trace_syscalls.c 15.2 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
static int syscall_name_to_nr(const char *name)
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
{
	int i;

	if (!syscalls_metadata)
		return -1;

	for (i = 0; i < NR_syscalls; i++) {
		if (syscalls_metadata[i]) {
			if (!strcmp(syscalls_metadata[i]->name, name))
				return i;
		}
	}
	return -1;
}

70
71
72
73
74
75
76
77
78
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;

79
	trace = (typeof(trace))ent;
80
81
	syscall = trace->nr;
	entry = syscall_nr_to_meta(syscall);
82

83
84
85
	if (!entry)
		goto end;

86
	if (entry->enter_event->id != ent->type) {
87
88
89
90
		WARN_ON_ONCE(1);
		goto end;
	}

91
92
93
94
95
96
	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 */
97
		if (trace_flags & TRACE_ITER_VERBOSE) {
98
99
100
101
102
			ret = trace_seq_printf(s, "%s ", entry->types[i]);
			if (!ret)
				return TRACE_TYPE_PARTIAL_LINE;
		}
		/* parameter values */
103
		ret = trace_seq_printf(s, "%s: %lx%s", entry->args[i],
104
				       trace->args[i],
105
				       i == entry->nb_args - 1 ? "" : ", ");
106
107
108
109
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}

110
111
112
113
	ret = trace_seq_putc(s, ')');
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

114
end:
115
116
117
118
	ret =  trace_seq_putc(s, '\n');
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

119
120
121
122
123
124
125
126
127
128
129
130
131
	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;

132
	trace = (typeof(trace))ent;
133
134
	syscall = trace->nr;
	entry = syscall_nr_to_meta(syscall);
135

136
137
138
139
140
	if (!entry) {
		trace_seq_printf(s, "\n");
		return TRACE_TYPE_HANDLED;
	}

141
	if (entry->exit_event->id != ent->type) {
142
143
144
145
		WARN_ON_ONCE(1);
		return TRACE_TYPE_UNHANDLED;
	}

146
147
148
149
150
151
152
153
	ret = trace_seq_printf(s, "%s -> 0x%lx\n", entry->name,
				trace->ret);
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

	return TRACE_TYPE_HANDLED;
}

154
155
156
157
158
extern char *__bad_type_size(void);

#define SYSCALL_FIELD(type, name)					\
	sizeof(type) != sizeof(trace.name) ?				\
		__bad_type_size() :					\
159
160
		#type, #name, offsetof(typeof(trace), name),		\
		sizeof(trace.name), is_signed_type(type)
161

162
int syscall_enter_format(struct ftrace_event_call *call, struct trace_seq *s)
163
164
{
	int i;
165
	int ret;
166
	struct syscall_metadata *entry = call->data;
167
168
	struct syscall_trace_enter trace;
	int offset = offsetof(struct syscall_trace_enter, args);
169

170
171
	ret = trace_seq_printf(s, "\tfield:%s %s;\toffset:%zu;\tsize:%zu;"
			       "\tsigned:%u;\n",
172
173
174
			       SYSCALL_FIELD(int, nr));
	if (!ret)
		return 0;
175
176
177
178
179
180

	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;
181
182
183
184
		ret = trace_seq_printf(s, "\toffset:%d;\tsize:%zu;"
				       "\tsigned:%u;\n", offset,
				       sizeof(unsigned long),
				       is_signed_type(unsigned long));
185
186
187
188
189
		if (!ret)
			return 0;
		offset += sizeof(unsigned long);
	}

190
	trace_seq_puts(s, "\nprint fmt: \"");
191
	for (i = 0; i < entry->nb_args; i++) {
192
		ret = trace_seq_printf(s, "%s: 0x%%0%zulx%s", entry->args[i],
193
				        sizeof(unsigned long),
194
					i == entry->nb_args - 1 ? "" : ", ");
195
196
197
		if (!ret)
			return 0;
	}
198
	trace_seq_putc(s, '"');
199
200

	for (i = 0; i < entry->nb_args; i++) {
201
202
		ret = trace_seq_printf(s, ", ((unsigned long)(REC->%s))",
				       entry->args[i]);
203
204
205
206
		if (!ret)
			return 0;
	}

207
	return trace_seq_putc(s, '\n');
208
209
}

210
211
212
213
214
215
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,
216
217
218
219
			       "\tfield:%s %s;\toffset:%zu;\tsize:%zu;"
			       "\tsigned:%u;\n"
			       "\tfield:%s %s;\toffset:%zu;\tsize:%zu;"
			       "\tsigned:%u;\n",
220
			       SYSCALL_FIELD(int, nr),
221
			       SYSCALL_FIELD(long, ret));
222
223
224
225
226
227
	if (!ret)
		return 0;

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

228
229
230
int syscall_enter_define_fields(struct ftrace_event_call *call)
{
	struct syscall_trace_enter trace;
231
	struct syscall_metadata *meta = call->data;
232
233
234
235
236
237
238
239
	int ret;
	int i;
	int offset = offsetof(typeof(trace), args);

	ret = trace_define_common_fields(call);
	if (ret)
		return ret;

240
241
242
243
	ret = trace_define_field(call, SYSCALL_FIELD(int, nr), FILTER_OTHER);
	if (ret)
		return ret;

244
	for (i = 0; i < meta->nb_args; i++) {
245
246
		ret = trace_define_field(call, meta->types[i],
					 meta->args[i], offset,
247
248
					 sizeof(unsigned long), 0,
					 FILTER_OTHER);
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
		offset += sizeof(unsigned long);
	}

	return ret;
}

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

	ret = trace_define_common_fields(call);
	if (ret)
		return ret;

264
265
266
267
	ret = trace_define_field(call, SYSCALL_FIELD(int, nr), FILTER_OTHER);
	if (ret)
		return ret;

268
	ret = trace_define_field(call, SYSCALL_FIELD(long, ret),
269
				 FILTER_OTHER);
270
271
272
273

	return ret;
}

274
void ftrace_syscall_enter(struct pt_regs *regs, long id)
275
{
276
277
278
	struct syscall_trace_enter *entry;
	struct syscall_metadata *sys_data;
	struct ring_buffer_event *event;
279
	struct ring_buffer *buffer;
280
	int size;
281
282
283
	int syscall_nr;

	syscall_nr = syscall_get_nr(current, regs);
284
285
	if (syscall_nr < 0)
		return;
286
287
	if (!test_bit(syscall_nr, enabled_enter_syscalls))
		return;
288

289
290
291
292
293
294
	sys_data = syscall_nr_to_meta(syscall_nr);
	if (!sys_data)
		return;

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

295
296
	event = trace_current_buffer_lock_reserve(&buffer,
			sys_data->enter_event->id, size, 0, 0);
297
298
299
300
301
302
303
	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);

304
305
306
	if (!filter_current_check_discard(buffer, sys_data->enter_event,
					  entry, event))
		trace_current_buffer_unlock_commit(buffer, event, 0, 0);
307
308
}

309
void ftrace_syscall_exit(struct pt_regs *regs, long ret)
310
{
311
312
313
	struct syscall_trace_exit *entry;
	struct syscall_metadata *sys_data;
	struct ring_buffer_event *event;
314
	struct ring_buffer *buffer;
315
316
317
	int syscall_nr;

	syscall_nr = syscall_get_nr(current, regs);
318
319
	if (syscall_nr < 0)
		return;
320
321
	if (!test_bit(syscall_nr, enabled_exit_syscalls))
		return;
322

323
324
325
326
	sys_data = syscall_nr_to_meta(syscall_nr);
	if (!sys_data)
		return;

327
328
	event = trace_current_buffer_lock_reserve(&buffer,
			sys_data->exit_event->id, sizeof(*entry), 0, 0);
329
330
331
332
333
334
335
	if (!event)
		return;

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

336
337
338
	if (!filter_current_check_discard(buffer, sys_data->exit_event,
					  entry, event))
		trace_current_buffer_unlock_commit(buffer, event, 0, 0);
339
340
}

341
int reg_event_syscall_enter(struct ftrace_event_call *call)
342
{
343
344
345
	int ret = 0;
	int num;

346
	num = ((struct syscall_metadata *)call->data)->syscall_nr;
347
	if (num < 0 || num >= NR_syscalls)
348
349
350
		return -ENOSYS;
	mutex_lock(&syscall_trace_lock);
	if (!sys_refcount_enter)
351
		ret = register_trace_sys_enter(ftrace_syscall_enter);
352
353
354
355
356
357
358
359
360
	if (ret) {
		pr_info("event trace: Could not activate"
				"syscall entry trace point");
	} else {
		set_bit(num, enabled_enter_syscalls);
		sys_refcount_enter++;
	}
	mutex_unlock(&syscall_trace_lock);
	return ret;
361
362
}

363
void unreg_event_syscall_enter(struct ftrace_event_call *call)
364
{
365
	int num;
366

367
	num = ((struct syscall_metadata *)call->data)->syscall_nr;
368
	if (num < 0 || num >= NR_syscalls)
369
370
371
372
373
		return;
	mutex_lock(&syscall_trace_lock);
	sys_refcount_enter--;
	clear_bit(num, enabled_enter_syscalls);
	if (!sys_refcount_enter)
374
		unregister_trace_sys_enter(ftrace_syscall_enter);
375
376
	mutex_unlock(&syscall_trace_lock);
}
377

378
int reg_event_syscall_exit(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_exit)
388
		ret = register_trace_sys_exit(ftrace_syscall_exit);
389
390
391
392
393
394
	if (ret) {
		pr_info("event trace: Could not activate"
				"syscall exit trace point");
	} else {
		set_bit(num, enabled_exit_syscalls);
		sys_refcount_exit++;
395
	}
396
397
398
	mutex_unlock(&syscall_trace_lock);
	return ret;
}
399

400
void unreg_event_syscall_exit(struct ftrace_event_call *call)
401
402
{
	int num;
403

404
	num = ((struct syscall_metadata *)call->data)->syscall_nr;
405
	if (num < 0 || num >= NR_syscalls)
406
407
408
409
410
		return;
	mutex_lock(&syscall_trace_lock);
	sys_refcount_exit--;
	clear_bit(num, enabled_exit_syscalls);
	if (!sys_refcount_exit)
411
		unregister_trace_sys_exit(ftrace_syscall_exit);
412
	mutex_unlock(&syscall_trace_lock);
413
}
414

415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
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);
431
432
433
434
		if (!meta)
			continue;

		meta->syscall_nr = i;
435
436
437
438
439
440
441
		syscalls_metadata[i] = meta;
	}

	return 0;
}
core_initcall(init_ftrace_syscalls);

442
#ifdef CONFIG_EVENT_PROFILE
443

444
445
static DECLARE_BITMAP(enabled_prof_enter_syscalls, NR_syscalls);
static DECLARE_BITMAP(enabled_prof_exit_syscalls, NR_syscalls);
446
447
448
449
450
451
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;
452
453
	struct syscall_trace_enter *rec;
	unsigned long flags;
454
	char *trace_buf;
455
	char *raw_data;
456
	int syscall_nr;
457
	int rctx;
458
	int size;
459
	int cpu;
460
461
462
463
464
465
466
467
468

	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;

469
470
471
472
473
	/* 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);

474
475
476
477
478
479
	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);
480

481
482
	rctx = perf_swevent_get_recursion_context();
	if (rctx < 0)
483
484
		goto end_recursion;

485
486
	cpu = smp_processor_id();

487
	trace_buf = rcu_dereference(perf_trace_buf);
488

489
	if (!trace_buf)
490
		goto end;
491

492
	raw_data = per_cpu_ptr(trace_buf, cpu);
493
494
495
496
497
498

	/* 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);
499
	rec->ent.type = sys_data->enter_event->id;
500
501
502
	rec->nr = syscall_nr;
	syscall_get_arguments(current, regs, 0, sys_data->nb_args,
			       (unsigned long *)&rec->args);
503
	perf_tp_event(sys_data->enter_event->id, 0, 1, rec, size);
504
505

end:
506
	perf_swevent_put_recursion_context(rctx);
507
end_recursion:
508
	local_irq_restore(flags);
509
510
511
512
513
514
515
516
}

int reg_prof_syscall_enter(char *name)
{
	int ret = 0;
	int num;

	num = syscall_name_to_nr(name);
517
	if (num < 0 || num >= NR_syscalls)
518
519
520
521
		return -ENOSYS;

	mutex_lock(&syscall_trace_lock);
	if (!sys_prof_refcount_enter)
522
		ret = register_trace_sys_enter(prof_syscall_enter);
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
	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;
}

void unreg_prof_syscall_enter(char *name)
{
	int num;

	num = syscall_name_to_nr(name);
539
	if (num < 0 || num >= NR_syscalls)
540
541
542
543
544
545
		return;

	mutex_lock(&syscall_trace_lock);
	sys_prof_refcount_enter--;
	clear_bit(num, enabled_prof_enter_syscalls);
	if (!sys_prof_refcount_enter)
546
		unregister_trace_sys_enter(prof_syscall_enter);
547
548
549
550
551
552
	mutex_unlock(&syscall_trace_lock);
}

static void prof_syscall_exit(struct pt_regs *regs, long ret)
{
	struct syscall_metadata *sys_data;
553
554
	struct syscall_trace_exit *rec;
	unsigned long flags;
555
	int syscall_nr;
556
	char *trace_buf;
557
	char *raw_data;
558
	int rctx;
559
560
	int size;
	int cpu;
561
562
563
564
565
566
567
568
569

	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;

570
571
572
	/* We can probably do that at build time */
	size = ALIGN(sizeof(*rec) + sizeof(u32), sizeof(u64));
	size -= sizeof(u32);
573

574
575
576
577
578
579
580
581
582
583
	/*
	 * 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);
584

585
586
	rctx = perf_swevent_get_recursion_context();
	if (rctx < 0)
587
588
		goto end_recursion;

589
590
	cpu = smp_processor_id();

591
	trace_buf = rcu_dereference(perf_trace_buf);
592

593
	if (!trace_buf)
594
595
		goto end;

596
	raw_data = per_cpu_ptr(trace_buf, cpu);
597
598
599
600
601
602
603

	/* 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);
604
	rec->ent.type = sys_data->exit_event->id;
605
606
607
	rec->nr = syscall_nr;
	rec->ret = syscall_get_return_value(current, regs);

608
	perf_tp_event(sys_data->exit_event->id, 0, 1, rec, size);
609
610

end:
611
	perf_swevent_put_recursion_context(rctx);
612
end_recursion:
613
	local_irq_restore(flags);
614
615
616
617
618
619
620
621
}

int reg_prof_syscall_exit(char *name)
{
	int ret = 0;
	int num;

	num = syscall_name_to_nr(name);
622
	if (num < 0 || num >= NR_syscalls)
623
624
625
626
		return -ENOSYS;

	mutex_lock(&syscall_trace_lock);
	if (!sys_prof_refcount_exit)
627
		ret = register_trace_sys_exit(prof_syscall_exit);
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
	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;
}

void unreg_prof_syscall_exit(char *name)
{
	int num;

	num = syscall_name_to_nr(name);
644
	if (num < 0 || num >= NR_syscalls)
645
646
647
648
649
650
		return;

	mutex_lock(&syscall_trace_lock);
	sys_prof_refcount_exit--;
	clear_bit(num, enabled_prof_exit_syscalls);
	if (!sys_prof_refcount_exit)
651
		unregister_trace_sys_exit(prof_syscall_exit);
652
653
654
655
656
657
	mutex_unlock(&syscall_trace_lock);
}

#endif