symbol.c 63.0 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12
#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>
13
#include <inttypes.h>
14
#include "build-id.h"
15
#include "debug.h"
16
#include "symbol.h"
17
#include "strlist.h"
18 19 20 21

#include <libelf.h>
#include <gelf.h>
#include <elf.h>
22
#include <limits.h>
23
#include <sys/utsname.h>
P
Peter Zijlstra 已提交
24

25
#ifndef KSYM_NAME_LEN
26
#define KSYM_NAME_LEN 256
27 28
#endif

29 30 31 32
#ifndef NT_GNU_BUILD_ID
#define NT_GNU_BUILD_ID 3
#endif

33
static bool dso__build_id_equal(const struct dso *dso, u8 *build_id);
34
static int elf_read_build_id(Elf *elf, void *bf, size_t size);
35
static void dsos__add(struct list_head *head, struct dso *dso);
36
static struct map *map__new2(u64 start, struct dso *dso, enum map_type type);
37
static int dso__load_kernel_sym(struct dso *dso, struct map *map,
38
				symbol_filter_t filter);
39
static int dso__load_guest_kernel_sym(struct dso *dso, struct map *map,
40
			symbol_filter_t filter);
41 42
static int vmlinux_path__nr_entries;
static char **vmlinux_path;
43

44
struct symbol_conf symbol_conf = {
45
	.exclude_other	  = true,
46 47
	.use_modules	  = true,
	.try_vmlinux_path = true,
48
	.annotate_src	  = 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
	symbol_type = toupper(symbol_type);

79 80 81
	switch (map_type) {
	case MAP__FUNCTION:
		return symbol_type == 'T' || symbol_type == 'W';
82
	case MAP__VARIABLE:
83
		return symbol_type == 'D';
84 85 86 87 88
	default:
		return false;
	}
}

89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174
static int prefix_underscores_count(const char *str)
{
	const char *tail = str;

	while (*tail == '_')
		tail++;

	return tail - str;
}

#define SYMBOL_A 0
#define SYMBOL_B 1

static int choose_best_symbol(struct symbol *syma, struct symbol *symb)
{
	s64 a;
	s64 b;

	/* Prefer a symbol with non zero length */
	a = syma->end - syma->start;
	b = symb->end - symb->start;
	if ((b == 0) && (a > 0))
		return SYMBOL_A;
	else if ((a == 0) && (b > 0))
		return SYMBOL_B;

	/* Prefer a non weak symbol over a weak one */
	a = syma->binding == STB_WEAK;
	b = symb->binding == STB_WEAK;
	if (b && !a)
		return SYMBOL_A;
	if (a && !b)
		return SYMBOL_B;

	/* Prefer a global symbol over a non global one */
	a = syma->binding == STB_GLOBAL;
	b = symb->binding == STB_GLOBAL;
	if (a && !b)
		return SYMBOL_A;
	if (b && !a)
		return SYMBOL_B;

	/* Prefer a symbol with less underscores */
	a = prefix_underscores_count(syma->name);
	b = prefix_underscores_count(symb->name);
	if (b > a)
		return SYMBOL_A;
	else if (a > b)
		return SYMBOL_B;

	/* If all else fails, choose the symbol with the longest name */
	if (strlen(syma->name) >= strlen(symb->name))
		return SYMBOL_A;
	else
		return SYMBOL_B;
}

static void symbols__fixup_duplicate(struct rb_root *symbols)
{
	struct rb_node *nd;
	struct symbol *curr, *next;

	nd = rb_first(symbols);

	while (nd) {
		curr = rb_entry(nd, struct symbol, rb_node);
again:
		nd = rb_next(&curr->rb_node);
		next = rb_entry(nd, struct symbol, rb_node);

		if (!nd)
			break;

		if (curr->start != next->start)
			continue;

		if (choose_best_symbol(curr, next) == SYMBOL_A) {
			rb_erase(&next->rb_node, symbols);
			goto again;
		} else {
			nd = rb_next(&curr->rb_node);
			rb_erase(&curr->rb_node, symbols);
		}
	}
}

175
static void symbols__fixup_end(struct rb_root *symbols)
176
{
177
	struct rb_node *nd, *prevnd = rb_first(symbols);
178
	struct symbol *curr, *prev;
179 180 181 182

	if (prevnd == NULL)
		return;

183 184
	curr = rb_entry(prevnd, struct symbol, rb_node);

185
	for (nd = rb_next(prevnd); nd; nd = rb_next(nd)) {
186 187
		prev = curr;
		curr = rb_entry(nd, struct symbol, rb_node);
188

189
		if (prev->end == prev->start && prev->end != curr->start)
190 191
			prev->end = curr->start - 1;
	}
192 193 194 195

	/* Last entry */
	if (curr->end == curr->start)
		curr->end = roundup(curr->start, 4096);
196 197
}

198
static void __map_groups__fixup_end(struct map_groups *mg, enum map_type type)
199 200
{
	struct map *prev, *curr;
201
	struct rb_node *nd, *prevnd = rb_first(&mg->maps[type]);
202 203 204 205 206 207 208 209 210 211

	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;
212
	}
213 214 215 216 217

	/*
	 * We still haven't the actual symbols, so guess the
	 * last map final address.
	 */
218
	curr->end = ~0ULL;
219 220
}

221
static void map_groups__fixup_end(struct map_groups *mg)
222 223 224
{
	int i;
	for (i = 0; i < MAP__NR_TYPES; ++i)
225
		__map_groups__fixup_end(mg, i);
226 227
}

228 229
static struct symbol *symbol__new(u64 start, u64 len, u8 binding,
				  const char *name)
230
{
231
	size_t namelen = strlen(name) + 1;
232 233 234
	struct symbol *sym = calloc(1, (symbol_conf.priv_size +
					sizeof(*sym) + namelen));
	if (sym == NULL)
235 236
		return NULL;

237
	if (symbol_conf.priv_size)
238
		sym = ((void *)sym) + symbol_conf.priv_size;
239

240 241 242 243
	sym->start   = start;
	sym->end     = len ? start + len - 1 : start;
	sym->binding = binding;
	sym->namelen = namelen - 1;
244

245 246 247
	pr_debug4("%s: %s %#" PRIx64 "-%#" PRIx64 "\n",
		  __func__, name, start, sym->end);
	memcpy(sym->name, name, namelen);
248

249
	return sym;
250 251
}

252
void symbol__delete(struct symbol *sym)
253
{
254
	free(((void *)sym) - symbol_conf.priv_size);
255 256
}

257
static size_t symbol__fprintf(struct symbol *sym, FILE *fp)
258
{
259
	return fprintf(fp, " %" PRIx64 "-%" PRIx64 " %c %s\n",
260 261 262 263
		       sym->start, sym->end,
		       sym->binding == STB_GLOBAL ? 'g' :
		       sym->binding == STB_LOCAL  ? 'l' : 'w',
		       sym->name);
264 265
}

266 267
size_t symbol__fprintf_symname_offs(const struct symbol *sym,
				    const struct addr_location *al, FILE *fp)
268
{
269 270 271 272 273 274 275 276 277 278 279 280 281
	unsigned long offset;
	size_t length;

	if (sym && sym->name) {
		length = fprintf(fp, "%s", sym->name);
		if (al) {
			offset = al->addr - sym->start;
			length += fprintf(fp, "+0x%lx", offset);
		}
		return length;
	} else
		return fprintf(fp, "[unknown]");
}
282

283 284 285
size_t symbol__fprintf_symname(const struct symbol *sym, FILE *fp)
{
	return symbol__fprintf_symname_offs(sym, NULL, fp);
286 287
}

288
void dso__set_long_name(struct dso *dso, char *name)
289
{
290 291
	if (name == NULL)
		return;
292 293
	dso->long_name = name;
	dso->long_name_len = strlen(name);
294 295
}

296
static void dso__set_short_name(struct dso *dso, const char *name)
297 298 299
{
	if (name == NULL)
		return;
300 301
	dso->short_name = name;
	dso->short_name_len = strlen(name);
302 303
}

304
static void dso__set_basename(struct dso *dso)
305
{
306
	dso__set_short_name(dso, basename(dso->long_name));
307 308
}

309
struct dso *dso__new(const char *name)
310
{
311
	struct dso *dso = calloc(1, sizeof(*dso) + strlen(name) + 1);
312

313
	if (dso != NULL) {
314
		int i;
315 316 317
		strcpy(dso->name, name);
		dso__set_long_name(dso, dso->name);
		dso__set_short_name(dso, dso->name);
318
		for (i = 0; i < MAP__NR_TYPES; ++i)
319 320 321 322 323 324 325
			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);
326 327
	}

328
	return dso;
329 330
}

331
static void symbols__delete(struct rb_root *symbols)
332 333
{
	struct symbol *pos;
334
	struct rb_node *next = rb_first(symbols);
335 336 337 338

	while (next) {
		pos = rb_entry(next, struct symbol, rb_node);
		next = rb_next(&pos->rb_node);
339
		rb_erase(&pos->rb_node, symbols);
340
		symbol__delete(pos);
341 342 343
	}
}

344
void dso__delete(struct dso *dso)
345
{
346 347
	int i;
	for (i = 0; i < MAP__NR_TYPES; ++i)
348 349 350 351 352 353
		symbols__delete(&dso->symbols[i]);
	if (dso->sname_alloc)
		free((char *)dso->short_name);
	if (dso->lname_alloc)
		free(dso->long_name);
	free(dso);
354 355
}

356
void dso__set_build_id(struct dso *dso, void *build_id)
357
{
358 359
	memcpy(dso->build_id, build_id, sizeof(dso->build_id));
	dso->has_build_id = 1;
360 361
}

362
static void symbols__insert(struct rb_root *symbols, struct symbol *sym)
363
{
364
	struct rb_node **p = &symbols->rb_node;
365
	struct rb_node *parent = NULL;
366
	const u64 ip = sym->start;
367 368 369 370 371 372 373 374 375 376 377
	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);
378
	rb_insert_color(&sym->rb_node, symbols);
379 380
}

381
static struct symbol *symbols__find(struct rb_root *symbols, u64 ip)
382 383 384
{
	struct rb_node *n;

385
	if (symbols == NULL)
386 387
		return NULL;

388
	n = symbols->rb_node;
389 390 391 392 393 394 395 396 397 398 399 400 401 402 403

	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;
}

404 405 406 407 408
struct symbol_name_rb_node {
	struct rb_node	rb_node;
	struct symbol	sym;
};

409
static void symbols__insert_by_name(struct rb_root *symbols, struct symbol *sym)
410
{
411
	struct rb_node **p = &symbols->rb_node;
412
	struct rb_node *parent = NULL;
413 414 415
	struct symbol_name_rb_node *symn, *s;

	symn = container_of(sym, struct symbol_name_rb_node, sym);
416 417 418 419 420 421 422 423 424 425

	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);
426
	rb_insert_color(&symn->rb_node, symbols);
427 428
}

429 430
static void symbols__sort_by_name(struct rb_root *symbols,
				  struct rb_root *source)
431 432 433 434 435
{
	struct rb_node *nd;

	for (nd = rb_first(source); nd; nd = rb_next(nd)) {
		struct symbol *pos = rb_entry(nd, struct symbol, rb_node);
436
		symbols__insert_by_name(symbols, pos);
437 438 439
	}
}

440 441
static struct symbol *symbols__find_by_name(struct rb_root *symbols,
					    const char *name)
442 443 444
{
	struct rb_node *n;

445
	if (symbols == NULL)
446 447
		return NULL;

448
	n = symbols->rb_node;
449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467

	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;
}

468
struct symbol *dso__find_symbol(struct dso *dso,
469
				enum map_type type, u64 addr)
470
{
471
	return symbols__find(&dso->symbols[type], addr);
472 473
}

474
struct symbol *dso__find_symbol_by_name(struct dso *dso, enum map_type type,
475 476
					const char *name)
{
477
	return symbols__find_by_name(&dso->symbol_names[type], name);
478 479
}

480
void dso__sort_by_name(struct dso *dso, enum map_type type)
481
{
482 483 484
	dso__set_sorted_by_name(dso, type);
	return symbols__sort_by_name(&dso->symbol_names[type],
				     &dso->symbols[type]);
485 486
}

487
int build_id__sprintf(const u8 *build_id, int len, char *bf)
488
{
489
	char *bid = bf;
490
	const u8 *raw = build_id;
491
	int i;
492

493 494 495 496 497 498
	for (i = 0; i < len; ++i) {
		sprintf(bid, "%02x", *raw);
		++raw;
		bid += 2;
	}

499
	return raw - build_id;
500 501
}

502
size_t dso__fprintf_buildid(struct dso *dso, FILE *fp)
503 504 505
{
	char sbuild_id[BUILD_ID_SIZE * 2 + 1];

506
	build_id__sprintf(dso->build_id, sizeof(dso->build_id), sbuild_id);
507 508 509
	return fprintf(fp, "%s", sbuild_id);
}

510 511
size_t dso__fprintf_symbols_by_name(struct dso *dso,
				    enum map_type type, FILE *fp)
512 513 514 515 516
{
	size_t ret = 0;
	struct rb_node *nd;
	struct symbol_name_rb_node *pos;

517
	for (nd = rb_first(&dso->symbol_names[type]); nd; nd = rb_next(nd)) {
518 519 520 521 522 523 524
		pos = rb_entry(nd, struct symbol_name_rb_node, rb_node);
		fprintf(fp, "%s\n", pos->sym.name);
	}

	return ret;
}

525
size_t dso__fprintf(struct dso *dso, enum map_type type, FILE *fp)
526 527
{
	struct rb_node *nd;
528
	size_t ret = fprintf(fp, "dso: %s (", dso->short_name);
529

530 531
	if (dso->short_name != dso->long_name)
		ret += fprintf(fp, "%s, ", dso->long_name);
532
	ret += fprintf(fp, "%s, %sloaded, ", map_type__name[type],
533 534
		       dso->loaded ? "" : "NOT ");
	ret += dso__fprintf_buildid(dso, fp);
535
	ret += fprintf(fp, ")\n");
536
	for (nd = rb_first(&dso->symbols[type]); nd; nd = rb_next(nd)) {
537 538
		struct symbol *pos = rb_entry(nd, struct symbol, rb_node);
		ret += symbol__fprintf(pos, fp);
539 540 541 542 543
	}

	return ret;
}

544 545
int kallsyms__parse(const char *filename, void *arg,
		    int (*process_symbol)(void *arg, const char *name,
546
					  char type, u64 start, u64 end))
547 548 549
{
	char *line = NULL;
	size_t n;
550
	int err = -1;
551
	FILE *file = fopen(filename, "r");
552 553 554 555

	if (file == NULL)
		goto out_failure;

556 557
	err = 0;

558
	while (!feof(file)) {
559
		u64 start;
560 561
		int line_len, len;
		char symbol_type;
562
		char *symbol_name;
563 564

		line_len = getline(&line, &n, file);
565
		if (line_len < 0 || !line)
566 567 568 569
			break;

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

570
		len = hex2u64(line, &start);
571 572 573 574 575

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

576
		symbol_type = line[len];
577 578 579
		len += 2;
		symbol_name = line + len;
		len = line_len - len;
580

581 582
		if (len >= KSYM_NAME_LEN) {
			err = -1;
583
			break;
584 585
		}

586 587 588 589 590 591 592 593 594
		/*
		 * module symbols are not sorted so we add all
		 * symbols with zero length and rely on
		 * symbols__fixup_end() to fix it up.
		 */
		err = process_symbol(arg, symbol_name,
				     symbol_type, start, start);
		if (err)
			break;
595 596 597 598
	}

	free(line);
	fclose(file);
599
	return err;
600 601 602 603 604

out_failure:
	return -1;
}

605 606 607 608 609
struct process_kallsyms_args {
	struct map *map;
	struct dso *dso;
};

610 611 612 613 614 615 616 617
static u8 kallsyms2elf_type(char type)
{
	if (type == 'W')
		return STB_WEAK;

	return isupper(type) ? STB_GLOBAL : STB_LOCAL;
}

618
static int map__process_kallsym_symbol(void *arg, const char *name,
619
				       char type, u64 start, u64 end)
620 621 622 623 624 625 626 627
{
	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;

628 629
	sym = symbol__new(start, end - start + 1,
			  kallsyms2elf_type(type), name);
630 631 632 633 634 635 636
	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);
637

638 639 640 641 642 643 644 645
	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.
 */
646
static int dso__load_all_kallsyms(struct dso *dso, const char *filename,
647
				  struct map *map)
648
{
649
	struct process_kallsyms_args args = { .map = map, .dso = dso, };
650
	return kallsyms__parse(filename, &args, map__process_kallsym_symbol);
651 652
}

653 654 655 656 657
/*
 * 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.
 */
658
static int dso__split_kallsyms(struct dso *dso, struct map *map,
659
			       symbol_filter_t filter)
660
{
661
	struct map_groups *kmaps = map__kmap(map)->kmaps;
662
	struct machine *machine = kmaps->machine;
663
	struct map *curr_map = map;
664
	struct symbol *pos;
665
	int count = 0, moved = 0;	
666
	struct rb_root *root = &dso->symbols[map->type];
667
	struct rb_node *next = rb_first(root);
668 669 670 671 672 673 674 675 676 677
	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) {
678
			if (!symbol_conf.use_modules)
679 680
				goto discard_symbol;

681 682
			*module++ = '\0';

683
			if (strcmp(curr_map->dso->short_name, module)) {
684
				if (curr_map != map &&
685
				    dso->kernel == DSO_TYPE_GUEST_KERNEL &&
686
				    machine__is_default_guest(machine)) {
687 688 689 690 691 692 693 694 695 696 697 698 699
					/*
					 * 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);
700
				if (curr_map == NULL) {
701
					pr_debug("%s/proc/{kallsyms,modules} "
702
					         "inconsistency while looking "
703
						 "for \"%s\" module!\n",
704
						 machine->root_dir, module);
705 706
					curr_map = map;
					goto discard_symbol;
707
				}
708

709
				if (curr_map->dso->loaded &&
710
				    !machine__is_default_guest(machine))
711
					goto discard_symbol;
712
			}
713 714 715 716
			/*
			 * So that we look just like we get from .ko files,
			 * i.e. not prelinked, relative to map->start.
			 */
717 718 719
			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) {
720
			char dso_name[PATH_MAX];
721
			struct dso *ndso;
722

723 724 725 726 727
			if (count == 0) {
				curr_map = map;
				goto filter_symbol;
			}

728
			if (dso->kernel == DSO_TYPE_GUEST_KERNEL)
729 730 731 732 733 734 735
				snprintf(dso_name, sizeof(dso_name),
					"[guest.kernel].%d",
					kernel_range++);
			else
				snprintf(dso_name, sizeof(dso_name),
					"[kernel].%d",
					kernel_range++);
736

737 738
			ndso = dso__new(dso_name);
			if (ndso == NULL)
739 740
				return -1;

741
			ndso->kernel = dso->kernel;
742

743
			curr_map = map__new2(pos->start, ndso, map->type);
744
			if (curr_map == NULL) {
745
				dso__delete(ndso);
746 747
				return -1;
			}
748

749
			curr_map->map_ip = curr_map->unmap_ip = identity__map_ip;
750
			map_groups__insert(kmaps, curr_map);
751 752
			++kernel_range;
		}
753
filter_symbol:
754
		if (filter && filter(curr_map, pos)) {
755
discard_symbol:		rb_erase(&pos->rb_node, root);
756
			symbol__delete(pos);
757
		} else {
758 759 760
			if (curr_map != map) {
				rb_erase(&pos->rb_node, root);
				symbols__insert(&curr_map->dso->symbols[curr_map->type], pos);
761 762 763
				++moved;
			} else
				++count;
764
		}
765 766
	}

767
	if (curr_map != map &&
768
	    dso->kernel == DSO_TYPE_GUEST_KERNEL &&
769
	    machine__is_default_guest(kmaps->machine)) {
770 771 772
		dso__set_loaded(curr_map->dso, curr_map->type);
	}

773
	return count + moved;
774
}
775

776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793
static bool symbol__restricted_filename(const char *filename,
					const char *restricted_filename)
{
	bool restricted = false;

	if (symbol_conf.kptr_restrict) {
		char *r = realpath(filename, NULL);

		if (r != NULL) {
			restricted = strcmp(r, restricted_filename) == 0;
			free(r);
			return restricted;
		}
	}

	return restricted;
}

794
int dso__load_kallsyms(struct dso *dso, const char *filename,
795
		       struct map *map, symbol_filter_t filter)
796
{
797 798 799
	if (symbol__restricted_filename(filename, "/proc/kallsyms"))
		return -1;

800
	if (dso__load_all_kallsyms(dso, filename, map) < 0)
801 802
		return -1;

803
	symbols__fixup_duplicate(&dso->symbols[map->type]);
804 805
	symbols__fixup_end(&dso->symbols[map->type]);

806 807
	if (dso->kernel == DSO_TYPE_GUEST_KERNEL)
		dso->symtab_type = SYMTAB__GUEST_KALLSYMS;
808
	else
809
		dso->symtab_type = SYMTAB__KALLSYMS;
810

811
	return dso__split_kallsyms(dso, map, filter);
812 813
}

814
static int dso__load_perf_map(struct dso *dso, struct map *map,
815
			      symbol_filter_t filter)
816 817 818 819 820 821
{
	char *line = NULL;
	size_t n;
	FILE *file;
	int nr_syms = 0;

822
	file = fopen(dso->long_name, "r");
823 824 825 826
	if (file == NULL)
		goto out_failure;

	while (!feof(file)) {
827
		u64 start, size;
828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851
		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;

852
		sym = symbol__new(start, size, STB_GLOBAL, line + len);
853 854 855 856

		if (sym == NULL)
			goto out_delete_line;

857
		if (filter && filter(map, sym))
858
			symbol__delete(sym);
859
		else {
860
			symbols__insert(&dso->symbols[map->type], sym);
861 862 863 864 865 866 867 868 869 870 871 872 873 874 875
			nr_syms++;
		}
	}

	free(line);
	fclose(file);

	return nr_syms;

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

876 877 878
/**
 * elf_symtab__for_each_symbol - iterate thru all the symbols
 *
879
 * @syms: struct elf_symtab instance to iterate
880
 * @idx: uint32_t idx
881 882
 * @sym: GElf_Sym iterator
 */
883 884 885 886
#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))
887 888 889 890 891 892 893 894 895 896

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 &&
897
	       sym->st_shndx != SHN_UNDEF;
898 899
}

900 901 902 903 904 905 906
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;
}

907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926
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;
}

927 928 929 930 931 932
static inline bool elf_sec__is_data(const GElf_Shdr *shdr,
				    const Elf_Data *secstrs)
{
	return strstr(elf_sec__name(shdr, secstrs), "data") != NULL;
}

933 934 935 936 937 938 939 940
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,
941
				    size_t *idx)
942 943 944 945 946 947 948 949 950 951
{
	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)) {
952 953
			if (idx)
				*idx = cnt;
954 955 956 957 958 959 960 961
			break;
		}
		++cnt;
	}

	return sec;
}

962 963 964 965 966 967 968 969 970 971
#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))

972 973 974 975 976 977 978
/*
 * 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).
 */
979
static int dso__synthesize_plt_symbols(struct  dso *dso, struct map *map,
980
				       symbol_filter_t filter)
981 982 983
{
	uint32_t nr_rel_entries, idx;
	GElf_Sym sym;
984
	u64 plt_offset;
985 986
	GElf_Shdr shdr_plt;
	struct symbol *f;
987
	GElf_Shdr shdr_rel_plt, shdr_dynsym;
988
	Elf_Data *reldata, *syms, *symstrs;
989 990 991
	Elf_Scn *scn_plt_rel, *scn_symstrs, *scn_dynsym;
	size_t dynsym_idx;
	GElf_Ehdr ehdr;
992
	char sympltname[1024];
993 994
	Elf *elf;
	int nr = 0, symidx, fd, err = 0;
995
	char name[PATH_MAX];
996

997
	snprintf(name, sizeof(name), "%s%s",
998
		 symbol_conf.symfs, dso->long_name);
999
	fd = open(name, O_RDONLY);
1000 1001 1002
	if (fd < 0)
		goto out;

1003
	elf = elf_begin(fd, PERF_ELF_C_READ_MMAP, NULL);
1004 1005 1006 1007 1008 1009 1010 1011 1012 1013
	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;
1014

1015
	scn_plt_rel = elf_section_by_name(elf, &ehdr, &shdr_rel_plt,
1016 1017
					  ".rela.plt", NULL);
	if (scn_plt_rel == NULL) {
1018
		scn_plt_rel = elf_section_by_name(elf, &ehdr, &shdr_rel_plt,
1019 1020
						  ".rel.plt", NULL);
		if (scn_plt_rel == NULL)
1021
			goto out_elf_end;
1022 1023
	}

1024 1025
	err = -1;

1026
	if (shdr_rel_plt.sh_link != dynsym_idx)
1027
		goto out_elf_end;
1028

1029 1030
	if (elf_section_by_name(elf, &ehdr, &shdr_plt, ".plt", NULL) == NULL)
		goto out_elf_end;
1031 1032

	/*
1033
	 * Fetch the relocation section to find the idxes to the GOT
1034 1035 1036 1037
	 * and the symbols in the .dynsym they refer to.
	 */
	reldata = elf_getdata(scn_plt_rel, NULL);
	if (reldata == NULL)
1038
		goto out_elf_end;
1039 1040 1041

	syms = elf_getdata(scn_dynsym, NULL);
	if (syms == NULL)
1042
		goto out_elf_end;
1043

1044
	scn_symstrs = elf_getscn(elf, shdr_dynsym.sh_link);
1045
	if (scn_symstrs == NULL)
1046
		goto out_elf_end;
1047 1048 1049

	symstrs = elf_getdata(scn_symstrs, NULL);
	if (symstrs == NULL)
1050
		goto out_elf_end;
1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066

	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,
1067
					STB_GLOBAL, sympltname);
1068
			if (!f)
1069
				goto out_elf_end;
1070

1071 1072 1073
			if (filter && filter(map, f))
				symbol__delete(f);
			else {
1074
				symbols__insert(&dso->symbols[map->type], f);
1075 1076
				++nr;
			}
1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088
		}
	} 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,
1089
					STB_GLOBAL, sympltname);
1090
			if (!f)
1091
				goto out_elf_end;
1092

1093 1094 1095
			if (filter && filter(map, f))
				symbol__delete(f);
			else {
1096
				symbols__insert(&dso->symbols[map->type], f);
1097 1098
				++nr;
			}
1099 1100 1101
		}
	}

1102 1103 1104 1105 1106 1107 1108 1109 1110
	err = 0;
out_elf_end:
	elf_end(elf);
out_close:
	close(fd);

	if (err == 0)
		return nr;
out:
1111
	pr_debug("%s: problems reading %s PLT info.\n",
1112
		 __func__, dso->long_name);
1113
	return 0;
1114 1115
}

1116
static bool elf_sym__is_a(GElf_Sym *sym, enum map_type type)
1117 1118 1119
{
	switch (type) {
	case MAP__FUNCTION:
1120
		return elf_sym__is_function(sym);
1121
	case MAP__VARIABLE:
1122
		return elf_sym__is_object(sym);
1123 1124 1125 1126 1127
	default:
		return false;
	}
}

1128 1129
static bool elf_sec__is_a(GElf_Shdr *shdr, Elf_Data *secstrs,
			  enum map_type type)
1130 1131 1132
{
	switch (type) {
	case MAP__FUNCTION:
1133
		return elf_sec__is_text(shdr, secstrs);
1134
	case MAP__VARIABLE:
1135
		return elf_sec__is_data(shdr, secstrs);
1136 1137 1138 1139 1140
	default:
		return false;
	}
}

1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159
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;
}

1160
static int dso__load_sym(struct dso *dso, struct map *map, const char *name,
1161 1162
			 int fd, symbol_filter_t filter, int kmodule,
			 int want_symtab)
1163
{
1164
	struct kmap *kmap = dso->kernel ? map__kmap(map) : NULL;
1165
	struct map *curr_map = map;
1166
	struct dso *curr_dso = dso;
1167
	Elf_Data *symstrs, *secstrs;
1168 1169
	uint32_t nr_syms;
	int err = -1;
1170
	uint32_t idx;
1171
	GElf_Ehdr ehdr;
1172 1173
	GElf_Shdr shdr, opdshdr;
	Elf_Data *syms, *opddata = NULL;
1174
	GElf_Sym sym;
1175
	Elf_Scn *sec, *sec_strndx, *opdsec;
1176
	Elf *elf;
1177
	int nr = 0;
1178
	size_t opdidx = 0;
1179

1180
	elf = elf_begin(fd, PERF_ELF_C_READ_MMAP, NULL);
1181
	if (elf == NULL) {
1182
		pr_debug("%s: cannot read %s ELF file.\n", __func__, name);
1183 1184 1185 1186
		goto out_close;
	}

	if (gelf_getehdr(elf, &ehdr) == NULL) {
1187
		pr_debug("%s: cannot get elf header.\n", __func__);
1188 1189 1190
		goto out_elf_end;
	}

1191
	/* Always reject images with a mismatched build-id: */
1192
	if (dso->has_build_id) {
1193 1194
		u8 build_id[BUILD_ID_SIZE];

1195
		if (elf_read_build_id(elf, build_id, BUILD_ID_SIZE) < 0)
1196 1197
			goto out_elf_end;

1198
		if (!dso__build_id_equal(dso, build_id))
1199 1200 1201
			goto out_elf_end;
	}

1202
	sec = elf_section_by_name(elf, &ehdr, &shdr, ".symtab", NULL);
1203
	if (sec == NULL) {
1204 1205 1206
		if (want_symtab)
			goto out_elf_end;

1207 1208
		sec = elf_section_by_name(elf, &ehdr, &shdr, ".dynsym", NULL);
		if (sec == NULL)
1209 1210
			goto out_elf_end;
	}
1211

1212
	opdsec = elf_section_by_name(elf, &ehdr, &opdshdr, ".opd", &opdidx);
1213 1214
	if (opdshdr.sh_type != SHT_PROGBITS)
		opdsec = NULL;
1215 1216 1217
	if (opdsec)
		opddata = elf_rawdata(opdsec, NULL);

1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229
	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;

1230 1231 1232 1233 1234
	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
1235
	if (secstrs == NULL)
1236 1237
		goto out_elf_end;

1238 1239
	nr_syms = shdr.sh_size / shdr.sh_entsize;

1240
	memset(&sym, 0, sizeof(sym));
1241 1242
	if (dso->kernel == DSO_TYPE_USER) {
		dso->adjust_symbols = (ehdr.e_type == ET_EXEC ||
1243 1244 1245
				elf_section_by_name(elf, &ehdr, &shdr,
						     ".gnu.prelink_undo",
						     NULL) != NULL);
1246 1247 1248
	} else {
		dso->adjust_symbols = 0;
	}
1249
	elf_symtab__for_each_symbol(syms, nr_syms, idx, sym) {
1250
		struct symbol *f;
1251
		const char *elf_name = elf_sym__name(&sym, symstrs);
1252
		char *demangled = NULL;
1253 1254
		int is_label = elf_sym__is_label(&sym);
		const char *section_name;
1255

1256 1257 1258
		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;
1259

1260
		if (!is_label && !elf_sym__is_a(&sym, map->type))
1261 1262
			continue;

1263 1264 1265 1266 1267 1268 1269 1270 1271 1272
		/* 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;
		}

1273 1274 1275 1276 1277 1278 1279
		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);
		}

1280 1281 1282 1283 1284
		sec = elf_getscn(elf, sym.st_shndx);
		if (!sec)
			goto out_elf_end;

		gelf_getshdr(sec, &shdr);
1285

1286
		if (is_label && !elf_sec__is_a(&shdr, secstrs, map->type))
1287 1288 1289
			continue;

		section_name = elf_sec__name(&shdr, secstrs);
1290

1291 1292 1293 1294 1295 1296 1297
		/* 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;

1298
		if (dso->kernel != DSO_TYPE_USER || kmodule) {
1299 1300 1301
			char dso_name[PATH_MAX];

			if (strcmp(section_name,
1302
				   (curr_dso->short_name +
1303
				    dso->short_name_len)) == 0)
1304 1305 1306 1307
				goto new_symbol;

			if (strcmp(section_name, ".text") == 0) {
				curr_map = map;
1308
				curr_dso = dso;
1309 1310 1311 1312
				goto new_symbol;
			}

			snprintf(dso_name, sizeof(dso_name),
1313
				 "%s%s", dso->short_name, section_name);
1314

1315
			curr_map = map_groups__find_by_name(kmap->kmaps, map->type, dso_name);
1316 1317 1318 1319 1320 1321
			if (curr_map == NULL) {
				u64 start = sym.st_value;

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

1322
				curr_dso = dso__new(dso_name);
1323 1324
				if (curr_dso == NULL)
					goto out_elf_end;
1325 1326 1327
				curr_dso->kernel = dso->kernel;
				curr_dso->long_name = dso->long_name;
				curr_dso->long_name_len = dso->long_name_len;
1328
				curr_map = map__new2(start, curr_dso,
1329
						     map->type);
1330 1331 1332 1333
				if (curr_map == NULL) {
					dso__delete(curr_dso);
					goto out_elf_end;
				}
1334 1335
				curr_map->map_ip = identity__map_ip;
				curr_map->unmap_ip = identity__map_ip;
1336
				curr_dso->symtab_type = dso->symtab_type;
1337
				map_groups__insert(kmap->kmaps, curr_map);
1338
				dsos__add(&dso->node, curr_dso);
1339
				dso__set_loaded(curr_dso, map->type);
1340 1341 1342 1343
			} else
				curr_dso = curr_map->dso;

			goto new_symbol;
1344 1345
		}

1346
		if (curr_dso->adjust_symbols) {
1347 1348
			pr_debug4("%s: adjusting symbol: st_value: %#" PRIx64 " "
				  "sh_addr: %#" PRIx64 " sh_offset: %#" PRIx64 "\n", __func__,
1349 1350
				  (u64)sym.st_value, (u64)shdr.sh_addr,
				  (u64)shdr.sh_offset);
1351
			sym.st_value -= shdr.sh_addr - shdr.sh_offset;
1352
		}
1353 1354 1355 1356 1357
		/*
		 * 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...
		 */
1358
		demangled = bfd_demangle(NULL, elf_name, DMGL_PARAMS | DMGL_ANSI);
1359
		if (demangled != NULL)
1360
			elf_name = demangled;
1361
new_symbol:
1362 1363
		f = symbol__new(sym.st_value, sym.st_size,
				GELF_ST_BIND(sym.st_info), elf_name);
1364
		free(demangled);
1365 1366 1367
		if (!f)
			goto out_elf_end;

1368
		if (filter && filter(curr_map, f))
1369
			symbol__delete(f);
1370
		else {
1371
			symbols__insert(&curr_dso->symbols[curr_map->type], f);
1372 1373
			nr++;
		}
1374 1375
	}

1376 1377 1378
	/*
	 * For misannotated, zeroed, ASM function sizes.
	 */
1379
	if (nr > 0) {
1380
		symbols__fixup_duplicate(&dso->symbols[map->type]);
1381
		symbols__fixup_end(&dso->symbols[map->type]);
1382 1383 1384 1385 1386 1387 1388 1389
		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);
		}
	}
1390 1391 1392 1393 1394 1395 1396
	err = nr;
out_elf_end:
	elf_end(elf);
out_close:
	return err;
}

1397
static bool dso__build_id_equal(const struct dso *dso, u8 *build_id)
1398
{
1399
	return memcmp(dso->build_id, build_id, sizeof(dso->build_id)) == 0;
1400 1401
}

1402
bool __dsos__read_build_ids(struct list_head *head, bool with_hits)
1403
{
1404
	bool have_build_id = false;
1405 1406
	struct dso *pos;

1407 1408 1409
	list_for_each_entry(pos, head, node) {
		if (with_hits && !pos->hit)
			continue;
1410 1411 1412 1413
		if (pos->has_build_id) {
			have_build_id = true;
			continue;
		}
1414 1415 1416 1417 1418
		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;
		}
1419
	}
1420

1421
	return have_build_id;
1422 1423
}

1424 1425 1426 1427 1428
/*
 * Align offset to 4 bytes as needed for note name and descriptor data.
 */
#define NOTE_ALIGN(n) (((n) + 3) & -4U)

1429
static int elf_read_build_id(Elf *elf, void *bf, size_t size)
1430
{
1431
	int err = -1;
1432 1433
	GElf_Ehdr ehdr;
	GElf_Shdr shdr;
1434
	Elf_Data *data;
1435
	Elf_Scn *sec;
1436
	Elf_Kind ek;
1437
	void *ptr;
1438

1439 1440 1441
	if (size < BUILD_ID_SIZE)
		goto out;

1442 1443
	ek = elf_kind(elf);
	if (ek != ELF_K_ELF)
1444
		goto out;
1445

1446
	if (gelf_getehdr(elf, &ehdr) == NULL) {
1447
		pr_err("%s: cannot get elf header.\n", __func__);
1448
		goto out;
1449 1450
	}

1451 1452
	sec = elf_section_by_name(elf, &ehdr, &shdr,
				  ".note.gnu.build-id", NULL);
1453 1454 1455 1456
	if (sec == NULL) {
		sec = elf_section_by_name(elf, &ehdr, &shdr,
					  ".notes", NULL);
		if (sec == NULL)
1457
			goto out;
1458
	}
1459

1460 1461
	data = elf_getdata(sec, NULL);
	if (data == NULL)
1462
		goto out;
1463 1464 1465 1466

	ptr = data->d_buf;
	while (ptr < (data->d_buf + data->d_size)) {
		GElf_Nhdr *nhdr = ptr;
1467 1468
		size_t namesz = NOTE_ALIGN(nhdr->n_namesz),
		       descsz = NOTE_ALIGN(nhdr->n_descsz);
1469 1470 1471 1472 1473 1474 1475 1476
		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) {
1477 1478 1479 1480
				size_t sz = min(size, descsz);
				memcpy(bf, ptr, sz);
				memset(bf + sz, 0, size - sz);
				err = descsz;
1481 1482 1483 1484 1485
				break;
			}
		}
		ptr += descsz;
	}
1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510

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);

1511 1512 1513 1514 1515 1516 1517
	elf_end(elf);
out_close:
	close(fd);
out:
	return err;
}

1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531
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;
1532
		size_t namesz, descsz;
1533 1534 1535 1536

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

1537 1538
		namesz = NOTE_ALIGN(nhdr.n_namesz);
		descsz = NOTE_ALIGN(nhdr.n_descsz);
1539 1540
		if (nhdr.n_type == NT_GNU_BUILD_ID &&
		    nhdr.n_namesz == sizeof("GNU")) {
1541
			if (read(fd, bf, namesz) != (ssize_t)namesz)
1542 1543
				break;
			if (memcmp(bf, "GNU", sizeof("GNU")) == 0) {
1544 1545 1546
				size_t sz = min(descsz, size);
				if (read(fd, build_id, sz) == (ssize_t)sz) {
					memset(build_id + sz, 0, size - sz);
1547 1548 1549
					err = 0;
					break;
				}
1550
			} else if (read(fd, bf, descsz) != (ssize_t)descsz)
1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562
				break;
		} else {
			int n = namesz + descsz;
			if (read(fd, bf, n) != n)
				break;
		}
	}
	close(fd);
out:
	return err;
}

1563
char dso__symtab_origin(const struct dso *dso)
1564 1565
{
	static const char origin[] = {
1566 1567 1568 1569 1570 1571 1572 1573 1574 1575
		[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',
1576 1577
	};

1578
	if (dso == NULL || dso->symtab_type == SYMTAB__NOT_FOUND)
1579
		return '!';
1580
	return origin[dso->symtab_type];
1581 1582
}

1583
int dso__load(struct dso *dso, struct map *map, symbol_filter_t filter)
1584
{
1585
	int size = PATH_MAX;
1586
	char *name;
1587 1588
	int ret = -1;
	int fd;
1589
	struct machine *machine;
1590
	const char *root_dir;
1591
	int want_symtab;
1592

1593
	dso__set_loaded(dso, map->type);
1594

1595 1596 1597 1598
	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);
1599

1600 1601
	if (map->groups && map->groups->machine)
		machine = map->groups->machine;
1602
	else
1603
		machine = NULL;
1604 1605

	name = malloc(size);
1606 1607 1608
	if (!name)
		return -1;

1609
	dso->adjust_symbols = 0;
1610

1611
	if (strncmp(dso->name, "/tmp/perf-", 10) == 0) {
1612 1613
		struct stat st;

1614
		if (lstat(dso->name, &st) < 0)
1615 1616 1617 1618 1619 1620 1621 1622
			return -1;

		if (st.st_uid && (st.st_uid != geteuid())) {
			pr_warning("File %s not owned by current user or root, "
				"ignoring it.\n", dso->name);
			return -1;
		}

1623 1624
		ret = dso__load_perf_map(dso, map, filter);
		dso->symtab_type = ret > 0 ? SYMTAB__JAVA_JIT :
1625
					      SYMTAB__NOT_FOUND;
1626 1627 1628
		return ret;
	}

1629 1630 1631 1632
	/* 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
	 */
1633 1634
	want_symtab = 1;
restart:
1635 1636 1637 1638
	for (dso->symtab_type = SYMTAB__BUILD_ID_CACHE;
	     dso->symtab_type != SYMTAB__NOT_FOUND;
	     dso->symtab_type++) {
		switch (dso->symtab_type) {
1639
		case SYMTAB__BUILD_ID_CACHE:
1640 1641
			/* skip the locally configured cache if a symfs is given */
			if (symbol_conf.symfs[0] ||
1642
			    (dso__build_id_filename(dso, name, size) == NULL)) {
1643
				continue;
1644
			}
1645
			break;
1646
		case SYMTAB__FEDORA_DEBUGINFO:
1647
			snprintf(name, size, "%s/usr/lib/debug%s.debug",
1648
				 symbol_conf.symfs, dso->long_name);
1649
			break;
1650
		case SYMTAB__UBUNTU_DEBUGINFO:
1651
			snprintf(name, size, "%s/usr/lib/debug%s",
1652
				 symbol_conf.symfs, dso->long_name);
1653
			break;
1654
		case SYMTAB__BUILDID_DEBUGINFO: {
1655 1656
			char build_id_hex[BUILD_ID_SIZE * 2 + 1];

1657
			if (!dso->has_build_id)
1658 1659
				continue;

1660 1661
			build_id__sprintf(dso->build_id,
					  sizeof(dso->build_id),
1662 1663
					  build_id_hex);
			snprintf(name, size,
1664 1665
				 "%s/usr/lib/debug/.build-id/%.2s/%s.debug",
				 symbol_conf.symfs, build_id_hex, build_id_hex + 2);
1666
			}
1667
			break;
1668
		case SYMTAB__SYSTEM_PATH_DSO:
1669
			snprintf(name, size, "%s%s",
1670
			     symbol_conf.symfs, dso->long_name);
1671
			break;
1672
		case SYMTAB__GUEST_KMODULE:
K
Kyle McMartin 已提交
1673 1674
			if (map->groups && machine)
				root_dir = machine->root_dir;
1675 1676
			else
				root_dir = "";
1677
			snprintf(name, size, "%s%s%s", symbol_conf.symfs,
1678
				 root_dir, dso->long_name);
1679 1680
			break;

1681
		case SYMTAB__SYSTEM_PATH_KMODULE:
1682
			snprintf(name, size, "%s%s", symbol_conf.symfs,
1683
				 dso->long_name);
1684
			break;
1685
		default:;
1686
		}
1687 1688

		/* Name is now the name of the next image to try */
1689
		fd = open(name, O_RDONLY);
1690 1691
		if (fd < 0)
			continue;
1692

1693
		ret = dso__load_sym(dso, map, name, fd, filter, 0,
1694 1695
				    want_symtab);
		close(fd);
1696

1697 1698 1699 1700 1701 1702
		/*
		 * Some people seem to have debuginfo files _WITHOUT_ debug
		 * info!?!?
		 */
		if (!ret)
			continue;
1703

1704
		if (ret > 0) {
1705 1706
			int nr_plt = dso__synthesize_plt_symbols(dso, map,
								 filter);
1707 1708 1709 1710
			if (nr_plt > 0)
				ret += nr_plt;
			break;
		}
1711
	}
1712

1713 1714 1715 1716 1717 1718 1719 1720 1721
	/*
	 * 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;
	}

1722
	free(name);
1723
	if (ret < 0 && strstr(dso->name, " (deleted)") != NULL)
1724
		return 0;
1725 1726 1727
	return ret;
}

1728
struct map *map_groups__find_by_name(struct map_groups *mg,
1729
				     enum map_type type, const char *name)
1730 1731 1732
{
	struct rb_node *nd;

1733
	for (nd = rb_first(&mg->maps[type]); nd; nd = rb_next(nd)) {
1734 1735
		struct map *map = rb_entry(nd, struct map, rb_node);

1736
		if (map->dso && strcmp(map->dso->short_name, name) == 0)
1737 1738 1739 1740 1741 1742
			return map;
	}

	return NULL;
}

1743 1744
static int dso__kernel_module_get_build_id(struct dso *dso,
					   const char *root_dir)
1745 1746 1747 1748 1749 1750
{
	char filename[PATH_MAX];
	/*
	 * kernel module short names are of the form "[module]" and
	 * we need just "module" here.
	 */
1751
	const char *name = dso->short_name + 1;
1752 1753

	snprintf(filename, sizeof(filename),
1754 1755
		 "%s/sys/module/%.*s/notes/.note.gnu.build-id",
		 root_dir, (int)strlen(name) - 1, name);
1756

1757 1758 1759
	if (sysfs__read_build_id(filename, dso->build_id,
				 sizeof(dso->build_id)) == 0)
		dso->has_build_id = true;
1760 1761 1762 1763

	return 0;
}

1764
static int map_groups__set_modules_path_dir(struct map_groups *mg,
1765
				const char *dir_name)
1766
{
1767
	struct dirent *dent;
1768
	DIR *dir = opendir(dir_name);
1769
	int ret = 0;
1770

1771
	if (!dir) {
1772
		pr_debug("%s: cannot open %s dir\n", __func__, dir_name);
1773 1774
		return -1;
	}
1775

1776 1777
	while ((dent = readdir(dir)) != NULL) {
		char path[PATH_MAX];
1778 1779 1780
		struct stat st;

		/*sshfs might return bad dent->d_type, so we have to stat*/
1781
		snprintf(path, sizeof(path), "%s/%s", dir_name, dent->d_name);
1782 1783
		if (stat(path, &st))
			continue;
1784

1785
		if (S_ISDIR(st.st_mode)) {
1786 1787 1788 1789
			if (!strcmp(dent->d_name, ".") ||
			    !strcmp(dent->d_name, ".."))
				continue;

1790
			ret = map_groups__set_modules_path_dir(mg, path);
1791 1792
			if (ret < 0)
				goto out;
1793 1794 1795 1796
		} else {
			char *dot = strrchr(dent->d_name, '.'),
			     dso_name[PATH_MAX];
			struct map *map;
1797
			char *long_name;
1798 1799 1800 1801 1802 1803

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

1804
			strxfrchar(dso_name, '-', '_');
1805 1806
			map = map_groups__find_by_name(mg, MAP__FUNCTION,
						       dso_name);
1807 1808 1809
			if (map == NULL)
				continue;

1810
			long_name = strdup(path);
1811 1812 1813 1814
			if (long_name == NULL) {
				ret = -1;
				goto out;
			}
1815
			dso__set_long_name(map->dso, long_name);
1816
			map->dso->lname_alloc = 1;
1817
			dso__kernel_module_get_build_id(map->dso, "");
1818 1819
		}
	}
1820

1821
out:
1822
	closedir(dir);
1823
	return ret;
1824
}
1825

1826
static char *get_kernel_version(const char *root_dir)
1827
{
1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852
	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);
}

1853
static int machine__set_modules_path(struct machine *machine)
1854 1855
{
	char *version;
1856
	char modules_path[PATH_MAX];
1857

1858
	version = get_kernel_version(machine->root_dir);
1859
	if (!version)
1860
		return -1;
1861

1862
	snprintf(modules_path, sizeof(modules_path), "%s/lib/modules/%s/kernel",
1863
		 machine->root_dir, version);
1864
	free(version);
1865

1866
	return map_groups__set_modules_path_dir(&machine->kmaps, modules_path);
1867 1868
}

1869 1870 1871 1872 1873
/*
 * 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.
 */
1874
static struct map *map__new2(u64 start, struct dso *dso, enum map_type type)
1875
{
1876 1877 1878
	struct map *map = calloc(1, (sizeof(*map) +
				     (dso->kernel ? sizeof(struct kmap) : 0)));
	if (map != NULL) {
1879
		/*
1880
		 * ->end will be filled after we load all the symbols
1881
		 */
1882
		map__init(map, type, start, 0, 0, dso);
1883
	}
1884

1885
	return map;
1886 1887
}

1888
struct map *machine__new_module(struct machine *machine, u64 start,
1889
				const char *filename)
1890 1891
{
	struct map *map;
1892
	struct dso *dso = __dsos__findnew(&machine->kernel_dsos, filename);
1893 1894 1895 1896 1897 1898 1899 1900

	if (dso == NULL)
		return NULL;

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

1901
	if (machine__is_host(machine))
1902
		dso->symtab_type = SYMTAB__SYSTEM_PATH_KMODULE;
1903
	else
1904
		dso->symtab_type = SYMTAB__GUEST_KMODULE;
1905
	map_groups__insert(&machine->kmaps, map);
1906 1907 1908
	return map;
}

1909
static int machine__create_modules(struct machine *machine)
1910 1911 1912
{
	char *line = NULL;
	size_t n;
1913
	FILE *file;
1914
	struct map *map;
1915 1916 1917
	const char *modules;
	char path[PATH_MAX];

1918
	if (machine__is_default_guest(machine))
1919 1920
		modules = symbol_conf.default_guest_modules;
	else {
1921
		sprintf(path, "%s/proc/modules", machine->root_dir);
1922 1923
		modules = path;
	}
1924

1925 1926 1927
	if (symbol__restricted_filename(path, "/proc/modules"))
		return -1;

1928
	file = fopen(modules, "r");
1929 1930
	if (file == NULL)
		return -1;
1931

1932 1933 1934 1935 1936
	while (!feof(file)) {
		char name[PATH_MAX];
		u64 start;
		char *sep;
		int line_len;
1937

1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959
		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);
1960
		map = machine__new_module(machine, start, name);
1961
		if (map == NULL)
1962
			goto out_delete_line;
1963
		dso__kernel_module_get_build_id(map->dso, machine->root_dir);
1964
	}
1965 1966 1967 1968

	free(line);
	fclose(file);

1969
	return machine__set_modules_path(machine);
1970 1971 1972 1973 1974

out_delete_line:
	free(line);
out_failure:
	return -1;
1975 1976
}

1977
int dso__load_vmlinux(struct dso *dso, struct map *map,
1978
		      const char *vmlinux, symbol_filter_t filter)
1979
{
1980
	int err = -1, fd;
1981
	char symfs_vmlinux[PATH_MAX];
1982

1983
	snprintf(symfs_vmlinux, sizeof(symfs_vmlinux), "%s%s",
1984 1985
		 symbol_conf.symfs, vmlinux);
	fd = open(symfs_vmlinux, O_RDONLY);
1986 1987 1988
	if (fd < 0)
		return -1;

1989 1990 1991
	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);
1992 1993
	close(fd);

1994
	if (err > 0)
1995
		pr_debug("Using %s for symbols\n", symfs_vmlinux);
1996

1997 1998 1999
	return err;
}

2000
int dso__load_vmlinux_path(struct dso *dso, struct map *map,
2001
			   symbol_filter_t filter)
2002 2003
{
	int i, err = 0;
2004
	char *filename;
2005 2006

	pr_debug("Looking at the vmlinux_path (%d entries long)\n",
2007 2008
		 vmlinux_path__nr_entries + 1);

2009
	filename = dso__build_id_filename(dso, NULL, 0);
2010
	if (filename != NULL) {
2011
		err = dso__load_vmlinux(dso, map, filename, filter);
2012
		if (err > 0) {
2013
			dso__set_long_name(dso, filename);
2014 2015 2016 2017
			goto out;
		}
		free(filename);
	}
2018 2019

	for (i = 0; i < vmlinux_path__nr_entries; ++i) {
2020
		err = dso__load_vmlinux(dso, map, vmlinux_path[i], filter);
2021
		if (err > 0) {
2022
			dso__set_long_name(dso, strdup(vmlinux_path[i]));
2023 2024 2025
			break;
		}
	}
2026
out:
2027 2028 2029
	return err;
}

2030
static int dso__load_kernel_sym(struct dso *dso, struct map *map,
2031
				symbol_filter_t filter)
2032
{
2033
	int err;
2034 2035
	const char *kallsyms_filename = NULL;
	char *kallsyms_allocated_filename = NULL;
2036
	/*
2037 2038
	 * 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.
2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050
	 *
	 * 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.
	 */
2051 2052 2053 2054 2055
	if (symbol_conf.kallsyms_name != NULL) {
		kallsyms_filename = symbol_conf.kallsyms_name;
		goto do_kallsyms;
	}

2056
	if (symbol_conf.vmlinux_name != NULL) {
2057
		err = dso__load_vmlinux(dso, map,
2058
					symbol_conf.vmlinux_name, filter);
2059
		if (err > 0) {
2060
			dso__set_long_name(dso,
2061 2062 2063 2064
					   strdup(symbol_conf.vmlinux_name));
			goto out_fixup;
		}
		return err;
2065
	}
2066 2067

	if (vmlinux_path != NULL) {
2068
		err = dso__load_vmlinux_path(dso, map, filter);
2069 2070
		if (err > 0)
			goto out_fixup;
2071 2072
	}

2073 2074 2075 2076
	/* do not try local files if a symfs was given */
	if (symbol_conf.symfs[0] != 0)
		return -1;

2077 2078 2079 2080 2081
	/*
	 * 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.
	 */
2082
	if (dso->has_build_id) {
2083
		u8 kallsyms_build_id[BUILD_ID_SIZE];
2084
		char sbuild_id[BUILD_ID_SIZE * 2 + 1];
2085 2086

		if (sysfs__read_build_id("/sys/kernel/notes", kallsyms_build_id,
2087
					 sizeof(kallsyms_build_id)) == 0) {
2088
			if (dso__build_id_equal(dso, kallsyms_build_id)) {
2089
				kallsyms_filename = "/proc/kallsyms";
2090
				goto do_kallsyms;
2091
			}
2092
		}
2093 2094 2095 2096
		/*
		 * Now look if we have it on the build-id cache in
		 * $HOME/.debug/[kernel.kallsyms].
		 */
2097
		build_id__sprintf(dso->build_id, sizeof(dso->build_id),
2098 2099 2100 2101
				  sbuild_id);

		if (asprintf(&kallsyms_allocated_filename,
			     "%s/.debug/[kernel.kallsyms]/%s",
2102 2103
			     getenv("HOME"), sbuild_id) == -1) {
			pr_err("Not enough memory for kallsyms file lookup\n");
2104
			return -1;
2105
		}
2106

2107 2108
		kallsyms_filename = kallsyms_allocated_filename;

2109
		if (access(kallsyms_filename, F_OK)) {
2110 2111
			pr_err("No kallsyms or vmlinux with build-id %s "
			       "was found\n", sbuild_id);
2112
			free(kallsyms_allocated_filename);
2113
			return -1;
2114
		}
2115 2116 2117 2118 2119
	} else {
		/*
		 * Last resort, if we don't have a build-id and couldn't find
		 * any vmlinux file, try the running kernel kallsyms table.
		 */
2120 2121
		kallsyms_filename = "/proc/kallsyms";
	}
2122

2123
do_kallsyms:
2124
	err = dso__load_kallsyms(dso, kallsyms_filename, map, filter);
2125 2126
	if (err > 0)
		pr_debug("Using %s for symbols\n", kallsyms_filename);
2127
	free(kallsyms_allocated_filename);
2128 2129

	if (err > 0) {
2130
out_fixup:
2131
		if (kallsyms_filename != NULL)
2132
			dso__set_long_name(dso, strdup("[kernel.kallsyms]"));
2133 2134
		map__fixup_start(map);
		map__fixup_end(map);
2135
	}
2136

2137 2138 2139
	return err;
}

2140 2141
static int dso__load_guest_kernel_sym(struct dso *dso, struct map *map,
				      symbol_filter_t filter)
2142 2143 2144
{
	int err;
	const char *kallsyms_filename = NULL;
2145
	struct machine *machine;
2146 2147 2148 2149 2150 2151
	char path[PATH_MAX];

	if (!map->groups) {
		pr_debug("Guest kernel map hasn't the point to groups\n");
		return -1;
	}
2152
	machine = map->groups->machine;
2153

2154
	if (machine__is_default_guest(machine)) {
2155 2156 2157 2158 2159 2160
		/*
		 * 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) {
2161
			err = dso__load_vmlinux(dso, map,
2162 2163 2164 2165 2166 2167 2168 2169
				symbol_conf.default_guest_vmlinux_name, filter);
			goto out_try_fixup;
		}

		kallsyms_filename = symbol_conf.default_guest_kallsyms;
		if (!kallsyms_filename)
			return -1;
	} else {
2170
		sprintf(path, "%s/proc/kallsyms", machine->root_dir);
2171 2172 2173
		kallsyms_filename = path;
	}

2174
	err = dso__load_kallsyms(dso, kallsyms_filename, map, filter);
2175 2176 2177 2178 2179 2180
	if (err > 0)
		pr_debug("Using %s for symbols\n", kallsyms_filename);

out_try_fixup:
	if (err > 0) {
		if (kallsyms_filename != NULL) {
2181
			machine__mmap_name(machine, path, sizeof(path));
2182
			dso__set_long_name(dso, strdup(path));
2183 2184 2185 2186 2187 2188 2189
		}
		map__fixup_start(map);
		map__fixup_end(map);
	}

	return err;
}
2190

2191
static void dsos__add(struct list_head *head, struct dso *dso)
2192
{
2193
	list_add_tail(&dso->node, head);
2194 2195
}

2196
static struct dso *dsos__find(struct list_head *head, const char *name)
2197 2198 2199
{
	struct dso *pos;

2200
	list_for_each_entry(pos, head, node)
2201
		if (strcmp(pos->long_name, name) == 0)
2202 2203 2204 2205
			return pos;
	return NULL;
}

2206
struct dso *__dsos__findnew(struct list_head *head, const char *name)
2207
{
2208
	struct dso *dso = dsos__find(head, name);
2209

2210
	if (!dso) {
2211
		dso = dso__new(name);
2212
		if (dso != NULL) {
2213
			dsos__add(head, dso);
2214 2215
			dso__set_basename(dso);
		}
2216
	}
2217 2218 2219 2220

	return dso;
}

2221
size_t __dsos__fprintf(struct list_head *head, FILE *fp)
2222 2223
{
	struct dso *pos;
2224
	size_t ret = 0;
2225

2226 2227 2228
	list_for_each_entry(pos, head, node) {
		int i;
		for (i = 0; i < MAP__NR_TYPES; ++i)
2229
			ret += dso__fprintf(pos, i, fp);
2230
	}
2231 2232

	return ret;
2233 2234
}

2235
size_t machines__fprintf_dsos(struct rb_root *machines, FILE *fp)
2236
{
2237
	struct rb_node *nd;
2238
	size_t ret = 0;
2239

2240
	for (nd = rb_first(machines); nd; nd = rb_next(nd)) {
2241
		struct machine *pos = rb_entry(nd, struct machine, rb_node);
2242 2243
		ret += __dsos__fprintf(&pos->kernel_dsos, fp);
		ret += __dsos__fprintf(&pos->user_dsos, fp);
2244
	}
2245 2246

	return ret;
2247 2248
}

2249 2250
static size_t __dsos__fprintf_buildid(struct list_head *head, FILE *fp,
				      bool with_hits)
2251 2252 2253 2254
{
	struct dso *pos;
	size_t ret = 0;

2255
	list_for_each_entry(pos, head, node) {
2256 2257
		if (with_hits && !pos->hit)
			continue;
2258
		ret += dso__fprintf_buildid(pos, fp);
2259
		ret += fprintf(fp, " %s\n", pos->long_name);
2260 2261 2262 2263
	}
	return ret;
}

2264 2265
size_t machine__fprintf_dsos_buildid(struct machine *machine, FILE *fp,
				     bool with_hits)
2266
{
2267 2268
	return __dsos__fprintf_buildid(&machine->kernel_dsos, fp, with_hits) +
	       __dsos__fprintf_buildid(&machine->user_dsos, fp, with_hits);
2269 2270
}

2271 2272
size_t machines__fprintf_dsos_buildid(struct rb_root *machines,
				      FILE *fp, bool with_hits)
2273
{
2274 2275 2276
	struct rb_node *nd;
	size_t ret = 0;

2277
	for (nd = rb_first(machines); nd; nd = rb_next(nd)) {
2278
		struct machine *pos = rb_entry(nd, struct machine, rb_node);
2279
		ret += machine__fprintf_dsos_buildid(pos, fp, with_hits);
2280 2281
	}
	return ret;
2282 2283
}

2284 2285 2286
static struct dso*
dso__kernel_findnew(struct machine *machine, const char *name,
		    const char *short_name, int dso_type)
2287
{
2288 2289 2290 2291
	/*
	 * The kernel dso could be created by build_id processing.
	 */
	struct dso *dso = __dsos__findnew(&machine->kernel_dsos, name);
2292

2293 2294 2295 2296
	/*
	 * We need to run this in all cases, since during the build_id
	 * processing we had no idea this was the kernel dso.
	 */
2297
	if (dso != NULL) {
2298 2299
		dso__set_short_name(dso, short_name);
		dso->kernel = dso_type;
2300 2301
	}

2302
	return dso;
2303 2304
}

2305
void dso__read_running_kernel_build_id(struct dso *dso, struct machine *machine)
2306
{
2307 2308
	char path[PATH_MAX];

2309
	if (machine__is_default_guest(machine))
2310
		return;
2311
	sprintf(path, "%s/sys/kernel/notes", machine->root_dir);
2312 2313 2314
	if (sysfs__read_build_id(path, dso->build_id,
				 sizeof(dso->build_id)) == 0)
		dso->has_build_id = true;
2315 2316
}

2317
static struct dso *machine__get_kernel(struct machine *machine)
2318
{
2319 2320
	const char *vmlinux_name = NULL;
	struct dso *kernel;
2321

2322
	if (machine__is_host(machine)) {
2323
		vmlinux_name = symbol_conf.vmlinux_name;
2324 2325 2326 2327 2328 2329
		if (!vmlinux_name)
			vmlinux_name = "[kernel.kallsyms]";

		kernel = dso__kernel_findnew(machine, vmlinux_name,
					     "[kernel]",
					     DSO_TYPE_KERNEL);
2330
	} else {
2331 2332
		char bf[PATH_MAX];

2333
		if (machine__is_default_guest(machine))
2334
			vmlinux_name = symbol_conf.default_guest_vmlinux_name;
2335 2336 2337 2338 2339 2340 2341
		if (!vmlinux_name)
			vmlinux_name = machine__mmap_name(machine, bf,
							  sizeof(bf));

		kernel = dso__kernel_findnew(machine, vmlinux_name,
					     "[guest.kernel]",
					     DSO_TYPE_GUEST_KERNEL);
2342
	}
2343

2344
	if (kernel != NULL && (!kernel->has_build_id))
2345
		dso__read_running_kernel_build_id(kernel, machine);
2346

2347 2348 2349
	return kernel;
}

2350 2351 2352 2353 2354
struct process_args {
	u64 start;
};

static int symbol__in_kernel(void *arg, const char *name,
2355
			     char type __used, u64 start, u64 end __used)
2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383
{
	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;
		}
	}

2384 2385 2386
	if (symbol__restricted_filename(filename, "/proc/kallsyms"))
		return 0;

2387 2388 2389 2390 2391 2392
	if (kallsyms__parse(filename, &args, symbol__in_kernel) <= 0)
		return 0;

	return args.start;
}

2393
int __machine__create_kernel_maps(struct machine *machine, struct dso *kernel)
2394
{
2395
	enum map_type type;
2396
	u64 start = machine__get_kernel_start_addr(machine);
2397

2398
	for (type = 0; type < MAP__NR_TYPES; ++type) {
2399 2400
		struct kmap *kmap;

2401 2402
		machine->vmlinux_maps[type] = map__new2(start, kernel, type);
		if (machine->vmlinux_maps[type] == NULL)
2403
			return -1;
2404

2405 2406 2407 2408 2409 2410 2411
		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]);
2412 2413 2414
	}

	return 0;
2415 2416
}

2417
void machine__destroy_kernel_maps(struct machine *machine)
2418 2419 2420 2421 2422 2423
{
	enum map_type type;

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

2424
		if (machine->vmlinux_maps[type] == NULL)
2425 2426
			continue;

2427 2428 2429
		kmap = map__kmap(machine->vmlinux_maps[type]);
		map_groups__remove(&machine->kmaps,
				   machine->vmlinux_maps[type]);
2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442
		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;
		}

2443 2444
		map__delete(machine->vmlinux_maps[type]);
		machine->vmlinux_maps[type] = NULL;
2445 2446 2447
	}
}

2448
int machine__create_kernel_maps(struct machine *machine)
2449
{
2450
	struct dso *kernel = machine__get_kernel(machine);
2451 2452

	if (kernel == NULL ||
2453
	    __machine__create_kernel_maps(machine, kernel) < 0)
2454 2455
		return -1;

2456
	if (symbol_conf.use_modules && machine__create_modules(machine) < 0)
2457 2458 2459 2460
		pr_debug("Problems creating module maps, continuing anyway...\n");
	/*
	 * Now that we have all the maps created, just set the ->end of them:
	 */
2461
	map_groups__fixup_end(&machine->kmaps);
2462 2463 2464
	return 0;
}

2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492
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;
2493 2494 2495 2496 2497 2498 2499 2500

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

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

2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524
	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;
}

2525
size_t machine__fprintf_vmlinux_path(struct machine *machine, FILE *fp)
2526 2527 2528
{
	int i;
	size_t printed = 0;
2529
	struct dso *kdso = machine->vmlinux_maps[MAP__FUNCTION]->dso;
2530 2531 2532 2533 2534 2535

	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);
	}
2536 2537

	for (i = 0; i < vmlinux_path__nr_entries; ++i)
2538 2539
		printed += fprintf(fp, "[%d] %s\n",
				   i + kdso->has_build_id, vmlinux_path[i]);
2540 2541 2542 2543

	return printed;
}

2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557
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;
}

2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576
static bool symbol__read_kptr_restrict(void)
{
	bool value = false;

	if (geteuid() != 0) {
		FILE *fp = fopen("/proc/sys/kernel/kptr_restrict", "r");
		if (fp != NULL) {
			char line[8];

			if (fgets(line, sizeof(line), fp) != NULL)
				value = atoi(line) != 0;

			fclose(fp);
		}
	}

	return value;
}

2577
int symbol__init(void)
2578
{
2579 2580
	const char *symfs;

2581 2582 2583
	if (symbol_conf.initialized)
		return 0;

2584 2585
	symbol_conf.priv_size = ALIGN(symbol_conf.priv_size, sizeof(u64));

2586
	elf_version(EV_CURRENT);
2587 2588 2589
	if (symbol_conf.sort_by_name)
		symbol_conf.priv_size += (sizeof(struct symbol_name_rb_node) -
					  sizeof(struct symbol));
2590

2591
	if (symbol_conf.try_vmlinux_path && vmlinux_path__init() < 0)
2592 2593
		return -1;

2594 2595 2596 2597 2598
	if (symbol_conf.field_sep && *symbol_conf.field_sep == '.') {
		pr_err("'.' is the only non valid --field-separator argument\n");
		return -1;
	}

2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610
	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;

2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622
	/*
	 * 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);

2623 2624
	symbol_conf.kptr_restrict = symbol__read_kptr_restrict();

2625
	symbol_conf.initialized = true;
2626
	return 0;
2627 2628 2629

out_free_comm_list:
	strlist__delete(symbol_conf.comm_list);
2630 2631
out_free_dso_list:
	strlist__delete(symbol_conf.dso_list);
2632
	return -1;
2633 2634
}

2635 2636
void symbol__exit(void)
{
2637 2638
	if (!symbol_conf.initialized)
		return;
2639 2640 2641 2642 2643
	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;
2644
	symbol_conf.initialized = false;
2645 2646
}

2647
int machines__create_kernel_maps(struct rb_root *machines, pid_t pid)
2648
{
2649
	struct machine *machine = machines__findnew(machines, pid);
2650

2651
	if (machine == NULL)
2652
		return -1;
2653

2654
	return machine__create_kernel_maps(machine);
2655
}
2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698

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;
}
2699

2700
int machines__create_guest_kernel_maps(struct rb_root *machines)
2701 2702 2703 2704 2705 2706 2707 2708 2709 2710
{
	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) {
2711
		machines__create_kernel_maps(machines, DEFAULT_GUEST_KERNEL_ID);
2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731
	}

	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;
			}
2732
			machines__create_kernel_maps(machines, pid);
2733 2734 2735 2736 2737 2738 2739
		}
failure:
		free(namelist);
	}

	return ret;
}
2740

2741
void machines__destroy_guest_kernel_maps(struct rb_root *machines)
2742
{
2743
	struct rb_node *next = rb_first(machines);
2744 2745 2746 2747 2748

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

		next = rb_next(&pos->rb_node);
2749
		rb_erase(&pos->rb_node, machines);
2750 2751 2752 2753
		machine__delete(pos);
	}
}

2754
int machine__load_kallsyms(struct machine *machine, const char *filename,
2755 2756
			   enum map_type type, symbol_filter_t filter)
{
2757
	struct map *map = machine->vmlinux_maps[type];
2758 2759 2760 2761 2762 2763 2764 2765 2766
	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.
		 */
2767
		__map_groups__fixup_end(&machine->kmaps, type);
2768 2769 2770 2771 2772
	}

	return ret;
}

2773
int machine__load_vmlinux_path(struct machine *machine, enum map_type type,
2774 2775
			       symbol_filter_t filter)
{
2776
	struct map *map = machine->vmlinux_maps[type];
2777 2778 2779 2780 2781 2782 2783 2784 2785
	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;
}