Merge tag 'sched_ext-for-7.2-rc3-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/sched_ext

Pull sched_ext fixes from Tejun Heo:

 - Lifecycle fixes for the new sub-scheduler support: two
   use-after-frees and an enable-failure path that left a
   half-initialized sub-scheduler linked.

 - Two dispatch-path locking bugs: a spurious scheduler abort from a
   migration race, and a lockdep splat from stale runqueue-lock
   tracking.

 - Callback and task-state fixes: stale scheduler-owned state on a task
   leaving SCX, a weight callback running after disable, and a bogus
   warning on core-scheduling forced idle.

 - On nohz_full, finite-slice tasks could miss the tick that expires
   their slice. Enable it when such a task is picked, with a selftest.

 - Smaller fixes: userspace CPU-mask helpers, ratelimited deprecation
   warnings, docs and a sparse annotation.

* tag 'sched_ext-for-7.2-rc3-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/sched_ext:
  sched_ext: Skip ops.set_weight() for disabled tasks
  tools/sched_ext: scx - Fix cmask_subset(), cmask_equal() and cmask_weight()
  sched_ext: Fix premature ops->priv publication in scx_alloc_and_add_sched()
  sched_ext: Record an error on errno-only sub-enable failure
  selftests/sched_ext: Verify nohz_full tick behavior
  sched_ext: Enable tick for finite slices on nohz_full
  sched_ext: Preserve rq tracking across local DSQ dispatch
  sched_ext: Documentation: Fix ops table header reference
  sched_ext: Don't warn on core-sched forced idle in put_prev_task_scx()
  sched_ext: Pin parent scx_sched across a child sub-scheduler's lifetime
  sched_ext: Annotate ksyncs with __rcu in alloc/free_kick_syncs()
  sched_ext: Check remote rq eligibility under task's rq lock
  sched_ext: Reset dsq_vtime and slice when a task leaves SCX
  sched_ext: Avoid flooding the log with deprecation warnings

7 files changed, 614 insertions, 95 deletions

diff --git a/Documentation/scheduler/sched-ext.rst b/Documentation/scheduler/sched-ext.rst
index 4b1ffd03f516..2771ea4cc14a 100644
--- a/Documentation/scheduler/sched-ext.rst
+++ b/Documentation/scheduler/sched-ext.rst
@@ -493,8 +493,9 @@ a freshly woken up task gets on a CPU.
 Where to Look
 =============
 
-* ``include/linux/sched/ext.h`` defines the core data structures, ops table
-  and constants.
+* ``include/linux/sched/ext.h`` defines the core data structures and
+  constants, while the ops table (``struct sched_ext_ops``) is defined in
+  ``kernel/sched/ext/internal.h``.
 
 * ``kernel/sched/ext/ext.c`` contains sched_ext core implementation and helpers.
   The functions prefixed with ``scx_bpf_`` can be called from the BPF
@@ -555,7 +556,8 @@ ABI Instability
 ===============
 
 The APIs provided by sched_ext to BPF schedulers programs have no stability
-guarantees. This includes the ops table callbacks and constants defined in
+guarantees. This includes the ops table callbacks defined in
+``kernel/sched/ext/internal.h`` and the constants defined in
 ``include/linux/sched/ext.h``, as well as the ``scx_bpf_`` kfuncs defined in
 ``kernel/sched/ext/ext.c`` and ``kernel/sched/ext/idle.c``.
 
diff --git a/kernel/sched/ext/ext.c b/kernel/sched/ext/ext.c
index 691d53fe0f64..e3fa7b2fac9d 100644
--- a/kernel/sched/ext/ext.c
+++ b/kernel/sched/ext/ext.c
@@ -479,6 +479,18 @@ static bool rq_is_open(struct rq *rq, u64 enq_flags)
  */
 DEFINE_PER_CPU(struct rq *, scx_locked_rq_state);
 
+static void switch_rq_lock(struct rq *from, struct rq *to)
+{
+	bool tracked = scx_locked_rq() == from;
+
+	if (tracked)
+		update_locked_rq(NULL);
+	raw_spin_rq_unlock(from);
+	raw_spin_rq_lock(to);
+	if (tracked)
+		update_locked_rq(to);
+}
+
 /*
  * Flipped on enable per sch->is_cid_type. Declared in internal.h so
  * subsystem inlines can read it.
@@ -2274,8 +2286,7 @@ static void move_remote_task_to_local_dsq(struct task_struct *p, u64 enq_flags,
 	deactivate_task(src_rq, p, 0);
 	set_task_cpu(p, cpu_of(dst_rq));
 
-	raw_spin_rq_unlock(src_rq);
-	raw_spin_rq_lock(dst_rq);
+	switch_rq_lock(src_rq, dst_rq);
 
 	/*
 	 * We want to pass scx-specific enq_flags but activate_task() will
@@ -2307,6 +2318,7 @@ static void move_remote_task_to_local_dsq(struct task_struct *p, u64 enq_flags,
  *   no to the BPF scheduler initiated migrations while offline.
  *
  * The caller must ensure that @p and @rq are on different CPUs.
+ * If enforce == true, caller must hold @p's rq lock.
  */
 static bool task_can_run_on_remote_rq(struct scx_sched *sch,
 				      struct task_struct *p, struct rq *rq,
@@ -2314,6 +2326,14 @@ static bool task_can_run_on_remote_rq(struct scx_sched *sch,
 {
 	s32 cpu = cpu_of(rq);
 
+	/*
+	 * To prevent races with @p still running on its old CPU while switching
+	 * out, make sure we're holding @p's rq lock so as not to risk
+	 * erroneously killing the BPF scheduler.
+	 */
+	if (enforce)
+		lockdep_assert_rq_held(task_rq(p));
+
 	WARN_ON_ONCE(task_cpu(p) == cpu);
 
 	/*
@@ -2581,13 +2601,6 @@ static void dispatch_to_local_dsq(struct scx_sched *sch, struct rq *rq,
 		return;
 	}
 
-	if (src_rq != dst_rq &&
-	    unlikely(!task_can_run_on_remote_rq(sch, p, dst_rq, true))) {
-		dispatch_enqueue(sch, rq, find_global_dsq(sch, task_cpu(p)), p,
-				 enq_flags | SCX_ENQ_CLEAR_OPSS | SCX_ENQ_GDSQ_FALLBACK);
-		return;
-	}
-
 	/*
 	 * @p is on a possibly remote @src_rq which we need to lock to move the
 	 * task. If dequeue is in progress, it'd be locking @src_rq and waiting
@@ -2606,14 +2619,14 @@ static void dispatch_to_local_dsq(struct scx_sched *sch, struct rq *rq,
 
 	/* switch to @src_rq lock */
 	if (locked_rq != src_rq) {
-		raw_spin_rq_unlock(locked_rq);
+		switch_rq_lock(locked_rq, src_rq);
 		locked_rq = src_rq;
-		raw_spin_rq_lock(src_rq);
 	}
 
 	/* task_rq couldn't have changed if we're still the holding cpu */
 	if (likely(p->scx.holding_cpu == raw_smp_processor_id()) &&
 	    !WARN_ON_ONCE(src_rq != task_rq(p))) {
+		bool fallback = false;
 		/*
 		 * If @p is staying on the same rq, there's no need to go
 		 * through the full deactivate/activate cycle. Optimize by
@@ -2623,6 +2636,11 @@ static void dispatch_to_local_dsq(struct scx_sched *sch, struct rq *rq,
 			p->scx.holding_cpu = -1;
 			dispatch_enqueue(sch, dst_rq, &dst_rq->scx.local_dsq, p,
 					 enq_flags);
+		} else if (unlikely(!task_can_run_on_remote_rq(sch, p, dst_rq, true))) {
+			p->scx.holding_cpu = -1;
+			fallback = true;
+			dispatch_enqueue(sch, src_rq, find_global_dsq(sch, task_cpu(p)),
+					 p, enq_flags | SCX_ENQ_GDSQ_FALLBACK);
 		} else {
 			move_remote_task_to_local_dsq(p, enq_flags,
 						      src_rq, dst_rq);
@@ -2631,15 +2649,13 @@ static void dispatch_to_local_dsq(struct scx_sched *sch, struct rq *rq,
 		}
 
 		/* if the destination CPU is idle, wake it up */
-		if (sched_class_above(p->sched_class, dst_rq->curr->sched_class))
+		if (!fallback && sched_class_above(p->sched_class, dst_rq->curr->sched_class))
 			resched_curr(dst_rq);
 	}
 
 	/* switch back to @rq lock */
-	if (locked_rq != rq) {
-		raw_spin_rq_unlock(locked_rq);
-		raw_spin_rq_lock(rq);
-	}
+	if (locked_rq != rq)
+		switch_rq_lock(locked_rq, rq);
 }
 
 /**
@@ -2970,24 +2986,38 @@ static void set_next_task_scx(struct rq *rq, struct task_struct *p, bool first)
 
 	/*
 	 * @p is getting newly scheduled or got kicked after someone updated its
-	 * slice. Refresh whether tick can be stopped. See scx_can_stop_tick().
+	 * slice. Update SCX_RQ_CAN_STOP_TICK to reflect whether the tick can be
+	 * stopped. See scx_can_stop_tick().
+	 *
+	 * Moreover, refresh the load_avgs just when transitioning in and out of
+	 * nohz. In the future, we might want to add a mechanism to update
+	 * load_avgs periodically on tick-stopped CPUs.
 	 */
-	if ((p->scx.slice == SCX_SLICE_INF) !=
-	    (bool)(rq->scx.flags & SCX_RQ_CAN_STOP_TICK)) {
-		if (p->scx.slice == SCX_SLICE_INF)
+	if (p->scx.slice == SCX_SLICE_INF) {
+		if (!(rq->scx.flags & SCX_RQ_CAN_STOP_TICK)) {
+			/*
+			 * Bypass mode always assigns finite slices, so @p
+			 * can't have an infinite slice while bypassing.
+			 * Therefore, sched_update_tick_dependency() can safely
+			 * evaluate the outgoing task.
+			 */
 			rq->scx.flags |= SCX_RQ_CAN_STOP_TICK;
-		else
-			rq->scx.flags &= ~SCX_RQ_CAN_STOP_TICK;
+			sched_update_tick_dependency(rq);
 
-		sched_update_tick_dependency(rq);
+			update_other_load_avgs(rq);
+		}
+	} else {
+		if (rq->scx.flags & SCX_RQ_CAN_STOP_TICK) {
+			rq->scx.flags &= ~SCX_RQ_CAN_STOP_TICK;
+			update_other_load_avgs(rq);
+		}
 
 		/*
-		 * For now, let's refresh the load_avgs just when transitioning
-		 * in and out of nohz. In the future, we might want to add a
-		 * mechanism which calls the following periodically on
-		 * tick-stopped CPUs.
+		 * @rq still references the outgoing scheduling context. A finite
+		 * slice is sufficient by itself to require the tick.
 		 */
-		update_other_load_avgs(rq);
+		if (tick_nohz_full_cpu(cpu_of(rq)))
+			tick_nohz_dep_set_cpu(cpu_of(rq), TICK_DEP_BIT_SCHED);
 	}
 }
 
@@ -3082,9 +3112,14 @@ static void put_prev_task_scx(struct rq *rq, struct task_struct *p,
 		 * sched_class, %SCX_OPS_ENQ_LAST must be set. Tell
 		 * ops.enqueue() that @p is the only one available for this cpu,
 		 * which should trigger an explicit follow-up scheduling event.
+		 *
+		 * Core scheduling can force this CPU idle while @p stays
+		 * runnable. @p's cookie then won't match the core's, so skip
+		 * the warning in that case.
 		 */
 		if (next && sched_class_above(&ext_sched_class, next->sched_class)) {
-			WARN_ON_ONCE(!(sch->ops.flags & SCX_OPS_ENQ_LAST));
+			WARN_ON_ONCE(sched_cpu_cookie_match(rq, p) &&
+				     !(sch->ops.flags & SCX_OPS_ENQ_LAST));
 			do_enqueue_task(rq, p, SCX_ENQ_LAST, -1);
 		} else {
 			do_enqueue_task(rq, p, 0, -1);
@@ -3648,6 +3683,13 @@ static void scx_disable_task(struct scx_sched *sch, struct task_struct *p)
 	scx_set_task_state(p, SCX_TASK_READY);
 
 	/*
+	 * Reset the SCX-managed fields when @p leaves the BPF scheduler's
+	 * control, after ops.disable() has observed their final values.
+	 */
+	p->scx.dsq_vtime = 0;
+	p->scx.slice = 0;
+
+	/*
 	 * Verify the task is not in BPF scheduler's custody. If flag
 	 * transitions are consistent, the flag should always be clear
 	 * here.
@@ -3925,6 +3967,17 @@ static void reweight_task_scx(struct rq *rq, struct task_struct *p,
 	if (task_dead_and_done(p))
 		return;
 
+	/*
+	 * When switching sched_class away from SCX, reweight_task_scx()
+	 * is called _after_ scx_disable_task(). Skip calling ops.set_weight()
+	 * since the BPF scheduler may have already forgotten the task in
+	 * ops.disable().
+	 * p->scx.weight will be recalculated in scx_enable_task() if the task
+	 * ever returns to SCX class.
+	 */
+	if (scx_get_task_state(p) != SCX_TASK_ENABLED)
+		return;
+
 	p->scx.weight = sched_weight_to_cgroup(scale_load_down(lw->weight));
 	if (SCX_HAS_OP(sch, set_weight))
 		SCX_CALL_OP_TASK(sch, set_weight, rq, p, p->scx.weight);
@@ -4301,6 +4354,15 @@ bool scx_can_stop_tick(struct rq *rq)
 	if (p->sched_class != &ext_sched_class)
 		return true;
 
+	/*
+	 * @rq->curr may still reference an outgoing EXT task after it has been
+	 * dequeued. If no EXT tasks are accounted on @rq, ignore its stale
+	 * slice state. If another task is dispatched from a DSQ,
+	 * set_next_task_scx() will update the dependency for the incoming task.
+	 */
+	if (!rq->scx.nr_running)
+		return true;
+
 	if (scx_bypassing(sch, cpu_of(rq)))
 		return false;
 
@@ -4901,6 +4963,8 @@ static void scx_sched_free_rcu_work(struct work_struct *work)
 		cgroup_put(sch_cgroup(sch));
 	if (sch->sub_kset)
 		kobject_put(&sch->sub_kset->kobj);
+	if (scx_parent(sch))
+		kobject_put(&scx_parent(sch)->kobj);
 #endif	/* CONFIG_EXT_SUB_SCHED */
 
 	for_each_possible_cpu(cpu) {
@@ -5672,7 +5736,7 @@ static void free_kick_syncs(void)
 	int cpu;
 
 	for_each_possible_cpu(cpu) {
-		struct scx_kick_syncs **ksyncs = per_cpu_ptr(&scx_kick_syncs, cpu);
+		struct scx_kick_syncs __rcu **ksyncs = per_cpu_ptr(&scx_kick_syncs, cpu);
 		struct scx_kick_syncs *to_free;
 
 		to_free = rcu_replace_pointer(*ksyncs, NULL, true);
@@ -6653,7 +6717,7 @@ static int alloc_kick_syncs(void)
 	 * can exceed percpu allocator limits on large machines.
 	 */
 	for_each_possible_cpu(cpu) {
-		struct scx_kick_syncs **ksyncs = per_cpu_ptr(&scx_kick_syncs, cpu);
+		struct scx_kick_syncs __rcu **ksyncs = per_cpu_ptr(&scx_kick_syncs, cpu);
 		struct scx_kick_syncs *new_ksyncs;
 
 		WARN_ON_ONCE(rcu_access_pointer(*ksyncs));
@@ -6825,11 +6889,6 @@ static struct scx_sched *scx_alloc_and_add_sched(struct scx_enable_cmd *cmd,
 		sch->ops = *cmd->ops;
 	}
 
-	rcu_assign_pointer(ops->priv, sch);
-
-	sch->kobj.kset = scx_kset;
-	INIT_LIST_HEAD(&sch->all);
-
 #ifdef CONFIG_EXT_SUB_SCHED
 	char *buf = kzalloc(PATH_MAX, GFP_KERNEL);
 	if (!buf) {
@@ -6847,13 +6906,32 @@ static struct scx_sched *scx_alloc_and_add_sched(struct scx_enable_cmd *cmd,
 	sch->cgrp = cgrp;
 	INIT_LIST_HEAD(&sch->children);
 	INIT_LIST_HEAD(&sch->sibling);
+#endif	/* CONFIG_EXT_SUB_SCHED */
 
-	if (parent)
+	/*
+	 * Publishing makes @sch visible to scx_prog_sched() readers. Failure
+	 * paths after this point must free @sch through kobject_put() whose
+	 * release path defers the actual freeing by an RCU grace period.
+	 */
+	rcu_assign_pointer(ops->priv, sch);
+
+	sch->kobj.kset = scx_kset;
+	INIT_LIST_HEAD(&sch->all);
+
+#ifdef CONFIG_EXT_SUB_SCHED
+	if (parent) {
+		/*
+		 * Pin @parent for @sch's lifetime. The kobject hierarchy pins
+		 * it only via @parent->sub_kset, which is dropped during
+		 * disable. Released in scx_sched_free_rcu_work().
+		 */
+		kobject_get(&parent->kobj);
 		ret = kobject_init_and_add(&sch->kobj, &scx_ktype,
 					   &parent->sub_kset->kobj,
 					   "sub-%llu", cgroup_id(cgrp));
-	else
+	} else {
 		ret = kobject_init_and_add(&sch->kobj, &scx_ktype, NULL, "root");
+	}
 
 	if (ret < 0) {
 		RCU_INIT_POINTER(ops->priv, NULL);
@@ -6895,7 +6973,6 @@ static struct scx_sched *scx_alloc_and_add_sched(struct scx_enable_cmd *cmd,
 
 #ifdef CONFIG_EXT_SUB_SCHED
 err_free_lb_resched:
-	RCU_INIT_POINTER(ops->priv, NULL);
 	free_cpumask_var(sch->bypass_lb_resched_cpumask);
 #endif
 err_free_lb_cpumask:
@@ -6988,7 +7065,7 @@ static int validate_ops(struct scx_sched *sch, const struct sched_ext_ops *ops)
 	 * run past the BPF allocation. Skip for cid-form.
 	 */
 	if (!sch->is_cid_type && (ops->cpu_acquire || ops->cpu_release))
-		pr_warn("ops->cpu_acquire/release() are deprecated, use sched_switch TP instead\n");
+		pr_warn_ratelimited("ops->cpu_acquire/release() are deprecated, use sched_switch TP instead\n");
 
 	/*
 	 * Sub-scheduler support is tied to the cid-form struct_ops. A sub-sched
@@ -7686,6 +7763,12 @@ err_unlock_and_disable:
 	percpu_up_write(&scx_fork_rwsem);
 err_disable:
 	mutex_unlock(&scx_enable_mutex);
+	/*
+	 * Some enable failures only return an errno (e.g. -ENOMEM from an
+	 * allocation) without calling scx_error(). Record it so
+	 * scx_flush_disable_work() runs the disable and ops.exit() fires.
+	 */
+	scx_error(sch, "scx_sub_enable() failed (%d)", ret);
 	scx_flush_disable_work(sch);
 	cmd->ret = 0;
 }
@@ -7806,7 +7889,7 @@ static int bpf_scx_btf_struct_access(struct bpf_verifier_log *log,
 		     off + size <= offsetofend(struct task_struct, scx.slice)) ||
 		    (off >= offsetof(struct task_struct, scx.dsq_vtime) &&
 		     off + size <= offsetofend(struct task_struct, scx.dsq_vtime))) {
-			pr_warn("sched_ext: Writing directly to p->scx.slice/dsq_vtime is deprecated, use scx_bpf_task_set_slice/dsq_vtime()");
+			pr_warn_ratelimited("sched_ext: Writing directly to p->scx.slice/dsq_vtime is deprecated, use scx_bpf_task_set_slice/dsq_vtime()\n");
 			return SCALAR_VALUE;
 		}
 
@@ -8796,10 +8879,8 @@ static bool scx_dsq_move(struct bpf_iter_scx_dsq_kern *kit,
 	in_balance = this_rq->scx.flags & SCX_RQ_IN_BALANCE;
 
 	if (in_balance) {
-		if (this_rq != src_rq) {
-			raw_spin_rq_unlock(this_rq);
-			raw_spin_rq_lock(src_rq);
-		}
+		if (this_rq != src_rq)
+			switch_rq_lock(this_rq, src_rq);
 	} else {
 		raw_spin_rq_lock(src_rq);
 	}
@@ -8831,10 +8912,8 @@ static bool scx_dsq_move(struct bpf_iter_scx_dsq_kern *kit,
 	dispatched = true;
 out:
 	if (in_balance) {
-		if (this_rq != locked_rq) {
-			raw_spin_rq_unlock(locked_rq);
-			raw_spin_rq_lock(this_rq);
-		}
+		if (this_rq != locked_rq)
+			switch_rq_lock(locked_rq, this_rq);
 	} else {
 		raw_spin_rq_unlock_irqrestore(locked_rq, flags);
 	}
diff --git a/kernel/sched/ext/internal.h b/kernel/sched/ext/internal.h
index 145272cb4d8a..673059fa9d72 100644
--- a/kernel/sched/ext/internal.h
+++ b/kernel/sched/ext/internal.h
@@ -1469,21 +1469,24 @@ static const char *scx_enable_state_str[] = {
  * The sched_ext core uses a "lock dancing" protocol coordinated by
  * p->scx.holding_cpu. When moving a task to a different rq:
  *
- *   1. Verify task can be moved (CPU affinity, migration_disabled, etc.)
- *   2. Set p->scx.holding_cpu to the current CPU
- *   3. Set task state to %SCX_OPSS_NONE; dequeue waits while DISPATCHING
+ *   1. Set p->scx.holding_cpu to the current CPU
+ *   2. Set task state to %SCX_OPSS_NONE; dequeue waits while DISPATCHING
  *      is set, so clearing DISPATCHING first prevents the circular wait
  *      (safe to lock the rq we need)
- *   4. Unlock the current CPU's rq
- *   5. Lock src_rq (where the task currently lives)
- *   6. Verify p->scx.holding_cpu == current CPU, if not, dequeue won the
+ *   3. Unlock the current CPU's rq
+ *   4. Lock src_rq (where the task currently lives)
+ *   5. Verify p->scx.holding_cpu == current CPU, if not, dequeue won the
  *      race (dequeue clears holding_cpu to -1 when it takes the task), in
  *      this case migration is aborted
- *   7. If src_rq == dst_rq: clear holding_cpu and enqueue directly
+ *   6. If src_rq == dst_rq: clear holding_cpu and enqueue directly
  *      into dst_rq's local DSQ (no lock swap needed)
- *   8. Otherwise: call move_remote_task_to_local_dsq(), which releases
- *      src_rq, locks dst_rq, and performs the deactivate/activate
- *      migration cycle (dst_rq is held on return)
+ *   7. Otherwise, verify under src_rq lock that the task can be moved to dst_rq
+ *      (CPU affinity, migration_disabled, etc.). If not, clear holding_cpu,
+ *      leave the task on src_rq, and enqueue it on the fallback DSQ.
+ *   8. Otherwise (i.e. if the task can be moved to dst_rq), call
+ *      move_remote_task_to_local_dsq(), which releases src_rq, locks dst_rq,
+ *      and performs the deactivate/activate migration cycle
+ *      (dst_rq is held on return)
  *   9. Unlock dst_rq and re-lock the current CPU's rq to restore
  *      the lock state expected by the caller
  *
diff --git a/tools/sched_ext/include/scx/cid.bpf.h b/tools/sched_ext/include/scx/cid.bpf.h
index db247e42fb45..6b0b4e41b288 100644
--- a/tools/sched_ext/include/scx/cid.bpf.h
+++ b/tools/sched_ext/include/scx/cid.bpf.h
@@ -391,7 +391,9 @@ static __always_inline bool cmask_equal(const struct scx_cmask __arena *a,
 
 	if (a->base != b->base || a->nr_cids != b->nr_cids)
 		return false;
-	nr_words = CMASK_NR_WORDS(a->nr_cids);
+	if (a->nr_cids == 0)
+		return true;
+	nr_words = (a->base + a->nr_cids - 1) / 64 - a->base / 64 + 1;
 
 	bpf_for(i, 0, CMASK_MAX_WORDS) {
 		if (i >= nr_words)
@@ -402,36 +404,6 @@ static __always_inline bool cmask_equal(const struct scx_cmask __arena *a,
 	return true;
 }
 
-/*
- * True iff every bit set in @a is also set in @b over the intersection of
- * their ranges. Bits of @a outside @b's range fail the test.
- */
-static __always_inline bool cmask_subset(const struct scx_cmask __arena *a,
-					 const struct scx_cmask __arena *b)
-{
-	u32 a_end = a->base + a->nr_cids;
-	u32 b_end = b->base + b->nr_cids;
-	u32 a_wbase = a->base / 64;
-	u32 b_wbase = b->base / 64;
-	u32 nr_words, i;
-
-	/* any bit of @a outside @b's range is a subset violation */
-	if (a->base < b->base || a_end > b_end)
-		return false;
-
-	nr_words = CMASK_NR_WORDS(a->nr_cids);
-	bpf_for(i, 0, CMASK_MAX_WORDS) {
-		u32 wi_b;
-
-		if (i >= nr_words)
-			break;
-		wi_b = a_wbase + i - b_wbase;
-		if (a->bits[i] & ~b->bits[wi_b])
-			return false;
-	}
-	return true;
-}
-
 /**
  * cmask_next_set - find the first set bit at or after @cid
  * @m: cmask to search
@@ -489,15 +461,65 @@ static __always_inline u32 cmask_first_set(const struct scx_cmask __arena *m)
 	     (cid) = cmask_next_set((m), (cid) + 1))
 
 /*
+ * True iff every bit set in @a is also set in @b. Matches the kernel-side
+ * scx_cmask_subset(): ranges don't need to nest, and set bits of @a outside
+ * @b's range fail the test.
+ */
+static __always_inline bool cmask_subset(const struct scx_cmask __arena *a,
+					 const struct scx_cmask __arena *b)
+{
+	u32 a_end = a->base + a->nr_cids;
+	u32 b_end = b->base + b->nr_cids;
+	u32 a_wbase = a->base / 64;
+	u32 b_wbase = b->base / 64;
+	u32 lo = a->base > b->base ? a->base : b->base;
+	u32 hi = a_end < b_end ? a_end : b_end;
+	u32 lo_word, hi_word, i;
+
+	/* set bits of @a outside @b's range can't be in @b */
+	if (a->base < b->base &&
+	    cmask_next_set(a, a->base) < (b->base < a_end ? b->base : a_end))
+		return false;
+	if (a_end > b_end &&
+	    cmask_next_set(a, a->base > b_end ? a->base : b_end) < a_end)
+		return false;
+
+	if (lo >= hi)
+		return true;
+
+	/*
+	 * Walk the words the range intersection spans. Plain word tests
+	 * suffice: the scans above guarantee @a has no set bit outside @b's
+	 * range and padding bits are kept clear by all cmask helpers.
+	 */
+	lo_word = lo / 64;
+	hi_word = (hi - 1) / 64;
+
+	bpf_for(i, 0, CMASK_MAX_WORDS) {
+		u32 w = lo_word + i;
+
+		if (w > hi_word)
+			break;
+		if (a->bits[w - a_wbase] & ~b->bits[w - b_wbase])
+			return false;
+	}
+	return true;
+}
+
+/*
  * Population count over [base, base + nr_cids). Padding bits in the head/tail
  * words are guaranteed zero by the mutating helpers, so a flat popcount over
- * all words is correct.
+ * the words the range spans is correct.
  */
 static __always_inline u32 cmask_weight(const struct scx_cmask __arena *m)
 {
-	u32 nr_words = CMASK_NR_WORDS(m->nr_cids), i;
+	u32 nr_words, i;
 	u32 count = 0;
 
+	if (!m->nr_cids)
+		return 0;
+	nr_words = (m->base + m->nr_cids - 1) / 64 - m->base / 64 + 1;
+
 	bpf_for(i, 0, CMASK_MAX_WORDS) {
 		if (i >= nr_words)
 			break;
diff --git a/tools/testing/selftests/sched_ext/Makefile b/tools/testing/selftests/sched_ext/Makefile
index 5d2dffca0e91..3cfe90e0f34f 100644
--- a/tools/testing/selftests/sched_ext/Makefile
+++ b/tools/testing/selftests/sched_ext/Makefile
@@ -176,6 +176,7 @@ auto-test-targets :=			\
 	maybe_null			\
 	minimal				\
 	non_scx_kfunc_deny		\
+	nohz_tick			\
 	numa				\
 	allowed_cpus			\
 	peek_dsq			\
diff --git a/tools/testing/selftests/sched_ext/nohz_tick.bpf.c b/tools/testing/selftests/sched_ext/nohz_tick.bpf.c
new file mode 100644
index 000000000000..6998c5dd6bcb
--- /dev/null
+++ b/tools/testing/selftests/sched_ext/nohz_tick.bpf.c
@@ -0,0 +1,65 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES
+ *
+ * Exercise tick dependency transitions between infinite and finite slices.
+ */
+#include <scx/common.bpf.h>
+
+char _license[] SEC("license") = "GPL";
+
+const volatile s32 test_cpu;
+bool finite_phase;
+u64 nr_inf_running;
+u64 nr_finite_running;
+u64 nr_finite_ticks;
+
+UEI_DEFINE(uei);
+
+s32 BPF_STRUCT_OPS(nohz_tick_select_cpu, struct task_struct *p, s32 prev_cpu,
+		   u64 wake_flags)
+{
+	return prev_cpu;
+}
+
+void BPF_STRUCT_OPS(nohz_tick_enqueue, struct task_struct *p, u64 enq_flags)
+{
+	u64 slice = finite_phase ? 1000000ULL : SCX_SLICE_INF;
+
+	scx_bpf_dsq_insert(p, SCX_DSQ_GLOBAL, slice, enq_flags);
+	if (enq_flags & SCX_ENQ_LAST)
+		scx_bpf_kick_cpu(test_cpu, SCX_KICK_IDLE);
+}
+
+void BPF_STRUCT_OPS(nohz_tick_running, struct task_struct *p)
+{
+	if (bpf_get_smp_processor_id() != test_cpu)
+		return;
+
+	if (finite_phase)
+		__sync_fetch_and_add(&nr_finite_running, 1);
+	else
+		__sync_fetch_and_add(&nr_inf_running, 1);
+}
+
+void BPF_STRUCT_OPS(nohz_tick_tick, struct task_struct *p)
+{
+	if (bpf_get_smp_processor_id() == test_cpu && finite_phase)
+		__sync_fetch_and_add(&nr_finite_ticks, 1);
+}
+
+void BPF_STRUCT_OPS(nohz_tick_exit, struct scx_exit_info *ei)
+{
+	UEI_RECORD(uei, ei);
+}
+
+SEC(".struct_ops.link")
+struct sched_ext_ops nohz_tick_ops = {
+	.select_cpu		= (void *)nohz_tick_select_cpu,
+	.enqueue		= (void *)nohz_tick_enqueue,
+	.running		= (void *)nohz_tick_running,
+	.tick			= (void *)nohz_tick_tick,
+	.exit			= (void *)nohz_tick_exit,
+	.name			= "nohz_tick",
+	.timeout_ms		= 1000U,
+};
diff --git a/tools/testing/selftests/sched_ext/nohz_tick.c b/tools/testing/selftests/sched_ext/nohz_tick.c
new file mode 100644
index 000000000000..028f54391c2c
--- /dev/null
+++ b/tools/testing/selftests/sched_ext/nohz_tick.c
@@ -0,0 +1,347 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES
+ *
+ * Validate that a finite-slice EXT task restarts the scheduler tick when it
+ * follows an infinite-slice EXT task and an idle interval on a NOHZ_FULL CPU.
+ */
+#define _GNU_SOURCE
+
+#include <bpf/bpf.h>
+#include <errno.h>
+#include <sched.h>
+#include <signal.h>
+#include <stdbool.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <sys/prctl.h>
+#include <sys/wait.h>
+#include <unistd.h>
+
+#include <scx/common.h>
+
+#include "nohz_tick.bpf.skel.h"
+#include "scx_test.h"
+
+#ifndef SCHED_EXT
+#define SCHED_EXT 7
+#endif
+
+#define MIN_FINITE_TICKS 3
+#define PHASE_TIMEOUT_MS 1000
+
+struct nohz_tick_ctx {
+	struct nohz_tick *skel;
+	cpu_set_t original_mask;
+	int test_cpu;
+};
+
+static int first_allowed_cpu(const cpu_set_t *mask, int first, int last)
+{
+	int cpu;
+
+	for (cpu = first; cpu <= last && cpu < CPU_SETSIZE; cpu++)
+		if (CPU_ISSET(cpu, mask))
+			return cpu;
+
+	return -1;
+}
+
+static int find_nohz_full_cpu(const cpu_set_t *allowed)
+{
+	char buf[4096], *cur, *end;
+	FILE *file;
+
+	file = fopen("/sys/devices/system/cpu/nohz_full", "r");
+	if (!file)
+		return -1;
+	if (!fgets(buf, sizeof(buf), file)) {
+		fclose(file);
+		return -1;
+	}
+	fclose(file);
+
+	cur = buf;
+	while (*cur) {
+		long first, last;
+		int cpu;
+
+		while (*cur == ' ' || *cur == '\t' || *cur == ',')
+			cur++;
+		if (*cur < '0' || *cur > '9')
+			break;
+
+		errno = 0;
+		first = strtol(cur, &end, 10);
+		if (errno || end == cur || first < 0 || first >= CPU_SETSIZE)
+			return -1;
+		cur = end;
+		last = first;
+		if (*cur == '-') {
+			cur++;
+			errno = 0;
+			last = strtol(cur, &end, 10);
+			if (errno || end == cur || last < first)
+				return -1;
+			cur = end;
+		}
+
+		cpu = first_allowed_cpu(allowed, first, last);
+		if (cpu >= 0)
+			return cpu;
+	}
+
+	return -1;
+}
+
+static pid_t start_worker(int cpu)
+{
+	struct sched_param param = {};
+	cpu_set_t mask;
+	pid_t parent;
+	pid_t pid;
+
+	parent = getpid();
+	pid = fork();
+	if (pid != 0)
+		return pid;
+	if (prctl(PR_SET_PDEATHSIG, SIGKILL) || getppid() != parent)
+		_exit(1);
+
+	/*
+	 * Become EXT before touching the target so it stays idle until wakeup.
+	 */
+	if (sched_setscheduler(0, SCHED_EXT, &param))
+		_exit(1);
+
+	CPU_ZERO(&mask);
+	CPU_SET(cpu, &mask);
+	if (sched_setaffinity(0, sizeof(mask), &mask))
+		_exit(1);
+
+	for (;;)
+		asm volatile("" ::: "memory");
+}
+
+static void stop_worker(pid_t pid)
+{
+	if (pid <= 0)
+		return;
+
+	kill(pid, SIGKILL);
+	waitpid(pid, NULL, 0);
+}
+
+static int pause_worker(pid_t pid)
+{
+	int status;
+
+	if (kill(pid, SIGSTOP))
+		return -errno;
+	if (waitpid(pid, &status, WUNTRACED) != pid)
+		return -errno;
+	if (!WIFSTOPPED(status))
+		return -ECHILD;
+
+	return 0;
+}
+
+static bool wait_for_counter(const u64 *counter, u64 value, int timeout_ms)
+{
+	int elapsed;
+
+	for (elapsed = 0; elapsed < timeout_ms; elapsed++) {
+		if (__atomic_load_n(counter, __ATOMIC_RELAXED) >= value)
+			return true;
+		usleep(1000);
+	}
+
+	return false;
+}
+
+static enum scx_test_status setup(void **ctx_ptr)
+{
+	struct nohz_tick_ctx *ctx;
+	cpu_set_t controller_mask;
+	int cpu;
+
+	ctx = calloc(1, sizeof(*ctx));
+	SCX_FAIL_IF(!ctx, "Failed to allocate context");
+	if (sched_getaffinity(0, sizeof(ctx->original_mask),
+			      &ctx->original_mask)) {
+		free(ctx);
+		SCX_FAIL("Failed to get affinity (%d)", errno);
+	}
+
+	cpu = find_nohz_full_cpu(&ctx->original_mask);
+	if (cpu < 0) {
+		fprintf(stderr, "SKIP: no allowed NOHZ_FULL CPU\n");
+		free(ctx);
+		return SCX_TEST_SKIP;
+	}
+
+	controller_mask = ctx->original_mask;
+	CPU_CLR(cpu, &controller_mask);
+	if (CPU_COUNT(&controller_mask) == 0) {
+		fprintf(stderr, "SKIP: no housekeeping CPU available\n");
+		free(ctx);
+		return SCX_TEST_SKIP;
+	}
+
+	ctx->test_cpu = cpu;
+	ctx->skel = nohz_tick__open();
+	if (!ctx->skel) {
+		free(ctx);
+		SCX_FAIL("Failed to open skeleton");
+	}
+
+	SCX_ENUM_INIT(ctx->skel);
+	ctx->skel->rodata->test_cpu = cpu;
+	ctx->skel->struct_ops.nohz_tick_ops->flags |= SCX_OPS_SWITCH_PARTIAL |
+							   SCX_OPS_ENQ_LAST;
+	if (nohz_tick__load(ctx->skel)) {
+		nohz_tick__destroy(ctx->skel);
+		free(ctx);
+		SCX_FAIL("Failed to load skeleton");
+	}
+
+	if (sched_setaffinity(0, sizeof(controller_mask), &controller_mask)) {
+		nohz_tick__destroy(ctx->skel);
+		free(ctx);
+		SCX_FAIL("Failed to move controller off CPU %d (%d)", cpu, errno);
+	}
+
+	*ctx_ptr = ctx;
+	return SCX_TEST_PASS;
+}
+
+static enum scx_test_status run(void *ctx_ptr)
+{
+	struct nohz_tick_ctx *ctx = ctx_ptr;
+	struct nohz_tick *skel = ctx->skel;
+	struct bpf_link *link = NULL;
+	enum scx_test_status status = SCX_TEST_FAIL;
+	pid_t finite_worker = -1;
+	pid_t inf_worker = -1;
+	u64 finite_running;
+	u64 finite_ticks;
+	int ret;
+
+	link = bpf_map__attach_struct_ops(skel->maps.nohz_tick_ops);
+	if (!link) {
+		SCX_ERR("Failed to attach scheduler");
+		goto out;
+	}
+
+	/*
+	 * Establish SCX_RQ_CAN_STOP_TICK with an infinite-slice task.
+	 */
+	inf_worker = start_worker(ctx->test_cpu);
+	if (inf_worker < 0) {
+		SCX_ERR("Failed to start infinite-slice worker (%d)", errno);
+		goto out;
+	}
+	if (!wait_for_counter(&skel->bss->nr_inf_running, 1,
+			      PHASE_TIMEOUT_MS)) {
+		SCX_ERR("Infinite-slice worker was not scheduled");
+		goto out;
+	}
+
+	/* Block without exiting so the rq retains the infinite-slice state. */
+	ret = pause_worker(inf_worker);
+	if (ret) {
+		SCX_ERR("Failed to stop infinite-slice worker (%d)", ret);
+		goto out;
+	}
+
+	/* Let the target enter idle with its tick stopped. */
+	usleep(100000);
+
+	/*
+	 * The next EXT task receives a finite slice and must restart the tick.
+	 */
+	__atomic_store_n(&skel->bss->finite_phase, true, __ATOMIC_RELEASE);
+	finite_worker = start_worker(ctx->test_cpu);
+	if (finite_worker < 0) {
+		SCX_ERR("Failed to start finite-slice worker (%d)", errno);
+		goto out;
+	}
+	if (!wait_for_counter(&skel->bss->nr_finite_running, 1,
+			      PHASE_TIMEOUT_MS)) {
+		SCX_ERR("Finite-slice worker was not scheduled");
+		goto out;
+	}
+	if (!wait_for_counter(&skel->bss->nr_finite_ticks, MIN_FINITE_TICKS,
+			      PHASE_TIMEOUT_MS)) {
+		SCX_ERR("Finite-slice worker received only %llu scheduler ticks",
+			(unsigned long long)skel->bss->nr_finite_ticks);
+		goto out;
+	}
+	stop_worker(finite_worker);
+	finite_worker = -1;
+
+	/*
+	 * Leave the CPU idle after a finite-slice task. The next finite-slice
+	 * task must restart the tick even though the slice type is unchanged.
+	 */
+	usleep(100000);
+	finite_running = __atomic_load_n(&skel->bss->nr_finite_running,
+					 __ATOMIC_RELAXED);
+	finite_ticks = __atomic_load_n(&skel->bss->nr_finite_ticks,
+				       __ATOMIC_RELAXED);
+
+	finite_worker = start_worker(ctx->test_cpu);
+	if (finite_worker < 0) {
+		SCX_ERR("Failed to start second finite-slice worker (%d)", errno);
+		goto out;
+	}
+	if (!wait_for_counter(&skel->bss->nr_finite_running,
+			      finite_running + 1, PHASE_TIMEOUT_MS)) {
+		SCX_ERR("Second finite-slice worker was not scheduled");
+		goto out;
+	}
+	if (!wait_for_counter(&skel->bss->nr_finite_ticks,
+			      finite_ticks + MIN_FINITE_TICKS,
+			      PHASE_TIMEOUT_MS)) {
+		SCX_ERR("Second finite-slice worker received only %llu scheduler ticks",
+			(unsigned long long)(skel->bss->nr_finite_ticks -
+					     finite_ticks));
+		goto out;
+	}
+
+	if (skel->data->uei.kind != EXIT_KIND(SCX_EXIT_NONE)) {
+		SCX_ERR("Scheduler exited unexpectedly (kind=%llu code=%lld)",
+			(unsigned long long)skel->data->uei.kind,
+			(long long)skel->data->uei.exit_code);
+		goto out;
+	}
+
+	fprintf(stderr, "CPU %d received %llu finite-slice ticks\n",
+		ctx->test_cpu,
+		(unsigned long long)skel->bss->nr_finite_ticks);
+	status = SCX_TEST_PASS;
+out:
+	stop_worker(finite_worker);
+	stop_worker(inf_worker);
+	if (link)
+		bpf_link__destroy(link);
+	return status;
+}
+
+static void cleanup(void *ctx_ptr)
+{
+	struct nohz_tick_ctx *ctx = ctx_ptr;
+
+	sched_setaffinity(0, sizeof(ctx->original_mask), &ctx->original_mask);
+	nohz_tick__destroy(ctx->skel);
+	free(ctx);
+}
+
+struct scx_test nohz_tick = {
+	.name = "nohz_tick",
+	.description = "Verify finite EXT slices restart the NOHZ_FULL tick",
+	.setup = setup,
+	.run = run,
+	.cleanup = cleanup,
+};
+REGISTER_SCX_TEST(&nohz_tick)