Closed Bug 1403457 Opened 7 years ago Closed 7 years ago

Intermittent bugs/579323-1.html == bugs/579323-1-ref.html | image comparison, max difference: 255, number of differing pixels: 10200

Categories

(Core :: Graphics: WebRender, defect, P1)

Product:
Core
Component:
Graphics: WebRender
Type:
defect

Tracking

()

Status:
VERIFIED FIXED
Milestone:
mozilla58
Tracking Flags:
Tracking Status
firefox58 --- fixed

People

(Reporter: intermittent-bug-filer, Assigned: kats)

References

Details

(Keywords: intermittent-failure, Whiteboard: [wr-mvp][stockwell disabled])

Component: Layout → Graphics: WebRender
Ok, I'll mark it failing for now and bisect what caused it.
Assignee: nobody → bugmail
Pushed by kgupta@mozilla.com: https://hg.mozilla.org/integration/mozilla-inbound/rev/c9d461081948 Mark a test as fails-if(webrender). r=me
This is a regression from bug 1383786. Try push with merge of 4d35361e1aec and 90fe4a380e8b: https://treeherder.mozilla.org/#/jobs?repo=try&revision=35a6c69a8399f177e29971f73ce210adf584e04a Try push with merge of 4d35361e1aec and 594f4ea5d802: https://treeherder.mozilla.org/#/jobs?repo=try&revision=9e98c2e5f97c024309a1594f37871b3bbbef56db Nical, fyi. It seems your patches combined with the layers-free flip caused this perma-fail.
Blocks: 1383786
Flags: needinfo?(nical.bugzilla)
Status: NEW → ASSIGNED
Priority: P5 → P1
Whiteboard: [wr-mvp]
Whiteboard: [wr-mvp] → [wr-mvp][stockwell disabled]
I don't think we need to leave this open. We have a big pile of fails-if(webrender) and skip-if(webrender) tests that I'm tracking externally in a spreadsheet, and we'll fix this as we tackle the rest of them.
Keywords: leave-open
https://hg.mozilla.org/mozilla-central/rev/c9d461081948
Status: ASSIGNED → RESOLVED
Closed: 7 years ago
status-firefox58: --- → fixed
Resolution: --- → FIXED
Target Milestone: --- → mozilla58
> Nical, fyi. It seems your patches combined with the layers-free flip caused this perma-fail. That's not surprising, my patch queue from that bug most likely changed timings a bit in the way images are received from the ImageBirdge. I think that the problem is that canvas frames should not be sent to the compositor process using ImageBridge as they are done with layers-free, because it breaks frame consistency (can't know when a canvas update will show up on screen). I bet a lot of the canvas reftests are flacky at the moment.
Flags: needinfo?(nical.bugzilla)
I am referring to bug 1396588.
Depends on: 1396588
OF shows this is down to zero now.
Status: RESOLVED → VERIFIED
You need to log in before you can comment on or make changes to this bug.