Update on MIDI lockup
Bill Schwab
squeak-dev at lists.squeakfoundation.org
Tue Oct 1 01:27:42 UTC 2002
Andreas,
I've all but proven that my "MIDI" lockup is in fact independent of
the MIDI primitives. At least on a particular machine, I can reliably
lock it up using only a simulated source of data.
Once the system locks up, the 3.2.2 (5-26-02) VM's Dump call
stack command always seems to show a particular background
process in various stages of activity, but never waiting on a
semaphore. My interpretation (contradictory opinions will be
cheerfully considered) is that the morphic main loop is being
deadlocked by another of my threads, or (much less likely I think)
is somehow failing to wake up from a wait for some other reason.
I couldn't find the call stack code in the VM source that I had on
hand; in looking for an update, I saw the new VM, and found the
call stack code in its source. More, I found printAllStacks(), which
sounds like the output that I want to see. Is it exposed somewhere
in the menus? If not, would it be as simple as exchanging
printAllStacks for printCallStack near the bottom of
sqWin32Prefs.c and recompiling?
Thanks for adding this stuff!
Bill
Wilhelm K. Schwab, Ph.D.
bills at anest4.anest.ufl.edu
(352) 846-1285
More information about the Squeak-dev
mailing list
|