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
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;
}

34
struct trace_event_functions enter_syscall_print_funcs = {
35
	.trace		= print_syscall_enter,
36
37
38
};

struct trace_event_functions exit_syscall_print_funcs = {
39
	.trace		= print_syscall_exit,
40
41
};

42
struct ftrace_event_class event_class_syscall_enter = {
43
44
45
46
47
	.system		= "syscalls",
	.reg		= syscall_enter_register,
	.define_fields	= syscall_enter_define_fields,
	.get_fields	= syscall_get_enter_fields,
	.raw_init	= init_syscall_trace,
48
49
50
};

struct ftrace_event_class event_class_syscall_exit = {
51
52
53
54
55
	.system		= "syscalls",
	.reg		= syscall_exit_register,
	.define_fields	= syscall_exit_define_fields,
	.fields		= LIST_HEAD_INIT(event_class_syscall_exit.fields),
	.raw_init	= init_syscall_trace,
56
57
};

58
59
extern struct syscall_metadata *__start_syscalls_metadata[];
extern struct syscall_metadata *__stop_syscalls_metadata[];
60
61
62

static struct syscall_metadata **syscalls_metadata;

63
64
static __init struct syscall_metadata *
find_syscall_meta(unsigned long syscall)
65
{
66
67
	struct syscall_metadata **start;
	struct syscall_metadata **stop;
68
69
70
	char str[KSYM_SYMBOL_LEN];


71
72
	start = __start_syscalls_metadata;
	stop = __stop_syscalls_metadata;
73
74
75
76
77
78
79
80
81
	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.
		 */
82
83
		if ((*start)->name && !strcmp((*start)->name + 3, str + 3))
			return *start;
84
85
86
87
88
89
90
91
92
93
94
95
	}
	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];
}

96
enum print_line_t
97
98
print_syscall_enter(struct trace_iterator *iter, int flags,
		    struct trace_event *event)
99
100
101
102
103
104
105
{
	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;

106
	trace = (typeof(trace))ent;
107
108
	syscall = trace->nr;
	entry = syscall_nr_to_meta(syscall);
109

110
111
112
	if (!entry)
		goto end;

113
	if (entry->enter_event->event.type != ent->type) {
114
115
116
117
		WARN_ON_ONCE(1);
		goto end;
	}

118
119
120
121
122
123
	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 */
124
		if (trace_flags & TRACE_ITER_VERBOSE) {
125
126
127
128
129
			ret = trace_seq_printf(s, "%s ", entry->types[i]);
			if (!ret)
				return TRACE_TYPE_PARTIAL_LINE;
		}
		/* parameter values */
130
		ret = trace_seq_printf(s, "%s: %lx%s", entry->args[i],
131
				       trace->args[i],
132
				       i == entry->nb_args - 1 ? "" : ", ");
133
134
135
136
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}

137
138
139
140
	ret = trace_seq_putc(s, ')');
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

141
end:
142
143
144
145
	ret =  trace_seq_putc(s, '\n');
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

146
147
148
149
	return TRACE_TYPE_HANDLED;
}

enum print_line_t
150
151
print_syscall_exit(struct trace_iterator *iter, int flags,
		   struct trace_event *event)
152
153
154
155
156
157
158
159
{
	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;

160
	trace = (typeof(trace))ent;
161
162
	syscall = trace->nr;
	entry = syscall_nr_to_meta(syscall);
163

164
165
166
167
168
	if (!entry) {
		trace_seq_printf(s, "\n");
		return TRACE_TYPE_HANDLED;
	}

169
	if (entry->exit_event->event.type != ent->type) {
170
171
172
173
		WARN_ON_ONCE(1);
		return TRACE_TYPE_UNHANDLED;
	}

174
175
176
177
178
179
180
181
	ret = trace_seq_printf(s, "%s -> 0x%lx\n", entry->name,
				trace->ret);
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

	return TRACE_TYPE_HANDLED;
}

182
183
184
185
186
extern char *__bad_type_size(void);

#define SYSCALL_FIELD(type, name)					\
	sizeof(type) != sizeof(trace.name) ?				\
		__bad_type_size() :					\
187
188
		#type, #name, offsetof(typeof(trace), name),		\
		sizeof(trace.name), is_signed_type(type)
189

190
191
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
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);
}

251
static int syscall_enter_define_fields(struct ftrace_event_call *call)
252
253
{
	struct syscall_trace_enter trace;
254
	struct syscall_metadata *meta = call->data;
255
256
257
258
	int ret;
	int i;
	int offset = offsetof(typeof(trace), args);

259
260
261
262
	ret = trace_define_field(call, SYSCALL_FIELD(int, nr), FILTER_OTHER);
	if (ret)
		return ret;

263
	for (i = 0; i < meta->nb_args; i++) {
264
265
		ret = trace_define_field(call, meta->types[i],
					 meta->args[i], offset,
266
267
					 sizeof(unsigned long), 0,
					 FILTER_OTHER);
268
269
270
271
272
273
		offset += sizeof(unsigned long);
	}

	return ret;
}

274
static int syscall_exit_define_fields(struct ftrace_event_call *call)
275
276
277
278
{
	struct syscall_trace_exit trace;
	int ret;

279
280
281
282
	ret = trace_define_field(call, SYSCALL_FIELD(int, nr), FILTER_OTHER);
	if (ret)
		return ret;

283
	ret = trace_define_field(call, SYSCALL_FIELD(long, ret),
284
				 FILTER_OTHER);
285
286
287
288

	return ret;
}

289
void ftrace_syscall_enter(void *ignore, struct pt_regs *regs, long id)
290
{
291
292
293
	struct syscall_trace_enter *entry;
	struct syscall_metadata *sys_data;
	struct ring_buffer_event *event;
294
	struct ring_buffer *buffer;
295
	int size;
296
297
298
	int syscall_nr;

	syscall_nr = syscall_get_nr(current, regs);
299
300
	if (syscall_nr < 0)
		return;
301
302
	if (!test_bit(syscall_nr, enabled_enter_syscalls))
		return;
303

304
305
306
307
308
309
	sys_data = syscall_nr_to_meta(syscall_nr);
	if (!sys_data)
		return;

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

310
	event = trace_current_buffer_lock_reserve(&buffer,
311
			sys_data->enter_event->event.type, size, 0, 0);
312
313
314
315
316
317
318
	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);

319
320
321
	if (!filter_current_check_discard(buffer, sys_data->enter_event,
					  entry, event))
		trace_current_buffer_unlock_commit(buffer, event, 0, 0);
322
323
}

324
void ftrace_syscall_exit(void *ignore, struct pt_regs *regs, long ret)
325
{
326
327
328
	struct syscall_trace_exit *entry;
	struct syscall_metadata *sys_data;
	struct ring_buffer_event *event;
329
	struct ring_buffer *buffer;
330
331
332
	int syscall_nr;

	syscall_nr = syscall_get_nr(current, regs);
333
334
	if (syscall_nr < 0)
		return;
335
336
	if (!test_bit(syscall_nr, enabled_exit_syscalls))
		return;
337

338
339
340
341
	sys_data = syscall_nr_to_meta(syscall_nr);
	if (!sys_data)
		return;

342
	event = trace_current_buffer_lock_reserve(&buffer,
343
			sys_data->exit_event->event.type, sizeof(*entry), 0, 0);
344
345
346
347
348
349
350
	if (!event)
		return;

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

351
352
353
	if (!filter_current_check_discard(buffer, sys_data->exit_event,
					  entry, event))
		trace_current_buffer_unlock_commit(buffer, event, 0, 0);
354
355
}

356
int reg_event_syscall_enter(struct ftrace_event_call *call)
357
{
358
359
360
	int ret = 0;
	int num;

361
	num = ((struct syscall_metadata *)call->data)->syscall_nr;
362
	if (num < 0 || num >= NR_syscalls)
363
364
365
		return -ENOSYS;
	mutex_lock(&syscall_trace_lock);
	if (!sys_refcount_enter)
366
		ret = register_trace_sys_enter(ftrace_syscall_enter, NULL);
367
	if (!ret) {
368
369
370
371
372
		set_bit(num, enabled_enter_syscalls);
		sys_refcount_enter++;
	}
	mutex_unlock(&syscall_trace_lock);
	return ret;
373
374
}

375
void unreg_event_syscall_enter(struct ftrace_event_call *call)
376
{
377
	int num;
378

379
	num = ((struct syscall_metadata *)call->data)->syscall_nr;
380
	if (num < 0 || num >= NR_syscalls)
381
382
383
384
385
		return;
	mutex_lock(&syscall_trace_lock);
	sys_refcount_enter--;
	clear_bit(num, enabled_enter_syscalls);
	if (!sys_refcount_enter)
386
		unregister_trace_sys_enter(ftrace_syscall_enter, NULL);
387
388
	mutex_unlock(&syscall_trace_lock);
}
389

390
int reg_event_syscall_exit(struct ftrace_event_call *call)
391
{
392
393
394
	int ret = 0;
	int num;

395
	num = ((struct syscall_metadata *)call->data)->syscall_nr;
396
	if (num < 0 || num >= NR_syscalls)
397
398
399
		return -ENOSYS;
	mutex_lock(&syscall_trace_lock);
	if (!sys_refcount_exit)
400
		ret = register_trace_sys_exit(ftrace_syscall_exit, NULL);
401
	if (!ret) {
402
403
		set_bit(num, enabled_exit_syscalls);
		sys_refcount_exit++;
404
	}
405
406
407
	mutex_unlock(&syscall_trace_lock);
	return ret;
}
408

409
void unreg_event_syscall_exit(struct ftrace_event_call *call)
410
411
{
	int num;
412

413
	num = ((struct syscall_metadata *)call->data)->syscall_nr;
414
	if (num < 0 || num >= NR_syscalls)
415
416
417
418
419
		return;
	mutex_lock(&syscall_trace_lock);
	sys_refcount_exit--;
	clear_bit(num, enabled_exit_syscalls);
	if (!sys_refcount_exit)
420
		unregister_trace_sys_exit(ftrace_syscall_exit, NULL);
421
	mutex_unlock(&syscall_trace_lock);
422
}
423

424
425
426
427
int init_syscall_trace(struct ftrace_event_call *call)
{
	int id;

428
429
430
	if (set_syscall_print_fmt(call) < 0)
		return -ENOMEM;

431
432
433
	id = trace_event_raw_init(call);

	if (id < 0) {
434
		free_syscall_print_fmt(call);
435
		return id;
436
	}
437
438

	return id;
439
440
}

441
442
443
444
445
unsigned long __init arch_syscall_addr(int nr)
{
	return (unsigned long)sys_call_table[nr];
}

446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
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);
462
463
464
465
		if (!meta)
			continue;

		meta->syscall_nr = i;
466
467
468
469
470
471
472
		syscalls_metadata[i] = meta;
	}

	return 0;
}
core_initcall(init_ftrace_syscalls);

473
#ifdef CONFIG_PERF_EVENTS
474

475
476
477
478
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;
479

480
static void perf_syscall_enter(void *ignore, struct pt_regs *regs, long id)
481
482
{
	struct syscall_metadata *sys_data;
483
	struct syscall_trace_enter *rec;
484
	struct hlist_head *head;
485
	int syscall_nr;
486
	int rctx;
487
	int size;
488
489

	syscall_nr = syscall_get_nr(current, regs);
490
	if (!test_bit(syscall_nr, enabled_perf_enter_syscalls))
491
492
493
494
495
496
		return;

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

497
498
499
500
501
	/* 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);

502
503
	if (WARN_ONCE(size > PERF_MAX_TRACE_SIZE,
		      "perf buffer not large enough"))
504
505
		return;

506
	rec = (struct syscall_trace_enter *)perf_trace_buf_prepare(size,
507
				sys_data->enter_event->event.type, regs, &rctx);
508
509
	if (!rec)
		return;
510
511
512
513

	rec->nr = syscall_nr;
	syscall_get_arguments(current, regs, 0, sys_data->nb_args,
			       (unsigned long *)&rec->args);
514

515
	head = this_cpu_ptr(sys_data->enter_event->perf_events);
516
	perf_trace_buf_submit(rec, size, rctx, 0, 1, regs, head);
517
518
}

519
int perf_sysenter_enable(struct ftrace_event_call *call)
520
521
522
523
{
	int ret = 0;
	int num;

524
	num = ((struct syscall_metadata *)call->data)->syscall_nr;
525
526

	mutex_lock(&syscall_trace_lock);
527
	if (!sys_perf_refcount_enter)
528
		ret = register_trace_sys_enter(perf_syscall_enter, NULL);
529
530
531
532
	if (ret) {
		pr_info("event trace: Could not activate"
				"syscall entry trace point");
	} else {
533
534
		set_bit(num, enabled_perf_enter_syscalls);
		sys_perf_refcount_enter++;
535
536
537
538
539
	}
	mutex_unlock(&syscall_trace_lock);
	return ret;
}

540
void perf_sysenter_disable(struct ftrace_event_call *call)
541
542
543
{
	int num;

544
	num = ((struct syscall_metadata *)call->data)->syscall_nr;
545
546

	mutex_lock(&syscall_trace_lock);
547
548
549
	sys_perf_refcount_enter--;
	clear_bit(num, enabled_perf_enter_syscalls);
	if (!sys_perf_refcount_enter)
550
		unregister_trace_sys_enter(perf_syscall_enter, NULL);
551
552
553
	mutex_unlock(&syscall_trace_lock);
}

554
static void perf_syscall_exit(void *ignore, struct pt_regs *regs, long ret)
555
556
{
	struct syscall_metadata *sys_data;
557
	struct syscall_trace_exit *rec;
558
	struct hlist_head *head;
559
	int syscall_nr;
560
	int rctx;
561
	int size;
562
563

	syscall_nr = syscall_get_nr(current, regs);
564
	if (!test_bit(syscall_nr, enabled_perf_exit_syscalls))
565
566
567
568
569
570
		return;

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

571
572
573
	/* We can probably do that at build time */
	size = ALIGN(sizeof(*rec) + sizeof(u32), sizeof(u64));
	size -= sizeof(u32);
574

575
576
577
578
	/*
	 * Impossible, but be paranoid with the future
	 * How to put this check outside runtime?
	 */
579
580
	if (WARN_ONCE(size > PERF_MAX_TRACE_SIZE,
		"exit event has grown above perf buffer size"))
581
582
		return;

583
	rec = (struct syscall_trace_exit *)perf_trace_buf_prepare(size,
584
				sys_data->exit_event->event.type, regs, &rctx);
585
586
	if (!rec)
		return;
587
588
589
590

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

591
	head = this_cpu_ptr(sys_data->exit_event->perf_events);
592
	perf_trace_buf_submit(rec, size, rctx, 0, 1, regs, head);
593
594
}

595
int perf_sysexit_enable(struct ftrace_event_call *call)
596
597
598
599
{
	int ret = 0;
	int num;

600
	num = ((struct syscall_metadata *)call->data)->syscall_nr;
601
602

	mutex_lock(&syscall_trace_lock);
603
	if (!sys_perf_refcount_exit)
604
		ret = register_trace_sys_exit(perf_syscall_exit, NULL);
605
606
	if (ret) {
		pr_info("event trace: Could not activate"
607
				"syscall exit trace point");
608
	} else {
609
610
		set_bit(num, enabled_perf_exit_syscalls);
		sys_perf_refcount_exit++;
611
612
613
614
615
	}
	mutex_unlock(&syscall_trace_lock);
	return ret;
}

616
void perf_sysexit_disable(struct ftrace_event_call *call)
617
618
619
{
	int num;

620
	num = ((struct syscall_metadata *)call->data)->syscall_nr;
621
622

	mutex_lock(&syscall_trace_lock);
623
624
625
	sys_perf_refcount_exit--;
	clear_bit(num, enabled_perf_exit_syscalls);
	if (!sys_perf_refcount_exit)
626
		unregister_trace_sys_exit(perf_syscall_exit, NULL);
627
628
629
	mutex_unlock(&syscall_trace_lock);
}

630
#endif /* CONFIG_PERF_EVENTS */
631

632
633
634
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
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;
}