From patchwork Fri Nov 4 14:22:59 2022 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Mel Gorman X-Patchwork-Id: 13032068 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from kanga.kvack.org (kanga.kvack.org [205.233.56.17]) by smtp.lore.kernel.org (Postfix) with ESMTP id EE18DC4332F for ; Fri, 4 Nov 2022 14:23:05 +0000 (UTC) Received: by kanga.kvack.org (Postfix) id 009458E0001; Fri, 4 Nov 2022 10:23:05 -0400 (EDT) Received: by kanga.kvack.org (Postfix, from userid 40) id EFBD96B0073; Fri, 4 Nov 2022 10:23:04 -0400 (EDT) X-Delivered-To: int-list-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix, from userid 63042) id D9CA68E0001; Fri, 4 Nov 2022 10:23:04 -0400 (EDT) X-Delivered-To: linux-mm@kvack.org Received: from relay.hostedemail.com (smtprelay0011.hostedemail.com [216.40.44.11]) by kanga.kvack.org (Postfix) with ESMTP id C6D8B6B0071 for ; Fri, 4 Nov 2022 10:23:04 -0400 (EDT) Received: from smtpin29.hostedemail.com (a10.router.float.18 [10.200.18.1]) by unirelay02.hostedemail.com (Postfix) with ESMTP id 8C6D812147F for ; Fri, 4 Nov 2022 14:23:04 +0000 (UTC) X-FDA: 80095976688.29.8DA42AC Received: from outbound-smtp11.blacknight.com (outbound-smtp11.blacknight.com [46.22.139.106]) by imf09.hostedemail.com (Postfix) with ESMTP id D398D140004 for ; Fri, 4 Nov 2022 14:23:03 +0000 (UTC) Received: from mail.blacknight.com (pemlinmail06.blacknight.ie [81.17.255.152]) by outbound-smtp11.blacknight.com (Postfix) with ESMTPS id 13D0D1C3F74 for ; Fri, 4 Nov 2022 14:23:02 +0000 (GMT) Received: (qmail 14488 invoked from network); 4 Nov 2022 14:23:01 -0000 Received: from unknown (HELO techsingularity.net) (mgorman@techsingularity.net@[84.203.198.246]) by 81.17.254.9 with ESMTPSA (AES256-SHA encrypted, authenticated); 4 Nov 2022 14:23:01 -0000 Date: Fri, 4 Nov 2022 14:22:59 +0000 From: Mel Gorman To: Andrew Morton Cc: Yu Zhao , Vlastimil Babka , Nicolas Saenz Julienne , Marcelo Tosatti , Michal Hocko , Hugh Dickins , Marek Szyprowski , LKML , Linux-MM Subject: [PATCH v2] mm/page_alloc: Leave IRQs enabled for per-cpu page allocations Message-ID: <20221104142259.5hohev5hzvwanbi2@techsingularity.net> MIME-Version: 1.0 Content-Disposition: inline ARC-Authentication-Results: i=1; imf09.hostedemail.com; dkim=none; spf=pass (imf09.hostedemail.com: domain of mgorman@techsingularity.net designates 46.22.139.106 as permitted sender) smtp.mailfrom=mgorman@techsingularity.net; dmarc=none ARC-Seal: i=1; s=arc-20220608; d=hostedemail.com; t=1667571784; a=rsa-sha256; cv=none; b=6MiECW/4hYRBurEUeix0LM504ZcjfTSYZUsZUSET1yfUYIKPrHPCAVmO72+5O/jFdN+hkD 03L/ovGl+4UMKGICfJ2G03e2jF01HU8V0aNAZLOEVSoc52ybanhcg8CzR97GRORMKM0IuD Exw3KDPzx+Gg33QQZ59339aDKgi2aDI= ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=hostedemail.com; s=arc-20220608; t=1667571784; h=from:from:sender:reply-to:subject:subject:date:date: message-id:message-id:to:to:cc:cc:mime-version:mime-version: content-type:content-type:content-transfer-encoding:in-reply-to: references; bh=CATb8fzi+B3Ye+bsTEQIc7UKBhRvvbBMksOqVe1XUaw=; b=sqL+y9q+1Hf+u0SrK5oXvaHw3hkwY1+2uRyZ3Pli+GcNvxbBTFl+/rWqqNJzD05wFNrrgX s52m2ZbpEJTSS5EbaeHEc5A8NNVvBFkFcxHBl/FGI1CSu8opU6W5f/RY94Xq/m+kHodDDX C7bfg7nwjTeSA8F6kka7Besedyv+Oio= X-Stat-Signature: y7xc343dnasa7pgwqmhaceyo434q4yti X-Rspamd-Queue-Id: D398D140004 Authentication-Results: imf09.hostedemail.com; dkim=none; spf=pass (imf09.hostedemail.com: domain of mgorman@techsingularity.net designates 46.22.139.106 as permitted sender) smtp.mailfrom=mgorman@techsingularity.net; dmarc=none X-Rspam-User: X-Rspamd-Server: rspam12 X-HE-Tag: 1667571783-931504 X-Bogosity: Ham, tests=bogofilter, spamicity=0.000000, version=1.2.4 Sender: owner-linux-mm@kvack.org Precedence: bulk X-Loop: owner-majordomo@kvack.org List-ID: Changelog since v1 o Use trylock in free_unref_page_list due to IO completion from softirq context The pcp_spin_lock_irqsave protecting the PCP lists is IRQ-safe as a task allocating from the PCP must not re-enter the allocator from IRQ context. In each instance where IRQ-reentrancy is possible, the lock is acquired using pcp_spin_trylock_irqsave() even though IRQs are disabled and re-entrancy is impossible. Demote the lock to pcp_spin_lock avoids an IRQ disable/enable in the common case at the cost of some IRQ allocations taking a slower path. If the PCP lists need to be refilled, the zone lock still needs to disable IRQs but that will only happen on PCP refill and drain. If an IRQ is raised when a PCP allocation is in progress, the trylock will fail and fallback to using the buddy lists directly. Note that this may not be a universal win if an interrupt-intensive workload also allocates heavily from interrupt context and contends heavily on the zone->lock as a result. [yuzhao@google.com: Reported lockdep issue on IO completion from softirq] Signed-off-by: Mel Gorman Reported-by: Hugh Dickins Signed-off-by: Mel Gorman --- mm/page_alloc.c | 122 ++++++++++++++++++++++++-------------------------------- 1 file changed, 53 insertions(+), 69 deletions(-) diff --git a/mm/page_alloc.c b/mm/page_alloc.c index e20ade858e71..ae410adf36fb 100644 --- a/mm/page_alloc.c +++ b/mm/page_alloc.c @@ -170,21 +170,12 @@ static DEFINE_MUTEX(pcp_batch_high_lock); _ret; \ }) -#define pcpu_spin_lock_irqsave(type, member, ptr, flags) \ +#define pcpu_spin_trylock(type, member, ptr) \ ({ \ type *_ret; \ pcpu_task_pin(); \ _ret = this_cpu_ptr(ptr); \ - spin_lock_irqsave(&_ret->member, flags); \ - _ret; \ -}) - -#define pcpu_spin_trylock_irqsave(type, member, ptr, flags) \ -({ \ - type *_ret; \ - pcpu_task_pin(); \ - _ret = this_cpu_ptr(ptr); \ - if (!spin_trylock_irqsave(&_ret->member, flags)) { \ + if (!spin_trylock(&_ret->member)) { \ pcpu_task_unpin(); \ _ret = NULL; \ } \ @@ -197,27 +188,16 @@ static DEFINE_MUTEX(pcp_batch_high_lock); pcpu_task_unpin(); \ }) -#define pcpu_spin_unlock_irqrestore(member, ptr, flags) \ -({ \ - spin_unlock_irqrestore(&ptr->member, flags); \ - pcpu_task_unpin(); \ -}) - /* struct per_cpu_pages specific helpers. */ #define pcp_spin_lock(ptr) \ pcpu_spin_lock(struct per_cpu_pages, lock, ptr) -#define pcp_spin_lock_irqsave(ptr, flags) \ - pcpu_spin_lock_irqsave(struct per_cpu_pages, lock, ptr, flags) - -#define pcp_spin_trylock_irqsave(ptr, flags) \ - pcpu_spin_trylock_irqsave(struct per_cpu_pages, lock, ptr, flags) +#define pcp_spin_trylock(ptr) \ + pcpu_spin_trylock(struct per_cpu_pages, lock, ptr) #define pcp_spin_unlock(ptr) \ pcpu_spin_unlock(lock, ptr) -#define pcp_spin_unlock_irqrestore(ptr, flags) \ - pcpu_spin_unlock_irqrestore(lock, ptr, flags) #ifdef CONFIG_USE_PERCPU_NUMA_NODE_ID DEFINE_PER_CPU(int, numa_node); EXPORT_PER_CPU_SYMBOL(numa_node); @@ -1546,6 +1526,7 @@ static void free_pcppages_bulk(struct zone *zone, int count, struct per_cpu_pages *pcp, int pindex) { + unsigned long flags; int min_pindex = 0; int max_pindex = NR_PCP_LISTS - 1; unsigned int order; @@ -1561,8 +1542,7 @@ static void free_pcppages_bulk(struct zone *zone, int count, /* Ensure requested pindex is drained first. */ pindex = pindex - 1; - /* Caller must hold IRQ-safe pcp->lock so IRQs are disabled. */ - spin_lock(&zone->lock); + spin_lock_irqsave(&zone->lock, flags); isolated_pageblocks = has_isolate_pageblock(zone); while (count > 0) { @@ -1610,7 +1590,7 @@ static void free_pcppages_bulk(struct zone *zone, int count, } while (count > 0 && !list_empty(list)); } - spin_unlock(&zone->lock); + spin_unlock_irqrestore(&zone->lock, flags); } static void free_one_page(struct zone *zone, @@ -3124,10 +3104,10 @@ static int rmqueue_bulk(struct zone *zone, unsigned int order, unsigned long count, struct list_head *list, int migratetype, unsigned int alloc_flags) { + unsigned long flags; int i, allocated = 0; - /* Caller must hold IRQ-safe pcp->lock so IRQs are disabled. */ - spin_lock(&zone->lock); + spin_lock_irqsave(&zone->lock, flags); for (i = 0; i < count; ++i) { struct page *page = __rmqueue(zone, order, migratetype, alloc_flags); @@ -3161,7 +3141,7 @@ static int rmqueue_bulk(struct zone *zone, unsigned int order, * pages added to the pcp list. */ __mod_zone_page_state(zone, NR_FREE_PAGES, -(i << order)); - spin_unlock(&zone->lock); + spin_unlock_irqrestore(&zone->lock, flags); return allocated; } @@ -3178,16 +3158,9 @@ void drain_zone_pages(struct zone *zone, struct per_cpu_pages *pcp) batch = READ_ONCE(pcp->batch); to_drain = min(pcp->count, batch); if (to_drain > 0) { - unsigned long flags; - - /* - * free_pcppages_bulk expects IRQs disabled for zone->lock - * so even though pcp->lock is not intended to be IRQ-safe, - * it's needed in this context. - */ - spin_lock_irqsave(&pcp->lock, flags); + spin_lock(&pcp->lock); free_pcppages_bulk(zone, to_drain, pcp, 0); - spin_unlock_irqrestore(&pcp->lock, flags); + spin_unlock(&pcp->lock); } } #endif @@ -3201,12 +3174,9 @@ static void drain_pages_zone(unsigned int cpu, struct zone *zone) pcp = per_cpu_ptr(zone->per_cpu_pageset, cpu); if (pcp->count) { - unsigned long flags; - - /* See drain_zone_pages on why this is disabling IRQs */ - spin_lock_irqsave(&pcp->lock, flags); + spin_lock(&pcp->lock); free_pcppages_bulk(zone, pcp->count, pcp, 0); - spin_unlock_irqrestore(&pcp->lock, flags); + spin_unlock(&pcp->lock); } } @@ -3472,7 +3442,6 @@ static void free_unref_page_commit(struct zone *zone, struct per_cpu_pages *pcp, */ void free_unref_page(struct page *page, unsigned int order) { - unsigned long flags; unsigned long __maybe_unused UP_flags; struct per_cpu_pages *pcp; struct zone *zone; @@ -3500,10 +3469,10 @@ void free_unref_page(struct page *page, unsigned int order) zone = page_zone(page); pcp_trylock_prepare(UP_flags); - pcp = pcp_spin_trylock_irqsave(zone->per_cpu_pageset, flags); + pcp = pcp_spin_trylock(zone->per_cpu_pageset); if (pcp) { free_unref_page_commit(zone, pcp, page, migratetype, order); - pcp_spin_unlock_irqrestore(pcp, flags); + pcp_spin_unlock(pcp); } else { free_one_page(zone, page, pfn, order, migratetype, FPI_NONE); } @@ -3515,10 +3484,10 @@ void free_unref_page(struct page *page, unsigned int order) */ void free_unref_page_list(struct list_head *list) { + unsigned long __maybe_unused UP_flags; struct page *page, *next; struct per_cpu_pages *pcp = NULL; struct zone *locked_zone = NULL; - unsigned long flags; int batch_count = 0; int migratetype; @@ -3547,11 +3516,26 @@ void free_unref_page_list(struct list_head *list) /* Different zone, different pcp lock. */ if (zone != locked_zone) { - if (pcp) - pcp_spin_unlock_irqrestore(pcp, flags); + if (pcp) { + pcp_spin_unlock(pcp); + pcp_trylock_finish(UP_flags); + } + /* + * trylock is necessary as pages may be getting freed + * from IRQ or SoftIRQ context after an IO completion. + */ + pcp_trylock_prepare(UP_flags); + pcp = pcp_spin_trylock(zone->per_cpu_pageset); + if (!pcp) { + pcp_trylock_finish(UP_flags); + list_del(&page->lru); + free_one_page(page_zone(page), page, + page_to_pfn(page), 0, migratetype, + FPI_NONE); + continue; + } locked_zone = zone; - pcp = pcp_spin_lock_irqsave(locked_zone->per_cpu_pageset, flags); } /* @@ -3566,18 +3550,23 @@ void free_unref_page_list(struct list_head *list) free_unref_page_commit(zone, pcp, page, migratetype, 0); /* - * Guard against excessive IRQ disabled times when we get - * a large list of pages to free. + * Guard against excessive IRQ disabled times when freeing + * a large list of pages. Lock will be reacquired if + * necessary on the next iteration. */ if (++batch_count == SWAP_CLUSTER_MAX) { - pcp_spin_unlock_irqrestore(pcp, flags); + pcp_spin_unlock(pcp); + pcp_trylock_finish(UP_flags); batch_count = 0; - pcp = pcp_spin_lock_irqsave(locked_zone->per_cpu_pageset, flags); + pcp = NULL; + locked_zone = NULL; } } - if (pcp) - pcp_spin_unlock_irqrestore(pcp, flags); + if (pcp) { + pcp_spin_unlock(pcp); + pcp_trylock_finish(UP_flags); + } } /* @@ -3778,15 +3767,11 @@ static struct page *rmqueue_pcplist(struct zone *preferred_zone, struct per_cpu_pages *pcp; struct list_head *list; struct page *page; - unsigned long flags; unsigned long __maybe_unused UP_flags; - /* - * spin_trylock may fail due to a parallel drain. In the future, the - * trylock will also protect against IRQ reentrancy. - */ + /* spin_trylock may fail due to a parallel drain or IRQ reentrancy. */ pcp_trylock_prepare(UP_flags); - pcp = pcp_spin_trylock_irqsave(zone->per_cpu_pageset, flags); + pcp = pcp_spin_trylock(zone->per_cpu_pageset); if (!pcp) { pcp_trylock_finish(UP_flags); return NULL; @@ -3800,7 +3785,7 @@ static struct page *rmqueue_pcplist(struct zone *preferred_zone, pcp->free_factor >>= 1; list = &pcp->lists[order_to_pindex(migratetype, order)]; page = __rmqueue_pcplist(zone, order, migratetype, alloc_flags, pcp, list); - pcp_spin_unlock_irqrestore(pcp, flags); + pcp_spin_unlock(pcp); pcp_trylock_finish(UP_flags); if (page) { __count_zid_vm_events(PGALLOC, page_zonenum(page), 1 << order); @@ -5368,7 +5353,6 @@ unsigned long __alloc_pages_bulk(gfp_t gfp, int preferred_nid, struct page **page_array) { struct page *page; - unsigned long flags; unsigned long __maybe_unused UP_flags; struct zone *zone; struct zoneref *z; @@ -5450,9 +5434,9 @@ unsigned long __alloc_pages_bulk(gfp_t gfp, int preferred_nid, if (unlikely(!zone)) goto failed; - /* Is a parallel drain in progress? */ + /* spin_trylock may fail due to a parallel drain or IRQ reentrancy. */ pcp_trylock_prepare(UP_flags); - pcp = pcp_spin_trylock_irqsave(zone->per_cpu_pageset, flags); + pcp = pcp_spin_trylock(zone->per_cpu_pageset); if (!pcp) goto failed_irq; @@ -5471,7 +5455,7 @@ unsigned long __alloc_pages_bulk(gfp_t gfp, int preferred_nid, if (unlikely(!page)) { /* Try and allocate at least one page */ if (!nr_account) { - pcp_spin_unlock_irqrestore(pcp, flags); + pcp_spin_unlock(pcp); goto failed_irq; } break; @@ -5486,7 +5470,7 @@ unsigned long __alloc_pages_bulk(gfp_t gfp, int preferred_nid, nr_populated++; } - pcp_spin_unlock_irqrestore(pcp, flags); + pcp_spin_unlock(pcp); pcp_trylock_finish(UP_flags); __count_zid_vm_events(PGALLOC, zone_idx(zone), nr_account);