> See pr# kern/14559 for a possible patch.

Ok, I tested with a random pair of my Orinoco Silver cards and I have one
where the timeout happens and a second where it doesn't happen.

Note that the suggested patch is a bit drastic and only fixes the problem
for cards where the timeout happens. On cards where it doesn't happen, the
card still blocks for nearly 400 ms, but then just in time becomes ready.

Blocking the kernel for this period is not acceptable, so we need to completely
rework the statistics gathering in the wi driver.

I have an idea how to do this and will look into it.

Martin