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

Re: can't detect NIC



> Susan G. Kleinmann wrote:
> > I'm tyring to install the i386 disks from 'testing' on a Toshiba 1805-S203,
> > to which I've added a Linksys "Combo" ethernet card.
> 
John H. Robinson wrote:
> you didn't mention which flavour (if any) you were using.

I just used the plain vanilla set of disks:
woody/main/disks-i386/3.0.13-2001-08-25/images-1.44/{rescue,root}.bin
and
woody/main/disks-i386/3.0.13-2001-08-25/images-1.44/driver-{1,2,3,4}.bin

> > I guess that somehow the system doesn't know that eth0 should be
> > identified with the pcnet_cs driver.  How can I tell it?
> 
> i found something similar with my Xircom CreditCard Ethernet CE3B-100BTX
> card - some versions (3.1.28) of pcmcia-card-services detect it as a
> memory card.

The version of cardmgr that is on the system is indeed 3.1.28.  

> however, i found that compiling the 3.1.22 modules, and using it with
> the 3.1.28 cardmgr (as is found on bf-3.0.13 and greater) provides
> satisfactory results.
I guess I'm a little confused here.  Do you mean that I should find a
very old version of pcmcia-source (3.1.22 rather than the current 3.1.28),
and then compile that (with some kernel or other)?  I take it that when/after
I do that I should place the result *somehow* on the rescue or root disks.
I emphasize "somehow" because today I ran into another set of problems
which I guess would have to be solved first:
I got the boot-floppies source and just ran 'make' on it, and then
tried to peer at the results.  I was able to mount the rescue.bin file on a 
loop device:
    losetup /dev/loop0 resc1440.bin
    mount /dev/loop0 /mnt
    umount /mnt
but I was not able to mount root.bin or root1440.bin the same way:
    losetup /dev/loop1 root.bin
    mount /dev/loop1 /mnt
mount: you must specify the filesystem type

When I guess 'ext2' I got the "wrong fs type, ..." message; ditto for msdos.
So I don't know how to check the contents of root.bin, or add/replace files
on it.

> to gain even more feedback as to what is going on, either start a shell,
> or go over to VC2 and start the shell there, and check the output of
> dmesg.

I gave cardmgr the option "-v", and then looked at the output of dmesg.
It looks as if the system is trying to put the card at IRQ 11.  
I notice that WindowsME (the other OS on this laptop right now), seems
to put the card at IRQ 5 at I/O 0x300.  I see the messages:
"PCI irq 11 test failed" and later
"axnet_cs: unable to read hardware net address for io base 0x300"

How can I force cardmgr to pick up pcnet_cs instead of axnet_cs?
These attempts (admittedly blunderbuss) failed:
-- Erase axnet_cs.o before cardmgr has a chance to find it;
   and erase the axnet line in modules.dep.  The system still failed
   to pick up pcnet_cs by itself.
-- Copy the pcnet_cs.o file to axnet_cs.o and copy the dependencies
   for pcnet_cs.o to the dependencies for axnet.o in modules.dep
   The system appeared not to have read the modules.dep file under 
   /target/lib, because it didn't properly pick up the depended-on modules.

If success here is just a matter of forcing cardmgr to try irq 5, then 
how do I do that?  In particular, what answer should I provide on one of
the dialog boxes related to setting up PCMCIA services?

In an entirely different approach:
I tried to circumvent the whole problem today by copying the base files
to some directory in the Windows partition, and then telling the 
installation script that I wanted to install the base system from a mounted
disk.  After a number of iterations, I got stuck at the "invalid 
release file" message.

At this point, I don't see a way to install Debian, short of using
the potato floppies.   Any alternative ideas appreciated.

TIA,
Susan



Reply to: