Date:   Wed, 22 Nov 2023 09:39:25 +0800
From:   Zhiguo Jiang <justinjiang@...o.com>
To:     Andrew Morton <akpm@...ux-foundation.org>, linux-mm@...ck.org,
        linux-kernel@...r.kernel.org
Cc:     Matthew Wilcox <willy@...radead.org>,
        Johannes Weiner <hannes@...xchg.org>,
        opensource.kernel@...o.com, Zhiguo Jiang <justinjiang@...o.com>
Subject: [PATCH v3] mm: ALLOC_HIGHATOMIC flag allocation issue

Update base on the latest Linus' tree or Andrew's tree.

Signed-off-by: Zhiguo Jiang <justinjiang@...o.com>
---

Changelog:
v1:
In case that alloc_flags contains ALLOC_HIGHATOMIC and alloc order
is order1/2/3/10 in rmqueue(), if pages are alloced successfully
from pcplist, a free pageblock will be also moved from the alloced
migratetype freelist to MIGRATE_HIGHATOMIC freelist, rather than
alloc from MIGRATE_HIGHATOMIC freelist firstly, so this will result
in an increasing number of pages on the MIGRATE_HIGHATOMIC freelist,
pages in other migratetype freelist are reduced and more likely to
allocation failure.

Currently the sequence of ALLOC_HIGHATOMIC allocation is:
pcplist --> rmqueue_bulk() --> rmqueue_buddy() MIGRATE_HIGHATOMIC
--> rmqueue_buddy() allocation migratetype.

Due to the fact that requesting pages from the pcplist is faster than
buddy, the sequence of modifying the ALLOC_HIGHATOMIC allocation is:
pcplist --> rmqueue_buddy() MIGRATE_HIGHATOMIC --> rmqueue_buddy()
allocation migratetype.

This patch can solve the failure problem of allocating other types of
pages due to excessive MIGRATE_HIGHATOMIC freelist reservations.

In comparative testing, cat /proc/pagetypeinfo and the HighAtomic
freelist size is:
Test without this patch:
Node 0, zone Normal, type HighAtomic 2369 771 138 15 0 0 0 0 0 0 0
Test with this patch:
Node 0, zone Normal, type HighAtomic 206 82 4 2 1 0 0 0 0 0 0

v2:
Update comments and modify variable highatomic_allocation to highatomic.

 mm/page_alloc.c | 37 +++++++++++++++++++------------------
 1 file changed, 19 insertions(+), 18 deletions(-)

diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 49890d00cc3c..693e86fc9850 100755
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -2851,9 +2851,9 @@ struct page *__rmqueue_pcplist(struct zone *zone, unsigned int order,
 			int alloced;
 
 			/*
-			 * If pcplist is empty and alloc_flags is with ALLOC_HIGHATOMIC,
-			 * it should alloc from buddy highatomic migrate freelist firstly
-			 * to ensure quick and successful allocation.
+			 * If pcplist is empty and alloc_flags contains
+			 * ALLOC_HIGHATOMIC, alloc from buddy highatomic
+			 * freelist first.
 			 */
 			if (alloc_flags & ALLOC_HIGHATOMIC)
 				goto out;
@@ -2927,7 +2927,7 @@ static inline
 struct page *rmqueue(struct zone *preferred_zone,
 			struct zone *zone, unsigned int order,
 			gfp_t gfp_flags, unsigned int alloc_flags,
-			int migratetype, bool *highatomc_allocation)
+			int migratetype, bool *highatomic)
 {
 	struct page *page;
 
@@ -2948,21 +2948,22 @@ struct page *rmqueue(struct zone *preferred_zone,
 							migratetype);
 
 	/*
-	 * The high-order atomic allocation pageblock reserved conditions:
+	 * The high-order atomic allocation pageblock reserved:
 	 *
-	 * If the high-order atomic allocation page is alloced from pcplist,
-	 * the highatomic pageblock does not need to be reserved, which can
-	 * void to migrate an increasing number of pages into buddy
-	 * MIGRATE_HIGHATOMIC freelist and lead to an increasing risk of
-	 * allocation failure on other buddy migrate freelists.
+	 * If the high-order atomic page is allocated from pcplist,
+	 * the highatomic pageblock does not need to be reserved,
+	 * which can avoid migrating an increasing number of pages
+	 * into buddy highatomic freelist and leading to an increased
+	 * risk of allocation failure on other migrate freelists in
+	 * buddy.
 	 *
-	 * If the high-order atomic allocation page is alloced from buddy
-	 * highatomic migrate freelist, regardless of whether the allocation
-	 * is successful or not, the highatomic pageblock can try to be
-	 * reserved.
+	 * If the high-order atomic page is allocated from buddy
+	 * highatomic freelist, regardless of whether the allocation
+	 * is successful or not, the highatomic pageblock can try to
+	 * be reserved.
 	 */
 	if (unlikely(alloc_flags & ALLOC_HIGHATOMIC))
-		*highatomc_allocation = true;
+		*highatomic = true;
 
 out:
 	/* Separate test+clear to avoid unnecessary atomics */
@@ -3234,7 +3235,7 @@ get_page_from_freelist(gfp_t gfp_mask, unsigned int order, int alloc_flags,
 	struct pglist_data *last_pgdat = NULL;
 	bool last_pgdat_dirty_ok = false;
 	bool no_fallback;
-	bool highatomc_allocation = false;
+	bool highatomic = false;
 
 retry:
 	/*
@@ -3366,7 +3367,7 @@ get_page_from_freelist(gfp_t gfp_mask, unsigned int order, int alloc_flags,
 
 try_this_zone:
 		page = rmqueue(ac->preferred_zoneref->zone, zone, order,
-				gfp_mask, alloc_flags, ac->migratetype, &highatomc_allocation);
+				gfp_mask, alloc_flags, ac->migratetype, &highatomic);
 		if (page) {
 			prep_new_page(page, order, gfp_mask, alloc_flags);
 
@@ -3374,7 +3375,7 @@ get_page_from_freelist(gfp_t gfp_mask, unsigned int order, int alloc_flags,
 			 * If this is a high-order atomic allocation then check
 			 * if the pageblock should be reserved for the future
 			 */
-			if (unlikely(highatomc_allocation))
+			if (unlikely(highatomic))
 				reserve_highatomic_pageblock(page, zone);
 
 			return page;
-- 
2.39.0

Powered by blists - more mailing lists