Commit 5bcc9f86ef09a933255ee66bd899d4601785dad5

Authored by Vlastimil Babka
Committed by Linus Torvalds
1 parent 9b857d26d0

mm/page_alloc: prevent MIGRATE_RESERVE pages from being misplaced

For the MIGRATE_RESERVE pages, it is useful when they do not get
misplaced on free_list of other migratetype, otherwise they might get
allocated prematurely and e.g.  fragment the MIGRATE_RESEVE pageblocks.
While this cannot be avoided completely when allocating new
MIGRATE_RESERVE pageblocks in min_free_kbytes sysctl handler, we should
prevent the misplacement where possible.

Currently, it is possible for the misplacement to happen when a
MIGRATE_RESERVE page is allocated on pcplist through rmqueue_bulk() as a
fallback for other desired migratetype, and then later freed back
through free_pcppages_bulk() without being actually used.  This happens
because free_pcppages_bulk() uses get_freepage_migratetype() to choose
the free_list, and rmqueue_bulk() calls set_freepage_migratetype() with
the *desired* migratetype and not the page's original MIGRATE_RESERVE
migratetype.

This patch fixes the problem by moving the call to
set_freepage_migratetype() from rmqueue_bulk() down to
__rmqueue_smallest() and __rmqueue_fallback() where the actual page's
migratetype (e.g.  from which free_list the page is taken from) is used.
Note that this migratetype might be different from the pageblock's
migratetype due to freepage stealing decisions.  This is OK, as page
stealing never uses MIGRATE_RESERVE as a fallback, and also takes care
to leave all MIGRATE_CMA pages on the correct freelist.

Therefore, as an additional benefit, the call to
get_pageblock_migratetype() from rmqueue_bulk() when CMA is enabled, can
be removed completely.  This relies on the fact that MIGRATE_CMA
pageblocks are created only during system init, and the above.  The
related is_migrate_isolate() check is also unnecessary, as memory
isolation has other ways to move pages between freelists, and drain pcp
lists containing pages that should be isolated.  The buffered_rmqueue()
can also benefit from calling get_freepage_migratetype() instead of
get_pageblock_migratetype().

Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Reported-by: Yong-Taek Lee <ytk.lee@samsung.com>
Reported-by: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>
Suggested-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Acked-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Suggested-by: Mel Gorman <mgorman@suse.de>
Acked-by: Minchan Kim <minchan@kernel.org>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Marek Szyprowski <m.szyprowski@samsung.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Michal Nazarewicz <mina86@mina86.com>
Cc: "Wang, Yalin" <Yalin.Wang@sonymobile.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Showing 1 changed file with 13 additions and 10 deletions Side-by-side Diff

... ... @@ -931,6 +931,7 @@
931 931 rmv_page_order(page);
932 932 area->nr_free--;
933 933 expand(zone, page, order, current_order, area, migratetype);
  934 + set_freepage_migratetype(page, migratetype);
934 935 return page;
935 936 }
936 937  
... ... @@ -1057,7 +1058,9 @@
1057 1058  
1058 1059 /*
1059 1060 * When borrowing from MIGRATE_CMA, we need to release the excess
1060   - * buddy pages to CMA itself.
  1061 + * buddy pages to CMA itself. We also ensure the freepage_migratetype
  1062 + * is set to CMA so it is returned to the correct freelist in case
  1063 + * the page ends up being not actually allocated from the pcp lists.
1061 1064 */
1062 1065 if (is_migrate_cma(fallback_type))
1063 1066 return fallback_type;
... ... @@ -1125,6 +1128,12 @@
1125 1128  
1126 1129 expand(zone, page, order, current_order, area,
1127 1130 new_type);
  1131 + /* The freepage_migratetype may differ from pageblock's
  1132 + * migratetype depending on the decisions in
  1133 + * try_to_steal_freepages. This is OK as long as it does
  1134 + * not differ for MIGRATE_CMA type.
  1135 + */
  1136 + set_freepage_migratetype(page, new_type);
1128 1137  
1129 1138 trace_mm_page_alloc_extfrag(page, order, current_order,
1130 1139 start_migratetype, migratetype, new_type);
... ... @@ -1175,7 +1184,7 @@
1175 1184 unsigned long count, struct list_head *list,
1176 1185 int migratetype, int cold)
1177 1186 {
1178   - int mt = migratetype, i;
  1187 + int i;
1179 1188  
1180 1189 spin_lock(&zone->lock);
1181 1190 for (i = 0; i < count; ++i) {
1182 1191  
... ... @@ -1196,14 +1205,8 @@
1196 1205 list_add(&page->lru, list);
1197 1206 else
1198 1207 list_add_tail(&page->lru, list);
1199   - if (IS_ENABLED(CONFIG_CMA)) {
1200   - mt = get_pageblock_migratetype(page);
1201   - if (!is_migrate_cma(mt) && !is_migrate_isolate(mt))
1202   - mt = migratetype;
1203   - }
1204   - set_freepage_migratetype(page, mt);
1205 1208 list = &page->lru;
1206   - if (is_migrate_cma(mt))
  1209 + if (is_migrate_cma(get_freepage_migratetype(page)))
1207 1210 __mod_zone_page_state(zone, NR_FREE_CMA_PAGES,
1208 1211 -(1 << order));
1209 1212 }
... ... @@ -1572,7 +1575,7 @@
1572 1575 if (!page)
1573 1576 goto failed;
1574 1577 __mod_zone_freepage_state(zone, -(1 << order),
1575   - get_pageblock_migratetype(page));
  1578 + get_freepage_migratetype(page));
1576 1579 }
1577 1580  
1578 1581 __mod_zone_page_state(zone, NR_ALLOC_BATCH, -(1 << order));