xref: /linux/tools/objtool/elf.c (revision 6fdcba32711044c35c0e1b094cbd8f3f0b4472c9) 
1 // SPDX-License-Identifier: GPL-2.0-or-later
2 /*
3  * elf.c - ELF access library
4  *
5  * Adapted from kpatch (https://github.com/dynup/kpatch):
6  * Copyright (C) 2013-2015 Josh Poimboeuf <jpoimboe@redhat.com>
7  * Copyright (C) 2014 Seth Jennings <sjenning@redhat.com>
8  */
9 
10 #include <sys/types.h>
11 #include <sys/stat.h>
12 #include <fcntl.h>
13 #include <stdio.h>
14 #include <stdlib.h>
15 #include <string.h>
16 #include <unistd.h>
17 #include <errno.h>
18 
19 #include "elf.h"
20 #include "warn.h"
21 
22 #define MAX_NAME_LEN 128
23 
24 struct section *find_section_by_name(struct elf *elf, const char *name)
25 {
26 	struct section *sec;
27 
28 	list_for_each_entry(sec, &elf->sections, list)
29 		if (!strcmp(sec->name, name))
30 			return sec;
31 
32 	return NULL;
33 }
34 
35 static struct section *find_section_by_index(struct elf *elf,
36 					     unsigned int idx)
37 {
38 	struct section *sec;
39 
40 	list_for_each_entry(sec, &elf->sections, list)
41 		if (sec->idx == idx)
42 			return sec;
43 
44 	return NULL;
45 }
46 
47 static struct symbol *find_symbol_by_index(struct elf *elf, unsigned int idx)
48 {
49 	struct section *sec;
50 	struct symbol *sym;
51 
52 	list_for_each_entry(sec, &elf->sections, list)
53 		hash_for_each_possible(sec->symbol_hash, sym, hash, idx)
54 			if (sym->idx == idx)
55 				return sym;
56 
57 	return NULL;
58 }
59 
60 struct symbol *find_symbol_by_offset(struct section *sec, unsigned long offset)
61 {
62 	struct symbol *sym;
63 
64 	list_for_each_entry(sym, &sec->symbol_list, list)
65 		if (sym->type != STT_SECTION &&
66 		    sym->offset == offset)
67 			return sym;
68 
69 	return NULL;
70 }
71 
72 struct symbol *find_symbol_by_name(struct elf *elf, const char *name)
73 {
74 	struct section *sec;
75 	struct symbol *sym;
76 
77 	list_for_each_entry(sec, &elf->sections, list)
78 		list_for_each_entry(sym, &sec->symbol_list, list)
79 			if (!strcmp(sym->name, name))
80 				return sym;
81 
82 	return NULL;
83 }
84 
85 struct symbol *find_symbol_containing(struct section *sec, unsigned long offset)
86 {
87 	struct symbol *sym;
88 
89 	list_for_each_entry(sym, &sec->symbol_list, list)
90 		if (sym->type != STT_SECTION &&
91 		    offset >= sym->offset && offset < sym->offset + sym->len)
92 			return sym;
93 
94 	return NULL;
95 }
96 
97 struct rela *find_rela_by_dest_range(struct section *sec, unsigned long offset,
98 				     unsigned int len)
99 {
100 	struct rela *rela;
101 	unsigned long o;
102 
103 	if (!sec->rela)
104 		return NULL;
105 
106 	for (o = offset; o < offset + len; o++)
107 		hash_for_each_possible(sec->rela->rela_hash, rela, hash, o)
108 			if (rela->offset == o)
109 				return rela;
110 
111 	return NULL;
112 }
113 
114 struct rela *find_rela_by_dest(struct section *sec, unsigned long offset)
115 {
116 	return find_rela_by_dest_range(sec, offset, 1);
117 }
118 
119 struct symbol *find_containing_func(struct section *sec, unsigned long offset)
120 {
121 	struct symbol *func;
122 
123 	list_for_each_entry(func, &sec->symbol_list, list)
124 		if (func->type == STT_FUNC && offset >= func->offset &&
125 		    offset < func->offset + func->len)
126 			return func;
127 
128 	return NULL;
129 }
130 
131 static int read_sections(struct elf *elf)
132 {
133 	Elf_Scn *s = NULL;
134 	struct section *sec;
135 	size_t shstrndx, sections_nr;
136 	int i;
137 
138 	if (elf_getshdrnum(elf->elf, &sections_nr)) {
139 		WARN_ELF("elf_getshdrnum");
140 		return -1;
141 	}
142 
143 	if (elf_getshdrstrndx(elf->elf, &shstrndx)) {
144 		WARN_ELF("elf_getshdrstrndx");
145 		return -1;
146 	}
147 
148 	for (i = 0; i < sections_nr; i++) {
149 		sec = malloc(sizeof(*sec));
150 		if (!sec) {
151 			perror("malloc");
152 			return -1;
153 		}
154 		memset(sec, 0, sizeof(*sec));
155 
156 		INIT_LIST_HEAD(&sec->symbol_list);
157 		INIT_LIST_HEAD(&sec->rela_list);
158 		hash_init(sec->rela_hash);
159 		hash_init(sec->symbol_hash);
160 
161 		list_add_tail(&sec->list, &elf->sections);
162 
163 		s = elf_getscn(elf->elf, i);
164 		if (!s) {
165 			WARN_ELF("elf_getscn");
166 			return -1;
167 		}
168 
169 		sec->idx = elf_ndxscn(s);
170 
171 		if (!gelf_getshdr(s, &sec->sh)) {
172 			WARN_ELF("gelf_getshdr");
173 			return -1;
174 		}
175 
176 		sec->name = elf_strptr(elf->elf, shstrndx, sec->sh.sh_name);
177 		if (!sec->name) {
178 			WARN_ELF("elf_strptr");
179 			return -1;
180 		}
181 
182 		if (sec->sh.sh_size != 0) {
183 			sec->data = elf_getdata(s, NULL);
184 			if (!sec->data) {
185 				WARN_ELF("elf_getdata");
186 				return -1;
187 			}
188 			if (sec->data->d_off != 0 ||
189 			    sec->data->d_size != sec->sh.sh_size) {
190 				WARN("unexpected data attributes for %s",
191 				     sec->name);
192 				return -1;
193 			}
194 		}
195 		sec->len = sec->sh.sh_size;
196 	}
197 
198 	/* sanity check, one more call to elf_nextscn() should return NULL */
199 	if (elf_nextscn(elf->elf, s)) {
200 		WARN("section entry mismatch");
201 		return -1;
202 	}
203 
204 	return 0;
205 }
206 
207 static int read_symbols(struct elf *elf)
208 {
209 	struct section *symtab, *sec;
210 	struct symbol *sym, *pfunc, *alias;
211 	struct list_head *entry, *tmp;
212 	int symbols_nr, i;
213 	char *coldstr;
214 
215 	symtab = find_section_by_name(elf, ".symtab");
216 	if (!symtab) {
217 		WARN("missing symbol table");
218 		return -1;
219 	}
220 
221 	symbols_nr = symtab->sh.sh_size / symtab->sh.sh_entsize;
222 
223 	for (i = 0; i < symbols_nr; i++) {
224 		sym = malloc(sizeof(*sym));
225 		if (!sym) {
226 			perror("malloc");
227 			return -1;
228 		}
229 		memset(sym, 0, sizeof(*sym));
230 		alias = sym;
231 
232 		sym->idx = i;
233 
234 		if (!gelf_getsym(symtab->data, i, &sym->sym)) {
235 			WARN_ELF("gelf_getsym");
236 			goto err;
237 		}
238 
239 		sym->name = elf_strptr(elf->elf, symtab->sh.sh_link,
240 				       sym->sym.st_name);
241 		if (!sym->name) {
242 			WARN_ELF("elf_strptr");
243 			goto err;
244 		}
245 
246 		sym->type = GELF_ST_TYPE(sym->sym.st_info);
247 		sym->bind = GELF_ST_BIND(sym->sym.st_info);
248 
249 		if (sym->sym.st_shndx > SHN_UNDEF &&
250 		    sym->sym.st_shndx < SHN_LORESERVE) {
251 			sym->sec = find_section_by_index(elf,
252 							 sym->sym.st_shndx);
253 			if (!sym->sec) {
254 				WARN("couldn't find section for symbol %s",
255 				     sym->name);
256 				goto err;
257 			}
258 			if (sym->type == STT_SECTION) {
259 				sym->name = sym->sec->name;
260 				sym->sec->sym = sym;
261 			}
262 		} else
263 			sym->sec = find_section_by_index(elf, 0);
264 
265 		sym->offset = sym->sym.st_value;
266 		sym->len = sym->sym.st_size;
267 
268 		/* sorted insert into a per-section list */
269 		entry = &sym->sec->symbol_list;
270 		list_for_each_prev(tmp, &sym->sec->symbol_list) {
271 			struct symbol *s;
272 
273 			s = list_entry(tmp, struct symbol, list);
274 
275 			if (sym->offset > s->offset) {
276 				entry = tmp;
277 				break;
278 			}
279 
280 			if (sym->offset == s->offset) {
281 				if (sym->len && sym->len == s->len && alias == sym)
282 					alias = s;
283 
284 				if (sym->len >= s->len) {
285 					entry = tmp;
286 					break;
287 				}
288 			}
289 		}
290 		sym->alias = alias;
291 		list_add(&sym->list, entry);
292 		hash_add(sym->sec->symbol_hash, &sym->hash, sym->idx);
293 	}
294 
295 	/* Create parent/child links for any cold subfunctions */
296 	list_for_each_entry(sec, &elf->sections, list) {
297 		list_for_each_entry(sym, &sec->symbol_list, list) {
298 			char pname[MAX_NAME_LEN + 1];
299 			size_t pnamelen;
300 			if (sym->type != STT_FUNC)
301 				continue;
302 			sym->pfunc = sym->cfunc = sym;
303 			coldstr = strstr(sym->name, ".cold");
304 			if (!coldstr)
305 				continue;
306 
307 			pnamelen = coldstr - sym->name;
308 			if (pnamelen > MAX_NAME_LEN) {
309 				WARN("%s(): parent function name exceeds maximum length of %d characters",
310 				     sym->name, MAX_NAME_LEN);
311 				return -1;
312 			}
313 
314 			strncpy(pname, sym->name, pnamelen);
315 			pname[pnamelen] = '\0';
316 			pfunc = find_symbol_by_name(elf, pname);
317 
318 			if (!pfunc) {
319 				WARN("%s(): can't find parent function",
320 				     sym->name);
321 				return -1;
322 			}
323 
324 			sym->pfunc = pfunc;
325 			pfunc->cfunc = sym;
326 
327 			/*
328 			 * Unfortunately, -fnoreorder-functions puts the child
329 			 * inside the parent.  Remove the overlap so we can
330 			 * have sane assumptions.
331 			 *
332 			 * Note that pfunc->len now no longer matches
333 			 * pfunc->sym.st_size.
334 			 */
335 			if (sym->sec == pfunc->sec &&
336 			    sym->offset >= pfunc->offset &&
337 			    sym->offset + sym->len == pfunc->offset + pfunc->len) {
338 				pfunc->len -= sym->len;
339 			}
340 		}
341 	}
342 
343 	return 0;
344 
345 err:
346 	free(sym);
347 	return -1;
348 }
349 
350 static int read_relas(struct elf *elf)
351 {
352 	struct section *sec;
353 	struct rela *rela;
354 	int i;
355 	unsigned int symndx;
356 
357 	list_for_each_entry(sec, &elf->sections, list) {
358 		if (sec->sh.sh_type != SHT_RELA)
359 			continue;
360 
361 		sec->base = find_section_by_name(elf, sec->name + 5);
362 		if (!sec->base) {
363 			WARN("can't find base section for rela section %s",
364 			     sec->name);
365 			return -1;
366 		}
367 
368 		sec->base->rela = sec;
369 
370 		for (i = 0; i < sec->sh.sh_size / sec->sh.sh_entsize; i++) {
371 			rela = malloc(sizeof(*rela));
372 			if (!rela) {
373 				perror("malloc");
374 				return -1;
375 			}
376 			memset(rela, 0, sizeof(*rela));
377 
378 			if (!gelf_getrela(sec->data, i, &rela->rela)) {
379 				WARN_ELF("gelf_getrela");
380 				return -1;
381 			}
382 
383 			rela->type = GELF_R_TYPE(rela->rela.r_info);
384 			rela->addend = rela->rela.r_addend;
385 			rela->offset = rela->rela.r_offset;
386 			symndx = GELF_R_SYM(rela->rela.r_info);
387 			rela->sym = find_symbol_by_index(elf, symndx);
388 			rela->sec = sec;
389 			if (!rela->sym) {
390 				WARN("can't find rela entry symbol %d for %s",
391 				     symndx, sec->name);
392 				return -1;
393 			}
394 
395 			list_add_tail(&rela->list, &sec->rela_list);
396 			hash_add(sec->rela_hash, &rela->hash, rela->offset);
397 
398 		}
399 	}
400 
401 	return 0;
402 }
403 
404 struct elf *elf_read(const char *name, int flags)
405 {
406 	struct elf *elf;
407 	Elf_Cmd cmd;
408 
409 	elf_version(EV_CURRENT);
410 
411 	elf = malloc(sizeof(*elf));
412 	if (!elf) {
413 		perror("malloc");
414 		return NULL;
415 	}
416 	memset(elf, 0, sizeof(*elf));
417 
418 	INIT_LIST_HEAD(&elf->sections);
419 
420 	elf->fd = open(name, flags);
421 	if (elf->fd == -1) {
422 		fprintf(stderr, "objtool: Can't open '%s': %s\n",
423 			name, strerror(errno));
424 		goto err;
425 	}
426 
427 	if ((flags & O_ACCMODE) == O_RDONLY)
428 		cmd = ELF_C_READ_MMAP;
429 	else if ((flags & O_ACCMODE) == O_RDWR)
430 		cmd = ELF_C_RDWR;
431 	else /* O_WRONLY */
432 		cmd = ELF_C_WRITE;
433 
434 	elf->elf = elf_begin(elf->fd, cmd, NULL);
435 	if (!elf->elf) {
436 		WARN_ELF("elf_begin");
437 		goto err;
438 	}
439 
440 	if (!gelf_getehdr(elf->elf, &elf->ehdr)) {
441 		WARN_ELF("gelf_getehdr");
442 		goto err;
443 	}
444 
445 	if (read_sections(elf))
446 		goto err;
447 
448 	if (read_symbols(elf))
449 		goto err;
450 
451 	if (read_relas(elf))
452 		goto err;
453 
454 	return elf;
455 
456 err:
457 	elf_close(elf);
458 	return NULL;
459 }
460 
461 struct section *elf_create_section(struct elf *elf, const char *name,
462 				   size_t entsize, int nr)
463 {
464 	struct section *sec, *shstrtab;
465 	size_t size = entsize * nr;
466 	Elf_Scn *s;
467 	Elf_Data *data;
468 
469 	sec = malloc(sizeof(*sec));
470 	if (!sec) {
471 		perror("malloc");
472 		return NULL;
473 	}
474 	memset(sec, 0, sizeof(*sec));
475 
476 	INIT_LIST_HEAD(&sec->symbol_list);
477 	INIT_LIST_HEAD(&sec->rela_list);
478 	hash_init(sec->rela_hash);
479 	hash_init(sec->symbol_hash);
480 
481 	list_add_tail(&sec->list, &elf->sections);
482 
483 	s = elf_newscn(elf->elf);
484 	if (!s) {
485 		WARN_ELF("elf_newscn");
486 		return NULL;
487 	}
488 
489 	sec->name = strdup(name);
490 	if (!sec->name) {
491 		perror("strdup");
492 		return NULL;
493 	}
494 
495 	sec->idx = elf_ndxscn(s);
496 	sec->len = size;
497 	sec->changed = true;
498 
499 	sec->data = elf_newdata(s);
500 	if (!sec->data) {
501 		WARN_ELF("elf_newdata");
502 		return NULL;
503 	}
504 
505 	sec->data->d_size = size;
506 	sec->data->d_align = 1;
507 
508 	if (size) {
509 		sec->data->d_buf = malloc(size);
510 		if (!sec->data->d_buf) {
511 			perror("malloc");
512 			return NULL;
513 		}
514 		memset(sec->data->d_buf, 0, size);
515 	}
516 
517 	if (!gelf_getshdr(s, &sec->sh)) {
518 		WARN_ELF("gelf_getshdr");
519 		return NULL;
520 	}
521 
522 	sec->sh.sh_size = size;
523 	sec->sh.sh_entsize = entsize;
524 	sec->sh.sh_type = SHT_PROGBITS;
525 	sec->sh.sh_addralign = 1;
526 	sec->sh.sh_flags = SHF_ALLOC;
527 
528 
529 	/* Add section name to .shstrtab (or .strtab for Clang) */
530 	shstrtab = find_section_by_name(elf, ".shstrtab");
531 	if (!shstrtab)
532 		shstrtab = find_section_by_name(elf, ".strtab");
533 	if (!shstrtab) {
534 		WARN("can't find .shstrtab or .strtab section");
535 		return NULL;
536 	}
537 
538 	s = elf_getscn(elf->elf, shstrtab->idx);
539 	if (!s) {
540 		WARN_ELF("elf_getscn");
541 		return NULL;
542 	}
543 
544 	data = elf_newdata(s);
545 	if (!data) {
546 		WARN_ELF("elf_newdata");
547 		return NULL;
548 	}
549 
550 	data->d_buf = sec->name;
551 	data->d_size = strlen(name) + 1;
552 	data->d_align = 1;
553 
554 	sec->sh.sh_name = shstrtab->len;
555 
556 	shstrtab->len += strlen(name) + 1;
557 	shstrtab->changed = true;
558 
559 	return sec;
560 }
561 
562 struct section *elf_create_rela_section(struct elf *elf, struct section *base)
563 {
564 	char *relaname;
565 	struct section *sec;
566 
567 	relaname = malloc(strlen(base->name) + strlen(".rela") + 1);
568 	if (!relaname) {
569 		perror("malloc");
570 		return NULL;
571 	}
572 	strcpy(relaname, ".rela");
573 	strcat(relaname, base->name);
574 
575 	sec = elf_create_section(elf, relaname, sizeof(GElf_Rela), 0);
576 	free(relaname);
577 	if (!sec)
578 		return NULL;
579 
580 	base->rela = sec;
581 	sec->base = base;
582 
583 	sec->sh.sh_type = SHT_RELA;
584 	sec->sh.sh_addralign = 8;
585 	sec->sh.sh_link = find_section_by_name(elf, ".symtab")->idx;
586 	sec->sh.sh_info = base->idx;
587 	sec->sh.sh_flags = SHF_INFO_LINK;
588 
589 	return sec;
590 }
591 
592 int elf_rebuild_rela_section(struct section *sec)
593 {
594 	struct rela *rela;
595 	int nr, idx = 0, size;
596 	GElf_Rela *relas;
597 
598 	nr = 0;
599 	list_for_each_entry(rela, &sec->rela_list, list)
600 		nr++;
601 
602 	size = nr * sizeof(*relas);
603 	relas = malloc(size);
604 	if (!relas) {
605 		perror("malloc");
606 		return -1;
607 	}
608 
609 	sec->data->d_buf = relas;
610 	sec->data->d_size = size;
611 
612 	sec->sh.sh_size = size;
613 
614 	idx = 0;
615 	list_for_each_entry(rela, &sec->rela_list, list) {
616 		relas[idx].r_offset = rela->offset;
617 		relas[idx].r_addend = rela->addend;
618 		relas[idx].r_info = GELF_R_INFO(rela->sym->idx, rela->type);
619 		idx++;
620 	}
621 
622 	return 0;
623 }
624 
625 int elf_write(struct elf *elf)
626 {
627 	struct section *sec;
628 	Elf_Scn *s;
629 
630 	/* Update section headers for changed sections: */
631 	list_for_each_entry(sec, &elf->sections, list) {
632 		if (sec->changed) {
633 			s = elf_getscn(elf->elf, sec->idx);
634 			if (!s) {
635 				WARN_ELF("elf_getscn");
636 				return -1;
637 			}
638 			if (!gelf_update_shdr(s, &sec->sh)) {
639 				WARN_ELF("gelf_update_shdr");
640 				return -1;
641 			}
642 		}
643 	}
644 
645 	/* Make sure the new section header entries get updated properly. */
646 	elf_flagelf(elf->elf, ELF_C_SET, ELF_F_DIRTY);
647 
648 	/* Write all changes to the file. */
649 	if (elf_update(elf->elf, ELF_C_WRITE) < 0) {
650 		WARN_ELF("elf_update");
651 		return -1;
652 	}
653 
654 	return 0;
655 }
656 
657 void elf_close(struct elf *elf)
658 {
659 	struct section *sec, *tmpsec;
660 	struct symbol *sym, *tmpsym;
661 	struct rela *rela, *tmprela;
662 
663 	if (elf->elf)
664 		elf_end(elf->elf);
665 
666 	if (elf->fd > 0)
667 		close(elf->fd);
668 
669 	list_for_each_entry_safe(sec, tmpsec, &elf->sections, list) {
670 		list_for_each_entry_safe(sym, tmpsym, &sec->symbol_list, list) {
671 			list_del(&sym->list);
672 			hash_del(&sym->hash);
673 			free(sym);
674 		}
675 		list_for_each_entry_safe(rela, tmprela, &sec->rela_list, list) {
676 			list_del(&rela->list);
677 			hash_del(&rela->hash);
678 			free(rela);
679 		}
680 		list_del(&sec->list);
681 		free(sec);
682 	}
683 
684 	free(elf);
685 }
686