tree.h 29.2 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
/*
 * Read-Copy Update mechanism for mutual exclusion (tree-based version)
 * Internal non-public definitions.
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
16 17
 * along with this program; if not, you can access it online at
 * http://www.gnu.org/licenses/gpl-2.0.html.
18 19 20 21 22 23 24 25 26
 *
 * Copyright IBM Corporation, 2008
 *
 * Author: Ingo Molnar <mingo@elte.hu>
 *	   Paul E. McKenney <paulmck@linux.vnet.ibm.com>
 */

#include <linux/cache.h>
#include <linux/spinlock.h>
27
#include <linux/rtmutex.h>
28 29 30
#include <linux/threads.h>
#include <linux/cpumask.h>
#include <linux/seqlock.h>
31
#include <linux/swait.h>
32
#include <linux/stop_machine.h>
33 34

/*
35 36
 * Define shape of hierarchy based on NR_CPUS, CONFIG_RCU_FANOUT, and
 * CONFIG_RCU_FANOUT_LEAF.
37
 * In theory, it should be possible to add more levels straightforwardly.
38 39
 * In practice, this did work well going from three levels to four.
 * Of course, your mileage may vary.
40
 */
41 42 43 44 45 46 47 48 49 50 51

#ifdef CONFIG_RCU_FANOUT
#define RCU_FANOUT CONFIG_RCU_FANOUT
#else /* #ifdef CONFIG_RCU_FANOUT */
# ifdef CONFIG_64BIT
# define RCU_FANOUT 64
# else
# define RCU_FANOUT 32
# endif
#endif /* #else #ifdef CONFIG_RCU_FANOUT */

52 53 54
#ifdef CONFIG_RCU_FANOUT_LEAF
#define RCU_FANOUT_LEAF CONFIG_RCU_FANOUT_LEAF
#else /* #ifdef CONFIG_RCU_FANOUT_LEAF */
55
#define RCU_FANOUT_LEAF 16
56 57 58
#endif /* #else #ifdef CONFIG_RCU_FANOUT_LEAF */

#define RCU_FANOUT_1	      (RCU_FANOUT_LEAF)
59 60 61
#define RCU_FANOUT_2	      (RCU_FANOUT_1 * RCU_FANOUT)
#define RCU_FANOUT_3	      (RCU_FANOUT_2 * RCU_FANOUT)
#define RCU_FANOUT_4	      (RCU_FANOUT_3 * RCU_FANOUT)
62

63
#if NR_CPUS <= RCU_FANOUT_1
64
#  define RCU_NUM_LVLS	      1
65
#  define NUM_RCU_LVL_0	      1
66
#  define NUM_RCU_NODES	      NUM_RCU_LVL_0
67 68 69
#  define NUM_RCU_LVL_INIT    { NUM_RCU_LVL_0 }
#  define RCU_NODE_NAME_INIT  { "rcu_node_0" }
#  define RCU_FQS_NAME_INIT   { "rcu_node_fqs_0" }
70
#elif NR_CPUS <= RCU_FANOUT_2
71
#  define RCU_NUM_LVLS	      2
72
#  define NUM_RCU_LVL_0	      1
73
#  define NUM_RCU_LVL_1	      DIV_ROUND_UP(NR_CPUS, RCU_FANOUT_1)
74
#  define NUM_RCU_NODES	      (NUM_RCU_LVL_0 + NUM_RCU_LVL_1)
75 76 77
#  define NUM_RCU_LVL_INIT    { NUM_RCU_LVL_0, NUM_RCU_LVL_1 }
#  define RCU_NODE_NAME_INIT  { "rcu_node_0", "rcu_node_1" }
#  define RCU_FQS_NAME_INIT   { "rcu_node_fqs_0", "rcu_node_fqs_1" }
78
#elif NR_CPUS <= RCU_FANOUT_3
79
#  define RCU_NUM_LVLS	      3
80
#  define NUM_RCU_LVL_0	      1
81 82
#  define NUM_RCU_LVL_1	      DIV_ROUND_UP(NR_CPUS, RCU_FANOUT_2)
#  define NUM_RCU_LVL_2	      DIV_ROUND_UP(NR_CPUS, RCU_FANOUT_1)
83
#  define NUM_RCU_NODES	      (NUM_RCU_LVL_0 + NUM_RCU_LVL_1 + NUM_RCU_LVL_2)
84 85 86
#  define NUM_RCU_LVL_INIT    { NUM_RCU_LVL_0, NUM_RCU_LVL_1, NUM_RCU_LVL_2 }
#  define RCU_NODE_NAME_INIT  { "rcu_node_0", "rcu_node_1", "rcu_node_2" }
#  define RCU_FQS_NAME_INIT   { "rcu_node_fqs_0", "rcu_node_fqs_1", "rcu_node_fqs_2" }
87
#elif NR_CPUS <= RCU_FANOUT_4
88
#  define RCU_NUM_LVLS	      4
89
#  define NUM_RCU_LVL_0	      1
90 91 92
#  define NUM_RCU_LVL_1	      DIV_ROUND_UP(NR_CPUS, RCU_FANOUT_3)
#  define NUM_RCU_LVL_2	      DIV_ROUND_UP(NR_CPUS, RCU_FANOUT_2)
#  define NUM_RCU_LVL_3	      DIV_ROUND_UP(NR_CPUS, RCU_FANOUT_1)
93
#  define NUM_RCU_NODES	      (NUM_RCU_LVL_0 + NUM_RCU_LVL_1 + NUM_RCU_LVL_2 + NUM_RCU_LVL_3)
94 95 96
#  define NUM_RCU_LVL_INIT    { NUM_RCU_LVL_0, NUM_RCU_LVL_1, NUM_RCU_LVL_2, NUM_RCU_LVL_3 }
#  define RCU_NODE_NAME_INIT  { "rcu_node_0", "rcu_node_1", "rcu_node_2", "rcu_node_3" }
#  define RCU_FQS_NAME_INIT   { "rcu_node_fqs_0", "rcu_node_fqs_1", "rcu_node_fqs_2", "rcu_node_fqs_3" }
97 98
#else
# error "CONFIG_RCU_FANOUT insufficient for NR_CPUS"
99
#endif /* #if (NR_CPUS) <= RCU_FANOUT_1 */
100

101 102 103
extern int rcu_num_lvls;
extern int rcu_num_nodes;

104 105 106 107
/*
 * Dynticks per-CPU state.
 */
struct rcu_dynticks {
108 109 110 111
	long long dynticks_nesting; /* Track irq/process nesting level. */
				    /* Process level is worth LLONG_MAX/2. */
	int dynticks_nmi_nesting;   /* Track NMI nesting level. */
	atomic_t dynticks;	    /* Even value for idle, else odd. */
112
	bool rcu_need_heavy_qs;     /* GP old, need heavy quiescent state. */
113
	unsigned long rcu_qs_ctr;   /* Light universal quiescent state ctr. */
114
	bool rcu_urgent_qs;	    /* GP old need light quiescent state. */
115 116 117 118 119 120 121 122
#ifdef CONFIG_NO_HZ_FULL_SYSIDLE
	long long dynticks_idle_nesting;
				    /* irq/process nesting level from idle. */
	atomic_t dynticks_idle;	    /* Even value for idle, else odd. */
				    /*  "Idle" excludes userspace execution. */
	unsigned long dynticks_idle_jiffies;
				    /* End of last non-NMI non-idle period. */
#endif /* #ifdef CONFIG_NO_HZ_FULL_SYSIDLE */
123
#ifdef CONFIG_RCU_FAST_NO_HZ
124
	bool all_lazy;		    /* Are all CPU's CBs lazy? */
125 126 127 128
	unsigned long nonlazy_posted;
				    /* # times non-lazy CBs posted to CPU. */
	unsigned long nonlazy_posted_snap;
				    /* idle-period nonlazy_posted snapshot. */
129 130
	unsigned long last_accelerate;
				    /* Last jiffy CBs were accelerated. */
131 132
	unsigned long last_advance_all;
				    /* Last jiffy CBs were all advanced. */
133
	int tick_nohz_enabled_snap; /* Previously seen value from sysfs. */
134
#endif /* #ifdef CONFIG_RCU_FAST_NO_HZ */
135 136
};

137 138 139 140
/* RCU's kthread states for tracing. */
#define RCU_KTHREAD_STOPPED  0
#define RCU_KTHREAD_RUNNING  1
#define RCU_KTHREAD_WAITING  2
141 142 143
#define RCU_KTHREAD_OFFCPU   3
#define RCU_KTHREAD_YIELDING 4
#define RCU_KTHREAD_MAX      4
144

145 146 147 148
/*
 * Definition for node within the RCU grace-period-detection hierarchy.
 */
struct rcu_node {
Boqun Feng's avatar
Boqun Feng committed
149 150 151
	raw_spinlock_t __private lock;	/* Root rcu_node's lock protects */
					/*  some rcu_state fields as well as */
					/*  following. */
152
	unsigned long gpnum;	/* Current grace period for this node. */
153 154
				/*  This will either be equal to or one */
				/*  behind the root rcu_node's gpnum. */
155
	unsigned long completed; /* Last GP completed for this node. */
156 157
				/*  This will either be equal to or one */
				/*  behind the root rcu_node's gpnum. */
158 159
	unsigned long qsmask;	/* CPUs or groups that need to switch in */
				/*  order for current grace period to proceed.*/
160 161 162 163
				/*  In leaf rcu_node, each bit corresponds to */
				/*  an rcu_data structure, otherwise, each */
				/*  bit corresponds to a child rcu_node */
				/*  structure. */
164
	unsigned long qsmaskinit;
165
				/* Per-GP initial value for qsmask. */
166 167 168 169
				/*  Initialized from ->qsmaskinitnext at the */
				/*  beginning of each grace period. */
	unsigned long qsmaskinitnext;
				/* Online CPUs for next grace period. */
170 171 172 173 174 175 176 177 178
	unsigned long expmask;	/* CPUs or groups that need to check in */
				/*  to allow the current expedited GP */
				/*  to complete. */
	unsigned long expmaskinit;
				/* Per-GP initial values for expmask. */
				/*  Initialized from ->expmaskinitnext at the */
				/*  beginning of each expedited GP. */
	unsigned long expmaskinitnext;
				/* Online CPUs for next expedited GP. */
179 180
				/*  Any CPU that has ever been online will */
				/*  have its bit set. */
181
	unsigned long grpmask;	/* Mask to apply to parent qsmask. */
182
				/*  Only one bit will be set in this mask. */
183 184 185 186
	int	grplo;		/* lowest-numbered CPU or group here. */
	int	grphi;		/* highest-numbered CPU or group here. */
	u8	grpnum;		/* CPU/group number for next level up. */
	u8	level;		/* root is at level 0. */
187 188 189 190
	bool	wait_blkd_tasks;/* Necessary to wait for blocked tasks to */
				/*  exit RCU read-side critical sections */
				/*  before propagating offline up the */
				/*  rcu_node tree? */
191
	struct rcu_node *parent;
192 193 194 195 196 197 198 199 200 201 202 203 204 205
	struct list_head blkd_tasks;
				/* Tasks blocked in RCU read-side critical */
				/*  section.  Tasks are placed at the head */
				/*  of this list and age towards the tail. */
	struct list_head *gp_tasks;
				/* Pointer to the first task blocking the */
				/*  current grace period, or NULL if there */
				/*  is no such task. */
	struct list_head *exp_tasks;
				/* Pointer to the first task blocking the */
				/*  current expedited grace period, or NULL */
				/*  if there is no such task.  If there */
				/*  is no current expedited grace period, */
				/*  then there can cannot be any such task. */
206 207 208 209 210 211 212 213
	struct list_head *boost_tasks;
				/* Pointer to first task that needs to be */
				/*  priority boosted, or NULL if no priority */
				/*  boosting is needed for this rcu_node */
				/*  structure.  If there are no tasks */
				/*  queued on this rcu_node structure that */
				/*  are blocking the current grace period, */
				/*  there can be no such task. */
214 215 216
	struct rt_mutex boost_mtx;
				/* Used only for the priority-boosting */
				/*  side effect, not as a lock. */
217 218 219 220 221
	unsigned long boost_time;
				/* When to start boosting (jiffies). */
	struct task_struct *boost_kthread_task;
				/* kthread that takes care of priority */
				/*  boosting for this rcu_node structure. */
222 223
	unsigned int boost_kthread_status;
				/* State of boost_kthread_task for tracing. */
224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242
	unsigned long n_tasks_boosted;
				/* Total number of tasks boosted. */
	unsigned long n_exp_boosts;
				/* Number of tasks boosted for expedited GP. */
	unsigned long n_normal_boosts;
				/* Number of tasks boosted for normal GP. */
	unsigned long n_balk_blkd_tasks;
				/* Refused to boost: no blocked tasks. */
	unsigned long n_balk_exp_gp_tasks;
				/* Refused to boost: nothing blocking GP. */
	unsigned long n_balk_boost_tasks;
				/* Refused to boost: already boosting. */
	unsigned long n_balk_notblocked;
				/* Refused to boost: RCU RS CS still running. */
	unsigned long n_balk_notyet;
				/* Refused to boost: not yet time. */
	unsigned long n_balk_nos;
				/* Refused to boost: not sure why, though. */
				/*  This can happen due to race conditions. */
243
#ifdef CONFIG_RCU_NOCB_CPU
244
	struct swait_queue_head nocb_gp_wq[2];
245 246
				/* Place for rcu_nocb_kthread() to wait GP. */
#endif /* #ifdef CONFIG_RCU_NOCB_CPU */
247 248
	int need_future_gp[2];
				/* Counts of upcoming no-CB GP requests. */
249
	raw_spinlock_t fqslock ____cacheline_internodealigned_in_smp;
250

251 252
	spinlock_t exp_lock ____cacheline_internodealigned_in_smp;
	unsigned long exp_seq_rq;
253
	wait_queue_head_t exp_wq[4];
254 255
} ____cacheline_internodealigned_in_smp;

256 257 258 259 260 261 262
/*
 * Bitmasks in an rcu_node cover the interval [grplo, grphi] of CPU IDs, and
 * are indexed relative to this interval rather than the global CPU ID space.
 * This generates the bit for a CPU in node-local masks.
 */
#define leaf_node_cpu_bit(rnp, cpu) (1UL << ((cpu) - (rnp)->grplo))

263 264 265 266 267 268
/*
 * Do a full breadth-first scan of the rcu_node structures for the
 * specified rcu_state structure.
 */
#define rcu_for_each_node_breadth_first(rsp, rnp) \
	for ((rnp) = &(rsp)->node[0]; \
269
	     (rnp) < &(rsp)->node[rcu_num_nodes]; (rnp)++)
270

271 272 273 274 275 276 277
/*
 * Do a breadth-first scan of the non-leaf rcu_node structures for the
 * specified rcu_state structure.  Note that if there is a singleton
 * rcu_node tree with but one rcu_node structure, this loop is a no-op.
 */
#define rcu_for_each_nonleaf_node_breadth_first(rsp, rnp) \
	for ((rnp) = &(rsp)->node[0]; \
278
	     (rnp) < (rsp)->level[rcu_num_lvls - 1]; (rnp)++)
279 280 281 282 283 284 285

/*
 * Scan the leaves of the rcu_node hierarchy for the specified 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.
 */
286
#define rcu_for_each_leaf_node(rsp, rnp) \
287 288
	for ((rnp) = (rsp)->level[rcu_num_lvls - 1]; \
	     (rnp) < &(rsp)->node[rcu_num_nodes]; (rnp)++)
289

290 291 292 293 294 295 296 297
/*
 * Iterate over all possible CPUs in a leaf RCU node.
 */
#define for_each_leaf_node_possible_cpu(rnp, cpu) \
	for ((cpu) = cpumask_next(rnp->grplo - 1, cpu_possible_mask); \
	     cpu <= rnp->grphi; \
	     cpu = cpumask_next((cpu), cpu_possible_mask))

298 299 300 301 302 303 304 305 306 307 308 309
/*
 * Union to allow "aggregate OR" operation on the need for a quiescent
 * state by the normal and expedited grace periods.
 */
union rcu_noqs {
	struct {
		u8 norm;
		u8 exp;
	} b; /* Bits. */
	u16 s; /* Set of bits, aggregate OR here. */
};

310 311 312 313 314 315 316 317 318 319
/* Index values for nxttail array in struct rcu_data. */
#define RCU_DONE_TAIL		0	/* Also RCU_WAIT head. */
#define RCU_WAIT_TAIL		1	/* Also RCU_NEXT_READY head. */
#define RCU_NEXT_READY_TAIL	2	/* Also RCU_NEXT head. */
#define RCU_NEXT_TAIL		3
#define RCU_NEXT_SIZE		4

/* Per-CPU data for read-copy update. */
struct rcu_data {
	/* 1) quiescent-state and grace-period handling : */
320
	unsigned long	completed;	/* Track rsp->completed gp number */
321
					/*  in order to detect GP end. */
322
	unsigned long	gpnum;		/* Highest gp number that this CPU */
323
					/*  is aware of having started. */
324 325
	unsigned long	rcu_qs_ctr_snap;/* Snapshot of rcu_qs_ctr to check */
					/*  for rcu_all_qs() invocations. */
326
	union rcu_noqs	cpu_no_qs;	/* No QSes yet for this CPU. */
327
	bool		core_needs_qs;	/* Core waits for quiesc state. */
328
	bool		beenonline;	/* CPU online at least once. */
329
	bool		gpwrap;		/* Possible gpnum/completed wrap. */
330 331
	struct rcu_node *mynode;	/* This CPU's leaf of hierarchy */
	unsigned long grpmask;		/* Mask to apply to leaf qsmask. */
332 333 334 335
	unsigned long	ticks_this_gp;	/* The number of scheduling-clock */
					/*  ticks this CPU has handled */
					/*  during and after the last grace */
					/* period it is aware of. */
336 337 338 339 340 341 342

	/* 2) batch handling */
	/*
	 * If nxtlist is not NULL, it is partitioned as follows.
	 * Any of the partitions might be empty, in which case the
	 * pointer to that partition will be equal to the pointer for
	 * the following partition.  When the list is empty, all of
343 344
	 * the nxttail elements point to the ->nxtlist pointer itself,
	 * which in that case is NULL.
345 346 347 348 349 350
	 *
	 * [nxtlist, *nxttail[RCU_DONE_TAIL]):
	 *	Entries that batch # <= ->completed
	 *	The grace period for these entries has completed, and
	 *	the other grace-period-completed entries may be moved
	 *	here temporarily in rcu_process_callbacks().
351 352 353 354 355 356 357 358
	 * [*nxttail[RCU_DONE_TAIL], *nxttail[RCU_WAIT_TAIL]):
	 *	Entries that batch # <= ->completed - 1: waiting for current GP
	 * [*nxttail[RCU_WAIT_TAIL], *nxttail[RCU_NEXT_READY_TAIL]):
	 *	Entries known to have arrived before current GP ended
	 * [*nxttail[RCU_NEXT_READY_TAIL], *nxttail[RCU_NEXT_TAIL]):
	 *	Entries that might have arrived after current GP ended
	 *	Note that the value of *nxttail[RCU_NEXT_TAIL] will
	 *	always be NULL, as this is the end of the list.
359 360 361
	 */
	struct rcu_head *nxtlist;
	struct rcu_head **nxttail[RCU_NEXT_SIZE];
362 363
	unsigned long	nxtcompleted[RCU_NEXT_SIZE];
					/* grace periods for sublists. */
364 365
	long		qlen_lazy;	/* # of lazy queued callbacks */
	long		qlen;		/* # of queued callbacks, incl lazy */
366 367
	long		qlen_last_fqs_check;
					/* qlen at last check for QS forcing */
368
	unsigned long	n_cbs_invoked;	/* count of RCU cbs invoked. */
369
	unsigned long	n_nocbs_invoked; /* count of no-CBs RCU cbs invoked. */
370 371
	unsigned long   n_cbs_orphaned; /* RCU cbs orphaned by dying CPU */
	unsigned long   n_cbs_adopted;  /* RCU cbs adopted from dying CPU */
372 373
	unsigned long	n_force_qs_snap;
					/* did other CPU force QS recently? */
374 375 376 377 378 379 380 381 382
	long		blimit;		/* Upper limit on a processed batch */

	/* 3) dynticks interface. */
	struct rcu_dynticks *dynticks;	/* Shared per-CPU dynticks state. */
	int dynticks_snap;		/* Per-GP tracking for dynticks. */

	/* 4) reasons this CPU needed to be kicked by force_quiescent_state */
	unsigned long dynticks_fqs;	/* Kicked due to dynticks idle. */
	unsigned long offline_fqs;	/* Kicked due to being offline. */
383 384 385
	unsigned long cond_resched_completed;
					/* Grace period that needs help */
					/*  from cond_resched(). */
386 387

	/* 5) __rcu_pending() statistics. */
388
	unsigned long n_rcu_pending;	/* rcu_pending() calls since boot. */
389
	unsigned long n_rp_core_needs_qs;
390
	unsigned long n_rp_report_qs;
391 392 393 394
	unsigned long n_rp_cb_ready;
	unsigned long n_rp_cpu_needs_gp;
	unsigned long n_rp_gp_completed;
	unsigned long n_rp_gp_started;
395
	unsigned long n_rp_nocb_defer_wakeup;
396
	unsigned long n_rp_need_nothing;
397

398
	/* 6) _rcu_barrier(), OOM callbacks, and expediting. */
399
	struct rcu_head barrier_head;
400 401 402
#ifdef CONFIG_RCU_FAST_NO_HZ
	struct rcu_head oom_head;
#endif /* #ifdef CONFIG_RCU_FAST_NO_HZ */
403
	atomic_long_t exp_workdone0;	/* # done by workqueue. */
404 405 406
	atomic_long_t exp_workdone1;	/* # done by others #1. */
	atomic_long_t exp_workdone2;	/* # done by others #2. */
	atomic_long_t exp_workdone3;	/* # done by others #3. */
407
	int exp_dynticks_snap;		/* Double-check need for IPI. */
408

409 410 411 412
	/* 7) Callback offloading. */
#ifdef CONFIG_RCU_NOCB_CPU
	struct rcu_head *nocb_head;	/* CBs waiting for kthread. */
	struct rcu_head **nocb_tail;
413 414
	atomic_long_t nocb_q_count;	/* # CBs waiting for nocb */
	atomic_long_t nocb_q_count_lazy; /*  invocation (all stages). */
415 416
	struct rcu_head *nocb_follower_head; /* CBs ready to invoke. */
	struct rcu_head **nocb_follower_tail;
417
	struct swait_queue_head nocb_wq; /* For nocb kthreads to sleep on. */
418
	struct task_struct *nocb_kthread;
419
	int nocb_defer_wakeup;		/* Defer wakeup of nocb_kthread. */
420 421 422 423 424

	/* The following fields are used by the leader, hence own cacheline. */
	struct rcu_head *nocb_gp_head ____cacheline_internodealigned_in_smp;
					/* CBs waiting for GP. */
	struct rcu_head **nocb_gp_tail;
425
	bool nocb_leader_sleep;		/* Is the nocb leader thread asleep? */
426 427 428 429 430 431
	struct rcu_data *nocb_next_follower;
					/* Next follower in wakeup chain. */

	/* The following fields are used by the follower, hence new cachline. */
	struct rcu_data *nocb_leader ____cacheline_internodealigned_in_smp;
					/* Leader CPU takes GP-end wakeups. */
432 433
#endif /* #ifdef CONFIG_RCU_NOCB_CPU */

434 435 436
	/* 8) RCU CPU stall data. */
	unsigned int softirq_snap;	/* Snapshot of softirq activity. */

437
	int cpu;
438
	struct rcu_state *rsp;
439 440
};

441 442 443 444 445
/* Values for nocb_defer_wakeup field in struct rcu_data. */
#define RCU_NOGP_WAKE_NOT	0
#define RCU_NOGP_WAKE		1
#define RCU_NOGP_WAKE_FORCE	2

446 447 448
#define RCU_JIFFIES_TILL_FORCE_QS (1 + (HZ > 250) + (HZ > 500))
					/* For jiffies_till_first_fqs and */
					/*  and jiffies_till_next_fqs. */
449

450 451 452 453 454 455 456
#define RCU_JIFFIES_FQS_DIV	256	/* Very large systems need more */
					/*  delay between bouts of */
					/*  quiescent-state forcing. */

#define RCU_STALL_RAT_DELAY	2	/* Allow other CPUs time to take */
					/*  at least one scheduling clock */
					/*  irq before ratting on them. */
457

458 459 460 461 462 463 464 465 466 467
#define rcu_wait(cond)							\
do {									\
	for (;;) {							\
		set_current_state(TASK_INTERRUPTIBLE);			\
		if (cond)						\
			break;						\
		schedule();						\
	}								\
	__set_current_state(TASK_RUNNING);				\
} while (0)
468 469 470 471 472 473 474 475 476 477 478 479 480

/*
 * RCU global state, including node hierarchy.  This hierarchy is
 * represented in "heap" form in a dense array.  The root (first level)
 * of the hierarchy is in ->node[0] (referenced by ->level[0]), the second
 * level in ->node[1] through ->node[m] (->node[1] referenced by ->level[1]),
 * and the third level in ->node[m+1] and following (->node[m+1] referenced
 * by ->level[2]).  The number of levels is determined by the number of
 * CPUs and by CONFIG_RCU_FANOUT.  Small systems will have a "hierarchy"
 * consisting of a single rcu_node.
 */
struct rcu_state {
	struct rcu_node node[NUM_RCU_NODES];	/* Hierarchy. */
481 482 483
	struct rcu_node *level[RCU_NUM_LVLS + 1];
						/* Hierarchy levels (+1 to */
						/*  shut bogus gcc warning) */
484
	struct rcu_data __percpu *rda;		/* pointer of percu rcu_data. */
485
	call_rcu_func_t call;			/* call_rcu() flavor. */
486
	int ncpus;				/* # CPUs seen so far. */
487 488 489

	/* The following fields are guarded by the root rcu_node's lock. */

490 491
	u8	boost ____cacheline_internodealigned_in_smp;
						/* Subject to priority boost. */
492 493
	unsigned long gpnum;			/* Current gp number. */
	unsigned long completed;		/* # of last completed gp. */
494
	struct task_struct *gp_kthread;		/* Task for grace periods. */
495
	struct swait_queue_head gp_wq;		/* Where GP task waits. */
496 497
	short gp_flags;				/* Commands for GP task. */
	short gp_state;				/* GP kthread sleep state. */
498

499
	/* End of fields guarded by root rcu_node's lock. */
500

501 502
	raw_spinlock_t orphan_lock ____cacheline_internodealigned_in_smp;
						/* Protect following fields. */
503 504 505 506 507 508 509 510
	struct rcu_head *orphan_nxtlist;	/* Orphaned callbacks that */
						/*  need a grace period. */
	struct rcu_head **orphan_nxttail;	/* Tail of above. */
	struct rcu_head *orphan_donelist;	/* Orphaned callbacks that */
						/*  are ready to invoke. */
	struct rcu_head **orphan_donetail;	/* Tail of above. */
	long qlen_lazy;				/* Number of lazy callbacks. */
	long qlen;				/* Total number of callbacks. */
511
	/* End of fields guarded by orphan_lock. */
512

513
	struct mutex barrier_mutex;		/* Guards barrier fields. */
514
	atomic_t barrier_cpu_count;		/* # CPUs waiting on. */
515
	struct completion barrier_completion;	/* Wake at barrier end. */
516
	unsigned long barrier_sequence;		/* ++ at start and end of */
517
						/*  _rcu_barrier(). */
518 519
	/* End of fields guarded by barrier_mutex. */

520
	struct mutex exp_mutex;			/* Serialize expedited GP. */
521
	struct mutex exp_wake_mutex;		/* Serialize wakeup. */
522
	unsigned long expedited_sequence;	/* Take a ticket. */
523
	atomic_t expedited_need_qs;		/* # CPUs left to check in. */
524
	struct swait_queue_head expedited_wq;	/* Wait for check-ins. */
525
	int ncpus_snap;				/* # CPUs seen last time. */
526

527 528
	unsigned long jiffies_force_qs;		/* Time at which to invoke */
						/*  force_quiescent_state(). */
529 530
	unsigned long jiffies_kick_kthreads;	/* Time at which to kick */
						/*  kthreads, if configured. */
531 532 533 534 535 536 537 538
	unsigned long n_force_qs;		/* Number of calls to */
						/*  force_quiescent_state(). */
	unsigned long n_force_qs_lh;		/* ~Number of calls leaving */
						/*  due to lock unavailable. */
	unsigned long n_force_qs_ngp;		/* Number of calls leaving */
						/*  due to no GP active. */
	unsigned long gp_start;			/* Time at which GP started, */
						/*  but in jiffies. */
539 540
	unsigned long gp_activity;		/* Time of last GP kthread */
						/*  activity in jiffies. */
541 542
	unsigned long jiffies_stall;		/* Time at which to check */
						/*  for CPU stalls. */
543 544
	unsigned long jiffies_resched;		/* Time at which to resched */
						/*  a reluctant CPU. */
545 546
	unsigned long n_force_qs_gpstart;	/* Snapshot of n_force_qs at */
						/*  GP start. */
547 548
	unsigned long gp_max;			/* Maximum GP duration in */
						/*  jiffies. */
549
	const char *name;			/* Name of structure. */
550
	char abbr;				/* Abbreviated name. */
551
	struct list_head flavors;		/* List of RCU flavors. */
552 553
};

554 555 556 557
/* Values for rcu_state structure's gp_flags field. */
#define RCU_GP_FLAG_INIT 0x1	/* Need grace-period initialization. */
#define RCU_GP_FLAG_FQS  0x2	/* Need grace-period quiescent-state forcing. */

558
/* Values for rcu_state structure's gp_state field. */
559
#define RCU_GP_IDLE	 0	/* Initial state and no GP in progress. */
560
#define RCU_GP_WAIT_GPS  1	/* Wait for grace-period start. */
561 562
#define RCU_GP_DONE_GPS  2	/* Wait done for grace-period start. */
#define RCU_GP_WAIT_FQS  3	/* Wait for force-quiescent-state time. */
563
#define RCU_GP_DOING_FQS 4	/* Wait done for force-quiescent-state time. */
564 565
#define RCU_GP_CLEANUP   5	/* Grace-period cleanup started. */
#define RCU_GP_CLEANED   6	/* Grace-period cleanup complete. */
566

567 568 569 570 571 572 573 574 575 576 577 578
#ifndef RCU_TREE_NONCORE
static const char * const gp_state_names[] = {
	"RCU_GP_IDLE",
	"RCU_GP_WAIT_GPS",
	"RCU_GP_DONE_GPS",
	"RCU_GP_WAIT_FQS",
	"RCU_GP_DOING_FQS",
	"RCU_GP_CLEANUP",
	"RCU_GP_CLEANED",
};
#endif /* #ifndef RCU_TREE_NONCORE */

579
extern struct list_head rcu_struct_flavors;
580 581

/* Sequence through rcu_state structures for each RCU flavor. */
582 583 584
#define for_each_rcu_flavor(rsp) \
	list_for_each_entry((rsp), &rcu_struct_flavors, flavors)

585 586 587
/*
 * RCU implementation internal declarations:
 */
588
extern struct rcu_state rcu_sched_state;
589 590 591

extern struct rcu_state rcu_bh_state;

592
#ifdef CONFIG_PREEMPT_RCU
593
extern struct rcu_state rcu_preempt_state;
594
#endif /* #ifdef CONFIG_PREEMPT_RCU */
595

596
int rcu_dynticks_snap(struct rcu_dynticks *rdtp);
597
bool rcu_eqs_special_set(int cpu);
598

599 600 601 602 603 604 605
#ifdef CONFIG_RCU_BOOST
DECLARE_PER_CPU(unsigned int, rcu_cpu_kthread_status);
DECLARE_PER_CPU(int, rcu_cpu_kthread_cpu);
DECLARE_PER_CPU(unsigned int, rcu_cpu_kthread_loops);
DECLARE_PER_CPU(char, rcu_cpu_has_work);
#endif /* #ifdef CONFIG_RCU_BOOST */

Paul E. McKenney's avatar
Paul E. McKenney committed
606
#ifndef RCU_TREE_NONCORE
607

608
/* Forward declarations for rcutree_plugin.h */
609
static void rcu_bootup_announce(void);
610
static void rcu_preempt_note_context_switch(void);
611
static int rcu_preempt_blocked_readers_cgp(struct rcu_node *rnp);
612
#ifdef CONFIG_HOTPLUG_CPU
613
static bool rcu_preempt_has_tasks(struct rcu_node *rnp);
614
#endif /* #ifdef CONFIG_HOTPLUG_CPU */
615
static void rcu_print_detail_task_stall(struct rcu_state *rsp);
616
static int rcu_print_task_stall(struct rcu_node *rnp);
617
static int rcu_print_task_exp_stall(struct rcu_node *rnp);
618
static void rcu_preempt_check_blocked_tasks(struct rcu_node *rnp);
619
static void rcu_preempt_check_callbacks(void);
620
void call_rcu(struct rcu_head *head, rcu_callback_t func);
621
static void __init __rcu_init_preempt(void);
622
static void rcu_initiate_boost(struct rcu_node *rnp, unsigned long flags);
623 624
static void rcu_preempt_boost_start_gp(struct rcu_node *rnp);
static void invoke_rcu_callbacks_kthread(void);
625
static bool rcu_is_callbacks_kthread(void);
626 627
#ifdef CONFIG_RCU_BOOST
static void rcu_preempt_do_callbacks(void);
628
static int rcu_spawn_one_boost_kthread(struct rcu_state *rsp,
Thomas Gleixner's avatar
Thomas Gleixner committed
629
						 struct rcu_node *rnp);
630
#endif /* #ifdef CONFIG_RCU_BOOST */
631
static void __init rcu_spawn_boost_kthreads(void);
632
static void rcu_prepare_kthreads(int cpu);
633
static void rcu_cleanup_after_idle(void);
634
static void rcu_prepare_for_idle(void);
635
static void rcu_idle_count_callbacks_posted(void);
636
static bool rcu_preempt_has_tasks(struct rcu_node *rnp);
637 638 639 640 641
static void print_cpu_stall_info_begin(void);
static void print_cpu_stall_info(struct rcu_state *rsp, int cpu);
static void print_cpu_stall_info_end(void);
static void zero_cpu_stall_ticks(struct rcu_data *rdp);
static void increment_cpu_stall_ticks(void);
642
static bool rcu_nocb_cpu_needs_barrier(struct rcu_state *rsp, int cpu);
643
static void rcu_nocb_gp_set(struct rcu_node *rnp, int nrq);
644 645
static struct swait_queue_head *rcu_nocb_gp_get(struct rcu_node *rnp);
static void rcu_nocb_gp_cleanup(struct swait_queue_head *sq);
646
static void rcu_init_one_nocb(struct rcu_node *rnp);
647
static bool __call_rcu_nocb(struct rcu_data *rdp, struct rcu_head *rhp,
648
			    bool lazy, unsigned long flags);
649
static bool rcu_nocb_adopt_orphan_cbs(struct rcu_state *rsp,
650 651
				      struct rcu_data *rdp,
				      unsigned long flags);
652
static int rcu_nocb_need_deferred_wakeup(struct rcu_data *rdp);
653
static void do_nocb_deferred_wakeup(struct rcu_data *rdp);
654
static void rcu_boot_init_nocb_percpu_data(struct rcu_data *rdp);
655 656 657 658 659
static void rcu_spawn_all_nocb_kthreads(int cpu);
static void __init rcu_spawn_nocb_kthreads(void);
#ifdef CONFIG_RCU_NOCB_CPU
static void __init rcu_organize_nocb_kthreads(struct rcu_state *rsp);
#endif /* #ifdef CONFIG_RCU_NOCB_CPU */
660
static void __maybe_unused rcu_kick_nohz_cpu(int cpu);
661
static bool init_nocb_callback_list(struct rcu_data *rdp);
662 663
static void rcu_sysidle_enter(int irq);
static void rcu_sysidle_exit(int irq);
664 665 666 667 668
static void rcu_sysidle_check_cpu(struct rcu_data *rdp, bool *isidle,
				  unsigned long *maxj);
static bool is_sysidle_rcu_state(struct rcu_state *rsp);
static void rcu_sysidle_report_gp(struct rcu_state *rsp, int isidle,
				  unsigned long maxj);
669
static void rcu_bind_gp_kthread(void);
670
static void rcu_sysidle_init_percpu_data(struct rcu_dynticks *rdtp);
671
static bool rcu_nohz_full_cpu(struct rcu_state *rsp);
672 673
static void rcu_dynticks_task_enter(void);
static void rcu_dynticks_task_exit(void);
674

Paul E. McKenney's avatar
Paul E. McKenney committed
675
#endif /* #ifndef RCU_TREE_NONCORE */
676 677

#ifdef CONFIG_RCU_TRACE
678
/* Read out queue lengths for tracing. */
679 680
static inline void rcu_nocb_q_lengths(struct rcu_data *rdp, long *ql, long *qll)
{
681 682 683
#ifdef CONFIG_RCU_NOCB_CPU
	*ql = atomic_long_read(&rdp->nocb_q_count);
	*qll = atomic_long_read(&rdp->nocb_q_count_lazy);
684 685 686 687
#else /* #ifdef CONFIG_RCU_NOCB_CPU */
	*ql = 0;
	*qll = 0;
#endif /* #else #ifdef CONFIG_RCU_NOCB_CPU */
688
}
689
#endif /* #ifdef CONFIG_RCU_TRACE */
690

691
/*
Boqun Feng's avatar
Boqun Feng committed
692
 * Wrappers for the rcu_node::lock acquire and release.
693 694 695 696 697 698 699 700
 *
 * 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().
Boqun Feng's avatar
Boqun Feng committed
701 702 703
 *
 * 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.
704 705 706
 */
static inline void raw_spin_lock_rcu_node(struct rcu_node *rnp)
{
Boqun Feng's avatar
Boqun Feng committed
707
	raw_spin_lock(&ACCESS_PRIVATE(rnp, lock));
708 709 710
	smp_mb__after_unlock_lock();
}

Boqun Feng's avatar
Boqun Feng committed
711 712 713 714 715
static inline void raw_spin_unlock_rcu_node(struct rcu_node *rnp)
{
	raw_spin_unlock(&ACCESS_PRIVATE(rnp, lock));
}

716 717
static inline void raw_spin_lock_irq_rcu_node(struct rcu_node *rnp)
{
Boqun Feng's avatar
Boqun Feng committed
718
	raw_spin_lock_irq(&ACCESS_PRIVATE(rnp, lock));
719 720 721
	smp_mb__after_unlock_lock();
}

Boqun Feng's avatar
Boqun Feng committed
722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737
static inline void raw_spin_unlock_irq_rcu_node(struct rcu_node *rnp)
{
	raw_spin_unlock_irq(&ACCESS_PRIVATE(rnp, lock));
}

#define raw_spin_lock_irqsave_rcu_node(rnp, flags)			\
do {									\
	typecheck(unsigned long, flags);				\
	raw_spin_lock_irqsave(&ACCESS_PRIVATE(rnp, lock), flags);	\
	smp_mb__after_unlock_lock();					\
} while (0)

#define raw_spin_unlock_irqrestore_rcu_node(rnp, flags)			\
do {									\
	typecheck(unsigned long, flags);				\
	raw_spin_unlock_irqrestore(&ACCESS_PRIVATE(rnp, lock), flags);	\
738 739 740 741
} while (0)

static inline bool raw_spin_trylock_rcu_node(struct rcu_node *rnp)
{
Boqun Feng's avatar
Boqun Feng committed
742
	bool locked = raw_spin_trylock(&ACCESS_PRIVATE(rnp, lock));
743 744 745 746 747

	if (locked)
		smp_mb__after_unlock_lock();
	return locked;
}