NEW 266999
resize-observer/contain-instrinsic-size-should-not-leak-nodes.html is flaky failure recently
https://bugs.webkit.org/show_bug.cgi?id=266999
Summary resize-observer/contain-instrinsic-size-should-not-leak-nodes.html is flaky f...
Fujii Hironori
Reported 2024-01-02 13:40:25 PST
resize-observer/contain-instrinsic-size-should-not-leak-nodes.html is flaky faiure recently History: https://results.webkit.org/?suite=layout-tests&test=resize-observer%2Fcontain-instrinsic-size-should-not-leak-nodes.html It seems that the following test run is the first failure. Buildbot: builder Apple-Sonoma-Release-WK2-Tests build 1068 : 272445@main https://build.webkit.org/#/builders/926/builds/1068 resize-observer/contain-instrinsic-size-should-not-leak-nodes.html [ Failure ] https://build.webkit.org/results/Apple-Sonoma-Release-WK2-Tests/272445@main%20(1068)/resize-observer/contain-instrinsic-size-should-not-leak-nodes-diff.txt --- /Volumes/Data/worker/Apple-Sonoma-Release-WK2-Tests/build/layout-test-results/resize-observer/contain-instrinsic-size-should-not-leak-nodes-expected.txt +++ /Volumes/Data/worker/Apple-Sonoma-Release-WK2-Tests/build/layout-test-results/resize-observer/contain-instrinsic-size-should-not-leak-nodes-actual.txt @@ -1,6 +1,6 @@ This tests element is with CIS auto and removing the element from the document. The element should become eligible for GC at some point. -PASS Remove nodes after observing -PASS Remove nodes after skipping content +FAIL Remove nodes after observing assert_equals: More than 20% of nodes should be collected expected true but got false +FAIL Remove nodes after skipping content assert_equals: More than 20% of nodes should be collected expected true but got false
Attachments
Fujii Hironori
Comment 1 2024-01-03 04:11:59 PST
This is reproducible with WinCairo Release on my PC. > python .\Tools\Scripts\run-webkit-tests --release --no-retry --iter=1000 -f --exit-after-n-f=1 resize-observer/contain-instrinsic-size-should-not-leak-nodes.html But, applying the following patch works around the problem. diff --git a/LayoutTests/resize-observer/contain-instrinsic-size-should-not-leak-nodes.html b/LayoutTests/resize-observer/contain-instrinsic-size-should-not-leak-nodes.html index d894254ce8e1..001cbc0cf3de 100644 --- a/LayoutTests/resize-observer/contain-instrinsic-size-should-not-leak-nodes.html +++ b/LayoutTests/resize-observer/contain-instrinsic-size-should-not-leak-nodes.html @@ -62,6 +62,7 @@ promise_test(async function () { await nextRendering(); removeNodes(); await nextRendering(); + await new Promise(resolve => { setTimeout(resolve, 1); }); gc(); await nextRendering(); assert_equals(internals.referencingNodeCount(document) < initialNodeCount + CISCount * 0.8, true, 'More than 20% of nodes should be collected'); @@ -75,6 +76,7 @@ promise_test(async function () { await nextRendering(); removeNodes(); await nextRendering(); + await new Promise(resolve => { setTimeout(resolve, 1); }); gc(); await nextRendering(); assert_equals(internals.referencingNodeCount(document) < initialNodeCount + CISCount * 0.8, true, 'More than 20% of nodes should be collected'); It recalls me another gc test which is also worked around by delaying gc (but#214673 comment#5).
Fujii Hironori
Comment 2 2024-01-03 04:12:31 PST Comment hidden (obsolete)
Fujii Hironori
Comment 3 2024-01-03 04:12:45 PST
Radar WebKit Bug Importer
Comment 4 2024-01-09 13:41:14 PST
Note You need to log in before you can comment on or make changes to this bug.