| > | In zssoft() near line 1016, the input is dispatched off to the
| > | input handler for the type of channel (kbd/ms/tty).  This is
| > | one place where a layered driver should just call an input
| > | handler using a function pointer set during attach...
| > 
| > this did the trick
| 
| Great.  I assume you also changed the NZS define as needed.

No, as best I could tell NZS is the number of zs chips, and
should stay 2. just commented out the one line in zsopen, and
the few in zssoft. I left the blocks in zsattach alone, as leaving
it didn't seem to hurt.


| Well, my (3/60) test machine has been pretty stable, so send
| bug reports.  (reproducible bugs preferred :-)
| 
| There is one strange bug I've been having trouble reproducing:
| When I copy a file from NFS to either NFS or SCSI, the file
| will often have a block of garbage where the garbage looks
| suspiciously like the client's Ethernet address.  I suspect
| the "le" driver, and may try switching to mycroft's latest.
| 
| Anyway, if this occurs on diskless machines, the result will
| be seriously broken programs (i.e. unexplained SIGSEGV's).

This could be it. I'm running a 3/50 completely diskless. At bootup
a few daemons drop core (no real pattern as to which) and after some
use the core droppings increase, to the point where anything typed at
a prompt responds with a seg. fault or illegal instr. and I reboot.


	--jeff