PCMCIA troubles

Dagmar d'Surreal no.spam at allowed.here
Sun Jan 12 14:56:29 PST 2003


On Fri, 2003-01-10 at 05:56, Aleksandr Guidrevitch wrote:
> > Actually this is normal.  Until the interface has been initialized with
> > an IP address, it's not going to appear in the output of
> > `/sbin/ifconfig` at all, and in some cases specifying the interface
> > explicitly like `/sbin/ifconfig eth0` is pretty meaningless as well. 
> > Perhaps you forgot to edit /etc/pcmcia/network.opts?  (hint hint)
> 
> No way I forgot :) Everything seems to be ok with config files
> 
> > If none of this works, upgrade pcmcia-cs to either 3.1.34 or 3.2.3 (my
> > own notebook just died recently, so I haven't tested the 3.2.x series)
> > and then post updated syslog output.
> 
> 3.2.3 is installed, is there a newer one :) ?

Something else is wrong with your installation then, since the kernel
log you posted showed you were running 3.1.22.  :/

> > BTW, you should consider updating pcmcia-cs to 3.1.34 (the latest 3.1.x
> > version) or 3.2.3 (the latest 3.2.x version).  I've gotten Linux going
> > on about 30 different types of notebooks and usually any problems I had
> > would disappear with an upgrade to the latest pcmcia-cs package.  
> hmm... lemme try to install 3.1.34 :)
> 
> 
> > You might also want to consider _not_ enabling the yenta module in the
> > kernel if it gives you fits.  On some older chipsets it still kills the
> > machine.  (You seem to be doing just fine with it tho.  :) )
> 
> Since I use kernel modules I'm expected to use yenta_sockets, at least 
> RH7.3 uses:)

It's not necessary.  The pcmcia-cs package was developed long before the
kernel got any yenta drivers, and basically operates perfectly fine
without any help from the kernel at all.

> Here is what I've found
> `/etc/pcmcia/network` fails at the line
> 
> grep_stab $1 <$STAB || usage
> 
> 
> grep_stab ()
> {
> local DEV
> while read SOCKET CLASS DRIVER INSTANCE DEV MAJOR MINOR ; do
> if ["$SOCKET" != "Socket" -a "$1" = "$DEV"] ; then return 0; fi
> done
> return 1
> }
> 
> of course grep_stab fails due to stab containing
> Socket 0: CardBus hotplug device
> Socket 1: empty
> 
> cardmgr starts ok, but seems it doesn't detect my card. On the other 
> hand `cardctl info 0` gives
> PRODID_1="Xircom"
> PRODID_2="CardBus Ethernet 10/100 + Modem 56"
> PRODID_3="CBEM56G"
> PRODID_4="1.03"
> MANFID=0105,0103
> FUNCID=6
> 
> `cardctl ident 0` gives the same thing
> product info: "Xircom", "CardBus Ethernet....
> manfid: 0X0105, 0x0103
> function 6
> 
> When I start /etc/rc.d/init.d/pcmcia start it produces the output but 
> doesn't beep even once (as expected), when I stop it - it beeps once 
> (the same when I plug/unplug the device).
> 
> Also, I have RH7.3 installed on the same laptop where my Xircom is 
> working ok. I've copied RH's /etc/ pcmcia/* as well as 
> /etc/rc.d/init.d/pcmcia - it doesn't help, the output is pretty the same :(
> 
> Where else can I dig ? Where the problem is ? The card type is not 
> detected by cardmgr ?

Nothing of what you're telling me leads me to believe that it's _not_
working.  It looks like you've just severely misconfigured it now.

Let's try this again.  Remove support for pcmcia from your kernel. 
Rebuild your kernel and be _sure_ you have eliminated all the stale
modules (rm -rf /lib/modules before you make modules_install if you have
to).  Remove everything from your /etc/pcmcia directory since you've
done something really bizarre with it.  Then rebuild and install the
latest pcmcia-cs package.  (After while I just got used to leaving it in
/usr/src/pcmcia -> /usr/src/pcmcia-3.1.34 right next to the kernel).

Things should then work just by executing `/etc/rc.d/rc.pcmcia start` or
wherever the init script landed.  I say again, you must edit
/etc/pcmcia/network.opts and assign it an IP address (dhcp is not
sufficient for reasons that should be obvious) or your network card will
never "appear".

-- 
Unsubscribe: send email to listar at linuxfromscratch.org
and put 'unsubscribe lfs-support' in the subject header of the message



More information about the lfs-support mailing list