Merge tag 'sched-urgent-2023-01-12' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull scheduler fixes from Ingo Molnar:

 - Fix scheduler frequency invariance bug related to overly long
   tickless periods triggering an integer overflow and disabling the
   feature.

 - Fix use-after-free bug in dup_user_cpus_ptr().

 - Fix do_set_cpus_allowed() deadlock scenarios related to calling
   kfree() with the pi_lock held. NOTE: the rcu_free() is the 'lazy'
   solution here - we looked at patches to free the structure after the
   pi_lock got dropped, but that looked quite a bit messier - and none
   of this is truly performance critical. We can revisit this if it's
   too lazy of a solution ...

* tag 'sched-urgent-2023-01-12' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  sched/core: Use kfree_rcu() in do_set_cpus_allowed()
  sched/core: Fix use-after-free bug in dup_user_cpus_ptr()
  sched/core: Fix arch_scale_freq_tick() on tickless systems

1 files changed, 60 insertions, 9 deletions

diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 25b582b6ee5f..bb1ee6d7bdde 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -2604,27 +2604,71 @@ void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask)
 		.user_mask = NULL,
 		.flags     = SCA_USER,	/* clear the user requested mask */
 	};
+	union cpumask_rcuhead {
+		cpumask_t cpumask;
+		struct rcu_head rcu;
+	};
 
 	__do_set_cpus_allowed(p, &ac);
-	kfree(ac.user_mask);
+
+	/*
+	 * Because this is called with p->pi_lock held, it is not possible
+	 * to use kfree() here (when PREEMPT_RT=y), therefore punt to using
+	 * kfree_rcu().
+	 */
+	kfree_rcu((union cpumask_rcuhead *)ac.user_mask, rcu);
+}
+
+static cpumask_t *alloc_user_cpus_ptr(int node)
+{
+	/*
+	 * See do_set_cpus_allowed() above for the rcu_head usage.
+	 */
+	int size = max_t(int, cpumask_size(), sizeof(struct rcu_head));
+
+	return kmalloc_node(size, GFP_KERNEL, node);
 }
 
 int dup_user_cpus_ptr(struct task_struct *dst, struct task_struct *src,
 		      int node)
 {
+	cpumask_t *user_mask;
 	unsigned long flags;
 
-	if (!src->user_cpus_ptr)
+	/*
+	 * Always clear dst->user_cpus_ptr first as their user_cpus_ptr's
+	 * may differ by now due to racing.
+	 */
+	dst->user_cpus_ptr = NULL;
+
+	/*
+	 * This check is racy and losing the race is a valid situation.
+	 * It is not worth the extra overhead of taking the pi_lock on
+	 * every fork/clone.
+	 */
+	if (data_race(!src->user_cpus_ptr))
 		return 0;
 
-	dst->user_cpus_ptr = kmalloc_node(cpumask_size(), GFP_KERNEL, node);
-	if (!dst->user_cpus_ptr)
+	user_mask = alloc_user_cpus_ptr(node);
+	if (!user_mask)
 		return -ENOMEM;
 
-	/* Use pi_lock to protect content of user_cpus_ptr */
+	/*
+	 * Use pi_lock to protect content of user_cpus_ptr
+	 *
+	 * Though unlikely, user_cpus_ptr can be reset to NULL by a concurrent
+	 * do_set_cpus_allowed().
+	 */
 	raw_spin_lock_irqsave(&src->pi_lock, flags);
-	cpumask_copy(dst->user_cpus_ptr, src->user_cpus_ptr);
+	if (src->user_cpus_ptr) {
+		swap(dst->user_cpus_ptr, user_mask);
+		cpumask_copy(dst->user_cpus_ptr, src->user_cpus_ptr);
+	}
 	raw_spin_unlock_irqrestore(&src->pi_lock, flags);
+
+	if (unlikely(user_mask))
+		kfree(user_mask);
+
 	return 0;
 }
 
@@ -3581,6 +3625,11 @@ static inline bool rq_has_pinned_tasks(struct rq *rq)
 	return false;
 }
 
+static inline cpumask_t *alloc_user_cpus_ptr(int node)
+{
+	return NULL;
+}
+
 #endif /* !CONFIG_SMP */
 
 static void
@@ -5504,7 +5553,9 @@ void scheduler_tick(void)
 	unsigned long thermal_pressure;
 	u64 resched_latency;
 
-	arch_scale_freq_tick();
+	if (housekeeping_cpu(cpu, HK_TYPE_TICK))
+		arch_scale_freq_tick();
+
 	sched_clock_tick();
 
 	rq_lock(rq, &rf);
@@ -8239,8 +8290,8 @@ long sched_setaffinity(pid_t pid, const struct cpumask *in_mask)
 	if (retval)
 		goto out_put_task;
 
-	user_mask = kmalloc(cpumask_size(), GFP_KERNEL);
-	if (!user_mask) {
+	user_mask = alloc_user_cpus_ptr(NUMA_NO_NODE);
+	if (IS_ENABLED(CONFIG_SMP) && !user_mask) {
 		retval = -ENOMEM;
 		goto out_put_task;
 	}