[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Woody with 2.4.18-bf2.4: network interface configuration



I'm very puzzled about how network device configuration is actually
done under Woody with the "of-the-shelf" kernel 2.4.18-bf2.4. The
reason why I need to know is I tried have compiled a custom 2.4.26
kernel (not with kernel-package, but the traditional way, with the
source from kernel.org). If I boot this new kernel, the network
interface configuration seems to fail. An extract from the startup
messages (tabs and the backslash added, the latter to indicate there
is no line break on the screen, just here):

        Setting up IP spoofing protection: rp_filter.
        modprobe: modprobe: Can't locate module net-pf-17 
 -->    Configuring network interfaces: modprobe: modprobe: Can't\
 -->    locate module net-pf-17
 -->    done.
        Starting portmap demon: portmap.


And the network won't work. This is not so when I boot the
"off-the-shelf" 2.4.18 kernel. The messages will instead go


        Setting up IP spoofing protection: rp_filter.
-->     Configuring network interfaces: SIOCADDRT: file exists
-->     done.
        Starting portmap demon: portmap.


and the network works (however, the " SIOCADDRT: file exists" part is
sometimes there and sometimes not). No mention of module net-pf-17.

I am not even sure whether this unfound module net-pf-17 has to do
with this or whether the modprobe error message is rather somehow
intermingled with this. I have found nothing to support the former
idea.

I have found that the message is generated by /etc/init.d/networking
which calls "ifup -a" in between. "ifup", called interactively, claims
that "interface eth0 [is] already configured". dhclient is completely
tacit, no matter what option specified. route, however, displays an
empty routing table (which looks normal under 2.4.18).

Calling "ifconfig eth0" does not change the situation, but displays
information which looks quite normal compared to that on the computer
I am writing on, and where the network works (however, the number of
RX and TX packets is very low (68 vs. 1927 and 4 vs. 2051), as is the
number of RX and TX bytes (8011/1060250 and 812/297367) - it seems
this simply reflects past network traffic, but then I'm wondering why
there is any traffic at all).

So I have almost no idea what is going on. I know two things: 1.) it
must have to do with the different kernel/modules combo, since
that is the only difference, and 2.) the problem is not a missing
driver for the card (eepro100), since that is present in both
combinations (and eepro100 is loaded with 2.4.26, but not used, while
with 2.4.18, it is loaded AND used).

Can anybody point me out where to look for the culprit?

Thanks for any help!

-- 


Florian v. Savigny

If you are going to reply in private, please be patient, as I only
check for mail something like once a week. - Si vous allez répondre
personellement, patientez s.v.p., car je ne lis les courriels
qu'environ une fois par semaine.



Reply to: