diff mbox series

[v2,1/2] secretmem/gup: don't check if page is secretmem without reference

Message ID 20210420131611.8259-2-rppt@kernel.org (mailing list archive)
State New
Headers show
Series secretmem: optimize page_is_secretmem() | expand

Commit Message

Mike Rapoport April 20, 2021, 1:16 p.m. UTC
From: Mike Rapoport <rppt@linux.ibm.com>

The check in gup_pte_range() whether a page belongs to a secretmem mapping
is performed before grabbing the page reference.

To avoid potential race move the check after try_grab_compound_head().

Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
---
 mm/gup.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

Comments

David Hildenbrand April 20, 2021, 1:19 p.m. UTC | #1
On 20.04.21 15:16, Mike Rapoport wrote:
> From: Mike Rapoport <rppt@linux.ibm.com>
> 
> The check in gup_pte_range() whether a page belongs to a secretmem mapping
> is performed before grabbing the page reference.
> 
> To avoid potential race move the check after try_grab_compound_head().
> 
> Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
> ---
>   mm/gup.c | 6 +++---
>   1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/mm/gup.c b/mm/gup.c
> index c3a17b189064..4b58c016e949 100644
> --- a/mm/gup.c
> +++ b/mm/gup.c
> @@ -2080,13 +2080,13 @@ static int gup_pte_range(pmd_t pmd, unsigned long addr, unsigned long end,
>   		VM_BUG_ON(!pfn_valid(pte_pfn(pte)));
>   		page = pte_page(pte);
>   
> -		if (page_is_secretmem(page))
> -			goto pte_unmap;
> -
>   		head = try_grab_compound_head(page, 1, flags);
>   		if (!head)
>   			goto pte_unmap;
>   
> +		if (page_is_secretmem(page))
> +			goto pte_unmap;
> +

Looking at the hunk below, I wonder if you're missing a put_compound_head().

(also, I'd do if unlikely(page_is_secretmem()) but that's a different 
discussion)

>   		if (unlikely(pte_val(pte) != pte_val(*ptep))) {
>   			put_compound_head(head, 1, flags);
>   			goto pte_unmap;
>
Mike Rapoport April 20, 2021, 1:27 p.m. UTC | #2
On Tue, Apr 20, 2021 at 03:19:56PM +0200, David Hildenbrand wrote:
> On 20.04.21 15:16, Mike Rapoport wrote:
> > From: Mike Rapoport <rppt@linux.ibm.com>
> > 
> > The check in gup_pte_range() whether a page belongs to a secretmem mapping
> > is performed before grabbing the page reference.
> > 
> > To avoid potential race move the check after try_grab_compound_head().
> > 
> > Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
> > ---
> >   mm/gup.c | 6 +++---
> >   1 file changed, 3 insertions(+), 3 deletions(-)
> > 
> > diff --git a/mm/gup.c b/mm/gup.c
> > index c3a17b189064..4b58c016e949 100644
> > --- a/mm/gup.c
> > +++ b/mm/gup.c
> > @@ -2080,13 +2080,13 @@ static int gup_pte_range(pmd_t pmd, unsigned long addr, unsigned long end,
> >   		VM_BUG_ON(!pfn_valid(pte_pfn(pte)));
> >   		page = pte_page(pte);
> > -		if (page_is_secretmem(page))
> > -			goto pte_unmap;
> > -
> >   		head = try_grab_compound_head(page, 1, flags);
> >   		if (!head)
> >   			goto pte_unmap;
> > +		if (page_is_secretmem(page))
> > +			goto pte_unmap;
> > +
> 
> Looking at the hunk below, I wonder if you're missing a put_compound_head().

Hmm, yes.
 
> (also, I'd do if unlikely(page_is_secretmem()) but that's a different
> discussion)

I don't mind, actually. I don't think there would be massive secretmem
usage soon.

> >   		if (unlikely(pte_val(pte) != pte_val(*ptep))) {
> >   			put_compound_head(head, 1, flags);
> >   			goto pte_unmap;
> > 
> 
> 
> -- 
> Thanks,
> 
> David / dhildenb
>
diff mbox series

Patch

diff --git a/mm/gup.c b/mm/gup.c
index c3a17b189064..4b58c016e949 100644
--- a/mm/gup.c
+++ b/mm/gup.c
@@ -2080,13 +2080,13 @@  static int gup_pte_range(pmd_t pmd, unsigned long addr, unsigned long end,
 		VM_BUG_ON(!pfn_valid(pte_pfn(pte)));
 		page = pte_page(pte);
 
-		if (page_is_secretmem(page))
-			goto pte_unmap;
-
 		head = try_grab_compound_head(page, 1, flags);
 		if (!head)
 			goto pte_unmap;
 
+		if (page_is_secretmem(page))
+			goto pte_unmap;
+
 		if (unlikely(pte_val(pte) != pte_val(*ptep))) {
 			put_compound_head(head, 1, flags);
 			goto pte_unmap;