Re: [patch 1/3] Percpu infrastructure to rebase the per cpu area to 0UL

[Rusty Russell <rusty@rustcorp.com.au>]

On Friday 30 November 2007 17:43:06 Christoph Lameter wrote:
> Support an option
>
> 	CONFIG_PERCPU_ZERO_BASED

Minor nitpicks aside, I like this patch (using AT() was a nice touch: I tried 
using section-relative relocs originally and it gave horrible results).

But this belongs in a completely separate patch, since it could well break 
archs which don't have NUMA stuff set up yet:

> Index: linux-2.6.24-rc3-mm2/init/main.c
> ===================================================================
> --- linux-2.6.24-rc3-mm2.orig/init/main.c	2007-11-29 22:06:01.607326159
> -0800 +++ linux-2.6.24-rc3-mm2/init/main.c	2007-11-29 22:06:22.754826440
> -0800 @@ -370,18 +370,19 @@ EXPORT_SYMBOL(__per_cpu_offset);
>
>  static void __init setup_per_cpu_areas(void)
>  {
> -	unsigned long size, i;
> -	char *ptr;
> -	unsigned long nr_possible_cpus = num_possible_cpus();
> +	unsigned long size;
> +	int cpu;
>
>  	/* Copy section for each CPU (we discard the original) */
>  	size = ALIGN(PERCPU_ENOUGH_ROOM, PAGE_SIZE);
> -	ptr = alloc_bootmem_pages(size * nr_possible_cpus);
>
> -	for_each_possible_cpu(i) {
> -		__per_cpu_offset[i] = ptr - __per_cpu_start;
> -		memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start);
> -		ptr += size;
> +	for_each_possible_cpu(cpu) {
> +		char *ptr;
> +
> +		ptr = alloc_bootmem_pages_node(NODE_DATA(cpu_to_node(cpu)),
> +								size);
> +		__per_cpu_offset[cpu] = ptr - __per_cpu_start;
> +		memcpy(ptr, __per_cpu_load, __per_cpu_size);
>  	}
>  }

Thanks!
Rusty.