sched_debug.c 11.9 KB
Newer Older
I
Ingo Molnar 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
/*
 * kernel/time/sched_debug.c
 *
 * 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
static DEFINE_SPINLOCK(sched_debug_lock);

I
Ingo Molnar 已提交
21 22 23 24 25 26 27 28 29 30 31 32
/*
 * 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
33 34 35
/*
 * Ease the printing of nsec fields:
 */
I
Ingo Molnar 已提交
36
static long long nsec_high(unsigned long long nsec)
I
Ingo Molnar 已提交
37
{
I
Ingo Molnar 已提交
38
	if ((long long)nsec < 0) {
I
Ingo Molnar 已提交
39 40 41 42 43 44 45 46 47
		nsec = -nsec;
		do_div(nsec, 1000000);
		return -nsec;
	}
	do_div(nsec, 1000000);

	return nsec;
}

I
Ingo Molnar 已提交
48
static unsigned long nsec_low(unsigned long long nsec)
I
Ingo Molnar 已提交
49
{
I
Ingo Molnar 已提交
50
	if ((long long)nsec < 0)
I
Ingo Molnar 已提交
51 52 53 54 55 56 57
		nsec = -nsec;

	return do_div(nsec, 1000000);
}

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

58
#ifdef CONFIG_FAIR_GROUP_SCHED
59
static void print_cfs_group_stats(struct seq_file *m, int cpu, struct task_group *tg)
60 61 62 63 64 65 66 67 68 69 70 71 72 73
{
	struct sched_entity *se = tg->se[cpu];
	if (!se)
		return;

#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))

	PN(se->exec_start);
	PN(se->vruntime);
	PN(se->sum_exec_runtime);
#ifdef CONFIG_SCHEDSTATS
74 75 76 77 78 79 80 81 82 83
	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);
84 85 86 87 88 89 90
#endif
	P(se->load.weight);
#undef PN
#undef P
}
#endif

91 92 93 94 95
#ifdef CONFIG_CGROUP_SCHED
static char group_path[PATH_MAX];

static char *task_group_path(struct task_group *tg)
{
96 97 98
	if (autogroup_path(tg, group_path, PATH_MAX))
		return group_path;

99 100 101 102 103 104 105 106 107 108 109 110
	/*
	 * 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
111
static void
112
print_task(struct seq_file *m, struct rq *rq, struct task_struct *p)
I
Ingo Molnar 已提交
113 114 115 116 117 118
{
	if (rq->curr == p)
		SEQ_printf(m, "R");
	else
		SEQ_printf(m, " ");

I
Ingo Molnar 已提交
119
	SEQ_printf(m, "%15s %5d %9Ld.%06ld %9Ld %5d ",
I
Ingo Molnar 已提交
120
		p->comm, p->pid,
I
Ingo Molnar 已提交
121
		SPLIT_NS(p->se.vruntime),
I
Ingo Molnar 已提交
122
		(long long)(p->nvcsw + p->nivcsw),
123
		p->prio);
I
Ingo Molnar 已提交
124
#ifdef CONFIG_SCHEDSTATS
125
	SEQ_printf(m, "%9Ld.%06ld %9Ld.%06ld %9Ld.%06ld",
I
Ingo Molnar 已提交
126 127
		SPLIT_NS(p->se.vruntime),
		SPLIT_NS(p->se.sum_exec_runtime),
128
		SPLIT_NS(p->se.statistics.sum_sleep_runtime));
I
Ingo Molnar 已提交
129
#else
130
	SEQ_printf(m, "%15Ld %15Ld %15Ld.%06ld %15Ld.%06ld %15Ld.%06ld",
I
Ingo Molnar 已提交
131
		0LL, 0LL, 0LL, 0L, 0LL, 0L, 0LL, 0L);
I
Ingo Molnar 已提交
132
#endif
133 134 135
#ifdef CONFIG_CGROUP_SCHED
	SEQ_printf(m, " %s", task_group_path(task_group(p)));
#endif
136 137

	SEQ_printf(m, "\n");
I
Ingo Molnar 已提交
138 139
}

140
static void print_rq(struct seq_file *m, struct rq *rq, int rq_cpu)
I
Ingo Molnar 已提交
141 142
{
	struct task_struct *g, *p;
P
Peter Zijlstra 已提交
143
	unsigned long flags;
I
Ingo Molnar 已提交
144 145 146

	SEQ_printf(m,
	"\nrunnable tasks:\n"
147 148
	"            task   PID         tree-key  switches  prio"
	"     exec-runtime         sum-exec        sum-sleep\n"
149
	"------------------------------------------------------"
150
	"----------------------------------------------------\n");
I
Ingo Molnar 已提交
151

P
Peter Zijlstra 已提交
152
	read_lock_irqsave(&tasklist_lock, flags);
I
Ingo Molnar 已提交
153 154

	do_each_thread(g, p) {
P
Peter Zijlstra 已提交
155
		if (!p->on_rq || task_cpu(p) != rq_cpu)
I
Ingo Molnar 已提交
156 157
			continue;

158
		print_task(m, rq, p);
I
Ingo Molnar 已提交
159 160
	} while_each_thread(g, p);

P
Peter Zijlstra 已提交
161
	read_unlock_irqrestore(&tasklist_lock, flags);
I
Ingo Molnar 已提交
162 163
}

164
void print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
I
Ingo Molnar 已提交
165
{
I
Ingo Molnar 已提交
166 167
	s64 MIN_vruntime = -1, min_vruntime, max_vruntime = -1,
		spread, rq0_min_vruntime, spread0;
168
	struct rq *rq = cpu_rq(cpu);
I
Ingo Molnar 已提交
169 170 171
	struct sched_entity *last;
	unsigned long flags;

172 173 174
#ifdef CONFIG_FAIR_GROUP_SCHED
	SEQ_printf(m, "\ncfs_rq[%d]:%s\n", cpu, task_group_path(cfs_rq->tg));
#else
175
	SEQ_printf(m, "\ncfs_rq[%d]:\n", cpu);
176
#endif
I
Ingo Molnar 已提交
177 178
	SEQ_printf(m, "  .%-30s: %Ld.%06ld\n", "exec_clock",
			SPLIT_NS(cfs_rq->exec_clock));
I
Ingo Molnar 已提交
179

180
	raw_spin_lock_irqsave(&rq->lock, flags);
I
Ingo Molnar 已提交
181
	if (cfs_rq->rb_leftmost)
182
		MIN_vruntime = (__pick_first_entity(cfs_rq))->vruntime;
I
Ingo Molnar 已提交
183 184 185
	last = __pick_last_entity(cfs_rq);
	if (last)
		max_vruntime = last->vruntime;
P
Peter Zijlstra 已提交
186
	min_vruntime = cfs_rq->min_vruntime;
187
	rq0_min_vruntime = cpu_rq(0)->cfs.min_vruntime;
188
	raw_spin_unlock_irqrestore(&rq->lock, flags);
I
Ingo Molnar 已提交
189 190 191 192 193 194
	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
195
	spread = max_vruntime - MIN_vruntime;
I
Ingo Molnar 已提交
196 197
	SEQ_printf(m, "  .%-30s: %Ld.%06ld\n", "spread",
			SPLIT_NS(spread));
I
Ingo Molnar 已提交
198
	spread0 = min_vruntime - rq0_min_vruntime;
I
Ingo Molnar 已提交
199 200
	SEQ_printf(m, "  .%-30s: %Ld.%06ld\n", "spread0",
			SPLIT_NS(spread0));
P
Peter Zijlstra 已提交
201
	SEQ_printf(m, "  .%-30s: %d\n", "nr_spread_over",
P
Peter Zijlstra 已提交
202
			cfs_rq->nr_spread_over);
P
Peter Zijlstra 已提交
203 204
	SEQ_printf(m, "  .%-30s: %ld\n", "nr_running", cfs_rq->nr_running);
	SEQ_printf(m, "  .%-30s: %ld\n", "load", cfs_rq->load.weight);
205 206
#ifdef CONFIG_FAIR_GROUP_SCHED
#ifdef CONFIG_SMP
P
Peter Zijlstra 已提交
207 208 209 210 211 212 213
	SEQ_printf(m, "  .%-30s: %Ld.%06ld\n", "load_avg",
			SPLIT_NS(cfs_rq->load_avg));
	SEQ_printf(m, "  .%-30s: %Ld.%06ld\n", "load_period",
			SPLIT_NS(cfs_rq->load_period));
	SEQ_printf(m, "  .%-30s: %ld\n", "load_contrib",
			cfs_rq->load_contribution);
	SEQ_printf(m, "  .%-30s: %d\n", "load_tg",
214
			atomic_read(&cfs_rq->tg->load_weight));
215
#endif
P
Peter Zijlstra 已提交
216

217
	print_cfs_group_stats(m, cpu, cfs_rq->tg);
218
#endif
I
Ingo Molnar 已提交
219 220
}

221 222
void print_rt_rq(struct seq_file *m, int cpu, struct rt_rq *rt_rq)
{
223 224 225
#ifdef CONFIG_RT_GROUP_SCHED
	SEQ_printf(m, "\nrt_rq[%d]:%s\n", cpu, task_group_path(rt_rq->tg));
#else
226
	SEQ_printf(m, "\nrt_rq[%d]:\n", cpu);
227
#endif
228 229 230 231 232 233 234 235 236 237 238 239 240 241 242

#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
}

243 244
extern __read_mostly int sched_clock_running;

245
static void print_cpu(struct seq_file *m, int cpu)
I
Ingo Molnar 已提交
246
{
247
	struct rq *rq = cpu_rq(cpu);
248
	unsigned long flags;
I
Ingo Molnar 已提交
249 250 251 252 253 254 255 256 257 258 259 260 261 262

#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

#define P(x) \
	SEQ_printf(m, "  .%-30s: %Ld\n", #x, (long long)(rq->x))
I
Ingo Molnar 已提交
263 264
#define PN(x) \
	SEQ_printf(m, "  .%-30s: %Ld.%06ld\n", #x, SPLIT_NS(rq->x))
I
Ingo Molnar 已提交
265 266 267

	P(nr_running);
	SEQ_printf(m, "  .%-30s: %lu\n", "load",
268
		   rq->load.weight);
I
Ingo Molnar 已提交
269 270 271
	P(nr_switches);
	P(nr_load_updates);
	P(nr_uninterruptible);
I
Ingo Molnar 已提交
272
	PN(next_balance);
I
Ingo Molnar 已提交
273
	P(curr->pid);
I
Ingo Molnar 已提交
274
	PN(clock);
I
Ingo Molnar 已提交
275 276 277 278 279 280
	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
281
#undef PN
I
Ingo Molnar 已提交
282

P
Peter Zijlstra 已提交
283 284
#ifdef CONFIG_SCHEDSTATS
#define P(n) SEQ_printf(m, "  .%-30s: %d\n", #n, rq->n);
M
Mike Galbraith 已提交
285
#define P64(n) SEQ_printf(m, "  .%-30s: %Ld\n", #n, rq->n);
P
Peter Zijlstra 已提交
286 287 288 289 290 291

	P(yld_count);

	P(sched_switch);
	P(sched_count);
	P(sched_goidle);
M
Mike Galbraith 已提交
292 293 294
#ifdef CONFIG_SMP
	P64(avg_idle);
#endif
P
Peter Zijlstra 已提交
295 296 297 298

	P(ttwu_count);
	P(ttwu_local);

299 300
	SEQ_printf(m, "  .%-30s: %d\n", "bkl_count",
				rq->rq_sched_info.bkl_count);
P
Peter Zijlstra 已提交
301 302

#undef P
303
#undef P64
P
Peter Zijlstra 已提交
304
#endif
305
	spin_lock_irqsave(&sched_debug_lock, flags);
306
	print_cfs_stats(m, cpu);
307
	print_rt_stats(m, cpu);
I
Ingo Molnar 已提交
308

309
	rcu_read_lock();
310
	print_rq(m, rq, cpu);
311 312
	rcu_read_unlock();
	spin_unlock_irqrestore(&sched_debug_lock, flags);
I
Ingo Molnar 已提交
313 314
}

315 316 317 318 319 320
static const char *sched_tunable_scaling_names[] = {
	"none",
	"logaritmic",
	"linear"
};

I
Ingo Molnar 已提交
321 322
static int sched_debug_show(struct seq_file *m, void *v)
{
323 324
	u64 ktime, sched_clk, cpu_clk;
	unsigned long flags;
I
Ingo Molnar 已提交
325 326
	int cpu;

327 328 329 330 331 332 333
	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
334 335 336 337
		init_utsname()->release,
		(int)strcspn(init_utsname()->version, " "),
		init_utsname()->version);

338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353
#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
354

I
Ingo Molnar 已提交
355
#define P(x) \
356
	SEQ_printf(m, "  .%-40s: %Ld\n", #x, (long long)(x))
I
Ingo Molnar 已提交
357
#define PN(x) \
358
	SEQ_printf(m, "  .%-40s: %Ld.%06ld\n", #x, SPLIT_NS(x))
I
Ingo Molnar 已提交
359
	PN(sysctl_sched_latency);
360
	PN(sysctl_sched_min_granularity);
I
Ingo Molnar 已提交
361
	PN(sysctl_sched_wakeup_granularity);
362
	P(sysctl_sched_child_runs_first);
I
Ingo Molnar 已提交
363 364 365 366
	P(sysctl_sched_features);
#undef PN
#undef P

367 368 369 370
	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
371
	for_each_online_cpu(cpu)
372
		print_cpu(m, cpu);
I
Ingo Molnar 已提交
373 374 375 376 377 378

	SEQ_printf(m, "\n");

	return 0;
}

379
static void sysrq_sched_debug_show(void)
I
Ingo Molnar 已提交
380 381 382 383 384 385 386 387 388
{
	sched_debug_show(NULL, NULL);
}

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

389
static const struct file_operations sched_debug_fops = {
I
Ingo Molnar 已提交
390 391 392
	.open		= sched_debug_open,
	.read		= seq_read,
	.llseek		= seq_lseek,
393
	.release	= single_release,
I
Ingo Molnar 已提交
394 395 396 397 398 399
};

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

400
	pe = proc_create("sched_debug", 0444, NULL, &sched_debug_fops);
I
Ingo Molnar 已提交
401 402 403 404 405 406 407 408 409
	if (!pe)
		return -ENOMEM;
	return 0;
}

__initcall(init_sched_debug_procfs);

void proc_sched_show_task(struct task_struct *p, struct seq_file *m)
{
410
	unsigned long nr_switches;
I
Ingo Molnar 已提交
411

412 413
	SEQ_printf(m, "%s (%d, #threads: %d)\n", p->comm, p->pid,
						get_nr_threads(p));
414 415
	SEQ_printf(m,
		"---------------------------------------------------------\n");
416 417
#define __P(F) \
	SEQ_printf(m, "%-35s:%21Ld\n", #F, (long long)F)
I
Ingo Molnar 已提交
418
#define P(F) \
419
	SEQ_printf(m, "%-35s:%21Ld\n", #F, (long long)p->F)
420 421
#define __PN(F) \
	SEQ_printf(m, "%-35s:%14Ld.%06ld\n", #F, SPLIT_NS((long long)F))
I
Ingo Molnar 已提交
422
#define PN(F) \
423
	SEQ_printf(m, "%-35s:%14Ld.%06ld\n", #F, SPLIT_NS((long long)p->F))
I
Ingo Molnar 已提交
424

I
Ingo Molnar 已提交
425 426 427
	PN(se.exec_start);
	PN(se.vruntime);
	PN(se.sum_exec_runtime);
I
Ingo Molnar 已提交
428

429 430
	nr_switches = p->nvcsw + p->nivcsw;

I
Ingo Molnar 已提交
431
#ifdef CONFIG_SCHEDSTATS
432 433 434 435 436 437 438 439 440 441 442 443
	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);
444
	P(sched_info.bkl_count);
445
	P(se.nr_migrations);
446 447 448 449 450 451 452 453 454 455 456 457 458 459
	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);
460 461 462 463 464 465 466 467 468 469 470

	{
		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;
471
		if (p->se.nr_migrations) {
R
Roman Zippel 已提交
472 473
			avg_per_cpu = div64_u64(avg_per_cpu,
						p->se.nr_migrations);
474
		} else {
475
			avg_per_cpu = -1LL;
476
		}
477 478 479 480

		__PN(avg_atom);
		__PN(avg_per_cpu);
	}
I
Ingo Molnar 已提交
481
#endif
482
	__P(nr_switches);
483
	SEQ_printf(m, "%-35s:%21Ld\n",
484 485 486 487
		   "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
488 489 490
	P(se.load.weight);
	P(policy);
	P(prio);
I
Ingo Molnar 已提交
491
#undef PN
492 493 494
#undef __PN
#undef P
#undef __P
I
Ingo Molnar 已提交
495 496

	{
497
		unsigned int this_cpu = raw_smp_processor_id();
I
Ingo Molnar 已提交
498 499
		u64 t0, t1;

500 501
		t0 = cpu_clock(this_cpu);
		t1 = cpu_clock(this_cpu);
502
		SEQ_printf(m, "%-35s:%21Ld\n",
I
Ingo Molnar 已提交
503 504 505 506 507 508
			   "clock-delta", (long long)(t1-t0));
	}
}

void proc_sched_set_task(struct task_struct *p)
{
I
Ingo Molnar 已提交
509
#ifdef CONFIG_SCHEDSTATS
510
	memset(&p->se.statistics, 0, sizeof(p->se.statistics));
I
Ingo Molnar 已提交
511
#endif
I
Ingo Molnar 已提交
512
}