Details
-
Improvement
-
Status: Resolved
-
Major
-
Resolution: Fixed
-
2.6.0SDK
-
None
Description
Jerry C discovered a hang failure - I was able to reproduce on a particular large machine. Traced to a race condition - where a "reserve" is found, and then a lock is acquired. There's a chance that between these 2 events, another thread can change the reserve to a real value - if this happens, a hang. Recheck the value after acquiring the lock.