Red Hat Bugzilla – Bug 26852
ethernet bootup 3c509
Last modified: 2014-03-16 22:18:52 EDT
From Bugzilla Helper:
User-Agent: Mozilla/4.76 [en] (X11; U; Linux 2.4.0-0.99.11 i586)
During boot up of interface eth0, insmod fails, and eth0 is thus not
If I use modprobe 3c509 at init 1 level or init 5, eth0 is made available
with no complaints. This was not a problem on ver 7.0
Steps to Reproduce:
1.Configure a 3Com 3c509 card (pnp type on an ISA bus)
2.Observe the messages when bringing up interface eth0 (or what ever eth?)
3.alternately boot to single user and issue modprobe 3c509 and continue the
In this later case there will be no complaints --or--
Allow boot to init 3 or 5 (the complaint will occur when bring up the eth
device. Use modprobe 3c509 and it will bring up eth without complaint.
Actual Results: In the default bootup, the eth device is not available
If you use modprobe the eth devices is correctly configured and available.
The complaint issued is 'insmod: module ..... not found'. The module is in
the correct location.
Expected Results: Bring up interface eth0 should have reported as OK,
rather than failed.
I also have tried the SMC-ULTRA card and in that instance, while I set
linuxconf to io=0x300, the 'bring up' fails and if I force eth0 up (
ifconfig eth0 up) it reports the io=310 :(( and of course the interface
does not work.
This may be a related problem. The SMC card is hardware jumper configured,
while the 3Com card is PNP configured
What does 'modprobe -v eth0' say?
command modprobe -v eth0
Note: /etc/modules.conf is more recent than
We (Red Hat) should really try to resolve this before next release.
I'm sorry, what does 'modprobe -v eth0' say, when the network device
isn't up and the module isn't loaded?
I have found that a conflict maybe configured between SB16 and eth0 (3c509).
In the file /etc/modules.conf, a line 'options eth0 io=0x300 irq=10' causes an
invalid io parm. I suspect PNP assigns an io address that is not the value set
I deleted the io parm in the line. The modified line is 'options eth0 irq=10'.
This boots without conflict.
If I eliminate the line, then I get a collision of IRQ (both SB16 and 3c509
attempt to use IRQ 5).
The fix appears to be to only set the 3c509 irq option.
Note: In linuxconf, I specify the module as 3c509. I do not specify the IO or
IRQ.. It seems that I should be specifying the IRQ ONLY.
This sounds like a configuration issue, in that that specific
io parameter will not work on your machine. That's not something we
really can fix on a global basis, if I'm reading this right.