Messages in this thread | | | Subject | Re: 2.6.9-rc3-mm2 | From | Rusty Russell <> | Date | Fri, 15 Oct 2004 12:05:25 +1000 |
| |
On Fri, 2004-10-15 at 11:44, Andrew Morton wrote: > Rusty Russell <rusty@rustcorp.com.au> wrote: > > Please, please please! Never use per_cpu(XXX, smp_processor_id())! > > Why?
Because that's what get_cpu_var() does, and an arch might not need to get the processor id to do it. As more things get per-cpu-ified, and smp_processor_id() for its own sake becomes more unusual, I expect archs to go to smp_processor_id() as a per-cpu variable, and a register holding the per-cpu offset. (This needs a real dynamic per-cpu allocator, as well, which I've been meaning to polish off).
> We were getting warnings from somewhere or other due to smp_processor_id() > within preemptible code - I don't recall the callsite.
That's weird, but implies bogosity in the caller. Covering it up like this is not necessarily a win.
Rusty. -- Anyone who quotes me in their signature is an idiot -- Rusty Russell
- 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/
| |