Summary: | Valgrind on Mac hangs on a race-condition | ||
---|---|---|---|
Product: | [Developer tools] valgrind | Reporter: | Nirnimesh <nirnimesh> |
Component: | memcheck | Assignee: | Rhys Kidd <rhyskidd> |
Status: | RESOLVED WORKSFORME | ||
Severity: | crash | CC: | glider, nirnimesh, njn, rhyskidd, timurrrr |
Priority: | NOR | ||
Version: | 3.5 SVN | ||
Target Milestone: | --- | ||
Platform: | Compiled Sources | ||
OS: | macOS | ||
Latest Commit: | Version Fixed In: | ||
Sentry Crash Report: |
Description
Nirnimesh
2009-04-15 00:14:40 UTC
Nirnimesh, Can you please give us a name of the test? The comments in $CHROMIUM_DIR/base/data/valgrind/base_unittests.gtest.txt state that the corresponding tests are TimerTest.RepeatingTimer and TimerTest.RepeatingTimer_Cancel. They seem to pass under the current revision of Valgrind used for testing Chromium. The two tests -- TimerTest.RepeatingTimer and TimerTest.RepeatingTimer_Cancel are currently disabled when running base_unittests. Alexander, did you run the test under valgrind without disabling them? They used to hang often but not always. I could enable them to see if recent versions of valgrind treats them with more love. I did, and got flaky results, too. Sorry for deception. I'll take a look at these tests to find out if we can reproduce the bug in a simpler way. Any possibility that you'e been able to produce a simpler test case that reproduces the issue? There's been significant improvements to Valgrind on OS X since this report, so it may well have been resolved in the interim. Not a chance, sorry. We've kept Valgrind running on some Mac tests on Chromium, but we haven't done any Valgrind development for several years, as the tool is now being superseded by compiler-based memory error detectors. Please feel free to close Chromium-related bugs in Valgrind on Mac as non-reproducible. |