[comp.sys.ibm.pc] PS/II Model 60, Interrupts, Dos

jrb@nbires.nbi.com (John Barr) (11/08/88)

	This is my second attempt at getting some info from one of you
	pseudo experts out there on the net. I didn't get a single reply 
	after my first attempt so I thought I'd try again.

	I'm having problems running a non-dedicated file server 
	on the PS/II (MS-Dos). We originally wrote it for the 
	AT and it has reached a somewhat mature state in both
	functionality and reliability. 

	It runs off of both INT 28(DosOk) and INT 70(timer)  on the AT. We had 
	hoped that this would also work on the PS/II but apparently our checks
	to determine if it is safe to make DOS calls when entered via
	INT 70 are not sufficient on the PS/II.

	In case you're wondering why INT 70(timer) instead of INT 1C (timer),
	it's because windows (among others) chains on INT 1C, and in doing so we
	can no longer determine exactly where (which segment) we were when 
	the interrupt occurred. That is to say we cannot determine where on the
	stack the segment address is unless we are the first interrupt in the
	chain. Hence the choice of INT 70.

	Currently when entered via INT 70 we check to see if

		1. We are re-entering ourselves/netbios/tcp.
		2. We were in IO.SYS (interrupted segment <= 070h)
		3. We were in an INT 21 call.
		4. We were in an INT 13 call.
		5  We were in ROM (interrupted segment >= 0a000h)
		6. We were in a Dos critical section(InDos Flag).

		before we attempt any Dos calls.


	I was wondering if anyone had any ideas as to what we were missing.
	If not does anyone know of a good technical book on the PS/II, I've
	not found the tech ref very helpful, and everything in the book store
	appears to be written for secretaries.

				

-- 

John R. Barr		
NBI Inc., Boulder, CO		jrb@nbires.UUCP or jrb@nbires.NBI.COM
(303) 938-2968