1842623 - Intermittent pixel-rounding/border-image-width-4.html == pixel-rounding/border-image-width-0.html | single tracking bug

Status: RESOLVED FIXED

(Core :: Graphics: WebRender, defect, P5)

Description

[Treeherder Bug Filer]

Filed by: sstanca [at] mozilla.com
Parsed log: https://treeherder.mozilla.org/logviewer?job_id=422133536&repo=autoland
Full log: https://firefox-ci-tc.services.mozilla.com/api/queue/v1/task/Aai87wXkQOeJgdZyL06JLg/runs/0/artifacts/public/logs/live_backing.log
Reftest URL: https://hg.mozilla.org/mozilla-central/raw-file/tip/layout/tools/reftest/reftest-analyzer.xhtml#logurl=https://firefox-ci-tc.services.mozilla.com/api/queue/v1/task/Aai87wXkQOeJgdZyL06JLg/runs/0/artifacts/public/logs/live_backing.log&only_show_unexpected=1

---

There are no visible differences.

Comment 1

[Noemi Erli[:noemi_erli]]

R+B range: https://treeherder.mozilla.org/jobs?repo=autoland&searchStr=Windows%2C11%2Cx86%2C22H2%2CWebRender%2Cdebug%2CReftests%2Cwith%2Cdcomp%2Ccompositing%2Ctest-windows11-32-2009-qr%2Fdebug-reftest-wr-dc3-c%2CR5&tochange=47f266144ce3d9bc035cc91770285c4702098cda&fromchange=6861b492d5412d5a6fe18fcc2d2b6260e7c95c83&selectedTaskRun=Uf6tNgX0SFG_WK2S-_a9NQ.0
This started with the change in https://bugzilla.mozilla.org/show_bug.cgi?id=1778032
Nicolas, could you please take a look at this?
Thanks

Comment 2

[BugBot [:suhaib / :marco/ :calixte]]

Set release status flags based on info from the regressing bug 1778032

Comment 3

[Nicolas Silva [:nical]]

This should be fixed by the patch in bug 1842857.

Comment 4

[Intermittent Failures Robot]

21 failures in 3309 pushes (0.006 failures/push) were associated with this bug in the last 7 days.

Repository breakdown:

• autoland: 14
• mozilla-central: 7

Platform and build breakdown:

• windows11-32-2009-qr: 5
  • debug: 2
  • opt: 3
• windows11-64-2009-qr: 5
  • opt: 5
• windows11-64-2009-asan-qr: 5
  • opt: 5
• windows11-64-2009-shippable-qr: 5
  • opt: 5
• windows11-32-2009-shippable-qr: 1
  • opt: 1

For more details, see:
https://treeherder.mozilla.org/intermittent-failures/bugdetails?bug=1842623&startday=2023-07-10&endday=2023-07-16&tree=all

Comment 5

[BugBot [:suhaib / :marco/ :calixte]]

Set release status flags based on info from the regressing bug 1778032

Comment 6

[Intermittent Failures Robot]

1 failures in 3276 pushes (0.0 failures/push) were associated with this bug in the last 7 days.

Repository breakdown:

• mozilla-central: 1

Platform and build breakdown:

• windows11-64-2009-mingwclang-qr: 1
  • debug: 1

For more details, see:
https://treeherder.mozilla.org/intermittent-failures/bugdetails?bug=1842623&startday=2023-07-31&endday=2023-08-06&tree=all

Comment 7

[Intermittent Failures Robot]

2 failures in 3076 pushes (0.001 failures/push) were associated with this bug in the last 7 days.

Repository breakdown:

• autoland: 1
• mozilla-beta: 1

Platform and build breakdown:

• windows11-32-2009-qr: 1
  • opt: 1
• windows11-64-2009-devedition-qr: 1
  • opt: 1

For more details, see:
https://treeherder.mozilla.org/intermittent-failures/bugdetails?bug=1842623&startday=2023-08-14&endday=2023-08-20&tree=all

Comment 8

[Intermittent Failures Robot]

1 failures in 2218 pushes (0.0 failures/push) were associated with this bug in the last 7 days.

Repository breakdown:

• autoland: 1

Platform and build breakdown:

• windows11-32-2009-qr: 1
  • opt: 1

For more details, see:
https://treeherder.mozilla.org/intermittent-failures/bugdetails?bug=1842623&startday=2023-08-21&endday=2023-08-27&tree=all

Comment 9

[Daniel Holbert [:dholbert]]

(In reply to Nicolas Silva [:nical] from comment #3)

This should be fixed by the patch in bug 1842857.

Apparently that patch (which adjusted the fuzzy tolerance) wasn't quite sufficient. That patch (https://hg.mozilla.org/mozilla-central/rev/3170f4f57d01 ) bumped the differing-pixel-count to 21, but in fact the failure sometimes has a bit more than that.

This log has 25 differing pixels for border-image-width-4.html and border-image-width-9.html:
https://treeherder.mozilla.org/logviewer?job_id=424702855&repo=mozilla-central&lineNumber=10824
That's the most I'm seeing here. The other recent logs have 22 as the differing-pixel-count.

We should probably just use 25 as the upper-limit instead of 21, I guess?

(These do look like "real" failures; there are border-image pixels that are substantially the wrong color here, not like in an antialiasing-fringe sort of way. I'm not sure if it's worth having a followup to investigate the actual failure or not? It looks like the failure is new since bug 1778032; before that, these tests had no fuzzy annotation at all.)

Comment 10

[Daniel Holbert [:dholbert]]

Attachment: Bug 1842623: Increase the fuzzy tolerance for border-image-width-4.html and -9.html to reflect reality. r?nical,lsalzman

Comment 11

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/a1fb2d995877
Increase the fuzzy tolerance for border-image-width-4.html and -9.html to reflect reality. r=lsalzman DONTBUILD

Comment 12

[Daniel Holbert [:dholbert]]

Attachment: the reftest test screenshot from the log in comment 9

(In reply to Daniel Holbert [:dholbert] from comment #9)

This log has 25 differing pixels for border-image-width-4.html and border-image-width-9.html:
https://treeherder.mozilla.org/logviewer?job_id=424702855&repo=mozilla-central&lineNumber=10824

For posterity, here's the reftest-analyzer view of that test failure:
https://hg.mozilla.org/mozilla-central/raw-file/tip/layout/tools/reftest/reftest-analyzer.xhtml#logurl=https://firefox-ci-tc.services.mozilla.com/api/queue/v1/task/ZsRl4vyASASxuOkX7t8wbA/runs/0/artifacts/public/logs/live_backing.log&only_show_unexpected=1

And in case the log expires or something, I'm attaching the test screenshot for the -4.html test-failure here, and I'll attach the reference screenshot as well.

Comment 13

[Daniel Holbert [:dholbert]]

Attachment: the reftest reference screenshot from the log in comment 9

Here's the corresponding "reference" screenshot that was shown in the reftest-analyzer UI.

Comment 14

[Daniel Holbert [:dholbert]]

Attachment: screencast of reftest-analyzer

nical do you think we should spin off a followup for the underlying issue here? (Are these intermittent mismatches concerning?)

For convenience, here's a screencast of what the reftest-analyzer comparison looks like for the failure in comment 9 / comment 12. As you can see in the pixel-grid in the second half of the video, the mismatching pixels are quite different in the testcase vs. the reference case.

(This might be easiest to see if you fullscreen the video when playing it; or just use the reftest-analyzer link directly.)

Comment 15

[Cosmin Sabou [:CosminS]]

https://hg.mozilla.org/mozilla-central/rev/a1fb2d995877

Comment 16

[Nicolas Silva [:nical]]

It's surprising to me that the issue is intermittent. As far as the actual differences are concerned, I'm not too worried because this test magnifies off-by-one differences in how repetitions are positioned/snapped. For web content I don't expect it to make a big difference in practice, so that puts figuring this out far down the priority list.