FreeTDS Development Group

[James Cameron <cameron AT stl.dec.com>]

Is there some test I can make to help trace this problem further?

For instance, I'd love to write a little C program that interfaces to
the TDS layer directly.  I could exclude PHP and Apache from the
problem.  Is there a good example in CVS to start with?  Making the
calls seems easy from the TDS unittests; it's processing the returned
data which looks hard.

Brian Bruns wrote:
> [...] the RH 7.2 upgrade crapped out on my dev box, and I'm having
> trouble getting the onboard sis900 network card going. (I had to do some
> convoluted thing last time, and i forget what it was now).  Anyway, I'll
> see what's up.

In the interests of encouraging you ...

- there should be an sis900 module in /lib/modules
  find /lib/modules/`uname -r` -name sis900\*

- at a guess, 'modprobe sis900' might make it work

- editing /etc/conf.modules or /etc/modules.conf (can't remember which)
  and adding an 'alias eth0 sis900' seems to be the thing to do

- searched the kernel source for 2.4.9 and found plenty of references,
  find . -type f -exec grep -i "sis900" {} /dev/null \;|more
  Documentation/Configure.help
  Documentation/networking/sis900.txt
  drivers/net/sis900.c
  drivers/net/sis900.h
  drivers/pci/pci.ids

- these matches scare me ... spelling errors ;-)
  drivers/net/sis900.h:enum sis900_tx_buffer_status {
  drivers/net/sis900.h:enum sis900_rx_bufer_status {

- then I saw this in the documentation and giggled ... defects do
  congregate in code, as it is the nature of coding that Bad Days
generate
  Bad Code.

   "2. A bug is hidden somewhere in the receive buffer management code,
       the bug causes NULL pointer reference in the kernel. This fault
is
       caught before bad things happen and reported with the message:
       eth0: NULL pointer encountered in Rx ring, skipping which can be
       viewed with dmesg or cat /var/log/message."

-- 
James Cameron                                      (cameron AT stl.dec.com)

"Did I ask for fries?"