symbol.c 59.5 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13
#define _GNU_SOURCE
#include <ctype.h>
#include <dirent.h>
#include <errno.h>
#include <libgen.h>
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <sys/param.h>
#include <fcntl.h>
#include <unistd.h>
14
#include <inttypes.h>
15
#include "build-id.h"
16
#include "debug.h"
17
#include "symbol.h"
18
#include "strlist.h"
19 20 21 22

#include <libelf.h>
#include <gelf.h>
#include <elf.h>
23
#include <limits.h>
24
#include <sys/utsname.h>
P
Peter Zijlstra 已提交
25

26 27 28 29
#ifndef KSYM_NAME_LEN
#define KSYM_NAME_LEN 128
#endif

30 31 32 33
#ifndef NT_GNU_BUILD_ID
#define NT_GNU_BUILD_ID 3
#endif

34
static bool dso__build_id_equal(const struct dso *dso, u8 *build_id);
35
static int elf_read_build_id(Elf *elf, void *bf, size_t size);
36
static void dsos__add(struct list_head *head, struct dso *dso);
37
static struct map *map__new2(u64 start, struct dso *dso, enum map_type type);
38
static int dso__load_kernel_sym(struct dso *dso, struct map *map,
39
				symbol_filter_t filter);
40
static int dso__load_guest_kernel_sym(struct dso *dso, struct map *map,
41
			symbol_filter_t filter);
42 43
static int vmlinux_path__nr_entries;
static char **vmlinux_path;
44

45
struct symbol_conf symbol_conf = {
46
	.exclude_other	  = true,
47 48
	.use_modules	  = true,
	.try_vmlinux_path = true,
49
	.symfs            = "",
50 51
};

52
int dso__name_len(const struct dso *dso)
53 54
{
	if (verbose)
55
		return dso->long_name_len;
56

57
	return dso->short_name_len;
58 59
}

60
bool dso__loaded(const struct dso *dso, enum map_type type)
61
{
62
	return dso->loaded & (1 << type);
63 64
}

65
bool dso__sorted_by_name(const struct dso *dso, enum map_type type)
66
{
67
	return dso->sorted_by_name & (1 << type);
68 69
}

70
static void dso__set_sorted_by_name(struct dso *dso, enum map_type type)
71
{
72
	dso->sorted_by_name |= (1 << type);
73 74
}

75
bool symbol_type__is_a(char symbol_type, enum map_type map_type)
76 77 78 79
{
	switch (map_type) {
	case MAP__FUNCTION:
		return symbol_type == 'T' || symbol_type == 'W';
80 81
	case MAP__VARIABLE:
		return symbol_type == 'D' || symbol_type == 'd';
82 83 84 85 86
	default:
		return false;
	}
}

87
static void symbols__fixup_end(struct rb_root *symbols)
88
{
89
	struct rb_node *nd, *prevnd = rb_first(symbols);
90
	struct symbol *curr, *prev;
91 92 93 94

	if (prevnd == NULL)
		return;

95 96
	curr = rb_entry(prevnd, struct symbol, rb_node);

97
	for (nd = rb_next(prevnd); nd; nd = rb_next(nd)) {
98 99
		prev = curr;
		curr = rb_entry(nd, struct symbol, rb_node);
100

101
		if (prev->end == prev->start && prev->end != curr->start)
102 103
			prev->end = curr->start - 1;
	}
104 105 106 107

	/* Last entry */
	if (curr->end == curr->start)
		curr->end = roundup(curr->start, 4096);
108 109
}

110
static void __map_groups__fixup_end(struct map_groups *mg, enum map_type type)
111 112
{
	struct map *prev, *curr;
113
	struct rb_node *nd, *prevnd = rb_first(&mg->maps[type]);
114 115 116 117 118 119 120 121 122 123

	if (prevnd == NULL)
		return;

	curr = rb_entry(prevnd, struct map, rb_node);

	for (nd = rb_next(prevnd); nd; nd = rb_next(nd)) {
		prev = curr;
		curr = rb_entry(nd, struct map, rb_node);
		prev->end = curr->start - 1;
124
	}
125 126 127 128 129

	/*
	 * We still haven't the actual symbols, so guess the
	 * last map final address.
	 */
130
	curr->end = ~0ULL;
131 132
}

133
static void map_groups__fixup_end(struct map_groups *mg)
134 135 136
{
	int i;
	for (i = 0; i < MAP__NR_TYPES; ++i)
137
		__map_groups__fixup_end(mg, i);
138 139
}

140 141
static struct symbol *symbol__new(u64 start, u64 len, u8 binding,
				  const char *name)
142
{
143
	size_t namelen = strlen(name) + 1;
144 145 146
	struct symbol *sym = calloc(1, (symbol_conf.priv_size +
					sizeof(*sym) + namelen));
	if (sym == NULL)
147 148
		return NULL;

149
	if (symbol_conf.priv_size)
150
		sym = ((void *)sym) + symbol_conf.priv_size;
151

152 153 154 155
	sym->start   = start;
	sym->end     = len ? start + len - 1 : start;
	sym->binding = binding;
	sym->namelen = namelen - 1;
156

157 158 159
	pr_debug4("%s: %s %#" PRIx64 "-%#" PRIx64 "\n",
		  __func__, name, start, sym->end);
	memcpy(sym->name, name, namelen);
160

161
	return sym;
162 163
}

164
void symbol__delete(struct symbol *sym)
165
{
166
	free(((void *)sym) - symbol_conf.priv_size);
167 168
}

169
static size_t symbol__fprintf(struct symbol *sym, FILE *fp)
170
{
171
	return fprintf(fp, " %" PRIx64 "-%" PRIx64 " %c %s\n",
172 173 174 175
		       sym->start, sym->end,
		       sym->binding == STB_GLOBAL ? 'g' :
		       sym->binding == STB_LOCAL  ? 'l' : 'w',
		       sym->name);
176 177
}

178
void dso__set_long_name(struct dso *dso, char *name)
179
{
180 181
	if (name == NULL)
		return;
182 183
	dso->long_name = name;
	dso->long_name_len = strlen(name);
184 185
}

186
static void dso__set_short_name(struct dso *dso, const char *name)
187 188 189
{
	if (name == NULL)
		return;
190 191
	dso->short_name = name;
	dso->short_name_len = strlen(name);
192 193
}

194
static void dso__set_basename(struct dso *dso)
195
{
196
	dso__set_short_name(dso, basename(dso->long_name));
197 198
}

199
struct dso *dso__new(const char *name)
200
{
201
	struct dso *dso = calloc(1, sizeof(*dso) + strlen(name) + 1);
202

203
	if (dso != NULL) {
204
		int i;
205 206 207
		strcpy(dso->name, name);
		dso__set_long_name(dso, dso->name);
		dso__set_short_name(dso, dso->name);
208
		for (i = 0; i < MAP__NR_TYPES; ++i)
209 210 211 212 213 214 215
			dso->symbols[i] = dso->symbol_names[i] = RB_ROOT;
		dso->symtab_type = SYMTAB__NOT_FOUND;
		dso->loaded = 0;
		dso->sorted_by_name = 0;
		dso->has_build_id = 0;
		dso->kernel = DSO_TYPE_USER;
		INIT_LIST_HEAD(&dso->node);
216 217
	}

218
	return dso;
219 220
}

221
static void symbols__delete(struct rb_root *symbols)
222 223
{
	struct symbol *pos;
224
	struct rb_node *next = rb_first(symbols);
225 226 227 228

	while (next) {
		pos = rb_entry(next, struct symbol, rb_node);
		next = rb_next(&pos->rb_node);
229
		rb_erase(&pos->rb_node, symbols);
230
		symbol__delete(pos);
231 232 233
	}
}

234
void dso__delete(struct dso *dso)
235
{
236 237
	int i;
	for (i = 0; i < MAP__NR_TYPES; ++i)
238 239 240 241 242 243
		symbols__delete(&dso->symbols[i]);
	if (dso->sname_alloc)
		free((char *)dso->short_name);
	if (dso->lname_alloc)
		free(dso->long_name);
	free(dso);
244 245
}

246
void dso__set_build_id(struct dso *dso, void *build_id)
247
{
248 249
	memcpy(dso->build_id, build_id, sizeof(dso->build_id));
	dso->has_build_id = 1;
250 251
}

252
static void symbols__insert(struct rb_root *symbols, struct symbol *sym)
253
{
254
	struct rb_node **p = &symbols->rb_node;
255
	struct rb_node *parent = NULL;
256
	const u64 ip = sym->start;
257 258 259 260 261 262 263 264 265 266 267
	struct symbol *s;

	while (*p != NULL) {
		parent = *p;
		s = rb_entry(parent, struct symbol, rb_node);
		if (ip < s->start)
			p = &(*p)->rb_left;
		else
			p = &(*p)->rb_right;
	}
	rb_link_node(&sym->rb_node, parent, p);
268
	rb_insert_color(&sym->rb_node, symbols);
269 270
}

271
static struct symbol *symbols__find(struct rb_root *symbols, u64 ip)
272 273 274
{
	struct rb_node *n;

275
	if (symbols == NULL)
276 277
		return NULL;

278
	n = symbols->rb_node;
279 280 281 282 283 284 285 286 287 288 289 290 291 292 293

	while (n) {
		struct symbol *s = rb_entry(n, struct symbol, rb_node);

		if (ip < s->start)
			n = n->rb_left;
		else if (ip > s->end)
			n = n->rb_right;
		else
			return s;
	}

	return NULL;
}

294 295 296 297 298
struct symbol_name_rb_node {
	struct rb_node	rb_node;
	struct symbol	sym;
};

299
static void symbols__insert_by_name(struct rb_root *symbols, struct symbol *sym)
300
{
301
	struct rb_node **p = &symbols->rb_node;
302
	struct rb_node *parent = NULL;
303 304 305
	struct symbol_name_rb_node *symn, *s;

	symn = container_of(sym, struct symbol_name_rb_node, sym);
306 307 308 309 310 311 312 313 314 315

	while (*p != NULL) {
		parent = *p;
		s = rb_entry(parent, struct symbol_name_rb_node, rb_node);
		if (strcmp(sym->name, s->sym.name) < 0)
			p = &(*p)->rb_left;
		else
			p = &(*p)->rb_right;
	}
	rb_link_node(&symn->rb_node, parent, p);
316
	rb_insert_color(&symn->rb_node, symbols);
317 318
}

319 320
static void symbols__sort_by_name(struct rb_root *symbols,
				  struct rb_root *source)
321 322 323 324 325
{
	struct rb_node *nd;

	for (nd = rb_first(source); nd; nd = rb_next(nd)) {
		struct symbol *pos = rb_entry(nd, struct symbol, rb_node);
326
		symbols__insert_by_name(symbols, pos);
327 328 329
	}
}

330 331
static struct symbol *symbols__find_by_name(struct rb_root *symbols,
					    const char *name)
332 333 334
{
	struct rb_node *n;

335
	if (symbols == NULL)
336 337
		return NULL;

338
	n = symbols->rb_node;
339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357

	while (n) {
		struct symbol_name_rb_node *s;
		int cmp;

		s = rb_entry(n, struct symbol_name_rb_node, rb_node);
		cmp = strcmp(name, s->sym.name);

		if (cmp < 0)
			n = n->rb_left;
		else if (cmp > 0)
			n = n->rb_right;
		else
			return &s->sym;
	}

	return NULL;
}

358
struct symbol *dso__find_symbol(struct dso *dso,
359
				enum map_type type, u64 addr)
360
{
361
	return symbols__find(&dso->symbols[type], addr);
362 363
}

364
struct symbol *dso__find_symbol_by_name(struct dso *dso, enum map_type type,
365 366
					const char *name)
{
367
	return symbols__find_by_name(&dso->symbol_names[type], name);
368 369
}

370
void dso__sort_by_name(struct dso *dso, enum map_type type)
371
{
372 373 374
	dso__set_sorted_by_name(dso, type);
	return symbols__sort_by_name(&dso->symbol_names[type],
				     &dso->symbols[type]);
375 376
}

377
int build_id__sprintf(const u8 *build_id, int len, char *bf)
378
{
379
	char *bid = bf;
380
	const u8 *raw = build_id;
381
	int i;
382

383 384 385 386 387 388
	for (i = 0; i < len; ++i) {
		sprintf(bid, "%02x", *raw);
		++raw;
		bid += 2;
	}

389
	return raw - build_id;
390 391
}

392
size_t dso__fprintf_buildid(struct dso *dso, FILE *fp)
393 394 395
{
	char sbuild_id[BUILD_ID_SIZE * 2 + 1];

396
	build_id__sprintf(dso->build_id, sizeof(dso->build_id), sbuild_id);
397 398 399
	return fprintf(fp, "%s", sbuild_id);
}

400 401
size_t dso__fprintf_symbols_by_name(struct dso *dso,
				    enum map_type type, FILE *fp)
402 403 404 405 406
{
	size_t ret = 0;
	struct rb_node *nd;
	struct symbol_name_rb_node *pos;

407
	for (nd = rb_first(&dso->symbol_names[type]); nd; nd = rb_next(nd)) {
408 409 410 411 412 413 414
		pos = rb_entry(nd, struct symbol_name_rb_node, rb_node);
		fprintf(fp, "%s\n", pos->sym.name);
	}

	return ret;
}

415
size_t dso__fprintf(struct dso *dso, enum map_type type, FILE *fp)
416 417
{
	struct rb_node *nd;
418
	size_t ret = fprintf(fp, "dso: %s (", dso->short_name);
419

420 421
	if (dso->short_name != dso->long_name)
		ret += fprintf(fp, "%s, ", dso->long_name);
422
	ret += fprintf(fp, "%s, %sloaded, ", map_type__name[type],
423 424
		       dso->loaded ? "" : "NOT ");
	ret += dso__fprintf_buildid(dso, fp);
425
	ret += fprintf(fp, ")\n");
426
	for (nd = rb_first(&dso->symbols[type]); nd; nd = rb_next(nd)) {
427 428
		struct symbol *pos = rb_entry(nd, struct symbol, rb_node);
		ret += symbol__fprintf(pos, fp);
429 430 431 432 433
	}

	return ret;
}

434 435
int kallsyms__parse(const char *filename, void *arg,
		    int (*process_symbol)(void *arg, const char *name,
436
					  char type, u64 start, u64 end))
437 438 439
{
	char *line = NULL;
	size_t n;
440 441 442 443
	int err = -1;
	u64 prev_start = 0;
	char prev_symbol_type = 0;
	char *prev_symbol_name;
444
	FILE *file = fopen(filename, "r");
445 446 447 448

	if (file == NULL)
		goto out_failure;

449 450 451 452 453 454
	prev_symbol_name = malloc(KSYM_NAME_LEN);
	if (prev_symbol_name == NULL)
		goto out_close;

	err = 0;

455
	while (!feof(file)) {
456
		u64 start;
457 458
		int line_len, len;
		char symbol_type;
459
		char *symbol_name;
460 461

		line_len = getline(&line, &n, file);
462
		if (line_len < 0 || !line)
463 464 465 466
			break;

		line[--line_len] = '\0'; /* \n */

467
		len = hex2u64(line, &start);
468 469 470 471 472 473

		len++;
		if (len + 2 >= line_len)
			continue;

		symbol_type = toupper(line[len]);
474 475 476
		len += 2;
		symbol_name = line + len;
		len = line_len - len;
477

478 479
		if (len >= KSYM_NAME_LEN) {
			err = -1;
480
			break;
481 482 483 484 485 486 487 488 489 490 491 492 493 494 495
		}

		if (prev_symbol_type) {
			u64 end = start;
			if (end != prev_start)
				--end;
			err = process_symbol(arg, prev_symbol_name,
					     prev_symbol_type, prev_start, end);
			if (err)
				break;
		}

		memcpy(prev_symbol_name, symbol_name, len + 1);
		prev_symbol_type = symbol_type;
		prev_start = start;
496 497
	}

498
	free(prev_symbol_name);
499
	free(line);
500
out_close:
501
	fclose(file);
502
	return err;
503 504 505 506 507

out_failure:
	return -1;
}

508 509 510 511 512
struct process_kallsyms_args {
	struct map *map;
	struct dso *dso;
};

513 514 515 516 517 518 519 520
static u8 kallsyms2elf_type(char type)
{
	if (type == 'W')
		return STB_WEAK;

	return isupper(type) ? STB_GLOBAL : STB_LOCAL;
}

521
static int map__process_kallsym_symbol(void *arg, const char *name,
522
				       char type, u64 start, u64 end)
523 524 525 526 527 528 529 530
{
	struct symbol *sym;
	struct process_kallsyms_args *a = arg;
	struct rb_root *root = &a->dso->symbols[a->map->type];

	if (!symbol_type__is_a(type, a->map->type))
		return 0;

531 532
	sym = symbol__new(start, end - start + 1,
			  kallsyms2elf_type(type), name);
533 534 535 536 537 538 539
	if (sym == NULL)
		return -ENOMEM;
	/*
	 * We will pass the symbols to the filter later, in
	 * map__split_kallsyms, when we have split the maps per module
	 */
	symbols__insert(root, sym);
540

541 542 543 544 545 546 547 548
	return 0;
}

/*
 * Loads the function entries in /proc/kallsyms into kernel_map->dso,
 * so that we can in the next step set the symbol ->end address and then
 * call kernel_maps__split_kallsyms.
 */
549
static int dso__load_all_kallsyms(struct dso *dso, const char *filename,
550
				  struct map *map)
551
{
552
	struct process_kallsyms_args args = { .map = map, .dso = dso, };
553
	return kallsyms__parse(filename, &args, map__process_kallsym_symbol);
554 555
}

556 557 558 559 560
/*
 * Split the symbols into maps, making sure there are no overlaps, i.e. the
 * kernel range is broken in several maps, named [kernel].N, as we don't have
 * the original ELF section names vmlinux have.
 */
561
static int dso__split_kallsyms(struct dso *dso, struct map *map,
562
			       symbol_filter_t filter)
563
{
564
	struct map_groups *kmaps = map__kmap(map)->kmaps;
565
	struct machine *machine = kmaps->machine;
566
	struct map *curr_map = map;
567
	struct symbol *pos;
568
	int count = 0, moved = 0;	
569
	struct rb_root *root = &dso->symbols[map->type];
570
	struct rb_node *next = rb_first(root);
571 572 573 574 575 576 577 578 579 580
	int kernel_range = 0;

	while (next) {
		char *module;

		pos = rb_entry(next, struct symbol, rb_node);
		next = rb_next(&pos->rb_node);

		module = strchr(pos->name, '\t');
		if (module) {
581
			if (!symbol_conf.use_modules)
582 583
				goto discard_symbol;

584 585
			*module++ = '\0';

586
			if (strcmp(curr_map->dso->short_name, module)) {
587
				if (curr_map != map &&
588
				    dso->kernel == DSO_TYPE_GUEST_KERNEL &&
589
				    machine__is_default_guest(machine)) {
590 591 592 593 594 595 596 597 598 599 600 601 602
					/*
					 * We assume all symbols of a module are
					 * continuous in * kallsyms, so curr_map
					 * points to a module and all its
					 * symbols are in its kmap. Mark it as
					 * loaded.
					 */
					dso__set_loaded(curr_map->dso,
							curr_map->type);
				}

				curr_map = map_groups__find_by_name(kmaps,
							map->type, module);
603
				if (curr_map == NULL) {
604
					pr_debug("%s/proc/{kallsyms,modules} "
605
					         "inconsistency while looking "
606
						 "for \"%s\" module!\n",
607
						 machine->root_dir, module);
608 609
					curr_map = map;
					goto discard_symbol;
610
				}
611

612
				if (curr_map->dso->loaded &&
613
				    !machine__is_default_guest(machine))
614
					goto discard_symbol;
615
			}
616 617 618 619
			/*
			 * So that we look just like we get from .ko files,
			 * i.e. not prelinked, relative to map->start.
			 */
620 621 622
			pos->start = curr_map->map_ip(curr_map, pos->start);
			pos->end   = curr_map->map_ip(curr_map, pos->end);
		} else if (curr_map != map) {
623
			char dso_name[PATH_MAX];
624
			struct dso *ndso;
625

626 627 628 629 630
			if (count == 0) {
				curr_map = map;
				goto filter_symbol;
			}

631
			if (dso->kernel == DSO_TYPE_GUEST_KERNEL)
632 633 634 635 636 637 638
				snprintf(dso_name, sizeof(dso_name),
					"[guest.kernel].%d",
					kernel_range++);
			else
				snprintf(dso_name, sizeof(dso_name),
					"[kernel].%d",
					kernel_range++);
639

640 641
			ndso = dso__new(dso_name);
			if (ndso == NULL)
642 643
				return -1;

644
			ndso->kernel = dso->kernel;
645

646
			curr_map = map__new2(pos->start, ndso, map->type);
647
			if (curr_map == NULL) {
648
				dso__delete(ndso);
649 650
				return -1;
			}
651

652
			curr_map->map_ip = curr_map->unmap_ip = identity__map_ip;
653
			map_groups__insert(kmaps, curr_map);
654 655
			++kernel_range;
		}
656
filter_symbol:
657
		if (filter && filter(curr_map, pos)) {
658
discard_symbol:		rb_erase(&pos->rb_node, root);
659
			symbol__delete(pos);
660
		} else {
661 662 663
			if (curr_map != map) {
				rb_erase(&pos->rb_node, root);
				symbols__insert(&curr_map->dso->symbols[curr_map->type], pos);
664 665 666
				++moved;
			} else
				++count;
667
		}
668 669
	}

670
	if (curr_map != map &&
671
	    dso->kernel == DSO_TYPE_GUEST_KERNEL &&
672
	    machine__is_default_guest(kmaps->machine)) {
673 674 675
		dso__set_loaded(curr_map->dso, curr_map->type);
	}

676
	return count + moved;
677
}
678

679
int dso__load_kallsyms(struct dso *dso, const char *filename,
680
		       struct map *map, symbol_filter_t filter)
681
{
682
	if (dso__load_all_kallsyms(dso, filename, map) < 0)
683 684
		return -1;

685 686
	if (dso->kernel == DSO_TYPE_GUEST_KERNEL)
		dso->symtab_type = SYMTAB__GUEST_KALLSYMS;
687
	else
688
		dso->symtab_type = SYMTAB__KALLSYMS;
689

690
	return dso__split_kallsyms(dso, map, filter);
691 692
}

693
static int dso__load_perf_map(struct dso *dso, struct map *map,
694
			      symbol_filter_t filter)
695 696 697 698 699 700
{
	char *line = NULL;
	size_t n;
	FILE *file;
	int nr_syms = 0;

701
	file = fopen(dso->long_name, "r");
702 703 704 705
	if (file == NULL)
		goto out_failure;

	while (!feof(file)) {
706
		u64 start, size;
707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730
		struct symbol *sym;
		int line_len, len;

		line_len = getline(&line, &n, file);
		if (line_len < 0)
			break;

		if (!line)
			goto out_failure;

		line[--line_len] = '\0'; /* \n */

		len = hex2u64(line, &start);

		len++;
		if (len + 2 >= line_len)
			continue;

		len += hex2u64(line + len, &size);

		len++;
		if (len + 2 >= line_len)
			continue;

731
		sym = symbol__new(start, size, STB_GLOBAL, line + len);
732 733 734 735

		if (sym == NULL)
			goto out_delete_line;

736
		if (filter && filter(map, sym))
737
			symbol__delete(sym);
738
		else {
739
			symbols__insert(&dso->symbols[map->type], sym);
740 741 742 743 744 745 746 747 748 749 750 751 752 753 754
			nr_syms++;
		}
	}

	free(line);
	fclose(file);

	return nr_syms;

out_delete_line:
	free(line);
out_failure:
	return -1;
}

755 756 757
/**
 * elf_symtab__for_each_symbol - iterate thru all the symbols
 *
758
 * @syms: struct elf_symtab instance to iterate
759
 * @idx: uint32_t idx
760 761
 * @sym: GElf_Sym iterator
 */
762 763 764 765
#define elf_symtab__for_each_symbol(syms, nr_syms, idx, sym) \
	for (idx = 0, gelf_getsym(syms, idx, &sym);\
	     idx < nr_syms; \
	     idx++, gelf_getsym(syms, idx, &sym))
766 767 768 769 770 771 772 773 774 775

static inline uint8_t elf_sym__type(const GElf_Sym *sym)
{
	return GELF_ST_TYPE(sym->st_info);
}

static inline int elf_sym__is_function(const GElf_Sym *sym)
{
	return elf_sym__type(sym) == STT_FUNC &&
	       sym->st_name != 0 &&
776
	       sym->st_shndx != SHN_UNDEF;
777 778
}

779 780 781 782 783 784 785
static inline bool elf_sym__is_object(const GElf_Sym *sym)
{
	return elf_sym__type(sym) == STT_OBJECT &&
		sym->st_name != 0 &&
		sym->st_shndx != SHN_UNDEF;
}

786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805
static inline int elf_sym__is_label(const GElf_Sym *sym)
{
	return elf_sym__type(sym) == STT_NOTYPE &&
		sym->st_name != 0 &&
		sym->st_shndx != SHN_UNDEF &&
		sym->st_shndx != SHN_ABS;
}

static inline const char *elf_sec__name(const GElf_Shdr *shdr,
					const Elf_Data *secstrs)
{
	return secstrs->d_buf + shdr->sh_name;
}

static inline int elf_sec__is_text(const GElf_Shdr *shdr,
					const Elf_Data *secstrs)
{
	return strstr(elf_sec__name(shdr, secstrs), "text") != NULL;
}

806 807 808 809 810 811
static inline bool elf_sec__is_data(const GElf_Shdr *shdr,
				    const Elf_Data *secstrs)
{
	return strstr(elf_sec__name(shdr, secstrs), "data") != NULL;
}

812 813 814 815 816 817 818 819
static inline const char *elf_sym__name(const GElf_Sym *sym,
					const Elf_Data *symstrs)
{
	return symstrs->d_buf + sym->st_name;
}

static Elf_Scn *elf_section_by_name(Elf *elf, GElf_Ehdr *ep,
				    GElf_Shdr *shp, const char *name,
820
				    size_t *idx)
821 822 823 824 825 826 827 828 829 830
{
	Elf_Scn *sec = NULL;
	size_t cnt = 1;

	while ((sec = elf_nextscn(elf, sec)) != NULL) {
		char *str;

		gelf_getshdr(sec, shp);
		str = elf_strptr(elf, ep->e_shstrndx, shp->sh_name);
		if (!strcmp(name, str)) {
831 832
			if (idx)
				*idx = cnt;
833 834 835 836 837 838 839 840
			break;
		}
		++cnt;
	}

	return sec;
}

841 842 843 844 845 846 847 848 849 850
#define elf_section__for_each_rel(reldata, pos, pos_mem, idx, nr_entries) \
	for (idx = 0, pos = gelf_getrel(reldata, 0, &pos_mem); \
	     idx < nr_entries; \
	     ++idx, pos = gelf_getrel(reldata, idx, &pos_mem))

#define elf_section__for_each_rela(reldata, pos, pos_mem, idx, nr_entries) \
	for (idx = 0, pos = gelf_getrela(reldata, 0, &pos_mem); \
	     idx < nr_entries; \
	     ++idx, pos = gelf_getrela(reldata, idx, &pos_mem))

851 852 853 854 855 856 857
/*
 * We need to check if we have a .dynsym, so that we can handle the
 * .plt, synthesizing its symbols, that aren't on the symtabs (be it
 * .dynsym or .symtab).
 * And always look at the original dso, not at debuginfo packages, that
 * have the PLT data stripped out (shdr_rel_plt.sh_type == SHT_NOBITS).
 */
858
static int dso__synthesize_plt_symbols(struct  dso *dso, struct map *map,
859
				       symbol_filter_t filter)
860 861 862
{
	uint32_t nr_rel_entries, idx;
	GElf_Sym sym;
863
	u64 plt_offset;
864 865
	GElf_Shdr shdr_plt;
	struct symbol *f;
866
	GElf_Shdr shdr_rel_plt, shdr_dynsym;
867
	Elf_Data *reldata, *syms, *symstrs;
868 869 870
	Elf_Scn *scn_plt_rel, *scn_symstrs, *scn_dynsym;
	size_t dynsym_idx;
	GElf_Ehdr ehdr;
871
	char sympltname[1024];
872 873
	Elf *elf;
	int nr = 0, symidx, fd, err = 0;
874
	char name[PATH_MAX];
875

876
	snprintf(name, sizeof(name), "%s%s",
877
		 symbol_conf.symfs, dso->long_name);
878
	fd = open(name, O_RDONLY);
879 880 881
	if (fd < 0)
		goto out;

882
	elf = elf_begin(fd, PERF_ELF_C_READ_MMAP, NULL);
883 884 885 886 887 888 889 890 891 892
	if (elf == NULL)
		goto out_close;

	if (gelf_getehdr(elf, &ehdr) == NULL)
		goto out_elf_end;

	scn_dynsym = elf_section_by_name(elf, &ehdr, &shdr_dynsym,
					 ".dynsym", &dynsym_idx);
	if (scn_dynsym == NULL)
		goto out_elf_end;
893

894
	scn_plt_rel = elf_section_by_name(elf, &ehdr, &shdr_rel_plt,
895 896
					  ".rela.plt", NULL);
	if (scn_plt_rel == NULL) {
897
		scn_plt_rel = elf_section_by_name(elf, &ehdr, &shdr_rel_plt,
898 899
						  ".rel.plt", NULL);
		if (scn_plt_rel == NULL)
900
			goto out_elf_end;
901 902
	}

903 904
	err = -1;

905
	if (shdr_rel_plt.sh_link != dynsym_idx)
906
		goto out_elf_end;
907

908 909
	if (elf_section_by_name(elf, &ehdr, &shdr_plt, ".plt", NULL) == NULL)
		goto out_elf_end;
910 911

	/*
912
	 * Fetch the relocation section to find the idxes to the GOT
913 914 915 916
	 * and the symbols in the .dynsym they refer to.
	 */
	reldata = elf_getdata(scn_plt_rel, NULL);
	if (reldata == NULL)
917
		goto out_elf_end;
918 919 920

	syms = elf_getdata(scn_dynsym, NULL);
	if (syms == NULL)
921
		goto out_elf_end;
922

923
	scn_symstrs = elf_getscn(elf, shdr_dynsym.sh_link);
924
	if (scn_symstrs == NULL)
925
		goto out_elf_end;
926 927 928

	symstrs = elf_getdata(scn_symstrs, NULL);
	if (symstrs == NULL)
929
		goto out_elf_end;
930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945

	nr_rel_entries = shdr_rel_plt.sh_size / shdr_rel_plt.sh_entsize;
	plt_offset = shdr_plt.sh_offset;

	if (shdr_rel_plt.sh_type == SHT_RELA) {
		GElf_Rela pos_mem, *pos;

		elf_section__for_each_rela(reldata, pos, pos_mem, idx,
					   nr_rel_entries) {
			symidx = GELF_R_SYM(pos->r_info);
			plt_offset += shdr_plt.sh_entsize;
			gelf_getsym(syms, symidx, &sym);
			snprintf(sympltname, sizeof(sympltname),
				 "%s@plt", elf_sym__name(&sym, symstrs));

			f = symbol__new(plt_offset, shdr_plt.sh_entsize,
946
					STB_GLOBAL, sympltname);
947
			if (!f)
948
				goto out_elf_end;
949

950 951 952
			if (filter && filter(map, f))
				symbol__delete(f);
			else {
953
				symbols__insert(&dso->symbols[map->type], f);
954 955
				++nr;
			}
956 957 958 959 960 961 962 963 964 965 966 967
		}
	} else if (shdr_rel_plt.sh_type == SHT_REL) {
		GElf_Rel pos_mem, *pos;
		elf_section__for_each_rel(reldata, pos, pos_mem, idx,
					  nr_rel_entries) {
			symidx = GELF_R_SYM(pos->r_info);
			plt_offset += shdr_plt.sh_entsize;
			gelf_getsym(syms, symidx, &sym);
			snprintf(sympltname, sizeof(sympltname),
				 "%s@plt", elf_sym__name(&sym, symstrs));

			f = symbol__new(plt_offset, shdr_plt.sh_entsize,
968
					STB_GLOBAL, sympltname);
969
			if (!f)
970
				goto out_elf_end;
971

972 973 974
			if (filter && filter(map, f))
				symbol__delete(f);
			else {
975
				symbols__insert(&dso->symbols[map->type], f);
976 977
				++nr;
			}
978 979 980
		}
	}

981 982 983 984 985 986 987 988 989
	err = 0;
out_elf_end:
	elf_end(elf);
out_close:
	close(fd);

	if (err == 0)
		return nr;
out:
990
	pr_debug("%s: problems reading %s PLT info.\n",
991
		 __func__, dso->long_name);
992
	return 0;
993 994
}

995
static bool elf_sym__is_a(GElf_Sym *sym, enum map_type type)
996 997 998
{
	switch (type) {
	case MAP__FUNCTION:
999
		return elf_sym__is_function(sym);
1000
	case MAP__VARIABLE:
1001
		return elf_sym__is_object(sym);
1002 1003 1004 1005 1006
	default:
		return false;
	}
}

1007 1008
static bool elf_sec__is_a(GElf_Shdr *shdr, Elf_Data *secstrs,
			  enum map_type type)
1009 1010 1011
{
	switch (type) {
	case MAP__FUNCTION:
1012
		return elf_sec__is_text(shdr, secstrs);
1013
	case MAP__VARIABLE:
1014
		return elf_sec__is_data(shdr, secstrs);
1015 1016 1017 1018 1019
	default:
		return false;
	}
}

1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038
static size_t elf_addr_to_index(Elf *elf, GElf_Addr addr)
{
	Elf_Scn *sec = NULL;
	GElf_Shdr shdr;
	size_t cnt = 1;

	while ((sec = elf_nextscn(elf, sec)) != NULL) {
		gelf_getshdr(sec, &shdr);

		if ((addr >= shdr.sh_addr) &&
		    (addr < (shdr.sh_addr + shdr.sh_size)))
			return cnt;

		++cnt;
	}

	return -1;
}

1039
static int dso__load_sym(struct dso *dso, struct map *map, const char *name,
1040 1041
			 int fd, symbol_filter_t filter, int kmodule,
			 int want_symtab)
1042
{
1043
	struct kmap *kmap = dso->kernel ? map__kmap(map) : NULL;
1044
	struct map *curr_map = map;
1045
	struct dso *curr_dso = dso;
1046
	Elf_Data *symstrs, *secstrs;
1047 1048
	uint32_t nr_syms;
	int err = -1;
1049
	uint32_t idx;
1050
	GElf_Ehdr ehdr;
1051 1052
	GElf_Shdr shdr, opdshdr;
	Elf_Data *syms, *opddata = NULL;
1053
	GElf_Sym sym;
1054
	Elf_Scn *sec, *sec_strndx, *opdsec;
1055
	Elf *elf;
1056
	int nr = 0;
1057
	size_t opdidx = 0;
1058

1059
	elf = elf_begin(fd, PERF_ELF_C_READ_MMAP, NULL);
1060
	if (elf == NULL) {
1061
		pr_debug("%s: cannot read %s ELF file.\n", __func__, name);
1062 1063 1064 1065
		goto out_close;
	}

	if (gelf_getehdr(elf, &ehdr) == NULL) {
1066
		pr_debug("%s: cannot get elf header.\n", __func__);
1067 1068 1069
		goto out_elf_end;
	}

1070
	/* Always reject images with a mismatched build-id: */
1071
	if (dso->has_build_id) {
1072 1073 1074 1075 1076 1077
		u8 build_id[BUILD_ID_SIZE];

		if (elf_read_build_id(elf, build_id,
				      BUILD_ID_SIZE) != BUILD_ID_SIZE)
			goto out_elf_end;

1078
		if (!dso__build_id_equal(dso, build_id))
1079 1080 1081
			goto out_elf_end;
	}

1082
	sec = elf_section_by_name(elf, &ehdr, &shdr, ".symtab", NULL);
1083
	if (sec == NULL) {
1084 1085 1086
		if (want_symtab)
			goto out_elf_end;

1087 1088
		sec = elf_section_by_name(elf, &ehdr, &shdr, ".dynsym", NULL);
		if (sec == NULL)
1089 1090
			goto out_elf_end;
	}
1091

1092 1093 1094 1095
	opdsec = elf_section_by_name(elf, &ehdr, &opdshdr, ".opd", &opdidx);
	if (opdsec)
		opddata = elf_rawdata(opdsec, NULL);

1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107
	syms = elf_getdata(sec, NULL);
	if (syms == NULL)
		goto out_elf_end;

	sec = elf_getscn(elf, shdr.sh_link);
	if (sec == NULL)
		goto out_elf_end;

	symstrs = elf_getdata(sec, NULL);
	if (symstrs == NULL)
		goto out_elf_end;

1108 1109 1110 1111 1112
	sec_strndx = elf_getscn(elf, ehdr.e_shstrndx);
	if (sec_strndx == NULL)
		goto out_elf_end;

	secstrs = elf_getdata(sec_strndx, NULL);
S
Stoyan Gaydarov 已提交
1113
	if (secstrs == NULL)
1114 1115
		goto out_elf_end;

1116 1117
	nr_syms = shdr.sh_size / shdr.sh_entsize;

1118
	memset(&sym, 0, sizeof(sym));
1119 1120
	if (dso->kernel == DSO_TYPE_USER) {
		dso->adjust_symbols = (ehdr.e_type == ET_EXEC ||
1121 1122 1123
				elf_section_by_name(elf, &ehdr, &shdr,
						     ".gnu.prelink_undo",
						     NULL) != NULL);
1124 1125 1126
	} else {
		dso->adjust_symbols = 0;
	}
1127
	elf_symtab__for_each_symbol(syms, nr_syms, idx, sym) {
1128
		struct symbol *f;
1129
		const char *elf_name = elf_sym__name(&sym, symstrs);
1130
		char *demangled = NULL;
1131 1132
		int is_label = elf_sym__is_label(&sym);
		const char *section_name;
1133

1134 1135 1136
		if (kmap && kmap->ref_reloc_sym && kmap->ref_reloc_sym->name &&
		    strcmp(elf_name, kmap->ref_reloc_sym->name) == 0)
			kmap->ref_reloc_sym->unrelocated_addr = sym.st_value;
1137

1138
		if (!is_label && !elf_sym__is_a(&sym, map->type))
1139 1140
			continue;

1141 1142 1143 1144 1145 1146 1147 1148 1149 1150
		/* Reject ARM ELF "mapping symbols": these aren't unique and
		 * don't identify functions, so will confuse the profile
		 * output: */
		if (ehdr.e_machine == EM_ARM) {
			if (!strcmp(elf_name, "$a") ||
			    !strcmp(elf_name, "$d") ||
			    !strcmp(elf_name, "$t"))
				continue;
		}

1151 1152 1153 1154 1155 1156 1157
		if (opdsec && sym.st_shndx == opdidx) {
			u32 offset = sym.st_value - opdshdr.sh_addr;
			u64 *opd = opddata->d_buf + offset;
			sym.st_value = *opd;
			sym.st_shndx = elf_addr_to_index(elf, sym.st_value);
		}

1158 1159 1160 1161 1162
		sec = elf_getscn(elf, sym.st_shndx);
		if (!sec)
			goto out_elf_end;

		gelf_getshdr(sec, &shdr);
1163

1164
		if (is_label && !elf_sec__is_a(&shdr, secstrs, map->type))
1165 1166 1167
			continue;

		section_name = elf_sec__name(&shdr, secstrs);
1168

1169 1170 1171 1172 1173 1174 1175
		/* On ARM, symbols for thumb functions have 1 added to
		 * the symbol address as a flag - remove it */
		if ((ehdr.e_machine == EM_ARM) &&
		    (map->type == MAP__FUNCTION) &&
		    (sym.st_value & 1))
			--sym.st_value;

1176
		if (dso->kernel != DSO_TYPE_USER || kmodule) {
1177 1178 1179
			char dso_name[PATH_MAX];

			if (strcmp(section_name,
1180
				   (curr_dso->short_name +
1181
				    dso->short_name_len)) == 0)
1182 1183 1184 1185
				goto new_symbol;

			if (strcmp(section_name, ".text") == 0) {
				curr_map = map;
1186
				curr_dso = dso;
1187 1188 1189 1190
				goto new_symbol;
			}

			snprintf(dso_name, sizeof(dso_name),
1191
				 "%s%s", dso->short_name, section_name);
1192

1193
			curr_map = map_groups__find_by_name(kmap->kmaps, map->type, dso_name);
1194 1195 1196 1197 1198 1199
			if (curr_map == NULL) {
				u64 start = sym.st_value;

				if (kmodule)
					start += map->start + shdr.sh_offset;

1200
				curr_dso = dso__new(dso_name);
1201 1202
				if (curr_dso == NULL)
					goto out_elf_end;
1203 1204 1205
				curr_dso->kernel = dso->kernel;
				curr_dso->long_name = dso->long_name;
				curr_dso->long_name_len = dso->long_name_len;
1206
				curr_map = map__new2(start, curr_dso,
1207
						     map->type);
1208 1209 1210 1211
				if (curr_map == NULL) {
					dso__delete(curr_dso);
					goto out_elf_end;
				}
1212 1213
				curr_map->map_ip = identity__map_ip;
				curr_map->unmap_ip = identity__map_ip;
1214
				curr_dso->symtab_type = dso->symtab_type;
1215
				map_groups__insert(kmap->kmaps, curr_map);
1216
				dsos__add(&dso->node, curr_dso);
1217
				dso__set_loaded(curr_dso, map->type);
1218 1219 1220 1221
			} else
				curr_dso = curr_map->dso;

			goto new_symbol;
1222 1223
		}

1224
		if (curr_dso->adjust_symbols) {
1225 1226
			pr_debug4("%s: adjusting symbol: st_value: %#" PRIx64 " "
				  "sh_addr: %#" PRIx64 " sh_offset: %#" PRIx64 "\n", __func__,
1227 1228
				  (u64)sym.st_value, (u64)shdr.sh_addr,
				  (u64)shdr.sh_offset);
1229
			sym.st_value -= shdr.sh_addr - shdr.sh_offset;
1230
		}
1231 1232 1233 1234 1235
		/*
		 * We need to figure out if the object was created from C++ sources
		 * DWARF DW_compile_unit has this, but we don't always have access
		 * to it...
		 */
1236
		demangled = bfd_demangle(NULL, elf_name, DMGL_PARAMS | DMGL_ANSI);
1237
		if (demangled != NULL)
1238
			elf_name = demangled;
1239
new_symbol:
1240 1241
		f = symbol__new(sym.st_value, sym.st_size,
				GELF_ST_BIND(sym.st_info), elf_name);
1242
		free(demangled);
1243 1244 1245
		if (!f)
			goto out_elf_end;

1246
		if (filter && filter(curr_map, f))
1247
			symbol__delete(f);
1248
		else {
1249
			symbols__insert(&curr_dso->symbols[curr_map->type], f);
1250 1251
			nr++;
		}
1252 1253
	}

1254 1255 1256
	/*
	 * For misannotated, zeroed, ASM function sizes.
	 */
1257
	if (nr > 0) {
1258
		symbols__fixup_end(&dso->symbols[map->type]);
1259 1260 1261 1262 1263 1264 1265 1266
		if (kmap) {
			/*
			 * We need to fixup this here too because we create new
			 * maps here, for things like vsyscall sections.
			 */
			__map_groups__fixup_end(kmap->kmaps, map->type);
		}
	}
1267 1268 1269 1270 1271 1272 1273
	err = nr;
out_elf_end:
	elf_end(elf);
out_close:
	return err;
}

1274
static bool dso__build_id_equal(const struct dso *dso, u8 *build_id)
1275
{
1276
	return memcmp(dso->build_id, build_id, sizeof(dso->build_id)) == 0;
1277 1278
}

1279
bool __dsos__read_build_ids(struct list_head *head, bool with_hits)
1280
{
1281
	bool have_build_id = false;
1282 1283
	struct dso *pos;

1284 1285 1286
	list_for_each_entry(pos, head, node) {
		if (with_hits && !pos->hit)
			continue;
1287 1288 1289 1290
		if (pos->has_build_id) {
			have_build_id = true;
			continue;
		}
1291 1292 1293 1294 1295
		if (filename__read_build_id(pos->long_name, pos->build_id,
					    sizeof(pos->build_id)) > 0) {
			have_build_id	  = true;
			pos->has_build_id = true;
		}
1296
	}
1297

1298
	return have_build_id;
1299 1300
}

1301 1302 1303 1304 1305
/*
 * Align offset to 4 bytes as needed for note name and descriptor data.
 */
#define NOTE_ALIGN(n) (((n) + 3) & -4U)

1306
static int elf_read_build_id(Elf *elf, void *bf, size_t size)
1307
{
1308
	int err = -1;
1309 1310
	GElf_Ehdr ehdr;
	GElf_Shdr shdr;
1311
	Elf_Data *data;
1312
	Elf_Scn *sec;
1313
	Elf_Kind ek;
1314
	void *ptr;
1315

1316 1317 1318
	if (size < BUILD_ID_SIZE)
		goto out;

1319 1320
	ek = elf_kind(elf);
	if (ek != ELF_K_ELF)
1321
		goto out;
1322

1323
	if (gelf_getehdr(elf, &ehdr) == NULL) {
1324
		pr_err("%s: cannot get elf header.\n", __func__);
1325
		goto out;
1326 1327
	}

1328 1329
	sec = elf_section_by_name(elf, &ehdr, &shdr,
				  ".note.gnu.build-id", NULL);
1330 1331 1332 1333
	if (sec == NULL) {
		sec = elf_section_by_name(elf, &ehdr, &shdr,
					  ".notes", NULL);
		if (sec == NULL)
1334
			goto out;
1335
	}
1336

1337 1338
	data = elf_getdata(sec, NULL);
	if (data == NULL)
1339
		goto out;
1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360

	ptr = data->d_buf;
	while (ptr < (data->d_buf + data->d_size)) {
		GElf_Nhdr *nhdr = ptr;
		int namesz = NOTE_ALIGN(nhdr->n_namesz),
		    descsz = NOTE_ALIGN(nhdr->n_descsz);
		const char *name;

		ptr += sizeof(*nhdr);
		name = ptr;
		ptr += namesz;
		if (nhdr->n_type == NT_GNU_BUILD_ID &&
		    nhdr->n_namesz == sizeof("GNU")) {
			if (memcmp(name, "GNU", sizeof("GNU")) == 0) {
				memcpy(bf, ptr, BUILD_ID_SIZE);
				err = BUILD_ID_SIZE;
				break;
			}
		}
		ptr += descsz;
	}
1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385

out:
	return err;
}

int filename__read_build_id(const char *filename, void *bf, size_t size)
{
	int fd, err = -1;
	Elf *elf;

	if (size < BUILD_ID_SIZE)
		goto out;

	fd = open(filename, O_RDONLY);
	if (fd < 0)
		goto out;

	elf = elf_begin(fd, PERF_ELF_C_READ_MMAP, NULL);
	if (elf == NULL) {
		pr_debug2("%s: cannot read %s ELF file.\n", __func__, filename);
		goto out_close;
	}

	err = elf_read_build_id(elf, bf, size);

1386 1387 1388 1389 1390 1391 1392
	elf_end(elf);
out_close:
	close(fd);
out:
	return err;
}

1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411
int sysfs__read_build_id(const char *filename, void *build_id, size_t size)
{
	int fd, err = -1;

	if (size < BUILD_ID_SIZE)
		goto out;

	fd = open(filename, O_RDONLY);
	if (fd < 0)
		goto out;

	while (1) {
		char bf[BUFSIZ];
		GElf_Nhdr nhdr;
		int namesz, descsz;

		if (read(fd, &nhdr, sizeof(nhdr)) != sizeof(nhdr))
			break;

1412 1413
		namesz = NOTE_ALIGN(nhdr.n_namesz);
		descsz = NOTE_ALIGN(nhdr.n_descsz);
1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436
		if (nhdr.n_type == NT_GNU_BUILD_ID &&
		    nhdr.n_namesz == sizeof("GNU")) {
			if (read(fd, bf, namesz) != namesz)
				break;
			if (memcmp(bf, "GNU", sizeof("GNU")) == 0) {
				if (read(fd, build_id,
				    BUILD_ID_SIZE) == BUILD_ID_SIZE) {
					err = 0;
					break;
				}
			} else if (read(fd, bf, descsz) != descsz)
				break;
		} else {
			int n = namesz + descsz;
			if (read(fd, bf, n) != n)
				break;
		}
	}
	close(fd);
out:
	return err;
}

1437
char dso__symtab_origin(const struct dso *dso)
1438 1439
{
	static const char origin[] = {
1440 1441 1442 1443 1444 1445 1446 1447 1448 1449
		[SYMTAB__KALLSYMS]	      = 'k',
		[SYMTAB__JAVA_JIT]	      = 'j',
		[SYMTAB__BUILD_ID_CACHE]      = 'B',
		[SYMTAB__FEDORA_DEBUGINFO]    = 'f',
		[SYMTAB__UBUNTU_DEBUGINFO]    = 'u',
		[SYMTAB__BUILDID_DEBUGINFO]   = 'b',
		[SYMTAB__SYSTEM_PATH_DSO]     = 'd',
		[SYMTAB__SYSTEM_PATH_KMODULE] = 'K',
		[SYMTAB__GUEST_KALLSYMS]      =  'g',
		[SYMTAB__GUEST_KMODULE]	      =  'G',
1450 1451
	};

1452
	if (dso == NULL || dso->symtab_type == SYMTAB__NOT_FOUND)
1453
		return '!';
1454
	return origin[dso->symtab_type];
1455 1456
}

1457
int dso__load(struct dso *dso, struct map *map, symbol_filter_t filter)
1458
{
1459
	int size = PATH_MAX;
1460
	char *name;
1461 1462
	int ret = -1;
	int fd;
1463
	struct machine *machine;
1464
	const char *root_dir;
1465
	int want_symtab;
1466

1467
	dso__set_loaded(dso, map->type);
1468

1469 1470 1471 1472
	if (dso->kernel == DSO_TYPE_KERNEL)
		return dso__load_kernel_sym(dso, map, filter);
	else if (dso->kernel == DSO_TYPE_GUEST_KERNEL)
		return dso__load_guest_kernel_sym(dso, map, filter);
1473

1474 1475
	if (map->groups && map->groups->machine)
		machine = map->groups->machine;
1476
	else
1477
		machine = NULL;
1478 1479

	name = malloc(size);
1480 1481 1482
	if (!name)
		return -1;

1483
	dso->adjust_symbols = 0;
1484

1485 1486 1487
	if (strncmp(dso->name, "/tmp/perf-", 10) == 0) {
		ret = dso__load_perf_map(dso, map, filter);
		dso->symtab_type = ret > 0 ? SYMTAB__JAVA_JIT :
1488
					      SYMTAB__NOT_FOUND;
1489 1490 1491
		return ret;
	}

1492 1493 1494 1495
	/* Iterate over candidate debug images.
	 * On the first pass, only load images if they have a full symtab.
	 * Failing that, do a second pass where we accept .dynsym also
	 */
1496 1497
	want_symtab = 1;
restart:
1498 1499 1500 1501
	for (dso->symtab_type = SYMTAB__BUILD_ID_CACHE;
	     dso->symtab_type != SYMTAB__NOT_FOUND;
	     dso->symtab_type++) {
		switch (dso->symtab_type) {
1502
		case SYMTAB__BUILD_ID_CACHE:
1503 1504
			/* skip the locally configured cache if a symfs is given */
			if (symbol_conf.symfs[0] ||
1505
			    (dso__build_id_filename(dso, name, size) == NULL)) {
1506
				continue;
1507
			}
1508
			break;
1509
		case SYMTAB__FEDORA_DEBUGINFO:
1510
			snprintf(name, size, "%s/usr/lib/debug%s.debug",
1511
				 symbol_conf.symfs, dso->long_name);
1512
			break;
1513
		case SYMTAB__UBUNTU_DEBUGINFO:
1514
			snprintf(name, size, "%s/usr/lib/debug%s",
1515
				 symbol_conf.symfs, dso->long_name);
1516
			break;
1517
		case SYMTAB__BUILDID_DEBUGINFO: {
1518 1519
			char build_id_hex[BUILD_ID_SIZE * 2 + 1];

1520
			if (!dso->has_build_id)
1521 1522
				continue;

1523 1524
			build_id__sprintf(dso->build_id,
					  sizeof(dso->build_id),
1525 1526
					  build_id_hex);
			snprintf(name, size,
1527 1528
				 "%s/usr/lib/debug/.build-id/%.2s/%s.debug",
				 symbol_conf.symfs, build_id_hex, build_id_hex + 2);
1529
			}
1530
			break;
1531
		case SYMTAB__SYSTEM_PATH_DSO:
1532
			snprintf(name, size, "%s%s",
1533
			     symbol_conf.symfs, dso->long_name);
1534
			break;
1535
		case SYMTAB__GUEST_KMODULE:
K
Kyle McMartin 已提交
1536 1537
			if (map->groups && machine)
				root_dir = machine->root_dir;
1538 1539
			else
				root_dir = "";
1540
			snprintf(name, size, "%s%s%s", symbol_conf.symfs,
1541
				 root_dir, dso->long_name);
1542 1543
			break;

1544
		case SYMTAB__SYSTEM_PATH_KMODULE:
1545
			snprintf(name, size, "%s%s", symbol_conf.symfs,
1546
				 dso->long_name);
1547
			break;
1548
		default:;
1549
		}
1550 1551

		/* Name is now the name of the next image to try */
1552
		fd = open(name, O_RDONLY);
1553 1554
		if (fd < 0)
			continue;
1555

1556
		ret = dso__load_sym(dso, map, name, fd, filter, 0,
1557 1558
				    want_symtab);
		close(fd);
1559

1560 1561 1562 1563 1564 1565
		/*
		 * Some people seem to have debuginfo files _WITHOUT_ debug
		 * info!?!?
		 */
		if (!ret)
			continue;
1566

1567
		if (ret > 0) {
1568 1569
			int nr_plt = dso__synthesize_plt_symbols(dso, map,
								 filter);
1570 1571 1572 1573
			if (nr_plt > 0)
				ret += nr_plt;
			break;
		}
1574
	}
1575

1576 1577 1578 1579 1580 1581 1582 1583 1584
	/*
	 * If we wanted a full symtab but no image had one,
	 * relax our requirements and repeat the search.
	 */
	if (ret <= 0 && want_symtab) {
		want_symtab = 0;
		goto restart;
	}

1585
	free(name);
1586
	if (ret < 0 && strstr(dso->name, " (deleted)") != NULL)
1587
		return 0;
1588 1589 1590
	return ret;
}

1591
struct map *map_groups__find_by_name(struct map_groups *mg,
1592
				     enum map_type type, const char *name)
1593 1594 1595
{
	struct rb_node *nd;

1596
	for (nd = rb_first(&mg->maps[type]); nd; nd = rb_next(nd)) {
1597 1598
		struct map *map = rb_entry(nd, struct map, rb_node);

1599
		if (map->dso && strcmp(map->dso->short_name, name) == 0)
1600 1601 1602 1603 1604 1605
			return map;
	}

	return NULL;
}

1606 1607
static int dso__kernel_module_get_build_id(struct dso *dso,
					   const char *root_dir)
1608 1609 1610 1611 1612 1613
{
	char filename[PATH_MAX];
	/*
	 * kernel module short names are of the form "[module]" and
	 * we need just "module" here.
	 */
1614
	const char *name = dso->short_name + 1;
1615 1616

	snprintf(filename, sizeof(filename),
1617 1618
		 "%s/sys/module/%.*s/notes/.note.gnu.build-id",
		 root_dir, (int)strlen(name) - 1, name);
1619

1620 1621 1622
	if (sysfs__read_build_id(filename, dso->build_id,
				 sizeof(dso->build_id)) == 0)
		dso->has_build_id = true;
1623 1624 1625 1626

	return 0;
}

1627
static int map_groups__set_modules_path_dir(struct map_groups *mg,
1628
				const char *dir_name)
1629
{
1630
	struct dirent *dent;
1631
	DIR *dir = opendir(dir_name);
1632
	int ret = 0;
1633

1634
	if (!dir) {
1635
		pr_debug("%s: cannot open %s dir\n", __func__, dir_name);
1636 1637
		return -1;
	}
1638

1639 1640
	while ((dent = readdir(dir)) != NULL) {
		char path[PATH_MAX];
1641 1642 1643 1644 1645 1646
		struct stat st;

		/*sshfs might return bad dent->d_type, so we have to stat*/
		sprintf(path, "%s/%s", dir_name, dent->d_name);
		if (stat(path, &st))
			continue;
1647

1648
		if (S_ISDIR(st.st_mode)) {
1649 1650 1651 1652 1653
			if (!strcmp(dent->d_name, ".") ||
			    !strcmp(dent->d_name, ".."))
				continue;

			snprintf(path, sizeof(path), "%s/%s",
1654
				 dir_name, dent->d_name);
1655
			ret = map_groups__set_modules_path_dir(mg, path);
1656 1657
			if (ret < 0)
				goto out;
1658 1659 1660 1661
		} else {
			char *dot = strrchr(dent->d_name, '.'),
			     dso_name[PATH_MAX];
			struct map *map;
1662
			char *long_name;
1663 1664 1665 1666 1667 1668

			if (dot == NULL || strcmp(dot, ".ko"))
				continue;
			snprintf(dso_name, sizeof(dso_name), "[%.*s]",
				 (int)(dot - dent->d_name), dent->d_name);

1669
			strxfrchar(dso_name, '-', '_');
1670 1671
			map = map_groups__find_by_name(mg, MAP__FUNCTION,
						       dso_name);
1672 1673 1674 1675
			if (map == NULL)
				continue;

			snprintf(path, sizeof(path), "%s/%s",
1676
				 dir_name, dent->d_name);
1677

1678
			long_name = strdup(path);
1679 1680 1681 1682
			if (long_name == NULL) {
				ret = -1;
				goto out;
			}
1683
			dso__set_long_name(map->dso, long_name);
1684
			map->dso->lname_alloc = 1;
1685
			dso__kernel_module_get_build_id(map->dso, "");
1686 1687
		}
	}
1688

1689
out:
1690
	closedir(dir);
1691
	return ret;
1692
}
1693

1694
static char *get_kernel_version(const char *root_dir)
1695
{
1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720
	char version[PATH_MAX];
	FILE *file;
	char *name, *tmp;
	const char *prefix = "Linux version ";

	sprintf(version, "%s/proc/version", root_dir);
	file = fopen(version, "r");
	if (!file)
		return NULL;

	version[0] = '\0';
	tmp = fgets(version, sizeof(version), file);
	fclose(file);

	name = strstr(version, prefix);
	if (!name)
		return NULL;
	name += strlen(prefix);
	tmp = strchr(name, ' ');
	if (tmp)
		*tmp = '\0';

	return strdup(name);
}

1721
static int machine__set_modules_path(struct machine *machine)
1722 1723
{
	char *version;
1724
	char modules_path[PATH_MAX];
1725

1726
	version = get_kernel_version(machine->root_dir);
1727
	if (!version)
1728
		return -1;
1729

1730
	snprintf(modules_path, sizeof(modules_path), "%s/lib/modules/%s/kernel",
1731
		 machine->root_dir, version);
1732
	free(version);
1733

1734
	return map_groups__set_modules_path_dir(&machine->kmaps, modules_path);
1735 1736
}

1737 1738 1739 1740 1741
/*
 * Constructor variant for modules (where we know from /proc/modules where
 * they are loaded) and for vmlinux, where only after we load all the
 * symbols we'll know where it starts and ends.
 */
1742
static struct map *map__new2(u64 start, struct dso *dso, enum map_type type)
1743
{
1744 1745 1746
	struct map *map = calloc(1, (sizeof(*map) +
				     (dso->kernel ? sizeof(struct kmap) : 0)));
	if (map != NULL) {
1747
		/*
1748
		 * ->end will be filled after we load all the symbols
1749
		 */
1750
		map__init(map, type, start, 0, 0, dso);
1751
	}
1752

1753
	return map;
1754 1755
}

1756
struct map *machine__new_module(struct machine *machine, u64 start,
1757
				const char *filename)
1758 1759
{
	struct map *map;
1760
	struct dso *dso = __dsos__findnew(&machine->kernel_dsos, filename);
1761 1762 1763 1764 1765 1766 1767 1768

	if (dso == NULL)
		return NULL;

	map = map__new2(start, dso, MAP__FUNCTION);
	if (map == NULL)
		return NULL;

1769
	if (machine__is_host(machine))
1770
		dso->symtab_type = SYMTAB__SYSTEM_PATH_KMODULE;
1771
	else
1772
		dso->symtab_type = SYMTAB__GUEST_KMODULE;
1773
	map_groups__insert(&machine->kmaps, map);
1774 1775 1776
	return map;
}

1777
static int machine__create_modules(struct machine *machine)
1778 1779 1780
{
	char *line = NULL;
	size_t n;
1781
	FILE *file;
1782
	struct map *map;
1783 1784 1785
	const char *modules;
	char path[PATH_MAX];

1786
	if (machine__is_default_guest(machine))
1787 1788
		modules = symbol_conf.default_guest_modules;
	else {
1789
		sprintf(path, "%s/proc/modules", machine->root_dir);
1790 1791
		modules = path;
	}
1792

1793
	file = fopen(modules, "r");
1794 1795
	if (file == NULL)
		return -1;
1796

1797 1798 1799 1800 1801
	while (!feof(file)) {
		char name[PATH_MAX];
		u64 start;
		char *sep;
		int line_len;
1802

1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824
		line_len = getline(&line, &n, file);
		if (line_len < 0)
			break;

		if (!line)
			goto out_failure;

		line[--line_len] = '\0'; /* \n */

		sep = strrchr(line, 'x');
		if (sep == NULL)
			continue;

		hex2u64(sep + 1, &start);

		sep = strchr(line, ' ');
		if (sep == NULL)
			continue;

		*sep = '\0';

		snprintf(name, sizeof(name), "[%s]", line);
1825
		map = machine__new_module(machine, start, name);
1826
		if (map == NULL)
1827
			goto out_delete_line;
1828
		dso__kernel_module_get_build_id(map->dso, machine->root_dir);
1829
	}
1830 1831 1832 1833

	free(line);
	fclose(file);

1834
	return machine__set_modules_path(machine);
1835 1836 1837 1838 1839

out_delete_line:
	free(line);
out_failure:
	return -1;
1840 1841
}

1842
int dso__load_vmlinux(struct dso *dso, struct map *map,
1843
		      const char *vmlinux, symbol_filter_t filter)
1844
{
1845
	int err = -1, fd;
1846
	char symfs_vmlinux[PATH_MAX];
1847

1848
	snprintf(symfs_vmlinux, sizeof(symfs_vmlinux), "%s%s",
1849 1850
		 symbol_conf.symfs, vmlinux);
	fd = open(symfs_vmlinux, O_RDONLY);
1851 1852 1853
	if (fd < 0)
		return -1;

1854 1855 1856
	dso__set_long_name(dso, (char *)vmlinux);
	dso__set_loaded(dso, map->type);
	err = dso__load_sym(dso, map, symfs_vmlinux, fd, filter, 0, 0);
1857 1858
	close(fd);

1859
	if (err > 0)
1860
		pr_debug("Using %s for symbols\n", symfs_vmlinux);
1861

1862 1863 1864
	return err;
}

1865
int dso__load_vmlinux_path(struct dso *dso, struct map *map,
1866
			   symbol_filter_t filter)
1867 1868
{
	int i, err = 0;
1869
	char *filename;
1870 1871

	pr_debug("Looking at the vmlinux_path (%d entries long)\n",
1872 1873
		 vmlinux_path__nr_entries + 1);

1874
	filename = dso__build_id_filename(dso, NULL, 0);
1875
	if (filename != NULL) {
1876
		err = dso__load_vmlinux(dso, map, filename, filter);
1877
		if (err > 0) {
1878
			dso__set_long_name(dso, filename);
1879 1880 1881 1882
			goto out;
		}
		free(filename);
	}
1883 1884

	for (i = 0; i < vmlinux_path__nr_entries; ++i) {
1885
		err = dso__load_vmlinux(dso, map, vmlinux_path[i], filter);
1886
		if (err > 0) {
1887
			dso__set_long_name(dso, strdup(vmlinux_path[i]));
1888 1889 1890
			break;
		}
	}
1891
out:
1892 1893 1894
	return err;
}

1895
static int dso__load_kernel_sym(struct dso *dso, struct map *map,
1896
				symbol_filter_t filter)
1897
{
1898
	int err;
1899 1900
	const char *kallsyms_filename = NULL;
	char *kallsyms_allocated_filename = NULL;
1901
	/*
1902 1903
	 * Step 1: if the user specified a kallsyms or vmlinux filename, use
	 * it and only it, reporting errors to the user if it cannot be used.
1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915
	 *
	 * For instance, try to analyse an ARM perf.data file _without_ a
	 * build-id, or if the user specifies the wrong path to the right
	 * vmlinux file, obviously we can't fallback to another vmlinux (a
	 * x86_86 one, on the machine where analysis is being performed, say),
	 * or worse, /proc/kallsyms.
	 *
	 * If the specified file _has_ a build-id and there is a build-id
	 * section in the perf.data file, we will still do the expected
	 * validation in dso__load_vmlinux and will bail out if they don't
	 * match.
	 */
1916 1917 1918 1919 1920
	if (symbol_conf.kallsyms_name != NULL) {
		kallsyms_filename = symbol_conf.kallsyms_name;
		goto do_kallsyms;
	}

1921
	if (symbol_conf.vmlinux_name != NULL) {
1922
		err = dso__load_vmlinux(dso, map,
1923
					symbol_conf.vmlinux_name, filter);
1924
		if (err > 0) {
1925
			dso__set_long_name(dso,
1926 1927 1928 1929
					   strdup(symbol_conf.vmlinux_name));
			goto out_fixup;
		}
		return err;
1930
	}
1931 1932

	if (vmlinux_path != NULL) {
1933
		err = dso__load_vmlinux_path(dso, map, filter);
1934 1935
		if (err > 0)
			goto out_fixup;
1936 1937
	}

1938 1939 1940 1941
	/* do not try local files if a symfs was given */
	if (symbol_conf.symfs[0] != 0)
		return -1;

1942 1943 1944 1945 1946
	/*
	 * Say the kernel DSO was created when processing the build-id header table,
	 * we have a build-id, so check if it is the same as the running kernel,
	 * using it if it is.
	 */
1947
	if (dso->has_build_id) {
1948
		u8 kallsyms_build_id[BUILD_ID_SIZE];
1949
		char sbuild_id[BUILD_ID_SIZE * 2 + 1];
1950 1951

		if (sysfs__read_build_id("/sys/kernel/notes", kallsyms_build_id,
1952
					 sizeof(kallsyms_build_id)) == 0) {
1953
			if (dso__build_id_equal(dso, kallsyms_build_id)) {
1954
				kallsyms_filename = "/proc/kallsyms";
1955
				goto do_kallsyms;
1956
			}
1957
		}
1958 1959 1960 1961
		/*
		 * Now look if we have it on the build-id cache in
		 * $HOME/.debug/[kernel.kallsyms].
		 */
1962
		build_id__sprintf(dso->build_id, sizeof(dso->build_id),
1963 1964 1965 1966
				  sbuild_id);

		if (asprintf(&kallsyms_allocated_filename,
			     "%s/.debug/[kernel.kallsyms]/%s",
1967 1968
			     getenv("HOME"), sbuild_id) == -1) {
			pr_err("Not enough memory for kallsyms file lookup\n");
1969
			return -1;
1970
		}
1971

1972 1973
		kallsyms_filename = kallsyms_allocated_filename;

1974
		if (access(kallsyms_filename, F_OK)) {
1975 1976
			pr_err("No kallsyms or vmlinux with build-id %s "
			       "was found\n", sbuild_id);
1977
			free(kallsyms_allocated_filename);
1978
			return -1;
1979
		}
1980 1981 1982 1983 1984
	} else {
		/*
		 * Last resort, if we don't have a build-id and couldn't find
		 * any vmlinux file, try the running kernel kallsyms table.
		 */
1985 1986
		kallsyms_filename = "/proc/kallsyms";
	}
1987

1988
do_kallsyms:
1989
	err = dso__load_kallsyms(dso, kallsyms_filename, map, filter);
1990 1991
	if (err > 0)
		pr_debug("Using %s for symbols\n", kallsyms_filename);
1992
	free(kallsyms_allocated_filename);
1993 1994

	if (err > 0) {
1995
out_fixup:
1996
		if (kallsyms_filename != NULL)
1997
			dso__set_long_name(dso, strdup("[kernel.kallsyms]"));
1998 1999
		map__fixup_start(map);
		map__fixup_end(map);
2000
	}
2001

2002 2003 2004
	return err;
}

2005 2006
static int dso__load_guest_kernel_sym(struct dso *dso, struct map *map,
				      symbol_filter_t filter)
2007 2008 2009
{
	int err;
	const char *kallsyms_filename = NULL;
2010
	struct machine *machine;
2011 2012 2013 2014 2015 2016
	char path[PATH_MAX];

	if (!map->groups) {
		pr_debug("Guest kernel map hasn't the point to groups\n");
		return -1;
	}
2017
	machine = map->groups->machine;
2018

2019
	if (machine__is_default_guest(machine)) {
2020 2021 2022 2023 2024 2025
		/*
		 * if the user specified a vmlinux filename, use it and only
		 * it, reporting errors to the user if it cannot be used.
		 * Or use file guest_kallsyms inputted by user on commandline
		 */
		if (symbol_conf.default_guest_vmlinux_name != NULL) {
2026
			err = dso__load_vmlinux(dso, map,
2027 2028 2029 2030 2031 2032 2033 2034
				symbol_conf.default_guest_vmlinux_name, filter);
			goto out_try_fixup;
		}

		kallsyms_filename = symbol_conf.default_guest_kallsyms;
		if (!kallsyms_filename)
			return -1;
	} else {
2035
		sprintf(path, "%s/proc/kallsyms", machine->root_dir);
2036 2037 2038
		kallsyms_filename = path;
	}

2039
	err = dso__load_kallsyms(dso, kallsyms_filename, map, filter);
2040 2041 2042 2043 2044 2045
	if (err > 0)
		pr_debug("Using %s for symbols\n", kallsyms_filename);

out_try_fixup:
	if (err > 0) {
		if (kallsyms_filename != NULL) {
2046
			machine__mmap_name(machine, path, sizeof(path));
2047
			dso__set_long_name(dso, strdup(path));
2048 2049 2050 2051 2052 2053 2054
		}
		map__fixup_start(map);
		map__fixup_end(map);
	}

	return err;
}
2055

2056
static void dsos__add(struct list_head *head, struct dso *dso)
2057
{
2058
	list_add_tail(&dso->node, head);
2059 2060
}

2061
static struct dso *dsos__find(struct list_head *head, const char *name)
2062 2063 2064
{
	struct dso *pos;

2065
	list_for_each_entry(pos, head, node)
2066
		if (strcmp(pos->long_name, name) == 0)
2067 2068 2069 2070
			return pos;
	return NULL;
}

2071
struct dso *__dsos__findnew(struct list_head *head, const char *name)
2072
{
2073
	struct dso *dso = dsos__find(head, name);
2074

2075
	if (!dso) {
2076
		dso = dso__new(name);
2077
		if (dso != NULL) {
2078
			dsos__add(head, dso);
2079 2080
			dso__set_basename(dso);
		}
2081
	}
2082 2083 2084 2085

	return dso;
}

2086
size_t __dsos__fprintf(struct list_head *head, FILE *fp)
2087 2088
{
	struct dso *pos;
2089
	size_t ret = 0;
2090

2091 2092 2093
	list_for_each_entry(pos, head, node) {
		int i;
		for (i = 0; i < MAP__NR_TYPES; ++i)
2094
			ret += dso__fprintf(pos, i, fp);
2095
	}
2096 2097

	return ret;
2098 2099
}

2100
size_t machines__fprintf_dsos(struct rb_root *machines, FILE *fp)
2101
{
2102
	struct rb_node *nd;
2103
	size_t ret = 0;
2104

2105
	for (nd = rb_first(machines); nd; nd = rb_next(nd)) {
2106
		struct machine *pos = rb_entry(nd, struct machine, rb_node);
2107 2108
		ret += __dsos__fprintf(&pos->kernel_dsos, fp);
		ret += __dsos__fprintf(&pos->user_dsos, fp);
2109
	}
2110 2111

	return ret;
2112 2113
}

2114 2115
static size_t __dsos__fprintf_buildid(struct list_head *head, FILE *fp,
				      bool with_hits)
2116 2117 2118 2119
{
	struct dso *pos;
	size_t ret = 0;

2120
	list_for_each_entry(pos, head, node) {
2121 2122
		if (with_hits && !pos->hit)
			continue;
2123
		ret += dso__fprintf_buildid(pos, fp);
2124
		ret += fprintf(fp, " %s\n", pos->long_name);
2125 2126 2127 2128
	}
	return ret;
}

2129 2130
size_t machine__fprintf_dsos_buildid(struct machine *machine, FILE *fp,
				     bool with_hits)
2131
{
2132 2133
	return __dsos__fprintf_buildid(&machine->kernel_dsos, fp, with_hits) +
	       __dsos__fprintf_buildid(&machine->user_dsos, fp, with_hits);
2134 2135
}

2136 2137
size_t machines__fprintf_dsos_buildid(struct rb_root *machines,
				      FILE *fp, bool with_hits)
2138
{
2139 2140 2141
	struct rb_node *nd;
	size_t ret = 0;

2142
	for (nd = rb_first(machines); nd; nd = rb_next(nd)) {
2143
		struct machine *pos = rb_entry(nd, struct machine, rb_node);
2144
		ret += machine__fprintf_dsos_buildid(pos, fp, with_hits);
2145 2146
	}
	return ret;
2147 2148
}

2149 2150
struct dso *dso__new_kernel(const char *name)
{
2151
	struct dso *dso = dso__new(name ?: "[kernel.kallsyms]");
2152

2153 2154 2155
	if (dso != NULL) {
		dso__set_short_name(dso, "[kernel]");
		dso->kernel = DSO_TYPE_KERNEL;
2156 2157
	}

2158
	return dso;
2159 2160
}

2161
static struct dso *dso__new_guest_kernel(struct machine *machine,
2162 2163
					const char *name)
{
2164
	char bf[PATH_MAX];
2165 2166 2167 2168 2169
	struct dso *dso = dso__new(name ?: machine__mmap_name(machine, bf,
							      sizeof(bf)));
	if (dso != NULL) {
		dso__set_short_name(dso, "[guest.kernel]");
		dso->kernel = DSO_TYPE_GUEST_KERNEL;
2170 2171
	}

2172
	return dso;
2173 2174
}

2175
void dso__read_running_kernel_build_id(struct dso *dso, struct machine *machine)
2176
{
2177 2178
	char path[PATH_MAX];

2179
	if (machine__is_default_guest(machine))
2180
		return;
2181
	sprintf(path, "%s/sys/kernel/notes", machine->root_dir);
2182 2183 2184
	if (sysfs__read_build_id(path, dso->build_id,
				 sizeof(dso->build_id)) == 0)
		dso->has_build_id = true;
2185 2186
}

2187
static struct dso *machine__create_kernel(struct machine *machine)
2188
{
2189 2190
	const char *vmlinux_name = NULL;
	struct dso *kernel;
2191

2192
	if (machine__is_host(machine)) {
2193 2194 2195
		vmlinux_name = symbol_conf.vmlinux_name;
		kernel = dso__new_kernel(vmlinux_name);
	} else {
2196
		if (machine__is_default_guest(machine))
2197
			vmlinux_name = symbol_conf.default_guest_vmlinux_name;
2198
		kernel = dso__new_guest_kernel(machine, vmlinux_name);
2199
	}
2200

2201
	if (kernel != NULL) {
2202 2203
		dso__read_running_kernel_build_id(kernel, machine);
		dsos__add(&machine->kernel_dsos, kernel);
2204
	}
2205 2206 2207
	return kernel;
}

2208 2209 2210 2211 2212
struct process_args {
	u64 start;
};

static int symbol__in_kernel(void *arg, const char *name,
2213
			     char type __used, u64 start, u64 end __used)
2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247
{
	struct process_args *args = arg;

	if (strchr(name, '['))
		return 0;

	args->start = start;
	return 1;
}

/* Figure out the start address of kernel map from /proc/kallsyms */
static u64 machine__get_kernel_start_addr(struct machine *machine)
{
	const char *filename;
	char path[PATH_MAX];
	struct process_args args;

	if (machine__is_host(machine)) {
		filename = "/proc/kallsyms";
	} else {
		if (machine__is_default_guest(machine))
			filename = (char *)symbol_conf.default_guest_kallsyms;
		else {
			sprintf(path, "%s/proc/kallsyms", machine->root_dir);
			filename = path;
		}
	}

	if (kallsyms__parse(filename, &args, symbol__in_kernel) <= 0)
		return 0;

	return args.start;
}

2248
int __machine__create_kernel_maps(struct machine *machine, struct dso *kernel)
2249
{
2250
	enum map_type type;
2251
	u64 start = machine__get_kernel_start_addr(machine);
2252

2253
	for (type = 0; type < MAP__NR_TYPES; ++type) {
2254 2255
		struct kmap *kmap;

2256 2257
		machine->vmlinux_maps[type] = map__new2(start, kernel, type);
		if (machine->vmlinux_maps[type] == NULL)
2258
			return -1;
2259

2260 2261 2262 2263 2264 2265 2266
		machine->vmlinux_maps[type]->map_ip =
			machine->vmlinux_maps[type]->unmap_ip =
				identity__map_ip;
		kmap = map__kmap(machine->vmlinux_maps[type]);
		kmap->kmaps = &machine->kmaps;
		map_groups__insert(&machine->kmaps,
				   machine->vmlinux_maps[type]);
2267 2268 2269
	}

	return 0;
2270 2271
}

2272
void machine__destroy_kernel_maps(struct machine *machine)
2273 2274 2275 2276 2277 2278
{
	enum map_type type;

	for (type = 0; type < MAP__NR_TYPES; ++type) {
		struct kmap *kmap;

2279
		if (machine->vmlinux_maps[type] == NULL)
2280 2281
			continue;

2282 2283 2284
		kmap = map__kmap(machine->vmlinux_maps[type]);
		map_groups__remove(&machine->kmaps,
				   machine->vmlinux_maps[type]);
2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297
		if (kmap->ref_reloc_sym) {
			/*
			 * ref_reloc_sym is shared among all maps, so free just
			 * on one of them.
			 */
			if (type == MAP__FUNCTION) {
				free((char *)kmap->ref_reloc_sym->name);
				kmap->ref_reloc_sym->name = NULL;
				free(kmap->ref_reloc_sym);
			}
			kmap->ref_reloc_sym = NULL;
		}

2298 2299
		map__delete(machine->vmlinux_maps[type]);
		machine->vmlinux_maps[type] = NULL;
2300 2301 2302
	}
}

2303
int machine__create_kernel_maps(struct machine *machine)
2304
{
2305
	struct dso *kernel = machine__create_kernel(machine);
2306 2307

	if (kernel == NULL ||
2308
	    __machine__create_kernel_maps(machine, kernel) < 0)
2309 2310
		return -1;

2311
	if (symbol_conf.use_modules && machine__create_modules(machine) < 0)
2312 2313 2314 2315
		pr_debug("Problems creating module maps, continuing anyway...\n");
	/*
	 * Now that we have all the maps created, just set the ->end of them:
	 */
2316
	map_groups__fixup_end(&machine->kmaps);
2317 2318 2319
	return 0;
}

2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347
static void vmlinux_path__exit(void)
{
	while (--vmlinux_path__nr_entries >= 0) {
		free(vmlinux_path[vmlinux_path__nr_entries]);
		vmlinux_path[vmlinux_path__nr_entries] = NULL;
	}

	free(vmlinux_path);
	vmlinux_path = NULL;
}

static int vmlinux_path__init(void)
{
	struct utsname uts;
	char bf[PATH_MAX];

	vmlinux_path = malloc(sizeof(char *) * 5);
	if (vmlinux_path == NULL)
		return -1;

	vmlinux_path[vmlinux_path__nr_entries] = strdup("vmlinux");
	if (vmlinux_path[vmlinux_path__nr_entries] == NULL)
		goto out_fail;
	++vmlinux_path__nr_entries;
	vmlinux_path[vmlinux_path__nr_entries] = strdup("/boot/vmlinux");
	if (vmlinux_path[vmlinux_path__nr_entries] == NULL)
		goto out_fail;
	++vmlinux_path__nr_entries;
2348 2349 2350 2351 2352 2353 2354 2355

	/* only try running kernel version if no symfs was given */
	if (symbol_conf.symfs[0] != 0)
		return 0;

	if (uname(&uts) < 0)
		return -1;

2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379
	snprintf(bf, sizeof(bf), "/boot/vmlinux-%s", uts.release);
	vmlinux_path[vmlinux_path__nr_entries] = strdup(bf);
	if (vmlinux_path[vmlinux_path__nr_entries] == NULL)
		goto out_fail;
	++vmlinux_path__nr_entries;
	snprintf(bf, sizeof(bf), "/lib/modules/%s/build/vmlinux", uts.release);
	vmlinux_path[vmlinux_path__nr_entries] = strdup(bf);
	if (vmlinux_path[vmlinux_path__nr_entries] == NULL)
		goto out_fail;
	++vmlinux_path__nr_entries;
	snprintf(bf, sizeof(bf), "/usr/lib/debug/lib/modules/%s/vmlinux",
		 uts.release);
	vmlinux_path[vmlinux_path__nr_entries] = strdup(bf);
	if (vmlinux_path[vmlinux_path__nr_entries] == NULL)
		goto out_fail;
	++vmlinux_path__nr_entries;

	return 0;

out_fail:
	vmlinux_path__exit();
	return -1;
}

2380
size_t machine__fprintf_vmlinux_path(struct machine *machine, FILE *fp)
2381 2382 2383
{
	int i;
	size_t printed = 0;
2384
	struct dso *kdso = machine->vmlinux_maps[MAP__FUNCTION]->dso;
2385 2386 2387 2388 2389 2390

	if (kdso->has_build_id) {
		char filename[PATH_MAX];
		if (dso__build_id_filename(kdso, filename, sizeof(filename)))
			printed += fprintf(fp, "[0] %s\n", filename);
	}
2391 2392

	for (i = 0; i < vmlinux_path__nr_entries; ++i)
2393 2394
		printed += fprintf(fp, "[%d] %s\n",
				   i + kdso->has_build_id, vmlinux_path[i]);
2395 2396 2397 2398

	return printed;
}

2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412
static int setup_list(struct strlist **list, const char *list_str,
		      const char *list_name)
{
	if (list_str == NULL)
		return 0;

	*list = strlist__new(true, list_str);
	if (!*list) {
		pr_err("problems parsing %s list\n", list_name);
		return -1;
	}
	return 0;
}

2413
int symbol__init(void)
2414
{
2415 2416
	const char *symfs;

2417 2418 2419
	if (symbol_conf.initialized)
		return 0;

2420 2421
	symbol_conf.priv_size = ALIGN(symbol_conf.priv_size, sizeof(u64));

2422
	elf_version(EV_CURRENT);
2423 2424 2425
	if (symbol_conf.sort_by_name)
		symbol_conf.priv_size += (sizeof(struct symbol_name_rb_node) -
					  sizeof(struct symbol));
2426

2427
	if (symbol_conf.try_vmlinux_path && vmlinux_path__init() < 0)
2428 2429
		return -1;

2430 2431 2432 2433 2434
	if (symbol_conf.field_sep && *symbol_conf.field_sep == '.') {
		pr_err("'.' is the only non valid --field-separator argument\n");
		return -1;
	}

2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446
	if (setup_list(&symbol_conf.dso_list,
		       symbol_conf.dso_list_str, "dso") < 0)
		return -1;

	if (setup_list(&symbol_conf.comm_list,
		       symbol_conf.comm_list_str, "comm") < 0)
		goto out_free_dso_list;

	if (setup_list(&symbol_conf.sym_list,
		       symbol_conf.sym_list_str, "symbol") < 0)
		goto out_free_comm_list;

2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458
	/*
	 * A path to symbols of "/" is identical to ""
	 * reset here for simplicity.
	 */
	symfs = realpath(symbol_conf.symfs, NULL);
	if (symfs == NULL)
		symfs = symbol_conf.symfs;
	if (strcmp(symfs, "/") == 0)
		symbol_conf.symfs = "";
	if (symfs != symbol_conf.symfs)
		free((void *)symfs);

2459
	symbol_conf.initialized = true;
2460
	return 0;
2461 2462 2463 2464 2465 2466

out_free_dso_list:
	strlist__delete(symbol_conf.dso_list);
out_free_comm_list:
	strlist__delete(symbol_conf.comm_list);
	return -1;
2467 2468
}

2469 2470
void symbol__exit(void)
{
2471 2472
	if (!symbol_conf.initialized)
		return;
2473 2474 2475 2476 2477
	strlist__delete(symbol_conf.sym_list);
	strlist__delete(symbol_conf.dso_list);
	strlist__delete(symbol_conf.comm_list);
	vmlinux_path__exit();
	symbol_conf.sym_list = symbol_conf.dso_list = symbol_conf.comm_list = NULL;
2478
	symbol_conf.initialized = false;
2479 2480
}

2481
int machines__create_kernel_maps(struct rb_root *machines, pid_t pid)
2482
{
2483
	struct machine *machine = machines__findnew(machines, pid);
2484

2485
	if (machine == NULL)
2486
		return -1;
2487

2488
	return machine__create_kernel_maps(machine);
2489
}
2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532

static int hex(char ch)
{
	if ((ch >= '0') && (ch <= '9'))
		return ch - '0';
	if ((ch >= 'a') && (ch <= 'f'))
		return ch - 'a' + 10;
	if ((ch >= 'A') && (ch <= 'F'))
		return ch - 'A' + 10;
	return -1;
}

/*
 * While we find nice hex chars, build a long_val.
 * Return number of chars processed.
 */
int hex2u64(const char *ptr, u64 *long_val)
{
	const char *p = ptr;
	*long_val = 0;

	while (*p) {
		const int hex_val = hex(*p);

		if (hex_val < 0)
			break;

		*long_val = (*long_val << 4) | hex_val;
		p++;
	}

	return p - ptr;
}

char *strxfrchar(char *s, char from, char to)
{
	char *p = s;

	while ((p = strchr(p, from)) != NULL)
		*p++ = to;

	return s;
}
2533

2534
int machines__create_guest_kernel_maps(struct rb_root *machines)
2535 2536 2537 2538 2539 2540 2541 2542 2543 2544
{
	int ret = 0;
	struct dirent **namelist = NULL;
	int i, items = 0;
	char path[PATH_MAX];
	pid_t pid;

	if (symbol_conf.default_guest_vmlinux_name ||
	    symbol_conf.default_guest_modules ||
	    symbol_conf.default_guest_kallsyms) {
2545
		machines__create_kernel_maps(machines, DEFAULT_GUEST_KERNEL_ID);
2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565
	}

	if (symbol_conf.guestmount) {
		items = scandir(symbol_conf.guestmount, &namelist, NULL, NULL);
		if (items <= 0)
			return -ENOENT;
		for (i = 0; i < items; i++) {
			if (!isdigit(namelist[i]->d_name[0])) {
				/* Filter out . and .. */
				continue;
			}
			pid = atoi(namelist[i]->d_name);
			sprintf(path, "%s/%s/proc/kallsyms",
				symbol_conf.guestmount,
				namelist[i]->d_name);
			ret = access(path, R_OK);
			if (ret) {
				pr_debug("Can't access file %s\n", path);
				goto failure;
			}
2566
			machines__create_kernel_maps(machines, pid);
2567 2568 2569 2570 2571 2572 2573
		}
failure:
		free(namelist);
	}

	return ret;
}
2574

2575
void machines__destroy_guest_kernel_maps(struct rb_root *machines)
2576
{
2577
	struct rb_node *next = rb_first(machines);
2578 2579 2580 2581 2582

	while (next) {
		struct machine *pos = rb_entry(next, struct machine, rb_node);

		next = rb_next(&pos->rb_node);
2583
		rb_erase(&pos->rb_node, machines);
2584 2585 2586 2587
		machine__delete(pos);
	}
}

2588
int machine__load_kallsyms(struct machine *machine, const char *filename,
2589 2590
			   enum map_type type, symbol_filter_t filter)
{
2591
	struct map *map = machine->vmlinux_maps[type];
2592 2593 2594 2595 2596 2597 2598 2599 2600
	int ret = dso__load_kallsyms(map->dso, filename, map, filter);

	if (ret > 0) {
		dso__set_loaded(map->dso, type);
		/*
		 * Since /proc/kallsyms will have multiple sessions for the
		 * kernel, with modules between them, fixup the end of all
		 * sections.
		 */
2601
		__map_groups__fixup_end(&machine->kmaps, type);
2602 2603 2604 2605 2606
	}

	return ret;
}

2607
int machine__load_vmlinux_path(struct machine *machine, enum map_type type,
2608 2609
			       symbol_filter_t filter)
{
2610
	struct map *map = machine->vmlinux_maps[type];
2611 2612 2613 2614 2615 2616 2617 2618 2619
	int ret = dso__load_vmlinux_path(map->dso, map, filter);

	if (ret > 0) {
		dso__set_loaded(map->dso, type);
		map__reloc_vmlinux(map);
	}

	return ret;
}