[ANNOUNCE] v4.18.16-rt9

From: Sebastian Andrzej Siewior
Date: Mon Oct 29 2018 - 07:59:10 EST

Dear RT folks!

I'm pleased to announce the v4.18.16-rt9 patch set.

Changes since v4.18.16-rt8:

- The RCU fix, which was introduced in v4.18.7-rt5, leads to a lockdep
warning during CPU hotplug. After a discussion with upstream it was
suggested to revert the change that lead to the problem in -RT.

Known issues
- A warning triggered in "rcu_note_context_switch" originated from
SyS_timer_gettime(). The issue was always there, it is now
visible. Reported by Grygorii Strashko and Daniel Wagner.

The delta patch against v4.18.16-rt8 is appended below and can be found here:


You can get this release via the git tree at:

git://git.kernel.org/pub/scm/linux/kernel/git/rt/linux-rt-devel.git v4.18.16-rt9

The RT patch against v4.18.16 can be found here:


The split quilt queue is available at:



diff --git a/kernel/rcu/tree_exp.h b/kernel/rcu/tree_exp.h
index a104cf91e6b90..d40708e8c5d6e 100644
--- a/kernel/rcu/tree_exp.h
+++ b/kernel/rcu/tree_exp.h
@@ -472,14 +472,12 @@ static void sync_rcu_exp_select_node_cpus(struct work_struct *wp)
static void sync_rcu_exp_select_cpus(struct rcu_state *rsp,
smp_call_func_t func)
- int cpu;
struct rcu_node *rnp;

trace_rcu_exp_grace_period(rsp->name, rcu_exp_gp_seq_endval(rsp), TPS("reset"));
trace_rcu_exp_grace_period(rsp->name, rcu_exp_gp_seq_endval(rsp), TPS("select"));

- cpus_read_lock();
/* Schedule work for each leaf rcu_node structure. */
rcu_for_each_leaf_node(rsp, rnp) {
rnp->exp_need_flush = false;
@@ -494,11 +492,7 @@ static void sync_rcu_exp_select_cpus(struct rcu_state *rsp,
INIT_WORK(&rnp->rew.rew_work, sync_rcu_exp_select_node_cpus);
- cpu = cpumask_next(rnp->grplo - 1, cpu_online_mask);
- /* If all offline, queue the work on an unbound CPU. */
- if (unlikely(cpu > rnp->grphi))
- queue_work_on(cpu, rcu_par_gp_wq, &rnp->rew.rew_work);
+ queue_work_on(rnp->grplo, rcu_par_gp_wq, &rnp->rew.rew_work);
rnp->exp_need_flush = true;

@@ -506,7 +500,6 @@ static void sync_rcu_exp_select_cpus(struct rcu_state *rsp,
rcu_for_each_leaf_node(rsp, rnp)
if (rnp->exp_need_flush)
- cpus_read_unlock();

static void synchronize_sched_expedited_wait(struct rcu_state *rsp)
diff --git a/localversion-rt b/localversion-rt
index 700c857efd9ba..22746d6390a42 100644
--- a/localversion-rt
+++ b/localversion-rt
@@ -1 +1 @@