From e0e3f42fd96cf830c61aa720f0abb4eef41c5ce3 Mon Sep 17 00:00:00 2001 From: Qian Cai Date: Fri, 14 Aug 2020 17:31:37 -0700 Subject: [PATCH] mm/memcontrol: fix a data race in scan count struct mem_cgroup_per_node mz.lru_zone_size[zone_idx][lru] could be accessed concurrently as noticed by KCSAN, BUG: KCSAN: data-race in lruvec_lru_size / mem_cgroup_update_lru_size write to 0xffff9c804ca285f8 of 8 bytes by task 50951 on cpu 12: mem_cgroup_update_lru_size+0x11c/0x1d0 mem_cgroup_update_lru_size at mm/memcontrol.c:1266 isolate_lru_pages+0x6a9/0xf30 shrink_active_list+0x123/0xcc0 shrink_lruvec+0x8fd/0x1380 shrink_node+0x317/0xd80 do_try_to_free_pages+0x1f7/0xa10 try_to_free_pages+0x26c/0x5e0 __alloc_pages_slowpath+0x458/0x1290 __alloc_pages_nodemask+0x3bb/0x450 alloc_pages_vma+0x8a/0x2c0 do_anonymous_page+0x170/0x700 __handle_mm_fault+0xc9f/0xd00 handle_mm_fault+0xfc/0x2f0 do_page_fault+0x263/0x6f9 page_fault+0x34/0x40 read to 0xffff9c804ca285f8 of 8 bytes by task 50964 on cpu 95: lruvec_lru_size+0xbb/0x270 mem_cgroup_get_zone_lru_size at include/linux/memcontrol.h:536 (inlined by) lruvec_lru_size at mm/vmscan.c:326 shrink_lruvec+0x1d0/0x1380 shrink_node+0x317/0xd80 do_try_to_free_pages+0x1f7/0xa10 try_to_free_pages+0x26c/0x5e0 __alloc_pages_slowpath+0x458/0x1290 __alloc_pages_nodemask+0x3bb/0x450 alloc_pages_current+0xa6/0x120 alloc_slab_page+0x3b1/0x540 allocate_slab+0x70/0x660 new_slab+0x46/0x70 ___slab_alloc+0x4ad/0x7d0 __slab_alloc+0x43/0x70 kmem_cache_alloc+0x2c3/0x420 getname_flags+0x4c/0x230 getname+0x22/0x30 do_sys_openat2+0x205/0x3b0 do_sys_open+0x9a/0xf0 __x64_sys_openat+0x62/0x80 do_syscall_64+0x91/0xb47 entry_SYSCALL_64_after_hwframe+0x49/0xbe Reported by Kernel Concurrency Sanitizer on: CPU: 95 PID: 50964 Comm: cc1 Tainted: G W O L 5.5.0-next-20200204+ #6 Hardware name: HPE ProLiant DL385 Gen10/ProLiant DL385 Gen10, BIOS A40 07/10/2019 The write is under lru_lock, but the read is done as lockless. The scan count is used to determine how aggressively the anon and file LRU lists should be scanned. Load tearing could generate an inefficient heuristic, so fix it by adding READ_ONCE() for the read. Signed-off-by: Qian Cai Signed-off-by: Andrew Morton Cc: Johannes Weiner Cc: Michal Hocko Cc: Vladimir Davydov Link: http://lkml.kernel.org/r/20200206034945.2481-1-cai@lca.pw Signed-off-by: Linus Torvalds --- include/linux/memcontrol.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h index 385237e4cb44b9..d0b036123c6abd 100644 --- a/include/linux/memcontrol.h +++ b/include/linux/memcontrol.h @@ -630,7 +630,7 @@ unsigned long mem_cgroup_get_zone_lru_size(struct lruvec *lruvec, struct mem_cgroup_per_node *mz; mz = container_of(lruvec, struct mem_cgroup_per_node, lruvec); - return mz->lru_zone_size[zone_idx][lru]; + return READ_ONCE(mz->lru_zone_size[zone_idx][lru]); } void mem_cgroup_handle_over_high(void);