[comp.unix.i386] 386 interrupts, pcnfs

montgome@udel.EDU (Kevin Montgomery) (08/31/89)

Does anybody run Sun's pcnfs software on PC's and a 386 system V machine??
I have been haveing trouble bringing such a system up, and would love to hear
from anybody who has been successfull.

In particular, the standard PC boards only seem to support the 'non cascaded'
interrupts on the AT.  How can you run 2 standard serial ports, a parralel
port, buss mouse, ethernet board, etc??  On the unix server side the drivers
might be smart enough to poll multiple devices on 1 interrupt line.
What can be about the pcnfs dos pc clients??


Any suggestions would be appreciated, (even to RTMF if you could steer
me the right way).


-- 
Kevin Montgomery

izen@amelia.nas.nasa.gov (Steven H. Izen) (08/31/89)

In your article <22902@louie.udel.EDU> you ask:

>In particular, the standard PC boards only seem to support the 'non cascaded'
>interrupts on the AT.  How can you run 2 standard serial ports, a parralel
>port, buss mouse, ethernet board, etc??  On the unix server side the drivers
>might be smart enough to poll multiple devices on 1 interrupt line.
>What can be about the pcnfs dos pc clients??


I had similar problems getting 2 serial ports, 2 parallel ports, a tape drive,
and a bus mouse all to work under 386/ix 2.0.1.  I ran out of interrupts as
well.  My solution was a hardware hack:  I redirected the interrupts from both
parallel ports to the cascaded interrupt controller, more specifically 11 and
12.  ISC's parallel driver works with these interrupts.  Their serial driver
can't handle this hack-I learned that the hard way.  Fortunately I did the 
hardware modification in such a way that restoring my serial board to interrupt
4 was easy. (I just flipped a dip switch).  
	Incidentally, the only side effect of the interrupt hack that I've
discovered so far is that under real-mode DOS (yuck, but I have to run
Mathematica under DOS) the Logitech mouse driver can't find any interrupt
jumper even though I have it on int 5.  (It does work under VPIX, however).
Hey ISC, this may be a first!  Something works under VPIX but not real-mode
DOS! :-)  Anyway, since DOS printing is not interrupt driven, the hack doesn't
affect printing under real-mode DOS.


Steve Izen

-- 
Steve Izen: {sun,decvax,uunet}!cwjcc!skybridge!izen386!steve
or steve%izen386.uucp@skybridge.scl.cwru.edu
or izen@cwru.cwru.edu

keithe@tekgvs.LABS.TEK.COM (Keith Ericson) (09/01/89)

In article <22902@louie.udel.EDU> montgome@udel.EDU (Kevin Montgomery) writes:
>
>Does anybody run Sun's pcnfs software on PC's and a 386 system V machine??
>I have been haveing trouble bringing such a system up, and would love to hear
>from anybody who has been successfull.

I have an Everex STEP/25  that runs either [MS-DOS+PC-NFS using a
WD8003E/PLUS] or AT&T SysVR3.2 (from Intel) with Wollongong's TCP/IP
and NFS (with the same WD8003E/PLUS, of course.)

The WD8003E/PLUS is jumpered to what it thinks is interrupt 2.  I
configured the PC-NFS software to use IRQ2 but the Wollongong to use
IRQ9.  It works with two/three comm ports (1&2, plus comm3 as long as I
don't try to use 1 & 3 at the same time: comm 2 is my serial mouse,
comm 1 is hooked to the inhouse communications system and comm3 is an
internal modem), and a printer port.

kEITHe