[PATCH] mm:rmap: use the pra.mapcount to do the check

[PATCH] mm:rmap: use the pra.mapcount to do the check

[Huang Shijie]

We have the pra.mapcount already, and there is no need to call
the page_mapped() which may do some complicated computing
for compound page.

Signed-off-by: Huang Shijie <sjhuang@iluvatar.ai>
---
 mm/rmap.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/mm/rmap.c b/mm/rmap.c
index 76c8dfd3ae1c..6c5843dddb5a 100644
--- a/mm/rmap.c
+++ b/mm/rmap.c
@@ -850,7 +850,7 @@ int page_referenced(struct page *page,
 	};
 
 	*vm_flags = 0;
-	if (!page_mapped(page))
+	if (!pra.mapcount)
 		return 0;
 
 	if (!page_rmapping(page))
-- 
2.17.1

Re: [PATCH] mm:rmap: use the pra.mapcount to do the check

[Mike Kravetz]

On 4/3/19 10:48 PM, Huang Shijie wrote:
> We have the pra.mapcount already, and there is no need to call
> the page_mapped() which may do some complicated computing
> for compound page.
> 
> Signed-off-by: Huang Shijie <sjhuang@iluvatar.ai>

This looks good to me.  I had to convince myself that there were no
issues if we were operating on a sub-page of a compound-page.  However,
Kirill is the expert here and would know of any subtle issues I may have
overlooked.

-- 
Mike Kravetz

> ---
>  mm/rmap.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/mm/rmap.c b/mm/rmap.c
> index 76c8dfd3ae1c..6c5843dddb5a 100644
> --- a/mm/rmap.c
> +++ b/mm/rmap.c
> @@ -850,7 +850,7 @@ int page_referenced(struct page *page,
>  	};
>  
>  	*vm_flags = 0;
> -	if (!page_mapped(page))
> +	if (!pra.mapcount)
>  		return 0;
>  
>  	if (!page_rmapping(page))
> 

Re: [PATCH] mm:rmap: use the pra.mapcount to do the check

[Andrew Morton]

On Thu, 4 Apr 2019 16:08:33 -0700 Mike Kravetz <mike.kravetz@oracle.com> wrote:

> On 4/3/19 10:48 PM, Huang Shijie wrote:
> > We have the pra.mapcount already, and there is no need to call
> > the page_mapped() which may do some complicated computing
> > for compound page.
> > 
> > Signed-off-by: Huang Shijie <sjhuang@iluvatar.ai>
> 
> This looks good to me.  I had to convince myself that there were no
> issues if we were operating on a sub-page of a compound-page.  However,
> Kirill is the expert here and would know of any subtle issues I may have
> overlooked.
> 
> -- 
> Mike Kravetz
> 
> > ---
> >  mm/rmap.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > diff --git a/mm/rmap.c b/mm/rmap.c
> > index 76c8dfd3ae1c..6c5843dddb5a 100644
> > --- a/mm/rmap.c
> > +++ b/mm/rmap.c
> > @@ -850,7 +850,7 @@ int page_referenced(struct page *page,
> >  	};
> >  
> >  	*vm_flags = 0;
> > -	if (!page_mapped(page))
> > +	if (!pra.mapcount)
> >  		return 0;
> >  
> >  	if (!page_rmapping(page))
> > 

Re: [PATCH] mm:rmap: use the pra.mapcount to do the check

[Kirill A. Shutemov]

On Thu, Apr 04, 2019 at 11:08:33PM +0000, Mike Kravetz wrote:
> On 4/3/19 10:48 PM, Huang Shijie wrote:
> > We have the pra.mapcount already, and there is no need to call
> > the page_mapped() which may do some complicated computing
> > for compound page.
> > 
> > Signed-off-by: Huang Shijie <sjhuang@iluvatar.ai>
> 
> This looks good to me.  I had to convince myself that there were no
> issues if we were operating on a sub-page of a compound-page.  However,
> Kirill is the expert here and would know of any subtle issues I may have
> overlooked.

Looks good to me.

Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

-- 
 Kirill A. Shutemov