From: Johannes Weiner <hannes@xxxxxxxxxxx> Subject: mm: keep separate anon and file statistics on page reclaim activity Having statistics on pages scanned and pages reclaimed for both anon and file pages makes it easier to evaluate changes to LRU balancing. While at it, clean up the stat-keeping mess for isolation, putback, reclaim stats etc. a bit: first the physical LRU operation (isolation and putback), followed by vmstats, reclaim_stats, and then vm events. Link: http://lkml.kernel.org/r/20200520232525.798933-3-hannes@xxxxxxxxxxx Signed-off-by: Johannes Weiner <hannes@xxxxxxxxxxx> Cc: Joonsoo Kim <iamjoonsoo.kim@xxxxxxx> Cc: Michal Hocko <mhocko@xxxxxxxx> Cc: Minchan Kim <minchan@xxxxxxxxxx> Cc: Rik van Riel <riel@xxxxxxxxxxx> Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> --- include/linux/vm_event_item.h | 4 ++++ mm/vmscan.c | 17 +++++++++-------- mm/vmstat.c | 4 ++++ 3 files changed, 17 insertions(+), 8 deletions(-) --- a/include/linux/vm_event_item.h~mm-keep-separate-anon-and-file-statistics-on-page-reclaim-activity +++ a/include/linux/vm_event_item.h @@ -35,6 +35,10 @@ enum vm_event_item { PGPGIN, PGPGOUT, PS PGSCAN_KSWAPD, PGSCAN_DIRECT, PGSCAN_DIRECT_THROTTLE, + PGSCAN_ANON, + PGSCAN_FILE, + PGSTEAL_ANON, + PGSTEAL_FILE, #ifdef CONFIG_NUMA PGSCAN_ZONE_RECLAIM_FAILED, #endif --- a/mm/vmscan.c~mm-keep-separate-anon-and-file-statistics-on-page-reclaim-activity +++ a/mm/vmscan.c @@ -1913,7 +1913,7 @@ shrink_inactive_list(unsigned long nr_to unsigned int nr_reclaimed = 0; unsigned long nr_taken; struct reclaim_stat stat; - int file = is_file_lru(lru); + bool file = is_file_lru(lru); enum vm_event_item item; struct pglist_data *pgdat = lruvec_pgdat(lruvec); struct zone_reclaim_stat *reclaim_stat = &lruvec->reclaim_stat; @@ -1941,11 +1941,12 @@ shrink_inactive_list(unsigned long nr_to __mod_node_page_state(pgdat, NR_ISOLATED_ANON + file, nr_taken); reclaim_stat->recent_scanned[file] += nr_taken; - item = current_is_kswapd() ? PGSCAN_KSWAPD : PGSCAN_DIRECT; if (!cgroup_reclaim(sc)) __count_vm_events(item, nr_scanned); __count_memcg_events(lruvec_memcg(lruvec), item, nr_scanned); + __count_vm_events(PGSCAN_ANON + file, nr_scanned); + spin_unlock_irq(&pgdat->lru_lock); if (nr_taken == 0) @@ -1956,16 +1957,16 @@ shrink_inactive_list(unsigned long nr_to spin_lock_irq(&pgdat->lru_lock); + move_pages_to_lru(lruvec, &page_list); + + __mod_node_page_state(pgdat, NR_ISOLATED_ANON + file, -nr_taken); + reclaim_stat->recent_rotated[0] += stat.nr_activate[0]; + reclaim_stat->recent_rotated[1] += stat.nr_activate[1]; item = current_is_kswapd() ? PGSTEAL_KSWAPD : PGSTEAL_DIRECT; if (!cgroup_reclaim(sc)) __count_vm_events(item, nr_reclaimed); __count_memcg_events(lruvec_memcg(lruvec), item, nr_reclaimed); - reclaim_stat->recent_rotated[0] += stat.nr_activate[0]; - reclaim_stat->recent_rotated[1] += stat.nr_activate[1]; - - move_pages_to_lru(lruvec, &page_list); - - __mod_node_page_state(pgdat, NR_ISOLATED_ANON + file, -nr_taken); + __count_vm_events(PGSTEAL_ANON + file, nr_reclaimed); spin_unlock_irq(&pgdat->lru_lock); --- a/mm/vmstat.c~mm-keep-separate-anon-and-file-statistics-on-page-reclaim-activity +++ a/mm/vmstat.c @@ -1203,6 +1203,10 @@ const char * const vmstat_text[] = { "pgscan_kswapd", "pgscan_direct", "pgscan_direct_throttle", + "pgscan_anon", + "pgscan_file", + "pgsteal_anon", + "pgsteal_file", #ifdef CONFIG_NUMA "zone_reclaim_failed", _
