[tulip-bug] Kingston kne111tx and recompiling tulip on Mandrake8.0

Keith Warno keith.warno@valaran.com
Fri, 13 Jul 2001 10:53:36 -0400


Donald Becker wrote:
> 
> On Thu, 12 Jul 2001, Keith Warno wrote:
> > Donald Becker wrote:
> > > The 2.4.6 kernel now defines SIOCGMIIPHY, but not the ioctl(SIOCSPARAMS)
> > > call that sets parameters such as the interrupt mitigation value (on the
> > > tulip) and debug level (on certain drivers).
> > >
> > > The work-around is to add an #endif/#ifdef to the kern_compat.h file
> >
> > OK, I did that, and also made the change to tulip.c ~line 2742 you
> > mentioned yesterday (changing dev_free_skb() called to
> > dev_kfree_skb_irq() to get rid of the bazillion syslog messages).  All
> > seems to be happy, but then again I've only been using it for the past
> > 10 minutes or so.  :)
> 
> Great!
> I'll put the updated version of kern_compat.h in the test directory.
> The drivers will updated during their new test cycle.
> 
> I apologize that we can't do testing against 2.4 ourselves, but all of
> the people here are testing the new cluster distribution.  It is still
> based on 2.2 for stability reasons.  I guess we need to hire a few more
> people...
> 
> Donald Becker                           becker@scyld.com
> Scyld Computing Corporation             http://www.scyld.com
> 410 Severn Ave. Suite 210               Second Generation Beowulf Clusters
> Annapolis MD 21403                      410-990-9993

Hmm, perhaps I spoke too soon.

Maybe an hour after using the driver, the LinkSys LNE100 v4 just died. 
Packets weren't going anywhere and unfortunately there were no messages
in syslog at all.

So I went back to using the kernel-distributed tulip driver (2.4.6),
maintained by Jeff Garzik.  This driver, too, didn't work out.  Syslog
reports:
NETDEV WATCHDOG: eth0: transmit timed out
eth0: PNIC2 transmit timed out, status e4260000, CSR6/7 82422002 /
effffbff CSR12 40a1d0cc, resetting...

I am CC'ing Jeff on this because I can't decypher the hex messages. :) 
Hmm, wait.  I see someone else in this forum has had a similar issue
with the same card?  And Jeff says there "is a generic PCI PM problem in
2.4.6, unrelated to tulip specifically, that has since been fixed"?  I
Suppose the issue would have to be outside of the driver, considering
I've never run into problems like this before.  :/

I will try out the 2.4.7-pre6 test patch and see if it makes a
difference in either or both drivers.

Regards,
kw
--                           
| Keith Warno                       cell: +1 609-209-5800
| http://www.valaran.com/           work: +1 609-716-7200 x243
| Valaran Corporation Penguin Guy   SMS : kw-mobile@valaran.com
+--------------------------------------------------------------//