When used with 'collect' on a fairly big MT application (NetBeans), the JVM may
crash in AsyncGetCallTrace(). All symptoms suggest that the stack overflowed.
% collect -j on java -server <app>
See Comments for a pstack generated stack trace.
BugTraq+ Release Management Values
COMMIT TO FIX:
Deferring to tiger.
I fixed one infinite recursion in stack walking in the Mantis-Update
batch of bug fixes.
I found the thread stuck in the for-loop that walks the sender chain
in forte_is_walkable_frame(). The for-loop tried walking the stack
and found a self-referential frame so we just kept walking the same set
of frames over and over.
I have not been able to reproduce the original failure mode. Marty
has agreed that we can consider this bug fixed.