[PATCH 1/1] sched/numa: Fix memory leak due to the overwritten vma->numab_state

[Adrian Huang]

From: Adrian Huang <ahuang12@xxxxxxxxxx>

[Problem Description]
When running the hackbench program of LTP, the following memory leak is
reported by kmemleak.

  # /opt/ltp/testcases/bin/hackbench 20 thread 1000
  Running with 20*40 (== 800) tasks.

  # dmesg | grep kmemleak
  ...
  kmemleak: 480 new suspected memory leaks (see /sys/kernel/debug/kmemleak)
  kmemleak: 665 new suspected memory leaks (see /sys/kernel/debug/kmemleak)

  # cat /sys/kernel/debug/kmemleak
  unreferenced object 0xffff888cd8ca2c40 (size 64):
    comm "hackbench", pid 17142, jiffies 4299780315
    hex dump (first 32 bytes):
      ac 74 49 00 01 00 00 00 4c 84 49 00 01 00 00 00  .tI.....L.I.....
      00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
    backtrace (crc bff18fd4):
      [<ffffffff81419a89>] __kmalloc_cache_noprof+0x2f9/0x3f0
      [<ffffffff8113f715>] task_numa_work+0x725/0xa00
      [<ffffffff8110f878>] task_work_run+0x58/0x90
      [<ffffffff81ddd9f8>] syscall_exit_to_user_mode+0x1c8/0x1e0
      [<ffffffff81dd78d5>] do_syscall_64+0x85/0x150
      [<ffffffff81e0012b>] entry_SYSCALL_64_after_hwframe+0x76/0x7e
  ...

  This issue can be consistently reproduced on three different servers:
    * a 448-core server
    * a 256-core server
    * a 192-core server

[Root Cause]
Since multiple threads are created by the hackbench program (along with
the command argument 'thread'), a shared vma might be accessed by two or
more cores simultaneously. When two or more cores observe that
vma->numab_state is NULL at the same time, vma->numab_state will be
overwritten.

Note that the command `/opt/ltp/testcases/bin/hackbench 50 process 1000`
cannot the reproduce the issue because of the fork() and COW. It is
verified with 200+ test runs.

[Solution]
Introduce a lock to make sure the atomic operation of the vma->numab_state
access.

Fixes: ef6a22b70f6d ("sched/numa: apply the scan delay to every new vma")
Reported-by: Jiwei Sun <sunjw10@xxxxxxxxxx>
Signed-off-by: Adrian Huang <ahuang12@xxxxxxxxxx>
---
 include/linux/mm.h       |  1 +
 include/linux/mm_types.h |  1 +
 kernel/sched/fair.c      | 17 ++++++++++++++++-
 3 files changed, 18 insertions(+), 1 deletion(-)

diff --git a/include/linux/mm.h b/include/linux/mm.h
index 61fff5d34ed5..a08e31ac53de 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -673,6 +673,7 @@ struct vm_operations_struct {
 static inline void vma_numab_state_init(struct vm_area_struct *vma)
 {
 	vma->numab_state = NULL;
+	mutex_init(&vma->numab_state_lock);
 }
 static inline void vma_numab_state_free(struct vm_area_struct *vma)
 {
diff --git a/include/linux/mm_types.h b/include/linux/mm_types.h
index 6e3bdf8e38bc..77eee89a89f5 100644
--- a/include/linux/mm_types.h
+++ b/include/linux/mm_types.h
@@ -768,6 +768,7 @@ struct vm_area_struct {
 #endif
 #ifdef CONFIG_NUMA_BALANCING
 	struct vma_numab_state *numab_state;	/* NUMA Balancing state */
+	struct mutex numab_state_lock;		/* NUMA Balancing state lock */
 #endif
 	struct vm_userfaultfd_ctx vm_userfaultfd_ctx;
 } __randomize_layout;
diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index c157d4860a3b..53e6383cd94e 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -3397,12 +3397,24 @@ static void task_numa_work(struct callback_head *work)
 			continue;
 		}
 
+		/*
+		 * In case of the shared vma, the vma->numab_state will be
+		 * overwritten if two or more cores observe vma->numab_state
+		 * is NULL at the same time. Make sure that only one core
+		 * allocates memory for vma->numab_state. This can prevent
+		 * the memory leak.
+		 */
+		if (!mutex_trylock(&vma->numab_state_lock))
+			continue;
+
 		/* Initialise new per-VMA NUMAB state. */
 		if (!vma->numab_state) {
 			vma->numab_state = kzalloc(sizeof(struct vma_numab_state),
 				GFP_KERNEL);
-			if (!vma->numab_state)
+			if (!vma->numab_state) {
+				mutex_unlock(&vma->numab_state_lock);
 				continue;
+			}
 
 			vma->numab_state->start_scan_seq = mm->numa_scan_seq;
 
@@ -3428,6 +3440,7 @@ static void task_numa_work(struct callback_head *work)
 		if (mm->numa_scan_seq && time_before(jiffies,
 						vma->numab_state->next_scan)) {
 			trace_sched_skip_vma_numa(mm, vma, NUMAB_SKIP_SCAN_DELAY);
+			mutex_unlock(&vma->numab_state_lock);
 			continue;
 		}
 
@@ -3440,6 +3453,8 @@ static void task_numa_work(struct callback_head *work)
 			vma->numab_state->pids_active[1] = 0;
 		}
 
+		mutex_unlock(&vma->numab_state_lock);
+
 		/* Do not rescan VMAs twice within the same sequence. */
 		if (vma->numab_state->prev_scan_seq == mm->numa_scan_seq) {
 			mm->numa_scan_offset = vma->vm_end;
-- 
2.34.1