The Linux Kernel Mailing List
 help / color / mirror / Atom feed
* [PATCH v2] sched/isolation: Enable cpuset_mutex/cpus_read_lock synchronization before freeing old housekeeping cpumask
@ 2026-11-17  5:18 Waiman Long
  2026-11-17  6:43 ` Guopeng Zhang
  2026-11-17  8:34 ` Ridong Chen
  0 siblings, 2 replies; 4+ messages in thread
From: Waiman Long @ 2026-11-17  5:18 UTC (permalink / raw)
  To: Ingo Molnar, Peter Zijlstra, Juri Lelli, Vincent Guittot,
	Dietmar Eggemann, Steven Rostedt, Ben Segall, Mel Gorman,
	Valentin Schneider, K Prateek Nayak, Tejun Heo,
	Frederic Weisbecker
  Cc: linux-kernel, Guopeng Zhang, Ridong Chen, Mathieu Desnoyers,
	Waiman Long

Commit a84097e625f2 ("cgroup/cpuset: Call housekeeping_update() without
holding cpus_read_lock") adds a new cpuset_top_mutex on top but drops
cpus_read_lock and cpuset_mutex before calling housekeeping_update()
which calls other functions that may acquire cpus_read_lock leading
to a possibility of deadlock if cpus_read_lock isn't dropped before
calling housekeeping_update().

This commit also changes lockdep_is_cpuset_held() to return true if
either cpuset_mutex or cpuset_top_mutex is held. lockdep_is_cpuset_held()
is used by housekeeping_dereference_check() as one of the conditions
that rcu_read_lock() is not needed to access HK_TYPE_DOMAIN
cpumask which can be changed by housekeeping_update() at run
time. The housekeeping_dereference_check() function also uses
lockdep_is_cpus_write_held() as another condition that can skip the
use of rcu_read_lock().

In reality, both cpuset_mutex and cpus_write_lock are now no longer
valid conditions to skip rcu_read_lock(). It is certainly possible
that there can be a parallel housekeeping_update() call going on if
only cpuset_mutex or cpus_write_lock is held. This can lead to a
potential use-after-free situation when HK_TYPE_DOMAIN housekeeping
cpumask is being accessed.

One possible solution is to hold rcu_read_lock for functions that may
be called with only cpuset_mutex or cpus_write_lock is held. Quite a
large number of functions will need to be modified if rcu_read_lock()
has to be added. A far simpler solution is to synchronize with other
cpuset_mutex and cpus_write_lock critical sections before freeing the old
housekeeping cpumask. This will ensure that code within those critical
sections will not have a UaF situation when accessing housekeeping
cpumask. This commit implements the latter solution and make the code
in lockdep_is_cpuset_held() and housekeeping_dereference_check() correct.

Fixes: a84097e625f2 ("cgroup/cpuset: Call housekeeping_update() without holding cpus_read_lock")
Signed-off-by: Waiman Long <longman@redhat.com>
---
 kernel/sched/isolation.c | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/kernel/sched/isolation.c b/kernel/sched/isolation.c
index 156025ef81b7..5634d6a597bd 100644
--- a/kernel/sched/isolation.c
+++ b/kernel/sched/isolation.c
@@ -159,6 +159,16 @@ int housekeeping_update(struct cpumask *isol_mask)
 	err = kthreads_update_housekeeping();
 	WARN_ON_ONCE(err < 0);
 
+	/*
+	 * Synchronize with cpuset_mutex and cpus_read_lock to make sure that
+	 * cpuset_mutex and cpus_write_lock critical sections afterward will
+	 * not have access to the old housekeeping cpumask.
+	 */
+	cpus_read_lock();
+	cpuset_lock();
+	cpuset_unlock();
+	cpus_read_unlock();
+
 	kfree(old);
 
 	return 0;
-- 
2.56.0


^ permalink raw reply related	[flat|nested] 4+ messages in thread

* Re: [PATCH v2] sched/isolation: Enable cpuset_mutex/cpus_read_lock synchronization before freeing old housekeeping cpumask
  2026-11-17  5:18 [PATCH v2] sched/isolation: Enable cpuset_mutex/cpus_read_lock synchronization before freeing old housekeeping cpumask Waiman Long
@ 2026-11-17  6:43 ` Guopeng Zhang
  2026-11-17  8:34 ` Ridong Chen
  1 sibling, 0 replies; 4+ messages in thread
From: Guopeng Zhang @ 2026-11-17  6:43 UTC (permalink / raw)
  To: Waiman Long, Ingo Molnar, Peter Zijlstra, Juri Lelli,
	Vincent Guittot, Dietmar Eggemann, Steven Rostedt, Ben Segall,
	Mel Gorman, Valentin Schneider, K Prateek Nayak, Tejun Heo,
	Frederic Weisbecker
  Cc: linux-kernel, Ridong Chen, Mathieu Desnoyers



在 2026/11/17 13:18, Waiman Long 写道:
> Commit a84097e625f2 ("cgroup/cpuset: Call housekeeping_update() without
> holding cpus_read_lock") adds a new cpuset_top_mutex on top but drops
> cpus_read_lock and cpuset_mutex before calling housekeeping_update()
> which calls other functions that may acquire cpus_read_lock leading
> to a possibility of deadlock if cpus_read_lock isn't dropped before
> calling housekeeping_update().
> 
> This commit also changes lockdep_is_cpuset_held() to return true if
> either cpuset_mutex or cpuset_top_mutex is held. lockdep_is_cpuset_held()
> is used by housekeeping_dereference_check() as one of the conditions
> that rcu_read_lock() is not needed to access HK_TYPE_DOMAIN
> cpumask which can be changed by housekeeping_update() at run
> time. The housekeeping_dereference_check() function also uses
> lockdep_is_cpus_write_held() as another condition that can skip the
> use of rcu_read_lock().
> 
> In reality, both cpuset_mutex and cpus_write_lock are now no longer
> valid conditions to skip rcu_read_lock(). It is certainly possible
> that there can be a parallel housekeeping_update() call going on if
> only cpuset_mutex or cpus_write_lock is held. This can lead to a
> potential use-after-free situation when HK_TYPE_DOMAIN housekeeping
> cpumask is being accessed.
> 
> One possible solution is to hold rcu_read_lock for functions that may
> be called with only cpuset_mutex or cpus_write_lock is held. Quite a
> large number of functions will need to be modified if rcu_read_lock()
> has to be added. A far simpler solution is to synchronize with other
> cpuset_mutex and cpus_write_lock critical sections before freeing the old
> housekeeping cpumask. This will ensure that code within those critical
> sections will not have a UaF situation when accessing housekeeping
> cpumask. This commit implements the latter solution and make the code
> in lockdep_is_cpuset_held() and housekeeping_dereference_check() correct.
> 
> Fixes: a84097e625f2 ("cgroup/cpuset: Call housekeeping_update() without holding cpus_read_lock")
> Signed-off-by: Waiman Long <longman@redhat.com>
> ---
>  kernel/sched/isolation.c | 10 ++++++++++
>  1 file changed, 10 insertions(+)
> 
> diff --git a/kernel/sched/isolation.c b/kernel/sched/isolation.c
> index 156025ef81b7..5634d6a597bd 100644
> --- a/kernel/sched/isolation.c
> +++ b/kernel/sched/isolation.c
> @@ -159,6 +159,16 @@ int housekeeping_update(struct cpumask *isol_mask)
>  	err = kthreads_update_housekeeping();
>  	WARN_ON_ONCE(err < 0);
>  
> +	/*
> +	 * Synchronize with cpuset_mutex and cpus_read_lock to make sure that
> +	 * cpuset_mutex and cpus_write_lock critical sections afterward will
> +	 * not have access to the old housekeeping cpumask.
> +	 */
> +	cpus_read_lock();
> +	cpuset_lock();
> +	cpuset_unlock();
> +	cpus_read_unlock();
> +
>  	kfree(old);
>  
>  	return 0;

Hi Waiman,

This looks good to me.

Nit: Could the comments in housekeeping_dereference_check() be updated
as well? They seem to be outdated after the locking changes.

Acked-by: Guopeng Zhang <zhangguopeng@kylinos.cn>

Thanks,
Guopeng

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH v2] sched/isolation: Enable cpuset_mutex/cpus_read_lock synchronization before freeing old housekeeping cpumask
  2026-11-17  5:18 [PATCH v2] sched/isolation: Enable cpuset_mutex/cpus_read_lock synchronization before freeing old housekeeping cpumask Waiman Long
  2026-11-17  6:43 ` Guopeng Zhang
@ 2026-11-17  8:34 ` Ridong Chen
  2026-11-17  8:39   ` [PATCH] sched/isolation: Publish new HK_TYPE_DOMAIN cpumask under cpuset_mutex Ridong
  1 sibling, 1 reply; 4+ messages in thread
From: Ridong Chen @ 2026-11-17  8:34 UTC (permalink / raw)
  To: Waiman Long, Ingo Molnar, Peter Zijlstra, Juri Lelli,
	Vincent Guittot, Dietmar Eggemann, Steven Rostedt, Ben Segall,
	Mel Gorman, Valentin Schneider, K Prateek Nayak, Tejun Heo,
	Frederic Weisbecker
  Cc: linux-kernel, Guopeng Zhang, Mathieu Desnoyers



On 11/17/2026 1:18 PM, Waiman Long wrote:
> Commit a84097e625f2 ("cgroup/cpuset: Call housekeeping_update() without
> holding cpus_read_lock") adds a new cpuset_top_mutex on top but drops
> cpus_read_lock and cpuset_mutex before calling housekeeping_update()
> which calls other functions that may acquire cpus_read_lock leading
> to a possibility of deadlock if cpus_read_lock isn't dropped before
> calling housekeeping_update().
> 
> This commit also changes lockdep_is_cpuset_held() to return true if
> either cpuset_mutex or cpuset_top_mutex is held. lockdep_is_cpuset_held()
> is used by housekeeping_dereference_check() as one of the conditions
> that rcu_read_lock() is not needed to access HK_TYPE_DOMAIN
> cpumask which can be changed by housekeeping_update() at run
> time. The housekeeping_dereference_check() function also uses
> lockdep_is_cpus_write_held() as another condition that can skip the
> use of rcu_read_lock().
> 
> In reality, both cpuset_mutex and cpus_write_lock are now no longer
> valid conditions to skip rcu_read_lock(). It is certainly possible
> that there can be a parallel housekeeping_update() call going on if
> only cpuset_mutex or cpus_write_lock is held. This can lead to a
> potential use-after-free situation when HK_TYPE_DOMAIN housekeeping
> cpumask is being accessed.
> 
> One possible solution is to hold rcu_read_lock for functions that may
> be called with only cpuset_mutex or cpus_write_lock is held. Quite a
> large number of functions will need to be modified if rcu_read_lock()
> has to be added. A far simpler solution is to synchronize with other
> cpuset_mutex and cpus_write_lock critical sections before freeing the old
> housekeeping cpumask. This will ensure that code within those critical
> sections will not have a UaF situation when accessing housekeeping
> cpumask. This commit implements the latter solution and make the code
> in lockdep_is_cpuset_held() and housekeeping_dereference_check() correct.
> 
> Fixes: a84097e625f2 ("cgroup/cpuset: Call housekeeping_update() without holding cpus_read_lock")
> Signed-off-by: Waiman Long <longman@redhat.com>
> ---
>   kernel/sched/isolation.c | 10 ++++++++++
>   1 file changed, 10 insertions(+)
> 
> diff --git a/kernel/sched/isolation.c b/kernel/sched/isolation.c
> index 156025ef81b7..5634d6a597bd 100644
> --- a/kernel/sched/isolation.c
> +++ b/kernel/sched/isolation.c
> @@ -159,6 +159,16 @@ int housekeeping_update(struct cpumask *isol_mask)
>   	err = kthreads_update_housekeeping();
>   	WARN_ON_ONCE(err < 0);
>   
> +	/*
> +	 * Synchronize with cpuset_mutex and cpus_read_lock to make sure that
> +	 * cpuset_mutex and cpus_write_lock critical sections afterward will
> +	 * not have access to the old housekeeping cpumask.
> +	 */
> +	cpus_read_lock();
> +	cpuset_lock();
> +	cpuset_unlock();
> +	cpus_read_unlock();
> +
>   	kfree(old);
>   
>   	return 0;

Hi Waiman,

Taking the locks and releasing them right away looks a bit odd to me.
It is not obvious what the empty critical section protects until one
reads the commit message.

Let me check that I understand the problem correctly.
housekeeping_dereference_check() lets readers holding cpus_write_lock
or cpuset_mutex access HK_TYPE_DOMAIN without rcu_read_lock(). After
commit a84097e625f2, the pointer is switched and the old cpumask is
freed after cpuset_mutex and cpus_read_lock have been dropped, and
synchronize_rcu() does not wait for those readers. For example:

   CPU A                                  CPU B
   rebuild_sched_domains()                cpuset_update_sd_hk_unlock()
                                            mutex_unlock(&cpuset_mutex)
                                            cpus_read_unlock()
     mutex_lock(&cpuset_mutex)
     p = housekeeping_cpumask(DOMAIN)
                                            housekeeping_update()
                                              rcu_assign_pointer()
                                              synchronize_rcu()
                                              kfree(old)
     access p <- UAF

Here CPU A reads the mask in asym_cpu_capacity_scan() or
dl_get_task_effective_cpus().

If that is right, the root cause is that the pointer is now switched
outside of those locks. So instead of adding a lock barrier before
kfree(), how about switching the pointer before the locks are
dropped? housekeeping_update() could be split into two parts:

  - housekeeping_publish(): allocate, check and publish the new mask,
    called with cpus_read_lock and cpuset_mutex held.

  - housekeeping_update(): synchronize_rcu(), the workqueue flushes and
    kfree(old), called after the locks are dropped as it is now.

Then a reader holding cpuset_mutex either finishes before the pointer
is switched, or only sees the new mask:

   CPU A                                  CPU B
                                          cpuset_update_sd_hk_unlock()
                                            housekeeping_publish()
                                              rcu_assign_pointer()
     mutex_lock(&cpuset_mutex) <- wait
                                            mutex_unlock(&cpuset_mutex)
                                            cpus_read_unlock()
     p = housekeeping_cpumask(DOMAIN)       housekeeping_update()
         <- new cpumask                       synchronize_rcu()
                                              kfree(old)

The same applies to cpus_write_lock readers, as cpus_read_lock is held
until the new mask is published. The workqueue flushes still run
without those locks, so the deadlock fixed by a84097e625f2 does not come
back.

I will send a patch in reply to this mail. Could you take a look and
tell me whether this approach works for you?

-- 
Best regards
Ridong


^ permalink raw reply	[flat|nested] 4+ messages in thread

* [PATCH] sched/isolation: Publish new HK_TYPE_DOMAIN cpumask under cpuset_mutex
  2026-11-17  8:34 ` Ridong Chen
@ 2026-11-17  8:39   ` Ridong
  0 siblings, 0 replies; 4+ messages in thread
From: Ridong @ 2026-11-17  8:39 UTC (permalink / raw)
  To: longman, frederic, tj, peterz, mingo, juri.lelli, vincent.guittot,
	dietmar.eggemann, rostedt, bsegall, mgorman, vschneid,
	kprateek.nayak, hannes, mkoutny
  Cc: linux-kernel, cgroups, guopeng.zhang, mathieu.desnoyers,
	Ridong Chen

From: Ridong Chen <chenridong@xiaomi.com>

Commit a84097e625f2 ("cgroup/cpuset: Call housekeeping_update() without
holding cpus_read_lock") calls housekeeping_update() after dropping
cpus_read_lock and cpuset_mutex, so the HK_TYPE_DOMAIN cpumask is now
switched and freed outside of those locks. But readers holding either
lock may still access it without rcu_read_lock() as allowed by
housekeeping_dereference_check(), and synchronize_rcu() doesn't wait
for them:

  CPU A                                  CPU B
  rebuild_sched_domains()                cpuset_update_sd_hk_unlock()
                                           mutex_unlock(&cpuset_mutex)
                                           cpus_read_unlock()
    mutex_lock(&cpuset_mutex)
    p = housekeeping_cpumask(DOMAIN)
                                           housekeeping_update()
                                             rcu_assign_pointer()
                                             synchronize_rcu()
                                             kfree(old)
    access p <- UAF

Fix it by moving the pointer switch into housekeeping_publish(), which
is called before dropping the locks. Then CPU A either finishes before
CPU B switches the pointer, or only sees the new cpumask:

  CPU A                                  CPU B
                                         cpuset_update_sd_hk_unlock()
                                           housekeeping_publish()
                                             rcu_assign_pointer()
    mutex_lock(&cpuset_mutex) <- wait
                                           mutex_unlock(&cpuset_mutex)
                                           cpus_read_unlock()
    p = housekeeping_cpumask(DOMAIN)       housekeeping_update()
        <- new cpumask                       synchronize_rcu()
                                             kfree(old)

The rest of housekeeping_update(), including the workqueue flushes,
still runs without those locks.

Fixes: a84097e625f2 ("cgroup/cpuset: Call housekeeping_update() without holding cpus_read_lock")
Signed-off-by: Ridong Chen <chenridong@xiaomi.com>
---
 include/linux/sched/isolation.h |  6 ++++--
 kernel/cgroup/cpuset.c          | 12 ++++++++----
 kernel/sched/isolation.c        | 23 ++++++++++++++++-------
 3 files changed, 28 insertions(+), 13 deletions(-)

diff --git a/include/linux/sched/isolation.h b/include/linux/sched/isolation.h
index cf0fd03dd7a24..09fabf42e3224 100644
--- a/include/linux/sched/isolation.h
+++ b/include/linux/sched/isolation.h
@@ -43,7 +43,8 @@ extern const struct cpumask *housekeeping_cpumask(enum hk_type type);
 extern bool housekeeping_enabled(enum hk_type type);
 extern void housekeeping_affine(struct task_struct *t, enum hk_type type);
 extern bool housekeeping_test_cpu(int cpu, enum hk_type type);
-extern int housekeeping_update(struct cpumask *isol_mask);
+extern struct cpumask *housekeeping_publish(struct cpumask *isol_mask);
+extern void housekeeping_update(struct cpumask *isol_mask, struct cpumask *old);
 extern void __init housekeeping_init(void);
 
 #else
@@ -71,7 +72,8 @@ static inline bool housekeeping_test_cpu(int cpu, enum hk_type type)
 	return true;
 }
 
-static inline int housekeeping_update(struct cpumask *isol_mask) { return 0; }
+static inline struct cpumask *housekeeping_publish(struct cpumask *isol_mask) { return NULL; }
+static inline void housekeeping_update(struct cpumask *isol_mask, struct cpumask *old) { }
 static inline void housekeeping_init(void) { }
 #endif /* CONFIG_CPU_ISOLATION */
 
diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c
index 8a8a424a389f3..1673bb25a797b 100644
--- a/kernel/cgroup/cpuset.c
+++ b/kernel/cgroup/cpuset.c
@@ -1472,17 +1472,21 @@ static void cpuset_update_sd_hk_unlock(void)
 		rebuild_sched_domains_locked();
 
 	if (update_housekeeping) {
+		struct cpumask *old;
+
 		update_housekeeping = false;
 		cpumask_copy(isolated_hk_cpus, isolated_cpus);
 
 		/*
-		 * housekeeping_update() is now called without holding
-		 * cpus_read_lock and cpuset_mutex. Only cpuset_top_mutex
-		 * is still being held for mutual exclusion.
+		 * Publish the new cpumask with cpus_read_lock and cpuset_mutex
+		 * held, but call housekeeping_update() without them. Only
+		 * cpuset_top_mutex is still being held for mutual exclusion.
 		 */
+		old = housekeeping_publish(isolated_hk_cpus);
 		mutex_unlock(&cpuset_mutex);
 		cpus_read_unlock();
-		WARN_ON_ONCE(housekeeping_update(isolated_hk_cpus));
+		if (!WARN_ON_ONCE(IS_ERR(old)))
+			housekeeping_update(isolated_hk_cpus, old);
 		mutex_unlock(&cpuset_top_mutex);
 	} else {
 		cpuset_full_unlock();
diff --git a/kernel/sched/isolation.c b/kernel/sched/isolation.c
index 156025ef81b75..9e202269e42f6 100644
--- a/kernel/sched/isolation.c
+++ b/kernel/sched/isolation.c
@@ -120,23 +120,26 @@ bool housekeeping_test_cpu(int cpu, enum hk_type type)
 }
 EXPORT_SYMBOL_GPL(housekeeping_test_cpu);
 
-int housekeeping_update(struct cpumask *isol_mask)
+/* Called with cpus_read_lock and cpuset_mutex held */
+struct cpumask *housekeeping_publish(struct cpumask *isol_mask)
 {
 	struct cpumask *trial, *old = NULL;
-	int err;
+
+	lockdep_assert_cpus_held();
+	lockdep_assert_cpuset_lock_held();
 
 	trial = kmalloc(cpumask_size(), GFP_KERNEL);
 	if (!trial)
-		return -ENOMEM;
+		return ERR_PTR(-ENOMEM);
 
 	cpumask_andnot(trial, housekeeping_cpumask(HK_TYPE_DOMAIN_BOOT), isol_mask);
 	if (!cpumask_intersects(trial, cpu_online_mask)) {
 		kfree(trial);
-		return -EINVAL;
+		return ERR_PTR(-EINVAL);
 	}
 
 	if (!housekeeping.flags)
-		static_branch_enable(&housekeeping_overridden);
+		static_branch_enable_cpuslocked(&housekeeping_overridden);
 
 	if (housekeeping.flags & HK_FLAG_DOMAIN)
 		old = housekeeping_cpumask_dereference(HK_TYPE_DOMAIN);
@@ -144,6 +147,14 @@ int housekeeping_update(struct cpumask *isol_mask)
 		WRITE_ONCE(housekeeping.flags, housekeeping.flags | HK_FLAG_DOMAIN);
 	rcu_assign_pointer(housekeeping.cpumasks[HK_TYPE_DOMAIN], trial);
 
+	return old;
+}
+
+/* Called without cpus_read_lock and cpuset_mutex held */
+void housekeeping_update(struct cpumask *isol_mask, struct cpumask *old)
+{
+	int err;
+
 	synchronize_rcu();
 
 	pci_probe_flush_workqueue();
@@ -160,8 +171,6 @@ int housekeeping_update(struct cpumask *isol_mask)
 	WARN_ON_ONCE(err < 0);
 
 	kfree(old);
-
-	return 0;
 }
 
 void __init housekeeping_init(void)
-- 
2.34.1


^ permalink raw reply related	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2026-11-17  8:39 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-11-17  5:18 [PATCH v2] sched/isolation: Enable cpuset_mutex/cpus_read_lock synchronization before freeing old housekeeping cpumask Waiman Long
2026-11-17  6:43 ` Guopeng Zhang
2026-11-17  8:34 ` Ridong Chen
2026-11-17  8:39   ` [PATCH] sched/isolation: Publish new HK_TYPE_DOMAIN cpumask under cpuset_mutex Ridong

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox