[comp.unix.i386] Ethernet boards for ISC 386/ix

lance@belltec.UUCP (Lance Norskog) (08/11/89)

In article <1423@hydra.gatech.EDU>, gb7@prism.gatech.EDU (Joe Bradley) writes:
> 
> The only other potential problem is that the ISC sysadm script for
> configuring host based TCP/IP wants to set an IRQ of 3 to 7 for this
> board.  Well, all of those are taken in my present configuration, and
> the board will support IRQ 2.  Thus, I edited the sysadm script to allow
> me to set an IRQ of 2 (and of course set the board likewise).  Assuming
> that's not causing a problem, then I guess I have to conclude the board
> just plain won't work in my machine because the clock rate is too high. 
> 

Interrupt 2 on the card is interrupt 9 in the operating system, so you 
want to sysadm with 9 and then set the WD interrupt jumper at the left-hand
end (interrupt 2).  The WD driver in Streamlined TCP figures out if this
is the problem and prints out a message on the console, I don't know if the
386/ix driver does.

The motherboard mhz is not the problem, the Dell machine should clock the bus
somewhere between 6-10mhz and the WD should work fine.  I think the WD cards 
come with a disk that has a diagnostic on it; did you run that?

The device driver for the WD board should do a complete check on the card
when the system comes up (the one in Streamlined TCP does).  You can test
the digital and analog signal path at various points and make sure the 
shared memory works.

The WD driver in 386/ix was written by Western Digital, and is available in
source form from them, if all else fails.

Lance Norskog
Sales Engineer
Streamlined Networks
415-659-1450

jr@frog.UUCP (John Richardson) (08/12/89)

In article <201@scooter.PLSN.CA.US>, carlson@scooter.PLSN.CA.US (Joe Carlson) writes:
> In article <1423@hydra.gatech.EDU>, gb7@prism.gatech.EDU (Joe Bradley) writes:
> > 
> > I just bought a Western Digital EtherCard Plus (WD 8003e) to use with
> > ISC 386/ix.  Unfortunately, the WD manual says it only works in machines
> > with clock rates of 16 MHz or less.  I have a Dell 310 running at 20
> > MHz.  I thought I would give it a try anyway.  Not surpisingly, it
> > doesn't work.  ........
> 
> 	Hmmm... I cannot find any such limitation mentioned in the WD8003E
> docs that I have (I have heard they will not work in machines that run the
> AT bus at .gt. 8Mhz).  I also have 2 Dell 310's running SCO XENIX and various
> TCP/IP packages all with WD8003E cards... all working just fine.

  I use a WD8003-EBT (32K) board in a 33 Mhz motherboard with out any
problems. I am running a version of our real time UNIX and have taken care
to insert delays between access to the control registers on the NIC chip
to give it time to recover.
  Another issue is that some motherboards will not pay attention to the
WAIT line, and if this is the case, an access while the NIC chip is DMA'ing
to local on board memory will not be held off properly.

					JR

wek@point.UUCP (Bill Kuykendall) (08/15/89)

>The only other potential problem is that the ISC sysadm script for
>configuring host based TCP/IP wants to set an IRQ of 3 to 7 for this
>board.  Well, all of those are taken in my present configuration, and
>the board will support IRQ 2.  Thus, I edited the sysadm script to allow
>me to set an IRQ of 2 (and of course set the board likewise).  Assuming
>that's not causing a problem, then I guess I have to conclude the board

Ya gotta watch that 'assuming' stuff!  IRQ2 may well be your problem.  I've
had a number of problems with IRQ2 under ISC 2.01.  Come to think of it, I
haven't gotten *anything* to work on IRQ2 since I upgraded from 1.0.6.  As a
test, pull something else out and try this board on another interupt.  If it
works, you can worry about what to keep later.

rcd@ico.ISC.COM (Dick Dunn) (08/16/89)

> >...board will support IRQ 2.  Thus, I edited the sysadm script to allow
> >me to set an IRQ of 2 (and of course set the board likewise)...
>...Ya gotta watch that 'assuming' stuff!  IRQ2 may well be your problem.  I've
> had a number of problems with IRQ2 under ISC 2.01.  Come to think of it, I
> haven't gotten *anything* to work on IRQ2 since I upgraded from 1.0.6...

In a strict sense, you can't use IRQ2, period.  IRQ2 is the line for the
second interrupt controller on an AT.  That is, there are two interrupt
controllers in the machine, but they're cascaded.  One can yank directly on
the 386 interrupt-request; the other yanks on an interrupt line of the
first.

Some of the old configuration code may have treated IRQ2 as if you'd said
IRQ9--that is, it may have bypatched the attempt to put a device on IRQ2
and put it on the second line on the slave controller, which is 9.
-- 
Dick Dunn     rcd@ico.isc.com    uucp: {ncar,nbires}!ico!rcd     (303)449-2870
   ...Are you making this up as you go along?

dennis@virtech.UUCP (Dennis P. Bednar) (08/17/89)

In article <2642@dell.dell.com>, james@raid.dell.com (James Van Artsdalen) writes:
> 
> There are a number of reasons why IRQ 2 may not work.  In particular,
> some video cards (Video 7 16 bit comes to mind) drive IRQ 2 both active
> *and* inactive.  Also, on an AT compatible, the bus line labelled IRQ 2
> is actually connected to IRQ 9, since the real IRQ 2 is used to cascade
> the second interrupt controller.
> 

I'm still a little unclear on how the two PIC's (Programmable
Interrupt Controller chips) are actually connected to the bus,
to one another, and to the CPU.

I have a vague idea that that there are 8 input IRQ lines per PIC,
but that when you cascade two PIC's you don't get all 16 IRQ levels
because of the daisy-chaining.   I suppose there is only one
INT_REQ (interrupt request) line that goes to the CPU, and
depending on the IRQ pin and how the PIC's were daisy-chained
you would get one one PIC responding with the interrupt vector
number information.  But I still can't picture how the bus is
connected to the PIC's and to the CPU.  Can someone draw a
figure or elaborate in a little more detail please?