From: Hisashi Hifumi Date: Tue, 16 Oct 2007 08:24:52 +0000 (-0700) Subject: mm: use pagevec to rotate reclaimable page X-Git-Tag: v2.6.24-rc1~1211 X-Git-Url: https://err.no/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=902aaed0d983dfd459fcb2b678608d4584782200;p=linux-2.6 mm: use pagevec to rotate reclaimable page 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. [akpm@linux-foundation.org: fix uninitalised var warning] [hugh@veritas.com: fix locking] [randy.dunlap@oracle.com: fix function declaration] [hugh@veritas.com: fix BUG at include/linux/mm.h:220!] [hugh@veritas.com: kill redundancy in rotate_reclaimable_page] [hugh@veritas.com: move_tail_pages into lru_add_drain] Signed-off-by: Hisashi Hifumi Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- diff --git a/mm/swap.c b/mm/swap.c index d93fa1e9ec..d034b2128d 100644 --- a/mm/swap.c +++ b/mm/swap.c @@ -32,6 +32,10 @@ /* How many pages do we try to swap or page in/out together? */ int page_cluster; +static DEFINE_PER_CPU(struct pagevec, lru_add_pvecs) = { 0, }; +static DEFINE_PER_CPU(struct pagevec, lru_add_active_pvecs) = { 0, }; +static DEFINE_PER_CPU(struct pagevec, lru_rotate_pvecs) = { 0, }; + /* * This path almost never happens for VM activity - pages are normally * freed via pagevecs. But it gets used by networking. @@ -91,24 +95,48 @@ void put_pages_list(struct list_head *pages) } 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; + + 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(&zone->lru_lock); + zone = pagezone; + spin_lock(&zone->lru_lock); + } + if (PageLRU(page) && !PageActive(page)) { + list_move_tail(&page->lru, &zone->inactive_list); + pgmoved++; + } + } + if (zone) + spin_unlock(&zone->lru_lock); + __count_vm_events(PGROTATED, pgmoved); + release_pages(pvec->pages, pvec->nr, pvec->cold); + pagevec_reinit(pvec); +} + /* * 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; + struct pagevec *pvec; unsigned long flags; if (PageLocked(page)) @@ -120,15 +148,16 @@ int rotate_reclaimable_page(struct page *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); + local_irq_save(flags); + pvec = &__get_cpu_var(lru_rotate_pvecs); + if (!pagevec_add(pvec, page)) + pagevec_move_tail(pvec); + local_irq_restore(flags); + if (!test_clear_page_writeback(page)) BUG(); - spin_unlock_irqrestore(&zone->lru_lock, flags); + return 0; } @@ -172,9 +201,6 @@ EXPORT_SYMBOL(mark_page_accessed); * lru_cache_add: add a page to the page lists * @page: the page to add */ -static DEFINE_PER_CPU(struct pagevec, lru_add_pvecs) = { 0, }; -static DEFINE_PER_CPU(struct pagevec, lru_add_active_pvecs) = { 0, }; - void fastcall lru_cache_add(struct page *page) { struct pagevec *pvec = &get_cpu_var(lru_add_pvecs); @@ -195,21 +221,37 @@ void fastcall lru_cache_add_active(struct page *page) put_cpu_var(lru_add_active_pvecs); } -static void __lru_add_drain(int cpu) +/* + * Drain pages out of the cpu's pagevecs. + * Either "cpu" is the current CPU, and preemption has already been + * disabled; or "cpu" is being hot-unplugged, and is already dead. + */ +static void drain_cpu_pagevecs(int cpu) { - struct pagevec *pvec = &per_cpu(lru_add_pvecs, cpu); + struct pagevec *pvec; - /* CPU is dead, so no locking needed. */ + pvec = &per_cpu(lru_add_pvecs, cpu); if (pagevec_count(pvec)) __pagevec_lru_add(pvec); + pvec = &per_cpu(lru_add_active_pvecs, cpu); if (pagevec_count(pvec)) __pagevec_lru_add_active(pvec); + + pvec = &per_cpu(lru_rotate_pvecs, cpu); + if (pagevec_count(pvec)) { + unsigned long flags; + + /* No harm done if a racing interrupt already did this */ + local_irq_save(flags); + pagevec_move_tail(pvec); + local_irq_restore(flags); + } } void lru_add_drain(void) { - __lru_add_drain(get_cpu()); + drain_cpu_pagevecs(get_cpu()); put_cpu(); } @@ -256,6 +298,7 @@ void release_pages(struct page **pages, int nr, int cold) int i; struct pagevec pages_to_free; struct zone *zone = NULL; + unsigned long uninitialized_var(flags); pagevec_init(&pages_to_free, cold); for (i = 0; i < nr; i++) { @@ -263,7 +306,7 @@ void release_pages(struct page **pages, int nr, int cold) if (unlikely(PageCompound(page))) { if (zone) { - spin_unlock_irq(&zone->lru_lock); + spin_unlock_irqrestore(&zone->lru_lock, flags); zone = NULL; } put_compound_page(page); @@ -277,9 +320,10 @@ void release_pages(struct page **pages, int nr, int cold) struct zone *pagezone = page_zone(page); if (pagezone != zone) { if (zone) - spin_unlock_irq(&zone->lru_lock); + spin_unlock_irqrestore(&zone->lru_lock, + flags); zone = pagezone; - spin_lock_irq(&zone->lru_lock); + spin_lock_irqsave(&zone->lru_lock, flags); } VM_BUG_ON(!PageLRU(page)); __ClearPageLRU(page); @@ -288,7 +332,7 @@ void release_pages(struct page **pages, int nr, int cold) if (!pagevec_add(&pages_to_free, page)) { if (zone) { - spin_unlock_irq(&zone->lru_lock); + spin_unlock_irqrestore(&zone->lru_lock, flags); zone = NULL; } __pagevec_free(&pages_to_free); @@ -296,7 +340,7 @@ void release_pages(struct page **pages, int nr, int cold) } } if (zone) - spin_unlock_irq(&zone->lru_lock); + spin_unlock_irqrestore(&zone->lru_lock, flags); pagevec_free(&pages_to_free); } @@ -489,7 +533,7 @@ static int cpu_swap_callback(struct notifier_block *nfb, if (action == CPU_DEAD || action == CPU_DEAD_FROZEN) { atomic_add(*committed, &vm_committed_space); *committed = 0; - __lru_add_drain((long)hcpu); + drain_cpu_pagevecs((long)hcpu); } return NOTIFY_OK; }