The patch titled vmscan: shrink_slab() requires the number of lru_pages, not the page order has been added to the -mm tree. Its filename is vmscan-shrink_slab-require-number-of-lru_pages-not-page-order.patch Before you just go and hit "reply", please: a) Consider who else should be cc'ed b) Prefer to cc a suitable mailing list as well c) Ideally: find the original patch on the mailing list and do a reply-to-all to that, adding suitable additional cc's *** Remember to use Documentation/SubmitChecklist when testing your code *** See http://userweb.kernel.org/~akpm/stuff/added-to-mm.txt to find out what to do about this The current -mm tree may be found at http://userweb.kernel.org/~akpm/mmotm/ ------------------------------------------------------ Subject: vmscan: shrink_slab() requires the number of lru_pages, not the page order From: KOSAKI Motohiro <kosaki.motohiro@xxxxxxxxxxxxxx> Presently shrink_slab() has the following scanning equation. lru_scanned max_pass basic_scan_objects = 4 x ------------- x ----------------------------- lru_pages shrinker->seeks (default:2) scan_objects = min(basic_scan_objects, max_pass * 2) If we pass very small value as lru_pages instead real number of lru pages, shrink_slab() drop much objects rather than necessary. And now, __zone_reclaim() pass 'order' as lru_pages by mistake. That produces a bad result. For example, if we receive very low memory pressure (scan = 32, order = 0), shrink_slab() via zone_reclaim() always drop _all_ icache/dcache objects. (see above equation, very small lru_pages make very big scan_objects result). This patch fixes it. Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@xxxxxxxxxxxxxx> Reviewed-by: Minchan Kim <minchan.kim@xxxxxxxxx> Acked-by: Christoph Lameter <cl@xxxxxxxxxxxxxxxxxxxx> Acked-by: Rik van Riel <riel@xxxxxxxxxx> Cc: Johannes Weiner <hannes@xxxxxxxxxxx> Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> --- mm/vmscan.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff -puN mm/vmscan.c~vmscan-shrink_slab-require-number-of-lru_pages-not-page-order mm/vmscan.c --- a/mm/vmscan.c~vmscan-shrink_slab-require-number-of-lru_pages-not-page-order +++ a/mm/vmscan.c @@ -2623,6 +2623,8 @@ static int __zone_reclaim(struct zone *z nr_slab_pages0 = zone_page_state(zone, NR_SLAB_RECLAIMABLE); if (nr_slab_pages0 > zone->min_slab_pages) { + unsigned long lru_pages = zone_reclaimable_pages(zone); + /* * shrink_slab() does not currently allow us to determine how * many pages were freed in this zone. So we take the current @@ -2633,7 +2635,7 @@ static int __zone_reclaim(struct zone *z * Note that shrink_slab will free memory on all zones and may * take a long time. */ - while (shrink_slab(sc.nr_scanned, gfp_mask, order) && + while (shrink_slab(sc.nr_scanned, gfp_mask, lru_pages) && (zone_page_state(zone, NR_SLAB_RECLAIMABLE) + nr_pages > nr_slab_pages0)) ; _ Patches currently in -mm which might be from kosaki.motohiro@xxxxxxxxxxxxxx are acpi-fix-unused-function-warning.patch serial-mmio32-support-for-8250_earlyc-fix-fix.patch security-add-const-to-security_task_setscheduler.patch sched-make-sched_param-argument-static-variables-in-some-sched_setscheduler-caller.patch mm-use-memdup_user.patch oom-check-pf_kthread-instead-of-mm-to-skip-kthreads.patch oom-pf_exiting-check-should-take-mm-into-account.patch oom-introduce-find_lock_task_mm-to-fix-mm-false-positives.patch oom-dump_tasks-use-find_lock_task_mm-too.patch oom-improve-commentary-in-dump_tasks.patch oom-dump_tasks-use-find_lock_task_mm-too-dump_tasks-use-find_lock_task_mm-too-fix.patch oom-give-current-access-to-memory-reserves-if-it-has-been-killed.patch oom-avoid-sending-exiting-tasks-a-sigkill.patch oom-filter-tasks-not-sharing-the-same-cpuset.patch oom-sacrifice-child-with-highest-badness-score-for-parent.patch oom-sacrifice-child-with-highest-badness-score-for-parent-protect-dereferencing-of-tasks-comm.patch oom-sacrifice-child-with-highest-badness-score-for-parent-fix.patch oom-select-task-from-tasklist-for-mempolicy-ooms.patch oom-select-task-from-tasklist-for-mempolicy-ooms-add-has_intersects_mems_allowed-uma-variant.patch oom-select-task-from-tasklist-for-mempolicy-ooms-introduce-find_lock_task_mm-to-fix-mm-false-positives-fix.patch oom-enable-oom-tasklist-dump-by-default.patch oom-avoid-oom-killer-for-lowmem-allocations.patch oom-extract-panic-helper-function.patch oom-remove-special-handling-for-pagefault-ooms.patch oom-move-sysctl-declarations-to-oomh.patch mm-rename-try_set_zone_oom-to-try_set_zonelist_oom.patch oom-remove-constraint-argument-from-select_bad_process-and-__out_of_memory.patch oom-fold-__out_of_memory-into-out_of_memory.patch mm-use-for_each_online_cpu-in-vmstat.patch mempolicy-reduce-stack-size-of-migrate_pages.patch mempolicy-reduce-stack-size-of-migrate_pages-fix.patch vmscan-zone_reclaim-dont-call-disable_swap_token.patch vmscan-recalculate-lru_pages-on-each-priority.patch vmscan-tracing-add-trace-events-for-kswapd-wakeup-sleeping-and-direct-reclaim.patch vmscan-tracing-add-trace-events-for-lru-page-isolation.patch vmscan-tracing-add-trace-event-when-a-page-is-written.patch vmscan-tracing-add-a-postprocessing-script-for-reclaim-related-ftrace-events.patch vmscan-kill-prev_priority-completely.patch vmscan-simplify-shrink_inactive_list.patch vmscan-remove-unnecessary-temporary-vars-in-do_try_to_free_pages.patch vmscan-set-up-pagevec-as-late-as-possible-in-shrink_inactive_list.patch vmscan-set-up-pagevec-as-late-as-possible-in-shrink_page_list.patch vmscan-update-isolated-page-counters-outside-of-main-path-in-shrink_inactive_list.patch oom-dont-try-to-kill-oom_unkillable-child.patch oom-dont-try-to-kill-oom_unkillable-child-checkpatch-fixes.patch oom-oom_kill_process-doesnt-select-kthread-child.patch oom-make-oom_unkillable_task-helper-function.patch oom-oom_kill_process-needs-to-check-that-p-is-unkillable.patch oom-proc-pid-oom_score-treat-kernel-thread-honestly.patch oom-kill-duplicate-oom_disable-check.patch oom-move-oom_disable-check-from-oom_kill_task-to-out_of_memory.patch oom-cleanup-has_intersects_mems_allowed.patch oom-remove-child-mm-check-from-oom_kill_process.patch oom-give-the-dying-task-a-higher-priority.patch oom-multi-threaded-process-coredump-dont-make-deadlock.patch rmap-add-exclusive-page-to-private-anon_vma-on-swapin.patch mm-set-vm_fault_write-in-do_swap_page.patch drivers-base-nodec-reduce-stack-usage-of-node_read_meminfo.patch vmscan-avoid-subtraction-of-unsigned-types.patch vmscan-protect-reading-of-reclaim_stat-with-lru_lock.patch vmscan-shrink_slab-require-number-of-lru_pages-not-page-order.patch reiser4.patch -- To unsubscribe from this list: send the line "unsubscribe mm-commits" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html