[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-Id: <20241108133139.25326-1-ahuang12@lenovo.com>
Date: Fri, 8 Nov 2024 21:31:39 +0800
From: Adrian Huang <adrianhuang0701@...il.com>
To: Ingo Molnar <mingo@...hat.com>,
Peter Zijlstra <peterz@...radead.org>,
Juri Lelli <juri.lelli@...hat.com>,
Vincent Guittot <vincent.guittot@...aro.org>,
Andrew Morton <akpm@...ux-foundation.org>
Cc: Dietmar Eggemann <dietmar.eggemann@....com>,
Steven Rostedt <rostedt@...dmis.org>,
Ben Segall <bsegall@...gle.com>,
Mel Gorman <mgorman@...e.de>,
Valentin Schneider <vschneid@...hat.com>,
Raghavendra K T <raghavendra.kt@....com>,
linux-mm@...ck.org,
linux-kernel@...r.kernel.org,
Adrian Huang <ahuang12@...ovo.com>,
Jiwei Sun <sunjw10@...ovo.com>
Subject: [PATCH 1/1] sched/numa: Fix memory leak due to the overwritten vma->numab_state
From: Adrian Huang <ahuang12@...ovo.com>
[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@...ovo.com>
Signed-off-by: Adrian Huang <ahuang12@...ovo.com>
---
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
Powered by blists - more mailing lists