Re: [PATCH 1/2] atomic.h atomic64_t standardization

[Mathieu Desnoyers <compudj@krystal.dyndns.org>]

* Paul Mundt (lethal@linux-sh.org) wrote:
> On Thu, Nov 30, 2006 at 10:11:53PM -0500, Mathieu Desnoyers wrote:
> > --- a/include/asm-generic/atomic.h
> > +++ b/include/asm-generic/atomic.h
> [snip]
> > +#if 0
> > +/* Atomic add unless is only effective on atomic_t on powerpc (at least) */
> > +static inline long atomic_long_add_unless(atomic_long_t *l, long a, long u)
> > +{
> > +	atomic_t *v = (atomic_t *)l;
> > +	
> > +	return atomic_add_unless(v, a, u);
> > +}
> > +
> > +static inline long atomic_long_inc_not_zero(atomic_long_t *l)
> > +{
> > +	atomic_t *v = (atomic_t *)l;
> > +	
> > +	return atomic_inc_not_zero(v);
> > +}
> > +#endif //0
> > +
> 
> Why is this in the patch?
> 

Oops, I forgot to remove these comments after I fixed it in the powerpc code.
Code for all other architectures will have to modified too : I just modified
i386, x86_64, mips, arm and powerpc.

Thanks for reporting.

Mathieu


Here is the fix :


--- a/include/asm-generic/atomic.h
+++ b/include/asm-generic/atomic.h
@@ -122,22 +122,19 @@ static inline long atomic_long_dec_retur
 	return atomic64_dec_return(v);
 }
 
-#if 0
-/* Atomic add unless is only effective on atomic_t on powerpc (at least) */
 static inline long atomic_long_add_unless(atomic_long_t *l, long a, long u)
 {
-	atomic_t *v = (atomic_t *)l;
+	atomic64_t *v = (atomic64_t *)l;
 	
-	return atomic_add_unless(v, a, u);
+	return atomic64_add_unless(v, a, u);
 }
 
 static inline long atomic_long_inc_not_zero(atomic_long_t *l)
 {
-	atomic_t *v = (atomic_t *)l;
+	atomic64_t *v = (atomic64_t *)l;
 	
-	return atomic_inc_not_zero(v);
+	return atomic64_inc_not_zero(v);
 }
-#endif //0
 
 #else
 


OpenPGP public key:              http://krystal.dyndns.org:8080/key/compudj.gpg
Key fingerprint:     8CD5 52C3 8E3C 4140 715F  BA06 3F25 A8FE 3BAE 9A68