mm, compaction: fix NR_ISOLATED_* stats for pfn based migration
authorMing Ling <ming.ling@spreadtrum.com>
Tue, 13 Dec 2016 00:42:26 +0000 (16:42 -0800)
committerLinus Torvalds <torvalds@linux-foundation.org>
Tue, 13 Dec 2016 02:55:07 +0000 (18:55 -0800)
Since commit bda807d44454 ("mm: migrate: support non-lru movable page
migration") isolate_migratepages_block) can isolate !PageLRU pages which
would acct_isolated account as NR_ISOLATED_*.  Accounting these non-lru
pages NR_ISOLATED_{ANON,FILE} doesn't make any sense and it can misguide
heuristics based on those counters such as pgdat_reclaimable_pages resp.
too_many_isolated which would lead to unexpected stalls during the
direct reclaim without any good reason.  Note that
__alloc_contig_migrate_range can isolate a lot of pages at once.

On mobile devices such as 512M ram android Phone, it may use a big zram
swap.  In some cases zram(zsmalloc) uses too many non-lru but
migratedable pages, such as:

      MemTotal: 468148 kB
      Normal free:5620kB
      Free swap:4736kB
      Total swap:409596kB
      ZRAM: 164616kB(zsmalloc non-lru pages)
      active_anon:60700kB
      inactive_anon:60744kB
      active_file:34420kB
      inactive_file:37532kB

Fix this by only accounting lru pages to NR_ISOLATED_* in
isolate_migratepages_block right after they were isolated and we still
know they were on LRU.  Drop acct_isolated because it is called after
the fact and we've lost that information.  Batching per-cpu counter
doesn't make much improvement anyway.  Also make sure that we uncharge
only LRU pages when putting them back on the LRU in
putback_movable_pages resp.  when unmap_and_move migrates the page.

[mhocko@suse.com: replace acct_isolated() with direct counting]
Fixes: bda807d44454 ("mm: migrate: support non-lru movable page migration")
Link: http://lkml.kernel.org/r/20161019080240.9682-1-mhocko@kernel.org
Signed-off-by: Ming Ling <ming.ling@spreadtrum.com>
Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Minchan Kim <minchan@kernel.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Joonsoo Kim <js1304@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/compaction.c
mm/migrate.c

index 0409a4ad6ea1363611d49269ecbb5ef88afe5c87..70e6bec46dc27f90c126c995cb36b8b50eb76ae8 100644 (file)
@@ -634,22 +634,6 @@ isolate_freepages_range(struct compact_control *cc,
        return pfn;
 }
 
-/* Update the number of anon and file isolated pages in the zone */
-static void acct_isolated(struct zone *zone, struct compact_control *cc)
-{
-       struct page *page;
-       unsigned int count[2] = { 0, };
-
-       if (list_empty(&cc->migratepages))
-               return;
-
-       list_for_each_entry(page, &cc->migratepages, lru)
-               count[!!page_is_file_cache(page)]++;
-
-       mod_node_page_state(zone->zone_pgdat, NR_ISOLATED_ANON, count[0]);
-       mod_node_page_state(zone->zone_pgdat, NR_ISOLATED_FILE, count[1]);
-}
-
 /* Similar to reclaim, but different enough that they don't share logic */
 static bool too_many_isolated(struct zone *zone)
 {
@@ -866,6 +850,8 @@ isolate_migratepages_block(struct compact_control *cc, unsigned long low_pfn,
 
                /* Successfully isolated */
                del_page_from_lru_list(page, lruvec, page_lru(page));
+               inc_node_page_state(page,
+                               NR_ISOLATED_ANON + page_is_file_cache(page));
 
 isolate_success:
                list_add(&page->lru, &cc->migratepages);
@@ -902,7 +888,6 @@ isolate_fail:
                                spin_unlock_irqrestore(zone_lru_lock(zone), flags);
                                locked = false;
                        }
-                       acct_isolated(zone, cc);
                        putback_movable_pages(&cc->migratepages);
                        cc->nr_migratepages = 0;
                        cc->last_migrated_pfn = 0;
@@ -988,7 +973,6 @@ isolate_migratepages_range(struct compact_control *cc, unsigned long start_pfn,
                if (cc->nr_migratepages == COMPACT_CLUSTER_MAX)
                        break;
        }
-       acct_isolated(cc->zone, cc);
 
        return pfn;
 }
@@ -1258,10 +1242,8 @@ static isolate_migrate_t isolate_migratepages(struct zone *zone,
                low_pfn = isolate_migratepages_block(cc, low_pfn,
                                                block_end_pfn, isolate_mode);
 
-               if (!low_pfn || cc->contended) {
-                       acct_isolated(zone, cc);
+               if (!low_pfn || cc->contended)
                        return ISOLATE_ABORT;
-               }
 
                /*
                 * Either we isolated something and proceed with migration. Or
@@ -1271,7 +1253,6 @@ static isolate_migrate_t isolate_migratepages(struct zone *zone,
                break;
        }
 
-       acct_isolated(zone, cc);
        /* Record where migration scanner will be restarted. */
        cc->migrate_pfn = low_pfn;
 
index 99250aee1ac166fd8d5bb03849041b2a07fc3f80..66ce6b490b13a9394411addc61ca43f819d5132b 100644 (file)
@@ -168,8 +168,6 @@ void putback_movable_pages(struct list_head *l)
                        continue;
                }
                list_del(&page->lru);
-               dec_node_page_state(page, NR_ISOLATED_ANON +
-                               page_is_file_cache(page));
                /*
                 * We isolated non-lru movable page so here we can use
                 * __PageMovable because LRU page's mapping cannot have
@@ -186,6 +184,8 @@ void putback_movable_pages(struct list_head *l)
                        put_page(page);
                } else {
                        putback_lru_page(page);
+                       dec_node_page_state(page, NR_ISOLATED_ANON +
+                                       page_is_file_cache(page));
                }
        }
 }
@@ -1121,8 +1121,15 @@ out:
                 * restored.
                 */
                list_del(&page->lru);
-               dec_node_page_state(page, NR_ISOLATED_ANON +
-                               page_is_file_cache(page));
+
+               /*
+                * Compaction can migrate also non-LRU pages which are
+                * not accounted to NR_ISOLATED_*. They can be recognized
+                * as __PageMovable
+                */
+               if (likely(!__PageMovable(page)))
+                       dec_node_page_state(page, NR_ISOLATED_ANON +
+                                       page_is_file_cache(page));
        }
 
        /*