> # Actually I bet it really doesn't matter how that code is written,
> # it probably isn't executed often enough to make a significant
> # difference to system speed, or maybe even process switch speed.
> 
> You might want to hedge your bets and give it a try...?

What, and remove all the fun from the debate :-)

Actually I don't have enough different types of systems to
do any realistic measurements.
I'd also find some other tweaks that are more significant.

Also my kernel tree has enough minor (probable) performance
improvements in it already - tracking current is getting to
be a nightmare.

	David

-- 
David Laight: david@l8s.co.uk