Message-ID: <20260124070814.806828-1-realwujing@gmail.com>
Date: Sat, 24 Jan 2026 02:08:14 -0500
From: Qiliang Yuan <realwujing@...il.com>
To: dianders@...omium.org
Cc: akpm@...ux-foundation.org,
	lihuafei1@...wei.com,
	linux-kernel@...r.kernel.org,
	mingo@...nel.org,
	mm-commits@...r.kernel.org,
	realwujing@...il.com,
	song@...nel.org,
	stable@...r.kernel.org,
	sunshx@...natelecom.cn,
	thorsten.blum@...ux.dev,
	wangjinchao600@...il.com,
	yangyicong@...ilicon.com,
	yuanql9@...natelecom.cn,
	zhangjn11@...natelecom.cn,
	linux-watchdog@...r.kernel.org
Subject: [PATCH v4] watchdog/hardlockup: Fix UAF in perf event cleanup due to migration race

Original analysis on Linux 4.19 showed a race condition in the hardlockup
detector's initialization phase. Specifically, during the early probe
phase, hardlockup_detector_perf_init() (renamed to
watchdog_hardlockup_probe() in newer kernels via commit d9b3629ade8e)
interacted with the per-cpu 'watchdog_ev' variable.

If the initializing task migrates to another CPU during this probe phase,
two issues arise:
1. The 'watchdog_ev' pointer on the original CPU is set but not cleared,
   leaving a stale pointer to a freed perf event.
2. The 'watchdog_ev' pointer on the new CPU might be incorrectly cleared.

Note: Although the logs below reference hardlockup_detector_perf_init(),
the same logic persists in the current watchdog_hardlockup_probe()
implementation.

This race condition was observed in console logs:
[23.038376] hardlockup_detector_perf_init 313 cur_cpu=2
...
[23.076385] hardlockup_detector_event_create 203 cpu(cur)=2 set watchdog_ev
...
[23.095788] perf_event_release_kernel 4623 cur_cpu=2
...
[23.116963] lockup_detector_reconfigure 577 cur_cpu=3

The log shows the task started on CPU 2, set watchdog_ev on CPU 2,
released the event on CPU 2, but then migrated to CPU 3 before the
cleanup logic could run. This left watchdog_ev on CPU 2 pointing to a
freed event, resulting in a UAF when later accessed:

[26.540732] BUG: KASAN: use-after-free in perf_event_ctx_lock_nested.isra.72+0x6b/0x140
[26.542442] Read of size 8 at addr ff110006b360d718 by task kworker/2:1/94

Fix this by refactoring hardlockup_detector_event_create() to return the
created perf event instead of directly assigning it to the per-cpu variable.
In the probe function, use an arbitrary CPU but ensure it remains
online via cpu_hotplug_disable() during the check.

Fixes: 930d8f8dbab9 ("watchdog/perf: adapt the watchdog_perf interface for async model")
Signed-off-by: Shouxin Sun <sunshx@...natelecom.cn>
Signed-off-by: Junnan Zhang <zhangjn11@...natelecom.cn>
Signed-off-by: Qiliang Yuan <realwujing@...il.com>
Signed-off-by: Qiliang Yuan <yuanql9@...natelecom.cn>
Cc: Song Liu <song@...nel.org>
Cc: Douglas Anderson <dianders@...omium.org>
Cc: Jinchao Wang <wangjinchao600@...il.com>
Cc: Wang Jinchao <wangjinchao600@...il.com>
Cc: <stable@...r.kernel.org>
---
v4:
- Add cpu_hotplug_disable() in watchdog_hardlockup_probe() to ensure the
  sampled CPU remains online during probing. 
- Update commit message to explain the relevance of 4.19 logs even
  though functions were renamed in modern kernels. 
v3:
- Refactor hardlockup_detector_event_create() to return the event pointer
  instead of directly assigning to per-cpu variables to fix the UAF.
- Restore PMU cycle fallback and unify the enable/probe paths.
v2:
- Add Cc: <stable@...r.kernel.org>.
v1:
- Avoid 'watchdog_ev' in probe path by manually creating and releasing a
  local perf event.
 kernel/watchdog_perf.c | 56 +++++++++++++++++++++++++-----------------
 1 file changed, 34 insertions(+), 22 deletions(-)

diff --git a/kernel/watchdog_perf.c b/kernel/watchdog_perf.c
index d3ca70e3c256..887b61c65c1b 100644
--- a/kernel/watchdog_perf.c
+++ b/kernel/watchdog_perf.c
@@ -17,6 +17,7 @@
 #include <linux/atomic.h>
 #include <linux/module.h>
 #include <linux/sched/debug.h>
+#include <linux/cpu.h>
 
 #include <asm/irq_regs.h>
 #include <linux/perf_event.h>
@@ -118,18 +119,11 @@ static void watchdog_overflow_callback(struct perf_event *event,
 	watchdog_hardlockup_check(smp_processor_id(), regs);
 }
 
-static int hardlockup_detector_event_create(void)
+static struct perf_event *hardlockup_detector_event_create(unsigned int cpu)
 {
-	unsigned int cpu;
 	struct perf_event_attr *wd_attr;
 	struct perf_event *evt;
 
-	/*
-	 * Preemption is not disabled because memory will be allocated.
-	 * Ensure CPU-locality by calling this in per-CPU kthread.
-	 */
-	WARN_ON(!is_percpu_thread());
-	cpu = raw_smp_processor_id();
 	wd_attr = &wd_hw_attr;
 	wd_attr->sample_period = hw_nmi_get_sample_period(watchdog_thresh);
 
@@ -143,14 +137,7 @@ static int hardlockup_detector_event_create(void)
 						       watchdog_overflow_callback, NULL);
 	}
 
-	if (IS_ERR(evt)) {
-		pr_debug("Perf event create on CPU %d failed with %ld\n", cpu,
-			 PTR_ERR(evt));
-		return PTR_ERR(evt);
-	}
-	WARN_ONCE(this_cpu_read(watchdog_ev), "unexpected watchdog_ev leak");
-	this_cpu_write(watchdog_ev, evt);
-	return 0;
+	return evt;
 }
 
 /**
@@ -159,17 +146,26 @@ static int hardlockup_detector_event_create(void)
  */
 void watchdog_hardlockup_enable(unsigned int cpu)
 {
+	struct perf_event *evt;
+
 	WARN_ON_ONCE(cpu != smp_processor_id());
 
-	if (hardlockup_detector_event_create())
+	evt = hardlockup_detector_event_create(cpu);
+	if (IS_ERR(evt)) {
+		pr_debug("Perf event create on CPU %d failed with %ld\n", cpu,
+			 PTR_ERR(evt));
 		return;
+	}
 
 	/* use original value for check */
 	if (!atomic_fetch_inc(&watchdog_cpus))
 		pr_info("Enabled. Permanently consumes one hw-PMU counter.\n");
 
+	WARN_ONCE(this_cpu_read(watchdog_ev), "unexpected watchdog_ev leak");
+	this_cpu_write(watchdog_ev, evt);
+
 	watchdog_init_timestamp();
-	perf_event_enable(this_cpu_read(watchdog_ev));
+	perf_event_enable(evt);
 }
 
 /**
@@ -263,19 +259,35 @@ bool __weak __init arch_perf_nmi_is_available(void)
  */
 int __init watchdog_hardlockup_probe(void)
 {
+	struct perf_event *evt;
+	unsigned int cpu;
 	int ret;
 
 	if (!arch_perf_nmi_is_available())
 		return -ENODEV;
 
-	ret = hardlockup_detector_event_create();
+	if (!hw_nmi_get_sample_period(watchdog_thresh))
+		return -EINVAL;
 
-	if (ret) {
+	/*
+	 * Test hardware PMU availability by creating a temporary perf event.
+	 * The requested CPU is arbitrary; preemption is not disabled, so
+	 * raw_smp_processor_id() is used. Surround with cpu_hotplug_disable()
+	 * to ensure the arbitrarily chosen CPU remains online during the check.
+	 * The event is released immediately.
+	 */
+	cpu_hotplug_disable();
+	cpu = raw_smp_processor_id();
+	evt = hardlockup_detector_event_create(cpu);
+	if (IS_ERR(evt)) {
 		pr_info("Perf NMI watchdog permanently disabled\n");
+		ret = PTR_ERR(evt);
 	} else {
-		perf_event_release_kernel(this_cpu_read(watchdog_ev));
-		this_cpu_write(watchdog_ev, NULL);
+		perf_event_release_kernel(evt);
+		ret = 0;
 	}
+	cpu_hotplug_enable();
+
 	return ret;
 }
 
-- 
2.51.0

Powered by blists - more mailing lists