Hi,

I read nubus.h, if_snvar.h and if_ae_nubus.c, now I understand the reason.

CABLETRON is a special case.

APPLE ETHERNET NB       DP93934 SONIC-T
APPLE ETHERNET LC       DP93934 SONIC-T

ASANTE MC3NB              DP83902 ST-NIC
ASANTE MacLC III          DP83902 ST-NIC
ASANTE MCLC                DP83902

CABLETRON E6119 (NB)  DP83902
CABLETRON E9119 (LC)  DP83932

At 9:06 AM -0500 04.1.15, Michael R.Zucca wrote:
> Maybe the ae and sn drivers can be modified to check more than vendor
> ID to verify that the card is really an 8390 vs a SONIC. The sn driver
> could also be modified to check for the CABLETRON vendor, which it
> currently does not.

I tried to set CABLETRON Vendor type in if_snvar.h and nubus.h,
but I found myself immediately "I have NO skill to do that" :)
I shall be very glad if I'm a tester for that.

Thanks a lot,
Kazu Inanaga


At 9:06 AM -0500 04.1.15, Michael R.Zucca wrote:
>On Jan 13, 2004, at 8:15 AM, Kazuyuki Inanaga wrote:
>
>>> ae0 at nubus0 slot e: failed to determine size of RAM.
>>
>> Cabletron E9119 DNI Ethernet Card (it seems same card as yours)
>> has DP83932 which is SONIC-based "sn*", not DP8390 based "ae*".
>> I wonder why NetBSD recognizes it as "ae*".
>
>The ae driver seems to check the vendor ID and not much else during the
>match phase. The sn driver also checks the vendor ID during the match
>phase but CABLETRON isn't one of the vendors in the current sources.
>Perhaps the ae driver spots a vendor ID of CABLETRON and assumes it can
>control the card, shutting out the sn driver.
>
>Maybe the ae and sn drivers can be modified to check more than vendor
>ID to verify that the card is really an 8390 vs a SONIC. The sn driver
>could also be modified to check for the CABLETRON vendor, which it
>currently does not.
>
>--
>----------------------------------------------
>  Michael Zucca - mrz5149@acm.org
>----------------------------------------------
>  "I'm too old to use Emacs." -- Rod MacDonald
>----------------------------------------------