Re: [RFC PATCH v2 03/20] mm: Provide pmd to pte_leaf_size()

[Oscar Salvador]

On Fri, May 17, 2024 at 08:59:57PM +0200, Christophe Leroy wrote:
> On powerpc 8xx, when a page is 8M size, the information is in the PMD
> entry. So provide it to pte_leaf_size().
> 
> Signed-off-by: Christophe Leroy <christophe.leroy@xxxxxxxxxx>

Overall looks good to me.

Would be nicer if we could left the arch code untouched.
I wanted to see how this would be if we go down that road and focus only 
on 8xx at the risk of being more esoteric.
pmd_pte_leaf_size() is a name of hell, but could be replaced
with __pte_leaf_size for example.

Worth it? Maybe not, anyway, just wanted to give it a go:


 diff --git a/arch/powerpc/include/asm/nohash/32/pte-8xx.h b/arch/powerpc/include/asm/nohash/32/pte-8xx.h
 index 137dc3c84e45..9e3fe6e1083f 100644
 --- a/arch/powerpc/include/asm/nohash/32/pte-8xx.h
 +++ b/arch/powerpc/include/asm/nohash/32/pte-8xx.h
 @@ -151,7 +151,7 @@ static inline unsigned long pgd_leaf_size(pgd_t pgd)
 
  #define pgd_leaf_size pgd_leaf_size
 
 -static inline unsigned long pte_leaf_size(pte_t pte)
 +static inline unsigned long pmd_pte_leaf_size(pte_t pte)
  {
         pte_basic_t val = pte_val(pte);
 
 @@ -162,7 +162,7 @@ static inline unsigned long pte_leaf_size(pte_t pte)
         return SZ_4K;
  }
 
 -#define pte_leaf_size pte_leaf_size
 +#define pmd_pte_leaf_size pmd_pte_leaf_size
 
  /*
   * On the 8xx, the page tables are a bit special. For 16k pages, we have
 diff --git a/include/linux/pgtable.h b/include/linux/pgtable.h
 index 18019f037bae..2bc2fe3b2b53 100644
 --- a/include/linux/pgtable.h
 +++ b/include/linux/pgtable.h
 @@ -1891,6 +1891,9 @@ typedef unsigned int pgtbl_mod_mask;
  #ifndef pte_leaf_size
  #define pte_leaf_size(x) PAGE_SIZE
  #endif
 +#ifndef pmd_pte_leaf_size
 +#define pmd_pte_leaf_size(x, y) pte_leaf_size(y)
 +#endif
 
  /*
   * We always define pmd_pfn for all archs as it's used in lots of generic
 diff --git a/kernel/events/core.c b/kernel/events/core.c
 index f0128c5ff278..e90a547d2fb2 100644
 --- a/kernel/events/core.c
 +++ b/kernel/events/core.c
 @@ -7596,7 +7596,7 @@ static u64 perf_get_pgtable_size(struct mm_struct *mm, unsigned long addr)
 
         pte = ptep_get_lockless(ptep);
         if (pte_present(pte))
 -               size = pte_leaf_size(pte);
 +               size = pmd_pte_leaf_size(pmd, pte);
         pte_unmap(ptep);
  #endif /* CONFIG_HAVE_GUP_FAST */

 

-- 
Oscar Salvador
SUSE Labs