trace_syscalls.c 16 KB
Newer Older
1
#include <trace/syscall.h>
2
#include <trace/events/syscalls.h>
3
#include <linux/slab.h>
4
#include <linux/kernel.h>
5
#include <linux/ftrace.h>
6
#include <linux/perf_event.h>
7
8
9
10
11
#include <asm/syscall.h>

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

12
static DEFINE_MUTEX(syscall_trace_lock);
13
14
static int sys_refcount_enter;
static int sys_refcount_exit;
15
16
static DECLARE_BITMAP(enabled_enter_syscalls, NR_syscalls);
static DECLARE_BITMAP(enabled_exit_syscalls, NR_syscalls);
17

18
19
20
21
22
static int syscall_enter_register(struct ftrace_event_call *event,
				 enum trace_reg type);
static int syscall_exit_register(struct ftrace_event_call *event,
				 enum trace_reg type);

23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
static int syscall_enter_define_fields(struct ftrace_event_call *call);
static int syscall_exit_define_fields(struct ftrace_event_call *call);

static struct list_head *
syscall_get_enter_fields(struct ftrace_event_call *call)
{
	struct syscall_metadata *entry = call->data;

	return &entry->enter_fields;
}

static struct list_head *
syscall_get_exit_fields(struct ftrace_event_call *call)
{
	struct syscall_metadata *entry = call->data;

	return &entry->exit_fields;
}

42
43
44
45
46
47
48
49
struct trace_event_functions enter_syscall_print_funcs = {
	.trace                  = print_syscall_enter,
};

struct trace_event_functions exit_syscall_print_funcs = {
	.trace                  = print_syscall_exit,
};

50
51
struct ftrace_event_class event_class_syscall_enter = {
	.system			= "syscalls",
52
53
54
	.reg			= syscall_enter_register,
	.define_fields		= syscall_enter_define_fields,
	.get_fields		= syscall_get_enter_fields,
55
	.raw_init		= init_syscall_trace,
56
57
58
59
};

struct ftrace_event_class event_class_syscall_exit = {
	.system			= "syscalls",
60
61
62
	.reg			= syscall_exit_register,
	.define_fields		= syscall_exit_define_fields,
	.get_fields		= syscall_get_exit_fields,
63
	.raw_init		= init_syscall_trace,
64
65
};

66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
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];
}

103
enum print_line_t
104
105
print_syscall_enter(struct trace_iterator *iter, int flags,
		    struct trace_event *event)
106
107
108
109
110
111
112
{
	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;

113
	trace = (typeof(trace))ent;
114
115
	syscall = trace->nr;
	entry = syscall_nr_to_meta(syscall);
116

117
118
119
	if (!entry)
		goto end;

120
	if (entry->enter_event->id != ent->type) {
121
122
123
124
		WARN_ON_ONCE(1);
		goto end;
	}

125
126
127
128
129
130
	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 */
131
		if (trace_flags & TRACE_ITER_VERBOSE) {
132
133
134
135
136
			ret = trace_seq_printf(s, "%s ", entry->types[i]);
			if (!ret)
				return TRACE_TYPE_PARTIAL_LINE;
		}
		/* parameter values */
137
		ret = trace_seq_printf(s, "%s: %lx%s", entry->args[i],
138
				       trace->args[i],
139
				       i == entry->nb_args - 1 ? "" : ", ");
140
141
142
143
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}

144
145
146
147
	ret = trace_seq_putc(s, ')');
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

148
end:
149
150
151
152
	ret =  trace_seq_putc(s, '\n');
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

153
154
155
156
	return TRACE_TYPE_HANDLED;
}

enum print_line_t
157
158
print_syscall_exit(struct trace_iterator *iter, int flags,
		   struct trace_event *event)
159
160
161
162
163
164
165
166
{
	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;

167
	trace = (typeof(trace))ent;
168
169
	syscall = trace->nr;
	entry = syscall_nr_to_meta(syscall);
170

171
172
173
174
175
	if (!entry) {
		trace_seq_printf(s, "\n");
		return TRACE_TYPE_HANDLED;
	}

176
	if (entry->exit_event->id != ent->type) {
177
178
179
180
		WARN_ON_ONCE(1);
		return TRACE_TYPE_UNHANDLED;
	}

181
182
183
184
185
186
187
188
	ret = trace_seq_printf(s, "%s -> 0x%lx\n", entry->name,
				trace->ret);
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

	return TRACE_TYPE_HANDLED;
}

189
190
191
192
193
extern char *__bad_type_size(void);

#define SYSCALL_FIELD(type, name)					\
	sizeof(type) != sizeof(trace.name) ?				\
		__bad_type_size() :					\
194
195
		#type, #name, offsetof(typeof(trace), name),		\
		sizeof(trace.name), is_signed_type(type)
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
255
256
257
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);
}

258
static int syscall_enter_define_fields(struct ftrace_event_call *call)
259
260
{
	struct syscall_trace_enter trace;
261
	struct syscall_metadata *meta = call->data;
262
263
264
265
	int ret;
	int i;
	int offset = offsetof(typeof(trace), args);

266
267
268
269
	ret = trace_define_field(call, SYSCALL_FIELD(int, nr), FILTER_OTHER);
	if (ret)
		return ret;

270
	for (i = 0; i < meta->nb_args; i++) {
271
272
		ret = trace_define_field(call, meta->types[i],
					 meta->args[i], offset,
273
274
					 sizeof(unsigned long), 0,
					 FILTER_OTHER);
275
276
277
278
279
280
		offset += sizeof(unsigned long);
	}

	return ret;
}

281
static int syscall_exit_define_fields(struct ftrace_event_call *call)
282
283
284
285
{
	struct syscall_trace_exit trace;
	int ret;

286
287
288
289
	ret = trace_define_field(call, SYSCALL_FIELD(int, nr), FILTER_OTHER);
	if (ret)
		return ret;

290
	ret = trace_define_field(call, SYSCALL_FIELD(long, ret),
291
				 FILTER_OTHER);
292
293
294
295

	return ret;
}

296
void ftrace_syscall_enter(void *ignore, struct pt_regs *regs, long id)
297
{
298
299
300
	struct syscall_trace_enter *entry;
	struct syscall_metadata *sys_data;
	struct ring_buffer_event *event;
301
	struct ring_buffer *buffer;
302
	int size;
303
304
305
	int syscall_nr;

	syscall_nr = syscall_get_nr(current, regs);
306
307
	if (syscall_nr < 0)
		return;
308
309
	if (!test_bit(syscall_nr, enabled_enter_syscalls))
		return;
310

311
312
313
314
315
316
	sys_data = syscall_nr_to_meta(syscall_nr);
	if (!sys_data)
		return;

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

317
318
	event = trace_current_buffer_lock_reserve(&buffer,
			sys_data->enter_event->id, size, 0, 0);
319
320
321
322
323
324
325
	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);

326
327
328
	if (!filter_current_check_discard(buffer, sys_data->enter_event,
					  entry, event))
		trace_current_buffer_unlock_commit(buffer, event, 0, 0);
329
330
}

331
void ftrace_syscall_exit(void *ignore, struct pt_regs *regs, long ret)
332
{
333
334
335
	struct syscall_trace_exit *entry;
	struct syscall_metadata *sys_data;
	struct ring_buffer_event *event;
336
	struct ring_buffer *buffer;
337
338
339
	int syscall_nr;

	syscall_nr = syscall_get_nr(current, regs);
340
341
	if (syscall_nr < 0)
		return;
342
343
	if (!test_bit(syscall_nr, enabled_exit_syscalls))
		return;
344

345
346
347
348
	sys_data = syscall_nr_to_meta(syscall_nr);
	if (!sys_data)
		return;

349
350
	event = trace_current_buffer_lock_reserve(&buffer,
			sys_data->exit_event->id, sizeof(*entry), 0, 0);
351
352
353
354
355
356
357
	if (!event)
		return;

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

358
359
360
	if (!filter_current_check_discard(buffer, sys_data->exit_event,
					  entry, event))
		trace_current_buffer_unlock_commit(buffer, event, 0, 0);
361
362
}

363
int reg_event_syscall_enter(struct ftrace_event_call *call)
364
{
365
366
367
	int ret = 0;
	int num;

368
	num = ((struct syscall_metadata *)call->data)->syscall_nr;
369
	if (num < 0 || num >= NR_syscalls)
370
371
372
		return -ENOSYS;
	mutex_lock(&syscall_trace_lock);
	if (!sys_refcount_enter)
373
		ret = register_trace_sys_enter(ftrace_syscall_enter, NULL);
374
	if (!ret) {
375
376
377
378
379
		set_bit(num, enabled_enter_syscalls);
		sys_refcount_enter++;
	}
	mutex_unlock(&syscall_trace_lock);
	return ret;
380
381
}

382
void unreg_event_syscall_enter(struct ftrace_event_call *call)
383
{
384
	int num;
385

386
	num = ((struct syscall_metadata *)call->data)->syscall_nr;
387
	if (num < 0 || num >= NR_syscalls)
388
389
390
391
392
		return;
	mutex_lock(&syscall_trace_lock);
	sys_refcount_enter--;
	clear_bit(num, enabled_enter_syscalls);
	if (!sys_refcount_enter)
393
		unregister_trace_sys_enter(ftrace_syscall_enter, NULL);
394
395
	mutex_unlock(&syscall_trace_lock);
}
396

397
int reg_event_syscall_exit(struct ftrace_event_call *call)
398
{
399
400
401
	int ret = 0;
	int num;

402
	num = ((struct syscall_metadata *)call->data)->syscall_nr;
403
	if (num < 0 || num >= NR_syscalls)
404
405
406
		return -ENOSYS;
	mutex_lock(&syscall_trace_lock);
	if (!sys_refcount_exit)
407
		ret = register_trace_sys_exit(ftrace_syscall_exit, NULL);
408
	if (!ret) {
409
410
		set_bit(num, enabled_exit_syscalls);
		sys_refcount_exit++;
411
	}
412
413
414
	mutex_unlock(&syscall_trace_lock);
	return ret;
}
415

416
void unreg_event_syscall_exit(struct ftrace_event_call *call)
417
418
{
	int num;
419

420
	num = ((struct syscall_metadata *)call->data)->syscall_nr;
421
	if (num < 0 || num >= NR_syscalls)
422
423
424
425
426
		return;
	mutex_lock(&syscall_trace_lock);
	sys_refcount_exit--;
	clear_bit(num, enabled_exit_syscalls);
	if (!sys_refcount_exit)
427
		unregister_trace_sys_exit(ftrace_syscall_exit, NULL);
428
	mutex_unlock(&syscall_trace_lock);
429
}
430

431
432
433
434
int init_syscall_trace(struct ftrace_event_call *call)
{
	int id;

435
436
437
	if (set_syscall_print_fmt(call) < 0)
		return -ENOMEM;

438
439
440
	id = trace_event_raw_init(call);

	if (id < 0) {
441
		free_syscall_print_fmt(call);
442
		return id;
443
	}
444
445

	return id;
446
447
}

448
449
450
451
452
unsigned long __init arch_syscall_addr(int nr)
{
	return (unsigned long)sys_call_table[nr];
}

453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
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);
469
470
471
472
		if (!meta)
			continue;

		meta->syscall_nr = i;
473
474
475
476
477
478
479
		syscalls_metadata[i] = meta;
	}

	return 0;
}
core_initcall(init_ftrace_syscalls);

480
#ifdef CONFIG_PERF_EVENTS
481

482
483
484
485
static DECLARE_BITMAP(enabled_perf_enter_syscalls, NR_syscalls);
static DECLARE_BITMAP(enabled_perf_exit_syscalls, NR_syscalls);
static int sys_perf_refcount_enter;
static int sys_perf_refcount_exit;
486

487
static void perf_syscall_enter(void *ignore, struct pt_regs *regs, long id)
488
489
{
	struct syscall_metadata *sys_data;
490
491
	struct syscall_trace_enter *rec;
	unsigned long flags;
492
	int syscall_nr;
493
	int rctx;
494
	int size;
495
496

	syscall_nr = syscall_get_nr(current, regs);
497
	if (!test_bit(syscall_nr, enabled_perf_enter_syscalls))
498
499
500
501
502
503
		return;

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

504
505
506
507
508
	/* 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);

509
510
	if (WARN_ONCE(size > PERF_MAX_TRACE_SIZE,
		      "perf buffer not large enough"))
511
512
		return;

513
	rec = (struct syscall_trace_enter *)perf_trace_buf_prepare(size,
514
515
516
				sys_data->enter_event->id, &rctx, &flags);
	if (!rec)
		return;
517
518
519
520

	rec->nr = syscall_nr;
	syscall_get_arguments(current, regs, 0, sys_data->nb_args,
			       (unsigned long *)&rec->args);
521
	perf_trace_buf_submit(rec, size, rctx, 0, 1, flags, regs);
522
523
}

524
int perf_sysenter_enable(struct ftrace_event_call *call)
525
526
527
528
{
	int ret = 0;
	int num;

529
	num = ((struct syscall_metadata *)call->data)->syscall_nr;
530
531

	mutex_lock(&syscall_trace_lock);
532
	if (!sys_perf_refcount_enter)
533
		ret = register_trace_sys_enter(perf_syscall_enter, NULL);
534
535
536
537
	if (ret) {
		pr_info("event trace: Could not activate"
				"syscall entry trace point");
	} else {
538
539
		set_bit(num, enabled_perf_enter_syscalls);
		sys_perf_refcount_enter++;
540
541
542
543
544
	}
	mutex_unlock(&syscall_trace_lock);
	return ret;
}

545
void perf_sysenter_disable(struct ftrace_event_call *call)
546
547
548
{
	int num;

549
	num = ((struct syscall_metadata *)call->data)->syscall_nr;
550
551

	mutex_lock(&syscall_trace_lock);
552
553
554
	sys_perf_refcount_enter--;
	clear_bit(num, enabled_perf_enter_syscalls);
	if (!sys_perf_refcount_enter)
555
		unregister_trace_sys_enter(perf_syscall_enter, NULL);
556
557
558
	mutex_unlock(&syscall_trace_lock);
}

559
static void perf_syscall_exit(void *ignore, struct pt_regs *regs, long ret)
560
561
{
	struct syscall_metadata *sys_data;
562
563
	struct syscall_trace_exit *rec;
	unsigned long flags;
564
	int syscall_nr;
565
	int rctx;
566
	int size;
567
568

	syscall_nr = syscall_get_nr(current, regs);
569
	if (!test_bit(syscall_nr, enabled_perf_exit_syscalls))
570
571
572
573
574
575
		return;

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

576
577
578
	/* We can probably do that at build time */
	size = ALIGN(sizeof(*rec) + sizeof(u32), sizeof(u64));
	size -= sizeof(u32);
579

580
581
582
583
	/*
	 * Impossible, but be paranoid with the future
	 * How to put this check outside runtime?
	 */
584
585
	if (WARN_ONCE(size > PERF_MAX_TRACE_SIZE,
		"exit event has grown above perf buffer size"))
586
587
		return;

588
	rec = (struct syscall_trace_exit *)perf_trace_buf_prepare(size,
589
590
591
				sys_data->exit_event->id, &rctx, &flags);
	if (!rec)
		return;
592
593
594
595

	rec->nr = syscall_nr;
	rec->ret = syscall_get_return_value(current, regs);

596
	perf_trace_buf_submit(rec, size, rctx, 0, 1, flags, regs);
597
598
}

599
int perf_sysexit_enable(struct ftrace_event_call *call)
600
601
602
603
{
	int ret = 0;
	int num;

604
	num = ((struct syscall_metadata *)call->data)->syscall_nr;
605
606

	mutex_lock(&syscall_trace_lock);
607
	if (!sys_perf_refcount_exit)
608
		ret = register_trace_sys_exit(perf_syscall_exit, NULL);
609
610
	if (ret) {
		pr_info("event trace: Could not activate"
611
				"syscall exit trace point");
612
	} else {
613
614
		set_bit(num, enabled_perf_exit_syscalls);
		sys_perf_refcount_exit++;
615
616
617
618
619
	}
	mutex_unlock(&syscall_trace_lock);
	return ret;
}

620
void perf_sysexit_disable(struct ftrace_event_call *call)
621
622
623
{
	int num;

624
	num = ((struct syscall_metadata *)call->data)->syscall_nr;
625
626

	mutex_lock(&syscall_trace_lock);
627
628
629
	sys_perf_refcount_exit--;
	clear_bit(num, enabled_perf_exit_syscalls);
	if (!sys_perf_refcount_exit)
630
		unregister_trace_sys_exit(perf_syscall_exit, NULL);
631
632
633
	mutex_unlock(&syscall_trace_lock);
}

634
#endif /* CONFIG_PERF_EVENTS */
635

636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
static int syscall_enter_register(struct ftrace_event_call *event,
				 enum trace_reg type)
{
	switch (type) {
	case TRACE_REG_REGISTER:
		return reg_event_syscall_enter(event);
	case TRACE_REG_UNREGISTER:
		unreg_event_syscall_enter(event);
		return 0;

#ifdef CONFIG_PERF_EVENTS
	case TRACE_REG_PERF_REGISTER:
		return perf_sysenter_enable(event);
	case TRACE_REG_PERF_UNREGISTER:
		perf_sysenter_disable(event);
		return 0;
#endif
	}
	return 0;
}

static int syscall_exit_register(struct ftrace_event_call *event,
				 enum trace_reg type)
{
	switch (type) {
	case TRACE_REG_REGISTER:
		return reg_event_syscall_exit(event);
	case TRACE_REG_UNREGISTER:
		unreg_event_syscall_exit(event);
		return 0;

#ifdef CONFIG_PERF_EVENTS
	case TRACE_REG_PERF_REGISTER:
		return perf_sysexit_enable(event);
	case TRACE_REG_PERF_UNREGISTER:
		perf_sysexit_disable(event);
		return 0;
#endif
	}
	return 0;
}