I used to have serious tx timeout problems with ath0.  After I added 

	hw.ath0.txintrperiod=1

to sysctl.conf, that stopped.  That's also the case when I run the PM
branch.  *However*...  After a suspend/resume cycle, I started seeing
the problem; in fact, the device eventually became unusable.  Is it
possible that some state isn't being saved?  (I tried reissuing the
sysctl, of course, and setting it to 5 and then 1; none of this helped.)

I only have one data point on this thus far, unlike the battery and AC
power situations, so it may be coincidence.  (This is all on a T42.)


		--Steve Bellovin, http://www.cs.columbia.edu/~smb