Re: tulip.c + Netgear FA511 + RH9 + 2.4 kernel + Omnibook 800CT = nothing back

Donald Becker <[email protected]> Thu, 14 Aug 2003 13:40:43 -0400 (EDT)
Newsgroups gmane.linux.drivers.tulip.general
Message-ID <Pine.LNX.4.44.0308141330360.1714-100000@beohost>
On Wed, 13 Aug 2003, Jon Fairbairn wrote:

> Date: Wed, 13 Aug 2003 21:17:39 +0100
> From: Jon Fairbairn <[email protected]>
> To: [email protected]
> Subject: [tulip] tulip.c + Netgear FA511 + RH9 + 2.4 kernel + Omnibook
>     800CT = nothing back
> 
> I'm trying to get the above PCMCIA ethernet combination to
> work, with limited success. When tulip.o loads it complains
> about the BIOS not having enabled the device and then
> reports
> 
> eth0: ADMtek Centaur-C rev 17 at 0xc283e000, 00:09:5B:1E:AD:D7, IRQ 5.

This looks good.  You have a Centaur CardBus chip made by ADMtek.  The
station address is reasonable, and you have an assigned IRQ.

> eth0: tulip_open() irq 5.
> eth0: MII link partner 0021, negotiated 0021.
> eth0: Done tulip_open(), CSR0 fff98000, CSR5 fc124010 CSR6 ff972113.
> eth0: Added filter for 01:00:5e:00:00:01  7fa32d9b bit 62.
> eth0: Comet link status 786d partner capability 0021.
> eth0: MII link partner 0021, negotiated 0021.

Good.  You are talking to a 10baseT-only link partner.

> eth0:  MII transceiver #1 config 3100 status 7849 advertising 05e1.

  7849  mean link beat was lost at some time in the past.  The link
  status is a "sticky bit", so we must read the register a second time
  to find the current status.

> eth0: Comet link status 786d partner capability 0021.

  786d means that you currently have valid link beat.

> I've put other diagnostic outputs in the same directory
> <URL:
> http://www.cl.cam.ac.uk/~jf/FA511/tulip.c-v0.97/2003-08-12/
> >, including mii-diag and tulip-diag. I've tried to make the
> filenames indicate the command-line arguments.

Run 'tulip-diag -a' after the interface is started, and watch for
changing counts on IRQ 5 in /proc/interface when you try to send
packets.  My first guess is that you have an IRQ problem, which might be
the PCMCIA (really "CardBus") control code.

> I'd guess that either I'm making a naïf mistake, or the
> eeprom in the card needs some poking, or it's significant
> that it's seeing the card as coming from a different
> manufacturer.

Nope, this is a regular ADMtek Comet CardBus card, and the EEPROM
programming looks reasonable.  It doesn't matter who puts the sticker on
the outside. 

> The installation is otherwise OK -- I can use the network if
> I plug in a 3Com 3C589D and use the appropriate driver,
> though the cable for that is wearing out, hence the new
> Netgear card.

A 3c589 is a PCMICA (16 bit) card, which uses different IRQ routing.

-- 
Donald Becker				[email protected]
Scyld Computing Corporation		http://www.scyld.com
914 Bay Ridge Road, Suite 220		Scyld Beowulf cluster system
Annapolis MD 21403			410-990-9993

_______________________________________________
tulip mailing list, [email protected]
To change to digest mode or unsubscribe visit
http://www.scyld.com/mailman/listinfo/tulip