Skip to contentSkip to conversation

shrink_page_list: save page_mapped() to local val

RFC PATCH In review Bob Liu 2 participants 2 messages Mar 24, 2010

Subsystems (3)
lore
Share & tools
lore

Bob Liu

author
#0 lore raw

RFC PATCH 1/1 shrink_page_list: save page_mapped() to local val

In funtion shrink_page_list(), page_mapped() is called several times,save it to local val to reduce atomic_read.

---
 mm/vmscan.c |    8 +++++---
 1 files changed, 5 insertions(+), 3 deletions(-)
mm/vmscan.c +5 -3
index 79c8098..08cc3ac 100644
@@ -637,6 +637,7 @@ static unsigned long shrink_page_list(struct list_head *page_list,
 		struct address_space *mapping;
 		struct page *page;
 		int may_enter_fs;
+		int page_mapcount;
 
 		cond_resched();
 
@@ -653,11 +654,12 @@ static unsigned long shrink_page_list(struct list_head *page_list,
 		if (unlikely(!page_evictable(page, NULL)))
 			goto cull_mlocked;
 
-		if (!sc->may_unmap && page_mapped(page))
+		page_mapcount = page_mapped(page);
+		if (!sc->may_unmap && page_mapcount)
 			goto keep_locked;
 
 		/* Double the slab pressure for mapped and swapcache pages */
-		if (page_mapped(page) || PageSwapCache(page))
+		if (page_mapcount || PageSwapCache(page))
 			sc->nr_scanned++;
 
 		may_enter_fs = (sc->gfp_mask & __GFP_FS) ||
@@ -707,7 +709,7 @@ static unsigned long shrink_page_list(struct list_head *page_list,
 		 * The page is mapped into the page tables of one or more
 		 * processes. Try to unmap it here.
 		 */
-		if (page_mapped(page) && mapping) {
+		if (page_mapcount && mapping) {
 			switch (try_to_unmap(page, TTU_UNMAP)) {
 			case SWAP_FAIL:
 				goto activate_locked;
Signature · 7 lines
1.5.6.3

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
maintainer
#1 lore raw

Hi,

Quoted 23 lines · Bob Liu

In funtion shrink_page_list(), page_mapped() is called several times,save it to local val to reduce atomic_read.

---
 mm/vmscan.c |    8 +++++---
 1 files changed, 5 insertions(+), 3 deletions(-)
mm/vmscan.c +1 -0
index 79c8098..08cc3ac 100644
@@ -637,6 +637,7 @@ static unsigned long shrink_page_list(struct list_head *page_list,
 		struct address_space *mapping;
 		struct page *page;
 		int may_enter_fs;
+		int page_mapcount;
 
 		cond_resched();
 
@@ -653,11 +654,12 @@ static unsigned long shrink_page_list(struct list_head *page_list,
 		if (unlikely(!page_evictable(page, NULL)))
 			goto cull_mlocked;

mm/vmscan.c static unsigned long shrink_page_list(struct list_head *page_list, · line 663

Quoted 10 lines
mm/vmscan.c +3 -2
 
-		if (!sc->may_unmap && page_mapped(page))
+		page_mapcount = page_mapped(page);
+		if (!sc->may_unmap && page_mapcount)
 			goto keep_locked;
 
 		/* Double the slab pressure for mapped and swapcache pages */
-		if (page_mapped(page) || PageSwapCache(page))
+		if (page_mapcount || PageSwapCache(page))
 			sc->nr_scanned++;

Note that the mapcount is unstable and might very well drop while this code runs.

The first two instances are close enough together that a change is unlikely, but between them and the below check before try_to_unmap() we might be even waiting for writeback to complete.

try_to_unmap() will figure it out, but it would lock the mapping first and then read the mapcount.

I am unsure whether the change is worth it.

mm/vmscan.c static unsigned long shrink_page_list(struct list_head *page_list, · line 715

Quoted 10 lines
mm/vmscan.c +1 -1
 		may_enter_fs = (sc->gfp_mask & __GFP_FS) ||
@@ -707,7 +709,7 @@ static unsigned long shrink_page_list(struct list_head *page_list,
 		 * The page is mapped into the page tables of one or more
 		 * processes. Try to unmap it here.
 		 */
-		if (page_mapped(page) && mapping) {
+		if (page_mapcount && mapping) {
 			switch (try_to_unmap(page, TTU_UNMAP)) {
 			case SWAP_FAIL:
 				goto activate_locked;

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>