mm: restore node stat checking in /proc/sys/vm/stat_refresh
authorHugh Dickins <hughd@google.com>
Wed, 5 May 2021 01:37:51 +0000 (18:37 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Wed, 5 May 2021 18:27:25 +0000 (11:27 -0700)
In v4.7 commit 52b6f46bc163 ("mm: /proc/sys/vm/stat_refresh to force
vmstat update") introduced vmstat_refresh(), with its vmstat underflow
checking; then in v4.8 commit 75ef71840539 ("mm, vmstat: add
infrastructure for per-node vmstats") split NR_VM_NODE_STAT_ITEMS out of
NR_VM_ZONE_STAT_ITEMS without updating vmstat_refresh(): so it has been
missing out much of the vmstat underflow checking ever since.

Reinstate it.

Thanks to Roman Gushchin <guro@fb.com> for tangentially pointing this out.

Link: https://lkml.kernel.org/r/alpine.LSU.2.11.2102251502240.13363@eggly.anvils
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: Roman Gushchin <guro@fb.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/vmstat.c

index 49a8456..14d0ba4 100644 (file)
@@ -1875,6 +1875,14 @@ int vmstat_refresh(struct ctl_table *table, int write,
                }
        }
 #endif
+       for (i = 0; i < NR_VM_NODE_STAT_ITEMS; i++) {
+               val = atomic_long_read(&vm_node_stat[i]);
+               if (val < 0) {
+                       pr_warn("%s: %s %ld\n",
+                               __func__, node_stat_name(i), val);
+                       err = -EINVAL;
+               }
+       }
        if (err)
                return err;
        if (write)