Re: [PATCH 1/2] sched/topology: Annotate RCU pointers properly

[Valentin Schneider]

On 03/01/24 13:56, Pierre Gondois wrote:
> Cleanup RCU-related spare errors by annotating RCU pointers.
>
> sched_domains_numa_distance:
>   error: incompatible types in comparison expression
>   (different address spaces):
>       int [noderef] __rcu *
>       int *
>
> sched_domains_numa_masks:
>   error: incompatible types in comparison expression
>   (different address spaces):
>       struct cpumask **[noderef] __rcu *
>       struct cpumask ***
>
> Signed-off-by: Pierre Gondois <pierre.gondois@xxxxxxx>

That's from when the NUMA topologies were made dynamic, which should be:
Fixes: 0fb3978b0aac ("sched/numa: Fix NUMA topology for systems with CPU-less nodes")
> ---
>  kernel/sched/topology.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/kernel/sched/topology.c b/kernel/sched/topology.c
> index 10d1391e7416..0342a4f41f09 100644
> --- a/kernel/sched/topology.c
> +++ b/kernel/sched/topology.c
> @@ -1542,8 +1542,8 @@ static int			sched_domains_numa_levels;
>  static int			sched_domains_curr_level;
>
>  int				sched_max_numa_distance;
> -static int			*sched_domains_numa_distance;
> -static struct cpumask		***sched_domains_numa_masks;
> +static int			__rcu *sched_domains_numa_distance;
> +static struct cpumask		** __rcu *sched_domains_numa_masks;

I understand that's what sparse is asking for, but that looks odd to me. We
use it as:

  rcu_assign_pointer(sched_domains_numa_masks, foo);

so why isn't it

  __rcu ***sched_domains_numa_masks;

?

This isn't a pointer to an RCU-protected array of masks, this is an
RCU-protected double array of masks.
>  #endif
>
>  /*
> --
> 2.25.1