trace_syscalls.c 13.2 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
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;

26
	trace = (typeof(trace))ent;
27 28
	syscall = trace->nr;
	entry = syscall_nr_to_meta(syscall);
29

30 31 32
	if (!entry)
		goto end;

33 34 35 36 37
	if (entry->enter_id != ent->type) {
		WARN_ON_ONCE(1);
		goto end;
	}

38 39 40 41 42 43
	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 */
44
		if (trace_flags & TRACE_ITER_VERBOSE) {
45 46 47 48 49
			ret = trace_seq_printf(s, "%s ", entry->types[i]);
			if (!ret)
				return TRACE_TYPE_PARTIAL_LINE;
		}
		/* parameter values */
50
		ret = trace_seq_printf(s, "%s: %lx%s", entry->args[i],
51
				       trace->args[i],
52
				       i == entry->nb_args - 1 ? "" : ", ");
53 54 55 56
		if (!ret)
			return TRACE_TYPE_PARTIAL_LINE;
	}

57 58 59 60
	ret = trace_seq_putc(s, ')');
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

61
end:
62 63 64 65
	ret =  trace_seq_putc(s, '\n');
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

66 67 68 69 70 71 72 73 74 75 76 77 78
	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;

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

83 84 85 86 87
	if (!entry) {
		trace_seq_printf(s, "\n");
		return TRACE_TYPE_HANDLED;
	}

88 89 90 91 92
	if (entry->exit_id != ent->type) {
		WARN_ON_ONCE(1);
		return TRACE_TYPE_UNHANDLED;
	}

93 94 95 96 97 98 99 100
	ret = trace_seq_printf(s, "%s -> 0x%lx\n", entry->name,
				trace->ret);
	if (!ret)
		return TRACE_TYPE_PARTIAL_LINE;

	return TRACE_TYPE_HANDLED;
}

101 102 103 104 105 106 107
extern char *__bad_type_size(void);

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

108
int syscall_enter_format(struct ftrace_event_call *call, struct trace_seq *s)
109 110 111
{
	int i;
	int nr;
112
	int ret;
113
	struct syscall_metadata *entry;
114 115
	struct syscall_trace_enter trace;
	int offset = offsetof(struct syscall_trace_enter, args);
116

117
	nr = syscall_name_to_nr(call->data);
118 119 120
	entry = syscall_nr_to_meta(nr);

	if (!entry)
121 122 123 124 125 126
		return 0;

	ret = trace_seq_printf(s, "\tfield:%s %s;\toffset:%zu;\tsize:%zu;\n",
			       SYSCALL_FIELD(int, nr));
	if (!ret)
		return 0;
127 128 129 130 131 132

	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;
133
		ret = trace_seq_printf(s, "\toffset:%d;\tsize:%zu;\n", offset,
134 135 136 137 138 139
				       sizeof(unsigned long));
		if (!ret)
			return 0;
		offset += sizeof(unsigned long);
	}

140
	trace_seq_puts(s, "\nprint fmt: \"");
141
	for (i = 0; i < entry->nb_args; i++) {
142
		ret = trace_seq_printf(s, "%s: 0x%%0%zulx%s", entry->args[i],
143
				        sizeof(unsigned long),
144
					i == entry->nb_args - 1 ? "" : ", ");
145 146 147
		if (!ret)
			return 0;
	}
148
	trace_seq_putc(s, '"');
149 150

	for (i = 0; i < entry->nb_args; i++) {
151 152
		ret = trace_seq_printf(s, ", ((unsigned long)(REC->%s))",
				       entry->args[i]);
153 154 155 156
		if (!ret)
			return 0;
	}

157
	return trace_seq_putc(s, '\n');
158 159
}

160 161 162 163 164 165 166 167 168
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,
			       "\tfield:%s %s;\toffset:%zu;\tsize:%zu;\n"
			       "\tfield:%s %s;\toffset:%zu;\tsize:%zu;\n",
			       SYSCALL_FIELD(int, nr),
169
			       SYSCALL_FIELD(long, ret));
170 171 172 173 174 175
	if (!ret)
		return 0;

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

176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197
int syscall_enter_define_fields(struct ftrace_event_call *call)
{
	struct syscall_trace_enter trace;
	struct syscall_metadata *meta;
	int ret;
	int nr;
	int i;
	int offset = offsetof(typeof(trace), args);

	nr = syscall_name_to_nr(call->data);
	meta = syscall_nr_to_meta(nr);

	if (!meta)
		return 0;

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

	for (i = 0; i < meta->nb_args; i++) {
		ret = trace_define_field(call, meta->types[i],
					 meta->args[i], offset,
198 199
					 sizeof(unsigned long), 0,
					 FILTER_OTHER);
200 201 202 203 204 205 206 207 208 209 210 211 212 213 214
		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;

215
	ret = trace_define_field(call, SYSCALL_FIELD(long, ret), 0,
216
				 FILTER_OTHER);
217 218 219 220

	return ret;
}

221
void ftrace_syscall_enter(struct pt_regs *regs, long id)
222
{
223 224 225
	struct syscall_trace_enter *entry;
	struct syscall_metadata *sys_data;
	struct ring_buffer_event *event;
226
	struct ring_buffer *buffer;
227
	int size;
228 229 230
	int syscall_nr;

	syscall_nr = syscall_get_nr(current, regs);
231 232
	if (syscall_nr < 0)
		return;
233 234
	if (!test_bit(syscall_nr, enabled_enter_syscalls))
		return;
235

236 237 238 239 240 241
	sys_data = syscall_nr_to_meta(syscall_nr);
	if (!sys_data)
		return;

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

242 243
	event = trace_current_buffer_lock_reserve(&buffer, sys_data->enter_id,
						  size, 0, 0);
244 245 246 247 248 249 250
	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);

251 252 253
	if (!filter_current_check_discard(buffer, sys_data->enter_event,
					  entry, event))
		trace_current_buffer_unlock_commit(buffer, event, 0, 0);
254 255
}

256
void ftrace_syscall_exit(struct pt_regs *regs, long ret)
257
{
258 259 260
	struct syscall_trace_exit *entry;
	struct syscall_metadata *sys_data;
	struct ring_buffer_event *event;
261
	struct ring_buffer *buffer;
262 263 264
	int syscall_nr;

	syscall_nr = syscall_get_nr(current, regs);
265 266
	if (syscall_nr < 0)
		return;
267 268
	if (!test_bit(syscall_nr, enabled_exit_syscalls))
		return;
269

270 271 272 273
	sys_data = syscall_nr_to_meta(syscall_nr);
	if (!sys_data)
		return;

274
	event = trace_current_buffer_lock_reserve(&buffer, sys_data->exit_id,
275 276 277 278 279 280 281 282
				sizeof(*entry), 0, 0);
	if (!event)
		return;

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

283 284 285
	if (!filter_current_check_discard(buffer, sys_data->exit_event,
					  entry, event))
		trace_current_buffer_unlock_commit(buffer, event, 0, 0);
286 287
}

288
int reg_event_syscall_enter(void *ptr)
289
{
290 291 292 293 294 295
	int ret = 0;
	int num;
	char *name;

	name = (char *)ptr;
	num = syscall_name_to_nr(name);
296
	if (num < 0 || num >= NR_syscalls)
297 298 299
		return -ENOSYS;
	mutex_lock(&syscall_trace_lock);
	if (!sys_refcount_enter)
300
		ret = register_trace_sys_enter(ftrace_syscall_enter);
301 302 303 304 305 306 307 308 309
	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;
310 311
}

312
void unreg_event_syscall_enter(void *ptr)
313
{
314 315
	int num;
	char *name;
316

317 318
	name = (char *)ptr;
	num = syscall_name_to_nr(name);
319
	if (num < 0 || num >= NR_syscalls)
320 321 322 323 324
		return;
	mutex_lock(&syscall_trace_lock);
	sys_refcount_enter--;
	clear_bit(num, enabled_enter_syscalls);
	if (!sys_refcount_enter)
325
		unregister_trace_sys_enter(ftrace_syscall_enter);
326 327
	mutex_unlock(&syscall_trace_lock);
}
328

329
int reg_event_syscall_exit(void *ptr)
330
{
331 332 333 334 335 336
	int ret = 0;
	int num;
	char *name;

	name = (char *)ptr;
	num = syscall_name_to_nr(name);
337
	if (num < 0 || num >= NR_syscalls)
338 339 340
		return -ENOSYS;
	mutex_lock(&syscall_trace_lock);
	if (!sys_refcount_exit)
341
		ret = register_trace_sys_exit(ftrace_syscall_exit);
342 343 344 345 346 347
	if (ret) {
		pr_info("event trace: Could not activate"
				"syscall exit trace point");
	} else {
		set_bit(num, enabled_exit_syscalls);
		sys_refcount_exit++;
348
	}
349 350 351
	mutex_unlock(&syscall_trace_lock);
	return ret;
}
352

353 354 355 356
void unreg_event_syscall_exit(void *ptr)
{
	int num;
	char *name;
357

358 359
	name = (char *)ptr;
	num = syscall_name_to_nr(name);
360
	if (num < 0 || num >= NR_syscalls)
361 362 363 364 365
		return;
	mutex_lock(&syscall_trace_lock);
	sys_refcount_exit--;
	clear_bit(num, enabled_exit_syscalls);
	if (!sys_refcount_exit)
366
		unregister_trace_sys_exit(ftrace_syscall_exit);
367
	mutex_unlock(&syscall_trace_lock);
368
}
369 370 371 372 373 374 375 376

struct trace_event event_syscall_enter = {
	.trace			= print_syscall_enter,
};

struct trace_event event_syscall_exit = {
	.trace			= print_syscall_exit,
};
377 378

#ifdef CONFIG_EVENT_PROFILE
379

380 381
static DECLARE_BITMAP(enabled_prof_enter_syscalls, NR_syscalls);
static DECLARE_BITMAP(enabled_prof_exit_syscalls, NR_syscalls);
382 383 384 385 386 387
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;
388 389 390
	struct syscall_trace_enter *rec;
	unsigned long flags;
	char *raw_data;
391
	int syscall_nr;
392
	int size;
393
	int cpu;
394 395 396 397 398 399 400 401 402

	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;

403 404 405 406 407
	/* 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);

408 409 410 411 412 413
	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);
414

415 416 417 418 419 420 421 422 423
	cpu = smp_processor_id();

	if (in_nmi())
		raw_data = rcu_dereference(trace_profile_buf_nmi);
	else
		raw_data = rcu_dereference(trace_profile_buf);

	if (!raw_data)
		goto end;
424

425 426 427 428 429 430 431 432 433 434 435
	raw_data = per_cpu_ptr(raw_data, cpu);

	/* 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);
	rec->ent.type = sys_data->enter_id;
	rec->nr = syscall_nr;
	syscall_get_arguments(current, regs, 0, sys_data->nb_args,
			       (unsigned long *)&rec->args);
436
	perf_tp_event(sys_data->enter_id, 0, 1, rec, size);
437 438 439

end:
	local_irq_restore(flags);
440 441 442 443 444 445 446 447
}

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

	num = syscall_name_to_nr(name);
448
	if (num < 0 || num >= NR_syscalls)
449 450 451 452
		return -ENOSYS;

	mutex_lock(&syscall_trace_lock);
	if (!sys_prof_refcount_enter)
453
		ret = register_trace_sys_enter(prof_syscall_enter);
454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469
	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);
470
	if (num < 0 || num >= NR_syscalls)
471 472 473 474 475 476
		return;

	mutex_lock(&syscall_trace_lock);
	sys_prof_refcount_enter--;
	clear_bit(num, enabled_prof_enter_syscalls);
	if (!sys_prof_refcount_enter)
477
		unregister_trace_sys_enter(prof_syscall_enter);
478 479 480 481 482 483
	mutex_unlock(&syscall_trace_lock);
}

static void prof_syscall_exit(struct pt_regs *regs, long ret)
{
	struct syscall_metadata *sys_data;
484 485
	struct syscall_trace_exit *rec;
	unsigned long flags;
486
	int syscall_nr;
487 488 489
	char *raw_data;
	int size;
	int cpu;
490 491 492 493 494 495 496 497 498

	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;

499 500 501
	/* We can probably do that at build time */
	size = ALIGN(sizeof(*rec) + sizeof(u32), sizeof(u64));
	size -= sizeof(u32);
502

503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534
	/*
	 * 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);
	cpu = smp_processor_id();

	if (in_nmi())
		raw_data = rcu_dereference(trace_profile_buf_nmi);
	else
		raw_data = rcu_dereference(trace_profile_buf);

	if (!raw_data)
		goto end;

	raw_data = per_cpu_ptr(raw_data, cpu);

	/* 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);
	rec->ent.type = sys_data->exit_id;
	rec->nr = syscall_nr;
	rec->ret = syscall_get_return_value(current, regs);

535
	perf_tp_event(sys_data->exit_id, 0, 1, rec, size);
536 537 538

end:
	local_irq_restore(flags);
539 540 541 542 543 544 545 546
}

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

	num = syscall_name_to_nr(name);
547
	if (num < 0 || num >= NR_syscalls)
548 549 550 551
		return -ENOSYS;

	mutex_lock(&syscall_trace_lock);
	if (!sys_prof_refcount_exit)
552
		ret = register_trace_sys_exit(prof_syscall_exit);
553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568
	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);
569
	if (num < 0 || num >= NR_syscalls)
570 571 572 573 574 575
		return;

	mutex_lock(&syscall_trace_lock);
	sys_prof_refcount_exit--;
	clear_bit(num, enabled_prof_exit_syscalls);
	if (!sys_prof_refcount_exit)
576
		unregister_trace_sys_exit(prof_syscall_exit);
577 578 579 580 581 582
	mutex_unlock(&syscall_trace_lock);
}

#endif