Re: [RFC PATCH 03/14] mm/mprotect: allow exclusive anon pages to be writable

[David Hildenbrand]

On 18.07.22 14:02, Nadav Amit wrote:
> From: Nadav Amit <namit@xxxxxxxxxx>
> 
> Anonymous pages might have the dirty bit clear, but this should not
> prevent mprotect from making them writable if they are exclusive.
> Therefore, skip the test whether the page is dirty in this case.
> 
> Cc: Andrea Arcangeli <aarcange@xxxxxxxxxx>
> Cc: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
> Cc: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
> Cc: Andy Lutomirski <luto@xxxxxxxxxx>
> Cc: Dave Hansen <dave.hansen@xxxxxxxxxxxxxxx>
> Cc: David Hildenbrand <david@xxxxxxxxxx>
> Cc: Peter Xu <peterx@xxxxxxxxxx>
> Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
> Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
> Cc: Will Deacon <will@xxxxxxxxxx>
> Cc: Yu Zhao <yuzhao@xxxxxxxxxx>
> Cc: Nick Piggin <npiggin@xxxxxxxxx>
> Signed-off-by: Nadav Amit <namit@xxxxxxxxxx>
> ---
>  mm/mprotect.c | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/mm/mprotect.c b/mm/mprotect.c
> index 34c2dfb68c42..da5b9bf8204f 100644
> --- a/mm/mprotect.c
> +++ b/mm/mprotect.c
> @@ -45,7 +45,7 @@ static inline bool can_change_pte_writable(struct vm_area_struct *vma,
>  
>  	VM_BUG_ON(!(vma->vm_flags & VM_WRITE) || pte_write(pte));
>  
> -	if (pte_protnone(pte) || !pte_dirty(pte))
> +	if (pte_protnone(pte))
>  		return false;
>  
>  	/* Do we need write faults for softdirty tracking? */
> @@ -66,7 +66,8 @@ static inline bool can_change_pte_writable(struct vm_area_struct *vma,
>  		page = vm_normal_page(vma, addr, pte);
>  		if (!page || !PageAnon(page) || !PageAnonExclusive(page))
>  			return false;
> -	}
> +	} else if (!pte_dirty(pte))
> +		return false;
>  
>  	return true;
>  }

When I wrote that code, I was wondering how often that would actually
happen in practice -- and if we care about optimizing that. Do you have
a gut feeling in which scenarios this would happen and if we care?

If the page is in the swapcache and was swapped out, you'd be requiring
a writeback even though nobody modified the page and possibly isn't
going to do so in the near future.

-- 
Thanks,

David / dhildenb