Closed Bug 1321707 Opened 3 years ago Closed 3 years ago
.xht == w3c-css/submitted/ui3/box-sizing-replaced-001-ref .xht | image comparison, max difference: 255, number of differing pixels: 900
Filed by: philringnalda [at] gmail.com https://treeherder.mozilla.org/logviewer.html#?job_id=40220193&repo=mozilla-inbound https://queue.taskcluster.net/v1/task/KnTSKKljSFWTarSilmlO0w/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/KnTSKKljSFWTarSilmlO0w/runs/0/artifacts/public/logs/live_backing.log&only_show_unexpected=1
this has spiked recently, and on android I did a few hundred retriggers: https://treeherder.mozilla.org/#/jobs?repo=autoland&filter-searchStr=android%20r2&tochange=11e5d90cbf4de0882a6759d7062fa62f73e61d99&fromchange=b29345665e1df21e46b54b40ffcc85610bdd6882&selectedJob=69043813 and it comes down to: https://hg.mozilla.org/integration/autoland/rev/30d0b78371d543b6691a4edbb1f57628917fd562 James, can you look at this?
This intermittent test failure happened from 2016/12, those are only happened on debug build... As comment 7 said it comes down to （landed on Jan）https://hg.mozilla.org/integration/autoland/rev/30d0b78371d543b6691a4edbb1f57628917fd562 The change just makes the MOZ_ASSERT to MOZ_RELEASE_ASSERT which makes us be more aggressive to detect the misusing of mozilla::Variant and we can see the intermittent failure started to occur on non-debug build. Since I'm not the layout guy, I cannot tell what happened. Hi Astley, Would you please help to dispatch this issue to a suitable person to see how the code related to Variant? Since the assertion is obvious to know there must have some error casting by ```Variant::as<ToSomeErrorType>``` Thank you.
Flags: needinfo?(jacheng) → needinfo?(aschen)
James, suppose the test will generate error casting log as you mentioned. But I didn't find it in relevant android tests failure provided in comment 7. I'll do a simple local test and see if the relevant log can be discovered and create some clues for further debugging. Leave ni for tracking.
I think the pattern is showing both android and linux* as culprits to this. Astley, do you need any help or more information to help solve this?
will disable this test until Astley has time to look at this. Astley, you can easily test this locally or on try by removing the edits in reftest.list.
Assignee: nobody → jmaher
Attachment #8834001 - Flags: review?(gbrown)
Comment on attachment 8834001 [details] [diff] [review] disable reftest on linux/android Review of attachment 8834001 [details] [diff] [review]: ----------------------------------------------------------------- Maybe add "bug 1321707" to the comment?
Attachment #8834001 - Flags: review?(gbrown) → review+
Pushed by firstname.lastname@example.org: https://hg.mozilla.org/integration/mozilla-inbound/rev/86e50de186e9 Intermittent w3c-css/submitted/ui3/box-sizing-replaced-001.xht skip on android/linux. r=gbrown
this isn't fixing everything, :ryanvm suspects order of skip-if vs other clauses in the reftest.list
Status: RESOLVED → REOPENED
Resolution: FIXED → ---
odd that these are failing now, I see android failing as well- this indicates a parser error. I verified with local hacking that skip-if() can come before fuzzy-if(), also with some logs where we have prior art of skip-if() showing up before fuzzy-if(). So I pushed to try a few times to see if this test case is getting skipped with various adjustments.
try server confirmed what :ryanvm said- there must be other clues to figure this out which I am missing- either way, this patch solves the immediate problem.
Attachment #8834575 - Flags: review?(gbrown)
Comment on attachment 8834575 [details] [diff] [review] fix reftest.list ordering of skip-if fuzzy-if Review of attachment 8834575 [details] [diff] [review]: ----------------------------------------------------------------- I don't understand why this is necessary, but I have seen similar unexpected issues....not surprised. Is there a bug for the generic manifest problem?
Attachment #8834575 - Flags: review?(gbrown) → review+
I filed bug 1337766 to investigate the manifest conditions.
Pushed by email@example.com: https://hg.mozilla.org/integration/mozilla-inbound/rev/20df6b744813 Intermittent w3c-css/submitted/ui3/box-sizing-replaced-001.xht. fix skip-if ordering in reftest.list. r=gbrown
3 years ago
Whiteboard: [stockwell disabled]
You need to log in before you can comment on or make changes to this bug.