Closed Bug 1460574 Opened 8 years ago Closed 7 years ago

Intermittent font-inflation/list-1.html == font-inflation/list-1-ref.html | image comparison, max difference: 255, number of differing pixels: 192211

Categories

(Core :: Graphics: WebRender, defect, P3)

Product:
Core
Component:
Graphics: WebRender
Type:
defect

Tracking

()

Status:
RESOLVED FIXED

People

(Reporter: intermittent-bug-filer, Unassigned)

References

Details

(Keywords: intermittent-failure, Whiteboard: [gfx-noted])

Filed by: ncsoregi [at] mozilla.com https://treeherder.mozilla.org/logviewer.html#?job_id=177839458&repo=mozilla-inbound https://queue.taskcluster.net/v1/task/CI0a8NtpTfqQkXP7VCOSsA/runs/0/artifacts/public/logs/live_backing.log https://hg.mozilla.org/mozilla-central/raw-file/tip/layout/tools/reftest/reftest-analyzer.xhtml#logurl=https://queue.taskcluster.net/v1/task/CI0a8NtpTfqQkXP7VCOSsA/runs/0/artifacts/public/logs/live_backing.log&only_show_unexpected=1 04:09:22 INFO - REFTEST TEST-START | file:///Users/cltbld/tasks/task_1525950312/build/tests/reftest/tests/layout/reftests/font-inflation/list-1.html == file:///Users/cltbld/tasks/task_1525950312/build/tests/reftest/tests/layout/reftests/font-inflation/list-1-ref.html 04:09:22 INFO - REFTEST INFO | RESTORE PREFERENCE pref(dom.meta-viewport.enabled,false) 04:09:22 INFO - REFTEST INFO | SET PREFERENCE pref(dom.meta-viewport.enabled,true) 04:09:22 INFO - REFTEST INFO | SET PREFERENCE pref(font.size.inflation.emPerLine,15) 04:09:22 INFO - REFTEST INFO | SET PREFERENCE pref(font.size.inflation.forceEnabled,true) 04:09:22 INFO - REFTEST INFO | SET PREFERENCE pref(font.size.inflation.lineThreshold,100) 04:09:22 INFO - REFTEST TEST-LOAD | file:///Users/cltbld/tasks/task_1525950312/build/tests/reftest/tests/layout/reftests/font-inflation/list-1.html | 56 / 89 (62%) 04:09:22 INFO - ++DOMWINDOW == 102 (0x12a9c1c00) [pid = 1920] [serial = 168] [outer = 0x11893a600] 04:09:22 INFO - REFTEST INFO | RESTORE PREFERENCE pref(font.size.inflation.lineThreshold,400) 04:09:22 INFO - REFTEST INFO | RESTORE PREFERENCE pref(font.size.inflation.forceEnabled,false) 04:09:22 INFO - REFTEST INFO | RESTORE PREFERENCE pref(font.size.inflation.emPerLine,0) 04:09:22 INFO - REFTEST INFO | RESTORE PREFERENCE pref(dom.meta-viewport.enabled,false) 04:09:22 INFO - REFTEST INFO | SET PREFERENCE pref(dom.meta-viewport.enabled,true) 04:09:22 INFO - REFTEST TEST-LOAD | file:///Users/cltbld/tasks/task_1525950312/build/tests/reftest/tests/layout/reftests/font-inflation/list-1-ref.html | 56 / 89 (62%) 04:09:22 INFO - ++DOMWINDOW == 103 (0x12a9bb800) [pid = 1920] [serial = 169] [outer = 0x11893a600] 04:09:23 INFO - REFTEST TEST-UNEXPECTED-FAIL | file:///Users/cltbld/tasks/task_1525950312/build/tests/reftest/tests/layout/reftests/font-inflation/list-1.html == file:///Users/cltbld/tasks/task_1525950312/build/tests/reftest/tests/layout/reftests/font-inflation/list-1-ref.html | image comparison, max difference: 255, number of differing pixels: 192211
Component: Layout: Text → Graphics: WebRender
Blocks: stage-wr-backlog
Whiteboard: [gfx-noted]
This seems to be pretty high frequency if not permanent. The analyser shows the test image is using black boxes for the glyphs but the reference is not. Considering the test page uses the Ahem font that's what I would expect so I'll looka little closer at what the test is actually doing.
Assignee: nobody → kats
Removing the spurious ahem-font goo seems to help: https://treeherder.mozilla.org/#/jobs?repo=try&selectedJob=206874410&revision=900130bfd3eccd5d28e496b91f40ce5aae6cb6b4 - the failures in this try push are all for something else. However it's odd that the test is rendering with the ahem font since there is no ahem.css file in that folder. Could it be that somehow we're picking up the ahem font from some cache?
Also why does this only happen on mac? Happens with and without WR although much more frequent with WR.
Could this possibly be being affected by the recent work to install Ahem on macOS test machines for the sake of web platform tests that depend on it? If it's installed locally, the "li { font-family: ahem; }" rule will pick it up despite the lack of any CSS file. In any case, the font-family rule in that testcase is clearly spurious, as there's no corresponding rule in the reference.
That sounds plausible. Do you know when that happened, or have a bug number for that work? I can put upa patch to remove the rule.
Patch is on bug 1500815, since the permafail is a recent issue and there were low-volume intermittents here before which will probably remain.
So far no more failures have showed up since the landing of bug 1500815. I'm turning this bug loose again.
Assignee: kats → nobody
Whiteboard: [gfx-noted][stockwell disable-recommended] → [gfx-noted]
There are no more occurrences in the last 7 days.
Whiteboard: [gfx-noted][stockwell disable-recommended] → [gfx-noted]
Whiteboard: [gfx-noted][stockwell disable-recommended] → [gfx-noted]
Priority: -- → P3

I think this was fixed by bug 1500815. All the ones after that seem to be on mozilla-release which didn't have the fix at the time.

Status: NEW → RESOLVED
Closed: 7 years ago
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.