Re: strange ne2000ish card on A1200

[is%netbsd.org@localhost]

Hi,

On Sat, Mar 28, 2020 at 10:49:31AM +0100, Karoly Balogh (Charlie/SGR) wrote:
> Hi,
> 
> On Fri, 27 Mar 2020, is%netbsd.org@localhost wrote:
> 
> > On Fri, Mar 27, 2020 at 11:16:12AM +0100, Karoly Balogh (Charlie/SGR) wrote:
> > > Hi,
> > >
> > > On Fri, 27 Mar 2020, is%beverly.kleinbus.org@localhost wrote:
> > >
> > > > Ah, I see (after reading sys/dev/ic/ne2000.c). I'd guess it's a
> > > > AX88190 variant, like the EDIMAX EP4101 which has the same product
> > > > ID of 0x1090,
> > > > (...)
> > > > in your patched if_ne_pcmcia.c to
> > > >
> > > >     { PCMCIA_VENDOR_INVALID, 0x1090,
> > > >       { "PCMCIA", "Fast-Ethernet", NULL, NULL},
> > > >       0, -1, { 0x00, 0x40, 0xf4 }, NE2000DVF_AX88190},
> >
> > semi-wild guess after staring at the AX88190 data sheet:
> > try to replace the -1 by 0x400.
> >
> With this change, I get again the "en0: can't match ethernet vendor code"
> error, using NE2000DVF_AX88190.

What do you get if you put

0xff, 0xff, 0xff instead of 0x00, 0x40, 0xf4?

Also it might be time to put

options PCMCIACISDEBUG

into the kernel configuration (e.g. into GENERIC.local), 
configure and build the kernel again, then

boot once into the debugger (see "man ddb")

db> examine pcmciacis_debug/l
db> write pcmciacis_debug 1
db> continue

which will print the metainformation on the card, as we manage to read,
incredibly verbose (but less raw than card_read did)...

Regards,
	-is