]>
Commit | Line | Data |
---|---|---|
64db4cff PM |
1 | /* |
2 | * Read-Copy Update mechanism for mutual exclusion | |
3 | * | |
4 | * This program is free software; you can redistribute it and/or modify | |
5 | * it under the terms of the GNU General Public License as published by | |
6 | * the Free Software Foundation; either version 2 of the License, or | |
7 | * (at your option) any later version. | |
8 | * | |
9 | * This program is distributed in the hope that it will be useful, | |
10 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
11 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
12 | * GNU General Public License for more details. | |
13 | * | |
14 | * You should have received a copy of the GNU General Public License | |
15 | * along with this program; if not, write to the Free Software | |
16 | * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. | |
17 | * | |
18 | * Copyright IBM Corporation, 2008 | |
19 | * | |
20 | * Authors: Dipankar Sarma <dipankar@in.ibm.com> | |
21 | * Manfred Spraul <manfred@colorfullife.com> | |
22 | * Paul E. McKenney <paulmck@linux.vnet.ibm.com> Hierarchical version | |
23 | * | |
24 | * Based on the original work by Paul McKenney <paulmck@us.ibm.com> | |
25 | * and inputs from Rusty Russell, Andrea Arcangeli and Andi Kleen. | |
26 | * | |
27 | * For detailed explanation of Read-Copy Update mechanism see - | |
28 | * Documentation/RCU | |
29 | */ | |
30 | #include <linux/types.h> | |
31 | #include <linux/kernel.h> | |
32 | #include <linux/init.h> | |
33 | #include <linux/spinlock.h> | |
34 | #include <linux/smp.h> | |
35 | #include <linux/rcupdate.h> | |
36 | #include <linux/interrupt.h> | |
37 | #include <linux/sched.h> | |
38 | #include <asm/atomic.h> | |
39 | #include <linux/bitops.h> | |
40 | #include <linux/module.h> | |
41 | #include <linux/completion.h> | |
42 | #include <linux/moduleparam.h> | |
43 | #include <linux/percpu.h> | |
44 | #include <linux/notifier.h> | |
45 | #include <linux/cpu.h> | |
46 | #include <linux/mutex.h> | |
47 | #include <linux/time.h> | |
48 | ||
49 | #ifdef CONFIG_DEBUG_LOCK_ALLOC | |
50 | static struct lock_class_key rcu_lock_key; | |
51 | struct lockdep_map rcu_lock_map = | |
52 | STATIC_LOCKDEP_MAP_INIT("rcu_read_lock", &rcu_lock_key); | |
53 | EXPORT_SYMBOL_GPL(rcu_lock_map); | |
54 | #endif | |
55 | ||
56 | /* Data structures. */ | |
57 | ||
58 | #define RCU_STATE_INITIALIZER(name) { \ | |
59 | .level = { &name.node[0] }, \ | |
60 | .levelcnt = { \ | |
61 | NUM_RCU_LVL_0, /* root of hierarchy. */ \ | |
62 | NUM_RCU_LVL_1, \ | |
63 | NUM_RCU_LVL_2, \ | |
64 | NUM_RCU_LVL_3, /* == MAX_RCU_LVLS */ \ | |
65 | }, \ | |
66 | .signaled = RCU_SIGNAL_INIT, \ | |
67 | .gpnum = -300, \ | |
68 | .completed = -300, \ | |
69 | .onofflock = __SPIN_LOCK_UNLOCKED(&name.onofflock), \ | |
70 | .fqslock = __SPIN_LOCK_UNLOCKED(&name.fqslock), \ | |
71 | .n_force_qs = 0, \ | |
72 | .n_force_qs_ngp = 0, \ | |
73 | } | |
74 | ||
75 | struct rcu_state rcu_state = RCU_STATE_INITIALIZER(rcu_state); | |
76 | DEFINE_PER_CPU(struct rcu_data, rcu_data); | |
77 | ||
78 | struct rcu_state rcu_bh_state = RCU_STATE_INITIALIZER(rcu_bh_state); | |
79 | DEFINE_PER_CPU(struct rcu_data, rcu_bh_data); | |
80 | ||
81 | #ifdef CONFIG_NO_HZ | |
82 | DEFINE_PER_CPU(struct rcu_dynticks, rcu_dynticks); | |
83 | #endif /* #ifdef CONFIG_NO_HZ */ | |
84 | ||
85 | static int blimit = 10; /* Maximum callbacks per softirq. */ | |
86 | static int qhimark = 10000; /* If this many pending, ignore blimit. */ | |
87 | static int qlowmark = 100; /* Once only this many pending, use blimit. */ | |
88 | ||
89 | static void force_quiescent_state(struct rcu_state *rsp, int relaxed); | |
90 | ||
91 | /* | |
92 | * Return the number of RCU batches processed thus far for debug & stats. | |
93 | */ | |
94 | long rcu_batches_completed(void) | |
95 | { | |
96 | return rcu_state.completed; | |
97 | } | |
98 | EXPORT_SYMBOL_GPL(rcu_batches_completed); | |
99 | ||
100 | /* | |
101 | * Return the number of RCU BH batches processed thus far for debug & stats. | |
102 | */ | |
103 | long rcu_batches_completed_bh(void) | |
104 | { | |
105 | return rcu_bh_state.completed; | |
106 | } | |
107 | EXPORT_SYMBOL_GPL(rcu_batches_completed_bh); | |
108 | ||
109 | /* | |
110 | * Does the CPU have callbacks ready to be invoked? | |
111 | */ | |
112 | static int | |
113 | cpu_has_callbacks_ready_to_invoke(struct rcu_data *rdp) | |
114 | { | |
115 | return &rdp->nxtlist != rdp->nxttail[RCU_DONE_TAIL]; | |
116 | } | |
117 | ||
118 | /* | |
119 | * Does the current CPU require a yet-as-unscheduled grace period? | |
120 | */ | |
121 | static int | |
122 | cpu_needs_another_gp(struct rcu_state *rsp, struct rcu_data *rdp) | |
123 | { | |
124 | /* ACCESS_ONCE() because we are accessing outside of lock. */ | |
125 | return *rdp->nxttail[RCU_DONE_TAIL] && | |
126 | ACCESS_ONCE(rsp->completed) == ACCESS_ONCE(rsp->gpnum); | |
127 | } | |
128 | ||
129 | /* | |
130 | * Return the root node of the specified rcu_state structure. | |
131 | */ | |
132 | static struct rcu_node *rcu_get_root(struct rcu_state *rsp) | |
133 | { | |
134 | return &rsp->node[0]; | |
135 | } | |
136 | ||
137 | #ifdef CONFIG_SMP | |
138 | ||
139 | /* | |
140 | * If the specified CPU is offline, tell the caller that it is in | |
141 | * a quiescent state. Otherwise, whack it with a reschedule IPI. | |
142 | * Grace periods can end up waiting on an offline CPU when that | |
143 | * CPU is in the process of coming online -- it will be added to the | |
144 | * rcu_node bitmasks before it actually makes it online. The same thing | |
145 | * can happen while a CPU is in the process of coming online. Because this | |
146 | * race is quite rare, we check for it after detecting that the grace | |
147 | * period has been delayed rather than checking each and every CPU | |
148 | * each and every time we start a new grace period. | |
149 | */ | |
150 | static int rcu_implicit_offline_qs(struct rcu_data *rdp) | |
151 | { | |
152 | /* | |
153 | * If the CPU is offline, it is in a quiescent state. We can | |
154 | * trust its state not to change because interrupts are disabled. | |
155 | */ | |
156 | if (cpu_is_offline(rdp->cpu)) { | |
157 | rdp->offline_fqs++; | |
158 | return 1; | |
159 | } | |
160 | ||
161 | /* The CPU is online, so send it a reschedule IPI. */ | |
162 | if (rdp->cpu != smp_processor_id()) | |
163 | smp_send_reschedule(rdp->cpu); | |
164 | else | |
165 | set_need_resched(); | |
166 | rdp->resched_ipi++; | |
167 | return 0; | |
168 | } | |
169 | ||
170 | #endif /* #ifdef CONFIG_SMP */ | |
171 | ||
172 | #ifdef CONFIG_NO_HZ | |
173 | static DEFINE_RATELIMIT_STATE(rcu_rs, 10 * HZ, 5); | |
174 | ||
175 | /** | |
176 | * rcu_enter_nohz - inform RCU that current CPU is entering nohz | |
177 | * | |
178 | * Enter nohz mode, in other words, -leave- the mode in which RCU | |
179 | * read-side critical sections can occur. (Though RCU read-side | |
180 | * critical sections can occur in irq handlers in nohz mode, a possibility | |
181 | * handled by rcu_irq_enter() and rcu_irq_exit()). | |
182 | */ | |
183 | void rcu_enter_nohz(void) | |
184 | { | |
185 | unsigned long flags; | |
186 | struct rcu_dynticks *rdtp; | |
187 | ||
188 | smp_mb(); /* CPUs seeing ++ must see prior RCU read-side crit sects */ | |
189 | local_irq_save(flags); | |
190 | rdtp = &__get_cpu_var(rcu_dynticks); | |
191 | rdtp->dynticks++; | |
192 | rdtp->dynticks_nesting--; | |
193 | WARN_ON_RATELIMIT(rdtp->dynticks & 0x1, &rcu_rs); | |
194 | local_irq_restore(flags); | |
195 | } | |
196 | ||
197 | /* | |
198 | * rcu_exit_nohz - inform RCU that current CPU is leaving nohz | |
199 | * | |
200 | * Exit nohz mode, in other words, -enter- the mode in which RCU | |
201 | * read-side critical sections normally occur. | |
202 | */ | |
203 | void rcu_exit_nohz(void) | |
204 | { | |
205 | unsigned long flags; | |
206 | struct rcu_dynticks *rdtp; | |
207 | ||
208 | local_irq_save(flags); | |
209 | rdtp = &__get_cpu_var(rcu_dynticks); | |
210 | rdtp->dynticks++; | |
211 | rdtp->dynticks_nesting++; | |
212 | WARN_ON_RATELIMIT(!(rdtp->dynticks & 0x1), &rcu_rs); | |
213 | local_irq_restore(flags); | |
214 | smp_mb(); /* CPUs seeing ++ must see later RCU read-side crit sects */ | |
215 | } | |
216 | ||
217 | /** | |
218 | * rcu_nmi_enter - inform RCU of entry to NMI context | |
219 | * | |
220 | * If the CPU was idle with dynamic ticks active, and there is no | |
221 | * irq handler running, this updates rdtp->dynticks_nmi to let the | |
222 | * RCU grace-period handling know that the CPU is active. | |
223 | */ | |
224 | void rcu_nmi_enter(void) | |
225 | { | |
226 | struct rcu_dynticks *rdtp = &__get_cpu_var(rcu_dynticks); | |
227 | ||
228 | if (rdtp->dynticks & 0x1) | |
229 | return; | |
230 | rdtp->dynticks_nmi++; | |
231 | WARN_ON_RATELIMIT(!(rdtp->dynticks_nmi & 0x1), &rcu_rs); | |
232 | smp_mb(); /* CPUs seeing ++ must see later RCU read-side crit sects */ | |
233 | } | |
234 | ||
235 | /** | |
236 | * rcu_nmi_exit - inform RCU of exit from NMI context | |
237 | * | |
238 | * If the CPU was idle with dynamic ticks active, and there is no | |
239 | * irq handler running, this updates rdtp->dynticks_nmi to let the | |
240 | * RCU grace-period handling know that the CPU is no longer active. | |
241 | */ | |
242 | void rcu_nmi_exit(void) | |
243 | { | |
244 | struct rcu_dynticks *rdtp = &__get_cpu_var(rcu_dynticks); | |
245 | ||
246 | if (rdtp->dynticks & 0x1) | |
247 | return; | |
248 | smp_mb(); /* CPUs seeing ++ must see prior RCU read-side crit sects */ | |
249 | rdtp->dynticks_nmi++; | |
250 | WARN_ON_RATELIMIT(rdtp->dynticks_nmi & 0x1, &rcu_rs); | |
251 | } | |
252 | ||
253 | /** | |
254 | * rcu_irq_enter - inform RCU of entry to hard irq context | |
255 | * | |
256 | * If the CPU was idle with dynamic ticks active, this updates the | |
257 | * rdtp->dynticks to let the RCU handling know that the CPU is active. | |
258 | */ | |
259 | void rcu_irq_enter(void) | |
260 | { | |
261 | struct rcu_dynticks *rdtp = &__get_cpu_var(rcu_dynticks); | |
262 | ||
263 | if (rdtp->dynticks_nesting++) | |
264 | return; | |
265 | rdtp->dynticks++; | |
266 | WARN_ON_RATELIMIT(!(rdtp->dynticks & 0x1), &rcu_rs); | |
267 | smp_mb(); /* CPUs seeing ++ must see later RCU read-side crit sects */ | |
268 | } | |
269 | ||
270 | /** | |
271 | * rcu_irq_exit - inform RCU of exit from hard irq context | |
272 | * | |
273 | * If the CPU was idle with dynamic ticks active, update the rdp->dynticks | |
274 | * to put let the RCU handling be aware that the CPU is going back to idle | |
275 | * with no ticks. | |
276 | */ | |
277 | void rcu_irq_exit(void) | |
278 | { | |
279 | struct rcu_dynticks *rdtp = &__get_cpu_var(rcu_dynticks); | |
280 | ||
281 | if (--rdtp->dynticks_nesting) | |
282 | return; | |
283 | smp_mb(); /* CPUs seeing ++ must see prior RCU read-side crit sects */ | |
284 | rdtp->dynticks++; | |
285 | WARN_ON_RATELIMIT(rdtp->dynticks & 0x1, &rcu_rs); | |
286 | ||
287 | /* If the interrupt queued a callback, get out of dyntick mode. */ | |
288 | if (__get_cpu_var(rcu_data).nxtlist || | |
289 | __get_cpu_var(rcu_bh_data).nxtlist) | |
290 | set_need_resched(); | |
291 | } | |
292 | ||
293 | /* | |
294 | * Record the specified "completed" value, which is later used to validate | |
295 | * dynticks counter manipulations. Specify "rsp->completed - 1" to | |
296 | * unconditionally invalidate any future dynticks manipulations (which is | |
297 | * useful at the beginning of a grace period). | |
298 | */ | |
299 | static void dyntick_record_completed(struct rcu_state *rsp, long comp) | |
300 | { | |
301 | rsp->dynticks_completed = comp; | |
302 | } | |
303 | ||
304 | #ifdef CONFIG_SMP | |
305 | ||
306 | /* | |
307 | * Recall the previously recorded value of the completion for dynticks. | |
308 | */ | |
309 | static long dyntick_recall_completed(struct rcu_state *rsp) | |
310 | { | |
311 | return rsp->dynticks_completed; | |
312 | } | |
313 | ||
314 | /* | |
315 | * Snapshot the specified CPU's dynticks counter so that we can later | |
316 | * credit them with an implicit quiescent state. Return 1 if this CPU | |
317 | * is already in a quiescent state courtesy of dynticks idle mode. | |
318 | */ | |
319 | static int dyntick_save_progress_counter(struct rcu_data *rdp) | |
320 | { | |
321 | int ret; | |
322 | int snap; | |
323 | int snap_nmi; | |
324 | ||
325 | snap = rdp->dynticks->dynticks; | |
326 | snap_nmi = rdp->dynticks->dynticks_nmi; | |
327 | smp_mb(); /* Order sampling of snap with end of grace period. */ | |
328 | rdp->dynticks_snap = snap; | |
329 | rdp->dynticks_nmi_snap = snap_nmi; | |
330 | ret = ((snap & 0x1) == 0) && ((snap_nmi & 0x1) == 0); | |
331 | if (ret) | |
332 | rdp->dynticks_fqs++; | |
333 | return ret; | |
334 | } | |
335 | ||
336 | /* | |
337 | * Return true if the specified CPU has passed through a quiescent | |
338 | * state by virtue of being in or having passed through an dynticks | |
339 | * idle state since the last call to dyntick_save_progress_counter() | |
340 | * for this same CPU. | |
341 | */ | |
342 | static int rcu_implicit_dynticks_qs(struct rcu_data *rdp) | |
343 | { | |
344 | long curr; | |
345 | long curr_nmi; | |
346 | long snap; | |
347 | long snap_nmi; | |
348 | ||
349 | curr = rdp->dynticks->dynticks; | |
350 | snap = rdp->dynticks_snap; | |
351 | curr_nmi = rdp->dynticks->dynticks_nmi; | |
352 | snap_nmi = rdp->dynticks_nmi_snap; | |
353 | smp_mb(); /* force ordering with cpu entering/leaving dynticks. */ | |
354 | ||
355 | /* | |
356 | * If the CPU passed through or entered a dynticks idle phase with | |
357 | * no active irq/NMI handlers, then we can safely pretend that the CPU | |
358 | * already acknowledged the request to pass through a quiescent | |
359 | * state. Either way, that CPU cannot possibly be in an RCU | |
360 | * read-side critical section that started before the beginning | |
361 | * of the current RCU grace period. | |
362 | */ | |
363 | if ((curr != snap || (curr & 0x1) == 0) && | |
364 | (curr_nmi != snap_nmi || (curr_nmi & 0x1) == 0)) { | |
365 | rdp->dynticks_fqs++; | |
366 | return 1; | |
367 | } | |
368 | ||
369 | /* Go check for the CPU being offline. */ | |
370 | return rcu_implicit_offline_qs(rdp); | |
371 | } | |
372 | ||
373 | #endif /* #ifdef CONFIG_SMP */ | |
374 | ||
375 | #else /* #ifdef CONFIG_NO_HZ */ | |
376 | ||
377 | static void dyntick_record_completed(struct rcu_state *rsp, long comp) | |
378 | { | |
379 | } | |
380 | ||
381 | #ifdef CONFIG_SMP | |
382 | ||
383 | /* | |
384 | * If there are no dynticks, then the only way that a CPU can passively | |
385 | * be in a quiescent state is to be offline. Unlike dynticks idle, which | |
386 | * is a point in time during the prior (already finished) grace period, | |
387 | * an offline CPU is always in a quiescent state, and thus can be | |
388 | * unconditionally applied. So just return the current value of completed. | |
389 | */ | |
390 | static long dyntick_recall_completed(struct rcu_state *rsp) | |
391 | { | |
392 | return rsp->completed; | |
393 | } | |
394 | ||
395 | static int dyntick_save_progress_counter(struct rcu_data *rdp) | |
396 | { | |
397 | return 0; | |
398 | } | |
399 | ||
400 | static int rcu_implicit_dynticks_qs(struct rcu_data *rdp) | |
401 | { | |
402 | return rcu_implicit_offline_qs(rdp); | |
403 | } | |
404 | ||
405 | #endif /* #ifdef CONFIG_SMP */ | |
406 | ||
407 | #endif /* #else #ifdef CONFIG_NO_HZ */ | |
408 | ||
409 | #ifdef CONFIG_RCU_CPU_STALL_DETECTOR | |
410 | ||
411 | static void record_gp_stall_check_time(struct rcu_state *rsp) | |
412 | { | |
413 | rsp->gp_start = jiffies; | |
414 | rsp->jiffies_stall = jiffies + RCU_SECONDS_TILL_STALL_CHECK; | |
415 | } | |
416 | ||
417 | static void print_other_cpu_stall(struct rcu_state *rsp) | |
418 | { | |
419 | int cpu; | |
420 | long delta; | |
421 | unsigned long flags; | |
422 | struct rcu_node *rnp = rcu_get_root(rsp); | |
423 | struct rcu_node *rnp_cur = rsp->level[NUM_RCU_LVLS - 1]; | |
424 | struct rcu_node *rnp_end = &rsp->node[NUM_RCU_NODES]; | |
425 | ||
426 | /* Only let one CPU complain about others per time interval. */ | |
427 | ||
428 | spin_lock_irqsave(&rnp->lock, flags); | |
429 | delta = jiffies - rsp->jiffies_stall; | |
430 | if (delta < RCU_STALL_RAT_DELAY || rsp->gpnum == rsp->completed) { | |
431 | spin_unlock_irqrestore(&rnp->lock, flags); | |
432 | return; | |
433 | } | |
434 | rsp->jiffies_stall = jiffies + RCU_SECONDS_TILL_STALL_RECHECK; | |
435 | spin_unlock_irqrestore(&rnp->lock, flags); | |
436 | ||
437 | /* OK, time to rat on our buddy... */ | |
438 | ||
439 | printk(KERN_ERR "INFO: RCU detected CPU stalls:"); | |
440 | for (; rnp_cur < rnp_end; rnp_cur++) { | |
441 | if (rnp_cur->qsmask == 0) | |
442 | continue; | |
443 | for (cpu = 0; cpu <= rnp_cur->grphi - rnp_cur->grplo; cpu++) | |
444 | if (rnp_cur->qsmask & (1UL << cpu)) | |
445 | printk(" %d", rnp_cur->grplo + cpu); | |
446 | } | |
447 | printk(" (detected by %d, t=%ld jiffies)\n", | |
448 | smp_processor_id(), (long)(jiffies - rsp->gp_start)); | |
449 | force_quiescent_state(rsp, 0); /* Kick them all. */ | |
450 | } | |
451 | ||
452 | static void print_cpu_stall(struct rcu_state *rsp) | |
453 | { | |
454 | unsigned long flags; | |
455 | struct rcu_node *rnp = rcu_get_root(rsp); | |
456 | ||
457 | printk(KERN_ERR "INFO: RCU detected CPU %d stall (t=%lu jiffies)\n", | |
458 | smp_processor_id(), jiffies - rsp->gp_start); | |
459 | dump_stack(); | |
460 | spin_lock_irqsave(&rnp->lock, flags); | |
461 | if ((long)(jiffies - rsp->jiffies_stall) >= 0) | |
462 | rsp->jiffies_stall = | |
463 | jiffies + RCU_SECONDS_TILL_STALL_RECHECK; | |
464 | spin_unlock_irqrestore(&rnp->lock, flags); | |
465 | set_need_resched(); /* kick ourselves to get things going. */ | |
466 | } | |
467 | ||
468 | static void check_cpu_stall(struct rcu_state *rsp, struct rcu_data *rdp) | |
469 | { | |
470 | long delta; | |
471 | struct rcu_node *rnp; | |
472 | ||
473 | delta = jiffies - rsp->jiffies_stall; | |
474 | rnp = rdp->mynode; | |
475 | if ((rnp->qsmask & rdp->grpmask) && delta >= 0) { | |
476 | ||
477 | /* We haven't checked in, so go dump stack. */ | |
478 | print_cpu_stall(rsp); | |
479 | ||
480 | } else if (rsp->gpnum != rsp->completed && | |
481 | delta >= RCU_STALL_RAT_DELAY) { | |
482 | ||
483 | /* They had two time units to dump stack, so complain. */ | |
484 | print_other_cpu_stall(rsp); | |
485 | } | |
486 | } | |
487 | ||
488 | #else /* #ifdef CONFIG_RCU_CPU_STALL_DETECTOR */ | |
489 | ||
490 | static void record_gp_stall_check_time(struct rcu_state *rsp) | |
491 | { | |
492 | } | |
493 | ||
494 | static void check_cpu_stall(struct rcu_state *rsp, struct rcu_data *rdp) | |
495 | { | |
496 | } | |
497 | ||
498 | #endif /* #else #ifdef CONFIG_RCU_CPU_STALL_DETECTOR */ | |
499 | ||
500 | /* | |
501 | * Update CPU-local rcu_data state to record the newly noticed grace period. | |
502 | * This is used both when we started the grace period and when we notice | |
503 | * that someone else started the grace period. | |
504 | */ | |
505 | static void note_new_gpnum(struct rcu_state *rsp, struct rcu_data *rdp) | |
506 | { | |
507 | rdp->qs_pending = 1; | |
508 | rdp->passed_quiesc = 0; | |
509 | rdp->gpnum = rsp->gpnum; | |
510 | rdp->n_rcu_pending_force_qs = rdp->n_rcu_pending + | |
511 | RCU_JIFFIES_TILL_FORCE_QS; | |
512 | } | |
513 | ||
514 | /* | |
515 | * Did someone else start a new RCU grace period start since we last | |
516 | * checked? Update local state appropriately if so. Must be called | |
517 | * on the CPU corresponding to rdp. | |
518 | */ | |
519 | static int | |
520 | check_for_new_grace_period(struct rcu_state *rsp, struct rcu_data *rdp) | |
521 | { | |
522 | unsigned long flags; | |
523 | int ret = 0; | |
524 | ||
525 | local_irq_save(flags); | |
526 | if (rdp->gpnum != rsp->gpnum) { | |
527 | note_new_gpnum(rsp, rdp); | |
528 | ret = 1; | |
529 | } | |
530 | local_irq_restore(flags); | |
531 | return ret; | |
532 | } | |
533 | ||
534 | /* | |
535 | * Start a new RCU grace period if warranted, re-initializing the hierarchy | |
536 | * in preparation for detecting the next grace period. The caller must hold | |
537 | * the root node's ->lock, which is released before return. Hard irqs must | |
538 | * be disabled. | |
539 | */ | |
540 | static void | |
541 | rcu_start_gp(struct rcu_state *rsp, unsigned long flags) | |
542 | __releases(rcu_get_root(rsp)->lock) | |
543 | { | |
544 | struct rcu_data *rdp = rsp->rda[smp_processor_id()]; | |
545 | struct rcu_node *rnp = rcu_get_root(rsp); | |
546 | struct rcu_node *rnp_cur; | |
547 | struct rcu_node *rnp_end; | |
548 | ||
549 | if (!cpu_needs_another_gp(rsp, rdp)) { | |
550 | spin_unlock_irqrestore(&rnp->lock, flags); | |
551 | return; | |
552 | } | |
553 | ||
554 | /* Advance to a new grace period and initialize state. */ | |
555 | rsp->gpnum++; | |
556 | rsp->signaled = RCU_GP_INIT; /* Hold off force_quiescent_state. */ | |
557 | rsp->jiffies_force_qs = jiffies + RCU_JIFFIES_TILL_FORCE_QS; | |
558 | rdp->n_rcu_pending_force_qs = rdp->n_rcu_pending + | |
559 | RCU_JIFFIES_TILL_FORCE_QS; | |
560 | record_gp_stall_check_time(rsp); | |
561 | dyntick_record_completed(rsp, rsp->completed - 1); | |
562 | note_new_gpnum(rsp, rdp); | |
563 | ||
564 | /* | |
565 | * Because we are first, we know that all our callbacks will | |
566 | * be covered by this upcoming grace period, even the ones | |
567 | * that were registered arbitrarily recently. | |
568 | */ | |
569 | rdp->nxttail[RCU_NEXT_READY_TAIL] = rdp->nxttail[RCU_NEXT_TAIL]; | |
570 | rdp->nxttail[RCU_WAIT_TAIL] = rdp->nxttail[RCU_NEXT_TAIL]; | |
571 | ||
572 | /* Special-case the common single-level case. */ | |
573 | if (NUM_RCU_NODES == 1) { | |
574 | rnp->qsmask = rnp->qsmaskinit; | |
c12172c0 | 575 | rsp->signaled = RCU_SIGNAL_INIT; /* force_quiescent_state OK. */ |
64db4cff PM |
576 | spin_unlock_irqrestore(&rnp->lock, flags); |
577 | return; | |
578 | } | |
579 | ||
580 | spin_unlock(&rnp->lock); /* leave irqs disabled. */ | |
581 | ||
582 | ||
583 | /* Exclude any concurrent CPU-hotplug operations. */ | |
584 | spin_lock(&rsp->onofflock); /* irqs already disabled. */ | |
585 | ||
586 | /* | |
587 | * Set the quiescent-state-needed bits in all the non-leaf RCU | |
588 | * nodes for all currently online CPUs. This operation relies | |
589 | * on the layout of the hierarchy within the rsp->node[] array. | |
590 | * Note that other CPUs will access only the leaves of the | |
591 | * hierarchy, which still indicate that no grace period is in | |
592 | * progress. In addition, we have excluded CPU-hotplug operations. | |
593 | * | |
594 | * We therefore do not need to hold any locks. Any required | |
595 | * memory barriers will be supplied by the locks guarding the | |
596 | * leaf rcu_nodes in the hierarchy. | |
597 | */ | |
598 | ||
599 | rnp_end = rsp->level[NUM_RCU_LVLS - 1]; | |
600 | for (rnp_cur = &rsp->node[0]; rnp_cur < rnp_end; rnp_cur++) | |
601 | rnp_cur->qsmask = rnp_cur->qsmaskinit; | |
602 | ||
603 | /* | |
604 | * Now set up the leaf nodes. Here we must be careful. First, | |
605 | * we need to hold the lock in order to exclude other CPUs, which | |
606 | * might be contending for the leaf nodes' locks. Second, as | |
607 | * soon as we initialize a given leaf node, its CPUs might run | |
608 | * up the rest of the hierarchy. We must therefore acquire locks | |
609 | * for each node that we touch during this stage. (But we still | |
610 | * are excluding CPU-hotplug operations.) | |
611 | * | |
612 | * Note that the grace period cannot complete until we finish | |
613 | * the initialization process, as there will be at least one | |
614 | * qsmask bit set in the root node until that time, namely the | |
615 | * one corresponding to this CPU. | |
616 | */ | |
617 | rnp_end = &rsp->node[NUM_RCU_NODES]; | |
618 | rnp_cur = rsp->level[NUM_RCU_LVLS - 1]; | |
619 | for (; rnp_cur < rnp_end; rnp_cur++) { | |
620 | spin_lock(&rnp_cur->lock); /* irqs already disabled. */ | |
621 | rnp_cur->qsmask = rnp_cur->qsmaskinit; | |
622 | spin_unlock(&rnp_cur->lock); /* irqs already disabled. */ | |
623 | } | |
624 | ||
625 | rsp->signaled = RCU_SIGNAL_INIT; /* force_quiescent_state now OK. */ | |
626 | spin_unlock_irqrestore(&rsp->onofflock, flags); | |
627 | } | |
628 | ||
629 | /* | |
630 | * Advance this CPU's callbacks, but only if the current grace period | |
631 | * has ended. This may be called only from the CPU to whom the rdp | |
632 | * belongs. | |
633 | */ | |
634 | static void | |
635 | rcu_process_gp_end(struct rcu_state *rsp, struct rcu_data *rdp) | |
636 | { | |
637 | long completed_snap; | |
638 | unsigned long flags; | |
639 | ||
640 | local_irq_save(flags); | |
641 | completed_snap = ACCESS_ONCE(rsp->completed); /* outside of lock. */ | |
642 | ||
643 | /* Did another grace period end? */ | |
644 | if (rdp->completed != completed_snap) { | |
645 | ||
646 | /* Advance callbacks. No harm if list empty. */ | |
647 | rdp->nxttail[RCU_DONE_TAIL] = rdp->nxttail[RCU_WAIT_TAIL]; | |
648 | rdp->nxttail[RCU_WAIT_TAIL] = rdp->nxttail[RCU_NEXT_READY_TAIL]; | |
649 | rdp->nxttail[RCU_NEXT_READY_TAIL] = rdp->nxttail[RCU_NEXT_TAIL]; | |
650 | ||
651 | /* Remember that we saw this grace-period completion. */ | |
652 | rdp->completed = completed_snap; | |
653 | } | |
654 | local_irq_restore(flags); | |
655 | } | |
656 | ||
657 | /* | |
658 | * Similar to cpu_quiet(), for which it is a helper function. Allows | |
659 | * a group of CPUs to be quieted at one go, though all the CPUs in the | |
660 | * group must be represented by the same leaf rcu_node structure. | |
661 | * That structure's lock must be held upon entry, and it is released | |
662 | * before return. | |
663 | */ | |
664 | static void | |
665 | cpu_quiet_msk(unsigned long mask, struct rcu_state *rsp, struct rcu_node *rnp, | |
666 | unsigned long flags) | |
667 | __releases(rnp->lock) | |
668 | { | |
669 | /* Walk up the rcu_node hierarchy. */ | |
670 | for (;;) { | |
671 | if (!(rnp->qsmask & mask)) { | |
672 | ||
673 | /* Our bit has already been cleared, so done. */ | |
674 | spin_unlock_irqrestore(&rnp->lock, flags); | |
675 | return; | |
676 | } | |
677 | rnp->qsmask &= ~mask; | |
678 | if (rnp->qsmask != 0) { | |
679 | ||
680 | /* Other bits still set at this level, so done. */ | |
681 | spin_unlock_irqrestore(&rnp->lock, flags); | |
682 | return; | |
683 | } | |
684 | mask = rnp->grpmask; | |
685 | if (rnp->parent == NULL) { | |
686 | ||
687 | /* No more levels. Exit loop holding root lock. */ | |
688 | ||
689 | break; | |
690 | } | |
691 | spin_unlock_irqrestore(&rnp->lock, flags); | |
692 | rnp = rnp->parent; | |
693 | spin_lock_irqsave(&rnp->lock, flags); | |
694 | } | |
695 | ||
696 | /* | |
697 | * Get here if we are the last CPU to pass through a quiescent | |
698 | * state for this grace period. Clean up and let rcu_start_gp() | |
699 | * start up the next grace period if one is needed. Note that | |
700 | * we still hold rnp->lock, as required by rcu_start_gp(), which | |
701 | * will release it. | |
702 | */ | |
703 | rsp->completed = rsp->gpnum; | |
704 | rcu_process_gp_end(rsp, rsp->rda[smp_processor_id()]); | |
705 | rcu_start_gp(rsp, flags); /* releases rnp->lock. */ | |
706 | } | |
707 | ||
708 | /* | |
709 | * Record a quiescent state for the specified CPU, which must either be | |
710 | * the current CPU or an offline CPU. The lastcomp argument is used to | |
711 | * make sure we are still in the grace period of interest. We don't want | |
712 | * to end the current grace period based on quiescent states detected in | |
713 | * an earlier grace period! | |
714 | */ | |
715 | static void | |
716 | cpu_quiet(int cpu, struct rcu_state *rsp, struct rcu_data *rdp, long lastcomp) | |
717 | { | |
718 | unsigned long flags; | |
719 | unsigned long mask; | |
720 | struct rcu_node *rnp; | |
721 | ||
722 | rnp = rdp->mynode; | |
723 | spin_lock_irqsave(&rnp->lock, flags); | |
724 | if (lastcomp != ACCESS_ONCE(rsp->completed)) { | |
725 | ||
726 | /* | |
727 | * Someone beat us to it for this grace period, so leave. | |
728 | * The race with GP start is resolved by the fact that we | |
729 | * hold the leaf rcu_node lock, so that the per-CPU bits | |
730 | * cannot yet be initialized -- so we would simply find our | |
731 | * CPU's bit already cleared in cpu_quiet_msk() if this race | |
732 | * occurred. | |
733 | */ | |
734 | rdp->passed_quiesc = 0; /* try again later! */ | |
735 | spin_unlock_irqrestore(&rnp->lock, flags); | |
736 | return; | |
737 | } | |
738 | mask = rdp->grpmask; | |
739 | if ((rnp->qsmask & mask) == 0) { | |
740 | spin_unlock_irqrestore(&rnp->lock, flags); | |
741 | } else { | |
742 | rdp->qs_pending = 0; | |
743 | ||
744 | /* | |
745 | * This GP can't end until cpu checks in, so all of our | |
746 | * callbacks can be processed during the next GP. | |
747 | */ | |
748 | rdp = rsp->rda[smp_processor_id()]; | |
749 | rdp->nxttail[RCU_NEXT_READY_TAIL] = rdp->nxttail[RCU_NEXT_TAIL]; | |
750 | ||
751 | cpu_quiet_msk(mask, rsp, rnp, flags); /* releases rnp->lock */ | |
752 | } | |
753 | } | |
754 | ||
755 | /* | |
756 | * Check to see if there is a new grace period of which this CPU | |
757 | * is not yet aware, and if so, set up local rcu_data state for it. | |
758 | * Otherwise, see if this CPU has just passed through its first | |
759 | * quiescent state for this grace period, and record that fact if so. | |
760 | */ | |
761 | static void | |
762 | rcu_check_quiescent_state(struct rcu_state *rsp, struct rcu_data *rdp) | |
763 | { | |
764 | /* If there is now a new grace period, record and return. */ | |
765 | if (check_for_new_grace_period(rsp, rdp)) | |
766 | return; | |
767 | ||
768 | /* | |
769 | * Does this CPU still need to do its part for current grace period? | |
770 | * If no, return and let the other CPUs do their part as well. | |
771 | */ | |
772 | if (!rdp->qs_pending) | |
773 | return; | |
774 | ||
775 | /* | |
776 | * Was there a quiescent state since the beginning of the grace | |
777 | * period? If no, then exit and wait for the next call. | |
778 | */ | |
779 | if (!rdp->passed_quiesc) | |
780 | return; | |
781 | ||
782 | /* Tell RCU we are done (but cpu_quiet() will be the judge of that). */ | |
783 | cpu_quiet(rdp->cpu, rsp, rdp, rdp->passed_quiesc_completed); | |
784 | } | |
785 | ||
786 | #ifdef CONFIG_HOTPLUG_CPU | |
787 | ||
788 | /* | |
789 | * Remove the outgoing CPU from the bitmasks in the rcu_node hierarchy | |
790 | * and move all callbacks from the outgoing CPU to the current one. | |
791 | */ | |
792 | static void __rcu_offline_cpu(int cpu, struct rcu_state *rsp) | |
793 | { | |
794 | int i; | |
795 | unsigned long flags; | |
796 | long lastcomp; | |
797 | unsigned long mask; | |
798 | struct rcu_data *rdp = rsp->rda[cpu]; | |
799 | struct rcu_data *rdp_me; | |
800 | struct rcu_node *rnp; | |
801 | ||
802 | /* Exclude any attempts to start a new grace period. */ | |
803 | spin_lock_irqsave(&rsp->onofflock, flags); | |
804 | ||
805 | /* Remove the outgoing CPU from the masks in the rcu_node hierarchy. */ | |
806 | rnp = rdp->mynode; | |
807 | mask = rdp->grpmask; /* rnp->grplo is constant. */ | |
808 | do { | |
809 | spin_lock(&rnp->lock); /* irqs already disabled. */ | |
810 | rnp->qsmaskinit &= ~mask; | |
811 | if (rnp->qsmaskinit != 0) { | |
812 | spin_unlock(&rnp->lock); /* irqs already disabled. */ | |
813 | break; | |
814 | } | |
815 | mask = rnp->grpmask; | |
816 | spin_unlock(&rnp->lock); /* irqs already disabled. */ | |
817 | rnp = rnp->parent; | |
818 | } while (rnp != NULL); | |
819 | lastcomp = rsp->completed; | |
820 | ||
821 | spin_unlock(&rsp->onofflock); /* irqs remain disabled. */ | |
822 | ||
823 | /* Being offline is a quiescent state, so go record it. */ | |
824 | cpu_quiet(cpu, rsp, rdp, lastcomp); | |
825 | ||
826 | /* | |
827 | * Move callbacks from the outgoing CPU to the running CPU. | |
828 | * Note that the outgoing CPU is now quiscent, so it is now | |
829 | * (uncharacteristically) safe to access it rcu_data structure. | |
830 | * Note also that we must carefully retain the order of the | |
831 | * outgoing CPU's callbacks in order for rcu_barrier() to work | |
832 | * correctly. Finally, note that we start all the callbacks | |
833 | * afresh, even those that have passed through a grace period | |
834 | * and are therefore ready to invoke. The theory is that hotplug | |
835 | * events are rare, and that if they are frequent enough to | |
836 | * indefinitely delay callbacks, you have far worse things to | |
837 | * be worrying about. | |
838 | */ | |
839 | rdp_me = rsp->rda[smp_processor_id()]; | |
840 | if (rdp->nxtlist != NULL) { | |
841 | *rdp_me->nxttail[RCU_NEXT_TAIL] = rdp->nxtlist; | |
842 | rdp_me->nxttail[RCU_NEXT_TAIL] = rdp->nxttail[RCU_NEXT_TAIL]; | |
843 | rdp->nxtlist = NULL; | |
844 | for (i = 0; i < RCU_NEXT_SIZE; i++) | |
845 | rdp->nxttail[i] = &rdp->nxtlist; | |
846 | rdp_me->qlen += rdp->qlen; | |
847 | rdp->qlen = 0; | |
848 | } | |
849 | local_irq_restore(flags); | |
850 | } | |
851 | ||
852 | /* | |
853 | * Remove the specified CPU from the RCU hierarchy and move any pending | |
854 | * callbacks that it might have to the current CPU. This code assumes | |
855 | * that at least one CPU in the system will remain running at all times. | |
856 | * Any attempt to offline -all- CPUs is likely to strand RCU callbacks. | |
857 | */ | |
858 | static void rcu_offline_cpu(int cpu) | |
859 | { | |
860 | __rcu_offline_cpu(cpu, &rcu_state); | |
861 | __rcu_offline_cpu(cpu, &rcu_bh_state); | |
862 | } | |
863 | ||
864 | #else /* #ifdef CONFIG_HOTPLUG_CPU */ | |
865 | ||
866 | static void rcu_offline_cpu(int cpu) | |
867 | { | |
868 | } | |
869 | ||
870 | #endif /* #else #ifdef CONFIG_HOTPLUG_CPU */ | |
871 | ||
872 | /* | |
873 | * Invoke any RCU callbacks that have made it to the end of their grace | |
874 | * period. Thottle as specified by rdp->blimit. | |
875 | */ | |
876 | static void rcu_do_batch(struct rcu_data *rdp) | |
877 | { | |
878 | unsigned long flags; | |
879 | struct rcu_head *next, *list, **tail; | |
880 | int count; | |
881 | ||
882 | /* If no callbacks are ready, just return.*/ | |
883 | if (!cpu_has_callbacks_ready_to_invoke(rdp)) | |
884 | return; | |
885 | ||
886 | /* | |
887 | * Extract the list of ready callbacks, disabling to prevent | |
888 | * races with call_rcu() from interrupt handlers. | |
889 | */ | |
890 | local_irq_save(flags); | |
891 | list = rdp->nxtlist; | |
892 | rdp->nxtlist = *rdp->nxttail[RCU_DONE_TAIL]; | |
893 | *rdp->nxttail[RCU_DONE_TAIL] = NULL; | |
894 | tail = rdp->nxttail[RCU_DONE_TAIL]; | |
895 | for (count = RCU_NEXT_SIZE - 1; count >= 0; count--) | |
896 | if (rdp->nxttail[count] == rdp->nxttail[RCU_DONE_TAIL]) | |
897 | rdp->nxttail[count] = &rdp->nxtlist; | |
898 | local_irq_restore(flags); | |
899 | ||
900 | /* Invoke callbacks. */ | |
901 | count = 0; | |
902 | while (list) { | |
903 | next = list->next; | |
904 | prefetch(next); | |
905 | list->func(list); | |
906 | list = next; | |
907 | if (++count >= rdp->blimit) | |
908 | break; | |
909 | } | |
910 | ||
911 | local_irq_save(flags); | |
912 | ||
913 | /* Update count, and requeue any remaining callbacks. */ | |
914 | rdp->qlen -= count; | |
915 | if (list != NULL) { | |
916 | *tail = rdp->nxtlist; | |
917 | rdp->nxtlist = list; | |
918 | for (count = 0; count < RCU_NEXT_SIZE; count++) | |
919 | if (&rdp->nxtlist == rdp->nxttail[count]) | |
920 | rdp->nxttail[count] = tail; | |
921 | else | |
922 | break; | |
923 | } | |
924 | ||
925 | /* Reinstate batch limit if we have worked down the excess. */ | |
926 | if (rdp->blimit == LONG_MAX && rdp->qlen <= qlowmark) | |
927 | rdp->blimit = blimit; | |
928 | ||
929 | local_irq_restore(flags); | |
930 | ||
931 | /* Re-raise the RCU softirq if there are callbacks remaining. */ | |
932 | if (cpu_has_callbacks_ready_to_invoke(rdp)) | |
933 | raise_softirq(RCU_SOFTIRQ); | |
934 | } | |
935 | ||
936 | /* | |
937 | * Check to see if this CPU is in a non-context-switch quiescent state | |
938 | * (user mode or idle loop for rcu, non-softirq execution for rcu_bh). | |
939 | * Also schedule the RCU softirq handler. | |
940 | * | |
941 | * This function must be called with hardirqs disabled. It is normally | |
942 | * invoked from the scheduling-clock interrupt. If rcu_pending returns | |
943 | * false, there is no point in invoking rcu_check_callbacks(). | |
944 | */ | |
945 | void rcu_check_callbacks(int cpu, int user) | |
946 | { | |
947 | if (user || | |
948 | (idle_cpu(cpu) && !in_softirq() && | |
949 | hardirq_count() <= (1 << HARDIRQ_SHIFT))) { | |
950 | ||
951 | /* | |
952 | * Get here if this CPU took its interrupt from user | |
953 | * mode or from the idle loop, and if this is not a | |
954 | * nested interrupt. In this case, the CPU is in | |
955 | * a quiescent state, so count it. | |
956 | * | |
957 | * No memory barrier is required here because both | |
958 | * rcu_qsctr_inc() and rcu_bh_qsctr_inc() reference | |
959 | * only CPU-local variables that other CPUs neither | |
960 | * access nor modify, at least not while the corresponding | |
961 | * CPU is online. | |
962 | */ | |
963 | ||
964 | rcu_qsctr_inc(cpu); | |
965 | rcu_bh_qsctr_inc(cpu); | |
966 | ||
967 | } else if (!in_softirq()) { | |
968 | ||
969 | /* | |
970 | * Get here if this CPU did not take its interrupt from | |
971 | * softirq, in other words, if it is not interrupting | |
972 | * a rcu_bh read-side critical section. This is an _bh | |
973 | * critical section, so count it. | |
974 | */ | |
975 | ||
976 | rcu_bh_qsctr_inc(cpu); | |
977 | } | |
978 | raise_softirq(RCU_SOFTIRQ); | |
979 | } | |
980 | ||
981 | #ifdef CONFIG_SMP | |
982 | ||
983 | /* | |
984 | * Scan the leaf rcu_node structures, processing dyntick state for any that | |
985 | * have not yet encountered a quiescent state, using the function specified. | |
986 | * Returns 1 if the current grace period ends while scanning (possibly | |
987 | * because we made it end). | |
988 | */ | |
989 | static int rcu_process_dyntick(struct rcu_state *rsp, long lastcomp, | |
990 | int (*f)(struct rcu_data *)) | |
991 | { | |
992 | unsigned long bit; | |
993 | int cpu; | |
994 | unsigned long flags; | |
995 | unsigned long mask; | |
996 | struct rcu_node *rnp_cur = rsp->level[NUM_RCU_LVLS - 1]; | |
997 | struct rcu_node *rnp_end = &rsp->node[NUM_RCU_NODES]; | |
998 | ||
999 | for (; rnp_cur < rnp_end; rnp_cur++) { | |
1000 | mask = 0; | |
1001 | spin_lock_irqsave(&rnp_cur->lock, flags); | |
1002 | if (rsp->completed != lastcomp) { | |
1003 | spin_unlock_irqrestore(&rnp_cur->lock, flags); | |
1004 | return 1; | |
1005 | } | |
1006 | if (rnp_cur->qsmask == 0) { | |
1007 | spin_unlock_irqrestore(&rnp_cur->lock, flags); | |
1008 | continue; | |
1009 | } | |
1010 | cpu = rnp_cur->grplo; | |
1011 | bit = 1; | |
1012 | for (; cpu <= rnp_cur->grphi; cpu++, bit <<= 1) { | |
1013 | if ((rnp_cur->qsmask & bit) != 0 && f(rsp->rda[cpu])) | |
1014 | mask |= bit; | |
1015 | } | |
1016 | if (mask != 0 && rsp->completed == lastcomp) { | |
1017 | ||
1018 | /* cpu_quiet_msk() releases rnp_cur->lock. */ | |
1019 | cpu_quiet_msk(mask, rsp, rnp_cur, flags); | |
1020 | continue; | |
1021 | } | |
1022 | spin_unlock_irqrestore(&rnp_cur->lock, flags); | |
1023 | } | |
1024 | return 0; | |
1025 | } | |
1026 | ||
1027 | /* | |
1028 | * Force quiescent states on reluctant CPUs, and also detect which | |
1029 | * CPUs are in dyntick-idle mode. | |
1030 | */ | |
1031 | static void force_quiescent_state(struct rcu_state *rsp, int relaxed) | |
1032 | { | |
1033 | unsigned long flags; | |
1034 | long lastcomp; | |
1035 | struct rcu_data *rdp = rsp->rda[smp_processor_id()]; | |
1036 | struct rcu_node *rnp = rcu_get_root(rsp); | |
1037 | u8 signaled; | |
1038 | ||
1039 | if (ACCESS_ONCE(rsp->completed) == ACCESS_ONCE(rsp->gpnum)) | |
1040 | return; /* No grace period in progress, nothing to force. */ | |
1041 | if (!spin_trylock_irqsave(&rsp->fqslock, flags)) { | |
1042 | rsp->n_force_qs_lh++; /* Inexact, can lose counts. Tough! */ | |
1043 | return; /* Someone else is already on the job. */ | |
1044 | } | |
1045 | if (relaxed && | |
1046 | (long)(rsp->jiffies_force_qs - jiffies) >= 0 && | |
1047 | (rdp->n_rcu_pending_force_qs - rdp->n_rcu_pending) >= 0) | |
1048 | goto unlock_ret; /* no emergency and done recently. */ | |
1049 | rsp->n_force_qs++; | |
1050 | spin_lock(&rnp->lock); | |
1051 | lastcomp = rsp->completed; | |
1052 | signaled = rsp->signaled; | |
1053 | rsp->jiffies_force_qs = jiffies + RCU_JIFFIES_TILL_FORCE_QS; | |
1054 | rdp->n_rcu_pending_force_qs = rdp->n_rcu_pending + | |
1055 | RCU_JIFFIES_TILL_FORCE_QS; | |
1056 | if (lastcomp == rsp->gpnum) { | |
1057 | rsp->n_force_qs_ngp++; | |
1058 | spin_unlock(&rnp->lock); | |
1059 | goto unlock_ret; /* no GP in progress, time updated. */ | |
1060 | } | |
1061 | spin_unlock(&rnp->lock); | |
1062 | switch (signaled) { | |
1063 | case RCU_GP_INIT: | |
1064 | ||
1065 | break; /* grace period still initializing, ignore. */ | |
1066 | ||
1067 | case RCU_SAVE_DYNTICK: | |
1068 | ||
1069 | if (RCU_SIGNAL_INIT != RCU_SAVE_DYNTICK) | |
1070 | break; /* So gcc recognizes the dead code. */ | |
1071 | ||
1072 | /* Record dyntick-idle state. */ | |
1073 | if (rcu_process_dyntick(rsp, lastcomp, | |
1074 | dyntick_save_progress_counter)) | |
1075 | goto unlock_ret; | |
1076 | ||
1077 | /* Update state, record completion counter. */ | |
1078 | spin_lock(&rnp->lock); | |
1079 | if (lastcomp == rsp->completed) { | |
1080 | rsp->signaled = RCU_FORCE_QS; | |
1081 | dyntick_record_completed(rsp, lastcomp); | |
1082 | } | |
1083 | spin_unlock(&rnp->lock); | |
1084 | break; | |
1085 | ||
1086 | case RCU_FORCE_QS: | |
1087 | ||
1088 | /* Check dyntick-idle state, send IPI to laggarts. */ | |
1089 | if (rcu_process_dyntick(rsp, dyntick_recall_completed(rsp), | |
1090 | rcu_implicit_dynticks_qs)) | |
1091 | goto unlock_ret; | |
1092 | ||
1093 | /* Leave state in case more forcing is required. */ | |
1094 | ||
1095 | break; | |
1096 | } | |
1097 | unlock_ret: | |
1098 | spin_unlock_irqrestore(&rsp->fqslock, flags); | |
1099 | } | |
1100 | ||
1101 | #else /* #ifdef CONFIG_SMP */ | |
1102 | ||
1103 | static void force_quiescent_state(struct rcu_state *rsp, int relaxed) | |
1104 | { | |
1105 | set_need_resched(); | |
1106 | } | |
1107 | ||
1108 | #endif /* #else #ifdef CONFIG_SMP */ | |
1109 | ||
1110 | /* | |
1111 | * This does the RCU processing work from softirq context for the | |
1112 | * specified rcu_state and rcu_data structures. This may be called | |
1113 | * only from the CPU to whom the rdp belongs. | |
1114 | */ | |
1115 | static void | |
1116 | __rcu_process_callbacks(struct rcu_state *rsp, struct rcu_data *rdp) | |
1117 | { | |
1118 | unsigned long flags; | |
1119 | ||
1120 | /* | |
1121 | * If an RCU GP has gone long enough, go check for dyntick | |
1122 | * idle CPUs and, if needed, send resched IPIs. | |
1123 | */ | |
1124 | if ((long)(ACCESS_ONCE(rsp->jiffies_force_qs) - jiffies) < 0 || | |
1125 | (rdp->n_rcu_pending_force_qs - rdp->n_rcu_pending) < 0) | |
1126 | force_quiescent_state(rsp, 1); | |
1127 | ||
1128 | /* | |
1129 | * Advance callbacks in response to end of earlier grace | |
1130 | * period that some other CPU ended. | |
1131 | */ | |
1132 | rcu_process_gp_end(rsp, rdp); | |
1133 | ||
1134 | /* Update RCU state based on any recent quiescent states. */ | |
1135 | rcu_check_quiescent_state(rsp, rdp); | |
1136 | ||
1137 | /* Does this CPU require a not-yet-started grace period? */ | |
1138 | if (cpu_needs_another_gp(rsp, rdp)) { | |
1139 | spin_lock_irqsave(&rcu_get_root(rsp)->lock, flags); | |
1140 | rcu_start_gp(rsp, flags); /* releases above lock */ | |
1141 | } | |
1142 | ||
1143 | /* If there are callbacks ready, invoke them. */ | |
1144 | rcu_do_batch(rdp); | |
1145 | } | |
1146 | ||
1147 | /* | |
1148 | * Do softirq processing for the current CPU. | |
1149 | */ | |
1150 | static void rcu_process_callbacks(struct softirq_action *unused) | |
1151 | { | |
1152 | /* | |
1153 | * Memory references from any prior RCU read-side critical sections | |
1154 | * executed by the interrupted code must be seen before any RCU | |
1155 | * grace-period manipulations below. | |
1156 | */ | |
1157 | smp_mb(); /* See above block comment. */ | |
1158 | ||
1159 | __rcu_process_callbacks(&rcu_state, &__get_cpu_var(rcu_data)); | |
1160 | __rcu_process_callbacks(&rcu_bh_state, &__get_cpu_var(rcu_bh_data)); | |
1161 | ||
1162 | /* | |
1163 | * Memory references from any later RCU read-side critical sections | |
1164 | * executed by the interrupted code must be seen after any RCU | |
1165 | * grace-period manipulations above. | |
1166 | */ | |
1167 | smp_mb(); /* See above block comment. */ | |
1168 | } | |
1169 | ||
1170 | static void | |
1171 | __call_rcu(struct rcu_head *head, void (*func)(struct rcu_head *rcu), | |
1172 | struct rcu_state *rsp) | |
1173 | { | |
1174 | unsigned long flags; | |
1175 | struct rcu_data *rdp; | |
1176 | ||
1177 | head->func = func; | |
1178 | head->next = NULL; | |
1179 | ||
1180 | smp_mb(); /* Ensure RCU update seen before callback registry. */ | |
1181 | ||
1182 | /* | |
1183 | * Opportunistically note grace-period endings and beginnings. | |
1184 | * Note that we might see a beginning right after we see an | |
1185 | * end, but never vice versa, since this CPU has to pass through | |
1186 | * a quiescent state betweentimes. | |
1187 | */ | |
1188 | local_irq_save(flags); | |
1189 | rdp = rsp->rda[smp_processor_id()]; | |
1190 | rcu_process_gp_end(rsp, rdp); | |
1191 | check_for_new_grace_period(rsp, rdp); | |
1192 | ||
1193 | /* Add the callback to our list. */ | |
1194 | *rdp->nxttail[RCU_NEXT_TAIL] = head; | |
1195 | rdp->nxttail[RCU_NEXT_TAIL] = &head->next; | |
1196 | ||
1197 | /* Start a new grace period if one not already started. */ | |
1198 | if (ACCESS_ONCE(rsp->completed) == ACCESS_ONCE(rsp->gpnum)) { | |
1199 | unsigned long nestflag; | |
1200 | struct rcu_node *rnp_root = rcu_get_root(rsp); | |
1201 | ||
1202 | spin_lock_irqsave(&rnp_root->lock, nestflag); | |
1203 | rcu_start_gp(rsp, nestflag); /* releases rnp_root->lock. */ | |
1204 | } | |
1205 | ||
1206 | /* Force the grace period if too many callbacks or too long waiting. */ | |
1207 | if (unlikely(++rdp->qlen > qhimark)) { | |
1208 | rdp->blimit = LONG_MAX; | |
1209 | force_quiescent_state(rsp, 0); | |
1210 | } else if ((long)(ACCESS_ONCE(rsp->jiffies_force_qs) - jiffies) < 0 || | |
1211 | (rdp->n_rcu_pending_force_qs - rdp->n_rcu_pending) < 0) | |
1212 | force_quiescent_state(rsp, 1); | |
1213 | local_irq_restore(flags); | |
1214 | } | |
1215 | ||
1216 | /* | |
1217 | * Queue an RCU callback for invocation after a grace period. | |
1218 | */ | |
1219 | void call_rcu(struct rcu_head *head, void (*func)(struct rcu_head *rcu)) | |
1220 | { | |
1221 | __call_rcu(head, func, &rcu_state); | |
1222 | } | |
1223 | EXPORT_SYMBOL_GPL(call_rcu); | |
1224 | ||
1225 | /* | |
1226 | * Queue an RCU for invocation after a quicker grace period. | |
1227 | */ | |
1228 | void call_rcu_bh(struct rcu_head *head, void (*func)(struct rcu_head *rcu)) | |
1229 | { | |
1230 | __call_rcu(head, func, &rcu_bh_state); | |
1231 | } | |
1232 | EXPORT_SYMBOL_GPL(call_rcu_bh); | |
1233 | ||
1234 | /* | |
1235 | * Check to see if there is any immediate RCU-related work to be done | |
1236 | * by the current CPU, for the specified type of RCU, returning 1 if so. | |
1237 | * The checks are in order of increasing expense: checks that can be | |
1238 | * carried out against CPU-local state are performed first. However, | |
1239 | * we must check for CPU stalls first, else we might not get a chance. | |
1240 | */ | |
1241 | static int __rcu_pending(struct rcu_state *rsp, struct rcu_data *rdp) | |
1242 | { | |
1243 | rdp->n_rcu_pending++; | |
1244 | ||
1245 | /* Check for CPU stalls, if enabled. */ | |
1246 | check_cpu_stall(rsp, rdp); | |
1247 | ||
1248 | /* Is the RCU core waiting for a quiescent state from this CPU? */ | |
1249 | if (rdp->qs_pending) | |
1250 | return 1; | |
1251 | ||
1252 | /* Does this CPU have callbacks ready to invoke? */ | |
1253 | if (cpu_has_callbacks_ready_to_invoke(rdp)) | |
1254 | return 1; | |
1255 | ||
1256 | /* Has RCU gone idle with this CPU needing another grace period? */ | |
1257 | if (cpu_needs_another_gp(rsp, rdp)) | |
1258 | return 1; | |
1259 | ||
1260 | /* Has another RCU grace period completed? */ | |
1261 | if (ACCESS_ONCE(rsp->completed) != rdp->completed) /* outside of lock */ | |
1262 | return 1; | |
1263 | ||
1264 | /* Has a new RCU grace period started? */ | |
1265 | if (ACCESS_ONCE(rsp->gpnum) != rdp->gpnum) /* outside of lock */ | |
1266 | return 1; | |
1267 | ||
1268 | /* Has an RCU GP gone long enough to send resched IPIs &c? */ | |
1269 | if (ACCESS_ONCE(rsp->completed) != ACCESS_ONCE(rsp->gpnum) && | |
1270 | ((long)(ACCESS_ONCE(rsp->jiffies_force_qs) - jiffies) < 0 || | |
1271 | (rdp->n_rcu_pending_force_qs - rdp->n_rcu_pending) < 0)) | |
1272 | return 1; | |
1273 | ||
1274 | /* nothing to do */ | |
1275 | return 0; | |
1276 | } | |
1277 | ||
1278 | /* | |
1279 | * Check to see if there is any immediate RCU-related work to be done | |
1280 | * by the current CPU, returning 1 if so. This function is part of the | |
1281 | * RCU implementation; it is -not- an exported member of the RCU API. | |
1282 | */ | |
1283 | int rcu_pending(int cpu) | |
1284 | { | |
1285 | return __rcu_pending(&rcu_state, &per_cpu(rcu_data, cpu)) || | |
1286 | __rcu_pending(&rcu_bh_state, &per_cpu(rcu_bh_data, cpu)); | |
1287 | } | |
1288 | ||
1289 | /* | |
1290 | * Check to see if any future RCU-related work will need to be done | |
1291 | * by the current CPU, even if none need be done immediately, returning | |
1292 | * 1 if so. This function is part of the RCU implementation; it is -not- | |
1293 | * an exported member of the RCU API. | |
1294 | */ | |
1295 | int rcu_needs_cpu(int cpu) | |
1296 | { | |
1297 | /* RCU callbacks either ready or pending? */ | |
1298 | return per_cpu(rcu_data, cpu).nxtlist || | |
1299 | per_cpu(rcu_bh_data, cpu).nxtlist; | |
1300 | } | |
1301 | ||
1302 | /* | |
1303 | * Initialize a CPU's per-CPU RCU data. We take this "scorched earth" | |
1304 | * approach so that we don't have to worry about how long the CPU has | |
1305 | * been gone, or whether it ever was online previously. We do trust the | |
1306 | * ->mynode field, as it is constant for a given struct rcu_data and | |
1307 | * initialized during early boot. | |
1308 | * | |
1309 | * Note that only one online or offline event can be happening at a given | |
1310 | * time. Note also that we can accept some slop in the rsp->completed | |
1311 | * access due to the fact that this CPU cannot possibly have any RCU | |
1312 | * callbacks in flight yet. | |
1313 | */ | |
1314 | static void | |
1315 | rcu_init_percpu_data(int cpu, struct rcu_state *rsp) | |
1316 | { | |
1317 | unsigned long flags; | |
1318 | int i; | |
1319 | long lastcomp; | |
1320 | unsigned long mask; | |
1321 | struct rcu_data *rdp = rsp->rda[cpu]; | |
1322 | struct rcu_node *rnp = rcu_get_root(rsp); | |
1323 | ||
1324 | /* Set up local state, ensuring consistent view of global state. */ | |
1325 | spin_lock_irqsave(&rnp->lock, flags); | |
1326 | lastcomp = rsp->completed; | |
1327 | rdp->completed = lastcomp; | |
1328 | rdp->gpnum = lastcomp; | |
1329 | rdp->passed_quiesc = 0; /* We could be racing with new GP, */ | |
1330 | rdp->qs_pending = 1; /* so set up to respond to current GP. */ | |
1331 | rdp->beenonline = 1; /* We have now been online. */ | |
1332 | rdp->passed_quiesc_completed = lastcomp - 1; | |
1333 | rdp->grpmask = 1UL << (cpu - rdp->mynode->grplo); | |
1334 | rdp->nxtlist = NULL; | |
1335 | for (i = 0; i < RCU_NEXT_SIZE; i++) | |
1336 | rdp->nxttail[i] = &rdp->nxtlist; | |
1337 | rdp->qlen = 0; | |
1338 | rdp->blimit = blimit; | |
1339 | #ifdef CONFIG_NO_HZ | |
1340 | rdp->dynticks = &per_cpu(rcu_dynticks, cpu); | |
1341 | #endif /* #ifdef CONFIG_NO_HZ */ | |
1342 | rdp->cpu = cpu; | |
1343 | spin_unlock(&rnp->lock); /* irqs remain disabled. */ | |
1344 | ||
1345 | /* | |
1346 | * A new grace period might start here. If so, we won't be part | |
1347 | * of it, but that is OK, as we are currently in a quiescent state. | |
1348 | */ | |
1349 | ||
1350 | /* Exclude any attempts to start a new GP on large systems. */ | |
1351 | spin_lock(&rsp->onofflock); /* irqs already disabled. */ | |
1352 | ||
1353 | /* Add CPU to rcu_node bitmasks. */ | |
1354 | rnp = rdp->mynode; | |
1355 | mask = rdp->grpmask; | |
1356 | do { | |
1357 | /* Exclude any attempts to start a new GP on small systems. */ | |
1358 | spin_lock(&rnp->lock); /* irqs already disabled. */ | |
1359 | rnp->qsmaskinit |= mask; | |
1360 | mask = rnp->grpmask; | |
1361 | spin_unlock(&rnp->lock); /* irqs already disabled. */ | |
1362 | rnp = rnp->parent; | |
1363 | } while (rnp != NULL && !(rnp->qsmaskinit & mask)); | |
1364 | ||
1365 | spin_unlock(&rsp->onofflock); /* irqs remain disabled. */ | |
1366 | ||
1367 | /* | |
1368 | * A new grace period might start here. If so, we will be part of | |
1369 | * it, and its gpnum will be greater than ours, so we will | |
1370 | * participate. It is also possible for the gpnum to have been | |
1371 | * incremented before this function was called, and the bitmasks | |
1372 | * to not be filled out until now, in which case we will also | |
1373 | * participate due to our gpnum being behind. | |
1374 | */ | |
1375 | ||
1376 | /* Since it is coming online, the CPU is in a quiescent state. */ | |
1377 | cpu_quiet(cpu, rsp, rdp, lastcomp); | |
1378 | local_irq_restore(flags); | |
1379 | } | |
1380 | ||
1381 | static void __cpuinit rcu_online_cpu(int cpu) | |
1382 | { | |
1383 | #ifdef CONFIG_NO_HZ | |
1384 | struct rcu_dynticks *rdtp = &per_cpu(rcu_dynticks, cpu); | |
1385 | ||
1386 | rdtp->dynticks_nesting = 1; | |
1387 | rdtp->dynticks |= 1; /* need consecutive #s even for hotplug. */ | |
1388 | rdtp->dynticks_nmi = (rdtp->dynticks_nmi + 1) & ~0x1; | |
1389 | #endif /* #ifdef CONFIG_NO_HZ */ | |
1390 | rcu_init_percpu_data(cpu, &rcu_state); | |
1391 | rcu_init_percpu_data(cpu, &rcu_bh_state); | |
1392 | open_softirq(RCU_SOFTIRQ, rcu_process_callbacks); | |
1393 | } | |
1394 | ||
1395 | /* | |
1396 | * Handle CPU online/offline notifcation events. | |
1397 | */ | |
1398 | static int __cpuinit rcu_cpu_notify(struct notifier_block *self, | |
1399 | unsigned long action, void *hcpu) | |
1400 | { | |
1401 | long cpu = (long)hcpu; | |
1402 | ||
1403 | switch (action) { | |
1404 | case CPU_UP_PREPARE: | |
1405 | case CPU_UP_PREPARE_FROZEN: | |
1406 | rcu_online_cpu(cpu); | |
1407 | break; | |
1408 | case CPU_DEAD: | |
1409 | case CPU_DEAD_FROZEN: | |
1410 | case CPU_UP_CANCELED: | |
1411 | case CPU_UP_CANCELED_FROZEN: | |
1412 | rcu_offline_cpu(cpu); | |
1413 | break; | |
1414 | default: | |
1415 | break; | |
1416 | } | |
1417 | return NOTIFY_OK; | |
1418 | } | |
1419 | ||
1420 | /* | |
1421 | * Compute the per-level fanout, either using the exact fanout specified | |
1422 | * or balancing the tree, depending on CONFIG_RCU_FANOUT_EXACT. | |
1423 | */ | |
1424 | #ifdef CONFIG_RCU_FANOUT_EXACT | |
1425 | static void __init rcu_init_levelspread(struct rcu_state *rsp) | |
1426 | { | |
1427 | int i; | |
1428 | ||
1429 | for (i = NUM_RCU_LVLS - 1; i >= 0; i--) | |
1430 | rsp->levelspread[i] = CONFIG_RCU_FANOUT; | |
1431 | } | |
1432 | #else /* #ifdef CONFIG_RCU_FANOUT_EXACT */ | |
1433 | static void __init rcu_init_levelspread(struct rcu_state *rsp) | |
1434 | { | |
1435 | int ccur; | |
1436 | int cprv; | |
1437 | int i; | |
1438 | ||
1439 | cprv = NR_CPUS; | |
1440 | for (i = NUM_RCU_LVLS - 1; i >= 0; i--) { | |
1441 | ccur = rsp->levelcnt[i]; | |
1442 | rsp->levelspread[i] = (cprv + ccur - 1) / ccur; | |
1443 | cprv = ccur; | |
1444 | } | |
1445 | } | |
1446 | #endif /* #else #ifdef CONFIG_RCU_FANOUT_EXACT */ | |
1447 | ||
1448 | /* | |
1449 | * Helper function for rcu_init() that initializes one rcu_state structure. | |
1450 | */ | |
1451 | static void __init rcu_init_one(struct rcu_state *rsp) | |
1452 | { | |
1453 | int cpustride = 1; | |
1454 | int i; | |
1455 | int j; | |
1456 | struct rcu_node *rnp; | |
1457 | ||
1458 | /* Initialize the level-tracking arrays. */ | |
1459 | ||
1460 | for (i = 1; i < NUM_RCU_LVLS; i++) | |
1461 | rsp->level[i] = rsp->level[i - 1] + rsp->levelcnt[i - 1]; | |
1462 | rcu_init_levelspread(rsp); | |
1463 | ||
1464 | /* Initialize the elements themselves, starting from the leaves. */ | |
1465 | ||
1466 | for (i = NUM_RCU_LVLS - 1; i >= 0; i--) { | |
1467 | cpustride *= rsp->levelspread[i]; | |
1468 | rnp = rsp->level[i]; | |
1469 | for (j = 0; j < rsp->levelcnt[i]; j++, rnp++) { | |
1470 | spin_lock_init(&rnp->lock); | |
1471 | rnp->qsmask = 0; | |
1472 | rnp->qsmaskinit = 0; | |
1473 | rnp->grplo = j * cpustride; | |
1474 | rnp->grphi = (j + 1) * cpustride - 1; | |
1475 | if (rnp->grphi >= NR_CPUS) | |
1476 | rnp->grphi = NR_CPUS - 1; | |
1477 | if (i == 0) { | |
1478 | rnp->grpnum = 0; | |
1479 | rnp->grpmask = 0; | |
1480 | rnp->parent = NULL; | |
1481 | } else { | |
1482 | rnp->grpnum = j % rsp->levelspread[i - 1]; | |
1483 | rnp->grpmask = 1UL << rnp->grpnum; | |
1484 | rnp->parent = rsp->level[i - 1] + | |
1485 | j / rsp->levelspread[i - 1]; | |
1486 | } | |
1487 | rnp->level = i; | |
1488 | } | |
1489 | } | |
1490 | } | |
1491 | ||
1492 | /* | |
1493 | * Helper macro for __rcu_init(). To be used nowhere else! | |
1494 | * Assigns leaf node pointers into each CPU's rcu_data structure. | |
1495 | */ | |
1496 | #define RCU_DATA_PTR_INIT(rsp, rcu_data) \ | |
1497 | do { \ | |
1498 | rnp = (rsp)->level[NUM_RCU_LVLS - 1]; \ | |
1499 | j = 0; \ | |
1500 | for_each_possible_cpu(i) { \ | |
1501 | if (i > rnp[j].grphi) \ | |
1502 | j++; \ | |
1503 | per_cpu(rcu_data, i).mynode = &rnp[j]; \ | |
1504 | (rsp)->rda[i] = &per_cpu(rcu_data, i); \ | |
1505 | } \ | |
1506 | } while (0) | |
1507 | ||
1508 | static struct notifier_block __cpuinitdata rcu_nb = { | |
1509 | .notifier_call = rcu_cpu_notify, | |
1510 | }; | |
1511 | ||
1512 | void __init __rcu_init(void) | |
1513 | { | |
1514 | int i; /* All used by RCU_DATA_PTR_INIT(). */ | |
1515 | int j; | |
1516 | struct rcu_node *rnp; | |
1517 | ||
1518 | printk(KERN_WARNING "Experimental hierarchical RCU implementation.\n"); | |
1519 | #ifdef CONFIG_RCU_CPU_STALL_DETECTOR | |
1520 | printk(KERN_INFO "RCU-based detection of stalled CPUs is enabled.\n"); | |
1521 | #endif /* #ifdef CONFIG_RCU_CPU_STALL_DETECTOR */ | |
1522 | rcu_init_one(&rcu_state); | |
1523 | RCU_DATA_PTR_INIT(&rcu_state, rcu_data); | |
1524 | rcu_init_one(&rcu_bh_state); | |
1525 | RCU_DATA_PTR_INIT(&rcu_bh_state, rcu_bh_data); | |
1526 | ||
1527 | for_each_online_cpu(i) | |
1528 | rcu_cpu_notify(&rcu_nb, CPU_UP_PREPARE, (void *)(long)i); | |
1529 | /* Register notifier for non-boot CPUs */ | |
1530 | register_cpu_notifier(&rcu_nb); | |
1531 | printk(KERN_WARNING "Experimental hierarchical RCU init done.\n"); | |
1532 | } | |
1533 | ||
1534 | module_param(blimit, int, 0); | |
1535 | module_param(qhimark, int, 0); | |
1536 | module_param(qlowmark, int, 0); |