On Thu, 20 Jan 2005 14:07:43 +0100
Wolfgang Solfrank <ws@tools.de> wrote:

> While in general this is the correct approach, the argument that
> "checking the specific CPU each time before entering power save mode
> would add a lot of overhead" in this specific case is rather bogus. 
> After all, we are about to turn CPU activity off, so we've got all the
> time in the world...

I would say that a rather cavalier approach to this portion of code is
asking for trouble. After all, it was not too long ago that it was
discovered that someone had assumed it was ok to have MSR[PSL_EE] _off_
before lowering the spl level in this very section. In the macppc
port this led to MP kernels not processing IPIs and a slew of interrupts
being missed. I'd rather be very rigorous in our treatment of this
section of code.

tim