trace_functions_graph.c 21.9 KB
Newer Older
1
2
3
/*
 *
 * Function graph tracer.
4
 * Copyright (c) 2008-2009 Frederic Weisbecker <fweisbec@gmail.com>
5
6
7
8
9
10
11
12
13
14
 * Mostly borrowed from function tracer which
 * is Copyright (c) Steven Rostedt <srostedt@redhat.com>
 *
 */
#include <linux/debugfs.h>
#include <linux/uaccess.h>
#include <linux/ftrace.h>
#include <linux/fs.h>

#include "trace.h"
15
#include "trace_output.h"
16

17
18
19
20
21
struct fgraph_data {
	pid_t		last_pid;
	int		depth;
};

22
#define TRACE_GRAPH_INDENT	2
23

24
/* Flag options */
25
#define TRACE_GRAPH_PRINT_OVERRUN	0x1
26
27
#define TRACE_GRAPH_PRINT_CPU		0x2
#define TRACE_GRAPH_PRINT_OVERHEAD	0x4
28
#define TRACE_GRAPH_PRINT_PROC		0x8
29
30
#define TRACE_GRAPH_PRINT_DURATION	0x10
#define TRACE_GRAPH_PRINT_ABS_TIME	0X20
31

32
static struct tracer_opt trace_opts[] = {
33
	/* Display overruns? (for self-debug purpose) */
34
35
36
37
38
	{ TRACER_OPT(funcgraph-overrun, TRACE_GRAPH_PRINT_OVERRUN) },
	/* Display CPU ? */
	{ TRACER_OPT(funcgraph-cpu, TRACE_GRAPH_PRINT_CPU) },
	/* Display Overhead ? */
	{ TRACER_OPT(funcgraph-overhead, TRACE_GRAPH_PRINT_OVERHEAD) },
39
40
	/* Display proc name/pid */
	{ TRACER_OPT(funcgraph-proc, TRACE_GRAPH_PRINT_PROC) },
41
42
43
44
	/* Display duration of execution */
	{ TRACER_OPT(funcgraph-duration, TRACE_GRAPH_PRINT_DURATION) },
	/* Display absolute time of an entry */
	{ TRACER_OPT(funcgraph-abstime, TRACE_GRAPH_PRINT_ABS_TIME) },
45
46
47
48
	{ } /* Empty entry */
};

static struct tracer_flags tracer_flags = {
49
	/* Don't display overruns and proc by default */
50
51
	.val = TRACE_GRAPH_PRINT_CPU | TRACE_GRAPH_PRINT_OVERHEAD |
	       TRACE_GRAPH_PRINT_DURATION,
52
53
54
	.opts = trace_opts
};

55
/* pid on the last trace processed */
56

57

58
59
/* Add a function return address to the trace stack on thread info.*/
int
60
61
ftrace_push_return_trace(unsigned long ret, unsigned long func, int *depth,
			 unsigned long frame_pointer)
62
{
63
	unsigned long long calltime;
64
65
66
67
68
	int index;

	if (!current->ret_stack)
		return -EBUSY;

69
70
71
72
73
74
	/*
	 * We must make sure the ret_stack is tested before we read
	 * anything else.
	 */
	smp_rmb();

75
76
77
78
79
80
	/* The return trace stack is full */
	if (current->curr_ret_stack == FTRACE_RETFUNC_DEPTH - 1) {
		atomic_inc(&current->trace_overrun);
		return -EBUSY;
	}

81
82
	calltime = trace_clock_local();

83
84
85
86
	index = ++current->curr_ret_stack;
	barrier();
	current->ret_stack[index].ret = ret;
	current->ret_stack[index].func = func;
87
	current->ret_stack[index].calltime = calltime;
88
	current->ret_stack[index].subtime = 0;
89
	current->ret_stack[index].fp = frame_pointer;
90
91
92
93
94
95
	*depth = index;

	return 0;
}

/* Retrieve a function return address to the trace stack on thread info.*/
96
static void
97
98
ftrace_pop_return_trace(struct ftrace_graph_ret *trace, unsigned long *ret,
			unsigned long frame_pointer)
99
100
101
102
103
104
105
106
107
108
109
110
111
{
	int index;

	index = current->curr_ret_stack;

	if (unlikely(index < 0)) {
		ftrace_graph_stop();
		WARN_ON(1);
		/* Might as well panic, otherwise we have no where to go */
		*ret = (unsigned long)panic;
		return;
	}

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
#ifdef CONFIG_HAVE_FUNCTION_GRAPH_FP_TEST
	/*
	 * The arch may choose to record the frame pointer used
	 * and check it here to make sure that it is what we expect it
	 * to be. If gcc does not set the place holder of the return
	 * address in the frame pointer, and does a copy instead, then
	 * the function graph trace will fail. This test detects this
	 * case.
	 *
	 * Currently, x86_32 with optimize for size (-Os) makes the latest
	 * gcc do the above.
	 */
	if (unlikely(current->ret_stack[index].fp != frame_pointer)) {
		ftrace_graph_stop();
		WARN(1, "Bad frame pointer: expected %lx, received %lx\n"
		     "  from func %pF return to %lx\n",
		     current->ret_stack[index].fp,
		     frame_pointer,
		     (void *)current->ret_stack[index].func,
		     current->ret_stack[index].ret);
		*ret = (unsigned long)panic;
		return;
	}
#endif

137
138
139
140
141
142
143
144
145
146
147
	*ret = current->ret_stack[index].ret;
	trace->func = current->ret_stack[index].func;
	trace->calltime = current->ret_stack[index].calltime;
	trace->overrun = atomic_read(&current->trace_overrun);
	trace->depth = index;
}

/*
 * Send the trace to the ring-buffer.
 * @return the original return address.
 */
148
unsigned long ftrace_return_to_handler(unsigned long frame_pointer)
149
150
151
152
{
	struct ftrace_graph_ret trace;
	unsigned long ret;

153
	ftrace_pop_return_trace(&trace, &ret, frame_pointer);
154
	trace.rettime = trace_clock_local();
155
	ftrace_graph_return(&trace);
156
157
	barrier();
	current->curr_ret_stack--;
158
159
160
161
162
163
164
165
166
167
168

	if (unlikely(!ret)) {
		ftrace_graph_stop();
		WARN_ON(1);
		/* Might as well panic. What else to do? */
		ret = (unsigned long)panic;
	}

	return ret;
}

169
170
static int graph_trace_init(struct trace_array *tr)
{
171
	int ret = register_ftrace_graph(&trace_graph_return,
172
					&trace_graph_entry);
173
174
175
176
177
	if (ret)
		return ret;
	tracing_start_cmdline_record();

	return 0;
178
179
180
181
}

static void graph_trace_reset(struct trace_array *tr)
{
182
183
	tracing_stop_cmdline_record();
	unregister_ftrace_graph();
184
185
}

186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
static inline int log10_cpu(int nb)
{
	if (nb / 100)
		return 3;
	if (nb / 10)
		return 2;
	return 1;
}

static enum print_line_t
print_graph_cpu(struct trace_seq *s, int cpu)
{
	int i;
	int ret;
	int log10_this = log10_cpu(cpu);
201
	int log10_all = log10_cpu(cpumask_weight(cpu_online_mask));
202
203


204
205
206
207
208
209
210
211
212
213
214
215
216
	/*
	 * Start with a space character - to make it stand out
	 * to the right a bit when trace output is pasted into
	 * email:
	 */
	ret = trace_seq_printf(s, " ");

	/*
	 * Tricky - we space the CPU field according to the max
	 * number of online CPUs. On a 2-cpu system it would take
	 * a maximum of 1 digit - on a 128 cpu system it would
	 * take up to 3 digits:
	 */
217
218
219
220
221
222
223
	for (i = 0; i < log10_all - log10_this; i++) {
		ret = trace_seq_printf(s, " ");
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}
	ret = trace_seq_printf(s, "%d) ", cpu);
	if (!ret)
224
225
		return TRACE_TYPE_PARTIAL_LINE;

226
227
228
	return TRACE_TYPE_HANDLED;
}

229
230
231
232
233
#define TRACE_GRAPH_PROCINFO_LENGTH	14

static enum print_line_t
print_graph_proc(struct trace_seq *s, pid_t pid)
{
234
	char comm[TASK_COMM_LEN];
235
236
	/* sign + log10(MAX_INT) + '\0' */
	char pid_str[11];
237
238
239
240
	int spaces = 0;
	int ret;
	int len;
	int i;
241

242
	trace_find_cmdline(pid, comm);
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
	comm[7] = '\0';
	sprintf(pid_str, "%d", pid);

	/* 1 stands for the "-" character */
	len = strlen(comm) + strlen(pid_str) + 1;

	if (len < TRACE_GRAPH_PROCINFO_LENGTH)
		spaces = TRACE_GRAPH_PROCINFO_LENGTH - len;

	/* First spaces to align center */
	for (i = 0; i < spaces / 2; i++) {
		ret = trace_seq_printf(s, " ");
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}

	ret = trace_seq_printf(s, "%s-%s", comm, pid_str);
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

	/* Last spaces to align center */
	for (i = 0; i < spaces - (spaces / 2); i++) {
		ret = trace_seq_printf(s, " ");
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}
	return TRACE_TYPE_HANDLED;
}

272

273
/* If the pid changed since the last trace, output this event */
274
static enum print_line_t
275
verif_pid(struct trace_seq *s, pid_t pid, int cpu, struct fgraph_data *data)
276
{
277
	pid_t prev_pid;
278
	pid_t *last_pid;
279
	int ret;
280

281
	if (!data)
282
283
		return TRACE_TYPE_HANDLED;

284
	last_pid = &(per_cpu_ptr(data, cpu)->last_pid);
285
286

	if (*last_pid == pid)
287
		return TRACE_TYPE_HANDLED;
288

289
290
	prev_pid = *last_pid;
	*last_pid = pid;
291

292
293
	if (prev_pid == -1)
		return TRACE_TYPE_HANDLED;
294
295
296
297
298
299
300
301
302
/*
 * Context-switch trace line:

 ------------------------------------------
 | 1)  migration/0--1  =>  sshd-1755
 ------------------------------------------

 */
	ret = trace_seq_printf(s,
303
		" ------------------------------------------\n");
304
	if (!ret)
305
		return TRACE_TYPE_PARTIAL_LINE;
306
307
308

	ret = print_graph_cpu(s, cpu);
	if (ret == TRACE_TYPE_PARTIAL_LINE)
309
		return TRACE_TYPE_PARTIAL_LINE;
310
311
312

	ret = print_graph_proc(s, prev_pid);
	if (ret == TRACE_TYPE_PARTIAL_LINE)
313
		return TRACE_TYPE_PARTIAL_LINE;
314
315
316

	ret = trace_seq_printf(s, " => ");
	if (!ret)
317
		return TRACE_TYPE_PARTIAL_LINE;
318
319
320

	ret = print_graph_proc(s, pid);
	if (ret == TRACE_TYPE_PARTIAL_LINE)
321
		return TRACE_TYPE_PARTIAL_LINE;
322
323
324
325

	ret = trace_seq_printf(s,
		"\n ------------------------------------------\n\n");
	if (!ret)
326
		return TRACE_TYPE_PARTIAL_LINE;
327

328
	return TRACE_TYPE_HANDLED;
329
330
}

331
332
static struct ftrace_graph_ret_entry *
get_return_for_leaf(struct trace_iterator *iter,
333
334
335
336
337
338
339
340
		struct ftrace_graph_ent_entry *curr)
{
	struct ring_buffer_iter *ring_iter;
	struct ring_buffer_event *event;
	struct ftrace_graph_ret_entry *next;

	ring_iter = iter->buffer_iter[iter->cpu];

341
342
343
344
345
346
347
348
349
	/* First peek to compare current entry and the next one */
	if (ring_iter)
		event = ring_buffer_iter_peek(ring_iter, NULL);
	else {
	/* We need to consume the current entry to see the next one */
		ring_buffer_consume(iter->tr->buffer, iter->cpu, NULL);
		event = ring_buffer_peek(iter->tr->buffer, iter->cpu,
					NULL);
	}
350
351

	if (!event)
352
		return NULL;
353
354
355
356

	next = ring_buffer_event_data(event);

	if (next->ent.type != TRACE_GRAPH_RET)
357
		return NULL;
358
359
360

	if (curr->ent.pid != next->ent.pid ||
			curr->graph_ent.func != next->ret.func)
361
		return NULL;
362

363
364
365
366
367
	/* this is a leaf, now advance the iterator */
	if (ring_iter)
		ring_buffer_read(ring_iter, NULL);

	return next;
368
369
}

370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
/* Signal a overhead of time execution to the output */
static int
print_graph_overhead(unsigned long long duration, struct trace_seq *s)
{
	/* If duration disappear, we don't need anything */
	if (!(tracer_flags.val & TRACE_GRAPH_PRINT_DURATION))
		return 1;

	/* Non nested entry or return */
	if (duration == -1)
		return trace_seq_printf(s, "  ");

	if (tracer_flags.val & TRACE_GRAPH_PRINT_OVERHEAD) {
		/* Duration exceeded 100 msecs */
		if (duration > 100000ULL)
			return trace_seq_printf(s, "! ");

		/* Duration exceeded 10 msecs */
		if (duration > 10000ULL)
			return trace_seq_printf(s, "+ ");
	}

	return trace_seq_printf(s, "  ");
}

395
396
397
398
399
400
401
402
403
404
405
static int print_graph_abs_time(u64 t, struct trace_seq *s)
{
	unsigned long usecs_rem;

	usecs_rem = do_div(t, NSEC_PER_SEC);
	usecs_rem /= 1000;

	return trace_seq_printf(s, "%5lu.%06lu |  ",
			(unsigned long)t, usecs_rem);
}

406
static enum print_line_t
407
print_graph_irq(struct trace_iterator *iter, unsigned long addr,
408
		enum trace_type type, int cpu, pid_t pid)
409
410
{
	int ret;
411
	struct trace_seq *s = &iter->seq;
412
413
414
415
416

	if (addr < (unsigned long)__irqentry_text_start ||
		addr >= (unsigned long)__irqentry_text_end)
		return TRACE_TYPE_UNHANDLED;

417
418
419
420
421
422
423
	/* Absolute time */
	if (tracer_flags.val & TRACE_GRAPH_PRINT_ABS_TIME) {
		ret = print_graph_abs_time(iter->ts, s);
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}

424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
	/* Cpu */
	if (tracer_flags.val & TRACE_GRAPH_PRINT_CPU) {
		ret = print_graph_cpu(s, cpu);
		if (ret == TRACE_TYPE_PARTIAL_LINE)
			return TRACE_TYPE_PARTIAL_LINE;
	}
	/* Proc */
	if (tracer_flags.val & TRACE_GRAPH_PRINT_PROC) {
		ret = print_graph_proc(s, pid);
		if (ret == TRACE_TYPE_PARTIAL_LINE)
			return TRACE_TYPE_PARTIAL_LINE;
		ret = trace_seq_printf(s, " | ");
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}
439

440
441
442
443
	/* No overhead */
	ret = print_graph_overhead(-1, s);
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;
444

445
446
447
448
449
450
451
452
453
454
455
456
	if (type == TRACE_GRAPH_ENT)
		ret = trace_seq_printf(s, "==========>");
	else
		ret = trace_seq_printf(s, "<==========");

	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

	/* Don't close the duration column if haven't one */
	if (tracer_flags.val & TRACE_GRAPH_PRINT_DURATION)
		trace_seq_printf(s, " |");
	ret = trace_seq_printf(s, "\n");
457
458
459
460
461

	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;
	return TRACE_TYPE_HANDLED;
}
462

463
464
enum print_line_t
trace_print_graph_duration(unsigned long long duration, struct trace_seq *s)
465
466
{
	unsigned long nsecs_rem = do_div(duration, 1000);
467
468
469
470
471
472
473
474
475
	/* log10(ULONG_MAX) + '\0' */
	char msecs_str[21];
	char nsecs_str[5];
	int ret, len;
	int i;

	sprintf(msecs_str, "%lu", (unsigned long) duration);

	/* Print msecs */
476
	ret = trace_seq_printf(s, "%s", msecs_str);
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

	len = strlen(msecs_str);

	/* Print nsecs (we don't want to exceed 7 numbers) */
	if (len < 7) {
		snprintf(nsecs_str, 8 - len, "%03lu", nsecs_rem);
		ret = trace_seq_printf(s, ".%s", nsecs_str);
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
		len += strlen(nsecs_str);
	}

	ret = trace_seq_printf(s, " us ");
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

	/* Print remaining spaces to fit the row's width */
	for (i = len; i < 7; i++) {
		ret = trace_seq_printf(s, " ");
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}
501
502
503
504
505
506
507
508
509
510
511
	return TRACE_TYPE_HANDLED;
}

static enum print_line_t
print_graph_duration(unsigned long long duration, struct trace_seq *s)
{
	int ret;

	ret = trace_print_graph_duration(duration, s);
	if (ret != TRACE_TYPE_HANDLED)
		return ret;
512
513
514
515
516

	ret = trace_seq_printf(s, "|  ");
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

517
	return TRACE_TYPE_HANDLED;
518
519
520
}

/* Case of a leaf function on its call entry */
521
static enum print_line_t
522
print_graph_entry_leaf(struct trace_iterator *iter,
523
524
		struct ftrace_graph_ent_entry *entry,
		struct ftrace_graph_ret_entry *ret_entry, struct trace_seq *s)
525
{
526
	struct fgraph_data *data = iter->private;
527
528
529
	struct ftrace_graph_ret *graph_ret;
	struct ftrace_graph_ent *call;
	unsigned long long duration;
530
	int ret;
531
	int i;
532

533
534
535
536
	graph_ret = &ret_entry->ret;
	call = &entry->graph_ent;
	duration = graph_ret->rettime - graph_ret->calltime;

537
538
539
540
541
542
543
544
545
546
547
548
	if (data) {
		int cpu = iter->cpu;
		int *depth = &(per_cpu_ptr(data, cpu)->depth);

		/*
		 * Comments display at + 1 to depth. Since
		 * this is a leaf function, keep the comments
		 * equal to this depth.
		 */
		*depth = call->depth - 1;
	}

549
	/* Overhead */
550
551
552
	ret = print_graph_overhead(duration, s);
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;
553
554

	/* Duration */
555
556
557
558
559
	if (tracer_flags.val & TRACE_GRAPH_PRINT_DURATION) {
		ret = print_graph_duration(duration, s);
		if (ret == TRACE_TYPE_PARTIAL_LINE)
			return TRACE_TYPE_PARTIAL_LINE;
	}
560

561
562
563
564
565
566
567
	/* Function */
	for (i = 0; i < call->depth * TRACE_GRAPH_INDENT; i++) {
		ret = trace_seq_printf(s, " ");
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}

568
	ret = trace_seq_printf(s, "%pf();\n", (void *)call->func);
569
570
571
572
573
574
575
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

	return TRACE_TYPE_HANDLED;
}

static enum print_line_t
576
577
578
print_graph_entry_nested(struct trace_iterator *iter,
			 struct ftrace_graph_ent_entry *entry,
			 struct trace_seq *s, int cpu)
579
580
{
	struct ftrace_graph_ent *call = &entry->graph_ent;
581
582
583
584
585
586
587
588
589
590
	struct fgraph_data *data = iter->private;
	int ret;
	int i;

	if (data) {
		int cpu = iter->cpu;
		int *depth = &(per_cpu_ptr(data, cpu)->depth);

		*depth = call->depth;
	}
591
592

	/* No overhead */
593
594
595
	ret = print_graph_overhead(-1, s);
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;
596

597
598
	/* No time */
	if (tracer_flags.val & TRACE_GRAPH_PRINT_DURATION) {
599
600
601
602
603
		ret = trace_seq_printf(s, "            |  ");
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}

604
	/* Function */
605
606
	for (i = 0; i < call->depth * TRACE_GRAPH_INDENT; i++) {
		ret = trace_seq_printf(s, " ");
607
608
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
609
610
	}

611
	ret = trace_seq_printf(s, "%pf() {\n", (void *)call->func);
612
613
614
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

615
616
617
618
619
	/*
	 * we already consumed the current entry to check the next one
	 * and see if this is a leaf.
	 */
	return TRACE_TYPE_NO_CONSUME;
620
621
}

622
static enum print_line_t
623
624
print_graph_prologue(struct trace_iterator *iter, struct trace_seq *s,
		     int type, unsigned long addr)
625
{
626
	struct fgraph_data *data = iter->private;
627
	struct trace_entry *ent = iter->ent;
628
629
	int cpu = iter->cpu;
	int ret;
630

631
	/* Pid */
632
	if (verif_pid(s, ent->pid, cpu, data) == TRACE_TYPE_PARTIAL_LINE)
633
634
		return TRACE_TYPE_PARTIAL_LINE;

635
636
637
638
639
640
	if (type) {
		/* Interrupt */
		ret = print_graph_irq(iter, addr, type, cpu, ent->pid);
		if (ret == TRACE_TYPE_PARTIAL_LINE)
			return TRACE_TYPE_PARTIAL_LINE;
	}
641

642
643
644
645
646
647
648
	/* Absolute time */
	if (tracer_flags.val & TRACE_GRAPH_PRINT_ABS_TIME) {
		ret = print_graph_abs_time(iter->ts, s);
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}

649
650
651
	/* Cpu */
	if (tracer_flags.val & TRACE_GRAPH_PRINT_CPU) {
		ret = print_graph_cpu(s, cpu);
652
653
654
655
656
657
		if (ret == TRACE_TYPE_PARTIAL_LINE)
			return TRACE_TYPE_PARTIAL_LINE;
	}

	/* Proc */
	if (tracer_flags.val & TRACE_GRAPH_PRINT_PROC) {
658
		ret = print_graph_proc(s, ent->pid);
659
660
661
662
		if (ret == TRACE_TYPE_PARTIAL_LINE)
			return TRACE_TYPE_PARTIAL_LINE;

		ret = trace_seq_printf(s, " | ");
663
664
665
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}
666

667
668
669
670
671
672
673
674
675
676
677
678
679
680
	return 0;
}

static enum print_line_t
print_graph_entry(struct ftrace_graph_ent_entry *field, struct trace_seq *s,
			struct trace_iterator *iter)
{
	int cpu = iter->cpu;
	struct ftrace_graph_ent *call = &field->graph_ent;
	struct ftrace_graph_ret_entry *leaf_ret;

	if (print_graph_prologue(iter, s, TRACE_GRAPH_ENT, call->func))
		return TRACE_TYPE_PARTIAL_LINE;

681
682
683
	leaf_ret = get_return_for_leaf(iter, field);
	if (leaf_ret)
		return print_graph_entry_leaf(iter, field, leaf_ret, s);
684
	else
685
		return print_graph_entry_nested(iter, field, s, cpu);
686
687
688

}

689
690
static enum print_line_t
print_graph_return(struct ftrace_graph_ret *trace, struct trace_seq *s,
691
		   struct trace_entry *ent, struct trace_iterator *iter)
692
{
693
	unsigned long long duration = trace->rettime - trace->calltime;
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
	struct fgraph_data *data = iter->private;
	pid_t pid = ent->pid;
	int cpu = iter->cpu;
	int ret;
	int i;

	if (data) {
		int cpu = iter->cpu;
		int *depth = &(per_cpu_ptr(data, cpu)->depth);

		/*
		 * Comments display at + 1 to depth. This is the
		 * return from a function, we now want the comments
		 * to display at the same level of the bracket.
		 */
		*depth = trace->depth - 1;
	}
711

712
	if (print_graph_prologue(iter, s, 0, 0))
713
714
		return TRACE_TYPE_PARTIAL_LINE;

715
	/* Overhead */
716
717
718
	ret = print_graph_overhead(duration, s);
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;
719
720

	/* Duration */
721
722
723
724
725
	if (tracer_flags.val & TRACE_GRAPH_PRINT_DURATION) {
		ret = print_graph_duration(duration, s);
		if (ret == TRACE_TYPE_PARTIAL_LINE)
			return TRACE_TYPE_PARTIAL_LINE;
	}
726
727

	/* Closing brace */
728
729
	for (i = 0; i < trace->depth * TRACE_GRAPH_INDENT; i++) {
		ret = trace_seq_printf(s, " ");
730
731
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
732
733
	}

734
	ret = trace_seq_printf(s, "}\n");
735
736
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;
737

738
	/* Overrun */
739
740
741
	if (tracer_flags.val & TRACE_GRAPH_PRINT_OVERRUN) {
		ret = trace_seq_printf(s, " (Overruns: %lu)\n",
					trace->overrun);
742
743
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
744
	}
745

746
	ret = print_graph_irq(iter, trace->func, TRACE_GRAPH_RET, cpu, pid);
747
748
749
	if (ret == TRACE_TYPE_PARTIAL_LINE)
		return TRACE_TYPE_PARTIAL_LINE;

750
751
752
	return TRACE_TYPE_HANDLED;
}

753
static enum print_line_t
754
755
print_graph_comment(struct trace_seq *s,  struct trace_entry *ent,
		    struct trace_iterator *iter)
756
{
757
	unsigned long sym_flags = (trace_flags & TRACE_ITER_SYM_MASK);
758
	struct fgraph_data *data = iter->private;
759
	struct trace_event *event;
760
	int depth = 0;
761
	int ret;
762
763
764
765
	int i;

	if (data)
		depth = per_cpu_ptr(data, iter->cpu)->depth;
766

767
	if (print_graph_prologue(iter, s, 0, 0))
768
769
		return TRACE_TYPE_PARTIAL_LINE;

770
	/* No overhead */
771
772
773
774
775
776
777
	ret = print_graph_overhead(-1, s);
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

	/* No time */
	if (tracer_flags.val & TRACE_GRAPH_PRINT_DURATION) {
		ret = trace_seq_printf(s, "            |  ");
778
779
780
781
782
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}

	/* Indentation */
783
784
	if (depth > 0)
		for (i = 0; i < (depth + 1) * TRACE_GRAPH_INDENT; i++) {
785
786
787
788
789
790
			ret = trace_seq_printf(s, " ");
			if (!ret)
				return TRACE_TYPE_PARTIAL_LINE;
		}

	/* The comment */
791
792
793
794
	ret = trace_seq_printf(s, "/* ");
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
	switch (iter->ent->type) {
	case TRACE_BPRINT:
		ret = trace_print_bprintk_msg_only(iter);
		if (ret != TRACE_TYPE_HANDLED)
			return ret;
		break;
	case TRACE_PRINT:
		ret = trace_print_printk_msg_only(iter);
		if (ret != TRACE_TYPE_HANDLED)
			return ret;
		break;
	default:
		event = ftrace_find_event(ent->type);
		if (!event)
			return TRACE_TYPE_UNHANDLED;

		ret = event->trace(iter, sym_flags);
		if (ret != TRACE_TYPE_HANDLED)
			return ret;
	}
815

816
817
818
819
820
821
	/* Strip ending newline */
	if (s->buffer[s->len - 1] == '\n') {
		s->buffer[s->len - 1] = '\0';
		s->len--;
	}

822
823
824
825
826
827
828
829
	ret = trace_seq_printf(s, " */\n");
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

	return TRACE_TYPE_HANDLED;
}


830
831
832
833
enum print_line_t
print_graph_function(struct trace_iterator *iter)
{
	struct trace_entry *entry = iter->ent;
834
	struct trace_seq *s = &iter->seq;
835

836
837
838
839
	switch (entry->type) {
	case TRACE_GRAPH_ENT: {
		struct ftrace_graph_ent_entry *field;
		trace_assign_type(field, entry);
840
		return print_graph_entry(field, s, iter);
841
842
843
844
	}
	case TRACE_GRAPH_RET: {
		struct ftrace_graph_ret_entry *field;
		trace_assign_type(field, entry);
845
		return print_graph_return(&field->ret, s, entry, iter);
846
847
	}
	default:
848
		return print_graph_comment(s, entry, iter);
849
	}
850
851

	return TRACE_TYPE_HANDLED;
852
853
}

854
855
856
857
static void print_graph_headers(struct seq_file *s)
{
	/* 1st line */
	seq_printf(s, "# ");
858
859
	if (tracer_flags.val & TRACE_GRAPH_PRINT_ABS_TIME)
		seq_printf(s, "     TIME       ");
860
	if (tracer_flags.val & TRACE_GRAPH_PRINT_CPU)
861
		seq_printf(s, "CPU");
862
	if (tracer_flags.val & TRACE_GRAPH_PRINT_PROC)
863
864
865
866
		seq_printf(s, "  TASK/PID      ");
	if (tracer_flags.val & TRACE_GRAPH_PRINT_DURATION)
		seq_printf(s, "  DURATION   ");
	seq_printf(s, "               FUNCTION CALLS\n");
867
868
869

	/* 2nd line */
	seq_printf(s, "# ");
870
871
	if (tracer_flags.val & TRACE_GRAPH_PRINT_ABS_TIME)
		seq_printf(s, "      |         ");
872
	if (tracer_flags.val & TRACE_GRAPH_PRINT_CPU)
873
		seq_printf(s, "|  ");
874
	if (tracer_flags.val & TRACE_GRAPH_PRINT_PROC)
875
876
877
878
		seq_printf(s, "  |    |        ");
	if (tracer_flags.val & TRACE_GRAPH_PRINT_DURATION)
		seq_printf(s, "   |   |      ");
	seq_printf(s, "               |   |   |   |\n");
879
}
880
881
882

static void graph_trace_open(struct trace_iterator *iter)
{
883
884
	/* pid and depth on the last trace processed */
	struct fgraph_data *data = alloc_percpu(struct fgraph_data);
885
886
	int cpu;

887
	if (!data)
888
889
890
		pr_warning("function graph tracer: not enough memory\n");
	else
		for_each_possible_cpu(cpu) {
891
892
			pid_t *pid = &(per_cpu_ptr(data, cpu)->last_pid);
			int *depth = &(per_cpu_ptr(data, cpu)->depth);
893
			*pid = -1;
894
			*depth = 0;
895
896
		}

897
	iter->private = data;
898
899
900
901
}

static void graph_trace_close(struct trace_iterator *iter)
{
902
	free_percpu(iter->private);
903
904
}

905
static struct tracer graph_trace __read_mostly = {
906
	.name		= "function_graph",
907
908
	.open		= graph_trace_open,
	.close		= graph_trace_close,
909
	.wait_pipe	= poll_wait_pipe,
910
911
	.init		= graph_trace_init,
	.reset		= graph_trace_reset,
912
913
	.print_line	= print_graph_function,
	.print_header	= print_graph_headers,
914
	.flags		= &tracer_flags,
915
916
917
#ifdef CONFIG_FTRACE_SELFTEST
	.selftest	= trace_selftest_startup_function_graph,
#endif
918
919
920
921
922
923
924
925
};

static __init int init_graph_trace(void)
{
	return register_tracer(&graph_trace);
}

device_initcall(init_graph_trace);