There is a wrong driver/module in pcitable for DEC 21142/43 network card:
Currently it is tulip but should be de4x5, here is the diff:
< 0x1011 0x0019 "tulip" "DEC|DECchip 21142/43"
> 0x1011 0x0019 "de4x5" "DEC|DECchip 21142/43"
if you check in /usr/src/linux/Documentation/networking/
you can see that de4x5 is a better choice for 21142/43 dec network cards.
Hmm... I'm not sure that that docummentation is particularly relevant; it's
more than a year out of date.
Does the de4x5 driver work for you where the tulip does not?
Yes, the de4x5 works for me fine.
also kudzu configured the device correctly with modified pcitable.
Regarding kernel documentation, if you search for:
grep 2114 de4x5.txt tulip.txt
de4x5.txt: all cards running the dc2114x series chips in particular. Cards
tulip.txt:| EtherPower 10/100 PCI(9332DST) |21140 |Pentium/UDB |
tulip.txt:| EtherWorks 100/10 PCI(DE500-XA) |21140 |Pentium |
tulip.txt:| LA100PCI-T |21140 |Pentium/UDB |
tulip.txt:| EM110 |21140 |Pentium |
I think this is clear enough that tulip is not the right choice.
The tulip.txt file hasn't been updated in a year and a half -
using it as a definitive list of the cards that it supports probably isn't
a good idea.
Does the tulip driver fail in some way that the de4x5 driver works?
Yes the documentation files are old, but also the source code
tulip.c is much older than de4x5.c:
last change in de4x5.c:
0.544 8-May-99 Fix for buggy SROM in Motorola embedded boards using
last change in tulip.c:
static const char version = "tulip.c:v0.89H 5/23/98
Why do you insist that this is not a problem?
de4x5 works tulip does not work for my card.
That's all I can say.
OK, that's the information I wanted (that the tulip did *not* work.)
Of course, it probably will work with the 0.91 driver. Fun.
Fixed in CVS, will make its way into the next kudzu release.