Bug 90896 – fast/js/dfg-cross-global-object-inline-new-array-literal.html appears to be racey

• Status: RESOLVED
• Resolution: CONFIGURATION CHANGED
• Component: DOM

[Rafael Weinstein]

The test itself looks to be always "passing" but the output order of the test appears to be racey WRT iframe & parent. Added abarth, eseidel and esphren as they have recently looked into a bug that was similar.

http://test-results.appspot.com/dashboards/flakiness_dashboard.html#showAllRuns=true&showExpectatioans=true&tests=fast%2Fjs%2Fdfg-cross-global-object-inline-new-array-literal.html

Test expects:

----

This tests that function inlining in the DFG JIT doesn't get confused about the global object to use for array allocation.

On success, you will see a series of "PASS" messages, followed by "TEST COMPLETE".


PASS successfullyParsed is true

TEST COMPLETE
PASS done() called with 6600

----
But sometimes gets
----


This tests that function inlining in the DFG JIT doesn't get confused about the global object to use for array allocation.

On success, you will see a series of "PASS" messages, followed by "TEST COMPLETE".


PASS done() called with 6600
PASS successfullyParsed is true

TEST COMPLETE

[Eric Seidel (no email)]

We should CC whoever added the test. :) They'll likely know more than any of us.

[Eric Seidel (no email)]

http://trac.webkit.org/browser/trunk/LayoutTests/fast/js/dfg-cross-global-object-inline-new-array-literal.html

fpizlo is your man!

[Elliott Sprehn]

This test doesn't actually look flaky. I see it failed one time a week ago, but I can't download the results. Looking at the test I don't see anything that looks like it would be flaky either.

Can we remove this expectations change?

[Ahmad Saleem]

This test seems to be moved from 'fast' to 'js/dom' and it does not have any platform specific expectation while using searchfox.org to check.

Do we need to look into anymore? Even Comment 03 said that it does not seem to be flaky and only failed once in a whole week.

If we have it as flaky or failing, we would have newer big for sure.

[Anne van Kesteren]

If the file is expected to normally pass everywhere we can close this, yeah.