Messages in this thread |  | | | Subject | Re: [patch] spinlocks: remove 'volatile' | | From | Andi Kleen <> | | Date | 08 Jul 2006 15:57:06 +0200 |
| |
Joe Korty <joe.korty@ccur.com> writes:
> On Fri, Jul 07, 2006 at 11:54:10PM -0400, Albert Cahalan wrote: > > That's all theoretical though. Today, gcc's volatile does > > not follow the C standard on modern hardware. Bummer. > > It'd be low-performance anyway though. > > But gcc would follow the standard if it emitted a 'lock' > insn on every volatile reference. It should at least > have an option to do that.
How do you define reference? While you could do locked mem++ you can't do locked mem *= 2 (or any other non trivial operation that doesn't direct map to an memory operand x86 instruction which allows lock prefixes)
IMHO the "barrier model" used by Linux and implemented by gcc is much more useful for multithreaded programming, and MMIO needs other mechanisms anyways.
I wish the C standard would adopt barriers.
-Andi - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
|  |