1 // SPDX-License-Identifier: (LGPL-2.1 OR BSD-2-Clause)
2
3 /*
4 * resolve_btfids scans Elf object for .BTF_ids section and resolves
5 * its symbols with BTF ID values.
6 *
7 * Each symbol points to 4 bytes data and is expected to have
8 * following name syntax:
9 *
10 * __BTF_ID__<type>__<symbol>[__<id>]
11 *
12 * type is:
13 *
14 * func - lookup BTF_KIND_FUNC symbol with <symbol> name
15 * and store its ID into the data:
16 *
17 * __BTF_ID__func__vfs_close__1:
18 * .zero 4
19 *
20 * struct - lookup BTF_KIND_STRUCT symbol with <symbol> name
21 * and store its ID into the data:
22 *
23 * __BTF_ID__struct__sk_buff__1:
24 * .zero 4
25 *
26 * union - lookup BTF_KIND_UNION symbol with <symbol> name
27 * and store its ID into the data:
28 *
29 * __BTF_ID__union__thread_union__1:
30 * .zero 4
31 *
32 * typedef - lookup BTF_KIND_TYPEDEF symbol with <symbol> name
33 * and store its ID into the data:
34 *
35 * __BTF_ID__typedef__pid_t__1:
36 * .zero 4
37 *
38 * set - store symbol size into first 4 bytes and sort following
39 * ID list
40 *
41 * __BTF_ID__set__list:
42 * .zero 4
43 * list:
44 * __BTF_ID__func__vfs_getattr__3:
45 * .zero 4
46 * __BTF_ID__func__vfs_fallocate__4:
47 * .zero 4
48 */
49
50 #define _GNU_SOURCE
51 #include <stdio.h>
52 #include <string.h>
53 #include <unistd.h>
54 #include <stdlib.h>
55 #include <libelf.h>
56 #include <gelf.h>
57 #include <sys/stat.h>
58 #include <fcntl.h>
59 #include <errno.h>
60 #include <linux/rbtree.h>
61 #include <linux/zalloc.h>
62 #include <linux/err.h>
63 #include <btf.h>
64 #include <libbpf.h>
65 #include <parse-options.h>
66
67 #define BTF_IDS_SECTION ".BTF_ids"
68 #define BTF_ID "__BTF_ID__"
69
70 #define BTF_STRUCT "struct"
71 #define BTF_UNION "union"
72 #define BTF_TYPEDEF "typedef"
73 #define BTF_FUNC "func"
74 #define BTF_SET "set"
75
76 #define ADDR_CNT 100
77
78 struct btf_id {
79 struct rb_node rb_node;
80 char *name;
81 union {
82 int id;
83 int cnt;
84 };
85 int addr_cnt;
86 Elf64_Addr addr[ADDR_CNT];
87 };
88
89 struct object {
90 const char *path;
91 const char *btf;
92
93 struct {
94 int fd;
95 Elf *elf;
96 Elf_Data *symbols;
97 Elf_Data *idlist;
98 int symbols_shndx;
99 int idlist_shndx;
100 size_t strtabidx;
101 unsigned long idlist_addr;
102 } efile;
103
104 struct rb_root sets;
105 struct rb_root structs;
106 struct rb_root unions;
107 struct rb_root typedefs;
108 struct rb_root funcs;
109
110 int nr_funcs;
111 int nr_structs;
112 int nr_unions;
113 int nr_typedefs;
114 };
115
116 static int verbose;
117
eprintf(int level,int var,const char * fmt,...)118 static int eprintf(int level, int var, const char *fmt, ...)
119 {
120 va_list args;
121 int ret = 0;
122
123 if (var >= level) {
124 va_start(args, fmt);
125 ret = vfprintf(stderr, fmt, args);
126 va_end(args);
127 }
128 return ret;
129 }
130
131 #ifndef pr_fmt
132 #define pr_fmt(fmt) fmt
133 #endif
134
135 #define pr_debug(fmt, ...) \
136 eprintf(1, verbose, pr_fmt(fmt), ##__VA_ARGS__)
137 #define pr_debugN(n, fmt, ...) \
138 eprintf(n, verbose, pr_fmt(fmt), ##__VA_ARGS__)
139 #define pr_debug2(fmt, ...) pr_debugN(2, pr_fmt(fmt), ##__VA_ARGS__)
140 #define pr_err(fmt, ...) \
141 eprintf(0, verbose, pr_fmt(fmt), ##__VA_ARGS__)
142 #define pr_info(fmt, ...) \
143 eprintf(0, verbose, pr_fmt(fmt), ##__VA_ARGS__)
144
is_btf_id(const char * name)145 static bool is_btf_id(const char *name)
146 {
147 return name && !strncmp(name, BTF_ID, sizeof(BTF_ID) - 1);
148 }
149
btf_id__find(struct rb_root * root,const char * name)150 static struct btf_id *btf_id__find(struct rb_root *root, const char *name)
151 {
152 struct rb_node *p = root->rb_node;
153 struct btf_id *id;
154 int cmp;
155
156 while (p) {
157 id = rb_entry(p, struct btf_id, rb_node);
158 cmp = strcmp(id->name, name);
159 if (cmp < 0)
160 p = p->rb_left;
161 else if (cmp > 0)
162 p = p->rb_right;
163 else
164 return id;
165 }
166 return NULL;
167 }
168
169 static struct btf_id*
btf_id__add(struct rb_root * root,char * name,bool unique)170 btf_id__add(struct rb_root *root, char *name, bool unique)
171 {
172 struct rb_node **p = &root->rb_node;
173 struct rb_node *parent = NULL;
174 struct btf_id *id;
175 int cmp;
176
177 while (*p != NULL) {
178 parent = *p;
179 id = rb_entry(parent, struct btf_id, rb_node);
180 cmp = strcmp(id->name, name);
181 if (cmp < 0)
182 p = &(*p)->rb_left;
183 else if (cmp > 0)
184 p = &(*p)->rb_right;
185 else
186 return unique ? NULL : id;
187 }
188
189 id = zalloc(sizeof(*id));
190 if (id) {
191 pr_debug("adding symbol %s\n", name);
192 id->name = name;
193 rb_link_node(&id->rb_node, parent, p);
194 rb_insert_color(&id->rb_node, root);
195 }
196 return id;
197 }
198
get_id(const char * prefix_end)199 static char *get_id(const char *prefix_end)
200 {
201 /*
202 * __BTF_ID__func__vfs_truncate__0
203 * prefix_end = ^
204 * pos = ^
205 */
206 int len = strlen(prefix_end);
207 int pos = sizeof("__") - 1;
208 char *p, *id;
209
210 if (pos >= len)
211 return NULL;
212
213 id = strdup(prefix_end + pos);
214 if (id) {
215 /*
216 * __BTF_ID__func__vfs_truncate__0
217 * id = ^
218 *
219 * cut the unique id part
220 */
221 p = strrchr(id, '_');
222 p--;
223 if (*p != '_') {
224 free(id);
225 return NULL;
226 }
227 *p = '\0';
228 }
229 return id;
230 }
231
add_set(struct object * obj,char * name)232 static struct btf_id *add_set(struct object *obj, char *name)
233 {
234 /*
235 * __BTF_ID__set__name
236 * name = ^
237 * id = ^
238 */
239 char *id = name + sizeof(BTF_SET "__") - 1;
240 int len = strlen(name);
241
242 if (id >= name + len) {
243 pr_err("FAILED to parse set name: %s\n", name);
244 return NULL;
245 }
246
247 return btf_id__add(&obj->sets, id, true);
248 }
249
add_symbol(struct rb_root * root,char * name,size_t size)250 static struct btf_id *add_symbol(struct rb_root *root, char *name, size_t size)
251 {
252 char *id;
253
254 id = get_id(name + size);
255 if (!id) {
256 pr_err("FAILED to parse symbol name: %s\n", name);
257 return NULL;
258 }
259
260 return btf_id__add(root, id, false);
261 }
262
263 /* Older libelf.h and glibc elf.h might not yet define the ELF compression types. */
264 #ifndef SHF_COMPRESSED
265 #define SHF_COMPRESSED (1 << 11) /* Section with compressed data. */
266 #endif
267
268 /*
269 * The data of compressed section should be aligned to 4
270 * (for 32bit) or 8 (for 64 bit) bytes. The binutils ld
271 * sets sh_addralign to 1, which makes libelf fail with
272 * misaligned section error during the update:
273 * FAILED elf_update(WRITE): invalid section alignment
274 *
275 * While waiting for ld fix, we fix the compressed sections
276 * sh_addralign value manualy.
277 */
compressed_section_fix(Elf * elf,Elf_Scn * scn,GElf_Shdr * sh)278 static int compressed_section_fix(Elf *elf, Elf_Scn *scn, GElf_Shdr *sh)
279 {
280 int expected = gelf_getclass(elf) == ELFCLASS32 ? 4 : 8;
281
282 if (!(sh->sh_flags & SHF_COMPRESSED))
283 return 0;
284
285 if (sh->sh_addralign == expected)
286 return 0;
287
288 pr_debug2(" - fixing wrong alignment sh_addralign %u, expected %u\n",
289 sh->sh_addralign, expected);
290
291 sh->sh_addralign = expected;
292
293 if (gelf_update_shdr(scn, sh) == 0) {
294 pr_err("FAILED cannot update section header: %s\n",
295 elf_errmsg(-1));
296 return -1;
297 }
298 return 0;
299 }
300
elf_collect(struct object * obj)301 static int elf_collect(struct object *obj)
302 {
303 Elf_Scn *scn = NULL;
304 size_t shdrstrndx;
305 int idx = 0;
306 Elf *elf;
307 int fd;
308
309 fd = open(obj->path, O_RDWR, 0666);
310 if (fd == -1) {
311 pr_err("FAILED cannot open %s: %s\n",
312 obj->path, strerror(errno));
313 return -1;
314 }
315
316 elf_version(EV_CURRENT);
317
318 elf = elf_begin(fd, ELF_C_RDWR_MMAP, NULL);
319 if (!elf) {
320 close(fd);
321 pr_err("FAILED cannot create ELF descriptor: %s\n",
322 elf_errmsg(-1));
323 return -1;
324 }
325
326 obj->efile.fd = fd;
327 obj->efile.elf = elf;
328
329 elf_flagelf(elf, ELF_C_SET, ELF_F_LAYOUT);
330
331 if (elf_getshdrstrndx(elf, &shdrstrndx) != 0) {
332 pr_err("FAILED cannot get shdr str ndx\n");
333 return -1;
334 }
335
336 /*
337 * Scan all the elf sections and look for save data
338 * from .BTF_ids section and symbols.
339 */
340 while ((scn = elf_nextscn(elf, scn)) != NULL) {
341 Elf_Data *data;
342 GElf_Shdr sh;
343 char *name;
344
345 idx++;
346 if (gelf_getshdr(scn, &sh) != &sh) {
347 pr_err("FAILED get section(%d) header\n", idx);
348 return -1;
349 }
350
351 name = elf_strptr(elf, shdrstrndx, sh.sh_name);
352 if (!name) {
353 pr_err("FAILED get section(%d) name\n", idx);
354 return -1;
355 }
356
357 data = elf_getdata(scn, 0);
358 if (!data) {
359 pr_err("FAILED to get section(%d) data from %s\n",
360 idx, name);
361 return -1;
362 }
363
364 pr_debug2("section(%d) %s, size %ld, link %d, flags %lx, type=%d\n",
365 idx, name, (unsigned long) data->d_size,
366 (int) sh.sh_link, (unsigned long) sh.sh_flags,
367 (int) sh.sh_type);
368
369 if (sh.sh_type == SHT_SYMTAB) {
370 obj->efile.symbols = data;
371 obj->efile.symbols_shndx = idx;
372 obj->efile.strtabidx = sh.sh_link;
373 } else if (!strcmp(name, BTF_IDS_SECTION)) {
374 obj->efile.idlist = data;
375 obj->efile.idlist_shndx = idx;
376 obj->efile.idlist_addr = sh.sh_addr;
377 }
378
379 if (compressed_section_fix(elf, scn, &sh))
380 return -1;
381 }
382
383 return 0;
384 }
385
symbols_collect(struct object * obj)386 static int symbols_collect(struct object *obj)
387 {
388 Elf_Scn *scn = NULL;
389 int n, i;
390 GElf_Shdr sh;
391 char *name;
392
393 scn = elf_getscn(obj->efile.elf, obj->efile.symbols_shndx);
394 if (!scn)
395 return -1;
396
397 if (gelf_getshdr(scn, &sh) != &sh)
398 return -1;
399
400 n = sh.sh_size / sh.sh_entsize;
401
402 /*
403 * Scan symbols and look for the ones starting with
404 * __BTF_ID__* over .BTF_ids section.
405 */
406 for (i = 0; i < n; i++) {
407 char *prefix;
408 struct btf_id *id;
409 GElf_Sym sym;
410
411 if (!gelf_getsym(obj->efile.symbols, i, &sym))
412 return -1;
413
414 if (sym.st_shndx != obj->efile.idlist_shndx)
415 continue;
416
417 name = elf_strptr(obj->efile.elf, obj->efile.strtabidx,
418 sym.st_name);
419
420 if (!is_btf_id(name))
421 continue;
422
423 /*
424 * __BTF_ID__TYPE__vfs_truncate__0
425 * prefix = ^
426 */
427 prefix = name + sizeof(BTF_ID) - 1;
428
429 /* struct */
430 if (!strncmp(prefix, BTF_STRUCT, sizeof(BTF_STRUCT) - 1)) {
431 obj->nr_structs++;
432 id = add_symbol(&obj->structs, prefix, sizeof(BTF_STRUCT) - 1);
433 /* union */
434 } else if (!strncmp(prefix, BTF_UNION, sizeof(BTF_UNION) - 1)) {
435 obj->nr_unions++;
436 id = add_symbol(&obj->unions, prefix, sizeof(BTF_UNION) - 1);
437 /* typedef */
438 } else if (!strncmp(prefix, BTF_TYPEDEF, sizeof(BTF_TYPEDEF) - 1)) {
439 obj->nr_typedefs++;
440 id = add_symbol(&obj->typedefs, prefix, sizeof(BTF_TYPEDEF) - 1);
441 /* func */
442 } else if (!strncmp(prefix, BTF_FUNC, sizeof(BTF_FUNC) - 1)) {
443 obj->nr_funcs++;
444 id = add_symbol(&obj->funcs, prefix, sizeof(BTF_FUNC) - 1);
445 /* set */
446 } else if (!strncmp(prefix, BTF_SET, sizeof(BTF_SET) - 1)) {
447 id = add_set(obj, prefix);
448 /*
449 * SET objects store list's count, which is encoded
450 * in symbol's size, together with 'cnt' field hence
451 * that - 1.
452 */
453 if (id)
454 id->cnt = sym.st_size / sizeof(int) - 1;
455 } else {
456 pr_err("FAILED unsupported prefix %s\n", prefix);
457 return -1;
458 }
459
460 if (!id)
461 return -ENOMEM;
462
463 if (id->addr_cnt >= ADDR_CNT) {
464 pr_err("FAILED symbol %s crossed the number of allowed lists\n",
465 id->name);
466 return -1;
467 }
468 id->addr[id->addr_cnt++] = sym.st_value;
469 }
470
471 return 0;
472 }
473
symbols_resolve(struct object * obj)474 static int symbols_resolve(struct object *obj)
475 {
476 int nr_typedefs = obj->nr_typedefs;
477 int nr_structs = obj->nr_structs;
478 int nr_unions = obj->nr_unions;
479 int nr_funcs = obj->nr_funcs;
480 int err, type_id;
481 struct btf *btf;
482 __u32 nr_types;
483
484 btf = btf__parse(obj->btf ?: obj->path, NULL);
485 err = libbpf_get_error(btf);
486 if (err) {
487 pr_err("FAILED: load BTF from %s: %s\n",
488 obj->btf ?: obj->path, strerror(-err));
489 return -1;
490 }
491
492 err = -1;
493 nr_types = btf__get_nr_types(btf);
494
495 /*
496 * Iterate all the BTF types and search for collected symbol IDs.
497 */
498 for (type_id = 1; type_id <= nr_types; type_id++) {
499 const struct btf_type *type;
500 struct rb_root *root;
501 struct btf_id *id;
502 const char *str;
503 int *nr;
504
505 type = btf__type_by_id(btf, type_id);
506 if (!type) {
507 pr_err("FAILED: malformed BTF, can't resolve type for ID %d\n",
508 type_id);
509 goto out;
510 }
511
512 if (btf_is_func(type) && nr_funcs) {
513 nr = &nr_funcs;
514 root = &obj->funcs;
515 } else if (btf_is_struct(type) && nr_structs) {
516 nr = &nr_structs;
517 root = &obj->structs;
518 } else if (btf_is_union(type) && nr_unions) {
519 nr = &nr_unions;
520 root = &obj->unions;
521 } else if (btf_is_typedef(type) && nr_typedefs) {
522 nr = &nr_typedefs;
523 root = &obj->typedefs;
524 } else
525 continue;
526
527 str = btf__name_by_offset(btf, type->name_off);
528 if (!str) {
529 pr_err("FAILED: malformed BTF, can't resolve name for ID %d\n",
530 type_id);
531 goto out;
532 }
533
534 id = btf_id__find(root, str);
535 if (id) {
536 if (id->id) {
537 pr_info("WARN: multiple IDs found for '%s': %d, %d - using %d\n",
538 str, id->id, type_id, id->id);
539 } else {
540 id->id = type_id;
541 (*nr)--;
542 }
543 }
544 }
545
546 err = 0;
547 out:
548 btf__free(btf);
549 return err;
550 }
551
id_patch(struct object * obj,struct btf_id * id)552 static int id_patch(struct object *obj, struct btf_id *id)
553 {
554 Elf_Data *data = obj->efile.idlist;
555 int *ptr = data->d_buf;
556 int i;
557
558 if (!id->id) {
559 pr_err("WARN: resolve_btfids: unresolved symbol %s\n", id->name);
560 }
561
562 for (i = 0; i < id->addr_cnt; i++) {
563 unsigned long addr = id->addr[i];
564 unsigned long idx = addr - obj->efile.idlist_addr;
565
566 pr_debug("patching addr %5lu: ID %7d [%s]\n",
567 idx, id->id, id->name);
568
569 if (idx >= data->d_size) {
570 pr_err("FAILED patching index %lu out of bounds %lu\n",
571 idx, data->d_size);
572 return -1;
573 }
574
575 idx = idx / sizeof(int);
576 ptr[idx] = id->id;
577 }
578
579 return 0;
580 }
581
__symbols_patch(struct object * obj,struct rb_root * root)582 static int __symbols_patch(struct object *obj, struct rb_root *root)
583 {
584 struct rb_node *next;
585 struct btf_id *id;
586
587 next = rb_first(root);
588 while (next) {
589 id = rb_entry(next, struct btf_id, rb_node);
590
591 if (id_patch(obj, id))
592 return -1;
593
594 next = rb_next(next);
595 }
596 return 0;
597 }
598
cmp_id(const void * pa,const void * pb)599 static int cmp_id(const void *pa, const void *pb)
600 {
601 const int *a = pa, *b = pb;
602
603 return *a - *b;
604 }
605
sets_patch(struct object * obj)606 static int sets_patch(struct object *obj)
607 {
608 Elf_Data *data = obj->efile.idlist;
609 int *ptr = data->d_buf;
610 struct rb_node *next;
611
612 next = rb_first(&obj->sets);
613 while (next) {
614 unsigned long addr, idx;
615 struct btf_id *id;
616 int *base;
617 int cnt;
618
619 id = rb_entry(next, struct btf_id, rb_node);
620 addr = id->addr[0];
621 idx = addr - obj->efile.idlist_addr;
622
623 /* sets are unique */
624 if (id->addr_cnt != 1) {
625 pr_err("FAILED malformed data for set '%s'\n",
626 id->name);
627 return -1;
628 }
629
630 idx = idx / sizeof(int);
631 base = &ptr[idx] + 1;
632 cnt = ptr[idx];
633
634 pr_debug("sorting addr %5lu: cnt %6d [%s]\n",
635 (idx + 1) * sizeof(int), cnt, id->name);
636
637 qsort(base, cnt, sizeof(int), cmp_id);
638
639 next = rb_next(next);
640 }
641 return 0;
642 }
643
symbols_patch(struct object * obj)644 static int symbols_patch(struct object *obj)
645 {
646 int err;
647
648 if (__symbols_patch(obj, &obj->structs) ||
649 __symbols_patch(obj, &obj->unions) ||
650 __symbols_patch(obj, &obj->typedefs) ||
651 __symbols_patch(obj, &obj->funcs) ||
652 __symbols_patch(obj, &obj->sets))
653 return -1;
654
655 if (sets_patch(obj))
656 return -1;
657
658 /* Set type to ensure endian translation occurs. */
659 obj->efile.idlist->d_type = ELF_T_WORD;
660
661 elf_flagdata(obj->efile.idlist, ELF_C_SET, ELF_F_DIRTY);
662
663 err = elf_update(obj->efile.elf, ELF_C_WRITE);
664 if (err < 0) {
665 pr_err("FAILED elf_update(WRITE): %s\n",
666 elf_errmsg(-1));
667 }
668
669 pr_debug("update %s for %s\n",
670 err >= 0 ? "ok" : "failed", obj->path);
671 return err < 0 ? -1 : 0;
672 }
673
674 static const char * const resolve_btfids_usage[] = {
675 "resolve_btfids [<options>] <ELF object>",
676 NULL
677 };
678
main(int argc,const char ** argv)679 int main(int argc, const char **argv)
680 {
681 bool no_fail = false;
682 struct object obj = {
683 .efile = {
684 .idlist_shndx = -1,
685 .symbols_shndx = -1,
686 },
687 .structs = RB_ROOT,
688 .unions = RB_ROOT,
689 .typedefs = RB_ROOT,
690 .funcs = RB_ROOT,
691 .sets = RB_ROOT,
692 };
693 struct option btfid_options[] = {
694 OPT_INCR('v', "verbose", &verbose,
695 "be more verbose (show errors, etc)"),
696 OPT_STRING(0, "btf", &obj.btf, "BTF data",
697 "BTF data"),
698 OPT_BOOLEAN(0, "no-fail", &no_fail,
699 "do not fail if " BTF_IDS_SECTION " section is not found"),
700 OPT_END()
701 };
702 int err = -1;
703
704 argc = parse_options(argc, argv, btfid_options, resolve_btfids_usage,
705 PARSE_OPT_STOP_AT_NON_OPTION);
706 if (argc != 1)
707 usage_with_options(resolve_btfids_usage, btfid_options);
708
709 obj.path = argv[0];
710
711 if (elf_collect(&obj))
712 goto out;
713
714 /*
715 * We did not find .BTF_ids section or symbols section,
716 * nothing to do..
717 */
718 if (obj.efile.idlist_shndx == -1 ||
719 obj.efile.symbols_shndx == -1) {
720 if (no_fail)
721 return 0;
722 pr_err("FAILED to find needed sections\n");
723 return -1;
724 }
725
726 if (symbols_collect(&obj))
727 goto out;
728
729 if (symbols_resolve(&obj))
730 goto out;
731
732 if (symbols_patch(&obj))
733 goto out;
734
735 err = 0;
736 out:
737 if (obj.efile.elf) {
738 elf_end(obj.efile.elf);
739 close(obj.efile.fd);
740 }
741 return err;
742 }
743
