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          * pos        =    ^
 203          */
 204         int len = strlen(prefix_end);
 205         int pos = sizeof("__") - 1;
 206         char *p, *id;
 207
 208         if (pos >= len)
 209                 return NULL;
 210
 211         id = strdup(prefix_end + pos);
 212         if (id) {
 213                 /*
 214                  * __BTF_ID__func__vfs_truncate__0
 215                  * id =            ^
 216                  *
 217                  * cut the unique id part
 218                  */
 219                 p = strrchr(id, '_');
 220                 p--;
 221                 if (*p != '_') {
 222                         free(id);
 223                         return NULL;
 224                 }
 225                 *p = '\0';
 226         }
 227         return id;
 228 }
 229
 230 static struct btf_id *add_set(struct object *obj, char *name)
 231 {
 232         /*
 233          * __BTF_ID__set__name
 234          * name =    ^
 235          * id   =         ^
 236          */
 237         char *id = name + sizeof(BTF_SET "__") - 1;
 238         int len = strlen(name);
 239
 240         if (id >= name + len) {
 241                 pr_err("FAILED to parse set name: %s\n", name);
 242                 return NULL;
 243         }
 244
 245         return btf_id__add(&obj->sets, id, true);
 246 }
 247
 248 static struct btf_id *add_symbol(struct rb_root *root, char *name, size_t size)
 249 {
 250         char *id;
 251
 252         id = get_id(name + size);
 253         if (!id) {
 254                 pr_err("FAILED to parse symbol name: %s\n", name);
 255                 return NULL;
 256         }
 257
 258         return btf_id__add(root, id, false);
 259 }
 260
 261 /*
 262  * The data of compressed section should be aligned to 4
 263  * (for 32bit) or 8 (for 64 bit) bytes. The binutils ld
 264  * sets sh_addralign to 1, which makes libelf fail with
 265  * misaligned section error during the update:
 266  *    FAILED elf_update(WRITE): invalid section alignment
 267  *
 268  * While waiting for ld fix, we fix the compressed sections
 269  * sh_addralign value manualy.
 270  */
 271 static int compressed_section_fix(Elf *elf, Elf_Scn *scn, GElf_Shdr *sh)
 272 {
 273         int expected = gelf_getclass(elf) == ELFCLASS32 ? 4 : 8;
 274
 275         if (!(sh->sh_flags & SHF_COMPRESSED))
 276                 return 0;
 277
 278         if (sh->sh_addralign == expected)
 279                 return 0;
 280
 281         pr_debug2(" - fixing wrong alignment sh_addralign %u, expected %u\n",
 282                   sh->sh_addralign, expected);
 283
 284         sh->sh_addralign = expected;
 285
 286         if (gelf_update_shdr(scn, sh) == 0) {
 287                 printf("FAILED cannot update section header: %s\n",
 288                         elf_errmsg(-1));
 289                 return -1;
 290         }
 291         return 0;
 292 }
 293
 294 static int elf_collect(struct object *obj)
 295 {
 296         Elf_Scn *scn = NULL;
 297         size_t shdrstrndx;
 298         int idx = 0;
 299         Elf *elf;
 300         int fd;
 301
 302         fd = open(obj->path, O_RDWR, 0666);
 303         if (fd == -1) {
 304                 pr_err("FAILED cannot open %s: %s\n",
 305                         obj->path, strerror(errno));
 306                 return -1;
 307         }
 308
 309         elf_version(EV_CURRENT);
 310
 311         elf = elf_begin(fd, ELF_C_RDWR_MMAP, NULL);
 312         if (!elf) {
 313                 pr_err("FAILED cannot create ELF descriptor: %s\n",
 314                         elf_errmsg(-1));
 315                 return -1;
 316         }
 317
 318         obj->efile.fd  = fd;
 319         obj->efile.elf = elf;
 320
 321         elf_flagelf(elf, ELF_C_SET, ELF_F_LAYOUT);
 322
 323         if (elf_getshdrstrndx(elf, &shdrstrndx) != 0) {
 324                 pr_err("FAILED cannot get shdr str ndx\n");
 325                 return -1;
 326         }
 327
 328         /*
 329          * Scan all the elf sections and look for save data
 330          * from .BTF_ids section and symbols.
 331          */
 332         while ((scn = elf_nextscn(elf, scn)) != NULL) {
 333                 Elf_Data *data;
 334                 GElf_Shdr sh;
 335                 char *name;
 336
 337                 idx++;
 338                 if (gelf_getshdr(scn, &sh) != &sh) {
 339                         pr_err("FAILED get section(%d) header\n", idx);
 340                         return -1;
 341                 }
 342
 343                 name = elf_strptr(elf, shdrstrndx, sh.sh_name);
 344                 if (!name) {
 345                         pr_err("FAILED get section(%d) name\n", idx);
 346                         return -1;
 347                 }
 348
 349                 data = elf_getdata(scn, 0);
 350                 if (!data) {
 351                         pr_err("FAILED to get section(%d) data from %s\n",
 352                                 idx, name);
 353                         return -1;
 354                 }
 355
 356                 pr_debug2("section(%d) %s, size %ld, link %d, flags %lx, type=%d\n",
 357                           idx, name, (unsigned long) data->d_size,
 358                           (int) sh.sh_link, (unsigned long) sh.sh_flags,
 359                           (int) sh.sh_type);
 360
 361                 if (sh.sh_type == SHT_SYMTAB) {
 362                         obj->efile.symbols       = data;
 363                         obj->efile.symbols_shndx = idx;
 364                         obj->efile.strtabidx     = sh.sh_link;
 365                 } else if (!strcmp(name, BTF_IDS_SECTION)) {
 366                         obj->efile.idlist       = data;
 367                         obj->efile.idlist_shndx = idx;
 368                         obj->efile.idlist_addr  = sh.sh_addr;
 369                 }
 370
 371                 if (compressed_section_fix(elf, scn, &sh))
 372                         return -1;
 373         }
 374
 375         return 0;
 376 }
 377
 378 static int symbols_collect(struct object *obj)
 379 {
 380         Elf_Scn *scn = NULL;
 381         int n, i, err = 0;
 382         GElf_Shdr sh;
 383         char *name;
 384
 385         scn = elf_getscn(obj->efile.elf, obj->efile.symbols_shndx);
 386         if (!scn)
 387                 return -1;
 388
 389         if (gelf_getshdr(scn, &sh) != &sh)
 390                 return -1;
 391
 392         n = sh.sh_size / sh.sh_entsize;
 393
 394         /*
 395          * Scan symbols and look for the ones starting with
 396          * __BTF_ID__* over .BTF_ids section.
 397          */
 398         for (i = 0; !err && i < n; i++) {
 399                 char *tmp, *prefix;
 400                 struct btf_id *id;
 401                 GElf_Sym sym;
 402                 int err = -1;
 403
 404                 if (!gelf_getsym(obj->efile.symbols, i, &sym))
 405                         return -1;
 406
 407                 if (sym.st_shndx != obj->efile.idlist_shndx)
 408                         continue;
 409
 410                 name = elf_strptr(obj->efile.elf, obj->efile.strtabidx,
 411                                   sym.st_name);
 412
 413                 if (!is_btf_id(name))
 414                         continue;
 415
 416                 /*
 417                  * __BTF_ID__TYPE__vfs_truncate__0
 418                  * prefix =  ^
 419                  */
 420                 prefix = name + sizeof(BTF_ID) - 1;
 421
 422                 /* struct */
 423                 if (!strncmp(prefix, BTF_STRUCT, sizeof(BTF_STRUCT) - 1)) {
 424                         obj->nr_structs++;
 425                         id = add_symbol(&obj->structs, prefix, sizeof(BTF_STRUCT) - 1);
 426                 /* union  */
 427                 } else if (!strncmp(prefix, BTF_UNION, sizeof(BTF_UNION) - 1)) {
 428                         obj->nr_unions++;
 429                         id = add_symbol(&obj->unions, prefix, sizeof(BTF_UNION) - 1);
 430                 /* typedef */
 431                 } else if (!strncmp(prefix, BTF_TYPEDEF, sizeof(BTF_TYPEDEF) - 1)) {
 432                         obj->nr_typedefs++;
 433                         id = add_symbol(&obj->typedefs, prefix, sizeof(BTF_TYPEDEF) - 1);
 434                 /* func */
 435                 } else if (!strncmp(prefix, BTF_FUNC, sizeof(BTF_FUNC) - 1)) {
 436                         obj->nr_funcs++;
 437                         id = add_symbol(&obj->funcs, prefix, sizeof(BTF_FUNC) - 1);
 438                 /* set */
 439                 } else if (!strncmp(prefix, BTF_SET, sizeof(BTF_SET) - 1)) {
 440                         id = add_set(obj, prefix);
 441                         /*
 442                          * SET objects store list's count, which is encoded
 443                          * in symbol's size, together with 'cnt' field hence
 444                          * that - 1.
 445                          */
 446                         if (id)
 447                                 id->cnt = sym.st_size / sizeof(int) - 1;
 448                 } else {
 449                         pr_err("FAILED unsupported prefix %s\n", prefix);
 450                         return -1;
 451                 }
 452
 453                 if (!id)
 454                         return -ENOMEM;
 455
 456                 if (id->addr_cnt >= ADDR_CNT) {
 457                         pr_err("FAILED symbol %s crossed the number of allowed lists",
 458                                 id->name);
 459                         return -1;
 460                 }
 461                 id->addr[id->addr_cnt++] = sym.st_value;
 462         }
 463
 464         return 0;
 465 }
 466
 467 static int symbols_resolve(struct object *obj)
 468 {
 469         int nr_typedefs = obj->nr_typedefs;
 470         int nr_structs  = obj->nr_structs;
 471         int nr_unions   = obj->nr_unions;
 472         int nr_funcs    = obj->nr_funcs;
 473         int err, type_id;
 474         struct btf *btf;
 475         __u32 nr;
 476
 477         btf = btf__parse(obj->btf ?: obj->path, NULL);
 478         err = libbpf_get_error(btf);
 479         if (err) {
 480                 pr_err("FAILED: load BTF from %s: %s",
 481                         obj->path, strerror(err));
 482                 return -1;
 483         }
 484
 485         err = -1;
 486         nr  = btf__get_nr_types(btf);
 487
 488         /*
 489          * Iterate all the BTF types and search for collected symbol IDs.
 490          */
 491         for (type_id = 1; type_id <= nr; type_id++) {
 492                 const struct btf_type *type;
 493                 struct rb_root *root;
 494                 struct btf_id *id;
 495                 const char *str;
 496                 int *nr;
 497
 498                 type = btf__type_by_id(btf, type_id);
 499                 if (!type) {
 500                         pr_err("FAILED: malformed BTF, can't resolve type for ID %d\n",
 501                                 type_id);
 502                         goto out;
 503                 }
 504
 505                 if (btf_is_func(type) && nr_funcs) {
 506                         nr   = &nr_funcs;
 507                         root = &obj->funcs;
 508                 } else if (btf_is_struct(type) && nr_structs) {
 509                         nr   = &nr_structs;
 510                         root = &obj->structs;
 511                 } else if (btf_is_union(type) && nr_unions) {
 512                         nr   = &nr_unions;
 513                         root = &obj->unions;
 514                 } else if (btf_is_typedef(type) && nr_typedefs) {
 515                         nr   = &nr_typedefs;
 516                         root = &obj->typedefs;
 517                 } else
 518                         continue;
 519
 520                 str = btf__name_by_offset(btf, type->name_off);
 521                 if (!str) {
 522                         pr_err("FAILED: malformed BTF, can't resolve name for ID %d\n",
 523                                 type_id);
 524                         goto out;
 525                 }
 526
 527                 id = btf_id__find(root, str);
 528                 if (id) {
 529                         id->id = type_id;
 530                         (*nr)--;
 531                 }
 532         }
 533
 534         err = 0;
 535 out:
 536         btf__free(btf);
 537         return err;
 538 }
 539
 540 static int id_patch(struct object *obj, struct btf_id *id)
 541 {
 542         Elf_Data *data = obj->efile.idlist;
 543         int *ptr = data->d_buf;
 544         int i;
 545
 546         if (!id->id) {
 547                 pr_err("FAILED unresolved symbol %s\n", id->name);
 548                 return -EINVAL;
 549         }
 550
 551         for (i = 0; i < id->addr_cnt; i++) {
 552                 unsigned long addr = id->addr[i];
 553                 unsigned long idx = addr - obj->efile.idlist_addr;
 554
 555                 pr_debug("patching addr %5lu: ID %7d [%s]\n",
 556                          idx, id->id, id->name);
 557
 558                 if (idx >= data->d_size) {
 559                         pr_err("FAILED patching index %lu out of bounds %lu\n",
 560                                 idx, data->d_size);
 561                         return -1;
 562                 }
 563
 564                 idx = idx / sizeof(int);
 565                 ptr[idx] = id->id;
 566         }
 567
 568         return 0;
 569 }
 570
 571 static int __symbols_patch(struct object *obj, struct rb_root *root)
 572 {
 573         struct rb_node *next;
 574         struct btf_id *id;
 575
 576         next = rb_first(root);
 577         while (next) {
 578                 id = rb_entry(next, struct btf_id, rb_node);
 579
 580                 if (id_patch(obj, id))
 581                         return -1;
 582
 583                 next = rb_next(next);
 584         }
 585         return 0;
 586 }
 587
 588 static int cmp_id(const void *pa, const void *pb)
 589 {
 590         const int *a = pa, *b = pb;
 591
 592         return *a - *b;
 593 }
 594
 595 static int sets_patch(struct object *obj)
 596 {
 597         Elf_Data *data = obj->efile.idlist;
 598         int *ptr = data->d_buf;
 599         struct rb_node *next;
 600
 601         next = rb_first(&obj->sets);
 602         while (next) {
 603                 unsigned long addr, idx;
 604                 struct btf_id *id;
 605                 int *base;
 606                 int cnt;
 607
 608                 id   = rb_entry(next, struct btf_id, rb_node);
 609                 addr = id->addr[0];
 610                 idx  = addr - obj->efile.idlist_addr;
 611
 612                 /* sets are unique */
 613                 if (id->addr_cnt != 1) {
 614                         pr_err("FAILED malformed data for set '%s'\n",
 615                                 id->name);
 616                         return -1;
 617                 }
 618
 619                 idx = idx / sizeof(int);
 620                 base = &ptr[idx] + 1;
 621                 cnt = ptr[idx];
 622
 623                 pr_debug("sorting  addr %5lu: cnt %6d [%s]\n",
 624                          (idx + 1) * sizeof(int), cnt, id->name);
 625
 626                 qsort(base, cnt, sizeof(int), cmp_id);
 627
 628                 next = rb_next(next);
 629         }
 630         return 0;
 631 }
 632
 633 static int symbols_patch(struct object *obj)
 634 {
 635         int err;
 636
 637         if (__symbols_patch(obj, &obj->structs)  ||
 638             __symbols_patch(obj, &obj->unions)   ||
 639             __symbols_patch(obj, &obj->typedefs) ||
 640             __symbols_patch(obj, &obj->funcs)    ||
 641             __symbols_patch(obj, &obj->sets))
 642                 return -1;
 643
 644         if (sets_patch(obj))
 645                 return -1;
 646
 647         elf_flagdata(obj->efile.idlist, ELF_C_SET, ELF_F_DIRTY);
 648
 649         err = elf_update(obj->efile.elf, ELF_C_WRITE);
 650         if (err < 0) {
 651                 pr_err("FAILED elf_update(WRITE): %s\n",
 652                         elf_errmsg(-1));
 653         }
 654
 655         pr_debug("update %s for %s\n",
 656                  err >= 0 ? "ok" : "failed", obj->path);
 657         return err < 0 ? -1 : 0;
 658 }
 659
 660 static const char * const resolve_btfids_usage[] = {
 661         "resolve_btfids [<options>] <ELF object>",
 662         NULL
 663 };
 664
 665 int main(int argc, const char **argv)
 666 {
 667         bool no_fail = false;
 668         struct object obj = {
 669                 .efile = {
 670                         .idlist_shndx  = -1,
 671                         .symbols_shndx = -1,
 672                 },
 673                 .structs  = RB_ROOT,
 674                 .unions   = RB_ROOT,
 675                 .typedefs = RB_ROOT,
 676                 .funcs    = RB_ROOT,
 677                 .sets     = RB_ROOT,
 678         };
 679         struct option btfid_options[] = {
 680                 OPT_INCR('v', "verbose", &verbose,
 681                          "be more verbose (show errors, etc)"),
 682                 OPT_STRING(0, "btf", &obj.btf, "BTF data",
 683                            "BTF data"),
 684                 OPT_BOOLEAN(0, "no-fail", &no_fail,
 685                            "do not fail if " BTF_IDS_SECTION " section is not found"),
 686                 OPT_END()
 687         };
 688         int err = -1;
 689
 690         argc = parse_options(argc, argv, btfid_options, resolve_btfids_usage,
 691                              PARSE_OPT_STOP_AT_NON_OPTION);
 692         if (argc != 1)
 693                 usage_with_options(resolve_btfids_usage, btfid_options);
 694
 695         obj.path = argv[0];
 696
 697         if (elf_collect(&obj))
 698                 goto out;
 699
 700         /*
 701          * We did not find .BTF_ids section or symbols section,
 702          * nothing to do..
 703          */
 704         if (obj.efile.idlist_shndx == -1 ||
 705             obj.efile.symbols_shndx == -1) {
 706                 if (no_fail)
 707                         return 0;
 708                 pr_err("FAILED to find needed sections\n");
 709                 return -1;
 710         }
 711
 712         if (symbols_collect(&obj))
 713                 goto out;
 714
 715         if (symbols_resolve(&obj))
 716                 goto out;
 717
 718         if (symbols_patch(&obj))
 719                 goto out;
 720
 721         err = 0;
 722 out:
 723         if (obj.efile.elf)
 724                 elf_end(obj.efile.elf);
 725         close(obj.efile.fd);
 726         return err;
 727 }