Steve Bellovin <smb@research.att.com> writes:

> I've been having the usual "thread program hanging" problem on -current.
> It's my perception -- certainly not confirmed by anything analytic -- 
> that the problem occurs more when galeon (the program where I usually 
> see this) has been idle and is paged out.  Not sure if that's a helpful 
> clue or not.  (My kernel and libraries are from about a week ago.  I 
> may install today's build this afternoon.)

Yeah, there is something fishy going on with the various recoveries
that have to happen when the SA subsystem needs to communicate with a
swapped-out app (or an app where certain pages, like the internal
pthread_t structures, are paged out). I've been looking in that
direction lately.

        - Nathan