diff mbox series

[merged] mm-compaction-count-pages-and-stop-correctly-during-page-isolation.patch removed from -mm tree

Message ID 20201116184106.hIvF6hfqf%akpm@linux-foundation.org
State New
Headers show
Series [merged] mm-compaction-count-pages-and-stop-correctly-during-page-isolation.patch removed from -mm tree | expand

Commit Message

Andrew Morton Nov. 16, 2020, 6:41 p.m. UTC
The patch titled
     Subject: mm/compaction: count pages and stop correctly during page isolation
has been removed from the -mm tree.  Its filename was
     mm-compaction-count-pages-and-stop-correctly-during-page-isolation.patch

This patch was dropped because it was merged into mainline or a subsystem tree

------------------------------------------------------
From: Zi Yan <ziy@nvidia.com>
Subject: mm/compaction: count pages and stop correctly during page isolation

In isolate_migratepages_block, when cc->alloc_contig is true, we are able
to isolate compound pages, nr_migratepages and nr_isolated did not count
compound pages correctly, causing us to isolate more pages than we
thought.  Count compound pages as the number of base pages they contain. 
Otherwise, we might be trapped in too_many_isolated while loop, since the
actual isolated pages can go up to COMPACT_CLUSTER_MAX*512=16384, where
COMPACT_CLUSTER_MAX is 32, since we stop isolation after
cc->nr_migratepages reaches to COMPACT_CLUSTER_MAX.

In addition, after we fix the issue above, cc->nr_migratepages could never
be equal to COMPACT_CLUSTER_MAX if compound pages are isolated, thus page
isolation could not stop as we intended.  Change the isolation stop
condition to >=.

The issue can be triggered as follows: In a system with 16GB memory and an
8GB CMA region reserved by hugetlb_cma, if we first allocate 10GB THPs and
mlock them (so some THPs are allocated in the CMA region and mlocked),
reserving 6 1GB hugetlb pages via
/sys/kernel/mm/hugepages/hugepages-1048576kB/nr_hugepages will get stuck
(looping in too_many_isolated function) until we kill either task.  With
the patch applied, oom will kill the application with 10GB THPs and let
hugetlb page reservation finish.

[ziy@nvidia.com: v3]
  Link: https://lkml.kernel.org/r/20201030183809.3616803-1-zi.yan@sent.com
Link: https://lkml.kernel.org/r/20201029200435.3386066-1-zi.yan@sent.com
Fixes: 1da2f328fa64 ("cmm,thp,compaction,cma: allow THP migration for CMA allocations")
Signed-off-by: Zi Yan <ziy@nvidia.com>
Reviewed-by: Yang Shi <shy828301@gmail.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Rik van Riel <riel@surriel.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
---

 mm/compaction.c |    8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)
diff mbox series

Patch

--- a/mm/compaction.c~mm-compaction-count-pages-and-stop-correctly-during-page-isolation
+++ a/mm/compaction.c
@@ -1012,8 +1012,8 @@  isolate_migratepages_block(struct compac
 
 isolate_success:
 		list_add(&page->lru, &cc->migratepages);
-		cc->nr_migratepages++;
-		nr_isolated++;
+		cc->nr_migratepages += compound_nr(page);
+		nr_isolated += compound_nr(page);
 
 		/*
 		 * Avoid isolating too much unless this block is being
@@ -1021,7 +1021,7 @@  isolate_success:
 		 * or a lock is contended. For contention, isolate quickly to
 		 * potentially remove one source of contention.
 		 */
-		if (cc->nr_migratepages == COMPACT_CLUSTER_MAX &&
+		if (cc->nr_migratepages >= COMPACT_CLUSTER_MAX &&
 		    !cc->rescan && !cc->contended) {
 			++low_pfn;
 			break;
@@ -1132,7 +1132,7 @@  isolate_migratepages_range(struct compac
 		if (!pfn)
 			break;
 
-		if (cc->nr_migratepages == COMPACT_CLUSTER_MAX)
+		if (cc->nr_migratepages >= COMPACT_CLUSTER_MAX)
 			break;
 	}