[tulip-bug] admtek al981/an985 bug workaround

[Dan Hollis]

I talked with the freebsd tulip driver author.
They have a workaround for an ADMtek comet tx bug that appears very
similar to the ADMtek centaur bug. From 'man 4 dc':

     The ADMtek AL981 chip (and possibly the AN985 as well) has been observed
     to sometimes wedge on transmit: this appears to happen when the driver
     queues a sequence of frames which cause it to wrap from the end of the
     the transmit descriptor ring back to the beginning.  The dc driver at-
     tempts to avoid this condition by not queing any frames past the end of
     the transmit ring during a single invocation of the dc_start() routine.
     This workaround has a negligible impact on transmit performance.

There are various other bugs listed some of which the current scyld driver
does not appear to handle.

-Dan