Hi! On Sat, 5 Nov 2011 01:24:05 +0100, Thomas Schwinge <[email protected]> wrote: > * sysdeps/hurd/pt-destroy-specific.c (__pthread_destroy_specific): Correct > logic for PTHREAD_KEY_INVALID slots.
In fact, this fixes the licq issue mentioned at <http://www.gnu.org/software/hurd/open_issues/libpthread_pthread_key_create_reuse.html>: [----------] 3 tests from ThreadSpecificData [ RUN ] ThreadSpecificData.create [ OK ] ThreadSpecificData.create (0 ms) [ RUN ] ThreadSpecificData.setAndGet [ OK ] ThreadSpecificData.setAndGet (0 ms) [ RUN ] ThreadSpecificData.dataDeletedWhenThreadExits /home/thomas/tmp/licq/licq-1.5.1/src/thread/tests/threadspecificdatatest.cpp:65: Failure Value of: DeleteLogger::myCount Actual: 0 Expected: 1 [ FAILED ] ThreadSpecificData.dataDeletedWhenThreadExits (0 ms) [----------] 3 tests from ThreadSpecificData (0 ms total) I wondered why licq -- a user program -- would contain tests for basic pthread primitives. It turns out, they have a C++ class library wrapping these, and that's what they're testing. Grüße, Thomas
pgpz3vW5oe6eF.pgp
Description: PGP signature
