sched_ext: Fixes for v6.19-rc3

- Fix uninitialized @ret on alloc_percpu() failure leading to ERR_PTR(0).
 
 - Fix PREEMPT_RT warning when bypass load balancer sends IPI to offline
   CPU by using resched_cpu() instead of resched_curr().
 
 - Fix comment referring to renamed function.
 
 - Update scx_show_state.py for scx_root and scx_aborting changes.
 -----BEGIN PGP SIGNATURE-----
 
 iIQEABYKACwWIQTfIjM1kS57o3GsC/uxYfJx3gVYGQUCaVHQAw4cdGpAa2VybmVs
 Lm9yZwAKCRCxYfJx3gVYGdJ2AP9nLMUa5Rw2hpcKCLvPjgkqe5fDpNteWrQB3ni9
 bu28jQD/XGMwooJbATlDEgCtFqCH74QbddqUABJxBw4FE8qpUgw=
 =hp6J
 -----END PGP SIGNATURE-----

Merge tag 'sched_ext-for-6.19-rc3-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/sched_ext

Pull sched_ext fixes from Tejun Heo:

 - Fix uninitialized @ret on alloc_percpu() failure leading to
   ERR_PTR(0)

 - Fix PREEMPT_RT warning when bypass load balancer sends IPI to offline
   CPU by using resched_cpu() instead of resched_curr()

 - Fix comment referring to renamed function

 - Update scx_show_state.py for scx_root and scx_aborting changes

* tag 'sched_ext-for-6.19-rc3-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/sched_ext:
  tools/sched_ext: update scx_show_state.py for scx_aborting change
  tools/sched_ext: fix scx_show_state.py for scx_root change
  sched_ext: Use the resched_cpu() to replace resched_curr() in the bypass_lb_node()
  sched_ext: Fix some comments in ext.c
  sched_ext: fix uninitialized ret on alloc_percpu() failure

crypto/seqiv.c

@@ -50,6 +50,7 @@ static int seqiv_aead_encrypt(struct aead_request *req)
 	struct aead_geniv_ctx *ctx = crypto_aead_ctx(geniv);
 	struct aead_request *subreq = aead_request_ctx(req);
 	crypto_completion_t compl;
+	bool unaligned_info;
 	void *data;
 	u8 *info;
 	unsigned int ivsize = 8;
@@ -68,8 +69,9 @@ static int seqiv_aead_encrypt(struct aead_request *req)
 		memcpy_sglist(req->dst, req->src,
 			      req->assoclen + req->cryptlen);
 
-	if (unlikely(!IS_ALIGNED((unsigned long)info,
-				 crypto_aead_alignmask(geniv) + 1))) {
+	unaligned_info = !IS_ALIGNED((unsigned long)info,
+				     crypto_aead_alignmask(geniv) + 1);
+	if (unlikely(unaligned_info)) {
 		info = kmemdup(req->iv, ivsize, req->base.flags &
 			       CRYPTO_TFM_REQ_MAY_SLEEP ? GFP_KERNEL :
 			       GFP_ATOMIC);
@@ -89,7 +91,7 @@ static int seqiv_aead_encrypt(struct aead_request *req)
 	scatterwalk_map_and_copy(info, req->dst, req->assoclen, ivsize, 1);
 
 	err = crypto_aead_encrypt(subreq);
-	if (unlikely(info != req->iv))
+	if (unlikely(unaligned_info))
 		seqiv_aead_encrypt_complete2(req, err);
 	return err;
 }

drivers/crypto/hisilicon/qm.c

@@ -991,7 +991,7 @@ static void qm_get_complete_eqe_num(struct hisi_qm *qm)
 		return;
 	poll_data = &qm->poll_data[cqn];
 
-	while (QM_EQE_PHASE(dw0) != qm->status.eqc_phase) {
+	do {
 		poll_data->qp_finish_id[eqe_num] = dw0 & QM_EQE_CQN_MASK;
 		eqe_num++;
 
@@ -1004,11 +1004,10 @@ static void qm_get_complete_eqe_num(struct hisi_qm *qm)
 			qm->status.eq_head++;
 		}
 
-		if (eqe_num == (eq_depth >> 1) - 1)
-			break;
-
 		dw0 = le32_to_cpu(eqe->dw0);
-	}
+		if (QM_EQE_PHASE(dw0) != qm->status.eqc_phase)
+			break;
+	} while (eqe_num < (eq_depth >> 1) - 1);
 
 	poll_data->eqe_num = eqe_num;
 	queue_work(qm->wq, &poll_data->work);

kernel/cgroup/cpuset.c

@@ -1668,7 +1668,14 @@ static int remote_partition_enable(struct cpuset *cs, int new_prs,
 static void remote_partition_disable(struct cpuset *cs, struct tmpmasks *tmp)
 {
 	WARN_ON_ONCE(!is_remote_partition(cs));
-	WARN_ON_ONCE(!cpumask_subset(cs->effective_xcpus, subpartitions_cpus));
+	/*
+	 * When a CPU is offlined, top_cpuset may end up with no available CPUs,
+	 * which should clear subpartitions_cpus. We should not emit a warning for this
+	 * scenario: the hierarchy is updated from top to bottom, so subpartitions_cpus
+	 * may already be cleared when disabling the partition.
+	 */
+	WARN_ON_ONCE(!cpumask_subset(cs->effective_xcpus, subpartitions_cpus) &&
+		     !cpumask_empty(subpartitions_cpus));
 
 	spin_lock_irq(&callback_lock);
 	cs->remote_partition = false;
@@ -3976,8 +3983,9 @@ retry:
 	if (remote || (is_partition_valid(cs) && is_partition_valid(parent)))
 		compute_partition_effective_cpumask(cs, &new_cpus);
 
-	if (remote && cpumask_empty(&new_cpus) &&
-	    partition_is_populated(cs, NULL)) {
+	if (remote && (cpumask_empty(subpartitions_cpus) ||
+			(cpumask_empty(&new_cpus) &&
+			 partition_is_populated(cs, NULL)))) {
 		cs->prs_err = PERR_HOTPLUG;
 		remote_partition_disable(cs, tmp);
 		compute_effective_cpumask(&new_cpus, cs, parent);
@@ -3990,9 +3998,12 @@ retry:
 	 * 1) empty effective cpus but not valid empty partition.
 	 * 2) parent is invalid or doesn't grant any cpus to child
 	 *    partitions.
+	 * 3) subpartitions_cpus is empty.
 	 */
-	if (is_local_partition(cs) && (!is_partition_valid(parent) ||
-				tasks_nocpu_error(parent, cs, &new_cpus)))
+	if (is_local_partition(cs) &&
+	    (!is_partition_valid(parent) ||
+	     tasks_nocpu_error(parent, cs, &new_cpus) ||
+	     cpumask_empty(subpartitions_cpus)))
 		partcmd = partcmd_invalidate;
 	/*
 	 * On the other hand, an invalid partition root may be transitioned

kernel/sched/ext.c

@@ -1577,7 +1577,7 @@ static bool dequeue_task_scx(struct rq *rq, struct task_struct *p, int deq_flags
 	 *
 	 * @p may go through multiple stopping <-> running transitions between
 	 * here and put_prev_task_scx() if task attribute changes occur while
-	 * balance_scx() leaves @rq unlocked. However, they don't contain any
+	 * balance_one() leaves @rq unlocked. However, they don't contain any
 	 * information meaningful to the BPF scheduler and can be suppressed by
 	 * skipping the callbacks if the task is !QUEUED.
 	 */
@@ -2372,7 +2372,7 @@ static void switch_class(struct rq *rq, struct task_struct *next)
 	 * preempted, and it regaining control of the CPU.
 	 *
 	 * ->cpu_release() complements ->cpu_acquire(), which is emitted the
-	 *  next time that balance_scx() is invoked.
+	 *  next time that balance_one() is invoked.
 	 */
 	if (!rq->scx.cpu_released) {
 		if (SCX_HAS_OP(sch, cpu_release)) {
@@ -2478,7 +2478,7 @@ do_pick_task_scx(struct rq *rq, struct rq_flags *rf, bool force_scx)
 	}
 
 	/*
-	 * If balance_scx() is telling us to keep running @prev, replenish slice
+	 * If balance_one() is telling us to keep running @prev, replenish slice
 	 * if necessary and keep running @prev. Otherwise, pop the first one
 	 * from the local DSQ.
 	 */
@@ -3956,13 +3956,8 @@ static void bypass_lb_node(struct scx_sched *sch, int node)
 					     nr_donor_target, nr_target);
 	}
 
-	for_each_cpu(cpu, resched_mask) {
-		struct rq *rq = cpu_rq(cpu);
-
-		raw_spin_rq_lock_irq(rq);
-		resched_curr(rq);
-		raw_spin_rq_unlock_irq(rq);
-	}
+	for_each_cpu(cpu, resched_mask)
+		resched_cpu(cpu);
 
 	for_each_cpu_and(cpu, cpu_online_mask, node_mask) {
 		u32 nr = READ_ONCE(cpu_rq(cpu)->scx.bypass_dsq.nr);
@@ -4025,7 +4020,7 @@ static DEFINE_TIMER(scx_bypass_lb_timer, scx_bypass_lb_timerfn);
  *
  * - ops.dispatch() is ignored.
  *
- * - balance_scx() does not set %SCX_RQ_BAL_KEEP on non-zero slice as slice
+ * - balance_one() does not set %SCX_RQ_BAL_KEEP on non-zero slice as slice
  *   can't be trusted. Whenever a tick triggers, the running task is rotated to
  *   the tail of the queue with core_sched_at touched.
  *
@@ -4783,8 +4778,10 @@ static struct scx_sched *scx_alloc_and_add_sched(struct sched_ext_ops *ops)
 	}
 
 	sch->pcpu = alloc_percpu(struct scx_sched_pcpu);
-	if (!sch->pcpu)
+	if (!sch->pcpu) {
+		ret = -ENOMEM;
 		goto err_free_gdsqs;
+	}
 
 	sch->helper = kthread_run_worker(0, "sched_ext_helper");
 	if (IS_ERR(sch->helper)) {
@@ -6067,7 +6064,7 @@ __bpf_kfunc bool scx_bpf_dsq_move_to_local(u64 dsq_id)
 		/*
 		 * A successfully consumed task can be dequeued before it starts
 		 * running while the CPU is trying to migrate other dispatched
-		 * tasks. Bump nr_tasks to tell balance_scx() to retry on empty
+		 * tasks. Bump nr_tasks to tell balance_one() to retry on empty
 		 * local DSQ.
 		 */
 		dspc->nr_tasks++;

tools/sched_ext/scx_show_state.py

@@ -27,16 +27,18 @@ def read_static_key(name):
 def state_str(state):
     return prog['scx_enable_state_str'][state].string_().decode()
 
-ops = prog['scx_ops']
+root = prog['scx_root']
 enable_state = read_atomic("scx_enable_state_var")
 
-print(f'ops           : {ops.name.string_().decode()}')
+if root:
+    print(f'ops           : {root.ops.name.string_().decode()}')
+else:
+    print('ops           : ')
 print(f'enabled       : {read_static_key("__scx_enabled")}')
 print(f'switching_all : {read_int("scx_switching_all")}')
 print(f'switched_all  : {read_static_key("__scx_switched_all")}')
 print(f'enable_state  : {state_str(enable_state)} ({enable_state})')
-print(f'in_softlockup : {prog["scx_in_softlockup"].value_()}')
-print(f'breather_depth: {read_atomic("scx_breather_depth")}')
+print(f'aborting      : {prog["scx_aborting"].value_()}')
 print(f'bypass_depth  : {prog["scx_bypass_depth"].value_()}')
 print(f'nr_rejected   : {read_atomic("scx_nr_rejected")}')
 print(f'enable_seq    : {read_atomic("scx_enable_seq")}')