Re: [PATCH v3 05/26] x86: move PAE PMD preallocation defines to header

[Brendan Jackman]

On Sat Aug 1, 2026 at 12:59 AM BST, Yosry Ahmed wrote:
> On Sun, Jul 26, 2026 at 10:22:38PM +0000, Brendan Jackman wrote:
>> In a subsequent patch these defines will need to be referenced from a new
>> file, move them as a separate patch for easy review.
>> 
>> The comments have style violations (personal pronouns etc), do not fix
>> them as this is just code movement.
>> 
>> No functional change intended.
>> 
>> Signed-off-by: Brendan Jackman <jackmanb@xxxxxxxxxx>
>
> AFAICT, this is only needed to expose PREALLOCATED_PMDS for
> mm_local_map_to_user() in patch #7. The only usage is a BUILD_BUG_ON().
>
> I wonder if it would be more preferrable to keep the definitions in
> pgtable.c and add a runtime check instead in mm_local_map_to_user(),
> something like:
>
> diff --git a/arch/x86/include/asm/mmu_context.h b/arch/x86/include/asm/mmu_context.h
> index 09b8d8e6a56ea..15055cb80092c 100644
> --- a/arch/x86/include/asm/mmu_context.h
> +++ b/arch/x86/include/asm/mmu_context.h
> @@ -268,7 +268,6 @@ static inline pmd_t *pgd_to_pmd_walk(pgd_t *pgd, unsigned long va)
>
>  static inline int mm_local_map_to_user(struct mm_struct *mm)
>  {
> -       BUILD_BUG_ON(!PREALLOCATED_PMDS);
>         pgd_t *k_pgd = pgd_offset(mm, MM_LOCAL_BASE_ADDR);
>         pgd_t *u_pgd = kernel_to_user_pgdp(k_pgd);
>         pmd_t *k_pmd, *u_pmd;
> @@ -277,6 +276,9 @@ static inline int mm_local_map_to_user(struct mm_struct *mm)
>         k_pmd = pgd_to_pmd_walk(k_pgd, MM_LOCAL_BASE_ADDR);
>         u_pmd = pgd_to_pmd_walk(u_pgd, MM_LOCAL_BASE_ADDR);
>
> +       if (WARN_ON_ONCE(!k_pmd || !u_pmd))
> +               return -EINVAL;
> +
>         BUILD_BUG_ON(MM_LOCAL_END_ADDR - MM_LOCAL_BASE_ADDR > PMD_SIZE);
>
>         /* Preallocate the PTE table so it can be shared. */

Yeah, I'd say that's fine. Will get back to you if I find another reason
why I thought the move was necessary...