debug.c 12.3 KB
Newer Older
I
Ingo Molnar 已提交
1
/*
2
 * kernel/sched/debug.c
I
Ingo Molnar 已提交
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
 *
 * Print the CFS rbtree
 *
 * Copyright(C) 2007, Red Hat, Inc., Ingo Molnar
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License version 2 as
 * published by the Free Software Foundation.
 */

#include <linux/proc_fs.h>
#include <linux/sched.h>
#include <linux/seq_file.h>
#include <linux/kallsyms.h>
#include <linux/utsname.h>

19 20
#include "sched.h"

21 22
static DEFINE_SPINLOCK(sched_debug_lock);

I
Ingo Molnar 已提交
23 24 25 26 27 28 29 30 31 32 33 34
/*
 * This allows printing both to /proc/sched_debug and
 * to the console
 */
#define SEQ_printf(m, x...)			\
 do {						\
	if (m)					\
		seq_printf(m, x);		\
	else					\
		printk(x);			\
 } while (0)

I
Ingo Molnar 已提交
35 36 37
/*
 * Ease the printing of nsec fields:
 */
I
Ingo Molnar 已提交
38
static long long nsec_high(unsigned long long nsec)
I
Ingo Molnar 已提交
39
{
I
Ingo Molnar 已提交
40
	if ((long long)nsec < 0) {
I
Ingo Molnar 已提交
41 42 43 44 45 46 47 48 49
		nsec = -nsec;
		do_div(nsec, 1000000);
		return -nsec;
	}
	do_div(nsec, 1000000);

	return nsec;
}

I
Ingo Molnar 已提交
50
static unsigned long nsec_low(unsigned long long nsec)
I
Ingo Molnar 已提交
51
{
I
Ingo Molnar 已提交
52
	if ((long long)nsec < 0)
I
Ingo Molnar 已提交
53 54 55 56 57 58 59
		nsec = -nsec;

	return do_div(nsec, 1000000);
}

#define SPLIT_NS(x) nsec_high(x), nsec_low(x)

60
#ifdef CONFIG_FAIR_GROUP_SCHED
61
static void print_cfs_group_stats(struct seq_file *m, int cpu, struct task_group *tg)
62 63 64 65 66 67 68 69
{
	struct sched_entity *se = tg->se[cpu];

#define P(F) \
	SEQ_printf(m, "  .%-30s: %lld\n", #F, (long long)F)
#define PN(F) \
	SEQ_printf(m, "  .%-30s: %lld.%06ld\n", #F, SPLIT_NS((long long)F))

70 71 72 73 74 75 76 77
	if (!se) {
		struct sched_avg *avg = &cpu_rq(cpu)->avg;
		P(avg->runnable_avg_sum);
		P(avg->runnable_avg_period);
		return;
	}


78 79 80 81
	PN(se->exec_start);
	PN(se->vruntime);
	PN(se->sum_exec_runtime);
#ifdef CONFIG_SCHEDSTATS
82 83 84 85 86 87 88 89 90 91
	PN(se->statistics.wait_start);
	PN(se->statistics.sleep_start);
	PN(se->statistics.block_start);
	PN(se->statistics.sleep_max);
	PN(se->statistics.block_max);
	PN(se->statistics.exec_max);
	PN(se->statistics.slice_max);
	PN(se->statistics.wait_max);
	PN(se->statistics.wait_sum);
	P(se->statistics.wait_count);
92 93
#endif
	P(se->load.weight);
94 95 96
#ifdef CONFIG_SMP
	P(se->avg.runnable_avg_sum);
	P(se->avg.runnable_avg_period);
97
	P(se->avg.load_avg_contrib);
98
	P(se->avg.decay_count);
99
#endif
100 101 102 103 104
#undef PN
#undef P
}
#endif

105 106 107 108 109
#ifdef CONFIG_CGROUP_SCHED
static char group_path[PATH_MAX];

static char *task_group_path(struct task_group *tg)
{
110 111 112
	if (autogroup_path(tg, group_path, PATH_MAX))
		return group_path;

113 114 115 116 117 118 119 120 121 122 123 124
	/*
	 * May be NULL if the underlying cgroup isn't fully-created yet
	 */
	if (!tg->css.cgroup) {
		group_path[0] = '\0';
		return group_path;
	}
	cgroup_path(tg->css.cgroup, group_path, PATH_MAX);
	return group_path;
}
#endif

I
Ingo Molnar 已提交
125
static void
126
print_task(struct seq_file *m, struct rq *rq, struct task_struct *p)
I
Ingo Molnar 已提交
127 128 129 130 131 132
{
	if (rq->curr == p)
		SEQ_printf(m, "R");
	else
		SEQ_printf(m, " ");

I
Ingo Molnar 已提交
133
	SEQ_printf(m, "%15s %5d %9Ld.%06ld %9Ld %5d ",
I
Ingo Molnar 已提交
134
		p->comm, p->pid,
I
Ingo Molnar 已提交
135
		SPLIT_NS(p->se.vruntime),
I
Ingo Molnar 已提交
136
		(long long)(p->nvcsw + p->nivcsw),
137
		p->prio);
I
Ingo Molnar 已提交
138
#ifdef CONFIG_SCHEDSTATS
139
	SEQ_printf(m, "%9Ld.%06ld %9Ld.%06ld %9Ld.%06ld",
I
Ingo Molnar 已提交
140 141
		SPLIT_NS(p->se.vruntime),
		SPLIT_NS(p->se.sum_exec_runtime),
142
		SPLIT_NS(p->se.statistics.sum_sleep_runtime));
I
Ingo Molnar 已提交
143
#else
144
	SEQ_printf(m, "%15Ld %15Ld %15Ld.%06ld %15Ld.%06ld %15Ld.%06ld",
I
Ingo Molnar 已提交
145
		0LL, 0LL, 0LL, 0L, 0LL, 0L, 0LL, 0L);
I
Ingo Molnar 已提交
146
#endif
147 148 149
#ifdef CONFIG_CGROUP_SCHED
	SEQ_printf(m, " %s", task_group_path(task_group(p)));
#endif
150 151

	SEQ_printf(m, "\n");
I
Ingo Molnar 已提交
152 153
}

154
static void print_rq(struct seq_file *m, struct rq *rq, int rq_cpu)
I
Ingo Molnar 已提交
155 156
{
	struct task_struct *g, *p;
P
Peter Zijlstra 已提交
157
	unsigned long flags;
I
Ingo Molnar 已提交
158 159 160

	SEQ_printf(m,
	"\nrunnable tasks:\n"
161 162
	"            task   PID         tree-key  switches  prio"
	"     exec-runtime         sum-exec        sum-sleep\n"
163
	"------------------------------------------------------"
164
	"----------------------------------------------------\n");
I
Ingo Molnar 已提交
165

P
Peter Zijlstra 已提交
166
	read_lock_irqsave(&tasklist_lock, flags);
I
Ingo Molnar 已提交
167 168

	do_each_thread(g, p) {
P
Peter Zijlstra 已提交
169
		if (!p->on_rq || task_cpu(p) != rq_cpu)
I
Ingo Molnar 已提交
170 171
			continue;

172
		print_task(m, rq, p);
I
Ingo Molnar 已提交
173 174
	} while_each_thread(g, p);

P
Peter Zijlstra 已提交
175
	read_unlock_irqrestore(&tasklist_lock, flags);
I
Ingo Molnar 已提交
176 177
}

178
void print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
I
Ingo Molnar 已提交
179
{
I
Ingo Molnar 已提交
180 181
	s64 MIN_vruntime = -1, min_vruntime, max_vruntime = -1,
		spread, rq0_min_vruntime, spread0;
182
	struct rq *rq = cpu_rq(cpu);
I
Ingo Molnar 已提交
183 184 185
	struct sched_entity *last;
	unsigned long flags;

186 187 188
#ifdef CONFIG_FAIR_GROUP_SCHED
	SEQ_printf(m, "\ncfs_rq[%d]:%s\n", cpu, task_group_path(cfs_rq->tg));
#else
189
	SEQ_printf(m, "\ncfs_rq[%d]:\n", cpu);
190
#endif
I
Ingo Molnar 已提交
191 192
	SEQ_printf(m, "  .%-30s: %Ld.%06ld\n", "exec_clock",
			SPLIT_NS(cfs_rq->exec_clock));
I
Ingo Molnar 已提交
193

194
	raw_spin_lock_irqsave(&rq->lock, flags);
I
Ingo Molnar 已提交
195
	if (cfs_rq->rb_leftmost)
196
		MIN_vruntime = (__pick_first_entity(cfs_rq))->vruntime;
I
Ingo Molnar 已提交
197 198 199
	last = __pick_last_entity(cfs_rq);
	if (last)
		max_vruntime = last->vruntime;
P
Peter Zijlstra 已提交
200
	min_vruntime = cfs_rq->min_vruntime;
201
	rq0_min_vruntime = cpu_rq(0)->cfs.min_vruntime;
202
	raw_spin_unlock_irqrestore(&rq->lock, flags);
I
Ingo Molnar 已提交
203 204 205 206 207 208
	SEQ_printf(m, "  .%-30s: %Ld.%06ld\n", "MIN_vruntime",
			SPLIT_NS(MIN_vruntime));
	SEQ_printf(m, "  .%-30s: %Ld.%06ld\n", "min_vruntime",
			SPLIT_NS(min_vruntime));
	SEQ_printf(m, "  .%-30s: %Ld.%06ld\n", "max_vruntime",
			SPLIT_NS(max_vruntime));
I
Ingo Molnar 已提交
209
	spread = max_vruntime - MIN_vruntime;
I
Ingo Molnar 已提交
210 211
	SEQ_printf(m, "  .%-30s: %Ld.%06ld\n", "spread",
			SPLIT_NS(spread));
I
Ingo Molnar 已提交
212
	spread0 = min_vruntime - rq0_min_vruntime;
I
Ingo Molnar 已提交
213 214
	SEQ_printf(m, "  .%-30s: %Ld.%06ld\n", "spread0",
			SPLIT_NS(spread0));
P
Peter Zijlstra 已提交
215
	SEQ_printf(m, "  .%-30s: %d\n", "nr_spread_over",
P
Peter Zijlstra 已提交
216
			cfs_rq->nr_spread_over);
217
	SEQ_printf(m, "  .%-30s: %d\n", "nr_running", cfs_rq->nr_running);
P
Peter Zijlstra 已提交
218
	SEQ_printf(m, "  .%-30s: %ld\n", "load", cfs_rq->load.weight);
219 220
#ifdef CONFIG_FAIR_GROUP_SCHED
#ifdef CONFIG_SMP
221 222
	SEQ_printf(m, "  .%-30s: %lld\n", "runnable_load_avg",
			cfs_rq->runnable_load_avg);
223 224
	SEQ_printf(m, "  .%-30s: %lld\n", "blocked_load_avg",
			cfs_rq->blocked_load_avg);
225 226 227 228
	SEQ_printf(m, "  .%-30s: %ld\n", "tg_load_avg",
			atomic64_read(&cfs_rq->tg->load_avg));
	SEQ_printf(m, "  .%-30s: %lld\n", "tg_load_contrib",
			cfs_rq->tg_load_contrib);
229 230 231 232
	SEQ_printf(m, "  .%-30s: %d\n", "tg_runnable_contrib",
			cfs_rq->tg_runnable_contrib);
	SEQ_printf(m, "  .%-30s: %d\n", "tg->runnable_avg",
			atomic_read(&cfs_rq->tg->runnable_avg));
233
#endif
P
Peter Zijlstra 已提交
234

235
	print_cfs_group_stats(m, cpu, cfs_rq->tg);
236
#endif
I
Ingo Molnar 已提交
237 238
}

239 240
void print_rt_rq(struct seq_file *m, int cpu, struct rt_rq *rt_rq)
{
241 242 243
#ifdef CONFIG_RT_GROUP_SCHED
	SEQ_printf(m, "\nrt_rq[%d]:%s\n", cpu, task_group_path(rt_rq->tg));
#else
244
	SEQ_printf(m, "\nrt_rq[%d]:\n", cpu);
245
#endif
246 247 248 249 250 251 252 253 254 255 256 257 258 259 260

#define P(x) \
	SEQ_printf(m, "  .%-30s: %Ld\n", #x, (long long)(rt_rq->x))
#define PN(x) \
	SEQ_printf(m, "  .%-30s: %Ld.%06ld\n", #x, SPLIT_NS(rt_rq->x))

	P(rt_nr_running);
	P(rt_throttled);
	PN(rt_time);
	PN(rt_runtime);

#undef PN
#undef P
}

261 262
extern __read_mostly int sched_clock_running;

263
static void print_cpu(struct seq_file *m, int cpu)
I
Ingo Molnar 已提交
264
{
265
	struct rq *rq = cpu_rq(cpu);
266
	unsigned long flags;
I
Ingo Molnar 已提交
267 268 269 270 271 272 273 274 275 276 277 278

#ifdef CONFIG_X86
	{
		unsigned int freq = cpu_khz ? : 1;

		SEQ_printf(m, "\ncpu#%d, %u.%03u MHz\n",
			   cpu, freq / 1000, (freq % 1000));
	}
#else
	SEQ_printf(m, "\ncpu#%d\n", cpu);
#endif

279 280 281 282 283 284 285 286
#define P(x)								\
do {									\
	if (sizeof(rq->x) == 4)						\
		SEQ_printf(m, "  .%-30s: %ld\n", #x, (long)(rq->x));	\
	else								\
		SEQ_printf(m, "  .%-30s: %Ld\n", #x, (long long)(rq->x));\
} while (0)

I
Ingo Molnar 已提交
287 288
#define PN(x) \
	SEQ_printf(m, "  .%-30s: %Ld.%06ld\n", #x, SPLIT_NS(rq->x))
I
Ingo Molnar 已提交
289 290 291

	P(nr_running);
	SEQ_printf(m, "  .%-30s: %lu\n", "load",
292
		   rq->load.weight);
I
Ingo Molnar 已提交
293 294 295
	P(nr_switches);
	P(nr_load_updates);
	P(nr_uninterruptible);
I
Ingo Molnar 已提交
296
	PN(next_balance);
I
Ingo Molnar 已提交
297
	P(curr->pid);
I
Ingo Molnar 已提交
298
	PN(clock);
I
Ingo Molnar 已提交
299 300 301 302 303 304
	P(cpu_load[0]);
	P(cpu_load[1]);
	P(cpu_load[2]);
	P(cpu_load[3]);
	P(cpu_load[4]);
#undef P
I
Ingo Molnar 已提交
305
#undef PN
I
Ingo Molnar 已提交
306

P
Peter Zijlstra 已提交
307 308
#ifdef CONFIG_SCHEDSTATS
#define P(n) SEQ_printf(m, "  .%-30s: %d\n", #n, rq->n);
M
Mike Galbraith 已提交
309
#define P64(n) SEQ_printf(m, "  .%-30s: %Ld\n", #n, rq->n);
P
Peter Zijlstra 已提交
310 311 312 313 314

	P(yld_count);

	P(sched_count);
	P(sched_goidle);
M
Mike Galbraith 已提交
315 316 317
#ifdef CONFIG_SMP
	P64(avg_idle);
#endif
P
Peter Zijlstra 已提交
318 319 320 321 322

	P(ttwu_count);
	P(ttwu_local);

#undef P
323
#undef P64
P
Peter Zijlstra 已提交
324
#endif
325
	spin_lock_irqsave(&sched_debug_lock, flags);
326
	print_cfs_stats(m, cpu);
327
	print_rt_stats(m, cpu);
I
Ingo Molnar 已提交
328

329
	rcu_read_lock();
330
	print_rq(m, rq, cpu);
331 332
	rcu_read_unlock();
	spin_unlock_irqrestore(&sched_debug_lock, flags);
I
Ingo Molnar 已提交
333 334
}

335 336 337 338 339 340
static const char *sched_tunable_scaling_names[] = {
	"none",
	"logaritmic",
	"linear"
};

I
Ingo Molnar 已提交
341 342
static int sched_debug_show(struct seq_file *m, void *v)
{
343 344
	u64 ktime, sched_clk, cpu_clk;
	unsigned long flags;
I
Ingo Molnar 已提交
345 346
	int cpu;

347 348 349 350 351 352 353
	local_irq_save(flags);
	ktime = ktime_to_ns(ktime_get());
	sched_clk = sched_clock();
	cpu_clk = local_clock();
	local_irq_restore(flags);

	SEQ_printf(m, "Sched Debug Version: v0.10, %s %.*s\n",
I
Ingo Molnar 已提交
354 355 356 357
		init_utsname()->release,
		(int)strcspn(init_utsname()->version, " "),
		init_utsname()->version);

358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373
#define P(x) \
	SEQ_printf(m, "%-40s: %Ld\n", #x, (long long)(x))
#define PN(x) \
	SEQ_printf(m, "%-40s: %Ld.%06ld\n", #x, SPLIT_NS(x))
	PN(ktime);
	PN(sched_clk);
	PN(cpu_clk);
	P(jiffies);
#ifdef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
	P(sched_clock_stable);
#endif
#undef PN
#undef P

	SEQ_printf(m, "\n");
	SEQ_printf(m, "sysctl_sched\n");
I
Ingo Molnar 已提交
374

I
Ingo Molnar 已提交
375
#define P(x) \
376
	SEQ_printf(m, "  .%-40s: %Ld\n", #x, (long long)(x))
I
Ingo Molnar 已提交
377
#define PN(x) \
378
	SEQ_printf(m, "  .%-40s: %Ld.%06ld\n", #x, SPLIT_NS(x))
I
Ingo Molnar 已提交
379
	PN(sysctl_sched_latency);
380
	PN(sysctl_sched_min_granularity);
I
Ingo Molnar 已提交
381
	PN(sysctl_sched_wakeup_granularity);
382
	P(sysctl_sched_child_runs_first);
I
Ingo Molnar 已提交
383 384 385 386
	P(sysctl_sched_features);
#undef PN
#undef P

387 388 389 390
	SEQ_printf(m, "  .%-40s: %d (%s)\n", "sysctl_sched_tunable_scaling",
		sysctl_sched_tunable_scaling,
		sched_tunable_scaling_names[sysctl_sched_tunable_scaling]);

I
Ingo Molnar 已提交
391
	for_each_online_cpu(cpu)
392
		print_cpu(m, cpu);
I
Ingo Molnar 已提交
393 394 395 396 397 398

	SEQ_printf(m, "\n");

	return 0;
}

399
void sysrq_sched_debug_show(void)
I
Ingo Molnar 已提交
400 401 402 403 404 405 406 407 408
{
	sched_debug_show(NULL, NULL);
}

static int sched_debug_open(struct inode *inode, struct file *filp)
{
	return single_open(filp, sched_debug_show, NULL);
}

409
static const struct file_operations sched_debug_fops = {
I
Ingo Molnar 已提交
410 411 412
	.open		= sched_debug_open,
	.read		= seq_read,
	.llseek		= seq_lseek,
413
	.release	= single_release,
I
Ingo Molnar 已提交
414 415 416 417 418 419
};

static int __init init_sched_debug_procfs(void)
{
	struct proc_dir_entry *pe;

420
	pe = proc_create("sched_debug", 0444, NULL, &sched_debug_fops);
I
Ingo Molnar 已提交
421 422 423 424 425 426 427 428 429
	if (!pe)
		return -ENOMEM;
	return 0;
}

__initcall(init_sched_debug_procfs);

void proc_sched_show_task(struct task_struct *p, struct seq_file *m)
{
430
	unsigned long nr_switches;
I
Ingo Molnar 已提交
431

432 433
	SEQ_printf(m, "%s (%d, #threads: %d)\n", p->comm, p->pid,
						get_nr_threads(p));
434 435
	SEQ_printf(m,
		"---------------------------------------------------------\n");
436 437
#define __P(F) \
	SEQ_printf(m, "%-35s:%21Ld\n", #F, (long long)F)
I
Ingo Molnar 已提交
438
#define P(F) \
439
	SEQ_printf(m, "%-35s:%21Ld\n", #F, (long long)p->F)
440 441
#define __PN(F) \
	SEQ_printf(m, "%-35s:%14Ld.%06ld\n", #F, SPLIT_NS((long long)F))
I
Ingo Molnar 已提交
442
#define PN(F) \
443
	SEQ_printf(m, "%-35s:%14Ld.%06ld\n", #F, SPLIT_NS((long long)p->F))
I
Ingo Molnar 已提交
444

I
Ingo Molnar 已提交
445 446 447
	PN(se.exec_start);
	PN(se.vruntime);
	PN(se.sum_exec_runtime);
I
Ingo Molnar 已提交
448

449 450
	nr_switches = p->nvcsw + p->nivcsw;

I
Ingo Molnar 已提交
451
#ifdef CONFIG_SCHEDSTATS
452 453 454 455 456 457 458 459 460 461 462 463
	PN(se.statistics.wait_start);
	PN(se.statistics.sleep_start);
	PN(se.statistics.block_start);
	PN(se.statistics.sleep_max);
	PN(se.statistics.block_max);
	PN(se.statistics.exec_max);
	PN(se.statistics.slice_max);
	PN(se.statistics.wait_max);
	PN(se.statistics.wait_sum);
	P(se.statistics.wait_count);
	PN(se.statistics.iowait_sum);
	P(se.statistics.iowait_count);
464
	P(se.nr_migrations);
465 466 467 468 469 470 471 472 473 474 475 476 477 478
	P(se.statistics.nr_migrations_cold);
	P(se.statistics.nr_failed_migrations_affine);
	P(se.statistics.nr_failed_migrations_running);
	P(se.statistics.nr_failed_migrations_hot);
	P(se.statistics.nr_forced_migrations);
	P(se.statistics.nr_wakeups);
	P(se.statistics.nr_wakeups_sync);
	P(se.statistics.nr_wakeups_migrate);
	P(se.statistics.nr_wakeups_local);
	P(se.statistics.nr_wakeups_remote);
	P(se.statistics.nr_wakeups_affine);
	P(se.statistics.nr_wakeups_affine_attempts);
	P(se.statistics.nr_wakeups_passive);
	P(se.statistics.nr_wakeups_idle);
479 480 481 482 483 484 485 486 487 488 489

	{
		u64 avg_atom, avg_per_cpu;

		avg_atom = p->se.sum_exec_runtime;
		if (nr_switches)
			do_div(avg_atom, nr_switches);
		else
			avg_atom = -1LL;

		avg_per_cpu = p->se.sum_exec_runtime;
490
		if (p->se.nr_migrations) {
R
Roman Zippel 已提交
491 492
			avg_per_cpu = div64_u64(avg_per_cpu,
						p->se.nr_migrations);
493
		} else {
494
			avg_per_cpu = -1LL;
495
		}
496 497 498 499

		__PN(avg_atom);
		__PN(avg_per_cpu);
	}
I
Ingo Molnar 已提交
500
#endif
501
	__P(nr_switches);
502
	SEQ_printf(m, "%-35s:%21Ld\n",
503 504 505 506
		   "nr_voluntary_switches", (long long)p->nvcsw);
	SEQ_printf(m, "%-35s:%21Ld\n",
		   "nr_involuntary_switches", (long long)p->nivcsw);

I
Ingo Molnar 已提交
507 508 509
	P(se.load.weight);
	P(policy);
	P(prio);
I
Ingo Molnar 已提交
510
#undef PN
511 512 513
#undef __PN
#undef P
#undef __P
I
Ingo Molnar 已提交
514 515

	{
516
		unsigned int this_cpu = raw_smp_processor_id();
I
Ingo Molnar 已提交
517 518
		u64 t0, t1;

519 520
		t0 = cpu_clock(this_cpu);
		t1 = cpu_clock(this_cpu);
521
		SEQ_printf(m, "%-35s:%21Ld\n",
I
Ingo Molnar 已提交
522 523 524 525 526 527
			   "clock-delta", (long long)(t1-t0));
	}
}

void proc_sched_set_task(struct task_struct *p)
{
I
Ingo Molnar 已提交
528
#ifdef CONFIG_SCHEDSTATS
529
	memset(&p->se.statistics, 0, sizeof(p->se.statistics));
I
Ingo Molnar 已提交
530
#endif
I
Ingo Molnar 已提交
531
}