Re: [PATCH] another minor bit of cpumask cleanup

[Andrew Morton <akpm@osdl.org>]

Paul Jackson <pj@sgi.com> wrote:
>
> Ingo Oeser pointed out to me in private email that one of the cpumask
> macros was broken - the macro for for_each_online_cpu() starts its loop
> with _any_ cpu from the provided mask, and only worries about restricting
> itself to _online_ cpus when looping to the next cpu:
> 
> include/linux/cpumask.h:
> > #define for_each_online_cpu(cpu, map)                                   \
> >         for (cpu = first_cpu_const(mk_cpumask_const(map));              \
> >                 cpu < NR_CPUS;                                          \
> >                 cpu = next_online_cpu(cpu,map))
> 
> Looking further, I see this macro is never used, and its subordinate
> inline macro next_online_cpu() used no where else.  What's more, it's
> redundant.  Calling it with a map of "cpu_online_map" (which you have to
> do, given it's broken thus) is just as good as calling the macro right
> above, "for_each_cpu()", with that same "cpu_online_map". Indeed the
> only uses of "for_each_cpu()", in arch/i386/mach-voyager/voyager_smp.c,
> do pass "cpu_online_map" explicitly, in 5 of 6 calls there from.
> 
> So, having found a piece of code that is broken, redundant and unused,
> I hereby off the following patch to remove it.

Generates rejects for my tree.  I already have three patches which alter
cpumask.h.

Please, hang onto it until we get things synced up a bit more.