Open Bug 1385288 Opened 2 years ago Updated 2 years ago

Intermittent parser/htmlparser/tests/reftest/bug582940-1.html == parser/htmlparser/tests/reftest/bug582940-1-ref.html | image comparison, max difference: 255, number of differing pixels: 647

Categories

(Core :: Layout, defect, P2)

defect

Tracking

()

Tracking Status
firefox58 --- disabled
firefox59 --- verified

People

(Reporter: intermittent-bug-filer, Unassigned)

References

Details

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

Attachments

(1 file)

Alphan, would you like to give it a try?
Flags: needinfo?(alchen)
Assignee: nobody → alchen
Flags: needinfo?(alchen)
Priority: -- → P2
thanks for working on this :alchen :)
this has 156 failures in the last 30 days, any update or eta?
Flags: needinfo?(alchen)
(In reply to Joel Maher ( :jmaher) (UTC-5) from comment #17)
> this has 156 failures in the last 30 days, any update or eta?

Sorry about the delay. I will have some updates next week.
Flags: needinfo?(alchen)
After investigation, the result is positive for parser part.
They both go to the right anchor. (Ref!)

However, there is displacement in showing the second "Filler" after "Ref!" which cause the image comparison failed.
It should be related to layout or render part.
(In reply to Alphan Chen [:alchen] from comment #20)
> After investigation, the result is positive for parser part.
> They both go to the right anchor. (Ref!)
>
The result can be found in the following link. 
https://hg.mozilla.org/mozilla-central/raw-file/tip/layout/tools/reftest/reftest-analyzer.xhtml#logurl=https://queue.taskcluster.net/v1/task/bKW5HaPOSjSZpyZHx9mKfw/runs/0/artifacts/public/logs/live_backing.log&only_show_unexpected=1
Attached image Failure Screen Shot
Two images are scrolled to the right anchor.
Hi Astley,
as I said in previous comments, the failure is not caused by purpose of this test.
There is displacement in showing some elements.
could you take a look about this bug?


The failure only happened on macosx.
https://brasstacks.mozilla.com/orangefactor/?display=BugCount&includefiltertype=quicksearch&includefilterdetailsexcludeResolved=false&includefilterdetailsexcludeDisabled=false&includefilterdetailsquicksearch=bug582940-1.html%20&includefilterdetailsnumbugs=0&includefilterdetailsresolvedIds=&excludefiltertype=quicksearch&excludefilterdetailsquicksearch=&excludefilterdetailsnumbugs=0&excludefilterdetailsresolvedIds=&key=platform&tree=all&startday=2017-10-19&endday=2017-10-26

In the beginning, only on macosx64-stylo.
(In reply to OrangeFactor Robot from comment #1)
> 3 failures in 1008 pushes (0.003 failures/push) were associated with this
> bug in the last 7 days.   
> 
> Platform breakdown:
> * macosx64-stylo: 3

(In reply to OrangeFactor Robot from comment #3)
> 10 failures in 888 pushes (0.011 failures/push) were associated with this
> bug in the last 7 days.   
> 
> Platform breakdown:
> * macosx64-stylo: 10

(In reply to OrangeFactor Robot from comment #4)
> 12 failures in 901 pushes (0.013 failures/push) were associated with this
> bug in the last 7 days.   
> 
> Platform breakdown:
> * macosx64-stylo: 12

(In reply to OrangeFactor Robot from comment #5)
> 12 failures in 949 pushes (0.013 failures/push) were associated with this
> bug in the last 7 days.   
> 
> Platform breakdown:
> * macosx64-stylo: 12

(In reply to OrangeFactor Robot from comment #7)
> 17 failures in 939 pushes (0.018 failures/push) were associated with this
> bug in the last 7 days.   
> 
> Platform breakdown:
> * macosx64-stylo: 17
Assignee: alchen → nobody
Component: XML → Layout
Flags: needinfo?(bmo)
(In reply to OrangeFactor Robot from comment #25)
> 70 failures in 912 pushes (0.077 failures/push) were associated with this
> bug in the last 7 days. 
> 
> This is the #16 most frequent failure this week.  
> 
> ** This failure happened more than 30 times this week! Resolving this bug is
> a high priority. **

Xidorn, do you have any idea why stylo could cause this failure intermittently?
Flags: needinfo?(bmo) → needinfo?(xidorn+moz)
Had a brief look at the comparison result, and don't immediately have any thought on why this happens. This may be related to font metrics being rounded differently, although I'm not sure why that could be nondeterministic. Maybe Manish has some insight about whether there is anything could go wrong with font-size.

If this is easy to reproduce on try server, maybe we can test with different font and see if this still happens for mitigation.
Flags: needinfo?(xidorn+moz)
Yeah, I don't know why this would happen.
Pushed by jmaher@mozilla.com:
https://hg.mozilla.org/integration/mozilla-inbound/rev/4f39d5874392
Disable parser/htmlparser/tests/reftest/bug582940-1.html on osx debug for frequent failures. r=me, a=testonly
I disabled this test for having >200 failures in the last 30 days, please remember to enable this when testing a fix
Keywords: leave-open
Whiteboard: [stockwell disable-recommended] → [stockwell disabled]
Pushed by jmaher@mozilla.com:
https://hg.mozilla.org/integration/mozilla-inbound/rev/d3f290bc08ee
Disable parser/htmlparser/tests/reftest/bug582940-1.html on osx for frequent failures. r=me, a=testonly
There are 90 failures in the last week.
All the failures occur on the OS X 10.10 platform.
The most affected build type is debug, but there are also 10 failures on opt.
Here is a recent log file:
https://treeherder.mozilla.org/logviewer.html#?repo=mozilla-central&job_id=143972900&lineNumber=34410

And a snippet from it:
02:55:33     INFO - [Child 738, Main Thread] WARNING: Unsupported TextureClient backend type: 'aMoz2DBackend != gfx::BackendType::SKIA && aMoz2DBackend != gfx::BackendType::DIRECT2D && aMoz2DBackend != gfx::BackendType::DIRECT2D1_1', file /builds/worker/workspace/build/src/gfx/layers/client/TextureClient.cpp, line 1255
34407
02:55:33     INFO - [Child 738, Main Thread] WARNING: Unsupported TextureClient backend type: 'aMoz2DBackend != gfx::BackendType::SKIA && aMoz2DBackend != gfx::BackendType::DIRECT2D && aMoz2DBackend != gfx::BackendType::DIRECT2D1_1', file /builds/worker/workspace/build/src/gfx/layers/client/TextureClient.cpp, line 1255
34408
02:55:33     INFO - [Child 738, Main Thread] WARNING: Unsupported TextureClient backend type: 'aMoz2DBackend != gfx::BackendType::SKIA && aMoz2DBackend != gfx::BackendType::DIRECT2D && aMoz2DBackend != gfx::BackendType::DIRECT2D1_1', file /builds/worker/workspace/build/src/gfx/layers/client/TextureClient.cpp, line 1255
34409
02:55:34     INFO - REFTEST INFO | REFTEST fuzzy test (0, 0) <= (255, 647) <= (2, 5)
34410
02:55:34     INFO - REFTEST TEST-UNEXPECTED-FAIL | file:///Users/cltbld/tasks/task_1510396492/build/tests/reftest/tests/parser/htmlparser/tests/reftest/bug582940-1.html == file:///Users/cltbld/tasks/task_1510396492/build/tests/reftest/tests/parser/htmlparser/tests/reftest/bug582940-1-ref.html | image comparison, max difference: 255, number of differing pixels: 647
Flags: needinfo?(bugs)
it looks like my attempt to disable this test didn't work:
https://hg.mozilla.org/mozilla-central/rev/d3f290bc08ee

I will figure it out, I think there is an issue with skip-if and fuzzy-if in the reftest manifest where fuzzy-if will negate any skip-if.
Flags: needinfo?(bugs)
ok, we cannot skip-if and fuzzy-if on the same line:
https://searchfox.org/mozilla-central/source/layout/tools/reftest/globals.jsm#42

fuzzy-if will take precedence and be used throwing out the skip-if clause- so the only thing that I know is to disable fully on all tests.
Pushed by jmaher@mozilla.com:
https://hg.mozilla.org/integration/mozilla-inbound/rev/dc331b8af3a7
Disable parser/htmlparser/tests/reftest/bug582940-1.html on osx for frequent failures. r=me, a=testonly
tried flipping skip-if and fuzzy-if, I am skeptical, but will watch and disable fully if this doesn't work.
Flags: needinfo?(jmaher)
my skepticism was uncalled for, this worked:
0:01:42     INFO - REFTEST TEST-START | file:///Users/cltbld/tasks/task_1510594843/build/tests/reftest/tests/parser/htmlparser/tests/reftest/bug582940-1.html == file:///Users/cltbld/tasks/task_1510594843/build/tests/reftest/tests/parser/htmlparser/tests/reftest/bug582940-1-ref.html
10:01:42     INFO - REFTEST TEST-SKIP | file:///Users/cltbld/tasks/task_1510594843/build/tests/reftest/tests/parser/htmlparser/tests/reftest/bug582940-1.html == file:///Users/cltbld/tasks/task_1510594843/build/tests/reftest/tests/parser/htmlparser/tests/reftest/bug582940-1-ref.html | 
10:01:42     INFO - REFTEST TEST-END | file:///Users/cltbld/tasks/task_1510594843/build/tests/reftest/tests/parser/htmlparser/tests/reftest/bug582940-1.html == file:///Users/cltbld/tasks/task_1510594843/build/tests/reftest/tests/parser/htmlparser/tests/reftest/bug582940-1-ref.html

I am going to assume this is righfully disabled and the failures should be gone now.
Flags: needinfo?(jmaher)
You need to log in before you can comment on or make changes to this bug.