Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > linux.kernel > #1630484 > unrolled thread

[PATCH -mm -v10 3/3] mm, THP, swap: Enable THP swap optimization only if has compound map

Started by"Huang, Ying" <ying.huang@intel.com>
First post2017-04-25 15:00 +0200
Last post2017-04-28 15:20 +0200
Articles 3 — 3 participants

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.


Contents

  [PATCH -mm -v10 3/3] mm, THP, swap: Enable THP swap optimization only if has compound map "Huang, Ying" <ying.huang@intel.com> - 2017-04-25 15:00 +0200
    Re: [PATCH -mm -v10 3/3] mm, THP, swap: Enable THP swap optimization  only if has compound map Johannes Weiner <hannes@cmpxchg.org> - 2017-04-25 23:50 +0200
      Re: [PATCH -mm -v10 3/3] mm, THP, swap: Enable THP swap optimization  only if has compound map "Kirill A. Shutemov" <kirill@shutemov.name> - 2017-04-28 15:20 +0200

#1630484 — [PATCH -mm -v10 3/3] mm, THP, swap: Enable THP swap optimization only if has compound map

From"Huang, Ying" <ying.huang@intel.com>
Date2017-04-25 15:00 +0200
Subject[PATCH -mm -v10 3/3] mm, THP, swap: Enable THP swap optimization only if has compound map
Message-ID<tA8eu-4Pu-29@gated-at.bofh.it>
From: Huang Ying <ying.huang@intel.com>

If there is no compound map for a THP (Transparent Huge Page), it is
possible that the map count of some sub-pages of the THP is 0.  So it
is better to split the THP before swapping out. In this way, the
sub-pages not mapped will be freed, and we can avoid the unnecessary
swap out operations for these sub-pages.

Cc: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
---
 mm/swap_state.c | 16 +++++++++++++---
 1 file changed, 13 insertions(+), 3 deletions(-)

diff --git a/mm/swap_state.c b/mm/swap_state.c
index 006d91d8fc53..13f83c6bb1b4 100644
--- a/mm/swap_state.c
+++ b/mm/swap_state.c
@@ -192,9 +192,19 @@ int add_to_swap(struct page *page, struct list_head *list)
 	VM_BUG_ON_PAGE(!PageLocked(page), page);
 	VM_BUG_ON_PAGE(!PageUptodate(page), page);
 
-	/* cannot split, skip it */
-	if (PageTransHuge(page) && !can_split_huge_page(page, NULL))
-		return 0;
+	if (PageTransHuge(page)) {
+		/* cannot split, skip it */
+		if (!can_split_huge_page(page, NULL))
+			return 0;
+		/*
+		 * Split pages without a PMD map right away. Chances
+		 * are some or all of the tail pages can be freed
+		 * without IO.
+		 */
+		if (!compound_mapcount(page) &&
+		    split_huge_page_to_list(page, list))
+			return 0;
+	}
 
 retry:
 	entry = get_swap_page(page);
-- 
2.11.0

[toc] | [next] | [standalone]


#1631026 — Re: [PATCH -mm -v10 3/3] mm, THP, swap: Enable THP swap optimization only if has compound map

FromJohannes Weiner <hannes@cmpxchg.org>
Date2017-04-25 23:50 +0200
SubjectRe: [PATCH -mm -v10 3/3] mm, THP, swap: Enable THP swap optimization only if has compound map
Message-ID<tAgvn-1HU-13@gated-at.bofh.it>
In reply to#1630484
On Tue, Apr 25, 2017 at 08:56:58PM +0800, Huang, Ying wrote:
> From: Huang Ying <ying.huang@intel.com>
> 
> If there is no compound map for a THP (Transparent Huge Page), it is
> possible that the map count of some sub-pages of the THP is 0.  So it
> is better to split the THP before swapping out. In this way, the
> sub-pages not mapped will be freed, and we can avoid the unnecessary
> swap out operations for these sub-pages.
> 
> Cc: Johannes Weiner <hannes@cmpxchg.org>
> Signed-off-by: "Huang, Ying" <ying.huang@intel.com>

Acked-by: Johannes Weiner <hannes@cmpxchg.org>

CC Kirill to double check the reasoning here

> ---
>  mm/swap_state.c | 16 +++++++++++++---
>  1 file changed, 13 insertions(+), 3 deletions(-)
> 
> diff --git a/mm/swap_state.c b/mm/swap_state.c
> index 006d91d8fc53..13f83c6bb1b4 100644
> --- a/mm/swap_state.c
> +++ b/mm/swap_state.c
> @@ -192,9 +192,19 @@ int add_to_swap(struct page *page, struct list_head *list)
>  	VM_BUG_ON_PAGE(!PageLocked(page), page);
>  	VM_BUG_ON_PAGE(!PageUptodate(page), page);
>  
> -	/* cannot split, skip it */
> -	if (PageTransHuge(page) && !can_split_huge_page(page, NULL))
> -		return 0;
> +	if (PageTransHuge(page)) {
> +		/* cannot split, skip it */
> +		if (!can_split_huge_page(page, NULL))
> +			return 0;
> +		/*
> +		 * Split pages without a PMD map right away. Chances
> +		 * are some or all of the tail pages can be freed
> +		 * without IO.
> +		 */
> +		if (!compound_mapcount(page) &&
> +		    split_huge_page_to_list(page, list))
> +			return 0;
> +	}
>  
>  retry:
>  	entry = get_swap_page(page);
> -- 
> 2.11.0

[toc] | [prev] | [next] | [standalone]


#1632868 — Re: [PATCH -mm -v10 3/3] mm, THP, swap: Enable THP swap optimization only if has compound map

From"Kirill A. Shutemov" <kirill@shutemov.name>
Date2017-04-28 15:20 +0200
SubjectRe: [PATCH -mm -v10 3/3] mm, THP, swap: Enable THP swap optimization only if has compound map
Message-ID<tBdYu-7Nk-17@gated-at.bofh.it>
In reply to#1631026
On Tue, Apr 25, 2017 at 05:46:18PM -0400, Johannes Weiner wrote:
> On Tue, Apr 25, 2017 at 08:56:58PM +0800, Huang, Ying wrote:
> > From: Huang Ying <ying.huang@intel.com>
> > 
> > If there is no compound map for a THP (Transparent Huge Page), it is
> > possible that the map count of some sub-pages of the THP is 0.  So it
> > is better to split the THP before swapping out. In this way, the
> > sub-pages not mapped will be freed, and we can avoid the unnecessary
> > swap out operations for these sub-pages.
> > 
> > Cc: Johannes Weiner <hannes@cmpxchg.org>
> > Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
> 
> Acked-by: Johannes Weiner <hannes@cmpxchg.org>
> 
> CC Kirill to double check the reasoning here

Looks good to me:

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

-- 
 Kirill A. Shutemov

[toc] | [prev] | [standalone]


Back to top | Article view | linux.kernel


csiph-web