Re: [git pull] cpus4096 fixes

[Mike Travis <travis@sgi.com>]

Ingo Molnar wrote:
...
> NUMA-locality might have been a valid argument in favor of the massive 
> array of constant cpumasks (common usage is to use it for the current 
> cpu), if it wasnt all stupidly allocated on the boot node:
> 
>         cpumask_of_cpu_map = alloc_bootmem_low(sizeof(cpumask_t) * nr_cpu_ids);
>         for (i = 0; i < nr_cpu_ids; i++)
>                 cpu_set(i, cpumask_of_cpu_map[i]);

The alternate method was for each cpu to contribute a cpumask_of_cpu
for itself in the percpu area.  I don't recall why that idea was rejected.

But yes, the set_cpus_allowed is the biggest receiver of cpumask_t's with
a single bit set.  I had also proposed a different interface allowing just
that, but I think it was rejected as being redundant.

Thanks,
Mike