In message <19398D273324D3118A2B0008C7E9A569037B47A7@SIT.platinum.corp.microsof
t.com>, "Adam Glass (Exchange)" writes:
>umm..you probably didn't need the cardbus code at all for this card.  i
>doubt anyone makes a cardbus ne2000 card...this is probably a pcmcia r2
>card.

I know, but I need the cardbus code to support my CF cards.

Anyway, I'm having a bear of a time figuring out *WHICH* "socket_disable"
gets called.  The problem is that some function with a name like
pcmcia_chip_socket_disable gets called.  This is a macro which refers to
the socket_disable member of a structure.  Unfortunately, my guesses (so
far) as to *which* "socket_disable" function it is have all been wrong.

I could just switch back to the pcic code, but I really want this to work
with CardBus.  At that point, NetBSD will have, so far as I can tell,
"complete" support for my laptop, including every kind of peripheral I want
to run on it.  Very cool!  ;-)

Ah-hah.  Duh.  It's in the pccbb code.  Anyway, it looks like the first write
to the socket is crashing.  Note that it works fine with anything but the
ne2000 in the socket...

The code in question is
	DPRINTF(("pccbb_pcmcia_socket_disable\n"));

	/* reset signal asserting... */

	intr = Pcic_read(ph, PCIC_INTR);
	intr &= ~PCIC_INTR_RESET;
	Pcic_write(ph, PCIC_INTR, intr);
	delay(2*1000);

Someone suggested it could just be an IRQ problem.  I didn't see any messages
about IRQ's; what am I looking for?

-s