trace_functions_graph.c 27.7 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
struct fgraph_cpu_data {
18 19
	pid_t		last_pid;
	int		depth;
20 21 22 23 24 25 26 27 28 29 30
	int		ignore;
};

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;
31 32
};

33
#define TRACE_GRAPH_INDENT	2
34

35
/* Flag options */
36
#define TRACE_GRAPH_PRINT_OVERRUN	0x1
37 38
#define TRACE_GRAPH_PRINT_CPU		0x2
#define TRACE_GRAPH_PRINT_OVERHEAD	0x4
39
#define TRACE_GRAPH_PRINT_PROC		0x8
40 41
#define TRACE_GRAPH_PRINT_DURATION	0x10
#define TRACE_GRAPH_PRINT_ABS_TIME	0X20
42

43
static struct tracer_opt trace_opts[] = {
44
	/* Display overruns? (for self-debug purpose) */
45 46 47 48 49
	{ 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) },
50 51
	/* Display proc name/pid */
	{ TRACER_OPT(funcgraph-proc, TRACE_GRAPH_PRINT_PROC) },
52 53 54 55
	/* 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) },
56 57 58 59
	{ } /* Empty entry */
};

static struct tracer_flags tracer_flags = {
60
	/* Don't display overruns and proc by default */
61 62
	.val = TRACE_GRAPH_PRINT_CPU | TRACE_GRAPH_PRINT_OVERHEAD |
	       TRACE_GRAPH_PRINT_DURATION,
63 64 65
	.opts = trace_opts
};

66
static struct trace_array *graph_array;
67

68

69 70
/* Add a function return address to the trace stack on thread info.*/
int
71 72
ftrace_push_return_trace(unsigned long ret, unsigned long func, int *depth,
			 unsigned long frame_pointer)
73
{
74
	unsigned long long calltime;
75 76 77 78 79
	int index;

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

80 81 82 83 84 85
	/*
	 * We must make sure the ret_stack is tested before we read
	 * anything else.
	 */
	smp_rmb();

86 87 88 89 90 91
	/* The return trace stack is full */
	if (current->curr_ret_stack == FTRACE_RETFUNC_DEPTH - 1) {
		atomic_inc(&current->trace_overrun);
		return -EBUSY;
	}

92 93
	calltime = trace_clock_local();

94 95 96 97
	index = ++current->curr_ret_stack;
	barrier();
	current->ret_stack[index].ret = ret;
	current->ret_stack[index].func = func;
98
	current->ret_stack[index].calltime = calltime;
99
	current->ret_stack[index].subtime = 0;
100
	current->ret_stack[index].fp = frame_pointer;
101 102 103 104 105 106
	*depth = index;

	return 0;
}

/* Retrieve a function return address to the trace stack on thread info.*/
107
static void
108 109
ftrace_pop_return_trace(struct ftrace_graph_ret *trace, unsigned long *ret,
			unsigned long frame_pointer)
110 111 112 113 114 115 116 117 118 119 120 121 122
{
	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;
	}

123 124 125 126 127 128 129 130 131 132 133 134 135 136 137
#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"
138
		     "  from func %ps return to %lx\n",
139 140 141 142 143 144 145 146 147
		     current->ret_stack[index].fp,
		     frame_pointer,
		     (void *)current->ret_stack[index].func,
		     current->ret_stack[index].ret);
		*ret = (unsigned long)panic;
		return;
	}
#endif

148 149 150 151 152 153 154 155 156 157 158
	*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.
 */
159
unsigned long ftrace_return_to_handler(unsigned long frame_pointer)
160 161 162 163
{
	struct ftrace_graph_ret trace;
	unsigned long ret;

164
	ftrace_pop_return_trace(&trace, &ret, frame_pointer);
165
	trace.rettime = trace_clock_local();
166
	ftrace_graph_return(&trace);
167 168
	barrier();
	current->curr_ret_stack--;
169 170 171 172 173 174 175 176 177 178 179

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

	return ret;
}

180 181 182 183 184 185 186
static int __trace_graph_entry(struct trace_array *tr,
				struct ftrace_graph_ent *trace,
				unsigned long flags,
				int pc)
{
	struct ftrace_event_call *call = &event_funcgraph_entry;
	struct ring_buffer_event *event;
187
	struct ring_buffer *buffer = tr->buffer;
188 189
	struct ftrace_graph_ent_entry *entry;

190
	if (unlikely(__this_cpu_read(per_cpu_var(ftrace_cpu_disabled))))
191 192
		return 0;

193
	event = trace_buffer_lock_reserve(buffer, TRACE_GRAPH_ENT,
194 195 196 197 198
					  sizeof(*entry), flags, pc);
	if (!event)
		return 0;
	entry	= ring_buffer_event_data(event);
	entry->graph_ent			= *trace;
199 200
	if (!filter_current_check_discard(buffer, call, entry, event))
		ring_buffer_unlock_commit(buffer, event);
201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217

	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;

218 219
	/* trace it when it is-nested-in or is a function enabled. */
	if (!(trace->depth || ftrace_graph_addr(trace->func)))
220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245
		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;
}

static void __trace_graph_return(struct trace_array *tr,
				struct ftrace_graph_ret *trace,
				unsigned long flags,
				int pc)
{
	struct ftrace_event_call *call = &event_funcgraph_exit;
	struct ring_buffer_event *event;
246
	struct ring_buffer *buffer = tr->buffer;
247 248
	struct ftrace_graph_ret_entry *entry;

249
	if (unlikely(__this_cpu_read(per_cpu_var(ftrace_cpu_disabled))))
250 251
		return;

252
	event = trace_buffer_lock_reserve(buffer, TRACE_GRAPH_RET,
253 254 255 256 257
					  sizeof(*entry), flags, pc);
	if (!event)
		return;
	entry	= ring_buffer_event_data(event);
	entry->ret				= *trace;
258 259
	if (!filter_current_check_discard(buffer, call, entry, event))
		ring_buffer_unlock_commit(buffer, event);
260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282
}

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);
}

283 284 285 286 287 288 289 290 291
void set_graph_array(struct trace_array *tr)
{
	graph_array = tr;

	/* Make graph_array visible before we start tracing */

	smp_mb();
}

292 293
static int graph_trace_init(struct trace_array *tr)
{
294 295
	int ret;

296
	set_graph_array(tr);
297 298
	ret = register_ftrace_graph(&trace_graph_return,
				    &trace_graph_entry);
299 300 301 302 303
	if (ret)
		return ret;
	tracing_start_cmdline_record();

	return 0;
304 305 306 307
}

static void graph_trace_reset(struct trace_array *tr)
{
308 309
	tracing_stop_cmdline_record();
	unregister_ftrace_graph();
310 311
}

312
static int max_bytes_for_cpu;
313 314 315 316 317 318

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

319 320 321 322 323
	/*
	 * Start with a space character - to make it stand out
	 * to the right a bit when trace output is pasted into
	 * email:
	 */
324
	ret = trace_seq_printf(s, " %*d) ", max_bytes_for_cpu, cpu);
325
	if (!ret)
326 327
		return TRACE_TYPE_PARTIAL_LINE;

328 329 330
	return TRACE_TYPE_HANDLED;
}

331 332 333 334 335
#define TRACE_GRAPH_PROCINFO_LENGTH	14

static enum print_line_t
print_graph_proc(struct trace_seq *s, pid_t pid)
{
336
	char comm[TASK_COMM_LEN];
337 338
	/* sign + log10(MAX_INT) + '\0' */
	char pid_str[11];
339 340 341 342
	int spaces = 0;
	int ret;
	int len;
	int i;
343

344
	trace_find_cmdline(pid, comm);
345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373
	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;
}

374

375 376 377
static enum print_line_t
print_graph_lat_fmt(struct trace_seq *s, struct trace_entry *entry)
{
378
	if (!trace_seq_putc(s, ' '))
379 380
		return 0;

381
	return trace_print_lat_fmt(s, entry);
382 383
}

384
/* If the pid changed since the last trace, output this event */
385
static enum print_line_t
386
verif_pid(struct trace_seq *s, pid_t pid, int cpu, struct fgraph_data *data)
387
{
388
	pid_t prev_pid;
389
	pid_t *last_pid;
390
	int ret;
391

392
	if (!data)
393 394
		return TRACE_TYPE_HANDLED;

395
	last_pid = &(per_cpu_ptr(data->cpu_data, cpu)->last_pid);
396 397

	if (*last_pid == pid)
398
		return TRACE_TYPE_HANDLED;
399

400 401
	prev_pid = *last_pid;
	*last_pid = pid;
402

403 404
	if (prev_pid == -1)
		return TRACE_TYPE_HANDLED;
405 406 407 408 409 410 411 412 413
/*
 * Context-switch trace line:

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

 */
	ret = trace_seq_printf(s,
414
		" ------------------------------------------\n");
415
	if (!ret)
416
		return TRACE_TYPE_PARTIAL_LINE;
417 418 419

	ret = print_graph_cpu(s, cpu);
	if (ret == TRACE_TYPE_PARTIAL_LINE)
420
		return TRACE_TYPE_PARTIAL_LINE;
421 422 423

	ret = print_graph_proc(s, prev_pid);
	if (ret == TRACE_TYPE_PARTIAL_LINE)
424
		return TRACE_TYPE_PARTIAL_LINE;
425 426 427

	ret = trace_seq_printf(s, " => ");
	if (!ret)
428
		return TRACE_TYPE_PARTIAL_LINE;
429 430 431

	ret = print_graph_proc(s, pid);
	if (ret == TRACE_TYPE_PARTIAL_LINE)
432
		return TRACE_TYPE_PARTIAL_LINE;
433 434 435 436

	ret = trace_seq_printf(s,
		"\n ------------------------------------------\n\n");
	if (!ret)
437
		return TRACE_TYPE_PARTIAL_LINE;
438

439
	return TRACE_TYPE_HANDLED;
440 441
}

442 443
static struct ftrace_graph_ret_entry *
get_return_for_leaf(struct trace_iterator *iter,
444 445
		struct ftrace_graph_ent_entry *curr)
{
446 447
	struct fgraph_data *data = iter->private;
	struct ring_buffer_iter *ring_iter = NULL;
448 449 450
	struct ring_buffer_event *event;
	struct ftrace_graph_ret_entry *next;

451 452 453 454 455 456 457 458
	/*
	 * 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 {
459

460 461 462 463 464 465 466 467 468 469 470 471 472 473
		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.
			 */
			ring_buffer_consume(iter->tr->buffer, iter->cpu, NULL);
			event = ring_buffer_peek(iter->tr->buffer, iter->cpu,
						 NULL);
		}
474

475 476 477 478
		if (!event)
			return NULL;

		next = ring_buffer_event_data(event);
479

480 481 482 483 484 485 486 487 488
		if (data) {
			/*
			 * Save current and next entries for later reference
			 * if the output fails.
			 */
			data->ent = *curr;
			data->ret = *next;
		}
	}
489 490

	if (next->ent.type != TRACE_GRAPH_RET)
491
		return NULL;
492 493 494

	if (curr->ent.pid != next->ent.pid ||
			curr->graph_ent.func != next->ret.func)
495
		return NULL;
496

497 498 499 500 501
	/* this is a leaf, now advance the iterator */
	if (ring_iter)
		ring_buffer_read(ring_iter, NULL);

	return next;
502 503
}

504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528
/* 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, "  ");
}

529 530 531 532 533 534 535 536 537 538 539
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);
}

540
static enum print_line_t
541
print_graph_irq(struct trace_iterator *iter, unsigned long addr,
542
		enum trace_type type, int cpu, pid_t pid)
543 544
{
	int ret;
545
	struct trace_seq *s = &iter->seq;
546 547 548 549 550

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

551 552 553 554 555 556 557
	/* 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;
	}

558 559 560 561 562 563
	/* 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;
	}
564

565 566 567 568 569 570 571 572 573
	/* 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;
	}
574

575 576 577 578
	/* No overhead */
	ret = print_graph_overhead(-1, s);
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;
579

580 581 582 583 584 585 586 587 588 589 590 591
	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");
592 593 594 595 596

	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;
	return TRACE_TYPE_HANDLED;
}
597

598 599
enum print_line_t
trace_print_graph_duration(unsigned long long duration, struct trace_seq *s)
600 601
{
	unsigned long nsecs_rem = do_div(duration, 1000);
602 603 604 605 606 607 608 609 610
	/* 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 */
611
	ret = trace_seq_printf(s, "%s", msecs_str);
612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635
	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;
	}
636 637 638 639 640 641 642 643 644 645 646
	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;
647 648 649 650 651

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

652
	return TRACE_TYPE_HANDLED;
653 654 655
}

/* Case of a leaf function on its call entry */
656
static enum print_line_t
657
print_graph_entry_leaf(struct trace_iterator *iter,
658 659
		struct ftrace_graph_ent_entry *entry,
		struct ftrace_graph_ret_entry *ret_entry, struct trace_seq *s)
660
{
661
	struct fgraph_data *data = iter->private;
662 663 664
	struct ftrace_graph_ret *graph_ret;
	struct ftrace_graph_ent *call;
	unsigned long long duration;
665
	int ret;
666
	int i;
667

668 669 670 671
	graph_ret = &ret_entry->ret;
	call = &entry->graph_ent;
	duration = graph_ret->rettime - graph_ret->calltime;

672 673
	if (data) {
		int cpu = iter->cpu;
674
		int *depth = &(per_cpu_ptr(data->cpu_data, cpu)->depth);
675 676 677 678 679 680 681 682 683

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

684
	/* Overhead */
685 686 687
	ret = print_graph_overhead(duration, s);
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;
688 689

	/* Duration */
690 691 692 693 694
	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;
	}
695

696 697 698 699 700 701 702
	/* Function */
	for (i = 0; i < call->depth * TRACE_GRAPH_INDENT; i++) {
		ret = trace_seq_printf(s, " ");
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}

703
	ret = trace_seq_printf(s, "%ps();\n", (void *)call->func);
704 705 706 707 708 709 710
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

	return TRACE_TYPE_HANDLED;
}

static enum print_line_t
711 712 713
print_graph_entry_nested(struct trace_iterator *iter,
			 struct ftrace_graph_ent_entry *entry,
			 struct trace_seq *s, int cpu)
714 715
{
	struct ftrace_graph_ent *call = &entry->graph_ent;
716 717 718 719 720 721
	struct fgraph_data *data = iter->private;
	int ret;
	int i;

	if (data) {
		int cpu = iter->cpu;
722
		int *depth = &(per_cpu_ptr(data->cpu_data, cpu)->depth);
723 724 725

		*depth = call->depth;
	}
726 727

	/* No overhead */
728 729 730
	ret = print_graph_overhead(-1, s);
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;
731

732 733
	/* No time */
	if (tracer_flags.val & TRACE_GRAPH_PRINT_DURATION) {
734 735 736 737 738
		ret = trace_seq_printf(s, "            |  ");
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}

739
	/* Function */
740 741
	for (i = 0; i < call->depth * TRACE_GRAPH_INDENT; i++) {
		ret = trace_seq_printf(s, " ");
742 743
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
744 745
	}

746
	ret = trace_seq_printf(s, "%ps() {\n", (void *)call->func);
747 748 749
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

750 751 752 753 754
	/*
	 * we already consumed the current entry to check the next one
	 * and see if this is a leaf.
	 */
	return TRACE_TYPE_NO_CONSUME;
755 756
}

757
static enum print_line_t
758 759
print_graph_prologue(struct trace_iterator *iter, struct trace_seq *s,
		     int type, unsigned long addr)
760
{
761
	struct fgraph_data *data = iter->private;
762
	struct trace_entry *ent = iter->ent;
763 764
	int cpu = iter->cpu;
	int ret;
765

766
	/* Pid */
767
	if (verif_pid(s, ent->pid, cpu, data) == TRACE_TYPE_PARTIAL_LINE)
768 769
		return TRACE_TYPE_PARTIAL_LINE;

770 771 772 773 774 775
	if (type) {
		/* Interrupt */
		ret = print_graph_irq(iter, addr, type, cpu, ent->pid);
		if (ret == TRACE_TYPE_PARTIAL_LINE)
			return TRACE_TYPE_PARTIAL_LINE;
	}
776

777 778 779 780 781 782 783
	/* 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;
	}

784 785 786
	/* Cpu */
	if (tracer_flags.val & TRACE_GRAPH_PRINT_CPU) {
		ret = print_graph_cpu(s, cpu);
787 788 789 790 791 792
		if (ret == TRACE_TYPE_PARTIAL_LINE)
			return TRACE_TYPE_PARTIAL_LINE;
	}

	/* Proc */
	if (tracer_flags.val & TRACE_GRAPH_PRINT_PROC) {
793
		ret = print_graph_proc(s, ent->pid);
794 795 796 797
		if (ret == TRACE_TYPE_PARTIAL_LINE)
			return TRACE_TYPE_PARTIAL_LINE;

		ret = trace_seq_printf(s, " | ");
798 799 800
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}
801

802 803 804 805 806 807 808
	/* 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;
	}

809 810 811 812 813 814 815
	return 0;
}

static enum print_line_t
print_graph_entry(struct ftrace_graph_ent_entry *field, struct trace_seq *s,
			struct trace_iterator *iter)
{
816
	struct fgraph_data *data = iter->private;
817 818
	struct ftrace_graph_ent *call = &field->graph_ent;
	struct ftrace_graph_ret_entry *leaf_ret;
819 820
	static enum print_line_t ret;
	int cpu = iter->cpu;
821 822 823 824

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

825 826
	leaf_ret = get_return_for_leaf(iter, field);
	if (leaf_ret)
827
		ret = print_graph_entry_leaf(iter, field, leaf_ret, s);
828
	else
829
		ret = print_graph_entry_nested(iter, field, s, cpu);
830

831 832 833 834 835 836 837 838 839 840 841 842 843
	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;
844 845
}

846 847
static enum print_line_t
print_graph_return(struct ftrace_graph_ret *trace, struct trace_seq *s,
848
		   struct trace_entry *ent, struct trace_iterator *iter)
849
{
850
	unsigned long long duration = trace->rettime - trace->calltime;
851 852 853 854 855 856 857
	struct fgraph_data *data = iter->private;
	pid_t pid = ent->pid;
	int cpu = iter->cpu;
	int ret;
	int i;

	if (data) {
858
		int *depth = &(per_cpu_ptr(data->cpu_data, cpu)->depth);
859 860 861 862 863 864 865 866

		/*
		 * 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;
	}
867

868
	if (print_graph_prologue(iter, s, 0, 0))
869 870
		return TRACE_TYPE_PARTIAL_LINE;

871
	/* Overhead */
872 873 874
	ret = print_graph_overhead(duration, s);
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;
875 876

	/* Duration */
877 878 879 880 881
	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;
	}
882 883

	/* Closing brace */
884 885
	for (i = 0; i < trace->depth * TRACE_GRAPH_INDENT; i++) {
		ret = trace_seq_printf(s, " ");
886 887
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
888 889
	}

890
	ret = trace_seq_printf(s, "}\n");
891 892
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;
893

894
	/* Overrun */
895 896 897
	if (tracer_flags.val & TRACE_GRAPH_PRINT_OVERRUN) {
		ret = trace_seq_printf(s, " (Overruns: %lu)\n",
					trace->overrun);
898 899
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
900
	}
901

902
	ret = print_graph_irq(iter, trace->func, TRACE_GRAPH_RET, cpu, pid);
903 904 905
	if (ret == TRACE_TYPE_PARTIAL_LINE)
		return TRACE_TYPE_PARTIAL_LINE;

906 907 908
	return TRACE_TYPE_HANDLED;
}

909
static enum print_line_t
910 911
print_graph_comment(struct trace_seq *s,  struct trace_entry *ent,
		    struct trace_iterator *iter)
912
{
913
	unsigned long sym_flags = (trace_flags & TRACE_ITER_SYM_MASK);
914
	struct fgraph_data *data = iter->private;
915
	struct trace_event *event;
916
	int depth = 0;
917
	int ret;
918 919 920
	int i;

	if (data)
921
		depth = per_cpu_ptr(data->cpu_data, iter->cpu)->depth;
922

923
	if (print_graph_prologue(iter, s, 0, 0))
924 925
		return TRACE_TYPE_PARTIAL_LINE;

926
	/* No overhead */
927 928 929 930 931 932 933
	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, "            |  ");
934 935 936 937 938
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}

	/* Indentation */
939 940
	if (depth > 0)
		for (i = 0; i < (depth + 1) * TRACE_GRAPH_INDENT; i++) {
941 942 943 944 945 946
			ret = trace_seq_printf(s, " ");
			if (!ret)
				return TRACE_TYPE_PARTIAL_LINE;
		}

	/* The comment */
947 948 949 950
	ret = trace_seq_printf(s, "/* ");
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970
	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;
	}
971

972 973 974 975 976 977
	/* Strip ending newline */
	if (s->buffer[s->len - 1] == '\n') {
		s->buffer[s->len - 1] = '\0';
		s->len--;
	}

978 979 980 981 982 983 984 985
	ret = trace_seq_printf(s, " */\n");
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

	return TRACE_TYPE_HANDLED;
}


986 987 988
enum print_line_t
print_graph_function(struct trace_iterator *iter)
{
989 990
	struct ftrace_graph_ent_entry *field;
	struct fgraph_data *data = iter->private;
991
	struct trace_entry *entry = iter->ent;
992
	struct trace_seq *s = &iter->seq;
993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015
	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;
		ret = print_graph_entry(field, s, iter);
		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;
	}
1016

1017 1018
	switch (entry->type) {
	case TRACE_GRAPH_ENT: {
1019 1020 1021 1022 1023 1024
		/*
		 * 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.
		 */
1025
		struct ftrace_graph_ent_entry saved;
1026
		trace_assign_type(field, entry);
1027 1028
		saved = *field;
		return print_graph_entry(&saved, s, iter);
1029 1030 1031 1032
	}
	case TRACE_GRAPH_RET: {
		struct ftrace_graph_ret_entry *field;
		trace_assign_type(field, entry);
1033
		return print_graph_return(&field->ret, s, entry, iter);
1034 1035
	}
	default:
1036
		return print_graph_comment(s, entry, iter);
1037
	}
1038 1039

	return TRACE_TYPE_HANDLED;
1040 1041
}

1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059
static void print_lat_header(struct seq_file *s)
{
	static const char spaces[] = "                "	/* 16 spaces */
		"    "					/* 4 spaces */
		"                 ";			/* 17 spaces */
	int size = 0;

	if (tracer_flags.val & TRACE_GRAPH_PRINT_ABS_TIME)
		size += 16;
	if (tracer_flags.val & TRACE_GRAPH_PRINT_CPU)
		size += 4;
	if (tracer_flags.val & TRACE_GRAPH_PRINT_PROC)
		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);
1060 1061
	seq_printf(s, "#%.*s||| / _-=> lock-depth      \n", size, spaces);
	seq_printf(s, "#%.*s|||| /                     \n", size, spaces);
1062 1063
}

1064 1065
static void print_graph_headers(struct seq_file *s)
{
1066 1067 1068 1069 1070
	int lat = trace_flags & TRACE_ITER_LATENCY_FMT;

	if (lat)
		print_lat_header(s);

1071
	/* 1st line */
1072
	seq_printf(s, "#");
1073 1074
	if (tracer_flags.val & TRACE_GRAPH_PRINT_ABS_TIME)
		seq_printf(s, "     TIME       ");
1075
	if (tracer_flags.val & TRACE_GRAPH_PRINT_CPU)
1076
		seq_printf(s, " CPU");
1077
	if (tracer_flags.val & TRACE_GRAPH_PRINT_PROC)
1078 1079
		seq_printf(s, "  TASK/PID       ");
	if (lat)
1080
		seq_printf(s, "|||||");
1081 1082 1083
	if (tracer_flags.val & TRACE_GRAPH_PRINT_DURATION)
		seq_printf(s, "  DURATION   ");
	seq_printf(s, "               FUNCTION CALLS\n");
1084 1085

	/* 2nd line */
1086
	seq_printf(s, "#");
1087 1088
	if (tracer_flags.val & TRACE_GRAPH_PRINT_ABS_TIME)
		seq_printf(s, "      |         ");
1089
	if (tracer_flags.val & TRACE_GRAPH_PRINT_CPU)
1090
		seq_printf(s, " |  ");
1091
	if (tracer_flags.val & TRACE_GRAPH_PRINT_PROC)
1092 1093
		seq_printf(s, "   |    |        ");
	if (lat)
1094
		seq_printf(s, "|||||");
1095 1096 1097
	if (tracer_flags.val & TRACE_GRAPH_PRINT_DURATION)
		seq_printf(s, "   |   |      ");
	seq_printf(s, "               |   |   |   |\n");
1098
}
1099 1100 1101

static void graph_trace_open(struct trace_iterator *iter)
{
1102
	/* pid and depth on the last trace processed */
1103
	struct fgraph_data *data;
1104 1105
	int cpu;

1106 1107 1108
	iter->private = NULL;

	data = kzalloc(sizeof(*data), GFP_KERNEL);
1109
	if (!data)
1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123
		goto out_err;

	data->cpu_data = alloc_percpu(struct fgraph_cpu_data);
	if (!data->cpu_data)
		goto out_err_free;

	for_each_possible_cpu(cpu) {
		pid_t *pid = &(per_cpu_ptr(data->cpu_data, cpu)->last_pid);
		int *depth = &(per_cpu_ptr(data->cpu_data, cpu)->depth);
		int *ignore = &(per_cpu_ptr(data->cpu_data, cpu)->ignore);
		*pid = -1;
		*depth = 0;
		*ignore = 0;
	}
1124

1125
	iter->private = data;
1126 1127 1128 1129 1130 1131 1132

	return;

 out_err_free:
	kfree(data);
 out_err:
	pr_warning("function graph tracer: not enough memory\n");
1133 1134 1135 1136
}

static void graph_trace_close(struct trace_iterator *iter)
{
1137 1138 1139 1140 1141 1142
	struct fgraph_data *data = iter->private;

	if (data) {
		free_percpu(data->cpu_data);
		kfree(data);
	}
1143 1144
}

1145
static struct tracer graph_trace __read_mostly = {
1146
	.name		= "function_graph",
1147
	.open		= graph_trace_open,
1148
	.pipe_open	= graph_trace_open,
1149
	.close		= graph_trace_close,
1150
	.pipe_close	= graph_trace_close,
1151
	.wait_pipe	= poll_wait_pipe,
1152 1153
	.init		= graph_trace_init,
	.reset		= graph_trace_reset,
1154 1155
	.print_line	= print_graph_function,
	.print_header	= print_graph_headers,
1156
	.flags		= &tracer_flags,
1157 1158 1159
#ifdef CONFIG_FTRACE_SELFTEST
	.selftest	= trace_selftest_startup_function_graph,
#endif
1160 1161 1162 1163
};

static __init int init_graph_trace(void)
{
1164 1165
	max_bytes_for_cpu = snprintf(NULL, 0, "%d", nr_cpu_ids - 1);

1166 1167 1168 1169
	return register_tracer(&graph_trace);
}

device_initcall(init_graph_trace);