> This is seriously broken.  Don't work around port-specific bugs
> in MI drivers this way.

Ok. I already filed a PR about the port-brokeness and actually forgot
an XXX in the commit message.

This will be restructured when the l1<->l2 interface gets into shape.
I intend do defer enabling of interrupts on the cards until the 
enable() function gets called (and disable them later via disable
for pcmcia and similar cards). This will handily solve this problem
as well.

In this case, it doesn't make a difference from the drivers point of
view (i.e. the work around doesn't imply any costs).


Martin