Re: [PATCH] mm/secretmem: disable under HIGHMEM

[Mike Rapoport]

On Wed, Jul 15, 2026 at 02:31:08PM +0200, David Hildenbrand (Arm) wrote:
> On 7/15/26 13:33, Mike Rapoport wrote:
> > On Wed, Jul 15, 2026 at 11:22:33AM +0000, Brendan Jackman wrote:
> >> On Mon Jul 6, 2026 at 8:42 AM UTC, David Hildenbrand (Arm) wrote:
> >>>
> >>> Right.
> >>>
> >>>
> >>> We could even limit it to 64BIT ;)
> >>
> >> I fear this in limbo, we have quite a wide range of opinions from "fix
> >> the broken configuration" all the way to "disable secretmem completely on
> >> 32bit". I'm not passionately committed to any one answer but I do think
> >> we need to pick something.
> >>
> >> Mike's position seems to be roughly "in the middle" of the spectrum, and
> > 
> > Right :)
> > Supporting 32-bit without HIGHMEM is easy, so I don't think we should limit
> > it to 64BIT at this point.
> 
> You mean
> 
> diff --git a/mm/secretmem.c b/mm/secretmem.c
> index 4877c262cb1f6..d29865075b6ea 100644
> --- a/mm/secretmem.c
> +++ b/mm/secretmem.c
> @@ -202,7 +202,7 @@ static struct file *secretmem_file_create(unsigned long flags)
>         if (IS_ERR(file))
>                 goto err_free_inode;
> 
> -       mapping_set_gfp_mask(inode->i_mapping, GFP_HIGHUSER);
> +       mapping_set_gfp_mask(inode->i_mapping, GFP_USER);
>         mapping_set_unevictable(inode->i_mapping);
> 
>         inode->i_op = &secretmem_iops;
> 
> 
> then?

No, I mean what Brendan's patch did:

 config SECRETMEM
        default y
        bool "Enable memfd_secret() system call" if EXPERT
-       depends on ARCH_HAS_SET_DIRECT_MAP
+       depends on ARCH_HAS_SET_DIRECT_MAP && !HIGHMEM

We can change gfp_mask as well for the sake of pedancy
 
> -- 
> Cheers,
> 
> David

-- 
Sincerely yours,
Mike.