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 static int elf_collect(struct object *obj)
 237 {
 238         Elf_Scn *scn = NULL;
 239         size_t shdrstrndx;
 240         int idx = 0;
 241         Elf *elf;
 242         int fd;
 243
 244         fd = open(obj->path, O_RDWR, 0666);
 245         if (fd == -1) {
 246                 pr_err("FAILED cannot open %s: %s\n",
 247                         obj->path, strerror(errno));
 248                 return -1;
 249         }
 250
 251         elf_version(EV_CURRENT);
 252
 253         elf = elf_begin(fd, ELF_C_RDWR_MMAP, NULL);
 254         if (!elf) {
 255                 pr_err("FAILED cannot create ELF descriptor: %s\n",
 256                         elf_errmsg(-1));
 257                 return -1;
 258         }
 259
 260         obj->efile.fd  = fd;
 261         obj->efile.elf = elf;
 262
 263         elf_flagelf(elf, ELF_C_SET, ELF_F_LAYOUT);
 264
 265         if (elf_getshdrstrndx(elf, &shdrstrndx) != 0) {
 266                 pr_err("FAILED cannot get shdr str ndx\n");
 267                 return -1;
 268         }
 269
 270         /*
 271          * Scan all the elf sections and look for save data
 272          * from .BTF_ids section and symbols.
 273          */
 274         while ((scn = elf_nextscn(elf, scn)) != NULL) {
 275                 Elf_Data *data;
 276                 GElf_Shdr sh;
 277                 char *name;
 278
 279                 idx++;
 280                 if (gelf_getshdr(scn, &sh) != &sh) {
 281                         pr_err("FAILED get section(%d) header\n", idx);
 282                         return -1;
 283                 }
 284
 285                 name = elf_strptr(elf, shdrstrndx, sh.sh_name);
 286                 if (!name) {
 287                         pr_err("FAILED get section(%d) name\n", idx);
 288                         return -1;
 289                 }
 290
 291                 data = elf_getdata(scn, 0);
 292                 if (!data) {
 293                         pr_err("FAILED to get section(%d) data from %s\n",
 294                                 idx, name);
 295                         return -1;
 296                 }
 297
 298                 pr_debug2("section(%d) %s, size %ld, link %d, flags %lx, type=%d\n",
 299                           idx, name, (unsigned long) data->d_size,
 300                           (int) sh.sh_link, (unsigned long) sh.sh_flags,
 301                           (int) sh.sh_type);
 302
 303                 if (sh.sh_type == SHT_SYMTAB) {
 304                         obj->efile.symbols       = data;
 305                         obj->efile.symbols_shndx = idx;
 306                         obj->efile.strtabidx     = sh.sh_link;
 307                 } else if (!strcmp(name, BTF_IDS_SECTION)) {
 308                         obj->efile.idlist       = data;
 309                         obj->efile.idlist_shndx = idx;
 310                         obj->efile.idlist_addr  = sh.sh_addr;
 311                 }
 312         }
 313
 314         return 0;
 315 }
 316
 317 static int symbols_collect(struct object *obj)
 318 {
 319         Elf_Scn *scn = NULL;
 320         int n, i, err = 0;
 321         GElf_Shdr sh;
 322         char *name;
 323
 324         scn = elf_getscn(obj->efile.elf, obj->efile.symbols_shndx);
 325         if (!scn)
 326                 return -1;
 327
 328         if (gelf_getshdr(scn, &sh) != &sh)
 329                 return -1;
 330
 331         n = sh.sh_size / sh.sh_entsize;
 332
 333         /*
 334          * Scan symbols and look for the ones starting with
 335          * __BTF_ID__* over .BTF_ids section.
 336          */
 337         for (i = 0; !err && i < n; i++) {
 338                 char *tmp, *prefix;
 339                 struct btf_id *id;
 340                 GElf_Sym sym;
 341                 int err = -1;
 342
 343                 if (!gelf_getsym(obj->efile.symbols, i, &sym))
 344                         return -1;
 345
 346                 if (sym.st_shndx != obj->efile.idlist_shndx)
 347                         continue;
 348
 349                 name = elf_strptr(obj->efile.elf, obj->efile.strtabidx,
 350                                   sym.st_name);
 351
 352                 if (!is_btf_id(name))
 353                         continue;
 354
 355                 /*
 356                  * __BTF_ID__TYPE__vfs_truncate__0
 357                  * prefix =  ^
 358                  */
 359                 prefix = name + sizeof(BTF_ID) - 1;
 360
 361                 /* struct */
 362                 if (!strncmp(prefix, BTF_STRUCT, sizeof(BTF_STRUCT) - 1)) {
 363                         obj->nr_structs++;
 364                         id = add_symbol(&obj->structs, prefix, sizeof(BTF_STRUCT) - 1);
 365                 /* union  */
 366                 } else if (!strncmp(prefix, BTF_UNION, sizeof(BTF_UNION) - 1)) {
 367                         obj->nr_unions++;
 368                         id = add_symbol(&obj->unions, prefix, sizeof(BTF_UNION) - 1);
 369                 /* typedef */
 370                 } else if (!strncmp(prefix, BTF_TYPEDEF, sizeof(BTF_TYPEDEF) - 1)) {
 371                         obj->nr_typedefs++;
 372                         id = add_symbol(&obj->typedefs, prefix, sizeof(BTF_TYPEDEF) - 1);
 373                 /* func */
 374                 } else if (!strncmp(prefix, BTF_FUNC, sizeof(BTF_FUNC) - 1)) {
 375                         obj->nr_funcs++;
 376                         id = add_symbol(&obj->funcs, prefix, sizeof(BTF_FUNC) - 1);
 377                 /* set */
 378                 } else if (!strncmp(prefix, BTF_SET, sizeof(BTF_SET) - 1)) {
 379                         id = add_symbol(&obj->sets, prefix, sizeof(BTF_SET) - 1);
 380                         /*
 381                          * SET objects store list's count, which is encoded
 382                          * in symbol's size, together with 'cnt' field hence
 383                          * that - 1.
 384                          */
 385                         if (id)
 386                                 id->cnt = sym.st_size / sizeof(int) - 1;
 387                 } else {
 388                         pr_err("FAILED unsupported prefix %s\n", prefix);
 389                         return -1;
 390                 }
 391
 392                 if (!id)
 393                         return -ENOMEM;
 394
 395                 if (id->addr_cnt >= ADDR_CNT) {
 396                         pr_err("FAILED symbol %s crossed the number of allowed lists",
 397                                 id->name);
 398                         return -1;
 399                 }
 400                 id->addr[id->addr_cnt++] = sym.st_value;
 401         }
 402
 403         return 0;
 404 }
 405
 406 static int symbols_resolve(struct object *obj)
 407 {
 408         int nr_typedefs = obj->nr_typedefs;
 409         int nr_structs  = obj->nr_structs;
 410         int nr_unions   = obj->nr_unions;
 411         int nr_funcs    = obj->nr_funcs;
 412         int err, type_id;
 413         struct btf *btf;
 414         __u32 nr;
 415
 416         btf = btf__parse(obj->btf ?: obj->path, NULL);
 417         err = libbpf_get_error(btf);
 418         if (err) {
 419                 pr_err("FAILED: load BTF from %s: %s",
 420                         obj->path, strerror(err));
 421                 return -1;
 422         }
 423
 424         err = -1;
 425         nr  = btf__get_nr_types(btf);
 426
 427         /*
 428          * Iterate all the BTF types and search for collected symbol IDs.
 429          */
 430         for (type_id = 1; type_id <= nr; type_id++) {
 431                 const struct btf_type *type;
 432                 struct rb_root *root;
 433                 struct btf_id *id;
 434                 const char *str;
 435                 int *nr;
 436
 437                 type = btf__type_by_id(btf, type_id);
 438                 if (!type) {
 439                         pr_err("FAILED: malformed BTF, can't resolve type for ID %d\n",
 440                                 type_id);
 441                         goto out;
 442                 }
 443
 444                 if (btf_is_func(type) && nr_funcs) {
 445                         nr   = &nr_funcs;
 446                         root = &obj->funcs;
 447                 } else if (btf_is_struct(type) && nr_structs) {
 448                         nr   = &nr_structs;
 449                         root = &obj->structs;
 450                 } else if (btf_is_union(type) && nr_unions) {
 451                         nr   = &nr_unions;
 452                         root = &obj->unions;
 453                 } else if (btf_is_typedef(type) && nr_typedefs) {
 454                         nr   = &nr_typedefs;
 455                         root = &obj->typedefs;
 456                 } else
 457                         continue;
 458
 459                 str = btf__name_by_offset(btf, type->name_off);
 460                 if (!str) {
 461                         pr_err("FAILED: malformed BTF, can't resolve name for ID %d\n",
 462                                 type_id);
 463                         goto out;
 464                 }
 465
 466                 id = btf_id__find(root, str);
 467                 if (id) {
 468                         id->id = type_id;
 469                         (*nr)--;
 470                 }
 471         }
 472
 473         err = 0;
 474 out:
 475         btf__free(btf);
 476         return err;
 477 }
 478
 479 static int id_patch(struct object *obj, struct btf_id *id)
 480 {
 481         Elf_Data *data = obj->efile.idlist;
 482         int *ptr = data->d_buf;
 483         int i;
 484
 485         if (!id->id) {
 486                 pr_err("FAILED unresolved symbol %s\n", id->name);
 487                 return -EINVAL;
 488         }
 489
 490         for (i = 0; i < id->addr_cnt; i++) {
 491                 unsigned long addr = id->addr[i];
 492                 unsigned long idx = addr - obj->efile.idlist_addr;
 493
 494                 pr_debug("patching addr %5lu: ID %7d [%s]\n",
 495                          idx, id->id, id->name);
 496
 497                 if (idx >= data->d_size) {
 498                         pr_err("FAILED patching index %lu out of bounds %lu\n",
 499                                 idx, data->d_size);
 500                         return -1;
 501                 }
 502
 503                 idx = idx / sizeof(int);
 504                 ptr[idx] = id->id;
 505         }
 506
 507         return 0;
 508 }
 509
 510 static int __symbols_patch(struct object *obj, struct rb_root *root)
 511 {
 512         struct rb_node *next;
 513         struct btf_id *id;
 514
 515         next = rb_first(root);
 516         while (next) {
 517                 id = rb_entry(next, struct btf_id, rb_node);
 518
 519                 if (id_patch(obj, id))
 520                         return -1;
 521
 522                 next = rb_next(next);
 523         }
 524         return 0;
 525 }
 526
 527 static int cmp_id(const void *pa, const void *pb)
 528 {
 529         const int *a = pa, *b = pb;
 530
 531         return *a - *b;
 532 }
 533
 534 static int sets_patch(struct object *obj)
 535 {
 536         Elf_Data *data = obj->efile.idlist;
 537         int *ptr = data->d_buf;
 538         struct rb_node *next;
 539
 540         next = rb_first(&obj->sets);
 541         while (next) {
 542                 unsigned long addr, idx;
 543                 struct btf_id *id;
 544                 int *base;
 545                 int cnt;
 546
 547                 id   = rb_entry(next, struct btf_id, rb_node);
 548                 addr = id->addr[0];
 549                 idx  = addr - obj->efile.idlist_addr;
 550
 551                 /* sets are unique */
 552                 if (id->addr_cnt != 1) {
 553                         pr_err("FAILED malformed data for set '%s'\n",
 554                                 id->name);
 555                         return -1;
 556                 }
 557
 558                 idx = idx / sizeof(int);
 559                 base = &ptr[idx] + 1;
 560                 cnt = ptr[idx];
 561
 562                 pr_debug("sorting  addr %5lu: cnt %6d [%s]\n",
 563                          (idx + 1) * sizeof(int), cnt, id->name);
 564
 565                 qsort(base, cnt, sizeof(int), cmp_id);
 566
 567                 next = rb_next(next);
 568         }
 569 }
 570
 571 static int symbols_patch(struct object *obj)
 572 {
 573         int err;
 574
 575         if (__symbols_patch(obj, &obj->structs)  ||
 576             __symbols_patch(obj, &obj->unions)   ||
 577             __symbols_patch(obj, &obj->typedefs) ||
 578             __symbols_patch(obj, &obj->funcs)    ||
 579             __symbols_patch(obj, &obj->sets))
 580                 return -1;
 581
 582         if (sets_patch(obj))
 583                 return -1;
 584
 585         elf_flagdata(obj->efile.idlist, ELF_C_SET, ELF_F_DIRTY);
 586
 587         err = elf_update(obj->efile.elf, ELF_C_WRITE);
 588         if (err < 0) {
 589                 pr_err("FAILED elf_update(WRITE): %s\n",
 590                         elf_errmsg(-1));
 591         }
 592
 593         pr_debug("update %s for %s\n",
 594                  err >= 0 ? "ok" : "failed", obj->path);
 595         return err < 0 ? -1 : 0;
 596 }
 597
 598 static const char * const resolve_btfids_usage[] = {
 599         "resolve_btfids [<options>] <ELF object>",
 600         NULL
 601 };
 602
 603 int main(int argc, const char **argv)
 604 {
 605         bool no_fail = false;
 606         struct object obj = {
 607                 .efile = {
 608                         .idlist_shndx  = -1,
 609                         .symbols_shndx = -1,
 610                 },
 611                 .structs  = RB_ROOT,
 612                 .unions   = RB_ROOT,
 613                 .typedefs = RB_ROOT,
 614                 .funcs    = RB_ROOT,
 615                 .sets     = RB_ROOT,
 616         };
 617         struct option btfid_options[] = {
 618                 OPT_INCR('v', "verbose", &verbose,
 619                          "be more verbose (show errors, etc)"),
 620                 OPT_STRING(0, "btf", &obj.btf, "BTF data",
 621                            "BTF data"),
 622                 OPT_BOOLEAN(0, "no-fail", &no_fail,
 623                            "do not fail if " BTF_IDS_SECTION " section is not found"),
 624                 OPT_END()
 625         };
 626         int err = -1;
 627
 628         argc = parse_options(argc, argv, btfid_options, resolve_btfids_usage,
 629                              PARSE_OPT_STOP_AT_NON_OPTION);
 630         if (argc != 1)
 631                 usage_with_options(resolve_btfids_usage, btfid_options);
 632
 633         obj.path = argv[0];
 634
 635         if (elf_collect(&obj))
 636                 goto out;
 637
 638         /*
 639          * We did not find .BTF_ids section or symbols section,
 640          * nothing to do..
 641          */
 642         if (obj.efile.idlist_shndx == -1 ||
 643             obj.efile.symbols_shndx == -1) {
 644                 if (no_fail)
 645                         return 0;
 646                 pr_err("FAILED to find needed sections\n");
 647                 return -1;
 648         }
 649
 650         if (symbols_collect(&obj))
 651                 goto out;
 652
 653         if (symbols_resolve(&obj))
 654                 goto out;
 655
 656         if (symbols_patch(&obj))
 657                 goto out;
 658
 659         err = 0;
 660 out:
 661         if (obj.efile.elf)
 662                 elf_end(obj.efile.elf);
 663         close(obj.efile.fd);
 664         return err;
 665 }