According to comment #106 it looks like the bug is still happening also in 5.3.0-52-generic, so maybe we have just reduced the probability to hit the bug.
Let's try a different approach. Assuming that the bug wasn't happening in -45 and it started to happen in -46, the bug is likely to be introduced by: 411b24761d28 "drm/i915: Schedue request retirement in execlists_user_end()" that was supposed to fix a power consumption issue. The alternative approach that I would like to try is to revert all the commits that I backported and revert also 411b24761d28. This would re- introduce the power consumption issue, but if we are not hitting the GPU HANG anymore, it would tell us the exact commit that introduced the bug and we can figure out an alternative way of fixing the power consumption problem. So, I've upload a new test kernel (5.3.0-52.46+lp1861395v1) with the changes mentioned above: https://kernel.ubuntu.com/~arighi/LP-1861395/5.3.0-52.46+lp1861395v1/ It would be great if someone could give it a try and report if they are still experiencing the GPU HANG bug also with this new test kernel. Thanks! -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/1872001 Title: 5.3.0-46-generic - i915 - frequent GPU hangs / resets rcs0 To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1872001/+subscriptions -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
