hi,

> > i think it's better to assert (tvmin <= tvmax) in TCPT_RANGESET.
> 
> That is *definitely* not a reasonable solution.  It moves the problem to an 
> obscure place where the effect isn't obvious, *and* it doesn't cause the 
> fallback to non-1323 RTT measurement.

i suggested it for debugging (ie. i meant KASSERT),
not as a solution, of course.

> Something like the attached might be okay.

looks good to me.

YAMAMOTO Takashi