trace_functions_graph.c 30.2 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
 * 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>
12
#include <linux/slab.h>
13 14 15
#include <linux/fs.h>

#include "trace.h"
16
#include "trace_output.h"
17

18
struct fgraph_cpu_data {
19 20
	pid_t		last_pid;
	int		depth;
21
	int		ignore;
22
	unsigned long	enter_funcs[FTRACE_RETFUNC_DEPTH];
23 24 25 26 27 28 29 30 31 32
};

struct fgraph_data {
	struct fgraph_cpu_data		*cpu_data;

	/* Place to preserve last processed entry. */
	struct ftrace_graph_ent_entry	ent;
	struct ftrace_graph_ret_entry	ret;
	int				failed;
	int				cpu;
33 34
};

35
#define TRACE_GRAPH_INDENT	2
36

37
/* Flag options */
38
#define TRACE_GRAPH_PRINT_OVERRUN	0x1
39 40
#define TRACE_GRAPH_PRINT_CPU		0x2
#define TRACE_GRAPH_PRINT_OVERHEAD	0x4
41
#define TRACE_GRAPH_PRINT_PROC		0x8
42
#define TRACE_GRAPH_PRINT_DURATION	0x10
43
#define TRACE_GRAPH_PRINT_ABS_TIME	0x20
44

45
static struct tracer_opt trace_opts[] = {
46
	/* Display overruns? (for self-debug purpose) */
47 48 49 50 51
	{ 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) },
52 53
	/* Display proc name/pid */
	{ TRACER_OPT(funcgraph-proc, TRACE_GRAPH_PRINT_PROC) },
54 55 56 57
	/* 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) },
58 59 60 61
	{ } /* Empty entry */
};

static struct tracer_flags tracer_flags = {
62
	/* Don't display overruns and proc by default */
63 64
	.val = TRACE_GRAPH_PRINT_CPU | TRACE_GRAPH_PRINT_OVERHEAD |
	       TRACE_GRAPH_PRINT_DURATION,
65 66 67
	.opts = trace_opts
};

68
static struct trace_array *graph_array;
69

70

71 72
/* Add a function return address to the trace stack on thread info.*/
int
73 74
ftrace_push_return_trace(unsigned long ret, unsigned long func, int *depth,
			 unsigned long frame_pointer)
75
{
76
	unsigned long long calltime;
77 78 79 80 81
	int index;

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

82 83 84 85 86 87
	/*
	 * We must make sure the ret_stack is tested before we read
	 * anything else.
	 */
	smp_rmb();

88 89 90 91 92 93
	/* The return trace stack is full */
	if (current->curr_ret_stack == FTRACE_RETFUNC_DEPTH - 1) {
		atomic_inc(&current->trace_overrun);
		return -EBUSY;
	}

94 95
	calltime = trace_clock_local();

96 97 98 99
	index = ++current->curr_ret_stack;
	barrier();
	current->ret_stack[index].ret = ret;
	current->ret_stack[index].func = func;
100
	current->ret_stack[index].calltime = calltime;
101
	current->ret_stack[index].subtime = 0;
102
	current->ret_stack[index].fp = frame_pointer;
103 104 105 106 107 108
	*depth = index;

	return 0;
}

/* Retrieve a function return address to the trace stack on thread info.*/
109
static void
110 111
ftrace_pop_return_trace(struct ftrace_graph_ret *trace, unsigned long *ret,
			unsigned long frame_pointer)
112 113 114 115 116 117 118 119 120 121 122 123 124
{
	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;
	}

125 126 127 128 129 130 131 132 133 134 135 136 137 138 139
#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"
140
		     "  from func %ps return to %lx\n",
141 142 143 144 145 146 147 148 149
		     current->ret_stack[index].fp,
		     frame_pointer,
		     (void *)current->ret_stack[index].func,
		     current->ret_stack[index].ret);
		*ret = (unsigned long)panic;
		return;
	}
#endif

150 151 152 153 154 155 156 157 158 159 160
	*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.
 */
161
unsigned long ftrace_return_to_handler(unsigned long frame_pointer)
162 163 164 165
{
	struct ftrace_graph_ret trace;
	unsigned long ret;

166
	ftrace_pop_return_trace(&trace, &ret, frame_pointer);
167
	trace.rettime = trace_clock_local();
168
	ftrace_graph_return(&trace);
169 170
	barrier();
	current->curr_ret_stack--;
171 172 173 174 175 176 177 178 179 180 181

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

	return ret;
}

182
int __trace_graph_entry(struct trace_array *tr,
183 184 185 186 187 188
				struct ftrace_graph_ent *trace,
				unsigned long flags,
				int pc)
{
	struct ftrace_event_call *call = &event_funcgraph_entry;
	struct ring_buffer_event *event;
189
	struct ring_buffer *buffer = tr->buffer;
190 191
	struct ftrace_graph_ent_entry *entry;

192
	if (unlikely(__this_cpu_read(ftrace_cpu_disabled)))
193 194
		return 0;

195
	event = trace_buffer_lock_reserve(buffer, TRACE_GRAPH_ENT,
196 197 198 199 200
					  sizeof(*entry), flags, pc);
	if (!event)
		return 0;
	entry	= ring_buffer_event_data(event);
	entry->graph_ent			= *trace;
201 202
	if (!filter_current_check_discard(buffer, call, entry, event))
		ring_buffer_unlock_commit(buffer, event);
203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219

	return 1;
}

int trace_graph_entry(struct ftrace_graph_ent *trace)
{
	struct trace_array *tr = graph_array;
	struct trace_array_cpu *data;
	unsigned long flags;
	long disabled;
	int ret;
	int cpu;
	int pc;

	if (!ftrace_trace_task(current))
		return 0;

220 221
	/* trace it when it is-nested-in or is a function enabled. */
	if (!(trace->depth || ftrace_graph_addr(trace->func)))
222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240
		return 0;

	local_irq_save(flags);
	cpu = raw_smp_processor_id();
	data = tr->data[cpu];
	disabled = atomic_inc_return(&data->disabled);
	if (likely(disabled == 1)) {
		pc = preempt_count();
		ret = __trace_graph_entry(tr, trace, flags, pc);
	} else {
		ret = 0;
	}

	atomic_dec(&data->disabled);
	local_irq_restore(flags);

	return ret;
}

241 242 243 244 245 246 247 248
int trace_graph_thresh_entry(struct ftrace_graph_ent *trace)
{
	if (tracing_thresh)
		return 1;
	else
		return trace_graph_entry(trace);
}

249
void __trace_graph_return(struct trace_array *tr,
250 251 252 253 254 255
				struct ftrace_graph_ret *trace,
				unsigned long flags,
				int pc)
{
	struct ftrace_event_call *call = &event_funcgraph_exit;
	struct ring_buffer_event *event;
256
	struct ring_buffer *buffer = tr->buffer;
257 258
	struct ftrace_graph_ret_entry *entry;

259
	if (unlikely(__this_cpu_read(ftrace_cpu_disabled)))
260 261
		return;

262
	event = trace_buffer_lock_reserve(buffer, TRACE_GRAPH_RET,
263 264 265 266 267
					  sizeof(*entry), flags, pc);
	if (!event)
		return;
	entry	= ring_buffer_event_data(event);
	entry->ret				= *trace;
268 269
	if (!filter_current_check_discard(buffer, call, entry, event))
		ring_buffer_unlock_commit(buffer, event);
270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292
}

void trace_graph_return(struct ftrace_graph_ret *trace)
{
	struct trace_array *tr = graph_array;
	struct trace_array_cpu *data;
	unsigned long flags;
	long disabled;
	int cpu;
	int pc;

	local_irq_save(flags);
	cpu = raw_smp_processor_id();
	data = tr->data[cpu];
	disabled = atomic_inc_return(&data->disabled);
	if (likely(disabled == 1)) {
		pc = preempt_count();
		__trace_graph_return(tr, trace, flags, pc);
	}
	atomic_dec(&data->disabled);
	local_irq_restore(flags);
}

293 294 295 296 297 298 299 300 301
void set_graph_array(struct trace_array *tr)
{
	graph_array = tr;

	/* Make graph_array visible before we start tracing */

	smp_mb();
}

302 303 304 305 306 307 308 309 310
void trace_graph_thresh_return(struct ftrace_graph_ret *trace)
{
	if (tracing_thresh &&
	    (trace->rettime - trace->calltime < tracing_thresh))
		return;
	else
		trace_graph_return(trace);
}

311 312
static int graph_trace_init(struct trace_array *tr)
{
313 314
	int ret;

315
	set_graph_array(tr);
316 317 318 319 320 321
	if (tracing_thresh)
		ret = register_ftrace_graph(&trace_graph_thresh_return,
					    &trace_graph_thresh_entry);
	else
		ret = register_ftrace_graph(&trace_graph_return,
					    &trace_graph_entry);
322 323 324 325 326
	if (ret)
		return ret;
	tracing_start_cmdline_record();

	return 0;
327 328 329 330
}

static void graph_trace_reset(struct trace_array *tr)
{
331 332
	tracing_stop_cmdline_record();
	unregister_ftrace_graph();
333 334
}

335
static int max_bytes_for_cpu;
336 337 338 339 340 341

static enum print_line_t
print_graph_cpu(struct trace_seq *s, int cpu)
{
	int ret;

342 343 344 345 346
	/*
	 * Start with a space character - to make it stand out
	 * to the right a bit when trace output is pasted into
	 * email:
	 */
347
	ret = trace_seq_printf(s, " %*d) ", max_bytes_for_cpu, cpu);
348
	if (!ret)
349 350
		return TRACE_TYPE_PARTIAL_LINE;

351 352 353
	return TRACE_TYPE_HANDLED;
}

354 355 356 357 358
#define TRACE_GRAPH_PROCINFO_LENGTH	14

static enum print_line_t
print_graph_proc(struct trace_seq *s, pid_t pid)
{
359
	char comm[TASK_COMM_LEN];
360 361
	/* sign + log10(MAX_INT) + '\0' */
	char pid_str[11];
362 363 364 365
	int spaces = 0;
	int ret;
	int len;
	int i;
366

367
	trace_find_cmdline(pid, comm);
368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396
	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;
}

397

398 399 400
static enum print_line_t
print_graph_lat_fmt(struct trace_seq *s, struct trace_entry *entry)
{
401
	if (!trace_seq_putc(s, ' '))
402 403
		return 0;

404
	return trace_print_lat_fmt(s, entry);
405 406
}

407
/* If the pid changed since the last trace, output this event */
408
static enum print_line_t
409
verif_pid(struct trace_seq *s, pid_t pid, int cpu, struct fgraph_data *data)
410
{
411
	pid_t prev_pid;
412
	pid_t *last_pid;
413
	int ret;
414

415
	if (!data)
416 417
		return TRACE_TYPE_HANDLED;

418
	last_pid = &(per_cpu_ptr(data->cpu_data, cpu)->last_pid);
419 420

	if (*last_pid == pid)
421
		return TRACE_TYPE_HANDLED;
422

423 424
	prev_pid = *last_pid;
	*last_pid = pid;
425

426 427
	if (prev_pid == -1)
		return TRACE_TYPE_HANDLED;
428 429 430 431 432 433 434 435 436
/*
 * Context-switch trace line:

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

 */
	ret = trace_seq_printf(s,
437
		" ------------------------------------------\n");
438
	if (!ret)
439
		return TRACE_TYPE_PARTIAL_LINE;
440 441 442

	ret = print_graph_cpu(s, cpu);
	if (ret == TRACE_TYPE_PARTIAL_LINE)
443
		return TRACE_TYPE_PARTIAL_LINE;
444 445 446

	ret = print_graph_proc(s, prev_pid);
	if (ret == TRACE_TYPE_PARTIAL_LINE)
447
		return TRACE_TYPE_PARTIAL_LINE;
448 449 450

	ret = trace_seq_printf(s, " => ");
	if (!ret)
451
		return TRACE_TYPE_PARTIAL_LINE;
452 453 454

	ret = print_graph_proc(s, pid);
	if (ret == TRACE_TYPE_PARTIAL_LINE)
455
		return TRACE_TYPE_PARTIAL_LINE;
456 457 458 459

	ret = trace_seq_printf(s,
		"\n ------------------------------------------\n\n");
	if (!ret)
460
		return TRACE_TYPE_PARTIAL_LINE;
461

462
	return TRACE_TYPE_HANDLED;
463 464
}

465 466
static struct ftrace_graph_ret_entry *
get_return_for_leaf(struct trace_iterator *iter,
467 468
		struct ftrace_graph_ent_entry *curr)
{
469 470
	struct fgraph_data *data = iter->private;
	struct ring_buffer_iter *ring_iter = NULL;
471 472 473
	struct ring_buffer_event *event;
	struct ftrace_graph_ret_entry *next;

474 475 476 477 478 479 480 481
	/*
	 * If the previous output failed to write to the seq buffer,
	 * then we just reuse the data from before.
	 */
	if (data && data->failed) {
		curr = &data->ent;
		next = &data->ret;
	} else {
482

483 484 485 486 487 488 489 490 491 492
		ring_iter = iter->buffer_iter[iter->cpu];

		/* 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.
			 */
493 494
			ring_buffer_consume(iter->tr->buffer, iter->cpu,
					    NULL, NULL);
495
			event = ring_buffer_peek(iter->tr->buffer, iter->cpu,
496
						 NULL, NULL);
497
		}
498

499 500 501 502
		if (!event)
			return NULL;

		next = ring_buffer_event_data(event);
503

504 505 506 507 508 509 510 511 512
		if (data) {
			/*
			 * Save current and next entries for later reference
			 * if the output fails.
			 */
			data->ent = *curr;
			data->ret = *next;
		}
	}
513 514

	if (next->ent.type != TRACE_GRAPH_RET)
515
		return NULL;
516 517 518

	if (curr->ent.pid != next->ent.pid ||
			curr->graph_ent.func != next->ret.func)
519
		return NULL;
520

521 522 523 524 525
	/* this is a leaf, now advance the iterator */
	if (ring_iter)
		ring_buffer_read(ring_iter, NULL);

	return next;
526 527
}

528 529
/* Signal a overhead of time execution to the output */
static int
530 531
print_graph_overhead(unsigned long long duration, struct trace_seq *s,
		     u32 flags)
532 533
{
	/* If duration disappear, we don't need anything */
534
	if (!(flags & TRACE_GRAPH_PRINT_DURATION))
535 536 537 538 539 540
		return 1;

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

541
	if (flags & TRACE_GRAPH_PRINT_OVERHEAD) {
542 543 544 545 546 547 548 549 550 551 552 553
		/* 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, "  ");
}

554 555 556 557 558 559 560 561 562 563 564
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);
}

565
static enum print_line_t
566
print_graph_irq(struct trace_iterator *iter, unsigned long addr,
567
		enum trace_type type, int cpu, pid_t pid, u32 flags)
568 569
{
	int ret;
570
	struct trace_seq *s = &iter->seq;
571 572 573 574 575

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

576
	/* Absolute time */
577
	if (flags & TRACE_GRAPH_PRINT_ABS_TIME) {
578 579 580 581 582
		ret = print_graph_abs_time(iter->ts, s);
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}

583
	/* Cpu */
584
	if (flags & TRACE_GRAPH_PRINT_CPU) {
585 586 587 588
		ret = print_graph_cpu(s, cpu);
		if (ret == TRACE_TYPE_PARTIAL_LINE)
			return TRACE_TYPE_PARTIAL_LINE;
	}
589

590
	/* Proc */
591
	if (flags & TRACE_GRAPH_PRINT_PROC) {
592 593 594 595 596 597 598
		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;
	}
599

600
	/* No overhead */
601
	ret = print_graph_overhead(-1, s, flags);
602 603
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;
604

605 606 607 608 609 610 611 612 613
	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 */
614
	if (flags & TRACE_GRAPH_PRINT_DURATION)
615 616
		trace_seq_printf(s, " |");
	ret = trace_seq_printf(s, "\n");
617 618 619 620 621

	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;
	return TRACE_TYPE_HANDLED;
}
622

623 624
enum print_line_t
trace_print_graph_duration(unsigned long long duration, struct trace_seq *s)
625 626
{
	unsigned long nsecs_rem = do_div(duration, 1000);
627 628 629 630 631 632 633 634 635
	/* 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 */
636
	ret = trace_seq_printf(s, "%s", msecs_str);
637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660
	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;
	}
661 662 663 664 665 666 667 668 669 670 671
	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;
672 673 674 675 676

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

677
	return TRACE_TYPE_HANDLED;
678 679 680
}

/* Case of a leaf function on its call entry */
681
static enum print_line_t
682
print_graph_entry_leaf(struct trace_iterator *iter,
683
		struct ftrace_graph_ent_entry *entry,
684 685
		struct ftrace_graph_ret_entry *ret_entry,
		struct trace_seq *s, u32 flags)
686
{
687
	struct fgraph_data *data = iter->private;
688 689 690
	struct ftrace_graph_ret *graph_ret;
	struct ftrace_graph_ent *call;
	unsigned long long duration;
691
	int ret;
692
	int i;
693

694 695 696 697
	graph_ret = &ret_entry->ret;
	call = &entry->graph_ent;
	duration = graph_ret->rettime - graph_ret->calltime;

698
	if (data) {
699
		struct fgraph_cpu_data *cpu_data;
700
		int cpu = iter->cpu;
701 702

		cpu_data = per_cpu_ptr(data->cpu_data, cpu);
703 704 705 706 707 708

		/*
		 * Comments display at + 1 to depth. Since
		 * this is a leaf function, keep the comments
		 * equal to this depth.
		 */
709 710 711 712 713
		cpu_data->depth = call->depth - 1;

		/* No need to keep this function around for this depth */
		if (call->depth < FTRACE_RETFUNC_DEPTH)
			cpu_data->enter_funcs[call->depth] = 0;
714 715
	}

716
	/* Overhead */
717
	ret = print_graph_overhead(duration, s, flags);
718 719
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;
720 721

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

728 729 730 731 732 733 734
	/* Function */
	for (i = 0; i < call->depth * TRACE_GRAPH_INDENT; i++) {
		ret = trace_seq_printf(s, " ");
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}

735
	ret = trace_seq_printf(s, "%ps();\n", (void *)call->func);
736 737 738 739 740 741 742
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

	return TRACE_TYPE_HANDLED;
}

static enum print_line_t
743 744
print_graph_entry_nested(struct trace_iterator *iter,
			 struct ftrace_graph_ent_entry *entry,
745
			 struct trace_seq *s, int cpu, u32 flags)
746 747
{
	struct ftrace_graph_ent *call = &entry->graph_ent;
748 749 750 751 752
	struct fgraph_data *data = iter->private;
	int ret;
	int i;

	if (data) {
753
		struct fgraph_cpu_data *cpu_data;
754 755
		int cpu = iter->cpu;

756 757 758 759 760 761
		cpu_data = per_cpu_ptr(data->cpu_data, cpu);
		cpu_data->depth = call->depth;

		/* Save this function pointer to see if the exit matches */
		if (call->depth < FTRACE_RETFUNC_DEPTH)
			cpu_data->enter_funcs[call->depth] = call->func;
762
	}
763 764

	/* No overhead */
765
	ret = print_graph_overhead(-1, s, flags);
766 767
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;
768

769
	/* No time */
770
	if (flags & TRACE_GRAPH_PRINT_DURATION) {
771 772 773 774 775
		ret = trace_seq_printf(s, "            |  ");
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}

776
	/* Function */
777 778
	for (i = 0; i < call->depth * TRACE_GRAPH_INDENT; i++) {
		ret = trace_seq_printf(s, " ");
779 780
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
781 782
	}

783
	ret = trace_seq_printf(s, "%ps() {\n", (void *)call->func);
784 785 786
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

787 788 789 790 791
	/*
	 * we already consumed the current entry to check the next one
	 * and see if this is a leaf.
	 */
	return TRACE_TYPE_NO_CONSUME;
792 793
}

794
static enum print_line_t
795
print_graph_prologue(struct trace_iterator *iter, struct trace_seq *s,
796
		     int type, unsigned long addr, u32 flags)
797
{
798
	struct fgraph_data *data = iter->private;
799
	struct trace_entry *ent = iter->ent;
800 801
	int cpu = iter->cpu;
	int ret;
802

803
	/* Pid */
804
	if (verif_pid(s, ent->pid, cpu, data) == TRACE_TYPE_PARTIAL_LINE)
805 806
		return TRACE_TYPE_PARTIAL_LINE;

807 808
	if (type) {
		/* Interrupt */
809
		ret = print_graph_irq(iter, addr, type, cpu, ent->pid, flags);
810 811 812
		if (ret == TRACE_TYPE_PARTIAL_LINE)
			return TRACE_TYPE_PARTIAL_LINE;
	}
813

814
	/* Absolute time */
815
	if (flags & TRACE_GRAPH_PRINT_ABS_TIME) {
816 817 818 819 820
		ret = print_graph_abs_time(iter->ts, s);
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}

821
	/* Cpu */
822
	if (flags & TRACE_GRAPH_PRINT_CPU) {
823
		ret = print_graph_cpu(s, cpu);
824 825 826 827 828
		if (ret == TRACE_TYPE_PARTIAL_LINE)
			return TRACE_TYPE_PARTIAL_LINE;
	}

	/* Proc */
829
	if (flags & TRACE_GRAPH_PRINT_PROC) {
830
		ret = print_graph_proc(s, ent->pid);
831 832 833 834
		if (ret == TRACE_TYPE_PARTIAL_LINE)
			return TRACE_TYPE_PARTIAL_LINE;

		ret = trace_seq_printf(s, " | ");
835 836 837
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}
838

839 840 841 842 843 844 845
	/* Latency format */
	if (trace_flags & TRACE_ITER_LATENCY_FMT) {
		ret = print_graph_lat_fmt(s, ent);
		if (ret == TRACE_TYPE_PARTIAL_LINE)
			return TRACE_TYPE_PARTIAL_LINE;
	}

846 847 848 849 850
	return 0;
}

static enum print_line_t
print_graph_entry(struct ftrace_graph_ent_entry *field, struct trace_seq *s,
851
			struct trace_iterator *iter, u32 flags)
852
{
853
	struct fgraph_data *data = iter->private;
854 855
	struct ftrace_graph_ent *call = &field->graph_ent;
	struct ftrace_graph_ret_entry *leaf_ret;
856 857
	static enum print_line_t ret;
	int cpu = iter->cpu;
858

859
	if (print_graph_prologue(iter, s, TRACE_GRAPH_ENT, call->func, flags))
860 861
		return TRACE_TYPE_PARTIAL_LINE;

862 863
	leaf_ret = get_return_for_leaf(iter, field);
	if (leaf_ret)
864
		ret = print_graph_entry_leaf(iter, field, leaf_ret, s, flags);
865
	else
866
		ret = print_graph_entry_nested(iter, field, s, cpu, flags);
867

868 869 870 871 872 873 874 875 876 877 878 879 880
	if (data) {
		/*
		 * If we failed to write our output, then we need to make
		 * note of it. Because we already consumed our entry.
		 */
		if (s->full) {
			data->failed = 1;
			data->cpu = cpu;
		} else
			data->failed = 0;
	}

	return ret;
881 882
}

883 884
static enum print_line_t
print_graph_return(struct ftrace_graph_ret *trace, struct trace_seq *s,
885 886
		   struct trace_entry *ent, struct trace_iterator *iter,
		   u32 flags)
887
{
888
	unsigned long long duration = trace->rettime - trace->calltime;
889 890 891
	struct fgraph_data *data = iter->private;
	pid_t pid = ent->pid;
	int cpu = iter->cpu;
892
	int func_match = 1;
893 894 895 896
	int ret;
	int i;

	if (data) {
897 898 899 900
		struct fgraph_cpu_data *cpu_data;
		int cpu = iter->cpu;

		cpu_data = per_cpu_ptr(data->cpu_data, cpu);
901 902 903 904 905 906

		/*
		 * 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.
		 */
907 908 909 910 911 912 913
		cpu_data->depth = trace->depth - 1;

		if (trace->depth < FTRACE_RETFUNC_DEPTH) {
			if (cpu_data->enter_funcs[trace->depth] != trace->func)
				func_match = 0;
			cpu_data->enter_funcs[trace->depth] = 0;
		}
914
	}
915

916
	if (print_graph_prologue(iter, s, 0, 0, flags))
917 918
		return TRACE_TYPE_PARTIAL_LINE;

919
	/* Overhead */
920
	ret = print_graph_overhead(duration, s, flags);
921 922
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;
923 924

	/* Duration */
925
	if (flags & TRACE_GRAPH_PRINT_DURATION) {
926 927 928 929
		ret = print_graph_duration(duration, s);
		if (ret == TRACE_TYPE_PARTIAL_LINE)
			return TRACE_TYPE_PARTIAL_LINE;
	}
930 931

	/* Closing brace */
932 933
	for (i = 0; i < trace->depth * TRACE_GRAPH_INDENT; i++) {
		ret = trace_seq_printf(s, " ");
934 935
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
936 937
	}

938 939 940 941 942 943 944 945 946 947 948
	/*
	 * If the return function does not have a matching entry,
	 * then the entry was lost. Instead of just printing
	 * the '}' and letting the user guess what function this
	 * belongs to, write out the function name.
	 */
	if (func_match) {
		ret = trace_seq_printf(s, "}\n");
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	} else {
949
		ret = trace_seq_printf(s, "} /* %ps */\n", (void *)trace->func);
950 951 952
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}
953

954
	/* Overrun */
955
	if (flags & TRACE_GRAPH_PRINT_OVERRUN) {
956 957
		ret = trace_seq_printf(s, " (Overruns: %lu)\n",
					trace->overrun);
958 959
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
960
	}
961

962 963
	ret = print_graph_irq(iter, trace->func, TRACE_GRAPH_RET,
			      cpu, pid, flags);
964 965 966
	if (ret == TRACE_TYPE_PARTIAL_LINE)
		return TRACE_TYPE_PARTIAL_LINE;

967 968 969
	return TRACE_TYPE_HANDLED;
}

970
static enum print_line_t
971 972
print_graph_comment(struct trace_seq *s, struct trace_entry *ent,
		    struct trace_iterator *iter, u32 flags)
973
{
974
	unsigned long sym_flags = (trace_flags & TRACE_ITER_SYM_MASK);
975
	struct fgraph_data *data = iter->private;
976
	struct trace_event *event;
977
	int depth = 0;
978
	int ret;
979 980 981
	int i;

	if (data)
982
		depth = per_cpu_ptr(data->cpu_data, iter->cpu)->depth;
983

984
	if (print_graph_prologue(iter, s, 0, 0, flags))
985 986
		return TRACE_TYPE_PARTIAL_LINE;

987
	/* No overhead */
988
	ret = print_graph_overhead(-1, s, flags);
989 990 991 992
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

	/* No time */
993
	if (flags & TRACE_GRAPH_PRINT_DURATION) {
994
		ret = trace_seq_printf(s, "            |  ");
995 996 997 998 999
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}

	/* Indentation */
1000 1001
	if (depth > 0)
		for (i = 0; i < (depth + 1) * TRACE_GRAPH_INDENT; i++) {
1002 1003 1004 1005 1006 1007
			ret = trace_seq_printf(s, " ");
			if (!ret)
				return TRACE_TYPE_PARTIAL_LINE;
		}

	/* The comment */
1008 1009 1010 1011
	ret = trace_seq_printf(s, "/* ");
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027
	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;

1028
		ret = event->funcs->trace(iter, sym_flags, event);
1029 1030 1031
		if (ret != TRACE_TYPE_HANDLED)
			return ret;
	}
1032

1033 1034 1035 1036 1037 1038
	/* Strip ending newline */
	if (s->buffer[s->len - 1] == '\n') {
		s->buffer[s->len - 1] = '\0';
		s->len--;
	}

1039 1040 1041 1042 1043 1044 1045 1046
	ret = trace_seq_printf(s, " */\n");
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

	return TRACE_TYPE_HANDLED;
}


1047
enum print_line_t
1048
print_graph_function_flags(struct trace_iterator *iter, u32 flags)
1049
{
1050 1051
	struct ftrace_graph_ent_entry *field;
	struct fgraph_data *data = iter->private;
1052
	struct trace_entry *entry = iter->ent;
1053
	struct trace_seq *s = &iter->seq;
1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068
	int cpu = iter->cpu;
	int ret;

	if (data && per_cpu_ptr(data->cpu_data, cpu)->ignore) {
		per_cpu_ptr(data->cpu_data, cpu)->ignore = 0;
		return TRACE_TYPE_HANDLED;
	}

	/*
	 * If the last output failed, there's a possibility we need
	 * to print out the missing entry which would never go out.
	 */
	if (data && data->failed) {
		field = &data->ent;
		iter->cpu = data->cpu;
1069
		ret = print_graph_entry(field, s, iter, flags);
1070 1071 1072 1073 1074 1075 1076
		if (ret == TRACE_TYPE_HANDLED && iter->cpu != cpu) {
			per_cpu_ptr(data->cpu_data, iter->cpu)->ignore = 1;
			ret = TRACE_TYPE_NO_CONSUME;
		}
		iter->cpu = cpu;
		return ret;
	}
1077

1078 1079
	switch (entry->type) {
	case TRACE_GRAPH_ENT: {
1080 1081 1082 1083 1084 1085
		/*
		 * print_graph_entry() may consume the current event,
		 * thus @field may become invalid, so we need to save it.
		 * sizeof(struct ftrace_graph_ent_entry) is very small,
		 * it can be safely saved at the stack.
		 */
1086
		struct ftrace_graph_ent_entry saved;
1087
		trace_assign_type(field, entry);
1088
		saved = *field;
1089
		return print_graph_entry(&saved, s, iter, flags);
1090 1091 1092 1093
	}
	case TRACE_GRAPH_RET: {
		struct ftrace_graph_ret_entry *field;
		trace_assign_type(field, entry);
1094
		return print_graph_return(&field->ret, s, entry, iter, flags);
1095
	}
1096 1097 1098 1099 1100
	case TRACE_STACK:
	case TRACE_FN:
		/* dont trace stack and functions as comments */
		return TRACE_TYPE_UNHANDLED;

1101
	default:
1102
		return print_graph_comment(s, entry, iter, flags);
1103
	}
1104 1105

	return TRACE_TYPE_HANDLED;
1106 1107
}

1108 1109 1110 1111 1112 1113
static enum print_line_t
print_graph_function(struct trace_iterator *iter)
{
	return print_graph_function_flags(iter, tracer_flags.val);
}

1114
static enum print_line_t
1115 1116
print_graph_function_event(struct trace_iterator *iter, int flags,
			   struct trace_event *event)
1117 1118 1119 1120
{
	return print_graph_function(iter);
}

1121
static void print_lat_header(struct seq_file *s, u32 flags)
1122 1123 1124 1125 1126 1127
{
	static const char spaces[] = "                "	/* 16 spaces */
		"    "					/* 4 spaces */
		"                 ";			/* 17 spaces */
	int size = 0;

1128
	if (flags & TRACE_GRAPH_PRINT_ABS_TIME)
1129
		size += 16;
1130
	if (flags & TRACE_GRAPH_PRINT_CPU)
1131
		size += 4;
1132
	if (flags & TRACE_GRAPH_PRINT_PROC)
1133 1134 1135 1136 1137 1138
		size += 17;

	seq_printf(s, "#%.*s  _-----=> irqs-off        \n", size, spaces);
	seq_printf(s, "#%.*s / _----=> need-resched    \n", size, spaces);
	seq_printf(s, "#%.*s| / _---=> hardirq/softirq \n", size, spaces);
	seq_printf(s, "#%.*s|| / _--=> preempt-depth   \n", size, spaces);
1139 1140
	seq_printf(s, "#%.*s||| / _-=> lock-depth      \n", size, spaces);
	seq_printf(s, "#%.*s|||| /                     \n", size, spaces);
1141 1142
}

1143
void print_graph_headers_flags(struct seq_file *s, u32 flags)
1144
{
1145 1146 1147
	int lat = trace_flags & TRACE_ITER_LATENCY_FMT;

	if (lat)
1148
		print_lat_header(s, flags);
1149

1150
	/* 1st line */
1151
	seq_printf(s, "#");
1152
	if (flags & TRACE_GRAPH_PRINT_ABS_TIME)
1153
		seq_printf(s, "     TIME       ");
1154
	if (flags & TRACE_GRAPH_PRINT_CPU)
1155
		seq_printf(s, " CPU");
1156
	if (flags & TRACE_GRAPH_PRINT_PROC)
1157 1158
		seq_printf(s, "  TASK/PID       ");
	if (lat)
1159
		seq_printf(s, "|||||");
1160
	if (flags & TRACE_GRAPH_PRINT_DURATION)
1161 1162
		seq_printf(s, "  DURATION   ");
	seq_printf(s, "               FUNCTION CALLS\n");
1163 1164

	/* 2nd line */
1165
	seq_printf(s, "#");
1166
	if (flags & TRACE_GRAPH_PRINT_ABS_TIME)
1167
		seq_printf(s, "      |         ");