From: Hisashi Hifumi While running some memory intensive load, system response deteriorated just after swap-out started. The cause of this problem is that when a PG_reclaim page is moved to the tail of the inactive LRU list in rotate_reclaimable_page(), lru_lock spin lock is acquired every page writeback . This deteriorates system performance and makes interrupt hold off time longer when swap-out started. Following patch solves this problem. I use pagevec in rotating reclaimable pages to mitigate LRU spin lock contention and reduce interrupt hold off time. I did a test that allocating and touching pages in multiple processes, and pinging to the test machine in flooding mode to measure response under memory intensive load. The test result is: -2.6.23-rc5 --- testmachine ping statistics --- 3000 packets transmitted, 3000 received, 0% packet loss, time 53222ms rtt min/avg/max/mdev = 0.074/0.652/172.228/7.176 ms, pipe 11, ipg/ewma 17.746/0.092 ms -2.6.23-rc5-patched --- testmachine ping statistics --- 3000 packets transmitted, 3000 received, 0% packet loss, time 51924ms rtt min/avg/max/mdev = 0.072/0.108/3.884/0.114 ms, pipe 2, ipg/ewma 17.314/0.091 ms Max round-trip-time was improved. The test machine spec is that 4CPU(3.16GHz, Hyper-threading enabled) 8GB memory , 8GB swap. I did ping test again to observe performance deterioration caused by taking a ref. -2.6.23-rc6-with-modifiedpatch --- testmachine ping statistics --- 3000 packets transmitted, 3000 received, 0% packet loss, time 53386ms rtt min/avg/max/mdev = 0.074/0.110/4.716/0.147 ms, pipe 2, ipg/ewma 17.801/0.129 ms The result for my original patch is as follows. -2.6.23-rc5-with-originalpatch --- testmachine ping statistics --- 3000 packets transmitted, 3000 received, 0% packet loss, time 51924ms rtt min/avg/max/mdev = 0.072/0.108/3.884/0.114 ms, pipe 2, ipg/ewma 17.314/0.091 ms The influence to response was small. Signed-off-by: Hisashi Hifumi Signed-off-by: Andrew Morton --- include/linux/swap.h | 1 mm/swap.c | 88 +++++++++++++++++++++++++++++++++-------- mm/vmscan.c | 1 3 files changed, 74 insertions(+), 16 deletions(-) diff -puN include/linux/swap.h~mm-use-pagevec-to-rotate-reclaimable-page include/linux/swap.h --- a/include/linux/swap.h~mm-use-pagevec-to-rotate-reclaimable-page +++ a/include/linux/swap.h @@ -185,6 +185,7 @@ extern void FASTCALL(mark_page_accessed( extern void lru_add_drain(void); extern int lru_add_drain_all(void); extern int rotate_reclaimable_page(struct page *page); +extern void move_tail_pages(void); extern void swap_setup(void); /* linux/mm/vmscan.c */ diff -puN mm/swap.c~mm-use-pagevec-to-rotate-reclaimable-page mm/swap.c --- a/mm/swap.c~mm-use-pagevec-to-rotate-reclaimable-page +++ a/mm/swap.c @@ -92,24 +92,62 @@ void put_pages_list(struct list_head *pa EXPORT_SYMBOL(put_pages_list); /* + * pagevec_move_tail() must be called with IRQ disabled. + * Otherwise this may cause nasty races. + */ +static void pagevec_move_tail(struct pagevec *pvec) +{ + int i; + int pgmoved = 0; + struct zone *zone = NULL; + unsigned long flags = 0; + + for (i = 0; i < pagevec_count(pvec); i++) { + struct page *page = pvec->pages[i]; + struct zone *pagezone = page_zone(page); + + if (pagezone != zone) { + if (zone) + spin_unlock_irqrestore(&zone->lru_lock, flags); + zone = pagezone; + spin_lock_irqsave(&zone->lru_lock, flags); + } + if (PageLRU(page) && !PageActive(page)) { + list_move_tail(&page->lru, &zone->inactive_list); + pgmoved++; + } + } + if (zone) + spin_unlock_irqrestore(&zone->lru_lock, flags); + __count_vm_events(PGROTATED, pgmoved); + release_pages(pvec->pages, pvec->nr, pvec->cold); + pagevec_reinit(pvec); +} + +static DEFINE_PER_CPU(struct pagevec, rotate_pvecs) = { 0, }; + +void move_tail_pages() +{ + unsigned long flags; + struct pagevec *pvec; + + local_irq_save(flags); + pvec = &__get_cpu_var(rotate_pvecs); + if (pagevec_count(pvec)) + pagevec_move_tail(pvec); + local_irq_restore(flags); +} + +/* * Writeback is about to end against a page which has been marked for immediate * reclaim. If it still appears to be reclaimable, move it to the tail of the - * inactive list. The page still has PageWriteback set, which will pin it. - * - * We don't expect many pages to come through here, so don't bother batching - * things up. - * - * To avoid placing the page at the tail of the LRU while PG_writeback is still - * set, this function will clear PG_writeback before performing the page - * motion. Do that inside the lru lock because once PG_writeback is cleared - * we may not touch the page. + * inactive list. * * Returns zero if it cleared PG_writeback. */ int rotate_reclaimable_page(struct page *page) { - struct zone *zone; - unsigned long flags; + struct pagevec *pvec; if (PageLocked(page)) return 1; @@ -120,15 +158,15 @@ int rotate_reclaimable_page(struct page if (!PageLRU(page)) return 1; - zone = page_zone(page); - spin_lock_irqsave(&zone->lru_lock, flags); if (PageLRU(page) && !PageActive(page)) { - list_move_tail(&page->lru, &zone->inactive_list); - __count_vm_event(PGROTATED); + page_cache_get(page); + pvec = &__get_cpu_var(rotate_pvecs); + if (!pagevec_add(pvec, page)) + pagevec_move_tail(pvec); } if (!test_clear_page_writeback(page)) BUG(); - spin_unlock_irqrestore(&zone->lru_lock, flags); + return 0; } @@ -493,6 +531,23 @@ static int cpu_swap_callback(struct noti } return NOTIFY_OK; } + +static int cpu_movetail_callback(struct notifier_block *nfb, + unsigned long action, void *hcpu) +{ + unsigned long flags; + struct pagevec *pvec; + + if (action == CPU_DEAD || action == CPU_DEAD_FROZEN) { + local_irq_save(flags); + pvec = &per_cpu(rotate_pvecs, (long)hcpu); + if (pagevec_count(pvec)) + pagevec_move_tail(pvec); + local_irq_restore(flags); + } + + return NOTIFY_OK; +} #endif /* CONFIG_HOTPLUG_CPU */ #endif /* CONFIG_SMP */ @@ -514,5 +569,6 @@ void __init swap_setup(void) */ #ifdef CONFIG_HOTPLUG_CPU hotcpu_notifier(cpu_swap_callback, 0); + hotcpu_notifier(cpu_movetail_callback, 0); #endif } diff -puN mm/vmscan.c~mm-use-pagevec-to-rotate-reclaimable-page mm/vmscan.c --- a/mm/vmscan.c~mm-use-pagevec-to-rotate-reclaimable-page +++ a/mm/vmscan.c @@ -792,6 +792,7 @@ static unsigned long shrink_inactive_lis pagevec_init(&pvec, 1); + move_tail_pages(); lru_add_drain(); spin_lock_irq(&zone->lru_lock); do { _