mm: Fix for negative counter: nr_file_hugepages

PATCH Applied Stefan Roesch 4 participants 4 messages Nov 8, 2023 – Nov 14, 2023

Subsystems (3)

Stable branch Memory management THP

lore

Share & tools

lore

Stefan Roesch

author

2023-11-08 17:15 UTC #0 lore raw

PATCH v3 1/1 mm: Fix for negative counter: nr_file_hugepages

While qualifiying the 6.4 release, the following warning was detected in messages:

vmstat_refresh: nr_file_hugepages -15664

The warning is caused by the incorrect updating of the NR_FILE_THPS counter in the function split_huge_page_to_list. The if case is checking for folio_test_swapbacked, but the else case is missing the check for folio_test_pmd_mappable. The other functions that manipulate the counter like __filemap_add_folio and filemap_unaccount_folio have the corresponding check.

I have a test case, which reproduces the problem. It can be found here:
https://github.com/sroeschus/testcase/blob/main/vmstat_refresh/madv.c

The test case reproduces on an XFS filesystem. Running the same test case on a BTRFS filesystem does not reproduce the problem.

AFAIK version 6.1 until 6.6 are affected by this problem.

Signed-off-by Stefan Roesch <[email protected]>

Co-debugged-by Johannes Weiner <[email protected]>

Acked-by Johannes Weiner <[email protected]>

Cc [email protected]

---
 mm/huge_memory.c | 16 +++++++++-------
 1 file changed, 9 insertions(+), 7 deletions(-)

mm/huge_memory.c +9 -7

index 064fbd90822b4..874000f97bfc1 100644
@@ -2737,13 +2737,15 @@ int split_huge_page_to_list(struct page *page, struct list_head *list)
 			int nr = folio_nr_pages(folio);
 
 			xas_split(&xas, folio, folio_order(folio));
-			if (folio_test_swapbacked(folio)) {
-				__lruvec_stat_mod_folio(folio, NR_SHMEM_THPS,
-							-nr);
-			} else {
-				__lruvec_stat_mod_folio(folio, NR_FILE_THPS,
-							-nr);
-				filemap_nr_thps_dec(mapping);
+			if (folio_test_pmd_mappable(folio)) {
+				if (folio_test_swapbacked(folio)) {
+					__lruvec_stat_mod_folio(folio,
+							NR_SHMEM_THPS, -nr);
+				} else {
+					__lruvec_stat_mod_folio(folio,
+							NR_FILE_THPS, -nr);
+					filemap_nr_thps_dec(mapping);
+				}
 			}
 		}

base-commit: ffc253263a1375a65fa6c9f62a893e9767fbebfa

Signature · 1 line

2.39.3

Matthew Wilcox

Reviewed-by on v3 1/1

2023-11-08 17:30 UTC #1 lore raw

Quoted 18 lines · Stefan Roesch

While qualifiying the 6.4 release, the following warning was detected in messages:
vmstat_refresh: nr_file_hugepages -15664
The warning is caused by the incorrect updating of the NR_FILE_THPS counter in the function split_huge_page_to_list. The if case is checking for folio_test_swapbacked, but the else case is missing the check for folio_test_pmd_mappable. The other functions that manipulate the counter like __filemap_add_folio and filemap_unaccount_folio have the corresponding check.
I have a test case, which reproduces the problem. It can be found here:
https://github.com/sroeschus/testcase/blob/main/vmstat_refresh/madv.c
The test case reproduces on an XFS filesystem. Running the same test case on a BTRFS filesystem does not reproduce the problem.

Quoted 6 lines

AFAIK version 6.1 until 6.6 are affected by this problem.
Signed-off-by Stefan Roesch <[email protected]>
Co-debugged-by Johannes Weiner <[email protected]>
Acked-by Johannes Weiner <[email protected]>
Cc [email protected]

Reviewed-by Matthew Wilcox (Oracle) <[email protected]>

David Hildenbrand

Reviewed-by on v3 1/1

2023-11-08 17:38 UTC #2 lore raw

In reply to Stefan Roesch

Quoted 42 lines · Stefan Roesch

While qualifiying the 6.4 release, the following warning was detected in messages:
vmstat_refresh: nr_file_hugepages -15664
The warning is caused by the incorrect updating of the NR_FILE_THPS counter in the function split_huge_page_to_list. The if case is checking for folio_test_swapbacked, but the else case is missing the check for folio_test_pmd_mappable. The other functions that manipulate the counter like __filemap_add_folio and filemap_unaccount_folio have the corresponding check.
I have a test case, which reproduces the problem. It can be found here:
https://github.com/sroeschus/testcase/blob/main/vmstat_refresh/madv.c
The test case reproduces on an XFS filesystem. Running the same test case on a BTRFS filesystem does not reproduce the problem.
AFAIK version 6.1 until 6.6 are affected by this problem.
Signed-off-by Stefan Roesch <[email protected]>
Co-debugged-by Johannes Weiner <[email protected]>
Acked-by Johannes Weiner <[email protected]>
Cc [email protected]
---
  mm/huge_memory.c | 16 +++++++++-------
  1 file changed, 9 insertions(+), 7 deletions(-)
mm/huge_memory.c +0 -6
index 064fbd90822b4..874000f97bfc1 100644
@@ -2737,13 +2737,15 @@ int split_huge_page_to_list(struct page *page, struct list_head *list)
  			int nr = folio_nr_pages(folio);
 
  			xas_split(&xas, folio, folio_order(folio));
-			if (folio_test_swapbacked(folio)) {
-				__lruvec_stat_mod_folio(folio, NR_SHMEM_THPS,
-							-nr);
-			} else {
-				__lruvec_stat_mod_folio(folio, NR_FILE_THPS,
-							-nr);

mm/huge_memory.c int split_huge_page_to_list(struct page *page, struct list_head *list) · line 2748

Quoted 10 lines

mm/huge_memory.c +9 -1

-				filemap_nr_thps_dec(mapping);
+			if (folio_test_pmd_mappable(folio)) {
+				if (folio_test_swapbacked(folio)) {
+					__lruvec_stat_mod_folio(folio,
+							NR_SHMEM_THPS, -nr);
+				} else {
+					__lruvec_stat_mod_folio(folio,
+							NR_FILE_THPS, -nr);
+					filemap_nr_thps_dec(mapping);
+				}

Reviewed-by David Hildenbrand <[email protected]>

Yes, that's the current state: update these counters only for (traditional, IOW PMD-sized) THP. What we'll do with non-pmd-sized THP remains to be discussed (Ryan had some ideas).

Signature · 3 lines

Cheers,

David / dhildenb

Yang Shi

Reviewed-by on v3 1/1

2023-11-14 01:42 UTC #3 lore raw

In reply to Stefan Roesch

Quoted 19 lines · Stefan Roesch

While qualifiying the 6.4 release, the following warning was detected in messages:
vmstat_refresh: nr_file_hugepages -15664
The warning is caused by the incorrect updating of the NR_FILE_THPS counter in the function split_huge_page_to_list. The if case is checking for folio_test_swapbacked, but the else case is missing the check for folio_test_pmd_mappable. The other functions that manipulate the counter like __filemap_add_folio and filemap_unaccount_folio have the corresponding check.
I have a test case, which reproduces the problem. It can be found here:
https://github.com/sroeschus/testcase/blob/main/vmstat_refresh/madv.c
The test case reproduces on an XFS filesystem. Running the same test case on a BTRFS filesystem does not reproduce the problem.

Quoted 6 lines

AFAIK version 6.1 until 6.6 are affected by this problem.
Signed-off-by Stefan Roesch <[email protected]>
Co-debugged-by Johannes Weiner <[email protected]>
Acked-by Johannes Weiner <[email protected]>
Cc [email protected]

Reviewed-by Yang Shi <[email protected]>

Quoted 37 lines

---
 mm/huge_memory.c | 16 +++++++++-------
 1 file changed, 9 insertions(+), 7 deletions(-)

mm/huge_memory.c +9 -7

index 064fbd90822b4..874000f97bfc1 100644
@@ -2737,13 +2737,15 @@ int split_huge_page_to_list(struct page *page, struct list_head *list)
                        int nr = folio_nr_pages(folio);
 
                        xas_split(&xas, folio, folio_order(folio));
-                       if (folio_test_swapbacked(folio)) {
-                               __lruvec_stat_mod_folio(folio, NR_SHMEM_THPS,
-                                                       -nr);
-                       } else {
-                               __lruvec_stat_mod_folio(folio, NR_FILE_THPS,
-                                                       -nr);
-                               filemap_nr_thps_dec(mapping);
+                       if (folio_test_pmd_mappable(folio)) {
+                               if (folio_test_swapbacked(folio)) {
+                                       __lruvec_stat_mod_folio(folio,
+                                                       NR_SHMEM_THPS, -nr);
+                               } else {
+                                       __lruvec_stat_mod_folio(folio,
+                                                       NR_FILE_THPS, -nr);
+                                       filemap_nr_thps_dec(mapping);
+                               }
                        }
                }

base-commit: ffc253263a1375a65fa6c9f62a893e9767fbebfa --
2.39.3