Received-SPF: pass (google.com: domain of linux-kernel+bounces-195441-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:40f1:3f00::1 as permitted sender) client-ip=2604:1380:40f1:3f00::1;
From: Frederic Weisbecker <frederic@kernel.org>
To: LKML <linux-kernel@vger.kernel.org>
Cc: Frederic Weisbecker <frederic@kernel.org>,
	Boqun Feng <boqun.feng@gmail.com>,
	Joel Fernandes <joel@joelfernandes.org>,
	Neeraj Upadhyay <neeraj.upadhyay@amd.com>,
	"Paul E . McKenney" <paulmck@kernel.org>,
	Uladzislau Rezki <urezki@gmail.com>,
	Zqiang <qiang.zhang1211@gmail.com>,
	rcu <rcu@vger.kernel.org>
Subject: [PATCH 03/11] rcu/nocb: Assert no callbacks while nocb kthread allocation fails
Date: Thu, 30 May 2024 15:45:44 +0200
Message-ID: <20240530134552.5467-4-frederic@kernel.org>
In-Reply-To: <20240530134552.5467-1-frederic@kernel.org>
References: <20240530134552.5467-1-frederic@kernel.org>
Precedence: bulk
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit

When a NOCB CPU fails to create a nocb kthread on bringup, the CPU is
then deoffloaded. The barrier mutex is locked at this stage. It is
typically used to protect against concurrent (de-)offloading and/or
concurrent rcu_barrier() that would otherwise risk a nocb locking
imbalance. However:

* rcu_barrier() can't run concurrently if it's the boot CPU on early
  boot-up.

* rcu_barrier() can run concurrently if it's a secondary CPU but it is
  expected to see 0 callbacks on this target because it's the first
  time it boots.

* (de-)offloading can't happen concurrently with smp_init(), as
  rcutorture is initialized later, at least not before device_initcall(),
  and userspace isn't available yet.

* (de-)offloading can't happen concurrently with cpu_up(), courtesy of
  cpu_hotplug_lock.

But:

* The lazy shrinker might run concurrently with cpu_up(). It shouldn't
  try to grab the nocb_lock and risk an imbalance due to lazy_len
  supposed to be 0 but be extra cautious.

* Also be cautious against resume from hibernation potential subtleties.

So keep the locking and add some assertions and comments.

Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
---
 kernel/rcu/tree_nocb.h | 14 +++++++++++---
 1 file changed, 11 insertions(+), 3 deletions(-)

diff --git a/kernel/rcu/tree_nocb.h b/kernel/rcu/tree_nocb.h
index f4112fc663a7..fdd0616f2fd1 100644
--- a/kernel/rcu/tree_nocb.h
+++ b/kernel/rcu/tree_nocb.h
@@ -1442,7 +1442,7 @@ static void rcu_spawn_cpu_nocb_kthread(int cpu)
 				"rcuog/%d", rdp_gp->cpu);
 		if (WARN_ONCE(IS_ERR(t), "%s: Could not start rcuo GP kthread, OOM is now expected behavior\n", __func__)) {
 			mutex_unlock(&rdp_gp->nocb_gp_kthread_mutex);
-			goto end;
+			goto err;
 		}
 		WRITE_ONCE(rdp_gp->nocb_gp_kthread, t);
 		if (kthread_prio)
@@ -1454,7 +1454,7 @@ static void rcu_spawn_cpu_nocb_kthread(int cpu)
 	t = kthread_create(rcu_nocb_cb_kthread, rdp,
 			   "rcuo%c/%d", rcu_state.abbr, cpu);
 	if (WARN_ONCE(IS_ERR(t), "%s: Could not start rcuo CB kthread, OOM is now expected behavior\n", __func__))
-		goto end;
+		goto err;
 
 	if (rcu_rdp_is_offloaded(rdp))
 		wake_up_process(t);
@@ -1467,7 +1467,15 @@ static void rcu_spawn_cpu_nocb_kthread(int cpu)
 	WRITE_ONCE(rdp->nocb_cb_kthread, t);
 	WRITE_ONCE(rdp->nocb_gp_kthread, rdp_gp->nocb_gp_kthread);
 	return;
-end:
+
+err:
+	/*
+	 * No need to protect against concurrent rcu_barrier()
+	 * because the number of callbacks should be 0 for a non-boot CPU,
+	 * therefore rcu_barrier() shouldn't even try to grab the nocb_lock.
+	 * But hold barrier_mutex to avoid nocb_lock imbalance from shrinker.
+	 */
+	WARN_ON_ONCE(system_state > SYSTEM_BOOTING && rcu_segcblist_n_cbs(&rdp->cblist));
 	mutex_lock(&rcu_state.barrier_mutex);
 	if (rcu_rdp_is_offloaded(rdp)) {
 		rcu_nocb_rdp_deoffload(rdp);
-- 
2.45.1