symbol.c 59.3 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 35
static bool dso__build_id_equal(const struct dso *self, u8 *build_id);
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 *self, struct map *map,
39
				symbol_filter_t filter);
40 41
static int dso__load_guest_kernel_sym(struct dso *self, struct map *map,
			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 53 54 55 56 57 58 59
int dso__name_len(const struct dso *self)
{
	if (verbose)
		return self->long_name_len;

	return self->short_name_len;
}

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

65 66 67 68 69 70 71 72 73 74
bool dso__sorted_by_name(const struct dso *self, enum map_type type)
{
	return self->sorted_by_name & (1 << type);
}

static void dso__set_sorted_by_name(struct dso *self, enum map_type type)
{
	self->sorted_by_name |= (1 << type);
}

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 *self)
88
{
89
	struct rb_node *nd, *prevnd = rb_first(self);
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 *self, enum map_type type)
111 112
{
	struct map *prev, *curr;
113
	struct rb_node *nd, *prevnd = rb_first(&self->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 *self)
134 135 136
{
	int i;
	for (i = 0; i < MAP__NR_TYPES; ++i)
137
		__map_groups__fixup_end(self, 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
	struct symbol *self = calloc(1, (symbol_conf.priv_size +
					 sizeof(*self) + namelen));
146
	if (self == NULL)
147 148
		return NULL;

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

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

157
	pr_debug4("%s: %s %#" PRIx64 "-%#" PRIx64 "\n", __func__, name, start, self->end);
158

159
	memcpy(self->name, name, namelen);
160 161 162 163

	return self;
}

164
void symbol__delete(struct symbol *self)
165
{
166
	free(((void *)self) - symbol_conf.priv_size);
167 168 169 170
}

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

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

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

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

199
struct dso *dso__new(const char *name)
200
{
201
	struct dso *self = calloc(1, sizeof(*self) + strlen(name) + 1);
202 203

	if (self != NULL) {
204
		int i;
205
		strcpy(self->name, name);
206
		dso__set_long_name(self, self->name);
207
		dso__set_short_name(self, self->name);
208
		for (i = 0; i < MAP__NR_TYPES; ++i)
209
			self->symbols[i] = self->symbol_names[i] = RB_ROOT;
210
		self->symtab_type = SYMTAB__NOT_FOUND;
211
		self->loaded = 0;
212
		self->sorted_by_name = 0;
213
		self->has_build_id = 0;
214
		self->kernel = DSO_TYPE_USER;
215
		INIT_LIST_HEAD(&self->node);
216 217 218 219 220
	}

	return self;
}

221
static void symbols__delete(struct rb_root *self)
222 223
{
	struct symbol *pos;
224
	struct rb_node *next = rb_first(self);
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, self);
230
		symbol__delete(pos);
231 232 233 234 235
	}
}

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

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

252
static void symbols__insert(struct rb_root *self, struct symbol *sym)
253
{
254
	struct rb_node **p = &self->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, self);
269 270
}

271
static struct symbol *symbols__find(struct rb_root *self, u64 ip)
272 273 274 275 276 277
{
	struct rb_node *n;

	if (self == NULL)
		return NULL;

278
	n = self->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;
}

294 295 296 297 298 299 300 301 302
struct symbol_name_rb_node {
	struct rb_node	rb_node;
	struct symbol	sym;
};

static void symbols__insert_by_name(struct rb_root *self, struct symbol *sym)
{
	struct rb_node **p = &self->rb_node;
	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);
306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357

	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);
	rb_insert_color(&symn->rb_node, self);
}

static void symbols__sort_by_name(struct rb_root *self, struct rb_root *source)
{
	struct rb_node *nd;

	for (nd = rb_first(source); nd; nd = rb_next(nd)) {
		struct symbol *pos = rb_entry(nd, struct symbol, rb_node);
		symbols__insert_by_name(self, pos);
	}
}

static struct symbol *symbols__find_by_name(struct rb_root *self, const char *name)
{
	struct rb_node *n;

	if (self == NULL)
		return NULL;

	n = self->rb_node;

	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;
}

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

362 363 364 365 366 367 368 369 370 371 372 373 374
struct symbol *dso__find_symbol_by_name(struct dso *self, enum map_type type,
					const char *name)
{
	return symbols__find_by_name(&self->symbol_names[type], name);
}

void dso__sort_by_name(struct dso *self, enum map_type type)
{
	dso__set_sorted_by_name(self, type);
	return symbols__sort_by_name(&self->symbol_names[type],
				     &self->symbols[type]);
}

375
int build_id__sprintf(const u8 *self, int len, char *bf)
376
{
377
	char *bid = bf;
378
	const u8 *raw = self;
379
	int i;
380

381 382 383 384 385 386 387 388 389
	for (i = 0; i < len; ++i) {
		sprintf(bid, "%02x", *raw);
		++raw;
		bid += 2;
	}

	return raw - self;
}

390
size_t dso__fprintf_buildid(struct dso *self, FILE *fp)
391 392 393 394
{
	char sbuild_id[BUILD_ID_SIZE * 2 + 1];

	build_id__sprintf(self->build_id, sizeof(self->build_id), sbuild_id);
395 396 397
	return fprintf(fp, "%s", sbuild_id);
}

398 399 400 401 402 403 404 405 406 407 408 409 410 411
size_t dso__fprintf_symbols_by_name(struct dso *self, enum map_type type, FILE *fp)
{
	size_t ret = 0;
	struct rb_node *nd;
	struct symbol_name_rb_node *pos;

	for (nd = rb_first(&self->symbol_names[type]); nd; nd = rb_next(nd)) {
		pos = rb_entry(nd, struct symbol_name_rb_node, rb_node);
		fprintf(fp, "%s\n", pos->sym.name);
	}

	return ret;
}

412
size_t dso__fprintf(struct dso *self, enum map_type type, FILE *fp)
413 414 415 416
{
	struct rb_node *nd;
	size_t ret = fprintf(fp, "dso: %s (", self->short_name);

417 418 419 420
	if (self->short_name != self->long_name)
		ret += fprintf(fp, "%s, ", self->long_name);
	ret += fprintf(fp, "%s, %sloaded, ", map_type__name[type],
		       self->loaded ? "" : "NOT ");
421
	ret += dso__fprintf_buildid(self, fp);
422
	ret += fprintf(fp, ")\n");
423 424 425
	for (nd = rb_first(&self->symbols[type]); nd; nd = rb_next(nd)) {
		struct symbol *pos = rb_entry(nd, struct symbol, rb_node);
		ret += symbol__fprintf(pos, fp);
426 427 428 429 430
	}

	return ret;
}

431 432
int kallsyms__parse(const char *filename, void *arg,
		    int (*process_symbol)(void *arg, const char *name,
433
					  char type, u64 start, u64 end))
434 435 436
{
	char *line = NULL;
	size_t n;
437 438 439 440
	int err = -1;
	u64 prev_start = 0;
	char prev_symbol_type = 0;
	char *prev_symbol_name;
441
	FILE *file = fopen(filename, "r");
442 443 444 445

	if (file == NULL)
		goto out_failure;

446 447 448 449 450 451
	prev_symbol_name = malloc(KSYM_NAME_LEN);
	if (prev_symbol_name == NULL)
		goto out_close;

	err = 0;

452
	while (!feof(file)) {
453
		u64 start;
454 455
		int line_len, len;
		char symbol_type;
456
		char *symbol_name;
457 458

		line_len = getline(&line, &n, file);
459
		if (line_len < 0 || !line)
460 461 462 463
			break;

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

464
		len = hex2u64(line, &start);
465 466 467 468 469 470

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

		symbol_type = toupper(line[len]);
471 472 473
		len += 2;
		symbol_name = line + len;
		len = line_len - len;
474

475 476
		if (len >= KSYM_NAME_LEN) {
			err = -1;
477
			break;
478 479 480 481 482 483 484 485 486 487 488 489 490 491 492
		}

		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;
493 494
	}

495
	free(prev_symbol_name);
496
	free(line);
497
out_close:
498
	fclose(file);
499
	return err;
500 501 502 503 504

out_failure:
	return -1;
}

505 506 507 508 509
struct process_kallsyms_args {
	struct map *map;
	struct dso *dso;
};

510 511 512 513 514 515 516 517
static u8 kallsyms2elf_type(char type)
{
	if (type == 'W')
		return STB_WEAK;

	return isupper(type) ? STB_GLOBAL : STB_LOCAL;
}

518
static int map__process_kallsym_symbol(void *arg, const char *name,
519
				       char type, u64 start, u64 end)
520 521 522 523 524 525 526 527
{
	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;

528 529
	sym = symbol__new(start, end - start + 1,
			  kallsyms2elf_type(type), name);
530 531 532 533 534 535 536
	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);
537

538 539 540 541 542 543 544 545
	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.
 */
546 547
static int dso__load_all_kallsyms(struct dso *self, const char *filename,
				  struct map *map)
548 549
{
	struct process_kallsyms_args args = { .map = map, .dso = self, };
550
	return kallsyms__parse(filename, &args, map__process_kallsym_symbol);
551 552
}

553 554 555 556 557
/*
 * 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.
 */
558
static int dso__split_kallsyms(struct dso *self, struct map *map,
559
			       symbol_filter_t filter)
560
{
561
	struct map_groups *kmaps = map__kmap(map)->kmaps;
562
	struct machine *machine = kmaps->machine;
563
	struct map *curr_map = map;
564
	struct symbol *pos;
565
	int count = 0, moved = 0;	
566 567
	struct rb_root *root = &self->symbols[map->type];
	struct rb_node *next = rb_first(root);
568 569 570 571 572 573 574 575 576 577
	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) {
578
			if (!symbol_conf.use_modules)
579 580
				goto discard_symbol;

581 582
			*module++ = '\0';

583
			if (strcmp(curr_map->dso->short_name, module)) {
584
				if (curr_map != map &&
585 586
				    self->kernel == DSO_TYPE_GUEST_KERNEL &&
				    machine__is_default_guest(machine)) {
587 588 589 590 591 592 593 594 595 596 597 598 599
					/*
					 * 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);
600
				if (curr_map == NULL) {
601
					pr_debug("%s/proc/{kallsyms,modules} "
602
					         "inconsistency while looking "
603
						 "for \"%s\" module!\n",
604
						 machine->root_dir, module);
605 606
					curr_map = map;
					goto discard_symbol;
607
				}
608

609
				if (curr_map->dso->loaded &&
610
				    !machine__is_default_guest(machine))
611
					goto discard_symbol;
612
			}
613 614 615 616
			/*
			 * So that we look just like we get from .ko files,
			 * i.e. not prelinked, relative to map->start.
			 */
617 618 619
			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) {
620 621 622
			char dso_name[PATH_MAX];
			struct dso *dso;

623 624 625 626 627
			if (count == 0) {
				curr_map = map;
				goto filter_symbol;
			}

628 629 630 631 632 633 634 635
			if (self->kernel == DSO_TYPE_GUEST_KERNEL)
				snprintf(dso_name, sizeof(dso_name),
					"[guest.kernel].%d",
					kernel_range++);
			else
				snprintf(dso_name, sizeof(dso_name),
					"[kernel].%d",
					kernel_range++);
636

637
			dso = dso__new(dso_name);
638 639 640
			if (dso == NULL)
				return -1;

641 642
			dso->kernel = self->kernel;

643
			curr_map = map__new2(pos->start, dso, map->type);
644
			if (curr_map == NULL) {
645 646 647
				dso__delete(dso);
				return -1;
			}
648

649
			curr_map->map_ip = curr_map->unmap_ip = identity__map_ip;
650
			map_groups__insert(kmaps, curr_map);
651 652
			++kernel_range;
		}
653
filter_symbol:
654
		if (filter && filter(curr_map, pos)) {
655
discard_symbol:		rb_erase(&pos->rb_node, root);
656
			symbol__delete(pos);
657
		} else {
658 659 660
			if (curr_map != map) {
				rb_erase(&pos->rb_node, root);
				symbols__insert(&curr_map->dso->symbols[curr_map->type], pos);
661 662 663
				++moved;
			} else
				++count;
664
		}
665 666
	}

667 668
	if (curr_map != map &&
	    self->kernel == DSO_TYPE_GUEST_KERNEL &&
669
	    machine__is_default_guest(kmaps->machine)) {
670 671 672
		dso__set_loaded(curr_map->dso, curr_map->type);
	}

673
	return count + moved;
674
}
675

676 677
int dso__load_kallsyms(struct dso *self, const char *filename,
		       struct map *map, symbol_filter_t filter)
678
{
679
	if (dso__load_all_kallsyms(self, filename, map) < 0)
680 681
		return -1;

682
	if (self->kernel == DSO_TYPE_GUEST_KERNEL)
683
		self->symtab_type = SYMTAB__GUEST_KALLSYMS;
684
	else
685
		self->symtab_type = SYMTAB__KALLSYMS;
686

687
	return dso__split_kallsyms(self, map, filter);
688 689
}

690
static int dso__load_perf_map(struct dso *self, struct map *map,
691
			      symbol_filter_t filter)
692 693 694 695 696 697
{
	char *line = NULL;
	size_t n;
	FILE *file;
	int nr_syms = 0;

698
	file = fopen(self->long_name, "r");
699 700 701 702
	if (file == NULL)
		goto out_failure;

	while (!feof(file)) {
703
		u64 start, size;
704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727
		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;

728
		sym = symbol__new(start, size, STB_GLOBAL, line + len);
729 730 731 732

		if (sym == NULL)
			goto out_delete_line;

733
		if (filter && filter(map, sym))
734
			symbol__delete(sym);
735
		else {
736
			symbols__insert(&self->symbols[map->type], sym);
737 738 739 740 741 742 743 744 745 746 747 748 749 750 751
			nr_syms++;
		}
	}

	free(line);
	fclose(file);

	return nr_syms;

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

752 753 754 755
/**
 * elf_symtab__for_each_symbol - iterate thru all the symbols
 *
 * @self: struct elf_symtab instance to iterate
756
 * @idx: uint32_t idx
757 758
 * @sym: GElf_Sym iterator
 */
759 760 761 762
#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))
763 764 765 766 767 768 769 770 771 772

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 &&
773
	       sym->st_shndx != SHN_UNDEF;
774 775
}

776 777 778 779 780 781 782
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;
}

783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802
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;
}

803 804 805 806 807 808
static inline bool elf_sec__is_data(const GElf_Shdr *shdr,
				    const Elf_Data *secstrs)
{
	return strstr(elf_sec__name(shdr, secstrs), "data") != NULL;
}

809 810 811 812 813 814 815 816
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,
817
				    size_t *idx)
818 819 820 821 822 823 824 825 826 827
{
	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)) {
828 829
			if (idx)
				*idx = cnt;
830 831 832 833 834 835 836 837
			break;
		}
		++cnt;
	}

	return sec;
}

838 839 840 841 842 843 844 845 846 847
#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))

848 849 850 851 852 853 854
/*
 * 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).
 */
855 856
static int dso__synthesize_plt_symbols(struct  dso *self, struct map *map,
				       symbol_filter_t filter)
857 858 859
{
	uint32_t nr_rel_entries, idx;
	GElf_Sym sym;
860
	u64 plt_offset;
861 862
	GElf_Shdr shdr_plt;
	struct symbol *f;
863
	GElf_Shdr shdr_rel_plt, shdr_dynsym;
864
	Elf_Data *reldata, *syms, *symstrs;
865 866 867
	Elf_Scn *scn_plt_rel, *scn_symstrs, *scn_dynsym;
	size_t dynsym_idx;
	GElf_Ehdr ehdr;
868
	char sympltname[1024];
869 870
	Elf *elf;
	int nr = 0, symidx, fd, err = 0;
871
	char name[PATH_MAX];
872

873 874 875
	snprintf(name, sizeof(name), "%s%s",
		 symbol_conf.symfs, self->long_name);
	fd = open(name, O_RDONLY);
876 877 878
	if (fd < 0)
		goto out;

879
	elf = elf_begin(fd, PERF_ELF_C_READ_MMAP, NULL);
880 881 882 883 884 885 886 887 888 889
	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;
890

891
	scn_plt_rel = elf_section_by_name(elf, &ehdr, &shdr_rel_plt,
892 893
					  ".rela.plt", NULL);
	if (scn_plt_rel == NULL) {
894
		scn_plt_rel = elf_section_by_name(elf, &ehdr, &shdr_rel_plt,
895 896
						  ".rel.plt", NULL);
		if (scn_plt_rel == NULL)
897
			goto out_elf_end;
898 899
	}

900 901
	err = -1;

902
	if (shdr_rel_plt.sh_link != dynsym_idx)
903
		goto out_elf_end;
904

905 906
	if (elf_section_by_name(elf, &ehdr, &shdr_plt, ".plt", NULL) == NULL)
		goto out_elf_end;
907 908

	/*
909
	 * Fetch the relocation section to find the idxes to the GOT
910 911 912 913
	 * and the symbols in the .dynsym they refer to.
	 */
	reldata = elf_getdata(scn_plt_rel, NULL);
	if (reldata == NULL)
914
		goto out_elf_end;
915 916 917

	syms = elf_getdata(scn_dynsym, NULL);
	if (syms == NULL)
918
		goto out_elf_end;
919

920
	scn_symstrs = elf_getscn(elf, shdr_dynsym.sh_link);
921
	if (scn_symstrs == NULL)
922
		goto out_elf_end;
923 924 925

	symstrs = elf_getdata(scn_symstrs, NULL);
	if (symstrs == NULL)
926
		goto out_elf_end;
927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942

	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,
943
					STB_GLOBAL, sympltname);
944
			if (!f)
945
				goto out_elf_end;
946

947 948 949
			if (filter && filter(map, f))
				symbol__delete(f);
			else {
950
				symbols__insert(&self->symbols[map->type], f);
951 952
				++nr;
			}
953 954 955 956 957 958 959 960 961 962 963 964
		}
	} 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,
965
					STB_GLOBAL, sympltname);
966
			if (!f)
967
				goto out_elf_end;
968

969 970 971
			if (filter && filter(map, f))
				symbol__delete(f);
			else {
972
				symbols__insert(&self->symbols[map->type], f);
973 974
				++nr;
			}
975 976 977
		}
	}

978 979 980 981 982 983 984 985 986
	err = 0;
out_elf_end:
	elf_end(elf);
out_close:
	close(fd);

	if (err == 0)
		return nr;
out:
987 988
	pr_debug("%s: problems reading %s PLT info.\n",
		 __func__, self->long_name);
989
	return 0;
990 991
}

992 993 994 995 996
static bool elf_sym__is_a(GElf_Sym *self, enum map_type type)
{
	switch (type) {
	case MAP__FUNCTION:
		return elf_sym__is_function(self);
997 998
	case MAP__VARIABLE:
		return elf_sym__is_object(self);
999 1000 1001 1002 1003 1004 1005 1006 1007 1008
	default:
		return false;
	}
}

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

1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034
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;
}

1035
static int dso__load_sym(struct dso *self, struct map *map, const char *name,
1036 1037
			 int fd, symbol_filter_t filter, int kmodule,
			 int want_symtab)
1038
{
1039
	struct kmap *kmap = self->kernel ? map__kmap(map) : NULL;
1040 1041
	struct map *curr_map = map;
	struct dso *curr_dso = self;
1042
	Elf_Data *symstrs, *secstrs;
1043 1044
	uint32_t nr_syms;
	int err = -1;
1045
	uint32_t idx;
1046
	GElf_Ehdr ehdr;
1047 1048
	GElf_Shdr shdr, opdshdr;
	Elf_Data *syms, *opddata = NULL;
1049
	GElf_Sym sym;
1050
	Elf_Scn *sec, *sec_strndx, *opdsec;
1051
	Elf *elf;
1052
	int nr = 0;
1053
	size_t opdidx = 0;
1054

1055
	elf = elf_begin(fd, PERF_ELF_C_READ_MMAP, NULL);
1056
	if (elf == NULL) {
1057
		pr_debug("%s: cannot read %s ELF file.\n", __func__, name);
1058 1059 1060 1061
		goto out_close;
	}

	if (gelf_getehdr(elf, &ehdr) == NULL) {
1062
		pr_debug("%s: cannot get elf header.\n", __func__);
1063 1064 1065
		goto out_elf_end;
	}

1066
	/* Always reject images with a mismatched build-id: */
1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077
	if (self->has_build_id) {
		u8 build_id[BUILD_ID_SIZE];

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

		if (!dso__build_id_equal(self, build_id))
			goto out_elf_end;
	}

1078
	sec = elf_section_by_name(elf, &ehdr, &shdr, ".symtab", NULL);
1079
	if (sec == NULL) {
1080 1081 1082
		if (want_symtab)
			goto out_elf_end;

1083 1084
		sec = elf_section_by_name(elf, &ehdr, &shdr, ".dynsym", NULL);
		if (sec == NULL)
1085 1086
			goto out_elf_end;
	}
1087

1088 1089 1090 1091
	opdsec = elf_section_by_name(elf, &ehdr, &opdshdr, ".opd", &opdidx);
	if (opdsec)
		opddata = elf_rawdata(opdsec, NULL);

1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103
	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;

1104 1105 1106 1107 1108
	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
1109
	if (secstrs == NULL)
1110 1111
		goto out_elf_end;

1112 1113
	nr_syms = shdr.sh_size / shdr.sh_entsize;

1114
	memset(&sym, 0, sizeof(sym));
1115
	if (self->kernel == DSO_TYPE_USER) {
1116
		self->adjust_symbols = (ehdr.e_type == ET_EXEC ||
1117 1118 1119
				elf_section_by_name(elf, &ehdr, &shdr,
						     ".gnu.prelink_undo",
						     NULL) != NULL);
1120 1121
	} else self->adjust_symbols = 0;

1122
	elf_symtab__for_each_symbol(syms, nr_syms, idx, sym) {
1123
		struct symbol *f;
1124
		const char *elf_name = elf_sym__name(&sym, symstrs);
1125
		char *demangled = NULL;
1126 1127
		int is_label = elf_sym__is_label(&sym);
		const char *section_name;
1128

1129 1130 1131
		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;
1132

1133
		if (!is_label && !elf_sym__is_a(&sym, map->type))
1134 1135
			continue;

1136 1137 1138 1139 1140 1141 1142 1143 1144 1145
		/* 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;
		}

1146 1147 1148 1149 1150 1151 1152
		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);
		}

1153 1154 1155 1156 1157
		sec = elf_getscn(elf, sym.st_shndx);
		if (!sec)
			goto out_elf_end;

		gelf_getshdr(sec, &shdr);
1158

1159
		if (is_label && !elf_sec__is_a(&shdr, secstrs, map->type))
1160 1161 1162
			continue;

		section_name = elf_sec__name(&shdr, secstrs);
1163

1164 1165 1166 1167 1168 1169 1170
		/* 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;

1171
		if (self->kernel != DSO_TYPE_USER || kmodule) {
1172 1173 1174
			char dso_name[PATH_MAX];

			if (strcmp(section_name,
1175 1176
				   (curr_dso->short_name +
				    self->short_name_len)) == 0)
1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187
				goto new_symbol;

			if (strcmp(section_name, ".text") == 0) {
				curr_map = map;
				curr_dso = self;
				goto new_symbol;
			}

			snprintf(dso_name, sizeof(dso_name),
				 "%s%s", self->short_name, section_name);

1188
			curr_map = map_groups__find_by_name(kmap->kmaps, map->type, dso_name);
1189 1190 1191 1192 1193 1194
			if (curr_map == NULL) {
				u64 start = sym.st_value;

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

1195
				curr_dso = dso__new(dso_name);
1196 1197
				if (curr_dso == NULL)
					goto out_elf_end;
1198
				curr_dso->kernel = self->kernel;
1199 1200
				curr_dso->long_name = self->long_name;
				curr_dso->long_name_len = self->long_name_len;
1201
				curr_map = map__new2(start, curr_dso,
1202
						     map->type);
1203 1204 1205 1206
				if (curr_map == NULL) {
					dso__delete(curr_dso);
					goto out_elf_end;
				}
1207 1208
				curr_map->map_ip = identity__map_ip;
				curr_map->unmap_ip = identity__map_ip;
1209
				curr_dso->symtab_type = self->symtab_type;
1210
				map_groups__insert(kmap->kmaps, curr_map);
1211
				dsos__add(&self->node, curr_dso);
1212
				dso__set_loaded(curr_dso, map->type);
1213 1214 1215 1216
			} else
				curr_dso = curr_map->dso;

			goto new_symbol;
1217 1218
		}

1219
		if (curr_dso->adjust_symbols) {
1220 1221
			pr_debug4("%s: adjusting symbol: st_value: %#" PRIx64 " "
				  "sh_addr: %#" PRIx64 " sh_offset: %#" PRIx64 "\n", __func__,
1222 1223
				  (u64)sym.st_value, (u64)shdr.sh_addr,
				  (u64)shdr.sh_offset);
1224
			sym.st_value -= shdr.sh_addr - shdr.sh_offset;
1225
		}
1226 1227 1228 1229 1230
		/*
		 * 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...
		 */
1231
		demangled = bfd_demangle(NULL, elf_name, DMGL_PARAMS | DMGL_ANSI);
1232
		if (demangled != NULL)
1233
			elf_name = demangled;
1234
new_symbol:
1235 1236
		f = symbol__new(sym.st_value, sym.st_size,
				GELF_ST_BIND(sym.st_info), elf_name);
1237
		free(demangled);
1238 1239 1240
		if (!f)
			goto out_elf_end;

1241
		if (filter && filter(curr_map, f))
1242
			symbol__delete(f);
1243
		else {
1244
			symbols__insert(&curr_dso->symbols[curr_map->type], f);
1245 1246
			nr++;
		}
1247 1248
	}

1249 1250 1251
	/*
	 * For misannotated, zeroed, ASM function sizes.
	 */
1252
	if (nr > 0) {
1253
		symbols__fixup_end(&self->symbols[map->type]);
1254 1255 1256 1257 1258 1259 1260 1261
		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);
		}
	}
1262 1263 1264 1265 1266 1267 1268
	err = nr;
out_elf_end:
	elf_end(elf);
out_close:
	return err;
}

1269 1270 1271 1272 1273
static bool dso__build_id_equal(const struct dso *self, u8 *build_id)
{
	return memcmp(self->build_id, build_id, sizeof(self->build_id)) == 0;
}

1274
bool __dsos__read_build_ids(struct list_head *head, bool with_hits)
1275
{
1276
	bool have_build_id = false;
1277 1278
	struct dso *pos;

1279 1280 1281
	list_for_each_entry(pos, head, node) {
		if (with_hits && !pos->hit)
			continue;
1282 1283 1284 1285
		if (pos->has_build_id) {
			have_build_id = true;
			continue;
		}
1286 1287 1288 1289 1290
		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;
		}
1291
	}
1292

1293
	return have_build_id;
1294 1295
}

1296 1297 1298 1299 1300
/*
 * Align offset to 4 bytes as needed for note name and descriptor data.
 */
#define NOTE_ALIGN(n) (((n) + 3) & -4U)

1301
static int elf_read_build_id(Elf *elf, void *bf, size_t size)
1302
{
1303
	int err = -1;
1304 1305
	GElf_Ehdr ehdr;
	GElf_Shdr shdr;
1306
	Elf_Data *data;
1307
	Elf_Scn *sec;
1308
	Elf_Kind ek;
1309
	void *ptr;
1310

1311 1312 1313
	if (size < BUILD_ID_SIZE)
		goto out;

1314 1315
	ek = elf_kind(elf);
	if (ek != ELF_K_ELF)
1316
		goto out;
1317

1318
	if (gelf_getehdr(elf, &ehdr) == NULL) {
1319
		pr_err("%s: cannot get elf header.\n", __func__);
1320
		goto out;
1321 1322
	}

1323 1324
	sec = elf_section_by_name(elf, &ehdr, &shdr,
				  ".note.gnu.build-id", NULL);
1325 1326 1327 1328
	if (sec == NULL) {
		sec = elf_section_by_name(elf, &ehdr, &shdr,
					  ".notes", NULL);
		if (sec == NULL)
1329
			goto out;
1330
	}
1331

1332 1333
	data = elf_getdata(sec, NULL);
	if (data == NULL)
1334
		goto out;
1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355

	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;
	}
1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380

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);

1381 1382 1383 1384 1385 1386 1387
	elf_end(elf);
out_close:
	close(fd);
out:
	return err;
}

1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406
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;

1407 1408
		namesz = NOTE_ALIGN(nhdr.n_namesz);
		descsz = NOTE_ALIGN(nhdr.n_descsz);
1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431
		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;
}

1432 1433 1434
char dso__symtab_origin(const struct dso *self)
{
	static const char origin[] = {
1435 1436 1437 1438 1439 1440 1441 1442 1443 1444
		[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',
1445 1446
	};

1447
	if (self == NULL || self->symtab_type == SYMTAB__NOT_FOUND)
1448
		return '!';
1449
	return origin[self->symtab_type];
1450 1451
}

1452
int dso__load(struct dso *self, struct map *map, symbol_filter_t filter)
1453
{
1454
	int size = PATH_MAX;
1455
	char *name;
1456 1457
	int ret = -1;
	int fd;
1458
	struct machine *machine;
1459
	const char *root_dir;
1460
	int want_symtab;
1461

1462
	dso__set_loaded(self, map->type);
1463

1464
	if (self->kernel == DSO_TYPE_KERNEL)
1465
		return dso__load_kernel_sym(self, map, filter);
1466 1467 1468
	else if (self->kernel == DSO_TYPE_GUEST_KERNEL)
		return dso__load_guest_kernel_sym(self, map, filter);

1469 1470
	if (map->groups && map->groups->machine)
		machine = map->groups->machine;
1471
	else
1472
		machine = NULL;
1473 1474

	name = malloc(size);
1475 1476 1477
	if (!name)
		return -1;

1478
	self->adjust_symbols = 0;
1479

1480
	if (strncmp(self->name, "/tmp/perf-", 10) == 0) {
1481
		ret = dso__load_perf_map(self, map, filter);
1482 1483
		self->symtab_type = ret > 0 ? SYMTAB__JAVA_JIT :
					      SYMTAB__NOT_FOUND;
1484 1485 1486
		return ret;
	}

1487 1488 1489 1490
	/* 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
	 */
1491 1492 1493
	want_symtab = 1;
restart:
	for (self->symtab_type = SYMTAB__BUILD_ID_CACHE;
1494 1495 1496 1497
	     self->symtab_type != SYMTAB__NOT_FOUND;
	     self->symtab_type++) {
		switch (self->symtab_type) {
		case SYMTAB__BUILD_ID_CACHE:
1498 1499 1500
			/* skip the locally configured cache if a symfs is given */
			if (symbol_conf.symfs[0] ||
			    (dso__build_id_filename(self, name, size) == NULL)) {
1501
				continue;
1502
			}
1503
			break;
1504
		case SYMTAB__FEDORA_DEBUGINFO:
1505 1506
			snprintf(name, size, "%s/usr/lib/debug%s.debug",
				 symbol_conf.symfs, self->long_name);
1507
			break;
1508
		case SYMTAB__UBUNTU_DEBUGINFO:
1509 1510
			snprintf(name, size, "%s/usr/lib/debug%s",
				 symbol_conf.symfs, self->long_name);
1511
			break;
1512
		case SYMTAB__BUILDID_DEBUGINFO: {
1513 1514 1515 1516 1517 1518 1519 1520 1521
			char build_id_hex[BUILD_ID_SIZE * 2 + 1];

			if (!self->has_build_id)
				continue;

			build_id__sprintf(self->build_id,
					  sizeof(self->build_id),
					  build_id_hex);
			snprintf(name, size,
1522 1523
				 "%s/usr/lib/debug/.build-id/%.2s/%s.debug",
				 symbol_conf.symfs, build_id_hex, build_id_hex + 2);
1524
			}
1525
			break;
1526
		case SYMTAB__SYSTEM_PATH_DSO:
1527 1528
			snprintf(name, size, "%s%s",
			     symbol_conf.symfs, self->long_name);
1529
			break;
1530
		case SYMTAB__GUEST_KMODULE:
K
Kyle McMartin 已提交
1531 1532
			if (map->groups && machine)
				root_dir = machine->root_dir;
1533 1534
			else
				root_dir = "";
1535 1536 1537 1538
			snprintf(name, size, "%s%s%s", symbol_conf.symfs,
				 root_dir, self->long_name);
			break;

1539
		case SYMTAB__SYSTEM_PATH_KMODULE:
1540 1541
			snprintf(name, size, "%s%s", symbol_conf.symfs,
				 self->long_name);
1542
			break;
1543
		default:;
1544
		}
1545 1546

		/* Name is now the name of the next image to try */
1547
		fd = open(name, O_RDONLY);
1548 1549
		if (fd < 0)
			continue;
1550

1551 1552 1553
		ret = dso__load_sym(self, map, name, fd, filter, 0,
				    want_symtab);
		close(fd);
1554

1555 1556 1557 1558 1559 1560
		/*
		 * Some people seem to have debuginfo files _WITHOUT_ debug
		 * info!?!?
		 */
		if (!ret)
			continue;
1561

1562 1563 1564 1565 1566 1567
		if (ret > 0) {
			int nr_plt = dso__synthesize_plt_symbols(self, map, filter);
			if (nr_plt > 0)
				ret += nr_plt;
			break;
		}
1568
	}
1569

1570 1571 1572 1573 1574 1575 1576 1577 1578
	/*
	 * 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;
	}

1579
	free(name);
1580 1581
	if (ret < 0 && strstr(self->name, " (deleted)") != NULL)
		return 0;
1582 1583 1584
	return ret;
}

1585 1586
struct map *map_groups__find_by_name(struct map_groups *self,
				     enum map_type type, const char *name)
1587 1588 1589
{
	struct rb_node *nd;

1590
	for (nd = rb_first(&self->maps[type]); nd; nd = rb_next(nd)) {
1591 1592
		struct map *map = rb_entry(nd, struct map, rb_node);

1593
		if (map->dso && strcmp(map->dso->short_name, name) == 0)
1594 1595 1596 1597 1598 1599
			return map;
	}

	return NULL;
}

1600 1601
static int dso__kernel_module_get_build_id(struct dso *self,
				const char *root_dir)
1602 1603 1604 1605 1606 1607 1608 1609 1610
{
	char filename[PATH_MAX];
	/*
	 * kernel module short names are of the form "[module]" and
	 * we need just "module" here.
	 */
	const char *name = self->short_name + 1;

	snprintf(filename, sizeof(filename),
1611 1612
		 "%s/sys/module/%.*s/notes/.note.gnu.build-id",
		 root_dir, (int)strlen(name) - 1, name);
1613 1614 1615 1616 1617 1618 1619 1620

	if (sysfs__read_build_id(filename, self->build_id,
				 sizeof(self->build_id)) == 0)
		self->has_build_id = true;

	return 0;
}

1621 1622
static int map_groups__set_modules_path_dir(struct map_groups *self,
				const char *dir_name)
1623
{
1624
	struct dirent *dent;
1625
	DIR *dir = opendir(dir_name);
1626
	int ret = 0;
1627

1628
	if (!dir) {
1629
		pr_debug("%s: cannot open %s dir\n", __func__, dir_name);
1630 1631
		return -1;
	}
1632

1633 1634
	while ((dent = readdir(dir)) != NULL) {
		char path[PATH_MAX];
1635 1636 1637 1638 1639 1640
		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;
1641

1642
		if (S_ISDIR(st.st_mode)) {
1643 1644 1645 1646 1647
			if (!strcmp(dent->d_name, ".") ||
			    !strcmp(dent->d_name, ".."))
				continue;

			snprintf(path, sizeof(path), "%s/%s",
1648
				 dir_name, dent->d_name);
1649 1650 1651
			ret = map_groups__set_modules_path_dir(self, path);
			if (ret < 0)
				goto out;
1652 1653 1654 1655
		} else {
			char *dot = strrchr(dent->d_name, '.'),
			     dso_name[PATH_MAX];
			struct map *map;
1656
			char *long_name;
1657 1658 1659 1660 1661 1662

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

1663
			strxfrchar(dso_name, '-', '_');
1664
			map = map_groups__find_by_name(self, MAP__FUNCTION, dso_name);
1665 1666 1667 1668
			if (map == NULL)
				continue;

			snprintf(path, sizeof(path), "%s/%s",
1669
				 dir_name, dent->d_name);
1670

1671
			long_name = strdup(path);
1672 1673 1674 1675
			if (long_name == NULL) {
				ret = -1;
				goto out;
			}
1676
			dso__set_long_name(map->dso, long_name);
1677
			map->dso->lname_alloc = 1;
1678
			dso__kernel_module_get_build_id(map->dso, "");
1679 1680
		}
	}
1681

1682
out:
1683
	closedir(dir);
1684
	return ret;
1685
}
1686

1687
static char *get_kernel_version(const char *root_dir)
1688
{
1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713
	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);
}

1714
static int machine__set_modules_path(struct machine *self)
1715 1716
{
	char *version;
1717
	char modules_path[PATH_MAX];
1718

1719
	version = get_kernel_version(self->root_dir);
1720
	if (!version)
1721
		return -1;
1722

1723
	snprintf(modules_path, sizeof(modules_path), "%s/lib/modules/%s/kernel",
1724
		 self->root_dir, version);
1725
	free(version);
1726

1727
	return map_groups__set_modules_path_dir(&self->kmaps, modules_path);
1728 1729
}

1730 1731 1732 1733 1734
/*
 * 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.
 */
1735
static struct map *map__new2(u64 start, struct dso *dso, enum map_type type)
1736
{
1737 1738
	struct map *self = calloc(1, (sizeof(*self) +
				      (dso->kernel ? sizeof(struct kmap) : 0)));
1739 1740
	if (self != NULL) {
		/*
1741
		 * ->end will be filled after we load all the symbols
1742
		 */
1743
		map__init(self, type, start, 0, 0, dso);
1744
	}
1745

1746 1747 1748
	return self;
}

1749 1750
struct map *machine__new_module(struct machine *self, u64 start,
				const char *filename)
1751 1752
{
	struct map *map;
1753
	struct dso *dso = __dsos__findnew(&self->kernel_dsos, filename);
1754 1755 1756 1757 1758 1759 1760 1761

	if (dso == NULL)
		return NULL;

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

1762
	if (machine__is_host(self))
1763
		dso->symtab_type = SYMTAB__SYSTEM_PATH_KMODULE;
1764
	else
1765
		dso->symtab_type = SYMTAB__GUEST_KMODULE;
1766
	map_groups__insert(&self->kmaps, map);
1767 1768 1769
	return map;
}

1770
static int machine__create_modules(struct machine *self)
1771 1772 1773
{
	char *line = NULL;
	size_t n;
1774
	FILE *file;
1775
	struct map *map;
1776 1777 1778
	const char *modules;
	char path[PATH_MAX];

1779
	if (machine__is_default_guest(self))
1780 1781
		modules = symbol_conf.default_guest_modules;
	else {
1782
		sprintf(path, "%s/proc/modules", self->root_dir);
1783 1784
		modules = path;
	}
1785

1786
	file = fopen(modules, "r");
1787 1788
	if (file == NULL)
		return -1;
1789

1790 1791 1792 1793 1794
	while (!feof(file)) {
		char name[PATH_MAX];
		u64 start;
		char *sep;
		int line_len;
1795

1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817
		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);
1818
		map = machine__new_module(self, start, name);
1819
		if (map == NULL)
1820
			goto out_delete_line;
1821
		dso__kernel_module_get_build_id(map->dso, self->root_dir);
1822
	}
1823 1824 1825 1826

	free(line);
	fclose(file);

1827
	return machine__set_modules_path(self);
1828 1829 1830 1831 1832

out_delete_line:
	free(line);
out_failure:
	return -1;
1833 1834
}

1835 1836
int dso__load_vmlinux(struct dso *self, struct map *map,
		      const char *vmlinux, symbol_filter_t filter)
1837
{
1838
	int err = -1, fd;
1839
	char symfs_vmlinux[PATH_MAX];
1840

1841
	snprintf(symfs_vmlinux, sizeof(symfs_vmlinux), "%s%s",
1842 1843
		 symbol_conf.symfs, vmlinux);
	fd = open(symfs_vmlinux, O_RDONLY);
1844 1845 1846
	if (fd < 0)
		return -1;

1847
	dso__set_long_name(self, (char *)vmlinux);
1848
	dso__set_loaded(self, map->type);
1849
	err = dso__load_sym(self, map, symfs_vmlinux, fd, filter, 0, 0);
1850 1851
	close(fd);

1852
	if (err > 0)
1853
		pr_debug("Using %s for symbols\n", symfs_vmlinux);
1854

1855 1856 1857
	return err;
}

1858
int dso__load_vmlinux_path(struct dso *self, struct map *map,
1859
			   symbol_filter_t filter)
1860 1861
{
	int i, err = 0;
1862
	char *filename;
1863 1864

	pr_debug("Looking at the vmlinux_path (%d entries long)\n",
1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875
		 vmlinux_path__nr_entries + 1);

	filename = dso__build_id_filename(self, NULL, 0);
	if (filename != NULL) {
		err = dso__load_vmlinux(self, map, filename, filter);
		if (err > 0) {
			dso__set_long_name(self, filename);
			goto out;
		}
		free(filename);
	}
1876 1877

	for (i = 0; i < vmlinux_path__nr_entries; ++i) {
1878
		err = dso__load_vmlinux(self, map, vmlinux_path[i], filter);
1879 1880 1881 1882 1883
		if (err > 0) {
			dso__set_long_name(self, strdup(vmlinux_path[i]));
			break;
		}
	}
1884
out:
1885 1886 1887
	return err;
}

1888
static int dso__load_kernel_sym(struct dso *self, struct map *map,
1889
				symbol_filter_t filter)
1890
{
1891
	int err;
1892 1893
	const char *kallsyms_filename = NULL;
	char *kallsyms_allocated_filename = NULL;
1894
	/*
1895 1896
	 * 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.
1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908
	 *
	 * 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.
	 */
1909 1910 1911 1912 1913
	if (symbol_conf.kallsyms_name != NULL) {
		kallsyms_filename = symbol_conf.kallsyms_name;
		goto do_kallsyms;
	}

1914
	if (symbol_conf.vmlinux_name != NULL) {
1915
		err = dso__load_vmlinux(self, map,
1916
					symbol_conf.vmlinux_name, filter);
1917 1918 1919 1920 1921 1922
		if (err > 0) {
			dso__set_long_name(self,
					   strdup(symbol_conf.vmlinux_name));
			goto out_fixup;
		}
		return err;
1923
	}
1924 1925

	if (vmlinux_path != NULL) {
1926
		err = dso__load_vmlinux_path(self, map, filter);
1927 1928
		if (err > 0)
			goto out_fixup;
1929 1930
	}

1931 1932 1933 1934
	/* do not try local files if a symfs was given */
	if (symbol_conf.symfs[0] != 0)
		return -1;

1935 1936 1937 1938 1939 1940 1941
	/*
	 * 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.
	 */
	if (self->has_build_id) {
		u8 kallsyms_build_id[BUILD_ID_SIZE];
1942
		char sbuild_id[BUILD_ID_SIZE * 2 + 1];
1943 1944

		if (sysfs__read_build_id("/sys/kernel/notes", kallsyms_build_id,
1945
					 sizeof(kallsyms_build_id)) == 0) {
1946 1947
			if (dso__build_id_equal(self, kallsyms_build_id)) {
				kallsyms_filename = "/proc/kallsyms";
1948
				goto do_kallsyms;
1949
			}
1950
		}
1951 1952 1953 1954
		/*
		 * Now look if we have it on the build-id cache in
		 * $HOME/.debug/[kernel.kallsyms].
		 */
1955 1956 1957 1958 1959
		build_id__sprintf(self->build_id, sizeof(self->build_id),
				  sbuild_id);

		if (asprintf(&kallsyms_allocated_filename,
			     "%s/.debug/[kernel.kallsyms]/%s",
1960 1961
			     getenv("HOME"), sbuild_id) == -1) {
			pr_err("Not enough memory for kallsyms file lookup\n");
1962
			return -1;
1963
		}
1964

1965 1966
		kallsyms_filename = kallsyms_allocated_filename;

1967
		if (access(kallsyms_filename, F_OK)) {
1968 1969
			pr_err("No kallsyms or vmlinux with build-id %s "
			       "was found\n", sbuild_id);
1970
			free(kallsyms_allocated_filename);
1971
			return -1;
1972
		}
1973 1974 1975 1976 1977
	} else {
		/*
		 * Last resort, if we don't have a build-id and couldn't find
		 * any vmlinux file, try the running kernel kallsyms table.
		 */
1978 1979
		kallsyms_filename = "/proc/kallsyms";
	}
1980

1981
do_kallsyms:
1982
	err = dso__load_kallsyms(self, kallsyms_filename, map, filter);
1983 1984
	if (err > 0)
		pr_debug("Using %s for symbols\n", kallsyms_filename);
1985
	free(kallsyms_allocated_filename);
1986 1987

	if (err > 0) {
1988
out_fixup:
1989
		if (kallsyms_filename != NULL)
1990
			dso__set_long_name(self, strdup("[kernel.kallsyms]"));
1991 1992
		map__fixup_start(map);
		map__fixup_end(map);
1993
	}
1994

1995 1996 1997
	return err;
}

1998 1999 2000 2001 2002
static int dso__load_guest_kernel_sym(struct dso *self, struct map *map,
				symbol_filter_t filter)
{
	int err;
	const char *kallsyms_filename = NULL;
2003
	struct machine *machine;
2004 2005 2006 2007 2008 2009
	char path[PATH_MAX];

	if (!map->groups) {
		pr_debug("Guest kernel map hasn't the point to groups\n");
		return -1;
	}
2010
	machine = map->groups->machine;
2011

2012
	if (machine__is_default_guest(machine)) {
2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027
		/*
		 * 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) {
			err = dso__load_vmlinux(self, map,
				symbol_conf.default_guest_vmlinux_name, filter);
			goto out_try_fixup;
		}

		kallsyms_filename = symbol_conf.default_guest_kallsyms;
		if (!kallsyms_filename)
			return -1;
	} else {
2028
		sprintf(path, "%s/proc/kallsyms", machine->root_dir);
2029 2030 2031 2032 2033 2034 2035 2036 2037 2038
		kallsyms_filename = path;
	}

	err = dso__load_kallsyms(self, kallsyms_filename, map, filter);
	if (err > 0)
		pr_debug("Using %s for symbols\n", kallsyms_filename);

out_try_fixup:
	if (err > 0) {
		if (kallsyms_filename != NULL) {
2039
			machine__mmap_name(machine, path, sizeof(path));
2040
			dso__set_long_name(self, strdup(path));
2041 2042 2043 2044 2045 2046 2047
		}
		map__fixup_start(map);
		map__fixup_end(map);
	}

	return err;
}
2048

2049
static void dsos__add(struct list_head *head, struct dso *dso)
2050
{
2051
	list_add_tail(&dso->node, head);
2052 2053
}

2054
static struct dso *dsos__find(struct list_head *head, const char *name)
2055 2056 2057
{
	struct dso *pos;

2058
	list_for_each_entry(pos, head, node)
2059
		if (strcmp(pos->long_name, name) == 0)
2060 2061 2062 2063
			return pos;
	return NULL;
}

2064
struct dso *__dsos__findnew(struct list_head *head, const char *name)
2065
{
2066
	struct dso *dso = dsos__find(head, name);
2067

2068
	if (!dso) {
2069
		dso = dso__new(name);
2070
		if (dso != NULL) {
2071
			dsos__add(head, dso);
2072 2073
			dso__set_basename(dso);
		}
2074
	}
2075 2076 2077 2078

	return dso;
}

2079
size_t __dsos__fprintf(struct list_head *head, FILE *fp)
2080 2081
{
	struct dso *pos;
2082
	size_t ret = 0;
2083

2084 2085 2086
	list_for_each_entry(pos, head, node) {
		int i;
		for (i = 0; i < MAP__NR_TYPES; ++i)
2087
			ret += dso__fprintf(pos, i, fp);
2088
	}
2089 2090

	return ret;
2091 2092
}

2093
size_t machines__fprintf_dsos(struct rb_root *self, FILE *fp)
2094
{
2095
	struct rb_node *nd;
2096
	size_t ret = 0;
2097

2098
	for (nd = rb_first(self); nd; nd = rb_next(nd)) {
2099
		struct machine *pos = rb_entry(nd, struct machine, rb_node);
2100 2101
		ret += __dsos__fprintf(&pos->kernel_dsos, fp);
		ret += __dsos__fprintf(&pos->user_dsos, fp);
2102
	}
2103 2104

	return ret;
2105 2106
}

2107 2108
static size_t __dsos__fprintf_buildid(struct list_head *head, FILE *fp,
				      bool with_hits)
2109 2110 2111 2112
{
	struct dso *pos;
	size_t ret = 0;

2113
	list_for_each_entry(pos, head, node) {
2114 2115
		if (with_hits && !pos->hit)
			continue;
2116
		ret += dso__fprintf_buildid(pos, fp);
2117
		ret += fprintf(fp, " %s\n", pos->long_name);
2118 2119 2120 2121
	}
	return ret;
}

2122 2123 2124 2125 2126 2127
size_t machine__fprintf_dsos_buildid(struct machine *self, FILE *fp, bool with_hits)
{
	return __dsos__fprintf_buildid(&self->kernel_dsos, fp, with_hits) +
	       __dsos__fprintf_buildid(&self->user_dsos, fp, with_hits);
}

2128
size_t machines__fprintf_dsos_buildid(struct rb_root *self, FILE *fp, bool with_hits)
2129
{
2130 2131 2132
	struct rb_node *nd;
	size_t ret = 0;

2133
	for (nd = rb_first(self); nd; nd = rb_next(nd)) {
2134
		struct machine *pos = rb_entry(nd, struct machine, rb_node);
2135
		ret += machine__fprintf_dsos_buildid(pos, fp, with_hits);
2136 2137
	}
	return ret;
2138 2139
}

2140 2141 2142 2143 2144
struct dso *dso__new_kernel(const char *name)
{
	struct dso *self = dso__new(name ?: "[kernel.kallsyms]");

	if (self != NULL) {
2145
		dso__set_short_name(self, "[kernel]");
2146 2147 2148 2149 2150 2151
		self->kernel = DSO_TYPE_KERNEL;
	}

	return self;
}

2152
static struct dso *dso__new_guest_kernel(struct machine *machine,
2153 2154
					const char *name)
{
2155 2156
	char bf[PATH_MAX];
	struct dso *self = dso__new(name ?: machine__mmap_name(machine, bf, sizeof(bf)));
2157 2158 2159 2160

	if (self != NULL) {
		dso__set_short_name(self, "[guest.kernel]");
		self->kernel = DSO_TYPE_GUEST_KERNEL;
2161 2162 2163 2164 2165
	}

	return self;
}

2166
void dso__read_running_kernel_build_id(struct dso *self, struct machine *machine)
2167
{
2168 2169
	char path[PATH_MAX];

2170
	if (machine__is_default_guest(machine))
2171
		return;
2172
	sprintf(path, "%s/sys/kernel/notes", machine->root_dir);
2173
	if (sysfs__read_build_id(path, self->build_id,
2174 2175 2176 2177
				 sizeof(self->build_id)) == 0)
		self->has_build_id = true;
}

2178
static struct dso *machine__create_kernel(struct machine *self)
2179
{
2180 2181
	const char *vmlinux_name = NULL;
	struct dso *kernel;
2182

2183
	if (machine__is_host(self)) {
2184 2185 2186
		vmlinux_name = symbol_conf.vmlinux_name;
		kernel = dso__new_kernel(vmlinux_name);
	} else {
2187
		if (machine__is_default_guest(self))
2188
			vmlinux_name = symbol_conf.default_guest_vmlinux_name;
2189
		kernel = dso__new_guest_kernel(self, vmlinux_name);
2190
	}
2191

2192
	if (kernel != NULL) {
2193 2194
		dso__read_running_kernel_build_id(kernel, self);
		dsos__add(&self->kernel_dsos, kernel);
2195
	}
2196 2197 2198
	return kernel;
}

2199 2200 2201 2202 2203
struct process_args {
	u64 start;
};

static int symbol__in_kernel(void *arg, const char *name,
2204
			     char type __used, u64 start, u64 end __used)
2205 2206 2207 2208 2209 2210 2211 2212 2213 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
{
	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;
}

2239
int __machine__create_kernel_maps(struct machine *self, struct dso *kernel)
2240
{
2241
	enum map_type type;
2242
	u64 start = machine__get_kernel_start_addr(self);
2243

2244
	for (type = 0; type < MAP__NR_TYPES; ++type) {
2245 2246
		struct kmap *kmap;

2247
		self->vmlinux_maps[type] = map__new2(start, kernel, type);
2248
		if (self->vmlinux_maps[type] == NULL)
2249
			return -1;
2250

2251 2252
		self->vmlinux_maps[type]->map_ip =
			self->vmlinux_maps[type]->unmap_ip = identity__map_ip;
2253

2254 2255 2256
		kmap = map__kmap(self->vmlinux_maps[type]);
		kmap->kmaps = &self->kmaps;
		map_groups__insert(&self->kmaps, self->vmlinux_maps[type]);
2257 2258 2259
	}

	return 0;
2260 2261
}

2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291
void machine__destroy_kernel_maps(struct machine *self)
{
	enum map_type type;

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

		if (self->vmlinux_maps[type] == NULL)
			continue;

		kmap = map__kmap(self->vmlinux_maps[type]);
		map_groups__remove(&self->kmaps, self->vmlinux_maps[type]);
		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;
		}

		map__delete(self->vmlinux_maps[type]);
		self->vmlinux_maps[type] = NULL;
	}
}

2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308
int machine__create_kernel_maps(struct machine *self)
{
	struct dso *kernel = machine__create_kernel(self);

	if (kernel == NULL ||
	    __machine__create_kernel_maps(self, kernel) < 0)
		return -1;

	if (symbol_conf.use_modules && machine__create_modules(self) < 0)
		pr_debug("Problems creating module maps, continuing anyway...\n");
	/*
	 * Now that we have all the maps created, just set the ->end of them:
	 */
	map_groups__fixup_end(&self->kmaps);
	return 0;
}

2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336
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;
2337 2338 2339 2340 2341 2342 2343 2344

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

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

2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368
	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;
}

2369
size_t machine__fprintf_vmlinux_path(struct machine *self, FILE *fp)
2370 2371 2372
{
	int i;
	size_t printed = 0;
2373 2374 2375 2376 2377 2378 2379
	struct dso *kdso = self->vmlinux_maps[MAP__FUNCTION]->dso;

	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);
	}
2380 2381

	for (i = 0; i < vmlinux_path__nr_entries; ++i)
2382 2383
		printed += fprintf(fp, "[%d] %s\n",
				   i + kdso->has_build_id, vmlinux_path[i]);
2384 2385 2386 2387

	return printed;
}

2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401
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;
}

2402
int symbol__init(void)
2403
{
2404 2405
	const char *symfs;

2406 2407 2408
	if (symbol_conf.initialized)
		return 0;

2409
	elf_version(EV_CURRENT);
2410 2411 2412
	if (symbol_conf.sort_by_name)
		symbol_conf.priv_size += (sizeof(struct symbol_name_rb_node) -
					  sizeof(struct symbol));
2413

2414
	if (symbol_conf.try_vmlinux_path && vmlinux_path__init() < 0)
2415 2416
		return -1;

2417 2418 2419 2420 2421
	if (symbol_conf.field_sep && *symbol_conf.field_sep == '.') {
		pr_err("'.' is the only non valid --field-separator argument\n");
		return -1;
	}

2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433
	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;

2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445
	/*
	 * 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);

2446
	symbol_conf.initialized = true;
2447
	return 0;
2448 2449 2450 2451 2452 2453

out_free_dso_list:
	strlist__delete(symbol_conf.dso_list);
out_free_comm_list:
	strlist__delete(symbol_conf.comm_list);
	return -1;
2454 2455
}

2456 2457
void symbol__exit(void)
{
2458 2459
	if (!symbol_conf.initialized)
		return;
2460 2461 2462 2463 2464
	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;
2465
	symbol_conf.initialized = false;
2466 2467
}

2468
int machines__create_kernel_maps(struct rb_root *self, pid_t pid)
2469
{
2470
	struct machine *machine = machines__findnew(self, pid);
2471

2472
	if (machine == NULL)
2473
		return -1;
2474

2475
	return machine__create_kernel_maps(machine);
2476
}
2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 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

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;
}
2520

2521
int machines__create_guest_kernel_maps(struct rb_root *self)
2522 2523 2524 2525 2526 2527 2528 2529 2530 2531
{
	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) {
2532
		machines__create_kernel_maps(self, DEFAULT_GUEST_KERNEL_ID);
2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552
	}

	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;
			}
2553
			machines__create_kernel_maps(self, pid);
2554 2555 2556 2557 2558 2559 2560
		}
failure:
		free(namelist);
	}

	return ret;
}
2561

2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574
void machines__destroy_guest_kernel_maps(struct rb_root *self)
{
	struct rb_node *next = rb_first(self);

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

		next = rb_next(&pos->rb_node);
		rb_erase(&pos->rb_node, self);
		machine__delete(pos);
	}
}

2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606
int machine__load_kallsyms(struct machine *self, const char *filename,
			   enum map_type type, symbol_filter_t filter)
{
	struct map *map = self->vmlinux_maps[type];
	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.
		 */
		__map_groups__fixup_end(&self->kmaps, type);
	}

	return ret;
}

int machine__load_vmlinux_path(struct machine *self, enum map_type type,
			       symbol_filter_t filter)
{
	struct map *map = self->vmlinux_maps[type];
	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;
}