On Sat, 20 Jan 2001, Allen Briggs wrote:

> Cha-ching.  To get things working with that video card, yeah, you just
> have to figure out how to clear the interrupts from that card.  It
> should be something pretty simple.  I wrote some instructions up at
> one point that explain what you need to do, but it sounds like you
> have a pretty good idea.  The other cards that have interrupt-clearing
> code are in src/sys/arch/mac68k/nubus/grf_nubus.c

Okay, I'll try and get this working soon. Thanks again. ^_^

> > > How are you checking for interrupts?
> > The last step in ae_nubus_attach() is to call add_nubus_intr(). Is this
> > what you are referring to?
> 
> No.  I was wondering how you knew that you weren't getting any nubus
> interrupts.

Well, something was wrong because I kept seeing "device timeout" messages.
And since adding a #define DEBUG at the top of if_ae_nubus.c got things
"working" (because it called the interrupt routine "by hand"), I figured
that the interrupts were being generated by the card (there were some
packets recieved), and that the interrupt was disappearing between there
and the ae driver (since the ae driver was obviously working).

> The behavior that you saw was a little odd in that I'm used to seeing
> the system wedge when the first ethernet interrupt is received because
> the video card is also interrupting and it keeps the nubus interrupt line
> asserted.
> 
> -allen

---------------------------
All programming can be viewed as an exercise.
---------------------------
Alastair Bridgewater
e-mail: nyef@softhome.net