Re: [RFC] mmiotrace full patch, preview 1

[Andy Whitcroft <apw@shadowen.org>]

On Tue, Feb 26, 2008 at 11:49:48AM +0100, Ingo Molnar wrote:
> 
> * Andy Whitcroft <apw@shadowen.org> wrote:
> 
> > Ok, so that would be the following, work for everyone?
> > 
> > WARNING: mutexes are preferred for single holder semaphores
> > #1: FILE: Z95.c:1:
> > +       DECLARE_MUTEX(&foo);
> > 
> > WARNING: mutexes are preferred for single holder semaphores
> > #3: FILE: Z95.c:3:
> > +       init_MUTEX(&foo);
> 
> yeah.
> 
>   Acked-by: Ingo Molnar <mingo@elte.hu>
> 
> also i guess init_MUTEX_LOCKED() should emit a "this should be a 
> completion" warning.

Thats easy enough.  Though your tone here implies its less definatly
wrong than the other use forms.  Do we want gentle language here?

	"consider using a completion"

> i guess non-DEFINE_SPINLOCK old-style spinlock definition:
> 
>   spinlock_t lock = SPIN_LOCK_UNLOCKED;
> 
> should emit a 'use DEFINE_SPINLOCK' warning as well?

Those (SPIN_LOCK_UNLOCKED & RW_LOCK_UNLOCKED) we already pick up and
indicate are deprecated.

-apw