Closed Bug 1842952 Opened 1 year ago Closed 1 year ago

Perma [tier 2] transform-3d/1157984-1.html != about:blank | image comparison, max difference: 0, number of differing pixels: 0

Categories

(Core :: Layout, defect, P5)

defect

Tracking

()

RESOLVED FIXED
117 Branch
Tracking Status
firefox-esr102 --- unaffected
firefox-esr115 --- unaffected
firefox115 --- unaffected
firefox116 --- unaffected
firefox117 --- fixed

People

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

References

(Regression)

Details

(Keywords: intermittent-failure, regression)

Attachments

(1 file)

Filed by: nfay [at] mozilla.com
Parsed log: https://treeherder.mozilla.org/logviewer?job_id=422354280&repo=autoland
Full log: https://firefox-ci-tc.services.mozilla.com/api/queue/v1/task/eD48RN8BQ_etB-dJHsjJvA/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/eD48RN8BQ_etB-dJHsjJvA/runs/0/artifacts/public/logs/live_backing.log&only_show_unexpected=1


[task 2023-07-12T00:13:37.252Z] 00:13:37     INFO -  REFTEST TEST-START | layout/reftests/transform-3d/1157984-1.html != about:blank
[task 2023-07-12T00:13:37.252Z] 00:13:37     INFO -  REFTEST TEST-LOAD | http://10.7.205.235:8854/tests/layout/reftests/transform-3d/1157984-1.html | 2162 / 2198 (98%)
[task 2023-07-12T00:13:37.252Z] 00:13:37  WARNING -  REFTEST TEST-UNEXPECTED-FAIL | layout/reftests/transform-3d/1157984-1.html != about:blank | image comparison, max difference: 0, number of differing pixels: 0
[task 2023-07-12T00:13:37.252Z] 00:13:37     INFO -  REFTEST   IMAGE: data:image/png;base64,iVBORw0K...
[task 2023-07-12T00:13:37.252Z] 00:13:37     INFO -  REFTEST INFO | Saved log: START http://10.7.205.235:8854/tests/layout/reftests/transform-3d/1157984-1.html
[task 2023-07-12T00:13:37.252Z] 00:13:37     INFO -  REFTEST INFO | Saved log: [CONTENT] OnDocumentLoad triggering AfterOnLoadScripts
[task 2023-07-12T00:13:37.252Z] 00:13:37     INFO -  REFTEST INFO | Saved log: Initializing canvas snapshot
[task 2023-07-12T00:13:37.252Z] 00:13:37     INFO -  REFTEST INFO | Saved log: DoDrawWindow 0,0,800,1000
[task 2023-07-12T00:13:37.252Z] 00:13:37     INFO -  REFTEST INFO | Saved log: [CONTENT] RecordResult fired
[task 2023-07-12T00:13:37.252Z] 00:13:37     INFO -  REFTEST INFO | Saved log: RecordResult fired
[task 2023-07-12T00:13:37.252Z] 00:13:37     INFO -  REFTEST INFO | Saved log: RecordResult fired
[task 2023-07-12T00:13:37.252Z] 00:13:37     INFO -  REFTEST TEST-END | layout/reftests/transform-3d/1157984-1.html != about:blank

:gw, since you are the author of the regressor, bug 1837902, could you take a look?

For more information, please visit BugBot documentation.

Flags: needinfo?(gwatson)

I get a network error for the parsed log and the reftest URL fails to load. The full log loads though. Any ideas what may cause that? Perhaps just a temporary issue?

Flags: needinfo?(nfay)

The network error issue for the parsed log is a common issue when the full log is opened first before the parsed one (not sure if there are other issues that could cause this). Regarding the reftest URL, not really sure what's going on there, it's not loading for me either.

Flags: needinfo?(nfay)

I tracked this down to a float accuracy problem when rendering this test on SWGL on AArch64.

If I log out the inputs to draw_perspective_spans on an android device, I see:

p 0 = 187.236938 74.136719 0.500000
p 1 = -23228420096.000000 -20980729856.000000 0.500000
p 2 = -7497301504.000000 6771831808.000000 0.500000
p 3 = 187.236938 525.863281 0.500000
l0 = -23228420096.000000 -20980729856.000000 l1=-7497301504.000000 6771831808.000000 [1 1 2 0]

This results in the spans that come from aa_span being 0 width, such as:

span 0 [-11335812096.000000 -1077.779053]
span 0 [-11335812096.000000 -1076.671875]
span 0 [-11335812096.000000 -1075.564697]
span 0 [-11335812096.000000 -1074.457520]
span 0 [-11335812096.000000 -1073.350342]
...

The large input values and the precision loss due to floating point (in)accuracy mean that the right edges are not as expected, and they fall outside the clip rect bounds.

Given that, I will mark this test as an expected failure for now on SWGL + android, and open a bug to get it fixed.

Flags: needinfo?(gwatson)
Assignee: nobody → gwatson
Status: NEW → ASSIGNED
Pushed by gwatson@mozilla.com: https://hg.mozilla.org/integration/autoland/rev/8f98c9ff33a8 Disable 1157984-1.html on swgl + android, due to float accuracy r=gfx-reviewers,lsalzman

Backed out for causing reftests failures.

  • Backout link
  • Push with failures
  • Failure Log
  • Failure line: REFTEST TEST-UNEXPECTED-PASS | layout/reftests/transform-3d/1157984-1.html != about:blank | image comparison, max difference: 255, number of differing pixels: 3600
Flags: needinfo?(gwatson)

Updated patch, will re-land now.

Flags: needinfo?(gwatson)
Pushed by gwatson@mozilla.com: https://hg.mozilla.org/integration/autoland/rev/f618f6a4a706 Disable 1157984-1.html on swgl + android, due to float accuracy r=gfx-reviewers,lsalzman
Status: ASSIGNED → RESOLVED
Closed: 1 year ago
Resolution: --- → FIXED
Target Milestone: --- → 117 Branch
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: