[PATCH tip/core/rcu 0/5] In-kernel programmatic grace-period expediting for v4.1

[PATCH tip/core/rcu 0/5] In-kernel programmatic grace-period expediting for v4.1

[Paul E. McKenney]

Hello!

This series provides an in-kernel API to expedite and unexpedite normal
RCU grace-period primitives such as synchronize_rcu().  It also provides
a Kconfig parameter that implicitly expedites at boot time, along with
a function that notes the end of in-kernel boot.  This last function is
intended to be invoked just before init is spawned.

1.	Provide rcu_expedite_gp() and rcu_unexpedite_gp().

2.	Add rcu_expedite_gp() and rcu_unexpedite_gp() to rcutorture.

3.	Change open-coded references ot the rcu_expedited variable to
	instead use the new rcu_gp_is_expedited() function.

4.	Add a CONFIG_RCU_EXPEDITE_BOOT Kconfig parameter that emulates
	a very early boot rcu_expedite_gp().  Also provide a new
	rcu_end_inkernel_boot() function that provides the corresponding
	rcu_unexpedite_gp() if CONFIG_RCU_EXPEDITE_BOOT.

5.	Make consistent use of variables in rcu_torture_writer().

							Thanx, Paul

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

 b/include/linux/rcupdate.h |   21 ++++++++++++++++
 b/init/Kconfig             |   13 +++++++++
 b/kernel/rcu/rcutorture.c  |   27 +++++++++++++++++++-
 b/kernel/rcu/srcu.c        |    2 -
 b/kernel/rcu/tree.c        |    9 +++---
 b/kernel/rcu/tree_plugin.h |    2 -
 b/kernel/rcu/update.c      |   59 ++++++++++++++++++++++++++++++++++++++++++++-
 7 files changed, 125 insertions(+), 8 deletions(-)

[PATCH tip/core/rcu 1/5] rcu: Provide rcu_expedite_gp() and rcu_unexpedite_gp()

[Paul E. McKenney]

From: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>

Currently, expediting of normal synchronous grace-period primitives
(synchronize_rcu() and friends) is controlled by the rcu_expedited()
boot/sysfs parameter.  This works well, but does not handle nesting.
This commit therefore provides rcu_expedite_gp() to enable expediting
and rcu_unexpedite_gp() to cancel a prior rcu_expedite_gp(), both of
which support nesting.

Reported-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 include/linux/rcupdate.h | 20 ++++++++++++++++++++
 kernel/rcu/update.c      | 48 ++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 68 insertions(+)

diff --git a/include/linux/rcupdate.h b/include/linux/rcupdate.h
index 78097491cd99..57a4d1f73a00 100644
--- a/include/linux/rcupdate.h
+++ b/include/linux/rcupdate.h
@@ -48,6 +48,26 @@
 
 extern int rcu_expedited; /* for sysctl */
 
+#ifdef CONFIG_TINY_RCU
+/* Tiny RCU doesn't expedite, as its purpose in life is instead to be tiny. */
+static inline bool rcu_gp_is_expedited(void)  /* Internal RCU use. */
+{
+	return false;
+}
+
+static inline void rcu_expedite_gp(void)
+{
+}
+
+static inline void rcu_unexpedite_gp(void)
+{
+}
+#else /* #ifdef CONFIG_TINY_RCU */
+bool rcu_gp_is_expedited(void);  /* Internal RCU use. */
+void rcu_expedite_gp(void);
+void rcu_unexpedite_gp(void);
+#endif /* #else #ifdef CONFIG_TINY_RCU */
+
 enum rcutorture_type {
 	RCU_FLAVOR,
 	RCU_BH_FLAVOR,
diff --git a/kernel/rcu/update.c b/kernel/rcu/update.c
index e0d31a345ee6..5f850823c187 100644
--- a/kernel/rcu/update.c
+++ b/kernel/rcu/update.c
@@ -62,6 +62,54 @@ MODULE_ALIAS("rcupdate");
 
 module_param(rcu_expedited, int, 0);
 
+#ifndef CONFIG_TINY_RCU
+
+static atomic_t rcu_expedited_nesting;
+
+/*
+ * Should normal grace-period primitives be expedited?  Intended for
+ * use within RCU.  Note that this function takes the rcu_expedited
+ * sysfs/boot variable into account as well as the rcu_expedite_gp()
+ * nesting.  So looping on rcu_unexpedite_gp() until rcu_gp_is_expedited()
+ * returns false is a -really- bad idea.
+ */
+bool rcu_gp_is_expedited(void)
+{
+	return rcu_expedited || atomic_read(&rcu_expedited_nesting);
+}
+EXPORT_SYMBOL_GPL(rcu_gp_is_expedited);
+
+/**
+ * rcu_expedite_gp - Expedite future RCU grace periods
+ *
+ * After a call to this function, future calls to synchronize_rcu() and
+ * friends act as the corresponding synchronize_rcu_expedited() function
+ * had instead been called.
+ */
+void rcu_expedite_gp(void)
+{
+	atomic_inc(&rcu_expedited_nesting);
+}
+EXPORT_SYMBOL_GPL(rcu_expedite_gp);
+
+/**
+ * rcu_unexpedite_gp - Cancel prior rcu_expedite_gp() invocation
+ *
+ * Undo a prior call to rcu_expedite_gp().  If all prior calls to
+ * rcu_expedite_gp() are undone by a subsequent call to rcu_unexpedite_gp(),
+ * and if the rcu_expedited sysfs/boot parameter is not set, then all
+ * subsequent calls to synchronize_rcu() and friends will return to
+ * their normal non-expedited behavior.
+ */
+void rcu_unexpedite_gp(void)
+{
+	atomic_dec(&rcu_expedited_nesting);
+}
+EXPORT_SYMBOL_GPL(rcu_unexpedite_gp);
+
+#endif /* #ifndef CONFIG_TINY_RCU */
+
+
 #ifdef CONFIG_PREEMPT_RCU
 
 /*
-- 
1.8.1.5

[PATCH tip/core/rcu 2/5] rcu: Add rcu_expedite_gp() and rcu_unexpedite_gp() to rcutorture

[Paul E. McKenney]

From: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 kernel/rcu/rcutorture.c | 25 +++++++++++++++++++++++++
 1 file changed, 25 insertions(+)

diff --git a/kernel/rcu/rcutorture.c b/kernel/rcu/rcutorture.c
index 30d42aa55d83..3833aa611ae7 100644
--- a/kernel/rcu/rcutorture.c
+++ b/kernel/rcu/rcutorture.c
@@ -853,6 +853,8 @@ rcu_torture_fqs(void *arg)
 static int
 rcu_torture_writer(void *arg)
 {
+	bool can_expedite = !rcu_gp_is_expedited();
+	int expediting = 0;
 	unsigned long gp_snap;
 	bool gp_cond1 = gp_cond, gp_exp1 = gp_exp, gp_normal1 = gp_normal;
 	bool gp_sync1 = gp_sync;
@@ -865,6 +867,12 @@ rcu_torture_writer(void *arg)
 	int nsynctypes = 0;
 
 	VERBOSE_TOROUT_STRING("rcu_torture_writer task started");
+	pr_alert("%s" TORTURE_FLAG
+		 " Grace periods expedited from boot/sysfs for %s,\n",
+		 torture_type, cur_ops->name);
+	pr_alert("%s" TORTURE_FLAG
+		 " Testing of dynamic grace-period expediting diabled.\n",
+		 torture_type);
 
 	/* Initialize synctype[] array.  If none set, take default. */
 	if (!gp_cond1 && !gp_exp1 && !gp_normal1 && !gp_sync)
@@ -949,9 +957,26 @@ rcu_torture_writer(void *arg)
 			}
 		}
 		rcutorture_record_progress(++rcu_torture_current_version);
+		/* Cycle through nesting levels of rcu_expedite_gp() calls. */
+		if (can_expedite &&
+		    !(torture_random(&rand) & 0xff & (!!expediting - 1))) {
+			WARN_ON_ONCE(expediting == 0 && rcu_gp_is_expedited());
+			if (expediting >= 0)
+				rcu_expedite_gp();
+			else
+				rcu_unexpedite_gp();
+			if (++expediting > 3)
+				expediting = -expediting;
+		}
 		rcu_torture_writer_state = RTWS_STUTTER;
 		stutter_wait("rcu_torture_writer");
 	} while (!torture_must_stop());
+	/* Reset expediting back to unexpedited. */
+	if (expediting > 0)
+		expediting = -expediting;
+	while (can_expedite && expediting++ < 0)
+		rcu_unexpedite_gp();
+	WARN_ON_ONCE(can_expedite && rcu_gp_is_expedited());
 	rcu_torture_writer_state = RTWS_STOPPING;
 	torture_kthread_stopping("rcu_torture_writer");
 	return 0;
-- 
1.8.1.5

[PATCH tip/core/rcu 3/5] rcu: Update from rcu_expedited variable to rcu_gp_is_expedited()

[Paul E. McKenney]

From: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>

This commit updates open-coded tests of the rcu_expedited variable
to instead use rcu_gp_is_expedited().

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 kernel/rcu/srcu.c        | 2 +-
 kernel/rcu/tree.c        | 9 +++++----
 kernel/rcu/tree_plugin.h | 2 +-
 3 files changed, 7 insertions(+), 6 deletions(-)

diff --git a/kernel/rcu/srcu.c b/kernel/rcu/srcu.c
index 445bf8ffe3fb..c871f07eff69 100644
--- a/kernel/rcu/srcu.c
+++ b/kernel/rcu/srcu.c
@@ -507,7 +507,7 @@ static void __synchronize_srcu(struct srcu_struct *sp, int trycount)
  */
 void synchronize_srcu(struct srcu_struct *sp)
 {
-	__synchronize_srcu(sp, rcu_expedited
+	__synchronize_srcu(sp, rcu_gp_is_expedited()
 			   ? SYNCHRONIZE_SRCU_EXP_TRYCOUNT
 			   : SYNCHRONIZE_SRCU_TRYCOUNT);
 }
diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
index 48d640ca1a05..4325fbe79d84 100644
--- a/kernel/rcu/tree.c
+++ b/kernel/rcu/tree.c
@@ -2954,7 +2954,7 @@ void synchronize_sched(void)
 			   "Illegal synchronize_sched() in RCU-sched read-side critical section");
 	if (rcu_blocking_is_gp())
 		return;
-	if (rcu_expedited)
+	if (rcu_gp_is_expedited())
 		synchronize_sched_expedited();
 	else
 		wait_rcu_gp(call_rcu_sched);
@@ -2981,7 +2981,7 @@ void synchronize_rcu_bh(void)
 			   "Illegal synchronize_rcu_bh() in RCU-bh read-side critical section");
 	if (rcu_blocking_is_gp())
 		return;
-	if (rcu_expedited)
+	if (rcu_gp_is_expedited())
 		synchronize_rcu_bh_expedited();
 	else
 		wait_rcu_gp(call_rcu_bh);
@@ -3660,11 +3660,12 @@ static int rcu_pm_notify(struct notifier_block *self,
 	case PM_HIBERNATION_PREPARE:
 	case PM_SUSPEND_PREPARE:
 		if (nr_cpu_ids <= 256) /* Expediting bad for large systems. */
-			rcu_expedited = 1;
+			rcu_expedite_gp();
 		break;
 	case PM_POST_HIBERNATION:
 	case PM_POST_SUSPEND:
-		rcu_expedited = 0;
+		if (nr_cpu_ids <= 256) /* Expediting bad for large systems. */
+			rcu_unexpedite_gp();
 		break;
 	default:
 		break;
diff --git a/kernel/rcu/tree_plugin.h b/kernel/rcu/tree_plugin.h
index 0a571e9a0f1d..63726b734d34 100644
--- a/kernel/rcu/tree_plugin.h
+++ b/kernel/rcu/tree_plugin.h
@@ -585,7 +585,7 @@ void synchronize_rcu(void)
 			   "Illegal synchronize_rcu() in RCU read-side critical section");
 	if (!rcu_scheduler_active)
 		return;
-	if (rcu_expedited)
+	if (rcu_gp_is_expedited())
 		synchronize_rcu_expedited();
 	else
 		wait_rcu_gp(call_rcu);
-- 
1.8.1.5

[PATCH tip/core/rcu 4/5] rcu: Add Kconfig option to expedite grace periods during boot

[Paul E. McKenney]

From: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>

This commit adds a CONFIG_RCU_EXPEDITE_BOOT Kconfig parameter
that emulates a very early boot rcu_expedite_gp().  A late-boot
call to rcu_end_inkernel_boot() will provide the corresponding
rcu_unexpedite_gp().  The late-boot call to rcu_end_inkernel_boot()
should be made just before init is spawned.

According to Arjan:

> To show the boot time, I'm using the timestamp of the "Write protecting"
> line, that's pretty much the last thing we print prior to ring 3 execution.
>
> A kernel with default RCU behavior (inside KVM, only virtual devices)
> looks like this:
>
> [    0.038724] Write protecting the kernel read-only data: 10240k
>
> a kernel with expedited RCU (using the command line option, so that I
> don't have to recompile between measurements and thus am completely
> oranges-to-oranges)
>
> [    0.031768] Write protecting the kernel read-only data: 10240k
>
> which, in percentage, is an 18% improvement.

Reported-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Arjan van de Ven <arjan@linux.intel.com>
---
 include/linux/rcupdate.h |  1 +
 init/Kconfig             | 13 +++++++++++++
 kernel/rcu/update.c      | 11 ++++++++++-
 3 files changed, 24 insertions(+), 1 deletion(-)

diff --git a/include/linux/rcupdate.h b/include/linux/rcupdate.h
index 57a4d1f73a00..b9f039b11d31 100644
--- a/include/linux/rcupdate.h
+++ b/include/linux/rcupdate.h
@@ -278,6 +278,7 @@ static inline int rcu_preempt_depth(void)
 
 /* Internal to kernel */
 void rcu_init(void);
+void rcu_end_inkernel_boot(void);
 void rcu_sched_qs(void);
 void rcu_bh_qs(void);
 void rcu_check_callbacks(int user);
diff --git a/init/Kconfig b/init/Kconfig
index f5dbc6d4261b..9a0592516f48 100644
--- a/init/Kconfig
+++ b/init/Kconfig
@@ -791,6 +791,19 @@ config RCU_NOCB_CPU_ALL
 
 endchoice
 
+config RCU_EXPEDITE_BOOT
+	bool
+	default n
+	help
+	  This option enables expedited grace periods at boot time,
+	  as if rcu_expedite_gp() had been invoked early in boot.
+	  The corresponding rcu_unexpedite_gp() is invoked from
+	  rcu_end_inkernel_boot(), which is intended to be invoked
+	  at the end of the kernel-only boot sequence, just before
+	  init is exec'ed.
+
+	  Accept the default if unsure.
+
 endmenu # "RCU Subsystem"
 
 config BUILD_BIN2C
diff --git a/kernel/rcu/update.c b/kernel/rcu/update.c
index 5f850823c187..7b12466f90bc 100644
--- a/kernel/rcu/update.c
+++ b/kernel/rcu/update.c
@@ -64,7 +64,8 @@ module_param(rcu_expedited, int, 0);
 
 #ifndef CONFIG_TINY_RCU
 
-static atomic_t rcu_expedited_nesting;
+static atomic_t rcu_expedited_nesting =
+	ATOMIC_INIT(IS_ENABLED(CONFIG_RCU_EXPEDITE_BOOT) ? 1 : 0);
 
 /*
  * Should normal grace-period primitives be expedited?  Intended for
@@ -109,6 +110,14 @@ EXPORT_SYMBOL_GPL(rcu_unexpedite_gp);
 
 #endif /* #ifndef CONFIG_TINY_RCU */
 
+/*
+ * Inform RCU of the end of the in-kernel boot sequence.
+ */
+void rcu_end_inkernel_boot(void)
+{
+	if (IS_ENABLED(CONFIG_RCU_EXPEDITE_BOOT))
+		rcu_unexpedite_gp();
+}
 
 #ifdef CONFIG_PREEMPT_RCU
 
-- 
1.8.1.5

[PATCH tip/core/rcu 5/5] rcutorture: Make consistent use of variables

[Paul E. McKenney]

From: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>

The "if" statement at the beginning of rcu_torture_writer() should
use the same set of variables.  In theory, this does not matter because
the corresponding variables (gp_sync and gp_sync1) have the same value
at this point in the code, but in practice such puzzles should be
removed.  This commit therefore makes the use of variables consistent.

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 kernel/rcu/rcutorture.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/kernel/rcu/rcutorture.c b/kernel/rcu/rcutorture.c
index 3833aa611ae7..8dbe27611ec3 100644
--- a/kernel/rcu/rcutorture.c
+++ b/kernel/rcu/rcutorture.c
@@ -875,7 +875,7 @@ rcu_torture_writer(void *arg)
 		 torture_type);
 
 	/* Initialize synctype[] array.  If none set, take default. */
-	if (!gp_cond1 && !gp_exp1 && !gp_normal1 && !gp_sync)
+	if (!gp_cond1 && !gp_exp1 && !gp_normal1 && !gp_sync1)
 		gp_cond1 = gp_exp1 = gp_normal1 = gp_sync1 = true;
 	if (gp_cond1 && cur_ops->get_state && cur_ops->cond_sync)
 		synctype[nsynctypes++] = RTWS_COND_GET;
-- 
1.8.1.5