Re: [PATCH v1 03/14] mm: add iomem vma selection for memory migration

[Felix Kuehling <felix.kuehling@amd.com>]

Am 2021-08-24 um 11:48 p.m. schrieb Alex Sierra:
> In this case, this is used to migrate pages from device memory, back to
> system memory. This particular device memory type should be accessible
> by the CPU, through IOMEM access. Typically, zone device public type
> memory falls into this category.
>
> Signed-off-by: Alex Sierra <alex.sierra@amd.com>
> ---
>  include/linux/migrate.h | 1 +
>  mm/migrate.c            | 3 ++-
>  2 files changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/include/linux/migrate.h b/include/linux/migrate.h
> index 4bb4e519e3f5..6b16f417384f 100644
> --- a/include/linux/migrate.h
> +++ b/include/linux/migrate.h
> @@ -156,6 +156,7 @@ static inline unsigned long migrate_pfn(unsigned long pfn)
>  enum migrate_vma_direction {
>  	MIGRATE_VMA_SELECT_SYSTEM = 1 << 0,
>  	MIGRATE_VMA_SELECT_DEVICE_PRIVATE = 1 << 1,
> +	MIGRATE_VMA_SELECT_IOMEM = 1 << 2,

How about calling this MIGRATE_VMA_SELECT_DEVICE_PUBLIC?


>  };
>  
>  struct migrate_vma {
> diff --git a/mm/migrate.c b/mm/migrate.c
> index e3a10e2a1bb3..d4ae2da99607 100644
> --- a/mm/migrate.c
> +++ b/mm/migrate.c
> @@ -2406,7 +2406,8 @@ static int migrate_vma_collect_pmd(pmd_t *pmdp,
>  			if (is_write_device_private_entry(entry))
>  				mpfn |= MIGRATE_PFN_WRITE;
>  		} else {
> -			if (!(migrate->flags & MIGRATE_VMA_SELECT_SYSTEM))
> +			if (!(migrate->flags & MIGRATE_VMA_SELECT_SYSTEM) &&
> +			    !(migrate->flags & MIGRATE_VMA_SELECT_IOMEM))
>  				goto next;

For MIGRATE_VMA_SELECT_IOMEM/DEVICE_PUBLIC, I think we should ensure the
pages are ZONE_DEVICE and we should also check the page owner for
symmetry with DEVICE_PRIVATE.

Regards,
  Felix


>  			pfn = pte_pfn(pte);
>  			if (is_zero_pfn(pfn)) {