[tulip-bug] PPC PCI Tulip resource conflict with X?
Tue May 7 11:19:00 2002
I'm new to the tulip-bug list, so hello to all the helpful correspondents.
I'm having a few problems with some Tulip-based cards on 3 PowerPC 5500s.
Currently, I am at my home computer and will have to work from memory --
screen dumps to come tomorrow.
The problem occurs with both a Debian (Woody) and Mandrake 8.2 install --
so we can safely assume I'm using somewhat modern drivers (versions to
follow) & kernels.
The cards work fine -- I can do a ping flood and get 100% packet return
success for thousands of packets -- until I start X windows. (Starting X
windows uses the network card, as the PowerMacs are X-terms. The X term
network is fine, there are iMacs using it atm).
At this point, X will get a little ways in to starting up, and then the
network card will hang. Pings in to the computer don't complete, pings out
don't complete. An ifdown eth0, rmmod tulip, insmod tulip, ifup eth0 is
required to get the network card going again. (This happens with the
dec4x5 driver as well as the tulip driver.)
dmesg shows a lot of transmit timed out, resetting... errors.
If I can't find anything else on the net, I intend to install Debian
(Potato) and see if an older kernel/driver will fix the problem -- this is
no solution though, as I do need more modern features.
Things seem to point to a resource conflict, because /proc/pci shows the
ATI Rage Pro video card at I/O address 0x400 [disabled] as well as the
network card at 0x400 [enabled].
I have little experience with debugging the hardware side of things, but I
am eager to learn. Any suggestions anyone can make based on this little
data? If not, no worries, there'll be screenfuls of logs posted tomorrow.
PS: (A prior reference to this issue was made in 1998. There are no follow
ups to the message, sadly.)
Thanks in advance,