rcu: Invoke rcu_spawn_core_kthreads() from rcu_spawn_gp_kthread()
authorPaul E. McKenney <paulmck@kernel.org>
Wed, 31 Mar 2021 17:59:05 +0000 (10:59 -0700)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Wed, 14 Jul 2021 14:56:33 +0000 (16:56 +0200)
commit728f23e53c65a90440fe16b270cdff4401433c38
tree98940e2f1f80b74a3bd43fbbaedee993fbf28947
parent66111dfe22b10284c517674bf69c844e1d8b6e73
rcu: Invoke rcu_spawn_core_kthreads() from rcu_spawn_gp_kthread()

[ Upstream commit 8e4b1d2bc198e34b48fc7cc3a3c5a2fcb269e271 ]

Currently, rcu_spawn_core_kthreads() is invoked via an early_initcall(),
which works, except that rcu_spawn_gp_kthread() is also invoked via an
early_initcall() and rcu_spawn_core_kthreads() relies on adjustments to
kthread_prio that are carried out by rcu_spawn_gp_kthread().  There is
no guaranttee of ordering among early_initcall() handlers, and thus no
guarantee that kthread_prio will be properly checked and range-limited
at the time that rcu_spawn_core_kthreads() needs it.

In most cases, this bug is harmless.  After all, the only reason that
rcu_spawn_gp_kthread() adjusts the value of kthread_prio is if the user
specified a nonsensical value for this boot parameter, which experience
indicates is rare.

Nevertheless, a bug is a bug.  This commit therefore causes the
rcu_spawn_core_kthreads() function to be invoked directly from
rcu_spawn_gp_kthread() after any needed adjustments to kthread_prio have
been carried out.

Fixes: 48d07c04b4cc ("rcu: Enable elimination of Tree-RCU softirq processing")
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
kernel/rcu/tree.c