To reproduce the bug: gcc none/tests/pth_blockedsig.c -o pth_blockedsig valgrind ./pth_blockedsig UNKNOWN __pthread_sigmask is unsupported. This warning will not be repeated. thread CHILD sending SIGUSR1 to thread MAIN SHOULD NOT BE HERE (SIGUSR1)!!!! ==66545== , while a native run produces the following output: glider$ ./pth_blockedsig thread CHILD sending SIGUSR1 to thread MAIN Nullgrind behaves the same way on this test.
This is what's currently preventing mysqld from running under valgrind on MacOSX 10.6
This can be reliably reproduced with the below commands: $ make check $ perl tests/vg_regtest memcheck/tests/thread_alloca none/tests/pth_blockedsig thread_alloca: valgrind -q --trace-children=yes ./thread_alloca 30 sh: line 1: 77175 Illegal instruction: 4 VALGRIND_LIB=/Users/rhyskidd/Documents/Coding/valgrind/trunk/.in_place VALGRIND_LIB_INNER=/Users/rhyskidd/Documents/Coding/valgrind/trunk/.in_place /Users/rhyskidd/Documents/Coding/valgrind/trunk/./coregrind/valgrind --command-line-only=yes --memcheck:leak-check=no --tool=memcheck -q --trace-children=yes ./thread_alloca 30 > thread_alloca.stdout.out 2> thread_alloca.stderr.out *** thread_alloca failed (stderr) *** pth_blockedsig: valgrind ./pth_blockedsig *** pth_blockedsig failed (stderr) *** == 2 tests, 2 stderr failures, 0 stdout failures, 0 stderrB failures, 0 stdoutB failures, 0 post failures == memcheck/tests/thread_alloca (stderr) none/tests/pth_blockedsig (stderr)
Created attachment 93559 [details] Proposed patch
Resolved with r15410.