Blame view
kernel/stop_machine.c
12.9 KB
1142d8102
|
1 2 3 4 5 6 7 8 9 |
/* * kernel/stop_machine.c * * Copyright (C) 2008, 2005 IBM Corporation. * Copyright (C) 2008, 2005 Rusty Russell rusty@rustcorp.com.au * Copyright (C) 2010 SUSE Linux Products GmbH * Copyright (C) 2010 Tejun Heo <tj@kernel.org> * * This file is released under the GPLv2 and any later version. |
e5582ca21
|
10 |
*/ |
1142d8102
|
11 |
#include <linux/completion.h> |
1da177e4c
|
12 |
#include <linux/cpu.h> |
1142d8102
|
13 |
#include <linux/init.h> |
ee527cd3a
|
14 15 |
#include <linux/kthread.h> #include <linux/module.h> |
1142d8102
|
16 |
#include <linux/percpu.h> |
ee527cd3a
|
17 18 |
#include <linux/sched.h> #include <linux/stop_machine.h> |
a12bb4447
|
19 |
#include <linux/interrupt.h> |
1142d8102
|
20 |
#include <linux/kallsyms.h> |
a12bb4447
|
21 |
|
1da177e4c
|
22 |
#include <asm/atomic.h> |
1142d8102
|
23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 |
/* * Structure to determine completion condition and record errors. May * be shared by works on different cpus. */ struct cpu_stop_done { atomic_t nr_todo; /* nr left to execute */ bool executed; /* actually executed? */ int ret; /* collected return value */ struct completion completion; /* fired if nr_todo reaches 0 */ }; /* the actual stopper, one per every possible cpu, enabled on online cpus */ struct cpu_stopper { spinlock_t lock; |
878ae1274
|
38 |
bool enabled; /* is this stopper enabled? */ |
1142d8102
|
39 40 |
struct list_head works; /* list of pending works */ struct task_struct *thread; /* stopper thread */ |
1142d8102
|
41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 |
}; static DEFINE_PER_CPU(struct cpu_stopper, cpu_stopper); static void cpu_stop_init_done(struct cpu_stop_done *done, unsigned int nr_todo) { memset(done, 0, sizeof(*done)); atomic_set(&done->nr_todo, nr_todo); init_completion(&done->completion); } /* signal completion unless @done is NULL */ static void cpu_stop_signal_done(struct cpu_stop_done *done, bool executed) { if (done) { if (executed) done->executed = true; if (atomic_dec_and_test(&done->nr_todo)) complete(&done->completion); } } /* queue @work to @stopper. if offline, @work is completed immediately */ static void cpu_stop_queue_work(struct cpu_stopper *stopper, struct cpu_stop_work *work) { unsigned long flags; spin_lock_irqsave(&stopper->lock, flags); if (stopper->enabled) { list_add_tail(&work->list, &stopper->works); wake_up_process(stopper->thread); } else cpu_stop_signal_done(work->done, false); spin_unlock_irqrestore(&stopper->lock, flags); } /** * stop_one_cpu - stop a cpu * @cpu: cpu to stop * @fn: function to execute * @arg: argument to @fn * * Execute @fn(@arg) on @cpu. @fn is run in a process context with * the highest priority preempting any task on the cpu and * monopolizing it. This function returns after the execution is * complete. * * This function doesn't guarantee @cpu stays online till @fn * completes. If @cpu goes down in the middle, execution may happen * partially or fully on different cpus. @fn should either be ready * for that or the caller should ensure that @cpu stays online until * this function completes. * * CONTEXT: * Might sleep. * * RETURNS: * -ENOENT if @fn(@arg) was not executed because @cpu was offline; * otherwise, the return value of @fn. */ int stop_one_cpu(unsigned int cpu, cpu_stop_fn_t fn, void *arg) { struct cpu_stop_done done; struct cpu_stop_work work = { .fn = fn, .arg = arg, .done = &done }; cpu_stop_init_done(&done, 1); cpu_stop_queue_work(&per_cpu(cpu_stopper, cpu), &work); wait_for_completion(&done.completion); return done.executed ? done.ret : -ENOENT; } /** * stop_one_cpu_nowait - stop a cpu but don't wait for completion * @cpu: cpu to stop * @fn: function to execute * @arg: argument to @fn * * Similar to stop_one_cpu() but doesn't wait for completion. The * caller is responsible for ensuring @work_buf is currently unused * and will remain untouched until stopper starts executing @fn. * * CONTEXT: * Don't care. */ void stop_one_cpu_nowait(unsigned int cpu, cpu_stop_fn_t fn, void *arg, struct cpu_stop_work *work_buf) { *work_buf = (struct cpu_stop_work){ .fn = fn, .arg = arg, }; cpu_stop_queue_work(&per_cpu(cpu_stopper, cpu), work_buf); } /* static data for stop_cpus */ static DEFINE_MUTEX(stop_cpus_mutex); static DEFINE_PER_CPU(struct cpu_stop_work, stop_cpus_work); int __stop_cpus(const struct cpumask *cpumask, cpu_stop_fn_t fn, void *arg) { struct cpu_stop_work *work; struct cpu_stop_done done; unsigned int cpu; /* initialize works and done */ for_each_cpu(cpu, cpumask) { work = &per_cpu(stop_cpus_work, cpu); work->fn = fn; work->arg = arg; work->done = &done; } cpu_stop_init_done(&done, cpumask_weight(cpumask)); /* * Disable preemption while queueing to avoid getting * preempted by a stopper which might wait for other stoppers * to enter @fn which can lead to deadlock. */ preempt_disable(); for_each_cpu(cpu, cpumask) cpu_stop_queue_work(&per_cpu(cpu_stopper, cpu), &per_cpu(stop_cpus_work, cpu)); preempt_enable(); wait_for_completion(&done.completion); return done.executed ? done.ret : -ENOENT; } /** * stop_cpus - stop multiple cpus * @cpumask: cpus to stop * @fn: function to execute * @arg: argument to @fn * * Execute @fn(@arg) on online cpus in @cpumask. On each target cpu, * @fn is run in a process context with the highest priority * preempting any task on the cpu and monopolizing it. This function * returns after all executions are complete. * * This function doesn't guarantee the cpus in @cpumask stay online * till @fn completes. If some cpus go down in the middle, execution * on the cpu may happen partially or fully on different cpus. @fn * should either be ready for that or the caller should ensure that * the cpus stay online until this function completes. * * All stop_cpus() calls are serialized making it safe for @fn to wait * for all cpus to start executing it. * * CONTEXT: * Might sleep. * * RETURNS: * -ENOENT if @fn(@arg) was not executed at all because all cpus in * @cpumask were offline; otherwise, 0 if all executions of @fn * returned 0, any non zero return value if any returned non zero. */ int stop_cpus(const struct cpumask *cpumask, cpu_stop_fn_t fn, void *arg) { int ret; /* static works are used, process one request at a time */ mutex_lock(&stop_cpus_mutex); ret = __stop_cpus(cpumask, fn, arg); mutex_unlock(&stop_cpus_mutex); return ret; } /** * try_stop_cpus - try to stop multiple cpus * @cpumask: cpus to stop * @fn: function to execute * @arg: argument to @fn * * Identical to stop_cpus() except that it fails with -EAGAIN if * someone else is already using the facility. * * CONTEXT: * Might sleep. * * RETURNS: * -EAGAIN if someone else is already stopping cpus, -ENOENT if * @fn(@arg) was not executed at all because all cpus in @cpumask were * offline; otherwise, 0 if all executions of @fn returned 0, any non * zero return value if any returned non zero. */ int try_stop_cpus(const struct cpumask *cpumask, cpu_stop_fn_t fn, void *arg) { int ret; /* static works are used, process one request at a time */ if (!mutex_trylock(&stop_cpus_mutex)) return -EAGAIN; ret = __stop_cpus(cpumask, fn, arg); mutex_unlock(&stop_cpus_mutex); return ret; } static int cpu_stopper_thread(void *data) { struct cpu_stopper *stopper = data; struct cpu_stop_work *work; int ret; repeat: set_current_state(TASK_INTERRUPTIBLE); /* mb paired w/ kthread_stop */ if (kthread_should_stop()) { __set_current_state(TASK_RUNNING); return 0; } work = NULL; spin_lock_irq(&stopper->lock); if (!list_empty(&stopper->works)) { work = list_first_entry(&stopper->works, struct cpu_stop_work, list); list_del_init(&work->list); } spin_unlock_irq(&stopper->lock); if (work) { cpu_stop_fn_t fn = work->fn; void *arg = work->arg; struct cpu_stop_done *done = work->done; |
ca51c5a76
|
265 |
char ksym_buf[KSYM_NAME_LEN] __maybe_unused; |
1142d8102
|
266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 |
__set_current_state(TASK_RUNNING); /* cpu stop callbacks are not allowed to sleep */ preempt_disable(); ret = fn(arg); if (ret) done->ret = ret; /* restore preemption and check it's still balanced */ preempt_enable(); WARN_ONCE(preempt_count(), "cpu_stop: %s(%p) leaked preempt count ", kallsyms_lookup((unsigned long)fn, NULL, NULL, NULL, ksym_buf), arg); cpu_stop_signal_done(done, true); } else schedule(); goto repeat; } |
34f971f6f
|
290 |
extern void sched_set_stop_task(int cpu, struct task_struct *stop); |
1142d8102
|
291 292 293 294 |
/* manage stopper for a cpu, mostly lifted from sched migration thread mgmt */ static int __cpuinit cpu_stop_cpu_callback(struct notifier_block *nfb, unsigned long action, void *hcpu) { |
1142d8102
|
295 296 |
unsigned int cpu = (unsigned long)hcpu; struct cpu_stopper *stopper = &per_cpu(cpu_stopper, cpu); |
1142d8102
|
297 298 299 300 301 302 |
struct task_struct *p; switch (action & ~CPU_TASKS_FROZEN) { case CPU_UP_PREPARE: BUG_ON(stopper->thread || stopper->enabled || !list_empty(&stopper->works)); |
969c79215
|
303 |
p = kthread_create(cpu_stopper_thread, stopper, "migration/%d", |
1142d8102
|
304 305 |
cpu); if (IS_ERR(p)) |
4ce6494db
|
306 |
return notifier_from_errno(PTR_ERR(p)); |
1142d8102
|
307 |
get_task_struct(p); |
34f971f6f
|
308 309 |
kthread_bind(p, cpu); sched_set_stop_task(cpu, p); |
1142d8102
|
310 311 312 313 |
stopper->thread = p; break; case CPU_ONLINE: |
1142d8102
|
314 315 316 317 318 319 320 321 322 323 |
/* strictly unnecessary, as first user will wake it */ wake_up_process(stopper->thread); /* mark enabled */ spin_lock_irq(&stopper->lock); stopper->enabled = true; spin_unlock_irq(&stopper->lock); break; #ifdef CONFIG_HOTPLUG_CPU case CPU_UP_CANCELED: |
54e88fad2
|
324 |
case CPU_POST_DEAD: |
9c6f7e43b
|
325 326 |
{ struct cpu_stop_work *work; |
34f971f6f
|
327 |
sched_set_stop_task(cpu, NULL); |
1142d8102
|
328 329 330 331 332 333 334 335 336 337 338 339 |
/* kill the stopper */ kthread_stop(stopper->thread); /* drain remaining works */ spin_lock_irq(&stopper->lock); list_for_each_entry(work, &stopper->works, list) cpu_stop_signal_done(work->done, false); stopper->enabled = false; spin_unlock_irq(&stopper->lock); /* release the stopper */ put_task_struct(stopper->thread); stopper->thread = NULL; break; |
9c6f7e43b
|
340 |
} |
1142d8102
|
341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 |
#endif } return NOTIFY_OK; } /* * Give it a higher priority so that cpu stopper is available to other * cpu notifiers. It currently shares the same priority as sched * migration_notifier. */ static struct notifier_block __cpuinitdata cpu_stop_cpu_notifier = { .notifier_call = cpu_stop_cpu_callback, .priority = 10, }; static int __init cpu_stop_init(void) { void *bcpu = (void *)(long)smp_processor_id(); unsigned int cpu; int err; for_each_possible_cpu(cpu) { struct cpu_stopper *stopper = &per_cpu(cpu_stopper, cpu); spin_lock_init(&stopper->lock); INIT_LIST_HEAD(&stopper->works); } /* start one for the boot cpu */ err = cpu_stop_cpu_callback(&cpu_stop_cpu_notifier, CPU_UP_PREPARE, bcpu); |
4ce6494db
|
373 |
BUG_ON(err != NOTIFY_OK); |
1142d8102
|
374 375 376 377 378 379 |
cpu_stop_cpu_callback(&cpu_stop_cpu_notifier, CPU_ONLINE, bcpu); register_cpu_notifier(&cpu_stop_cpu_notifier); return 0; } early_initcall(cpu_stop_init); |
1da177e4c
|
380 |
|
bbf1bb3ee
|
381 |
#ifdef CONFIG_STOP_MACHINE |
ffdb5976c
|
382 |
/* This controls the threads on each CPU. */ |
1da177e4c
|
383 |
enum stopmachine_state { |
ffdb5976c
|
384 385 386 |
/* Dummy starting state for thread. */ STOPMACHINE_NONE, /* Awaiting everyone to be scheduled. */ |
1da177e4c
|
387 |
STOPMACHINE_PREPARE, |
ffdb5976c
|
388 |
/* Disable interrupts. */ |
1da177e4c
|
389 |
STOPMACHINE_DISABLE_IRQ, |
ffdb5976c
|
390 |
/* Run the function */ |
5c2aed622
|
391 |
STOPMACHINE_RUN, |
ffdb5976c
|
392 |
/* Exit */ |
1da177e4c
|
393 394 |
STOPMACHINE_EXIT, }; |
5c2aed622
|
395 |
struct stop_machine_data { |
3fc1f1e27
|
396 397 398 399 400 401 402 403 |
int (*fn)(void *); void *data; /* Like num_online_cpus(), but hotplug cpu uses us, so we need this. */ unsigned int num_threads; const struct cpumask *active_cpus; enum stopmachine_state state; atomic_t thread_ack; |
ffdb5976c
|
404 |
}; |
5c2aed622
|
405 |
|
3fc1f1e27
|
406 407 |
static void set_state(struct stop_machine_data *smdata, enum stopmachine_state newstate) |
1da177e4c
|
408 |
{ |
ffdb5976c
|
409 |
/* Reset ack counter. */ |
3fc1f1e27
|
410 |
atomic_set(&smdata->thread_ack, smdata->num_threads); |
ffdb5976c
|
411 |
smp_wmb(); |
3fc1f1e27
|
412 |
smdata->state = newstate; |
1da177e4c
|
413 |
} |
ffdb5976c
|
414 |
/* Last one to ack a state moves to the next state. */ |
3fc1f1e27
|
415 |
static void ack_state(struct stop_machine_data *smdata) |
1da177e4c
|
416 |
{ |
3fc1f1e27
|
417 418 |
if (atomic_dec_and_test(&smdata->thread_ack)) set_state(smdata, smdata->state + 1); |
1da177e4c
|
419 |
} |
3fc1f1e27
|
420 421 |
/* This is the cpu_stop function which stops the CPU. */ static int stop_machine_cpu_stop(void *data) |
1da177e4c
|
422 |
{ |
3fc1f1e27
|
423 |
struct stop_machine_data *smdata = data; |
ffdb5976c
|
424 |
enum stopmachine_state curstate = STOPMACHINE_NONE; |
3fc1f1e27
|
425 426 427 428 429 430 431 |
int cpu = smp_processor_id(), err = 0; bool is_active; if (!smdata->active_cpus) is_active = cpu == cpumask_first(cpu_online_mask); else is_active = cpumask_test_cpu(cpu, smdata->active_cpus); |
c9583e55f
|
432 |
|
ffdb5976c
|
433 434 435 |
/* Simple state machine */ do { /* Chill out and ensure we re-read stopmachine_state. */ |
3401a61e1
|
436 |
cpu_relax(); |
3fc1f1e27
|
437 438 |
if (smdata->state != curstate) { curstate = smdata->state; |
ffdb5976c
|
439 440 441 442 443 444 |
switch (curstate) { case STOPMACHINE_DISABLE_IRQ: local_irq_disable(); hard_irq_disable(); break; case STOPMACHINE_RUN: |
3fc1f1e27
|
445 446 |
if (is_active) err = smdata->fn(smdata->data); |
ffdb5976c
|
447 448 449 450 |
break; default: break; } |
3fc1f1e27
|
451 |
ack_state(smdata); |
ffdb5976c
|
452 453 |
} } while (curstate != STOPMACHINE_EXIT); |
1da177e4c
|
454 |
|
1da177e4c
|
455 |
local_irq_enable(); |
3fc1f1e27
|
456 |
return err; |
1da177e4c
|
457 |
} |
41c7bb958
|
458 |
int __stop_machine(int (*fn)(void *), void *data, const struct cpumask *cpus) |
1da177e4c
|
459 |
{ |
3fc1f1e27
|
460 461 462 463 464 465 466 |
struct stop_machine_data smdata = { .fn = fn, .data = data, .num_threads = num_online_cpus(), .active_cpus = cpus }; /* Set the initial state and stop all online cpus. */ set_state(&smdata, STOPMACHINE_PREPARE); return stop_cpus(cpu_online_mask, stop_machine_cpu_stop, &smdata); |
1da177e4c
|
467 |
} |
41c7bb958
|
468 |
int stop_machine(int (*fn)(void *), void *data, const struct cpumask *cpus) |
1da177e4c
|
469 |
{ |
1da177e4c
|
470 471 472 |
int ret; /* No CPUs can come up or down during this. */ |
86ef5c9a8
|
473 |
get_online_cpus(); |
eeec4fad9
|
474 |
ret = __stop_machine(fn, data, cpus); |
86ef5c9a8
|
475 |
put_online_cpus(); |
1da177e4c
|
476 477 |
return ret; } |
eeec4fad9
|
478 |
EXPORT_SYMBOL_GPL(stop_machine); |
bbf1bb3ee
|
479 480 |
#endif /* CONFIG_STOP_MACHINE */ |