Hi all,

On Thu, 4 Jun 2020 16:52:46 +1000 Stephen Rothwell <sfr@canb.auug.org.au> wrote:
>
> diff --git a/arch/powerpc/include/asm/book3s/64/pgtable.h b/arch/powerpc/include/asm/book3s/64/pgtable.h
> index 25c3cb8272c0..a6799723cd98 100644
> --- a/arch/powerpc/include/asm/book3s/64/pgtable.h
> +++ b/arch/powerpc/include/asm/book3s/64/pgtable.h
> @@ -1008,6 +1008,12 @@ extern struct page *p4d_page(p4d_t p4d);
>  #define pud_page_vaddr(pud)	__va(pud_val(pud) & ~PUD_MASKED_BITS)
>  #define p4d_page_vaddr(p4d)	__va(p4d_val(p4d) & ~P4D_MASKED_BITS)
>  
> +static inline unsigned long pgd_index(unsigned long address)
> +{
> +	return (address >> PGDIR_SHIFT) & (PTRS_PER_PGD - 1);
> +}
> +#define pgd_index pgd_index
> +
>  #define pte_ERROR(e) \
>  	pr_err("%s:%d: bad pte %08lx.\n", __FILE__, __LINE__, pte_val(e))
>  #define pmd_ERROR(e) \

I have added that hunk to linux-next for tomorrow as a fix for
mm-consolidate-pgd_index-and-pgd_offset_k-definitions.

Its not strickly necessary, but Michael expressed a preference for the
inline function.  I was wondering if pgd_index "Must be a compile-time
constant" on one (or a few) architectures, then why not leave the
default as an inline function and special case it as a macro where
needed ...

-- 
Cheers,
Stephen Rothwell