Re: Firewire on Linux-2.4.20

[Richard B. Johnson]

On Tue, 11 Mar 2003, Ben Collins wrote:

> > # cd drivers/ieee*
> > # ls *.o
> > csr.o	     hosts.o	 ieee1394_core.o	  nodemgr.o   pcilynx.o  sbp2.o
> > highlevel.o  ieee1394.o  ieee1394_transactions.o  ohci1394.o  raw1394.o
> > # ls -la pcilynx.o
> > -rw-r--r--   1 root     root        16432 Mar 11 15:30 pcilynx.o
> > # depmod pcilynx.o
> > depmod: *** Unresolved symbols in pcilynx.o
> > pcilynx.o:
> 
> > pcilynx.o: unresolved symbol i2c_transfer
> > pcilynx.o: unresolved symbol i2c_bit_del_bus
> > pcilynx.o: unresolved symbol i2c_bit_add_bus
> 
> Everything built fine this time. Redo "make modules" with the SUBDIRS
> option to get the rest built.
> 
> I strongly believe this is something local to your system. Maybe your
> system clock is wrong and it is confusing make. Maybe you don't have all
> the right tools installed. I'm not realy sure, but I do know it is not a
> problem in ieee1394 itself.
> 

Well it seems that I2C has to be configured into the kernel, not
a module, for it to get built. Something seems screwed up with
the configuration. I started, again, with a new kernel tree, configured
from scratch, answering all the questions in `make config` <whew>,
instead of `make oldconfig` with the previous ".config" file. At
least it looks like it might build the stuff. I am currently building
the base kernel (bzImage) again. The kernel builds fine. That's what
I am running. I just can't get all my stuff working because the ieee1394
hardware driver won't build.

The system clock is within a few seconds of NIST and is not set
during the day. It gets set at 2:00 AM every morning by crond. The
hardware clock is updated afterwards.

Cheers,
Dick Johnson
Penguin : Linux version 2.4.20 on an i686 machine (797.90 BogoMips).
Why is the government concerned about the lunatic fringe? Think about it.