[PATCH] use msleep() for RTAS delays

[Arnd Bergmann]

On Thursday 01 June 2006 07:39, Paul Mackerras wrote:
> 
> > rtas_call_waitbusy(...);
> 
> That's a good idea...

Do we also need an atomic version of that? Maybe there are
places where we actually need to call a slow rtas call
under a spinlock.

At the very least, rtas_call_waitbusy() should have a
might_sleep() in it to find those places.

	Arnd <><