Hello,

I'm developing a set of lowelevel drivers for the A2410 "University of
Lowell" card.

My sources are as of 1995 Mar 6.

The ite_ul.c creates instructions for a server program running on the
TMS34010 processor on the A2410, and writes them into a circular
buffer (on the card).

If using busy waiting:

fetch_new_value(writeptr);
newwriteptr := writeptr+1;
fetch_new_value(readptr);
while (newwriteptr == readptr) {
	fetch_new_value(readptr);
}
write_stuff(source,*writeptr);
put_writeptr_to_card(newwriteptr);

the buffer codes seems to work under even under heavy (ab)use.

If using a tsleep() call:

...
while (newwriteptr == readptr) {
	tsleep((void *)gsp,0,"ulwait",hz/10); 
	fetch_new_value(readptr);
}
...

and I interrupt a ls -lR /usr | dd of=/dev/ttye4, the system hangs. No
disk activity, no display change (both of them), no cursor reaction on
mouse movement. ctrl-A-A helps.

The above code is called from the 6 ul_somethingorother() functions,
which are called by ite.c.

Should I protect the stuff above by splxxx() for some suitable xxx?
Which one? 

Are there any known problems with tsleep() fixed after Mar 6?

Regards,
	Ignatios Souvatzis