this will make no difference outside of intel_engine_cmd_parser(). The patch I've created is linked below, and I think it will solve your specific issue. If you would like, I can submit it to the upstream mailing list, as the change would have to be accepted there first before it can added to the Ubuntu kernel. Regardless, I will upload a testing kernel with this change to my personal PPA which you can test to see if the patch works for you. I'll link it here once it has built with instructions on how to test it. -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/2161098 Title: Random hard freezes caused by i915 kernel page fault in eb_relocate_parse_slow on MacBookAir6,2 (Intel HD 5000)acBookAir6,2 (Intel HD 5000) To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2161098/+subscriptions -- ubuntu-bugs mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
