I have tried a recent de driver (970703) as a fix to my crc error
problems.  This compiled flawlessly in the current tree and largely
solved the problem in PR#3734.

I still get a couple of bad packets at boot time from the machine I'm
testing the new kernel in (as seen from my server).

Jul  9 21:32:29 tequila /netbsd: de0: receive: 00:80:48:e9:0a:3a: bad crc
Jul  9 21:32:36 tequila /netbsd: de0: receive: 00:80:48:e9:0a:3a: bad crc

They seem to be matched on the transmitting end by the errors:

de0: abnormal interrupt: transmit underflow (raising TX threshold to 96|256)
de0: abnormal interrupt: transmit underflow (raising TX threshold to 8|512)

That an underflow produces a CRC error is consistent with my world
view. :)

I'm guessing this is the result of an efficiency hack to reduce
transmit interrupt numbers or the like?

The transmitting machine is a 486dx2/66, probably close to the lowest
end machine that would be expected to have one of these cards,
especially running at 100Mb/s).

In any case, if the new driver is integrated into the tree, PR#3734
can be closed. (Yay!)

Dave.