Bug 192607

Summary: Valgrind internal error while running Chromium's ObserverListThreadSafeTest unit test
Product: [Developer tools] valgrind Reporter: leiz
Component: generalAssignee: Rhys Kidd <rhyskidd>
Status: CONFIRMED ---    
Severity: crash CC: m.wege, njn, rhyskidd, stasiek-j
Priority: NOR    
Version: 3.5 SVN   
Target Milestone: wanted3.6.0   
Platform: Compiled Sources   
OS: macOS   
See Also: https://bugs.kde.org/show_bug.cgi?id=234814
Latest Commit: Version Fixed In:
Sentry Crash Report:

Description leiz 2009-05-13 22:18:05 UTC 
This is related to Chromium bug 7477. [1] I'm getting Valgrind internal errors on a 2nd gen Mac Pro running OS X 10.5.6. I'm using Xcode 3.1.2.

When I run my unit test under Valgrind, every so often I get a Valgrind internal error:

VALGRIND INTERNAL ERROR: Valgrind received a signal 11 (SIGSEGV) - exiting
si_code=1;  Faulting address: 0x7A4A010;  sp: 0xf4928ce0
</valgrindoutput>

> valgrind: the 'impossible' happened:
Killed by fatal signal
<stack>
<frame>
<ip>0xF00B6275</ip>
</frame>


To reproduce this bug:

1. Build Chromium per instructions here. [2] I tried with r15968.
2. Install Valgrind. I compile my copy from SVN source with r9779
3. Enable the ObserverListThreadSafeTest.CrossThreadObserver test by removing its entry from base/data/valgrind/base_unittests.gtest.txt
4. Run ./tools/valgrind/chrome_tests.sh -t base --gtest_filter=ObserverListThreadSafeTest.CrossThreadObserver

I did a loop of 100 runs. In 8 of the runs, I hit the internal error.

[1] http://code.google.com/p/chromium/issues/detail?id=7477
[2] http://code.google.com/p/chromium/wiki/MacBuildInstructions
Comment 1 Jacopo De Simoi 2010-01-09 11:35:01 UTC 
*** Bug 221053 has been marked as a duplicate of this bug. ***
Comment 2 Jacopo De Simoi 2010-01-09 11:36:09 UTC 
*** Bug 221253 has been marked as a duplicate of this bug. ***
Comment 3 Jacopo De Simoi 2010-01-09 11:42:02 UTC 
*** Bug 221622 has been marked as a duplicate of this bug. ***
Comment 4 Rhys Kidd 2015-07-13 01:57:26 UTC 
Note: A very similar bug report with si_code can be reliably reproduced with BZ#234814, using an available in-tree regression test.

Is anyone able to confirm if this bug is still present with current Chromium, current Valgrind SVN and the ObserverListThreadSafeTest.CrossThreadObserver test?
Comment 5 leiz 2015-08-21 05:46:41 UTC 
I just tested on Linux, but then I remembered this is a Mac bug. I'll have to find a suitable Mac to test with.