Skip to contentSkip to conversation

sched/deadline: fix task_struct reference leak

PATCH Applied Wander Lairson Costa 3 participants 4 messages Jun 20, 2024 – Jul 1, 2024 Scheduler

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 sched/deadline: fix task_struct reference leak

During the execution of the following stress test with linux-rt:

stress-ng --cyclic 30 --timeout 30 --minimize --quiet

kmemleak frequently reported a memory leak concerning the task_struct:

unreferenced object 0xffff8881305b8000 (size 16136):
comm "stress-ng", pid 614, jiffies 4294883961 (age 286.412s) object hex dump (first 32 bytes):

    02 40 00 00 00 00 00 00 00 00 00 00 00 00 00 00  .@..............
    00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
  debug hex dump (first 16 bytes):
    53 09 00 00 00 00 00 00 00 00 00 00 00 00 00 00  S...............
  backtrace:
    [<00000000046b6790>] dup_task_struct+0x30/0x540
    [<00000000c5ca0f0b>] copy_process+0x3d9/0x50e0
    [<00000000ced59777>] kernel_clone+0xb0/0x770
    [<00000000a50befdc>] __do_sys_clone+0xb6/0xf0
    [<000000001dbf2008>] do_syscall_64+0x5d/0xf0
    [<00000000552900ff>] entry_SYSCALL_64_after_hwframe+0x6e/0x76

The issue occurs in start_dl_timer(), which increments the task_struct reference count and sets a timer. The timer callback, dl_task_timer, is supposed to decrement the reference count upon expiration. However, if enqueue_task_dl() is called before the timer expires and cancels it, the reference count is not decremented, leading to the leak.

This patch fixes the reference leak by ensuring the task_struct reference count is properly decremented when the timer is canceled.

---

Changelog:

v2:
* Add the fixes tag
* Add a comment about canceling the timer while the callback was running

Fixes feff2e65efd8 ("sched/deadline: Unthrottle PI boosted threads while enqueuing")
---
 kernel/sched/deadline.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
kernel/sched/deadline.c +6 -1
index c75d1307d86d..9bedd148f007 100644
@@ -1804,8 +1804,13 @@ static void enqueue_task_dl(struct rq *rq, struct task_struct *p, int flags)
 			 * The replenish timer needs to be canceled. No
 			 * problem if it fires concurrently: boosted threads
 			 * are ignored in dl_task_timer().
+			 *
+			 * If the timer callback was running (hrtimer_try_to_cancel == -1),
+			 * it will eventually call put_task_struct().
 			 */
-			hrtimer_try_to_cancel(&p->dl.dl_timer);
+			if (hrtimer_try_to_cancel(&p->dl.dl_timer) == 1 &&
+			    !dl_server(&p->dl))
+				put_task_struct(p);
 			p->dl.dl_throttled = 0;
 		}
 	} else if (!dl_prio(p->normal_prio)) {
Signature · 1 line
2.45.2
maintainer Acked-by on v2 1/1
#1 lore raw
Quoted 30 lines · Wander Lairson Costa

During the execution of the following stress test with linux-rt:

stress-ng --cyclic 30 --timeout 30 --minimize --quiet

kmemleak frequently reported a memory leak concerning the task_struct:

unreferenced object 0xffff8881305b8000 (size 16136):
comm "stress-ng", pid 614, jiffies 4294883961 (age 286.412s) object hex dump (first 32 bytes):

    02 40 00 00 00 00 00 00 00 00 00 00 00 00 00 00  .@..............
    00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
  debug hex dump (first 16 bytes):
    53 09 00 00 00 00 00 00 00 00 00 00 00 00 00 00  S...............
  backtrace:
    [<00000000046b6790>] dup_task_struct+0x30/0x540
    [<00000000c5ca0f0b>] copy_process+0x3d9/0x50e0
    [<00000000ced59777>] kernel_clone+0xb0/0x770
    [<00000000a50befdc>] __do_sys_clone+0xb6/0xf0
    [<000000001dbf2008>] do_syscall_64+0x5d/0xf0
    [<00000000552900ff>] entry_SYSCALL_64_after_hwframe+0x6e/0x76

The issue occurs in start_dl_timer(), which increments the task_struct reference count and sets a timer. The timer callback, dl_task_timer, is supposed to decrement the reference count upon expiration. However, if enqueue_task_dl() is called before the timer expires and cancels it, the reference count is not decremented, leading to the leak.

This patch fixes the reference leak by ensuring the task_struct reference count is properly decremented when the timer is canceled.

Quoted 7 lines

---

Changelog:

v2:
* Add the fixes tag
* Add a comment about canceling the timer while the callback was running

Uh, looks like these bits above should come after the SoB section so that they are ignored when applying the patch? But, maybe, Peter, you can fix that if you decide to take this version?

Quoted 3 lines
Fixes feff2e65efd8 ("sched/deadline: Unthrottle PI boosted threads while enqueuing")

Apart from that

Thanks!
Juri

Quoted 26 lines
---
 kernel/sched/deadline.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
kernel/sched/deadline.c +6 -1
index c75d1307d86d..9bedd148f007 100644
@@ -1804,8 +1804,13 @@ static void enqueue_task_dl(struct rq *rq, struct task_struct *p, int flags)
 			 * The replenish timer needs to be canceled. No
 			 * problem if it fires concurrently: boosted threads
 			 * are ignored in dl_task_timer().
+			 *
+			 * If the timer callback was running (hrtimer_try_to_cancel == -1),
+			 * it will eventually call put_task_struct().
 			 */
-			hrtimer_try_to_cancel(&p->dl.dl_timer);
+			if (hrtimer_try_to_cancel(&p->dl.dl_timer) == 1 &&
+			    !dl_server(&p->dl))
+				put_task_struct(p);
 			p->dl.dl_throttled = 0;
 		}
 	} else if (!dl_prio(p->normal_prio)) {
Signature · 1 line
2.45.2
#2 lore raw
In reply to Wander Lairson Costa
The following commit has been merged into the sched/urgent branch of tip:

The following commit has been merged into the sched/urgent branch of tip:

Commit-ID:     a7accf658efa4fa5b04a74f21863833fd737f469
Gitweb:        https://git.kernel.org/tip/a7accf658efa4fa5b04a74f21863833fd737f469
Author:        Wander Lairson Costa <[email protected]>
AuthorDate:    Thu, 20 Jun 2024 09:56:17 -03:00
Committer:     Peter Zijlstra <[email protected]>
CommitterDate: Tue, 25 Jun 2024 10:43:41 +02:00

sched/deadline: fix task_struct reference leak

During the execution of the following stress test with linux-rt:

stress-ng --cyclic 30 --timeout 30 --minimize --quiet

kmemleak frequently reported a memory leak concerning the task_struct:

unreferenced object 0xffff8881305b8000 (size 16136):
comm "stress-ng", pid 614, jiffies 4294883961 (age 286.412s) object hex dump (first 32 bytes):

    02 40 00 00 00 00 00 00 00 00 00 00 00 00 00 00  .@..............
    00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
  debug hex dump (first 16 bytes):
    53 09 00 00 00 00 00 00 00 00 00 00 00 00 00 00  S...............
  backtrace:
    [<00000000046b6790>] dup_task_struct+0x30/0x540
    [<00000000c5ca0f0b>] copy_process+0x3d9/0x50e0
    [<00000000ced59777>] kernel_clone+0xb0/0x770
    [<00000000a50befdc>] __do_sys_clone+0xb6/0xf0
    [<000000001dbf2008>] do_syscall_64+0x5d/0xf0
    [<00000000552900ff>] entry_SYSCALL_64_after_hwframe+0x6e/0x76

The issue occurs in start_dl_timer(), which increments the task_struct reference count and sets a timer. The timer callback, dl_task_timer, is supposed to decrement the reference count upon expiration. However, if enqueue_task_dl() is called before the timer expires and cancels it, the reference count is not decremented, leading to the leak.

This patch fixes the reference leak by ensuring the task_struct reference count is properly decremented when the timer is canceled.

---
 kernel/sched/deadline.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
kernel/sched/deadline.c +6 -1
index c75d130..9bedd14 100644
@@ -1804,8 +1804,13 @@ static void enqueue_task_dl(struct rq *rq, struct task_struct *p, int flags)
 			 * The replenish timer needs to be canceled. No
 			 * problem if it fires concurrently: boosted threads
 			 * are ignored in dl_task_timer().
+			 *
+			 * If the timer callback was running (hrtimer_try_to_cancel == -1),
+			 * it will eventually call put_task_struct().
 			 */
-			hrtimer_try_to_cancel(&p->dl.dl_timer);
+			if (hrtimer_try_to_cancel(&p->dl.dl_timer) == 1 &&
+			    !dl_server(&p->dl))
+				put_task_struct(p);
 			p->dl.dl_throttled = 0;
 		}
 	} else if (!dl_prio(p->normal_prio)) {
#3 lore raw

[tip: sched/urgent] sched/deadline: Fix task_struct reference leak

In reply to Wander Lairson Costa
The following commit has been merged into the sched/urgent branch of tip:

The following commit has been merged into the sched/urgent branch of tip:

Commit-ID:     b58652db66c910c2245f5bee7deca41c12d707b9
Gitweb:        https://git.kernel.org/tip/b58652db66c910c2245f5bee7deca41c12d707b9
Author:        Wander Lairson Costa <[email protected]>
AuthorDate:    Thu, 20 Jun 2024 09:56:17 -03:00
Committer:     Peter Zijlstra <[email protected]>
CommitterDate: Mon, 01 Jul 2024 13:01:44 +02:00

sched/deadline: Fix task_struct reference leak

During the execution of the following stress test with linux-rt:

stress-ng --cyclic 30 --timeout 30 --minimize --quiet

kmemleak frequently reported a memory leak concerning the task_struct:

unreferenced object 0xffff8881305b8000 (size 16136):
comm "stress-ng", pid 614, jiffies 4294883961 (age 286.412s) object hex dump (first 32 bytes):

    02 40 00 00 00 00 00 00 00 00 00 00 00 00 00 00  .@..............
    00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
  debug hex dump (first 16 bytes):
    53 09 00 00 00 00 00 00 00 00 00 00 00 00 00 00  S...............
  backtrace:
    [<00000000046b6790>] dup_task_struct+0x30/0x540
    [<00000000c5ca0f0b>] copy_process+0x3d9/0x50e0
    [<00000000ced59777>] kernel_clone+0xb0/0x770
    [<00000000a50befdc>] __do_sys_clone+0xb6/0xf0
    [<000000001dbf2008>] do_syscall_64+0x5d/0xf0
    [<00000000552900ff>] entry_SYSCALL_64_after_hwframe+0x6e/0x76

The issue occurs in start_dl_timer(), which increments the task_struct reference count and sets a timer. The timer callback, dl_task_timer, is supposed to decrement the reference count upon expiration. However, if enqueue_task_dl() is called before the timer expires and cancels it, the reference count is not decremented, leading to the leak.

This patch fixes the reference leak by ensuring the task_struct reference count is properly decremented when the timer is canceled.

Fixes feff2e65efd8 ("sched/deadline: Unthrottle PI boosted threads while enqueuing")
---
 kernel/sched/deadline.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
kernel/sched/deadline.c +6 -1
index c75d130..9bedd14 100644
@@ -1804,8 +1804,13 @@ static void enqueue_task_dl(struct rq *rq, struct task_struct *p, int flags)
 			 * The replenish timer needs to be canceled. No
 			 * problem if it fires concurrently: boosted threads
 			 * are ignored in dl_task_timer().
+			 *
+			 * If the timer callback was running (hrtimer_try_to_cancel == -1),
+			 * it will eventually call put_task_struct().
 			 */
-			hrtimer_try_to_cancel(&p->dl.dl_timer);
+			if (hrtimer_try_to_cancel(&p->dl.dl_timer) == 1 &&
+			    !dl_server(&p->dl))
+				put_task_struct(p);
 			p->dl.dl_throttled = 0;
 		}
 	} else if (!dl_prio(p->normal_prio)) {