trace.c 98 KB
Newer Older
1
2
3
4
5
6
7
8
9
10
11
12
13
/*
 * ring buffer based function tracer
 *
 * Copyright (C) 2007-2008 Steven Rostedt <srostedt@redhat.com>
 * Copyright (C) 2008 Ingo Molnar <mingo@redhat.com>
 *
 * Originally taken from the RT patch by:
 *    Arnaldo Carvalho de Melo <acme@redhat.com>
 *
 * Based on code from the latency_tracer, that is:
 *  Copyright (C) 2004-2006 Ingo Molnar
 *  Copyright (C) 2004 William Lee Irwin III
 */
14
#include <linux/ring_buffer.h>
15
#include <linux/utsrelease.h>
16
17
#include <linux/stacktrace.h>
#include <linux/writeback.h>
18
19
#include <linux/kallsyms.h>
#include <linux/seq_file.h>
20
#include <linux/notifier.h>
21
#include <linux/irqflags.h>
22
#include <linux/debugfs.h>
23
#include <linux/pagemap.h>
24
25
26
#include <linux/hardirq.h>
#include <linux/linkage.h>
#include <linux/uaccess.h>
27
#include <linux/kprobes.h>
28
29
30
#include <linux/ftrace.h>
#include <linux/module.h>
#include <linux/percpu.h>
31
#include <linux/splice.h>
32
#include <linux/kdebug.h>
33
#include <linux/string.h>
34
35
#include <linux/ctype.h>
#include <linux/init.h>
36
#include <linux/poll.h>
37
38
#include <linux/gfp.h>
#include <linux/fs.h>
Ingo Molnar's avatar
Ingo Molnar committed
39

40
#include "trace.h"
41
#include "trace_output.h"
42

43
44
#define TRACE_BUFFER_FLAGS	(RB_FL_OVERWRITE)

45
unsigned long __read_mostly	tracing_max_latency;
46
47
unsigned long __read_mostly	tracing_thresh;

48
49
50
51
52
53
/*
 * On boot up, the ring buffer is set to the minimum size, so that
 * we do not waste memory on systems that are not using tracing.
 */
static int ring_buffer_expanded;

54
55
/*
 * We need to change this state when a selftest is running.
56
57
 * A selftest will lurk into the ring-buffer to count the
 * entries inserted during the selftest although some concurrent
58
 * insertions into the ring-buffer such as trace_printk could occurred
59
60
 * at the same time, giving false positive or negative results.
 */
61
static bool __read_mostly tracing_selftest_running;
62

63
64
65
66
67
/*
 * If a tracer is running, we do not want to run SELFTEST.
 */
static bool __read_mostly tracing_selftest_disabled;

68
69
70
71
72
73
74
75
76
77
78
79
80
81
/* For tracers that don't implement custom flags */
static struct tracer_opt dummy_tracer_opt[] = {
	{ }
};

static struct tracer_flags dummy_tracer_flags = {
	.val = 0,
	.opts = dummy_tracer_opt
};

static int dummy_set_flag(u32 old_flags, u32 bit, int set)
{
	return 0;
}
82
83
84
85
86
87
88

/*
 * Kill all tracing for good (never come back).
 * It is initialized to 1 but will turn to zero if the initialization
 * of the tracer is successful. But that is the only place that sets
 * this back to zero.
 */
89
static int tracing_disabled = 1;
90

91
92
93
94
95
96
97
98
99
100
101
102
103
104
static DEFINE_PER_CPU(local_t, ftrace_cpu_disabled);

static inline void ftrace_disable_cpu(void)
{
	preempt_disable();
	local_inc(&__get_cpu_var(ftrace_cpu_disabled));
}

static inline void ftrace_enable_cpu(void)
{
	local_dec(&__get_cpu_var(ftrace_cpu_disabled));
	preempt_enable();
}

105
static cpumask_var_t __read_mostly	tracing_buffer_mask;
106

107
108
109
/* Define which cpu buffers are currently read in trace_pipe */
static cpumask_var_t			tracing_reader_cpumask;

110
#define for_each_tracing_cpu(cpu)	\
111
	for_each_cpu(cpu, tracing_buffer_mask)
112

113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
/*
 * ftrace_dump_on_oops - variable to dump ftrace buffer on oops
 *
 * If there is an oops (or kernel panic) and the ftrace_dump_on_oops
 * is set, then ftrace_dump is called. This will output the contents
 * of the ftrace buffers to the console.  This is very useful for
 * capturing traces that lead to crashes and outputing it to a
 * serial console.
 *
 * It is default off, but you can enable it with either specifying
 * "ftrace_dump_on_oops" in the kernel command line, or setting
 * /proc/sys/kernel/ftrace_dump_on_oops to true.
 */
int ftrace_dump_on_oops;

128
129
130
131
132
static int tracing_set_tracer(const char *buf);

#define BOOTUP_TRACER_SIZE		100
static char bootup_tracer_buf[BOOTUP_TRACER_SIZE] __initdata;
static char *default_bootup_tracer;
133
134
135

static int __init set_ftrace(char *str)
{
136
137
	strncpy(bootup_tracer_buf, str, BOOTUP_TRACER_SIZE);
	default_bootup_tracer = bootup_tracer_buf;
138
139
	/* We are using ftrace early, expand it */
	ring_buffer_expanded = 1;
140
141
	return 1;
}
142
__setup("ftrace=", set_ftrace);
143

144
145
146
147
148
149
static int __init set_ftrace_dump_on_oops(char *str)
{
	ftrace_dump_on_oops = 1;
	return 1;
}
__setup("ftrace_dump_on_oops", set_ftrace_dump_on_oops);
Steven Rostedt's avatar
Steven Rostedt committed
150

151
unsigned long long ns2usecs(cycle_t nsec)
152
153
154
155
156
157
{
	nsec += 500;
	do_div(nsec, 1000);
	return nsec;
}

Steven Rostedt's avatar
Steven Rostedt committed
158
159
160
161
162
163
164
165
166
167
168
169
/*
 * The global_trace is the descriptor that holds the tracing
 * buffers for the live tracing. For each CPU, it contains
 * a link list of pages that will store trace entries. The
 * page descriptor of the pages in the memory is used to hold
 * the link list by linking the lru item in the page descriptor
 * to each of the pages in the buffer per CPU.
 *
 * For each active CPU there is a data field that holds the
 * pages for the buffer for that CPU. Each CPU has the same number
 * of pages allocated for its buffer.
 */
170
171
172
173
static struct trace_array	global_trace;

static DEFINE_PER_CPU(struct trace_array_cpu, global_trace_cpu);

174
175
176
177
178
int filter_current_check_discard(struct ftrace_event_call *call, void *rec,
				 struct ring_buffer_event *event)
{
	return filter_check_discard(call, rec, global_trace.buffer, event);
}
179
EXPORT_SYMBOL_GPL(filter_current_check_discard);
180

181
182
183
184
185
186
187
188
189
190
191
192
193
cycle_t ftrace_now(int cpu)
{
	u64 ts;

	/* Early boot up does not have a buffer yet */
	if (!global_trace.buffer)
		return trace_clock_local();

	ts = ring_buffer_time_stamp(global_trace.buffer, cpu);
	ring_buffer_normalize_time_stamp(global_trace.buffer, cpu, &ts);

	return ts;
}
194

Steven Rostedt's avatar
Steven Rostedt committed
195
196
197
198
199
200
201
202
203
204
/*
 * The max_tr is used to snapshot the global_trace when a maximum
 * latency is reached. Some tracers will use this to store a maximum
 * trace while it continues examining live traces.
 *
 * The buffers for the max_tr are set up the same as the global_trace.
 * When a snapshot is taken, the link list of the max_tr is swapped
 * with the link list of the global_trace and the buffers are reset for
 * the global_trace so the tracing can continue.
 */
205
206
207
208
static struct trace_array	max_tr;

static DEFINE_PER_CPU(struct trace_array_cpu, max_data);

Steven Rostedt's avatar
Steven Rostedt committed
209
/* tracer_enabled is used to toggle activation of a tracer */
210
static int			tracer_enabled = 1;
Steven Rostedt's avatar
Steven Rostedt committed
211

212
213
214
215
216
217
218
219
220
221
222
223
224
/**
 * tracing_is_enabled - return tracer_enabled status
 *
 * This function is used by other tracers to know the status
 * of the tracer_enabled flag.  Tracers may use this function
 * to know if it should enable their features when starting
 * up. See irqsoff tracer for an example (start_irqsoff_tracer).
 */
int tracing_is_enabled(void)
{
	return tracer_enabled;
}

Steven Rostedt's avatar
Steven Rostedt committed
225
/*
226
227
228
 * trace_buf_size is the size in bytes that is allocated
 * for a buffer. Note, the number of bytes is always rounded
 * to page size.
229
230
231
232
233
 *
 * This number is purposely set to a low number of 16384.
 * If the dump on oops happens, it will be much appreciated
 * to not have to wait for all that output. Anyway this can be
 * boot time and run time configurable.
Steven Rostedt's avatar
Steven Rostedt committed
234
 */
235
#define TRACE_BUF_SIZE_DEFAULT	1441792UL /* 16384 * 88 (sizeof(entry)) */
236

237
static unsigned long		trace_buf_size = TRACE_BUF_SIZE_DEFAULT;
238

Steven Rostedt's avatar
Steven Rostedt committed
239
/* trace_types holds a link list of available tracers. */
240
static struct tracer		*trace_types __read_mostly;
Steven Rostedt's avatar
Steven Rostedt committed
241
242

/* current_trace points to the tracer that is currently active */
243
static struct tracer		*current_trace __read_mostly;
Steven Rostedt's avatar
Steven Rostedt committed
244
245
246
247
248
249

/*
 * max_tracer_type_len is used to simplify the allocating of
 * buffers to read userspace tracer names. We keep track of
 * the longest tracer name registered.
 */
250
251
static int			max_tracer_type_len;

Steven Rostedt's avatar
Steven Rostedt committed
252
253
254
255
256
257
/*
 * trace_types_lock is used to protect the trace_types list.
 * This lock is also used to keep user access serialized.
 * Accesses from userspace will grab this lock while userspace
 * activities happen inside the kernel.
 */
258
static DEFINE_MUTEX(trace_types_lock);
Steven Rostedt's avatar
Steven Rostedt committed
259
260

/* trace_wait is a waitqueue for tasks blocked on trace_poll */
261
262
static DECLARE_WAIT_QUEUE_HEAD(trace_wait);

263
/* trace_flags holds trace_options default values */
264
unsigned long trace_flags = TRACE_ITER_PRINT_PARENT | TRACE_ITER_PRINTK |
265
266
	TRACE_ITER_ANNOTATE | TRACE_ITER_CONTEXT_INFO | TRACE_ITER_SLEEP_TIME |
	TRACE_ITER_GRAPH_TIME;
267

Steven Rostedt's avatar
Steven Rostedt committed
268
269
270
271
272
273
/**
 * trace_wake_up - wake up tasks waiting for trace input
 *
 * Simply wakes up any task that is blocked on the trace_wait
 * queue. These is used with trace_poll for tasks polling the trace.
 */
274
275
void trace_wake_up(void)
{
Ingo Molnar's avatar
Ingo Molnar committed
276
277
278
279
280
	/*
	 * The runqueue_is_locked() can fail, but this is the best we
	 * have for now:
	 */
	if (!(trace_flags & TRACE_ITER_BLOCK) && !runqueue_is_locked())
281
282
		wake_up(&trace_wait);
}
283

284
static int __init set_buf_size(char *str)
285
{
286
	unsigned long buf_size;
287
288
	int ret;

289
290
	if (!str)
		return 0;
291
	ret = strict_strtoul(str, 0, &buf_size);
292
	/* nr_entries can not be zero */
293
	if (ret < 0 || buf_size == 0)
294
		return 0;
295
	trace_buf_size = buf_size;
296
297
	return 1;
}
298
__setup("trace_buf_size=", set_buf_size);
299

Steven Rostedt's avatar
Steven Rostedt committed
300
301
302
303
304
unsigned long nsecs_to_usecs(unsigned long nsecs)
{
	return nsecs / 1000;
}

Steven Rostedt's avatar
Steven Rostedt committed
305
/* These must match the bit postions in trace_iterator_flags */
306
307
308
309
310
static const char *trace_options[] = {
	"print-parent",
	"sym-offset",
	"sym-addr",
	"verbose",
Ingo Molnar's avatar
Ingo Molnar committed
311
	"raw",
312
	"hex",
Ingo Molnar's avatar
Ingo Molnar committed
313
	"bin",
314
	"block",
Ingo Molnar's avatar
Ingo Molnar committed
315
	"stacktrace",
Ingo Molnar's avatar
Ingo Molnar committed
316
	"sched-tree",
317
	"trace_printk",
318
	"ftrace_preempt",
319
	"branch",
320
	"annotate",
321
	"userstacktrace",
322
	"sym-userobj",
323
	"printk-msg-only",
324
	"context-info",
325
	"latency-format",
326
	"global-clock",
327
	"sleep-time",
328
	"graph-time",
329
330
331
	NULL
};

Steven Rostedt's avatar
Steven Rostedt committed
332
333
334
335
336
337
338
339
340
/*
 * ftrace_max_lock is used to protect the swapping of buffers
 * when taking a max snapshot. The buffers themselves are
 * protected by per_cpu spinlocks. But the action of the swap
 * needs its own lock.
 *
 * This is defined as a raw_spinlock_t in order to help
 * with performance when lockdep debugging is enabled.
 */
341
342
static raw_spinlock_t ftrace_max_lock =
	(raw_spinlock_t)__RAW_SPIN_LOCK_UNLOCKED;
343
344
345
346

/*
 * Copy the new maximum trace into the separate maximum-trace
 * structure. (this way the maximum trace is permanently saved,
347
 * for later retrieval via /sys/kernel/debug/tracing/latency_trace)
348
 */
Ingo Molnar's avatar
Ingo Molnar committed
349
static void
350
351
352
353
354
355
356
357
358
359
360
361
__update_max_tr(struct trace_array *tr, struct task_struct *tsk, int cpu)
{
	struct trace_array_cpu *data = tr->data[cpu];

	max_tr.cpu = cpu;
	max_tr.time_start = data->preempt_timestamp;

	data = max_tr.data[cpu];
	data->saved_latency = tracing_max_latency;

	memcpy(data->comm, tsk->comm, TASK_COMM_LEN);
	data->pid = tsk->pid;
362
	data->uid = task_uid(tsk);
363
364
365
366
367
	data->nice = tsk->static_prio - 20 - MAX_RT_PRIO;
	data->policy = tsk->policy;
	data->rt_priority = tsk->rt_priority;

	/* record this tasks comm */
368
	tracing_record_cmdline(tsk);
369
370
}

371
372
373
374
375
ssize_t trace_seq_to_user(struct trace_seq *s, char __user *ubuf, size_t cnt)
{
	int len;
	int ret;

376
377
378
	if (!cnt)
		return 0;

379
380
381
382
383
384
385
	if (s->len <= s->readpos)
		return -EBUSY;

	len = s->len - s->readpos;
	if (cnt > len)
		cnt = len;
	ret = copy_to_user(ubuf, s->buffer + s->readpos, cnt);
386
	if (ret == cnt)
387
388
		return -EFAULT;

389
390
	cnt -= ret;

391
	s->readpos += cnt;
392
	return cnt;
393
394
}

395
static ssize_t trace_seq_to_buffer(struct trace_seq *s, void *buf, size_t cnt)
396
397
398
399
400
401
402
403
404
405
406
407
408
409
{
	int len;
	void *ret;

	if (s->len <= s->readpos)
		return -EBUSY;

	len = s->len - s->readpos;
	if (cnt > len)
		cnt = len;
	ret = memcpy(buf, s->buffer + s->readpos, cnt);
	if (!ret)
		return -EFAULT;

410
	s->readpos += cnt;
411
412
413
	return cnt;
}

Steven Rostedt's avatar
Steven Rostedt committed
414
415
416
417
418
419
420
421
422
/**
 * update_max_tr - snapshot all trace buffers from global_trace to max_tr
 * @tr: tracer
 * @tsk: the task with the latency
 * @cpu: The cpu that initiated the trace.
 *
 * Flip the buffers between the @tr and the max_tr and record information
 * about which task was the cause of this latency.
 */
Ingo Molnar's avatar
Ingo Molnar committed
423
void
424
425
update_max_tr(struct trace_array *tr, struct task_struct *tsk, int cpu)
{
426
	struct ring_buffer *buf = tr->buffer;
427

428
	WARN_ON_ONCE(!irqs_disabled());
429
	__raw_spin_lock(&ftrace_max_lock);
430
431
432
433

	tr->buffer = max_tr.buffer;
	max_tr.buffer = buf;

434
	ftrace_disable_cpu();
435
	ring_buffer_reset(tr->buffer);
436
	ftrace_enable_cpu();
437
438

	__update_max_tr(tr, tsk, cpu);
439
	__raw_spin_unlock(&ftrace_max_lock);
440
441
442
443
444
445
446
}

/**
 * update_max_tr_single - only copy one trace over, and reset the rest
 * @tr - tracer
 * @tsk - task with the latency
 * @cpu - the cpu of the buffer to copy.
Steven Rostedt's avatar
Steven Rostedt committed
447
448
 *
 * Flip the trace of a single CPU buffer between the @tr and the max_tr.
449
 */
Ingo Molnar's avatar
Ingo Molnar committed
450
void
451
452
update_max_tr_single(struct trace_array *tr, struct task_struct *tsk, int cpu)
{
453
	int ret;
454

455
	WARN_ON_ONCE(!irqs_disabled());
456
	__raw_spin_lock(&ftrace_max_lock);
457

458
459
	ftrace_disable_cpu();

460
461
462
	ring_buffer_reset(max_tr.buffer);
	ret = ring_buffer_swap_cpu(max_tr.buffer, tr->buffer, cpu);

463
464
	ftrace_enable_cpu();

465
	WARN_ON_ONCE(ret && ret != -EAGAIN);
466
467

	__update_max_tr(tr, tsk, cpu);
468
	__raw_spin_unlock(&ftrace_max_lock);
469
470
}

Steven Rostedt's avatar
Steven Rostedt committed
471
472
473
474
475
476
/**
 * register_tracer - register a tracer with the ftrace system.
 * @type - the plugin for the tracer
 *
 * Register a new plugin tracer.
 */
477
int register_tracer(struct tracer *type)
478
479
__releases(kernel_lock)
__acquires(kernel_lock)
480
481
482
483
484
485
486
487
488
489
{
	struct tracer *t;
	int len;
	int ret = 0;

	if (!type->name) {
		pr_info("Tracer must have a name\n");
		return -1;
	}

Ingo Molnar's avatar
Ingo Molnar committed
490
491
492
493
494
495
496
	/*
	 * When this gets called we hold the BKL which means that
	 * preemption is disabled. Various trace selftests however
	 * need to disable and enable preemption for successful tests.
	 * So we drop the BKL here and grab it after the tests again.
	 */
	unlock_kernel();
497
	mutex_lock(&trace_types_lock);
Ingo Molnar's avatar
Ingo Molnar committed
498

499
500
	tracing_selftest_running = true;

501
502
503
504
505
506
507
508
509
510
	for (t = trace_types; t; t = t->next) {
		if (strcmp(type->name, t->name) == 0) {
			/* already found */
			pr_info("Trace %s already registered\n",
				type->name);
			ret = -1;
			goto out;
		}
	}

511
512
513
514
515
516
517
	if (!type->set_flag)
		type->set_flag = &dummy_set_flag;
	if (!type->flags)
		type->flags = &dummy_tracer_flags;
	else
		if (!type->flags->opts)
			type->flags->opts = dummy_tracer_opt;
518
519
520
	if (!type->wait_pipe)
		type->wait_pipe = default_wait_pipe;

521

Steven Rostedt's avatar
Steven Rostedt committed
522
#ifdef CONFIG_FTRACE_STARTUP_TEST
523
	if (type->selftest && !tracing_selftest_disabled) {
Steven Rostedt's avatar
Steven Rostedt committed
524
525
526
		struct tracer *saved_tracer = current_trace;
		struct trace_array *tr = &global_trace;
		int i;
527

Steven Rostedt's avatar
Steven Rostedt committed
528
529
530
531
532
533
534
		/*
		 * Run a selftest on this tracer.
		 * Here we reset the trace buffer, and set the current
		 * tracer to be this tracer. The tracer can then run some
		 * internal tracing to verify that everything is in order.
		 * If we fail, we do not register this tracer.
		 */
Ingo Molnar's avatar
Ingo Molnar committed
535
		for_each_tracing_cpu(i)
536
			tracing_reset(tr, i);
Ingo Molnar's avatar
Ingo Molnar committed
537

Steven Rostedt's avatar
Steven Rostedt committed
538
539
540
541
542
543
544
545
546
547
		current_trace = type;
		/* the test is responsible for initializing and enabling */
		pr_info("Testing tracer %s: ", type->name);
		ret = type->selftest(type, tr);
		/* the test is responsible for resetting too */
		current_trace = saved_tracer;
		if (ret) {
			printk(KERN_CONT "FAILED!\n");
			goto out;
		}
Steven Rostedt's avatar
Steven Rostedt committed
548
		/* Only reset on passing, to avoid touching corrupted buffers */
Ingo Molnar's avatar
Ingo Molnar committed
549
		for_each_tracing_cpu(i)
550
			tracing_reset(tr, i);
Ingo Molnar's avatar
Ingo Molnar committed
551

Steven Rostedt's avatar
Steven Rostedt committed
552
553
554
555
		printk(KERN_CONT "PASSED\n");
	}
#endif

556
557
558
559
560
	type->next = trace_types;
	trace_types = type;
	len = strlen(type->name);
	if (len > max_tracer_type_len)
		max_tracer_type_len = len;
Steven Rostedt's avatar
Steven Rostedt committed
561

562
 out:
563
	tracing_selftest_running = false;
564
565
	mutex_unlock(&trace_types_lock);

Steven Rostedt's avatar
Steven Rostedt committed
566
567
568
569
570
571
572
573
574
575
576
577
	if (ret || !default_bootup_tracer)
		goto out_unlock;

	if (strncmp(default_bootup_tracer, type->name, BOOTUP_TRACER_SIZE))
		goto out_unlock;

	printk(KERN_INFO "Starting tracer '%s'\n", type->name);
	/* Do we want this tracer to start on bootup? */
	tracing_set_tracer(type->name);
	default_bootup_tracer = NULL;
	/* disable other selftests, since this will break it. */
	tracing_selftest_disabled = 1;
578
#ifdef CONFIG_FTRACE_STARTUP_TEST
Steven Rostedt's avatar
Steven Rostedt committed
579
580
	printk(KERN_INFO "Disabling FTRACE selftests due to running tracer '%s'\n",
	       type->name);
581
582
#endif

Steven Rostedt's avatar
Steven Rostedt committed
583
 out_unlock:
584
	lock_kernel();
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
	return ret;
}

void unregister_tracer(struct tracer *type)
{
	struct tracer **t;
	int len;

	mutex_lock(&trace_types_lock);
	for (t = &trace_types; *t; t = &(*t)->next) {
		if (*t == type)
			goto found;
	}
	pr_info("Trace %s not registered\n", type->name);
	goto out;

 found:
	*t = (*t)->next;
603
604
605
606
607
608
609
610
611

	if (type == current_trace && tracer_enabled) {
		tracer_enabled = 0;
		tracing_stop();
		if (current_trace->stop)
			current_trace->stop(&global_trace);
		current_trace = &nop_trace;
	}

612
613
614
615
616
617
618
619
620
621
622
623
624
	if (strlen(type->name) != max_tracer_type_len)
		goto out;

	max_tracer_type_len = 0;
	for (t = &trace_types; *t; t = &(*t)->next) {
		len = strlen((*t)->name);
		if (len > max_tracer_type_len)
			max_tracer_type_len = len;
	}
 out:
	mutex_unlock(&trace_types_lock);
}

625
void tracing_reset(struct trace_array *tr, int cpu)
626
{
627
	ftrace_disable_cpu();
628
	ring_buffer_reset_cpu(tr->buffer, cpu);
629
	ftrace_enable_cpu();
630
631
}

632
633
634
635
636
637
638
639
640
641
void tracing_reset_online_cpus(struct trace_array *tr)
{
	int cpu;

	tr->time_start = ftrace_now(tr->cpu);

	for_each_online_cpu(cpu)
		tracing_reset(tr, cpu);
}

642
643
644
645
646
647
648
649
650
651
void tracing_reset_current(int cpu)
{
	tracing_reset(&global_trace, cpu);
}

void tracing_reset_current_online_cpus(void)
{
	tracing_reset_online_cpus(&global_trace);
}

652
#define SAVED_CMDLINES 128
653
#define NO_CMDLINE_MAP UINT_MAX
654
655
656
657
static unsigned map_pid_to_cmdline[PID_MAX_DEFAULT+1];
static unsigned map_cmdline_to_pid[SAVED_CMDLINES];
static char saved_cmdlines[SAVED_CMDLINES][TASK_COMM_LEN];
static int cmdline_idx;
658
static raw_spinlock_t trace_cmdline_lock = __RAW_SPIN_LOCK_UNLOCKED;
659
660

/* temporary disable recording */
661
static atomic_t trace_record_cmdline_disabled __read_mostly;
662
663
664

static void trace_init_cmdlines(void)
{
665
666
	memset(&map_pid_to_cmdline, NO_CMDLINE_MAP, sizeof(map_pid_to_cmdline));
	memset(&map_cmdline_to_pid, NO_CMDLINE_MAP, sizeof(map_cmdline_to_pid));
667
668
669
	cmdline_idx = 0;
}

670
671
672
static int trace_stop_count;
static DEFINE_SPINLOCK(tracing_start_lock);

673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
/**
 * ftrace_off_permanent - disable all ftrace code permanently
 *
 * This should only be called when a serious anomally has
 * been detected.  This will turn off the function tracing,
 * ring buffers, and other tracing utilites. It takes no
 * locks and can be called from any context.
 */
void ftrace_off_permanent(void)
{
	tracing_disabled = 1;
	ftrace_stop();
	tracing_off_permanent();
}

688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
/**
 * tracing_start - quick start of the tracer
 *
 * If tracing is enabled but was stopped by tracing_stop,
 * this will start the tracer back up.
 */
void tracing_start(void)
{
	struct ring_buffer *buffer;
	unsigned long flags;

	if (tracing_disabled)
		return;

	spin_lock_irqsave(&tracing_start_lock, flags);
703
704
705
706
707
708
	if (--trace_stop_count) {
		if (trace_stop_count < 0) {
			/* Someone screwed up their debugging */
			WARN_ON_ONCE(1);
			trace_stop_count = 0;
		}
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
		goto out;
	}


	buffer = global_trace.buffer;
	if (buffer)
		ring_buffer_record_enable(buffer);

	buffer = max_tr.buffer;
	if (buffer)
		ring_buffer_record_enable(buffer);

	ftrace_start();
 out:
	spin_unlock_irqrestore(&tracing_start_lock, flags);
}

/**
 * tracing_stop - quick stop of the tracer
 *
 * Light weight way to stop tracing. Use in conjunction with
 * tracing_start.
 */
void tracing_stop(void)
{
	struct ring_buffer *buffer;
	unsigned long flags;

	ftrace_stop();
	spin_lock_irqsave(&tracing_start_lock, flags);
	if (trace_stop_count++)
		goto out;

	buffer = global_trace.buffer;
	if (buffer)
		ring_buffer_record_disable(buffer);

	buffer = max_tr.buffer;
	if (buffer)
		ring_buffer_record_disable(buffer);

 out:
	spin_unlock_irqrestore(&tracing_start_lock, flags);
}

Ingo Molnar's avatar
Ingo Molnar committed
754
void trace_stop_cmdline_recording(void);
755

Ingo Molnar's avatar
Ingo Molnar committed
756
static void trace_save_cmdline(struct task_struct *tsk)
757
{
758
	unsigned pid, idx;
759
760
761
762
763
764
765
766
767
768

	if (!tsk->pid || unlikely(tsk->pid > PID_MAX_DEFAULT))
		return;

	/*
	 * It's not the end of the world if we don't get
	 * the lock, but we also don't want to spin
	 * nor do we want to disable interrupts,
	 * so if we miss here, then better luck next time.
	 */
769
	if (!__raw_spin_trylock(&trace_cmdline_lock))
770
771
772
		return;

	idx = map_pid_to_cmdline[tsk->pid];
773
	if (idx == NO_CMDLINE_MAP) {
774
775
		idx = (cmdline_idx + 1) % SAVED_CMDLINES;

776
777
778
779
780
781
782
783
784
		/*
		 * Check whether the cmdline buffer at idx has a pid
		 * mapped. We are going to overwrite that entry so we
		 * need to clear the map_pid_to_cmdline. Otherwise we
		 * would read the new comm for the old pid.
		 */
		pid = map_cmdline_to_pid[idx];
		if (pid != NO_CMDLINE_MAP)
			map_pid_to_cmdline[pid] = NO_CMDLINE_MAP;
785

786
		map_cmdline_to_pid[idx] = tsk->pid;
787
788
789
790
791
792
793
		map_pid_to_cmdline[tsk->pid] = idx;

		cmdline_idx = idx;
	}

	memcpy(&saved_cmdlines[idx], tsk->comm, TASK_COMM_LEN);

794
	__raw_spin_unlock(&trace_cmdline_lock);
795
796
}

797
void trace_find_cmdline(int pid, char comm[])
798
799
800
{
	unsigned map;

801
802
803
804
	if (!pid) {
		strcpy(comm, "<idle>");
		return;
	}
805

806
807
808
809
	if (pid > PID_MAX_DEFAULT) {
		strcpy(comm, "<...>");
		return;
	}
810

811
	preempt_disable();
812
	__raw_spin_lock(&trace_cmdline_lock);
813
	map = map_pid_to_cmdline[pid];
814
815
816
817
	if (map != NO_CMDLINE_MAP)
		strcpy(comm, saved_cmdlines[map]);
	else
		strcpy(comm, "<...>");
818

819
	__raw_spin_unlock(&trace_cmdline_lock);
820
	preempt_enable();
821
822
}

Ingo Molnar's avatar
Ingo Molnar committed
823
void tracing_record_cmdline(struct task_struct *tsk)
824
{
825
826
	if (atomic_read(&trace_record_cmdline_disabled) || !tracer_enabled ||
	    !tracing_is_on())
827
828
829
830
831
		return;

	trace_save_cmdline(tsk);
}

832
void
833
834
tracing_generic_entry_update(struct trace_entry *entry, unsigned long flags,
			     int pc)
835
836
837
{
	struct task_struct *tsk = current;

838
839
	entry->preempt_count		= pc & 0xff;
	entry->pid			= (tsk) ? tsk->pid : 0;
840
	entry->tgid			= (tsk) ? tsk->tgid : 0;
841
	entry->flags =
842
#ifdef CONFIG_TRACE_IRQFLAGS_SUPPORT
843
		(irqs_disabled_flags(flags) ? TRACE_FLAG_IRQS_OFF : 0) |
844
845
846
#else
		TRACE_FLAG_IRQS_NOSUPPORT |
#endif
847
848
849
850
851
		((pc & HARDIRQ_MASK) ? TRACE_FLAG_HARDIRQ : 0) |
		((pc & SOFTIRQ_MASK) ? TRACE_FLAG_SOFTIRQ : 0) |
		(need_resched() ? TRACE_FLAG_NEED_RESCHED : 0);
}

852
struct ring_buffer_event *trace_buffer_lock_reserve(struct trace_array *tr,
853
						    int type,
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
						    unsigned long len,
						    unsigned long flags, int pc)
{
	struct ring_buffer_event *event;

	event = ring_buffer_lock_reserve(tr->buffer, len);
	if (event != NULL) {
		struct trace_entry *ent = ring_buffer_event_data(event);

		tracing_generic_entry_update(ent, flags, pc);
		ent->type = type;
	}

	return event;
}
static void ftrace_trace_stack(struct trace_array *tr,
			       unsigned long flags, int skip, int pc);
static void ftrace_trace_userstack(struct trace_array *tr,
				   unsigned long flags, int pc);

874
875
876
877
static inline void __trace_buffer_unlock_commit(struct trace_array *tr,
					struct ring_buffer_event *event,
					unsigned long flags, int pc,
					int wake)
878
879
880
881
882
{
	ring_buffer_unlock_commit(tr->buffer, event);

	ftrace_trace_stack(tr, flags, 6, pc);
	ftrace_trace_userstack(tr, flags, pc);
883
884
885
886
887
888
889
890
891
892

	if (wake)
		trace_wake_up();
}

void trace_buffer_unlock_commit(struct trace_array *tr,
					struct ring_buffer_event *event,
					unsigned long flags, int pc)
{
	__trace_buffer_unlock_commit(tr, event, flags, pc, 1);
893
894
}

895
struct ring_buffer_event *
896
trace_current_buffer_lock_reserve(int type, unsigned long len,
897
898
899
900
901
				  unsigned long flags, int pc)
{
	return trace_buffer_lock_reserve(&global_trace,
					 type, len, flags, pc);
}
902
EXPORT_SYMBOL_GPL(trace_current_buffer_lock_reserve);
903
904
905
906

void trace_current_buffer_unlock_commit(struct ring_buffer_event *event,
					unsigned long flags, int pc)
{
907
	__trace_buffer_unlock_commit(&global_trace, event, flags, pc, 1);
908
}
909
EXPORT_SYMBOL_GPL(trace_current_buffer_unlock_commit);
910
911
912
913

void trace_nowake_buffer_unlock_commit(struct ring_buffer_event *event,
					unsigned long flags, int pc)
{
914
915
	__trace_buffer_unlock_commit(&global_trace, event, flags, pc, 0);
}
916
EXPORT_SYMBOL_GPL(trace_nowake_buffer_unlock_commit);
917
918
919
920

void trace_current_buffer_discard_commit(struct ring_buffer_event *event)
{
	ring_buffer_discard_commit(global_trace.buffer, event);
921
}
922
EXPORT_SYMBOL_GPL(trace_current_buffer_discard_commit);
923

Ingo Molnar's avatar
Ingo Molnar committed
924
void
925
trace_function(struct trace_array *tr,
926
927
	       unsigned long ip, unsigned long parent_ip, unsigned long flags,
	       int pc)
928
{
929
	struct ftrace_event_call *call = &event_function;
930
	struct ring_buffer_event *event;
931
	struct ftrace_entry *entry;
932

933
934
935
936
	/* If we are reading the ring buffer, don't trace */
	if (unlikely(local_read(&__get_cpu_var(ftrace_cpu_disabled))))
		return;

937
938
	event = trace_buffer_lock_reserve(tr, TRACE_FN, sizeof(*entry),
					  flags, pc);
939
940
941
	if (!event)
		return;
	entry	= ring_buffer_event_data(event);
942
943
	entry->ip			= ip;
	entry->parent_ip		= parent_ip;
944

945
946
	if (!filter_check_discard(call, entry, tr->buffer, event))
		ring_buffer_unlock_commit(tr->buffer, event);
947
948
}

949
#ifdef CONFIG_FUNCTION_GRAPH_TRACER
950
static int __trace_graph_entry(struct trace_array *tr,
951
952
953
954
				struct ftrace_graph_ent *trace,
				unsigned long flags,
				int pc)
{
955
	struct ftrace_event_call *call = &event_funcgraph_entry;
956
957
958
959
	struct ring_buffer_event *event;
	struct ftrace_graph_ent_entry *entry;

	if (unlikely(local_read(&__get_cpu_var(ftrace_cpu_disabled))))
960
		return 0;
961

962
963
	event = trace_buffer_lock_reserve(&global_trace, TRACE_GRAPH_ENT,
					  sizeof(*entry), flags, pc);
964
	if (!event)
965
		return 0;
966
967
	entry	= ring_buffer_event_data(event);
	entry->graph_ent			= *trace;
968
969
	if (!filter_current_check_discard(call, entry, event))
		ring_buffer_unlock_commit(global_trace.buffer, event);
970
971

	return 1;
972
973
974
}

static void __trace_graph_return(struct trace_array *tr,
975
				struct ftrace_graph_ret *trace,
976
977
978
				unsigned long flags,
				int pc)
{
979
	struct ftrace_event_call *call = &event_funcgraph_exit;
980
	struct ring_buffer_event *event;
981
	struct ftrace_graph_ret_entry *entry;
982
983
984
985

	if (unlikely(local_read(&__get_cpu_var(ftrace_cpu_disabled))))
		return;

986
987
	event = trace_buffer_lock_reserve(&global_trace, TRACE_GRAPH_RET,
					  sizeof(*entry), flags, pc);
988
989
990
	if (!event)
		return;
	entry	= ring_buffer_event_data(event);
991
	entry->ret				= *trace;
992
993
	if (!filter_current_check_discard(call, entry, event))
		ring_buffer_unlock_commit(global_trace.buffer, event);
994
995
996
}
#endif

Ingo Molnar's avatar
Ingo Molnar committed
997
void
Ingo Molnar's avatar
Ingo Molnar committed
998
ftrace(struct trace_array *tr, struct trace_array_cpu *data,
999
1000
       unsigned long ip, unsigned long parent_ip, unsigned long flags,
       int pc)
Ingo Molnar's avatar
Ingo Molnar committed
1001
1002
{
	if (likely(!atomic_read(&data->disabled)))
1003
		trace_function(tr, ip, parent_ip, flags, pc);
Ingo Molnar's avatar
Ingo Molnar committed
1004
1005
}

1006
1007
1008
static void __ftrace_trace_stack(struct trace_array *tr,
				 unsigned long flags,
				 int skip, int pc)
Ingo Molnar's avatar
Ingo Molnar committed
1009
{
1010
#ifdef CONFIG_STACKTRACE
1011
	struct ftrace_event_call *call = &event_kernel_stack;
1012
	struct ring_buffer_event *event;
1013
	struct stack_entry *entry;
Ingo Molnar's avatar
Ingo Molnar committed
1014
1015
	struct stack_trace trace;

1016
1017
	event = trace_buffer_lock_reserve(tr, TRACE_STACK,
					  sizeof(*entry), flags, pc);
1018
1019
1020
	if (!event)
		return;
	entry	= ring_buffer_event_data(event);
1021
	memset(&entry->caller, 0, sizeof(entry->caller));
Ingo Molnar's avatar
Ingo Molnar committed
1022
1023
1024
1025

	trace.nr_entries	= 0;
	trace.max_entries	= FTRACE_STACK_ENTRIES;
	trace.skip		= skip;
1026
	trace.entries		= entry->caller;
Ingo Molnar's avatar
Ingo Molnar committed
1027
1028

	save_stack_trace(&trace);
1029
1030
	if (!filter_check_discard(call, entry, tr->buffer, event))
		ring_buffer_unlock_commit(tr->buffer, event);
1031
#endif
Ingo Molnar's avatar
Ingo Molnar committed
1032
1033
}

1034
1035
1036
1037
1038
1039
1040
static void ftrace_trace_stack(struct trace_array *tr,
			       unsigned long flags,
			       int skip, int pc)
{
	if (!(trace_flags & TRACE_ITER_STACKTRACE))
		return;

1041
	__ftrace_trace_stack(tr, flags, skip, pc);
1042
1043
}

1044
1045
void __trace_stack(struct trace_array *tr,
		   unsigned long flags,
1046
		   int skip, int pc)
1047
{
1048
	__ftrace_trace_stack(tr, flags, skip, pc);
1049
1050
}

1051
static void ftrace_trace_userstack(struct trace_array *tr,
1052
				   unsigned long flags, int pc)
1053
{
1054
#ifdef CONFIG_STACKTRACE
1055
	struct ftrace_event_call *call = &event_user_stack;
1056
	struct ring_buffer_event *event;
1057
1058
1059
1060
1061
1062
	struct userstack_entry *entry;
	struct stack_trace trace;

	if (!(trace_flags & TRACE_ITER_USERSTACKTRACE))
		return;

1063
1064
	event = trace_buffer_lock_reserve(tr, TRACE_USER_STACK,
					  sizeof(*entry), flags, pc);
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
	if (!event)
		return;
	entry	= ring_buffer_event_data(event);

	memset(&entry->caller, 0, sizeof(entry->caller));

	trace.nr_entries	= 0;
	trace.max_entries	= FTRACE_STACK_ENTRIES;
	trace.skip		= 0;
	trace.entries		= entry->caller;

	save_stack_trace_user(&trace);
1077
1078
	if (!filter_check_discard(call, entry, tr->buffer, event))
		ring_buffer_unlock_commit(tr->buffer, event);
1079
#endif
1080
1081
}

1082
1083
#ifdef UNUSED
static void __trace_userstack(struct trace_array *tr, unsigned long flags)
1084
{
1085
	ftrace_trace_userstack(tr, flags, preempt_count());
1086
}
1087
#endif /* UNUSED */
1088

1089
static void
1090
ftrace_trace_special(void *__tr,
1091
1092
		     unsigned long arg1, unsigned long arg2, unsigned long arg3,
		     int pc)
Ingo Molnar's avatar
Ingo Molnar committed
1093
{
1094
	struct ring_buffer_event *event;
Ingo Molnar's avatar
Ingo Molnar committed
1095
	struct trace_array *tr = __tr;
1096
	struct special_entry *entry;
Ingo Molnar's avatar
Ingo Molnar committed
1097

1098
1099
	event = trace_buffer_lock_reserve(tr, TRACE_SPECIAL,
					  sizeof(*entry), 0, pc);
1100
1101
1102
	if (!event)
		return;
	entry	= ring_buffer_event_data(event);
1103
1104
1105
	entry->arg1			= arg1;
	entry->arg2			= arg2;
	entry->arg3			= arg3;
1106
	trace_buffer_unlock_commit(tr, event, 0, pc);
Ingo Molnar's avatar
Ingo Molnar committed
1107
1108
}

1109
1110
1111
1112
void
__trace_special(void *__tr, void *__data,
		unsigned long arg1, unsigned long arg2, unsigned long arg3)
{
1113
	ftrace_trace_special(__tr, arg1, arg2, arg3, preempt_count());
1114
1115
}

Ingo Molnar's avatar
Ingo Molnar committed
1116
void
1117
tracing_sched_switch_trace(struct trace_array *tr,
Ingo Molnar's avatar
Ingo Molnar committed
1118
1119
			   struct task_struct *prev,
			   struct task_struct *next,
1120
			   unsigned long flags, int pc)
1121
{
1122
	struct ftrace_event_call *call = &event_context_switch;
1123
	struct ring_buffer_event *event;
1124
	struct ctx_switch_entry *entry;
1125

1126
1127
	event = trace_buffer_lock_reserve(tr, TRACE_CTX,
					  sizeof(*entry), flags, pc);
1128
1129
1130
	if (!event)
		return;
	entry	= ring_buffer_event_data(event);
1131
1132
1133
1134
1135
1136
1137
	entry->prev_pid			= prev->pid;
	entry->prev_prio		= prev->prio;
	entry->prev_state		= prev->state;
	entry->next_pid			= next->pid;
	entry->next_prio		= next->prio;
	entry->next_state		= next->state;
	entry->next_cpu	= task_cpu(next);
1138

1139
1140
	if (!filter_check_discard(call, entry, tr->buffer, event))
		trace_buffer_unlock_commit(tr, event, flags, pc);
1141
1142
}

1143
1144
void
tracing_sched_wakeup_trace(struct trace_array *tr,
Ingo Molnar's avatar
Ingo Molnar committed
1145
1146
			   struct task_struct *wakee,
			   struct task_struct *curr,
1147
			   unsigned long flags, int pc)
1148
{
1149
	struct ftrace_event_call *call = &event_wakeup;
1150
	struct ring_buffer_event *event;