tools/bpf/resolve_btfids/main.c

   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
 118 int eprintf(int level, int var, const char *fmt, ...)
 119 {
 120         va_list args;
 121         int ret;
 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
 143 static bool is_btf_id(const char *name)
 144 {
 145         return name && !strncmp(name, BTF_ID, sizeof(BTF_ID) - 1);
 146 }
 147
 148 static struct btf_id *btf_id__find(struct rb_root *root, const char *name)
 149 {
 150         struct rb_node *p = root->rb_node;
 151         struct btf_id *id;
 152         int cmp;
 153
 154         while (p) {
 155                 id = rb_entry(p, struct btf_id, rb_node);
 156                 cmp = strcmp(id->name, name);
 157                 if (cmp < 0)
 158                         p = p->rb_left;
 159                 else if (cmp > 0)
 160                         p = p->rb_right;
 161                 else
 162                         return id;
 163         }
 164         return NULL;
 165 }
 166
 167 static struct btf_id*
 168 btf_id__add(struct rb_root *root, char *name, bool unique)
 169 {
 170         struct rb_node **p = &root->rb_node;
 171         struct rb_node *parent = NULL;
 172         struct btf_id *id;
 173         int cmp;
 174
 175         while (*p != NULL) {
 176                 parent = *p;
 177                 id = rb_entry(parent, struct btf_id, rb_node);
 178                 cmp = strcmp(id->name, name);
 179                 if (cmp < 0)
 180                         p = &(*p)->rb_left;
 181                 else if (cmp > 0)
 182                         p = &(*p)->rb_right;
 183                 else
 184                         return unique ? NULL : id;
 185         }
 186
 187         id = zalloc(sizeof(*id));
 188         if (id) {
 189                 pr_debug("adding symbol %s\n", name);
 190                 id->name = name;
 191                 rb_link_node(&id->rb_node, parent, p);
 192                 rb_insert_color(&id->rb_node, root);
 193         }
 194         return id;
 195 }
 196
 197 static char *get_id(const char *prefix_end)
 198 {
 199         /*
 200          * __BTF_ID__func__vfs_truncate__0
 201          * prefix_end =  ^
 202          */
 203         char *p, *id = strdup(prefix_end + sizeof("__") - 1);
 204
 205         if (id) {
 206                 /*
 207                  * __BTF_ID__func__vfs_truncate__0
 208                  * id =            ^
 209                  *
 210                  * cut the unique id part
 211                  */
 212                 p = strrchr(id, '_');
 213                 p--;
 214                 if (*p != '_') {
 215                         free(id);
 216                         return NULL;
 217                 }
 218                 *p = '\0';
 219         }
 220         return id;
 221 }
 222
 223 static struct btf_id *add_symbol(struct rb_root *root, char *name, size_t size)
 224 {
 225         char *id;
 226
 227         id = get_id(name + size);
 228         if (!id) {
 229                 pr_err("FAILED to parse symbol name: %s\n", name);
 230                 return NULL;
 231         }
 232
 233         return btf_id__add(root, id, false);
 234 }
 235
 236 /*
 237  * The data of compressed section should be aligned to 4
 238  * (for 32bit) or 8 (for 64 bit) bytes. The binutils ld
 239  * sets sh_addralign to 1, which makes libelf fail with
 240  * misaligned section error during the update:
 241  *    FAILED elf_update(WRITE): invalid section alignment
 242  *
 243  * While waiting for ld fix, we fix the compressed sections
 244  * sh_addralign value manualy.
 245  */
 246 static int compressed_section_fix(Elf *elf, Elf_Scn *scn, GElf_Shdr *sh)
 247 {
 248         int expected = gelf_getclass(elf) == ELFCLASS32 ? 4 : 8;
 249
 250         if (!(sh->sh_flags & SHF_COMPRESSED))
 251                 return 0;
 252
 253         if (sh->sh_addralign == expected)
 254                 return 0;
 255
 256         pr_debug2(" - fixing wrong alignment sh_addralign %u, expected %u\n",
 257                   sh->sh_addralign, expected);
 258
 259         sh->sh_addralign = expected;
 260
 261         if (gelf_update_shdr(scn, sh) == 0) {
 262                 printf("FAILED cannot update section header: %s\n",
 263                         elf_errmsg(-1));
 264                 return -1;
 265         }
 266         return 0;
 267 }
 268
 269 static int elf_collect(struct object *obj)
 270 {
 271         Elf_Scn *scn = NULL;
 272         size_t shdrstrndx;
 273         int idx = 0;
 274         Elf *elf;
 275         int fd;
 276
 277         fd = open(obj->path, O_RDWR, 0666);
 278         if (fd == -1) {
 279                 pr_err("FAILED cannot open %s: %s\n",
 280                         obj->path, strerror(errno));
 281                 return -1;
 282         }
 283
 284         elf_version(EV_CURRENT);
 285
 286         elf = elf_begin(fd, ELF_C_RDWR_MMAP, NULL);
 287         if (!elf) {
 288                 pr_err("FAILED cannot create ELF descriptor: %s\n",
 289                         elf_errmsg(-1));
 290                 return -1;
 291         }
 292
 293         obj->efile.fd  = fd;
 294         obj->efile.elf = elf;
 295
 296         elf_flagelf(elf, ELF_C_SET, ELF_F_LAYOUT);
 297
 298         if (elf_getshdrstrndx(elf, &shdrstrndx) != 0) {
 299                 pr_err("FAILED cannot get shdr str ndx\n");
 300                 return -1;
 301         }
 302
 303         /*
 304          * Scan all the elf sections and look for save data
 305          * from .BTF_ids section and symbols.
 306          */
 307         while ((scn = elf_nextscn(elf, scn)) != NULL) {
 308                 Elf_Data *data;
 309                 GElf_Shdr sh;
 310                 char *name;
 311
 312                 idx++;
 313                 if (gelf_getshdr(scn, &sh) != &sh) {
 314                         pr_err("FAILED get section(%d) header\n", idx);
 315                         return -1;
 316                 }
 317
 318                 name = elf_strptr(elf, shdrstrndx, sh.sh_name);
 319                 if (!name) {
 320                         pr_err("FAILED get section(%d) name\n", idx);
 321                         return -1;
 322                 }
 323
 324                 data = elf_getdata(scn, 0);
 325                 if (!data) {
 326                         pr_err("FAILED to get section(%d) data from %s\n",
 327                                 idx, name);
 328                         return -1;
 329                 }
 330
 331                 pr_debug2("section(%d) %s, size %ld, link %d, flags %lx, type=%d\n",
 332                           idx, name, (unsigned long) data->d_size,
 333                           (int) sh.sh_link, (unsigned long) sh.sh_flags,
 334                           (int) sh.sh_type);
 335
 336                 if (sh.sh_type == SHT_SYMTAB) {
 337                         obj->efile.symbols       = data;
 338                         obj->efile.symbols_shndx = idx;
 339                         obj->efile.strtabidx     = sh.sh_link;
 340                 } else if (!strcmp(name, BTF_IDS_SECTION)) {
 341                         obj->efile.idlist       = data;
 342                         obj->efile.idlist_shndx = idx;
 343                         obj->efile.idlist_addr  = sh.sh_addr;
 344                 }
 345
 346                 if (compressed_section_fix(elf, scn, &sh))
 347                         return -1;
 348         }
 349
 350         return 0;
 351 }
 352
 353 static int symbols_collect(struct object *obj)
 354 {
 355         Elf_Scn *scn = NULL;
 356         int n, i, err = 0;
 357         GElf_Shdr sh;
 358         char *name;
 359
 360         scn = elf_getscn(obj->efile.elf, obj->efile.symbols_shndx);
 361         if (!scn)
 362                 return -1;
 363
 364         if (gelf_getshdr(scn, &sh) != &sh)
 365                 return -1;
 366
 367         n = sh.sh_size / sh.sh_entsize;
 368
 369         /*
 370          * Scan symbols and look for the ones starting with
 371          * __BTF_ID__* over .BTF_ids section.
 372          */
 373         for (i = 0; !err && i < n; i++) {
 374                 char *tmp, *prefix;
 375                 struct btf_id *id;
 376                 GElf_Sym sym;
 377                 int err = -1;
 378
 379                 if (!gelf_getsym(obj->efile.symbols, i, &sym))
 380                         return -1;
 381
 382                 if (sym.st_shndx != obj->efile.idlist_shndx)
 383                         continue;
 384
 385                 name = elf_strptr(obj->efile.elf, obj->efile.strtabidx,
 386                                   sym.st_name);
 387
 388                 if (!is_btf_id(name))
 389                         continue;
 390
 391                 /*
 392                  * __BTF_ID__TYPE__vfs_truncate__0
 393                  * prefix =  ^
 394                  */
 395                 prefix = name + sizeof(BTF_ID) - 1;
 396
 397                 /* struct */
 398                 if (!strncmp(prefix, BTF_STRUCT, sizeof(BTF_STRUCT) - 1)) {
 399                         obj->nr_structs++;
 400                         id = add_symbol(&obj->structs, prefix, sizeof(BTF_STRUCT) - 1);
 401                 /* union  */
 402                 } else if (!strncmp(prefix, BTF_UNION, sizeof(BTF_UNION) - 1)) {
 403                         obj->nr_unions++;
 404                         id = add_symbol(&obj->unions, prefix, sizeof(BTF_UNION) - 1);
 405                 /* typedef */
 406                 } else if (!strncmp(prefix, BTF_TYPEDEF, sizeof(BTF_TYPEDEF) - 1)) {
 407                         obj->nr_typedefs++;
 408                         id = add_symbol(&obj->typedefs, prefix, sizeof(BTF_TYPEDEF) - 1);
 409                 /* func */
 410                 } else if (!strncmp(prefix, BTF_FUNC, sizeof(BTF_FUNC) - 1)) {
 411                         obj->nr_funcs++;
 412                         id = add_symbol(&obj->funcs, prefix, sizeof(BTF_FUNC) - 1);
 413                 /* set */
 414                 } else if (!strncmp(prefix, BTF_SET, sizeof(BTF_SET) - 1)) {
 415                         id = add_symbol(&obj->sets, prefix, sizeof(BTF_SET) - 1);
 416                         /*
 417                          * SET objects store list's count, which is encoded
 418                          * in symbol's size, together with 'cnt' field hence
 419                          * that - 1.
 420                          */
 421                         if (id)
 422                                 id->cnt = sym.st_size / sizeof(int) - 1;
 423                 } else {
 424                         pr_err("FAILED unsupported prefix %s\n", prefix);
 425                         return -1;
 426                 }
 427
 428                 if (!id)
 429                         return -ENOMEM;
 430
 431                 if (id->addr_cnt >= ADDR_CNT) {
 432                         pr_err("FAILED symbol %s crossed the number of allowed lists",
 433                                 id->name);
 434                         return -1;
 435                 }
 436                 id->addr[id->addr_cnt++] = sym.st_value;
 437         }
 438
 439         return 0;
 440 }
 441
 442 static int symbols_resolve(struct object *obj)
 443 {
 444         int nr_typedefs = obj->nr_typedefs;
 445         int nr_structs  = obj->nr_structs;
 446         int nr_unions   = obj->nr_unions;
 447         int nr_funcs    = obj->nr_funcs;
 448         int err, type_id;
 449         struct btf *btf;
 450         __u32 nr;
 451
 452         btf = btf__parse(obj->btf ?: obj->path, NULL);
 453         err = libbpf_get_error(btf);
 454         if (err) {
 455                 pr_err("FAILED: load BTF from %s: %s",
 456                         obj->path, strerror(err));
 457                 return -1;
 458         }
 459
 460         err = -1;
 461         nr  = btf__get_nr_types(btf);
 462
 463         /*
 464          * Iterate all the BTF types and search for collected symbol IDs.
 465          */
 466         for (type_id = 1; type_id <= nr; type_id++) {
 467                 const struct btf_type *type;
 468                 struct rb_root *root;
 469                 struct btf_id *id;
 470                 const char *str;
 471                 int *nr;
 472
 473                 type = btf__type_by_id(btf, type_id);
 474                 if (!type) {
 475                         pr_err("FAILED: malformed BTF, can't resolve type for ID %d\n",
 476                                 type_id);
 477                         goto out;
 478                 }
 479
 480                 if (btf_is_func(type) && nr_funcs) {
 481                         nr   = &nr_funcs;
 482                         root = &obj->funcs;
 483                 } else if (btf_is_struct(type) && nr_structs) {
 484                         nr   = &nr_structs;
 485                         root = &obj->structs;
 486                 } else if (btf_is_union(type) && nr_unions) {
 487                         nr   = &nr_unions;
 488                         root = &obj->unions;
 489                 } else if (btf_is_typedef(type) && nr_typedefs) {
 490                         nr   = &nr_typedefs;
 491                         root = &obj->typedefs;
 492                 } else
 493                         continue;
 494
 495                 str = btf__name_by_offset(btf, type->name_off);
 496                 if (!str) {
 497                         pr_err("FAILED: malformed BTF, can't resolve name for ID %d\n",
 498                                 type_id);
 499                         goto out;
 500                 }
 501
 502                 id = btf_id__find(root, str);
 503                 if (id) {
 504                         id->id = type_id;
 505                         (*nr)--;
 506                 }
 507         }
 508
 509         err = 0;
 510 out:
 511         btf__free(btf);
 512         return err;
 513 }
 514
 515 static int id_patch(struct object *obj, struct btf_id *id)
 516 {
 517         Elf_Data *data = obj->efile.idlist;
 518         int *ptr = data->d_buf;
 519         int i;
 520
 521         if (!id->id) {
 522                 pr_err("FAILED unresolved symbol %s\n", id->name);
 523                 return -EINVAL;
 524         }
 525
 526         for (i = 0; i < id->addr_cnt; i++) {
 527                 unsigned long addr = id->addr[i];
 528                 unsigned long idx = addr - obj->efile.idlist_addr;
 529
 530                 pr_debug("patching addr %5lu: ID %7d [%s]\n",
 531                          idx, id->id, id->name);
 532
 533                 if (idx >= data->d_size) {
 534                         pr_err("FAILED patching index %lu out of bounds %lu\n",
 535                                 idx, data->d_size);
 536                         return -1;
 537                 }
 538
 539                 idx = idx / sizeof(int);
 540                 ptr[idx] = id->id;
 541         }
 542
 543         return 0;
 544 }
 545
 546 static int __symbols_patch(struct object *obj, struct rb_root *root)
 547 {
 548         struct rb_node *next;
 549         struct btf_id *id;
 550
 551         next = rb_first(root);
 552         while (next) {
 553                 id = rb_entry(next, struct btf_id, rb_node);
 554
 555                 if (id_patch(obj, id))
 556                         return -1;
 557
 558                 next = rb_next(next);
 559         }
 560         return 0;
 561 }
 562
 563 static int cmp_id(const void *pa, const void *pb)
 564 {
 565         const int *a = pa, *b = pb;
 566
 567         return *a - *b;
 568 }
 569
 570 static int sets_patch(struct object *obj)
 571 {
 572         Elf_Data *data = obj->efile.idlist;
 573         int *ptr = data->d_buf;
 574         struct rb_node *next;
 575
 576         next = rb_first(&obj->sets);
 577         while (next) {
 578                 unsigned long addr, idx;
 579                 struct btf_id *id;
 580                 int *base;
 581                 int cnt;
 582
 583                 id   = rb_entry(next, struct btf_id, rb_node);
 584                 addr = id->addr[0];
 585                 idx  = addr - obj->efile.idlist_addr;
 586
 587                 /* sets are unique */
 588                 if (id->addr_cnt != 1) {
 589                         pr_err("FAILED malformed data for set '%s'\n",
 590                                 id->name);
 591                         return -1;
 592                 }
 593
 594                 idx = idx / sizeof(int);
 595                 base = &ptr[idx] + 1;
 596                 cnt = ptr[idx];
 597
 598                 pr_debug("sorting  addr %5lu: cnt %6d [%s]\n",
 599                          (idx + 1) * sizeof(int), cnt, id->name);
 600
 601                 qsort(base, cnt, sizeof(int), cmp_id);
 602
 603                 next = rb_next(next);
 604         }
 605         return 0;
 606 }
 607
 608 static int symbols_patch(struct object *obj)
 609 {
 610         int err;
 611
 612         if (__symbols_patch(obj, &obj->structs)  ||
 613             __symbols_patch(obj, &obj->unions)   ||
 614             __symbols_patch(obj, &obj->typedefs) ||
 615             __symbols_patch(obj, &obj->funcs)    ||
 616             __symbols_patch(obj, &obj->sets))
 617                 return -1;
 618
 619         if (sets_patch(obj))
 620                 return -1;
 621
 622         elf_flagdata(obj->efile.idlist, ELF_C_SET, ELF_F_DIRTY);
 623
 624         err = elf_update(obj->efile.elf, ELF_C_WRITE);
 625         if (err < 0) {
 626                 pr_err("FAILED elf_update(WRITE): %s\n",
 627                         elf_errmsg(-1));
 628         }
 629
 630         pr_debug("update %s for %s\n",
 631                  err >= 0 ? "ok" : "failed", obj->path);
 632         return err < 0 ? -1 : 0;
 633 }
 634
 635 static const char * const resolve_btfids_usage[] = {
 636         "resolve_btfids [<options>] <ELF object>",
 637         NULL
 638 };
 639
 640 int main(int argc, const char **argv)
 641 {
 642         bool no_fail = false;
 643         struct object obj = {
 644                 .efile = {
 645                         .idlist_shndx  = -1,
 646                         .symbols_shndx = -1,
 647                 },
 648                 .structs  = RB_ROOT,
 649                 .unions   = RB_ROOT,
 650                 .typedefs = RB_ROOT,
 651                 .funcs    = RB_ROOT,
 652                 .sets     = RB_ROOT,
 653         };
 654         struct option btfid_options[] = {
 655                 OPT_INCR('v', "verbose", &verbose,
 656                          "be more verbose (show errors, etc)"),
 657                 OPT_STRING(0, "btf", &obj.btf, "BTF data",
 658                            "BTF data"),
 659                 OPT_BOOLEAN(0, "no-fail", &no_fail,
 660                            "do not fail if " BTF_IDS_SECTION " section is not found"),
 661                 OPT_END()
 662         };
 663         int err = -1;
 664
 665         argc = parse_options(argc, argv, btfid_options, resolve_btfids_usage,
 666                              PARSE_OPT_STOP_AT_NON_OPTION);
 667         if (argc != 1)
 668                 usage_with_options(resolve_btfids_usage, btfid_options);
 669
 670         obj.path = argv[0];
 671
 672         if (elf_collect(&obj))
 673                 goto out;
 674
 675         /*
 676          * We did not find .BTF_ids section or symbols section,
 677          * nothing to do..
 678          */
 679         if (obj.efile.idlist_shndx == -1 ||
 680             obj.efile.symbols_shndx == -1) {
 681                 if (no_fail)
 682                         return 0;
 683                 pr_err("FAILED to find needed sections\n");
 684                 return -1;
 685         }
 686
 687         if (symbols_collect(&obj))
 688                 goto out;
 689
 690         if (symbols_resolve(&obj))
 691                 goto out;
 692
 693         if (symbols_patch(&obj))
 694                 goto out;
 695
 696         err = 0;
 697 out:
 698         if (obj.efile.elf)
 699                 elf_end(obj.efile.elf);
 700         close(obj.efile.fd);
 701         return err;
 702 }