Re: ordered memory access

Horst von Brand (vonbrand@inf.utfsm.cl)
Thu, 30 Sep 1999 16:52:04 -0400


Andrea Arcangeli <andrea@suse.de> said:
> On 30 Sep 1999, Jes Sorensen wrote:

[...]

> >Intuitively I think it makes sense to do so, and I doubt it will be a
> >performance hit at all, but I am just guessing here.

> On alpha you'll avoid some mb/wmb additional asm instruction ;).
>
> If you don't test atomically the retval it make no sense to me to enforce
> ordering so the alpha implementation looks fine to me.

What if you intend to use something like:

atomic_inc(lock);
/* Do something */

and the /* Do something */ is moved _before_ the atomic_inc(lock)? For this
to be of any use in this way, you'd need the atomic_inc() and its ilk to
act as barriers. So it depends on the use to which they are put (or are
intended to be put).

-- 
Dr. Horst H. von Brand                       mailto:vonbrand@inf.utfsm.cl
Departamento de Informatica                     Fono: +56 32 654431
Universidad Tecnica Federico Santa Maria              +56 32 654239
Casilla 110-V, Valparaiso, Chile                Fax:  +56 32 797513

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/