Commit abfc0107 authored by Peter Zijlstra's avatar Peter Zijlstra
Browse files

sched: Fix do_set_cpus_allowed() locking 



All callers of do_set_cpus_allowed() only take p->pi_lock, which is
not sufficient to actually change the cpumask. Again, this is mostly
ok in these cases, but it results in unnecessarily complicated
reasoning.

Furthermore, there is no reason what so ever to not just take all the
required locks, so do just that.

Signed-off-by: default avatarPeter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: default avatarJuri Lelli <juri.lelli@redhat.com>
Acked-by: default avatarTejun Heo <tj@kernel.org>
Acked-by: default avatarVincent Guittot <vincent.guittot@linaro.org>
parent 942b8db9

kernel/kthread.c

+5 −10
Original line number Diff line number Diff line
@@ -593,18 +593,16 @@ EXPORT_SYMBOL(kthread_create_on_node); 


static void __kthread_bind_mask(struct task_struct *p, const struct cpumask *mask, unsigned int state)

{

	unsigned long flags;



	if (!wait_task_inactive(p, state)) {

		WARN_ON(1);

		return;

	}



	/* It's safe because the task is inactive. */

	raw_spin_lock_irqsave(&p->pi_lock, flags);

	scoped_guard (raw_spinlock_irqsave, &p->pi_lock)

		do_set_cpus_allowed(p, mask);



	/* It's safe because the task is inactive. */

	p->flags |= PF_NO_SETAFFINITY;

	raw_spin_unlock_irqrestore(&p->pi_lock, flags);

}



static void __kthread_bind(struct task_struct *p, unsigned int cpu, unsigned int state)
@@ -857,7 +855,6 @@ int kthread_affine_preferred(struct task_struct *p, const struct cpumask *mask) 
{

	struct kthread *kthread = to_kthread(p);

	cpumask_var_t affinity;

	unsigned long flags;

	int ret = 0;



	if (!wait_task_inactive(p, TASK_UNINTERRUPTIBLE) || kthread->started) {
@@ -882,10 +879,8 @@ int kthread_affine_preferred(struct task_struct *p, const struct cpumask *mask) 
	list_add_tail(&kthread->hotplug_node, &kthreads_hotplug);

	kthread_fetch_affinity(kthread, affinity);



	/* It's safe because the task is inactive. */

	raw_spin_lock_irqsave(&p->pi_lock, flags);

	scoped_guard (raw_spinlock_irqsave, &p->pi_lock)

		do_set_cpus_allowed(p, affinity);

	raw_spin_unlock_irqrestore(&p->pi_lock, flags);



	mutex_unlock(&kthreads_hotplug_lock);

out:

kernel/sched/core.c

+7 −14
Original line number Diff line number Diff line
@@ -2668,18 +2668,14 @@ __do_set_cpus_allowed(struct task_struct *p, struct affinity_context *ctx) 
	bool queued, running;



	lockdep_assert_held(&p->pi_lock);

	lockdep_assert_rq_held(rq);



	queued = task_on_rq_queued(p);

	running = task_current_donor(rq, p);



	if (queued) {

		/*

		 * Because __kthread_bind() calls this on blocked tasks without

		 * holding rq->lock.

		 */

		lockdep_assert_rq_held(rq);

	if (queued)

		dequeue_task(rq, p, DEQUEUE_SAVE | DEQUEUE_NOCLOCK);

	}



	if (running)

		put_prev_task(rq, p);
@@ -2708,7 +2704,10 @@ void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask) 
		struct rcu_head rcu;

	};



	scoped_guard (__task_rq_lock, p) {

		update_rq_clock(scope.rq);

		__do_set_cpus_allowed(p, &ac);

	}



	/*

	 * Because this is called with p->pi_lock held, it is not possible
@@ -3483,12 +3482,6 @@ static int select_fallback_rq(int cpu, struct task_struct *p) 
			}

			fallthrough;

		case possible:

			/*

			 * XXX When called from select_task_rq() we only

			 * hold p->pi_lock and again violate locking order.

			 *

			 * More yuck to audit.

			 */

			do_set_cpus_allowed(p, task_cpu_fallback_mask(p));

			state = fail;

			break;

kernel/sched/sched.h

+5 −0
Original line number Diff line number Diff line
@@ -1847,6 +1847,11 @@ DEFINE_LOCK_GUARD_1(task_rq_lock, struct task_struct, 
		    task_rq_unlock(_T->rq, _T->lock, &_T->rf),

		    struct rq *rq; struct rq_flags rf)



DEFINE_LOCK_GUARD_1(__task_rq_lock, struct task_struct,

		    _T->rq = __task_rq_lock(_T->lock, &_T->rf),

		    __task_rq_unlock(_T->rq, &_T->rf),

		    struct rq *rq; struct rq_flags rf)



static inline void rq_lock_irqsave(struct rq *rq, struct rq_flags *rf)

	__acquires(rq->lock)

{