Yahhh. I was finally able to reproduce the problem - running Stephen's
16MB make -j5 buildworld test overnight.
I haven't found the exact cause yet, and I suspect that Alan's patch will
not fix it (but I'll try it if I exhaust other possibilities). There is
plenty of free memory and only one process is blocked in a weird state
(objtrm). No deadlock condition exists anywhere. Alan's patch
will fix another potential low-memory deadlock (it allows pageouts to be
performed on an exiting objecting), though.
-Matt
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
