Re: [PATCH 1/3] High-res-timers part 1 (core) take 20

[george anzinger <george@mvista.com>]

Arjan van de Ven wrote:
> 
> On Sun, Dec 08, 2002 at 11:03:13PM -0800, george anzinger wrote:
> > Arjan van de Ven wrote:
> > >
> > > On Sun, 2002-12-08 at 08:46, george anzinger wrote:
> > >
> > > > +/*
> > > > + * Here is an SMP helping macro...
> > > > + */
> > > > +#ifdef CONFIG_SMP
> > > > +#define IF_SMP(a) a
> > > > +#else
> > > > +#define IF_SMP(a)
> > > > +#endif
> > >
> > > ehmmmmm personally I would consider any need of this ugly and evil
> > >
> > > > +     IF_SMP(if (old_base && (new_base != old_base))
> > > > +            spin_unlock(&old_base->lock);
> > > > +             )
> > >
> > > Like here..... SMP dependent ifdef's of spinlock usage... shudder
> > >
> > Well it does seem like a waste to do spinlock ordering code
> > on a UP system...
> 
> that's why spinlocks are effectively nops on UP.
> What you say is true of just about every spinlock user, and no
> they shouldn't all do some IF_SMP() thing; the spinlock itself should be
> (and is) zero on UP

But with preemption, they really are not nops on UP...
-- 
George Anzinger   george@mvista.com
High-res-timers: 
http://sourceforge.net/projects/high-res-timers/
Preemption patch:
http://www.kernel.org/pub/linux/kernel/people/rml