cpu.c 54.9 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9
/* CPU control.
 * (C) 2001, 2002, 2003, 2004 Rusty Russell
 *
 * This code is licenced under the GPL.
 */
#include <linux/proc_fs.h>
#include <linux/smp.h>
#include <linux/init.h>
#include <linux/notifier.h>
10
#include <linux/sched/signal.h>
11
#include <linux/sched/hotplug.h>
12
#include <linux/sched/task.h>
13
#include <linux/sched/smt.h>
L
Linus Torvalds 已提交
14 15
#include <linux/unistd.h>
#include <linux/cpu.h>
16 17
#include <linux/oom.h>
#include <linux/rcupdate.h>
18
#include <linux/export.h>
19
#include <linux/bug.h>
L
Linus Torvalds 已提交
20 21
#include <linux/kthread.h>
#include <linux/stop_machine.h>
22
#include <linux/mutex.h>
23
#include <linux/gfp.h>
24
#include <linux/suspend.h>
25
#include <linux/lockdep.h>
26
#include <linux/tick.h>
27
#include <linux/irq.h>
28
#include <linux/nmi.h>
29
#include <linux/smpboot.h>
30
#include <linux/relay.h>
31
#include <linux/slab.h>
32
#include <linux/percpu-rwsem.h>
33

34
#include <trace/events/power.h>
35 36
#define CREATE_TRACE_POINTS
#include <trace/events/cpuhp.h>
L
Linus Torvalds 已提交
37

38 39
#include "smpboot.h"

40 41 42 43
/**
 * cpuhp_cpu_state - Per cpu hotplug state storage
 * @state:	The current cpu state
 * @target:	The target state
44 45
 * @thread:	Pointer to the hotplug thread
 * @should_run:	Thread should execute
46
 * @rollback:	Perform a rollback
47 48 49
 * @single:	Single callback invocation
 * @bringup:	Single callback bringup or teardown selector
 * @cb_state:	The state for a single callback (install/uninstall)
50
 * @result:	Result of the operation
51 52
 * @done_up:	Signal completion to the issuer of the task for cpu-up
 * @done_down:	Signal completion to the issuer of the task for cpu-down
53 54 55 56
 */
struct cpuhp_cpu_state {
	enum cpuhp_state	state;
	enum cpuhp_state	target;
57
	enum cpuhp_state	fail;
58 59 60
#ifdef CONFIG_SMP
	struct task_struct	*thread;
	bool			should_run;
61
	bool			rollback;
62 63
	bool			single;
	bool			bringup;
64
	bool			booted_once;
65
	struct hlist_node	*node;
66
	struct hlist_node	*last;
67 68
	enum cpuhp_state	cb_state;
	int			result;
69 70
	struct completion	done_up;
	struct completion	done_down;
71
#endif
72 73
};

74 75 76
static DEFINE_PER_CPU(struct cpuhp_cpu_state, cpuhp_state) = {
	.fail = CPUHP_INVALID,
};
77

78
#if defined(CONFIG_LOCKDEP) && defined(CONFIG_SMP)
79 80 81 82 83 84
static struct lockdep_map cpuhp_state_up_map =
	STATIC_LOCKDEP_MAP_INIT("cpuhp_state-up", &cpuhp_state_up_map);
static struct lockdep_map cpuhp_state_down_map =
	STATIC_LOCKDEP_MAP_INIT("cpuhp_state-down", &cpuhp_state_down_map);


85
static inline void cpuhp_lock_acquire(bool bringup)
86 87 88 89
{
	lock_map_acquire(bringup ? &cpuhp_state_up_map : &cpuhp_state_down_map);
}

90
static inline void cpuhp_lock_release(bool bringup)
91 92 93 94 95
{
	lock_map_release(bringup ? &cpuhp_state_up_map : &cpuhp_state_down_map);
}
#else

96 97
static inline void cpuhp_lock_acquire(bool bringup) { }
static inline void cpuhp_lock_release(bool bringup) { }
98

99 100
#endif

101 102 103 104 105
/**
 * cpuhp_step - Hotplug state machine step
 * @name:	Name of the step
 * @startup:	Startup function of the step
 * @teardown:	Teardown function of the step
106
 * @cant_stop:	Bringup/teardown can't be stopped at this step
107 108
 */
struct cpuhp_step {
109 110
	const char		*name;
	union {
111 112 113 114
		int		(*single)(unsigned int cpu);
		int		(*multi)(unsigned int cpu,
					 struct hlist_node *node);
	} startup;
115
	union {
116 117 118 119
		int		(*single)(unsigned int cpu);
		int		(*multi)(unsigned int cpu,
					 struct hlist_node *node);
	} teardown;
120 121 122
	struct hlist_head	list;
	bool			cant_stop;
	bool			multi_instance;
123 124
};

125
static DEFINE_MUTEX(cpuhp_state_mutex);
126
static struct cpuhp_step cpuhp_hp_states[];
127

128 129
static struct cpuhp_step *cpuhp_get_step(enum cpuhp_state state)
{
130
	return cpuhp_hp_states + state;
131 132
}

133 134 135
/**
 * cpuhp_invoke_callback _ Invoke the callbacks for a given state
 * @cpu:	The cpu for which the callback should be invoked
136
 * @state:	The state to do callbacks for
137
 * @bringup:	True if the bringup callback should be invoked
138 139
 * @node:	For multi-instance, do a single entry callback for install/remove
 * @lastp:	For multi-instance rollback, remember how far we got
140
 *
141
 * Called from cpu hotplug and from the state register machinery.
142
 */
143
static int cpuhp_invoke_callback(unsigned int cpu, enum cpuhp_state state,
144 145
				 bool bringup, struct hlist_node *node,
				 struct hlist_node **lastp)
146 147
{
	struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
148
	struct cpuhp_step *step = cpuhp_get_step(state);
149 150 151 152
	int (*cbm)(unsigned int cpu, struct hlist_node *node);
	int (*cb)(unsigned int cpu);
	int ret, cnt;

153 154 155 156 157 158 159 160 161
	if (st->fail == state) {
		st->fail = CPUHP_INVALID;

		if (!(bringup ? step->startup.single : step->teardown.single))
			return 0;

		return -EAGAIN;
	}

162
	if (!step->multi_instance) {
163
		WARN_ON_ONCE(lastp && *lastp);
164
		cb = bringup ? step->startup.single : step->teardown.single;
165 166
		if (!cb)
			return 0;
167
		trace_cpuhp_enter(cpu, st->target, state, cb);
168
		ret = cb(cpu);
169
		trace_cpuhp_exit(cpu, st->state, state, ret);
170 171
		return ret;
	}
172
	cbm = bringup ? step->startup.multi : step->teardown.multi;
173 174 175 176 177
	if (!cbm)
		return 0;

	/* Single invocation for instance add/remove */
	if (node) {
178
		WARN_ON_ONCE(lastp && *lastp);
179 180 181 182 183 184 185 186 187
		trace_cpuhp_multi_enter(cpu, st->target, state, cbm, node);
		ret = cbm(cpu, node);
		trace_cpuhp_exit(cpu, st->state, state, ret);
		return ret;
	}

	/* State transition. Invoke on all instances */
	cnt = 0;
	hlist_for_each(node, &step->list) {
188 189 190
		if (lastp && node == *lastp)
			break;

191 192 193
		trace_cpuhp_multi_enter(cpu, st->target, state, cbm, node);
		ret = cbm(cpu, node);
		trace_cpuhp_exit(cpu, st->state, state, ret);
194 195 196 197 198 199 200
		if (ret) {
			if (!lastp)
				goto err;

			*lastp = node;
			return ret;
		}
201 202
		cnt++;
	}
203 204
	if (lastp)
		*lastp = NULL;
205 206 207
	return 0;
err:
	/* Rollback the instances if one failed */
208
	cbm = !bringup ? step->startup.multi : step->teardown.multi;
209 210 211 212 213 214
	if (!cbm)
		return ret;

	hlist_for_each(node, &step->list) {
		if (!cnt--)
			break;
215 216 217 218 219 220 221 222

		trace_cpuhp_multi_enter(cpu, st->target, state, cbm, node);
		ret = cbm(cpu, node);
		trace_cpuhp_exit(cpu, st->state, state, ret);
		/*
		 * Rollback must not fail,
		 */
		WARN_ON_ONCE(ret);
223 224 225 226
	}
	return ret;
}

227
#ifdef CONFIG_SMP
228 229 230 231 232 233 234 235 236
static bool cpuhp_is_ap_state(enum cpuhp_state state)
{
	/*
	 * The extra check for CPUHP_TEARDOWN_CPU is only for documentation
	 * purposes as that state is handled explicitly in cpu_down.
	 */
	return state > CPUHP_BRINGUP_CPU && state != CPUHP_TEARDOWN_CPU;
}

237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256
static inline void wait_for_ap_thread(struct cpuhp_cpu_state *st, bool bringup)
{
	struct completion *done = bringup ? &st->done_up : &st->done_down;
	wait_for_completion(done);
}

static inline void complete_ap_thread(struct cpuhp_cpu_state *st, bool bringup)
{
	struct completion *done = bringup ? &st->done_up : &st->done_down;
	complete(done);
}

/*
 * The former STARTING/DYING states, ran with IRQs disabled and must not fail.
 */
static bool cpuhp_is_atomic_state(enum cpuhp_state state)
{
	return CPUHP_AP_IDLE_DEAD <= state && state < CPUHP_AP_ONLINE;
}

257
/* Serializes the updates to cpu_online_mask, cpu_present_mask */
258
static DEFINE_MUTEX(cpu_add_remove_lock);
259 260
bool cpuhp_tasks_frozen;
EXPORT_SYMBOL_GPL(cpuhp_tasks_frozen);
L
Linus Torvalds 已提交
261

262
/*
263 264
 * The following two APIs (cpu_maps_update_begin/done) must be used when
 * attempting to serialize the updates to cpu_online_mask & cpu_present_mask.
265 266 267 268 269 270 271 272 273 274
 */
void cpu_maps_update_begin(void)
{
	mutex_lock(&cpu_add_remove_lock);
}

void cpu_maps_update_done(void)
{
	mutex_unlock(&cpu_add_remove_lock);
}
L
Linus Torvalds 已提交
275

276 277
/*
 * If set, cpu_up and cpu_down will return -EBUSY and do nothing.
278 279 280 281
 * Should always be manipulated under cpu_add_remove_lock
 */
static int cpu_hotplug_disabled;

282 283
#ifdef CONFIG_HOTPLUG_CPU

284
DEFINE_STATIC_PERCPU_RWSEM(cpu_hotplug_lock);
285

286
void cpus_read_lock(void)
287
{
288
	percpu_down_read(&cpu_hotplug_lock);
289
}
290
EXPORT_SYMBOL_GPL(cpus_read_lock);
291

292 293 294 295 296 297
int cpus_read_trylock(void)
{
	return percpu_down_read_trylock(&cpu_hotplug_lock);
}
EXPORT_SYMBOL_GPL(cpus_read_trylock);

298
void cpus_read_unlock(void)
299
{
300
	percpu_up_read(&cpu_hotplug_lock);
301
}
302
EXPORT_SYMBOL_GPL(cpus_read_unlock);
303

304
void cpus_write_lock(void)
305
{
306
	percpu_down_write(&cpu_hotplug_lock);
307
}
308

309
void cpus_write_unlock(void)
310
{
311
	percpu_up_write(&cpu_hotplug_lock);
312 313
}

314
void lockdep_assert_cpus_held(void)
315
{
316 317 318 319 320 321 322 323 324
	/*
	 * We can't have hotplug operations before userspace starts running,
	 * and some init codepaths will knowingly not take the hotplug lock.
	 * This is all valid, so mute lockdep until it makes sense to report
	 * unheld locks.
	 */
	if (system_state < SYSTEM_RUNNING)
		return;

325
	percpu_rwsem_assert_held(&cpu_hotplug_lock);
326
}
327

328 329 330 331 332 333 334 335 336 337
/*
 * Wait for currently running CPU hotplug operations to complete (if any) and
 * disable future CPU hotplug (from sysfs). The 'cpu_add_remove_lock' protects
 * the 'cpu_hotplug_disabled' flag. The same lock is also acquired by the
 * hotplug path before performing hotplug operations. So acquiring that lock
 * guarantees mutual exclusion from any currently running hotplug operations.
 */
void cpu_hotplug_disable(void)
{
	cpu_maps_update_begin();
338
	cpu_hotplug_disabled++;
339 340
	cpu_maps_update_done();
}
341
EXPORT_SYMBOL_GPL(cpu_hotplug_disable);
342

343 344 345 346 347 348 349
static void __cpu_hotplug_enable(void)
{
	if (WARN_ONCE(!cpu_hotplug_disabled, "Unbalanced cpu hotplug enable\n"))
		return;
	cpu_hotplug_disabled--;
}

350 351 352
void cpu_hotplug_enable(void)
{
	cpu_maps_update_begin();
353
	__cpu_hotplug_enable();
354 355
	cpu_maps_update_done();
}
356
EXPORT_SYMBOL_GPL(cpu_hotplug_enable);
357
#endif	/* CONFIG_HOTPLUG_CPU */
358

359 360 361 362 363 364
/*
 * Architectures that need SMT-specific errata handling during SMT hotplug
 * should override this.
 */
void __weak arch_smt_update(void) { }

365 366
#ifdef CONFIG_HOTPLUG_SMT
enum cpuhp_smt_control cpu_smt_control __read_mostly = CPU_SMT_ENABLED;
367

368
void __init cpu_smt_disable(bool force)
369
{
370 371 372 373 374
	if (cpu_smt_control == CPU_SMT_FORCE_DISABLED ||
		cpu_smt_control == CPU_SMT_NOT_SUPPORTED)
		return;

	if (force) {
375 376
		pr_info("SMT: Force disabled\n");
		cpu_smt_control = CPU_SMT_FORCE_DISABLED;
377 378
	} else {
		cpu_smt_control = CPU_SMT_DISABLED;
379
	}
380 381
}

382 383
/*
 * The decision whether SMT is supported can only be done after the full
384
 * CPU identification. Called from architecture code.
385 386 387
 */
void __init cpu_smt_check_topology(void)
{
388
	if (!topology_smt_supported())
389 390 391
		cpu_smt_control = CPU_SMT_NOT_SUPPORTED;
}

392 393 394
static int __init smt_cmdline_disable(char *str)
{
	cpu_smt_disable(str && !strcmp(str, "force"));
395 396 397 398 399 400
	return 0;
}
early_param("nosmt", smt_cmdline_disable);

static inline bool cpu_smt_allowed(unsigned int cpu)
{
401
	if (cpu_smt_control == CPU_SMT_ENABLED)
402 403
		return true;

404
	if (topology_is_primary_thread(cpu))
405 406 407 408 409 410 411 412 413 414 415 416 417 418
		return true;

	/*
	 * On x86 it's required to boot all logical CPUs at least once so
	 * that the init code can get a chance to set CR4.MCE on each
	 * CPU. Otherwise, a broadacasted MCE observing CR4.MCE=0b on any
	 * core will shutdown the machine.
	 */
	return !per_cpu(cpuhp_state, cpu).booted_once;
}
#else
static inline bool cpu_smt_allowed(unsigned int cpu) { return true; }
#endif

419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467
static inline enum cpuhp_state
cpuhp_set_state(struct cpuhp_cpu_state *st, enum cpuhp_state target)
{
	enum cpuhp_state prev_state = st->state;

	st->rollback = false;
	st->last = NULL;

	st->target = target;
	st->single = false;
	st->bringup = st->state < target;

	return prev_state;
}

static inline void
cpuhp_reset_state(struct cpuhp_cpu_state *st, enum cpuhp_state prev_state)
{
	st->rollback = true;

	/*
	 * If we have st->last we need to undo partial multi_instance of this
	 * state first. Otherwise start undo at the previous state.
	 */
	if (!st->last) {
		if (st->bringup)
			st->state--;
		else
			st->state++;
	}

	st->target = prev_state;
	st->bringup = !st->bringup;
}

/* Regular hotplug invocation of the AP hotplug thread */
static void __cpuhp_kick_ap(struct cpuhp_cpu_state *st)
{
	if (!st->single && st->state == st->target)
		return;

	st->result = 0;
	/*
	 * Make sure the above stores are visible before should_run becomes
	 * true. Paired with the mb() above in cpuhp_thread_fun()
	 */
	smp_mb();
	st->should_run = true;
	wake_up_process(st->thread);
468
	wait_for_ap_thread(st, st->bringup);
469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484
}

static int cpuhp_kick_ap(struct cpuhp_cpu_state *st, enum cpuhp_state target)
{
	enum cpuhp_state prev_state;
	int ret;

	prev_state = cpuhp_set_state(st, target);
	__cpuhp_kick_ap(st);
	if ((ret = st->result)) {
		cpuhp_reset_state(st, prev_state);
		__cpuhp_kick_ap(st);
	}

	return ret;
}
485

486 487 488 489
static int bringup_wait_for_ap(unsigned int cpu)
{
	struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);

490
	/* Wait for the CPU to reach CPUHP_AP_ONLINE_IDLE */
491
	wait_for_ap_thread(st, true);
492 493
	if (WARN_ON_ONCE((!cpu_online(cpu))))
		return -ECANCELED;
494 495 496 497 498

	/* Unpark the stopper thread and the hotplug thread of the target cpu */
	stop_machine_unpark(cpu);
	kthread_unpark(st->thread);

499 500 501 502 503 504 505 506 507 508
	/*
	 * SMT soft disabling on X86 requires to bring the CPU out of the
	 * BIOS 'wait for SIPI' state in order to set the CR4.MCE bit.  The
	 * CPU marked itself as booted_once in cpu_notify_starting() so the
	 * cpu_smt_allowed() check will now return false if this is not the
	 * primary sibling.
	 */
	if (!cpu_smt_allowed(cpu))
		return -ECANCELED;

509 510 511 512
	if (st->target <= CPUHP_AP_ONLINE_IDLE)
		return 0;

	return cpuhp_kick_ap(st, st->target);
513 514
}

515 516 517 518 519
static int bringup_cpu(unsigned int cpu)
{
	struct task_struct *idle = idle_thread_get(cpu);
	int ret;

520 521 522 523 524 525 526
	/*
	 * Some architectures have to walk the irq descriptors to
	 * setup the vector space for the cpu which comes online.
	 * Prevent irq alloc/free across the bringup.
	 */
	irq_lock_sparse();

527 528
	/* Arch-specific enabling code. */
	ret = __cpu_up(cpu, idle);
529
	irq_unlock_sparse();
530
	if (ret)
531
		return ret;
532
	return bringup_wait_for_ap(cpu);
533 534
}

535 536 537 538
/*
 * Hotplug state machine related functions
 */

539
static void undo_cpu_up(unsigned int cpu, struct cpuhp_cpu_state *st)
540
{
541 542
	for (st->state--; st->state > st->target; st->state--)
		cpuhp_invoke_callback(cpu, st->state, false, NULL, NULL);
543 544
}

545 546 547 548 549 550 551 552 553 554 555 556 557 558
static inline bool can_rollback_cpu(struct cpuhp_cpu_state *st)
{
	if (IS_ENABLED(CONFIG_HOTPLUG_CPU))
		return true;
	/*
	 * When CPU hotplug is disabled, then taking the CPU down is not
	 * possible because takedown_cpu() and the architecture and
	 * subsystem specific mechanisms are not available. So the CPU
	 * which would be completely unplugged again needs to stay around
	 * in the current state.
	 */
	return st->state <= CPUHP_BRINGUP_CPU;
}

559
static int cpuhp_up_callbacks(unsigned int cpu, struct cpuhp_cpu_state *st,
560
			      enum cpuhp_state target)
561 562 563 564 565 566
{
	enum cpuhp_state prev_state = st->state;
	int ret = 0;

	while (st->state < target) {
		st->state++;
567
		ret = cpuhp_invoke_callback(cpu, st->state, true, NULL, NULL);
568
		if (ret) {
569 570 571 572
			if (can_rollback_cpu(st)) {
				st->target = prev_state;
				undo_cpu_up(cpu, st);
			}
573 574 575 576 577 578
			break;
		}
	}
	return ret;
}

579 580 581 582 583 584 585
/*
 * The cpu hotplug threads manage the bringup and teardown of the cpus
 */
static void cpuhp_create(unsigned int cpu)
{
	struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);

586 587
	init_completion(&st->done_up);
	init_completion(&st->done_down);
588 589 590 591 592 593 594 595 596 597 598 599
}

static int cpuhp_should_run(unsigned int cpu)
{
	struct cpuhp_cpu_state *st = this_cpu_ptr(&cpuhp_state);

	return st->should_run;
}

/*
 * Execute teardown/startup callbacks on the plugged cpu. Also used to invoke
 * callbacks when a state gets [un]installed at runtime.
600 601 602 603 604 605 606 607 608 609
 *
 * Each invocation of this function by the smpboot thread does a single AP
 * state callback.
 *
 * It has 3 modes of operation:
 *  - single: runs st->cb_state
 *  - up:     runs ++st->state, while st->state < st->target
 *  - down:   runs st->state--, while st->state > st->target
 *
 * When complete or on error, should_run is cleared and the completion is fired.
610 611 612 613
 */
static void cpuhp_thread_fun(unsigned int cpu)
{
	struct cpuhp_cpu_state *st = this_cpu_ptr(&cpuhp_state);
614 615
	bool bringup = st->bringup;
	enum cpuhp_state state;
616

617 618 619
	if (WARN_ON_ONCE(!st->should_run))
		return;

620
	/*
621 622
	 * ACQUIRE for the cpuhp_should_run() load of ->should_run. Ensures
	 * that if we see ->should_run we also see the rest of the state.
623 624 625
	 */
	smp_mb();

626
	cpuhp_lock_acquire(bringup);
627

628
	if (st->single) {
629 630 631 632 633 634 635 636
		state = st->cb_state;
		st->should_run = false;
	} else {
		if (bringup) {
			st->state++;
			state = st->state;
			st->should_run = (st->state < st->target);
			WARN_ON_ONCE(st->state > st->target);
637
		} else {
638 639 640 641
			state = st->state;
			st->state--;
			st->should_run = (st->state > st->target);
			WARN_ON_ONCE(st->state < st->target);
642
		}
643 644 645 646 647 648 649 650
	}

	WARN_ON_ONCE(!cpuhp_is_ap_state(state));

	if (cpuhp_is_atomic_state(state)) {
		local_irq_disable();
		st->result = cpuhp_invoke_callback(cpu, state, bringup, st->node, &st->last);
		local_irq_enable();
651

652 653 654 655
		/*
		 * STARTING/DYING must not fail!
		 */
		WARN_ON_ONCE(st->result);
656
	} else {
657 658 659 660 661 662 663 664 665 666 667
		st->result = cpuhp_invoke_callback(cpu, state, bringup, st->node, &st->last);
	}

	if (st->result) {
		/*
		 * If we fail on a rollback, we're up a creek without no
		 * paddle, no way forward, no way back. We loose, thanks for
		 * playing.
		 */
		WARN_ON_ONCE(st->rollback);
		st->should_run = false;
668
	}
669

670
	cpuhp_lock_release(bringup);
671 672

	if (!st->should_run)
673
		complete_ap_thread(st, bringup);
674 675 676
}

/* Invoke a single callback on a remote cpu */
677
static int
678 679
cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state, bool bringup,
			 struct hlist_node *node)
680 681
{
	struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
682
	int ret;
683 684 685 686

	if (!cpu_online(cpu))
		return 0;

687 688 689 690 691
	cpuhp_lock_acquire(false);
	cpuhp_lock_release(false);

	cpuhp_lock_acquire(true);
	cpuhp_lock_release(true);
692

693 694 695 696 697
	/*
	 * If we are up and running, use the hotplug thread. For early calls
	 * we invoke the thread function directly.
	 */
	if (!st->thread)
698
		return cpuhp_invoke_callback(cpu, state, bringup, node, NULL);
699

700 701 702 703 704
	st->rollback = false;
	st->last = NULL;

	st->node = node;
	st->bringup = bringup;
705
	st->cb_state = state;
706 707
	st->single = true;

708
	__cpuhp_kick_ap(st);
709 710

	/*
711
	 * If we failed and did a partial, do a rollback.
712
	 */
713 714 715 716 717 718 719
	if ((ret = st->result) && st->last) {
		st->rollback = true;
		st->bringup = !bringup;

		__cpuhp_kick_ap(st);
	}

720 721 722 723 724
	/*
	 * Clean up the leftovers so the next hotplug operation wont use stale
	 * data.
	 */
	st->node = st->last = NULL;
725
	return ret;
726 727 728 729 730
}

static int cpuhp_kick_ap_work(unsigned int cpu)
{
	struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
731 732
	enum cpuhp_state prev_state = st->state;
	int ret;
733

734 735 736 737 738
	cpuhp_lock_acquire(false);
	cpuhp_lock_release(false);

	cpuhp_lock_acquire(true);
	cpuhp_lock_release(true);
739 740 741 742 743 744

	trace_cpuhp_enter(cpu, st->target, prev_state, cpuhp_kick_ap_work);
	ret = cpuhp_kick_ap(st, st->target);
	trace_cpuhp_exit(cpu, st->state, prev_state, ret);

	return ret;
745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761
}

static struct smp_hotplug_thread cpuhp_threads = {
	.store			= &cpuhp_state.thread,
	.create			= &cpuhp_create,
	.thread_should_run	= cpuhp_should_run,
	.thread_fn		= cpuhp_thread_fun,
	.thread_comm		= "cpuhp/%u",
	.selfparking		= true,
};

void __init cpuhp_threads_init(void)
{
	BUG_ON(smpboot_register_percpu_thread(&cpuhp_threads));
	kthread_unpark(this_cpu_read(cpuhp_state.thread));
}

762
#ifdef CONFIG_HOTPLUG_CPU
763 764 765 766 767 768 769 770 771 772 773 774
/**
 * clear_tasks_mm_cpumask - Safely clear tasks' mm_cpumask for a CPU
 * @cpu: a CPU id
 *
 * This function walks all processes, finds a valid mm struct for each one and
 * then clears a corresponding bit in mm's cpumask.  While this all sounds
 * trivial, there are various non-obvious corner cases, which this function
 * tries to solve in a safe manner.
 *
 * Also note that the function uses a somewhat relaxed locking scheme, so it may
 * be called only for an already offlined CPU.
 */
775 776 777 778 779 780 781 782 783 784 785
void clear_tasks_mm_cpumask(int cpu)
{
	struct task_struct *p;

	/*
	 * This function is called after the cpu is taken down and marked
	 * offline, so its not like new tasks will ever get this cpu set in
	 * their mm mask. -- Peter Zijlstra
	 * Thus, we may use rcu_read_lock() here, instead of grabbing
	 * full-fledged tasklist_lock.
	 */
786
	WARN_ON(cpu_online(cpu));
787 788 789 790
	rcu_read_lock();
	for_each_process(p) {
		struct task_struct *t;

791 792 793 794
		/*
		 * Main thread might exit, but other threads may still have
		 * a valid mm. Find one.
		 */
795 796 797 798 799 800 801 802 803
		t = find_lock_task_mm(p);
		if (!t)
			continue;
		cpumask_clear_cpu(cpu, mm_cpumask(t->mm));
		task_unlock(t);
	}
	rcu_read_unlock();
}

L
Linus Torvalds 已提交
804
/* Take this CPU down. */
805
static int take_cpu_down(void *_param)
L
Linus Torvalds 已提交
806
{
807 808
	struct cpuhp_cpu_state *st = this_cpu_ptr(&cpuhp_state);
	enum cpuhp_state target = max((int)st->target, CPUHP_AP_OFFLINE);
809
	int err, cpu = smp_processor_id();
810
	int ret;
L
Linus Torvalds 已提交
811 812 813 814

	/* Ensure this CPU doesn't handle any more interrupts. */
	err = __cpu_disable();
	if (err < 0)
Z
Zwane Mwaikambo 已提交
815
		return err;
L
Linus Torvalds 已提交
816

817 818 819 820 821 822
	/*
	 * We get here while we are in CPUHP_TEARDOWN_CPU state and we must not
	 * do this step again.
	 */
	WARN_ON(st->state != CPUHP_TEARDOWN_CPU);
	st->state--;
823
	/* Invoke the former CPU_DYING callbacks */
824 825 826 827 828 829 830
	for (; st->state > target; st->state--) {
		ret = cpuhp_invoke_callback(cpu, st->state, false, NULL, NULL);
		/*
		 * DYING must not fail!
		 */
		WARN_ON_ONCE(ret);
	}
831

832 833
	/* Give up timekeeping duties */
	tick_handover_do_timer();
834
	/* Park the stopper thread */
835
	stop_machine_park(cpu);
Z
Zwane Mwaikambo 已提交
836
	return 0;
L
Linus Torvalds 已提交
837 838
}

839
static int takedown_cpu(unsigned int cpu)
L
Linus Torvalds 已提交
840
{
841
	struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
842
	int err;
L
Linus Torvalds 已提交
843

844
	/* Park the smpboot threads */
845 846
	kthread_park(per_cpu_ptr(&cpuhp_state, cpu)->thread);

847
	/*
848 849
	 * Prevent irq alloc/free while the dying cpu reorganizes the
	 * interrupt affinities.
850
	 */
851
	irq_lock_sparse();
852

853 854 855
	/*
	 * So now all preempt/rcu users must observe !cpu_active().
	 */
856
	err = stop_machine_cpuslocked(take_cpu_down, NULL, cpumask_of(cpu));
857
	if (err) {
858
		/* CPU refused to die */
859
		irq_unlock_sparse();
860 861
		/* Unpark the hotplug thread so we can rollback there */
		kthread_unpark(per_cpu_ptr(&cpuhp_state, cpu)->thread);
862
		return err;
863
	}
864
	BUG_ON(cpu_online(cpu));
L
Linus Torvalds 已提交
865

866
	/*
867 868
	 * The teardown callback for CPUHP_AP_SCHED_STARTING will have removed
	 * all runnable tasks from the CPU, there's only the idle task left now
869
	 * that the migration thread is done doing the stop_machine thing.
P
Peter Zijlstra 已提交
870 871
	 *
	 * Wait for the stop thread to go away.
872
	 */
873
	wait_for_ap_thread(st, false);
874
	BUG_ON(st->state != CPUHP_AP_IDLE_DEAD);
L
Linus Torvalds 已提交
875

876 877 878
	/* Interrupts are moved away from the dying cpu, reenable alloc/free */
	irq_unlock_sparse();

879
	hotplug_cpu__broadcast_tick_pull(cpu);
L
Linus Torvalds 已提交
880 881 882
	/* This actually kills the CPU. */
	__cpu_die(cpu);

883
	tick_cleanup_dead_cpu(cpu);
884
	rcutree_migrate_callbacks(cpu);
885 886
	return 0;
}
L
Linus Torvalds 已提交
887

888 889 890 891
static void cpuhp_complete_idle_dead(void *arg)
{
	struct cpuhp_cpu_state *st = arg;

892
	complete_ap_thread(st, false);
893 894
}

895 896 897 898 899
void cpuhp_report_idle_dead(void)
{
	struct cpuhp_cpu_state *st = this_cpu_ptr(&cpuhp_state);

	BUG_ON(st->state != CPUHP_AP_OFFLINE);
900
	rcu_report_dead(smp_processor_id());
901 902 903 904 905 906 907
	st->state = CPUHP_AP_IDLE_DEAD;
	/*
	 * We cannot call complete after rcu_report_dead() so we delegate it
	 * to an online cpu.
	 */
	smp_call_function_single(cpumask_first(cpu_online_mask),
				 cpuhp_complete_idle_dead, st, 0);
908 909
}

910 911
static void undo_cpu_down(unsigned int cpu, struct cpuhp_cpu_state *st)
{
912 913
	for (st->state++; st->state < st->target; st->state++)
		cpuhp_invoke_callback(cpu, st->state, true, NULL, NULL);
914 915 916 917 918 919 920 921 922 923 924 925
}

static int cpuhp_down_callbacks(unsigned int cpu, struct cpuhp_cpu_state *st,
				enum cpuhp_state target)
{
	enum cpuhp_state prev_state = st->state;
	int ret = 0;

	for (; st->state > target; st->state--) {
		ret = cpuhp_invoke_callback(cpu, st->state, false, NULL, NULL);
		if (ret) {
			st->target = prev_state;
926 927
			if (st->state < prev_state)
				undo_cpu_down(cpu, st);
928 929 930 931 932
			break;
		}
	}
	return ret;
}
933

934
/* Requires cpu_add_remove_lock to be held */
935 936
static int __ref _cpu_down(unsigned int cpu, int tasks_frozen,
			   enum cpuhp_state target)
937
{
938 939
	struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
	int prev_state, ret = 0;
940 941 942 943

	if (num_online_cpus() == 1)
		return -EBUSY;

944
	if (!cpu_present(cpu))
945 946
		return -EINVAL;

947
	cpus_write_lock();
948 949 950

	cpuhp_tasks_frozen = tasks_frozen;

951
	prev_state = cpuhp_set_state(st, target);
952 953 954 955
	/*
	 * If the current CPU state is in the range of the AP hotplug thread,
	 * then we need to kick the thread.
	 */
956
	if (st->state > CPUHP_TEARDOWN_CPU) {
957
		st->target = max((int)target, CPUHP_TEARDOWN_CPU);
958 959 960 961 962 963 964 965 966 967 968 969
		ret = cpuhp_kick_ap_work(cpu);
		/*
		 * The AP side has done the error rollback already. Just
		 * return the error code..
		 */
		if (ret)
			goto out;

		/*
		 * We might have stopped still in the range of the AP hotplug
		 * thread. Nothing to do anymore.
		 */
970
		if (st->state > CPUHP_TEARDOWN_CPU)
971
			goto out;
972 973

		st->target = target;
974 975
	}
	/*
976
	 * The AP brought itself down to CPUHP_TEARDOWN_CPU. So we need
977 978
	 * to do the further cleanups.
	 */
979
	ret = cpuhp_down_callbacks(cpu, st, target);
980
	if (ret && st->state == CPUHP_TEARDOWN_CPU && st->state < prev_state) {
981 982
		cpuhp_reset_state(st, prev_state);
		__cpuhp_kick_ap(st);
983
	}
984

985
out:
986
	cpus_write_unlock();
987 988 989 990 991
	/*
	 * Do post unplug cleanup. This is still protected against
	 * concurrent CPU hotplug via cpu_add_remove_lock.
	 */
	lockup_detector_cleanup();
992
	arch_smt_update();
993
	return ret;
994 995
}

996 997 998 999 1000 1001 1002
static int cpu_down_maps_locked(unsigned int cpu, enum cpuhp_state target)
{
	if (cpu_hotplug_disabled)
		return -EBUSY;
	return _cpu_down(cpu, 0, target);
}

1003
static int do_cpu_down(unsigned int cpu, enum cpuhp_state target)
1004
{
1005
	int err;
1006

1007
	cpu_maps_update_begin();
1008
	err = cpu_down_maps_locked(cpu, target);
1009
	cpu_maps_update_done();
L
Linus Torvalds 已提交
1010 1011
	return err;
}
1012

1013 1014 1015 1016
int cpu_down(unsigned int cpu)
{
	return do_cpu_down(cpu, CPUHP_OFFLINE);
}
1017
EXPORT_SYMBOL(cpu_down);
1018 1019 1020

#else
#define takedown_cpu		NULL
L
Linus Torvalds 已提交
1021 1022
#endif /*CONFIG_HOTPLUG_CPU*/

1023
/**
1024
 * notify_cpu_starting(cpu) - Invoke the callbacks on the starting CPU
1025 1026 1027 1028 1029 1030 1031 1032 1033
 * @cpu: cpu that just started
 *
 * It must be called by the arch code on the new cpu, before the new cpu
 * enables interrupts and before the "boot" cpu returns from __cpu_up().
 */
void notify_cpu_starting(unsigned int cpu)
{
	struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
	enum cpuhp_state target = min((int)st->target, CPUHP_AP_ONLINE);
1034
	int ret;
1035

1036
	rcu_cpu_starting(cpu);	/* Enables RCU usage on this CPU. */
1037
	st->booted_once = true;
1038 1039
	while (st->state < target) {
		st->state++;
1040 1041 1042 1043 1044
		ret = cpuhp_invoke_callback(cpu, st->state, true, NULL, NULL);
		/*
		 * STARTING must not fail!
		 */
		WARN_ON_ONCE(ret);
1045 1046 1047
	}
}

1048
/*
1049 1050 1051
 * Called from the idle task. Wake up the controlling task which brings the
 * stopper and the hotplug thread of the upcoming CPU up and then delegates
 * the rest of the online bringup to the hotplug thread.
1052
 */
1053
void cpuhp_online_idle(enum cpuhp_state state)
1054
{
1055 1056 1057 1058 1059 1060 1061
	struct cpuhp_cpu_state *st = this_cpu_ptr(&cpuhp_state);

	/* Happens for the boot cpu */
	if (state != CPUHP_AP_ONLINE_IDLE)
		return;

	st->state = CPUHP_AP_ONLINE_IDLE;
1062
	complete_ap_thread(st, true);
1063 1064
}

1065
/* Requires cpu_add_remove_lock to be held */
1066
static int _cpu_up(unsigned int cpu, int tasks_frozen, enum cpuhp_state target)
L
Linus Torvalds 已提交
1067
{
1068
	struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
1069
	struct task_struct *idle;
1070
	int ret = 0;
L
Linus Torvalds 已提交
1071

1072
	cpus_write_lock();
1073

1074
	if (!cpu_present(cpu)) {
1075 1076 1077 1078
		ret = -EINVAL;
		goto out;
	}

1079 1080 1081 1082 1083
	/*
	 * The caller of do_cpu_up might have raced with another
	 * caller. Ignore it for now.
	 */
	if (st->state >= target)
1084
		goto out;
1085 1086 1087 1088 1089 1090 1091 1092

	if (st->state == CPUHP_OFFLINE) {
		/* Let it fail before we try to bring the cpu up */
		idle = idle_thread_get(cpu);
		if (IS_ERR(idle)) {
			ret = PTR_ERR(idle);
			goto out;
		}
1093
	}
1094

1095 1096
	cpuhp_tasks_frozen = tasks_frozen;

1097
	cpuhp_set_state(st, target);
1098 1099 1100 1101
	/*
	 * If the current CPU state is in the range of the AP hotplug thread,
	 * then we need to kick the thread once more.
	 */
1102
	if (st->state > CPUHP_BRINGUP_CPU) {
1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113
		ret = cpuhp_kick_ap_work(cpu);
		/*
		 * The AP side has done the error rollback already. Just
		 * return the error code..
		 */
		if (ret)
			goto out;
	}

	/*
	 * Try to reach the target state. We max out on the BP at
1114
	 * CPUHP_BRINGUP_CPU. After that the AP hotplug thread is
1115 1116
	 * responsible for bringing it up to the target state.
	 */
1117
	target = min((int)target, CPUHP_BRINGUP_CPU);
1118
	ret = cpuhp_up_callbacks(cpu, st, target);
1119
out:
1120
	cpus_write_unlock();
1121
	arch_smt_update();
1122 1123 1124
	return ret;
}

1125
static int do_cpu_up(unsigned int cpu, enum cpuhp_state target)
1126 1127
{
	int err = 0;
1128

R
Rusty Russell 已提交
1129
	if (!cpu_possible(cpu)) {
1130 1131
		pr_err("can't online cpu %d because it is not configured as may-hotadd at boot time\n",
		       cpu);
1132
#if defined(CONFIG_IA64)
1133
		pr_err("please check additional_cpus= boot parameter\n");
1134 1135 1136
#endif
		return -EINVAL;
	}
1137

1138 1139 1140
	err = try_online_node(cpu_to_node(cpu));
	if (err)
		return err;
1141

1142
	cpu_maps_update_begin();
1143 1144

	if (cpu_hotplug_disabled) {
1145
		err = -EBUSY;
1146 1147
		goto out;
	}
1148 1149 1150 1151
	if (!cpu_smt_allowed(cpu)) {
		err = -EPERM;
		goto out;
	}
1152

1153
	err = _cpu_up(cpu, 0, target);
1154
out:
1155
	cpu_maps_update_done();
1156 1157
	return err;
}
1158 1159 1160 1161 1162

int cpu_up(unsigned int cpu)
{
	return do_cpu_up(cpu, CPUHP_ONLINE);
}
P
Paul E. McKenney 已提交
1163
EXPORT_SYMBOL_GPL(cpu_up);
1164

1165
#ifdef CONFIG_PM_SLEEP_SMP
R
Rusty Russell 已提交
1166
static cpumask_var_t frozen_cpus;
1167

1168
int freeze_secondary_cpus(int primary)
1169
{
1170
	int cpu, error = 0;
1171

1172
	cpu_maps_update_begin();
1173 1174
	if (!cpu_online(primary))
		primary = cpumask_first(cpu_online_mask);
1175 1176
	/*
	 * We take down all of the non-boot CPUs in one shot to avoid races
1177 1178
	 * with the userspace trying to use the CPU hotplug at the same time
	 */
R
Rusty Russell 已提交
1179
	cpumask_clear(frozen_cpus);
1180

1181
	pr_info("Disabling non-boot CPUs ...\n");
1182
	for_each_online_cpu(cpu) {
1183
		if (cpu == primary)
1184
			continue;
1185
		trace_suspend_resume(TPS("CPU_OFF"), cpu, true);
1186
		error = _cpu_down(cpu, 1, CPUHP_OFFLINE);
1187
		trace_suspend_resume(TPS("CPU_OFF"), cpu, false);
1188
		if (!error)
R
Rusty Russell 已提交
1189
			cpumask_set_cpu(cpu, frozen_cpus);
1190
		else {
1191
			pr_err("Error taking CPU%d down: %d\n", cpu, error);
1192 1193 1194
			break;
		}
	}
1195

1196
	if (!error)
1197
		BUG_ON(num_online_cpus() > 1);
1198
	else
1199
		pr_err("Non-boot CPUs are not disabled\n");
1200 1201 1202 1203 1204 1205 1206 1207

	/*
	 * Make sure the CPUs won't be enabled by someone else. We need to do
	 * this even in case of failure as all disable_nonboot_cpus() users are
	 * supposed to do enable_nonboot_cpus() on the failure path.
	 */
	cpu_hotplug_disabled++;

1208
	cpu_maps_update_done();
1209 1210 1211
	return error;
}

1212 1213 1214 1215 1216 1217 1218 1219
void __weak arch_enable_nonboot_cpus_begin(void)
{
}

void __weak arch_enable_nonboot_cpus_end(void)
{
}

1220
void enable_nonboot_cpus(void)
1221 1222 1223 1224
{
	int cpu, error;

	/* Allow everyone to use the CPU hotplug again */
1225
	cpu_maps_update_begin();
1226
	__cpu_hotplug_enable();
R
Rusty Russell 已提交
1227
	if (cpumask_empty(frozen_cpus))
1228
		goto out;
1229

1230
	pr_info("Enabling non-boot CPUs ...\n");
1231 1232 1233

	arch_enable_nonboot_cpus_begin();

R
Rusty Russell 已提交
1234
	for_each_cpu(cpu, frozen_cpus) {
1235
		trace_suspend_resume(TPS("CPU_ON"), cpu, true);
1236
		error = _cpu_up(cpu, 1, CPUHP_ONLINE);
1237
		trace_suspend_resume(TPS("CPU_ON"), cpu, false);
1238
		if (!error) {
1239
			pr_info("CPU%d is up\n", cpu);
1240 1241
			continue;
		}
1242
		pr_warn("Error taking CPU%d up: %d\n", cpu, error);
1243
	}
1244 1245 1246

	arch_enable_nonboot_cpus_end();

R
Rusty Russell 已提交
1247
	cpumask_clear(frozen_cpus);
1248
out:
1249
	cpu_maps_update_done();
L
Linus Torvalds 已提交
1250
}
R
Rusty Russell 已提交
1251

1252
static int __init alloc_frozen_cpus(void)
R
Rusty Russell 已提交
1253 1254 1255 1256 1257 1258
{
	if (!alloc_cpumask_var(&frozen_cpus, GFP_KERNEL|__GFP_ZERO))
		return -ENOMEM;
	return 0;
}
core_initcall(alloc_frozen_cpus);
1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278

/*
 * When callbacks for CPU hotplug notifications are being executed, we must
 * ensure that the state of the system with respect to the tasks being frozen
 * or not, as reported by the notification, remains unchanged *throughout the
 * duration* of the execution of the callbacks.
 * Hence we need to prevent the freezer from racing with regular CPU hotplug.
 *
 * This synchronization is implemented by mutually excluding regular CPU
 * hotplug and Suspend/Hibernate call paths by hooking onto the Suspend/
 * Hibernate notifications.
 */
static int
cpu_hotplug_pm_callback(struct notifier_block *nb,
			unsigned long action, void *ptr)
{
	switch (action) {

	case PM_SUSPEND_PREPARE:
	case PM_HIBERNATION_PREPARE:
1279
		cpu_hotplug_disable();
1280 1281 1282 1283
		break;

	case PM_POST_SUSPEND:
	case PM_POST_HIBERNATION:
1284
		cpu_hotplug_enable();
1285 1286 1287 1288 1289 1290 1291 1292 1293 1294
		break;

	default:
		return NOTIFY_DONE;
	}

	return NOTIFY_OK;
}


1295
static int __init cpu_hotplug_pm_sync_init(void)
1296
{
1297 1298 1299 1300 1301
	/*
	 * cpu_hotplug_pm_callback has higher priority than x86
	 * bsp_pm_callback which depends on cpu_hotplug_pm_callback
	 * to disable cpu hotplug to avoid cpu hotplug race.
	 */
1302 1303 1304 1305 1306
	pm_notifier(cpu_hotplug_pm_callback, 0);
	return 0;
}
core_initcall(cpu_hotplug_pm_sync_init);

1307
#endif /* CONFIG_PM_SLEEP_SMP */
1308

1309 1310
int __boot_cpu_id;

1311
#endif /* CONFIG_SMP */
1312

1313
/* Boot processor state steps */
1314
static struct cpuhp_step cpuhp_hp_states[] = {
1315 1316
	[CPUHP_OFFLINE] = {
		.name			= "offline",
1317 1318
		.startup.single		= NULL,
		.teardown.single	= NULL,
1319 1320 1321
	},
#ifdef CONFIG_SMP
	[CPUHP_CREATE_THREADS]= {
1322
		.name			= "threads:prepare",
1323 1324
		.startup.single		= smpboot_create_threads,
		.teardown.single	= NULL,
1325
		.cant_stop		= true,
1326
	},
1327
	[CPUHP_PERF_PREPARE] = {
1328 1329 1330
		.name			= "perf:prepare",
		.startup.single		= perf_event_init_cpu,
		.teardown.single	= perf_event_exit_cpu,
1331
	},
1332
	[CPUHP_WORKQUEUE_PREP] = {
1333 1334 1335
		.name			= "workqueue:prepare",
		.startup.single		= workqueue_prepare_cpu,
		.teardown.single	= NULL,
1336
	},
1337
	[CPUHP_HRTIMERS_PREPARE] = {
1338 1339 1340
		.name			= "hrtimers:prepare",
		.startup.single		= hrtimers_prepare_cpu,
		.teardown.single	= hrtimers_dead_cpu,
1341
	},
1342
	[CPUHP_SMPCFD_PREPARE] = {
1343
		.name			= "smpcfd:prepare",
1344 1345
		.startup.single		= smpcfd_prepare_cpu,
		.teardown.single	= smpcfd_dead_cpu,
1346
	},
1347 1348 1349 1350 1351
	[CPUHP_RELAY_PREPARE] = {
		.name			= "relay:prepare",
		.startup.single		= relay_prepare_cpu,
		.teardown.single	= NULL,
	},
1352 1353 1354 1355
	[CPUHP_SLAB_PREPARE] = {
		.name			= "slab:prepare",
		.startup.single		= slab_prepare_cpu,
		.teardown.single	= slab_dead_cpu,
1356
	},
1357
	[CPUHP_RCUTREE_PREP] = {
1358
		.name			= "RCU/tree:prepare",
1359 1360
		.startup.single		= rcutree_prepare_cpu,
		.teardown.single	= rcutree_dead_cpu,
1361
	},
1362 1363 1364 1365 1366
	/*
	 * On the tear-down path, timers_dead_cpu() must be invoked
	 * before blk_mq_queue_reinit_notify() from notify_dead(),
	 * otherwise a RCU stall occurs.
	 */
1367
	[CPUHP_TIMERS_PREPARE] = {
1368
		.name			= "timers:prepare",
1369
		.startup.single		= timers_prepare_cpu,
1370
		.teardown.single	= timers_dead_cpu,
1371
	},
1372
	/* Kicks the plugged cpu into life */
1373 1374
	[CPUHP_BRINGUP_CPU] = {
		.name			= "cpu:bringup",
1375 1376
		.startup.single		= bringup_cpu,
		.teardown.single	= NULL,
1377
		.cant_stop		= true,
1378
	},
1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390
	/* Final state before CPU kills itself */
	[CPUHP_AP_IDLE_DEAD] = {
		.name			= "idle:dead",
	},
	/*
	 * Last state before CPU enters the idle loop to die. Transient state
	 * for synchronization.
	 */
	[CPUHP_AP_OFFLINE] = {
		.name			= "ap:offline",
		.cant_stop		= true,
	},
1391 1392 1393
	/* First state is scheduler control. Interrupts are disabled */
	[CPUHP_AP_SCHED_STARTING] = {
		.name			= "sched:starting",
1394 1395
		.startup.single		= sched_cpu_starting,
		.teardown.single	= sched_cpu_dying,
1396
	},
1397
	[CPUHP_AP_RCUTREE_DYING] = {
1398
		.name			= "RCU/tree:dying",
1399 1400
		.startup.single		= NULL,
		.teardown.single	= rcutree_dying_cpu,
1401
	},
1402 1403 1404 1405 1406
	[CPUHP_AP_SMPCFD_DYING] = {
		.name			= "smpcfd:dying",
		.startup.single		= NULL,
		.teardown.single	= smpcfd_dying_cpu,
	},
1407 1408 1409 1410 1411
	/* Entry state on starting. Interrupts enabled from here on. Transient
	 * state for synchronsization */
	[CPUHP_AP_ONLINE] = {
		.name			= "ap:online",
	},
1412 1413 1414 1415 1416 1417 1418 1419 1420 1421
	/*
	 * Handled on controll processor until the plugged processor manages
	 * this itself.
	 */
	[CPUHP_TEARDOWN_CPU] = {
		.name			= "cpu:teardown",
		.startup.single		= NULL,
		.teardown.single	= takedown_cpu,
		.cant_stop		= true,
	},
1422
	/* Handle smpboot threads park/unpark */
1423
	[CPUHP_AP_SMPBOOT_THREADS] = {
1424
		.name			= "smpboot/threads:online",
1425
		.startup.single		= smpboot_unpark_threads,
1426
		.teardown.single	= smpboot_park_threads,
1427
	},
1428 1429 1430 1431 1432
	[CPUHP_AP_IRQ_AFFINITY_ONLINE] = {
		.name			= "irq/affinity:online",
		.startup.single		= irq_affinity_online_cpu,
		.teardown.single	= NULL,
	},
1433
	[CPUHP_AP_PERF_ONLINE] = {
1434 1435 1436
		.name			= "perf:online",
		.startup.single		= perf_event_init_cpu,
		.teardown.single	= perf_event_exit_cpu,
1437
	},
1438 1439 1440 1441 1442
	[CPUHP_AP_WATCHDOG_ONLINE] = {
		.name			= "lockup_detector:online",
		.startup.single		= lockup_detector_online_cpu,
		.teardown.single	= lockup_detector_offline_cpu,
	},
1443
	[CPUHP_AP_WORKQUEUE_ONLINE] = {
1444 1445 1446
		.name			= "workqueue:online",
		.startup.single		= workqueue_online_cpu,
		.teardown.single	= workqueue_offline_cpu,
1447
	},
1448
	[CPUHP_AP_RCUTREE_ONLINE] = {
1449
		.name			= "RCU/tree:online",
1450 1451
		.startup.single		= rcutree_online_cpu,
		.teardown.single	= rcutree_offline_cpu,
1452
	},
1453
#endif
1454 1455 1456 1457
	/*
	 * The dynamically registered state space is here
	 */

1458 1459 1460 1461
#ifdef CONFIG_SMP
	/* Last state is scheduler control setting the cpu active */
	[CPUHP_AP_ACTIVE] = {
		.name			= "sched:active",
1462 1463
		.startup.single		= sched_cpu_activate,
		.teardown.single	= sched_cpu_deactivate,
1464 1465 1466
	},
#endif

1467
	/* CPU is fully up and running. */
1468 1469
	[CPUHP_ONLINE] = {
		.name			= "online",
1470 1471
		.startup.single		= NULL,
		.teardown.single	= NULL,
1472 1473 1474
	},
};

1475 1476 1477 1478 1479 1480 1481 1482
/* Sanity check for callbacks */
static int cpuhp_cb_check(enum cpuhp_state state)
{
	if (state <= CPUHP_OFFLINE || state >= CPUHP_ONLINE)
		return -EINVAL;
	return 0;
}

1483 1484 1485 1486 1487 1488 1489
/*
 * Returns a free for dynamic slot assignment of the Online state. The states
 * are protected by the cpuhp_slot_states mutex and an empty slot is identified
 * by having no name assigned.
 */
static int cpuhp_reserve_state(enum cpuhp_state state)
{
1490 1491
	enum cpuhp_state i, end;
	struct cpuhp_step *step;
1492

1493 1494
	switch (state) {
	case CPUHP_AP_ONLINE_DYN:
1495
		step = cpuhp_hp_states + CPUHP_AP_ONLINE_DYN;
1496 1497 1498
		end = CPUHP_AP_ONLINE_DYN_END;
		break;
	case CPUHP_BP_PREPARE_DYN:
1499
		step = cpuhp_hp_states + CPUHP_BP_PREPARE_DYN;
1500 1501 1502 1503 1504 1505 1506 1507
		end = CPUHP_BP_PREPARE_DYN_END;
		break;
	default:
		return -EINVAL;
	}

	for (i = state; i <= end; i++, step++) {
		if (!step->name)
1508 1509 1510 1511 1512 1513 1514 1515 1516 1517
			return i;
	}
	WARN(1, "No more dynamic states available for CPU hotplug\n");
	return -ENOSPC;
}

static int cpuhp_store_callbacks(enum cpuhp_state state, const char *name,
				 int (*startup)(unsigned int cpu),
				 int (*teardown)(unsigned int cpu),
				 bool multi_instance)
1518 1519 1520
{
	/* (Un)Install the callbacks for further cpu hotplug operations */
	struct cpuhp_step *sp;
1521
	int ret = 0;
1522

1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533
	/*
	 * If name is NULL, then the state gets removed.
	 *
	 * CPUHP_AP_ONLINE_DYN and CPUHP_BP_PREPARE_DYN are handed out on
	 * the first allocation from these dynamic ranges, so the removal
	 * would trigger a new allocation and clear the wrong (already
	 * empty) state, leaving the callbacks of the to be cleared state
	 * dangling, which causes wreckage on the next hotplug operation.
	 */
	if (name && (state == CPUHP_AP_ONLINE_DYN ||
		     state == CPUHP_BP_PREPARE_DYN)) {
1534 1535
		ret = cpuhp_reserve_state(state);
		if (ret < 0)
1536
			return ret;
1537 1538
		state = ret;
	}
1539
	sp = cpuhp_get_step(state);
1540 1541 1542
	if (name && sp->name)
		return -EBUSY;

1543 1544
	sp->startup.single = startup;
	sp->teardown.single = teardown;
1545
	sp->name = name;
1546 1547
	sp->multi_instance = multi_instance;
	INIT_HLIST_HEAD(&sp->list);
1548
	return ret;
1549 1550 1551 1552
}

static void *cpuhp_get_teardown_cb(enum cpuhp_state state)
{
1553
	return cpuhp_get_step(state)->teardown.single;
1554 1555 1556 1557 1558 1559
}

/*
 * Call the startup/teardown function for a step either on the AP or
 * on the current CPU.
 */
1560 1561
static int cpuhp_issue_call(int cpu, enum cpuhp_state state, bool bringup,
			    struct hlist_node *node)
1562
{
1563
	struct cpuhp_step *sp = cpuhp_get_step(state);
1564 1565
	int ret;

1566 1567 1568 1569
	/*
	 * If there's nothing to do, we done.
	 * Relies on the union for multi_instance.
	 */
1570 1571
	if ((bringup && !sp->startup.single) ||
	    (!bringup && !sp->teardown.single))
1572 1573 1574 1575 1576
		return 0;
	/*
	 * The non AP bound callbacks can fail on bringup. On teardown
	 * e.g. module removal we crash for now.
	 */
1577 1578
#ifdef CONFIG_SMP
	if (cpuhp_is_ap_state(state))
1579
		ret = cpuhp_invoke_ap_callback(cpu, state, bringup, node);
1580
	else
1581
		ret = cpuhp_invoke_callback(cpu, state, bringup, node, NULL);
1582
#else
1583
	ret = cpuhp_invoke_callback(cpu, state, bringup, node, NULL);
1584
#endif
1585 1586 1587 1588 1589 1590 1591 1592 1593 1594
	BUG_ON(ret && !bringup);
	return ret;
}

/*
 * Called from __cpuhp_setup_state on a recoverable failure.
 *
 * Note: The teardown callbacks for rollback are not allowed to fail!
 */
static void cpuhp_rollback_install(int failedcpu, enum cpuhp_state state,
1595
				   struct hlist_node *node)
1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608
{
	int cpu;

	/* Roll back the already executed steps on the other cpus */
	for_each_present_cpu(cpu) {
		struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
		int cpustate = st->state;

		if (cpu >= failedcpu)
			break;

		/* Did we invoke the startup call on that cpu ? */
		if (cpustate >= state)
1609
			cpuhp_issue_call(cpu, state, false, node);
1610 1611 1612
	}
}

1613 1614 1615
int __cpuhp_state_add_instance_cpuslocked(enum cpuhp_state state,
					  struct hlist_node *node,
					  bool invoke)
1616 1617 1618 1619 1620
{
	struct cpuhp_step *sp;
	int cpu;
	int ret;

1621 1622
	lockdep_assert_cpus_held();

1623 1624 1625 1626
	sp = cpuhp_get_step(state);
	if (sp->multi_instance == false)
		return -EINVAL;

1627
	mutex_lock(&cpuhp_state_mutex);
1628

1629
	if (!invoke || !sp->startup.multi)
1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644
		goto add_node;

	/*
	 * Try to call the startup callback for each present cpu
	 * depending on the hotplug state of the cpu.
	 */
	for_each_present_cpu(cpu) {
		struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
		int cpustate = st->state;

		if (cpustate < state)
			continue;

		ret = cpuhp_issue_call(cpu, state, true, node);
		if (ret) {
1645
			if (sp->teardown.multi)
1646
				cpuhp_rollback_install(cpu, state, node);
1647
			goto unlock;
1648 1649 1650 1651 1652
		}
	}
add_node:
	ret = 0;
	hlist_add_head(node, &sp->list);
1653
unlock:
1654
	mutex_unlock(&cpuhp_state_mutex);
1655 1656 1657 1658 1659 1660 1661 1662 1663 1664
	return ret;
}

int __cpuhp_state_add_instance(enum cpuhp_state state, struct hlist_node *node,
			       bool invoke)
{
	int ret;

	cpus_read_lock();
	ret = __cpuhp_state_add_instance_cpuslocked(state, node, invoke);
1665
	cpus_read_unlock();
1666 1667 1668 1669
	return ret;
}
EXPORT_SYMBOL_GPL(__cpuhp_state_add_instance);

1670
/**
1671
 * __cpuhp_setup_state_cpuslocked - Setup the callbacks for an hotplug machine state
1672 1673 1674 1675 1676 1677 1678
 * @state:		The state to setup
 * @invoke:		If true, the startup function is invoked for cpus where
 *			cpu state >= @state
 * @startup:		startup callback function
 * @teardown:		teardown callback function
 * @multi_instance:	State is set up for multiple instances which get
 *			added afterwards.
1679
 *
1680
 * The caller needs to hold cpus read locked while calling this function.
1681 1682 1683 1684 1685
 * Returns:
 *   On success:
 *      Positive state number if @state is CPUHP_AP_ONLINE_DYN
 *      0 for all other states
 *   On failure: proper (negative) error code
1686
 */
1687 1688 1689 1690 1691
int __cpuhp_setup_state_cpuslocked(enum cpuhp_state state,
				   const char *name, bool invoke,
				   int (*startup)(unsigned int cpu),
				   int (*teardown)(unsigned int cpu),
				   bool multi_instance)
1692 1693
{
	int cpu, ret = 0;
1694
	bool dynstate;
1695

1696 1697
	lockdep_assert_cpus_held();

1698 1699 1700
	if (cpuhp_cb_check(state) || !name)
		return -EINVAL;

1701
	mutex_lock(&cpuhp_state_mutex);
1702

1703 1704
	ret = cpuhp_store_callbacks(state, name, startup, teardown,
				    multi_instance);
1705

1706 1707 1708 1709 1710 1711
	dynstate = state == CPUHP_AP_ONLINE_DYN;
	if (ret > 0 && dynstate) {
		state = ret;
		ret = 0;
	}

1712
	if (ret || !invoke || !startup)
1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725
		goto out;

	/*
	 * Try to call the startup callback for each present cpu
	 * depending on the hotplug state of the cpu.
	 */
	for_each_present_cpu(cpu) {
		struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
		int cpustate = st->state;

		if (cpustate < state)
			continue;

1726
		ret = cpuhp_issue_call(cpu, state, true, NULL);
1727
		if (ret) {
1728
			if (teardown)
1729 1730
				cpuhp_rollback_install(cpu, state, NULL);
			cpuhp_store_callbacks(state, NULL, NULL, NULL, false);
1731 1732 1733 1734
			goto out;
		}
	}
out:
1735
	mutex_unlock(&cpuhp_state_mutex);
1736 1737 1738 1739
	/*
	 * If the requested state is CPUHP_AP_ONLINE_DYN, return the
	 * dynamically allocated state in case of success.
	 */
1740
	if (!ret && dynstate)
1741 1742 1743
		return state;
	return ret;
}
1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759
EXPORT_SYMBOL(__cpuhp_setup_state_cpuslocked);

int __cpuhp_setup_state(enum cpuhp_state state,
			const char *name, bool invoke,
			int (*startup)(unsigned int cpu),
			int (*teardown)(unsigned int cpu),
			bool multi_instance)
{
	int ret;

	cpus_read_lock();
	ret = __cpuhp_setup_state_cpuslocked(state, name, invoke, startup,
					     teardown, multi_instance);
	cpus_read_unlock();
	return ret;
}
1760 1761
EXPORT_SYMBOL(__cpuhp_setup_state);

1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772
int __cpuhp_state_remove_instance(enum cpuhp_state state,
				  struct hlist_node *node, bool invoke)
{
	struct cpuhp_step *sp = cpuhp_get_step(state);
	int cpu;

	BUG_ON(cpuhp_cb_check(state));

	if (!sp->multi_instance)
		return -EINVAL;

1773
	cpus_read_lock();
1774 1775
	mutex_lock(&cpuhp_state_mutex);

1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793
	if (!invoke || !cpuhp_get_teardown_cb(state))
		goto remove;
	/*
	 * Call the teardown callback for each present cpu depending
	 * on the hotplug state of the cpu. This function is not
	 * allowed to fail currently!
	 */
	for_each_present_cpu(cpu) {
		struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
		int cpustate = st->state;

		if (cpustate >= state)
			cpuhp_issue_call(cpu, state, false, node);
	}

remove:
	hlist_del(node);
	mutex_unlock(&cpuhp_state_mutex);
1794
	cpus_read_unlock();
1795 1796 1797 1798

	return 0;
}
EXPORT_SYMBOL_GPL(__cpuhp_state_remove_instance);
1799

1800
/**
1801
 * __cpuhp_remove_state_cpuslocked - Remove the callbacks for an hotplug machine state
1802 1803 1804 1805
 * @state:	The state to remove
 * @invoke:	If true, the teardown function is invoked for cpus where
 *		cpu state >= @state
 *
1806
 * The caller needs to hold cpus read locked while calling this function.
1807 1808 1809
 * The teardown callback is currently not allowed to fail. Think
 * about module removal!
 */
1810
void __cpuhp_remove_state_cpuslocked(enum cpuhp_state state, bool invoke)
1811
{
1812
	struct cpuhp_step *sp = cpuhp_get_step(state);
1813 1814 1815 1816
	int cpu;

	BUG_ON(cpuhp_cb_check(state));

1817
	lockdep_assert_cpus_held();
1818

1819
	mutex_lock(&cpuhp_state_mutex);
1820 1821 1822 1823 1824 1825 1826
	if (sp->multi_instance) {
		WARN(!hlist_empty(&sp->list),
		     "Error: Removing state %d which has instances left.\n",
		     state);
		goto remove;
	}

1827
	if (!invoke || !cpuhp_get_teardown_cb(state))
1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839
		goto remove;

	/*
	 * Call the teardown callback for each present cpu depending
	 * on the hotplug state of the cpu. This function is not
	 * allowed to fail currently!
	 */
	for_each_present_cpu(cpu) {
		struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
		int cpustate = st->state;

		if (cpustate >= state)
1840
			cpuhp_issue_call(cpu, state, false, NULL);
1841 1842
	}
remove:
1843
	cpuhp_store_callbacks(state, NULL, NULL, NULL, false);
1844
	mutex_unlock(&cpuhp_state_mutex);
1845 1846 1847 1848 1849 1850 1851
}
EXPORT_SYMBOL(__cpuhp_remove_state_cpuslocked);

void __cpuhp_remove_state(enum cpuhp_state state, bool invoke)
{
	cpus_read_lock();
	__cpuhp_remove_state_cpuslocked(state, invoke);
1852
	cpus_read_unlock();
1853 1854 1855
}
EXPORT_SYMBOL(__cpuhp_remove_state);

1856 1857 1858 1859 1860 1861 1862 1863 1864 1865
#if defined(CONFIG_SYSFS) && defined(CONFIG_HOTPLUG_CPU)
static ssize_t show_cpuhp_state(struct device *dev,
				struct device_attribute *attr, char *buf)
{
	struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, dev->id);

	return sprintf(buf, "%d\n", st->state);
}
static DEVICE_ATTR(state, 0444, show_cpuhp_state, NULL);

1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894
static ssize_t write_cpuhp_target(struct device *dev,
				  struct device_attribute *attr,
				  const char *buf, size_t count)
{
	struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, dev->id);
	struct cpuhp_step *sp;
	int target, ret;

	ret = kstrtoint(buf, 10, &target);
	if (ret)
		return ret;

#ifdef CONFIG_CPU_HOTPLUG_STATE_CONTROL
	if (target < CPUHP_OFFLINE || target > CPUHP_ONLINE)
		return -EINVAL;
#else
	if (target != CPUHP_OFFLINE && target != CPUHP_ONLINE)
		return -EINVAL;
#endif

	ret = lock_device_hotplug_sysfs();
	if (ret)
		return ret;

	mutex_lock(&cpuhp_state_mutex);
	sp = cpuhp_get_step(target);
	ret = !sp->name || sp->cant_stop ? -EINVAL : 0;
	mutex_unlock(&cpuhp_state_mutex);
	if (ret)
1895
		goto out;
1896 1897 1898 1899 1900

	if (st->state < target)
		ret = do_cpu_up(dev->id, target);
	else
		ret = do_cpu_down(dev->id, target);
1901
out:
1902 1903 1904 1905
	unlock_device_hotplug();
	return ret ? ret : count;
}

1906 1907 1908 1909 1910 1911 1912
static ssize_t show_cpuhp_target(struct device *dev,
				 struct device_attribute *attr, char *buf)
{
	struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, dev->id);

	return sprintf(buf, "%d\n", st->target);
}
1913
static DEVICE_ATTR(target, 0644, show_cpuhp_target, write_cpuhp_target);
1914

1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959

static ssize_t write_cpuhp_fail(struct device *dev,
				struct device_attribute *attr,
				const char *buf, size_t count)
{
	struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, dev->id);
	struct cpuhp_step *sp;
	int fail, ret;

	ret = kstrtoint(buf, 10, &fail);
	if (ret)
		return ret;

	/*
	 * Cannot fail STARTING/DYING callbacks.
	 */
	if (cpuhp_is_atomic_state(fail))
		return -EINVAL;

	/*
	 * Cannot fail anything that doesn't have callbacks.
	 */
	mutex_lock(&cpuhp_state_mutex);
	sp = cpuhp_get_step(fail);
	if (!sp->startup.single && !sp->teardown.single)
		ret = -EINVAL;
	mutex_unlock(&cpuhp_state_mutex);
	if (ret)
		return ret;

	st->fail = fail;

	return count;
}

static ssize_t show_cpuhp_fail(struct device *dev,
			       struct device_attribute *attr, char *buf)
{
	struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, dev->id);

	return sprintf(buf, "%d\n", st->fail);
}

static DEVICE_ATTR(fail, 0644, show_cpuhp_fail, write_cpuhp_fail);

1960 1961 1962
static struct attribute *cpuhp_cpu_attrs[] = {
	&dev_attr_state.attr,
	&dev_attr_target.attr,
1963
	&dev_attr_fail.attr,
1964 1965 1966
	NULL
};

1967
static const struct attribute_group cpuhp_cpu_attr_group = {
1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979
	.attrs = cpuhp_cpu_attrs,
	.name = "hotplug",
	NULL
};

static ssize_t show_cpuhp_states(struct device *dev,
				 struct device_attribute *attr, char *buf)
{
	ssize_t cur, res = 0;
	int i;

	mutex_lock(&cpuhp_state_mutex);
1980
	for (i = CPUHP_OFFLINE; i <= CPUHP_ONLINE; i++) {
1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998
		struct cpuhp_step *sp = cpuhp_get_step(i);

		if (sp->name) {
			cur = sprintf(buf, "%3d: %s\n", i, sp->name);
			buf += cur;
			res += cur;
		}
	}
	mutex_unlock(&cpuhp_state_mutex);
	return res;
}
static DEVICE_ATTR(states, 0444, show_cpuhp_states, NULL);

static struct attribute *cpuhp_cpu_root_attrs[] = {
	&dev_attr_states.attr,
	NULL
};

1999
static const struct attribute_group cpuhp_cpu_root_attr_group = {
2000 2001 2002 2003 2004
	.attrs = cpuhp_cpu_root_attrs,
	.name = "hotplug",
	NULL
};

2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
#ifdef CONFIG_HOTPLUG_SMT

static const char *smt_states[] = {
	[CPU_SMT_ENABLED]		= "on",
	[CPU_SMT_DISABLED]		= "off",
	[CPU_SMT_FORCE_DISABLED]	= "forceoff",
	[CPU_SMT_NOT_SUPPORTED]		= "notsupported",
};

static ssize_t
show_smt_control(struct device *dev, struct device_attribute *attr, char *buf)
{
	return snprintf(buf, PAGE_SIZE - 2, "%s\n", smt_states[cpu_smt_control]);
}

static void cpuhp_offline_cpu_device(unsigned int cpu)
{
	struct device *dev = get_cpu_device(cpu);

	dev->offline = true;
	/* Tell user space about the state change */
	kobject_uevent(&dev->kobj, KOBJ_OFFLINE);
}

2029 2030 2031 2032 2033 2034 2035 2036 2037
static void cpuhp_online_cpu_device(unsigned int cpu)
{
	struct device *dev = get_cpu_device(cpu);

	dev->offline = false;
	/* Tell user space about the state change */
	kobject_uevent(&dev->kobj, KOBJ_ONLINE);
}

2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063
static int cpuhp_smt_disable(enum cpuhp_smt_control ctrlval)
{
	int cpu, ret = 0;

	cpu_maps_update_begin();
	for_each_online_cpu(cpu) {
		if (topology_is_primary_thread(cpu))
			continue;
		ret = cpu_down_maps_locked(cpu, CPUHP_OFFLINE);
		if (ret)
			break;
		/*
		 * As this needs to hold the cpu maps lock it's impossible
		 * to call device_offline() because that ends up calling
		 * cpu_down() which takes cpu maps lock. cpu maps lock
		 * needs to be held as this might race against in kernel
		 * abusers of the hotplug machinery (thermal management).
		 *
		 * So nothing would update device:offline state. That would
		 * leave the sysfs entry stale and prevent onlining after
		 * smt control has been changed to 'off' again. This is
		 * called under the sysfs hotplug lock, so it is properly
		 * serialized against the regular offline usage.
		 */
		cpuhp_offline_cpu_device(cpu);
	}
2064
	if (!ret) {
2065
		cpu_smt_control = ctrlval;
2066 2067
		arch_smt_update();
	}
2068 2069 2070 2071
	cpu_maps_update_done();
	return ret;
}

2072
static int cpuhp_smt_enable(void)
2073
{
2074 2075
	int cpu, ret = 0;

2076 2077
	cpu_maps_update_begin();
	cpu_smt_control = CPU_SMT_ENABLED;
2078
	arch_smt_update();
2079 2080 2081 2082 2083 2084 2085 2086 2087 2088
	for_each_present_cpu(cpu) {
		/* Skip online CPUs and CPUs on offline nodes */
		if (cpu_online(cpu) || !node_online(cpu_to_node(cpu)))
			continue;
		ret = _cpu_up(cpu, 0, CPUHP_ONLINE);
		if (ret)
			break;
		/* See comment in cpuhp_smt_disable() */
		cpuhp_online_cpu_device(cpu);
	}
2089
	cpu_maps_update_done();
2090
	return ret;
2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120
}

static ssize_t
store_smt_control(struct device *dev, struct device_attribute *attr,
		  const char *buf, size_t count)
{
	int ctrlval, ret;

	if (sysfs_streq(buf, "on"))
		ctrlval = CPU_SMT_ENABLED;
	else if (sysfs_streq(buf, "off"))
		ctrlval = CPU_SMT_DISABLED;
	else if (sysfs_streq(buf, "forceoff"))
		ctrlval = CPU_SMT_FORCE_DISABLED;
	else
		return -EINVAL;

	if (cpu_smt_control == CPU_SMT_FORCE_DISABLED)
		return -EPERM;

	if (cpu_smt_control == CPU_SMT_NOT_SUPPORTED)
		return -ENODEV;

	ret = lock_device_hotplug_sysfs();
	if (ret)
		return ret;

	if (ctrlval != cpu_smt_control) {
		switch (ctrlval) {
		case CPU_SMT_ENABLED:
2121
			ret = cpuhp_smt_enable();
2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165
			break;
		case CPU_SMT_DISABLED:
		case CPU_SMT_FORCE_DISABLED:
			ret = cpuhp_smt_disable(ctrlval);
			break;
		}
	}

	unlock_device_hotplug();
	return ret ? ret : count;
}
static DEVICE_ATTR(control, 0644, show_smt_control, store_smt_control);

static ssize_t
show_smt_active(struct device *dev, struct device_attribute *attr, char *buf)
{
	bool active = topology_max_smt_threads() > 1;

	return snprintf(buf, PAGE_SIZE - 2, "%d\n", active);
}
static DEVICE_ATTR(active, 0444, show_smt_active, NULL);

static struct attribute *cpuhp_smt_attrs[] = {
	&dev_attr_control.attr,
	&dev_attr_active.attr,
	NULL
};

static const struct attribute_group cpuhp_smt_attr_group = {
	.attrs = cpuhp_smt_attrs,
	.name = "smt",
	NULL
};

static int __init cpu_smt_state_init(void)
{
	return sysfs_create_group(&cpu_subsys.dev_root->kobj,
				  &cpuhp_smt_attr_group);
}

#else
static inline int cpu_smt_state_init(void) { return 0; }
#endif

2166 2167 2168 2169
static int __init cpuhp_sysfs_init(void)
{
	int cpu, ret;

2170 2171 2172 2173
	ret = cpu_smt_state_init();
	if (ret)
		return ret;

2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192
	ret = sysfs_create_group(&cpu_subsys.dev_root->kobj,
				 &cpuhp_cpu_root_attr_group);
	if (ret)
		return ret;

	for_each_possible_cpu(cpu) {
		struct device *dev = get_cpu_device(cpu);

		if (!dev)
			continue;
		ret = sysfs_create_group(&dev->kobj, &cpuhp_cpu_attr_group);
		if (ret)
			return ret;
	}
	return 0;
}
device_initcall(cpuhp_sysfs_init);
#endif

2193 2194 2195 2196
/*
 * cpu_bit_bitmap[] is a special, "compressed" data structure that
 * represents all NR_CPUS bits binary values of 1<<nr.
 *
R
Rusty Russell 已提交
2197
 * It is used by cpumask_of() to get a constant address to a CPU
2198 2199
 * mask value that has a single bit set only.
 */
2200

2201
/* cpu_bit_bitmap[0] is empty - so we can back into it */
2202
#define MASK_DECLARE_1(x)	[x+1][0] = (1UL << (x))
2203 2204 2205
#define MASK_DECLARE_2(x)	MASK_DECLARE_1(x), MASK_DECLARE_1(x+1)
#define MASK_DECLARE_4(x)	MASK_DECLARE_2(x), MASK_DECLARE_2(x+2)
#define MASK_DECLARE_8(x)	MASK_DECLARE_4(x), MASK_DECLARE_4(x+4)
2206

2207 2208 2209 2210 2211 2212 2213
const unsigned long cpu_bit_bitmap[BITS_PER_LONG+1][BITS_TO_LONGS(NR_CPUS)] = {

	MASK_DECLARE_8(0),	MASK_DECLARE_8(8),
	MASK_DECLARE_8(16),	MASK_DECLARE_8(24),
#if BITS_PER_LONG > 32
	MASK_DECLARE_8(32),	MASK_DECLARE_8(40),
	MASK_DECLARE_8(48),	MASK_DECLARE_8(56),
2214 2215
#endif
};
2216
EXPORT_SYMBOL_GPL(cpu_bit_bitmap);
2217 2218 2219

const DECLARE_BITMAP(cpu_all_bits, NR_CPUS) = CPU_BITS_ALL;
EXPORT_SYMBOL(cpu_all_bits);
2220 2221

#ifdef CONFIG_INIT_ALL_POSSIBLE
2222
struct cpumask __cpu_possible_mask __read_mostly
2223
	= {CPU_BITS_ALL};
2224
#else
2225
struct cpumask __cpu_possible_mask __read_mostly;
2226
#endif
2227
EXPORT_SYMBOL(__cpu_possible_mask);
2228

2229 2230
struct cpumask __cpu_online_mask __read_mostly;
EXPORT_SYMBOL(__cpu_online_mask);
2231

2232 2233
struct cpumask __cpu_present_mask __read_mostly;
EXPORT_SYMBOL(__cpu_present_mask);
2234

2235 2236
struct cpumask __cpu_active_mask __read_mostly;
EXPORT_SYMBOL(__cpu_active_mask);
2237 2238 2239

void init_cpu_present(const struct cpumask *src)
{
2240
	cpumask_copy(&__cpu_present_mask, src);
2241 2242 2243 2244
}

void init_cpu_possible(const struct cpumask *src)
{
2245
	cpumask_copy(&__cpu_possible_mask, src);
2246 2247 2248 2249
}

void init_cpu_online(const struct cpumask *src)
{
2250
	cpumask_copy(&__cpu_online_mask, src);
2251
}
2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264

/*
 * Activate the first processor.
 */
void __init boot_cpu_init(void)
{
	int cpu = smp_processor_id();

	/* Mark the boot cpu "present", "online" etc for SMP and UP case */
	set_cpu_online(cpu, true);
	set_cpu_active(cpu, true);
	set_cpu_present(cpu, true);
	set_cpu_possible(cpu, true);
2265 2266 2267 2268

#ifdef CONFIG_SMP
	__boot_cpu_id = cpu;
#endif
2269 2270 2271 2272 2273
}

/*
 * Must be called _AFTER_ setting up the per_cpu areas
 */
2274
void __init boot_cpu_hotplug_init(void)
2275
{
2276
#ifdef CONFIG_SMP
2277
	this_cpu_write(cpuhp_state.booted_once, true);
2278
#endif
2279
	this_cpu_write(cpuhp_state.state, CPUHP_ONLINE);
2280
}
2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295

enum cpu_mitigations cpu_mitigations __ro_after_init = CPU_MITIGATIONS_AUTO;

static int __init mitigations_parse_cmdline(char *arg)
{
	if (!strcmp(arg, "off"))
		cpu_mitigations = CPU_MITIGATIONS_OFF;
	else if (!strcmp(arg, "auto"))
		cpu_mitigations = CPU_MITIGATIONS_AUTO;
	else if (!strcmp(arg, "auto,nosmt"))
		cpu_mitigations = CPU_MITIGATIONS_AUTO_NOSMT;

	return 0;
}
early_param("mitigations", mitigations_parse_cmdline);