Re: [PATCH] mm: mempolicy: fix policy_nodemask() for MPOL_PREFERRED_MANY case

[Feng Tang <feng.tang@intel.com>]

On Wed, Aug 03, 2022 at 07:28:59PM +0800, Michal Hocko wrote:
> On Thu 04-08-22 01:14:32, Feng Tang wrote:
> [...]
> > Ok, I change it as below:
> 
> Wouldn't it be better to make this allowed_mems_nr specific to be
> explicit about the intention?
 
Yes, it is.

> Not that I feel strongly about that.
> 
> > ---
> >  mm/hugetlb.c | 28 +++++++++++++++++++++++-----
> >  1 file changed, 23 insertions(+), 5 deletions(-)
> 
> Not even compile tested
>  include/linux/mempolicy.h | 12 ------------
>  mm/hugetlb.c              | 24 ++++++++++++++++++++----
>  2 files changed, 20 insertions(+), 16 deletions(-)
> 
> diff --git a/include/linux/mempolicy.h b/include/linux/mempolicy.h
> index 668389b4b53d..e38b0ef20b8b 100644
> --- a/include/linux/mempolicy.h
> +++ b/include/linux/mempolicy.h
> @@ -151,13 +151,6 @@ extern bool mempolicy_in_oom_domain(struct task_struct *tsk,
>  				const nodemask_t *mask);
>  extern nodemask_t *policy_nodemask(gfp_t gfp, struct mempolicy *policy);
>  
> -static inline nodemask_t *policy_nodemask_current(gfp_t gfp)
> -{
> -	struct mempolicy *mpol = get_task_policy(current);
> -
> -	return policy_nodemask(gfp, mpol);
> -}
> -
>  extern unsigned int mempolicy_slab_node(void);
>  
>  extern enum zone_type policy_zone;
> @@ -294,11 +287,6 @@ static inline void mpol_put_task_policy(struct task_struct *task)
>  {
>  }
>  
> -static inline nodemask_t *policy_nodemask_current(gfp_t gfp)
> -{
> -	return NULL;
> -}
> -
>  static inline bool mpol_is_preferred_many(struct mempolicy *pol)
>  {
>  	return  false;
> diff --git a/mm/hugetlb.c b/mm/hugetlb.c
> index a18c071c294e..6cacbc9b15a1 100644
> --- a/mm/hugetlb.c
> +++ b/mm/hugetlb.c
> @@ -4330,18 +4330,34 @@ static int __init default_hugepagesz_setup(char *s)
>  }
>  __setup("default_hugepagesz=", default_hugepagesz_setup);
>  
> +struct mempolicy *policy_mbind_nodemask(gfp_t gfp)
> +{
> +#ifdef CONFIG_MEMPOLICY
> +	struct mempolicy *mpol = get_task_policy(current);
> +
> +	/*
> +	 * only enforce MBIND which overlaps with cpuset policy (from policy_nodemask)
> +	 * specifically for hugetlb case
> +	 */
> +	if (mpol->mode == MPOL_BIND &&
> +		(apply_policy_zone(mpol, gfp_zone(gfp)) &&
> +		 cpuset_nodemask_valid_mems_allowed(&policy->nodes))
> +		return &mpol->nodes;
> +#endif
> +	return NULL;

I saw the logic is not changed, and it confused me that if there is
no qualified node, it will still return NULL which effectively equals
node_states[N_MEMORY], while I think it should return a all zero
nodemasks.

Thanks,
Feng

> +}
> +
>  static unsigned int allowed_mems_nr(struct hstate *h)
>  {
>  	int node;
>  	unsigned int nr = 0;
> -	nodemask_t *mpol_allowed;
> +	nodemask_t *mbind_nodemask;
>  	unsigned int *array = h->free_huge_pages_node;
>  	gfp_t gfp_mask = htlb_alloc_mask(h);
>  
> -	mpol_allowed = policy_nodemask_current(gfp_mask);
> -
> +	mbind_nodemask = policy_mbind_nodemask(gfp_mask);
>  	for_each_node_mask(node, cpuset_current_mems_allowed) {
> -		if (!mpol_allowed || node_isset(node, *mpol_allowed))
> +		if (!mbind_nodemask || node_isset(node, *mbind_nodemask))
>  			nr += array[node];
>  	}
>  
> -- 
> Michal Hocko
> SUSE Labs