[vortex] Problem with 3com 10/100 Mini PCI adapter (3c556B?)under 2.2.19

[Andrew Morton]

"Akihiko YAMAMOTO ($B;3K\L@I'(B)" wrote:
> 
> ;; "star>" == star@eos.hokudai.ac.jp (Akihiko YAMAMOTO ($B;3K\L@I'(B)) wrote:
> 
> star> I have not tried his drivers under 2.4.6 because compilation
> star> failed.
> 
> I tried Donald's drivers (-Re) under 2.4.6.  Compilation was
> successful, but loadable module cannot be installed properly...
> 
> # /sbin/insmod /lib/modules/2.4.6-a/kernel/drivers/net/3c59x-donald.o
> /lib/modules/2.4.6-a/kernel/drivers/net/3c59x-donald.o: unresolved symbol pci_write_config_byte
> /lib/modules/2.4.6-a/kernel/drivers/net/3c59x-donald.o: unresolved symbol enable_irq
> /lib/modules/2.4.6-a/kernel/drivers/net/3c59x-donald.o: unresolved symbol eth_type_trans
> ...
> ...
> 
> Am I wrong in compilation or installation of Donald's driver
> under kernel 2.4.6?

Those symbols should definitely be resolvable - make sure the
base kernel is compiled without module versioning, and that
it's a uniprocessor build.

You could experiment with mii-diag while the NIC is online.
It has various options for resetting the MII, restarting 
autonegotiation, etc.

Also, I'd be interested in seeing the log output from

	modprobe 3c59x debug=7
	ifup eth0

under 2.4.

We're seriously needing documentation for the 556B.