trace_syscalls.c 16.4 KB
Newer Older
1
#include <trace/syscall.h>
2
#include <trace/events/syscalls.h>
3
#include <linux/kernel.h>
4
#include <linux/ftrace.h>
5
#include <linux/perf_event.h>
6 7 8 9 10
#include <asm/syscall.h>

#include "trace_output.h"
#include "trace.h"

11
static DEFINE_MUTEX(syscall_trace_lock);
12 13
static int sys_refcount_enter;
static int sys_refcount_exit;
14 15
static DECLARE_BITMAP(enabled_enter_syscalls, NR_syscalls);
static DECLARE_BITMAP(enabled_exit_syscalls, NR_syscalls);
16

17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53
extern unsigned long __start_syscalls_metadata[];
extern unsigned long __stop_syscalls_metadata[];

static struct syscall_metadata **syscalls_metadata;

static struct syscall_metadata *find_syscall_meta(unsigned long syscall)
{
	struct syscall_metadata *start;
	struct syscall_metadata *stop;
	char str[KSYM_SYMBOL_LEN];


	start = (struct syscall_metadata *)__start_syscalls_metadata;
	stop = (struct syscall_metadata *)__stop_syscalls_metadata;
	kallsyms_lookup(syscall, NULL, NULL, NULL, str);

	for ( ; start < stop; start++) {
		/*
		 * Only compare after the "sys" prefix. Archs that use
		 * syscall wrappers may have syscalls symbols aliases prefixed
		 * with "SyS" instead of "sys", leading to an unwanted
		 * mismatch.
		 */
		if (start->name && !strcmp(start->name + 3, str + 3))
			return start;
	}
	return NULL;
}

static struct syscall_metadata *syscall_nr_to_meta(int nr)
{
	if (!syscalls_metadata || nr >= NR_syscalls || nr < 0)
		return NULL;

	return syscalls_metadata[nr];
}

54 55 56 57 58 59 60 61 62
enum print_line_t
print_syscall_enter(struct trace_iterator *iter, int flags)
{
	struct trace_seq *s = &iter->seq;
	struct trace_entry *ent = iter->ent;
	struct syscall_trace_enter *trace;
	struct syscall_metadata *entry;
	int i, ret, syscall;

63
	trace = (typeof(trace))ent;
64 65
	syscall = trace->nr;
	entry = syscall_nr_to_meta(syscall);
66

67 68 69
	if (!entry)
		goto end;

70
	if (entry->enter_event->id != ent->type) {
71 72 73 74
		WARN_ON_ONCE(1);
		goto end;
	}

75 76 77 78 79 80
	ret = trace_seq_printf(s, "%s(", entry->name);
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

	for (i = 0; i < entry->nb_args; i++) {
		/* parameter types */
81
		if (trace_flags & TRACE_ITER_VERBOSE) {
82 83 84 85 86
			ret = trace_seq_printf(s, "%s ", entry->types[i]);
			if (!ret)
				return TRACE_TYPE_PARTIAL_LINE;
		}
		/* parameter values */
87
		ret = trace_seq_printf(s, "%s: %lx%s", entry->args[i],
88
				       trace->args[i],
89
				       i == entry->nb_args - 1 ? "" : ", ");
90 91 92 93
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}

94 95 96 97
	ret = trace_seq_putc(s, ')');
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

98
end:
99 100 101 102
	ret =  trace_seq_putc(s, '\n');
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

103 104 105 106 107 108 109 110 111 112 113 114 115
	return TRACE_TYPE_HANDLED;
}

enum print_line_t
print_syscall_exit(struct trace_iterator *iter, int flags)
{
	struct trace_seq *s = &iter->seq;
	struct trace_entry *ent = iter->ent;
	struct syscall_trace_exit *trace;
	int syscall;
	struct syscall_metadata *entry;
	int ret;

116
	trace = (typeof(trace))ent;
117 118
	syscall = trace->nr;
	entry = syscall_nr_to_meta(syscall);
119

120 121 122 123 124
	if (!entry) {
		trace_seq_printf(s, "\n");
		return TRACE_TYPE_HANDLED;
	}

125
	if (entry->exit_event->id != ent->type) {
126 127 128 129
		WARN_ON_ONCE(1);
		return TRACE_TYPE_UNHANDLED;
	}

130 131 132 133 134 135 136 137
	ret = trace_seq_printf(s, "%s -> 0x%lx\n", entry->name,
				trace->ret);
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

	return TRACE_TYPE_HANDLED;
}

138 139 140 141 142
extern char *__bad_type_size(void);

#define SYSCALL_FIELD(type, name)					\
	sizeof(type) != sizeof(trace.name) ?				\
		__bad_type_size() :					\
143 144
		#type, #name, offsetof(typeof(trace), name),		\
		sizeof(trace.name), is_signed_type(type)
145

146
int syscall_enter_format(struct ftrace_event_call *call, struct trace_seq *s)
147 148
{
	int i;
149
	int ret;
150
	struct syscall_metadata *entry = call->data;
151 152
	struct syscall_trace_enter trace;
	int offset = offsetof(struct syscall_trace_enter, args);
153

154 155
	ret = trace_seq_printf(s, "\tfield:%s %s;\toffset:%zu;\tsize:%zu;"
			       "\tsigned:%u;\n",
156 157 158
			       SYSCALL_FIELD(int, nr));
	if (!ret)
		return 0;
159 160 161 162 163 164

	for (i = 0; i < entry->nb_args; i++) {
		ret = trace_seq_printf(s, "\tfield:%s %s;", entry->types[i],
				        entry->args[i]);
		if (!ret)
			return 0;
165 166 167 168
		ret = trace_seq_printf(s, "\toffset:%d;\tsize:%zu;"
				       "\tsigned:%u;\n", offset,
				       sizeof(unsigned long),
				       is_signed_type(unsigned long));
169 170 171 172 173
		if (!ret)
			return 0;
		offset += sizeof(unsigned long);
	}

174
	trace_seq_puts(s, "\nprint fmt: \"");
175
	for (i = 0; i < entry->nb_args; i++) {
176
		ret = trace_seq_printf(s, "%s: 0x%%0%zulx%s", entry->args[i],
177
				        sizeof(unsigned long),
178
					i == entry->nb_args - 1 ? "" : ", ");
179 180 181
		if (!ret)
			return 0;
	}
182
	trace_seq_putc(s, '"');
183 184

	for (i = 0; i < entry->nb_args; i++) {
185 186
		ret = trace_seq_printf(s, ", ((unsigned long)(REC->%s))",
				       entry->args[i]);
187 188 189 190
		if (!ret)
			return 0;
	}

191
	return trace_seq_putc(s, '\n');
192 193
}

194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254
static
int  __set_enter_print_fmt(struct syscall_metadata *entry, char *buf, int len)
{
	int i;
	int pos = 0;

	/* When len=0, we just calculate the needed length */
#define LEN_OR_ZERO (len ? len - pos : 0)

	pos += snprintf(buf + pos, LEN_OR_ZERO, "\"");
	for (i = 0; i < entry->nb_args; i++) {
		pos += snprintf(buf + pos, LEN_OR_ZERO, "%s: 0x%%0%zulx%s",
				entry->args[i], sizeof(unsigned long),
				i == entry->nb_args - 1 ? "" : ", ");
	}
	pos += snprintf(buf + pos, LEN_OR_ZERO, "\"");

	for (i = 0; i < entry->nb_args; i++) {
		pos += snprintf(buf + pos, LEN_OR_ZERO,
				", ((unsigned long)(REC->%s))", entry->args[i]);
	}

#undef LEN_OR_ZERO

	/* return the length of print_fmt */
	return pos;
}

static int set_syscall_print_fmt(struct ftrace_event_call *call)
{
	char *print_fmt;
	int len;
	struct syscall_metadata *entry = call->data;

	if (entry->enter_event != call) {
		call->print_fmt = "\"0x%lx\", REC->ret";
		return 0;
	}

	/* First: called with 0 length to calculate the needed length */
	len = __set_enter_print_fmt(entry, NULL, 0);

	print_fmt = kmalloc(len + 1, GFP_KERNEL);
	if (!print_fmt)
		return -ENOMEM;

	/* Second: actually write the @print_fmt */
	__set_enter_print_fmt(entry, print_fmt, len + 1);
	call->print_fmt = print_fmt;

	return 0;
}

static void free_syscall_print_fmt(struct ftrace_event_call *call)
{
	struct syscall_metadata *entry = call->data;

	if (entry->enter_event == call)
		kfree(call->print_fmt);
}

255 256 257 258 259 260
int syscall_exit_format(struct ftrace_event_call *call, struct trace_seq *s)
{
	int ret;
	struct syscall_trace_exit trace;

	ret = trace_seq_printf(s,
261 262 263 264
			       "\tfield:%s %s;\toffset:%zu;\tsize:%zu;"
			       "\tsigned:%u;\n"
			       "\tfield:%s %s;\toffset:%zu;\tsize:%zu;"
			       "\tsigned:%u;\n",
265
			       SYSCALL_FIELD(int, nr),
266
			       SYSCALL_FIELD(long, ret));
267 268 269 270 271 272
	if (!ret)
		return 0;

	return trace_seq_printf(s, "\nprint fmt: \"0x%%lx\", REC->ret\n");
}

273 274 275
int syscall_enter_define_fields(struct ftrace_event_call *call)
{
	struct syscall_trace_enter trace;
276
	struct syscall_metadata *meta = call->data;
277 278 279 280
	int ret;
	int i;
	int offset = offsetof(typeof(trace), args);

281 282 283 284
	ret = trace_define_field(call, SYSCALL_FIELD(int, nr), FILTER_OTHER);
	if (ret)
		return ret;

285
	for (i = 0; i < meta->nb_args; i++) {
286 287
		ret = trace_define_field(call, meta->types[i],
					 meta->args[i], offset,
288 289
					 sizeof(unsigned long), 0,
					 FILTER_OTHER);
290 291 292 293 294 295 296 297 298 299 300
		offset += sizeof(unsigned long);
	}

	return ret;
}

int syscall_exit_define_fields(struct ftrace_event_call *call)
{
	struct syscall_trace_exit trace;
	int ret;

301 302 303 304
	ret = trace_define_field(call, SYSCALL_FIELD(int, nr), FILTER_OTHER);
	if (ret)
		return ret;

305
	ret = trace_define_field(call, SYSCALL_FIELD(long, ret),
306
				 FILTER_OTHER);
307 308 309 310

	return ret;
}

311
void ftrace_syscall_enter(struct pt_regs *regs, long id)
312
{
313 314 315
	struct syscall_trace_enter *entry;
	struct syscall_metadata *sys_data;
	struct ring_buffer_event *event;
316
	struct ring_buffer *buffer;
317
	int size;
318 319 320
	int syscall_nr;

	syscall_nr = syscall_get_nr(current, regs);
321 322
	if (syscall_nr < 0)
		return;
323 324
	if (!test_bit(syscall_nr, enabled_enter_syscalls))
		return;
325

326 327 328 329 330 331
	sys_data = syscall_nr_to_meta(syscall_nr);
	if (!sys_data)
		return;

	size = sizeof(*entry) + sizeof(unsigned long) * sys_data->nb_args;

332 333
	event = trace_current_buffer_lock_reserve(&buffer,
			sys_data->enter_event->id, size, 0, 0);
334 335 336 337 338 339 340
	if (!event)
		return;

	entry = ring_buffer_event_data(event);
	entry->nr = syscall_nr;
	syscall_get_arguments(current, regs, 0, sys_data->nb_args, entry->args);

341 342 343
	if (!filter_current_check_discard(buffer, sys_data->enter_event,
					  entry, event))
		trace_current_buffer_unlock_commit(buffer, event, 0, 0);
344 345
}

346
void ftrace_syscall_exit(struct pt_regs *regs, long ret)
347
{
348 349 350
	struct syscall_trace_exit *entry;
	struct syscall_metadata *sys_data;
	struct ring_buffer_event *event;
351
	struct ring_buffer *buffer;
352 353 354
	int syscall_nr;

	syscall_nr = syscall_get_nr(current, regs);
355 356
	if (syscall_nr < 0)
		return;
357 358
	if (!test_bit(syscall_nr, enabled_exit_syscalls))
		return;
359

360 361 362 363
	sys_data = syscall_nr_to_meta(syscall_nr);
	if (!sys_data)
		return;

364 365
	event = trace_current_buffer_lock_reserve(&buffer,
			sys_data->exit_event->id, sizeof(*entry), 0, 0);
366 367 368 369 370 371 372
	if (!event)
		return;

	entry = ring_buffer_event_data(event);
	entry->nr = syscall_nr;
	entry->ret = syscall_get_return_value(current, regs);

373 374 375
	if (!filter_current_check_discard(buffer, sys_data->exit_event,
					  entry, event))
		trace_current_buffer_unlock_commit(buffer, event, 0, 0);
376 377
}

378
int reg_event_syscall_enter(struct ftrace_event_call *call)
379
{
380 381 382
	int ret = 0;
	int num;

383
	num = ((struct syscall_metadata *)call->data)->syscall_nr;
384
	if (num < 0 || num >= NR_syscalls)
385 386 387
		return -ENOSYS;
	mutex_lock(&syscall_trace_lock);
	if (!sys_refcount_enter)
388
		ret = register_trace_sys_enter(ftrace_syscall_enter);
389
	if (!ret) {
390 391 392 393 394
		set_bit(num, enabled_enter_syscalls);
		sys_refcount_enter++;
	}
	mutex_unlock(&syscall_trace_lock);
	return ret;
395 396
}

397
void unreg_event_syscall_enter(struct ftrace_event_call *call)
398
{
399
	int num;
400

401
	num = ((struct syscall_metadata *)call->data)->syscall_nr;
402
	if (num < 0 || num >= NR_syscalls)
403 404 405 406 407
		return;
	mutex_lock(&syscall_trace_lock);
	sys_refcount_enter--;
	clear_bit(num, enabled_enter_syscalls);
	if (!sys_refcount_enter)
408
		unregister_trace_sys_enter(ftrace_syscall_enter);
409 410
	mutex_unlock(&syscall_trace_lock);
}
411

412
int reg_event_syscall_exit(struct ftrace_event_call *call)
413
{
414 415 416
	int ret = 0;
	int num;

417
	num = ((struct syscall_metadata *)call->data)->syscall_nr;
418
	if (num < 0 || num >= NR_syscalls)
419 420 421
		return -ENOSYS;
	mutex_lock(&syscall_trace_lock);
	if (!sys_refcount_exit)
422
		ret = register_trace_sys_exit(ftrace_syscall_exit);
423
	if (!ret) {
424 425
		set_bit(num, enabled_exit_syscalls);
		sys_refcount_exit++;
426
	}
427 428 429
	mutex_unlock(&syscall_trace_lock);
	return ret;
}
430

431
void unreg_event_syscall_exit(struct ftrace_event_call *call)
432 433
{
	int num;
434

435
	num = ((struct syscall_metadata *)call->data)->syscall_nr;
436
	if (num < 0 || num >= NR_syscalls)
437 438 439 440 441
		return;
	mutex_lock(&syscall_trace_lock);
	sys_refcount_exit--;
	clear_bit(num, enabled_exit_syscalls);
	if (!sys_refcount_exit)
442
		unregister_trace_sys_exit(ftrace_syscall_exit);
443
	mutex_unlock(&syscall_trace_lock);
444
}
445

446 447 448 449
int init_syscall_trace(struct ftrace_event_call *call)
{
	int id;

450 451 452
	if (set_syscall_print_fmt(call) < 0)
		return -ENOMEM;

453
	id = register_ftrace_event(call->event);
454 455
	if (!id) {
		free_syscall_print_fmt(call);
456
		return -ENODEV;
457
	}
458 459 460 461 462
	call->id = id;
	INIT_LIST_HEAD(&call->fields);
	return 0;
}

463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478
int __init init_ftrace_syscalls(void)
{
	struct syscall_metadata *meta;
	unsigned long addr;
	int i;

	syscalls_metadata = kzalloc(sizeof(*syscalls_metadata) *
					NR_syscalls, GFP_KERNEL);
	if (!syscalls_metadata) {
		WARN_ON(1);
		return -ENOMEM;
	}

	for (i = 0; i < NR_syscalls; i++) {
		addr = arch_syscall_addr(i);
		meta = find_syscall_meta(addr);
479 480 481 482
		if (!meta)
			continue;

		meta->syscall_nr = i;
483 484 485 486 487 488 489
		syscalls_metadata[i] = meta;
	}

	return 0;
}
core_initcall(init_ftrace_syscalls);

490
#ifdef CONFIG_EVENT_PROFILE
491

492 493
static DECLARE_BITMAP(enabled_prof_enter_syscalls, NR_syscalls);
static DECLARE_BITMAP(enabled_prof_exit_syscalls, NR_syscalls);
494 495 496 497 498 499
static int sys_prof_refcount_enter;
static int sys_prof_refcount_exit;

static void prof_syscall_enter(struct pt_regs *regs, long id)
{
	struct syscall_metadata *sys_data;
500 501
	struct syscall_trace_enter *rec;
	unsigned long flags;
502
	char *trace_buf;
503
	char *raw_data;
504
	int syscall_nr;
505
	int rctx;
506
	int size;
507
	int cpu;
508 509 510 511 512 513 514 515 516

	syscall_nr = syscall_get_nr(current, regs);
	if (!test_bit(syscall_nr, enabled_prof_enter_syscalls))
		return;

	sys_data = syscall_nr_to_meta(syscall_nr);
	if (!sys_data)
		return;

517 518 519 520 521
	/* get the size after alignment with the u32 buffer size field */
	size = sizeof(unsigned long) * sys_data->nb_args + sizeof(*rec);
	size = ALIGN(size + sizeof(u32), sizeof(u64));
	size -= sizeof(u32);

522 523 524 525 526 527
	if (WARN_ONCE(size > FTRACE_MAX_PROFILE_SIZE,
		      "profile buffer not large enough"))
		return;

	/* Protect the per cpu buffer, begin the rcu read side */
	local_irq_save(flags);
528

529 530
	rctx = perf_swevent_get_recursion_context();
	if (rctx < 0)
531 532
		goto end_recursion;

533 534
	cpu = smp_processor_id();

535
	trace_buf = rcu_dereference(perf_trace_buf);
536

537
	if (!trace_buf)
538
		goto end;
539

540
	raw_data = per_cpu_ptr(trace_buf, cpu);
541 542 543 544 545 546

	/* zero the dead bytes from align to not leak stack to user */
	*(u64 *)(&raw_data[size - sizeof(u64)]) = 0ULL;

	rec = (struct syscall_trace_enter *) raw_data;
	tracing_generic_entry_update(&rec->ent, 0, 0);
547
	rec->ent.type = sys_data->enter_event->id;
548 549 550
	rec->nr = syscall_nr;
	syscall_get_arguments(current, regs, 0, sys_data->nb_args,
			       (unsigned long *)&rec->args);
551
	perf_tp_event(sys_data->enter_event->id, 0, 1, rec, size);
552 553

end:
554
	perf_swevent_put_recursion_context(rctx);
555
end_recursion:
556
	local_irq_restore(flags);
557 558
}

559
int prof_sysenter_enable(struct ftrace_event_call *call)
560 561 562 563
{
	int ret = 0;
	int num;

564
	num = ((struct syscall_metadata *)call->data)->syscall_nr;
565 566 567

	mutex_lock(&syscall_trace_lock);
	if (!sys_prof_refcount_enter)
568
		ret = register_trace_sys_enter(prof_syscall_enter);
569 570 571 572 573 574 575 576 577 578 579
	if (ret) {
		pr_info("event trace: Could not activate"
				"syscall entry trace point");
	} else {
		set_bit(num, enabled_prof_enter_syscalls);
		sys_prof_refcount_enter++;
	}
	mutex_unlock(&syscall_trace_lock);
	return ret;
}

580
void prof_sysenter_disable(struct ftrace_event_call *call)
581 582 583
{
	int num;

584
	num = ((struct syscall_metadata *)call->data)->syscall_nr;
585 586 587 588 589

	mutex_lock(&syscall_trace_lock);
	sys_prof_refcount_enter--;
	clear_bit(num, enabled_prof_enter_syscalls);
	if (!sys_prof_refcount_enter)
590
		unregister_trace_sys_enter(prof_syscall_enter);
591 592 593 594 595 596
	mutex_unlock(&syscall_trace_lock);
}

static void prof_syscall_exit(struct pt_regs *regs, long ret)
{
	struct syscall_metadata *sys_data;
597 598
	struct syscall_trace_exit *rec;
	unsigned long flags;
599
	int syscall_nr;
600
	char *trace_buf;
601
	char *raw_data;
602
	int rctx;
603 604
	int size;
	int cpu;
605 606 607 608 609 610 611 612 613

	syscall_nr = syscall_get_nr(current, regs);
	if (!test_bit(syscall_nr, enabled_prof_exit_syscalls))
		return;

	sys_data = syscall_nr_to_meta(syscall_nr);
	if (!sys_data)
		return;

614 615 616
	/* We can probably do that at build time */
	size = ALIGN(sizeof(*rec) + sizeof(u32), sizeof(u64));
	size -= sizeof(u32);
617

618 619 620 621 622 623 624 625 626 627
	/*
	 * Impossible, but be paranoid with the future
	 * How to put this check outside runtime?
	 */
	if (WARN_ONCE(size > FTRACE_MAX_PROFILE_SIZE,
		"exit event has grown above profile buffer size"))
		return;

	/* Protect the per cpu buffer, begin the rcu read side */
	local_irq_save(flags);
628

629 630
	rctx = perf_swevent_get_recursion_context();
	if (rctx < 0)
631 632
		goto end_recursion;

633 634
	cpu = smp_processor_id();

635
	trace_buf = rcu_dereference(perf_trace_buf);
636

637
	if (!trace_buf)
638 639
		goto end;

640
	raw_data = per_cpu_ptr(trace_buf, cpu);
641 642 643 644 645 646 647

	/* zero the dead bytes from align to not leak stack to user */
	*(u64 *)(&raw_data[size - sizeof(u64)]) = 0ULL;

	rec = (struct syscall_trace_exit *)raw_data;

	tracing_generic_entry_update(&rec->ent, 0, 0);
648
	rec->ent.type = sys_data->exit_event->id;
649 650 651
	rec->nr = syscall_nr;
	rec->ret = syscall_get_return_value(current, regs);

652
	perf_tp_event(sys_data->exit_event->id, 0, 1, rec, size);
653 654

end:
655
	perf_swevent_put_recursion_context(rctx);
656
end_recursion:
657
	local_irq_restore(flags);
658 659
}

660
int prof_sysexit_enable(struct ftrace_event_call *call)
661 662 663 664
{
	int ret = 0;
	int num;

665
	num = ((struct syscall_metadata *)call->data)->syscall_nr;
666 667 668

	mutex_lock(&syscall_trace_lock);
	if (!sys_prof_refcount_exit)
669
		ret = register_trace_sys_exit(prof_syscall_exit);
670 671 672 673 674 675 676 677 678 679 680
	if (ret) {
		pr_info("event trace: Could not activate"
				"syscall entry trace point");
	} else {
		set_bit(num, enabled_prof_exit_syscalls);
		sys_prof_refcount_exit++;
	}
	mutex_unlock(&syscall_trace_lock);
	return ret;
}

681
void prof_sysexit_disable(struct ftrace_event_call *call)
682 683 684
{
	int num;

685
	num = ((struct syscall_metadata *)call->data)->syscall_nr;
686 687 688 689 690

	mutex_lock(&syscall_trace_lock);
	sys_prof_refcount_exit--;
	clear_bit(num, enabled_prof_exit_syscalls);
	if (!sys_prof_refcount_exit)
691
		unregister_trace_sys_exit(prof_syscall_exit);
692 693 694 695 696 697
	mutex_unlock(&syscall_trace_lock);
}

#endif