The Linux Kernel Mailing List
 help / color / mirror / Atom feed
From: Ridong Chen <ridong.chen@linux.dev>
To: Waiman Long <longman@redhat.com>,
	frederic@kernel.org, tj@kernel.org, peterz@infradead.org,
	mingo@redhat.com, juri.lelli@redhat.com,
	vincent.guittot@linaro.org, dietmar.eggemann@arm.com,
	rostedt@goodmis.org, bsegall@google.com, mgorman@suse.de,
	vschneid@redhat.com, kprateek.nayak@amd.com, hannes@cmpxchg.org,
	mkoutny@suse.com
Cc: linux-kernel@vger.kernel.org, cgroups@vger.kernel.org,
	guopeng.zhang@linux.dev, mathieu.desnoyers@efficios.com,
	Ridong Chen <chenridong@xiaomi.com>
Subject: Re: [PATCH] sched/isolation: Publish new HK_TYPE_DOMAIN cpumask under cpuset_mutex
Date: Wed, 18 Nov 2026 08:03:35 +0800	[thread overview]
Message-ID: <4d3e46fb-a374-4a9b-a47e-1c661b7f0f37@linux.dev> (raw)
In-Reply-To: <36d49371-31de-4305-89c4-2ce64dd6848c@redhat.com>



On 11/18/2026 12:15 AM, Waiman Long wrote:
> On 11/17/26 3:39 AM, Ridong wrote:
>> 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)
> 
> Yes, that should work too.
> 
> Acked-by: Waiman Long <longman@redhat.com>
> 
> I will let other decide which one do they want.
> 
Thanks.

-- 
Best regards
Ridong


  reply	other threads:[~2026-11-18  0:03 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
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
2026-11-17 16:15     ` Waiman Long
2026-11-18  0:03       ` Ridong Chen [this message]
2026-11-17 21:18     ` Tejun Heo
2026-11-17 16:02   ` [PATCH v2] sched/isolation: Enable cpuset_mutex/cpus_read_lock synchronization before freeing old housekeeping cpumask Waiman Long

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=4d3e46fb-a374-4a9b-a47e-1c661b7f0f37@linux.dev \
    --to=ridong.chen@linux.dev \
    --cc=bsegall@google.com \
    --cc=cgroups@vger.kernel.org \
    --cc=chenridong@xiaomi.com \
    --cc=dietmar.eggemann@arm.com \
    --cc=frederic@kernel.org \
    --cc=guopeng.zhang@linux.dev \
    --cc=hannes@cmpxchg.org \
    --cc=juri.lelli@redhat.com \
    --cc=kprateek.nayak@amd.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=longman@redhat.com \
    --cc=mathieu.desnoyers@efficios.com \
    --cc=mgorman@suse.de \
    --cc=mingo@redhat.com \
    --cc=mkoutny@suse.com \
    --cc=peterz@infradead.org \
    --cc=rostedt@goodmis.org \
    --cc=tj@kernel.org \
    --cc=vincent.guittot@linaro.org \
    --cc=vschneid@redhat.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox