Closed Bug 1356013 Opened 4 years ago Closed 4 years ago

Intermittent font-face/media-query-add-1.html == http://localhost:39929/1492020038431/238/font-face/media-query-add-1.html | image comparison, max difference: 255, number of differing pixels: 239

Categories

(Core :: CSS Parsing and Computation, defect, P5)

defect

Tracking

()

RESOLVED WORKSFORME

People

(Reporter: intermittent-bug-filer, Unassigned)

References

Details

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

Attachments

(1 file)

Priority: -- → P5
as we are hitting 100+/week and no information in the bug regarding a pending fix, lets disable this until we can get a fix
Attachment #8859996 - Flags: review?(gbrown)
Attachment #8859996 - Flags: review?(gbrown) → review+
Pushed by jmaher@mozilla.com:
https://hg.mozilla.org/integration/mozilla-inbound/rev/b4c3713d1933
Intermittent font-face/media-query-add-1.html. skip on stylo. r=gbrown
Keywords: leave-open
Whiteboard: [stockwell needswork] → [stockwell disabled]
Sorry for the radio silence here. This is probably uncovered by those media query fixes you mentioned, but seems more related to @font-face (there are similar disabled tests right before it).

I'm not aware of the details of that, Xidorn, any idea here?
Flags: needinfo?(emilio+bugs) → needinfo?(xidorn+moz)
@font-face rules are gathered from ServoStyleSet::AppendFontFaceRules() [1] for Gecko code, and they are collected in Stylist::rebuild() [2], so we need to rebuild Stylist for device change to make it work properly.

It seems to me that mStylistMayNeedRebuild isn't set for device changes, and thus MaybeRebuildStylist doesn't seem to actually update the stylist when calling AppendFontFaceRules. Could that be an issue?

Anything outside AppendFontFaceRules() should be shared between Gecko and Stylo for layout, and thus shouldn't cause such mismatch I suppose...


[1] http://searchfox.org/mozilla-central/rev/06b337ab22d40e12c4cde8040b23147ea636cc6e/layout/style/ServoStyleSet.cpp#1050
[2] http://searchfox.org/mozilla-central/rev/06b337ab22d40e12c4cde8040b23147ea636cc6e/servo/components/style/stylist.rs#306
Flags: needinfo?(xidorn+moz) → needinfo?(emilio+bugs)
Hmm... how doesn't it?

In nsPresContext::MediumFeatureValuesChanged, we end up calling RebuildAllStyleData[1].

That goes, through a series of hops, to ServoStyleSet::RebuildData(), which calls into servo and ends up at reset_device[2], which does a synchronous flush. So I'm not sure how could that happen.

That being said, the code has changed a bit lately, and this test has been disabled for ~20 days already, so perhaps it's not an issue anymore and we should try to re-enable it?

[1]: http://searchfox.org/mozilla-central/rev/06b337ab22d40e12c4cde8040b23147ea636cc6e/layout/base/nsPresContext.cpp#2071
[2]: http://searchfox.org/mozilla-central/rev/06b337ab22d40e12c4cde8040b23147ea636cc6e/servo/components/style/gecko/data.rs#93
Flags: needinfo?(emilio+bugs) → needinfo?(xidorn+moz)
Locally I'm seeing an unexpected scrollbar...

Perhaps this test consistently fails because of that :/

Given it is still not passing, I guess it isn't worth trying re-enabling it at the moment.
Flags: needinfo?(xidorn+moz)
No longer blocks: stylo-bustage
Resolving intermittent Stylo test failure bugs that haven't recurred within the last month.
Status: NEW → RESOLVED
Closed: 4 years ago
Resolution: --- → WORKSFORME
Removing leave-open keyword from resolved bugs, per :sylvestre.
Keywords: leave-open
You need to log in before you can comment on or make changes to this bug.