[Prism54-users] Help: SMC2802W - firmware loading fails, eth0 does not start

[Karl-Heinz Herrmann]

On Wed, 14 Jan 2004 01:18:43 -0500
Milan Zimmermann <milan.zimmermann@sympatico.ca> wrote:

> Jens,
> 
> Thanks for your reply - I am getting back to this a week later, was
> gone last week.
> 
> I do have firmware.agent in /etc/hotplug, it is executable, still the
> same problem. I am completely confused as to what to try next, I
> double checked everything I read in the docs... 
> 
> I'd like to get some suggestion on how to debug this problem, for
> example, can I put some echos inside the firmware.agent and see what
> the $DEVPATH and $FIRMWARE are set to?  Where do they come from when
> firmware.agent is executed? If I set the DEBUG=yes, where is the
> output going to be? 


Hmm... what Distribution is that? I had a SuSE 7.2 or 3 and there was a
hotplug and firmware.agent in place. They have it in /etc/hotplug with
an unused /etc/hotplug.d/ next to it. My firmware goes to
/usr/lib/hotplug/firmware/isl3890 (check place and spelling again -- got
me fooled for some time, number twister in the 38...)

Then I think the firmware was loading once I try to get eth1 up. But
this was not working on insertion, only by doing the ifconfig myself.
I hacked the net.agent to actually get to the place where a network is
setup and instead of fooling around with more script I just issued the
iwconfig/ifconfig. 

The script sources another source where msg() and debug_msg() are
defined. They in term use "logger" and this should therfore go to sylogd
and be logged somewhere in the syslog. For temporary debugging until I
found the if [] which was preventing to even reach the place I was
looking at I put direct "logger" debug messages like "entering script
xy...", "valid net device, interface is ...", "trying the ifconfig",
etc. 

Mind that this can look quite different on another distri but the
general holds true.

 Hope this helps,


K.-H.