I asked about this problem before Christmas, but then thought all my
problems were solved by version 0.90f - obviously they were not.
We have a load of Compaq servers which we use, all with 2 Kingston
KNE-100TX cards in (and also the on-board TLAN driver which we tend not to
use since its unreliable and eats CPU).
On Compaq 1600R & 850R servers all ethernet cards are seen on every boot
with all the kernel variants we have built (in general we run exactly the
same kernel on all our servers).
We have one kernel which when booted sees all 3 ethernet cards on an 1850
server (eth0,1 are KNE-100TXs, eth2 is the TLAN).
Another kernel, built with one slightly different option only sees 2
ethernet devices (eth0 - KNE-100TX, eth1 is now the TLAN).
In all cases /proc/pci is sensible and shows all the devices (sample at
end).
The only theory we have been able to come up with is:-
- The 1850 has a faster PCI bridge than the other systems and
a generally different PCI bridge arrangement
- The alignment of the tulip code is different between the working
and non-working code (both are still paragraph aligned though)
- maybe just maybe the second probe, which will be running from cache,
is just that bit too fast for the system sometimes....
An earlier version of the kernel, with a tulip driver earlier than 0.90f
saw one card sometimes and both cards the other times... mostly
randomly....
Any clues as to what is happening would be nice....
Any ideas to fix it would be even nicer :-)
Nigel.
Working kernel has the following tulip symbols:-
0019ed10 T tulip_probe
0019efb8 t tulip_probe1
001a02d4 t tulip_open
001a11f8 t tulip_timer
001a21f4 t tulip_tx_timeout
001a2464 t tulip_init_ring
001a2548 t tulip_start_xmit
001a2650 t tulip_interrupt
001a2ad4 t tulip_rx
001a2e1c t tulip_close
001a2f5c t tulip_get_stats
00209478 d tulip_tbl
0020957c d root_tulip_dev
Broken kernel has these symbols:-
0019ee40 T tulip_probe
0019f0e8 t tulip_probe1
001a0404 t tulip_open
001a1328 t tulip_timer
001a2324 t tulip_tx_timeout
001a2594 t tulip_init_ring
001a2678 t tulip_start_xmit
001a2780 t tulip_interrupt
001a2c04 t tulip_rx
001a2f4c t tulip_close
001a308c t tulip_get_stats
002095ac d tulip_tbl
[ Nigel.Metheringham@theplanet.net - Systems Software Engineer ]
[ Tel : +44 113 207 6112 Fax : +44 113 234 6065 ]
[ Real life is but a pale imitation of a Dilbert strip ]
[ We're recruiting http://www.theplanet.net/profile/recruit.htm ]
This archive was generated by hypermail 2b29 : Wed Jan 12 2000 - 12:00:22 EST