Subject: Re: BRAD vs current vs Ethernet
To: None <macbsd-development@NetBSD.ORG>
From: John D. Smerdon <jds@smerdon.plymouth.mi.us>
List: macbsd-development
Date: 02/19/1995 12:22:33
At 12:07 PM 2/19/95, I wrote:
>I just compiled a current kernel from a sup on 2/17.  The current kernel
>displays "starting network" and then hangs.

I made some random changes to the kernel based on Brad's patches to 1.0 and
the current sources.

In via.c in nubus_intr(), Brad had logic to ignore nubus interrupts that
were pointing at slot_noint().  On my SE/30, slotitab[i] is called the
first time when i = 0 and 5.  Adding the "ignore" logic in similar places
in the current code, it successfully boots.

During the boot process, ae0 displays a timeout message.  I placed a debug
printf() in ae_intr and it is getting called once for unit 0.

The ethernet interface doesn't work after it finishes booting.

In case you don't have it, Brad's code was:
>long nubus_intr(bit)
>int bit;
>{
>        int i, mask, ignore;
>
>        ignore = 0;
>        do {
>            mask = 1;
>            for (i = 0; i < 6; i++) {
>                if ((via_reg(VIA2, vBufA) & mask) == 0) {
>                        (*slotitab[i])(slotutab[i], i+9);
>                        if (slotitab[i] == slot_noint)
>                                ignore |= mask;
>                }
>                mask <<= 1;
>            }
>        } while (((via_reg(VIA2, vBufA) & 0x3f) | ignore) != 0x3f);
>
>        return 1;
>}

--
John D. Smerdon; Plymouth, Michigan; Contents are my opinion.
Internet: jds@smerdon.plymouth.mi.us; UUCP: umcc!smerdon!jds