Re: [PATCH 2/6] mm/gup: don't pin migrated cma pages in movable zone

[John Hubbard]

On 12/1/20 9:23 PM, Pavel Tatashin wrote:
> In order not to fragment CMA the pinned pages are migrated. However,
> they are migrated to ZONE_MOVABLE, which also should not have pinned pages.
>
> Remove __GFP_MOVABLE, so pages can be migrated to zones where pinning
> is allowed.
>
> Signed-off-by: Pavel Tatashin <pasha.tatashin@xxxxxxxxxx>
> ---
>   mm/gup.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/mm/gup.c b/mm/gup.c
> index cdb8b9eeb016..3a76c005a3e2 100644
> --- a/mm/gup.c
> +++ b/mm/gup.c
> @@ -1610,7 +1610,7 @@ static long check_and_migrate_cma_pages(struct mm_struct *mm,
>   	long ret = nr_pages;
>   	struct migration_target_control mtc = {
>   		.nid = NUMA_NO_NODE,
> -		.gfp_mask = GFP_USER | __GFP_MOVABLE | __GFP_NOWARN,
> +		.gfp_mask = GFP_USER | __GFP_NOWARN,
>   	};
>   
>   check_again:

Reviewed-by: John Hubbard <jhubbard@xxxxxxxxxx>

...while I was here, I noticed this, which is outside the scope of your patchset, but
I thought I'd just mention it anyway in case anyone cares:

static inline bool is_migrate_movable(int mt)
{
	return is_migrate_cma(mt) || mt == MIGRATE_MOVABLE;
}

...that really should take an "enum migratetype mt" instead of an int, I think.

thanks,
--
John Hubbard
NVIDIA