session.c 51.1 KB
Newer Older
1
#include <linux/kernel.h>
2
#include <traceevent/event-parse.h>
3

4
#include <byteswap.h>
5 6
#include <unistd.h>
#include <sys/types.h>
7
#include <sys/mman.h>
8

9 10
#include "evlist.h"
#include "evsel.h"
11
#include "session.h"
12
#include "tool.h"
13
#include "sort.h"
14
#include "util.h"
15
#include "cpumap.h"
16
#include "perf_regs.h"
17
#include "asm/bug.h"
18
#include "auxtrace.h"
19

20 21 22 23 24
static int perf_session__deliver_event(struct perf_session *session,
				       union perf_event *event,
				       struct perf_sample *sample,
				       struct perf_tool *tool,
				       u64 file_offset);
25

26
static int perf_session__open(struct perf_session *session)
27
{
28
	struct perf_data_file *file = session->file;
29

30
	if (perf_session__read_header(session) < 0) {
31
		pr_err("incompatible file format (rerun with -v to learn more)");
32
		return -1;
33 34
	}

35 36 37
	if (perf_data_file__is_pipe(file))
		return 0;

38
	if (!perf_evlist__valid_sample_type(session->evlist)) {
39
		pr_err("non matching sample_type");
40
		return -1;
41 42
	}

43
	if (!perf_evlist__valid_sample_id_all(session->evlist)) {
44
		pr_err("non matching sample_id_all");
45
		return -1;
46 47
	}

48
	if (!perf_evlist__valid_read_format(session->evlist)) {
49
		pr_err("non matching read_format");
50
		return -1;
51 52
	}

53 54 55
	return 0;
}

56
void perf_session__set_id_hdr_size(struct perf_session *session)
57
{
58 59 60
	u16 id_hdr_size = perf_evlist__id_hdr_size(session->evlist);

	machines__set_id_hdr_size(&session->machines, id_hdr_size);
61 62
}

63
int perf_session__create_kernel_maps(struct perf_session *session)
64
{
65
	int ret = machine__create_kernel_maps(&session->machines.host);
66 67

	if (ret >= 0)
68
		ret = machines__create_guest_kernel_maps(&session->machines);
69 70 71
	return ret;
}

72
static void perf_session__destroy_kernel_maps(struct perf_session *session)
73
{
74
	machines__destroy_kernel_maps(&session->machines);
75 76
}

77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95
static bool perf_session__has_comm_exec(struct perf_session *session)
{
	struct perf_evsel *evsel;

	evlist__for_each(session->evlist, evsel) {
		if (evsel->attr.comm_exec)
			return true;
	}

	return false;
}

static void perf_session__set_comm_exec(struct perf_session *session)
{
	bool comm_exec = perf_session__has_comm_exec(session);

	machines__set_comm_exec(&session->machines, comm_exec);
}

96
static int ordered_events__deliver_event(struct ordered_events *oe,
97
					 struct ordered_event *event)
98
{
99 100 101 102 103 104 105 106 107 108
	struct perf_sample sample;
	struct perf_session *session = container_of(oe, struct perf_session,
						    ordered_events);
	int ret = perf_evlist__parse_sample(session->evlist, event->event, &sample);

	if (ret) {
		pr_err("Can't parse sample, err = %d\n", ret);
		return ret;
	}

109 110
	return perf_session__deliver_event(session, event->event, &sample,
					   session->tool, event->file_offset);
111 112
}

113 114
struct perf_session *perf_session__new(struct perf_data_file *file,
				       bool repipe, struct perf_tool *tool)
115
{
116
	struct perf_session *session = zalloc(sizeof(*session));
117

118
	if (!session)
119 120
		goto out;

121
	session->repipe = repipe;
122
	session->tool   = tool;
123
	INIT_LIST_HEAD(&session->auxtrace_index);
124
	machines__init(&session->machines);
125
	ordered_events__init(&session->ordered_events, ordered_events__deliver_event);
126

127 128
	if (file) {
		if (perf_data_file__open(file))
129
			goto out_delete;
130

131
		session->file = file;
132 133

		if (perf_data_file__is_read(file)) {
134
			if (perf_session__open(session) < 0)
135 136
				goto out_close;

137
			perf_session__set_id_hdr_size(session);
138
			perf_session__set_comm_exec(session);
139 140 141 142
		}
	}

	if (!file || perf_data_file__is_write(file)) {
143 144
		/*
		 * In O_RDONLY mode this will be performed when reading the
145
		 * kernel MMAP event, in perf_event__process_mmap().
146
		 */
147
		if (perf_session__create_kernel_maps(session) < 0)
148
			pr_warning("Cannot read kernel map\n");
149
	}
150

151
	if (tool && tool->ordering_requires_timestamps &&
152
	    tool->ordered_events && !perf_evlist__sample_id_all(session->evlist)) {
153
		dump_printf("WARNING: No sample_id_all support, falling back to unordered processing\n");
154
		tool->ordered_events = false;
155
	}
156

157
	return session;
158 159 160 161

 out_close:
	perf_data_file__close(file);
 out_delete:
162
	perf_session__delete(session);
163
 out:
164
	return NULL;
165 166
}

167 168
static void perf_session__delete_threads(struct perf_session *session)
{
169
	machine__delete_threads(&session->machines.host);
170 171
}

172 173
static void perf_session_env__delete(struct perf_session_env *env)
{
174 175 176 177 178 179 180 181 182 183 184 185
	zfree(&env->hostname);
	zfree(&env->os_release);
	zfree(&env->version);
	zfree(&env->arch);
	zfree(&env->cpu_desc);
	zfree(&env->cpuid);

	zfree(&env->cmdline);
	zfree(&env->sibling_cores);
	zfree(&env->sibling_threads);
	zfree(&env->numa_nodes);
	zfree(&env->pmu_mappings);
186 187
}

188
void perf_session__delete(struct perf_session *session)
189
{
190
	auxtrace__free(session);
191
	auxtrace_index__free(&session->auxtrace_index);
192 193 194 195 196 197 198
	perf_session__destroy_kernel_maps(session);
	perf_session__delete_threads(session);
	perf_session_env__delete(&session->header.env);
	machines__exit(&session->machines);
	if (session->file)
		perf_data_file__close(session->file);
	free(session);
199
}
200

201 202 203
static int process_event_synth_tracing_data_stub(struct perf_tool *tool
						 __maybe_unused,
						 union perf_event *event
204 205 206
						 __maybe_unused,
						 struct perf_session *session
						__maybe_unused)
207 208 209 210 211
{
	dump_printf(": unhandled!\n");
	return 0;
}

212 213
static int process_event_synth_attr_stub(struct perf_tool *tool __maybe_unused,
					 union perf_event *event __maybe_unused,
214 215
					 struct perf_evlist **pevlist
					 __maybe_unused)
216 217 218 219 220
{
	dump_printf(": unhandled!\n");
	return 0;
}

221 222 223 224 225
static int process_event_sample_stub(struct perf_tool *tool __maybe_unused,
				     union perf_event *event __maybe_unused,
				     struct perf_sample *sample __maybe_unused,
				     struct perf_evsel *evsel __maybe_unused,
				     struct machine *machine __maybe_unused)
226 227 228 229 230
{
	dump_printf(": unhandled!\n");
	return 0;
}

231 232 233 234
static int process_event_stub(struct perf_tool *tool __maybe_unused,
			      union perf_event *event __maybe_unused,
			      struct perf_sample *sample __maybe_unused,
			      struct machine *machine __maybe_unused)
235 236 237 238 239
{
	dump_printf(": unhandled!\n");
	return 0;
}

240 241 242 243 244 245 246 247
static int process_build_id_stub(struct perf_tool *tool __maybe_unused,
				 union perf_event *event __maybe_unused,
				 struct perf_session *session __maybe_unused)
{
	dump_printf(": unhandled!\n");
	return 0;
}

248 249
static int process_finished_round_stub(struct perf_tool *tool __maybe_unused,
				       union perf_event *event __maybe_unused,
250
				       struct ordered_events *oe __maybe_unused)
251 252 253 254 255
{
	dump_printf(": unhandled!\n");
	return 0;
}

256
static int process_finished_round(struct perf_tool *tool,
257
				  union perf_event *event,
258
				  struct ordered_events *oe);
259

A
Adrian Hunter 已提交
260 261 262 263 264 265 266 267 268
static int process_id_index_stub(struct perf_tool *tool __maybe_unused,
				 union perf_event *event __maybe_unused,
				 struct perf_session *perf_session
				 __maybe_unused)
{
	dump_printf(": unhandled!\n");
	return 0;
}

269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302
static int process_event_auxtrace_info_stub(struct perf_tool *tool __maybe_unused,
				union perf_event *event __maybe_unused,
				struct perf_session *session __maybe_unused)
{
	dump_printf(": unhandled!\n");
	return 0;
}

static int skipn(int fd, off_t n)
{
	char buf[4096];
	ssize_t ret;

	while (n > 0) {
		ret = read(fd, buf, min(n, (off_t)sizeof(buf)));
		if (ret <= 0)
			return ret;
		n -= ret;
	}

	return 0;
}

static s64 process_event_auxtrace_stub(struct perf_tool *tool __maybe_unused,
				       union perf_event *event,
				       struct perf_session *session
				       __maybe_unused)
{
	dump_printf(": unhandled!\n");
	if (perf_data_file__is_pipe(session->file))
		skipn(perf_data_file__fd(session->file), event->auxtrace.size);
	return event->auxtrace.size;
}

303 304 305 306 307 308 309 310 311
static
int process_event_auxtrace_error_stub(struct perf_tool *tool __maybe_unused,
				      union perf_event *event __maybe_unused,
				      struct perf_session *session __maybe_unused)
{
	dump_printf(": unhandled!\n");
	return 0;
}

312
void perf_tool__fill_defaults(struct perf_tool *tool)
313
{
314 315 316 317
	if (tool->sample == NULL)
		tool->sample = process_event_sample_stub;
	if (tool->mmap == NULL)
		tool->mmap = process_event_stub;
318 319
	if (tool->mmap2 == NULL)
		tool->mmap2 = process_event_stub;
320 321 322 323 324 325 326 327
	if (tool->comm == NULL)
		tool->comm = process_event_stub;
	if (tool->fork == NULL)
		tool->fork = process_event_stub;
	if (tool->exit == NULL)
		tool->exit = process_event_stub;
	if (tool->lost == NULL)
		tool->lost = perf_event__process_lost;
328 329
	if (tool->aux == NULL)
		tool->aux = perf_event__process_aux;
330 331 332 333 334 335 336 337 338 339 340
	if (tool->read == NULL)
		tool->read = process_event_sample_stub;
	if (tool->throttle == NULL)
		tool->throttle = process_event_stub;
	if (tool->unthrottle == NULL)
		tool->unthrottle = process_event_stub;
	if (tool->attr == NULL)
		tool->attr = process_event_synth_attr_stub;
	if (tool->tracing_data == NULL)
		tool->tracing_data = process_event_synth_tracing_data_stub;
	if (tool->build_id == NULL)
341
		tool->build_id = process_build_id_stub;
342
	if (tool->finished_round == NULL) {
343
		if (tool->ordered_events)
344
			tool->finished_round = process_finished_round;
345
		else
346
			tool->finished_round = process_finished_round_stub;
347
	}
A
Adrian Hunter 已提交
348 349
	if (tool->id_index == NULL)
		tool->id_index = process_id_index_stub;
350 351 352 353
	if (tool->auxtrace_info == NULL)
		tool->auxtrace_info = process_event_auxtrace_info_stub;
	if (tool->auxtrace == NULL)
		tool->auxtrace = process_event_auxtrace_stub;
354 355
	if (tool->auxtrace_error == NULL)
		tool->auxtrace_error = process_event_auxtrace_error_stub;
356
}
357

358 359 360 361 362 363 364 365 366 367
static void swap_sample_id_all(union perf_event *event, void *data)
{
	void *end = (void *) event + event->header.size;
	int size = end - data;

	BUG_ON(size % sizeof(u64));
	mem_bswap_64(data, size);
}

static void perf_event__all64_swap(union perf_event *event,
368
				   bool sample_id_all __maybe_unused)
369
{
370 371
	struct perf_event_header *hdr = &event->header;
	mem_bswap_64(hdr + 1, event->header.size - sizeof(*hdr));
372 373
}

374
static void perf_event__comm_swap(union perf_event *event, bool sample_id_all)
375
{
376 377
	event->comm.pid = bswap_32(event->comm.pid);
	event->comm.tid = bswap_32(event->comm.tid);
378 379 380 381

	if (sample_id_all) {
		void *data = &event->comm.comm;

382
		data += PERF_ALIGN(strlen(data) + 1, sizeof(u64));
383 384
		swap_sample_id_all(event, data);
	}
385 386
}

387 388
static void perf_event__mmap_swap(union perf_event *event,
				  bool sample_id_all)
389
{
390 391 392 393 394
	event->mmap.pid	  = bswap_32(event->mmap.pid);
	event->mmap.tid	  = bswap_32(event->mmap.tid);
	event->mmap.start = bswap_64(event->mmap.start);
	event->mmap.len	  = bswap_64(event->mmap.len);
	event->mmap.pgoff = bswap_64(event->mmap.pgoff);
395 396 397 398

	if (sample_id_all) {
		void *data = &event->mmap.filename;

399
		data += PERF_ALIGN(strlen(data) + 1, sizeof(u64));
400 401
		swap_sample_id_all(event, data);
	}
402 403
}

404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422
static void perf_event__mmap2_swap(union perf_event *event,
				  bool sample_id_all)
{
	event->mmap2.pid   = bswap_32(event->mmap2.pid);
	event->mmap2.tid   = bswap_32(event->mmap2.tid);
	event->mmap2.start = bswap_64(event->mmap2.start);
	event->mmap2.len   = bswap_64(event->mmap2.len);
	event->mmap2.pgoff = bswap_64(event->mmap2.pgoff);
	event->mmap2.maj   = bswap_32(event->mmap2.maj);
	event->mmap2.min   = bswap_32(event->mmap2.min);
	event->mmap2.ino   = bswap_64(event->mmap2.ino);

	if (sample_id_all) {
		void *data = &event->mmap2.filename;

		data += PERF_ALIGN(strlen(data) + 1, sizeof(u64));
		swap_sample_id_all(event, data);
	}
}
423
static void perf_event__task_swap(union perf_event *event, bool sample_id_all)
424
{
425 426 427 428 429
	event->fork.pid	 = bswap_32(event->fork.pid);
	event->fork.tid	 = bswap_32(event->fork.tid);
	event->fork.ppid = bswap_32(event->fork.ppid);
	event->fork.ptid = bswap_32(event->fork.ptid);
	event->fork.time = bswap_64(event->fork.time);
430 431 432

	if (sample_id_all)
		swap_sample_id_all(event, &event->fork + 1);
433 434
}

435
static void perf_event__read_swap(union perf_event *event, bool sample_id_all)
436
{
437 438 439 440 441 442
	event->read.pid		 = bswap_32(event->read.pid);
	event->read.tid		 = bswap_32(event->read.tid);
	event->read.value	 = bswap_64(event->read.value);
	event->read.time_enabled = bswap_64(event->read.time_enabled);
	event->read.time_running = bswap_64(event->read.time_running);
	event->read.id		 = bswap_64(event->read.id);
443 444 445

	if (sample_id_all)
		swap_sample_id_all(event, &event->read + 1);
446 447
}

448 449 450 451 452 453 454 455 456 457
static void perf_event__aux_swap(union perf_event *event, bool sample_id_all)
{
	event->aux.aux_offset = bswap_64(event->aux.aux_offset);
	event->aux.aux_size   = bswap_64(event->aux.aux_size);
	event->aux.flags      = bswap_64(event->aux.flags);

	if (sample_id_all)
		swap_sample_id_all(event, &event->aux + 1);
}

458 459 460 461 462 463 464 465 466 467 468
static void perf_event__throttle_swap(union perf_event *event,
				      bool sample_id_all)
{
	event->throttle.time	  = bswap_64(event->throttle.time);
	event->throttle.id	  = bswap_64(event->throttle.id);
	event->throttle.stream_id = bswap_64(event->throttle.stream_id);

	if (sample_id_all)
		swap_sample_id_all(event, &event->throttle + 1);
}

469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500
static u8 revbyte(u8 b)
{
	int rev = (b >> 4) | ((b & 0xf) << 4);
	rev = ((rev & 0xcc) >> 2) | ((rev & 0x33) << 2);
	rev = ((rev & 0xaa) >> 1) | ((rev & 0x55) << 1);
	return (u8) rev;
}

/*
 * XXX this is hack in attempt to carry flags bitfield
 * throught endian village. ABI says:
 *
 * Bit-fields are allocated from right to left (least to most significant)
 * on little-endian implementations and from left to right (most to least
 * significant) on big-endian implementations.
 *
 * The above seems to be byte specific, so we need to reverse each
 * byte of the bitfield. 'Internet' also says this might be implementation
 * specific and we probably need proper fix and carry perf_event_attr
 * bitfield flags in separate data file FEAT_ section. Thought this seems
 * to work for now.
 */
static void swap_bitfield(u8 *p, unsigned len)
{
	unsigned i;

	for (i = 0; i < len; i++) {
		*p = revbyte(*p);
		p++;
	}
}

501 502 503 504 505 506 507 508 509 510 511 512 513
/* exported for swapping attributes in file header */
void perf_event__attr_swap(struct perf_event_attr *attr)
{
	attr->type		= bswap_32(attr->type);
	attr->size		= bswap_32(attr->size);
	attr->config		= bswap_64(attr->config);
	attr->sample_period	= bswap_64(attr->sample_period);
	attr->sample_type	= bswap_64(attr->sample_type);
	attr->read_format	= bswap_64(attr->read_format);
	attr->wakeup_events	= bswap_32(attr->wakeup_events);
	attr->bp_type		= bswap_32(attr->bp_type);
	attr->bp_addr		= bswap_64(attr->bp_addr);
	attr->bp_len		= bswap_64(attr->bp_len);
514 515 516
	attr->branch_sample_type = bswap_64(attr->branch_sample_type);
	attr->sample_regs_user	 = bswap_64(attr->sample_regs_user);
	attr->sample_stack_user  = bswap_32(attr->sample_stack_user);
517
	attr->aux_watermark	 = bswap_32(attr->aux_watermark);
518 519

	swap_bitfield((u8 *) (&attr->read_format + 1), sizeof(u64));
520 521
}

522
static void perf_event__hdr_attr_swap(union perf_event *event,
523
				      bool sample_id_all __maybe_unused)
524 525 526
{
	size_t size;

527
	perf_event__attr_swap(&event->attr.attr);
528

529 530 531
	size = event->header.size;
	size -= (void *)&event->attr.id - (void *)event;
	mem_bswap_64(event->attr.id, size);
532 533
}

534
static void perf_event__event_type_swap(union perf_event *event,
535
					bool sample_id_all __maybe_unused)
536
{
537 538
	event->event_type.event_type.event_id =
		bswap_64(event->event_type.event_type.event_id);
539 540
}

541
static void perf_event__tracing_data_swap(union perf_event *event,
542
					  bool sample_id_all __maybe_unused)
543
{
544
	event->tracing_data.size = bswap_32(event->tracing_data.size);
545 546
}

547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569
static void perf_event__auxtrace_info_swap(union perf_event *event,
					   bool sample_id_all __maybe_unused)
{
	size_t size;

	event->auxtrace_info.type = bswap_32(event->auxtrace_info.type);

	size = event->header.size;
	size -= (void *)&event->auxtrace_info.priv - (void *)event;
	mem_bswap_64(event->auxtrace_info.priv, size);
}

static void perf_event__auxtrace_swap(union perf_event *event,
				      bool sample_id_all __maybe_unused)
{
	event->auxtrace.size      = bswap_64(event->auxtrace.size);
	event->auxtrace.offset    = bswap_64(event->auxtrace.offset);
	event->auxtrace.reference = bswap_64(event->auxtrace.reference);
	event->auxtrace.idx       = bswap_32(event->auxtrace.idx);
	event->auxtrace.tid       = bswap_32(event->auxtrace.tid);
	event->auxtrace.cpu       = bswap_32(event->auxtrace.cpu);
}

570 571 572 573 574 575 576 577 578 579 580
static void perf_event__auxtrace_error_swap(union perf_event *event,
					    bool sample_id_all __maybe_unused)
{
	event->auxtrace_error.type = bswap_32(event->auxtrace_error.type);
	event->auxtrace_error.code = bswap_32(event->auxtrace_error.code);
	event->auxtrace_error.cpu  = bswap_32(event->auxtrace_error.cpu);
	event->auxtrace_error.pid  = bswap_32(event->auxtrace_error.pid);
	event->auxtrace_error.tid  = bswap_32(event->auxtrace_error.tid);
	event->auxtrace_error.ip   = bswap_64(event->auxtrace_error.ip);
}

581 582
typedef void (*perf_event__swap_op)(union perf_event *event,
				    bool sample_id_all);
583

584 585
static perf_event__swap_op perf_event__swap_ops[] = {
	[PERF_RECORD_MMAP]		  = perf_event__mmap_swap,
586
	[PERF_RECORD_MMAP2]		  = perf_event__mmap2_swap,
587 588 589 590 591
	[PERF_RECORD_COMM]		  = perf_event__comm_swap,
	[PERF_RECORD_FORK]		  = perf_event__task_swap,
	[PERF_RECORD_EXIT]		  = perf_event__task_swap,
	[PERF_RECORD_LOST]		  = perf_event__all64_swap,
	[PERF_RECORD_READ]		  = perf_event__read_swap,
592 593
	[PERF_RECORD_THROTTLE]		  = perf_event__throttle_swap,
	[PERF_RECORD_UNTHROTTLE]	  = perf_event__throttle_swap,
594
	[PERF_RECORD_SAMPLE]		  = perf_event__all64_swap,
595
	[PERF_RECORD_AUX]		  = perf_event__aux_swap,
596
	[PERF_RECORD_HEADER_ATTR]	  = perf_event__hdr_attr_swap,
597 598 599
	[PERF_RECORD_HEADER_EVENT_TYPE]	  = perf_event__event_type_swap,
	[PERF_RECORD_HEADER_TRACING_DATA] = perf_event__tracing_data_swap,
	[PERF_RECORD_HEADER_BUILD_ID]	  = NULL,
A
Adrian Hunter 已提交
600
	[PERF_RECORD_ID_INDEX]		  = perf_event__all64_swap,
601 602
	[PERF_RECORD_AUXTRACE_INFO]	  = perf_event__auxtrace_info_swap,
	[PERF_RECORD_AUXTRACE]		  = perf_event__auxtrace_swap,
603
	[PERF_RECORD_AUXTRACE_ERROR]	  = perf_event__auxtrace_error_swap,
604
	[PERF_RECORD_HEADER_MAX]	  = NULL,
605 606
};

607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645
/*
 * When perf record finishes a pass on every buffers, it records this pseudo
 * event.
 * We record the max timestamp t found in the pass n.
 * Assuming these timestamps are monotonic across cpus, we know that if
 * a buffer still has events with timestamps below t, they will be all
 * available and then read in the pass n + 1.
 * Hence when we start to read the pass n + 2, we can safely flush every
 * events with timestamps below t.
 *
 *    ============ PASS n =================
 *       CPU 0         |   CPU 1
 *                     |
 *    cnt1 timestamps  |   cnt2 timestamps
 *          1          |         2
 *          2          |         3
 *          -          |         4  <--- max recorded
 *
 *    ============ PASS n + 1 ==============
 *       CPU 0         |   CPU 1
 *                     |
 *    cnt1 timestamps  |   cnt2 timestamps
 *          3          |         5
 *          4          |         6
 *          5          |         7 <---- max recorded
 *
 *      Flush every events below timestamp 4
 *
 *    ============ PASS n + 2 ==============
 *       CPU 0         |   CPU 1
 *                     |
 *    cnt1 timestamps  |   cnt2 timestamps
 *          6          |         8
 *          7          |         9
 *          -          |         10
 *
 *      Flush every events below timestamp 7
 *      etc...
 */
646
static int process_finished_round(struct perf_tool *tool __maybe_unused,
647
				  union perf_event *event __maybe_unused,
648
				  struct ordered_events *oe)
649
{
650
	return ordered_events__flush(oe, OE_FLUSH__ROUND);
651 652
}

653 654
int perf_session__queue_event(struct perf_session *s, union perf_event *event,
			      struct perf_sample *sample, u64 file_offset)
655
{
656
	return ordered_events__queue(&s->ordered_events, event, sample, file_offset);
657
}
658

K
Kan Liang 已提交
659
static void callchain__lbr_callstack_printf(struct perf_sample *sample)
660
{
K
Kan Liang 已提交
661 662 663
	struct ip_callchain *callchain = sample->callchain;
	struct branch_stack *lbr_stack = sample->branch_stack;
	u64 kernel_callchain_nr = callchain->nr;
664
	unsigned int i;
665

K
Kan Liang 已提交
666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715
	for (i = 0; i < kernel_callchain_nr; i++) {
		if (callchain->ips[i] == PERF_CONTEXT_USER)
			break;
	}

	if ((i != kernel_callchain_nr) && lbr_stack->nr) {
		u64 total_nr;
		/*
		 * LBR callstack can only get user call chain,
		 * i is kernel call chain number,
		 * 1 is PERF_CONTEXT_USER.
		 *
		 * The user call chain is stored in LBR registers.
		 * LBR are pair registers. The caller is stored
		 * in "from" register, while the callee is stored
		 * in "to" register.
		 * For example, there is a call stack
		 * "A"->"B"->"C"->"D".
		 * The LBR registers will recorde like
		 * "C"->"D", "B"->"C", "A"->"B".
		 * So only the first "to" register and all "from"
		 * registers are needed to construct the whole stack.
		 */
		total_nr = i + 1 + lbr_stack->nr + 1;
		kernel_callchain_nr = i + 1;

		printf("... LBR call chain: nr:%" PRIu64 "\n", total_nr);

		for (i = 0; i < kernel_callchain_nr; i++)
			printf("..... %2d: %016" PRIx64 "\n",
			       i, callchain->ips[i]);

		printf("..... %2d: %016" PRIx64 "\n",
		       (int)(kernel_callchain_nr), lbr_stack->entries[0].to);
		for (i = 0; i < lbr_stack->nr; i++)
			printf("..... %2d: %016" PRIx64 "\n",
			       (int)(i + kernel_callchain_nr + 1), lbr_stack->entries[i].from);
	}
}

static void callchain__printf(struct perf_evsel *evsel,
			      struct perf_sample *sample)
{
	unsigned int i;
	struct ip_callchain *callchain = sample->callchain;

	if (has_branch_callstack(evsel))
		callchain__lbr_callstack_printf(sample);

	printf("... FP chain: nr:%" PRIu64 "\n", callchain->nr);
716

K
Kan Liang 已提交
717
	for (i = 0; i < callchain->nr; i++)
718
		printf("..... %2d: %016" PRIx64 "\n",
K
Kan Liang 已提交
719
		       i, callchain->ips[i]);
720 721
}

722 723 724 725 726 727 728 729 730 731 732 733
static void branch_stack__printf(struct perf_sample *sample)
{
	uint64_t i;

	printf("... branch stack: nr:%" PRIu64 "\n", sample->branch_stack->nr);

	for (i = 0; i < sample->branch_stack->nr; i++)
		printf("..... %2"PRIu64": %016" PRIx64 " -> %016" PRIx64 "\n",
			i, sample->branch_stack->entries[i].from,
			sample->branch_stack->entries[i].to);
}

734 735 736 737 738 739 740 741 742 743 744 745
static void regs_dump__printf(u64 mask, u64 *regs)
{
	unsigned rid, i = 0;

	for_each_set_bit(rid, (unsigned long *) &mask, sizeof(mask) * 8) {
		u64 val = regs[i++];

		printf(".... %-5s 0x%" PRIx64 "\n",
		       perf_reg_name(rid), val);
	}
}

746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771
static const char *regs_abi[] = {
	[PERF_SAMPLE_REGS_ABI_NONE] = "none",
	[PERF_SAMPLE_REGS_ABI_32] = "32-bit",
	[PERF_SAMPLE_REGS_ABI_64] = "64-bit",
};

static inline const char *regs_dump_abi(struct regs_dump *d)
{
	if (d->abi > PERF_SAMPLE_REGS_ABI_64)
		return "unknown";

	return regs_abi[d->abi];
}

static void regs__printf(const char *type, struct regs_dump *regs)
{
	u64 mask = regs->mask;

	printf("... %s regs: mask 0x%" PRIx64 " ABI %s\n",
	       type,
	       mask,
	       regs_dump_abi(regs));

	regs_dump__printf(mask, regs->regs);
}

772
static void regs_user__printf(struct perf_sample *sample)
773 774 775
{
	struct regs_dump *user_regs = &sample->user_regs;

776 777 778 779 780 781 782 783 784 785
	if (user_regs->regs)
		regs__printf("user", user_regs);
}

static void regs_intr__printf(struct perf_sample *sample)
{
	struct regs_dump *intr_regs = &sample->intr_regs;

	if (intr_regs->regs)
		regs__printf("intr", intr_regs);
786 787 788 789 790 791 792 793
}

static void stack_user__printf(struct stack_dump *dump)
{
	printf("... ustack: size %" PRIu64 ", offset 0x%x\n",
	       dump->size, dump->offset);
}

794
static void perf_evlist__print_tstamp(struct perf_evlist *evlist,
795
				       union perf_event *event,
796
				       struct perf_sample *sample)
797
{
798
	u64 sample_type = __perf_evlist__combined_sample_type(evlist);
799

800
	if (event->header.type != PERF_RECORD_SAMPLE &&
801
	    !perf_evlist__sample_id_all(evlist)) {
802 803 804 805
		fputs("-1 -1 ", stdout);
		return;
	}

806
	if ((sample_type & PERF_SAMPLE_CPU))
807 808
		printf("%u ", sample->cpu);

809
	if (sample_type & PERF_SAMPLE_TIME)
810
		printf("%" PRIu64 " ", sample->time);
811 812
}

813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842
static void sample_read__printf(struct perf_sample *sample, u64 read_format)
{
	printf("... sample_read:\n");

	if (read_format & PERF_FORMAT_TOTAL_TIME_ENABLED)
		printf("...... time enabled %016" PRIx64 "\n",
		       sample->read.time_enabled);

	if (read_format & PERF_FORMAT_TOTAL_TIME_RUNNING)
		printf("...... time running %016" PRIx64 "\n",
		       sample->read.time_running);

	if (read_format & PERF_FORMAT_GROUP) {
		u64 i;

		printf(".... group nr %" PRIu64 "\n", sample->read.group.nr);

		for (i = 0; i < sample->read.group.nr; i++) {
			struct sample_read_value *value;

			value = &sample->read.group.values[i];
			printf("..... id %016" PRIx64
			       ", value %016" PRIx64 "\n",
			       value->id, value->value);
		}
	} else
		printf("..... id %016" PRIx64 ", value %016" PRIx64 "\n",
			sample->read.one.id, sample->read.one.value);
}

843
static void dump_event(struct perf_evlist *evlist, union perf_event *event,
844
		       u64 file_offset, struct perf_sample *sample)
845 846 847 848
{
	if (!dump_trace)
		return;

849 850
	printf("\n%#" PRIx64 " [%#x]: event: %d\n",
	       file_offset, event->header.size, event->header.type);
851 852 853 854

	trace_event(event);

	if (sample)
855
		perf_evlist__print_tstamp(evlist, event, sample);
856

857
	printf("%#" PRIx64 " [%#x]: PERF_RECORD_%s", file_offset,
858
	       event->header.size, perf_event__name(event->header.type));
859 860
}

861
static void dump_sample(struct perf_evsel *evsel, union perf_event *event,
862
			struct perf_sample *sample)
863
{
864 865
	u64 sample_type;

866 867 868
	if (!dump_trace)
		return;

869
	printf("(IP, 0x%x): %d/%d: %#" PRIx64 " period: %" PRIu64 " addr: %#" PRIx64 "\n",
870
	       event->header.misc, sample->pid, sample->tid, sample->ip,
871
	       sample->period, sample->addr);
872

873
	sample_type = evsel->attr.sample_type;
874 875

	if (sample_type & PERF_SAMPLE_CALLCHAIN)
K
Kan Liang 已提交
876
		callchain__printf(evsel, sample);
877

K
Kan Liang 已提交
878
	if ((sample_type & PERF_SAMPLE_BRANCH_STACK) && !has_branch_callstack(evsel))
879
		branch_stack__printf(sample);
880 881

	if (sample_type & PERF_SAMPLE_REGS_USER)
882
		regs_user__printf(sample);
883

884 885 886
	if (sample_type & PERF_SAMPLE_REGS_INTR)
		regs_intr__printf(sample);

887 888
	if (sample_type & PERF_SAMPLE_STACK_USER)
		stack_user__printf(&sample->user_stack);
889 890 891

	if (sample_type & PERF_SAMPLE_WEIGHT)
		printf("... weight: %" PRIu64 "\n", sample->weight);
892 893 894

	if (sample_type & PERF_SAMPLE_DATA_SRC)
		printf(" . data_src: 0x%"PRIx64"\n", sample->data_src);
895

896 897 898
	if (sample_type & PERF_SAMPLE_TRANSACTION)
		printf("... transaction: %" PRIx64 "\n", sample->transaction);

899 900
	if (sample_type & PERF_SAMPLE_READ)
		sample_read__printf(sample, evsel->attr.read_format);
901 902
}

903
static struct machine *machines__find_for_cpumode(struct machines *machines,
904 905
					       union perf_event *event,
					       struct perf_sample *sample)
906 907
{
	const u8 cpumode = event->header.misc & PERF_RECORD_MISC_CPUMODE_MASK;
908
	struct machine *machine;
909

910 911 912
	if (perf_guest &&
	    ((cpumode == PERF_RECORD_MISC_GUEST_KERNEL) ||
	     (cpumode == PERF_RECORD_MISC_GUEST_USER))) {
913 914
		u32 pid;

915 916
		if (event->header.type == PERF_RECORD_MMAP
		    || event->header.type == PERF_RECORD_MMAP2)
917 918
			pid = event->mmap.pid;
		else
919
			pid = sample->pid;
920

921
		machine = machines__find(machines, pid);
922
		if (!machine)
923
			machine = machines__find(machines, DEFAULT_GUEST_KERNEL_ID);
924
		return machine;
925
	}
926

927
	return &machines->host;
928 929
}

930
static int deliver_sample_value(struct perf_evlist *evlist,
931 932 933 934 935 936
				struct perf_tool *tool,
				union perf_event *event,
				struct perf_sample *sample,
				struct sample_read_value *v,
				struct machine *machine)
{
937
	struct perf_sample_id *sid = perf_evlist__id2sid(evlist, v->id);
938 939 940 941 942 943 944 945

	if (sid) {
		sample->id     = v->id;
		sample->period = v->value - sid->period;
		sid->period    = v->value;
	}

	if (!sid || sid->evsel == NULL) {
946
		++evlist->stats.nr_unknown_id;
947 948 949 950 951 952
		return 0;
	}

	return tool->sample(tool, event, sample, sid->evsel, machine);
}

953
static int deliver_sample_group(struct perf_evlist *evlist,
954 955 956 957 958 959 960 961 962
				struct perf_tool *tool,
				union  perf_event *event,
				struct perf_sample *sample,
				struct machine *machine)
{
	int ret = -EINVAL;
	u64 i;

	for (i = 0; i < sample->read.group.nr; i++) {
963
		ret = deliver_sample_value(evlist, tool, event, sample,
964 965 966 967 968 969 970 971 972 973
					   &sample->read.group.values[i],
					   machine);
		if (ret)
			break;
	}

	return ret;
}

static int
974
 perf_evlist__deliver_sample(struct perf_evlist *evlist,
975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990
			     struct perf_tool *tool,
			     union  perf_event *event,
			     struct perf_sample *sample,
			     struct perf_evsel *evsel,
			     struct machine *machine)
{
	/* We know evsel != NULL. */
	u64 sample_type = evsel->attr.sample_type;
	u64 read_format = evsel->attr.read_format;

	/* Standard sample delievery. */
	if (!(sample_type & PERF_SAMPLE_READ))
		return tool->sample(tool, event, sample, evsel, machine);

	/* For PERF_SAMPLE_READ we have either single or group mode. */
	if (read_format & PERF_FORMAT_GROUP)
991
		return deliver_sample_group(evlist, tool, event, sample,
992 993
					    machine);
	else
994
		return deliver_sample_value(evlist, tool, event, sample,
995 996 997
					    &sample->read.one, machine);
}

998 999 1000 1001 1002
static int machines__deliver_event(struct machines *machines,
				   struct perf_evlist *evlist,
				   union perf_event *event,
				   struct perf_sample *sample,
				   struct perf_tool *tool, u64 file_offset)
1003
{
1004
	struct perf_evsel *evsel;
1005
	struct machine *machine;
1006

1007
	dump_event(evlist, event, file_offset, sample);
1008

1009
	evsel = perf_evlist__id2evsel(evlist, sample->id);
1010

1011
	machine = machines__find_for_cpumode(machines, event, sample);
1012

1013 1014
	switch (event->header.type) {
	case PERF_RECORD_SAMPLE:
1015
		dump_sample(evsel, event, sample);
1016
		if (evsel == NULL) {
1017
			++evlist->stats.nr_unknown_id;
1018
			return 0;
1019
		}
1020
		if (machine == NULL) {
1021
			++evlist->stats.nr_unprocessable_samples;
1022
			return 0;
1023
		}
1024
		return perf_evlist__deliver_sample(evlist, tool, event, sample, evsel, machine);
1025
	case PERF_RECORD_MMAP:
1026
		return tool->mmap(tool, event, sample, machine);
1027 1028
	case PERF_RECORD_MMAP2:
		return tool->mmap2(tool, event, sample, machine);
1029
	case PERF_RECORD_COMM:
1030
		return tool->comm(tool, event, sample, machine);
1031
	case PERF_RECORD_FORK:
1032
		return tool->fork(tool, event, sample, machine);
1033
	case PERF_RECORD_EXIT:
1034
		return tool->exit(tool, event, sample, machine);
1035
	case PERF_RECORD_LOST:
1036
		if (tool->lost == perf_event__process_lost)
1037
			evlist->stats.total_lost += event->lost.lost;
1038
		return tool->lost(tool, event, sample, machine);
1039
	case PERF_RECORD_READ:
1040
		return tool->read(tool, event, sample, evsel, machine);
1041
	case PERF_RECORD_THROTTLE:
1042
		return tool->throttle(tool, event, sample, machine);
1043
	case PERF_RECORD_UNTHROTTLE:
1044
		return tool->unthrottle(tool, event, sample, machine);
1045 1046
	case PERF_RECORD_AUX:
		return tool->aux(tool, event, sample, machine);
1047
	default:
1048
		++evlist->stats.nr_unknown_events;
1049 1050 1051 1052
		return -1;
	}
}

1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070
static int perf_session__deliver_event(struct perf_session *session,
				       union perf_event *event,
				       struct perf_sample *sample,
				       struct perf_tool *tool,
				       u64 file_offset)
{
	int ret;

	ret = auxtrace__process_event(session, event, sample, tool);
	if (ret < 0)
		return ret;
	if (ret > 0)
		return 0;

	return machines__deliver_event(&session->machines, session->evlist,
				       event, sample, tool, file_offset);
}

1071 1072 1073
static s64 perf_session__process_user_event(struct perf_session *session,
					    union perf_event *event,
					    u64 file_offset)
1074
{
1075
	struct ordered_events *oe = &session->ordered_events;
1076
	struct perf_tool *tool = session->tool;
1077
	int fd = perf_data_file__fd(session->file);
1078 1079
	int err;

1080
	dump_event(session->evlist, event, file_offset, NULL);
1081

1082
	/* These events are processed right away */
1083
	switch (event->header.type) {
1084
	case PERF_RECORD_HEADER_ATTR:
1085
		err = tool->attr(tool, event, &session->evlist);
1086
		if (err == 0) {
1087
			perf_session__set_id_hdr_size(session);
1088 1089
			perf_session__set_comm_exec(session);
		}
1090
		return err;
1091 1092 1093 1094 1095 1096
	case PERF_RECORD_HEADER_EVENT_TYPE:
		/*
		 * Depreceated, but we need to handle it for sake
		 * of old data files create in pipe mode.
		 */
		return 0;
1097 1098
	case PERF_RECORD_HEADER_TRACING_DATA:
		/* setup for reading amidst mmap */
1099
		lseek(fd, file_offset, SEEK_SET);
1100
		return tool->tracing_data(tool, event, session);
1101
	case PERF_RECORD_HEADER_BUILD_ID:
1102
		return tool->build_id(tool, event, session);
1103
	case PERF_RECORD_FINISHED_ROUND:
1104
		return tool->finished_round(tool, event, oe);
A
Adrian Hunter 已提交
1105 1106
	case PERF_RECORD_ID_INDEX:
		return tool->id_index(tool, event, session);
1107 1108 1109 1110 1111 1112
	case PERF_RECORD_AUXTRACE_INFO:
		return tool->auxtrace_info(tool, event, session);
	case PERF_RECORD_AUXTRACE:
		/* setup for reading amidst mmap */
		lseek(fd, file_offset + event->header.size, SEEK_SET);
		return tool->auxtrace(tool, event, session);
1113
	case PERF_RECORD_AUXTRACE_ERROR:
1114
		perf_session__auxtrace_error_inc(session, event);
1115
		return tool->auxtrace_error(tool, event, session);
1116
	default:
1117
		return -EINVAL;
1118
	}
1119 1120
}

1121 1122
int perf_session__deliver_synth_event(struct perf_session *session,
				      union perf_event *event,
1123
				      struct perf_sample *sample)
1124
{
1125
	struct perf_evlist *evlist = session->evlist;
1126
	struct perf_tool *tool = session->tool;
1127 1128

	events_stats__inc(&evlist->stats, event->header.type);
1129 1130

	if (event->header.type >= PERF_RECORD_USER_TYPE_START)
1131
		return perf_session__process_user_event(session, event, 0);
1132

1133
	return machines__deliver_event(&session->machines, evlist, event, sample, tool, 0);
1134 1135
}

1136 1137 1138 1139 1140 1141 1142 1143 1144
static void event_swap(union perf_event *event, bool sample_id_all)
{
	perf_event__swap_op swap;

	swap = perf_event__swap_ops[event->header.type];
	if (swap)
		swap(event, sample_id_all);
}

1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199
int perf_session__peek_event(struct perf_session *session, off_t file_offset,
			     void *buf, size_t buf_sz,
			     union perf_event **event_ptr,
			     struct perf_sample *sample)
{
	union perf_event *event;
	size_t hdr_sz, rest;
	int fd;

	if (session->one_mmap && !session->header.needs_swap) {
		event = file_offset - session->one_mmap_offset +
			session->one_mmap_addr;
		goto out_parse_sample;
	}

	if (perf_data_file__is_pipe(session->file))
		return -1;

	fd = perf_data_file__fd(session->file);
	hdr_sz = sizeof(struct perf_event_header);

	if (buf_sz < hdr_sz)
		return -1;

	if (lseek(fd, file_offset, SEEK_SET) == (off_t)-1 ||
	    readn(fd, &buf, hdr_sz) != (ssize_t)hdr_sz)
		return -1;

	event = (union perf_event *)buf;

	if (session->header.needs_swap)
		perf_event_header__bswap(&event->header);

	if (event->header.size < hdr_sz)
		return -1;

	rest = event->header.size - hdr_sz;

	if (readn(fd, &buf, rest) != (ssize_t)rest)
		return -1;

	if (session->header.needs_swap)
		event_swap(event, perf_evlist__sample_id_all(session->evlist));

out_parse_sample:

	if (sample && event->header.type < PERF_RECORD_USER_TYPE_START &&
	    perf_evlist__parse_sample(session->evlist, event, sample))
		return -1;

	*event_ptr = event;

	return 0;
}

1200
static s64 perf_session__process_event(struct perf_session *session,
1201
				       union perf_event *event, u64 file_offset)
1202
{
1203
	struct perf_evlist *evlist = session->evlist;
1204
	struct perf_tool *tool = session->tool;
1205
	struct perf_sample sample;
1206 1207
	int ret;

1208
	if (session->header.needs_swap)
1209
		event_swap(event, perf_evlist__sample_id_all(evlist));
1210 1211 1212 1213

	if (event->header.type >= PERF_RECORD_HEADER_MAX)
		return -EINVAL;

1214
	events_stats__inc(&evlist->stats, event->header.type);
1215 1216

	if (event->header.type >= PERF_RECORD_USER_TYPE_START)
1217
		return perf_session__process_user_event(session, event, file_offset);
1218

1219 1220 1221
	/*
	 * For all kernel events we get the sample data
	 */
1222
	ret = perf_evlist__parse_sample(evlist, event, &sample);
1223 1224
	if (ret)
		return ret;
1225

1226
	if (tool->ordered_events) {
1227
		ret = perf_session__queue_event(session, event, &sample, file_offset);
1228 1229 1230 1231
		if (ret != -ETIME)
			return ret;
	}

1232 1233
	return perf_session__deliver_event(session, event, &sample, tool,
					   file_offset);
1234 1235
}

1236
void perf_event_header__bswap(struct perf_event_header *hdr)
1237
{
1238 1239 1240
	hdr->type = bswap_32(hdr->type);
	hdr->misc = bswap_16(hdr->misc);
	hdr->size = bswap_16(hdr->size);
1241 1242
}

1243 1244
struct thread *perf_session__findnew(struct perf_session *session, pid_t pid)
{
1245
	return machine__findnew_thread(&session->machines.host, -1, pid);
1246 1247
}

1248
static struct thread *perf_session__register_idle_thread(struct perf_session *session)
1249
{
1250
	struct thread *thread;
1251

1252
	thread = machine__findnew_thread(&session->machines.host, 0, 0);
1253
	if (thread == NULL || thread__set_comm(thread, "swapper", 0)) {
1254 1255 1256 1257 1258 1259 1260
		pr_err("problem inserting idle task.\n");
		thread = NULL;
	}

	return thread;
}

1261
static void perf_session__warn_about_errors(const struct perf_session *session)
1262
{
1263 1264 1265 1266
	const struct events_stats *stats = &session->evlist->stats;
	const struct ordered_events *oe = &session->ordered_events;

	if (session->tool->lost == perf_event__process_lost &&
1267
	    stats->nr_events[PERF_RECORD_LOST] != 0) {
1268 1269
		ui__warning("Processed %d events and lost %d chunks!\n\n"
			    "Check IO/CPU overload!\n\n",
1270 1271
			    stats->nr_events[0],
			    stats->nr_events[PERF_RECORD_LOST]);
1272 1273
	}

1274
	if (stats->nr_unknown_events != 0) {
1275 1276 1277 1278 1279
		ui__warning("Found %u unknown events!\n\n"
			    "Is this an older tool processing a perf.data "
			    "file generated by a more recent tool?\n\n"
			    "If that is not the case, consider "
			    "reporting to linux-kernel@vger.kernel.org.\n\n",
1280
			    stats->nr_unknown_events);
1281 1282
	}

1283
	if (stats->nr_unknown_id != 0) {
1284
		ui__warning("%u samples with id not present in the header\n",
1285
			    stats->nr_unknown_id);
1286 1287
	}

1288
	if (stats->nr_invalid_chains != 0) {
1289 1290 1291
		ui__warning("Found invalid callchains!\n\n"
			    "%u out of %u events were discarded for this reason.\n\n"
			    "Consider reporting to linux-kernel@vger.kernel.org.\n\n",
1292 1293
			    stats->nr_invalid_chains,
			    stats->nr_events[PERF_RECORD_SAMPLE]);
1294
	}
1295

1296
	if (stats->nr_unprocessable_samples != 0) {
1297 1298
		ui__warning("%u unprocessable samples recorded.\n"
			    "Do you have a KVM guest running and not using 'perf kvm'?\n",
1299
			    stats->nr_unprocessable_samples);
1300
	}
1301

1302 1303
	if (oe->nr_unordered_events != 0)
		ui__warning("%u out of order events recorded.\n", oe->nr_unordered_events);
1304 1305

	events_stats__auxtrace_error_warn(stats);
1306 1307
}

1308 1309
volatile int session_done;

1310
static int __perf_session__process_pipe_events(struct perf_session *session)
1311
{
1312
	struct ordered_events *oe = &session->ordered_events;
1313
	struct perf_tool *tool = session->tool;
1314
	int fd = perf_data_file__fd(session->file);
1315 1316 1317
	union perf_event *event;
	uint32_t size, cur_size = 0;
	void *buf = NULL;
1318
	s64 skip = 0;
1319
	u64 head;
1320
	ssize_t err;
1321 1322
	void *p;

1323
	perf_tool__fill_defaults(tool);
1324 1325

	head = 0;
1326 1327 1328 1329 1330
	cur_size = sizeof(union perf_event);

	buf = malloc(cur_size);
	if (!buf)
		return -errno;
1331
more:
1332
	event = buf;
1333
	err = readn(fd, event, sizeof(struct perf_event_header));
1334 1335 1336 1337 1338 1339 1340 1341
	if (err <= 0) {
		if (err == 0)
			goto done;

		pr_err("failed to read event header\n");
		goto out_err;
	}

1342
	if (session->header.needs_swap)
1343
		perf_event_header__bswap(&event->header);
1344

1345
	size = event->header.size;
1346 1347 1348 1349
	if (size < sizeof(struct perf_event_header)) {
		pr_err("bad event header size\n");
		goto out_err;
	}
1350

1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361
	if (size > cur_size) {
		void *new = realloc(buf, size);
		if (!new) {
			pr_err("failed to allocate memory to read event\n");
			goto out_err;
		}
		buf = new;
		cur_size = size;
		event = buf;
	}
	p = event;
1362 1363
	p += sizeof(struct perf_event_header);

1364
	if (size - sizeof(struct perf_event_header)) {
1365
		err = readn(fd, p, size - sizeof(struct perf_event_header));
1366 1367 1368 1369 1370
		if (err <= 0) {
			if (err == 0) {
				pr_err("unexpected end of event stream\n");
				goto done;
			}
1371

1372 1373 1374
			pr_err("failed to read event data\n");
			goto out_err;
		}
1375 1376
	}

1377
	if ((skip = perf_session__process_event(session, event, head)) < 0) {
1378
		pr_err("%#" PRIx64 " [%#x]: failed to process type: %d\n",
1379
		       head, event->header.size, event->header.type);
1380 1381
		err = -EINVAL;
		goto out_err;
1382 1383 1384 1385 1386 1387 1388 1389 1390 1391
	}

	head += size;

	if (skip > 0)
		head += skip;

	if (!session_done())
		goto more;
done:
1392
	/* do the final flush for ordered samples */
1393
	err = ordered_events__flush(oe, OE_FLUSH__FINAL);
1394 1395 1396
	if (err)
		goto out_err;
	err = auxtrace__flush_events(session, tool);
1397
out_err:
1398
	free(buf);
1399
	perf_session__warn_about_errors(session);
1400
	ordered_events__free(&session->ordered_events);
1401
	auxtrace__free_events(session);
1402 1403 1404
	return err;
}

1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422
static union perf_event *
fetch_mmaped_event(struct perf_session *session,
		   u64 head, size_t mmap_size, char *buf)
{
	union perf_event *event;

	/*
	 * Ensure we have enough space remaining to read
	 * the size of the event in the headers.
	 */
	if (head + sizeof(event->header) > mmap_size)
		return NULL;

	event = (union perf_event *)(buf + head);

	if (session->header.needs_swap)
		perf_event_header__bswap(&event->header);

1423 1424 1425 1426
	if (head + event->header.size > mmap_size) {
		/* We're not fetching the event so swap back again */
		if (session->header.needs_swap)
			perf_event_header__bswap(&event->header);
1427
		return NULL;
1428
	}
1429 1430 1431 1432

	return event;
}

1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444
/*
 * On 64bit we can mmap the data file in one go. No need for tiny mmap
 * slices. On 32bit we use 32MB.
 */
#if BITS_PER_LONG == 64
#define MMAP_SIZE ULLONG_MAX
#define NUM_MMAPS 1
#else
#define MMAP_SIZE (32 * 1024 * 1024ULL)
#define NUM_MMAPS 128
#endif

1445 1446
static int __perf_session__process_events(struct perf_session *session,
					  u64 data_offset, u64 data_size,
1447
					  u64 file_size)
1448
{
1449
	struct ordered_events *oe = &session->ordered_events;
1450
	struct perf_tool *tool = session->tool;
1451
	int fd = perf_data_file__fd(session->file);
1452
	u64 head, page_offset, file_offset, file_pos, size;
1453
	int err, mmap_prot, mmap_flags, map_idx = 0;
1454
	size_t	mmap_size;
1455
	char *buf, *mmaps[NUM_MMAPS];
1456
	union perf_event *event;
1457
	struct ui_progress prog;
1458
	s64 skip;
1459

1460
	perf_tool__fill_defaults(tool);
1461

1462 1463 1464
	page_offset = page_size * (data_offset / page_size);
	file_offset = page_offset;
	head = data_offset - page_offset;
1465

1466
	if (data_size && (data_offset + data_size < file_size))
1467 1468
		file_size = data_offset + data_size;

1469
	ui_progress__init(&prog, file_size, "Processing events...");
1470

1471
	mmap_size = MMAP_SIZE;
1472
	if (mmap_size > file_size) {
1473
		mmap_size = file_size;
1474 1475
		session->one_mmap = true;
	}
1476

1477 1478
	memset(mmaps, 0, sizeof(mmaps));

1479 1480 1481
	mmap_prot  = PROT_READ;
	mmap_flags = MAP_SHARED;

1482
	if (session->header.needs_swap) {
1483 1484 1485
		mmap_prot  |= PROT_WRITE;
		mmap_flags = MAP_PRIVATE;
	}
1486
remap:
1487
	buf = mmap(NULL, mmap_size, mmap_prot, mmap_flags, fd,
1488
		   file_offset);
1489 1490 1491 1492 1493
	if (buf == MAP_FAILED) {
		pr_err("failed to mmap file\n");
		err = -errno;
		goto out_err;
	}
1494 1495
	mmaps[map_idx] = buf;
	map_idx = (map_idx + 1) & (ARRAY_SIZE(mmaps) - 1);
1496
	file_pos = file_offset + head;
1497 1498 1499 1500
	if (session->one_mmap) {
		session->one_mmap_addr = buf;
		session->one_mmap_offset = file_offset;
	}
1501 1502

more:
1503 1504
	event = fetch_mmaped_event(session, head, mmap_size, buf);
	if (!event) {
1505 1506 1507 1508
		if (mmaps[map_idx]) {
			munmap(mmaps[map_idx], mmap_size);
			mmaps[map_idx] = NULL;
		}
1509

1510 1511 1512
		page_offset = page_size * (head / page_size);
		file_offset += page_offset;
		head -= page_offset;
1513 1514 1515 1516 1517
		goto remap;
	}

	size = event->header.size;

1518
	if (size < sizeof(struct perf_event_header) ||
1519
	    (skip = perf_session__process_event(session, event, file_pos)) < 0) {
1520 1521 1522 1523 1524
		pr_err("%#" PRIx64 " [%#x]: failed to process type: %d\n",
		       file_offset + head, event->header.size,
		       event->header.type);
		err = -EINVAL;
		goto out_err;
1525 1526
	}

1527 1528 1529
	if (skip)
		size += skip;

1530
	head += size;
1531
	file_pos += size;
1532

1533
	ui_progress__update(&prog, size);
1534

1535
	if (session_done())
1536
		goto out;
1537

1538
	if (file_pos < file_size)
1539
		goto more;
1540

1541
out:
1542
	/* do the final flush for ordered samples */
1543
	err = ordered_events__flush(oe, OE_FLUSH__FINAL);
1544 1545 1546
	if (err)
		goto out_err;
	err = auxtrace__flush_events(session, tool);
1547
out_err:
N
Namhyung Kim 已提交
1548
	ui_progress__finish();
1549
	perf_session__warn_about_errors(session);
1550
	ordered_events__free(&session->ordered_events);
1551
	auxtrace__free_events(session);
1552
	session->one_mmap = false;
1553 1554
	return err;
}
1555

1556
int perf_session__process_events(struct perf_session *session)
1557
{
1558
	u64 size = perf_data_file__size(session->file);
1559 1560
	int err;

1561
	if (perf_session__register_idle_thread(session) == NULL)
1562 1563
		return -ENOMEM;

1564 1565 1566
	if (!perf_data_file__is_pipe(session->file))
		err = __perf_session__process_events(session,
						     session->header.data_offset,
1567
						     session->header.data_size, size);
1568
	else
1569
		err = __perf_session__process_pipe_events(session);
1570

1571 1572 1573
	return err;
}

1574
bool perf_session__has_traces(struct perf_session *session, const char *msg)
1575
{
1576 1577
	struct perf_evsel *evsel;

1578
	evlist__for_each(session->evlist, evsel) {
1579 1580
		if (evsel->attr.type == PERF_TYPE_TRACEPOINT)
			return true;
1581 1582
	}

1583 1584
	pr_err("No trace sample to read. Did you call 'perf %s'?\n", msg);
	return false;
1585
}
1586

1587 1588
int maps__set_kallsyms_ref_reloc_sym(struct map **maps,
				     const char *symbol_name, u64 addr)
1589 1590
{
	char *bracket;
1591
	enum map_type i;
1592 1593 1594 1595 1596
	struct ref_reloc_sym *ref;

	ref = zalloc(sizeof(struct ref_reloc_sym));
	if (ref == NULL)
		return -ENOMEM;
1597

1598 1599 1600
	ref->name = strdup(symbol_name);
	if (ref->name == NULL) {
		free(ref);
1601
		return -ENOMEM;
1602
	}
1603

1604
	bracket = strchr(ref->name, ']');
1605 1606 1607
	if (bracket)
		*bracket = '\0';

1608
	ref->addr = addr;
1609 1610

	for (i = 0; i < MAP__NR_TYPES; ++i) {
1611
		struct kmap *kmap = map__kmap(maps[i]);
1612 1613 1614

		if (!kmap)
			continue;
1615
		kmap->ref_reloc_sym = ref;
1616 1617
	}

1618 1619
	return 0;
}
1620

1621
size_t perf_session__fprintf_dsos(struct perf_session *session, FILE *fp)
1622
{
1623
	return machines__fprintf_dsos(&session->machines, fp);
1624
}
1625

1626
size_t perf_session__fprintf_dsos_buildid(struct perf_session *session, FILE *fp,
1627
					  bool (skip)(struct dso *dso, int parm), int parm)
1628
{
1629
	return machines__fprintf_dsos_buildid(&session->machines, fp, skip, parm);
1630
}
1631 1632 1633

size_t perf_session__fprintf_nr_events(struct perf_session *session, FILE *fp)
{
1634 1635 1636 1637 1638 1639 1640
	size_t ret;
	const char *msg = "";

	if (perf_header__has_feat(&session->header, HEADER_AUXTRACE))
		msg = " (excludes AUX area (e.g. instruction trace) decoded / synthesized events)";

	ret = fprintf(fp, "Aggregated stats:%s\n", msg);
1641

1642
	ret += events_stats__fprintf(&session->evlist->stats, fp);
1643 1644
	return ret;
}
1645

1646 1647 1648 1649 1650 1651
size_t perf_session__fprintf(struct perf_session *session, FILE *fp)
{
	/*
	 * FIXME: Here we have to actually print all the machines in this
	 * session, not just the host...
	 */
1652
	return machine__fprintf(&session->machines.host, fp);
1653 1654
}

1655 1656 1657 1658 1659
struct perf_evsel *perf_session__find_first_evtype(struct perf_session *session,
					      unsigned int type)
{
	struct perf_evsel *pos;

1660
	evlist__for_each(session->evlist, pos) {
1661 1662 1663 1664 1665 1666
		if (pos->attr.type == type)
			return pos;
	}
	return NULL;
}

1667
void perf_evsel__print_ip(struct perf_evsel *evsel, struct perf_sample *sample,
1668
			  struct addr_location *al,
1669
			  unsigned int print_opts, unsigned int stack_depth)
1670 1671
{
	struct callchain_cursor_node *node;
1672 1673 1674 1675
	int print_ip = print_opts & PRINT_IP_OPT_IP;
	int print_sym = print_opts & PRINT_IP_OPT_SYM;
	int print_dso = print_opts & PRINT_IP_OPT_DSO;
	int print_symoffset = print_opts & PRINT_IP_OPT_SYMOFFSET;
1676
	int print_oneline = print_opts & PRINT_IP_OPT_ONELINE;
1677
	int print_srcline = print_opts & PRINT_IP_OPT_SRCLINE;
1678
	char s = print_oneline ? ' ' : '\t';
1679 1680

	if (symbol_conf.use_callchain && sample->callchain) {
1681
		struct addr_location node_al;
1682

1683 1684 1685
		if (thread__resolve_callchain(al->thread, evsel,
					      sample, NULL, NULL,
					      PERF_MAX_STACK_DEPTH) != 0) {
1686 1687 1688 1689
			if (verbose)
				error("Failed to resolve callchain. Skipping\n");
			return;
		}
1690
		callchain_cursor_commit(&callchain_cursor);
1691

1692 1693 1694
		if (print_symoffset)
			node_al = *al;

1695
		while (stack_depth) {
1696 1697
			u64 addr = 0;

1698
			node = callchain_cursor_current(&callchain_cursor);
1699 1700 1701
			if (!node)
				break;

1702 1703 1704
			if (node->sym && node->sym->ignore)
				goto next;

1705
			if (print_ip)
1706
				printf("%c%16" PRIx64, s, node->ip);
1707

1708 1709 1710
			if (node->map)
				addr = node->map->map_ip(node->map, node->ip);

1711
			if (print_sym) {
1712
				printf(" ");
1713
				if (print_symoffset) {
1714
					node_al.addr = addr;
1715 1716
					node_al.map  = node->map;
					symbol__fprintf_symname_offs(node->sym, &node_al, stdout);
1717 1718
				} else
					symbol__fprintf_symname(node->sym, stdout);
1719
			}
1720

1721
			if (print_dso) {
1722
				printf(" (");
1723
				map__fprintf_dsoname(node->map, stdout);
1724
				printf(")");
1725
			}
1726

1727 1728 1729 1730
			if (print_srcline)
				map__fprintf_srcline(node->map, addr, "\n  ",
						     stdout);

1731 1732
			if (!print_oneline)
				printf("\n");
1733

1734
			stack_depth--;
1735 1736
next:
			callchain_cursor_advance(&callchain_cursor);
1737 1738 1739
		}

	} else {
1740
		if (al->sym && al->sym->ignore)
1741 1742
			return;

1743 1744 1745
		if (print_ip)
			printf("%16" PRIx64, sample->ip);

1746
		if (print_sym) {
1747
			printf(" ");
1748
			if (print_symoffset)
1749
				symbol__fprintf_symname_offs(al->sym, al,
1750 1751
							     stdout);
			else
1752
				symbol__fprintf_symname(al->sym, stdout);
1753 1754 1755
		}

		if (print_dso) {
1756
			printf(" (");
1757
			map__fprintf_dsoname(al->map, stdout);
1758
			printf(")");
1759
		}
1760 1761 1762

		if (print_srcline)
			map__fprintf_srcline(al->map, al->addr, "\n  ", stdout);
1763 1764
	}
}
1765 1766 1767 1768

int perf_session__cpu_bitmap(struct perf_session *session,
			     const char *cpu_list, unsigned long *cpu_bitmap)
{
1769
	int i, err = -1;
1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786
	struct cpu_map *map;

	for (i = 0; i < PERF_TYPE_MAX; ++i) {
		struct perf_evsel *evsel;

		evsel = perf_session__find_first_evtype(session, i);
		if (!evsel)
			continue;

		if (!(evsel->attr.sample_type & PERF_SAMPLE_CPU)) {
			pr_err("File does not contain CPU events. "
			       "Remove -c option to proceed.\n");
			return -1;
		}
	}

	map = cpu_map__new(cpu_list);
1787 1788 1789 1790
	if (map == NULL) {
		pr_err("Invalid cpu_list\n");
		return -1;
	}
1791 1792 1793 1794 1795 1796 1797

	for (i = 0; i < map->nr; i++) {
		int cpu = map->map[i];

		if (cpu >= MAX_NR_CPUS) {
			pr_err("Requested CPU %d too large. "
			       "Consider raising MAX_NR_CPUS\n", cpu);
1798
			goto out_delete_map;
1799 1800 1801 1802 1803
		}

		set_bit(cpu, cpu_bitmap);
	}

1804 1805 1806 1807 1808
	err = 0;

out_delete_map:
	cpu_map__delete(map);
	return err;
1809
}
1810 1811 1812 1813 1814

void perf_session__fprintf_info(struct perf_session *session, FILE *fp,
				bool full)
{
	struct stat st;
1815
	int fd, ret;
1816 1817 1818 1819

	if (session == NULL || fp == NULL)
		return;

1820 1821
	fd = perf_data_file__fd(session->file);

1822
	ret = fstat(fd, &st);
1823 1824 1825 1826 1827 1828 1829 1830
	if (ret == -1)
		return;

	fprintf(fp, "# ========\n");
	fprintf(fp, "# captured on: %s", ctime(&st.st_ctime));
	perf_header__fprintf_info(session, fp, full);
	fprintf(fp, "# ========\n#\n");
}
1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841


int __perf_session__set_tracepoints_handlers(struct perf_session *session,
					     const struct perf_evsel_str_handler *assocs,
					     size_t nr_assocs)
{
	struct perf_evsel *evsel;
	size_t i;
	int err;

	for (i = 0; i < nr_assocs; i++) {
1842 1843 1844 1845 1846
		/*
		 * Adding a handler for an event not in the session,
		 * just ignore it.
		 */
		evsel = perf_evlist__find_tracepoint_by_name(session->evlist, assocs[i].name);
1847
		if (evsel == NULL)
1848
			continue;
1849 1850

		err = -EEXIST;
1851
		if (evsel->handler != NULL)
1852
			goto out;
1853
		evsel->handler = assocs[i].handler;
1854 1855 1856 1857 1858 1859
	}

	err = 0;
out:
	return err;
}
A
Adrian Hunter 已提交
1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913

int perf_event__process_id_index(struct perf_tool *tool __maybe_unused,
				 union perf_event *event,
				 struct perf_session *session)
{
	struct perf_evlist *evlist = session->evlist;
	struct id_index_event *ie = &event->id_index;
	size_t i, nr, max_nr;

	max_nr = (ie->header.size - sizeof(struct id_index_event)) /
		 sizeof(struct id_index_entry);
	nr = ie->nr;
	if (nr > max_nr)
		return -EINVAL;

	if (dump_trace)
		fprintf(stdout, " nr: %zu\n", nr);

	for (i = 0; i < nr; i++) {
		struct id_index_entry *e = &ie->entries[i];
		struct perf_sample_id *sid;

		if (dump_trace) {
			fprintf(stdout,	" ... id: %"PRIu64, e->id);
			fprintf(stdout,	"  idx: %"PRIu64, e->idx);
			fprintf(stdout,	"  cpu: %"PRId64, e->cpu);
			fprintf(stdout,	"  tid: %"PRId64"\n", e->tid);
		}

		sid = perf_evlist__id2sid(evlist, e->id);
		if (!sid)
			return -ENOENT;
		sid->idx = e->idx;
		sid->cpu = e->cpu;
		sid->tid = e->tid;
	}
	return 0;
}

int perf_event__synthesize_id_index(struct perf_tool *tool,
				    perf_event__handler_t process,
				    struct perf_evlist *evlist,
				    struct machine *machine)
{
	union perf_event *ev;
	struct perf_evsel *evsel;
	size_t nr = 0, i = 0, sz, max_nr, n;
	int err;

	pr_debug2("Synthesizing id index\n");

	max_nr = (UINT16_MAX - sizeof(struct id_index_event)) /
		 sizeof(struct id_index_entry);

1914
	evlist__for_each(evlist, evsel)
A
Adrian Hunter 已提交
1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926
		nr += evsel->ids;

	n = nr > max_nr ? max_nr : nr;
	sz = sizeof(struct id_index_event) + n * sizeof(struct id_index_entry);
	ev = zalloc(sz);
	if (!ev)
		return -ENOMEM;

	ev->id_index.header.type = PERF_RECORD_ID_INDEX;
	ev->id_index.header.size = sz;
	ev->id_index.nr = n;

1927
	evlist__for_each(evlist, evsel) {
A
Adrian Hunter 已提交
1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967
		u32 j;

		for (j = 0; j < evsel->ids; j++) {
			struct id_index_entry *e;
			struct perf_sample_id *sid;

			if (i >= n) {
				err = process(tool, ev, NULL, machine);
				if (err)
					goto out_err;
				nr -= n;
				i = 0;
			}

			e = &ev->id_index.entries[i++];

			e->id = evsel->id[j];

			sid = perf_evlist__id2sid(evlist, e->id);
			if (!sid) {
				free(ev);
				return -ENOENT;
			}

			e->idx = sid->idx;
			e->cpu = sid->cpu;
			e->tid = sid->tid;
		}
	}

	sz = sizeof(struct id_index_event) + nr * sizeof(struct id_index_entry);
	ev->id_index.header.size = sz;
	ev->id_index.nr = nr;

	err = process(tool, ev, NULL, machine);
out_err:
	free(ev);

	return err;
}