Closed Bug 1354832 Opened 3 years ago Closed 2 years ago

Intermittent bugs/368020-1.html == bugs/368020-1-ref.html | image comparison, max difference: 32, number of differing pixels: 33

Categories

(Core :: Layout, defect)

defect
Not set

Tracking

()

RESOLVED FIXED
mozilla59
Tracking Status
firefox58 --- fixed
firefox59 --- fixed

People

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

References

Details

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

Attachments

(1 file)

There were 33 failures related to this bug in the last 7 days. This occurs on Linux opt, asan and debug.

Reftest analyzer is not showing relevant data for this. Here's an example of the log:
https://treeherder.mozilla.org/logviewer.html#?repo=autoland&job_id=142446545&lineNumber=9413
Analyzer example:
https://hg.mozilla.org/mozilla-central/raw-file/tip/layout/tools/reftest/reftest-analyzer.xhtml#logurl=https://queue.taskcluster.net/v1/task/NOp_4tiZTLm5sldksv3FVw/runs/0/artifacts/public/logs/live_backing.log&only_show_unexpected=1

:jet can you take a look at this since it spiked in the last few days?
Flags: needinfo?(bugs)
Whiteboard: [stockwell needswork]
This test is already a fuzzy-if within the pixel thresholds reported in this intermittent failure (max difference: 32, number of differing pixels: 33)

fuzzy-if(winWidget||gtkWidget,32,34) fuzzy-if(skiaContent,1,28) 

http://searchfox.org/mozilla-central/source/layout/reftests/bugs/reftest.list#601

I'd normally just add more fuzz, but that won't fix this if the reftest harness is intermittently ignoring the specified fuzz parameters. Mats: can you tell what's going on here? Thx!
Flags: needinfo?(bugs) → needinfo?(mats)
Line 9412 says: "INFO | REFTEST fuzzy test (0, 0) <= (32, 33) <= (1, 28)".
I'm guessing both gtkWidget AND skiaContent are true and we selected the latter fuzzy-if.

I find "azureSkia":true,"skiaContent":true in the log which seems to indicate that's
what happened.
Flags: needinfo?(mats)
I looked into this intermittent failure a bit.

In my environment(Ubuntu 17.04 + XVFB), I can reproduce this phenomenon with low probability. (about 0.0199. [4/200])
If I move the 2nd condition to primary condition, this test passes always. (Try count is 20000)

(In reply to Mats Palmgren (:mats) from comment #30)
> Line 9412 says: "INFO | REFTEST fuzzy test (0, 0) <= (32, 33) <= (1, 28)".
> I'm guessing both gtkWidget AND skiaContent are true and we selected the
> latter fuzzy-if.
> 
> I find "azureSkia":true,"skiaContent":true in the log which seems to
> indicate that's
> what happened.

A refests use last fuzzy-if condition if we specify multiple condition and these condition is true.[1][2]

[1] https://searchfox.org/mozilla-central/rev/30ead7d1ae5bf95b8bc0fd67b950cd46ca05e32c/layout/tools/reftest/manifest.jsm#202-206
[2] https://searchfox.org/mozilla-central/rev/30ead7d1ae5bf95b8bc0fd67b950cd46ca05e32c/layout/tools/reftest/manifest.jsm#384-387

So I think we should modify fuzzy-if condition order of this test in this bug.

Mats:
This fuzzy-if condition has modified in bug 1179543.
In this test environment(i.e. azureSkia:true and skiaContent:true), I think we can use the gtkWidget condition, Is this correct?
Flags: needinfo?(mats)
Yeah, maybe we should just swap the order, like so:
fuzzy-if(skiaContent,1,28) fuzzy-if(winWidget||gtkWidget,32,34)
Flags: needinfo?(mats)
(In reply to Mats Palmgren (:mats) from comment #32)
> Yeah, maybe we should just swap the order, like so:
> fuzzy-if(skiaContent,1,28) fuzzy-if(winWidget||gtkWidget,32,34)

Thanks.

Try:
https://treeherder.mozilla.org/#/jobs?repo=try&revision=ed95c7e42c57c8cdfeea795e294473e72f3c7e7d

Assuming a green try run and my environment, I guess this change is effective to this bug.
Comment on attachment 8927690 [details]
Bug 1354832 - Sort a fuzzy-if condition of bugs/368020-1.html to descending order.

https://reviewboard.mozilla.org/r/198456/#review203988
Attachment #8927690 - Flags: review?(mats) → review+
Pushed by mantaroh@gmail.com:
https://hg.mozilla.org/integration/autoland/rev/fb7f987c1a28
Sort a fuzzy-if condition of bugs/368020-1.html to descending order. r=mats
https://hg.mozilla.org/mozilla-central/rev/fb7f987c1a28
Status: NEW → RESOLVED
Closed: 2 years ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla59
Duplicate of this bug: 1413132
Assignee: nobody → mantaroh
You need to log in before you can comment on or make changes to this bug.