> 
> Would using something similiar to the delay mechanism from arm26/iobus/ioc.c 
> work?   Or are you thinking of some mechanism that uses the 2nd iomd timer 
> (which I don't think we use for anything currently)  Using the arm26 method 
> would actually be more portable to the other arm32 architectures.

The second timer should really be used for an independent stathz.  
Unfortunately, when I enabled that, my machine would hang after a few 
hours of use.  Never had a chance to track down why.

Basing stathz on hz when a second timer is available and unused is a bad 
idea -- the code says so...

R.