On Sun, 30 Aug 1998, Jason Thorpe wrote:

> On Sun, 30 Aug 1998 02:48:07 -0700 (PDT) 
>  Todd Whitesel <toddpw@best.com> wrote:
> 
>  > NetBSD/arm32 on my CATS system doesn't want to probe either of these boards;
>  > the firmware does not recognize the card as an ne0, and 'config' on arm32
>  > thinks that isapnp* and sb* are unknown devices, and won't let me add them
>  > to the kernel. I thought these were MI device drivers, so any ideas about
>  > how I might get them added to my arm32 kernel would be appreciated.
> 
> For the NE2000 problem, "take a look at how many chips there are that
> claim to be an NE2000" in sys/dev/pci/if_ne_pci.c
> 
> chances are the CATS firmware doesn't match all of those vendor/product
> combos.

Yep.

> Re. isapnp... you have to have Very Current sources to even have a
> prayer of using isapnp on an arm32 system... better press on that
> portmaster to get the CATS support code into the mainline :-)

I'm already flat ;-)

Actually the CATS/EBSA285 code is almost ready to drop into -current.
There are several problems outstanding that need fixing. Only one of these
is actually holding up the commit.

1. Interrupt / spl / soft int changes are currently breaking shark
kernels. RiscPC kernels are working fine but I am getting odd timeout
problems with shark kernels. (Once this is fixed the generic arm32 changes
can be committed).

2. I can only read from the raw floppy device. Reading the block device is
resulting is garbage data. (This does not prevent the CATS/EBSA285 code
being committed).

3. The CATS / EBSA285 1.3F sources have a bus mastering IDE driver that
will not be committed. This will mean a loss of performance on disk I/O
until the NetBSD MI driver supports it.

Cheers,
				Mark