Re: [RFC 2/3] sched: add type checks to for_each_cpu_mask()

[Andrew Morton <akpm@linux-foundation.org>]

On Sat, 14 Apr 2012 18:14:44 -0400
Sasha Levin <levinsasha928@gmail.com> wrote:

> Add type checks to assert that 'mask' is 'struct cpumask *'. This check
> would have detected the bug fixed in e3831ed ("sched: Fix incorrect usage
> of for_each_cpu_mask() in select_fallback_rq()"):
> 
> kernel/sched/core.c: In function 'select_fallback_rq':
> kernel/sched/core.c:1273:2: warning: comparison of distinct pointer types lacks a cast
> kernel/sched/core.c:1284:3: warning: comparison of distinct pointer types lacks a cast
> 
> ...
>
> @@ -809,6 +810,7 @@ int __next_cpu(int n, const cpumask_t *srcp);
>  #define next_cpu(n, src)	__next_cpu((n), &(src))
>  #define any_online_cpu(mask) cpumask_any_and(&mask, cpu_online_mask)
>  #define for_each_cpu_mask(cpu, mask)			\
> +	typecheck(struct cpumask *, (mask));		\
>  	for ((cpu) = -1;				\
>  		(cpu) = next_cpu((cpu), (mask)),	\
>  		(cpu) < NR_CPUS; )

and int __next_cpu(int n, const cpumask_t *srcp);

I'm mystified.  Why didn't the next_cpu() call generate a warning when
passed the wrong type?