[Bug 92] threading is broken (most visible on SMP)

http://sablevm.org/bugs/show_bug.cgi?id=92





------- Additional Comments From [email protected]  2005-01-29 01:10 -------
Interesting!  You found a bug.

This should probably look like (note: no SMP support here):

          /* notify owning thread that inflation is requested */
          _svmm_mutex_lock (owner->contention.owner.mutex);

          old_flag = owner->contention.owner.flag;

          owner->contention.owner.flag = 1;

/** MODIFICATION HERE **/
          /* re-read the instance's lockword, now that the thread
             has been notified */
          old_lockword = instance->lockword;
          
          if (_svmf_lockword_is_thin (old_lockword) &&
              _svmf_lockword_get_thinlock_id (old_lockword) ==
              owner->thread.thinlock_id)
/** END OF MODIFICATION **/
            {
              /* the thinlock owner is still the same and has been notified */

              notified = JNI_TRUE;

              /* add this thread into the lock owner's wait_list */
              env->contention.requester.wait_list_next =
                owner->contention.owner.wait_list;
              owner->contention.owner.wait_list = env;

              *(env->contention.requester.jobject) = instance;
            }
          else
            {
/** MODIFICATION HERE **/
              /* either the owner has changed or the lock is not thin anymore,
                 so restore the contention flag to its original value */
/** END OF MODIFICATION **/

              notified = JNI_FALSE;

              owner->contention.owner.flag = old_flag;
            }

          _svmm_mutex_unlock ();


I guess all of the code of monitor_enter/exit should be audited to find other
similar errors, if any.




------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.
You are the QA contact for the bug, or are watching the QA contact.