[thirdparty/linux.git] / kernel / rcu / rcu.h

/* SPDX-License-Identifier: GPL-2.0+ */
/*
 * Read-Copy Update definitions shared among RCU implementations.
 *
 * Copyright IBM Corporation, 2011
 *
 * Author: Paul E. McKenney <paulmck@linux.ibm.com>
 */

#ifndef __LINUX_RCU_H
#define __LINUX_RCU_H

#include <trace/events/rcu.h>

/* Offset to allow distinguishing irq vs. task-based idle entry/exit. */
#define DYNTICK_IRQ_NONIDLE	((LONG_MAX / 2) + 1)


/*
 * Grace-period counter management.
 */

#define RCU_SEQ_CTR_SHIFT	2
#define RCU_SEQ_STATE_MASK	((1 << RCU_SEQ_CTR_SHIFT) - 1)

/*
 * Return the counter portion of a sequence number previously returned
 * by rcu_seq_snap() or rcu_seq_current().
 */
static inline unsigned long rcu_seq_ctr(unsigned long s)
{
	return s >> RCU_SEQ_CTR_SHIFT;
}

/*
 * Return the state portion of a sequence number previously returned
 * by rcu_seq_snap() or rcu_seq_current().
 */
static inline int rcu_seq_state(unsigned long s)
{
	return s & RCU_SEQ_STATE_MASK;
}

/*
 * Set the state portion of the pointed-to sequence number.
 * The caller is responsible for preventing conflicting updates.
 */
static inline void rcu_seq_set_state(unsigned long *sp, int newstate)
{
	WARN_ON_ONCE(newstate & ~RCU_SEQ_STATE_MASK);
	WRITE_ONCE(*sp, (*sp & ~RCU_SEQ_STATE_MASK) + newstate);
}

/* Adjust sequence number for start of update-side operation. */
static inline void rcu_seq_start(unsigned long *sp)
{
	WRITE_ONCE(*sp, *sp + 1);
	smp_mb(); /* Ensure update-side operation after counter increment. */
	WARN_ON_ONCE(rcu_seq_state(*sp) != 1);
}

/* Compute the end-of-grace-period value for the specified sequence number. */
static inline unsigned long rcu_seq_endval(unsigned long *sp)
{
	return (*sp | RCU_SEQ_STATE_MASK) + 1;
}

/* Adjust sequence number for end of update-side operation. */
static inline void rcu_seq_end(unsigned long *sp)
{
	smp_mb(); /* Ensure update-side operation before counter increment. */
	WARN_ON_ONCE(!rcu_seq_state(*sp));
	WRITE_ONCE(*sp, rcu_seq_endval(sp));
}

/*
 * rcu_seq_snap - Take a snapshot of the update side's sequence number.
 *
 * This function returns the earliest value of the grace-period sequence number
 * that will indicate that a full grace period has elapsed since the current
 * time.  Once the grace-period sequence number has reached this value, it will
 * be safe to invoke all callbacks that have been registered prior to the
 * current time. This value is the current grace-period number plus two to the
 * power of the number of low-order bits reserved for state, then rounded up to
 * the next value in which the state bits are all zero.
 */
static inline unsigned long rcu_seq_snap(unsigned long *sp)
{
	unsigned long s;

	s = (READ_ONCE(*sp) + 2 * RCU_SEQ_STATE_MASK + 1) & ~RCU_SEQ_STATE_MASK;
	smp_mb(); /* Above access must not bleed into critical section. */
	return s;
}

/* Return the current value the update side's sequence number, no ordering. */
static inline unsigned long rcu_seq_current(unsigned long *sp)
{
	return READ_ONCE(*sp);
}

/*
 * Given a snapshot from rcu_seq_snap(), determine whether or not the
 * corresponding update-side operation has started.
 */
static inline bool rcu_seq_started(unsigned long *sp, unsigned long s)
{
	return ULONG_CMP_LT((s - 1) & ~RCU_SEQ_STATE_MASK, READ_ONCE(*sp));
}

/*
 * Given a snapshot from rcu_seq_snap(), determine whether or not a
 * full update-side operation has occurred.
 */
static inline bool rcu_seq_done(unsigned long *sp, unsigned long s)
{
	return ULONG_CMP_GE(READ_ONCE(*sp), s);
}

/*
 * Has a grace period completed since the time the old gp_seq was collected?
 */
static inline bool rcu_seq_completed_gp(unsigned long old, unsigned long new)
{
	return ULONG_CMP_LT(old, new & ~RCU_SEQ_STATE_MASK);
}

/*
 * Has a grace period started since the time the old gp_seq was collected?
 */
static inline bool rcu_seq_new_gp(unsigned long old, unsigned long new)
{
	return ULONG_CMP_LT((old + RCU_SEQ_STATE_MASK) & ~RCU_SEQ_STATE_MASK,
			    new);
}

/*
 * Roughly how many full grace periods have elapsed between the collection
 * of the two specified grace periods?
 */
static inline unsigned long rcu_seq_diff(unsigned long new, unsigned long old)
{
	unsigned long rnd_diff;

	if (old == new)
		return 0;
	/*
	 * Compute the number of grace periods (still shifted up), plus
	 * one if either of new and old is not an exact grace period.
	 */
	rnd_diff = (new & ~RCU_SEQ_STATE_MASK) -
		   ((old + RCU_SEQ_STATE_MASK) & ~RCU_SEQ_STATE_MASK) +
		   ((new & RCU_SEQ_STATE_MASK) || (old & RCU_SEQ_STATE_MASK));
	if (ULONG_CMP_GE(RCU_SEQ_STATE_MASK, rnd_diff))
		return 1; /* Definitely no grace period has elapsed. */
	return ((rnd_diff - RCU_SEQ_STATE_MASK - 1) >> RCU_SEQ_CTR_SHIFT) + 2;
}

/*
 * debug_rcu_head_queue()/debug_rcu_head_unqueue() are used internally
 * by call_rcu() and rcu callback execution, and are therefore not part
 * of the RCU API. These are in rcupdate.h because they are used by all
 * RCU implementations.
 */

#ifdef CONFIG_DEBUG_OBJECTS_RCU_HEAD
# define STATE_RCU_HEAD_READY	0
# define STATE_RCU_HEAD_QUEUED	1

extern struct debug_obj_descr rcuhead_debug_descr;

static inline int debug_rcu_head_queue(struct rcu_head *head)
{
	int r1;

	r1 = debug_object_activate(head, &rcuhead_debug_descr);
	debug_object_active_state(head, &rcuhead_debug_descr,
				  STATE_RCU_HEAD_READY,
				  STATE_RCU_HEAD_QUEUED);
	return r1;
}

static inline void debug_rcu_head_unqueue(struct rcu_head *head)
{
	debug_object_active_state(head, &rcuhead_debug_descr,
				  STATE_RCU_HEAD_QUEUED,
				  STATE_RCU_HEAD_READY);
	debug_object_deactivate(head, &rcuhead_debug_descr);
}
#else	/* !CONFIG_DEBUG_OBJECTS_RCU_HEAD */
static inline int debug_rcu_head_queue(struct rcu_head *head)
{
	return 0;
}

static inline void debug_rcu_head_unqueue(struct rcu_head *head)
{
}
#endif	/* #else !CONFIG_DEBUG_OBJECTS_RCU_HEAD */

extern int rcu_cpu_stall_suppress_at_boot;

static inline bool rcu_stall_is_suppressed_at_boot(void)
{
	return rcu_cpu_stall_suppress_at_boot && !rcu_inkernel_boot_has_ended();
}

#ifdef CONFIG_RCU_STALL_COMMON

extern int rcu_cpu_stall_ftrace_dump;
extern int rcu_cpu_stall_suppress;
extern int rcu_cpu_stall_timeout;
int rcu_jiffies_till_stall_check(void);

static inline bool rcu_stall_is_suppressed(void)
{
	return rcu_stall_is_suppressed_at_boot() || rcu_cpu_stall_suppress;
}

#define rcu_ftrace_dump_stall_suppress() \
do { \
	if (!rcu_cpu_stall_suppress) \
		rcu_cpu_stall_suppress = 3; \
} while (0)

#define rcu_ftrace_dump_stall_unsuppress() \
do { \
	if (rcu_cpu_stall_suppress == 3) \
		rcu_cpu_stall_suppress = 0; \
} while (0)

#else /* #endif #ifdef CONFIG_RCU_STALL_COMMON */

static inline bool rcu_stall_is_suppressed(void)
{
	return rcu_stall_is_suppressed_at_boot();
}
#define rcu_ftrace_dump_stall_suppress()
#define rcu_ftrace_dump_stall_unsuppress()
#endif /* #ifdef CONFIG_RCU_STALL_COMMON */

/*
 * Strings used in tracepoints need to be exported via the
 * tracing system such that tools like perf and trace-cmd can
 * translate the string address pointers to actual text.
 */
#define TPS(x)  tracepoint_string(x)

/*
 * Dump the ftrace buffer, but only one time per callsite per boot.
 */
#define rcu_ftrace_dump(oops_dump_mode) \
do { \
	static atomic_t ___rfd_beenhere = ATOMIC_INIT(0); \
	\
	if (!atomic_read(&___rfd_beenhere) && \
	    !atomic_xchg(&___rfd_beenhere, 1)) { \
		tracing_off(); \
		rcu_ftrace_dump_stall_suppress(); \
		ftrace_dump(oops_dump_mode); \
		rcu_ftrace_dump_stall_unsuppress(); \
	} \
} while (0)

void rcu_early_boot_tests(void);
void rcu_test_sync_prims(void);

/*
 * This function really isn't for public consumption, but RCU is special in
 * that context switches can allow the state machine to make progress.
 */
extern void resched_cpu(int cpu);

#if defined(CONFIG_SRCU) || !defined(CONFIG_TINY_RCU)

#include <linux/rcu_node_tree.h>

extern int rcu_num_lvls;
extern int num_rcu_lvl[];
extern int rcu_num_nodes;
static bool rcu_fanout_exact;
static int rcu_fanout_leaf;

/*
 * Compute the per-level fanout, either using the exact fanout specified
 * or balancing the tree, depending on the rcu_fanout_exact boot parameter.
 */
static inline void rcu_init_levelspread(int *levelspread, const int *levelcnt)
{
	int i;

	for (i = 0; i < RCU_NUM_LVLS; i++)
		levelspread[i] = INT_MIN;
	if (rcu_fanout_exact) {
		levelspread[rcu_num_lvls - 1] = rcu_fanout_leaf;
		for (i = rcu_num_lvls - 2; i >= 0; i--)
			levelspread[i] = RCU_FANOUT;
	} else {
		int ccur;
		int cprv;

		cprv = nr_cpu_ids;
		for (i = rcu_num_lvls - 1; i >= 0; i--) {
			ccur = levelcnt[i];
			levelspread[i] = (cprv + ccur - 1) / ccur;
			cprv = ccur;
		}
	}
}

/* Returns a pointer to the first leaf rcu_node structure. */
#define rcu_first_leaf_node() (rcu_state.level[rcu_num_lvls - 1])

/* Is this rcu_node a leaf? */
#define rcu_is_leaf_node(rnp) ((rnp)->level == rcu_num_lvls - 1)

/* Is this rcu_node the last leaf? */
#define rcu_is_last_leaf_node(rnp) ((rnp) == &rcu_state.node[rcu_num_nodes - 1])

/*
 * Do a full breadth-first scan of the {s,}rcu_node structures for the
 * specified state structure (for SRCU) or the only rcu_state structure
 * (for RCU).
 */
#define srcu_for_each_node_breadth_first(sp, rnp) \
	for ((rnp) = &(sp)->node[0]; \
	     (rnp) < &(sp)->node[rcu_num_nodes]; (rnp)++)
#define rcu_for_each_node_breadth_first(rnp) \
	srcu_for_each_node_breadth_first(&rcu_state, rnp)

/*
 * Scan the leaves of the rcu_node hierarchy for the rcu_state structure.
 * Note that if there is a singleton rcu_node tree with but one rcu_node
 * structure, this loop -will- visit the rcu_node structure.  It is still
 * a leaf node, even if it is also the root node.
 */
#define rcu_for_each_leaf_node(rnp) \
	for ((rnp) = rcu_first_leaf_node(); \
	     (rnp) < &rcu_state.node[rcu_num_nodes]; (rnp)++)

/*
 * Iterate over all possible CPUs in a leaf RCU node.
 */
#define for_each_leaf_node_possible_cpu(rnp, cpu) \
	for (WARN_ON_ONCE(!rcu_is_leaf_node(rnp)), \
	     (cpu) = cpumask_next((rnp)->grplo - 1, cpu_possible_mask); \
	     (cpu) <= rnp->grphi; \
	     (cpu) = cpumask_next((cpu), cpu_possible_mask))

/*
 * Iterate over all CPUs in a leaf RCU node's specified mask.
 */
#define rcu_find_next_bit(rnp, cpu, mask) \
	((rnp)->grplo + find_next_bit(&(mask), BITS_PER_LONG, (cpu)))
#define for_each_leaf_node_cpu_mask(rnp, cpu, mask) \
	for (WARN_ON_ONCE(!rcu_is_leaf_node(rnp)), \
	     (cpu) = rcu_find_next_bit((rnp), 0, (mask)); \
	     (cpu) <= rnp->grphi; \
	     (cpu) = rcu_find_next_bit((rnp), (cpu) + 1 - (rnp->grplo), (mask)))

/*
 * Wrappers for the rcu_node::lock acquire and release.
 *
 * Because the rcu_nodes form a tree, the tree traversal locking will observe
 * different lock values, this in turn means that an UNLOCK of one level
 * followed by a LOCK of another level does not imply a full memory barrier;
 * and most importantly transitivity is lost.
 *
 * In order to restore full ordering between tree levels, augment the regular
 * lock acquire functions with smp_mb__after_unlock_lock().
 *
 * As ->lock of struct rcu_node is a __private field, therefore one should use
 * these wrappers rather than directly call raw_spin_{lock,unlock}* on ->lock.
 */
#define raw_spin_lock_rcu_node(p)					\
do {									\
	raw_spin_lock(&ACCESS_PRIVATE(p, lock));			\
	smp_mb__after_unlock_lock();					\
} while (0)

#define raw_spin_unlock_rcu_node(p) raw_spin_unlock(&ACCESS_PRIVATE(p, lock))

#define raw_spin_lock_irq_rcu_node(p)					\
do {									\
	raw_spin_lock_irq(&ACCESS_PRIVATE(p, lock));			\
	smp_mb__after_unlock_lock();					\
} while (0)

#define raw_spin_unlock_irq_rcu_node(p)					\
	raw_spin_unlock_irq(&ACCESS_PRIVATE(p, lock))

#define raw_spin_lock_irqsave_rcu_node(p, flags)			\
do {									\
	raw_spin_lock_irqsave(&ACCESS_PRIVATE(p, lock), flags);	\
	smp_mb__after_unlock_lock();					\
} while (0)

#define raw_spin_unlock_irqrestore_rcu_node(p, flags)			\
	raw_spin_unlock_irqrestore(&ACCESS_PRIVATE(p, lock), flags)

#define raw_spin_trylock_rcu_node(p)					\
({									\
	bool ___locked = raw_spin_trylock(&ACCESS_PRIVATE(p, lock));	\
									\
	if (___locked)							\
		smp_mb__after_unlock_lock();				\
	___locked;							\
})

#define raw_lockdep_assert_held_rcu_node(p)				\
	lockdep_assert_held(&ACCESS_PRIVATE(p, lock))

#endif /* #if defined(CONFIG_SRCU) || !defined(CONFIG_TINY_RCU) */

#ifdef CONFIG_SRCU
void srcu_init(void);
#else /* #ifdef CONFIG_SRCU */
static inline void srcu_init(void) { }
#endif /* #else #ifdef CONFIG_SRCU */

#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_normal(void) { return true; }
static inline bool rcu_gp_is_expedited(void) { return false; }
static inline void rcu_expedite_gp(void) { }
static inline void rcu_unexpedite_gp(void) { }
static inline void rcu_request_urgent_qs_task(struct task_struct *t) { }
#else /* #ifdef CONFIG_TINY_RCU */
bool rcu_gp_is_normal(void);     /* Internal RCU use. */
bool rcu_gp_is_expedited(void);  /* Internal RCU use. */
void rcu_expedite_gp(void);
void rcu_unexpedite_gp(void);
void rcupdate_announce_bootup_oddness(void);
void show_rcu_tasks_gp_kthreads(void);
void rcu_request_urgent_qs_task(struct task_struct *t);
#endif /* #else #ifdef CONFIG_TINY_RCU */

#define RCU_SCHEDULER_INACTIVE	0
#define RCU_SCHEDULER_INIT	1
#define RCU_SCHEDULER_RUNNING	2

enum rcutorture_type {
	RCU_FLAVOR,
	RCU_TASKS_FLAVOR,
	RCU_TASKS_RUDE_FLAVOR,
	RCU_TASKS_TRACING_FLAVOR,
	RCU_TRIVIAL_FLAVOR,
	SRCU_FLAVOR,
	INVALID_RCU_FLAVOR
};

#if defined(CONFIG_TREE_RCU)
void rcutorture_get_gp_data(enum rcutorture_type test_type, int *flags,
			    unsigned long *gp_seq);
void do_trace_rcu_torture_read(const char *rcutorturename,
			       struct rcu_head *rhp,
			       unsigned long secs,
			       unsigned long c_old,
			       unsigned long c);
void rcu_gp_set_torture_wait(int duration);
#else
static inline void rcutorture_get_gp_data(enum rcutorture_type test_type,
					  int *flags, unsigned long *gp_seq)
{
	*flags = 0;
	*gp_seq = 0;
}
#ifdef CONFIG_RCU_TRACE
void do_trace_rcu_torture_read(const char *rcutorturename,
			       struct rcu_head *rhp,
			       unsigned long secs,
			       unsigned long c_old,
			       unsigned long c);
#else
#define do_trace_rcu_torture_read(rcutorturename, rhp, secs, c_old, c) \
	do { } while (0)
#endif
static inline void rcu_gp_set_torture_wait(int duration) { }
#endif

#if IS_ENABLED(CONFIG_RCU_TORTURE_TEST) || IS_MODULE(CONFIG_RCU_TORTURE_TEST)
long rcutorture_sched_setaffinity(pid_t pid, const struct cpumask *in_mask);
#endif

#ifdef CONFIG_TINY_SRCU

static inline void srcutorture_get_gp_data(enum rcutorture_type test_type,
					   struct srcu_struct *sp, int *flags,
					   unsigned long *gp_seq)
{
	if (test_type != SRCU_FLAVOR)
		return;
	*flags = 0;
	*gp_seq = sp->srcu_idx;
}

#elif defined(CONFIG_TREE_SRCU)

void srcutorture_get_gp_data(enum rcutorture_type test_type,
			     struct srcu_struct *sp, int *flags,
			     unsigned long *gp_seq);

#endif

#ifdef CONFIG_TINY_RCU
static inline bool rcu_dynticks_zero_in_eqs(int cpu, int *vp) { return false; }
static inline unsigned long rcu_get_gp_seq(void) { return 0; }
static inline unsigned long rcu_exp_batches_completed(void) { return 0; }
static inline unsigned long
srcu_batches_completed(struct srcu_struct *sp) { return 0; }
static inline void rcu_force_quiescent_state(void) { }
static inline void show_rcu_gp_kthreads(void) { }
static inline int rcu_get_gp_kthreads_prio(void) { return 0; }
static inline void rcu_fwd_progress_check(unsigned long j) { }
#else /* #ifdef CONFIG_TINY_RCU */
bool rcu_dynticks_zero_in_eqs(int cpu, int *vp);
unsigned long rcu_get_gp_seq(void);
unsigned long rcu_exp_batches_completed(void);
unsigned long srcu_batches_completed(struct srcu_struct *sp);
void show_rcu_gp_kthreads(void);
int rcu_get_gp_kthreads_prio(void);
void rcu_fwd_progress_check(unsigned long j);
void rcu_force_quiescent_state(void);
extern struct workqueue_struct *rcu_gp_wq;
extern struct workqueue_struct *rcu_par_gp_wq;
#endif /* #else #ifdef CONFIG_TINY_RCU */

#ifdef CONFIG_RCU_NOCB_CPU
bool rcu_is_nocb_cpu(int cpu);
void rcu_bind_current_to_nocb(void);
#else
static inline bool rcu_is_nocb_cpu(int cpu) { return false; }
static inline void rcu_bind_current_to_nocb(void) { }
#endif

#endif /* __LINUX_RCU_H */
Commit	Line	Data
b5b11890	1	/* SPDX-License-Identifier: GPL-2.0+ */
29c00b4a PM	2	/*
	3	* Read-Copy Update definitions shared among RCU implementations.
	4	*
29c00b4a PM	5	* Copyright IBM Corporation, 2011
29c00b4a PM	6	*
b5b11890	7	* Author: Paul E. McKenney <paulmck@linux.ibm.com>
29c00b4a PM	8	*/
	9
	10	#ifndef __LINUX_RCU_H
	11	#define __LINUX_RCU_H
	12
5cb5c6e1	13	#include <trace/events/rcu.h>
e99033c5	14
c2d8089d	15	/* Offset to allow distinguishing irq vs. task-based idle entry/exit. */
84585aa8	16	#define DYNTICK_IRQ_NONIDLE ((LONG_MAX / 2) + 1)
6136d6e4	17
2e8c28c2 PM	18
	19	/*
	20	* Grace-period counter management.
	21	*/
	22
f1ec57a4	23	#define RCU_SEQ_CTR_SHIFT 2
031aeee0 PM	24	#define RCU_SEQ_STATE_MASK ((1 << RCU_SEQ_CTR_SHIFT) - 1)
	25
	26	/*
	27	* Return the counter portion of a sequence number previously returned
	28	* by rcu_seq_snap() or rcu_seq_current().
	29	*/
	30	static inline unsigned long rcu_seq_ctr(unsigned long s)
	31	{
	32	return s >> RCU_SEQ_CTR_SHIFT;
	33	}
	34
	35	/*
	36	* Return the state portion of a sequence number previously returned
	37	* by rcu_seq_snap() or rcu_seq_current().
	38	*/
	39	static inline int rcu_seq_state(unsigned long s)
	40	{
	41	return s & RCU_SEQ_STATE_MASK;
	42	}
	43
80a7956f PM	44	/*
	45	* Set the state portion of the pointed-to sequence number.
	46	* The caller is responsible for preventing conflicting updates.
	47	*/
	48	static inline void rcu_seq_set_state(unsigned long *sp, int newstate)
	49	{
	50	WARN_ON_ONCE(newstate & ~RCU_SEQ_STATE_MASK);
	51	WRITE_ONCE(sp, (sp & ~RCU_SEQ_STATE_MASK) + newstate);
	52	}
	53
2e8c28c2 PM	54	/* Adjust sequence number for start of update-side operation. */
	55	static inline void rcu_seq_start(unsigned long *sp)
	56	{
	57	WRITE_ONCE(sp, sp + 1);
	58	smp_mb(); /* Ensure update-side operation after counter increment. */
031aeee0	59	WARN_ON_ONCE(rcu_seq_state(*sp) != 1);
2e8c28c2 PM	60	}
2e8c28c2 PM	61
9a414201 PM	62	/* Compute the end-of-grace-period value for the specified sequence number. */
	63	static inline unsigned long rcu_seq_endval(unsigned long *sp)
	64	{
	65	return (*sp \| RCU_SEQ_STATE_MASK) + 1;
	66	}
	67
2e8c28c2 PM	68	/* Adjust sequence number for end of update-side operation. */
	69	static inline void rcu_seq_end(unsigned long *sp)
	70	{
	71	smp_mb(); /* Ensure update-side operation before counter increment. */
031aeee0	72	WARN_ON_ONCE(!rcu_seq_state(*sp));
9a414201	73	WRITE_ONCE(*sp, rcu_seq_endval(sp));
2e8c28c2 PM	74	}
2e8c28c2 PM	75
0d805a70 JFG	76	/*
	77	* rcu_seq_snap - Take a snapshot of the update side's sequence number.
	78	*
	79	* This function returns the earliest value of the grace-period sequence number
	80	* that will indicate that a full grace period has elapsed since the current
	81	* time. Once the grace-period sequence number has reached this value, it will
	82	* be safe to invoke all callbacks that have been registered prior to the
	83	* current time. This value is the current grace-period number plus two to the
	84	* power of the number of low-order bits reserved for state, then rounded up to
	85	* the next value in which the state bits are all zero.
	86	*/
2e8c28c2 PM	87	static inline unsigned long rcu_seq_snap(unsigned long *sp)
	88	{
	89	unsigned long s;
	90
031aeee0	91	s = (READ_ONCE(sp) + 2 RCU_SEQ_STATE_MASK + 1) & ~RCU_SEQ_STATE_MASK;
2e8c28c2 PM	92	smp_mb(); /* Above access must not bleed into critical section. */
	93	return s;
	94	}
	95
8660b7d8 PM	96	/* Return the current value the update side's sequence number, no ordering. */
	97	static inline unsigned long rcu_seq_current(unsigned long *sp)
	98	{
	99	return READ_ONCE(*sp);
	100	}
	101
2e3e5e55 PM	102	/*
	103	* Given a snapshot from rcu_seq_snap(), determine whether or not the
	104	* corresponding update-side operation has started.
	105	*/
	106	static inline bool rcu_seq_started(unsigned long *sp, unsigned long s)
	107	{
	108	return ULONG_CMP_LT((s - 1) & ~RCU_SEQ_STATE_MASK, READ_ONCE(*sp));
	109	}
	110
2e8c28c2 PM	111	/*
	112	* Given a snapshot from rcu_seq_snap(), determine whether or not a
	113	* full update-side operation has occurred.
	114	*/
	115	static inline bool rcu_seq_done(unsigned long *sp, unsigned long s)
	116	{
	117	return ULONG_CMP_GE(READ_ONCE(*sp), s);
	118	}
	119
67e14c1e PM	120	/*
	121	* Has a grace period completed since the time the old gp_seq was collected?
	122	*/
	123	static inline bool rcu_seq_completed_gp(unsigned long old, unsigned long new)
	124	{
	125	return ULONG_CMP_LT(old, new & ~RCU_SEQ_STATE_MASK);
	126	}
	127
	128	/*
	129	* Has a grace period started since the time the old gp_seq was collected?
	130	*/
	131	static inline bool rcu_seq_new_gp(unsigned long old, unsigned long new)
	132	{
	133	return ULONG_CMP_LT((old + RCU_SEQ_STATE_MASK) & ~RCU_SEQ_STATE_MASK,
	134	new);
	135	}
	136
d7219312 PM	137	/*
	138	* Roughly how many full grace periods have elapsed between the collection
	139	* of the two specified grace periods?
	140	*/
	141	static inline unsigned long rcu_seq_diff(unsigned long new, unsigned long old)
	142	{
2ee5aca5 PM	143	unsigned long rnd_diff;
	144
	145	if (old == new)
	146	return 0;
	147	/*
	148	* Compute the number of grace periods (still shifted up), plus
	149	* one if either of new and old is not an exact grace period.
	150	*/
	151	rnd_diff = (new & ~RCU_SEQ_STATE_MASK) -
	152	((old + RCU_SEQ_STATE_MASK) & ~RCU_SEQ_STATE_MASK) +
	153	((new & RCU_SEQ_STATE_MASK) \|\| (old & RCU_SEQ_STATE_MASK));
	154	if (ULONG_CMP_GE(RCU_SEQ_STATE_MASK, rnd_diff))
	155	return 1; /* Definitely no grace period has elapsed. */
	156	return ((rnd_diff - RCU_SEQ_STATE_MASK - 1) >> RCU_SEQ_CTR_SHIFT) + 2;
d7219312 PM	157	}
d7219312 PM	158
29c00b4a PM	159	/*
29c00b4a PM	160	* debug_rcu_head_queue()/debug_rcu_head_unqueue() are used internally
7f87c036 PM	161	* by call_rcu() and rcu callback execution, and are therefore not part
	162	* of the RCU API. These are in rcupdate.h because they are used by all
	163	* RCU implementations.
29c00b4a PM	164	*/
	165
	166	#ifdef CONFIG_DEBUG_OBJECTS_RCU_HEAD
	167	# define STATE_RCU_HEAD_READY 0
	168	# define STATE_RCU_HEAD_QUEUED 1
	169
	170	extern struct debug_obj_descr rcuhead_debug_descr;
	171
ae150184	172	static inline int debug_rcu_head_queue(struct rcu_head *head)
29c00b4a	173	{
ae150184 PM	174	int r1;
	175
	176	r1 = debug_object_activate(head, &rcuhead_debug_descr);
29c00b4a PM	177	debug_object_active_state(head, &rcuhead_debug_descr,
	178	STATE_RCU_HEAD_READY,
	179	STATE_RCU_HEAD_QUEUED);
ae150184	180	return r1;
29c00b4a PM	181	}
	182
	183	static inline void debug_rcu_head_unqueue(struct rcu_head *head)
	184	{
	185	debug_object_active_state(head, &rcuhead_debug_descr,
	186	STATE_RCU_HEAD_QUEUED,
	187	STATE_RCU_HEAD_READY);
	188	debug_object_deactivate(head, &rcuhead_debug_descr);
	189	}
	190	#else /* !CONFIG_DEBUG_OBJECTS_RCU_HEAD */
ae150184	191	static inline int debug_rcu_head_queue(struct rcu_head *head)
29c00b4a	192	{
ae150184	193	return 0;
29c00b4a PM	194	}
	195
	196	static inline void debug_rcu_head_unqueue(struct rcu_head *head)
	197	{
	198	}
	199	#endif /* #else !CONFIG_DEBUG_OBJECTS_RCU_HEAD */
	200
58c53360 PM	201	extern int rcu_cpu_stall_suppress_at_boot;
	202
	203	static inline bool rcu_stall_is_suppressed_at_boot(void)
	204	{
	205	return rcu_cpu_stall_suppress_at_boot && !rcu_inkernel_boot_has_ended();
	206	}
	207
6bfc09e2 PM	208	#ifdef CONFIG_RCU_STALL_COMMON
6bfc09e2 PM	209
cdc694b2	210	extern int rcu_cpu_stall_ftrace_dump;
6bfc09e2	211	extern int rcu_cpu_stall_suppress;
10462d6f	212	extern int rcu_cpu_stall_timeout;
6bfc09e2 PM	213	int rcu_jiffies_till_stall_check(void);
6bfc09e2 PM	214
58c53360 PM	215	static inline bool rcu_stall_is_suppressed(void)
	216	{
	217	return rcu_stall_is_suppressed_at_boot() \|\| rcu_cpu_stall_suppress;
	218	}
	219
f22ce091 PM	220	#define rcu_ftrace_dump_stall_suppress() \
	221	do { \
	222	if (!rcu_cpu_stall_suppress) \
	223	rcu_cpu_stall_suppress = 3; \
	224	} while (0)
	225
	226	#define rcu_ftrace_dump_stall_unsuppress() \
	227	do { \
	228	if (rcu_cpu_stall_suppress == 3) \
	229	rcu_cpu_stall_suppress = 0; \
	230	} while (0)
	231
	232	#else /* #endif #ifdef CONFIG_RCU_STALL_COMMON */
58c53360 PM	233
	234	static inline bool rcu_stall_is_suppressed(void)
	235	{
	236	return rcu_stall_is_suppressed_at_boot();
	237	}
f22ce091 PM	238	#define rcu_ftrace_dump_stall_suppress()
f22ce091 PM	239	#define rcu_ftrace_dump_stall_unsuppress()
6bfc09e2 PM	240	#endif /* #ifdef CONFIG_RCU_STALL_COMMON */
6bfc09e2 PM	241
0d752924 PM	242	/*
	243	* Strings used in tracepoints need to be exported via the
	244	* tracing system such that tools like perf and trace-cmd can
	245	* translate the string address pointers to actual text.
	246	*/
	247	#define TPS(x) tracepoint_string(x)
	248
b8989b76 PM	249	/*
	250	* Dump the ftrace buffer, but only one time per callsite per boot.
	251	*/
	252	#define rcu_ftrace_dump(oops_dump_mode) \
	253	do { \
	254	static atomic_t ___rfd_beenhere = ATOMIC_INIT(0); \
	255	\
	256	if (!atomic_read(&___rfd_beenhere) && \
83b6ca1f PM	257	!atomic_xchg(&___rfd_beenhere, 1)) { \
83b6ca1f PM	258	tracing_off(); \
f22ce091	259	rcu_ftrace_dump_stall_suppress(); \
b8989b76	260	ftrace_dump(oops_dump_mode); \
f22ce091	261	rcu_ftrace_dump_stall_unsuppress(); \
83b6ca1f	262	} \
b8989b76 PM	263	} while (0)
b8989b76 PM	264
aa23c6fb	265	void rcu_early_boot_tests(void);
52d7e48b	266	void rcu_test_sync_prims(void);
aa23c6fb	267
5f6130fa LJ	268	/*
	269	* This function really isn't for public consumption, but RCU is special in
	270	* that context switches can allow the state machine to make progress.
	271	*/
	272	extern void resched_cpu(int cpu);
	273
822175e7	274	#if defined(CONFIG_SRCU) \|\| !defined(CONFIG_TINY_RCU)
2b34c43c PM	275
	276	#include <linux/rcu_node_tree.h>
	277
	278	extern int rcu_num_lvls;
e95d68d2	279	extern int num_rcu_lvl[];
2b34c43c PM	280	extern int rcu_num_nodes;
	281	static bool rcu_fanout_exact;
	282	static int rcu_fanout_leaf;
	283
	284	/*
	285	* Compute the per-level fanout, either using the exact fanout specified
	286	* or balancing the tree, depending on the rcu_fanout_exact boot parameter.
	287	*/
	288	static inline void rcu_init_levelspread(int levelspread, const int levelcnt)
	289	{
	290	int i;
	291
36b5dae6 PM	292	for (i = 0; i < RCU_NUM_LVLS; i++)
36b5dae6 PM	293	levelspread[i] = INT_MIN;
2b34c43c PM	294	if (rcu_fanout_exact) {
	295	levelspread[rcu_num_lvls - 1] = rcu_fanout_leaf;
	296	for (i = rcu_num_lvls - 2; i >= 0; i--)
	297	levelspread[i] = RCU_FANOUT;
	298	} else {
	299	int ccur;
	300	int cprv;
	301
	302	cprv = nr_cpu_ids;
	303	for (i = rcu_num_lvls - 1; i >= 0; i--) {
	304	ccur = levelcnt[i];
	305	levelspread[i] = (cprv + ccur - 1) / ccur;
	306	cprv = ccur;
	307	}
	308	}
	309	}
	310
7f87c036	311	/* Returns a pointer to the first leaf rcu_node structure. */
aedf4ba9	312	#define rcu_first_leaf_node() (rcu_state.level[rcu_num_lvls - 1])
5b4c11d5 PM	313
	314	/* Is this rcu_node a leaf? */
	315	#define rcu_is_leaf_node(rnp) ((rnp)->level == rcu_num_lvls - 1)
	316
5257514d	317	/* Is this rcu_node the last leaf? */
aedf4ba9	318	#define rcu_is_last_leaf_node(rnp) ((rnp) == &rcu_state.node[rcu_num_nodes - 1])
5257514d	319
efbe451d	320	/*
aedf4ba9	321	* Do a full breadth-first scan of the {s,}rcu_node structures for the
7f87c036 PM	322	* specified state structure (for SRCU) or the only rcu_state structure
7f87c036 PM	323	* (for RCU).
efbe451d	324	*/
aedf4ba9 PM	325	#define srcu_for_each_node_breadth_first(sp, rnp) \
	326	for ((rnp) = &(sp)->node[0]; \
	327	(rnp) < &(sp)->node[rcu_num_nodes]; (rnp)++)
	328	#define rcu_for_each_node_breadth_first(rnp) \
	329	srcu_for_each_node_breadth_first(&rcu_state, rnp)
efbe451d PM	330
efbe451d PM	331	/*
7f87c036 PM	332	* Scan the leaves of the rcu_node hierarchy for the rcu_state structure.
	333	* Note that if there is a singleton rcu_node tree with but one rcu_node
	334	* structure, this loop -will- visit the rcu_node structure. It is still
	335	* a leaf node, even if it is also the root node.
efbe451d	336	*/
aedf4ba9 PM	337	#define rcu_for_each_leaf_node(rnp) \
	338	for ((rnp) = rcu_first_leaf_node(); \
	339	(rnp) < &rcu_state.node[rcu_num_nodes]; (rnp)++)
efbe451d PM	340
	341	/*
	342	* Iterate over all possible CPUs in a leaf RCU node.
	343	*/
	344	#define for_each_leaf_node_possible_cpu(rnp, cpu) \
82dd8419 PM	345	for (WARN_ON_ONCE(!rcu_is_leaf_node(rnp)), \
82dd8419 PM	346	(cpu) = cpumask_next((rnp)->grplo - 1, cpu_possible_mask); \
65963d24 PM	347	(cpu) <= rnp->grphi; \
	348	(cpu) = cpumask_next((cpu), cpu_possible_mask))
	349
	350	/*
	351	* Iterate over all CPUs in a leaf RCU node's specified mask.
	352	*/
	353	#define rcu_find_next_bit(rnp, cpu, mask) \
	354	((rnp)->grplo + find_next_bit(&(mask), BITS_PER_LONG, (cpu)))
	355	#define for_each_leaf_node_cpu_mask(rnp, cpu, mask) \
82dd8419 PM	356	for (WARN_ON_ONCE(!rcu_is_leaf_node(rnp)), \
82dd8419 PM	357	(cpu) = rcu_find_next_bit((rnp), 0, (mask)); \
65963d24 PM	358	(cpu) <= rnp->grphi; \
65963d24 PM	359	(cpu) = rcu_find_next_bit((rnp), (cpu) + 1 - (rnp->grplo), (mask)))
efbe451d	360
83d40bd3 PM	361	/*
	362	* Wrappers for the rcu_node::lock acquire and release.
	363	*
	364	* Because the rcu_nodes form a tree, the tree traversal locking will observe
	365	* different lock values, this in turn means that an UNLOCK of one level
	366	* followed by a LOCK of another level does not imply a full memory barrier;
	367	* and most importantly transitivity is lost.
	368	*
	369	* In order to restore full ordering between tree levels, augment the regular
	370	* lock acquire functions with smp_mb__after_unlock_lock().
	371	*
	372	* As ->lock of struct rcu_node is a __private field, therefore one should use
	373	* these wrappers rather than directly call raw_spin_{lock,unlock}* on ->lock.
	374	*/
	375	#define raw_spin_lock_rcu_node(p) \
	376	do { \
	377	raw_spin_lock(&ACCESS_PRIVATE(p, lock)); \
	378	smp_mb__after_unlock_lock(); \
	379	} while (0)
	380
	381	#define raw_spin_unlock_rcu_node(p) raw_spin_unlock(&ACCESS_PRIVATE(p, lock))
	382
	383	#define raw_spin_lock_irq_rcu_node(p) \
	384	do { \
	385	raw_spin_lock_irq(&ACCESS_PRIVATE(p, lock)); \
	386	smp_mb__after_unlock_lock(); \
	387	} while (0)
	388
	389	#define raw_spin_unlock_irq_rcu_node(p) \
	390	raw_spin_unlock_irq(&ACCESS_PRIVATE(p, lock))
	391
4e4bea74	392	#define raw_spin_lock_irqsave_rcu_node(p, flags) \
83d40bd3	393	do { \
4e4bea74	394	raw_spin_lock_irqsave(&ACCESS_PRIVATE(p, lock), flags); \
83d40bd3 PM	395	smp_mb__after_unlock_lock(); \
	396	} while (0)
	397
4e4bea74	398	#define raw_spin_unlock_irqrestore_rcu_node(p, flags) \
a32e01ee	399	raw_spin_unlock_irqrestore(&ACCESS_PRIVATE(p, lock), flags)
83d40bd3 PM	400
	401	#define raw_spin_trylock_rcu_node(p) \
	402	({ \
	403	bool ___locked = raw_spin_trylock(&ACCESS_PRIVATE(p, lock)); \
	404	\
	405	if (___locked) \
	406	smp_mb__after_unlock_lock(); \
	407	___locked; \
	408	})
	409
a32e01ee MW	410	#define raw_lockdep_assert_held_rcu_node(p) \
	411	lockdep_assert_held(&ACCESS_PRIVATE(p, lock))
	412
822175e7	413	#endif /* #if defined(CONFIG_SRCU) \|\| !defined(CONFIG_TINY_RCU) */
2b34c43c	414
e0fcba9a PM	415	#ifdef CONFIG_SRCU
	416	void srcu_init(void);
	417	#else /* #ifdef CONFIG_SRCU */
	418	static inline void srcu_init(void) { }
	419	#endif /* #else #ifdef CONFIG_SRCU */
	420
25c36329 PM	421	#ifdef CONFIG_TINY_RCU
25c36329 PM	422	/* Tiny RCU doesn't expedite, as its purpose in life is instead to be tiny. */
7414fac0 PM	423	static inline bool rcu_gp_is_normal(void) { return true; }
	424	static inline bool rcu_gp_is_expedited(void) { return false; }
	425	static inline void rcu_expedite_gp(void) { }
	426	static inline void rcu_unexpedite_gp(void) { }
bfbd767d	427	static inline void rcu_request_urgent_qs_task(struct task_struct *t) { }
25c36329 PM	428	#else /* #ifdef CONFIG_TINY_RCU */
	429	bool rcu_gp_is_normal(void); /* Internal RCU use. */
	430	bool rcu_gp_is_expedited(void); /* Internal RCU use. */
	431	void rcu_expedite_gp(void);
	432	void rcu_unexpedite_gp(void);
	433	void rcupdate_announce_bootup_oddness(void);
e21408ce	434	void show_rcu_tasks_gp_kthreads(void);
bfbd767d	435	void rcu_request_urgent_qs_task(struct task_struct *t);
25c36329 PM	436	#endif /* #else #ifdef CONFIG_TINY_RCU */
25c36329 PM	437
82118249 PM	438	#define RCU_SCHEDULER_INACTIVE 0
	439	#define RCU_SCHEDULER_INIT 1
	440	#define RCU_SCHEDULER_RUNNING 2
	441
cad7b389 PM	442	enum rcutorture_type {
cad7b389 PM	443	RCU_FLAVOR,
cad7b389	444	RCU_TASKS_FLAVOR,
3d6e43c7	445	RCU_TASKS_RUDE_FLAVOR,
c1a76c0b	446	RCU_TASKS_TRACING_FLAVOR,
c682db55	447	RCU_TRIVIAL_FLAVOR,
cad7b389 PM	448	SRCU_FLAVOR,
	449	INVALID_RCU_FLAVOR
	450	};
	451
b3e627d3	452	#if defined(CONFIG_TREE_RCU)
cad7b389	453	void rcutorture_get_gp_data(enum rcutorture_type test_type, int *flags,
aebc8264	454	unsigned long *gp_seq);
cad7b389 PM	455	void do_trace_rcu_torture_read(const char *rcutorturename,
	456	struct rcu_head *rhp,
	457	unsigned long secs,
	458	unsigned long c_old,
	459	unsigned long c);
55b2dcf5	460	void rcu_gp_set_torture_wait(int duration);
cad7b389 PM	461	#else
cad7b389 PM	462	static inline void rcutorture_get_gp_data(enum rcutorture_type test_type,
aebc8264	463	int flags, unsigned long gp_seq)
cad7b389 PM	464	{
cad7b389 PM	465	*flags = 0;
aebc8264	466	*gp_seq = 0;
cad7b389	467	}
cad7b389 PM	468	#ifdef CONFIG_RCU_TRACE
	469	void do_trace_rcu_torture_read(const char *rcutorturename,
	470	struct rcu_head *rhp,
	471	unsigned long secs,
	472	unsigned long c_old,
	473	unsigned long c);
	474	#else
	475	#define do_trace_rcu_torture_read(rcutorturename, rhp, secs, c_old, c) \
	476	do { } while (0)
	477	#endif
55b2dcf5	478	static inline void rcu_gp_set_torture_wait(int duration) { }
cad7b389 PM	479	#endif
cad7b389 PM	480
c682db55 PM	481	#if IS_ENABLED(CONFIG_RCU_TORTURE_TEST) \|\| IS_MODULE(CONFIG_RCU_TORTURE_TEST)
	482	long rcutorture_sched_setaffinity(pid_t pid, const struct cpumask *in_mask);
	483	#endif
	484
cad7b389 PM	485	#ifdef CONFIG_TINY_SRCU
	486
	487	static inline void srcutorture_get_gp_data(enum rcutorture_type test_type,
	488	struct srcu_struct sp, int flags,
aebc8264	489	unsigned long *gp_seq)
cad7b389 PM	490	{
	491	if (test_type != SRCU_FLAVOR)
	492	return;
	493	*flags = 0;
aebc8264	494	*gp_seq = sp->srcu_idx;
cad7b389 PM	495	}
	496
	497	#elif defined(CONFIG_TREE_SRCU)
	498
	499	void srcutorture_get_gp_data(enum rcutorture_type test_type,
	500	struct srcu_struct sp, int flags,
aebc8264	501	unsigned long *gp_seq);
cad7b389	502
cad7b389 PM	503	#endif
cad7b389 PM	504
e3c8d51e	505	#ifdef CONFIG_TINY_RCU
7d0c9c50	506	static inline bool rcu_dynticks_zero_in_eqs(int cpu, int *vp) { return false; }
17ef2fe9	507	static inline unsigned long rcu_get_gp_seq(void) { return 0; }
7414fac0	508	static inline unsigned long rcu_exp_batches_completed(void) { return 0; }
7414fac0 PM	509	static inline unsigned long
	510	srcu_batches_completed(struct srcu_struct *sp) { return 0; }
	511	static inline void rcu_force_quiescent_state(void) { }
7414fac0	512	static inline void show_rcu_gp_kthreads(void) { }
4babd855	513	static inline int rcu_get_gp_kthreads_prio(void) { return 0; }
e0aff973	514	static inline void rcu_fwd_progress_check(unsigned long j) { }
e3c8d51e	515	#else /* #ifdef CONFIG_TINY_RCU */
7d0c9c50	516	bool rcu_dynticks_zero_in_eqs(int cpu, int *vp);
17ef2fe9	517	unsigned long rcu_get_gp_seq(void);
e3c8d51e	518	unsigned long rcu_exp_batches_completed(void);
5a0465e1	519	unsigned long srcu_batches_completed(struct srcu_struct *sp);
e3c8d51e	520	void show_rcu_gp_kthreads(void);
4babd855	521	int rcu_get_gp_kthreads_prio(void);
e0aff973	522	void rcu_fwd_progress_check(unsigned long j);
e3c8d51e	523	void rcu_force_quiescent_state(void);
ad7c946b	524	extern struct workqueue_struct *rcu_gp_wq;
25f3d7ef	525	extern struct workqueue_struct *rcu_par_gp_wq;
e3c8d51e PM	526	#endif /* #else #ifdef CONFIG_TINY_RCU */
e3c8d51e PM	527
44c65ff2	528	#ifdef CONFIG_RCU_NOCB_CPU
3d54f798	529	bool rcu_is_nocb_cpu(int cpu);
5ab7ab83	530	void rcu_bind_current_to_nocb(void);
3d54f798 PM	531	#else
3d54f798 PM	532	static inline bool rcu_is_nocb_cpu(int cpu) { return false; }
5ab7ab83	533	static inline void rcu_bind_current_to_nocb(void) { }
3d54f798 PM	534	#endif
3d54f798 PM	535
29c00b4a	536	#endif /* __LINUX_RCU_H */