[PATCH rcu 0/18] RCU no-CBs CPU updates for v5.17

[PATCH rcu 0/18] RCU no-CBs CPU updates for v5.17

[Paul E. McKenney]

Hello!

This series provides RCU no-CB CPUs updates, most notably the ability
to offload CPUs that are in de-offloaded state at boot time.

1.	Tighten rcu_advance_cbs_nowake() checks.

2.	Make local rcu_nocb_lock_irqsave() safe against concurrent
	deoffloading, courtesy of Frederic Weisbecker.

3.	Prepare state machine for a new step, courtesy of Frederic
	Weisbecker.

4.	Invoke rcu_core() at the start of deoffloading, courtesy of
	Frederic Weisbecker.

5.	Make rcu_core() callbacks acceleration preempt-safe, courtesy
	of Thomas Gleixner.

6.	Make rcu_core() callbacks acceleration (de-)offloading safe,
	courtesy of Frederic Weisbecker.

7.	Check a stable offloaded state to manipulate qlen_last_fqs_check,
	courtesy of Frederic Weisbecker.

8.	Use appropriate rcu_nocb_lock_irqsave(), courtesy of Frederic
	Weisbecker.

9.	Limit number of softirq callbacks only on softirq, courtesy of
	Frederic Weisbecker.

10.	Fix callbacks processing time limit retaining cond_resched(),
	courtesy of Frederic Weisbecker.

11.	Apply callbacks processing time limit only on softirq, courtesy
	of Frederic Weisbecker.

12.	Don't invoke local rcu core on callback overload from nocb
	kthread, courtesy of Frederic Weisbecker.

13.	Remove rcu_node structure from nocb list when de-offloaded,
	courtesy of Frederic Weisbecker.

14.	Prepare nocb_cb_wait() to start with a non-offloaded rdp,
	courtesy of Frederic Weisbecker.

15.	Optimize kthreads and rdp initialization, courtesy of Frederic
	Weisbecker.

16.	Create kthreads on all CPUs if "rcu_nocbs=" or "nohz_full="
	are passed, courtesy of Frederic Weisbecker.

17.	Allow empty "rcu_nocbs" kernel parameter, courtesy of Frederic
	Weisbecker.

18.	Merge rcu_spawn_cpu_nocb_kthread() and
	rcu_spawn_one_nocb_kthread(), courtesy of Frederic Weisbecker.

Note that #17 might be updated given some ongoing work by Yury Norov
to support "none" for bitmaps, including the cpumask taken by the
rcu_nocbs kernel-boot parameter.

						Thanx, Paul

------------------------------------------------------------------------

 b/Documentation/admin-guide/kernel-parameters.txt |   37 +++--
 b/include/linux/rcu_segcblist.h                   |   37 +++--
 b/kernel/rcu/rcu_segcblist.c                      |    6 
 b/kernel/rcu/rcu_segcblist.h                      |   12 +
 b/kernel/rcu/tree.c                               |    7 -
 b/kernel/rcu/tree.h                               |   16 +-
 b/kernel/rcu/tree_nocb.h                          |   24 ++-
 include/linux/rcu_segcblist.h                     |   14 ++
 kernel/rcu/rcu_segcblist.c                        |    6 
 kernel/rcu/tree.c                                 |  117 +++++++++++-------
 kernel/rcu/tree.h                                 |    7 -
 kernel/rcu/tree_nocb.h                            |  140 ++++++++++++++--------
 12 files changed, 275 insertions(+), 148 deletions(-)

[PATCH rcu 01/18] rcu: Tighten rcu_advance_cbs_nowake() checks

[Paul E. McKenney]

Currently, rcu_advance_cbs_nowake() checks that a grace period is in
progress, however, that grace period could end just after the check.
This commit rechecks that a grace period is still in progress the lock.
The grace period cannot end while the current CPU's rcu_node structure's
->lock is held, thus avoiding false positives from the WARN_ON_ONCE().

As Daniel Vacek noted, it is not necessary for the rcu_node structure
to have a CPU that has not yet passed through its quiescent state.

Tested-By: Guillaume Morin <guillaume@morinfr.org>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
---
 kernel/rcu/tree.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
index ef8d36f580fc3..8706b30c2ac88 100644
--- a/kernel/rcu/tree.c
+++ b/kernel/rcu/tree.c
@@ -1590,10 +1590,11 @@ static void __maybe_unused rcu_advance_cbs_nowake(struct rcu_node *rnp,
 						  struct rcu_data *rdp)
 {
 	rcu_lockdep_assert_cblist_protected(rdp);
-	if (!rcu_seq_state(rcu_seq_current(&rnp->gp_seq)) ||
-	    !raw_spin_trylock_rcu_node(rnp))
+	if (!rcu_seq_state(rcu_seq_current(&rnp->gp_seq)) || !raw_spin_trylock_rcu_node(rnp))
 		return;
-	WARN_ON_ONCE(rcu_advance_cbs(rnp, rdp));
+	// The grace period cannot end while we hold the rcu_node lock.
+	if (rcu_seq_state(rcu_seq_current(&rnp->gp_seq)))
+		WARN_ON_ONCE(rcu_advance_cbs(rnp, rdp));
 	raw_spin_unlock_rcu_node(rnp);
 }
 
-- 
2.31.1.189.g2e36527f23

[PATCH rcu 02/18] rcu/nocb: Make local rcu_nocb_lock_irqsave() safe against concurrent deoffloading

[Paul E. McKenney]

From: Frederic Weisbecker <frederic@kernel.org>

rcu_nocb_lock_irqsave() can be preempted between the call to
rcu_segcblist_is_offloaded() and the actual locking. This matters now
that rcu_core() is preemptible on PREEMPT_RT and the (de-)offloading
process can interrupt the softirq or the rcuc kthread.

As a result we may locklessly call into code that requires nocb locking.
In practice this is a problem while we accelerate callbacks on rcu_core().

Simply disabling interrupts before (instead of after) checking the NOCB
offload state fixes the issue.

Reported-and-tested-by: Valentin Schneider <valentin.schneider@arm.com>
Tested-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Cc: Valentin Schneider <valentin.schneider@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Joel Fernandes <joel@joelfernandes.org>
Cc: Boqun Feng <boqun.feng@gmail.com>
Cc: Neeraj Upadhyay <neeraju@codeaurora.org>
Cc: Uladzislau Rezki <urezki@gmail.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
---
 kernel/rcu/tree.h | 16 ++++++++++------
 1 file changed, 10 insertions(+), 6 deletions(-)

diff --git a/kernel/rcu/tree.h b/kernel/rcu/tree.h
index 305cf6aeb4086..4f6c67b3ccd5e 100644
--- a/kernel/rcu/tree.h
+++ b/kernel/rcu/tree.h
@@ -447,12 +447,16 @@ static void rcu_nocb_unlock_irqrestore(struct rcu_data *rdp,
 static void rcu_lockdep_assert_cblist_protected(struct rcu_data *rdp);
 #ifdef CONFIG_RCU_NOCB_CPU
 static void __init rcu_organize_nocb_kthreads(void);
-#define rcu_nocb_lock_irqsave(rdp, flags)				\
-do {									\
-	if (!rcu_segcblist_is_offloaded(&(rdp)->cblist))		\
-		local_irq_save(flags);					\
-	else								\
-		raw_spin_lock_irqsave(&(rdp)->nocb_lock, (flags));	\
+
+/*
+ * Disable IRQs before checking offloaded state so that local
+ * locking is safe against concurrent de-offloading.
+ */
+#define rcu_nocb_lock_irqsave(rdp, flags)			\
+do {								\
+	local_irq_save(flags);					\
+	if (rcu_segcblist_is_offloaded(&(rdp)->cblist))	\
+		raw_spin_lock(&(rdp)->nocb_lock);		\
 } while (0)
 #else /* #ifdef CONFIG_RCU_NOCB_CPU */
 #define rcu_nocb_lock_irqsave(rdp, flags) local_irq_save(flags)
-- 
2.31.1.189.g2e36527f23

[PATCH rcu 03/18] rcu/nocb: Prepare state machine for a new step

[Paul E. McKenney]

From: Frederic Weisbecker <frederic@kernel.org>

Currently SEGCBLIST_SOFTIRQ_ONLY is a bit of an exception among the
segcblist flags because it is an exclusive state that doesn't mix up
with the other flags. Remove it in favour of:

_ A flag specifying that rcu_core() needs to perform callbacks execution
  and acceleration

and

_ A flag specifying we want the nocb lock to be held in any needed
  circumstances

This clarifies the code and is more flexible: It allows to have a state
where rcu_core() runs with locking while offloading hasn't started yet.
This is a necessary step to prepare for triggering rcu_core() at the
very beginning of the de-offloading process so that rcu_core() won't
dismiss work while being preempted by the de-offloading process, at
least not without a pending subsequent rcu_core() that will quickly
catch up.

Reviewed-by: Valentin Schneider <Valentin.Schneider@arm.com>
Tested-by: Valentin Schneider <valentin.schneider@arm.com>
Tested-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Joel Fernandes <joel@joelfernandes.org>
Cc: Boqun Feng <boqun.feng@gmail.com>
Cc: Neeraj Upadhyay <neeraju@codeaurora.org>
Cc: Uladzislau Rezki <urezki@gmail.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
---
 include/linux/rcu_segcblist.h | 37 ++++++++++++++++++++++-------------
 kernel/rcu/rcu_segcblist.c    |  6 +++---
 kernel/rcu/rcu_segcblist.h    | 12 +++++++-----
 kernel/rcu/tree.c             |  2 +-
 kernel/rcu/tree_nocb.h        | 24 +++++++++++++++--------
 5 files changed, 50 insertions(+), 31 deletions(-)

diff --git a/include/linux/rcu_segcblist.h b/include/linux/rcu_segcblist.h
index 3db96c4f45fd4..812961b1d0642 100644
--- a/include/linux/rcu_segcblist.h
+++ b/include/linux/rcu_segcblist.h
@@ -69,7 +69,7 @@ struct rcu_cblist {
  *
  *
  *  ----------------------------------------------------------------------------
- *  |                         SEGCBLIST_SOFTIRQ_ONLY                           |
+ *  |                              SEGCBLIST_RCU_CORE                          |
  *  |                                                                          |
  *  |  Callbacks processed by rcu_core() from softirqs or local                |
  *  |  rcuc kthread, without holding nocb_lock.                                |
@@ -77,7 +77,7 @@ struct rcu_cblist {
  *                                         |
  *                                         v
  *  ----------------------------------------------------------------------------
- *  |                        SEGCBLIST_OFFLOADED                               |
+ *  |       SEGCBLIST_RCU_CORE | SEGCBLIST_LOCKING | SEGCBLIST_OFFLOADED       |
  *  |                                                                          |
  *  | Callbacks processed by rcu_core() from softirqs or local                 |
  *  | rcuc kthread, while holding nocb_lock. Waking up CB and GP kthreads,     |
@@ -89,7 +89,9 @@ struct rcu_cblist {
  *                        |                                 |
  *                        v                                 v
  *  ---------------------------------------  ----------------------------------|
- *  |        SEGCBLIST_OFFLOADED |        |  |     SEGCBLIST_OFFLOADED |       |
+ *  |        SEGCBLIST_RCU_CORE   |       |  |     SEGCBLIST_RCU_CORE   |      |
+ *  |        SEGCBLIST_LOCKING    |       |  |     SEGCBLIST_LOCKING    |      |
+ *  |        SEGCBLIST_OFFLOADED  |       |  |     SEGCBLIST_OFFLOADED  |      |
  *  |        SEGCBLIST_KTHREAD_CB         |  |     SEGCBLIST_KTHREAD_GP        |
  *  |                                     |  |                                 |
  *  |                                     |  |                                 |
@@ -104,9 +106,10 @@ struct rcu_cblist {
  *                                         |
  *                                         v
  *  |--------------------------------------------------------------------------|
- *  |                           SEGCBLIST_OFFLOADED |                          |
- *  |                           SEGCBLIST_KTHREAD_CB |                         |
- *  |                           SEGCBLIST_KTHREAD_GP                           |
+ *  |                           SEGCBLIST_LOCKING    |                         |
+ *  |                           SEGCBLIST_OFFLOADED  |                         |
+ *  |                           SEGCBLIST_KTHREAD_GP |                         |
+ *  |                           SEGCBLIST_KTHREAD_CB                           |
  *  |                                                                          |
  *  |   Kthreads handle callbacks holding nocb_lock, local rcu_core() stops    |
  *  |   handling callbacks. Enable bypass queueing.                            |
@@ -120,7 +123,8 @@ struct rcu_cblist {
  *
  *
  *  |--------------------------------------------------------------------------|
- *  |                           SEGCBLIST_OFFLOADED |                          |
+ *  |                           SEGCBLIST_LOCKING    |                         |
+ *  |                           SEGCBLIST_OFFLOADED  |                         |
  *  |                           SEGCBLIST_KTHREAD_CB |                         |
  *  |                           SEGCBLIST_KTHREAD_GP                           |
  *  |                                                                          |
@@ -130,6 +134,8 @@ struct rcu_cblist {
  *                                      |
  *                                      v
  *  |--------------------------------------------------------------------------|
+ *  |                           SEGCBLIST_RCU_CORE   |                         |
+ *  |                           SEGCBLIST_LOCKING    |                         |
  *  |                           SEGCBLIST_KTHREAD_CB |                         |
  *  |                           SEGCBLIST_KTHREAD_GP                           |
  *  |                                                                          |
@@ -143,7 +149,9 @@ struct rcu_cblist {
  *                     |                                 |
  *                     v                                 v
  *  ---------------------------------------------------------------------------|
- *  |                                                                          |
+ *  |                                     |                                    |
+ *  |        SEGCBLIST_RCU_CORE |         |       SEGCBLIST_RCU_CORE |         |
+ *  |        SEGCBLIST_LOCKING  |         |       SEGCBLIST_LOCKING  |         |
  *  |        SEGCBLIST_KTHREAD_CB         |       SEGCBLIST_KTHREAD_GP         |
  *  |                                     |                                    |
  *  | GP kthread woke up and              |   CB kthread woke up and           |
@@ -159,7 +167,7 @@ struct rcu_cblist {
  *                                      |
  *                                      v
  *  ----------------------------------------------------------------------------
- *  |                                   0                                      |
+ *  |                SEGCBLIST_RCU_CORE | SEGCBLIST_LOCKING                    |
  *  |                                                                          |
  *  | Callbacks processed by rcu_core() from softirqs or local                 |
  *  | rcuc kthread, while holding nocb_lock. Forbid nocb_timer to be armed.    |
@@ -168,17 +176,18 @@ struct rcu_cblist {
  *                                      |
  *                                      v
  *  ----------------------------------------------------------------------------
- *  |                         SEGCBLIST_SOFTIRQ_ONLY                           |
+ *  |                         SEGCBLIST_RCU_CORE                               |
  *  |                                                                          |
  *  |  Callbacks processed by rcu_core() from softirqs or local                |
  *  |  rcuc kthread, without holding nocb_lock.                                |
  *  ----------------------------------------------------------------------------
  */
 #define SEGCBLIST_ENABLED	BIT(0)
-#define SEGCBLIST_SOFTIRQ_ONLY	BIT(1)
-#define SEGCBLIST_KTHREAD_CB	BIT(2)
-#define SEGCBLIST_KTHREAD_GP	BIT(3)
-#define SEGCBLIST_OFFLOADED	BIT(4)
+#define SEGCBLIST_RCU_CORE	BIT(1)
+#define SEGCBLIST_LOCKING	BIT(2)
+#define SEGCBLIST_KTHREAD_CB	BIT(3)
+#define SEGCBLIST_KTHREAD_GP	BIT(4)
+#define SEGCBLIST_OFFLOADED	BIT(5)
 
 struct rcu_segcblist {
 	struct rcu_head *head;
diff --git a/kernel/rcu/rcu_segcblist.c b/kernel/rcu/rcu_segcblist.c
index aaa111237b602..c07aab6e39ef2 100644
--- a/kernel/rcu/rcu_segcblist.c
+++ b/kernel/rcu/rcu_segcblist.c
@@ -261,14 +261,14 @@ void rcu_segcblist_disable(struct rcu_segcblist *rsclp)
 }
 
 /*
- * Mark the specified rcu_segcblist structure as offloaded.
+ * Mark the specified rcu_segcblist structure as offloaded (or not)
  */
 void rcu_segcblist_offload(struct rcu_segcblist *rsclp, bool offload)
 {
 	if (offload) {
-		rcu_segcblist_clear_flags(rsclp, SEGCBLIST_SOFTIRQ_ONLY);
-		rcu_segcblist_set_flags(rsclp, SEGCBLIST_OFFLOADED);
+		rcu_segcblist_set_flags(rsclp, SEGCBLIST_LOCKING | SEGCBLIST_OFFLOADED);
 	} else {
+		rcu_segcblist_set_flags(rsclp, SEGCBLIST_RCU_CORE);
 		rcu_segcblist_clear_flags(rsclp, SEGCBLIST_OFFLOADED);
 	}
 }
diff --git a/kernel/rcu/rcu_segcblist.h b/kernel/rcu/rcu_segcblist.h
index 9a19328ff2514..e373fbe44da5e 100644
--- a/kernel/rcu/rcu_segcblist.h
+++ b/kernel/rcu/rcu_segcblist.h
@@ -80,11 +80,14 @@ static inline bool rcu_segcblist_is_enabled(struct rcu_segcblist *rsclp)
 	return rcu_segcblist_test_flags(rsclp, SEGCBLIST_ENABLED);
 }
 
-/* Is the specified rcu_segcblist offloaded, or is SEGCBLIST_SOFTIRQ_ONLY set? */
+/*
+ * Is the specified rcu_segcblist NOCB offloaded (or in the middle of the
+ * [de]offloading process)?
+ */
 static inline bool rcu_segcblist_is_offloaded(struct rcu_segcblist *rsclp)
 {
 	if (IS_ENABLED(CONFIG_RCU_NOCB_CPU) &&
-	    !rcu_segcblist_test_flags(rsclp, SEGCBLIST_SOFTIRQ_ONLY))
+	    rcu_segcblist_test_flags(rsclp, SEGCBLIST_LOCKING))
 		return true;
 
 	return false;
@@ -92,9 +95,8 @@ static inline bool rcu_segcblist_is_offloaded(struct rcu_segcblist *rsclp)
 
 static inline bool rcu_segcblist_completely_offloaded(struct rcu_segcblist *rsclp)
 {
-	int flags = SEGCBLIST_KTHREAD_CB | SEGCBLIST_KTHREAD_GP | SEGCBLIST_OFFLOADED;
-
-	if (IS_ENABLED(CONFIG_RCU_NOCB_CPU) && (rsclp->flags & flags) == flags)
+	if (IS_ENABLED(CONFIG_RCU_NOCB_CPU) &&
+	    !rcu_segcblist_test_flags(rsclp, SEGCBLIST_RCU_CORE))
 		return true;
 
 	return false;
diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
index 8706b30c2ac88..e905d7e4ddb91 100644
--- a/kernel/rcu/tree.c
+++ b/kernel/rcu/tree.c
@@ -79,7 +79,7 @@ static DEFINE_PER_CPU_SHARED_ALIGNED(struct rcu_data, rcu_data) = {
 	.dynticks_nmi_nesting = DYNTICK_IRQ_NONIDLE,
 	.dynticks = ATOMIC_INIT(1),
 #ifdef CONFIG_RCU_NOCB_CPU
-	.cblist.flags = SEGCBLIST_SOFTIRQ_ONLY,
+	.cblist.flags = SEGCBLIST_RCU_CORE,
 #endif
 };
 static struct rcu_state rcu_state = {
diff --git a/kernel/rcu/tree_nocb.h b/kernel/rcu/tree_nocb.h
index 368ef7b9af4fe..b3e07d0bfbbf8 100644
--- a/kernel/rcu/tree_nocb.h
+++ b/kernel/rcu/tree_nocb.h
@@ -1000,12 +1000,12 @@ static long rcu_nocb_rdp_deoffload(void *arg)
 	 */
 	rcu_nocb_lock_irqsave(rdp, flags);
 	/*
-	 * Theoretically we could set SEGCBLIST_SOFTIRQ_ONLY after the nocb
+	 * Theoretically we could clear SEGCBLIST_LOCKING after the nocb
 	 * lock is released but how about being paranoid for once?
 	 */
-	rcu_segcblist_set_flags(cblist, SEGCBLIST_SOFTIRQ_ONLY);
+	rcu_segcblist_clear_flags(cblist, SEGCBLIST_LOCKING);
 	/*
-	 * With SEGCBLIST_SOFTIRQ_ONLY, we can't use
+	 * Without SEGCBLIST_LOCKING, we can't use
 	 * rcu_nocb_unlock_irqrestore() anymore.
 	 */
 	raw_spin_unlock_irqrestore(&rdp->nocb_lock, flags);
@@ -1058,14 +1058,14 @@ static long rcu_nocb_rdp_offload(void *arg)
 
 	pr_info("Offloading %d\n", rdp->cpu);
 	/*
-	 * Can't use rcu_nocb_lock_irqsave() while we are in
-	 * SEGCBLIST_SOFTIRQ_ONLY mode.
+	 * Can't use rcu_nocb_lock_irqsave() before SEGCBLIST_LOCKING
+	 * is set.
 	 */
 	raw_spin_lock_irqsave(&rdp->nocb_lock, flags);
 
 	/*
 	 * We didn't take the nocb lock while working on the
-	 * rdp->cblist in SEGCBLIST_SOFTIRQ_ONLY mode.
+	 * rdp->cblist with SEGCBLIST_LOCKING cleared (pure softirq/rcuc mode).
 	 * Every modifications that have been done previously on
 	 * rdp->cblist must be visible remotely by the nocb kthreads
 	 * upon wake up after reading the cblist flags.
@@ -1084,6 +1084,14 @@ static long rcu_nocb_rdp_offload(void *arg)
 			      rcu_segcblist_test_flags(cblist, SEGCBLIST_KTHREAD_CB) &&
 			      rcu_segcblist_test_flags(cblist, SEGCBLIST_KTHREAD_GP));
 
+	/*
+	 * All kthreads are ready to work, we can finally relieve rcu_core() and
+	 * enable nocb bypass.
+	 */
+	rcu_nocb_lock_irqsave(rdp, flags);
+	rcu_segcblist_clear_flags(cblist, SEGCBLIST_RCU_CORE);
+	rcu_nocb_unlock_irqrestore(rdp, flags);
+
 	return ret;
 }
 
@@ -1154,8 +1162,8 @@ void __init rcu_init_nohz(void)
 		if (rcu_segcblist_empty(&rdp->cblist))
 			rcu_segcblist_init(&rdp->cblist);
 		rcu_segcblist_offload(&rdp->cblist, true);
-		rcu_segcblist_set_flags(&rdp->cblist, SEGCBLIST_KTHREAD_CB);
-		rcu_segcblist_set_flags(&rdp->cblist, SEGCBLIST_KTHREAD_GP);
+		rcu_segcblist_set_flags(&rdp->cblist, SEGCBLIST_KTHREAD_CB | SEGCBLIST_KTHREAD_GP);
+		rcu_segcblist_clear_flags(&rdp->cblist, SEGCBLIST_RCU_CORE);
 	}
 	rcu_organize_nocb_kthreads();
 }
-- 
2.31.1.189.g2e36527f23

[PATCH rcu 04/18] rcu/nocb: Invoke rcu_core() at the start of deoffloading

[Paul E. McKenney]

From: Frederic Weisbecker <frederic@kernel.org>

On PREEMPT_RT, if rcu_core() is preempted by the de-offloading process,
some work, such as callbacks acceleration and invocation, may be left
unattended due to the volatile checks on the offloaded state.

In the worst case this work is postponed until the next rcu_pending()
check that can take a jiffy to reach, which can be a problem in case
of callbacks flooding.

Solve that with invoking rcu_core() early in the de-offloading process.
This way any work dismissed by an ongoing rcu_core() call fooled by
a preempting deoffloading process will be caught up by a nearby future
recall to rcu_core(), this time fully aware of the de-offloading state.

Tested-by: Valentin Schneider <valentin.schneider@arm.com>
Tested-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Cc: Valentin Schneider <valentin.schneider@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Joel Fernandes <joel@joelfernandes.org>
Cc: Boqun Feng <boqun.feng@gmail.com>
Cc: Neeraj Upadhyay <neeraju@codeaurora.org>
Cc: Uladzislau Rezki <urezki@gmail.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
---
 include/linux/rcu_segcblist.h | 14 ++++++++++++++
 kernel/rcu/rcu_segcblist.c    |  6 ++----
 kernel/rcu/tree.c             | 17 +++++++++++++++++
 kernel/rcu/tree_nocb.h        |  9 +++++++++
 4 files changed, 42 insertions(+), 4 deletions(-)

diff --git a/include/linux/rcu_segcblist.h b/include/linux/rcu_segcblist.h
index 812961b1d0642..659d13a7ddaaa 100644
--- a/include/linux/rcu_segcblist.h
+++ b/include/linux/rcu_segcblist.h
@@ -136,6 +136,20 @@ struct rcu_cblist {
  *  |--------------------------------------------------------------------------|
  *  |                           SEGCBLIST_RCU_CORE   |                         |
  *  |                           SEGCBLIST_LOCKING    |                         |
+ *  |                           SEGCBLIST_OFFLOADED  |                         |
+ *  |                           SEGCBLIST_KTHREAD_CB |                         |
+ *  |                           SEGCBLIST_KTHREAD_GP                           |
+ *  |                                                                          |
+ *  |   CB/GP kthreads handle callbacks holding nocb_lock, local rcu_core()    |
+ *  |   handles callbacks concurrently. Bypass enqueue is enabled.             |
+ *  |   Invoke RCU core so we make sure not to preempt it in the middle with   |
+ *  |   leaving some urgent work unattended within a jiffy.                    |
+ *  ----------------------------------------------------------------------------
+ *                                      |
+ *                                      v
+ *  |--------------------------------------------------------------------------|
+ *  |                           SEGCBLIST_RCU_CORE   |                         |
+ *  |                           SEGCBLIST_LOCKING    |                         |
  *  |                           SEGCBLIST_KTHREAD_CB |                         |
  *  |                           SEGCBLIST_KTHREAD_GP                           |
  *  |                                                                          |
diff --git a/kernel/rcu/rcu_segcblist.c b/kernel/rcu/rcu_segcblist.c
index c07aab6e39ef2..81145c3ece25f 100644
--- a/kernel/rcu/rcu_segcblist.c
+++ b/kernel/rcu/rcu_segcblist.c
@@ -265,12 +265,10 @@ void rcu_segcblist_disable(struct rcu_segcblist *rsclp)
  */
 void rcu_segcblist_offload(struct rcu_segcblist *rsclp, bool offload)
 {
-	if (offload) {
+	if (offload)
 		rcu_segcblist_set_flags(rsclp, SEGCBLIST_LOCKING | SEGCBLIST_OFFLOADED);
-	} else {
-		rcu_segcblist_set_flags(rsclp, SEGCBLIST_RCU_CORE);
+	else
 		rcu_segcblist_clear_flags(rsclp, SEGCBLIST_OFFLOADED);
-	}
 }
 
 /*
diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
index e905d7e4ddb91..a329adfece86e 100644
--- a/kernel/rcu/tree.c
+++ b/kernel/rcu/tree.c
@@ -2707,6 +2707,23 @@ static __latent_entropy void rcu_core(void)
 	unsigned long flags;
 	struct rcu_data *rdp = raw_cpu_ptr(&rcu_data);
 	struct rcu_node *rnp = rdp->mynode;
+	/*
+	 * On RT rcu_core() can be preempted when IRQs aren't disabled.
+	 * Therefore this function can race with concurrent NOCB (de-)offloading
+	 * on this CPU and the below condition must be considered volatile.
+	 * However if we race with:
+	 *
+	 * _ Offloading:   In the worst case we accelerate or process callbacks
+	 *                 concurrently with NOCB kthreads. We are guaranteed to
+	 *                 call rcu_nocb_lock() if that happens.
+	 *
+	 * _ Deoffloading: In the worst case we miss callbacks acceleration or
+	 *                 processing. This is fine because the early stage
+	 *                 of deoffloading invokes rcu_core() after setting
+	 *                 SEGCBLIST_RCU_CORE. So we guarantee that we'll process
+	 *                 what could have been dismissed without the need to wait
+	 *                 for the next rcu_pending() check in the next jiffy.
+	 */
 	const bool do_batch = !rcu_segcblist_completely_offloaded(&rdp->cblist);
 
 	if (cpu_is_offline(smp_processor_id()))
diff --git a/kernel/rcu/tree_nocb.h b/kernel/rcu/tree_nocb.h
index b3e07d0bfbbf8..2461fe8d0c23a 100644
--- a/kernel/rcu/tree_nocb.h
+++ b/kernel/rcu/tree_nocb.h
@@ -990,6 +990,15 @@ static long rcu_nocb_rdp_deoffload(void *arg)
 	 * will refuse to put anything into the bypass.
 	 */
 	WARN_ON_ONCE(!rcu_nocb_flush_bypass(rdp, NULL, jiffies));
+	/*
+	 * Start with invoking rcu_core() early. This way if the current thread
+	 * happens to preempt an ongoing call to rcu_core() in the middle,
+	 * leaving some work dismissed because rcu_core() still thinks the rdp is
+	 * completely offloaded, we are guaranteed a nearby future instance of
+	 * rcu_core() to catch up.
+	 */
+	rcu_segcblist_set_flags(cblist, SEGCBLIST_RCU_CORE);
+	invoke_rcu_core();
 	ret = rdp_offload_toggle(rdp, false, flags);
 	swait_event_exclusive(rdp->nocb_state_wq,
 			      !rcu_segcblist_test_flags(cblist, SEGCBLIST_KTHREAD_CB |
-- 
2.31.1.189.g2e36527f23

[PATCH rcu 05/18] rcu/nocb: Make rcu_core() callbacks acceleration preempt-safe

[Paul E. McKenney]

From: Thomas Gleixner <tglx@linutronix.de>

While reporting a quiescent state for a given CPU, rcu_core() takes
advantage of the freshly loaded grace period sequence number and the
locked rnp to accelerate the callbacks whose sequence number have been
assigned a stale value.

This action is only necessary when the rdp isn't offloaded, otherwise
the NOCB kthreads already take care of the callbacks progression.

However the check for the offloaded state is volatile because it is
performed outside the IRQs disabled section. It's possible for the
offloading process to preempt rcu_core() at that point on PREEMPT_RT.

This is dangerous because rcu_core() may end up accelerating callbacks
concurrently with NOCB kthreads without appropriate locking.

Fix this with moving the offloaded check inside the rnp locking section.

Reported-and-tested-by: Valentin Schneider <valentin.schneider@arm.com>
Reviewed-by: Valentin Schneider <valentin.schneider@arm.com>
Tested-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Joel Fernandes <joel@joelfernandes.org>
Cc: Boqun Feng <boqun.feng@gmail.com>
Cc: Neeraj Upadhyay <neeraju@codeaurora.org>
Cc: Uladzislau Rezki <urezki@gmail.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
---
 kernel/rcu/tree.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
index a329adfece86e..5985698f33414 100644
--- a/kernel/rcu/tree.c
+++ b/kernel/rcu/tree.c
@@ -2278,7 +2278,6 @@ rcu_report_qs_rdp(struct rcu_data *rdp)
 	unsigned long flags;
 	unsigned long mask;
 	bool needwake = false;
-	const bool offloaded = rcu_rdp_is_offloaded(rdp);
 	struct rcu_node *rnp;
 
 	WARN_ON_ONCE(rdp->cpu != smp_processor_id());
@@ -2305,8 +2304,10 @@ rcu_report_qs_rdp(struct rcu_data *rdp)
 		/*
 		 * This GP can't end until cpu checks in, so all of our
 		 * callbacks can be processed during the next GP.
+		 *
+		 * NOCB kthreads have their own way to deal with that.
 		 */
-		if (!offloaded)
+		if (!rcu_rdp_is_offloaded(rdp))
 			needwake = rcu_accelerate_cbs(rnp, rdp);
 
 		rcu_disable_urgency_upon_qs(rdp);
-- 
2.31.1.189.g2e36527f23

[PATCH rcu 06/18] rcu/nocb: Make rcu_core() callbacks acceleration (de-)offloading safe

[Paul E. McKenney]

From: Frederic Weisbecker <frederic@kernel.org>

When callbacks are offloaded, the NOCB kthreads handle the callbacks
progression on behalf of rcu_core().

However during the (de-)offloading process, the kthread may not be
entirely up to the task. As a result some callbacks grace period
sequence number may remain stale for a while because rcu_core() won't
take care of them either.

Fix this with forcing callbacks acceleration from rcu_core() as long
as the offloading process isn't complete.

Reported-and-tested-by: Valentin Schneider <valentin.schneider@arm.com>
Tested-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Cc: Valentin Schneider <valentin.schneider@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Joel Fernandes <joel@joelfernandes.org>
Cc: Boqun Feng <boqun.feng@gmail.com>
Cc: Neeraj Upadhyay <neeraju@codeaurora.org>
Cc: Uladzislau Rezki <urezki@gmail.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
---
 kernel/rcu/tree.c | 18 ++++++++++++++++--
 1 file changed, 16 insertions(+), 2 deletions(-)

diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
index 5985698f33414..cb9abb80377fe 100644
--- a/kernel/rcu/tree.c
+++ b/kernel/rcu/tree.c
@@ -2278,6 +2278,7 @@ rcu_report_qs_rdp(struct rcu_data *rdp)
 	unsigned long flags;
 	unsigned long mask;
 	bool needwake = false;
+	bool needacc = false;
 	struct rcu_node *rnp;
 
 	WARN_ON_ONCE(rdp->cpu != smp_processor_id());
@@ -2305,16 +2306,29 @@ rcu_report_qs_rdp(struct rcu_data *rdp)
 		 * This GP can't end until cpu checks in, so all of our
 		 * callbacks can be processed during the next GP.
 		 *
-		 * NOCB kthreads have their own way to deal with that.
+		 * NOCB kthreads have their own way to deal with that...
 		 */
-		if (!rcu_rdp_is_offloaded(rdp))
+		if (!rcu_rdp_is_offloaded(rdp)) {
 			needwake = rcu_accelerate_cbs(rnp, rdp);
+		} else if (!rcu_segcblist_completely_offloaded(&rdp->cblist)) {
+			/*
+			 * ...but NOCB kthreads may miss or delay callbacks acceleration
+			 * if in the middle of a (de-)offloading process.
+			 */
+			needacc = true;
+		}
 
 		rcu_disable_urgency_upon_qs(rdp);
 		rcu_report_qs_rnp(mask, rnp, rnp->gp_seq, flags);
 		/* ^^^ Released rnp->lock */
 		if (needwake)
 			rcu_gp_kthread_wake();
+
+		if (needacc) {
+			rcu_nocb_lock_irqsave(rdp, flags);
+			rcu_accelerate_cbs_unlocked(rnp, rdp);
+			rcu_nocb_unlock_irqrestore(rdp, flags);
+		}
 	}
 }
 
-- 
2.31.1.189.g2e36527f23

[PATCH rcu 07/18] rcu/nocb: Check a stable offloaded state to manipulate qlen_last_fqs_check

[Paul E. McKenney]

From: Frederic Weisbecker <frederic@kernel.org>

It's not entirely obvious why rdp->qlen_last_fqs_check is updated before
processing the queue only on offloaded rdp. There can be different
effect to that, either in favour of triggering the force quiescent state
path or not. For example:

1) If the number of callbacks has decreased since the last
   rdp->qlen_last_fqs_check update (because we recently called
   rcu_do_batch() and we executed below qhimark callbacks) and the number
   of processed callbacks on a subsequent do_batch() arranges for
   exceeding qhimark on non-offloaded but not on offloaded setup, then we
   may spare a later run to the force quiescent state
   slow path on __call_rcu_nocb_wake(), as compared to the non-offloaded
   counterpart scenario.

   Here is such an offloaded scenario instance:

    qhimark = 1000
    rdp->last_qlen_last_fqs_check = 3000
    rcu_segcblist_n_cbs(rdp) = 2000

    rcu_do_batch() {
        if (offloaded)
            rdp->last_qlen_fqs_check = rcu_segcblist_n_cbs(rdp) // 2000
        // run 1000 callback
        rcu_segcblist_n_cbs(rdp) = 1000
        // Not updating rdp->qlen_last_fqs_check
        if (count < rdp->qlen_last_fqs_check - qhimark)
            rdp->qlen_last_fqs_check = count;
    }

    call_rcu() * 1001 {
        __call_rcu_nocb_wake() {
            // not taking the fqs slowpath:
            // rcu_segcblist_n_cbs(rdp) == 2001
            // rdp->qlen_last_fqs_check == 2000
            // qhimark == 1000
            if (len > rdp->qlen_last_fqs_check + qhimark)
                ...
    }

    In the case of a non-offloaded scenario, rdp->qlen_last_fqs_check
    would be 1000 and the fqs slowpath would have executed.

2) If the number of callbacks has increased since the last
   rdp->qlen_last_fqs_check update (because we recently queued below
   qhimark callbacks) and the number of callbacks executed in rcu_do_batch()
   doesn't exceed qhimark for either offloaded or non-offloaded setup,
   then it's possible that the offloaded scenario later run the force
   quiescent state slow path on __call_rcu_nocb_wake() while the
   non-offloaded doesn't.

    qhimark = 1000
    rdp->last_qlen_last_fqs_check = 3000
    rcu_segcblist_n_cbs(rdp) = 2000

    rcu_do_batch() {
        if (offloaded)
            rdp->last_qlen_last_fqs_check = rcu_segcblist_n_cbs(rdp) // 2000
        // run 100 callbacks
        // concurrent queued 100
        rcu_segcblist_n_cbs(rdp) = 2000
        // Not updating rdp->qlen_last_fqs_check
        if (count < rdp->qlen_last_fqs_check - qhimark)
            rdp->qlen_last_fqs_check = count;
    }

    call_rcu() * 1001 {
        __call_rcu_nocb_wake() {
            // Taking the fqs slowpath:
            // rcu_segcblist_n_cbs(rdp) == 3001
            // rdp->qlen_last_fqs_check == 2000
            // qhimark == 1000
            if (len > rdp->qlen_last_fqs_check + qhimark)
                ...
    }

    In the case of a non-offloaded scenario, rdp->qlen_last_fqs_check
    would be 3000 and the fqs slowpath would have executed.

The reason for updating rdp->qlen_last_fqs_check when invoking callbacks
for offloaded CPUs is that there is usually no point in waking up either
the rcuog or rcuoc kthreads while in this state.  After all, both threads
are prohibited from indefinite sleeps.

The exception is when some huge number of callbacks are enqueued while
rcu_do_batch() is in the midst of invoking, in which case interrupting
the rcuog kthread's timed sleep might get more callbacks set up for the
next grace period.

Reported-and-tested-by: Valentin Schneider <valentin.schneider@arm.com>
Tested-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Original-patch-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Cc: Valentin Schneider <valentin.schneider@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Joel Fernandes <joel@joelfernandes.org>
Cc: Boqun Feng <boqun.feng@gmail.com>
Cc: Neeraj Upadhyay <neeraju@codeaurora.org>
Cc: Uladzislau Rezki <urezki@gmail.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
---
 kernel/rcu/tree.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
index cb9abb80377fe..4cbfc4e4fa9ed 100644
--- a/kernel/rcu/tree.c
+++ b/kernel/rcu/tree.c
@@ -2498,7 +2498,7 @@ static void rcu_do_batch(struct rcu_data *rdp)
 	trace_rcu_batch_start(rcu_state.name,
 			      rcu_segcblist_n_cbs(&rdp->cblist), bl);
 	rcu_segcblist_extract_done_cbs(&rdp->cblist, &rcl);
-	if (offloaded)
+	if (rcu_rdp_is_offloaded(rdp))
 		rdp->qlen_last_fqs_check = rcu_segcblist_n_cbs(&rdp->cblist);
 
 	trace_rcu_segcb_stats(&rdp->cblist, TPS("SegCbDequeued"));
-- 
2.31.1.189.g2e36527f23

[PATCH rcu 08/18] rcu/nocb: Use appropriate rcu_nocb_lock_irqsave()

[Paul E. McKenney]

From: Frederic Weisbecker <frederic@kernel.org>

Instead of hardcoding IRQ save and nocb lock, use the consolidated
API (and fix a comment as per Valentin Schneider's suggestion).

Reviewed-by: Valentin Schneider <valentin.schneider@arm.com>
Tested-by: Valentin Schneider <valentin.schneider@arm.com>
Tested-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Cc: Valentin Schneider <valentin.schneider@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Joel Fernandes <joel@joelfernandes.org>
Cc: Boqun Feng <boqun.feng@gmail.com>
Cc: Neeraj Upadhyay <neeraju@codeaurora.org>
Cc: Uladzislau Rezki <urezki@gmail.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
---
 kernel/rcu/tree.c | 8 +++-----
 1 file changed, 3 insertions(+), 5 deletions(-)

diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
index 4cbfc4e4fa9ed..20587d035d03b 100644
--- a/kernel/rcu/tree.c
+++ b/kernel/rcu/tree.c
@@ -2478,12 +2478,11 @@ static void rcu_do_batch(struct rcu_data *rdp)
 	}
 
 	/*
-	 * Extract the list of ready callbacks, disabling to prevent
+	 * Extract the list of ready callbacks, disabling IRQs to prevent
 	 * races with call_rcu() from interrupt handlers.  Leave the
 	 * callback counts, as rcu_barrier() needs to be conservative.
 	 */
-	local_irq_save(flags);
-	rcu_nocb_lock(rdp);
+	rcu_nocb_lock_irqsave(rdp, flags);
 	WARN_ON_ONCE(cpu_is_offline(smp_processor_id()));
 	pending = rcu_segcblist_n_cbs(&rdp->cblist);
 	div = READ_ONCE(rcu_divisor);
@@ -2546,8 +2545,7 @@ static void rcu_do_batch(struct rcu_data *rdp)
 		}
 	}
 
-	local_irq_save(flags);
-	rcu_nocb_lock(rdp);
+	rcu_nocb_lock_irqsave(rdp, flags);
 	rdp->n_cbs_invoked += count;
 	trace_rcu_batch_end(rcu_state.name, count, !!rcl.head, need_resched(),
 			    is_idle_task(current), rcu_is_callbacks_kthread());
-- 
2.31.1.189.g2e36527f23

[PATCH rcu 09/18] rcu/nocb: Limit number of softirq callbacks only on softirq

[Paul E. McKenney]

From: Frederic Weisbecker <frederic@kernel.org>

The current condition to limit the number of callbacks executed in a
row checks the offloaded state of the rdp. Not only is it volatile
but it is also misleading: the rcu_core() may well be executing
callbacks concurrently with NOCB kthreads, and the offloaded state
would then be verified on both cases. As a result the limit would
spuriously not apply anymore on softirq while in the middle of
(de-)offloading process.

Fix and clarify the condition with those constraints in mind:

_ If callbacks are processed either by rcuc or NOCB kthread, the call
  to cond_resched_tasks_rcu_qs() is enough to take care of the overload.

_ If instead callbacks are processed by softirqs:
  * If need_resched(), exit the callbacks processing
  * Otherwise if CPU is idle we can continue
  * Otherwise exit because a softirq shouldn't interrupt a task for too
    long nor deprive other pending softirq vectors of the CPU.

Tested-by: Valentin Schneider <valentin.schneider@arm.com>
Tested-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Cc: Valentin Schneider <valentin.schneider@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Joel Fernandes <joel@joelfernandes.org>
Cc: Boqun Feng <boqun.feng@gmail.com>
Cc: Neeraj Upadhyay <neeraju@codeaurora.org>
Cc: Uladzislau Rezki <urezki@gmail.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
---
 kernel/rcu/tree.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
index 20587d035d03b..4f4d9ea6a9cb5 100644
--- a/kernel/rcu/tree.c
+++ b/kernel/rcu/tree.c
@@ -2525,9 +2525,8 @@ static void rcu_do_batch(struct rcu_data *rdp)
 		/*
 		 * Stop only if limit reached and CPU has something to do.
 		 */
-		if (count >= bl && !offloaded &&
-		    (need_resched() ||
-		     (!is_idle_task(current) && !rcu_is_callbacks_kthread())))
+		if (count >= bl && in_serving_softirq() &&
+		    (need_resched() || !is_idle_task(current)))
 			break;
 		if (unlikely(tlimit)) {
 			/* only call local_clock() every 32 callbacks */
-- 
2.31.1.189.g2e36527f23

[PATCH rcu 10/18] rcu: Fix callbacks processing time limit retaining cond_resched()

[Paul E. McKenney]

From: Frederic Weisbecker <frederic@kernel.org>

The callbacks processing time limit makes sure we are not exceeding a
given amount of time executing the queue.

However its "continue" clause bypasses the cond_resched() call on
rcuc and NOCB kthreads, delaying it until we reach the limit, which can
be very long...

Make sure the scheduler has a higher priority than the time limit.

Reviewed-by: Valentin Schneider <valentin.schneider@arm.com>
Tested-by: Valentin Schneider <valentin.schneider@arm.com>
Tested-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Cc: Valentin Schneider <valentin.schneider@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Joel Fernandes <joel@joelfernandes.org>
Cc: Boqun Feng <boqun.feng@gmail.com>
Cc: Neeraj Upadhyay <neeraju@codeaurora.org>
Cc: Uladzislau Rezki <urezki@gmail.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
---
 kernel/rcu/tree.c | 25 +++++++++++++++----------
 1 file changed, 15 insertions(+), 10 deletions(-)

diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
index 4f4d9ea6a9cb5..74e210ccfd382 100644
--- a/kernel/rcu/tree.c
+++ b/kernel/rcu/tree.c
@@ -2525,9 +2525,21 @@ static void rcu_do_batch(struct rcu_data *rdp)
 		/*
 		 * Stop only if limit reached and CPU has something to do.
 		 */
-		if (count >= bl && in_serving_softirq() &&
-		    (need_resched() || !is_idle_task(current)))
-			break;
+		if (in_serving_softirq()) {
+			if (count >= bl && (need_resched() || !is_idle_task(current)))
+				break;
+		} else {
+			local_bh_enable();
+			lockdep_assert_irqs_enabled();
+			cond_resched_tasks_rcu_qs();
+			lockdep_assert_irqs_enabled();
+			local_bh_disable();
+		}
+
+		/*
+		 * Make sure we don't spend too much time here and deprive other
+		 * softirq vectors of CPU cycles.
+		 */
 		if (unlikely(tlimit)) {
 			/* only call local_clock() every 32 callbacks */
 			if (likely((count & 31) || local_clock() < tlimit))
@@ -2535,13 +2547,6 @@ static void rcu_do_batch(struct rcu_data *rdp)
 			/* Exceeded the time limit, so leave. */
 			break;
 		}
-		if (!in_serving_softirq()) {
-			local_bh_enable();
-			lockdep_assert_irqs_enabled();
-			cond_resched_tasks_rcu_qs();
-			lockdep_assert_irqs_enabled();
-			local_bh_disable();
-		}
 	}
 
 	rcu_nocb_lock_irqsave(rdp, flags);
-- 
2.31.1.189.g2e36527f23

[PATCH rcu 11/18] rcu: Apply callbacks processing time limit only on softirq

[Paul E. McKenney]

From: Frederic Weisbecker <frederic@kernel.org>

Time limit only makes sense when callbacks are serviced in softirq mode
because:

_ In case we need to get back to the scheduler,
  cond_resched_tasks_rcu_qs() is called after each callback.

_ In case some other softirq vector needs the CPU, the call to
  local_bh_enable() before cond_resched_tasks_rcu_qs() takes care about
  them via a call to do_softirq().

Therefore, make sure the time limit only applies to softirq mode.

Reviewed-by: Valentin Schneider <valentin.schneider@arm.com>
Tested-by: Valentin Schneider <valentin.schneider@arm.com>
Tested-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Cc: Valentin Schneider <valentin.schneider@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Joel Fernandes <joel@joelfernandes.org>
Cc: Boqun Feng <boqun.feng@gmail.com>
Cc: Neeraj Upadhyay <neeraju@codeaurora.org>
Cc: Uladzislau Rezki <urezki@gmail.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
---
 kernel/rcu/tree.c | 25 ++++++++++++-------------
 1 file changed, 12 insertions(+), 13 deletions(-)

diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
index 74e210ccfd382..1b31fad9f1a0a 100644
--- a/kernel/rcu/tree.c
+++ b/kernel/rcu/tree.c
@@ -2488,7 +2488,7 @@ static void rcu_do_batch(struct rcu_data *rdp)
 	div = READ_ONCE(rcu_divisor);
 	div = div < 0 ? 7 : div > sizeof(long) * 8 - 2 ? sizeof(long) * 8 - 2 : div;
 	bl = max(rdp->blimit, pending >> div);
-	if (unlikely(bl > 100)) {
+	if (in_serving_softirq() && unlikely(bl > 100)) {
 		long rrn = READ_ONCE(rcu_resched_ns);
 
 		rrn = rrn < NSEC_PER_MSEC ? NSEC_PER_MSEC : rrn > NSEC_PER_SEC ? NSEC_PER_SEC : rrn;
@@ -2528,6 +2528,17 @@ static void rcu_do_batch(struct rcu_data *rdp)
 		if (in_serving_softirq()) {
 			if (count >= bl && (need_resched() || !is_idle_task(current)))
 				break;
+			/*
+			 * Make sure we don't spend too much time here and deprive other
+			 * softirq vectors of CPU cycles.
+			 */
+			if (unlikely(tlimit)) {
+				/* only call local_clock() every 32 callbacks */
+				if (likely((count & 31) || local_clock() < tlimit))
+					continue;
+				/* Exceeded the time limit, so leave. */
+				break;
+			}
 		} else {
 			local_bh_enable();
 			lockdep_assert_irqs_enabled();
@@ -2535,18 +2546,6 @@ static void rcu_do_batch(struct rcu_data *rdp)
 			lockdep_assert_irqs_enabled();
 			local_bh_disable();
 		}
-
-		/*
-		 * Make sure we don't spend too much time here and deprive other
-		 * softirq vectors of CPU cycles.
-		 */
-		if (unlikely(tlimit)) {
-			/* only call local_clock() every 32 callbacks */
-			if (likely((count & 31) || local_clock() < tlimit))
-				continue;
-			/* Exceeded the time limit, so leave. */
-			break;
-		}
 	}
 
 	rcu_nocb_lock_irqsave(rdp, flags);
-- 
2.31.1.189.g2e36527f23

[PATCH rcu 12/18] rcu/nocb: Don't invoke local rcu core on callback overload from nocb kthread

[Paul E. McKenney]

From: Frederic Weisbecker <frederic@kernel.org>

rcu_core() tries to ensure that its self-invocation in case of callbacks
overload only happen in softirq/rcuc mode. Indeed it doesn't make sense
to trigger local RCU core from nocb_cb kthread since it can execute
on a CPU different from the target rdp. Also in case of overload, the
nocb_cb kthread simply iterates a new loop of callbacks processing.

However the "offloaded" check that aims at preventing misplaced
rcu_core() invocations is wrong. First of all that state is volatile
and second: softirq/rcuc can execute while the target rdp is offloaded.
As a result rcu_core() can be invoked on the wrong CPU while in the
process of (de-)offloading.

Fix that with moving the rcu_core() self-invocation to rcu_core() itself,
irrespective of the rdp offloaded state.

Tested-by: Valentin Schneider <valentin.schneider@arm.com>
Tested-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Cc: Valentin Schneider <valentin.schneider@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Joel Fernandes <joel@joelfernandes.org>
Cc: Boqun Feng <boqun.feng@gmail.com>
Cc: Neeraj Upadhyay <neeraju@codeaurora.org>
Cc: Uladzislau Rezki <urezki@gmail.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
---
 kernel/rcu/tree.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
index 1b31fad9f1a0a..03c038f324a09 100644
--- a/kernel/rcu/tree.c
+++ b/kernel/rcu/tree.c
@@ -2460,7 +2460,6 @@ static void rcu_do_batch(struct rcu_data *rdp)
 	int div;
 	bool __maybe_unused empty;
 	unsigned long flags;
-	const bool offloaded = rcu_rdp_is_offloaded(rdp);
 	struct rcu_head *rhp;
 	struct rcu_cblist rcl = RCU_CBLIST_INITIALIZER(rcl);
 	long bl, count = 0;
@@ -2582,9 +2581,6 @@ static void rcu_do_batch(struct rcu_data *rdp)
 
 	rcu_nocb_unlock_irqrestore(rdp, flags);
 
-	/* Re-invoke RCU core processing if there are callbacks remaining. */
-	if (!offloaded && rcu_segcblist_ready_cbs(&rdp->cblist))
-		invoke_rcu_core();
 	tick_dep_clear_task(current, TICK_DEP_BIT_RCU);
 }
 
@@ -2771,8 +2767,12 @@ static __latent_entropy void rcu_core(void)
 
 	/* If there are callbacks ready, invoke them. */
 	if (do_batch && rcu_segcblist_ready_cbs(&rdp->cblist) &&
-	    likely(READ_ONCE(rcu_scheduler_fully_active)))
+	    likely(READ_ONCE(rcu_scheduler_fully_active))) {
 		rcu_do_batch(rdp);
+		/* Re-invoke RCU core processing if there are callbacks remaining. */
+		if (rcu_segcblist_ready_cbs(&rdp->cblist))
+			invoke_rcu_core();
+	}
 
 	/* Do any needed deferred wakeups of rcuo kthreads. */
 	do_nocb_deferred_wakeup(rdp);
-- 
2.31.1.189.g2e36527f23

[PATCH rcu 13/18] rcu/nocb: Remove rcu_node structure from nocb list when de-offloaded

[Paul E. McKenney]

From: Frederic Weisbecker <frederic@kernel.org>

The nocb_gp_wait() function iterates over all CPUs in its group,
including even those CPUs that have been de-offloaded.  This is of
course suboptimal, especially if none of the CPUs within the group are
currently offloaded.  This will become even more of a problem once a
nocb kthread is created for all possible CPUs.

Therefore use a standard double linked list to link all the offloaded
rcu_data structures and safely add or delete these structure as we
offload or de-offload them, respectively.

Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Cc: Neeraj Upadhyay <quic_neeraju@quicinc.com>
Cc: Boqun Feng <boqun.feng@gmail.com>
Cc: Uladzislau Rezki <urezki@gmail.com>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Joel Fernandes <joel@joelfernandes.org>
Tested-by: Juri Lelli <juri.lelli@redhat.com>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
---
 kernel/rcu/tree.h      |  7 +++++--
 kernel/rcu/tree_nocb.h | 45 +++++++++++++++++++++++++++++++++++-------
 2 files changed, 43 insertions(+), 9 deletions(-)

diff --git a/kernel/rcu/tree.h b/kernel/rcu/tree.h
index 4f6c67b3ccd5e..5884380f4039c 100644
--- a/kernel/rcu/tree.h
+++ b/kernel/rcu/tree.h
@@ -227,8 +227,11 @@ struct rcu_data {
 	struct swait_queue_head nocb_gp_wq; /* For nocb kthreads to sleep on. */
 	bool nocb_cb_sleep;		/* Is the nocb CB thread asleep? */
 	struct task_struct *nocb_cb_kthread;
-	struct rcu_data *nocb_next_cb_rdp;
-					/* Next rcu_data in wakeup chain. */
+	struct list_head nocb_head_rdp; /*
+					 * Head of rcu_data list in wakeup chain,
+					 * if rdp_gp.
+					 */
+	struct list_head nocb_entry_rdp; /* rcu_data node in wakeup chain. */
 
 	/* The following fields are used by CB kthread, hence new cacheline. */
 	struct rcu_data *nocb_gp_rdp ____cacheline_internodealigned_in_smp;
diff --git a/kernel/rcu/tree_nocb.h b/kernel/rcu/tree_nocb.h
index 2461fe8d0c23a..8e94a5344afed 100644
--- a/kernel/rcu/tree_nocb.h
+++ b/kernel/rcu/tree_nocb.h
@@ -625,7 +625,21 @@ static void nocb_gp_wait(struct rcu_data *my_rdp)
 	 * and the global grace-period kthread are awakened if needed.
 	 */
 	WARN_ON_ONCE(my_rdp->nocb_gp_rdp != my_rdp);
-	for (rdp = my_rdp; rdp; rdp = rdp->nocb_next_cb_rdp) {
+	/*
+	 * An rcu_data structure is removed from the list after its
+	 * CPU is de-offloaded and added to the list before that CPU is
+	 * (re-)offloaded.  If the following loop happens to be referencing
+	 * that rcu_data structure during the time that the corresponding
+	 * CPU is de-offloaded and then immediately re-offloaded, this
+	 * loop's rdp pointer will be carried to the end of the list by
+	 * the resulting pair of list operations.  This can cause the loop
+	 * to skip over some of the rcu_data structures that were supposed
+	 * to have been scanned.  Fortunately a new iteration through the
+	 * entire loop is forced after a given CPU's rcu_data structure
+	 * is added to the list, so the skipped-over rcu_data structures
+	 * won't be ignored for long.
+	 */
+	list_for_each_entry_rcu(rdp, &my_rdp->nocb_head_rdp, nocb_entry_rdp, 1) {
 		bool needwake_state = false;
 
 		if (!nocb_gp_enabled_cb(rdp))
@@ -1003,6 +1017,8 @@ static long rcu_nocb_rdp_deoffload(void *arg)
 	swait_event_exclusive(rdp->nocb_state_wq,
 			      !rcu_segcblist_test_flags(cblist, SEGCBLIST_KTHREAD_CB |
 							SEGCBLIST_KTHREAD_GP));
+	/* Stop nocb_gp_wait() from iterating over this structure. */
+	list_del_rcu(&rdp->nocb_entry_rdp);
 	/*
 	 * Lock one last time to acquire latest callback updates from kthreads
 	 * so we can later handle callbacks locally without locking.
@@ -1066,6 +1082,17 @@ static long rcu_nocb_rdp_offload(void *arg)
 		return -EINVAL;
 
 	pr_info("Offloading %d\n", rdp->cpu);
+
+	/*
+	 * Cause future nocb_gp_wait() invocations to iterate over
+	 * structure, resetting ->nocb_gp_sleep and waking up the related
+	 * "rcuog".  Since nocb_gp_wait() in turn locks ->nocb_gp_lock
+	 * before setting ->nocb_gp_sleep again, we are guaranteed to
+	 * iterate this newly added structure before "rcuog" goes to
+	 * sleep again.
+	 */
+	list_add_tail_rcu(&rdp->nocb_entry_rdp, &rdp->nocb_gp_rdp->nocb_head_rdp);
+
 	/*
 	 * Can't use rcu_nocb_lock_irqsave() before SEGCBLIST_LOCKING
 	 * is set.
@@ -1268,7 +1295,6 @@ static void __init rcu_organize_nocb_kthreads(void)
 	int nl = 0;  /* Next GP kthread. */
 	struct rcu_data *rdp;
 	struct rcu_data *rdp_gp = NULL;  /* Suppress misguided gcc warn. */
-	struct rcu_data *rdp_prev = NULL;
 
 	if (!cpumask_available(rcu_nocb_mask))
 		return;
@@ -1288,8 +1314,8 @@ static void __init rcu_organize_nocb_kthreads(void)
 			/* New GP kthread, set up for CBs & next GP. */
 			gotnocbs = true;
 			nl = DIV_ROUND_UP(rdp->cpu + 1, ls) * ls;
-			rdp->nocb_gp_rdp = rdp;
 			rdp_gp = rdp;
+			INIT_LIST_HEAD(&rdp->nocb_head_rdp);
 			if (dump_tree) {
 				if (!firsttime)
 					pr_cont("%s\n", gotnocbscbs
@@ -1302,12 +1328,11 @@ static void __init rcu_organize_nocb_kthreads(void)
 		} else {
 			/* Another CB kthread, link to previous GP kthread. */
 			gotnocbscbs = true;
-			rdp->nocb_gp_rdp = rdp_gp;
-			rdp_prev->nocb_next_cb_rdp = rdp;
 			if (dump_tree)
 				pr_cont(" %d", cpu);
 		}
-		rdp_prev = rdp;
+		rdp->nocb_gp_rdp = rdp_gp;
+		list_add_tail(&rdp->nocb_entry_rdp, &rdp_gp->nocb_head_rdp);
 	}
 	if (gotnocbs && dump_tree)
 		pr_cont("%s\n", gotnocbscbs ? "" : " (self only)");
@@ -1369,6 +1394,7 @@ static void show_rcu_nocb_state(struct rcu_data *rdp)
 {
 	char bufw[20];
 	char bufr[20];
+	struct rcu_data *nocb_next_rdp;
 	struct rcu_segcblist *rsclp = &rdp->cblist;
 	bool waslocked;
 	bool wassleep;
@@ -1376,11 +1402,16 @@ static void show_rcu_nocb_state(struct rcu_data *rdp)
 	if (rdp->nocb_gp_rdp == rdp)
 		show_rcu_nocb_gp_state(rdp);
 
+	nocb_next_rdp = list_next_or_null_rcu(&rdp->nocb_gp_rdp->nocb_head_rdp,
+					      &rdp->nocb_entry_rdp,
+					      typeof(*rdp),
+					      nocb_entry_rdp);
+
 	sprintf(bufw, "%ld", rsclp->gp_seq[RCU_WAIT_TAIL]);
 	sprintf(bufr, "%ld", rsclp->gp_seq[RCU_NEXT_READY_TAIL]);
 	pr_info("   CB %d^%d->%d %c%c%c%c%c%c F%ld L%ld C%d %c%c%s%c%s%c%c q%ld %c CPU %d%s\n",
 		rdp->cpu, rdp->nocb_gp_rdp->cpu,
-		rdp->nocb_next_cb_rdp ? rdp->nocb_next_cb_rdp->cpu : -1,
+		nocb_next_rdp ? nocb_next_rdp->cpu : -1,
 		"kK"[!!rdp->nocb_cb_kthread],
 		"bB"[raw_spin_is_locked(&rdp->nocb_bypass_lock)],
 		"cC"[!!atomic_read(&rdp->nocb_lock_contended)],
-- 
2.31.1.189.g2e36527f23

[PATCH rcu 14/18] rcu/nocb: Prepare nocb_cb_wait() to start with a non-offloaded rdp

[Paul E. McKenney]

From: Frederic Weisbecker <frederic@kernel.org>

In order to be able to toggle the offloaded state from cpusets, a nocb
kthread will need to be created for all possible CPUs whenever either
of the "rcu_nocbs=" or "nohz_full=" parameters are specified.

Therefore, the nocb_cb_wait() kthread must be prepared to start running
on a de-offloaded rdp.  To accomplish this, simply move the sleeping
condition to the beginning of the nocb_cb_wait() function, which prevents
this kthread from attempting to invoke callbacks before the corresponding
CPU is offloaded.

Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Cc: Neeraj Upadhyay <quic_neeraju@quicinc.com>
Cc: Boqun Feng <boqun.feng@gmail.com>
Cc: Uladzislau Rezki <urezki@gmail.com>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Joel Fernandes <joel@joelfernandes.org>
Tested-by: Juri Lelli <juri.lelli@redhat.com>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
---
 kernel/rcu/tree_nocb.h | 23 ++++++++++++-----------
 1 file changed, 12 insertions(+), 11 deletions(-)

diff --git a/kernel/rcu/tree_nocb.h b/kernel/rcu/tree_nocb.h
index 8e94a5344afed..cd596cf4ee79a 100644
--- a/kernel/rcu/tree_nocb.h
+++ b/kernel/rcu/tree_nocb.h
@@ -803,6 +803,18 @@ static void nocb_cb_wait(struct rcu_data *rdp)
 	bool can_sleep = true;
 	struct rcu_node *rnp = rdp->mynode;
 
+	do {
+		swait_event_interruptible_exclusive(rdp->nocb_cb_wq,
+						    nocb_cb_wait_cond(rdp));
+
+		// VVV Ensure CB invocation follows _sleep test.
+		if (smp_load_acquire(&rdp->nocb_cb_sleep)) { // ^^^
+			WARN_ON(signal_pending(current));
+			trace_rcu_nocb_wake(rcu_state.name, rdp->cpu, TPS("WokeEmpty"));
+		}
+	} while (!nocb_cb_can_run(rdp));
+
+
 	local_irq_save(flags);
 	rcu_momentary_dyntick_idle();
 	local_irq_restore(flags);
@@ -855,17 +867,6 @@ static void nocb_cb_wait(struct rcu_data *rdp)
 
 	if (needwake_state)
 		swake_up_one(&rdp->nocb_state_wq);
-
-	do {
-		swait_event_interruptible_exclusive(rdp->nocb_cb_wq,
-						    nocb_cb_wait_cond(rdp));
-
-		// VVV Ensure CB invocation follows _sleep test.
-		if (smp_load_acquire(&rdp->nocb_cb_sleep)) { // ^^^
-			WARN_ON(signal_pending(current));
-			trace_rcu_nocb_wake(rcu_state.name, rdp->cpu, TPS("WokeEmpty"));
-		}
-	} while (!nocb_cb_can_run(rdp));
 }
 
 /*
-- 
2.31.1.189.g2e36527f23

[PATCH rcu 15/18] rcu/nocb: Optimize kthreads and rdp initialization

[Paul E. McKenney]

From: Frederic Weisbecker <frederic@kernel.org>

Currently cpumask_available() is used to prevent from unwanted NOCB
initialization.  However if neither "rcu_nocbs=" nor "nohz_full="
parameters are passed to a kernel built with CONFIG_CPUMASK_OFFSTACK=n,
the initialization path is still taken, running through all sorts of
needless operations and iterations on an empty cpumask.

Fix this by relying on a real initialization state instead.  This also
optimizes kthread creation, preventing needless iteration over all online
CPUs when the kernel is booted without any offloaded CPUs.

Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Cc: Neeraj Upadhyay <quic_neeraju@quicinc.com>
Cc: Boqun Feng <boqun.feng@gmail.com>
Cc: Uladzislau Rezki <urezki@gmail.com>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Joel Fernandes <joel@joelfernandes.org>
Tested-by: Juri Lelli <juri.lelli@redhat.com>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
---
 kernel/rcu/tree_nocb.h | 24 +++++++++++++++++-------
 1 file changed, 17 insertions(+), 7 deletions(-)

diff --git a/kernel/rcu/tree_nocb.h b/kernel/rcu/tree_nocb.h
index cd596cf4ee79a..42092de677054 100644
--- a/kernel/rcu/tree_nocb.h
+++ b/kernel/rcu/tree_nocb.h
@@ -60,6 +60,9 @@ static inline bool rcu_current_is_nocb_kthread(struct rcu_data *rdp)
  * Parse the boot-time rcu_nocb_mask CPU list from the kernel parameters.
  * If the list is invalid, a warning is emitted and all CPUs are offloaded.
  */
+
+static bool rcu_nocb_is_setup;
+
 static int __init rcu_nocb_setup(char *str)
 {
 	alloc_bootmem_cpumask_var(&rcu_nocb_mask);
@@ -67,6 +70,7 @@ static int __init rcu_nocb_setup(char *str)
 		pr_warn("rcu_nocbs= bad CPU range, all CPUs set\n");
 		cpumask_setall(rcu_nocb_mask);
 	}
+	rcu_nocb_is_setup = true;
 	return 1;
 }
 __setup("rcu_nocbs=", rcu_nocb_setup);
@@ -1167,13 +1171,17 @@ void __init rcu_init_nohz(void)
 		need_rcu_nocb_mask = true;
 #endif /* #if defined(CONFIG_NO_HZ_FULL) */
 
-	if (!cpumask_available(rcu_nocb_mask) && need_rcu_nocb_mask) {
-		if (!zalloc_cpumask_var(&rcu_nocb_mask, GFP_KERNEL)) {
-			pr_info("rcu_nocb_mask allocation failed, callback offloading disabled.\n");
-			return;
+	if (need_rcu_nocb_mask) {
+		if (!cpumask_available(rcu_nocb_mask)) {
+			if (!zalloc_cpumask_var(&rcu_nocb_mask, GFP_KERNEL)) {
+				pr_info("rcu_nocb_mask allocation failed, callback offloading disabled.\n");
+				return;
+			}
 		}
+		rcu_nocb_is_setup = true;
 	}
-	if (!cpumask_available(rcu_nocb_mask))
+
+	if (!rcu_nocb_is_setup)
 		return;
 
 #if defined(CONFIG_NO_HZ_FULL)
@@ -1275,8 +1283,10 @@ static void __init rcu_spawn_nocb_kthreads(void)
 {
 	int cpu;
 
-	for_each_online_cpu(cpu)
-		rcu_spawn_cpu_nocb_kthread(cpu);
+	if (rcu_nocb_is_setup) {
+		for_each_online_cpu(cpu)
+			rcu_spawn_cpu_nocb_kthread(cpu);
+	}
 }
 
 /* How many CB CPU IDs per GP kthread?  Default of -1 for sqrt(nr_cpu_ids). */
-- 
2.31.1.189.g2e36527f23

[PATCH rcu 16/18] rcu/nocb: Create kthreads on all CPUs if "rcu_nocbs=" or "nohz_full=" are passed

[Paul E. McKenney]

From: Frederic Weisbecker <frederic@kernel.org>

In order to be able to (de-)offload any CPU using cpusets in the future,
create the NOCB data structures for all possible CPUs.  For now this is
done only as long as the "rcu_nocbs=" or "nohz_full=" kernel parameters
are passed to avoid the unnecessary overhead for most users.

Note that the rcuog and rcuoc kthreads are not created until at least
one of the corresponding CPUs comes online.  This approach avoids the
creation of excess kthreads when firmware lies about the number of CPUs
present on the system.

Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Cc: Neeraj Upadhyay <quic_neeraju@quicinc.com>
Cc: Boqun Feng <boqun.feng@gmail.com>
Cc: Uladzislau Rezki <urezki@gmail.com>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Joel Fernandes <joel@joelfernandes.org>
Tested-by: Juri Lelli <juri.lelli@redhat.com>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
---
 kernel/rcu/tree_nocb.h | 14 ++++++--------
 1 file changed, 6 insertions(+), 8 deletions(-)

diff --git a/kernel/rcu/tree_nocb.h b/kernel/rcu/tree_nocb.h
index 42092de677054..f580a6b2e74e3 100644
--- a/kernel/rcu/tree_nocb.h
+++ b/kernel/rcu/tree_nocb.h
@@ -1237,11 +1237,8 @@ static void rcu_spawn_one_nocb_kthread(int cpu)
 	struct rcu_data *rdp_gp;
 	struct task_struct *t;
 
-	/*
-	 * If this isn't a no-CBs CPU or if it already has an rcuo kthread,
-	 * then nothing to do.
-	 */
-	if (!rcu_is_nocb_cpu(cpu) || rdp->nocb_cb_kthread)
+	/* If there already is an rcuo kthread, then nothing to do. */
+	if (rdp->nocb_cb_kthread)
 		return;
 
 	/* If we didn't spawn the GP kthread first, reorganize! */
@@ -1269,7 +1266,7 @@ static void rcu_spawn_one_nocb_kthread(int cpu)
  */
 static void rcu_spawn_cpu_nocb_kthread(int cpu)
 {
-	if (rcu_scheduler_fully_active)
+	if (rcu_scheduler_fully_active && rcu_nocb_is_setup)
 		rcu_spawn_one_nocb_kthread(cpu);
 }
 
@@ -1319,7 +1316,7 @@ static void __init rcu_organize_nocb_kthreads(void)
 	 * Should the corresponding CPU come online in the future, then
 	 * we will spawn the needed set of rcu_nocb_kthread() kthreads.
 	 */
-	for_each_cpu(cpu, rcu_nocb_mask) {
+	for_each_possible_cpu(cpu) {
 		rdp = per_cpu_ptr(&rcu_data, cpu);
 		if (rdp->cpu >= nl) {
 			/* New GP kthread, set up for CBs & next GP. */
@@ -1343,7 +1340,8 @@ static void __init rcu_organize_nocb_kthreads(void)
 				pr_cont(" %d", cpu);
 		}
 		rdp->nocb_gp_rdp = rdp_gp;
-		list_add_tail(&rdp->nocb_entry_rdp, &rdp_gp->nocb_head_rdp);
+		if (cpumask_test_cpu(cpu, rcu_nocb_mask))
+			list_add_tail(&rdp->nocb_entry_rdp, &rdp_gp->nocb_head_rdp);
 	}
 	if (gotnocbs && dump_tree)
 		pr_cont("%s\n", gotnocbscbs ? "" : " (self only)");
-- 
2.31.1.189.g2e36527f23

[PATCH rcu 17/18] rcu/nocb: Allow empty "rcu_nocbs" kernel parameter

[Paul E. McKenney]

From: Frederic Weisbecker <frederic@kernel.org>

Allow the rcu_nocbs kernel parameter to be specified just by itself,
without specifying any CPUs.  This allows systems administrators to use
"rcu_nocbs" to specify that none of the CPUs are to be offloaded at boot
time, but than any of them may be offloaded at runtime via cpusets.

In contrast, if the "rcu_nocbs" or "nohz_full" kernel parameters are not
specified at all, then not only are none of the CPUs offloaded at boot,
none of them can be offloaded at runtime, either.

While in the area, modernize the description of the "rcuo" kthreads'
naming scheme.

Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Cc: Neeraj Upadhyay <quic_neeraju@quicinc.com>
Cc: Boqun Feng <boqun.feng@gmail.com>
Cc: Uladzislau Rezki <urezki@gmail.com>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Joel Fernandes <joel@joelfernandes.org>
Tested-by: Juri Lelli <juri.lelli@redhat.com>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
---
 .../admin-guide/kernel-parameters.txt         | 37 ++++++++++++-------
 kernel/rcu/tree_nocb.h                        | 10 +++--
 2 files changed, 30 insertions(+), 17 deletions(-)

diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
index 9725c546a0d46..8b5da10f932e2 100644
--- a/Documentation/admin-guide/kernel-parameters.txt
+++ b/Documentation/admin-guide/kernel-parameters.txt
@@ -4343,19 +4343,30 @@
 				Disable the Correctable Errors Collector,
 				see CONFIG_RAS_CEC help text.
 
-	rcu_nocbs=	[KNL]
-			The argument is a cpu list, as described above.
-
-			In kernels built with CONFIG_RCU_NOCB_CPU=y, set
-			the specified list of CPUs to be no-callback CPUs.
-			Invocation of these CPUs' RCU callbacks will be
-			offloaded to "rcuox/N" kthreads created for that
-			purpose, where "x" is "p" for RCU-preempt, and
-			"s" for RCU-sched, and "N" is the CPU number.
-			This reduces OS jitter on the offloaded CPUs,
-			which can be useful for HPC and real-time
-			workloads.  It can also improve energy efficiency
-			for asymmetric multiprocessors.
+	rcu_nocbs[=cpu-list]
+			[KNL] The optional argument is a cpu list,
+			as described above.
+
+			In kernels built with CONFIG_RCU_NOCB_CPU=y,
+			enable the no-callback CPU mode, which prevents
+			such CPUs' callbacks from being invoked in
+			softirq context.  Invocation of such CPUs' RCU
+			callbacks will instead be offloaded to "rcuox/N"
+			kthreads created for that purpose, where "x" is
+			"p" for RCU-preempt, "s" for RCU-sched, and "g"
+			for the kthreads that mediate grace periods; and
+			"N" is the CPU number. This reduces OS jitter on
+			the offloaded CPUs, which can be useful for HPC
+			and real-time workloads.  It can also improve
+			energy efficiency for asymmetric multiprocessors.
+
+			If a cpulist is passed as an argument, the specified
+			list of	CPUs is set to no-callback mode from boot.
+
+			Otherwise, if the '=' sign and the cpulist
+			arguments are omitted, no CPU will be set to
+			no-callback mode from boot but the mode may be
+			toggled at runtime via cpusets.
 
 	rcu_nocb_poll	[KNL]
 			Rather than requiring that offloaded CPUs
diff --git a/kernel/rcu/tree_nocb.h b/kernel/rcu/tree_nocb.h
index f580a6b2e74e3..0c1802ce4764c 100644
--- a/kernel/rcu/tree_nocb.h
+++ b/kernel/rcu/tree_nocb.h
@@ -66,14 +66,16 @@ static bool rcu_nocb_is_setup;
 static int __init rcu_nocb_setup(char *str)
 {
 	alloc_bootmem_cpumask_var(&rcu_nocb_mask);
-	if (cpulist_parse(str, rcu_nocb_mask)) {
-		pr_warn("rcu_nocbs= bad CPU range, all CPUs set\n");
-		cpumask_setall(rcu_nocb_mask);
+	if (*str == '=') {
+		if (cpulist_parse(++str, rcu_nocb_mask)) {
+			pr_warn("rcu_nocbs= bad CPU range, all CPUs set\n");
+			cpumask_setall(rcu_nocb_mask);
+		}
 	}
 	rcu_nocb_is_setup = true;
 	return 1;
 }
-__setup("rcu_nocbs=", rcu_nocb_setup);
+__setup("rcu_nocbs", rcu_nocb_setup);
 
 static int __init parse_rcu_nocb_poll(char *arg)
 {
-- 
2.31.1.189.g2e36527f23

[PATCH rcu 18/18] rcu/nocb: Merge rcu_spawn_cpu_nocb_kthread() and rcu_spawn_one_nocb_kthread()

[Paul E. McKenney]

From: Frederic Weisbecker <frederic@kernel.org>

The rcu_spawn_one_nocb_kthread() function is called only from
rcu_spawn_cpu_nocb_kthread().  Therefore, inline the former into
the latter, saving a few lines of code.

Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Cc: Neeraj Upadhyay <quic_neeraju@quicinc.com>
Cc: Boqun Feng <boqun.feng@gmail.com>
Cc: Uladzislau Rezki <urezki@gmail.com>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Joel Fernandes <joel@joelfernandes.org>
Tested-by: Juri Lelli <juri.lelli@redhat.com>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
---
 kernel/rcu/tree_nocb.h | 15 ++++-----------
 1 file changed, 4 insertions(+), 11 deletions(-)

diff --git a/kernel/rcu/tree_nocb.h b/kernel/rcu/tree_nocb.h
index 0c1802ce4764c..eeafb546a7a09 100644
--- a/kernel/rcu/tree_nocb.h
+++ b/kernel/rcu/tree_nocb.h
@@ -1233,12 +1233,15 @@ static void __init rcu_boot_init_nocb_percpu_data(struct rcu_data *rdp)
  * rcuo CB kthread, spawn it.  Additionally, if the rcuo GP kthread
  * for this CPU's group has not yet been created, spawn it as well.
  */
-static void rcu_spawn_one_nocb_kthread(int cpu)
+static void rcu_spawn_cpu_nocb_kthread(int cpu)
 {
 	struct rcu_data *rdp = per_cpu_ptr(&rcu_data, cpu);
 	struct rcu_data *rdp_gp;
 	struct task_struct *t;
 
+	if (!rcu_scheduler_fully_active || !rcu_nocb_is_setup)
+		return;
+
 	/* If there already is an rcuo kthread, then nothing to do. */
 	if (rdp->nocb_cb_kthread)
 		return;
@@ -1262,16 +1265,6 @@ static void rcu_spawn_one_nocb_kthread(int cpu)
 	WRITE_ONCE(rdp->nocb_gp_kthread, rdp_gp->nocb_gp_kthread);
 }
 
-/*
- * If the specified CPU is a no-CBs CPU that does not already have its
- * rcuo kthread, spawn it.
- */
-static void rcu_spawn_cpu_nocb_kthread(int cpu)
-{
-	if (rcu_scheduler_fully_active && rcu_nocb_is_setup)
-		rcu_spawn_one_nocb_kthread(cpu);
-}
-
 /*
  * Once the scheduler is running, spawn rcuo kthreads for all online
  * no-CBs CPUs.  This assumes that the early_initcall()s happen before
-- 
2.31.1.189.g2e36527f23

Re: [PATCH rcu 01/18] rcu: Tighten rcu_advance_cbs_nowake() checks

[Frederic Weisbecker]

On Wed, Dec 01, 2021 at 04:28:55PM -0800, Paul E. McKenney wrote:
> Currently, rcu_advance_cbs_nowake() checks that a grace period is in
> progress, however, that grace period could end just after the check.
> This commit rechecks that a grace period is still in progress the lock.

*while holding the node lock.

> The grace period cannot end while the current CPU's rcu_node structure's
> ->lock is held, thus avoiding false positives from the WARN_ON_ONCE().
> 
> As Daniel Vacek noted, it is not necessary for the rcu_node structure
> to have a CPU that has not yet passed through its quiescent state.
> 
> Tested-By: Guillaume Morin <guillaume@morinfr.org>
> Signed-off-by: Paul E. McKenney <paulmck@kernel.org>

Thanks!

> ---
>  kernel/rcu/tree.c | 7 ++++---
>  1 file changed, 4 insertions(+), 3 deletions(-)
> 
> diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
> index ef8d36f580fc3..8706b30c2ac88 100644
> --- a/kernel/rcu/tree.c
> +++ b/kernel/rcu/tree.c
> @@ -1590,10 +1590,11 @@ static void __maybe_unused rcu_advance_cbs_nowake(struct rcu_node *rnp,
>  						  struct rcu_data *rdp)
>  {
>  	rcu_lockdep_assert_cblist_protected(rdp);
> -	if (!rcu_seq_state(rcu_seq_current(&rnp->gp_seq)) ||
> -	    !raw_spin_trylock_rcu_node(rnp))
> +	if (!rcu_seq_state(rcu_seq_current(&rnp->gp_seq)) || !raw_spin_trylock_rcu_node(rnp))
>  		return;
> -	WARN_ON_ONCE(rcu_advance_cbs(rnp, rdp));
> +	// The grace period cannot end while we hold the rcu_node lock.
> +	if (rcu_seq_state(rcu_seq_current(&rnp->gp_seq)))
> +		WARN_ON_ONCE(rcu_advance_cbs(rnp, rdp));
>  	raw_spin_unlock_rcu_node(rnp);
>  }
>  
> -- 
> 2.31.1.189.g2e36527f23
> 

Re: [PATCH rcu 01/18] rcu: Tighten rcu_advance_cbs_nowake() checks

[Paul E. McKenney]

On Thu, Dec 02, 2021 at 01:41:42AM +0100, Frederic Weisbecker wrote:
> On Wed, Dec 01, 2021 at 04:28:55PM -0800, Paul E. McKenney wrote:
> > Currently, rcu_advance_cbs_nowake() checks that a grace period is in
> > progress, however, that grace period could end just after the check.
> > This commit rechecks that a grace period is still in progress the lock.
> 
> *while holding the node lock.

Good eyes, thank you!  I will fix on next rebase.

							Thanx, Paul

> > The grace period cannot end while the current CPU's rcu_node structure's
> > ->lock is held, thus avoiding false positives from the WARN_ON_ONCE().
> > 
> > As Daniel Vacek noted, it is not necessary for the rcu_node structure
> > to have a CPU that has not yet passed through its quiescent state.
> > 
> > Tested-By: Guillaume Morin <guillaume@morinfr.org>
> > Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
> 
> Thanks!
> 
> > ---
> >  kernel/rcu/tree.c | 7 ++++---
> >  1 file changed, 4 insertions(+), 3 deletions(-)
> > 
> > diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
> > index ef8d36f580fc3..8706b30c2ac88 100644
> > --- a/kernel/rcu/tree.c
> > +++ b/kernel/rcu/tree.c
> > @@ -1590,10 +1590,11 @@ static void __maybe_unused rcu_advance_cbs_nowake(struct rcu_node *rnp,
> >  						  struct rcu_data *rdp)
> >  {
> >  	rcu_lockdep_assert_cblist_protected(rdp);
> > -	if (!rcu_seq_state(rcu_seq_current(&rnp->gp_seq)) ||
> > -	    !raw_spin_trylock_rcu_node(rnp))
> > +	if (!rcu_seq_state(rcu_seq_current(&rnp->gp_seq)) || !raw_spin_trylock_rcu_node(rnp))
> >  		return;
> > -	WARN_ON_ONCE(rcu_advance_cbs(rnp, rdp));
> > +	// The grace period cannot end while we hold the rcu_node lock.
> > +	if (rcu_seq_state(rcu_seq_current(&rnp->gp_seq)))
> > +		WARN_ON_ONCE(rcu_advance_cbs(rnp, rdp));
> >  	raw_spin_unlock_rcu_node(rnp);
> >  }
> >  
> > -- 
> > 2.31.1.189.g2e36527f23
> > 

Re: [PATCH rcu 0/18] RCU no-CBs CPU updates for v5.17

[Yury Norov]

On Wed, Dec 01, 2021 at 04:28:48PM -0800, Paul E. McKenney wrote:
> Hello!

...

> 17.	Allow empty "rcu_nocbs" kernel parameter, courtesy of Frederic
> 	Weisbecker.

...

> Note that #17 might be updated given some ongoing work by Yury Norov
> to support "none" for bitmaps, including the cpumask taken by the
> rcu_nocbs kernel-boot parameter.

Hi Paul,

This is the only work that is needed to support 'none':

https://lkml.org/lkml/2021/11/24/2642

From the last discussion, it's not clear is this 'none' needed or not.
If needed, I'll submit the patch.

Thanks,
Yury

Re: [PATCH rcu 0/18] RCU no-CBs CPU updates for v5.17

[Paul E. McKenney]

On Wed, Dec 01, 2021 at 06:03:47PM -0800, Yury Norov wrote:
> On Wed, Dec 01, 2021 at 04:28:48PM -0800, Paul E. McKenney wrote:
> > Hello!
> 
> ...
> 
> > 17.	Allow empty "rcu_nocbs" kernel parameter, courtesy of Frederic
> > 	Weisbecker.
> 
> ...
> 
> > Note that #17 might be updated given some ongoing work by Yury Norov
> > to support "none" for bitmaps, including the cpumask taken by the
> > rcu_nocbs kernel-boot parameter.
> 
> Hi Paul,
> 
> This is the only work that is needed to support 'none':
> 
> https://lkml.org/lkml/2021/11/24/2642
> 
> >From the last discussion, it's not clear is this 'none' needed or not.
> If needed, I'll submit the patch.

The question is whether rcu_nocbs= semantics like this are reasonable.
In other words, will systems administrators be OK with the idea of
differentiation between "rcu_nocbs=none" on the one hand and not
specifying rcu_nocbs at all on the other.

Thoughts?

							Thanx, Paul

------------------------------------------------------------------------

	rcu_nocbs=	[KNL]
			The argument is a cpu list, as described above.

			In kernels built with CONFIG_RCU_NOCB_CPU=y,
			enable the no-callback CPU mode, which prevents
			such CPUs' callbacks from being invoked in
			softirq context.  Invocation of such CPUs' RCU
			callbacks will instead be offloaded to "rcuox/N"
			kthreads created for that purpose, where "x" is
			"p" for RCU-preempt, "s" for RCU-sched, and "g"
			for the kthreads that mediate grace periods; and
			"N" is the CPU number. This reduces OS jitter on
			the offloaded CPUs, which can be useful for HPC
			and real-time workloads.  It can also improve
			energy efficiency, especially for asymmetric
			multiprocessors.

			The cpulist specifies which are to be set to
			no-callback mode from boot.  In particular,
			"rcu_nocbs=none" specifies that no CPUs are in
			no-callback mode at boot, but that any of them
			can be offloaded at runtime.  In contrast, in
			kernels that boot with no rcu_nocbs kernel boot
			parameter, CPUs are not only in no-callback mode
			at boot, but they remain that way throughout.