netdev
[Top] [All Lists]

Re: [RFC] multiple eth0 mixed PCI/ISA init

To: Stephen Hemminger <shemminger@xxxxxxxx>
Subject: Re: [RFC] multiple eth0 mixed PCI/ISA init
From: Zoltan Farkas <zfarkas@xxxxxxxxxxxxxxx>
Date: Sat, 20 Dec 2003 12:33:34 +0100 (MET)
Cc: Jeff Garzik <jgarzik@xxxxxxxxx>, Alexander Viro <aviro@xxxxxxxxxx>, <netdev@xxxxxxxxxxx>
In-reply-to: <20031219151415.1cfdbe66.shemminger@xxxxxxxx>
Sender: netdev-bounce@xxxxxxxxxxx
On Fri, 19 Dec 2003, Stephen Hemminger wrote:

> PCI is getting initialized before the old ISA probing code, so your PCI card
> is grabbing eth0 before the ISA card gets started.
>
> You might be able workaround the problem probably by doing:
>       netdev=0,1,0,0,eth0
> which should cause the ISA probe code to skip eth0 and go on to eth1
>
> This patch should fix 2.6.0 (have similar but different fix for 
> net-drivers-2.5-exp).
> I don't have an ISA card to actually test this code path.
>
> diff -Nru a/drivers/net/Space.c b/drivers/net/Space.c
> --- a/drivers/net/Space.c     Fri Dec 19 15:10:50 2003
> +++ b/drivers/net/Space.c     Fri Dec 19 15:10:50 2003
> @@ -350,7 +350,7 @@
>        * Backwards compatibility - historically an I/O base of 1 was
>        * used to indicate not to probe for this ethN interface
>        */
> -     if (dev->base_addr == 1) {
> +     if (__dev_get_by_name(dev->name) || dev->base_addr == 1) {
>               free_netdev(dev);
>               return -ENXIO;
>       }
>
Thanks a lot! This has fixed my problem. Now my ISA card initializes as
eth1.
Cheers,

        Zoltan


<Prev in Thread] Current Thread [Next in Thread>