Closed Bug 1337131 Opened 8 years ago Closed 8 years ago

Enable restarting the GPU process when it dies

Categories

(Core :: Graphics: Layers, defect)

defect
Not set
normal

Tracking

()

RESOLVED FIXED
mozilla55
Tracking Status
firefox55 --- fixed

People

(Reporter: dvander, Assigned: dvander)

References

Details

Attachments

(1 file, 1 obsolete file)

I don't think there's any reason not to do this now - restarting has worked since the initial rollout, but we held off to minimize the test surface of the initial release.
Attached patch bug1331761.patch (obsolete) — Splinter Review
This sets the maximum number of restarts to 3 (the default is 0).
Attachment #8834133 - Flags: review?(milan)
Attached patch bug1337131.patchSplinter Review
Err, wrong patch.
Attachment #8834133 - Attachment is obsolete: true
Attachment #8834133 - Flags: review?(milan)
Attachment #8834135 - Flags: review?(milan)
Comment on attachment 8834135 [details] [diff] [review] bug1337131.patch Review of attachment 8834135 [details] [diff] [review]: ----------------------------------------------------------------- Where will we collect the telemetry for the GPU process startup time on restarts? One reason I can see for not doing this is if we have a very long GPU process startup time - it would lead to much longer pause from the user's point of view. Perhaps we need the "start SW compositor while waiting" with more urgency now? Better information on how far apart the restarts are would also help (telemetry again.) Perhaps it's worth leaving this nightly only until we sort these out? Also, do we have a bug for the follow up on a slightly more complex heuristic that takes into account frequency, and perhaps also disables the GPU process permanently until version/driver change?
Attachment #8834135 - Flags: review?(milan) → review+
(In reply to Milan Sreckovic [:milan] from comment #3) > Comment on attachment 8834135 [details] [diff] [review] > bug1337131.patch > > Review of attachment 8834135 [details] [diff] [review]: > ----------------------------------------------------------------- > > Where will we collect the telemetry for the GPU process startup time on > restarts? One reason I can see for not doing this is if we have a very long > GPU process startup time - it would lead to much longer pause from the > user's point of view. Perhaps we need the "start SW compositor while > waiting" with more urgency now? Better information on how far apart the > restarts are would also help (telemetry again.) Perhaps it's worth leaving > this nightly only until we sort these out? That's a good point. We do have the timeout, which is currently set to 3s. Is it good enough to rely on that, and maybe have a way to test that it works? > Also, do we have a bug for the follow up on a slightly more complex > heuristic that takes into account frequency, and perhaps also disables the > GPU process permanently until version/driver change? No, I will file that shortly.
Non-urgent ni? for thoughts on comment #4.
Flags: needinfo?(milan)
(In reply to David Anderson [:dvander] from comment #4) > ... > > That's a good point. We do have the timeout, which is currently set to 3s. > Is it good enough to rely on that, and maybe have a way to test that it > works? I'd say that is good enough, especially for restarts.
Flags: needinfo?(milan)
If this didn't land yet (or I suppose even if it did), can we have it stay on nightly and only ride 55? There were some issues discovered when softvision was testing with this option set to 3 and forced GPU process kills; want to make sure we have a chance to address those.
Flags: needinfo?(dvander)
Sure, this did not land yet. I'll hold off. Could those issues be filed & block this bug?
Flags: needinfo?(dvander)
Pushed by danderson@mozilla.com: https://hg.mozilla.org/integration/mozilla-inbound/rev/f553a59096ad Enable restarting the GPU process when it dies. (bug 1337131, r=milan)
Status: ASSIGNED → RESOLVED
Closed: 8 years ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla55
Depends on: 1369097
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: