Merge branches 'doc.2022.08.31b', 'fixes.2022.08.31b', 'kvfree.2022.08.31b',...

Merge branches 'doc.2022.08.31b', 'fixes.2022.08.31b', 'kvfree.2022.08.31b', 'nocb.2022.09.01a', 'poll.2022.08.31b', 'poll-srcu.2022.08.31b' and 'tasks.2022.08.31b' into HEAD

doc.2022.08.31b: Documentation updates
fixes.2022.08.31b: Miscellaneous fixes
kvfree.2022.08.31b: kvfree_rcu() updates
nocb.2022.09.01a: NOCB CPU updates
poll.2022.08.31b: Full-oldstate RCU polling grace-period API
poll-srcu.2022.08.31b: Polled SRCU grace-period updates
tasks.2022.08.31b: Tasks RCU updates

include/linux/rcupdate.h

@@ -42,7 +42,31 @@ void call_rcu(struct rcu_head *head, rcu_callback_t func);
 void rcu_barrier_tasks(void);
 void rcu_barrier_tasks_rude(void);
 void synchronize_rcu(void);
+
+struct rcu_gp_oldstate;
 unsigned long get_completed_synchronize_rcu(void);
+void get_completed_synchronize_rcu_full(struct rcu_gp_oldstate *rgosp);
+
+// Maximum number of unsigned long values corresponding to
+// not-yet-completed RCU grace periods.
+#define NUM_ACTIVE_RCU_POLL_OLDSTATE 2
+
+/**
+ * same_state_synchronize_rcu - Are two old-state values identical?
+ * @oldstate1: First old-state value.
+ * @oldstate2: Second old-state value.
+ *
+ * The two old-state values must have been obtained from either
+ * get_state_synchronize_rcu(), start_poll_synchronize_rcu(), or
+ * get_completed_synchronize_rcu().  Returns @true if the two values are
+ * identical and @false otherwise.  This allows structures whose lifetimes
+ * are tracked by old-state values to push these values to a list header,
+ * allowing those structures to be slightly smaller.
+ */
+static inline bool same_state_synchronize_rcu(unsigned long oldstate1, unsigned long oldstate2)
+{
+	return oldstate1 == oldstate2;
+}
 
 #ifdef CONFIG_PREEMPT_RCU
 
@@ -496,13 +520,21 @@ do {									      \
  * against NULL.  Although rcu_access_pointer() may also be used in cases
  * where update-side locks prevent the value of the pointer from changing,
  * you should instead use rcu_dereference_protected() for this use case.
+ * Within an RCU read-side critical section, there is little reason to
+ * use rcu_access_pointer().
+ *
+ * It is usually best to test the rcu_access_pointer() return value
+ * directly in order to avoid accidental dereferences being introduced
+ * by later inattentive changes.  In other words, assigning the
+ * rcu_access_pointer() return value to a local variable results in an
+ * accident waiting to happen.
  *
  * It is also permissible to use rcu_access_pointer() when read-side
- * access to the pointer was removed at least one grace period ago, as
- * is the case in the context of the RCU callback that is freeing up
- * the data, or after a synchronize_rcu() returns.  This can be useful
- * when tearing down multi-linked structures after a grace period
- * has elapsed.
+ * access to the pointer was removed at least one grace period ago, as is
+ * the case in the context of the RCU callback that is freeing up the data,
+ * or after a synchronize_rcu() returns.  This can be useful when tearing
+ * down multi-linked structures after a grace period has elapsed.  However,
+ * rcu_dereference_protected() is normally preferred for this use case.
  */
 #define rcu_access_pointer(p) __rcu_access_pointer((p), __UNIQUE_ID(rcu), __rcu)
 

include/linux/rcutiny.h

@@ -14,25 +14,75 @@
 
 #include <asm/param.h> /* for HZ */
 
+struct rcu_gp_oldstate {
+	unsigned long rgos_norm;
+};
+
+// Maximum number of rcu_gp_oldstate values corresponding to
+// not-yet-completed RCU grace periods.
+#define NUM_ACTIVE_RCU_POLL_FULL_OLDSTATE 2
+
+/*
+ * Are the two oldstate values the same?  See the Tree RCU version for
+ * docbook header.
+ */
+static inline bool same_state_synchronize_rcu_full(struct rcu_gp_oldstate *rgosp1,
+						   struct rcu_gp_oldstate *rgosp2)
+{
+	return rgosp1->rgos_norm == rgosp2->rgos_norm;
+}
+
 unsigned long get_state_synchronize_rcu(void);
+
+static inline void get_state_synchronize_rcu_full(struct rcu_gp_oldstate *rgosp)
+{
+	rgosp->rgos_norm = get_state_synchronize_rcu();
+}
+
 unsigned long start_poll_synchronize_rcu(void);
+
+static inline void start_poll_synchronize_rcu_full(struct rcu_gp_oldstate *rgosp)
+{
+	rgosp->rgos_norm = start_poll_synchronize_rcu();
+}
+
 bool poll_state_synchronize_rcu(unsigned long oldstate);
 
+static inline bool poll_state_synchronize_rcu_full(struct rcu_gp_oldstate *rgosp)
+{
+	return poll_state_synchronize_rcu(rgosp->rgos_norm);
+}
+
 static inline void cond_synchronize_rcu(unsigned long oldstate)
 {
 	might_sleep();
 }
 
+static inline void cond_synchronize_rcu_full(struct rcu_gp_oldstate *rgosp)
+{
+	cond_synchronize_rcu(rgosp->rgos_norm);
+}
+
 static inline unsigned long start_poll_synchronize_rcu_expedited(void)
 {
 	return start_poll_synchronize_rcu();
 }
 
+static inline void start_poll_synchronize_rcu_expedited_full(struct rcu_gp_oldstate *rgosp)
+{
+	rgosp->rgos_norm = start_poll_synchronize_rcu_expedited();
+}
+
 static inline void cond_synchronize_rcu_expedited(unsigned long oldstate)
 {
 	cond_synchronize_rcu(oldstate);
 }
 
+static inline void cond_synchronize_rcu_expedited_full(struct rcu_gp_oldstate *rgosp)
+{
+	cond_synchronize_rcu_expedited(rgosp->rgos_norm);
+}
+
 extern void rcu_barrier(void);
 
 static inline void synchronize_rcu_expedited(void)

include/linux/rcutree.h

@@ -40,12 +40,52 @@ bool rcu_eqs_special_set(int cpu);
 void rcu_momentary_dyntick_idle(void);
 void kfree_rcu_scheduler_running(void);
 bool rcu_gp_might_be_stalled(void);
+
+struct rcu_gp_oldstate {
+	unsigned long rgos_norm;
+	unsigned long rgos_exp;
+};
+
+// Maximum number of rcu_gp_oldstate values corresponding to
+// not-yet-completed RCU grace periods.
+#define NUM_ACTIVE_RCU_POLL_FULL_OLDSTATE 4
+
+/**
+ * same_state_synchronize_rcu_full - Are two old-state values identical?
+ * @rgosp1: First old-state value.
+ * @rgosp2: Second old-state value.
+ *
+ * The two old-state values must have been obtained from either
+ * get_state_synchronize_rcu_full(), start_poll_synchronize_rcu_full(),
+ * or get_completed_synchronize_rcu_full().  Returns @true if the two
+ * values are identical and @false otherwise.  This allows structures
+ * whose lifetimes are tracked by old-state values to push these values
+ * to a list header, allowing those structures to be slightly smaller.
+ *
+ * Note that equality is judged on a bitwise basis, so that an
+ * @rcu_gp_oldstate structure with an already-completed state in one field
+ * will compare not-equal to a structure with an already-completed state
+ * in the other field.  After all, the @rcu_gp_oldstate structure is opaque
+ * so how did such a situation come to pass in the first place?
+ */
+static inline bool same_state_synchronize_rcu_full(struct rcu_gp_oldstate *rgosp1,
+						   struct rcu_gp_oldstate *rgosp2)
+{
+	return rgosp1->rgos_norm == rgosp2->rgos_norm && rgosp1->rgos_exp == rgosp2->rgos_exp;
+}
+
 unsigned long start_poll_synchronize_rcu_expedited(void);
+void start_poll_synchronize_rcu_expedited_full(struct rcu_gp_oldstate *rgosp);
 void cond_synchronize_rcu_expedited(unsigned long oldstate);
+void cond_synchronize_rcu_expedited_full(struct rcu_gp_oldstate *rgosp);
 unsigned long get_state_synchronize_rcu(void);
+void get_state_synchronize_rcu_full(struct rcu_gp_oldstate *rgosp);
 unsigned long start_poll_synchronize_rcu(void);
+void start_poll_synchronize_rcu_full(struct rcu_gp_oldstate *rgosp);
 bool poll_state_synchronize_rcu(unsigned long oldstate);
+bool poll_state_synchronize_rcu_full(struct rcu_gp_oldstate *rgosp);
 void cond_synchronize_rcu(unsigned long oldstate);
+void cond_synchronize_rcu_full(struct rcu_gp_oldstate *rgosp);
 
 bool rcu_is_idle_cpu(int cpu);
 

include/linux/srcutiny.h

@@ -15,10 +15,10 @@
 
 struct srcu_struct {
 	short srcu_lock_nesting[2];	/* srcu_read_lock() nesting depth. */
-	unsigned short srcu_idx;	/* Current reader array element in bit 0x2. */
-	unsigned short srcu_idx_max;	/* Furthest future srcu_idx request. */
 	u8 srcu_gp_running;		/* GP workqueue running? */
 	u8 srcu_gp_waiting;		/* GP waiting for readers? */
+	unsigned long srcu_idx;		/* Current reader array element in bit 0x2. */
+	unsigned long srcu_idx_max;	/* Furthest future srcu_idx request. */
 	struct swait_queue_head srcu_wq;
 					/* Last srcu_read_unlock() wakes GP. */
 	struct rcu_head *srcu_cb_head;	/* Pending callbacks: Head. */
@@ -82,10 +82,12 @@ static inline void srcu_torture_stats_print(struct srcu_struct *ssp,
 	int idx;
 
 	idx = ((data_race(READ_ONCE(ssp->srcu_idx)) + 1) & 0x2) >> 1;
-	pr_alert("%s%s Tiny SRCU per-CPU(idx=%d): (%hd,%hd)\n",
+	pr_alert("%s%s Tiny SRCU per-CPU(idx=%d): (%hd,%hd) gp: %lu->%lu\n",
 		 tt, tf, idx,
 		 data_race(READ_ONCE(ssp->srcu_lock_nesting[!idx])),
-		 data_race(READ_ONCE(ssp->srcu_lock_nesting[idx])));
+		 data_race(READ_ONCE(ssp->srcu_lock_nesting[idx])),
+		 data_race(READ_ONCE(ssp->srcu_idx)),
+		 data_race(READ_ONCE(ssp->srcu_idx_max)));
 }
 
 #endif

kernel/rcu/rcutorture.c

@@ -84,10 +84,15 @@ torture_param(int, fwd_progress_holdoff, 60, "Time between forward-progress test
 torture_param(bool, fwd_progress_need_resched, 1, "Hide cond_resched() behind need_resched()");
 torture_param(bool, gp_cond, false, "Use conditional/async GP wait primitives");
 torture_param(bool, gp_cond_exp, false, "Use conditional/async expedited GP wait primitives");
+torture_param(bool, gp_cond_full, false, "Use conditional/async full-state GP wait primitives");
+torture_param(bool, gp_cond_exp_full, false,
+		    "Use conditional/async full-stateexpedited GP wait primitives");
 torture_param(bool, gp_exp, false, "Use expedited GP wait primitives");
 torture_param(bool, gp_normal, false, "Use normal (non-expedited) GP wait primitives");
 torture_param(bool, gp_poll, false, "Use polling GP wait primitives");
 torture_param(bool, gp_poll_exp, false, "Use polling expedited GP wait primitives");
+torture_param(bool, gp_poll_full, false, "Use polling full-state GP wait primitives");
+torture_param(bool, gp_poll_exp_full, false, "Use polling full-state expedited GP wait primitives");
 torture_param(bool, gp_sync, false, "Use synchronous GP wait primitives");
 torture_param(int, irqreader, 1, "Allow RCU readers from irq handlers");
 torture_param(int, leakpointer, 0, "Leak pointer dereferences from readers");
@@ -194,16 +199,24 @@ static int rcu_torture_writer_state;
 #define RTWS_DEF_FREE		3
 #define RTWS_EXP_SYNC		4
 #define RTWS_COND_GET		5
-#define RTWS_COND_GET_EXP	6
-#define RTWS_COND_SYNC		7
-#define RTWS_COND_SYNC_EXP	8
-#define RTWS_POLL_GET		9
-#define RTWS_POLL_GET_EXP	10
-#define RTWS_POLL_WAIT		11
-#define RTWS_POLL_WAIT_EXP	12
-#define RTWS_SYNC		13
-#define RTWS_STUTTER		14
-#define RTWS_STOPPING		15
+#define RTWS_COND_GET_FULL	6
+#define RTWS_COND_GET_EXP	7
+#define RTWS_COND_GET_EXP_FULL	8
+#define RTWS_COND_SYNC		9
+#define RTWS_COND_SYNC_FULL	10
+#define RTWS_COND_SYNC_EXP	11
+#define RTWS_COND_SYNC_EXP_FULL	12
+#define RTWS_POLL_GET		13
+#define RTWS_POLL_GET_FULL	14
+#define RTWS_POLL_GET_EXP	15
+#define RTWS_POLL_GET_EXP_FULL	16
+#define RTWS_POLL_WAIT		17
+#define RTWS_POLL_WAIT_FULL	18
+#define RTWS_POLL_WAIT_EXP	19
+#define RTWS_POLL_WAIT_EXP_FULL	20
+#define RTWS_SYNC		21
+#define RTWS_STUTTER		22
+#define RTWS_STOPPING		23
 static const char * const rcu_torture_writer_state_names[] = {
 	"RTWS_FIXED_DELAY",
 	"RTWS_DELAY",
@@ -211,13 +224,21 @@ static const char * const rcu_torture_writer_state_names[] = {
 	"RTWS_DEF_FREE",
 	"RTWS_EXP_SYNC",
 	"RTWS_COND_GET",
+	"RTWS_COND_GET_FULL",
 	"RTWS_COND_GET_EXP",
+	"RTWS_COND_GET_EXP_FULL",
 	"RTWS_COND_SYNC",
+	"RTWS_COND_SYNC_FULL",
 	"RTWS_COND_SYNC_EXP",
+	"RTWS_COND_SYNC_EXP_FULL",
 	"RTWS_POLL_GET",
+	"RTWS_POLL_GET_FULL",
 	"RTWS_POLL_GET_EXP",
+	"RTWS_POLL_GET_EXP_FULL",
 	"RTWS_POLL_WAIT",
+	"RTWS_POLL_WAIT_FULL",
 	"RTWS_POLL_WAIT_EXP",
+	"RTWS_POLL_WAIT_EXP_FULL",
 	"RTWS_SYNC",
 	"RTWS_STUTTER",
 	"RTWS_STOPPING",
@@ -332,13 +353,21 @@ struct rcu_torture_ops {
 	void (*exp_sync)(void);
 	unsigned long (*get_gp_state_exp)(void);
 	unsigned long (*start_gp_poll_exp)(void);
+	void (*start_gp_poll_exp_full)(struct rcu_gp_oldstate *rgosp);
 	bool (*poll_gp_state_exp)(unsigned long oldstate);
 	void (*cond_sync_exp)(unsigned long oldstate);
+	void (*cond_sync_exp_full)(struct rcu_gp_oldstate *rgosp);
 	unsigned long (*get_gp_state)(void);
+	void (*get_gp_state_full)(struct rcu_gp_oldstate *rgosp);
 	unsigned long (*get_gp_completed)(void);
+	void (*get_gp_completed_full)(struct rcu_gp_oldstate *rgosp);
 	unsigned long (*start_gp_poll)(void);
+	void (*start_gp_poll_full)(struct rcu_gp_oldstate *rgosp);
 	bool (*poll_gp_state)(unsigned long oldstate);
+	bool (*poll_gp_state_full)(struct rcu_gp_oldstate *rgosp);
+	bool (*poll_need_2gp)(bool poll, bool poll_full);
 	void (*cond_sync)(unsigned long oldstate);
+	void (*cond_sync_full)(struct rcu_gp_oldstate *rgosp);
 	call_rcu_func_t call;
 	void (*cb_barrier)(void);
 	void (*fqs)(void);
@@ -489,6 +518,11 @@ static void rcu_sync_torture_init(void)
 	INIT_LIST_HEAD(&rcu_torture_removed);
 }
 
+static bool rcu_poll_need_2gp(bool poll, bool poll_full)
+{
+	return poll;
+}
+
 static struct rcu_torture_ops rcu_ops = {
 	.ttype			= RCU_FLAVOR,
 	.init			= rcu_sync_torture_init,
@@ -502,12 +536,19 @@ static struct rcu_torture_ops rcu_ops = {
 	.sync			= synchronize_rcu,
 	.exp_sync		= synchronize_rcu_expedited,
 	.get_gp_state		= get_state_synchronize_rcu,
+	.get_gp_state_full	= get_state_synchronize_rcu_full,
 	.get_gp_completed	= get_completed_synchronize_rcu,
+	.get_gp_completed_full	= get_completed_synchronize_rcu_full,
 	.start_gp_poll		= start_poll_synchronize_rcu,
+	.start_gp_poll_full	= start_poll_synchronize_rcu_full,
 	.poll_gp_state		= poll_state_synchronize_rcu,
+	.poll_gp_state_full	= poll_state_synchronize_rcu_full,
+	.poll_need_2gp		= rcu_poll_need_2gp,
 	.cond_sync		= cond_synchronize_rcu,
+	.cond_sync_full		= cond_synchronize_rcu_full,
 	.get_gp_state_exp	= get_state_synchronize_rcu,
 	.start_gp_poll_exp	= start_poll_synchronize_rcu_expedited,
+	.start_gp_poll_exp_full	= start_poll_synchronize_rcu_expedited_full,
 	.poll_gp_state_exp	= poll_state_synchronize_rcu,
 	.cond_sync_exp		= cond_synchronize_rcu_expedited,
 	.call			= call_rcu,
@@ -709,6 +750,9 @@ static struct rcu_torture_ops srcud_ops = {
 	.deferred_free	= srcu_torture_deferred_free,
 	.sync		= srcu_torture_synchronize,
 	.exp_sync	= srcu_torture_synchronize_expedited,
+	.get_gp_state	= srcu_torture_get_gp_state,
+	.start_gp_poll	= srcu_torture_start_gp_poll,
+	.poll_gp_state	= srcu_torture_poll_gp_state,
 	.call		= srcu_torture_call,
 	.cb_barrier	= srcu_torture_barrier,
 	.stats		= srcu_torture_stats,
@@ -1148,15 +1192,35 @@ static int nsynctypes;
  */
 static void rcu_torture_write_types(void)
 {
-	bool gp_cond1 = gp_cond, gp_cond_exp1 = gp_cond_exp, gp_exp1 = gp_exp;
-	bool gp_poll_exp1 = gp_poll_exp, gp_normal1 = gp_normal, gp_poll1 = gp_poll;
-	bool gp_sync1 = gp_sync;
+	bool gp_cond1 = gp_cond, gp_cond_exp1 = gp_cond_exp, gp_cond_full1 = gp_cond_full;
+	bool gp_cond_exp_full1 = gp_cond_exp_full, gp_exp1 = gp_exp, gp_poll_exp1 = gp_poll_exp;
+	bool gp_poll_exp_full1 = gp_poll_exp_full, gp_normal1 = gp_normal, gp_poll1 = gp_poll;
+	bool gp_poll_full1 = gp_poll_full, gp_sync1 = gp_sync;
 
 	/* Initialize synctype[] array.  If none set, take default. */
-	if (!gp_cond1 && !gp_cond_exp1 && !gp_exp1 && !gp_poll_exp &&
-	    !gp_normal1 && !gp_poll1 && !gp_sync1)
-		gp_cond1 = gp_cond_exp1 = gp_exp1 = gp_poll_exp1 =
-			   gp_normal1 = gp_poll1 = gp_sync1 = true;
+	if (!gp_cond1 &&
+	    !gp_cond_exp1 &&
+	    !gp_cond_full1 &&
+	    !gp_cond_exp_full1 &&
+	    !gp_exp1 &&
+	    !gp_poll_exp1 &&
+	    !gp_poll_exp_full1 &&
+	    !gp_normal1 &&
+	    !gp_poll1 &&
+	    !gp_poll_full1 &&
+	    !gp_sync1) {
+		gp_cond1 = true;
+		gp_cond_exp1 = true;
+		gp_cond_full1 = true;
+		gp_cond_exp_full1 = true;
+		gp_exp1 = true;
+		gp_poll_exp1 = true;
+		gp_poll_exp_full1 = true;
+		gp_normal1 = true;
+		gp_poll1 = true;
+		gp_poll_full1 = true;
+		gp_sync1 = true;
+	}
 	if (gp_cond1 && cur_ops->get_gp_state && cur_ops->cond_sync) {
 		synctype[nsynctypes++] = RTWS_COND_GET;
 		pr_info("%s: Testing conditional GPs.\n", __func__);
@@ -1169,6 +1233,19 @@ static void rcu_torture_write_types(void)
 	} else if (gp_cond_exp && (!cur_ops->get_gp_state_exp || !cur_ops->cond_sync_exp)) {
 		pr_alert("%s: gp_cond_exp without primitives.\n", __func__);
 	}
+	if (gp_cond_full1 && cur_ops->get_gp_state && cur_ops->cond_sync_full) {
+		synctype[nsynctypes++] = RTWS_COND_GET_FULL;
+		pr_info("%s: Testing conditional full-state GPs.\n", __func__);
+	} else if (gp_cond_full && (!cur_ops->get_gp_state || !cur_ops->cond_sync_full)) {
+		pr_alert("%s: gp_cond_full without primitives.\n", __func__);
+	}
+	if (gp_cond_exp_full1 && cur_ops->get_gp_state_exp && cur_ops->cond_sync_exp_full) {
+		synctype[nsynctypes++] = RTWS_COND_GET_EXP_FULL;
+		pr_info("%s: Testing conditional full-state expedited GPs.\n", __func__);
+	} else if (gp_cond_exp_full &&
+		   (!cur_ops->get_gp_state_exp || !cur_ops->cond_sync_exp_full)) {
+		pr_alert("%s: gp_cond_exp_full without primitives.\n", __func__);
+	}
 	if (gp_exp1 && cur_ops->exp_sync) {
 		synctype[nsynctypes++] = RTWS_EXP_SYNC;
 		pr_info("%s: Testing expedited GPs.\n", __func__);
@@ -1187,12 +1264,25 @@ static void rcu_torture_write_types(void)
 	} else if (gp_poll && (!cur_ops->start_gp_poll || !cur_ops->poll_gp_state)) {
 		pr_alert("%s: gp_poll without primitives.\n", __func__);
 	}
+	if (gp_poll_full1 && cur_ops->start_gp_poll_full && cur_ops->poll_gp_state_full) {
+		synctype[nsynctypes++] = RTWS_POLL_GET_FULL;
+		pr_info("%s: Testing polling full-state GPs.\n", __func__);
+	} else if (gp_poll_full && (!cur_ops->start_gp_poll_full || !cur_ops->poll_gp_state_full)) {
+		pr_alert("%s: gp_poll_full without primitives.\n", __func__);
+	}
 	if (gp_poll_exp1 && cur_ops->start_gp_poll_exp && cur_ops->poll_gp_state_exp) {
 		synctype[nsynctypes++] = RTWS_POLL_GET_EXP;
 		pr_info("%s: Testing polling expedited GPs.\n", __func__);
 	} else if (gp_poll_exp && (!cur_ops->start_gp_poll_exp || !cur_ops->poll_gp_state_exp)) {
 		pr_alert("%s: gp_poll_exp without primitives.\n", __func__);
 	}
+	if (gp_poll_exp_full1 && cur_ops->start_gp_poll_exp_full && cur_ops->poll_gp_state_full) {
+		synctype[nsynctypes++] = RTWS_POLL_GET_EXP_FULL;
+		pr_info("%s: Testing polling full-state expedited GPs.\n", __func__);
+	} else if (gp_poll_exp_full &&
+		   (!cur_ops->start_gp_poll_exp_full || !cur_ops->poll_gp_state_full)) {
+		pr_alert("%s: gp_poll_exp_full without primitives.\n", __func__);
+	}
 	if (gp_sync1 && cur_ops->sync) {
 		synctype[nsynctypes++] = RTWS_SYNC;
 		pr_info("%s: Testing normal GPs.\n", __func__);
@@ -1201,6 +1291,40 @@ static void rcu_torture_write_types(void)
 	}
 }
 
+/*
+ * Do the specified rcu_torture_writer() synchronous grace period,
+ * while also testing out the polled APIs.  Note well that the single-CPU
+ * grace-period optimizations must be accounted for.
+ */
+static void do_rtws_sync(struct torture_random_state *trsp, void (*sync)(void))
+{
+	unsigned long cookie;
+	struct rcu_gp_oldstate cookie_full;
+	bool dopoll;
+	bool dopoll_full;
+	unsigned long r = torture_random(trsp);
+
+	dopoll = cur_ops->get_gp_state && cur_ops->poll_gp_state && !(r & 0x300);
+	dopoll_full = cur_ops->get_gp_state_full && cur_ops->poll_gp_state_full && !(r & 0xc00);
+	if (dopoll || dopoll_full)
+		cpus_read_lock();
+	if (dopoll)
+		cookie = cur_ops->get_gp_state();
+	if (dopoll_full)
+		cur_ops->get_gp_state_full(&cookie_full);
+	if (cur_ops->poll_need_2gp && cur_ops->poll_need_2gp(dopoll, dopoll_full))
+		sync();
+	sync();
+	WARN_ONCE(dopoll && !cur_ops->poll_gp_state(cookie),
+		  "%s: Cookie check 3 failed %pS() online %*pbl.",
+		  __func__, sync, cpumask_pr_args(cpu_online_mask));
+	WARN_ONCE(dopoll_full && !cur_ops->poll_gp_state_full(&cookie_full),
+		  "%s: Cookie check 4 failed %pS() online %*pbl",
+		  __func__, sync, cpumask_pr_args(cpu_online_mask));
+	if (dopoll || dopoll_full)
+		cpus_read_unlock();
+}
+
 /*
  * RCU torture writer kthread.  Repeatedly substitutes a new structure
  * for that pointed to by rcu_torture_current, freeing the old structure
@@ -1212,8 +1336,10 @@ rcu_torture_writer(void *arg)
 	bool boot_ended;
 	bool can_expedite = !rcu_gp_is_expedited() && !rcu_gp_is_normal();
 	unsigned long cookie;
+	struct rcu_gp_oldstate cookie_full;
 	int expediting = 0;
 	unsigned long gp_snap;
+	struct rcu_gp_oldstate gp_snap_full;
 	int i;
 	int idx;
 	int oldnice = task_nice(current);
@@ -1261,11 +1387,12 @@ rcu_torture_writer(void *arg)
 			atomic_inc(&rcu_torture_wcount[i]);
 			WRITE_ONCE(old_rp->rtort_pipe_count,
 				   old_rp->rtort_pipe_count + 1);
+
+			// Make sure readers block polled grace periods.
 			if (cur_ops->get_gp_state && cur_ops->poll_gp_state) {
 				idx = cur_ops->readlock();
 				cookie = cur_ops->get_gp_state();
-				WARN_ONCE(rcu_torture_writer_state != RTWS_DEF_FREE &&
-					  cur_ops->poll_gp_state(cookie),
+				WARN_ONCE(cur_ops->poll_gp_state(cookie),
 					  "%s: Cookie check 1 failed %s(%d) %lu->%lu\n",
 					  __func__,
 					  rcu_torture_writer_state_getname(),
@@ -1277,6 +1404,21 @@ rcu_torture_writer(void *arg)
 				}
 				cur_ops->readunlock(idx);
 			}
+			if (cur_ops->get_gp_state_full && cur_ops->poll_gp_state_full) {
+				idx = cur_ops->readlock();
+				cur_ops->get_gp_state_full(&cookie_full);
+				WARN_ONCE(cur_ops->poll_gp_state_full(&cookie_full),
+					  "%s: Cookie check 5 failed %s(%d) online %*pbl\n",
+					  __func__,
+					  rcu_torture_writer_state_getname(),
+					  rcu_torture_writer_state,
+					  cpumask_pr_args(cpu_online_mask));
+				if (cur_ops->get_gp_completed_full) {
+					cur_ops->get_gp_completed_full(&cookie_full);
+					WARN_ON_ONCE(!cur_ops->poll_gp_state_full(&cookie_full));
+				}
+				cur_ops->readunlock(idx);
+			}
 			switch (synctype[torture_random(&rand) % nsynctypes]) {
 			case RTWS_DEF_FREE:
 				rcu_torture_writer_state = RTWS_DEF_FREE;
@@ -1284,12 +1426,7 @@ rcu_torture_writer(void *arg)
 				break;
 			case RTWS_EXP_SYNC:
 				rcu_torture_writer_state = RTWS_EXP_SYNC;
-				if (cur_ops->get_gp_state && cur_ops->poll_gp_state)
-					cookie = cur_ops->get_gp_state();
-				cur_ops->exp_sync();
-				cur_ops->exp_sync();
-				if (cur_ops->get_gp_state && cur_ops->poll_gp_state)
-					WARN_ON_ONCE(!cur_ops->poll_gp_state(cookie));
+				do_rtws_sync(&rand, cur_ops->exp_sync);
 				rcu_torture_pipe_update(old_rp);
 				break;
 			case RTWS_COND_GET:
@@ -1308,6 +1445,22 @@ rcu_torture_writer(void *arg)
 				cur_ops->cond_sync_exp(gp_snap);
 				rcu_torture_pipe_update(old_rp);
 				break;
+			case RTWS_COND_GET_FULL:
+				rcu_torture_writer_state = RTWS_COND_GET_FULL;
+				cur_ops->get_gp_state_full(&gp_snap_full);
+				torture_hrtimeout_jiffies(torture_random(&rand) % 16, &rand);
+				rcu_torture_writer_state = RTWS_COND_SYNC_FULL;
+				cur_ops->cond_sync_full(&gp_snap_full);
+				rcu_torture_pipe_update(old_rp);
+				break;
+			case RTWS_COND_GET_EXP_FULL:
+				rcu_torture_writer_state = RTWS_COND_GET_EXP_FULL;
+				cur_ops->get_gp_state_full(&gp_snap_full);
+				torture_hrtimeout_jiffies(torture_random(&rand) % 16, &rand);
+				rcu_torture_writer_state = RTWS_COND_SYNC_EXP_FULL;
+				cur_ops->cond_sync_exp_full(&gp_snap_full);
+				rcu_torture_pipe_update(old_rp);
+				break;
 			case RTWS_POLL_GET:
 				rcu_torture_writer_state = RTWS_POLL_GET;
 				gp_snap = cur_ops->start_gp_poll();
@@ -1317,6 +1470,15 @@ rcu_torture_writer(void *arg)
 								  &rand);
 				rcu_torture_pipe_update(old_rp);
 				break;
+			case RTWS_POLL_GET_FULL:
+				rcu_torture_writer_state = RTWS_POLL_GET_FULL;
+				cur_ops->start_gp_poll_full(&gp_snap_full);
+				rcu_torture_writer_state = RTWS_POLL_WAIT_FULL;
+				while (!cur_ops->poll_gp_state_full(&gp_snap_full))
+					torture_hrtimeout_jiffies(torture_random(&rand) % 16,
+								  &rand);
+				rcu_torture_pipe_update(old_rp);
+				break;
 			case RTWS_POLL_GET_EXP:
 				rcu_torture_writer_state = RTWS_POLL_GET_EXP;
 				gp_snap = cur_ops->start_gp_poll_exp();
@@ -1326,14 +1488,18 @@ rcu_torture_writer(void *arg)
 								  &rand);
 				rcu_torture_pipe_update(old_rp);
 				break;
+			case RTWS_POLL_GET_EXP_FULL:
+				rcu_torture_writer_state = RTWS_POLL_GET_EXP_FULL;
+				cur_ops->start_gp_poll_exp_full(&gp_snap_full);
+				rcu_torture_writer_state = RTWS_POLL_WAIT_EXP_FULL;
+				while (!cur_ops->poll_gp_state_full(&gp_snap_full))
+					torture_hrtimeout_jiffies(torture_random(&rand) % 16,
+								  &rand);
+				rcu_torture_pipe_update(old_rp);
+				break;
 			case RTWS_SYNC:
 				rcu_torture_writer_state = RTWS_SYNC;
-				if (cur_ops->get_gp_state && cur_ops->poll_gp_state)
-					cookie = cur_ops->get_gp_state();
-				cur_ops->sync();
-				cur_ops->sync();
-				if (cur_ops->get_gp_state && cur_ops->poll_gp_state)
-					WARN_ON_ONCE(!cur_ops->poll_gp_state(cookie));
+				do_rtws_sync(&rand, cur_ops->sync);
 				rcu_torture_pipe_update(old_rp);
 				break;
 			default:
@@ -1400,6 +1566,7 @@ static int
 rcu_torture_fakewriter(void *arg)
 {
 	unsigned long gp_snap;
+	struct rcu_gp_oldstate gp_snap_full;
 	DEFINE_TORTURE_RANDOM(rand);
 
 	VERBOSE_TOROUT_STRING("rcu_torture_fakewriter task started");
@@ -1438,6 +1605,16 @@ rcu_torture_fakewriter(void *arg)
 				torture_hrtimeout_jiffies(torture_random(&rand) % 16, &rand);
 				cur_ops->cond_sync_exp(gp_snap);
 				break;
+			case RTWS_COND_GET_FULL:
+				cur_ops->get_gp_state_full(&gp_snap_full);
+				torture_hrtimeout_jiffies(torture_random(&rand) % 16, &rand);
+				cur_ops->cond_sync_full(&gp_snap_full);
+				break;
+			case RTWS_COND_GET_EXP_FULL:
+				cur_ops->get_gp_state_full(&gp_snap_full);
+				torture_hrtimeout_jiffies(torture_random(&rand) % 16, &rand);
+				cur_ops->cond_sync_exp_full(&gp_snap_full);
+				break;
 			case RTWS_POLL_GET:
 				gp_snap = cur_ops->start_gp_poll();
 				while (!cur_ops->poll_gp_state(gp_snap)) {
@@ -1445,6 +1622,13 @@ rcu_torture_fakewriter(void *arg)
 								  &rand);
 				}
 				break;
+			case RTWS_POLL_GET_FULL:
+				cur_ops->start_gp_poll_full(&gp_snap_full);
+				while (!cur_ops->poll_gp_state_full(&gp_snap_full)) {
+					torture_hrtimeout_jiffies(torture_random(&rand) % 16,
+								  &rand);
+				}
+				break;
 			case RTWS_POLL_GET_EXP:
 				gp_snap = cur_ops->start_gp_poll_exp();
 				while (!cur_ops->poll_gp_state_exp(gp_snap)) {
@@ -1452,6 +1636,13 @@ rcu_torture_fakewriter(void *arg)
 								  &rand);
 				}
 				break;
+			case RTWS_POLL_GET_EXP_FULL:
+				cur_ops->start_gp_poll_exp_full(&gp_snap_full);
+				while (!cur_ops->poll_gp_state_full(&gp_snap_full)) {
+					torture_hrtimeout_jiffies(torture_random(&rand) % 16,
+								  &rand);
+				}
+				break;
 			case RTWS_SYNC:
 				cur_ops->sync();
 				break;
@@ -1715,7 +1906,9 @@ rcutorture_loop_extend(int *readstate, struct torture_random_state *trsp,
  */
 static bool rcu_torture_one_read(struct torture_random_state *trsp, long myid)
 {
+	bool checkpolling = !(torture_random(trsp) & 0xfff);
 	unsigned long cookie;
+	struct rcu_gp_oldstate cookie_full;
 	int i;
 	unsigned long started;
 	unsigned long completed;
@@ -1731,8 +1924,12 @@ static bool rcu_torture_one_read(struct torture_random_state *trsp, long myid)
 	WARN_ON_ONCE(!rcu_is_watching());
 	newstate = rcutorture_extend_mask(readstate, trsp);
 	rcutorture_one_extend(&readstate, newstate, trsp, rtrsp++);
-	if (cur_ops->get_gp_state && cur_ops->poll_gp_state)
-		cookie = cur_ops->get_gp_state();
+	if (checkpolling) {
+		if (cur_ops->get_gp_state && cur_ops->poll_gp_state)
+			cookie = cur_ops->get_gp_state();
+		if (cur_ops->get_gp_state_full && cur_ops->poll_gp_state_full)
+			cur_ops->get_gp_state_full(&cookie_full);
+	}
 	started = cur_ops->get_gp_seq();
 	ts = rcu_trace_clock_local();
 	p = rcu_dereference_check(rcu_torture_current,
@@ -1766,13 +1963,22 @@ static bool rcu_torture_one_read(struct torture_random_state *trsp, long myid)
 	}
 	__this_cpu_inc(rcu_torture_batch[completed]);
 	preempt_enable();
-	if (cur_ops->get_gp_state && cur_ops->poll_gp_state)
-		WARN_ONCE(cur_ops->poll_gp_state(cookie),
-			  "%s: Cookie check 2 failed %s(%d) %lu->%lu\n",
-			  __func__,
-			  rcu_torture_writer_state_getname(),
-			  rcu_torture_writer_state,
-			  cookie, cur_ops->get_gp_state());
+	if (checkpolling) {
+		if (cur_ops->get_gp_state && cur_ops->poll_gp_state)
+			WARN_ONCE(cur_ops->poll_gp_state(cookie),
+				  "%s: Cookie check 2 failed %s(%d) %lu->%lu\n",
+				  __func__,
+				  rcu_torture_writer_state_getname(),
+				  rcu_torture_writer_state,
+				  cookie, cur_ops->get_gp_state());
+		if (cur_ops->get_gp_state_full && cur_ops->poll_gp_state_full)
+			WARN_ONCE(cur_ops->poll_gp_state_full(&cookie_full),
+				  "%s: Cookie check 6 failed %s(%d) online %*pbl\n",
+				  __func__,
+				  rcu_torture_writer_state_getname(),
+				  rcu_torture_writer_state,
+				  cpumask_pr_args(cpu_online_mask));
+	}
 	rcutorture_one_extend(&readstate, 0, trsp, rtrsp);
 	WARN_ON_ONCE(readstate);
 	// This next splat is expected behavior if leakpointer, especially
@@ -2600,12 +2806,12 @@ static int rcutorture_oom_notify(struct notifier_block *self,
 	for (i = 0; i < fwd_progress; i++)
 		ncbs += rcu_torture_fwd_prog_cbfree(&rfp[i]);
 	pr_info("%s: Freed %lu RCU callbacks.\n", __func__, ncbs);
-	rcu_barrier();
+	cur_ops->cb_barrier();
 	ncbs = 0;
 	for (i = 0; i < fwd_progress; i++)
 		ncbs += rcu_torture_fwd_prog_cbfree(&rfp[i]);
 	pr_info("%s: Freed %lu RCU callbacks.\n", __func__, ncbs);
-	rcu_barrier();
+	cur_ops->cb_barrier();
 	ncbs = 0;
 	for (i = 0; i < fwd_progress; i++)
 		ncbs += rcu_torture_fwd_prog_cbfree(&rfp[i]);

kernel/rcu/srcutiny.c

@@ -117,7 +117,7 @@ void srcu_drive_gp(struct work_struct *wp)
 	struct srcu_struct *ssp;
 
 	ssp = container_of(wp, struct srcu_struct, srcu_work);
-	if (ssp->srcu_gp_running || USHORT_CMP_GE(ssp->srcu_idx, READ_ONCE(ssp->srcu_idx_max)))
+	if (ssp->srcu_gp_running || ULONG_CMP_GE(ssp->srcu_idx, READ_ONCE(ssp->srcu_idx_max)))
 		return; /* Already running or nothing to do. */
 
 	/* Remove recently arrived callbacks and wait for readers. */
@@ -150,17 +150,17 @@ void srcu_drive_gp(struct work_struct *wp)
 	 * straighten that out.
 	 */
 	WRITE_ONCE(ssp->srcu_gp_running, false);
-	if (USHORT_CMP_LT(ssp->srcu_idx, READ_ONCE(ssp->srcu_idx_max)))
+	if (ULONG_CMP_LT(ssp->srcu_idx, READ_ONCE(ssp->srcu_idx_max)))
 		schedule_work(&ssp->srcu_work);
 }
 EXPORT_SYMBOL_GPL(srcu_drive_gp);
 
 static void srcu_gp_start_if_needed(struct srcu_struct *ssp)
 {
-	unsigned short cookie;
+	unsigned long cookie;
 
 	cookie = get_state_synchronize_srcu(ssp);
-	if (USHORT_CMP_GE(READ_ONCE(ssp->srcu_idx_max), cookie))
+	if (ULONG_CMP_GE(READ_ONCE(ssp->srcu_idx_max), cookie))
 		return;
 	WRITE_ONCE(ssp->srcu_idx_max, cookie);
 	if (!READ_ONCE(ssp->srcu_gp_running)) {
@@ -215,7 +215,7 @@ unsigned long get_state_synchronize_srcu(struct srcu_struct *ssp)
 	barrier();
 	ret = (READ_ONCE(ssp->srcu_idx) + 3) & ~0x1;
 	barrier();
-	return ret & USHRT_MAX;
+	return ret;
 }
 EXPORT_SYMBOL_GPL(get_state_synchronize_srcu);
 
@@ -240,10 +240,10 @@ EXPORT_SYMBOL_GPL(start_poll_synchronize_srcu);
  */
 bool poll_state_synchronize_srcu(struct srcu_struct *ssp, unsigned long cookie)
 {
-	bool ret = USHORT_CMP_GE(READ_ONCE(ssp->srcu_idx), cookie);
+	unsigned long cur_s = READ_ONCE(ssp->srcu_idx);
 
 	barrier();
-	return ret;
+	return ULONG_CMP_GE(cur_s, cookie) || ULONG_CMP_LT(cur_s, cookie - 3);
 }
 EXPORT_SYMBOL_GPL(poll_state_synchronize_srcu);
 

kernel/rcu/tasks.h

@@ -560,7 +560,7 @@ static int __noreturn rcu_tasks_kthread(void *arg)
 static void synchronize_rcu_tasks_generic(struct rcu_tasks *rtp)
 {
 	/* Complain if the scheduler has not started.  */
-	RCU_LOCKDEP_WARN(rcu_scheduler_active == RCU_SCHEDULER_INACTIVE,
+	WARN_ONCE(rcu_scheduler_active == RCU_SCHEDULER_INACTIVE,
 			 "synchronize_rcu_tasks called too soon");
 
 	// If the grace-period kthread is running, use it.
@@ -1500,6 +1500,7 @@ static void rcu_tasks_trace_pregp_step(struct list_head *hop)
 		if (rcu_tasks_trace_pertask_prep(t, true))
 			trc_add_holdout(t, hop);
 		rcu_read_unlock();
+		cond_resched_tasks_rcu_qs();
 	}
 
 	// Only after all running tasks have been accounted for is it
@@ -1520,6 +1521,7 @@ static void rcu_tasks_trace_pregp_step(struct list_head *hop)
 			raw_spin_lock_irqsave_rcu_node(rtpcp, flags);
 		}
 		raw_spin_unlock_irqrestore_rcu_node(rtpcp, flags);
+		cond_resched_tasks_rcu_qs();
 	}
 
 	// Re-enable CPU hotplug now that the holdout list is populated.
@@ -1619,6 +1621,7 @@ static void check_all_holdout_tasks_trace(struct list_head *hop,
 			trc_del_holdout(t);
 		else if (needreport)
 			show_stalled_task_trace(t, firstreport);
+		cond_resched_tasks_rcu_qs();
 	}
 
 	// Re-enable CPU hotplug now that the holdout list scan has completed.

kernel/rcu/tiny.c

@@ -158,6 +158,10 @@ void synchronize_rcu(void)
 }
 EXPORT_SYMBOL_GPL(synchronize_rcu);
 
+static void tiny_rcu_leak_callback(struct rcu_head *rhp)
+{
+}
+
 /*
  * Post an RCU callback to be invoked after the end of an RCU grace
  * period.  But since we have but one CPU, that would be after any
@@ -165,9 +169,20 @@ EXPORT_SYMBOL_GPL(synchronize_rcu);
  */
 void call_rcu(struct rcu_head *head, rcu_callback_t func)
 {
+	static atomic_t doublefrees;
 	unsigned long flags;
 
-	debug_rcu_head_queue(head);
+	if (debug_rcu_head_queue(head)) {
+		if (atomic_inc_return(&doublefrees) < 4) {
+			pr_err("%s(): Double-freed CB %p->%pS()!!!  ", __func__, head, head->func);
+			mem_dump_obj(head);
+		}
+
+		if (!__is_kvfree_rcu_offset((unsigned long)head->func))
+			WRITE_ONCE(head->func, tiny_rcu_leak_callback);
+		return;
+	}
+
 	head->func = func;
 	head->next = NULL;
 
@@ -183,6 +198,16 @@ void call_rcu(struct rcu_head *head, rcu_callback_t func)
 }
 EXPORT_SYMBOL_GPL(call_rcu);
 
+/*
+ * Store a grace-period-counter "cookie".  For more information,
+ * see the Tree RCU header comment.
+ */
+void get_completed_synchronize_rcu_full(struct rcu_gp_oldstate *rgosp)
+{
+	rgosp->rgos_norm = RCU_GET_STATE_COMPLETED;
+}
+EXPORT_SYMBOL_GPL(get_completed_synchronize_rcu_full);
+
 /*
  * Return a grace-period-counter "cookie".  For more information,
  * see the Tree RCU header comment.

kernel/rcu/tree.c

@@ -76,6 +76,7 @@
 /* Data structures. */
 
 static DEFINE_PER_CPU_SHARED_ALIGNED(struct rcu_data, rcu_data) = {
+	.gpwrap = true,
 #ifdef CONFIG_RCU_NOCB_CPU
 	.cblist.flags = SEGCBLIST_RCU_CORE,
 #endif
@@ -1755,6 +1756,8 @@ static noinline void rcu_gp_cleanup(void)
 			dump_blkd_tasks(rnp, 10);
 		WARN_ON_ONCE(rnp->qsmask);
 		WRITE_ONCE(rnp->gp_seq, new_gp_seq);
+		if (!rnp->parent)
+			smp_mb(); // Order against failing poll_state_synchronize_rcu_full().
 		rdp = this_cpu_ptr(&rcu_data);
 		if (rnp == rdp->mynode)
 			needgp = __note_gp_changes(rnp, rdp) || needgp;
@@ -2341,8 +2344,8 @@ void rcu_sched_clock_irq(int user)
 	rcu_flavor_sched_clock_irq(user);
 	if (rcu_pending(user))
 		invoke_rcu_core();
-	if (user)
-		rcu_tasks_classic_qs(current, false);
+	if (user || rcu_is_cpu_rrupt_from_idle())
+		rcu_note_voluntary_context_switch(current);
 	lockdep_assert_irqs_disabled();
 
 	trace_rcu_utilization(TPS("End scheduler-tick"));
@@ -2832,7 +2835,7 @@ EXPORT_SYMBOL_GPL(call_rcu);
 
 
 /* Maximum number of jiffies to wait before draining a batch. */
-#define KFREE_DRAIN_JIFFIES (HZ / 50)
+#define KFREE_DRAIN_JIFFIES (5 * HZ)
 #define KFREE_N_BATCHES 2
 #define FREE_N_CHANNELS 2
 
@@ -3093,6 +3096,21 @@ need_offload_krc(struct kfree_rcu_cpu *krcp)
 	return !!krcp->head;
 }
 
+static void
+schedule_delayed_monitor_work(struct kfree_rcu_cpu *krcp)
+{
+	long delay, delay_left;
+
+	delay = READ_ONCE(krcp->count) >= KVFREE_BULK_MAX_ENTR ? 1:KFREE_DRAIN_JIFFIES;
+	if (delayed_work_pending(&krcp->monitor_work)) {
+		delay_left = krcp->monitor_work.timer.expires - jiffies;
+		if (delay < delay_left)
+			mod_delayed_work(system_wq, &krcp->monitor_work, delay);
+		return;
+	}
+	queue_delayed_work(system_wq, &krcp->monitor_work, delay);
+}
+
 /*
  * This function is invoked after the KFREE_DRAIN_JIFFIES timeout.
  */
@@ -3150,7 +3168,7 @@ static void kfree_rcu_monitor(struct work_struct *work)
 	// work to repeat an attempt. Because previous batches are
 	// still in progress.
 	if (need_offload_krc(krcp))
-		schedule_delayed_work(&krcp->monitor_work, KFREE_DRAIN_JIFFIES);
+		schedule_delayed_monitor_work(krcp);
 
 	raw_spin_unlock_irqrestore(&krcp->lock, flags);
 }
@@ -3183,15 +3201,16 @@ static void fill_page_cache_func(struct work_struct *work)
 		bnode = (struct kvfree_rcu_bulk_data *)
 			__get_free_page(GFP_KERNEL | __GFP_NORETRY | __GFP_NOMEMALLOC | __GFP_NOWARN);
 
-		if (bnode) {
-			raw_spin_lock_irqsave(&krcp->lock, flags);
-			pushed = put_cached_bnode(krcp, bnode);
-			raw_spin_unlock_irqrestore(&krcp->lock, flags);
+		if (!bnode)
+			break;
 
-			if (!pushed) {
-				free_page((unsigned long) bnode);
-				break;
-			}
+		raw_spin_lock_irqsave(&krcp->lock, flags);
+		pushed = put_cached_bnode(krcp, bnode);
+		raw_spin_unlock_irqrestore(&krcp->lock, flags);
+
+		if (!pushed) {
+			free_page((unsigned long) bnode);
+			break;
 		}
 	}
 
@@ -3338,7 +3357,7 @@ void kvfree_call_rcu(struct rcu_head *head, rcu_callback_t func)
 
 	// Set timer to drain after KFREE_DRAIN_JIFFIES.
 	if (rcu_scheduler_active == RCU_SCHEDULER_RUNNING)
-		schedule_delayed_work(&krcp->monitor_work, KFREE_DRAIN_JIFFIES);
+		schedule_delayed_monitor_work(krcp);
 
 unlock_return:
 	krc_this_cpu_unlock(krcp, flags);
@@ -3371,7 +3390,7 @@ kfree_rcu_shrink_count(struct shrinker *shrink, struct shrink_control *sc)
 		atomic_set(&krcp->backoff_page_cache_fill, 1);
 	}
 
-	return count;
+	return count == 0 ? SHRINK_EMPTY : count;
 }
 
 static unsigned long
@@ -3414,49 +3433,27 @@ void __init kfree_rcu_scheduler_running(void)
 
 		raw_spin_lock_irqsave(&krcp->lock, flags);
 		if (need_offload_krc(krcp))
-			schedule_delayed_work_on(cpu, &krcp->monitor_work, KFREE_DRAIN_JIFFIES);
+			schedule_delayed_monitor_work(krcp);
 		raw_spin_unlock_irqrestore(&krcp->lock, flags);
 	}
 }
 
 /*
  * During early boot, any blocking grace-period wait automatically
- * implies a grace period.  Later on, this is never the case for PREEMPTION.
+ * implies a grace period.
  *
- * However, because a context switch is a grace period for !PREEMPTION, any
- * blocking grace-period wait automatically implies a grace period if
- * there is only one CPU online at any point time during execution of
- * either synchronize_rcu() or synchronize_rcu_expedited().  It is OK to
- * occasionally incorrectly indicate that there are multiple CPUs online
- * when there was in fact only one the whole time, as this just adds some
- * overhead: RCU still operates correctly.
+ * Later on, this could in theory be the case for kernels built with
+ * CONFIG_SMP=y && CONFIG_PREEMPTION=y running on a single CPU, but this
+ * is not a common case.  Furthermore, this optimization would cause
+ * the rcu_gp_oldstate structure to expand by 50%, so this potential
+ * grace-period optimization is ignored once the scheduler is running.
  */
 static int rcu_blocking_is_gp(void)
 {
-	int ret;
-
-	// Invoking preempt_model_*() too early gets a splat.
-	if (rcu_scheduler_active == RCU_SCHEDULER_INACTIVE ||
-	    preempt_model_full() || preempt_model_rt())
-		return rcu_scheduler_active == RCU_SCHEDULER_INACTIVE;
+	if (rcu_scheduler_active != RCU_SCHEDULER_INACTIVE)
+		return false;
 	might_sleep();  /* Check for RCU read-side critical section. */
-	preempt_disable();
-	/*
-	 * If the rcu_state.n_online_cpus counter is equal to one,
-	 * there is only one CPU, and that CPU sees all prior accesses
-	 * made by any CPU that was online at the time of its access.
-	 * Furthermore, if this counter is equal to one, its value cannot
-	 * change until after the preempt_enable() below.
-	 *
-	 * Furthermore, if rcu_state.n_online_cpus is equal to one here,
-	 * all later CPUs (both this one and any that come online later
-	 * on) are guaranteed to see all accesses prior to this point
-	 * in the code, without the need for additional memory barriers.
-	 * Those memory barriers are provided by CPU-hotplug code.
-	 */
-	ret = READ_ONCE(rcu_state.n_online_cpus) <= 1;
-	preempt_enable();
-	return ret;
+	return true;
 }
 
 /**
@@ -3499,29 +3496,58 @@ static int rcu_blocking_is_gp(void)
  */
 void synchronize_rcu(void)
 {
+	unsigned long flags;
+	struct rcu_node *rnp;
+
 	RCU_LOCKDEP_WARN(lock_is_held(&rcu_bh_lock_map) ||
 			 lock_is_held(&rcu_lock_map) ||
 			 lock_is_held(&rcu_sched_lock_map),
 			 "Illegal synchronize_rcu() in RCU read-side critical section");
-	if (rcu_blocking_is_gp()) {
-		// Note well that this code runs with !PREEMPT && !SMP.
-		// In addition, all code that advances grace periods runs at
-		// process level.  Therefore, this normal GP overlaps with
-		// other normal GPs only by being fully nested within them,
-		// which allows reuse of ->gp_seq_polled_snap.
-		rcu_poll_gp_seq_start_unlocked(&rcu_state.gp_seq_polled_snap);
-		rcu_poll_gp_seq_end_unlocked(&rcu_state.gp_seq_polled_snap);
-		if (rcu_init_invoked())
-			cond_resched_tasks_rcu_qs();
-		return;  // Context allows vacuous grace periods.
+	if (!rcu_blocking_is_gp()) {
+		if (rcu_gp_is_expedited())
+			synchronize_rcu_expedited();
+		else
+			wait_rcu_gp(call_rcu);
+		return;
 	}
-	if (rcu_gp_is_expedited())
-		synchronize_rcu_expedited();
-	else
-		wait_rcu_gp(call_rcu);
+
+	// Context allows vacuous grace periods.
+	// Note well that this code runs with !PREEMPT && !SMP.
+	// In addition, all code that advances grace periods runs at
+	// process level.  Therefore, this normal GP overlaps with other
+	// normal GPs only by being fully nested within them, which allows
+	// reuse of ->gp_seq_polled_snap.
+	rcu_poll_gp_seq_start_unlocked(&rcu_state.gp_seq_polled_snap);
+	rcu_poll_gp_seq_end_unlocked(&rcu_state.gp_seq_polled_snap);
+
+	// Update the normal grace-period counters to record
+	// this grace period, but only those used by the boot CPU.
+	// The rcu_scheduler_starting() will take care of the rest of
+	// these counters.
+	local_irq_save(flags);
+	WARN_ON_ONCE(num_online_cpus() > 1);
+	rcu_state.gp_seq += (1 << RCU_SEQ_CTR_SHIFT);
+	for (rnp = this_cpu_ptr(&rcu_data)->mynode; rnp; rnp = rnp->parent)
+		rnp->gp_seq_needed = rnp->gp_seq = rcu_state.gp_seq;
+	local_irq_restore(flags);
 }
 EXPORT_SYMBOL_GPL(synchronize_rcu);
 
+/**
+ * get_completed_synchronize_rcu_full - Return a full pre-completed polled state cookie
+ * @rgosp: Place to put state cookie
+ *
+ * Stores into @rgosp a value that will always be treated by functions
+ * like poll_state_synchronize_rcu_full() as a cookie whose grace period
+ * has already completed.
+ */
+void get_completed_synchronize_rcu_full(struct rcu_gp_oldstate *rgosp)
+{
+	rgosp->rgos_norm = RCU_GET_STATE_COMPLETED;
+	rgosp->rgos_exp = RCU_GET_STATE_COMPLETED;
+}
+EXPORT_SYMBOL_GPL(get_completed_synchronize_rcu_full);
+
 /**
  * get_state_synchronize_rcu - Snapshot current RCU state
  *
@@ -3541,21 +3567,42 @@ unsigned long get_state_synchronize_rcu(void)
 EXPORT_SYMBOL_GPL(get_state_synchronize_rcu);
 
 /**
- * start_poll_synchronize_rcu - Snapshot and start RCU grace period
+ * get_state_synchronize_rcu_full - Snapshot RCU state, both normal and expedited
+ * @rgosp: location to place combined normal/expedited grace-period state
  *
- * Returns a cookie that is used by a later call to cond_synchronize_rcu()
- * or poll_state_synchronize_rcu() to determine whether or not a full
- * grace period has elapsed in the meantime.  If the needed grace period
- * is not already slated to start, notifies RCU core of the need for that
- * grace period.
+ * Places the normal and expedited grace-period states in @rgosp.  This
+ * state value can be passed to a later call to cond_synchronize_rcu_full()
+ * or poll_state_synchronize_rcu_full() to determine whether or not a
+ * grace period (whether normal or expedited) has elapsed in the meantime.
+ * The rcu_gp_oldstate structure takes up twice the memory of an unsigned
+ * long, but is guaranteed to see all grace periods.  In contrast, the
+ * combined state occupies less memory, but can sometimes fail to take
+ * grace periods into account.
  *
- * Interrupts must be enabled for the case where it is necessary to awaken
- * the grace-period kthread.
+ * This does not guarantee that the needed grace period will actually
+ * start.
  */
-unsigned long start_poll_synchronize_rcu(void)
+void get_state_synchronize_rcu_full(struct rcu_gp_oldstate *rgosp)
+{
+	struct rcu_node *rnp = rcu_get_root();
+
+	/*
+	 * Any prior manipulation of RCU-protected data must happen
+	 * before the loads from ->gp_seq and ->expedited_sequence.
+	 */
+	smp_mb();  /* ^^^ */
+	rgosp->rgos_norm = rcu_seq_snap(&rnp->gp_seq);
+	rgosp->rgos_exp = rcu_seq_snap(&rcu_state.expedited_sequence);
+}
+EXPORT_SYMBOL_GPL(get_state_synchronize_rcu_full);
+
+/*
+ * Helper function for start_poll_synchronize_rcu() and
+ * start_poll_synchronize_rcu_full().
+ */
+static void start_poll_synchronize_rcu_common(void)
 {
 	unsigned long flags;
-	unsigned long gp_seq = get_state_synchronize_rcu();
 	bool needwake;
 	struct rcu_data *rdp;
 	struct rcu_node *rnp;
@@ -3575,17 +3622,57 @@ unsigned long start_poll_synchronize_rcu(void)
 	raw_spin_unlock_irqrestore_rcu_node(rnp, flags);
 	if (needwake)
 		rcu_gp_kthread_wake();
+}
+
+/**
+ * start_poll_synchronize_rcu - Snapshot and start RCU grace period
+ *
+ * Returns a cookie that is used by a later call to cond_synchronize_rcu()
+ * or poll_state_synchronize_rcu() to determine whether or not a full
+ * grace period has elapsed in the meantime.  If the needed grace period
+ * is not already slated to start, notifies RCU core of the need for that
+ * grace period.
+ *
+ * Interrupts must be enabled for the case where it is necessary to awaken
+ * the grace-period kthread.
+ */
+unsigned long start_poll_synchronize_rcu(void)
+{
+	unsigned long gp_seq = get_state_synchronize_rcu();
+
+	start_poll_synchronize_rcu_common();
 	return gp_seq;
 }
 EXPORT_SYMBOL_GPL(start_poll_synchronize_rcu);
 
 /**
- * poll_state_synchronize_rcu - Conditionally wait for an RCU grace period
+ * start_poll_synchronize_rcu_full - Take a full snapshot and start RCU grace period
+ * @rgosp: value from get_state_synchronize_rcu_full() or start_poll_synchronize_rcu_full()
  *
+ * Places the normal and expedited grace-period states in *@rgos.  This
+ * state value can be passed to a later call to cond_synchronize_rcu_full()
+ * or poll_state_synchronize_rcu_full() to determine whether or not a
+ * grace period (whether normal or expedited) has elapsed in the meantime.
+ * If the needed grace period is not already slated to start, notifies
+ * RCU core of the need for that grace period.
+ *
+ * Interrupts must be enabled for the case where it is necessary to awaken
+ * the grace-period kthread.
+ */
+void start_poll_synchronize_rcu_full(struct rcu_gp_oldstate *rgosp)
+{
+	get_state_synchronize_rcu_full(rgosp);
+
+	start_poll_synchronize_rcu_common();
+}
+EXPORT_SYMBOL_GPL(start_poll_synchronize_rcu_full);
+
+/**
+ * poll_state_synchronize_rcu - Has the specified RCU grace period completed?
  * @oldstate: value from get_state_synchronize_rcu() or start_poll_synchronize_rcu()
  *
  * If a full RCU grace period has elapsed since the earlier call from
- * which oldstate was obtained, return @true, otherwise return @false.
+ * which @oldstate was obtained, return @true, otherwise return @false.
  * If @false is returned, it is the caller's responsibility to invoke this
  * function later on until it does return @true.  Alternatively, the caller
  * can explicitly wait for a grace period, for example, by passing @oldstate
@@ -3594,10 +3681,11 @@ EXPORT_SYMBOL_GPL(start_poll_synchronize_rcu);
  * Yes, this function does not take counter wrap into account.
  * But counter wrap is harmless.  If the counter wraps, we have waited for
  * more than a billion grace periods (and way more on a 64-bit system!).
- * Those needing to keep oldstate values for very long time periods
- * (many hours even on 32-bit systems) should check them occasionally
- * and either refresh them or set a flag indicating that the grace period
- * has completed.
+ * Those needing to keep old state values for very long time periods
+ * (many hours even on 32-bit systems) should check them occasionally and
+ * either refresh them or set a flag indicating that the grace period has
+ * completed.  Alternatively, they can use get_completed_synchronize_rcu()
+ * to get a guaranteed-completed grace-period state.
  *
  * This function provides the same memory-ordering guarantees that
  * would be provided by a synchronize_rcu() that was invoked at the call
@@ -3616,8 +3704,56 @@ bool poll_state_synchronize_rcu(unsigned long oldstate)
 EXPORT_SYMBOL_GPL(poll_state_synchronize_rcu);
 
 /**
- * cond_synchronize_rcu - Conditionally wait for an RCU grace period
+ * poll_state_synchronize_rcu_full - Has the specified RCU grace period completed?
+ * @rgosp: value from get_state_synchronize_rcu_full() or start_poll_synchronize_rcu_full()
  *
+ * If a full RCU grace period has elapsed since the earlier call from
+ * which *rgosp was obtained, return @true, otherwise return @false.
+ * If @false is returned, it is the caller's responsibility to invoke this
+ * function later on until it does return @true.  Alternatively, the caller
+ * can explicitly wait for a grace period, for example, by passing @rgosp
+ * to cond_synchronize_rcu() or by directly invoking synchronize_rcu().
+ *
+ * Yes, this function does not take counter wrap into account.
+ * But counter wrap is harmless.  If the counter wraps, we have waited
+ * for more than a billion grace periods (and way more on a 64-bit
+ * system!).  Those needing to keep rcu_gp_oldstate values for very
+ * long time periods (many hours even on 32-bit systems) should check
+ * them occasionally and either refresh them or set a flag indicating
+ * that the grace period has completed.  Alternatively, they can use
+ * get_completed_synchronize_rcu_full() to get a guaranteed-completed
+ * grace-period state.
+ *
+ * This function provides the same memory-ordering guarantees that would
+ * be provided by a synchronize_rcu() that was invoked at the call to
+ * the function that provided @rgosp, and that returned at the end of this
+ * function.  And this guarantee requires that the root rcu_node structure's
+ * ->gp_seq field be checked instead of that of the rcu_state structure.
+ * The problem is that the just-ending grace-period's callbacks can be
+ * invoked between the time that the root rcu_node structure's ->gp_seq
+ * field is updated and the time that the rcu_state structure's ->gp_seq
+ * field is updated.  Therefore, if a single synchronize_rcu() is to
+ * cause a subsequent poll_state_synchronize_rcu_full() to return @true,
+ * then the root rcu_node structure is the one that needs to be polled.
+ */
+bool poll_state_synchronize_rcu_full(struct rcu_gp_oldstate *rgosp)
+{
+	struct rcu_node *rnp = rcu_get_root();
+
+	smp_mb(); // Order against root rcu_node structure grace-period cleanup.
+	if (rgosp->rgos_norm == RCU_GET_STATE_COMPLETED ||
+	    rcu_seq_done_exact(&rnp->gp_seq, rgosp->rgos_norm) ||
+	    rgosp->rgos_exp == RCU_GET_STATE_COMPLETED ||
+	    rcu_seq_done_exact(&rcu_state.expedited_sequence, rgosp->rgos_exp)) {
+		smp_mb(); /* Ensure GP ends before subsequent accesses. */
+		return true;
+	}
+	return false;
+}
+EXPORT_SYMBOL_GPL(poll_state_synchronize_rcu_full);
+
+/**
+ * cond_synchronize_rcu - Conditionally wait for an RCU grace period
  * @oldstate: value from get_state_synchronize_rcu(), start_poll_synchronize_rcu(), or start_poll_synchronize_rcu_expedited()
  *
  * If a full RCU grace period has elapsed since the earlier call to
@@ -3641,6 +3777,33 @@ void cond_synchronize_rcu(unsigned long oldstate)
 }
 EXPORT_SYMBOL_GPL(cond_synchronize_rcu);
 
+/**
+ * cond_synchronize_rcu_full - Conditionally wait for an RCU grace period
+ * @rgosp: value from get_state_synchronize_rcu_full(), start_poll_synchronize_rcu_full(), or start_poll_synchronize_rcu_expedited_full()
+ *
+ * If a full RCU grace period has elapsed since the call to
+ * get_state_synchronize_rcu_full(), start_poll_synchronize_rcu_full(),
+ * or start_poll_synchronize_rcu_expedited_full() from which @rgosp was
+ * obtained, just return.  Otherwise, invoke synchronize_rcu() to wait
+ * for a full grace period.
+ *
+ * Yes, this function does not take counter wrap into account.
+ * But counter wrap is harmless.  If the counter wraps, we have waited for
+ * more than 2 billion grace periods (and way more on a 64-bit system!),
+ * so waiting for a couple of additional grace periods should be just fine.
+ *
+ * This function provides the same memory-ordering guarantees that
+ * would be provided by a synchronize_rcu() that was invoked at the call
+ * to the function that provided @rgosp and that returned at the end of
+ * this function.
+ */
+void cond_synchronize_rcu_full(struct rcu_gp_oldstate *rgosp)
+{
+	if (!poll_state_synchronize_rcu_full(rgosp))
+		synchronize_rcu();
+}
+EXPORT_SYMBOL_GPL(cond_synchronize_rcu_full);
+
 /*
  * Check to see if there is any immediate RCU-related work to be done by
  * the current CPU, returning 1 if so and zero otherwise.  The checks are
@@ -4312,9 +4475,20 @@ early_initcall(rcu_spawn_gp_kthread);
  */
 void rcu_scheduler_starting(void)
 {
+	unsigned long flags;
+	struct rcu_node *rnp;
+
 	WARN_ON(num_online_cpus() != 1);
 	WARN_ON(nr_context_switches() > 0);
 	rcu_test_sync_prims();
+
+	// Fix up the ->gp_seq counters.
+	local_irq_save(flags);
+	rcu_for_each_node_breadth_first(rnp)
+		rnp->gp_seq_needed = rnp->gp_seq = rcu_state.gp_seq;
+	local_irq_restore(flags);
+
+	// Switch out of early boot mode.
 	rcu_scheduler_active = RCU_SCHEDULER_INIT;
 	rcu_test_sync_prims();
 }

kernel/rcu/tree_exp.h

@@ -828,11 +828,13 @@ static void rcu_exp_handler(void *unused)
 {
 	struct rcu_data *rdp = this_cpu_ptr(&rcu_data);
 	struct rcu_node *rnp = rdp->mynode;
+	bool preempt_bh_enabled = !(preempt_count() & (PREEMPT_MASK | SOFTIRQ_MASK));
 
 	if (!(READ_ONCE(rnp->expmask) & rdp->grpmask) ||
 	    __this_cpu_read(rcu_data.cpu_no_qs.b.exp))
 		return;
-	if (rcu_is_cpu_rrupt_from_idle()) {
+	if (rcu_is_cpu_rrupt_from_idle() ||
+	    (IS_ENABLED(CONFIG_PREEMPT_COUNT) && preempt_bh_enabled)) {
 		rcu_report_exp_rdp(this_cpu_ptr(&rcu_data));
 		return;
 	}
@@ -906,6 +908,7 @@ static int rcu_print_task_exp_stall(struct rcu_node *rnp)
 void synchronize_rcu_expedited(void)
 {
 	bool boottime = (rcu_scheduler_active == RCU_SCHEDULER_INIT);
+	unsigned long flags;
 	struct rcu_exp_work rew;
 	struct rcu_node *rnp;
 	unsigned long s;
@@ -924,8 +927,11 @@ void synchronize_rcu_expedited(void)
 		// them, which allows reuse of ->gp_seq_polled_exp_snap.
 		rcu_poll_gp_seq_start_unlocked(&rcu_state.gp_seq_polled_exp_snap);
 		rcu_poll_gp_seq_end_unlocked(&rcu_state.gp_seq_polled_exp_snap);
-		if (rcu_init_invoked())
-			cond_resched();
+
+		local_irq_save(flags);
+		WARN_ON_ONCE(num_online_cpus() > 1);
+		rcu_state.expedited_sequence += (1 << RCU_SEQ_CTR_SHIFT);
+		local_irq_restore(flags);
 		return;  // Context allows vacuous grace periods.
 	}
 
@@ -1027,6 +1033,24 @@ unsigned long start_poll_synchronize_rcu_expedited(void)
 }
 EXPORT_SYMBOL_GPL(start_poll_synchronize_rcu_expedited);
 
+/**
+ * start_poll_synchronize_rcu_expedited_full - Take a full snapshot and start expedited grace period
+ * @rgosp: Place to put snapshot of grace-period state
+ *
+ * Places the normal and expedited grace-period states in rgosp.  This
+ * state value can be passed to a later call to cond_synchronize_rcu_full()
+ * or poll_state_synchronize_rcu_full() to determine whether or not a
+ * grace period (whether normal or expedited) has elapsed in the meantime.
+ * If the needed expedited grace period is not already slated to start,
+ * initiates that grace period.
+ */
+void start_poll_synchronize_rcu_expedited_full(struct rcu_gp_oldstate *rgosp)
+{
+	get_state_synchronize_rcu_full(rgosp);
+	(void)start_poll_synchronize_rcu_expedited();
+}
+EXPORT_SYMBOL_GPL(start_poll_synchronize_rcu_expedited_full);
+
 /**
  * cond_synchronize_rcu_expedited - Conditionally wait for an expedited RCU grace period
  *
@@ -1053,3 +1077,30 @@ void cond_synchronize_rcu_expedited(unsigned long oldstate)
 		synchronize_rcu_expedited();
 }
 EXPORT_SYMBOL_GPL(cond_synchronize_rcu_expedited);
+
+/**
+ * cond_synchronize_rcu_expedited_full - Conditionally wait for an expedited RCU grace period
+ * @rgosp: value from get_state_synchronize_rcu_full(), start_poll_synchronize_rcu_full(), or start_poll_synchronize_rcu_expedited_full()
+ *
+ * If a full RCU grace period has elapsed since the call to
+ * get_state_synchronize_rcu_full(), start_poll_synchronize_rcu_full(),
+ * or start_poll_synchronize_rcu_expedited_full() from which @rgosp was
+ * obtained, just return.  Otherwise, invoke synchronize_rcu_expedited()
+ * to wait for a full grace period.
+ *
+ * Yes, this function does not take counter wrap into account.
+ * But counter wrap is harmless.  If the counter wraps, we have waited for
+ * more than 2 billion grace periods (and way more on a 64-bit system!),
+ * so waiting for a couple of additional grace periods should be just fine.
+ *
+ * This function provides the same memory-ordering guarantees that
+ * would be provided by a synchronize_rcu() that was invoked at the call
+ * to the function that provided @rgosp and that returned at the end of
+ * this function.
+ */
+void cond_synchronize_rcu_expedited_full(struct rcu_gp_oldstate *rgosp)
+{
+	if (!poll_state_synchronize_rcu_full(rgosp))
+		synchronize_rcu_expedited();
+}
+EXPORT_SYMBOL_GPL(cond_synchronize_rcu_expedited_full);

kernel/rcu/tree_nocb.h

@@ -1111,7 +1111,7 @@ int rcu_nocb_cpu_deoffload(int cpu)
 			if (!ret)
 				cpumask_clear_cpu(cpu, rcu_nocb_mask);
 		} else {
-			pr_info("NOCB: Can't CB-deoffload an offline CPU\n");
+			pr_info("NOCB: Cannot CB-deoffload offline CPU %d\n", rdp->cpu);
 			ret = -EINVAL;
 		}
 	}
@@ -1196,7 +1196,7 @@ int rcu_nocb_cpu_offload(int cpu)
 			if (!ret)
 				cpumask_set_cpu(cpu, rcu_nocb_mask);
 		} else {
-			pr_info("NOCB: Can't CB-offload an offline CPU\n");
+			pr_info("NOCB: Cannot CB-offload offline CPU %d\n", rdp->cpu);
 			ret = -EINVAL;
 		}
 	}
@@ -1452,8 +1452,8 @@ static void show_rcu_nocb_gp_state(struct rcu_data *rdp)
 		(long)rdp->nocb_gp_seq,
 		rnp->grplo, rnp->grphi, READ_ONCE(rdp->nocb_gp_loops),
 		rdp->nocb_gp_kthread ? task_state_to_char(rdp->nocb_gp_kthread) : '.',
-		rdp->nocb_cb_kthread ? (int)task_cpu(rdp->nocb_gp_kthread) : -1,
-		show_rcu_should_be_on_cpu(rdp->nocb_cb_kthread));
+		rdp->nocb_gp_kthread ? (int)task_cpu(rdp->nocb_gp_kthread) : -1,
+		show_rcu_should_be_on_cpu(rdp->nocb_gp_kthread));
 }
 
 /* Dump out nocb kthread state for the specified rcu_data structure. */
@@ -1497,7 +1497,7 @@ static void show_rcu_nocb_state(struct rcu_data *rdp)
 		".B"[!!rcu_cblist_n_cbs(&rdp->nocb_bypass)],
 		rcu_segcblist_n_cbs(&rdp->cblist),
 		rdp->nocb_cb_kthread ? task_state_to_char(rdp->nocb_cb_kthread) : '.',
-		rdp->nocb_cb_kthread ? (int)task_cpu(rdp->nocb_gp_kthread) : -1,
+		rdp->nocb_cb_kthread ? (int)task_cpu(rdp->nocb_cb_kthread) : -1,
 		show_rcu_should_be_on_cpu(rdp->nocb_cb_kthread));
 
 	/* It is OK for GP kthreads to have GP state. */

kernel/rcu/tree_plugin.h

@@ -641,7 +641,8 @@ static void rcu_read_unlock_special(struct task_struct *t)
 
 		expboost = (t->rcu_blocked_node && READ_ONCE(t->rcu_blocked_node->exp_tasks)) ||
 			   (rdp->grpmask & READ_ONCE(rnp->expmask)) ||
-			   IS_ENABLED(CONFIG_RCU_STRICT_GRACE_PERIOD) ||
+			   (IS_ENABLED(CONFIG_RCU_STRICT_GRACE_PERIOD) &&
+			   ((rdp->grpmask & READ_ONCE(rnp->qsmask)) || t->rcu_blocked_node)) ||
 			   (IS_ENABLED(CONFIG_RCU_BOOST) && irqs_were_disabled &&
 			    t->rcu_blocked_node);
 		// Need to defer quiescent state until everything is enabled.
@@ -718,9 +719,6 @@ static void rcu_flavor_sched_clock_irq(int user)
 	struct task_struct *t = current;
 
 	lockdep_assert_irqs_disabled();
-	if (user || rcu_is_cpu_rrupt_from_idle()) {
-		rcu_note_voluntary_context_switch(current);
-	}
 	if (rcu_preempt_depth() > 0 ||
 	    (preempt_count() & (PREEMPT_MASK | SOFTIRQ_MASK))) {
 		/* No QS, force context switch if deferred. */
@@ -824,6 +822,7 @@ void rcu_read_unlock_strict(void)
 	if (irqs_disabled() || preempt_count() || !rcu_state.gp_kthread)
 		return;
 	rdp = this_cpu_ptr(&rcu_data);
+	rdp->cpu_no_qs.b.norm = false;
 	rcu_report_qs_rdp(rdp);
 	udelay(rcu_unlock_delay);
 }
@@ -869,7 +868,7 @@ void rcu_all_qs(void)
 
 	if (!raw_cpu_read(rcu_data.rcu_urgent_qs))
 		return;
-	preempt_disable();
+	preempt_disable();  // For CONFIG_PREEMPT_COUNT=y kernels
 	/* Load rcu_urgent_qs before other flags. */
 	if (!smp_load_acquire(this_cpu_ptr(&rcu_data.rcu_urgent_qs))) {
 		preempt_enable();
@@ -931,10 +930,13 @@ static notrace bool rcu_preempt_need_deferred_qs(struct task_struct *t)
 	return false;
 }
 
-// Except that we do need to respond to a request by an expedited grace
-// period for a quiescent state from this CPU.  Note that requests from
-// tasks are handled when removing the task from the blocked-tasks list
-// below.
+// Except that we do need to respond to a request by an expedited
+// grace period for a quiescent state from this CPU.  Note that in
+// non-preemptible kernels, there can be no context switches within RCU
+// read-side critical sections, which in turn means that the leaf rcu_node
+// structure's blocked-tasks list is always empty.  is therefore no need to
+// actually check it.  Instead, a quiescent state from this CPU suffices,
+// and this function is only called from such a quiescent state.
 notrace void rcu_preempt_deferred_qs(struct task_struct *t)
 {
 	struct rcu_data *rdp = this_cpu_ptr(&rcu_data);
@@ -972,7 +974,6 @@ static void rcu_flavor_sched_clock_irq(int user)
 		 * neither access nor modify, at least not while the
 		 * corresponding CPU is online.
 		 */
-
 		rcu_qs();
 	}
 }
@@ -1238,8 +1239,11 @@ static void rcu_boost_kthread_setaffinity(struct rcu_node *rnp, int outgoingcpu)
 		    cpu != outgoingcpu)
 			cpumask_set_cpu(cpu, cm);
 	cpumask_and(cm, cm, housekeeping_cpumask(HK_TYPE_RCU));
-	if (cpumask_empty(cm))
+	if (cpumask_empty(cm)) {
 		cpumask_copy(cm, housekeeping_cpumask(HK_TYPE_RCU));
+		if (outgoingcpu >= 0)
+			cpumask_clear_cpu(outgoingcpu, cm);
+	}
 	set_cpus_allowed_ptr(t, cm);
 	mutex_unlock(&rnp->boost_kthread_mutex);
 	free_cpumask_var(cm);

kernel/rcu/tree_stall.h

@@ -368,7 +368,7 @@ static void rcu_dump_cpu_stacks(void)
 			if (rnp->qsmask & leaf_node_cpu_bit(rnp, cpu)) {
 				if (cpu_is_offline(cpu))
 					pr_err("Offline CPU %d blocking current GP.\n", cpu);
-				else if (!trigger_single_cpu_backtrace(cpu))
+				else
 					dump_cpu_task(cpu);
 			}
 		raw_spin_unlock_irqrestore_rcu_node(rnp, flags);
@@ -511,8 +511,7 @@ static void rcu_check_gp_kthread_starvation(void)
 					pr_err("RCU GP kthread last ran on offline CPU %d.\n", cpu);
 				} else  {
 					pr_err("Stack dump where RCU GP kthread last ran:\n");
-					if (!trigger_single_cpu_backtrace(cpu))
-						dump_cpu_task(cpu);
+					dump_cpu_task(cpu);
 				}
 			}
 			wake_up_process(gpk);

kernel/sched/core.c

@@ -73,6 +73,7 @@
 
 #include <uapi/linux/sched/types.h>
 
+#include <asm/irq_regs.h>
 #include <asm/switch_to.h>
 #include <asm/tlb.h>
 
@@ -11183,6 +11184,19 @@ struct cgroup_subsys cpu_cgrp_subsys = {
 
 void dump_cpu_task(int cpu)
 {
+	if (cpu == smp_processor_id() && in_hardirq()) {
+		struct pt_regs *regs;
+
+		regs = get_irq_regs();
+		if (regs) {
+			show_regs(regs);
+			return;
+		}
+	}
+
+	if (trigger_single_cpu_backtrace(cpu))
+		return;
+
 	pr_info("Task dump for CPU %d:\n", cpu);
 	sched_show_task(cpu_curr(cpu));
 }

kernel/smp.c

@@ -370,8 +370,7 @@ static bool csd_lock_wait_toolong(struct __call_single_data *csd, u64 ts0, u64 *
 	if (cpu >= 0) {
 		if (static_branch_unlikely(&csdlock_debug_extended))
 			csd_lock_print_extended(csd, cpu);
-		if (!trigger_single_cpu_backtrace(cpu))
-			dump_cpu_task(cpu);
+		dump_cpu_task(cpu);
 		if (!cpu_cur_csd) {
 			pr_alert("csd: Re-sending CSD lock (#%d) IPI from CPU#%02d to CPU#%02d\n", *bug_id, raw_smp_processor_id(), cpu);
 			arch_send_call_function_single_ipi(cpu);