trace_functions_graph.c 33.4 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 19 20
/* When set, irq functions will be ignored */
static int ftrace_graph_skip_irqs;

21
struct fgraph_cpu_data {
22 23
	pid_t		last_pid;
	int		depth;
24
	int		depth_irq;
25
	int		ignore;
26
	unsigned long	enter_funcs[FTRACE_RETFUNC_DEPTH];
27 28 29
};

struct fgraph_data {
30
	struct fgraph_cpu_data __percpu *cpu_data;
31 32 33 34 35 36

	/* Place to preserve last processed entry. */
	struct ftrace_graph_ent_entry	ent;
	struct ftrace_graph_ret_entry	ret;
	int				failed;
	int				cpu;
37 38
};

39
#define TRACE_GRAPH_INDENT	2
40

41
/* Flag options */
42
#define TRACE_GRAPH_PRINT_OVERRUN	0x1
43 44
#define TRACE_GRAPH_PRINT_CPU		0x2
#define TRACE_GRAPH_PRINT_OVERHEAD	0x4
45
#define TRACE_GRAPH_PRINT_PROC		0x8
46
#define TRACE_GRAPH_PRINT_DURATION	0x10
47
#define TRACE_GRAPH_PRINT_ABS_TIME	0x20
48
#define TRACE_GRAPH_PRINT_IRQS		0x40
49

50
static struct tracer_opt trace_opts[] = {
51
	/* Display overruns? (for self-debug purpose) */
52 53 54 55 56
	{ 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) },
57 58
	/* Display proc name/pid */
	{ TRACER_OPT(funcgraph-proc, TRACE_GRAPH_PRINT_PROC) },
59 60 61 62
	/* 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) },
63 64
	/* Display interrupts */
	{ TRACER_OPT(funcgraph-irqs, TRACE_GRAPH_PRINT_IRQS) },
65 66 67 68
	{ } /* Empty entry */
};

static struct tracer_flags tracer_flags = {
69
	/* Don't display overruns and proc by default */
70
	.val = TRACE_GRAPH_PRINT_CPU | TRACE_GRAPH_PRINT_OVERHEAD |
71
	       TRACE_GRAPH_PRINT_DURATION | TRACE_GRAPH_PRINT_IRQS,
72 73 74
	.opts = trace_opts
};

75
static struct trace_array *graph_array;
76

77

78 79
/* Add a function return address to the trace stack on thread info.*/
int
80 81
ftrace_push_return_trace(unsigned long ret, unsigned long func, int *depth,
			 unsigned long frame_pointer)
82
{
83
	unsigned long long calltime;
84 85 86 87 88
	int index;

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

89 90 91 92 93 94
	/*
	 * We must make sure the ret_stack is tested before we read
	 * anything else.
	 */
	smp_rmb();

95 96 97 98 99 100
	/* The return trace stack is full */
	if (current->curr_ret_stack == FTRACE_RETFUNC_DEPTH - 1) {
		atomic_inc(&current->trace_overrun);
		return -EBUSY;
	}

101 102
	calltime = trace_clock_local();

103 104 105 106
	index = ++current->curr_ret_stack;
	barrier();
	current->ret_stack[index].ret = ret;
	current->ret_stack[index].func = func;
107
	current->ret_stack[index].calltime = calltime;
108
	current->ret_stack[index].subtime = 0;
109
	current->ret_stack[index].fp = frame_pointer;
110 111 112 113 114 115
	*depth = index;

	return 0;
}

/* Retrieve a function return address to the trace stack on thread info.*/
116
static void
117 118
ftrace_pop_return_trace(struct ftrace_graph_ret *trace, unsigned long *ret,
			unsigned long frame_pointer)
119 120 121 122 123 124 125 126 127 128 129 130 131
{
	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;
	}

132 133 134 135 136 137 138 139 140 141 142 143 144 145 146
#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"
147
		     "  from func %ps return to %lx\n",
148 149 150 151 152 153 154 155 156
		     current->ret_stack[index].fp,
		     frame_pointer,
		     (void *)current->ret_stack[index].func,
		     current->ret_stack[index].ret);
		*ret = (unsigned long)panic;
		return;
	}
#endif

157 158 159 160 161 162 163 164 165 166 167
	*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.
 */
168
unsigned long ftrace_return_to_handler(unsigned long frame_pointer)
169 170 171 172
{
	struct ftrace_graph_ret trace;
	unsigned long ret;

173
	ftrace_pop_return_trace(&trace, &ret, frame_pointer);
174
	trace.rettime = trace_clock_local();
175
	ftrace_graph_return(&trace);
176 177
	barrier();
	current->curr_ret_stack--;
178 179 180 181 182 183 184 185 186 187 188

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

	return ret;
}

189
int __trace_graph_entry(struct trace_array *tr,
190 191 192 193 194 195
				struct ftrace_graph_ent *trace,
				unsigned long flags,
				int pc)
{
	struct ftrace_event_call *call = &event_funcgraph_entry;
	struct ring_buffer_event *event;
196
	struct ring_buffer *buffer = tr->buffer;
197 198
	struct ftrace_graph_ent_entry *entry;

199
	if (unlikely(__this_cpu_read(ftrace_cpu_disabled)))
200 201
		return 0;

202
	event = trace_buffer_lock_reserve(buffer, TRACE_GRAPH_ENT,
203 204 205 206 207
					  sizeof(*entry), flags, pc);
	if (!event)
		return 0;
	entry	= ring_buffer_event_data(event);
	entry->graph_ent			= *trace;
208 209
	if (!filter_current_check_discard(buffer, call, entry, event))
		ring_buffer_unlock_commit(buffer, event);
210 211 212 213

	return 1;
}

214 215 216 217 218 219 220 221
static inline int ftrace_graph_ignore_irqs(void)
{
	if (!ftrace_graph_skip_irqs)
		return 0;

	return in_irq();
}

222 223 224 225 226 227 228 229 230 231 232 233 234
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;

235
	/* trace it when it is-nested-in or is a function enabled. */
236 237
	if (!(trace->depth || ftrace_graph_addr(trace->func)) ||
	      ftrace_graph_ignore_irqs())
238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256
		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;
}

257 258 259 260 261 262 263 264
int trace_graph_thresh_entry(struct ftrace_graph_ent *trace)
{
	if (tracing_thresh)
		return 1;
	else
		return trace_graph_entry(trace);
}

265
void __trace_graph_return(struct trace_array *tr,
266 267 268 269 270 271
				struct ftrace_graph_ret *trace,
				unsigned long flags,
				int pc)
{
	struct ftrace_event_call *call = &event_funcgraph_exit;
	struct ring_buffer_event *event;
272
	struct ring_buffer *buffer = tr->buffer;
273 274
	struct ftrace_graph_ret_entry *entry;

275
	if (unlikely(__this_cpu_read(ftrace_cpu_disabled)))
276 277
		return;

278
	event = trace_buffer_lock_reserve(buffer, TRACE_GRAPH_RET,
279 280 281 282 283
					  sizeof(*entry), flags, pc);
	if (!event)
		return;
	entry	= ring_buffer_event_data(event);
	entry->ret				= *trace;
284 285
	if (!filter_current_check_discard(buffer, call, entry, event))
		ring_buffer_unlock_commit(buffer, event);
286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308
}

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);
}

309 310 311 312 313 314 315 316 317
void set_graph_array(struct trace_array *tr)
{
	graph_array = tr;

	/* Make graph_array visible before we start tracing */

	smp_mb();
}

318 319 320 321 322 323 324 325 326
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);
}

327 328
static int graph_trace_init(struct trace_array *tr)
{
329 330
	int ret;

331
	set_graph_array(tr);
332 333 334 335 336 337
	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);
338 339 340 341 342
	if (ret)
		return ret;
	tracing_start_cmdline_record();

	return 0;
343 344 345 346
}

static void graph_trace_reset(struct trace_array *tr)
{
347 348
	tracing_stop_cmdline_record();
	unregister_ftrace_graph();
349 350
}

351
static int max_bytes_for_cpu;
352 353 354 355 356 357

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

358 359 360 361 362
	/*
	 * Start with a space character - to make it stand out
	 * to the right a bit when trace output is pasted into
	 * email:
	 */
363
	ret = trace_seq_printf(s, " %*d) ", max_bytes_for_cpu, cpu);
364
	if (!ret)
365 366
		return TRACE_TYPE_PARTIAL_LINE;

367 368 369
	return TRACE_TYPE_HANDLED;
}

370 371 372 373 374
#define TRACE_GRAPH_PROCINFO_LENGTH	14

static enum print_line_t
print_graph_proc(struct trace_seq *s, pid_t pid)
{
375
	char comm[TASK_COMM_LEN];
376 377
	/* sign + log10(MAX_INT) + '\0' */
	char pid_str[11];
378 379 380 381
	int spaces = 0;
	int ret;
	int len;
	int i;
382

383
	trace_find_cmdline(pid, comm);
384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412
	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;
}

413

414 415 416
static enum print_line_t
print_graph_lat_fmt(struct trace_seq *s, struct trace_entry *entry)
{
417
	if (!trace_seq_putc(s, ' '))
418 419
		return 0;

420
	return trace_print_lat_fmt(s, entry);
421 422
}

423
/* If the pid changed since the last trace, output this event */
424
static enum print_line_t
425
verif_pid(struct trace_seq *s, pid_t pid, int cpu, struct fgraph_data *data)
426
{
427
	pid_t prev_pid;
428
	pid_t *last_pid;
429
	int ret;
430

431
	if (!data)
432 433
		return TRACE_TYPE_HANDLED;

434
	last_pid = &(per_cpu_ptr(data->cpu_data, cpu)->last_pid);
435 436

	if (*last_pid == pid)
437
		return TRACE_TYPE_HANDLED;
438

439 440
	prev_pid = *last_pid;
	*last_pid = pid;
441

442 443
	if (prev_pid == -1)
		return TRACE_TYPE_HANDLED;
444 445 446 447 448 449 450 451 452
/*
 * Context-switch trace line:

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

 */
	ret = trace_seq_printf(s,
453
		" ------------------------------------------\n");
454
	if (!ret)
455
		return TRACE_TYPE_PARTIAL_LINE;
456 457 458

	ret = print_graph_cpu(s, cpu);
	if (ret == TRACE_TYPE_PARTIAL_LINE)
459
		return TRACE_TYPE_PARTIAL_LINE;
460 461 462

	ret = print_graph_proc(s, prev_pid);
	if (ret == TRACE_TYPE_PARTIAL_LINE)
463
		return TRACE_TYPE_PARTIAL_LINE;
464 465 466

	ret = trace_seq_printf(s, " => ");
	if (!ret)
467
		return TRACE_TYPE_PARTIAL_LINE;
468 469 470

	ret = print_graph_proc(s, pid);
	if (ret == TRACE_TYPE_PARTIAL_LINE)
471
		return TRACE_TYPE_PARTIAL_LINE;
472 473 474 475

	ret = trace_seq_printf(s,
		"\n ------------------------------------------\n\n");
	if (!ret)
476
		return TRACE_TYPE_PARTIAL_LINE;
477

478
	return TRACE_TYPE_HANDLED;
479 480
}

481 482
static struct ftrace_graph_ret_entry *
get_return_for_leaf(struct trace_iterator *iter,
483 484
		struct ftrace_graph_ent_entry *curr)
{
485 486
	struct fgraph_data *data = iter->private;
	struct ring_buffer_iter *ring_iter = NULL;
487 488 489
	struct ring_buffer_event *event;
	struct ftrace_graph_ret_entry *next;

490 491 492 493 494 495 496 497
	/*
	 * 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 {
498

499 500 501 502 503 504 505 506 507 508
		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.
			 */
509 510
			ring_buffer_consume(iter->tr->buffer, iter->cpu,
					    NULL, NULL);
511
			event = ring_buffer_peek(iter->tr->buffer, iter->cpu,
512
						 NULL, NULL);
513
		}
514

515 516 517 518
		if (!event)
			return NULL;

		next = ring_buffer_event_data(event);
519

520 521 522 523 524 525
		if (data) {
			/*
			 * Save current and next entries for later reference
			 * if the output fails.
			 */
			data->ent = *curr;
526 527 528 529 530 531 532 533 534
			/*
			 * If the next event is not a return type, then
			 * we only care about what type it is. Otherwise we can
			 * safely copy the entire event.
			 */
			if (next->ent.type == TRACE_GRAPH_RET)
				data->ret = *next;
			else
				data->ret.ent.type = next->ent.type;
535 536
		}
	}
537 538

	if (next->ent.type != TRACE_GRAPH_RET)
539
		return NULL;
540 541 542

	if (curr->ent.pid != next->ent.pid ||
			curr->graph_ent.func != next->ret.func)
543
		return NULL;
544

545 546 547 548 549
	/* this is a leaf, now advance the iterator */
	if (ring_iter)
		ring_buffer_read(ring_iter, NULL);

	return next;
550 551
}

552 553
/* Signal a overhead of time execution to the output */
static int
554 555
print_graph_overhead(unsigned long long duration, struct trace_seq *s,
		     u32 flags)
556 557
{
	/* If duration disappear, we don't need anything */
558
	if (!(flags & TRACE_GRAPH_PRINT_DURATION))
559 560 561 562 563 564
		return 1;

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

565
	if (flags & TRACE_GRAPH_PRINT_OVERHEAD) {
566 567 568 569 570 571 572 573 574 575 576 577
		/* 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, "  ");
}

578 579 580 581 582 583 584 585 586 587 588
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);
}

589
static enum print_line_t
590
print_graph_irq(struct trace_iterator *iter, unsigned long addr,
591
		enum trace_type type, int cpu, pid_t pid, u32 flags)
592 593
{
	int ret;
594
	struct trace_seq *s = &iter->seq;
595 596 597 598 599

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

600
	/* Absolute time */
601
	if (flags & TRACE_GRAPH_PRINT_ABS_TIME) {
602 603 604 605 606
		ret = print_graph_abs_time(iter->ts, s);
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}

607
	/* Cpu */
608
	if (flags & TRACE_GRAPH_PRINT_CPU) {
609 610 611 612
		ret = print_graph_cpu(s, cpu);
		if (ret == TRACE_TYPE_PARTIAL_LINE)
			return TRACE_TYPE_PARTIAL_LINE;
	}
613

614
	/* Proc */
615
	if (flags & TRACE_GRAPH_PRINT_PROC) {
616 617 618 619 620 621 622
		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;
	}
623

624
	/* No overhead */
625
	ret = print_graph_overhead(-1, s, flags);
626 627
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;
628

629 630 631 632 633 634 635 636 637
	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 */
638
	if (flags & TRACE_GRAPH_PRINT_DURATION)
639 640
		trace_seq_printf(s, " |");
	ret = trace_seq_printf(s, "\n");
641 642 643 644 645

	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;
	return TRACE_TYPE_HANDLED;
}
646

647 648
enum print_line_t
trace_print_graph_duration(unsigned long long duration, struct trace_seq *s)
649 650
{
	unsigned long nsecs_rem = do_div(duration, 1000);
651 652 653 654 655 656 657 658 659
	/* 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 */
660
	ret = trace_seq_printf(s, "%s", msecs_str);
661 662 663 664 665 666 667
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

	len = strlen(msecs_str);

	/* Print nsecs (we don't want to exceed 7 numbers) */
	if (len < 7) {
668 669 670
		size_t slen = min_t(size_t, sizeof(nsecs_str), 8UL - len);

		snprintf(nsecs_str, slen, "%03lu", nsecs_rem);
671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686
		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;
	}
687 688 689 690 691 692 693 694 695 696 697
	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;
698 699 700 701 702

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

703
	return TRACE_TYPE_HANDLED;
704 705 706
}

/* Case of a leaf function on its call entry */
707
static enum print_line_t
708
print_graph_entry_leaf(struct trace_iterator *iter,
709
		struct ftrace_graph_ent_entry *entry,
710 711
		struct ftrace_graph_ret_entry *ret_entry,
		struct trace_seq *s, u32 flags)
712
{
713
	struct fgraph_data *data = iter->private;
714 715 716
	struct ftrace_graph_ret *graph_ret;
	struct ftrace_graph_ent *call;
	unsigned long long duration;
717
	int ret;
718
	int i;
719

720 721 722 723
	graph_ret = &ret_entry->ret;
	call = &entry->graph_ent;
	duration = graph_ret->rettime - graph_ret->calltime;

724
	if (data) {
725
		struct fgraph_cpu_data *cpu_data;
726
		int cpu = iter->cpu;
727 728

		cpu_data = per_cpu_ptr(data->cpu_data, cpu);
729 730 731 732 733 734

		/*
		 * Comments display at + 1 to depth. Since
		 * this is a leaf function, keep the comments
		 * equal to this depth.
		 */
735 736 737 738 739
		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;
740 741
	}

742
	/* Overhead */
743
	ret = print_graph_overhead(duration, s, flags);
744 745
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;
746 747

	/* Duration */
748
	if (flags & TRACE_GRAPH_PRINT_DURATION) {
749 750 751 752
		ret = print_graph_duration(duration, s);
		if (ret == TRACE_TYPE_PARTIAL_LINE)
			return TRACE_TYPE_PARTIAL_LINE;
	}
753

754 755 756 757 758 759 760
	/* Function */
	for (i = 0; i < call->depth * TRACE_GRAPH_INDENT; i++) {
		ret = trace_seq_printf(s, " ");
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}

761
	ret = trace_seq_printf(s, "%ps();\n", (void *)call->func);
762 763 764 765 766 767 768
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

	return TRACE_TYPE_HANDLED;
}

static enum print_line_t
769 770
print_graph_entry_nested(struct trace_iterator *iter,
			 struct ftrace_graph_ent_entry *entry,
771
			 struct trace_seq *s, int cpu, u32 flags)
772 773
{
	struct ftrace_graph_ent *call = &entry->graph_ent;
774 775 776 777 778
	struct fgraph_data *data = iter->private;
	int ret;
	int i;

	if (data) {
779
		struct fgraph_cpu_data *cpu_data;
780 781
		int cpu = iter->cpu;

782 783 784 785 786 787
		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;
788
	}
789 790

	/* No overhead */
791
	ret = print_graph_overhead(-1, s, flags);
792 793
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;
794

795
	/* No time */
796
	if (flags & TRACE_GRAPH_PRINT_DURATION) {
797 798 799 800 801
		ret = trace_seq_printf(s, "            |  ");
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}

802
	/* Function */
803 804
	for (i = 0; i < call->depth * TRACE_GRAPH_INDENT; i++) {
		ret = trace_seq_printf(s, " ");
805 806
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
807 808
	}

809
	ret = trace_seq_printf(s, "%ps() {\n", (void *)call->func);
810 811 812
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

813 814 815 816 817
	/*
	 * we already consumed the current entry to check the next one
	 * and see if this is a leaf.
	 */
	return TRACE_TYPE_NO_CONSUME;
818 819
}

820
static enum print_line_t
821
print_graph_prologue(struct trace_iterator *iter, struct trace_seq *s,
822
		     int type, unsigned long addr, u32 flags)
823
{
824
	struct fgraph_data *data = iter->private;
825
	struct trace_entry *ent = iter->ent;
826 827
	int cpu = iter->cpu;
	int ret;
828

829
	/* Pid */
830
	if (verif_pid(s, ent->pid, cpu, data) == TRACE_TYPE_PARTIAL_LINE)
831 832
		return TRACE_TYPE_PARTIAL_LINE;

833 834
	if (type) {
		/* Interrupt */
835
		ret = print_graph_irq(iter, addr, type, cpu, ent->pid, flags);
836 837 838
		if (ret == TRACE_TYPE_PARTIAL_LINE)
			return TRACE_TYPE_PARTIAL_LINE;
	}
839

840
	/* Absolute time */
841
	if (flags & TRACE_GRAPH_PRINT_ABS_TIME) {
842 843 844 845 846
		ret = print_graph_abs_time(iter->ts, s);
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}

847
	/* Cpu */
848
	if (flags & TRACE_GRAPH_PRINT_CPU) {
849
		ret = print_graph_cpu(s, cpu);
850 851 852 853 854
		if (ret == TRACE_TYPE_PARTIAL_LINE)
			return TRACE_TYPE_PARTIAL_LINE;
	}

	/* Proc */
855
	if (flags & TRACE_GRAPH_PRINT_PROC) {
856
		ret = print_graph_proc(s, ent->pid);
857 858 859 860
		if (ret == TRACE_TYPE_PARTIAL_LINE)
			return TRACE_TYPE_PARTIAL_LINE;

		ret = trace_seq_printf(s, " | ");
861 862 863
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}
864

865 866 867 868 869 870 871
	/* 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;
	}

872 873 874
	return 0;
}

875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890
/*
 * Entry check for irq code
 *
 * returns 1 if
 *  - we are inside irq code
 *  - we just extered irq code
 *
 * retunns 0 if
 *  - funcgraph-interrupts option is set
 *  - we are not inside irq code
 */
static int
check_irq_entry(struct trace_iterator *iter, u32 flags,
		unsigned long addr, int depth)
{
	int cpu = iter->cpu;
891
	int *depth_irq;
892 893
	struct fgraph_data *data = iter->private;

894 895 896 897 898 899 900
	/*
	 * If we are either displaying irqs, or we got called as
	 * a graph event and private data does not exist,
	 * then we bypass the irq check.
	 */
	if ((flags & TRACE_GRAPH_PRINT_IRQS) ||
	    (!data))
901 902
		return 0;

903 904
	depth_irq = &(per_cpu_ptr(data->cpu_data, cpu)->depth_irq);

905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936
	/*
	 * We are inside the irq code
	 */
	if (*depth_irq >= 0)
		return 1;

	if ((addr < (unsigned long)__irqentry_text_start) ||
	    (addr >= (unsigned long)__irqentry_text_end))
		return 0;

	/*
	 * We are entering irq code.
	 */
	*depth_irq = depth;
	return 1;
}

/*
 * Return check for irq code
 *
 * returns 1 if
 *  - we are inside irq code
 *  - we just left irq code
 *
 * returns 0 if
 *  - funcgraph-interrupts option is set
 *  - we are not inside irq code
 */
static int
check_irq_return(struct trace_iterator *iter, u32 flags, int depth)
{
	int cpu = iter->cpu;
937
	int *depth_irq;
938 939
	struct fgraph_data *data = iter->private;

940 941 942 943 944 945 946
	/*
	 * If we are either displaying irqs, or we got called as
	 * a graph event and private data does not exist,
	 * then we bypass the irq check.
	 */
	if ((flags & TRACE_GRAPH_PRINT_IRQS) ||
	    (!data))
947 948
		return 0;

949 950
	depth_irq = &(per_cpu_ptr(data->cpu_data, cpu)->depth_irq);

951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976
	/*
	 * We are not inside the irq code.
	 */
	if (*depth_irq == -1)
		return 0;

	/*
	 * We are inside the irq code, and this is returning entry.
	 * Let's not trace it and clear the entry depth, since
	 * we are out of irq code.
	 *
	 * This condition ensures that we 'leave the irq code' once
	 * we are out of the entry depth. Thus protecting us from
	 * the RETURN entry loss.
	 */
	if (*depth_irq >= depth) {
		*depth_irq = -1;
		return 1;
	}

	/*
	 * We are inside the irq code, and this is not the entry.
	 */
	return 1;
}

977 978
static enum print_line_t
print_graph_entry(struct ftrace_graph_ent_entry *field, struct trace_seq *s,
979
			struct trace_iterator *iter, u32 flags)
980
{
981
	struct fgraph_data *data = iter->private;
982 983
	struct ftrace_graph_ent *call = &field->graph_ent;
	struct ftrace_graph_ret_entry *leaf_ret;
984 985
	static enum print_line_t ret;
	int cpu = iter->cpu;
986

987 988 989
	if (check_irq_entry(iter, flags, call->func, call->depth))
		return TRACE_TYPE_HANDLED;

990
	if (print_graph_prologue(iter, s, TRACE_GRAPH_ENT, call->func, flags))
991 992
		return TRACE_TYPE_PARTIAL_LINE;

993 994
	leaf_ret = get_return_for_leaf(iter, field);
	if (leaf_ret)
995
		ret = print_graph_entry_leaf(iter, field, leaf_ret, s, flags);
996
	else
997
		ret = print_graph_entry_nested(iter, field, s, cpu, flags);
998

999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011
	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;
1012 1013
}

1014 1015
static enum print_line_t
print_graph_return(struct ftrace_graph_ret *trace, struct trace_seq *s,
1016 1017
		   struct trace_entry *ent, struct trace_iterator *iter,
		   u32 flags)
1018
{
1019
	unsigned long long duration = trace->rettime - trace->calltime;
1020 1021 1022
	struct fgraph_data *data = iter->private;
	pid_t pid = ent->pid;
	int cpu = iter->cpu;
1023
	int func_match = 1;
1024 1025 1026
	int ret;
	int i;

1027 1028 1029
	if (check_irq_return(iter, flags, trace->depth))
		return TRACE_TYPE_HANDLED;

1030
	if (data) {
1031 1032 1033 1034
		struct fgraph_cpu_data *cpu_data;
		int cpu = iter->cpu;

		cpu_data = per_cpu_ptr(data->cpu_data, cpu);
1035 1036 1037 1038 1039 1040

		/*
		 * 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.
		 */
1041 1042 1043 1044 1045 1046 1047
		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;
		}
1048
	}
1049

1050
	if (print_graph_prologue(iter, s, 0, 0, flags))
1051 1052
		return TRACE_TYPE_PARTIAL_LINE;

1053
	/* Overhead */
1054
	ret = print_graph_overhead(duration, s, flags);
1055 1056
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;
1057 1058

	/* Duration */
1059
	if (flags & TRACE_GRAPH_PRINT_DURATION) {
1060 1061 1062 1063
		ret = print_graph_duration(duration, s);
		if (ret == TRACE_TYPE_PARTIAL_LINE)
			return TRACE_TYPE_PARTIAL_LINE;
	}
1064 1065

	/* Closing brace */
1066 1067
	for (i = 0; i < trace->depth * TRACE_GRAPH_INDENT; i++) {
		ret = trace_seq_printf(s, " ");
1068 1069
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
1070 1071
	}

1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082
	/*
	 * 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 {
1083
		ret = trace_seq_printf(s, "} /* %ps */\n", (void *)trace->func);
1084 1085 1086
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}
1087

1088
	/* Overrun */
1089
	if (flags & TRACE_GRAPH_PRINT_OVERRUN) {
1090 1091
		ret = trace_seq_printf(s, " (Overruns: %lu)\n",
					trace->overrun);
1092 1093
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
1094
	}
1095

1096 1097
	ret = print_graph_irq(iter, trace->func, TRACE_GRAPH_RET,
			      cpu, pid, flags);
1098 1099 1100
	if (ret == TRACE_TYPE_PARTIAL_LINE)
		return TRACE_TYPE_PARTIAL_LINE;

1101 1102 1103
	return TRACE_TYPE_HANDLED;
}

1104
static enum print_line_t
1105 1106
print_graph_comment(struct trace_seq *s, struct trace_entry *ent,
		    struct trace_iterator *iter, u32 flags)
1107
{
1108
	unsigned long sym_flags = (trace_flags & TRACE_ITER_SYM_MASK);
1109
	struct fgraph_data *data = iter->private;
1110
	struct trace_event *event;
1111
	int depth = 0;
1112
	int ret;
1113 1114 1115
	int i;

	if (data)
1116
		depth = per_cpu_ptr(data->cpu_data, iter->cpu)->depth;
1117

1118
	if (print_graph_prologue(iter, s, 0, 0, flags))
1119 1120
		return TRACE_TYPE_PARTIAL_LINE;

1121
	/* No overhead */
1122
	ret = print_graph_overhead(-1, s, flags);
1123 1124 1125 1126
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

	/* No time */
1127
	if (flags & TRACE_GRAPH_PRINT_DURATION) {
1128
		ret = trace_seq_printf(s, "            |  ");
1129 1130 1131 1132 1133
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}

	/* Indentation */
1134 1135
	if (depth > 0)
		for (i = 0; i < (depth + 1) * TRACE_GRAPH_INDENT; i++) {
1136 1137 1138 1139 1140 1141
			ret = trace_seq_printf(s, " ");
			if (!ret)
				return TRACE_TYPE_PARTIAL_LINE;
		}

	/* The comment */
1142 1143 1144 1145
	ret = trace_seq_printf(s, "/* ");
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161
	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;

1162
		ret = event->funcs->trace(iter, sym_flags, event);
1163 1164 1165
		if (ret != TRACE_TYPE_HANDLED)
			return ret;
	}
1166

1167 1168 1169 1170 1171 1172
	/* Strip ending newline */
	if (s->buffer[s->len - 1] == '\n') {
		s->buffer[s->len - 1] = '\0';
		s->len--;
	}

1173 1174 1175 1176 1177 1178 1179 1180
	ret = trace_seq_printf(s, " */\n");
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

	return TRACE_TYPE_HANDLED;
}


1181
enum print_line_t
1182
print_graph_function_flags(struct trace_iterator *iter, u32 flags)
1183
{
1184 1185
	struct ftrace_graph_ent_entry *field;
	struct fgraph_data *data = iter->private;
1186
	struct trace_entry *entry = iter->ent;
1187
	struct trace_seq *s = &iter->seq;
1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202
	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;
1203
		ret = print_graph_entry(field, s, iter, flags);
1204 1205 1206 1207 1208 1209 1210
		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;
	}
1211

1212 1213
	switch (entry->type) {
	case TRACE_GRAPH_ENT: {
1214 1215 1216 1217 1218 1219
		/*
		 * 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.
		 */
1220
		struct ftrace_graph_ent_entry saved;
1221
		trace_assign_type(field, entry);
1222
		saved = *field;
1223
		return print_graph_entry(&saved, s, iter, flags);
1224 1225 1226 1227
	}
	case TRACE_GRAPH_RET: {
		struct ftrace_graph_ret_entry *field;
		trace_assign_type(field, entry);
1228
		return print_graph_return(&field->ret, s, entry, iter, flags);