mm/page_alloc: fix freeing static percpu memory

The size of struct per_cpu_zonestat can be 0 on !SMP && !NUMA.  In that
case, zone->per_cpu_zonestats will always equal to boot_zonestats.  But in
zone_pcp_reset(), zone->per_cpu_zonestats is freed via free_percpu()
directly without checking against boot_zonestats first.  boot_zonestats
will be released by free_percpu() unexpectedly.

Link: https://lkml.kernel.org/r/20220916072257.9639-7-linmiaohe@huawei.com
Fixes: 28f836b6 ("mm/page_alloc: split per cpu page lists and zone stats")
Signed-off-by: NMiaohe Lin <linmiaohe@huawei.com>
Reviewed-by: NDavid Hildenbrand <david@redhat.com>
Reviewed-by: NOscar Salvador <osalvador@suse.de>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>

mm/page_alloc.c

@@ -9513,9 +9513,11 @@ void zone_pcp_reset(struct zone *zone)
 			drain_zonestat(zone, pzstats);
 		}
 		free_percpu(zone->per_cpu_pageset);
-		free_percpu(zone->per_cpu_zonestats);
 		zone->per_cpu_pageset = &boot_pageset;
-		zone->per_cpu_zonestats = &boot_zonestats;
+		if (zone->per_cpu_zonestats != &boot_zonestats) {
+			free_percpu(zone->per_cpu_zonestats);
+			zone->per_cpu_zonestats = &boot_zonestats;
+		}
 	}
 }